Laboratory Directed Research and Development Annual Report for 2012

Home , Perfluorodecalin, Tin selenide

DISCLAIMER This report was prepared as an account of work sponsored by an agency of the United States Government. Neither the United States Government nor any agency thereof, nor Battelle Memorial Institute, nor any of their employees, makes any warranty, express or implied, or assumes any legal liability or responsibility for the accuracy, completeness, or usefulness of any information, apparatus, product, or process disclosed, or represents that its use would not infringe privately owned rights. Reference herein to any specific commercial product, process, or service by trade name, trademark, manufacturer, or otherwise does not necessarily constitute or imply its endorse- ment, recommendation, or favoring by the United States Government or any agency thereof, or Battelle Memorial Institute. The views and opinions of authors expressed herein do not necessarily state or reflect those of the United States Government or any agency thereof. PACIFIC NORTHWEST NATIONAL LABORATORY operated by BATTELLE for the UNITED STATES DEPARTMENT OF ENERGY under Contract DE-AC05-76RL01830

Printed in the United States of America

Available to DOE and DOE contractors from the Office of Scientific and Technical Information, P.O. Box 62, Oak Ridge, TN 37831-0062; ph: (865) 576-8401 fax: (865) 576-5728 email: [email protected]

Available to the public from the National Technical Information Service, U.S. Department of Commerce, 5285 Port Royal Rd., Springfield, VA 22161 ph: (800) 553-6847 fax: (703) 605-6900 email: [email protected] online ordering: http://www.ntis.gov/ordering.htm

This document was printed on recycled paper. PNNL-22324

Laboratory Directored Research and Development Annual Report

Fiscal Year 2012

March 2013

Prepared for the U.S. Department of Energy under Contract DE-AC05-76RL01830

Pacific Northwest National Laboratory Richland, Washington 99352 Contents Laboratory Director’s Message vii

Advanced Sensors and Instrumentation Advanced Environmental Sampling Technology for Safeguards and Proliferation Detection ...... 1 Advanced Surface Mass Spectrometry for Characterization of Explosives 3 Development of a Low-Cost Reliable Barometric Air Flow Station for Air-Handling Units 6 Forensic Analysis with Time Stamped Coincidence Sampled Gamma Radiation Spectroscopy 7 Glider-Based Passive Acoustic Detection System Capability Development ...... 8 In Situ High-Pressure X-Ray Diffraction Investigation of Caprock Mineral Reactions With Water Solvated

in Supercritical CO2 ...... 9 Low Frequency Electromagnetic Interrogation Techniques for Container Content Signature Detection 11 Low-Noise, Low-Background Electronics ...... 13 Neutron Signature Detection Requirements for Identifying Warhead Configurations ...... 14 Novel Detector Configurations 15 Optical Upconversion for Passive and Active Millimeter and Terahertz Imaging 16 Prototype Fast Neutron Detector ...... 18

Biological Sciences Analysis of Functional Potential from Metagenome Data 20 Aspergillus niger as a Platform for Exploitation of the Advanced Biofuel Producing Potential of Filamentous Fungi . . 21 Automated Thermal Image Processing to Identify and Enumerate Targets of Interest 23 Characterization of Signaling Networks in Single Cells ...... 24 Combining Proteomic Technologies to Create a Platform for Spatiotemporal Enzyme Activity Profiling ...... 25 Community Diversity and Functional Redundancy of Cellulytic Microbial Communities in Soil Aggregates 27 Correlative High Resolution Imaging and Spectroscopy to Characterize the Structure and Biogeochemical Function of Microbial Biofilms ...... 28 Development of Functionalized Nanoporous Materials for Bioenergy and Biomedical Applications ...... 30 Directed Strain Evolution through Riboswitch-Controlled Regulatory Circuit ...... 31 Enhanced Bioremediation of Uranium Contamination at DOE Sites Via Novel Directed Evolution of Uranium Specific Reductase on Bacteriophage Surface 32 Higher-Throughput, More Sensitive Stable Isotope Probing ...... 33 Identification of Functional Proteins Relevant to Bioenergy and Disease Pathology by Multiplexed Activity-Based Protein Profiling ...... 35 Integrated Nano-Scale Imaging for Investigating Applications and Implications of Nanomaterials ...... 36 Mapping and Characterization of Organic Matter in Soil Aggregates using Laser-Ablation Sampling ...... 38 Massively Parallel Sequencing Technology for the Forensic Identification of the “Unknown” Biological Threat Agent(s) Recovered from the Biological Crime Scene ...... 39

Microbial Processes Accompanying Deep Geologic CO2 Sequestration ...... 41 Micro-Fluidic Models for Studying Microbial Communities—Integration of Micro-Fluidic Model Experimentation, Multimodal Imaging, and Modeling ...... 42 Microscale Spectroscopic Analyses of Cellulose Degradation and Uptake by a Microbial Community 44 Mining the Data from Research on Dogs Exposed to Internally-deposited Radionuclides ...... 45 Multiscale Simulation of Microbial Carbon Transformation in Soils: Connecting Intra- and Inter-Aggregate Scales 46 Next Generation Software for Automated Structural Identification of Metabolites 47 Optofluidics and Microfluidics for Exploring Biofuel Production at the Single Cell and Molecule Levels ...... 48 Predicting the Response of Complex Biological Systems ...... 49 Proteomics Measurements of Functional Redundancy and Stability Testing of Cellulose Degrading Anaerobic Microbial Communities Within Engineered Bioreactors ...... 50 2012 Annual Report ii Laboratory Directed Research & Development Pulmonary Injury from Acute Events Related to Nuclear Energy Production 52 Statistically Significant Forensic Fingerprinting: Protein Analysis of Biological Agents 53 Understanding the Processes that Govern Subsurface Microbial Communities 54

Chemistry

Biodegradable Linear Polycarbonates Through Catalytic Polymerizations of CO2 Utilizing Switchable Ionic Liquids . . 56 Characterization of Catalyst Materials in the Electron and Atom-Probe Microscopes ...... 58 Characterization of Energy Storage Systems Using In Situ and Ex Situ Nuclear Magnetic Resonance Spectroscopy . . . 60 Chemical Imaging Analysis of Environmental Particles ...... 61

CO2 Separation Scale-Up ...... 62 Conversion of Biomass to Jet Fuels ...... 64 Development and Deployment of Reactive Surface Area Measurement Capabilities Using Nuclear Magnetic Resonance Spectroscopy ...... 65 Development of Bifunctional Electrocatalysts for Rechargeable Lithium-Air Batteries 66 Development of Inorganic Water Oxidation Electrocatalysts ...... 67 Development of New Soft Ionization Mass Spectrometry Approaches for Spatial Imaging of Complex Chemical and Biological Systems ...... 68 Development of Preparative Mass Spectrometry for the Creation of Novel Catalyst Materials 69 Environmental Chamber Studies of the Interactions of Anthropogenic and Biogenic VOCs ...... 70 Exploitation of Kinetic Processes in Gas Separations ...... 72 Fundamentals of Carbonate Formation: Interactions of Carbon Dioxide with Supported Metal Oxide Clusters 73

In Situ Molecular-Scale Investigations of Reactions between Supercritical CO2 and Minerals Relevant to Geological Carbon Storage 74 Increased Sensitivity and Improved Quantification of Th and U in Particles by SIMS ...... 75 Light Source Photocathode Performance and Development ...... 76 Molecular Structure and Interaction at Aqueous, Non-Aqueous Liquid Interfaces and Catalytic Solid Surfaces 77 Multiscale Synthetic Studies Targeted Towards the Development of Nanostructured

Heterogeneous CO2 Reduction Catalysts ...... 79 Non-Metal Activation of Hydrogen for Energy Storage in Chemical Bonds ...... 80 OnSpec Advanced Condition Monitoring and Fundamental Research of Aggregation Precursors in Nuclear Materials Reprocessing ...... 81 Oxygen Optode for Chemical Imaging in Microfluidic Microbial Models ...... 82 Probing Composition and Structure of Polarizable Reaction Mixtures Inside the Pores of Supported Metal-oxo Catalysts ...... 84 Probing Structural Dynamics with High Spatial and Temporal Resolution ...... 85 Quantitative Imaging of Atomic Scale Chemistry Changes at Interfaces ...... 86 The Statistical Mechanics of Complex Process in Bulk and Interfacial Environments ...... 88

Earth and Space Sciences Computational Framework for Diagnostics, Validation and Intercomparison of Numerical Simulators for Geologic Sequestration ...... 90 Decision Support Research for Integrated Regional Earth System Modeling ...... 91 Development of Coupled Flow, Thermal and Geomechanical Capability for Carbon Sequestration ...... 92 Development of Prototype Integrated Earth System and Environmental System Models ...... 93 Development of Regional-Global Climate Assessment Model 95 Exploration of Human and Environmental System Interactions due to Renewable Technology Penetration in the Midwest Pilot Region ...... 96 Integrated Regional Earth System Model (iRESM) Prototype Regional Testbed Specification and Selection ...... 98

Micromodel Pore-Scale Studies of Caprock-Sealing Efficiency and Trapping Mechanisms Related to CO2 Sequestration .100 Microscale Reconstruction of Biogeochemical Substrates Using Combined X-ray Tomography and Scanning Electron Microscopy 102 2012 Annual Report

iii Laboratory Directed Research & Development Non-destructive Challenge Measurements of Environmental Radiotracers 103 Numerically Robust Climate Simulation Through Improved Interaction between Model Components ...... 104

Predicting the Feasibility of Geologic Co-Sequestration of CO2, SOx and NOx Under a Broad Range of Conditions . . .105 Quantitative Framework Development to Support Effective Monitoring for Metal and Radionuclide Contamination . 107 Simultaneous 14C and T Dating: A Case Study Using Soil Organic Matter ...... 108 Tank Residual Waste Stabilization to Reduce Contaminant Release ...... 109 Translating Microbial Ecology Advances into Applied Tools for Remediation and Monitoring 110 Uncertainty Quantification and Risk Assessment Pipeline for Carbon Sequestration 111 Understanding the Sources and Consequences of Uncertainties ...... 112

Energy Supply and Use A Multi-Layer Data-Driven Advanced Reasoning Tool for Smart Grid Integrated Information Systems ...... 115 A Statistical State Prediction Methodology to Improve Reliability and Efficiency of Power System Operation . . . . 117 Cooperative Control of Distributed Energy Resources for Grid Support 118 Decision Support for Future Power Grid Organizations ...... 119 Development of a Regional Energy and Infrastructure Systems Framework ...... 120 Development of Rechargeable Li/air Batteries ...... 122 Development of Regional Agriculture-Land Use Models ...... 123 Forward Osmosis System for Integrated Desalination and Carbon Dioxide Separation ...... 124 Future Power Grid Control Paradigm ...... 125 Geological Sequestration Software Suite: Numerical Model Development ...... 126 High-Capacity Reversible Metal Hydride Air Battery ...... 128 Modeling of Distributed Energy Resources in the Smart Grid 129 Novel Carbon Capture Materials 130 Novel In-Situ Monitoring Framework for Environmental Degradation of Reactor Structural and Fuel Materials . . . 132 Quantify the “State of Health” of Nuclear Structure/Materials with an Inverse Resonance Inspection Algorithm 133 Restructuring FEDS for Continued Growth and Expanding Markets: GUI Update 136 Stabilized Li Metal Anode for Li Batteries ...... 137 Stable and High Voltage Electrolytes for Lithium Ion Batteries 138

Engineering and Manufacturing Processes Development of Functional Membranes for Improved Selectivity ...... 140

Materials Science and Technology Advanced Sorptive and Signature Indicating Materials for Ultra-Trace Proliferation Detection ...... 143 Development of Graphene/Ionic Liquid Hybrid Material for Ultracapacitors 145 Development of Hard X-Ray Emission Spectroscopy Nanoprobe ...... 146 Facet Specific Chemistry of Noble Metal Nanoparticles Using an Enhanced Scattering Infrared Scattering Near-Field Optical Microscope ...... 147 Hybrid Electrodes for Next Generation High Energy Ultracapacitors ...... 148 Imaging the Nucleation and Growth of Nanoparticles in Solution ...... 149 Integrating Multimodal Chemical Imaging Instrumentation by Data Reduction and Resolution Merge ...... 150 Integration of Molecular Imaging Techniques to Probe the Photoinduced Charge Transfer in Semiconductor Quantum Dots-Polymer Hybrid Solar Cells ...... 151 Materials and Methods for Low Cost Photovoltaic Manufacturing ...... 152 Meta-Material Coatings for Daylighting Windows ...... 153 Multiphysics Capability Development and Application to Magnesium Alloys ...... 154 Multiscale Modeling of Energy Storage Materials ...... 155 Photocathode Development for Next-Generation Light Sources ...... 157

2012 Annual Report Rare Earth-less Permanent Magnet ...... 158 iv Laboratory Directed Research & Development Site Specific Atomic Resolution Probing of Structure-Property Relationship Under Dynamic and/or Operando Conditions Using In Situ and Ex Situ Chemical Imaging Based on Multi-Instrument Approach 159 Three-Dimensional Structured Composite Nanomaterials for Energy Storage 161

Mathematics and Computing Sciences A Distributed Systems Architecture for the Power Grid 163 A Multi-Modal Integration Framework for Chemical Imaging 164 A Scalable Fault Tolerance Infrastructure and Algorithms with Programming Models and Scientific Applications 166 Actionable Visualization Tools For Power Grid Situation Awareness 168 Advanced Optimizations for Extreme-Scale Homogeneous Systems 169 Advanced Statistical Network Models for the Integration of Experimental and Open Source Textual Data for Bioforensic Analyses 171 Compressive Sensing for Threat Detection 172 Computational Optimization and Predictive Simulation for Synthetic Biology 173 Cyber Security Experimentation and Repeatability 174 Cyber/Physical Security Vulnerability Assessment Integration 175 Data-Intensive Algorithms for Bioinformatics-Inspired Signal Detection 176 Developing Functionality and Performance Enhancements to the Global Array Toolkit 177 Development of Parallel Multi-Reference Coupled Cluster Capabilities 178 Enabling Hypothesis Driven Research and Discovery in Extreme Data 179 Enabling the Meaningful Exploitation of Integrated Regional Earth Systems Data 180 Encrypted CPU Instruction Stream 183 Experts Inundated with Data: the Biomarker Problem 184 Extreme Scale Capabilities for Subsurface Science and Engineering 185 Geological Sequestration Software Suite Framework 186 GRADIENT: Graph Analytic Approach for Discovering Irregular Events - Nascent and Temporal 187 Identifying Cyber Assets and Inferring Criticality 188 Immense Social Media Stream Analytics 189 Integrating Advanced Optimizations for Extreme Scale Systems 190 Integrating Power and Performance Modeling for Exascale Systems 191 Intelligent Networked Sensors Capable of Autonomous, Adaptive Operations in the Electric Power System 194 Linear Algebra Solvers and Associated Matrix-Vector Kernels for Power Grid Simulations 196 Mapping Molecular Dynamics Algorithmic Parallelism to Heterogeneous Architectures 197 Multimedia Analysis of Cyber Data 198 Multi-Resolution Data Model and Directed Data Reduction, Reconstruction and Aggregation 199 Multiscale Models for Microbial Communities 201 Multi-Source Signatures of Nuclear Programs 202 Nanoscale-Macroscale Three-Dimensional Integration Using High Performance Computing 203 Next Generation Network Simulations for Power System Applications 205 Real-time High-Performance Computing Infrastructure for Next-Generation Power Grid Analysis 207 Scalable Sensor Data Management Middleware 208 Semantic Workflows for Signature Discovery 210 Signature Discovery Analytic Framework 211 Signature Quality Metrics 212 Single Node Optimizations for Extreme Scale Systems 213 SoftShuffle: A Game Changer for Secure Software System Implementation 215 Synergistic Integration of Feature Recognition and Analysis for Chemical Imaging Data 216 Targeting Extreme Scale Computational Challenges with Heterogeneous Systems 217 TASCEL: An Execution Model for Task-Based Optimizations 218

Visualizing Uncertainty in Conceptual and Numerical Models for Geological Sequestration 219 2012 Annual Report

v Laboratory Directed Research & Development Nuclear Science and Engineering A Virtual Testing Toolbox for Predicting the Properties and Behavior of Multiphase Materials in Disposal Environments ...... 221 Advanced Condition Monitoring of Microstructure Evolution and Property Degradation in Current and Advanced Nuclear Reactor Alloys ...... 223 Advanced Nondestructive Assay for Safeguards 224 Argon-39 Measurement ...... 226 Background Reduction for Nonmetallic Nuclear Detection Components 227 BazookaSPECT Neutron Imager ...... 228 Characterization of a 14 MeV Neutron Generator and Measurement of Fission Products Produced 229 Exploiting Correlated Radiographic and Passive Signatures for Threat Detection in Cargo 231 Isotopic Ratio Fluence Monitors for Canadian Deuterium Uranium (CANDU) and Pebble Bed Modular Reactor (PBMR) Plutonium Production Verification 232 Novel N/gamma Flux Monitoring Materials for Safeguards and Proliferation Detection 233 Three Dimensional Neutronics Analysis Capabilities for Nuclear Archeology Applications 234 Ultra-Low-Background Gas Measurement: Building an Advanced Capability 235 Uranium Enrichment Facility Signature Exploitation ...... 236

Physics Absolute Gas Counting Measurement Techniques 238 Highly Efficient and Cost Effective Gamma Detection Arrays for FRIB ...... 239 Low Energy Threshold Germanium Detectors and Science ...... 240 Mu2e September Beam Test 241 Nuclear Astrophysics Measurements using High-efficiency HPGe Arrays ...... 242 Search for New Physics at the Intensity Frontier ...... 243 Ultra-Precise Electron Spectroscopy to Measure the Neutrino Mass ...... 244 2012 Annual Report vi Laboratory Directed Research & Development Laboratory Director’s Message

The mission of Pacific Northwest National Laboratory is to transform the world through courageous discovery and innovation. Every day, PNNL scientists and engineers set out to fulfill the promise of that mission by advancing scientific discovery and delivering critical solutions to the challenges that face our nation. The science and technology we work with, create, and improve upon inspires and enables the world to live prosperously, safely, and securely. Our investments in Laboratory Directed Research and Development are essential to our ability to realize this mission and advance our strategy. LDRD investments help us nurture science and technology capabilities while capitalizing on the breadth and depth of talent which our staff possess. This report describes how we conduct our LDRD program in compliance with DOE objectives and guidelines. We use rigorous internal and external peer review to maintain the scientific value and soundness of the research enabled by our LDRD program. With great pride in our researchers’ accomplishments, I present PNNL’s Fiscal Year 2012 Laboratory Directed Research and Development Annual Report.

Michael Kluse Director, PNNL 2012 Annual Report

vii Laboratory Directed Research & Development Advanced Sensors and Instrumentation Advanced Environmental Sampling Technology for Safeguards and Proliferation Detection Norman C. Anheier, Jr. femtogram levels and with abundance sensitivity below The overall goal of this project is to produce that achievable with conventional mass spectrometry improved sample collection and analysis (this task was completed in FY 2009). technologies for detecting proliferation The unique selectivity provided by high-resolution laser signatures and to verify facility declarations technique is expected to reduce or remove complex sample by making determinations about the preparation chemistry substantially, resulting in a faster absence of undeclared nuclear activity. analysis time. Finally, we will evaluate laser ablation, absorbance ratio spectrometry (LAARS) for undeclared enriched uranium detection and bias defect measurement Environmental sampling is a critical verification tool for within uranium enrichment plants. An automated, unat- implementing safeguards and detecting proliferation activ- tended environmental aerosol sample collection approach ities. Under formal agreements, safeguards inspectors are will be developed and combined with the uranium isotope granted access to nuclear sites to collect samples. Prolifera- ratio analysis system to detect enrichment facility misuse tion detection and verification of peaceful nuclear activi- in a timely manner. ties are critical national security mission activities, Continuing from FY 2009, progress on the first project task underpinned by sophisticated technology that requires included evaluating additional new materials for increas- continuous capability improvements as the existing threats ing the collection efficiencies of key analytes of interest grow and new threats emerge. The expected outcomes of that are relevant to safeguards and proliferation detection. this research are to develop and evaluate new environmen- A polymer synthesis laboratory capability was set up, and tal sample and destructive assay (DA) collection technolo- several batches of a new methacrylic copolymer were gies to improve collection efficiency, significantly reduce synthesized and evaluated for their ability to adhere partic- the time and cost associated with subsequent laboratory ulates. Additionally, several commercially available con- analysis, and to demonstrate that PNNL-developed Laser ductive polymers were evaluated using SIMS to determine Ablation, Absorbance Ratio Spectrometry can be used in a which one(s) would be ideal for direct sample analysis purpose built configuration to perform unattended near (based on degree of conductivity and strength of the real-time monitoring of large enrichment facilities. analytical signal, processed by the SIMS instrument). Our project approach is to develop advanced sample col- It was determined that polyethylenedioxothiophene- lection, sample analysis, and automated, unattended mon- poly(styrene) sulfonate (PEDOT-PSS) performed optimally. itoring techniques. The project consists of three main tasks. In FY 2012, we continued to develop a Raman and fluores- First, we will develop improved sample collection methods cence spectroscopic reference library for uranium contain- and technologies for nuclear facility inspectors. Analyte- ing compounds using the optical microscopies developed specific conductive films will be developed that can directly in FY 2011. A series of both be deposited onto secondary uranium containing ore sam- ion mass spectrometry (SIMS) ples (standard reference substrates, with the goal of material purchased from significantly reducing the cost New Brunswick National and time associated with lab- Laboratory and the Canada oratory analyses. Next, we will Centre for Mineral and develop and evaluate a high- Energy Technology) as well resolution isotopic finger- as synthetic uranium con- printing system for rapid taining compounds were screening of minor uranium The LAARS-ES instrument used to analyze uranium enrichment acquired. The acquired isotopes (234, 235, and 236) in collected environmental samples. After sample collection, the samples were analyzed stand-alone environmental aerosol sampling system is inserted at ultratrace levels. It is using BET gas sorption expected that minor isotopes into the LAARS-ES instrument for rapid uranium particulate sample assay. The LAARS-ES instrument can be designed for deployment analysis to determine can be directly quantified at within a standard IAEA instrument rack. particle size distributions, PN09005/2173

1 Advanced Sensors and Instrumentation XRD for phase identification and quantification, Raman spectroscopy and fluorescence spectroscopy. The developed spectral library was used to demonstrate our ability to identify uranium compounds in mixed samples. Under our third task, we evaluated the feasibility to adapt the PNNL-developed LAARS into a uranium enrichment facility safeguards technology. During FY 2012, the LAARS- ES deployment scenario was updated to feature a stand- The LAARS-DA sampling assembly shown attached to a alone aerosol collector based on a rotating drum impactor conventional gas centrifuge enrichment plant (GCEP) sampling (RDI). LAARS-ES aerosol collectors are secured in tamper- tap. The sampling assembly is based on a modified commercial indicating enclosures and staged at strategic locations vacuum ball valve. throughout the GCEP. The collector drums are then retrieved by inspectors during onsite inspections and Studies were conducted on thin film NaF sorbent materials that were demonstrated to have excellent UF chemical inserted into the LAARS-ES instrument for immediate assay. 6 trapping properties. LAARS-DA measurements produced This year, LAARS-ES experimental studies characterized pre- strong absorbance signals and excellent abundance pared uranium particle samples. This study evaluated the accuracy on a sample that contained only 5–15 µg effects of heavy sample loading and particle retention on total uranium. the surface of the substrate. The RDI nozzle was redesigned to provide laminar flow within the nozzle and uniform The Task 1 Raman and fluorescence uranium study velocity across the nozzle exit, with the goal of improving resulted in two papers published in FY 2012. Significant small particle capture efficiency. The particle size distribu- Task 3 technical progress was reported through several tion captured by impaction is determined by the flow rate publications and presentations, including the following: through the RDI and by the impactor nozzle outlet width 1) two papers were prepared and presentations delivered rd and the gap between the rotating drum and nozzle. Simu- to the 53 Annual INMM Meeting in Orlando, FL, 2) an lation results, using a 0.25-mm nozzle width and 2-mm invited papers was published in the Journal of Nuclear gap between the nozzle and drum, suggested > 90% cap- Material Management, and 3) a technology brief titled “NIS ture efficiency down to 100 nm particle diameter at Uses Cutting-Edge Science and Technology to Advance the 20 LPM. The RDI performance was characterized using U.S. Nonproliferation Mission” was published in the NNSA hematite (density 5.3 g/cm3) aerosols having a 2.3 µm size Office of Nonproliferation and International Security High- distribution mean. The particle capture efficiency was lights publication. characterized using an aerodynamic particle sizer to measure the RDI inlet and outlet particle size distributions. The collection efficiency for submicron particles is significantly improved (>65% at 0.5 µm) compared to the previous design. A new LAARS application, LAARS-DA, was also evaluated. LAARS-DA offers on-site enrichment DA sample collection and bias defect measurement. DA samples are collected using a custom sampling device attached to a conventional sample tap. A chemisorbent film captures a few micro-

grams of gaseous UF6 in just minutes. The collected DA sample is then analyzed onsite using the LAARS-DA instrument. DA results are determined in a matter of minutes at sufficient accuracy to support reliable bias defect conclusions while greatly reducing DA sample volume, analysis time, and cost. PN09005/2173

2 Advanced Sensors and Instrumentation Advanced Surface Mass Spectrometry for Characterization of Explosives Christine Mahoney transform ion cyclotron resonance mass spectrometry Our government is always in need of precise (FT-ICR-MS), nano-desorption electrospray ionization mass tools that can rapidly and accurately charac- spectrometry (nano-DESI), inductively-coupled plasma mass spectrometry (ICP-MS), and high-throughput particle terize explosive compositions. At PNNL’s analysis with an ASPEX secondary electron microscope Environmental Molecular Sciences Labora- (SEM) and energy dispersive x-ray (EDX) spectroscopy. As a tory (EMSL), we have state-of-the-art facili- starting point, we obtained a series of composition C-4 ties containing a unique suite of analytical plastic explosive samples from across the country, through tools, some of which are not located any- collaborations with the Bureau of Alcohol, Tobacco, Fire- where else in the world. Developing these arms and Explosives. These samples are commercially facilities and tools for national security available, and have been characterized extensively using interests will help our government both GC-MS in previous work. The goal was then to use our suite of advanced metrology tools for detailed characteriza- prevent and respond to terrorist acts in a tion and differentiation of the C-4 samples based on a com- timely manner. bination of unique organic signatures and trace elemental compositions. We intend PNNL to be a center of excellence in addressing Composition C-4. Composition C-4 plastic explosive is com- problems relevant to national security. As such, we are cur- prised of ~90% RDX, which serves as the active explosive rently developing several novel methods available only ingredient in sample. However, C-4 also contains many through EMSL, a state-of-the-art PNNL instrument user other additives, such as a poly(isobutylene; (PIB) binder, facility, for national security applications. Previous results process oils, di-isoctylsebacate (DOS), or di-capryl adipate indicated that signatures of C-4 samples from around the (DCA) plasticizers, HMX explosive bi-products, and many world could be distinguished from one another using imag- contaminants and/or species that may be unique to a par- ing mass spectrometry. This study highlighted time-of-flight ticular region, source, or lot. secondary ion mass spectrometry (ToF-SIMS) as a powerful C-4 Sample Preparation and Method Development. Both chemical imaging forensics tool for direct characterization bulk analysis and imaging microanalysis methods were and differentiation of explosive components (trace and developed for characterization of 17 different C-4 samples, bulk). It was shown that ToF-SIMS could readily differenti- where sample preparation methods were optimized for ate between C-4 samples of varying origin based on their each method utilized: explosive and additive compositions. However, only 3 samples were analyzed in this work. Continuing on this theme, • Bulk Characterization Methods (e.g., ESI-MS and ICP- 17 well-characterized composition C-4 explosive samples MS): The entire material was dissolved in a solvent, such were obtained and used as a model system for method as THF or acetonitrile before analysis. development and advanced statistical analysis using a • Imaging Microanalysis Methods (e.g., ToF-SIMS and Bayesian integrated statistical analysis approach. Prelimi- nano-DESI): A nanoimprinter was used to dry transfer nary results indicate small compositional variations material directly to a Si wafer substrate by serving as a between C-4 samples can potentially be used in combina- constant pressure stamper. Solution cast films were also tion with robust data analysis procedures for categorization prepared by depositing 50 µl of 2% C-4 solution in THF of C-4 samples. Most importantly, an infrastructure has onto 1.5 × 1.5 cm substrates. been developed for future collaborations between EMSL After preparation, samples were distributed to various and the National Security Directorate (NSD) at PNNL, a experts for method development. A full data set was relationship particularly beneficial for the integrity of our obtained from all instrumentation utilized. national security. C-4 Characterization of Organic Constituents. We can We are currently developing several novel instrumentations organize the methods described above into organic analy- both at EMSL and throughout PNNL for simultaneous ele- sis methods (such as ToF-SIMS, ESI-MS XPS and nano-DESI), mental and molecular characterization of materials of and elemental/isotopic analysis methods (such as EDX, interest to national security. These methods include ToF- magnetic sector SIMS and ICP-MS). For FY 2012, most of our SIMS, x-ray photoelectron spectroscopy (XPS), Fourier PN12007/2408

3 Advanced Sensors and Instrumentation focus has been on organic characterization and analysis methods. However, methods for inorganic analysis will be introduced briefly. ToF-SIMS is one of the primary imaging methods used for organic characterization. ToF-SIMS uses a highly focused + ion beam probe (in this case, Bi3 ) to sputter material from a surface, which is subsequently analyzed by a ToF mass analyzer. High spatial resolution mass spectral imaging of molecular, isotopic and elemental species is therefore possible with sub-micron spatial resolutions, and excellent sensitivities (ppm – ppb detection limits). Large area maps were obtained from all the C-4 nanoimprinted samples. The mass spectral data was subsequently analyzed with the aid of principal components analysis (PCA), a multivariate statistical analysis approach used to display differences and similarities in large datasets. Differences were observed in the compositions and distributions, particularly for certain commercial samples (Hall explosives) and for certain years of production (2001, 1975, and 1989 in particular). These differences were attributed to the changing compositions of PIB, DOS, RDX, and HMX. The Hall explosives sample had the most significant differences. First, there was no evidence of HMX in the mass spectral data acquired from the Hall explosives samples, which alludes to the type of RDX processing utilized by this company. There are two primary processing methods used to manufacture RDX. The first is the rare woolrich process originally developed in England that yields RDX as the primary product. The Bachman process is much more common due to it being much cheaper to produce. This process, however, can yield up to 20% of HMX explosive as a bi-product. The Hall explosives sample was also the only sample containing peaks characteristic of DOS, as opposed to DCA for a plasticizer. Example of C-4 characterization process using ToF-SIMS. Top panel (a) shows examples of large area molecular image analysis Other mass spectrometry methods employed in this study of a model C-4 fingerprint. Si+ (m/z 28), NO2- (m/z 46) and two (nano-DESI and ESI-MS) yielded similar results as was molecular ion overlays are shown. Overlay 1: HMX (green, m/z 268) observed for ToF-SIMS. Additional signatures were vs. RDX (pink, m/z 342) explosives, and Overlay 2: PIB binder (green, observed, particularly for nano-DESI but have not yet been m/z 97), DCA plasticizer (pink, m/z 427), and nitrate explosives (blue, m/z 46). The bottom panel (b) shows the results from principal identified. XPS is capable of determining atomic concentra- components analysis of the dataset. Each point (score) in the plot is tions and chemical states of the elements present in the associated with a particular sample, where different scores will have sample. XPS found significant differences for samples made different mass spectra. For this example, the “Hall explosives” sample in 2001, which exhibited decreases in nitrogen content rel- was significantly different from the other samples, as indicated by ative to the other samples. This is consistent with SIMS the very different scores. Corresponding changes in the mass spectral results, showing increased binder contents in these particu- data indicated that there was a different plasticizer used in the production of this sample (DOS). lar samples. However, specificity of the method was insufficient for further differentiation. FY 2013; the complementary elemental analysis methods C-4 Characterization of Elemental and Isotopic Composi- will significantly improve the accuracy, precision, and spec- tions. Elemental and isotopic analysis using a combination ificity of results. of SEM/EDX, dynamic SIMS (with a magnetic sector instead of ToF), and ICP-MS is ongoing. For these samples, bulk Data Analysis. Data analysis and integration methods are a analysis of trace elemental compositions is performed with key feature of this project. Preliminary data analysis and ICP-MS. Insoluble particulate matter is also being charac- integration approaches have been determined and will be terized with a combination of SEM/EDX and SIMS. It is applied to the data sets in FY 2013. Briefly, partial least expected that these experiments will be completed in early squares (discriminant analysis [PLS-DA]) will be used for PN12007/2408

4 Advanced Sensors and Instrumentation investigations on each individual datasets. The results of Finally, two new capabilities being developed for analysis PLS-DA are ideal in that they are presented as probabilities of microbial samples: a state-of-the-art high spatial resolu- that a particular sample belongs to a particular class. For tion imaging mass spectrometer for analysis of single the current example, results would report on the probabil- microbial cells (nano-SIMS) and a gas cluster ion beam ity that an unknown C-4 sample belongs to a particular (GCIB) source for low damage sputtering and 3-D analysis manufacturer or lot. These probabilities will be combined of organic and biological materials. using Bayesian integrated statistics for more accurate cate- In FY 2013, we intend to complete our elemental analysis gorization. Determinations of classification accuracy, speci- of C-4 and focus on data integration procedures. Once in ficity, and selectivity will be calculated, making this place, these procedures likely will be used for all future method admissible in a court of law. The combination of forensics investigations. We also intend to utilize a high multi-modalities is anticipated to increase significantly the mass resolution SIMS instrument (SIMS with FT-ICR for ana- accuracy and specificity of the result. lyzer instead of a ToF) for improved specificity and mass Microbial Analysis. Infrastructure development for micro- accuracy. At least three different publications will be sub- bial forensics work at EMSL has also been started. Prelimi- mitted to reputable journals. We will then switch our focus nary work has been demonstrated, where several samples onto microbial forensics of single cells, where we will of E-Coli grown in Luria Bertani and tryptic soy broth develop our newer technologies (e.g., nanoSIMS) to address media were compared with ToF-SIMS and XPS. Preliminary significant problems in the microbial forensics field. 3-D data show significant differences in the resultant mass analysis in single cell microbial systems will be demon- spectral data from bacteria prepared in these different strated using advanced ion beam technology. Finally, single media. However, a better understanding of the prominant cell proteomics will be explored with a combination of mass spectral features is required. ICP-MS has also proven microfluids approaches and matrix-assisted SIMS (e.g., SIMS to be invaluable for trace elemental analysis of media with MALDI matrices applied to surface). compositions and has helped identify important mechanistic aspects of bacterial growth in different media sources. PN12007/2408

5 Advanced Sensors and Instrumentation PN12021/2422 6 the damper. Theapproach buildsanairflow station using rotateBarometric dampers from changesinpressure across aphysicalprototype device. anddeveloping HVAC industry ity tosolvethiscritical,well recognized problem inthe The outcomefocused ondemonstrating thePNNLcapabil- practices. savings compared toindustry significant energy ventilation ratetive control, whichinturnwouldleadto for airhandlingunitsthat- wouldenabledevice cost-effec tion), reliable barometric-operated airflow measurement compared with $1500~$3000 listpricefor existingflow sta- demonstrate alow-cost(targetlistpricelessthan$500, wastodevelop, objective test,and position. Theprimary ventilation whentheoutsideairdamperisat minimum prototype hardware solutiontargetedat widespread over- problem. andtesteda We developed ment efficiency able airflow measurement solutiontothebuildingequip- reli alow-cost,highly - The project focused ondeveloping hasbeeninitiated tosolvethisproblem.no effort ing tocommercial buildingswhenoperating properly, yet economizer, that provides automatically free adevice cool- results. Apotentialkeycomponentistheair-side survey energy, basedonU.S. Environmental Protection Agency true need,savings wouldbeapproximately 0.3quadof bring rates inalignmentwiththecurrent standards and were outdoorairsupply reduced to age minimumrate of theaver systemfor measuringventilation rates.integral If U.S. few buildingshave an useandhealth,very energy ventilation rates on Despite substantialinfluencesof pant health,andoccupantcomfort orindoorenvironment. consumption,occu- stantial influenceonbuildingenergy building ventilation (i.e., providing outdoorair)hasasub- consumption.Specifically,that leadtoincreased energy failure statesment running andcomponent ininefficient Often, pooroperation andmaintenanceresults inequip - improvements inbuildings. efficiency the potentialof globalclimatetion, ormitigating changeneedstoconsider consump- reducing energy aging theelectricpowergrid, inman- focused onimprovingAny seriouseffort efficiency Darrel Hatley Air FlowStationforAir-Handling Units Development ofaLow-CostReliableBarometric Advanced Sensorsand Instrumentation problems withcross-disciplinarysolutions. PNNL’s abilitytoaddresshighlyimpactful developing solutionsanddemonstrate ing technologiestocapturetheleadin integrated designandapplicationofemerg- economizer operationsreliablethrough low-cost barometricairflowstationtomake This projectdevelopedahighlyreliable, - flow error range, an industry acceptable errorflow rate. error range, anindustry maximum outdoorair equation withresults within5%of andasecondorder polynomial device using thistypeof Results demonstrated that airflow rates canbepredicted eliminated damper weightings. withdifferent marily Damper hysteresis wasthemainchallenges butwaspri- measurement prototype systeminareal airhandlingunit. The project teamdesigned,tested,and validated the flow 1355CFM). door airflow rate of themaximummeasured outline with±67.5CFM(±5%of thetrended secondorder polynomial was withinrangeof largest issuetoovercome. However, measured data mostof damper open(airflow increase). Hysteresis turnedouttobe teresis between damperclose(airflow decrease) and measured airflow rate andthedamperpositionwithhys- flows. The results showedthe over acompleterangeof adjustedinsmallincrementsmanually toallowtesting speedwas drive variable frequency fan Thesupply level. toanegligible tem wassealedtoreduce theairleakage ±0.5%)asthereference.turer’s Thesys- of rated accuracy flow station andwasconnectedtoa - flow hood(manufac A testingbedwasinstalledintheRTL540withabuilt-in profiles complicate further themeasurements. theoutsideairintake anditsimpactonvelocity tion of measurements. Thegeometry, turbulence, andstratifica- aPascal),whichislowfor accurate water (afractionof of aninch used tomeasurethousandthsof airspeed,isonly themoving air,ties, thedynamicpressure of whichisoften from beingdrawn intotheairhandler. Attheselowveloci- keptintentionally lowinorder tominimizerainandsnow are challengingbecauseoutdoorairintake velocitiesare airflow measurements rates.rate First, measurements of andaccu- There are significantobstaclestocost-effective measurement solutions. improved maintenance, andmore accurate airflow equal,oreasierinstallation, greater costeffective, provides technology hot-wire anemometer),thisnew (i.e., pitot-tubeflow station and with existingtechnology dling unitswithoutdoorairdamperinstalled.Compared promising results. Thetargetmarket includesallairhan- culated. The proof-of-concept testwascompletedwith damper positionsensed,theoutdoorairintake canbecal- Based onacorrelation betweenmeasured airflow and tobuildingautomation system. the signalwasinterfaced sensorand tions. Thedamperpositionissensedbyarotary (RTL)540testlab andcalibrated airflow Laboratory sta- ogy tested inatypicalairhandlerusingtheResearch Technol- tomeasure ventilation sensors rate. Thesystemwas rotary barometric and (ordraftregulator) dampers off-the-shelf Forensic Analysis with Time Stamped Coincidence Sampled Gamma Radiation Spectroscopy

Larry Chilton

safety requirements for radiochemical separations are sig- This project developed a novel data analysis nificant. method that takes advantage of established We used data collected from irradiated samples with coin- knowledge about the spectral features of cidence gamma ray spectroscopy to investigate differentiat- short-lived radioisotopes and exploits ing characteristics of short-lived isotopes associated with changes in relative abundances over time to their half-life and combined with their spectral features. increase the confidence of identifications. When successful, this method decreases the amount of time necessary to resolve the contents of a complex mixture, quite possibly reducing the need for the radiochemi- Often, cutting-edge sensors produce measurement data cal separation. Exploiting time series of spectra for that require novel analysis for signature discovery and identification and quantification of isotopes in a mixture detection. This is certainly the case for the application on required that the instrument efficiency and dead time be which we focused, which was to develop methods to detect characterized. This information was then used to adjust the the signatures of radio-isotopes in data collected by the a apparent starting quantities and half lives for isotopes of direct simultaneous measurement (DSM) instrument. PNNL interest. The failure to make these adjustments would compiled terabytes of time-stamped coincidence-sampled likely have resulted in underestimates of the initial quan- (TSCS) gamma-ray spectral data with the DSM consisting of tity and an over-estimate of the half-life. fission product gamma-ray measurements within which The dead time and efficiency for the instrument was char- valuable nuclear forensic information is contained. How- acterized, and experimental data were adjusted for these ever, no spectral analysis tools exist to extract this impor- parameters. Consequently, these data were analyzed with tant information. Of particular interests are short-lived time-series methods to estimate isotope half-life by model- isotopes (SLIs) and the accurate characterization of isotopes ing the change in count rate over time for specific spectral in the presence of interference. features associated with the isotope of interest. The results Current state-of-the-art approaches for gamma-ray spectral for 140Ba, 95Zr, and 144Ce were very encouraging, as the com- analysis are designed for time-integrated spectra. As TSCS bination of known spectral activity channels and half-life data sets are summed to time-integrated gamma spectra estimation using data in time series for known present iso- for analysis, the process ignores added information pro- topes resolved the half-life within a few percent. Equally duced by time-stamping and coincidence-sampling. Devel- important, when the isotopes are not present, half-life esti- opers and users of the DSM are confident that maximizing mates do not indicate their presence. As a result, two the time-series and coincidence-sampled data will likely papers are being prepared for peer reviewed publication: yield new insight into the isotopic composition of material one addresses the statistical methods developed for isotope samples, especially for short-lived isotopes. Importantly, identification, and the other presents approaches to system accurately identifying these isotopes in environmental independent dead time estimation and correction. samples supports several national security missions, including nuclear nonproliferation, forensics, and safeguards and security. Identifying the isotopes in a fission sample is often used in forensic analysis to gain information about the origin of the material. The typical approach to the problem is gamma ray spectroscopy. The challenge is that complex mixtures contain many isotopes with the same or overlapping characteristic peaks and other ambiguous identifying features such that some isotope identifications are highly uncertain or not tractable. Traditional solutions often involve radiochemical separations of the samples into sim- pler mixtures, which is not optimal because the time necessary for separations may result in some isotopes decaying Eighty-five isotopes have spectral peaks at 537 keV, but only one, to point below instrument sensitivity. Also, the facility and 140Ba, also has a half-life of 12.74 days. PN12033/2434

7 Advanced Sensors and Instrumentation PN12036/2437 8 prototype detection system affixed to the glider.prototype detectionsystem affixed the andrangeof and 3)assessingthedetection efficiency southernresident killer whale (SRKW)calls, detection of analgorithm to conduct automated system, 2)developing acousticdetection ing andbuildingaprototype passive Key technicalgoals throughout theyear included1)design- data canbeobtainedremotelywere notfeasibleinFY2012. thedetectionsystemwithglidersothat to integrate automated detectionontheWaveof Gliderplatform. Efforts acousticdetectionsystemcapableof-concept for apassive thisproject demonstrated theproof- Commencement of sea conditions. aircraft, andcanbelimitedby whichmay becostly onshipsand are collectedby trainedhumanobservers species presence, distribution,abundance). Currently, data (i.e., development ting processes for windenergy offshore projectneeded tofacilitate sitingandenvironmental permit- tion system(PADS) for collectingdata onwhalepopulations acousticdetec- anautonomousglider-based passive use of the thiswork istoadvance Thegoal of and permitting. use patterns), whichcanbeamajorobstacletoproject siting distribution,abundance, populations and migration, (e.g., solutions toresolve onmarinemammal knowledgegaps sources requires energy innovative hydrokinetic (MHK) windandmarine offshore To of development facilitate thedata. thelimitationsprotocols, of andunderstanding survey developing systemsintegration, lection andanalysis, them apracticalapplication, includingautomating data col- improvements tothesesystemsare stillneededtomake collectdata continuously.can essentially However, significant withunmannedoceanvehiclesthat detection technology researchers exploringways have tocoupleacoustic begun and may becostprohibitive. To helpaddress these issues, are inherentlydangerous, have poordetectionefficiency which aerialsurveys, is gathered viashipboard orlow-level Traditionally, information onmarinemammalpopulations Kyle B.Larson System CapabilityDevelopment Glider-Based PassiveAcousticDetection Advanced Sensorsand Instrumentation standing standing of marinemammalpopulations. improving thescientific community’s under with traditional ship-based surveysaswell ducing thecosts,risks,andbiasesassociated fits ofthisresearchincludesignificantlyre- offshore projectcitingandpermitting.Bene- to canbeamajorobstacle populations, which knowledge gapsaboutmarinemammal the Wave Glider™tohelpresolvecritical mous, wave-poweredsensorplatformcalled PNNL isexploringwaystoutilizeanautono-

distances lessthan200m. Bay, ≥80% at isgenerally WA indicate that detectionefficiency three broadcast driftsconductedinSequim results of Preliminary monitoring marine mammalpopulations. required inorder for thisapplication to beausefultoolfor vehicle, thissituation isultimately autonomous survey fully theWave Glidersuchthat constraints of itcanfunctionasa thedetection system tooperate grating withinthedesign potential for improving detectionrange. Inaddition tointe- expected detectionrange, indicating that there is good The system’s ability todetectSRKWcallsexceeded its ontheWavewhale calldetectionsuccessfully Gliderplatform. We were automated able todemonstrate proof-of-concept of 600 m. 200 mandthemaximumdetectionrangeisapproximately was at ≥80%)is approximately in whichdetectionefficiency detectionrange(i.e., therange suggest theaverage effective thesetests results of attempted toholdposition.Preliminary SRKWcallsfrom asilentdriftingboat whiletheglider ings of evaluated bybroadcasting andrange)was record- (efficiency ditioning hardware tooptimizeSNRs.Detectionperformance tional improvements were madebymodifyingthesignalcon- before assessingdetectionperformance ontheglider. Addi- improve algorithm performance anddetectionthresholds andfieldtestswere conductedto laboratory Preliminary detections. processing verification of this process ensured adequate recording qualityfor post- candidaterecord whalecalls.Ultimately, thefullduration of andprovidedsystem tolistencontinuously theability to which employed real-time looprecording that allowedthe methodology,SRKW callsusingsignal-to-noiseratio (SNR) toperform detectionof teries. Analgorithm wasdeveloped detector,software, energy-based andtworechargeable bat- asingle-board computer,face/amplifiers, data processing noise, broad-band hydrophones, signalconditioninginter omni-directional, lowself- and builtinFY2012,consistingof acousticdetectionsystemwasdesigned A prototype passive

- In Situ High-Pressure X-Ray Diffraction Investigation of Caprock Mineral Reactions

With Water Solvated in Supercritical CO2 H. Todd Schaef mentary techniques including isotopic labeling and molec- The objective of this project is to develop a ular dynamic simulations. high-pressure x-ray diffraction capability to advance our understanding of reaction processes and kinetics at the fluid-solid interface between water solvated in supercritical

carbon dioxide (scCO2) and minerals with special emphasis on layer silicates.

Deep saline reservoirs with confining layers (cap rocks) are considered primary targets for carbon storage by most countries. Post-injected CO2 will reside largely as a water- rich supercritical buoyant fluid constrained from vertical migration by a low-permeability cap rock seal. This process allows time for slow trapping mechanisms (e.g., CO2 dissolution in pore water, precipitation of carbonate minerals) In situ HXRD graphs depicting interlayer spacing changes when exposing variable hydrated sodium saturated montmorillonite to to sequester CO2 from the system. Despite this well- scCO at 50°C 90 bar. accepted conceptual model, most research on flow, trans- 2 port, and chemical reaction processes has remained in aqueous-dominated geologic systems, which will not In FY 2012, we used in situ HXRD to examine the carbonation of a pyroxenoid mineral, wollastonite [CaSiO ], in represent the vast majority of rock volume where scCO 4 2 contact with wet scCO at pressures (90, 125, 160 bar) and is stored. Moreover, recent PNNL work indicates that 2 chemical reaction processes between water solvated corresponding temperatures (50°, 55°, and 75°C) selected to simulate reservoir conditions. Exposing wollastonite to in the scCO2 phase and geologic media are significant. anhydrous scCO2 resulted in no detectable carbonation. This project is developing and conducting proof-of-concept However, the addition of water above ~50% saturation testing using high-pressure micro-focused x-ray diffraction level of the scCO2 resulted in partial carbonation through (HXRD) capabilities to investigate fundamental reaction the formation of calcite, detectable within a few hours at processes occurring in mineral systems exposed to water all conditions. Increases in water content correlated to solvated in scCO2. The in situ probe utilizes x-rays to exam- increases in carbonation. Tests conducted below the 50% ine transformation reactions of minerals and their kinetics water threshold did not produce detectable carbonation by when exposed to water bearing scCO2 fluids under pressure, the in situ XRD method. Oxygen-18 enriched water was temperature, and fluid composition conditions relevant to used to track mineral surface hydration processes occurring geologic sequestration of CO . 2 in the wet scCO2 environment. Designing a reactor compatible with x-rays and acid gas Reaction processes occurring with clay minerals (montmo- mixtures requires special materials and unique design cri- rillonite, kaolinite, and chlorite) common to caprock for- teria. Additionally, the design must account for beam con- mations were explored through varying amounts of figuration, geometries at low angles, and x-ray attenuation dissolved water in the scCO2. These experiments suggest a by dense gas mixtures. During the first year of the project, high degree of stability for kaolinites and chlorites when in a high pressure reactor was designed, fabricated, and contact with wet scCO2; no detectable structural changes tested at pressures and temperatures relevant to geologic were detected with the in situ HXRD technique. However, sequestration. In the project’s second year, we conducted characterized by their ability to accept water into their testing on pure minerals, including brucite, montmorillon- interlayer, montmorillonite clays proved dynamic when ite, and forsterite. Activities occurring in year three focused exposed to scCO2. Depending on the interlayer cation and on integrating experimental measurements with comple- hydration state, exposure to anhydrous scCO2 produced a PN10044/2286

9 Advanced Sensors and Instrumentation 10 PN10044/2286 were least affected whenexposedanhydrouswere scCO leastaffected hydration)changes. Lowerhydration states (~1water of structural change resulting inmeasurable mineralsvolume tact withanhydrous scCO were mostsusceptibletodehydration process whenincon- hydration montmorillonites containing2or3layers of togeologicsequestration. Bycontrast, conditions relevant uptake of water bythescCO uptake of However,spacing. andfollowing theprocess isreversible, the interlayer interlayer water andsubsequentcollapse of exposure withanhydrous scCO ring tocaprock seals.Insomeenvironments, initial base for predicting potentialpermeability changesoccur experimentscontinued toprovide adata- These typesof clay volume. through molecular dynamicsimulations, the itappears occur andthushelpmaintaincaprock integrity. Moreover, Advanced Sensorsand Instrumentation 2 andoftenproduced decreased 2 fluid, swelling of clays will swelling of fluid, 2 will cause dehydration of willcausedehydration of 2 under - cell hasscheduledfor finaltestingbefore project end. To expandthisinsituHXRD capability, asecondgeneration tions betweenmontmorillonitesandwater bearingscCO November andcompletingthefinalmanuscript oninterac- Greenhouse GasConference (GHGT11)inKyoto, Japan in willfocusdesigns. Immediate onpresenting efforts at the ited high-pressure experimentsonsecondgeneration through finalizingpublicationsmarily andconductinglim- maximize exposure andpromote theinsitucapability pri- Proposed research for FY2013are activities designedto the long-term fate of geologically stored geologically CO of the long-termfate studies tohelpenable betterpredictions of these kinds of HXRD experimentswithscCO ity, andwater-content dependencedemonstrated byinsitu trasts inreaction rates, product formation, mineralstabil - into theinterlayer andleadingtoitsexpansion. Thecon- presence of wetscCO presence of 2 facilitates the incorporation of water theincorporation facilitates of 2 highlight the importance of of highlighttheimportance 2 . 2 . Low Frequency Electromagnetic Interrogation Techniques for Container Content Signature Detection A. Mark Jones The electromagnetic field distribution and coil impedance This project performed a simulation-based response produced during a given container interrogation is dependent upon multiple factors such as the position, investigation of the use of low-frequency orientation, size, shape, surface area, and metal type of the electromagnetic signals to detect concealed internal contents. From the complex interaction between radiation threats in large, closed metal con- these variables, it is difficult to formulate a single figure of tainers via the impedance signature produced merit that can be used to describe the sensitivity of the coil by eddy currents induced in conducting method for general combinations of metal containers and shielding materials. Depending on the contained objects. The ability to efficiently perform system- concept of operations and container physical atic investigations of these system parameters using simulation software is a key enabler in understanding how the and electromagnetic properties, it was deter- technology can be applied. mined that this technology has the potential For coil geometry, simulation results showed that encircling to enable rapid volumetric inspection of coils yielded better overall detection of centrally located containers to assist with detecting illicit objects inside large containers than pancake or surface coil or diverted nuclear materials. configurations. In frequency range, the magnetic field penetration of container walls occurred for frequencies under Research previously conducted at PNNL has demonstrated 2 kHz for the majority of configurations analyzed. With con- that the low frequency electromagnetic (EM) response of a tainer penetration, metal containers can be constructed sealed nuclear weapons component storage container inter- from materials that exhibit magnetic or non-magnetic rogated with an encircling coil is a strong function of its properties. A magnetic container exhibits a high magnetic contents, and can be used to form a distinct signature permeability that significantly decreases the ability of mag- which can confirm the presence of specific items. A natural netic fields to interact with contained objects. The con- extension of this research is to investigate the scalability tainer wall thickness is another factor that affects field and sensitivity of this technique to determine if it can be penetration, and the skin depth equation can be used to used to detect illicit nuclear materials concealed within understand field penetration as a function of frequency large cargo containers. When properly normalized, the and both container wall thickness and electromagnetic resulting signatures could provide material and geometric properties. In contained object geometry, the largest coil information that is complementary to conventional modali- responses are produced by objects with a high electrical ties such as radiation detection. conductivity, high magnetic permeability, and large cross- sectional area along the direction perpendicular to the This project explored the use of low-frequency EM energy primary magnetic field. The interaction with the object to interrogate large metal containers that contain radiation is governed by the physics of Foucault currents induced threats shielded by conducting materials. Examples of con- on a conducting object. ducting radiation shielding materials are lead, tungsten, and iron. Unknowns included the sensitivity of the signa- An extensive set of EM simulations was performed for inter- tures to variations in container and internal object size, modal cargo containers (IMCC) and air cargo containers composition, and geometry. We evaluated frequency ranges (unit load devices) with metallic shielding geometries calcu- and sensor configurations which maximized the response of lated to be necessary to conceal certain amounts of nuclear the coil system for various inspection scenarios. Because materials. Numerous 3D models were analyzed for various experimental characterization of a complete set of these coil configurations, container material properties, and orthogonal parameters is unrealistic, we employed shielding material geometries and locations. The sensitivity advanced finite element electromagnetic field simulation of the coil response was quantified as a function of these software to investigate numerous detection configurations. parameters as well as the interrogation frequency. As an We validated the accuracy of the EM simulation software example, a detailed study of detection sensitivity was con- using coil impedance measurements for a set of conducting ducted for lead and carbon steel shielding materials located spheres placed inside a closed AT-400R nuclear material in the center of an 8 ft × 8 ft × 10 ft IMCC. The shielding storage container. Excellent agreement was obtained material was represented by a solid rectangular prism with between the simulated and measured results. a height and width of 1 ft and a length ranging from 1 ft to PN11047/2369 11 Advanced Sensors and Instrumentation 12 PN11047/2369 Materials Management. We presented technicalresults cou- Nuclear of accepted for 2012Journal publication inthefall were operational parameters avarietyof function of for ahypotheticalnuclear material storagecontainerasa coilresponse oursimulation studyof cations. Theresults of included progress inarmscontrol treaty verification appli- We accomplished ourintendedtasksonthisproject, which an Excalibur munitionsstoragecontainer. this methodontheAT-400R nuclear storagecontainerand resonant coilmethod. Measurements were performed using magnitude at compared theresonant tothenon- frequency improved byatcoil system.Sensitivity leastoneorder of byaddingin theoperating bandwidth capacitance tothe ments. Thismethodintroduces aresonance at afrequency experi- proposed, simulated, andverifiedusinglaboratory A methodfor increasingwas the coilmethodsensitivity metaldrums are constructed fromof carbonsteel. however,sensitivity; duetomaterial costs,thevastmajority magnetic stainlesssteeldrums wouldimprove detection non- precluded detectionviathecoilmethod.Theuseof andcontainerproperty variations noiselevels background manyitemsexploredsurements. duringthestudy, For lon steeldrum wasquantifiedusingsimulations andmea- - surrogates orientations placedindifferent insidea30-gal screen for threat items.Thecoilresponse for variousthreat taining high-valueliquidorpowdercargo to are difficult interest insecurityscreening since sealedmetaldrums con- high ponents insidelargesteeldrums. Steeldrums are of ture couldbeusedtodetectthreats withsmall metalcom- thecoilsigna- We alsocompletedastudytodetermine if thecoilmethod. of sensitivity increasescomposite aircargo thedetection containers confirmed that thetrend toward adoptingmetal-framed that containmetallicshieldingobjects.Wecontainers cling coilsusedtoinspectmetallicandcompositeaircargo encir Simulation theresponse studiesalsoinvestigated of niques were proposed tocompensate for thiseffect. wasaddressed, tech- unlessthisfactor tivity andseveral found toimpose significantlimitations ondetectionsensi- properties were alsostudied.Temperature changeswere measurement sitesoncontainerandmaterial electrical expectedtemperature at differentials thetwo of effects mate thesmallest detectable changesinobjectmass.The tainer andapplied tonoiselesssimulation results toesti- small-scale coilmeasurements withaDOE-STD-3013con- werematerial. quantifiedusing Statistical noiselevels order asthose expectedtoconcealintroduced illicitnuclear be usedtheoreticallytodetectmasschangesonthesame Results determinedthat anoptimizedencircling coilcould at theorigination anddestination ports. could bedetectedbycomparingcoilresponses measured materials duringtransit theintroduction of determine if wasincluded.Thegoalused incargo wasto containers beamfloorincluded; andgridded construction commonly steel walls;cornerpostsanddoorlatches were not corrugated wassimplifiedwithsmoothsteelinplaceof try 4 ft.To thesimulation facilitate study, thecontainergeome- Advanced Sensorsand Instrumentation

sensitivity studies.Spheresensitivity diameteris10cminallmodelsshown. Representative finiteelementmodelgeometriesusedincoil control treaty verification. ties for securityscreening andarms supporting objectives toenhancePNNL’sdemonstrated amethodology capabili- theinspectionmethod.Project results and improvement of measurements valuable insightinto thecapabilities togain bined powerfulsimulation studiesandappropriate lab national important securityapplications. Wevariety of com- theinductioncoilmethodfor a theuseof to investigate In summary, thisproject performed fundamentalresearch nuclear material. shield asubstantialquantityof to leadmaterial inamounts necessary the presence of container demonstrated that theEMcoilmethodcandetect materials. Simulation results for a5fttallstainlesssteel shielding materials couldbeconcealedthrough theuseof alone mayfor verification beinsufficient sincethe nuclear custody.sion from thechainof measurements Radiation empty andcontainnonuclear material intendedfor diver are toverifythat actually largemetalcontainers necessary During theweapons dismantlementprocess, itwilllikelybe materials. no information disposition about theisotopicnature of with thematerial nuclear structure, thecoilmethodreveals forms. Inaddition, becausemagneticfieldsdonotinteract methods that cannotdistinguishbetweenoxide andmetal container, toradiation detection whichiscomplementary the applied fieldandthe result isthesameasanempty metals andoxides astheoxide forms donot interactwith strated that itispossibletodistinguishbetweendifferent custody.chain of example, simulation For results demon- material inthe diverted of nium dispositionandprevention declared to beusedforforms duringpluto- verification of This publication outlinedthepotentialfor thecoilmethod Nuclear Materials Managementconference 2012. inJuly considerations at the53rd Institute of pled withpolicy - Low-Noise, Low-Background Electronics Brent A. VanDevender tor itself, and the stray capacitance of the grounded conduc- This project will establish a PNNL world-class tors used as structural elements of the system. Advances in capability for low-noise low-background detector and electronics technologies have reduced the first two sources to such a level that the latter source can no electronics to support the ultra-sensitive longer be neglected in the effort to reduce total noise con- nuclear measurements often demanded tributions from capacitance. To account for this, we have by national security and basic science developed the capability to model the electrical capacitance applications. of realistically detailed detector systems with finite-element analysis software. Electrical modeling is closely coupled to the mechanical design process for a fast Electronics noise restricts the sensitivity and accuracy of virtual-prototyping cycle that can minimize radiation detection systems but can be abated by careful stray capacitance without the need design and layout of electronic circuits following well- to build and test many known principles. However, these practices alone are iterations of an insufficient to meet much less exceed the state-of-the-art. expensive system. Attention must be paid to geometries and materials in the environment in which the detection system operates. Any Electric field electrical conductors in the environment will result in stray modeling also capacitance that couples into the circuit in ways that can supports the defy the intuition of even the most seasoned electronics analysis of ultra- engineer. These parasitic capacitances are a source of noise low background at best, and they frequently result in the complete malfunc- proportional tioning of circuits (e.g., feedback amplifiers driven into counter (ULBPC) oscillation). The problem is exacerbated if any of the stray performance. capacitance results from electric fields in dielectric material. ULBPCs developed to detect The sensitivity of nuclear measurements is often limited by 37Ar in support levels of electronic noise and/or intrinsic radioactivity in of on-site inspec- construction materials of detector systems. This is especially tions for verifica- true in contemporary national security and basic science tion of nuclear applications, where signal events of interest may be small test-ban treaties and/or very rare and thus easily obscured by noise and have a lower The electic potential inside a PNNL- backgrounds. Expertise in low-noise analog electronics efficiency than designed low-background low-capacitance design germane to radiation detectors is increasingly scarce, expected. The germanium detector cryostat on a plane especially in the United States. We will remedy this situation through the central axis (blue = ground, cause is hypothe- red = high voltage). by building a comprehensive capability at PNNL to support sized as being low-noise analog electronics design for radiation detector complicated electric field structures near the ends that systems. We take a system-level view of this capability that were not considered in the original design. Field modeling extends from the initial charge creation in sensitive detector is being used to inform future designs in a similar virtual elements, through the amplifier electronics to digitization prototyping cycle. for data acquisition. PNNL is a world leader in the production and use of ultra-low background materials for the With our team increasing its background in electronics for construction of detector systems, and we will extend radiation detectors and improving the history of capability that capability to the electronics used near sensitive development, we will continue to address immediate chal- detector elements. lenges while continuing to build capability in FY 2013. Our long-term capability consists of the equipment required to Two immediate problems to meet our low-noise low-back- outfit an electronics laboratory. We are establishing a dedi- ground electronics goals are identified: stray electrical cated hardware test stand for evaluating noise sources in capacitance in germanium detector cryostats and poorly part through developing support of low-capacitance, low- understood performance of ultra-low background propor- dielectric-loss design through field modeling. In addition, tional counters. The noise in a typical amplifier used with we will further implement the integrated software frame- germanium detectors increases with increasing capacitance work to serve slow control, data acquisition, and analysis at its input. In addition to capacitors that are part of the functions for the test stand. amplifier circuit, this includes the capacitance of the detec- PN12046/2447 13 Advanced Sensors and Instrumentation 14 PN11057/2379 clearly distinguish theconfigurations.clearly spectra energy thedetectedfast-neutron ences inshape of configuration from another. Inthe remaining cases,differ the emittedspectrum todistinguishone reliably suffices response of ied, thegross-count portion inthefast-neutron and practicalmeasurement times.Inalmost allcasesstud- one anotherusingexistingneutron detection technology weapon andshieldingare readilydiscriminated fromof themodelingstudysuggestthat configurations different of real-world Results differentiation detectors. for avarietyof confidenceintype representinglight chart of thedegree comparisons,resulting inastop- put through aseriesof The weapon typesandshieldingconfigurations were then mechanism for template matching ortypedifferentiation. using this detection software todeterminethepotentialof The matrix wasthenusedinconjunctionwithchange interest. tions of matrix totrackthemanypotentialscenariosandconfigura- amodeling werediscussions withexperts usedtodevelop the emittedneutron spectrum. Careful andmany analyses materials on andconfigurations todeterminetheireffect types, whichhave beenplacedwithinvariousshielding detailedMCNP5modelsfor fivedifferent weaponsof U.S. Our project work inFY2011focused ontheconstruction armscontrol. toolssupporting nation’s setof neutron signature wouldbeasignificantaddition toour tem that canidentifywarheadconfigurations basedonits tomeetourwarheadcountingneeds.Asys- development tures andtofurnishguidelinesfor future research and to extractusefulinformation from warheadneutron signa- apracticalneutron detectionsystem assessing theability of result inacoherent anddefensiblescientificbasis for usingneutron signatures. Thestudywill the feasibilityof calculations andmeasurements tostudy combination of verification measurements. Thisproject isemploying a and neutron signatures usedfor have notbeencommonly rial inthewarheadsemitsneutrons inaddition togammas, nuclear warheads.Themate signaturesthe gamma-ray - of Arms control verification measurements exploit commonly Jacob M.Benz Warhead Configurations Identifying for Requirements Detection Signature Neutron Advanced Sensorsand Instrumentation ent configurations. and accuratediscriminationbetweendiffer figurations, whichcouldprovidereliable tron signaturesforidentifyingwarheadcon- This studyinvestigatesthedetectionofneu- ons dependonverificationmeasurements. Agreements forthecontrolofnuclearweap-

- - placement of theTLDchips asdetectionmedia. placement of neutron spectrometer thepassive showingtheflexible 3D designof measurement time. neutron spectra afterareasonable varying sources of ter. Results indicated that itwaspossible todiscriminate theneutron spectrome- characterized detectorresponse of configurations theactual modeledinFY2011byinserting the measurement results andalsothevariousweapon then puttothetest.Changedetectionwasperformed on eter todiscriminate betweendeclared configurations was thisneutron spectrom- of theeffectiveness of Final analysis themeasured tomodeleddoseresponses. of createdto-dose equivalent inthemagnitude adifference dependentdose- energy issues related of totheconversion theresponse. Outstanding well intheoverallshape of modeledtomeasured configurations agreed comparison of readout beingintroduced uncertainties duringreadout. A 10mrad (100mrem). Thisthreshold minimizespotential sen insuchaway that eachchipreceives aminimumof energies(0.02eVto20MeV).Chiplocations are cho- varying thechipstoaccommodate aneutron fieldof placement of Additionally, thedesignallowedfor theflexibilityin deposited in eachTLDchip.total doseorequivalent spectrometer. Therefore, detectorresponseasa isgiven The designutilizedTLDchipsasthedetectionmediain between fission,thermalized,and(a,n)neutron spectra. to determinewhetherthespectrometer coulddistinguish sources waschosen erated Cf-252,andPuBe. Thisrangeof spectrometer. Thesources includedbare Cf-252,D2Omod- the surements were characterization performed of tobegin fabricate thespectrometer. mea- Upondelivery, aseriesof with amechanicalengineeringdesignteamtoand neutron spectrometer.terize apassive PNNLpartnered FY2012wastodesign,fabricate, andcharac- The focus of Novel Detector Configurations Marty Keillor We examined a wide variety of potential near-term This project sponsored the production of a research paths to enhance PNNL capabilities in ultra- wide range of new radiation detection con- sensitive radiation detection. The effort identified promising new detector developments that capitalize on existing cepts that have the potential to lead to sig- PNNL detector technology strength areas: intrinsic germa- nificant sensitivity improvements over the nium spectrometers, ultra-low-background gas propor- existing technology. The objective was build- tional counters, scintillators, coincidence systems (e.g., ing capability in existing PNNL strength areas germanium arrays, b–g systems), and cadmium zinc tellu- and expanding our repertoire of highly sen- ride arrays. Research also considered areas ripe for prog- sitive radiometric measurement systems. ress to improve availability of ultra-pure materials. We anticipated reaping several of the following benefits from this study: 1) promising technologies that represent oppor- Across a broad spectrum of basic and applied science, tunities for further sensitivity improvements; 2) prioritized ultra-sensitive measurements of radiation are used to research topic areas for consideration of near future fund- answer critical questions. In the arena of neutrino proper- ing; and 3) initial performance modeling for the most ties and dark matter, as few as several events each year in a promising candidate detector technologies. large experiment hold the key to discovery and observa- This project sponsored the production of a wide range of tion. In fields such as nuclear treaty monitoring or environ- radiation detection concepts with the potential to lead to mental carbon cycle detail probing, ultra-sensitive sensitivity improvements over existing technology. Individ- measurements offer the ability to obtain answers from ual researchers developed 5–10 page concepts with back- smaller samples and provide a more precise quantification ground information and the outline of a new proposition. of measured indicators. Specifically, the world’s most sensi- The focus of these studies was on advancing existing tive nuclear measurement systems rely on techniques technology or configuring existing technology in new developed to reduce or discriminate unwanted background ways, rather than developing fundamentally new detec- signals. This project maps a near-term approach for detectors. Brainstorming sessions were held to consider the tor science, focusing on the set of challenges that must be concepts under development and to provide a broader overcome to enable a new generation of sensitivity and view of each problem. diversity of application for ultra-sensitive measurements. Deliverables completed under the project included the pro- Recently, PNNL pursued a relatively focused set of ultra- duction of an FY 2013 Call for Proposal text in the area of sensitive radiometric measurement technologies. For ultra-sensitive measurements systems. Ten proposals were radiopure materials, these include methods development submitted based on that call. A total of 3 ideas were con- for the production of ultra-pure copper and the assay of a sidered to require further concept development, 6 were limited selection of commercial materials. For detector judged to have significant potential for funding through technology production, successful research projects devel- other venues without LDRD investment, and one was oped single-crystal cryostats and arrays of high-purity ger- selected for LDRD funding by the laboratory through its manium crystal, ultra-low-background gas proportional normal review and approval process. The second deliver- counters, new glass scintillator materials, and cadmium able is a final report detailing the concepts developed zinc telluride arrays. In addition, b–g coincidence detection under this project, which is in progress. systems for sensitive measurements of gas samples (e.g., xenon) have been a focus area, along with the develop- One significant outcome of this effort was determining that ment of sophisticated gas separation and purification sys- the method pursued resulted in concepts that were mainly tems to prepare samples for measurement. evolutionary, taking the next step in the strength areas of our researchers. While this generates useful research topics, This one-year research effort combined a literature study it does not tend to produce the preferred higher-risk, of recent advances in radiation detection techniques with higher-reward research projects. Brainstorming on this an analysis of PNNL strengths and weaknesses in the arena topic led us to a new approach that we believe will assist of ultra-sensitive radiometric measurement technologies. in stimulating innovative thought and intriguing new The effort sought to identify those areas where PNNL was radiation detection concepts. This method, which includes poised to move the state-of-the-art forward; an effort documenting the weaknesses of current detection technol- was also made to recognize areas of weakness where ogies along with the desired improvement, will be pursued targeted hiring could provide a significant boost to both in the final report of this project and in the coming PNNL capabilities.

year by laboratory management. PN12056/2457

15 Advanced Sensors and Instrumentation 16 PN11060/2382 cation of electro-optical polymer materials electro-optical anddevices. polymer cation of More- infabri importantly, withexpertise PNNLboastsstaff lithiumniobate materials. more traditionalcrystalline circuit technologiescomparedbility withintegrated tothe lower cost,simplerfabrication methodsaswellcompati- design withRFmicrostrip input. electro-optical ourpolymer-based modulatorGraphical rendering of width intothesub-THzregionandlowerV betterperformance (widermodulationalso offer band- employed at manyairports. Polymer-based modulators millimeterwave imagingportal as thePNNL-developed systemssuch fieldedactive lation toincrease resolution of (RF) modu- canbeusedfor widebandradiofrequency tors thelocaloscillator.distribution of Polymer-based modula- andlow-lossfiber high powertelecommunications lasers, opticaloscillator generation efficient using waveguide), (1000× lowerlosspermetercompared tomillimeter-wave compactandlow-loss telecomfibers exploits theuseof syntheticapertureUsing opticalupconversion, imaging lenging usingconventionalmillimeter-wave technology. smallantennasbutitsimplementation ischal- line array of can increase spatial resolution byexploitingalargebase- tion thanopticalapproaches. Syntheticaperture imaging lowerspatial resoluopaque toopticalimagingbutoffer - Millimeter-wave signalscanpenetrate manyobscurants Douglas L.McMakin Active MillimeterandTerahertz Imaging Optical UpconversionforPassiveand Advanced Sensorsand Instrumentation ing techniques. frequencies toexploitopticalsignalprocess- vert millimeter-wave energytooptical based electro-opticalcomponentstoupcon- Our objectiveistodevelopactivepolymer- π ), potentially ), potentially developed a process to pole our active polymer layer polymer a process using topole ouractive developed 4µm. Additionally, therequiredto achieve thicknessof we to pattern sputteredfollowed byplating silveronpolymer S-bend. To aprocess form theRFwaveguide, wedeveloped disclosure) RFmicrostrip usinganinverted with waveguide ceeded todesignanovelstructure (resulting inapatent SU-8 dopedwithDR19inaguest/hostsystemandpro - Therefore,smooth layer byspincasting. wesubstituted This wasduetoitsviscousnature andreluctance toform a unacceptable roughness inourthermosetcore material. opticallosspercmdueto showedexcessive evaluation a PDMSstamperfor rapid processing turnaround butcloser design for lateral confinement.Theribwasembossedusing finement withsinglemodeoperation rib usinganinverted During FY2012,wedemonstrated lateral con- andvertical characterizationour opticalwaveguide facility. using waveguides in onedimensionfabricated polymer ical modelinganddemonstrated singlemodewave guiding We usingnumer opticalwaveguide alsodesignedapassive chose acompatible thermally-cured epoxy for thecladding. mer for thecore and withhighchromophore loading, - athermosetpoly In FY2011,weselectedanddeveloped lower costRFcomponentsandtestequipment. 94 GHzbutchangedtoa10-GHzdesigninorder touse power. Originally, wesoughttodesignourmodulator at imagingsystemswithreducedpassive size, weight,and ture systemsthat canmake possiblemobilevideo-rate erate astable opticalcarrierenables stable syntheticaper theincidentbeam.Theability togen- polarization state of the relative power, preserve fully spectralresponse, and - RF waveguide. TheparameterizedRFsignalwouldfaith microwave andmillimeter-wave signalsinjectedintothe lator componentsthat permitall-opticalprocessing of RFmodu- signal. Theproject woulddemonstrate theuseof to theiropticalrepresentation anRF undermodulation of microwave signals of and demonstrated theupconversion electro-optical modulator, fabricated aphase modulator, mer system(cladding-core-cladding) suitable for anactive - apoly At project completion, wesoughttohave developed masseffect. weapons of production of ter-wave essentialtodetectthemovementand imagers higher resolution all-weather syntheticaperture millime- millimeter-wave componentstoenable compactand loss, size, weight,andpowercompared toconventional routed, combined, andprocessed withreduced signal into anopticalsignalrepresentation. Opticalsignalscanbe directly optical modulator microwave toupconvert energy The project goal wastodesign,build,anddemonstrate an - - the corona method, and demonstrated the expected resul- Regrettably, there was insufficient time once these neces- tant birefringence using ellipsometry and prism coupling. sary preliminary steps were accomplished to integrate all Poling is required to align the as-deposited chromophore of the above building blocks into a working demonstration molecules to elicit their non-linear optical behavior in the device. However, we established a capability to design, fab- presence of the polarized optical field and modulated elec- ricate and characterize polymer waveguides, developed a tric field produced by co-propagating RF signal in the process to obtain and fabricate electro-optically active microstrip. We also obtained RF test probes to inject test polymer layers, pattern small featured metal patterns on signals into the micron-scale RF waveguide from RF signal polymer, corona pole to achieve optical activity, and also generators and demonstrated their use and designed a test design and test novel RF waveguides to couple modulation method using a length of test RF microstrip. into the optical upconversion device. The strong partner- ships developed between team members, staff develop- By project’s end, we demonstrated the building blocks of ment, and new capabilities made possible by this project each step necessary to fabricate the polymer electro-optical will undoubtedly impact future projects vital to our modulator: a core/cladding system, inverted rib waveguide national security. design, method to fabricate the inverted rib using emboss- ing as well as reactive ion etching, method to pattern the silver microstrip and plate additional silver to necessary thickness, method to corona pole the active layer, and test fixtures and procedures to inject the microwave signal into the device to demonstrate operation. PN11060/2382

17 Advanced Sensors and Instrumentation 18 PN12064/2465 Conceptual 20-slab designwitharrows indicating incomingparticles. stray light.Inset: of alighttightbox tominimizetheeffect inside of measurements. Bothslabs areing andlightoutputlaboratory located Two thinscintillator slabs connectedtoMPPCsusedintheinitialtim- required detector. toscalealargerand more efficient project thepath willallowamore precise determination of thesecondyear.the endof Thelessons learnedfrom this that willbetestedinareal beamenvironmentdevice by firm ourmodelpredictions. We expecta proof-of-concept tocon- tor andtestingkey componentsinthelaboratory have focused Ourefforts onmodelingthisdetec- sensitive. rates overwhichthedetectoris allowing awiderrangeof run at highratesthe lightcollectionandcanindividually ing information. Thesmallslabs have lesstimespread on thetim- highspeedlightsensorwhichpreserves individual thinslabs eachwithitsown Our conceptutilizesdozens of tor smearingoutthelightsignal. thedetec- andthesheer sizeof multipleevents pile upof showthat approacheling efforts tobeinadequate dueto thelight.Ourmod- read fast outof tor andfocused onvery plasticscintilla- used large, singlepiecesof efforts Previous noise. thegamma-ray ing topulltheneutron signaloutof plasticscintillator andusesprecision tim- tions inlayers of timescalesfordifferent neutron interac- andgamma-ray the cial nuclear materials. Thisdetectortakes of advantage spe- interrogationmay of active have arole inthefuture of neutronsA high-rate-capable, fast non-volatile detectorof Sean C.Stave Prototype FastNeutron Detector Advanced Sensorsand Instrumentation material thatcanbeusedanywhere. using aplasticscintillator, anon-volatile ity todistinguishneutronsfromgammarays We aredevelopingadetectorwiththeabil-

neutrons while increasing and gamma-rays the efficiency. throughout theyear toimprove theseparation between tion. Itisexpectedthat algorithm work willadvance thefinal applica- mimic anenvironment closertothat of neutron sources, withlater neutron beamtestingthat will and testinga20-slab systemwithgamma expect tobegin scale upfrom thecurrent 2-slab 2013,we system. Byearly to FY 2012.Someelectronics designwork willbenecessary for thisapplication yet required significantspendingin digitizer.per channelof Nanosecondscaletimingiscritical iscost brought togetherandtested.Thelimitingfactor In FY2013,allcomponentsfor a20-slab modulewillbe desired sensitivity. thedetector whileyieldingthe the sizeandcomplexityof digital pulseprocessing) were brought togethertoreduce tiple cutting-edgecomponents(MPPCs,waveform digitizers, butresults background the gamma-ray Mul- are promising. the digitalsignalsrequires more work inFY2013toreduce based onthetimebetweenpulses.Theprocessing of slabs. Neutrons couldbedistinguishedfrom gamma-rays A time-of-flightmeasurement wasperformed usingthetwo 2MeVandup.detect neutrons withenergiesof to cient (<3nsFWHM)andthelightoutputwassufficient confirmed that thetiming - resolution wasmore thansuffi andneutrongamma-ray sources were used,andtesting digitizer tocapture Both signalsfrom thelightsensors. lator slabs separated by 30 cmanda500MHzwaveform testingwasperformed usingtwoplasticscintil- Laboratory timing resolution. intrinsic andhave sufficient photon sensitivity lower voltagesthanphotomultipliertubes,have single measure 3mm ×3mm,run Thesesensors atsors. much (MPPCs)werepixel chosenasthelightsen- photoncounters theslabs, smallmulti- not work. Due tothecompactsizeof small orthetimingresolution theconceptwill istoocoarse, thelightsignalistoo that light.If with thecollectionof between 2and10MeVthetimingresolution associated expected lightoutputfor andneutron gamma-ray energies Two key properties were identifiedandstudied:the electronic readout andbettertimingproperties. of channels had betterlightcollectionbuttheslabs hadfewer design usingthintubeswasalsostudied.Thetubular tled onduetotheirsuperiorperformance buta asagroup for thedetector. thebestgeometry gate Thinslabs were set- Modeling studieswere performed usingGEANT4toinvesti- Biological Sciences Analysis of Functional Potential from Metagenome Data Lee Ann McCue quality). In addition, we are porting the PhyloFacts database In recent years, the technology to generate of annotated gene families to our computer system, and DNA sequence data has out-paced the devel- have identified an additional source of gene function information that will enhance the annotations currently in Phy- opment of analytical tools for these data. loFacts. Specifically, we are designing a computer script to Novel computational methods to analyze associate the functional annotations of genes contained in the large quantity of sequence data being the SEED database with those in the PhyloFacts database. generated are needed, and will be essential In addition, we have identified a dataset on which we will for DNA-based studies designed to decipher test the approach. Illumina sequence data consisting of the functional role of microbial communities approximately 8 million sequence reads for a defined mix- in the environment. ture of genomic DNA from 22 microbes are available. We are expanding the PhyloFacts database to ensure that the species of interest present in our test data are represented Microbes exist and function in communities – studies of in the database, and in so doing, identifying methods and natural environmental or human-associated sites – often infrastructure that we will need to continue to expand the revealing hundreds to thousands of co-existing microbial PhyloFacts database so that there is a broad representation species. Interactions between microbes in a community of species. This will be key to ensuring the utility of our are complex, as is the impact the community as a whole approach in the analysis of microbial community sequence has on its environment. Thus, deciphering microbial data from diverse environments. community activities in their natural environments is a prerequisite to understanding their functional role in In FY 2013, we will finish porting and expanding the Phylo- environmental processes. Facts database on PNNL computing systems as well as installing additional software tools needed for the analysis Studies of natural microbial communities have increasingly pipeline. Moreover, a significant effort will be committed to taken advantage of the advanced technologies now avail- determining the sensitivity and specificity of the approach able to sequence DNA in a high-throughput manner. It is for predicting the taxonomic and functional composition of now feasible to sequence the DNA extracted from natural the test data set using this approach. microbial communities consisting of many species. Such DNA samples consist of a composite mixture of genomes from all the community members (i.e., a metagenome). While some progress has been made using metagenomic data to infer the presence and abundance of microbial taxonomic groups, inferring their functional contribution to the community has proven challenging. This research will develop a computational analysis approach for analyzing individual (unassembled) DNA sequence reads generated by current high-throughput sequencing technologies to provide information on the taxonomic composition and functional potential of the DNA source. This approach will utilize several established computational tools, combined with a database and novel software developed by our collaborator, Dr. Kimmen Sjölander. The advantage of this approach will be its ability to provide functional and broad-level taxonomic information for a much greater fraction of the metagenome than is currently possible using existing computational methods. This project has been active for 5 months, during which time we have focused on identifying and installing on local Flowchart of the computational analysis approach for metagenome computer systems the necessary tools for standard pro- data. Existing software tools being incorporated are shown in the

PN12008/2409 cesses in our analysis (e.g., filtering DNA sequence reads for hashed boxes.

20 Biological Sciences Aspergillus niger as a Platform for Exploitation of the Advanced Biofuel Producing Potential of Filamentous Fungi Kenneth S. Bruno have developed the capability for advanced biofuels pro- This project is developing new tools that will duced through fermentation. provide the means to discover novel poly- In FY 2011, we successfully generated strains that express ketide hydrocarbons that can be produced metabolite genes from other fungi. We demonstrated that by fermentation. this single gene can generate the compound microperfura- none (mpf). In addition, we successfully developed and tested a means to perform metabolite analyses using gas A number of pathways to energy-dense, infrastructure- chromotagraphy, as we were easily able to delineate the compatible hydrocarbon biofuels and biofuel precursors presence of a novel peak not present within control sam- are possible. Three biological routes to these compounds ples. Development of this tool was a goal for our project for are through fatty acid, isoprenoid, and polyketide synthesis. the first year of the project. This approach can greatly accel- Fatty acids consist of long chains of hydrocarbons readily erate the throughput in and allow for the rapid screening of converted to fuels such as diesel. Isoprenoids are secondary transformants. metabolite compounds whose natural material has been produced for decades in antibiotics and as therapeutic agents. Through genomics and advances in molecular biology, high levels of secondary metabolite production have been sought as a source of advanced biofuel. Like isoprenoids, polyketides have 10 to 50 carbon atoms that vary in degree of oxygenation (hydroxyl and ketone groups) saturation and cyclization. Traditionally, filamentous fungi have been a valuable source of secondary metabolites, but they have yet to be explored as potential biofuels. Our goal will be to build a flexible “chassis” that will allow us to determine rapidly the key genes found in native hydrocarbon producing organisms and express these in a fungal production system with an extensive track record of industrial usage. We will amplify the polyketide synthase (PKS) genes from available fungal genome sequences and constitutively express them by homolo- Advanced biofuels metabolic production process using the cell as the production plant. gous integration into Aspergillus niger. Com- pounds produced by these strains will be analyzed by solid phase micro extraction (SPME) and the In FY 2012, we successfully generated more strains that can structure of the compounds determined by mass spectros- produce biofuel compounds. Compounds known as highly copy and nuclear magnetic resonance. Molecules with desir- reduced polyketides hold promise as new hydrocarbon bio- able properties will be selected for optimization to increase fuels because of their lower oxygen content. We have taken yield. We also aim to discover novel polyketides from fila- highly reducing polyketide synthase (hrPKS) genes from mentous fungi, which could have a tremendous impact on other filamentous fungi and placed them into our produc- biofuels. If polyketides can be made from complex biomass tion system. These compounds are derived from common without preliminary processing into simple sugars, we will metabolic building blocks such as Acetyl CoA. Understand- PN11008/2330

21 Biological Sciences 22 PN11008/2330 and translated intotheCOBRAtoolbox tofor production of established metabolic modelfortook apreviously A.niger from thenormalmetabolism. substrates We that are of part compound codes for asynthase that cangenerate thempf strain containingthemicagenefrom A.nidulans.Thisgene molecule. Usingourhostchassissystem,wegenerated a asmallmpf In ourapproach, weusedtheproduction of changes toincrease production. function. Thisapproach willallowpredicting genetic mass balanceapproach toproduction tiedtoenzymatic metabolic modelingcapabilitiesadvanced for adetailed that willincrease productiontechnology capacity. We also inimplementing isimportant theseprecursors amounts of ing howoursystemmay beimproved toproduce higher Biological Sciences

production chassis. ourbiofuel resource for expandingtheutilityof compounds byoursystem.Thesetoolsare aninvaluable volatile method wasusedtotestfor theproduction of This rapid tosupport throughput screening. analysis During thelastfiscal year, wealsotestedSPME for higher mass flow from oneproduct toanother. choose genesthat canbeupordownregulated toshiftthe themodelmay helpus application acids.Further organic of small canoccurinparallelwithproduction of mpf tion of A. niger overproduces citricacid.Data indicate that produc - wasgenerated underconditionsinwhich ducted where mpf glucose,in converting oursubstrate. Experimentswere con- eachenzymatic reaction togenerate involved amap of mpf

Automated Thermal Image Processing to Identify and Enumerate Targets of Interest Corey A. Duberstein video over the course of a fixed time window into a single Our objective was to explore automatic image. The resulting image then contains the history of a methods to differentiate birds and bats target’s motion, or its track, through the camera’s FOV. from other observed phenomena captured VPS is usually accomplished with a dedicated hardware on digital infrared video. device, but we wrote our own software to perform the VPS processing. The algorithm then detects tracks in the set of VPS images produced from a video recording. Conceptually, Previous research indicated that the use of infrared video a track is composed of a series of objects. An object is a spa- cameras was an effective method to survey the sky for birds tially connected group of pixels that had peak values in the and bats. Unlike traditional methods that require a human same frame. Individual pixels are first grouped into objects observer recording events as they are observed, video using a form of region growing that was tailored to this recording provides a real-time archive of what was observed application. Objects are then combined into tracks using and could be conducted at remote locations. However, perceptual grouping, a general method of image processing the identification of observed phenomena still requires a inspired by human visual perception. In our algorithm, sim- trained observer viewing the video, which becomes both ilar objects that lie in a line or along a curve are grouped time consuming and expensive and, like traditional meth- together as a track. The bulk of the algorithm development ods, is still prone to observer bias. The wind energy commu- work was defining “similar” and “lie in a line or along a nity expressed the desire for a tool that would standardize curve” in an appropriate mathematical form. methods for and reduce cost of assessing wildlife popula- Tracks are reported in a text file for import into various tions in a semi-remote setting. Automated processing and software formats for further analysis. Each track is identified classification of birds and bats recorded on infrared video by the time it starts in the recorded video for independent would enable remote assessment of wildlife populations verification with observer annotations. The text file also potentially impacted from wind energy development, contains a number of measures of each track, such as the enable informed decision-making in a timely manner, mean size and intensity of the objects in the track, and and thereby lower the environmental impact and cost the sinuosity of the track. The sinuosity is a measure of of alternative energy development. the change in direction between successive objects in The outcome of our research is software that would allow a track. Preliminary testing showed that the algorithm for the rapid processing of infrared video by converting detected 96% of the tracks of flying birds and bats visible digital video into numerical information. To accomplish in the video. this, we devised methods to condense recorded video and identify thermal tracks of observed phenomena from an existing infrared video library. The effectiveness and robustness of thermal track identification methods and data extraction techniques were statistically evaluated. A new algorithm for automatically detecting birds and bats in infrared video was designed and implemented in Matlab. The algorithm was developed using a dataset provided by Dr. Sid Gauthreaux of Clemson University. The algorithm design is a unique combination of video peak store (VPS) processing, region growing, and perceptual grouping techniques. Birds, bats, and other warm targets moving through Conversion of digital infrared video into tabular attributes and then the camera’s field of view (FOV) produce bright spots in the classified into functional groups (i.e., sea bird, songbird, bat, insect) video that change position from frame to frame. VPS is the based on thermal track characteristics. process of storing the peak intensity of each pixel in the PN12010/2411

23 Biological Sciences 24 PN11011/2333 parameters for cellattachmentparameters theinflu- byinvestigating arrayed cell group. We identifiedcritical alsosuccessfully singlecellsamong an enable preciseof spatially lysis to buffers that of enabledprototype gradients devices delivered cells,andcreated capture effectively strategies, cell RNA usingmultiple quantitation. We devices developed for single culturing cellsinmicrofluidics devices aspects of allcritical of analysis FY2012,weperformed extensive For inasinglecell. activity andtranscriptionfactor cell cycle use andallowedfor dualcolorimagingfor quantitating celllinesforclonal populations future (singlecellderived) thesystem.Finally,throughput of wecreated four different increasing two-fluidinterfaces, the for precise control of geometries. Additionally, that allowed wedesigned devices “capture different and captured wall” usingavariety of capture cells,whichwere typsinized,loadedintodevices, Wedevices. microfluidic alsocreated designsto several singlecellsusingmicroscopy andmicrofluidicand analyze In FY2011,weoptimizedmultipleapproaches tocapture appropriately linkthemtocellularbehavior. theinherent variabilityto understand amongcellsand need tomeasure molecularchangesat thesinglecelllevel Thus,therenotypic (i.e., cellresponse) isacritical level. to capture thedynamicsoccurringatare thephe- necessary studies have demonstrated that singlecellmeasurements elegant Afew molecular changesat thesinglecelllevel. cell responses hasbeenlimitedbytheinability tomeasure themechanisticbasisfor of our conceptualunderstanding that mediate manybiologicalprocesses.events”) However, apparent heterogeneity incellresponse (socalled“rare- tounravel the wecanbegin (rather thanpopulation) level, insignalingat differences thecell tion. Byunderstanding inthehomogenouspopula- to-cell variability that existeven obscurechanges at cell- theunderlying thepopulation level human benefit.However, traditionalassays that measure thesecommunitiesandengineerthemfor behavior of RNA andprotein)(e.g., islikely toimprove predicting the their environment. Studyingassociated molecularchanges Cell communitiesexhibitcomplexresponses tochangesin Bryan E.Linggi in SingleCells Characterization ofSignalingNetworks Biological Sciences responses andcommunitybehavior. advance ourunderstandingofcellular network variability“states”willgreatly terization andenumerationofsignaling molecular measurementresolution.Charac- tems todecreasesampleinputandincrease works insinglecellsusingmicrofluidicsys- Our experimentscharacterizesignalingnet-

we anticipate submittingat leasttwopublications. responsesdifferential duetocelldensity. From thiswork, multiplecells,anddetermine mechanismsfor gle versus incell response identify differences micropatterning, insin - andoptimizecell In FY2013,wewillcharacterize efficiency S-phaseentry. affects interactions andhowitultimately valuable data concerninghowcellsare modulated by these laminin andE-cadherin.Theseexperimentsshouldyield celladhesion,including of orinhibitors soluble stimulators ables that canbepreciselymodulated usingthesecelland onmicropatterns.these cellsgrown There are manyvari- . We areexperimentsusing entry nowpositionedtobegin microenvironmentonS-phase Identify influenceof cytometry.flow cellswereFUCCI/GFP enrichedfor using highexpressors were generated toexpress aGFPprotein. Finally, dual enter S-phase. To identifyallcells(notjustS-phase), cellsthat cent reporter cellidentification that of allowslive Thisisafluoresthat reporter expressFUCCI. - thecellcycle We additionally created highexpressing clones celllines . Create S-phaseentry celllinesforphenotypeanalysisof customizable.easily 50 µmfrom centertocenter. Thesesizesanddistancesare 20,30,40and50µmdiametercircles withaspaceof of pluronic acid.Wehydrophobic created solutionof patterns theplate inwhichcellscannotbindiscoated witha der of laminin orfibronectin, inwhichcellscanbind.The remain- philic patterns toattach anyhydrophilic substrate, suchas hydro- ~2µmof that allowspatterns withinaprecision of procedure PDMSmold for patterning cellsusinganegative andmicroenvironment.We cell geometry optimizeda techniquestoconstrain Develop cellmicropatterning aims are notedbelow. singlecellphenotype. Thus,ourrevised of for analysis apromising development single cellsandinsteadpursue from for RNAin efforts thisproject, analysis weswitched Considering theseresults andthetimelineresources cells. of optimalgroups wouldallowselectionof device the Parallelization of betweencellRNAdifference levels. cells.Thissituation couldbeacritical analyzed area of deadordyingcellsinthelocal of would allowfor effects as increasing which throughput for thisdecreased survival, Multiple nextstepscouldovercome thislimitation, such robust cellssuchasHela. even viability isa feature of downstream3)thislow to therequirements analysis; of (~30cells)are whenlownumbers delivered due the device attaching cellin of a~30%survival only 2) wecouldachieve three findings:1)celldensitywascriticalinthisexperiment; and supplements,celldensity. Asaresult, wemade typsinization timeandconcentration, mediatype ence of

Combining Proteomic Technologies to Create a Platform for Spatiotemporal Enzyme Activity Profiling Susan D. Wiedner to infection with A. fumigatus on a subcellular level and to We seek to combine the proteomic technolo- find enzymatic biomarkers for infection. gies of activity-based protein profiling Our plan in FY 2011 included studying A. fumigatus with (ABPP) and subcellular proteomics. Individu- ABPS using liquid chromatography-mass spectrometry (LC- ally, these two technologies study different MS). Global cell lysates of A. fumigatus and non-pathogenic “fractions” of the proteome and in combina- related fungi Neosartorya fischeri and A. clavatus grown in tion would provide spatiotemporal informa- the presence or absence of human serum were treated with a novel general cysteine reactive activity-based probe and a tion of the functionally active proteome. new serine hydrolase specific probe. Sample preparation, LC-MS analysis, and bioinformatics data analysis were used to identify the tagged enzymes. The active proteome of the A BPP and organellar proteomics are techniques used in three aspergilli were compared. proteomics to decrease the complexity of the natural cellular proteome, which often contains thousands of proteins. During FY 2012, 48 datasets spanning two culture condi- Specifically, ABPP isolates the active portion of the pro- tions (HS or no HS), two time points (24 h or 48 h), and teome by tagging functionally active enzymes with chemi- three aspergilli were analyzed. We found that human cal activity based probes. In contrast, organellar proteomics reduces complexity by focusing on the proteome of subcellular com- partments. Combining ABPP and organellar proteomics into the spatiotemporal enzyme activity profile (STEAP) will provide information about the functionality of active proteome on the organelle level. On a cellular level, this information would help elucidate fundamental biological processes such as protein translocation, signaling, and cellular response to environmental stress. The STEAP platform will be used in systems biology to compare biologically relevant proteomes such as diseased systems. The following describes progress made in developing STEAP to investigate spatiotemporal enzyme activity of fungi and host response to the lung disease invasive aspergillosis (IA). IA is caused by the pathogenic filamentous fungus Aspergillus fumigatus and is a deadly, costly infection often found in immunocom- promised patients. To date, there is no definitive diagnostic test for IA except postmortem Differences in enzyme reactivity of A. fumiatus over time, dependent on culture autopsy, yet early detection and diagnosis is condition. KEGG pathway = All proteins with [fc]>2 were mapped onto KEGG glycolysis/ critical. Further, a better understanding of the gluconeogenesis, TC cycle, pyruvate metabolism, glyoxylate cycle, and oxidative virulence factors of A. fumigatus and how phosphorylation pathways; yellow = processes occurring in the mitochondria; orange host enzymatic function responds to infection = processes in cytosol; arrows = pathway products are used in other pathways; heat would increase the potential to find ways to maps within each pathway = fold change of 0.5 to 7.5 (log2 scale) calculated for enzyme reactivity over time for each condition (i.e., AMM+HS_24 h vs. AMM+HS_48 h combat this disease. Our goal is to develop and AMM_24 H vs. AMM_48 h); grey boxes = proteins that did not make a 10% protein and use STEAP to investigate a host response coverage cut-off; * = protein maps to more than one KEGG pathway. PN10011/2253

25 Biological Sciences 26 PN10011/2253 is able to evade cellular A.fumigatusisable toevade degradation; microbial defense bymacromolecule andprotein play akeyrole andacidicorganelles in The lysosome host-pathogen interactions. A. fumigatustothestudyof During FY2013,wewillshiftourfocus from thestudyof a hostinfectionmodel. identifyingpathogenic proteins orbiomarkers in chance of host proteome inaninfectionmodelandwillincrease the the of thepathogen analysis willfacilitate the proteome of thethree fungiisbeingprepared. of ison of Theanalysis A.fumigatusandanotherdescribingthecompar of activity publication describingfunctional resulted inahighlevel to infection,suchaslipidsignalingpathways. Thisanalysis fumigatus butnottheotherfungithat may becontributing processes. Ourdata suggestsfunctionsinA. and energy metabolic (includingfermentation) increased reactivityof HS;HSinduced ent overtimeinthepresence orabsence of relatedclosely fungi.Further, protein reactivitywasdiffer reactivity withinHSculture betweenthe wasdifferent serum altered greatly protein reactivity, andthat protein Biological Sciences - - conidia. A.fumigatus inducedbytheuptakeenzyme activity of changesinhost cell infectionmodel(A549)toinvestigate STEAPpathway onanepithelial then implementournew techniques.We will progression (STEAP)usingavariety of reactiveproteins overtimeanddisease allow monitoringof will enrichment.Thistechnology based andorganellar theproteome byactivity- reducing thecomplexityof tively reactiveproteins- weareenriching acidicorganelle effec Bytaggingand intact incelllabeling for opticalimaging. unknownenzymes.Inaddition, wewillinvestigate of ery tate enzyme class coverageinorder tooptimizethediscov- - protease warheadtofacili andageneralcysteine analysis, for LC-MS orafluorophore analysis (rhodamine) for in-gel enrichmenthandle(biotin) eitheranaffinity poration of andendosomes,analkynemoietyfor incor get lysosomes probes willincorporate anacidotropic weakaminetotar Ournew targetacidicorganelles. ABPsthat specifically new germinate and escape leadingtoinfection.We willdevelop defense byresidingwhere inacidicorganelles itcanthen - - Community Diversity and Functional Redundancy of Cellulytic Microbial Communities in Soil Aggregates Vanessa L. Bailey as a whole, and resiliency to perturbation at the fine scale An understanding of the role of natural of soil aggregates less than 1000 µm in diameter. In prior project years, we focused on the metabolism of cellulose, microbial communities in the global carbon earth’s most abundant natural biopolymer, and developed cycle is fundamental to our ability to miti- a highly sensitive fluorescence-based cellulase (b-glucosi- gate the effects of greenhouse gases. To gain dase) enzyme assay to reveal the functional capability and this knowledge, we must bridge the current resiliency of the community to perturbation. We also devel- knowledge gap between the community oped DNA isolation techniques that produced sufficient structure (who’s there?) and the functional high-quality DNA for pyrosequencing of the 16S ribosomal potential (what are they doing?) of microbial RNA gene from individual aggregates. communities. Microscale enzyme assays measuring b-glucosidase activity of individual aggregates showed that the per-volume enzyme activities in the smallest aggregates varied far more Soils are highly complex systems in terms of their composi- than the per-volume biomass of those aggregates, suggest- tion, dynamics, and heterogeneity, with an estimated 109 ing that either the smaller aggregates were favoring micro- bacteria and up to 1 km of fungal hyphae in 1 g of soil. bial communities that were more responsive to substrate Aggregates are the building blocks of soils, and highly stable additions or that potentially active extracellular enzymes individual microaggregates (less than 250 µm diameter) were protected within the aggregates. Our analysis of the bind together to form more dynamic (i.e., shorter-lived), microbial communities present in individual soil aggregates larger (up to and greater than 2 mm diameter) aggregates. revealed that the species richness of an aggregate was thor- These individual and compound aggregates form different oughly sampled with only a few hundred 16S sequences. habitat types and are likely to feature different microbial Analysis of the b-diversity of the aggregates showed that community structures and functions. Additionally, the each aggregate had a relatively unique community domi- smaller, stable microaggregates are the location of the most nated by only a few different microbial types, with these protected, stable carbon in the soil. As these different types dominant microbes varying between aggregates. of soil aggregates are what comprise the fabric of an intact Integrating these approaches, we performed these assays soil, it is clear that the fine-scale aggregates exert a strong in tandem during FY 2012, determining the b-glucosidase influence on the behavior of the whole soil. activity of aggregates and examining the microbial commu- With recent advances in DNA sequencing technologies and nity by 16S pyrosequencing. The richness and diversity of microscale enzyme assays, it is now possible to study the the microbial communities associated with high b-glucosi- diversity of soil microbial communities, activities of com- dase activities were not significantly different from those of munity members, functional redundancy of the community the low b-glucosidase activity communities; however, analyses of variance clearly show that the communities of these two groups differ. The separation of groups is partially driven by differential abundances of members of the Chitin- ophagaceae family. Interestingly, this separation is based on differential abundances, suggesting that either closely related strains not differentiable via 16S sequencing are functionally different or the presence of appropriate substrates in the high activity aggregates enhances the abundance and activity of key organisms that may be present but not stimulated in the low activity aggregates. This research has developed new approaches to link community phylogenetic composition with functional potential. As we demonstrated that this link in soil communities is capable of cellulose degradation, this finding will advance Non-metric multidimensional scaling ordination of the community our fundamental understanding of the global carbon cycle structures of high (green) and low (red) activity samples; symbols in the terrestrial biosphere and partly fulfill the DOE are scaled to reflect the abundance of the significantly differentially mission to address energy and environmental challenges represented (q < 1e-06) member of the family Chitinophagaceae. PN09016/2184 associated with CO2 emissions. 27 Biological Sciences 28 PN11015/2337 resolution visualizations and compositional data of biofilmsresolution visualizations andcompositional data of Together, thesetechniqueswillprovide detailed, high- thebiofilm. microscopy imagesand3Dreconstructions of ment-specific distributions which correspond toelectron spectroscopy (nanoSIMS)toproduce high-sensitivity, ele- ionmass x-ray microscopy (STXM),andnano-secondary fluorescence micro-imaging (µXRF),scanningtransmission with synchrotron-based computedmicrotomography, x-ray electron microscopy capabilitiescryogenic willbecorrelated microscopy withsoftx-ray tomography. Conventionaland light/fluorescence laser scanningmicroscopy andcryo EPScomposition:confocal tures knowledge of andgain niques willbeusedtointerpret biofilm community struc- initsnearest-to-native state. Thefollowingsurveyed tech- biogeochemical interactions,biofilm communities willbe techniquestoelicitthecomplexmicrobial and analytical imaging and multi-scaleapproachUsing amulti-faceted, of influencelarger,how microorganisms pore-scale processes. andprovides insightinto those at themicroorganism-level connects themolecular-scale biogeochemicalprocesses to substrates (i.e.,ganic metalionsandmineralsurfaces) EET.bly howbiofilm EPSinteractswithinor Understanding played akeyrole inmetalcapture, precipitation, andpossi- teins andnanoparticulate-reduced metalsuggestedthat EPS EETpromicrobial- metalreduction. Thejuxtaposition of hydrated 3DEPSwasproduced ahighly during imaging, transfer (EET).Usinghigh-resolution electron microscopy geochemical reactions includingextracellularelectron may play roles microbial infacilitating interactionsandbio- referred substance (EPS)that toasextracellularpolymeric matrix commonly characterizedorganic apoorly posed of biofilms. Inaddition tomicrobial cells,biofilms are com- instructured live microorganisms communitiestermed of In bothnatural andengineered environments, themajority Matthew J.Marshall of Microbial Biofilms Structure andBiogeochemicalFunction and Spectroscopy toCharacterizethe Correlative HighResolutionImaging Biological Sciences sequestration. radionuclides inthesubsurfaceorcarbon cesses suchasthefateandtransportof biofilm influencesonbiogeochemicalpro- microbial biofilmsandseekstounderstand analysis, andfunctionalcharacterizationof for state-of-scienceimaging,compositional This researchdevelopscorrelatedcapabilities

- that average UO was confirmedwhenx-ray absorption spectroscopy found thehigh-resolution electron microscopy.resolution of This discovery of biogenicUO of discovery transmission electron microscopy-enabled atomic-scale aberration-correctedThe recent addition Titanscanning/ of ties suchasreoxidation may andtransport beinfluenced. usingelectron microscopy,observed suggestingthat proper with UO cellandEPSmatrix associated Shewanella Amira 3Dreconstruction of reduced uranium(UO nascent thesizeof During FY2010,ourperceptions of geochemical reactions intheirenvironment. in theirnearest-to-native influencelocalbio- state asthey and theEPSmatrix associated withUO cells involved theserialsectioningand3Dreconstruction of cellsandtheUO Shewanella step to understanding nm-scale interactions of UO nm-scaleinteractionsof step tounderstanding smallerthan1nm.Anotherimportant nanoparticles tion of collected during EM analysis islimitedand complicated by collected duringEManalysis nately, chemicalinformation that canbe theamountof resolved studiesusingEM.Unfortu- biofilm morphology high-resolution,to thinsectioningandfacilitates depth- eter plastic(GelBond™)substrates. Thissubstrate isamiable ~200 µmon5-mm-diam- fermenter (CDFF)toathicknessof taining lactate andfumarate inaconstant-depth(bio)film cultures were usinganaerobic grown minimalmediumcon- films Shewanella for imagingstudiesandchemicalanalysis, reproducible generating highly bio- In FY2012,webegan 2 . Thecellcanbe seenincloseproximity withUO 2 nanoparticle size was smaller than sizewassmallerthan nanoparticle 2 ) nanoparticles were) nanoparticles limitedbythe 2 - andtheobserva nanoparticles 2 -EPS matrix are inprogress. 2 . Reconstructions of . Reconstructions of 2 -EPS. 2 with EPS withEPS

- the addition of chemicals for EM fixation, dehydration, biofilms on complex matrixes. Biofilms grown in the CDFF staining, and embedding. or on the surface of a hollow fiber were examined using high resolution x-ray microtomography (0.74 μm/pixel sam- Correlative soft x-ray-based imaging techniques present an pling) and with a Talbot interferometer (2.00 μm period excellent opportunity to gain chemical and/or elemental fringe pattern). In these studies, biofilm samples were information from ultrathin biofilm sections at relevant spa- imaged dehydrated or in a fully hydrated form. To distin- tial resolutions. A limitation to obtaining high-quality data guish biological material against higher-Z substrates, an is the addition of traditional sample processing chemicals osmium contrasting agent (e.g., stain) was added to some such as aldehydes that interfere with carbon edge spectra samples to differentiate between organic materials and sub- during STXM or produce interfering infrared signatures. We strates. Preliminary results were obtained on the character- developed a new cryosample preparation technique that istics of several phase contrast methods: absorption/ omits traditional sample processing chemicals that produce propagation-based, Fourier x-ray scattering radiography data interference. EM analysis confirmed that excellent (single grating), and Talbot interferometry (double grating depth-resolved biofilm morphology and cell ultrastructures with phase stepping) with tomography data acquisition. Our were well preserved. Samples were then imaged by STXM, findings suggested that phase contrast enhances the visibil- µXRF, or nanoSIMS to obtain nm-scale chemical information ity of unstained hydrated biofilms (relative to absorption that could be correlated to high resolution EM images. contrast), while scatter provides little additional information We investigated carbon chemistry using STXM and found on the hydrated samples. Using osmium as a contrast agent several differences between the cell surface and EPS enhances visibility for absorption-based imaging. For rela- matrix-encompassing biofilm cells. This high-resolution tively thin biofilms (~50 μm or less thick), spatial resolution visualization of biofilm chemistry at the nm scale will loss from the use of a 2-μm fringe pattern must be weighed help determine how biofilms influence local against the ability to image without a contrast agent. biogeochemical reactions. Through collaboration with another project, 3D reconstruc- For pore- and/or community-scale investigations, a variety tions and segmentation of reconstructed slices have visual- of surface imaging tools (i.e., secondary electrons or helium ized pores (or channels) in the hydrated biofilm. These ions) are available. However, understanding biofilm struc- pores indicate regions where water, nutrients, or electro- ture in complex 3D environments such as a soil particle is lytes can flow through a biofilm and may influence pore- considerably more difficult. X-ray microtomography can scale biogeochemical reactions such as contaminant fate provide 3D maps of pore structure on complex matrixes and transport. For FY 2012, we significantly narrowed the such as soil, but obtaining sufficient contrast to image low-Z technology gap between EM and x-ray capabilities. Our organic material against a higher-Z substrate makes detect- experiments have facilitated the initial construction of ing biofilms difficult. Phase contrast imaging can provide both micrometer and nm scale biofilm models. For FY 2013 enhanced contrast of low-Z materials and some methods experiments, we will extend this research and integrate the provide texture-based contrast modes as well. During knowledge from both the micro- and nano-scale models FY 2012 x-ray experiments, we used three physically distinct into a single high-resolution, 3D chemical image of a signatures—absorption, phase, and scatter—for imaging biofilm community.

Left: Reconstruction of a single slice on an x-ray absorption microtomography experiment showing an osmium stained biofilm imaged in water at 13keV and 0.7 um/pixel resolution. The substrate and biofilm are visible in the water region (inside a pipette tip). Right: Segmentation of 50 reconstructed slices of biofilm (representing ~35 µm thick [z-direction]). Water channels have been selected and clearly illustrate inter- connectivity of the channels throughout the biofilm. Internal structure suggests connectivity from outside of biofilm to bottom (i.e., surface attached face). Insets illustrate water channels in green and gross biofilm morphology in white. PN11015/2337

29 Biological Sciences 30 PN11026/2348 and NanoTiO prepared wecharacterized uniquely NanoClay with FMS, functionalnaomaterials.BesidesworkingDevelopment of detailed below. rials. Ourprogress duringFY2012inthefollowing tasksis basedonfunctionalnanostructuredmolecular activity mate- bio- of biomoleculeswithhighefficiency loading densityof transformational technologicalplatform toallowfor high and national security. We have beenworking tobuilda capabilities inbiologicalsciences,bioenergy, environment, biomedical applications enhancePNNL’s andaccordingly and biomoleculesfor targetedbioenergy active and highly method. TheNADH regeneration system wassuccessfully bio electron donorcharacterizedbyspectroelectrochemical electron(LD)-catalyzed transferreaction toregenerate the NADH vialipoamidedehydrogenase situ regeneration of tional nanomaterials.We andreliable established fast in andbiomedicalapplicationsbased onfunc- Bioenergy formediators biomedicalapplications. oxidephene andgraphene withmRNA andelectron transfer applications. Wecal andenergy - alsofunctionalizedgra forsuper highprotein biomedi- loadingdensitypotentially porous structures, area largesurface (upto1000m have uniqueproperties, includinglowtoxicity, opennano- loss.Mesoporousturation silicas andconcomitantactivity linking/polymerization, whichresult insomeprotein dena- cross-structures (encapsulation), conditionsof andharsh smallpore thesupports, sizes,non-openpore areaface of immobilization approaches from lowavailable ismainly sur conventional of immobilization. Ingeneral,lowefficiency free enzymesinsolutionpriorto much lessthanthat of lessor immobilizedenzymesisusually of specific activity Usingconventionalimmobilization supports. methods,the biomoleculeimmobilization usingsolid of There isahistory Chenghong Lei Applications Materials forBioenergyandBiomedical Development ofFunctionalizedNanoporous als (NanoClay, NanoTiO mesoporous nanomateri silicas(FMS)andotheralternative - Our goalsfunctionalized have focused developing onfurther and delivery. loading andfor optimizingthebiomolecularinteraction for high tunable pore chemistry size, andflexiblesurface Biological Sciences and biomedicalapplications. ized nanoporousmaterialsforbioenergy of activebiomoleculesbasedonfunctional- rial platformthatallowshighloadingdensity We aredevelopingatransformationalmate- 2 nanoparticles stabilized nanoparticles at neutralpHfor 2 , and graphene) for highly loaded loaded for, andgraphene) highly 2 /g), /g),

- phosphorus agents. - biosensor for detecting organo asensitive fabrication of fer mediator, Prussian bluenanocubes, andenzymesfor nanosheets were alsofunctionalizedwithan electron trans- Graphene cellimaging. andfor invivo for genedelivery oxidegraphene nanosheetsare excellent nano-platform good response for theirtargets.Theresults indicate that cells.Theprobesdelivered theprobes showa intoliving oxideized thegraphene withmRNA probes andsuccessfully drugs underphysiologicalconditions.We alsofunctional- nanoTiO preparedand otherproteins onouruniquely nanoclay and agents, immunoglobulinG(IgG)for cancerimmunotherapy, nerve organophosphate against butyrylcholinesterase Finally, weobtainedsuperhighprotein loadingdensityof temperature thanthat insideoroutside. atconformation ahigher vianon-covalentinteractioneven onstrate that FMScanhoard theprotein native inanearly GOX therein. conformation Ourresults native nearly dem- of maintainedthe andaccordingly and oligomeric aggregation resisted thethermaldenaturationGOX, butFMSefficiently thefree formational of changesandoligomeric aggregation demonstrated that theheating resulted insignificantcon- GOX trophoresis andintrinsicfluorescence emission of underthesameincubation. elec- Native allactivity nearly the GOX entrapment, whilethefree GOX in solution lost priorto itsinitialactivity GOX compositemaintained89%of at 60°Cfor 1h,thereleased GOX from thethusheated FMS- We found that aftertheFMS-GOX compositewasincubated (NH entrapped inaminopropyl-functionalized mesoporous silica Glucose oxidase andlargely (GOX) wasspontaneously under thesametreatment. ity, whilethefree denatured protein waspermanently conformation native can stillmaintainitsnearly - andactiv released from theheated protein-nanoporous composite FMS vianon-covalentinteraction.Surprisingly, theprotein aprotein canbehoarded in conformation of andactivity undertheheated native condition,thenearly that even theear. bymeasuringtheweightof inflammation We found of test involvesinjectioninthemouseearandobservation for cancerimmunotherapy. potentially (one cytokine) The withFMS-loadedIL-22 we obtainedthebestinvivoefficacy In collaboration withDr. BooninBrussels, Belgium, Thierry carbondioxide tousefulcompounds. of version mate dehydrogenase andalcoholdehydrogenase for con- NADH withforway tocombine theinsituregeneration of pyruvate tolactate. of Weversion have alsoexplored the coupled withlactate dehydrogenase for theenzymatic con- 2 -FMS) at 20°Cviaadominantelectrostatic interaction. 2 . We observed the slow release of theseprotein theslow release. We of observed

- Directed Strain Evolution through Riboswitch-Controlled Regulatory Circuit Alex S. Beliaev these regulatory circuits and applying selective pressure, the The project explores state-of-the-art meta- system will yield strains that have evolved to overproduce bolic engineering approaches for scalable the riboswitch-specific metabolite. The proof-of-principle work proposed herein will exploit the applicability of ribo- and cost-efficient production of biofuels and switch-controlled circuits for directed strain evolution and bioproducts. This is accomplished through product yield optimization. The expected outcomes of this development of new synthetic biology tools project include 1) engineering of synthetic regulatory cir- that will drastically accelerate the process of cuits; 2) developing approaches and toolkits for directed metabolic engineering using principles of strain evolution; and 3) conceptual developing of metabolic directed evolution. engineering and synthetic biology strategies to apply to the design of novel platforms for cost-effective biofuel production processes. Current metabolic engineering strategies involve modifica- Despite a mid-year start, the project has made significant tion of specific targets (reactions, pathways), that lead to sig- strides towards developing synthetic regulatory modules nificant improvements in product yield and productivity. As that can modulate gene expression in a ligand-dependent this approach requires clear understanding of the organ- manner. For the proof-of-principle demonstration, we ism’s genetic and metabolic makeup, its success is often selected the naturally occurring riboswitch of Shewanella confounded by unanticipated regulatory features, feed- oneidensis MR-1 involved in cobalamin (vitamin B12) biosyn- backs, or interactions. In contrast to direct engineering, thesis and used it to build a set of synthetic transcriptional strain evolution does not require a priori knowledge of the repressor molecules (LacI or TetR) whose 5-foot untranslated system and takes advantage of genetic diversity and evolu- region contained the riboswitch sequence. These constructs tionary forces. With few exceptions, however, the selective were subsequently incorporated into our synthetic two- pressure can rarely enhance product yields, as natural selec- component circuit to control the expression of reporter tion forces will target the functions essential for survival. (GFP, YFP) or antibiotic marker (kanamycin or gentamycin) cassettes in a ligand-dependent manner. The latter is We propose to address this issue by using riboswitch-con- achieved through translational attenuation of LacI and TetR trolled synthetic circuits that transform cellular input into a by cobalamine, which positively affects the expression of signal that triggers expression of growth-essential function. the output module (reporter or antibiotic markers). We have By placing genes essential for survival under the control of also modified the phenotype of the host organsm (i.e., S. oneidensis MR-1) to start testing the dynamics of the constructed circuit function and evaluating its performance based on the response curve quantitatively linking ligand (cobalamin) concentration and protein levels. The synthetic riboswitch-controlled circuit will be modified as necessary to display a broad dynamic range, low background response levels, and specific sensitivity integrated into S. oneidensis MR-1. The ability of the riboswitch to confer antibiotic resistance only in the presence of cobalamin will be used to drive the evolution process toward strains displaying (significantly) elevated levels of vitamin B12 biosynthesis. If we are successful in demonstrating the utility of synthetic riboswitches for directed strain evolution, we propose to initiate the development of synthetic regulatory circuits for detection specific metabolites. These Development of riboswitch-controlled cobalamine-dependent synthetic should include efforts that will utilize and strengthen regulatory circuit. (A) In the absence of cobalamine, the LacI or TetR PNNL expertise in computational biology, molecular protein shuts down gene expression by binding to tac or tetO promoter. modeling, and combinatorial chemistry. The range of (B) When cobalamine is present, the sequestration of ribosome binding applications involving synthetic riboswitch-based regu- site (RBS) blocks the translation of LacI/TetR subsequently derepressesing latory circuits include biofuels production, biological the tac- or tetO-driven gene expression. monitoring, and national security. PN12025/2426 31 Biological Sciences 32 PN11033/2355 nium fieldexperiments. engineeringrecombinantfacilitate phage for ura- Geobactersp. important environmentally (YF-12)to uranylreductases optimized in and expression of be isolated andoptimalbacteriophageinfection abundantnaturally bacteriophagesat RifleIFRC can that canbedeployed example, at DOE sites.For wasteusing bacteriophages radioactive cleanup of comes includeanovelprototype techniquefor potential long-termfield application. Expectedout- metals, bothanovelscientificbreakthrougha and cleanup withapplications tootherredox-sensitive bioremediationnew approaches for uranium enhanced.Thisworkshould begreatly willcreate U(VI)toU(IV) reduction of therate of riophage vectors, are deliveredinto environmental bacteriumthrough bacte- nique. Whentheselecteduraniumspecific reductase gene(s) usingphagedisplayU(VI) through tech- directed evolution This research specific willcreate active reductases for highly increases U(VI). and thereoxidation inoxygen levels of ther, acetate injectionsmay result termination in of uraniumreduction. Fur limittherate andextentof tially by iron reduction, andsulfate whosepresence may poten- rather, thissoilsystemisdominated of thebiogeochemistry Uraniumisnotanessentialnutrientface. for bacteria; brought about through acetate injectionsintothesubsur and immobilizedbybacterial-mediated bioreduction Many toxic metalsandradionuclides canbeprecipitated nium orotherradionuclide wastesat contaminated areas. microbes ura- isapowerfulmethodfor bioremediation of uble U(VI)tolesssolubleU(IV)through environmental sol- areas posingthreats population. Reductionof tocivilian intolarger disperse willeventually quences becausethey environmental conse- soluble wasteswillcausesevere cant environmental problem. Withoutamendment,these solubleradionuclides isasignifi - nium orotherlong-lived, groundwater,Contamination of soils,andsedimentsbyura- Thomas C.Squier Reductase onBacteriophageSurface Directed EvolutionofUraniumSpecific Novel Contamination atDOESitesVia Enhanced Bioremediation ofUranium Biological Sciences mation toenhanceU(VI)bioremediation. Research Challenge(IFRC)andusethisinfor face contaminatedRifleIntegratedField bacterial interactionsintheuraniumsubsur to studysystematicallythebacteriophage- We proposeamultidisciplinarycollaboration (red) tostandards: uranylnitrate (blue)andUO T7-Ab with uranium(VI).B. andcomparisonof Theintegration purifiedT7-Ab from themountsectionsamplescollectedafter incubation of U(IV)presentedXANES analysis ontheT7-A b of uranium (IV) presented on the surface of T7-phagebearing Ab uranium (IV)presented of onthesurface of - - - - from each subunit of Ab from eachsubunitof copper andiron. Itispossiblethat oneelectron isdonated proteinits precursor interactwithmetalionssuchaszinc, amyloid-beta(Ap)protein. Apand of abnormal aggregation manuscript). Alzheimer’s diseaseisassociated withthe andEnvironmental Microbiology in humanbrain(Applied likeuranium similarly theamyloidpeptide(Ab suggestedthesepeptides may reducetide sequenceanlysis chromatecapable anduraniumbioremediation. of Thepep- bacteriophagemutants ments have identifiedseveral pled uraniumabsorption, anduranylreduction measure- recombinant phagemutations for chromate reduction, cou- of of pools andfinalstrains.High-throughput screeningfirst and bacteria strainsGeobactersulfurreducens, containingcDNA from In FY2012,wecreated phage-library exhibiting Ab TEM andXAFSfound that recombinant T7bacteriophage peptide (Ab engineer T7bacteriophagebearingUraRandab chromate reduction. Finally, wedemonstrated anability to from thesepoolsshowedpromisingsmall peptidesderived for chromate/uranium reduction andfound that several established threefully randomizedmutation poolscreening the coordinates 3S2Y). tothePDBdatabase (file We success- UraRandsubmitted structureWe of solvedthex-ray crystal phage inUraRfrom GluconacetobacterhanseniiATCC 23769. uranium reductions andfound thebestcandidate for further In FY2011,wetestedthree metalreductases different for a putative radical(Ab a putative mate anduraniumreduction. showedsignificant potential for chro polypeptide catalytic - ner. Ourstudiesfound therecombinant phagebearingthe (MR-1), primarily screening oneidenesis(MR-1),primarily the Shewanella 2) capable of chromate2) capable anduraniumreduction. of 2 canreduce U(VI)toU(IV)at thephagesurface. :Ab , electron donation from Ab +•) inametaliondependentman- phagesamples.A.Theuranium map 2 (red), suggestedUvalencestate (M21), riflensis(M21), G. phage samples phagesamples peptide. ) aggrigation ) aggrigation –amyloid –amyloid creates phage phage

Higher-Throughput, More Sensitive Stable Isotope Probing Helen W. Kreuzer spectrometry (LA-C-IRMS) method to be described in the lit- This project develops sensitive, higher- erature, reducing the sample size required for analysis by throughput stable isotope probing nearly 1,000-fold (see figure). A paper describing this system was published in Rapid Communications in Mass Spectrome- technology, including more sensitive instru- try and attracted national attention as highlighted in sev- mentation for measuring isotope ratios and eral media outlets, including ScienceNOW and Wired. We methods for targeting specific microbes in a received additional funding from an external client to mixed population. modify the instrument for measuring O isotopes and were invited to collaborate on two fish ecology research proposals, both of which have been funded. This instrument is also An experimental approach that allows identification of being used to analyze C metabolism in natural microbial microorganisms within a community that consumes a par- mats as part of an ongoing DOE-funded research project. ticular substrate, stable isotope probing has been a breakthrough technology that permits a culture-independent association of function with phylogenetic groups. We developed improved methods for stable isotope probing to enable both higher sample throughput and greater sensitivity. These methods will enable outcomes such as: • identifying key processes controlling biogeochemical cycles that sustain all life and build the knowledge base required to predict the role of microbial communities in mitigation and adaptation to global change and subsurface contamination • developing new approaches to environmentally sustainable bioenergy systems and other industrial processes. At present, the sensitivity of stable isotope probing is limited by the methodology, which selects stable isotopically labeled nucleic acids by buoyant density centrifugation. For Single hair with three holes made by laser ablation. Each hole provided a labeled molecule to separate from its unlabeled brethren, sufficient material for several carbon isotope ratio measurements. it must incorporate a significant percentage (e.g., 20%) of the labeled atoms, which can obscure the detection of early incorporation events in a community. Our approach is to In addition to the LA-C-IRMS system, we developed a isolate ribosomal (r)RNA from a community exposed to a capillary absorption spectrometer (CAS) for measuring the 13 isotopic content of CO . This device is the most sensitive C-labeled substrate of interest, subject it to a sequence 2 (phylogeny)-based separation, and measure the 13C content instrument ever developed for measuring the stable isotopic content of CO and offers the potential for analyzing the of sub-populations of the rRNA directly. Direct measure- 2 ment of the rRNA isotopic content will provide greatly C from a few microbial cells. A manuscript describing this enhanced sensitivity (less than 0.1% enrichment will be system was published in Reviews of Scientific Instruments, detectable), permitting the association of phylogenetic where it was noted as a research highlight. Further, we groups of micro organisms with the earliest events in the showed that the CAS could be coupled to a laser ablation breakdown of cellulose and enabling finer time-resolution unit and measurements made in flow-through mode, analysis of the process. Further, our approach should another key technological innovation. greatly decrease the time required for analysis. This technology was deemed so promising that it received Technology development. One part of this project funding from the Environmental Molecular Sciences Labora- focused on developing new technology for C isotope ratios tory (EMSL) via a partner proposal, in which CETAC (a manu- to decrease sample size requirements, important for the facturer of laser ablation systems) and Opto-Knowledge analysis of microbial communities. We developed the most Systems Inc. (OKSI-a manufacturer of capillary optical wave- sensitive laser ablation-combustion-isotope ratio mass guides) donated parts and staff labor. EMSL invested in a PN09036/2204

33 Biological Sciences 34 PN09036/2204 ready for use. made isotoperatio measurements, andtheinstrument is withtheisotoperatio massspectrometer and this device more thanstandard technology. sensitive We interfaced approximately 1,000× 10ngC, theoretical of sensitivity has a (such ascells)from aqueoussolutions.Theinterface in dissolvedsolutes(suchasRNA) orsuspendedparticles forconstructed measuringCisotoperatios aninterface insmallamounts. Wesolution andtypically therefore itis recoveredhybridizationa orbyspecificcutting, in Whether thetargetRNA iscaptured bymagnetic bead thisseparation.chromatography toachieve electrophoresis andhighperformance liquid capillary theirsmallsize. Wefrom thepopulation used becauseof interest, whichgenerates fragmentsthat canbeseparated of (DNAzymes) from theorganism(s) tocutRNA selectively approach usessequence-specificdeoxyribonucleozymes mixture. Pioneered byateammember, thesecond from acomplex atargetorganism capturedfully RNA of tematic studies toincrease yieldandspecificitysuccess- beendescribed,weconductedsys- method hadpreviously capture specificRNA moleculesfrom amixture. Whilethis nucleotide to probes linked tomagneticbeadsselectively approaches for usesoligo accomplishingthisgoal:- thefirst molecules asproxies We two for theorganisms. developed these based ontheiruniquebasesequencesandanalyze RNA usingapproaches moleculesfrom thoseorganisms cific microbes withinamicrobial communitywastoisolate - spe forTargeting analyzing specificmicrobes.Ourstrategy scientific community. an EMSLcapability available for routine usebythe of imagingsystemwillbepart 2µ.ThisLA-CAS of level carbon,oxygen andnitrogen at the isotopic imagingof thiswillallow with theCASsystemstillunderdevelopment, Photon Machineswithsizeresolution downto2µ.Coupled state-of-the-art G2excimer laserablation systemfrom Biological Sciences

chemical cycles that sustainalllife. chemical cycles outbiogeo- thesecommunitiesinteracttocarry of bers howmem- understanding this project toward thegoal of tor. Together, under techniquesdeveloped wewillapply tor, andanotherteammemberwasnamedakey contribu- - the project andwasmadeaninternalco-principalinvestiga The project leadwasinvitedtobringisotopicapproaches to tions betweenmicrobes innatural andmodelcommunities. logical andEnvironmental Research that focuses oninterac- Bio- long-term research fundedbyDOE’s program of Office project. PNNLhasa another large-scalemicrobial ecology Additionally, theproject teamhasbeenincorporated into Snake basins. River water more managementintheColumbiaand efficient salmon,whichwillleadto migrating of thermal history Oisotoperatios infishotolithscouldbeusedto trackthe of specificmeasurementsLight todeterminewhetherspatially led toanotherproposal (nowfunded)toIdahoPower and instrument for Oisotoperatio measurements. Inturn,this ablation-combustion-isotope ratio mass spectrometer received U.S. State fundingtomodifythelaser Department crash intheherringpopulation there. Inaddition, we whata ledto Pacifictounderstand intheNorth ductivity marinepro - inretrospective studiesof theLA-C-IRMS use of Research Board about aproject collaboration ontestingthe team wasapproached Pacific andfundedbytheNorth sources. Our systemhasgarneredfundingfrom diverse etry The laserablation-combustion-isotope ratio massspectrom- scientific communityviatheEMSL. cells.Thiscapability willbemadeavailableor afew tothe individual resolvedatspatially thescaleof isotopicanalysis ises toprovide rapid auniqueultra-sensitive, methodfor andpromspectrometer- isundergoing development further ongoing absorption projects. Thelaserablation-capillary oped underthisproject have beenincorporated intoseveral Future- directions. Thetechnologiesandmethodsdevel Identification of Functional Proteins Relevant to Bioenergy and Disease Pathology by Multiplexed Activity-Based Protein Profiling Aaron T. Wright organism’s viability and host invasion, persistence, and This project develops a sensitive, broadly pathogenesis. We annotated its adenosine proteins, revealing widespread novel nucleotide binding function. It is well applicable chemical proteomics technology– known that the computational prediction of protein func- multiplexed activity-based protein profiling– tion is often error-prone and incomplete. In Mtb, ~25% of for characterizing functional enzyme all genes have no predicted function and are annotated as activities from myriad organisms. Applying hypothetical proteins, severely limiting the understanding this method improves our understanding of Mtb pathogenicity. We employed MABPP to probe, anno- of lung pathogen pathogenesis and tate, and validate nucleotide-binding proteins in Mtb that experimentally validated prior in silico predictions of biosynthetic energy production. > 250 proteins and identified 73 hypothetical proteins as novel nucleotide binders and chose proteins with diverse To isolate, enrich, and characterize functional proteins in unrelated sequences for an expanded view of adenosine complex proteomes, activity-based protein profiling has nucleotide binding. Further, several nucleotide probe emerged as a technology in the evolution of post-genomic binders are essential or taxonomically limited, suggesting proteomics. In this approach, chemical probes are applied specialized functions in mycobacterial physiology to complex biological systems to directly report on the func- and pathogenicity. tional activity of enzymes within native physiological set- The second phase of analysis was identifying a broad swath tings. To realize the rapid, sensitive individual measurement of functionally reactive enzymes in A. fumigatus that we of multiple enzymes from unrelated families simultaneously cultured in nutrient-rich media/media supplemented in native samples, we transformed this technology into a with human serum. The latter was used to identify fungal quantitative, multiplexed in situ approach. response to the presence of human proteins. Multiplexed In FY 2012, we demonstrated that our multiplexed activity- ABPP revealed significant changes in A. fumigatus metabo- based protein profiling (MABPP) mass spectrometry-based lism and stress response during culture with human serum proteomic approach is capable of contributing to widely dif- over time. It is likely that this pathogen adapts its metabolic ferent biological systems. We applied technology to provide activity within the host lung and fights immune host functional annotation and gain a greater understanding of response by similar functions that we identified in human key enzymatic functions in the lung pathogens Mycobacte- serum culture. We continued by comparing results with rium tuberculosis (Mtb) (bacterial) and Aspergillus fumigatus A. fumigatus to two closely related but non-pathogenic (fungal). We developed novel chemical probes to identify fungi, Neosartorya fischeri and A. clavatus. MABPP revealed protein targets, including cytochrome P450 enzymes, of anti- several important biological functions and pathways fungal agents in Mtb and A. fumigatus. In a biofuel project, upregulated in A. fumigatus compared to non-pathogenic we applied a multiplexed probe approach to identify the organisms when cultured in the presence of human serum. complementary suite of hydrolytic enzymes that facilitate As with earlier studies, environmental adaptation and lignocellulose degradation in the fungus Trichoderma reesei. metabolic response are key functions that facilitate the pathogenic capability of A. fumigatus. Importantly, we In addition to important biological research findings, we identified potential biomarkers of host infection, currently developed novel, improved, technical, and data analysis fea- a desperate clinical need, as the most common, accurate tures of activity-based protein profiling, including 1) a large method of post-mortem infection identification. Our find- chemical probe suite to profile broadly numerous enzyme ings in prokaryotic Mtb and eukaryotic A. fumigatus functions in prokaryotic and eukaryotic organisms; 2) quan- contribute to worldwide efforts to understand and titative data analysis methods, including accurate mass and eradicate these lung infections. time tag MABPP (AMT-ABPP); 3) aligned proteogenomics and hidden Markov modeling with MABPP for improved In FY 2013, we will focus on identifying probe-labeled functional annotation of organisms; and 4) improved peptides using our cleavable approach for both health sample preparation techniques and data normalization and energy relevant research. For bioenergy production, for improved data quality. we will do LC-MS based analysis of secreted enzymes from multiple strains of the lignocellulose degrading fungi, We made significant progress toward understanding the Trichoderma reesei and will characterize hydrocarbon functional proteolytic enzymes in Mtb that contribute to the biosynthetic enzymes in B. braunii. PN10042/2284 35 Biological Sciences 36 PN11043/2365 A. STXMimagestaken onthe11.0.2-beamlineat showingTiO theALS, materi- andmedicine. Theuniqueproperties of in industry nanotechnologies needed toaccelerate safeapplications of cells,providing thedataand theirinteractionswithliving imaging approaches nanoparticles individual toinvestigate We new commencedthisproject inFY2011bydeveloping hybrid solarcells. the of control thatand transport theefficiency essentially ments that determineexciton formation, chargeseparation, align- variations intheelectronic level structure andenergy applications withafocus energy onstudyingphoto-induced matrix. Inturn,thisenablespolymer theiruseinsolar quantumdotdistributioninthe individual standing of troscopy andimagingapproaches toprovide abetterunder Additionally, chemical spec- survival. integrate theseefforts theirbehavior andfunctionincellresponse and standing of under within thecellularenvironment, leadingtoanew nanoparticles aswellindividual plexes, andorganelles specificproteins, molecularcom- the spatial distributionof insightinto molecular-level intact cell.Thisapproach offers cellularprocessestigating withnanometerresolution inthe withx-ray for imagingandtomography inves- ing integrated This project provides 3Dhigh-resolution fluorescence imag- Galya Orr Implications ofNanomaterials for InvestigatingApplicationsand Integrated Nano-ScaleImaging C. andD. showingthetitaniumedge around (yellow), 458 eV. asmall cluster(red) nanoparticle andanindividual C. Spectraof B. Elementalmap for titanium inthearea marked withnmresolution; clusters nanoparticle bythebluesquare inA,showingstructures of Biological Sciences applications innanotechnology. polymers toenableandacceleratetheirsafe interactions withbiologicalsystemsand that willsupportinvestigatingnanomaterial chemical andstructuralimagingapproaches We havebeendevelopingnewintegrated - - 2 nanoparticles (black)accumulated nanoparticles intheperi-nuclear region; resolution using scanningtransmissionx-ray microscopy for thissystemwith20nm investigating mental parameters exposure vides information tonanoscalematerials and irrelevant andpromental conditions,whichleadstolargerparticles - nanomaterials toagglomerate underexperi- of tendency by current limitedresearch technologiesandbythestrong theserelationships have beenchallenged of Investigations properties that make toxic aparticle orbiocompatible. confusionexistsabout the butagreat dealof observed, systemshasbeen onliving effects als toimposeadverse engineered nanomaterithese applications.- Thepotentialof human healthandtheenvironment haslimitedtheuseof nanomaterials on cations, but the unknownimpactof als at thenanoscale have beenexplored for multipleappli - cells exposedtoTiO andcell.Usingalveolarepithelial between nanoparticle thecomplex,reciprocal relationships of understanding environment. Theseapproaches provide unprecedented intheintracellular theparticle about thechemicalstate of information intheintactalveolarcelltogain nanoparticles approaches individual toinvestigate entists, wedeveloped mechanisms.Alsoincollaborationunderlying withALS sci- or biocompatible properties anddelineate nanoparticle the toidentifytoxic nanoparticles individual of actions andfate methods have beenapplied thecellularinter toinvestigate elles intheintactcellwithnanometerresolution. These - withinorgan nanoparticles individual structural imagesof approaches toobtain3Dchemicaland been developing (ALS) at Lawrence Berkeley National Laboratory, wehave techniques andworking LightSource withtheAdvanced Maximizing ourexperienceinsingle-moleculefluorescence in vivo. 2 nanoparticles, weestablished nanoparticles, experi- - (STXM), including the growth of cells on 200 nm silicon environments. To investigate this possibility, we have been nitride windows either when submersed in growth media or and will continue in FY 2013 to use STXM to determine the at the air liquid interface. oxidation state of CeO2 nanoparticles inside the cell and determine whether any changes occur in the surrounding In FY 2012, we developed analysis tools specific to our ques- intracellular environment. Positively charged nanoparticles tions. TiO nanoparticles are relatively non-toxic but can be 2 are more toxic than negatively charged particles that are photo-activated to produce reactive oxygen species (ROS) otherwise identical. To understand the underlying mecha- that have potential to induce cytotoxicity. We currently nism, we modified nanoparticles with primary or tertiary expose cells to photo-activated particles to identify relation- amines, both creating particles with positive surface charge ships between cellular response and particle properties, and but causing toxicity at different levels. We currently use determine whether and where in the cell-oxidative damage STXM to determine whether the functional groups are shed occurs. Ceria nanoparticles were reported to have protective or modified inside organelles to understand the differences and toxic effects, conflicting results that might reflect a shift in toxicity. from cerium+4 to a more active cerium+3 in different cellular PN11043/2365

37 Biological Sciences 38 PN12047/2448 of chemical classes of Ccompounds corresponding to chemicalclassesof of hypha at a10micron resolution, yielding spatial distributions asinglefungal variations inCspeciation alongthelengthof to distinguishspatial LA-AMS theability of chosen asatestof mediawere onanagar hyphaegrown soil samples.Fungal down to10microns hasbeenvalidated onhomogeneous for obtainingmolecularspeciation at aspatial resolution LA-AMS Theviability of soilaggregates. resolved sampling of figuring theexistinglaser ablation instrument for thespatially A late for FYstart thisproject, ourfocus hasbeenoncon- Cat specificlocations. soilorganic of andrelative age(young/labile vs.old/recalcitrant) C, organic soil microbes,relationship location of betweenlocation of avenuescapability for examining the opensavarietyof 1-10 micron anddepthresolution aslittle100nm.This molecular information at alateral resolution between and soilopacity, andtomap andproduce elementsspatially scale observation potential toovercome theproblems of Cinsoil.Thesesophisticated methodshave the variation of tostudythespeciation andspatial andLA-CAS LA-AMS of thisproject existingtechnologies istoapply of The objective up thantoexaminethelarger, spatial-scale relationships. appropriate,scopic scaleishighly making iteasiertoscale at themicro- mechanisticdrivers of improved understanding correct modelpredictions soilsystem.An inaperturbed representquately mechanisticprocesses normay leadto theserelationships intomodelsmay notade- transfer of use empiricaldata collectedat batch andfieldscales,direct Asmostmodelingapproaches scale inintactsoilaggregates. tions andphysical-chemicalcontrols that occurat themicron- distribu- underlying of withnoclearunderstanding analysis, Cinsoilisbasedalmostentirely onbulksample organic C deposition,microbial utilization, andsequestration of andrelationshipcycling toclimate of change, current views carbon millimeter scale. Althoughcriticaltoapicture of insoilmustoccurat thesub- Ccycling of understanding interacting withtheirenvironment. Atrue mechanistic microorganisms afunctionof insoilisprimarily C cycling M. LizabethAlexander Ablation Sampling Matter inSoilAggregates usingLaser- Mapping andCharacterizationofOrganic Biological Sciences organic Catspecificlocations. soil organicC,andtherelativeageof between locationofmicrobes, new waystoexaminetherelationship We areapplyingexistingtechnologiesin

C bindsmicro- intomacro-aggregates. whereat interfaces labile micro-aggregate identifiable by oxidizing andweathering microorganisms reactions and C willbefound marginsmore at readilyaccessible aggregate C willbeusedtoexaminethehypothesisthat theyoungest inthosedistributions.Isotopicdistributionsof phology mor mass relative toothercarbonspeciesandtherole of microbial betweenthelocationimprove of understanding whichwill multiplesoilaggregates, cal, andisotopicmaps of junction withx-ray toconstruct tomography spatial, chemi- the chemicalimages.3Dimagingwillbeperformed incon- into isotopic information willbeintegrated from theLA-CAS and systemwillbeevaluated, Improvements totheLA-CAS to allowfor generating nearreal-time chemicalmages. and Aerodyne Research. willbeautomated Data analysis Colorado of software withourcollaborators at theUniversity theimaging of In FY2013,wewillcompletedevelopment habitats containing10-100microbes. of the level toallowisotopicimagingat ity tothe100femtomolelevel focus- onimproving molarsensitiv system, withtheprimary mapped aswell. Progress hasalsobeenmadeontheLA-CAS quantitieswillbe C/Nandotherrelevant of Ratios AMS. Ccompoundsasdeterminedbythe to create animageof 20 microns. Initial data are usedtoadapt existingsoftware ~140×120microns at aspatialan area resolution of of Data have beenacquired over for asinglesoilaggregate software required hasbeenvalidated. toperform LA-AMS have beenmounted,andopticalimaging soil aggregates morphological variations. Recently,observed water-stable 10 microns. Total data acquisitiontime wasunder10min. atcontaining compoundsobtained at aspatial LA-AMS resolution of carbon-andnitrogen- classesof several The lineardistributionof

Massively Parallel Sequencing Technology for the Forensic Identification of the “Unknown” Biological Threat Agent(s) Recovered from the Biological Crime Scene Rachel A. Bartholomew

We are using publicly available open source tools to assem- A major challenge to law enforcement and/or ble the MPS reads into contigs. We have tested Ray, which the first responder community is correctly works as-is for Illumina and Roche MPS data but does not identifying the unknown agent at a biologi- properly assemble color space data from Applied Biosys- cal crime scene. One means of identification tems. For this reason, we switched to ABySS, which is also publicly available and can assemble color space data. Con- is through use of massively parallel DNA tinued research into available assembly programs may be sequencing (MPS). The aim of our study is to useful, as our goal is to develop a species identification develop and test computational tools for method that can be used regardless of the MPS platform analysis of DNA sequencing data, with the used to generate the data. Similar in silico restriction frag- particular goal of identifying an unknown(s) ment sets were generated for the assembled contigs of our in a biocrime sample. “unknown” sample and the reference genomes of organisms of interest (using genomic data available in the public domain). Closely related organisms were included to exam- Many current nucleic acid assays and immunoassays for ine the sensitivity and specificity of the method. We then agent identification are limited by the requirement of basic applied classification methods to compare the “unknown” knowledge of the unknown in order to identify or confirm fragment set to the reference sets. Our goal is to determine identification of the sample. We propose to address this probability of the species (e.g., identification), given a set problem by developing and testing computational tools to of references and short read fragments. demonstrate proof of principle agent identification of labo- Publicly available MPS data from Bacillus anthracis, B. ratory samples using an MPS-based approach that requires cereus, and B. subtilis were utilized for the initial proof of no such knowledge and would provide an alternative and principle demonstration of our general approach. These potentially more specific means of identification of an data were appropriate for initial development and testing; unknown(s). The computational tools will provide innova- however, one of the challenges of using data only from the tive means for the identification of an unknown agent, public domain is that the available MPS data did not neces- where the test can be conducted without specific a priori sarily have an exact match (species and strain) with an knowledge of the target. The aim for these new tools is to available complete genome. Only in the case of B. subtilis focus on providing statistical probabilities of an organism did we have matching MPS and complete genome data for of interest being present rather than determining the B. subtilis ssp. subtilis str. 168. Analysis of the B. subtilis complete sequence of that organism. To accomplish the above goal, we developed bioin- formatic tools in FY 2011 that will identify microorganisms with high sensitivity and specificity using only MPS data. Our general approach took short MPS reads from a sample of an unknown organism, assembled these reads into contigs, and performed an in silico restriction digest to produce a set of sequence fragments. The ends of the contigs were not bound by restriction sites and were therefore not included in the fragment set. Identification was accomplished by matching fragments from the sample restriction fragments for a reference set of the complete genomes of microorganisms of interest. Analysis flowgram for massively parallel sequencing data. PN11050/2372

39 Biological Sciences 40 PN11050/2372 rather thanasimplematching percentage. viding aprobability for match eachgenome’s toasample scoring techniquebasedonBayesian probability, thuspro- a have focusedsequences. Recentefforts ondeveloping similar whenthesegenomeshave highly genomes, even discriminate betweenhighandlowpathogenicconsistently U112 (100%).Theseresults showthat ourapproach can rectly identifiedthepathogenic strain(>99.995%)andthe (similar sequencebutnon-pathogenic). Ourmethodcor tularensis holarcticaLVS andF. (pathogenic) novicida U112 the correct genome. We generated reads for Francisella genome, allowingustoassembleandmatch thesereads to producesThis program simulated reads from aspecified etogeneratednaa software simulated packag MPSdata. more thoroughly, weuseddwgsim inthefreelyavailable To testourability todiscriminate betweensimilargenomes were obtainedwithB. cereus. cis strainswithavailable completegenomes.Similarresults equalprobabilityroughly B. anthra- matches totheseveral theB. anthracisMPSdata resulted in of genome. Analysis percentage (<0.04)for anyotherBacillusorYersinia ing (0.81)totheB. subtilis168genomeandamuchlower 168 MPSdata (Illumina)resulted inhighpercentage match- Biological Sciences - tion withNAU resulting from thisdata set. lus anthracisresearcher. We anticipate afollow-on publica- (NAU) ledbyPaulKeim,University aworld-renowned Bacil- sis wasperformed byourcollaborator Arizona Northern generated MPS data usingtheSOLiDplatform.- Data analy sion toNucleicAcidsResearchonline. AlsoinFY2012,we isinpreparationsuite andMPSdata analysis for submis- not justacontigorread. Amanuscript detailingthetool- relationshipunknown genome’s toknowngenomes,and methods, canprovide aconfidencemeasure onan and,unlike other thanBLAST match. Ourtoolsuite isfaster confidenceinthe method provided ahighermetricof thecorrect Whilebothmethodsgave match,(BLAST). our approaches suchasBasicLocalAlignmentSearch Tool used toolsuiteandothercommonly PNNL-developed were matched backtothereferences genomesusingthe thedataset, thefragments in silicorestriction digestof and 24Francisellagenomes.Uponassembly consisting of therefined toolsuiteonanexpandeddata set power of the MPSbioinformatic toolsuiteanddemonstrated the In FY2012,weincorporated theBayesian calculations into Microbial Processes Accompanying Deep

Geologic CO2 Sequestration Michael J. Wilkins

spheric pressure, 37°C), and pressurized (25 bar, 37°C) condi- This project aims to increase understanding tions. As this research progresses, the response of biomass to increasing pressures will be determined. Additional NMR of the impact of geologic CO2 sequestration on microbial biomass and viability. Deep experiments are also being planned utilizing a methanogenic (methane-forming) bacterium. These microorganisms can gen- subsurface locations for CO storage contain 2 erate methane (CH ) from H and CO , and as such are of great microbial populations that may impact gas 4 2 2 interest in CO2 storage environments where CO2 will be in generation, caprock integrity, and other great excess and H2 may be present from the fermentation of 13 parameters that need to be better under- organic material. By using C-labeled CO2 we anticipate being stood when planning and modeling the able to track the activity of methanogenic bacteria under a behavior of these sites. range of pressures found in deep subsurface environments. In tandem with high-pressure NMR investigations, the physio-

logical state of D. vulgaris under elevated CO2 pressures has The capture and storage of carbon dioxide in deep geologic been investigated at the mRNA level using a shotgun RNASEQ formations represents one of the most promising options approach. A comprehensive suite of samples is currently being for mitigating impacts of greenhouse gases on global warm- sequenced for metatranscriptomics; biomass was exposed to ing, owing to the potentially large capacity of the forma- CO2 pressures of 25 bar, 50 bar, and 80 bar before being flash tions and their broad regional availability. CO2 is injected frozen to preserve mRNA signatures that may indicate poten- into formation fluids, nominally saline aquifers, as a super- tial stress responses to CO2 pressures. While preliminary data critical fluid (scCO ) at depths >2500 m that maintain its 2 has indicated that some aspects of CO2 stress response is supercritical state. Studies on “deep biosphere” have dem- shared across other stresses (e.g., pH, metal exposure), other onstrated that microbial communities are present in all up-regulated genes suggest that there may be some novel environments that have been suggested as suitable loca- aspects to the cellular response. In particular, the up-regula- tions for geologic carbon dioxide sequestration. In some tion of a series of hypothetical genes warrants further atten- industries, scCO2 is used as a sterilizing agent; however, field tion. The sequencing of additional metatranscriptomic trials have detected microbial activity after CO2 injection, samples will allow gene network responses to be elucidated suggesting that indigenous bacterial strains are able to tol- and may reveal roles for specific poorly annotated genes. erate these extreme conditions. The persistence of bacteria in these environments is important because these microor- During FY 2013, our plans involve the analysis of metatranscriptomic datasets, continued experiments utilizing both ganisms may affect long-term storage of CO2 in a number of ways. Bacteria may act as reaction sites for increased rates SRB and methanogenic strains in tandem with high-pressure of mineral precipitation, generate mineral precipitates as a NMR, and the pursuit of collaborative research opportunities result of their metabolism, and may also generate methane in this field that may lead to programmatic funding. gas within the storage location. We have focused our current studies on the physiology and metabolism of a model subsurface sulfate-reducing bacterium (SRB), Desulfovibrio vulgaris. SRB are commonly found in deep subsurface environments that are promising locations for CO2 storage. This research has leveraged unique experimental capabilities that PNNL has developed. High- pressure nuclear magnetic resonance (NMR) analyses of growing D. vulgaris cultures have been performed to determine the in situ metabolic state of these cells under varying 13 CO2 partial pressures. Biomass is incubated with C-lactate solution that acts as a source of both carbon and electrons for growth; as the biomass utilizes this compound, increasing concentrations of daughter products (e.g., acetate, bicarbonate) can be monitored to determine rates of A temporal series of NMR scans showing the utilization of lactate by metabolism. To date, we have reproducibly demonstrated D. vulgaris and conversion to acetate and CO (as bicarbonate) under microbial growth in this system under both ambient (atmo- 2 25 bar CO2 at 37°C. PN11052/2374

41 Biological Sciences 42 PN10055/2297 tion. Approximately 10% of atmospheric CO tion. Approximately 10%of magnitude largerthananthropogenic fossil fuelcombus- is thesecond-largestterrestrial Cfluxandhasanorder of carbonperyear, At75–100billionmetrictonsof cycle. it theterrestrial carbon least constrainedcomponentof Cellulose deposits (white) degraded over a 30-hr period by the action of acellulose enzymemixture underconstantflow conditions. overa30-hrperiodbytheactionof Cellulose deposits(white)degraded CO ally-generated microbi- Thefluxof transformations betweenreservoirs. the soil, withmicrobial communitiesmediating manyof intheatmosphere,voirs oceans,plants,and subsurface, isaninterplayglobal carboncycle betweencarbonreser The compound onEarth. and themostcommonorganic plants structural componentof Cellulose istheprimary Michael J.Wilkins and Modeling Model Experimentation,MultimodalImaging, Communities—Integration ofMicro-Fluidic Micro-Fluidic ModelsforStudyingMicrobial vironments. environmental Thesedifferent conditions can biogeochemical processes produces local microendiffusive - and advective withinpore spaceswhereoccurs arangeof environments, cellulosedegradation Within subsurface soil respiration tomodelorpredict. remain difficult the soilannually, yet thetemporalandspatial dynamicsof Biological Sciences land usepolicy. implications forbothclimatechangeand ing carboncyclingintheenvironment,with development ofmodelingtoolsforpredict- understanding ofsoilcarbonfluxiskeyfor degradation attheporescale.Abetter logies toinvestigatemicrobialcellulose coupled withmulti-modalimagingtechno- This projectusesnovelmicrofluidicplatforms 2 from soiltoatmosphere remains the 2 cycles through cycles

- aimed at developing thefollowing: aimed at developing inthisprojectthese data are sets.Ourgeneralefforts into otherprojects beingintegrated ing from anumber of withtechnologiesaris- approaches for terrestrial Ccycling, these experimentscanthenbeusedtorefine modeling Dataspatial obtainedfrom scalesandchemicalgradients. across microbial cellulosedegradation controls onrates of configured toimitate soilpore spaces,thisproject studies Using theaforementioned novelplatforms that canbe globalcarboncycle. ability topredict andmodelaspectsof isfundamentaltoour inthesubsurface lose degradation cellu- pore-scale microbial therates processes of affecting manner.scale inanon-destructive these Understanding ogies tointerrogate theprocesses occurringat thispore multipleimaging-basedtechnol- enable theapplication of environments. Inaddition,areface they engineered to proxiesthese microfluidic modelsare effective for subsur resulting inspecificpore sizesandpore throats. Therefore, columnsthat vertical actasgrains, aseriesof sisting of Microfluidic modelsaredegradation. small flow cellscon- cellulose andthesubsequentrates of within thesubsurface microbial communities boththedistributionof affect • • • community composition. andmicrobial cellulosedegradation rates of affects insolubleelectron donor acceptor and(initially) electron andthespatialhabitat couplingof diversity in howdifferences of an improved understanding cells substrates andproduction of consumptionof imagingof multimodal nondestructive carbonbymicrobial communities insoluble polymeric micro-fluidic models for studyingthetransformation of - In FY 2012, our work focused on tracking the degradation degrade some of the cellulose, and cause aggregation of cellulose within the microfluidic pore networks, and of the remaining cellulose, potentially within a bio- monitoring bacterial growth associated with this carbon mass-extra polymeric substance matrix. utilization. Significant progress has been made in develop- These achievements to date have set the foundation for ing these experimental areas: continued studies on cellulose degradation, with a focus on • Cellulose nanocrystals that are spatially distributed into determining rates of cellulose degradation within specific “resource islands” within the pore networks have been regions of the pore networks. In the near-term, we aim to degraded using cellulose enzyme mixtures (Celluclast) utilize the environmental enrichment culture (described for over 30 hours. These studies in heterogeneous pore earlier) within these experiments to determine cellulose- networks have suggested that deposited cellulose in degradation rates using indigenous microorganisms. Subse- regions of the model that are exposed to lower flow quent work will incorporate oxygen-sensing probes in the rates may be more resistant to degradation by the microfluidic structure (as either thin films or microparti- enzyme cocktail, although the precise mechanism for cles) so that the impact of micron-scale oxygen gradients this observation is as yet unknown. on the rates of cellulose degradation can be assessed. Addi- • A common soil microorganism Flavobacterium johnso- tional higher resolution imaging of labeled cellulose nano- niae was been inoculated and grown within these pore crystals using a new microscope will allow more accurate networks. These experiments have progressed over a cellulose degradation rates to be inferred. Data generated period of several weeks that include multiple growth from this project will be directly incorporated into pore- cycles. Over this period, little cellulose degradation was scale modeling approaches for better predicting these pro- observed, suggesting that these cells were able to sur- cesses in the environment. vive by utilizing only a very small fraction of the total C within the pore network. Calculations of the mass of C needed to support this biomass supported this conclusion. • Another common soil microorganism Cytophaga hutchinsonii was inoculated into a pore network, and appeared to show more activity against the cellulose deposits. Over 30 days, cellulose was partly degraded, and then mobilized within the pore spaces. These species are thought to produce more cellulose enzymes than F. johnsoniae, and may explain the cellulose degradation observed in this experiment. • An environmental enrichment was started using Cellulose degradation within a pore network that was inoculated with the a Palouse soil sample as a starter inoculum in common soil microorganism Cytophaga johnsonii. Cellulose deposits (green) Dubos media with cellulose nanocrystals as the decrease in size over time before being mobilized by advective flow and move carbon source. This enrichment was able to through pore spaces. Biomass is not visible in this series of images. PN10055/2297

43 Biological Sciences 44 PN09047/2215 12 spectroscopy.Raman Allspecies were usingboth grown speciescanbedifferentiated by whether thedifferent onthesamemediainorder ) grown todiscern ces griseus microbial species(Xylanimonascellulosilytica andStreptomy- sure We tosilvernanoparticles. alsocollected data for other expo- signature timeof appeared tochangeasafunctionof guanine, andadenine. TheRaman tryptophan, cation of for E.coli,includingtheidentifi- observed bases previously the literature for aromatic aminoacids andnucleic acid wasconsistentwithdata signature in weobserved Raman (DH5a spectrafor Escherichiacoli single, live (SERS) todevelop Initially, spectroscopy Raman weusedsurface-enhanced these locations infinedetail. spatially microbial activity,a given andweneedtobeable tomap cal and/orchemicalproperties often control the location of environments resulting from overlapping inphysi- gradients thanhasbeenreported Micro elsewhere. more- sensitively through microbial toor biomoleculesat ascaleequivalent andreproducibly trackcellulosemetabolism can sensitively spectroscopicthe research techniques onwhetherRaman specieswillbeusedfor simplicitytoinitiategle cellulytic cellulosecompletely,species that degrade unilaterally sin- the communityorhowsubstrate isrespired. Despitethefew substrate into andassimilation of totheentry organization undefined community;that is,whetherthere isapattern or knowledge about howtransformations occurinanative, and variousproducts comeout.However, there islittle the environment, inthat substrate goes intothecommunity are they oftentreatedcesses. Inshort, as“blackboxes” in metabolic pro- various scalestomediate awiderangeof tally, cellsthat interactat are they acomplex network of Microbial tostudy. communitiesare difficult Fundamen- Vanessa L.Bailey Microbial Community Cellulose DegradationandUptakebya Microscale Spectroscopic Analysesof organized systems. cellular cellulyticenzymesincomplexspatially of substratesbymicroorganismsusingextra- understanding theacquisitionanduptake scales willprovideanewfoundationfor able fashion.Complementaryworkatmicro- has notbeendirectlyobservedinameasur and communitiesispoorlyunderstood derived carbonintoindividualmicroorganisms The acquisitionandincorporationofcellulose- C- and Biological Sciences ) in minimal media with silver nanoparticles. The The ) inminimalmediawithsilvernanoparticles. 13 C-glucose to detect signatures of carbonuptake.C-glucose todetect signatures of

predominately comprised of di202. predominately comprisedof di 307,blue=glass slide. PLS-DA maps indicate that is green fromgreen=di thepeakidentification. 202,or Red=di181, spectralmapping resulting mount.InsetshowstheRaman dry mixed isolate selected of region Magnified whitelightimageof di202 anddi307isolates. however, PLS-DA only isable todiscriminate betweenthe spectraandthoseobtainedfrom PLS-DA;from theRaman excellent directly betweenthemaps agreement derived map themicrobialdestructively community. There was spectroscopy tonon- usedRaman and di307again thethree isolates di181,di202, equalabundanceroughly of In FY2012,weexaminedamixed cultures consistingof journal 9 unknownspresented. Theseresults were publishedinthe PLS-DA the classification was able to identifycorrectly8of isolates withnoidentifyinginformation. Working blind,the these spectrafrom afreshcation, setof wecollectedRaman 3.0%.To only challengetheclassifi- of a standard deviation 96.3%andwith of single class,withaclassification accuracy ing thesporulated spectrafor di202withina andvegetative (PLS-DA)Analysis separated allthree correctlyplac- groups, di202 anddi307isolates. LeastSquares Partial Discriminant thedi181isolates from the showed acleardiscrimination of the61samples of forms. Aprincipalcomponentanalysis microbial spectra distinguishedthedifferent speciesand spectroscopy of andsubsequent statistical analysis Raman di202 pleomorphic andspore having states. bothvegetative rates growth andhabits,and di307)exhibiteddifferent with cutterant dump sites.Theseisolates (di181,di202, from leaf Wisconsin Cameron lab of obtained through theUniversity ity. uncharacterizedmicrobial isolates We usedasuiteof applied signature processing- algorithms toenhancesensitiv and not SERS),optimizedspectralacquisitionparameters, We spectroscopy have(i.e., todirect sincemigrated Raman . Analytical andBioanalyticalChemistry

Mining the Data from Research on Dogs Exposed to Internally-deposited Radionuclides William F. Morgan Fibrotic scar tissue effectively encapsulated a portion of the Given concerns for human exposure in radi- plutonium and prevented its clearance from the body or translocation to other tissues and diminished dose to organ ation-contaminated cleanup areas (e.g., the parenchyma. Alpha radiation activity from deposited pluto- Hanford Site and Fukushima), the increasing nium in Human Case 0269 was observed primarily along use of radiation in medicine and industry, the sub-pleural regions while no alpha activity was seen in and the potential for dirty bombs and other the tracheobronchial lymph nodes of this individual. How- forms of nuclear terrorism, we are analyzing ever, relatively high activity levels in the tracheobronchial archival studies performed to understand lymph nodes of the beagles indicated that the lymphatic the potential health effects from radiation system was effective in clearing deposited plutonium from the lung tissues. In both the human case and in beagle exposure in a large animal model system. dogs, the appearance of retained plutonium within the Further, we are comparing the results to respiratory tract was inconsistent with current biokinetic occupational and accidental exposures of models of clearance for soluble forms of plutonium. Bound nuclear workers in an effort to understand plutonium can have a marked effect on the dose to the the biological effects of exposure to ionizing lungs, and any subsequent radiation exposure has the radiations, with an emphasis on internal- potential increase in cancer risk. emitting radionucleotides. For this study, we have completed the pedigree analysis of irradiated dogs to investigate the potential for genetic susceptibility to radiation health effects. Initially, we identified Life span studies of beagle dogs commenced in the early the primary cause of death and then determined whether 1950s to examine the effect of internally deposited radionu- there was a genetic component to radiation-induced dis- clides on a large animal. These studies were funded by the ease. With the aid of PNNL investigators utilizing programs U.S. Atomic Energy Commission and later by DOE. With a that analyze high-throughput proteomic data, we devised cost in excess of $200 million ($2 billion in today’s dollars), protocols for determining pedigree analysis and are cur- the experiments involved over 8000 dogs and lasted until rently studying potential genetic susceptibility in the dog 1993. The beagle was chosen because it has several positive population. A part of this study serves as a student’s characteristics, including an easy-going demeanor, optimum Master’s thesis, and some of our results have been size, appropriate life span (~14 year mean), and physiologi- published in Radiation Research. cal and anatomical similarities to humans. This breed also naturally exhibits similar cancers at the same frequency as humans. The information collected represents a unique resource; significantly, data from these animal studies can be compared with information obtained from humans accidentally or occupationally exposed to ionizing radiations. Several of these individuals have voluntarily chosen to donate their bodies to science. The long-term retention of inhaled soluble forms of plutonium raises concerns as to the potential health effects in persons working in nuclear energy or the nuclear weapons program. The distributions of long-term retained inhaled 239 plutonium-nitrate [ Pu (NO3)4] deposited in the lungs of an accidentally exposed nuclear worker (Human Case 0269) and in the lungs of experimentally exposed beagle dogs with varying initial lung depositions were determined via An autoradiograph of a paraffin embedded section of lung tissue autoradiographs of selected histological lung, lymph node, from a beagle dog that had inhaled plutonium. The plutonium trachea, and nasal turbinate tissue sections. These studies deposits in the lung and the autoradiograph illustrates the amount showed that both the human and dogs had a non-uniform of plutonium deposited and the alpha tracks emitted from the distribution of plutonium throughout the lung tissue. deposited plutonium are clearly visible. PN10057/2299

45 Biological Sciences 46 PN12052/2453 microbe-facilitated carbon cycling interrestrialmicrobe-facilitated ecosystems. carboncycling upscalingschemeto constructProgressive multiscalemodelsfor the sent withinthelargerpore of network theeffects Microscale simulations are notable repre toadequately - pores andsoilaggregates. thousands tomillionsof the systemishomogenizedandkinetics are averaged over orat particle) thebulkscale wherepore orsoilaggregate microbes see;”for within asingle example, specifically studies oftenfocus eitheronthemicroscale (“the world soilcarbontransformationBacteria-facilitated dynamics within soils. management practicesthat maximizecarbonstorage turnover inthesoilcarbonpoolare neededfor selecting carbon that control therates thefactors of understanding ronmental conditionsonsoilmicrobial communitiesand bon pool.Therefore, envi- ability toquantifytheimpactsof mediatedmicroorganism carbonturnoverinthesoilcar that control therates thefactors of of our understanding toimprove flow andtransport at largerscalesisnecessary within micro-scale environments tothecontrols exertedby microbial communitiesacting that linkthedynamicsof multiscalemodels of anaerobic metabolism. Development methaneduringonetypeof piration andtherelease of carbondioxide duringres leadtotherelease- ultimately of cellulosic materials intomore readily accessibleforms that carbonsuchaspolymeric recalcitrant forms of degrading by major role intheterrestrial primarily carboncycle, amount stored play Soilmicroorganisms invegetation. a carbonintheatmosphere anddoublethe amount of estimated tobelargerthanthe global carboncycle, The Earth’s carboninthe soilsrepresent amajor poolof Haluk Resat and Inter-Aggregate Scales Transformation inSoils:ConnectingIntra- Multiscale SimulationofMicrobial Carbon Biological Sciences inter-aggregate flowandtransport. aggregates) tothecontrolsexertedby very smallscales(withinindividualsoil acting withinenvironmentsthatvaryat dynamics ofmicroorganismcommunities multiscale modelingapproachestolinkthe The objectiveofthisprojectistodevelop

- conditions. regions for saturation under partial aerobic organisms making thoseareasaggregates, more favorable growth the tion willcreate of air-water nearthesurface interfaces Thissitua- pores duringsoildrying. smaller intra-aggregate poresinter-aggregate are expectedtodrainbefore the larger typically effects, capillary In addition, because of according tolocalcarbonandoxygenorganisms availability. patchy betweenanaerobic bacterialgrowth andaerobic airandwater filled pores causes co-existenceof aggregates, saturated thehypothesesthatinvestigate inpartially micro- approaches willbeusedto During FY2013,developed microbial reaction kinetics hascommenced. namics software for pore scaleflow calculations toinclude and thework hydrody onmodifyingthesmoothedparticle - studyresultsbased onindividual-based hasbeeninitiated, models level whole, aggregate of rized asthedevelopment 2months.Progress duringthisperiodcanbesumma- only This project for late started inFY2012andhasbeenactive our groups. tational requirements by previously that were developed rapid progress usingexistingprototypes for compu- different atwo-pronged approachpursue that willallowustomake are required tostudythesecases.Thisproject willtherefore That numerical is,there different methodsthat are several saturation (multiplefluidphases,air, andwater) conditions. for fullsaturation (onefluidphase, andpartial water only) soil processesinter-aggregate have requirements different Computationaltransport. methodsthat coupleintra-and accommodate bothpartially-saturated flow and nutrient microbial communitiestopore-scale modelsthat can carbon-degrading saturatedaggregate-scale, soilmodelof capability willbebuiltbycouplingourindividual-based, modelsandassociated simulation ourproject, thenew For transformation insoils. that mightbecontrolled tomanipulatefactors carbon toidentifyenvironmental these twoscalesisnecessary with microenvironments. Amultiscaleapproach that links microbial communitiesandtheirinteraction dynamics of tions are notable represent toadequately themechanistic Ontheotherhand,macroscopic (averaged) simula- cycling. carbon information overallrates of neededfor estimation of and microbes. Therefore, cannotprovide thelarge-scale they nutrients water andassociated of transport the flow of variability soilmicroenvironments between different and Next Generation Software for Automated Structural Identification of Metabolites Lars J. Kangas During FY 2011, 25 unique chemical reaction templates were This project will reduce the time and cost of developed that support the algorithm with how bonds and biomarker discovery by developing a soft- atoms should rearrange when bonds are cleaved in the modeled mass spectrometer. The programming code was devel- ware product to enable the high throughput oped, tested, and put into the algorithm that makes use of and accurate identification of metabolites these reaction templates. Modeling the correct chemical reac- and other small molecules. Successful devel- tions will enable us to remove a molecule fragment charge opment of the software will facilitate studies model, developed to temporarily remove false positives lipid of fundamental biology and the discovery of identifications that were experienced due to the algorithm disease biomarkers. not knowing which fragments were neutral and which were charged. The real experimental lipid data analyzed included both the spectrometer data that the algorithm uses and data Currently, the structural identification of unknown metabo- from liquid chromatography. Models were developed to pre- lites detected in liquid chromatography-mass spectrometry dict lipid elution times in reverse phase chromatography to studies is based on manual interpretation of their tandem increase the belief in the correct lipid identifications. mass spectra, which is laborious and typically does not yield The algorithm requires two major functionalities for the next a complete identification. The inability to identify endoge- year. First, the prediction of fragment charges will be created nous metabolites and other small molecules rapidly and from the chemical reactions that were developed in FY 2011. globally has significantly limited the discovery of novel bio- Second, the more challenging problem is to have the algo- markers of disease and drug toxicity in metabolomics stud- rithm accept different adduct ions; to date, it has searched ies. An algorithm that models the fragmentation pathways for hydrogen adducts and only a few specific lipids and of molecules in mass spectrometry and generates in silico ammonium adducts. The algorithm needs more generally ion spectra for putative metabolites is required. With a to search for all possible adduct ions for every lipid. large database of these spectra, unknown metabolites can be identified by comparing the observed spectra There were multiple goals and challenges for this project in against the database. FY 2012. We aimed to develop a method to calculate charge locations in molecules. The software algorithm generates in The software product we propose is anticipated to be of silico spectra; knowing charge locations allow it to generate substantial value to laboratories in the medical diagnostic, more accurate spectra, which would in turn enable more biotechnology, and pharmaceutical industries, as it will accurate metabolite identifications. A method was designed reduce the time and cost of elucidating mechanisms of dis- and tested using the NWChem software to compute Gibbs ease pathogenesis and drug toxicity. In addition, the tool free energies for molecules, and from that the proton would be indispensible for biofuels research laboratories, affinities that provide information about what atoms in as the high-throughput structural elucidation of metabolites molecules are more likely to be protonated to give the could greatly facilitate bioengineering experiments that molecules a charge. have as their goal the creation of microbiota that are more efficient in producing hydrogen or ethanol. Specifically in FY 2012, we adopted the software to process NH4 adduct molecules, not just the protonated molecules In FY 2010, we showed that identifying metabolites accu- from the previous year. We extended the software by adding rately, rapidly, and as automatically is possible using a soft- non-lipid metabolites to our development set. This increased ware algorithm. An early working version of the algorithm the molecules in our dataset from 97 to 314. This new metab- was completed, tested, and trained using spectra from olite data came from metabolite standards that had been 22 lipid species (metabolites). A set of 45 test lipids were analyzed at PNNL. In addition, we requested and received purchased, analyzed, and identified using the algorithm in the large metabolite database from the National Institute of conjunction with LIPID MAPS, a database with over 22,000 Standards and Technology to add to our dataset. We are also lipid structures provided by a consortium of lipidomics core promised to receive another dataset from Colorado State Uni- laboratories. Of the 45 test lipids, 39 were positively identi- versity as this data are generated in 2012. Another effort in fied, and five were deemed second most likely identifica- FY 2012 has been to generate chemical reaction templates tions. The last lipid only made it to the middle of a short that are used in the software to configure correctly all atoms/ list of possible identifications. bonds in molecules that experience bond cleavages in our software simulation of a mass spectrometer. PN10061/2303

47 Biological Sciences 48 PN12005/2406 ods have been identified, along with gas impermeableods have beenidentified, along with with experience for thesespecificcelltypesbasedonprevious or theircontrolled population number have beendesigned trapping and nanofluidicassays singlesuchcells capable of yeast). at Micro- Washington (e.g., Bioenergy State University hutchinsonii) aswellat theCenterfor Bioproducts and research Cytophaga inmicrobial communitiesat PNNL(e.g., glucose processing have microorganisms beenidentified for and cellulosedegrading tomicroorganisms, With regard temporal resolution. strong enoughtobedetectedbychemicalimagingat high ment, aswellidentifyingtheinteractiontypethat is that ininteractingwiththeenviron areganisms - efficient into FY2012,progress hasbeenmadeonselectingmicroor commencementapproximatelySince theproject’s halfway omicresearchactive platforms at PNNL. to interactionkinetics, aswellintegration imaging of chemical and nanofluidics for cellhandlingandtrapping, isbasedonmicro- methodology Theproject’s than others. speciesbehave will beextracted,aswellwhycertain better molecular technologiestoenergy, fuelproduction kinetics singlecellor inaccessible inbulkstudies.Byapplying will beunmasked, where becomeotherwise these behaviors Inthisway,entity level. structural ortemporalbehaviors cellsandenzymesdowntothesingle trolled population of The present acon- proposal explores biofuelproduction of these alternatives. andtoleranceto related of lot tounderstand toefficiencies duringthepastdecade, there advanced isstilla stantially these processes hassub- of our understanding ever, if even critical toourcurrent transportation infrastructure. How - biodieselandethanol) arebased. Thelatter especially (e.g., sources, whichcanbesolar-,able wind-,orbiofuel- energy - renew sustainable necessitates developing growth further demands,andfinite fossil fuel increasingreserves, energy sources.accessing reliable Duetoclimate changes, energy prosperity basedon isheavily andgrowth Our society’s Andreas E.Vasdekis Molecule Levels Biofuel Production attheSingleCelland Optofluidics andMicrofluidics forExploring Biological Sciences unmasked. revealed, andbetter-performing traitsare process timinganddetailsareuniquely and moleculelevels.Attheselevels,cellular rates atsingle(orcontrollednumberof)cell This projectwillmeasurebiofuelproduction E. coli.Appropriate highresolution meth - lithographic - channel inpolystyrene; (c)afluorescent (andO (b)animprinted microfluidic 3µmand1µm,respectively; of hutchinsonii withanaverage lengthandwidth (a) Cytophaga will explored andassessed. orsinglecells infew ’omicanalysis and thepossibilitiesof enzymes andsubstrates willbedesignedandfabricated, In parallel,assays for mapping theinteractionbetween imaged eitherinfluorescence ornon-linearopticalmode. will beidentified.Cell-environment interactionswillbe for plasticmicrofluidic andnanofluidiccircuit fabrication modling methods,thealternative, more rapid techniques tions willbeselected.Inaddition tothermalnanoimprint adhere andexhibitstrong environment tosurfaces interac- rene willbefabricated andtested.Cellsthat donotstrongly will beperformed. Trapping circuits- inPDMSandpolysty FY2013,singlecelltrapping inmicro- For andnano-fluidics date. to theplanar(i.e., 2D)films that have beenemployed to tribution canbecontrolled inthree dimensions,incontrast designed, andinitialtestsshowthat thechromophore dis- inthemicrochannelchromophores have been efficiently In thelatter approach, oxygen sensing methodstointegrate labeled cellulose aswelltime-domainlifetimeimaging. involvesthefluorescent imagingof Aparallelstrategy faces. focus onminimizingthenoisecontributionfrommarily sur micro-spectroscopic studieshave been initiated, whichpri- non-linear approaches have beenassessed thusfar. Raman cell-environment interactions, thechemicalimagingof For anaerobic processes. vironment of thermoplastics)for controllingmaterials themicroen (e.g., - integrated onapolystyrenemicrochannel. integrated 2 -sensing) film

- Predicting the Response of Complex Biological Systems Karin D. Rodland nents. Application of systems biology tools to such complex This project integrates data from multiple model systems is another goal of this project. Finally, if the imaging and mass spectrometry (MS)-based goal is prediction of response, then there is an urgent need to apply new computational tools that can accurately simu- technologies to predict complex biological late the behavior of complex biological systems over time system response to external stressors. In and space, and then extrapolate to predict the response addition to furthering PNNL’s capabilities in to perturbation. systems biology, we will explore applications During FY 2012, predictive capabilities successfully developed relevant to human health. under this project include integrating 2D and 3D imaging technologies to understand spatial factors in biological function; identifying modifications to proteins that monitor the Recent technical improvements in genomics, high-through- response to oxidative stress to predict changes in function or put proteomics and transcriptomics, and other analytical disease; developing unique nanomaterials for targeting medi- technologies provide biologists with powerful tools for cal isotopes; developing bioreactors for co-culturing of aero- describing the behavior of simple cellular systems. What has bic and anaerobic organisms; and applying activity-based been missing is predictive capabilities that allow manipula- protein profiling to understand developmental changes in tion of biological systems for desired outputs in bioenergy, enzyme function. The specific tasks undertaken in FY 2012 bioremediation, and mitigation of harm in response to and their outcomes are described below. stressors such as ionizing radiation. Achieving this predictive capacity will require both continued technological improve- Multi-modal imaging of complex biological systems. We ments in experimental systems and in analytical tools, applied unique PNNL capabilities in imaging MS (nanoDESI including imaging and developing new computational and MALDI) to identify the specific brain regions affected by approaches and mathematical models that allow molecular- exposure to neurotoxicants. In addition, we demonstrated level data to be scaled to cells, cell communities, tissues, high spatial resolution (12 µm) and sensitivity (0.2 fmol). and organisms. The goal is allowing us to predict, manipu- Characterize the protein modifications indicative of late, and potentially design multi-cellular systems that con- inflammatory responses. For this task, we first demonstrated tribute to DOE goals in bioenergy, contaminant fate and the ability to detect and quantify multiple brominated pro- transport, carbon sequestration, and global climate change. teins produced following activation of an immune response. This project applies unique PNNL capabilities in ’omics Next, we identified autoantibodies that are indicative of cel- measurements, imaging, and computational modeling and lular destruction in Type I diabetes. Finally, we have demon- simulation to promote the transition of biological science strated the ability to detect and quantify addition of from merely describing biological phenomena to predicting, glutathionine to proteins as a mediator of inflammation. and eventually controlling, the response of complex biolo- Develop and test nanomaterials for delivery of medical gical systems to perturbation, whether in the form of envi- isotopes in radioimmunotherapy. We produced antibody- ronmental stressors, changes in climate, or deliberate conjugate MnFe3O4 nanoparticles and loaded them with manipulation to enhance performance (e.g., production Ra-223. In addition, we obtained data on leaching of Ra-223 of biofuels). Chief among the capabilities required are from nanoparticles in relevant biological matrices. improved imaging systems that incorporate PNNL strengths in MS, nuclear magnetic resonance (NMR) spectroscopy, and Develop a bioreactor capable of co-culturing anaerobic emerging strengths in chemical imaging; one of the specific bacteria and aerobic host cells. We demonstrated the ability subtasks in this proposal focuses on combining mass spec- to co-culture aerobic host cells (human intestinal epithelial) trometry and NMR to provide a 2D image of biochemical and commensal bacteria (Lactoacillus reuteri) in a 3D culture changes in the brain. Integrating and interpreting data that system. We also engineered a membrane system to maintain arise from such combined imaging modalities will require differing oxygen concentrations on either side of a bioreactor. application of novel computational tools for registering and Determine changes in detoxifying enzymes in early devel- adjusting multiple inputs. While existing research has often opment to assess risk of fetal and neonatal exposure. We focused on very simple model systems consisting of one developed activity-based probes for the enzymes that detoxify organism or cell type, it is increasingly clear that predicting polycyclic aromatic hydrocarbons and measured the activity the behavior of real biological systems requires model sys- of these enzymes in early and late neonatal stages and in tems the incorporate realistic levels of cellular complexity, mature mice, demonstrating decreased detoxifying capability while allowing accurate measurement of individual compo- in the very young. PN12061/2462 49 Biological Sciences 50 PN10069/2311 a microbial community. theseenzymes within of measuring theredundancy enzymesandenable novelcellulose degrading of willallowfor thediscovery using massspectrometry theseenzymes andanalysis enzymes. Theenrichmentof these siteof tothe active probes that bindirreversibly proteomics of cellulosic enzymesthrough thedevelopment of theredundancy on asolutiontodetermineempirically but have dissimilargenomicsequencehomology. We focus thisactivity tion hasproduced manyenzymescapable of - community presents achallengeinthat convergentevolu enzymeswithinamicrobial cellulosedegrading of dancy recalcitrant cellulosicmaterial. Measuring theredun- of munity provide stability andsynergismfor thebreakdown theseenzymeswithinamicrobial com- of and redundancy ing second-generation biofuels.Theenzymatic makeup resource- for making develop itaprimary 6-carbon sugars, 5-and plantbiomassiscomposedof 70%of Roughly cellulose. community tocontinue todegrade actsasa“buffer,”tional redundancy allowingthe anenvironmental upset,func- Given lulose degradation. processes important innature,the stability suchascel- of iscriticalto isthat functionalredundancy general belief the communityisknownasfunctionalredundancy. The same functionbutproduced populations bydifferent in outthe carrying capable enzymecatalysts of number of within thecommunitycanperform thesameprocess. The work synergistically, oftenbecause multiplepopulations Unlike microbial asinglemicroorganism, communities improves ourknowledgeabout howbiofuelsare produced. standing howmicrobial communitiesperform thisprocess cellulosicbiomass.Under munities isthebreakdown of providedIn nature, bymicrobial com- service animportant Stephen J.Callister Bioreactors Engineered Within Communities Microbial Anaerobic Degrading Testing Cellulose Stability and of Redundancy Functional of Measurements Proteomics Biological Sciences and environmentalchange. correlated tofunctionalredundancy sure thestabilityofcellulosedegradation degradation. Thesedevelopmentswillmea- to biofuelsproductionviainsolublecellulose approaches fordiscoveringenzymesrelevant based proteomicsprobesandbioinformatics We aredevelopingchemicaloractivity-

- supply protein identifications.supply Resultsfrom the application to proteins wereusingmass spectrometry thenanalyzed F. labeled proteins andunlabeled succinogenes . Samplesof probes for andasubsetof thermocellum each probe for C. within cellbiomasswere extractedandlabeled using protein mixtures secreted mediaandfound togrowth mediacontainingglucoseandcellulose.growth Complex to testprobe selectivity. Eachbacteriumwascultured ona andFibrobactersuccinogenes , werethermocellum chosen bacteria,Clostridium In FY2012,twocellulosedegrading cleave b-glycosidic bonds. almost allshowedgeneralpreference for enzymesthat asingleenzymetomultipleenzymes), ficity (labeling of eachprobe.evaluate Whileeachprobe variedinitsspeci- testsusedto glycoside hydrolase enzymes,withseveral Testing performed oncommercially available wasinitially ity for exo-andendo-gluconasesaswellb-glycosidases. activity,glycoside hydrolysis andhave anticipated specific- totarget were synthesized,designedgenerally chemically 13probes FY2011,atotalof cutter antdumppile. For enzymes withinbacteriarecently isolated from aleaf The probe wasusedtoidentifycellulose-degrading for cellulose, asopposedtoothercarbohydrates.affinity [glucose] moleculestogetherincellulose)andhasahigh breaking b-glycosidic bonds(that connectsugar ble of enzymes showsthat theprobe isspecifictoenzymescapa- commercially five probes.available Initialtestingagainst theremainingsents abreakthrough toenable synthesisof thesis stepstomake thisprobe waschallengingandrepre - probe designisnotoriginal,workingthe first outthesyn- During FY2010,sixchemicalprobes were designed.While using massspectrometry. bind tothechemicalprobes are extractedandanalyzed microbial community, enzymesthat cellulose-degrading from a proteins harvested probe inacomplexmixture of thissubstance. Byplacinga degrading enzymes capable of bindtothe oped toimitate celluloseandirreversibly measurements are- performed viachemicalprobes devel sudden environmental upsets.Functionalredundancy under tocellulosedegradation the changeinredundancy ing cellulose. Thesemeasurements enable ustocorrelate - degrad capablebial communitieshousedinbioreactors of associated withmicro functionalredundancy - amount of Specifically, methodstomeasure weare the developing of probes to C. thermocellum are currently being written up lected for proteomics analysis. During the next stage of the for scientific publication, while the data acquired for F. suc- experiment, the volumetric flow rate through the bioreac- cinogenes are being analyzed with help from a University of tor was increased, altering the population makeup of the Wisconsin-Madison laboratory. Depending on the probe, a microbial community and theoretically altering the func- range of proteins (41 to 219) selective for carbohydrates tional redundancy. Finally, the same pH adjustment was were identified, including 7 proteins having an undefined again applied to the daughter reactor, and samples were role in cellulose processing. The enrichment of these 7 pro- collected as described above. pH is a critical environmental teins using these probes from the complex protein mixture factor in maintaining a healthy anaerobic microbial com- could aid in hypothesis development concerning their role. munity. The implementation of algorithms for analysis of liquid During FY 2013, measurements made during the bioreactor chromatography mass spectrometry probe data and global experiment will be analyzed and data generated from the proteomics data derived from microbial communities have samples collected for proteomics analysis. Additionally, the been completed, and a scientific paper describing these use of MultAlign will play a critical role in this analysis for algorithms has been submitted for peer review. These algo- measuring differences in functional redundancy. rithms are part of MultiAlign, a software tool used to process liquid chromatography mass spectrometry data. Enhancements made to MultiAlign include a dataset to dataset alignment capability of mass and elution time features, and a trace-back capability that allows the targeting of important mass and elution time features for tandem mass spectra data generation, if tandem mass spectra were not generated. Tandem mass spectra are critical for peptide sequence assignment and protein identification. These bioinformatics enhancements are significant in that they allow the proteomics analysis of both probe labeled proteins and complex protein mixtures from microbial communities without the initial need of genomic sequence information. In addition to testing the chemical probes against known cellulose degrading organisms present in a liquid culture environment, tests were carried out on the ability of the probes to label carbohydrate active enzymes within the soil environment. Crude extracts of the important cellulose degrading fungi Trichoderma reesei were added to a soil and successfully labeled. The action of native cellulases in the soil is a critical component in the cycling and sequestration of carbon. Testing of the probes’ ability to perform in the soil is ongoing. To test whether functional redundancy correlates with the stability of cellulose degradation, a systematic perturbation of cellulose degrading anaerobic microbial communities housed within engineered bioreactors has been undertaken. This experiment comprised of sampling a parent and daughter bioreactor to assess natural variability in cellulose degradation prior to the onset of a temporary pH The cellulosome of C. thermocellum is composed of several proteins adjustment in the daughter bioreactor. During this pertur- involved in the degradation of cellulose. The chemical probes synthesized as part of this research project are designed to label bation and just following the re-establishment of pre-per- and enrich these proteins from a complex mixture of proteins not turbation bioreactor conditions, several measurements involved carbohydrate enzymatic activities. were made, including enzyme kinetics, volatile fatty acid production, and gas production. Samples were also col- PN10069/2311

51 Biological Sciences 52 PN12065/2466 rat, a possible indicator of radiation-induced lungdisease.rat, apossibleindicator of heterogeneity the irradiated inthecontrol rat ishigher thanthat of arat exposedto15Gyradiation. Theventilation A similarsliceof a control rat. Right: thelungsof from thedynamicCTimagesof aventilation mapthe greatest). calculated Left:Acoronal sliceof showsmall increase, (on arainbowscale, warmcolors coolcolors Ventilation maps showairvolumeincrease onavoxel-by-voxel basis radiation doses. thechronic responseto gauge toarangeof Our initialtaskincludedexaminingadose-response study lungradiation thisnature. damagefor future workof of areliablesis. Thisproject animalmodel willalsodevelop - imagingandimageanaly indynamicpulmonary advances that thisproject willresulttranslatable inclinically technical physiological response We anddiseasecourse. anticipate translatable thebasic tohumanscriticalunderstanding adiagnosticapproachradiation onthelungsanddevelop of theeffects tounderstand ease models,wecanbegin changes duetoradiation exposure. Bystudyinganimaldis- mechanical thefundamentalpulmonary of understanding animproved thisproject istodevelop of The objective dynamic imagingapproach. mechanical changesthat may bedetectedusinganew caninduce detected byconventionalimagingorhistology Recent work at PNNLsuggeststhat lungdiseasethat isnot exchange. mechanical function,whichiscriticaltogas have examinedradiation exposure impactonwhole-organ damage from high-doseradiation oncology. studies Few cer occurrence from low-dosemedicalexamsoracutetissue long-termcan- radiation have exposure focused effects very pulmonary workers andlargepopulations. Moststudiesof result insignificantradiation release andexposure toplant such asthatCatastrophic inFukushima,Japan can events Rick Jacob to NuclearEnergyProduction Pulmonary Injuryfrom AcuteEventsRelated Biological Sciences and emergencyresponse. timely, effective treatmentsinclinical mechanics tofacilitatedevelopmentof of radiationexposureeffects onpulmonary imaging toenhanceunderstandingarange This projectwilldevelopstate-of-the-art

will becollected asbefore, will beadded. andhistology PFTdata andbiosamples the 0,6,and15 Gydoselevels. include This willbedonewithagreater n,butwewillonly ing at the6-monthandadd asecond8-monthtimepoint. In FY2013,wewillvalidate theradiation exposure andimag- subtle, radiation-induced lungdisease. indicator of sensitive our results suggest that dynamic 4D-CTimagingprovides a the biomarkers have nevertheless, notyet beenevaluated; disease. Itisalsonoted that understood thepoorly tion of - complexevolu represents a snapshot in thedynamic, only in manyPFTresults. that We the6-monthtimepoint observe showedthegreatestlow-dose group variation from control chord compliance, PFTs alsodidnotcorrelate withdose;the quasistatic correlate cantly withdose. Withtheexception of disease(measured inHounsfieldunits),didnotsignifi - of ingly, CTimageintensity, aconventionalandclinicalmetric decreasinggeneity generally withincreasing dose. Interest - showed higherheterogeneity thantheexposed,withhetero- Unexpectedly, withdoselevel. (p<0.05) cantly thecontrol rats variation correlated signifi- of measured bythecoefficient radiation dose. We of founddictive that map heterogeneity ventilation maps madefrom dynamicCTimagingwere pre- thecurrent year’s findingof The mostimportant work isthat collagencontentandweight. excised of for analysis sampleswereand saliva collected.Finally, thelungswere lated rats, afterwhichblood,bronchial-alveolar lavage fluid, venti- 100 mstemporalresolution mechanically onthelive, use. Mostimportantly, 4D-CTimagingwasperformed at for thefree-breathing PFT, andwereceived trainingonits tized rats. Anew, wasacquired wholebodyplethysmograph mechanical properties usingforced inanesthe- maneuvers breathing unanesthetizedrats, andtheothermeasured lung tion tests(PFT):onemeasured breathing infree- parameters func- the rats pulmonary withtwodifferent were evaluated refinedically through process. aniterative After6months, andnon-rigidregistrationprocessing were parameters empir sis usingexisting4D-CTdata. Inparticular, pre-registration significant progress- wasmadeinimageprocessing andanaly During the6monthsbetweenexposure anddata collection, was usedtominimizecollateral exposure. beam. Also, a12-inch-thickcustomleadbeamcollimator shape that enabled repeatable the positioninginfront of were madethatholders conformed totherats’ sizeand animals were ineachgroup. To exposure, facilitate custom experimentation, which were anddata analysis unaware of suchthat thosedoingtheanimalhandling, blind fashion radiation fromgamma a60-Cosource. Thiswasdoneina of 0,6,9,12,or15Gy(n=5eachgroup) one-time doseof rats.25 femaleSprague-Dawley Thethoraxwasexposedtoa four-dimensional (4D-CT)imagingof computedtomography thisyear wastheradiation exposureThe centerpiece and of

- Statistically Significant Forensic Fingerprinting: Protein Analysis of Biological Agents Brian H. Clowers match between an unknown and a laboratory-constructed This effort aims to develop a framework and fingerprint. While definitive matches are rarely obtained, the metric capable of statistically distinguishing level of association between two samples can be assessed using this approach, and a broad range of statistically robust between highly congruent forensic samples relationships may be constructed. One novel aspect of this using residual growth medium protein computation is developing a new sample indexing method profiles. or hashing algorithm that seeks to rapidly identify statistically relevant matches across sample fingerprints. Fundamentally sound statistical methodology is the founda- In FY 2011, three potential biological threat agents were cul- tion of any robust forensic method. While many different tivated under differing conditions, inactivated and stored for sample characterization approaches exist, difficulty in sam- future project work. Using a handling and data acquisition ple comparison is often encountered based on domain spe- protocol designed to extract trace protein impurities, these cific analytical methodologies. This effort aims to develop a samples provide core data used to characterize the forensic statistical comparison framework that integrates laboratory analysis approach. Sample analysis was also conducted with trials with an algorithm that directly identifies similar and an emphasis on carbohydrate and lipid signatures. Because dissimilar features between samples. Most importantly, the the goal is developing a robust statistical tool geared toward resulting methodology will provide forensic analysts with forensic investigations, we established a flexible database likelihood matching statistics between known and unknown architecture capable of capturing a wide range of relevant samples, while preserving many of the fine subtleties of the data sets. While the capacity exists to accommodate a variety differences between samples. The merit of this approach will of data sets, the current database format has been optimized be demonstrated using a series of laboratory experiments to incorporate raw and processed protein data and has the aimed at supporting the bioforensics community. unique capacity to collapse results across different experimental classes. With this tangible database format, we Within the range of observed media proteins, this project examined a suite of hashing algorithms to develop an leverages the frequency of amino acid observation for optimized matching scheme. These “similarity hashes” known samples which serve as an underlying, emergent sig- offer the benefit of rapid searching while returning nature to be employed for future sample comparisons. Each statistically meaningful likelihoods. peptide feature is weighted based upon its relative frequency of observation across the entire media library and, within a Applying our matching approach to a database comprised of given comparison, these weights provide a likelihood of a protein frequency profiles for different organisms and their respective culture conditions, it was possible to distinguish between three different source conditions for an unknown organism. Using a similar approach in FY 2012, an additional strain of Yersinia pestis (lcr-) was cultivated using an extended set of media growth conditions compared to FY 2011. Using the peptide profiling and signature hashing approach a high degree of similarity was observed between protein profiles derived from preparations using similar starting materials. This exercise further demonstrated the ability of this approach to associate known media formulations with residual components found in a biological preparation. During the second half of FY 2012, the signature curation, statistical algorithm, and database architecture were intener- ated into a web-based interface that enables researchers and analysts to examine protein profile associations at all levels, including access to raw data. Using a range of internal and Similarity matrix for observed protein signatures across different culture conditions. The clustering dendrogram highlights media external sample resources, this effort has developed a flexi- formulations with high degrees of similarity based on residual ble sample comparison framework. With emergent growth media components. Detailed signature comparison provides a media peptide profiles as a demonstration case, a tangible mechanism to validate manually calculated associations and forensic analysis tool developed at PNNL for the broader peptide identification. Strong correlations exist between similar bioforensic community is specifically directed at use cases protein types demonstrating promise in a forensic setting. related to biothreat attribution. PN11073/2395 53 Biological Sciences 54 PN11078/2400 interaction between river intrusion andsediment composition. interaction betweenriver geologically, intrusion. Thetwositesdiffer of suggestinganimportant community composition.Atasecond site(B),there wasnoinfluence demonstrates shiftmicrobial intrusion that cansubstantially river intrusion (A) of stage. withincreasing Increasing levels deviations which intrudes intotheHanford 300Area at highriver subsurface theColumbiaRiver, of the change(through time)intheelevation stochasticcommunitydynamics.Horizontal axis: composition given community composition(through time)andtheexpectedchangein changeinmicrobial Vertical betweenthe observed axis:difference necessary describe theconstituentprocesses quantitatively process. Thisapproach to fails infer thedominantclassof studiesonly patterns. Althoughuseful,mostcontemporary research focuses ontheprocesses that govern observed specieschangesacross locations). Currentnumber of howmanyspeciesarearea inagiven andhowthe (e.g., Historically, ecologicalcommunityresearch isdescriptive microbial communities. governs thecompositionof what understand lize contaminants,itisvitalthat wefirst toremove orstabi- through ecosystemsandhoweffectively howmicrobe-influenced contaminantsmove understand weare to ecologicalcommunities.Assuch,if position of influencedbythecom- isstrongly elementalcycling) (e.g., been repeatedly demonstrated that ecosystemfunction toremediateefforts contaminated locations. Ithasalso environmental contaminantsand ence human-derived Itcomesasnosurprise thatcal cycles. microbes caninflu- natural influencingbiogeochemi- ecosystemsbyheavily of Microbial communities play acentralrole inthefunctioning James C.Stegen Subsurface Microbial Communities Understanding theProcesses thatGovern Biological Sciences across theUnitedStatesandbeyond. through remediationofcontaminatedsites improving environmentalandhumanhealth change. Theresultingtoolswillbecrucialfor space andinresponsetoenvironmental community compositionandfunctionthrough retical toolscapableofpredictingmicrobial This projectultimatelyaimstodeveloptheo-

community composition andmetabolic function. microbialused topredict spatial andtemporalpatterns of simulation studies willprovide models thefoundation of geochemistry. Information through theseempirical gained microbial metabolic function,hydrological dynamics,and among touncoverlinkages generated willbeevaluated microbial communitycomposition.Thedata currentlybeing tools tocharacterizehowspeciesinteractionsinfluence interrogate existingdataset withadditional novelstatistical toward predictive simulation models.Inaddition, wewill community compositionthrough spaceandtimework In FY2013,wewillcontinue tofocus onunderstanding user proposal. arecentlyapproved EMSL withinthecontextof analyzed collected andprocessed; thematerial from thesewillbe (23) taken across have akey environmental been gradient andGeoChipdata. Additionalmetaproteomic, samples To date, wehave four datasets that includemetagenomic, theircompositionand metabolic function. ties intermsof andmicrobial communi- spatial variation ingeochemistry 300 Area that usesamulti-’omicapproach tocharacterize Design andexecute fieldcampaignintheHanford alarge gradients. govern spatial biodiversity theprocessesaimed at that understanding toecosystemfunctionandasimulation study biodiversity collaborations includeanempiricalstudylinking microbial governed bysimilar, deterministicprocesses. Two additional showed that across ecosystemsmicrobial communitiesare attools proved robust essentialinarriving inferences andwe Thenew analysis. ecosystemsincomparative broad rangeof taskwere inthefirst a applied to tistical toolsdeveloped sta- including researchers from ChinaandFinland.Thenew collaboration isinternationalmunities. Onenew inscope, that govern microbial factors com- focus onunderstanding collaborationswithinandoutsidePNNLthat Develop new variable) that govern variation incommunitycomposition. thesystem(includinganunmeasured major features of results were collated intoaconceptualmodeldescribing composition from theHanford 300area The subsurface. microbialapplied community toa pre-existing dataset of that influence microbial communities.Thesetoolswere ecologicalprocesses andidentifyunmeasured variables of statisticaltoolsthat Develop new quantify theinfluences as describedindetailbelow. Through FY2012, thisproject focused onthree majortasks as amodelsystem. ulation approaches, usingtheHanford 300Area subsurface experimental,andstochasticsim- by couplingobservational, predictivemodels.Thisproject fillsfor that developing void

Chemistry Biodegradable Linear Polycarbonates

Through Catalytic Polymerizations of CO2 Utilizing Switchable Ionic Liquids David J. Heldebrant Such polymerizations require high pressures needed to provide the available [CO ] in solution and suffer from This project is exploring the direct use of CO2 2 (chemically fixated) in CO -capture solvent chain termination of the growing polymer. Higher concen- 2 trations and activity of the carbonate could increase the systems toward developing more energy kinetics but also the molecular weight of polymers. Chain efficient and CO -neutral processes to 2 termination occurs when polycarbonate chains backbite produce plastics and fuels. and cyclize to form kinetically and thermo-dynamically favored cyclic carbonates. Alkyl carbonates could retard backbiting by using an excess of alkyl carbonate in solution The objective of this project was to demonstrate that alkyl to trans-carboxylate to the catalyst or the growing polymer carbonate ionic liquids can be CO mimics for CO catalysis 2 2 chain. Such a system could kinetically drown out the intra- as they act as a CO carrier, concentrator, and solvent, but 2 molecular cyclization thus enhancing molecular weight. most importantly as a chemical surrogate. Alkyl carbonate Many groups have focused on studying advanced catalysts ionic liquids contain a 0.5 mole fraction of “CO ” in solu- 2 for polymerizations but the majority of them operate on tion, offering unprecedented concentrations of CO at 2 CO such that regardless of catalyst, the reactivity and selec- atmospheric pressure. In contrast, conventional catalysis 2 tivity are similar as the reagents and conditions are the using CO requires high pressures (often > 50 ATM) to 2 same. The lone variable left to change in CO catalysis is achieve such concentrations. Alkyl carbonate ionic liquids 2 the CO itself. Changing the reagent could alter reaction continuously capture and activate CO . Being polar, they 2 2 selectivity, kinetics and catalyst design, and open doors for allow stabilization of charged/polar intermediates common new products that are unavailable starting from CO . in catalytic reactions such as polymerizations. Alkyl carbon- 2 Recently, the groups of Milstein and Samford have demon- ates are negatively charged, favoring complexation to strated just that, showing that carbonates can be hydroge- homogeneous catalysts. The SP2 hybridization of the alkyl nated directly to methanol, and their results prove that carbonate enhances the electrophilicity of the central car- carbonates can be viable reagents in catalysis. bon, making it more reactive than neutral CO2. Alkyl carbonates also are predicted to entail unique free energy Polymerizations of alkylcarbonates and epoxides.

profiles and likely new reactive pathways than to CO2. Polymerizations of CO2 and propylene oxide were performed with varied loadings of diazabiciclo[5.4.0]undec- It was hypothesized that alkyl carbonates were could 7-ene hexylcarbonate [DBUH+]C H OCO -] ionic liquids enhance catalytic polymerizations of CO and epoxides. 6 12 2 2 using zinc glutarate and chromium salen chloride catalysts. Un-optimized chromium catalysts were chosen to determine if transition metal catalysts were tolerant of alkylcarbonate chemistry and to discern the difference in reactivity

and selectivity between neutral CO2 and alkylcarbonates. As expected, the Lewis acid catalyst zinc glutarate was deactivated regardless of alkylcarbonate loadings, which is attributed to alcohol poisoning of the catalyst. The transition metal chromium salen Cl was more tolerant of DBU and 1-hexanol + - but only if [DBUH ]C6H12OCO2 ] Proposed mechanism of metal-free propylene carbonate (PC) synthesis from alkylcarbonate was formed prior to propylene precursors and PO. oxide (PO) addition. PN12013/2414

56 Chemistry + - The loading of [DBUH ]C6H12OCO2 ] had a direct impact on the ester to form PC The cyclization liberates 1-hexanol the selectivity, isolated yield, and the molecular weight of which can then further react with DBU and CO2 to reform + - produced polymer. Similarly to the observations of Dar- [DBUH ]C6H12OCO2 ] and the cycle continues. The mecha- rensbourgh with phosphine and azide co-catalysts, we find nism and reactivity of the alkylcarbonate towards epoxides one catalyst equivalent of the alkylcarbonate the optimal was confirmed using molecular theory. As predicted the loading with respect to isolated yield of polymer and selec- free energy pathways were lower for the alkylcarbonate tivity toward PPC. Increased loadings of alkylcarbonate than neutral CO2 with respect to carboxylation chemistry. above two catalyst equivalents shut down the polymeriza- The production of PC from an alkylcarbonate at 1 ATM of tion as PC began to be exclusively produced. pressure in the absence of a catalyst has never been Metal-free conversion of alkylcarbonates and PO to PC observed before. This program developed a new and cost- + - at atmospheric conditions. If [DBUH ]C6H12OCO2 ] is used effective synthetic methodology to produce PC straight in non-catalyst equivalent amounts the reaction exclusively from CO2 contained from a capture solvent. Ultimately, this favors the formation of PC in quantitative yield. At first we chemistry was shown in this study to capture, concentrate, believed the catalyst participated in the production of PC, and convert CO2 into useful products such as PC and poten- but the reaction proceeds at the same rate even in the tially PPC polymers. While during this project we were able absence of catalyst. Our results showed that PC is the to show this unique reactivity only towards cyclic carbon- favored product whether catalyst is present or not. The ate formation, we believe that the fundamentals of alkyl- + - reaction between [DBUH ]C6H12OCO2 ] and PO occurred at carbonate chemistry have been demonstrated, and our pressures as low as 1 ATM at temperatures as low as 40°C, data will direct the development of catalysts that could albeit slowly. utilize the reactivity of alkylcarbonates for polymerization reactions or other chemical conversions of CO into value- We observed a mechanism consistent with acid promoted 2 added products, including fuels. ring opening of PO by a proton transfer from [DBUH+] followed by ester formation, and subsequent cyclization of PN12013/2414

57 Chemistry 58 PN12015/2416 dently withacross sectionapproachingdently aZ This allowseachatom tobeconsidered toscatter indepen- betweenatomic coherent columnsinthespecimen. effects on ahighangleannular dark fielddetector, averaging is collectedat highanglesandoveralarge angular range form aZ-contrastimage, thescattered intensity(electrons) theZ-contrastimagingtechnique intheSTEM.Touse of inthisproject willmake theexperimentalanalysis Much of lossspectroscopy.electron energy Chemical information canbeobtainedintheSTEMusing (STEM) combinedwithatom probe (APT). tomography imaging inthescanningtransmissionelectron microscope atomic resolution and3Dimagingaswellin-situfluid processes that We aidsindesigningabettercatalyst. employ thecatalytic of scales allowsfor fundamentalunderstanding significance. Thecharacterization onthe atomic andnano- industrial design andcontrollably synthesizeamaterial of beingable to aspectof loop isanextremelyimportant Thecharacterization-synthesis zeolitecatalysts. standing of thefundamental under niques onthenanoscaletofurther tech- andcombineadvanced This project aims toadvance Ilke Arslan Electron andAtom-Probe Microscopes Characterization ofCatalystMaterialsinthe as small10×nm in3D, material canbeanalyzed volumesof of for example rapidly.rials usingSTEMhasdeveloped largerange Afairly mate- ETtoinorganic the nanomaterial. Theapplication of reconstruction algorithms toform a3Drepresentation of tiltangles; theimagesaredifferent thencombinedusing imagesare taken at In electron (ET),aseriesof tomography during theexperiment. so that atomic sitescanbeidentifiedunambiguously intheimage(unlike conventionalTEM) contrast reversals image where changes infocus andthicknessdonotcause yieldsanincoherent imaging). Thisdetectiongeometry on atomic number (andhencethenameZ-contrast 300 nm reconstruction methods. andalignment images acquired, tiltrangeachieved, 1 nmto~5dependingon thevolumesize, number of more environmentallyfriendlyfuel. energy resourcestoprovidelessexpensive, the conversionofrenewableandconventional and processesstudiedinthisprojectrelateto by car, plane,andboat.Thecatalyticmaterials fuel isextremelyimportantfortransportation With ourcurrentglobalbusinesseconomy, Chemistry 3 , with the 3D resolution varying from justunder , withthe3Dresolution varying 3 to as large as 300 × 300 × toaslarge300× 2 dependence dependence

there are no reports of images of small metal clusters in in smallmetalclusters imagesof there are noreports of microscopists, yet toour knowledge, caught theattention of extremely thinlayers. Thethin-layer zeolites have alreadyof appealing tobulkzeolites iszeolitesintheform alternative throughout thesamples. An themetalclusters structures of isdetermining zeolitesassupport thechallengesof One of denced byEXAFSspectra). - tendtoreact at(the metalprecursors theAlsites,asevi well-definedbondingsites tunity tocontrol thenumber of andcagestheoppor theirsurfaces theregularity of of asaconsequence metal clusters supported the structures of allowsprecisezeolites becauseazeolitesupport control of Ir(CO) Fe(acac) Ru,Os,Rh,andIr.metals isFe, are Suitable precursors of Thefamily toform clusters. and have astrong tendency metalsthat aresimilartoeachother chemically plexes of from onsupports com- bimetallicclusters the formation of delaminating zeolitesistoinvestigate The ultimate goal of with nanometerspatial resolution. Alatoms (acidsites) thelocation anddensityof understand tion imagingand3DETwillalsobecombinedwithAPTto thezeolitesinitsaqueousenvironment. Highresolu- ture of fluids intheelectron thestruc microscope- tounderstand imagingthrough of technology We new are afairly applying thezeolitesinwater. of andchemistry mental morphology inwater,occurs thefunda- tounderstand itisimportant oline rangehydrocarbons andmethanol.Asthisreaction - phenolicoiltoproduce gas lignin-derived of the conversion process in3D. for ZSM-5isbeingstudiedastheacidcatalyst thedelamination alkylation, andunderstanding cracking, delaminated materials are usefulfor processes suchas processes.tions, openingupthematerials These for new area allowsaccessfor largerbulkymoleculestocreate reac- intact.Thishighersurface sitesonthesurface the active areazeolite sheetstoprovide ahighersurface whilekeeping material pullingapart the for delamination, withthegoal of be delaminated, andZSM-5.MCM-22isusedasaprecursor lite materials: MCM-22andrelated layered zeolitesthat can zeo- These methodsare combinedandapplied totwosetsof andcompensated. each methodcanbeunderstood and ETare directions; indifferent therefore, of theartifacts techniquesuchasET.non-destructive inAPT Theartifacts most accurate way todothisisbymatching in3Dwitha aprioribyAPTisthereforerial tobeanalyzed crucial. The mate- thevolumeof resolution. Knowingthe3Dstructure of tions. However, canlimittheoverall artifacts anumber of duces sub-nanoscalespatial andconcurrent chemicalresolu- evaporationAPT isacontrolled techniquethat field pro- 2 (acac). The support for these metal clusters will be willbe for(acac). Thesupport thesemetalclusters 3 , H 2 Ru(CO) 4 , andH 2 Os(CO) 4 , Rh(CO) 2 (acac), and (acac), and - them. These ultrathin zeolites (some are exfoliated samples) At the final stage of synthesis, the material is calcined. The are appealing supports for our clusters, and we expect to be heat treatment does something phenomenal in that the able to obtain high-quality 3D images of the samples and zeolite sheets come apart in curvy layers, and the surface atomic-resolution images of the clusters in them (and even area increases. The distance between clusters of the curvy of the zeolite frame), as all the clusters will be located near sheets is larger (up to several hundred nanometers), thereby sample surfaces. providing ample space for bulky molecule reactions. This is the first demonstration of a direct 3D visualization of the To realize this goal, the zeolite support must be fully char- delamination process. acterized first in order to understand the morphology and determine whether the synthesis is successful. This is what With the support synthesis established, FY 2013 research we have focused on in the first year. We have done a thor- will focus on the synthesis of small metal clusters ough analysis of the 3D morphology of the zeolite at three (~4 atoms) and their insertion into the zeolite. Their loca- stages of chemical and heat treatment to understand the tion on the zeolite surface will provide us with an indication delamination mechanism. To date, there has been no pub- of the active (acid) site densities and locations and will be a lished literature on a direct and 3D visualization of the challenging task for both synthesis and characterization. delamination process. With zeolites for biomass conversion, we provided the first images of the ZSM-5 zeolite imaged in-situ in water. As expected, we find that the cluster size of particles when suspended in water is smaller than when the material is dried. Upon analysis of ZSM-5 on a dry grid with different temperature treatments, we find that the sample boiled to 250°C is more stable under the electron beam. The instability of zeolites under the electron beam is generally attributed to either the presence of trapped water in the pores, or unstable cations in the structure, but this has not been proven before. It is an open question since the first Nature publication in 1980. With our new liquid stage capability, we have established this year for zeolites, and we will be working to answer this question of damage mechanism by imaging several different temperature treated zeolites in-situ in water. In FY 2013, we plan to do an unprecedented experiment, 3D morphology of the zeolite at three stages of chemical and heat which is the desilication of a zeolite, by developing a heat- treatment ing mechanism for the liquid stage and observing it in-situ in the electron microscope. The desilication will be per- The synthesis is performed by the Katz group at the Univer- formed in a liquid with a high pH, and will be visible as sity of California-Berkeley. The starting material is MCM-22, the formation of small pores that should combine to make a precursor material. STEM tomography was performed on large bubbles in the material. We will combine this with this material, and it is found that there are many thin 3D and atomic scale imaging for a full understanding of sheets of zeolite that are bound together in large, flat clus- the desilication process. ters. The distance between the sheets is the pore size, which limits larger bulky molecules from entering. The next step is a chemical treatment, which in this case is performed at pH 9 using an aqueous solution of cetyltrimethy-lammonium bromide, tetrabutylammonium fluoride, and tetrabutylammonium chloride at 353 K. Note that this is a much lower pH than zeolite delamination in the literature (pH 13), which yields no detectable amorphous areas that inhibit catalytic performance, such as those found in materials such as ITQ-2. The 3D imaging performed at this stage shows that the zeolitic layers are still flat and in clusters, but the distance between the layers is slightly larger, suggesting swelling and breaking of bonds between layers. PN12015/2416

59 Chemistry 60 PN11010/2332 (broad peak2),atransitionLi structure between Li-metalandLi Li-S battery; (c) The stacked plot of theinsitu (c)Thestacked plotof Li-S battery; ona duringdata acquisition; (b)Thecharge-dischargecurve battery charge anddischargeof Our insituNMRcapability containinganNMRprobe equippedwithaCHI-600Dfor real time dation membranes,and mechanismsintheelectrolytes, - andtheperformance degra the detailedredox chemistry We uniqueinsituNMRcapabilities are tostudy developing lifegoals toobtainmarketmance andcycle penetration. thereby require tomeetperfor greater understanding and have beenoptimizedusingempiricalmethodsand combined withrather complexelectrochemical reactions (RFB) andmoltensaltsinsodium-metalhalidebattery, as concentrated solutionsinredox electrolyte flow battery electrochemical environments involveaggressive such ally batteries usu- Stationary storagedevices. large-scale energy thetopchoicesfor storage hasbeenrecognized asoneof and fundamentalresearch are needed.Electrochemical development breakthroughs inbothbattery Revolutionary the knowntechnologieshasreached thesecombinedgoals. andsafefor operation.charge cycles) Currently, noneof (more 10–15years imately than2000–4000chargedis- be lessthan$150/KWhwithalifetimegreater thanapprox- andsystemshould thestoragetechnology ultimate costof long-term,large-scaleapplications, For the electrical grid. powermanagementinmodern improving thequalityof andfor intermittentrenewable energy widespread useof storageisneededfor the energy Large-scale stationary Jian ZhiHu Resonance Spectroscopy Using InSituandExNuclearMagnetic Characterization ofEnergyStorageSystems to themagnetic field (shoulderpeak4), respectively. where thepeakassignmentsare Lifrom andsolubleLi-Sspecies(narrow electrolytes peak1),Li the spectraselectedfrom (c)andat thecorresponding charge-dischargetimepointslabeled in(b), Chemistry batteries, andLi-ionLi-airbatteries. in theelectrodes,electrolytesofstationary the performancedegradationmechanisms tro-chemistry iontransportmechanismsand advance theunderstandingofcomplexelec- include existingdevelopedtechniques,to magnetic resonance(NMR)capabilities,to We aredevelopinguniqueinsitunuclear 7 Li NMR spectra as a function of time;(d)Highlights LiNMRspectraasafunctionof - 2 S (peak 3),andLi-metalparallel S (peak

battery can beused.Goodresults were obtained usinga battery or planar plastic Li-S batteries where eitheracylindrical an insitustatic NMRcapability Li-ion/ for live investigating Li-air, research inFY2012,wedeveloped andLi-ionbattery In situNMRcapabilitydevelopment. Inaddition toV-RFB, C,forPhysical Li-ionbatteries. Chemistry of Power SourcesandoneinJournal of one inJournal Power Sources for Li-airbatteries andtwopublications, have of resulted publishedintheJournal intwoarticles and Li-airbatteries, suchthat ourresearch andactivities mechanismsinbothLi-ion theiontransport understanding Li-ion andLi-airbatteries.We madesignificantprogress densityinV-RFB.tions that ledtoa70%increase inenergy mixed acidsolu- tion specification andthermalstability of solu- of understanding amolecularlevel our studiesoffered ings resulted Specifically, insixpublishedjournalarticles. temperature. at elevated mechanism degradation Ourfind- ated withV-RFB. performance Thegoal istounderstand andmembranesassoci- NMR studiesonnovelelectrolytes andtemperatureNMR investigations insitumulti-nuclear ex situmulti-nuclear (including (REB).WeRedox performed flowbattery comprehensive stability storagesystems. for electrochemical energy andpowerdensitieslong-term improved energy batteries with stationary new forimportant developing and at theelectrode-electrolytes Thisisespecially interface. ionsbothinsideelectrodes nisms andthechemicalstatus of detailed molecularinformation about mecha- iontransport the complexchemistry. yields Thiscombinedinvestigation modeling capabilities have beenemployed tounderstand 21.1 ultra-high fieldof Tesla andNMRcomputational and magneticfieldsfrom 7.05to copy withawiderangeof on flow-batteries, Li-ionandLi-air. NMRspectros- Advanced batteries emphasis withaparticular stationary electrodes of 2 S discharge cycles. obtainedduringcharge- chemistry theelectro- reaction dynamicsof The insituNMRcapability allows receiver of thesignalgenerated. receiver of electrodes and the into battery tion magneticfieldpenetrating by aplasticholdertoallowexcita- the stainlesssteelcontainerreplaced coincelldesign,with cial battery design isclosertothereal commer ple (seefigure). OurinsituNMRcell asanexam- Li-Sbattery cylindrical 17 O, O, 51 V, V, 1 H, 29 Si, and 35 Cl) - Chemical Imaging Analysis of Environmental Particles Alexander Laskin This project focuses on developing complementary The production and use of energy often leads experimental setups that allow chemical imaging of to the generation of atmospheric particles changes in particle morphology and composition in reactive environments. The research and development efforts that impact a broad range of environmental are performed along two major directions: first, in the issues, including atmospheric composition development and application of micro-reactor cells for in and chemistry, the Earth’s radiation balance, situ studies of gas-particle reactions using complementary polar ozone depletion, and health effects. To methods of spectro-microscopy; and second, in the devel- address DOE concerns about the safe produc- opment and application of particle depth-profiling tion and use of energy, we are developing methods by single-particle mass spectrometry. and applying novel experimental approaches In FY 2011, we evaluated depth-profiling by single particle to improve our understanding of the origin, mass spectrometry, where data indicated that soot particles fate, and chemical and physical properties of generated at different temperatures have remarkably dif- environmental particles. ferent physical and chemical properties. Complementary offline, high-resolution mass spectrometry analysis of soot samples allowed us to provide molecular speciation of It is of great interest to understand how the chemical and aliphatic compounds. For FY 2012, we continued our suc- morphological microstructure of environmental particles cessful collaboration with Advanced Light Source (ALS) affect physicochemical properties such as chemical reactiv- beamline scientists to develop a micro-reaction cell for in ity, thermodynamics of gas-particle systems, and the hygro- situ x-ray microscopy studies of gas-particle reactions. We scopic and optical properties of aerosols. Fundamental utilized a novel environmental aerosol chamber for experi- understanding of these properties and their possible evolu- mentally controlled synthesis of internally mixed NaCl/ tion in time requires advanced analytical approaches for secondary organic aerosol (SOA) and we characterized the chemical imaging of particles on the scale of 10–100 nm in composition, shape, and morphology of NaCl/SOA particles, reactive environments. Environmental particles are multi- studying their hygroscopic transformations using micro- component; therefore, a detailed understanding of their reaction cell and spectro-microscopy techniques as well as complex chemistry will reduce uncertainties related to in situ single-particle mass spectrometry. We also designed their environmental effects. Application of complementary and constructed a new vacuum ultraviolet (VUV) ionization analytical methods is needed to provide comprehensive module for advanced depth-profiling of particles with a information ranging from microscopy level details of indi- single-particle mass spectrometer. We conducted labora- vidual particles to advanced molecular characterization tory studies focused on the chemistry of individual parti- of complex molecules comprising particulate matter. Par- cles where we demonstrated critical need for the methods ticular attention is given to the application of a range of of chemical imaging analysis for fundamental understand- analytical techniques to experimental studies of heteroge- ing particle internal structures and their effects on hygro- neous gas-particle reactions pertinent to atmospheric scopic properties, heterogeneous reactivity, and ice and occupational environments, technologies related to nucleation properties. energy production and emissions control, and particle In FY 2013, we will apply complementary chemical imaging health effects. techniques in the following laboratory studies: 1) NaCl/SOA and mineral dust/SOA particle chemistry; 2) chemical imaging analysis of iron solubilization processes; 3) fresh soot particles and soot particles aged in atmospheric-like environment; and 4) soot and SOA with an emphasis on evaporation kinetics of semi-volatile compounds. Additionally, we will continue development and modification of the micro-reaction cells for temperature-controlled experiments, and we will will apply VUV depth-profiling approach to mixed organic/inorganic model system with different morphologies.

Color-coded SEM images showing internal structures of mixed NaCl/

SOA particles. PN11012/2334 61 Chemistry 62 PN10010/2252 larger scaleinaCO are applied in the context of aCO are applied in thecontextof Allthreeress capabilities toward engineering-scaletesting. prog- (byPNNLor others) efforts as material development testing materials inlargerquantitiesandat increased scale test hardware andmethodsbringsauniquecapability for and continuous-scale Thissetof liquidsolventtest cart. ted-wall column,continuous-scale solidsorbenttestcart, testsystems:awet- three primary ect in2010consistedof promising CO capabilitiesperformance overlife. for Developing themost including representative impactsonmaterial attrition and for operating testingmaterials regime, underarelevant Currently available technicalapproaches for CO Dale A.King CO CO new, meansof developing A drawback cost-effective of based electricityandfuelsintheUnitedStates. coal- canimproveshift reactors theeconomicviability of from room temperature totemperatures compatible with technologies for CO suchasmonoethanolamine(MEA).These liquid adsorbers include pressure CO adsorption andcapture swing of The CO baseline standard iszeolite13x. solidadsorbents,theperformancewill becompared. For nomic removal of CO nomic removal of technologies that enable separate orsimultaneouseco- inmaking attractive. anyprocess New tors economically selectivity, andtheprocess- complexityare fac important separation MEA)cost,itsoperating agent(notably lifeand non-pressure adsorptionagent. For approaches, swing the theseparation consumptionduringregeneration of energy lowmassfluxes in theseparation unitsandhigh factors: ature andpressure. Thehighcostresults from twokey ature, are costly, andrequire intemper substantialswings against whichotherliquidphaseCO against highest priority, MEAcanbeusedasa baselinetechnology post-combustion capture,goal. For whichiscurrently the Chemistry mobile CO This projectisdevelopingaflexible,modular, CO of scaledtestingcapabilitiestopotential other researchfacilitiestoprovidearange purpose istodevelopcollaborationswith tive toconventionalbaselinematerials.The materials fromPNNLandotherentitiesrela- and evaluatingtheperformanceofcapture 2 capture from fossil fuelsources islimitedcapabilities 2 capturematerials. 2 capture underthisproj testcapability- developed 2 SeparationScale-Up 2 capture technologiesandtesting themona 2 capturematerialtestcapability 2 2 capture test laboratory isaprincipal capture testlaboratory capture operate nearambienttemper 2 , H 2 S (andCOS),andHgoverarange 2 capture material screen- 2 capture technologies 2 capture 2 by

- - circumference of a vertical metalcolumnwithCO avertical circumference of the wetted-wallcolumn,solventflows alongthe downward Duringtestingon loading/regeneration cycles. of numbers operating conditionsandthrough relevant large ally enableous-scale testcarts characterization underindustri- materials are available),higher quantitiesof thecontinu - initialscreeningrial characterization. (andonce Following arerately) applied tosolidsorbentsfor fundamentalmate- sepa- the wetted-wallcolumn.Similarmethods(developed used to characterize the performance of baselineCO used tocharacterizetheperformance of conditions. InFY2011,thesetestsystemswere refined and performance datasive andmore representative operating material more candidates comprehen usingprogressively - the viability of toevaluate ing (TRLGate) methodology be necessary to achieve theanticipated CO toachieve be necessary approach allowsustoexplore configurations expectedto adsorption between0.012and4.8bar. Thetwo-bed ambient and200°Cthepressure/vacuum swing adsorption performedeach, withthermalswing between 1.5L between5and20slpmusingtwobedsof feed flow of thetwo.combination of canoperate Thesolidcart witha orpressure/vacuumthermal swing approaches swing ora through characteristics.Thebedscanberegenerated using loadingprofileenable andbreak characterization- of thebedoutletto thecompositionof sample continuously saturator. andagas trollers, Amassspectrometer isusedto mass flow con- bottled gases, asetof through theuseof gas compositionsimulates acoalflue tion. Inletgas filled withamaterial candidate inapacked-bed configura- twocolumns The continuous-scale consistsof solidtestcart continuous-scale capabilities. liquidsolventtestcart thewetted-wallcolumnand was acquired asaresult of and Fossil Energy This work wasfundedbyDOE’s of Office kinetics and CO more suchasreaction fundamentalmaterial parameters liquidsolvents, andselectionof inthedevelopment Early operation. ture materials overextendedperiodsof developed CO developed wall columnwasusedtocollectkinetic data onthePNNL- carbonate withmobilizedenzymes.InFY2012,thewetted- also usedonarelated project tocharacterizepotassium and equilibriumdata onMEAandMDEAsolvents.Itwas wall columnwasusedinFY2011tocollectdetailedkinetic Thewetted- describing sorption kinetics canbedeveloped. flowing pastthewettedcolumnperimeter, relationships compositionchanges after Bymonitoringgas vent andgas. enables area awell-controlled interface betweenthesol- flowing upwarding gas around thecolumn.This apparatus through theloadingandregeneration steps.Thesolidcart bed temperature profile andpressure drop are monitored requirements.purity levels Inadditioncompositions, togas 2 BOLs inateamingarrangementwithFluor. 2 loadingcapacities using were evaluated 2 capture and 2 -contain- 2 cap - was completed in FY 2010. It was then used in 2011 to Following testing, the system was updated to improve solid characterize the performance of zeolite 13x in a dual-bed and solvent off-site testing capabilities. Of particular con- configuration using both thermal and vacuum swing regen- cern was the need to prevent moisture from condensing in eration. Because the performance of zeolites tends to be the lines. Minor flue gas constituents such as SO2, H2S, and affected adversely by the presence of water in the flue gas, Hg dissolve in the liquid water rather than staying in the both dry and humid gas testing was conducted to quantify gas phase to interact with the CO2 sorbent material. To the impact of water. The solid sorbent cart was also tested address this concern, heated flex lines, a compressor with in multiple campaigns both in the Sigma V carbon capture capability of internal heating, and high temperature laboratory with simulated flue gas and at the flue gas test- flowmeters capable of heat tracing were purchased and ing facility at the PGE coal plant in Boardman, OR with assembled on a new “off-site testing” cart. actual flue gas. In addition to the cart testing and upgrades, considerable For liquid solvents, the continuous-scale solvent test cart work was performed to develop collaborations that would includes both absorber and stripping columns. Liquid sor- position PNNL for future CO2 capture opportunities. Meet- bent candidates are routed through the contactor to absorb ings were held with the National Carbon Capture Center

CO2 from a simulated flue gas or actual flue gas. Solvent (NCCC) and plans made to allow the solvent cart to be can be continuously loaded by circulating through the shipped to their testing facility and evaluate it with their absorber alone or can be routed through a heat recupera- flue gas. The PNNL test cart would provide an early TRL tor before routing through the CO2 stripping column, as evaluation tool at a much smaller scale to augment their would be done in a full-scale process. Trim heaters and larger scale capability and fill their gap between laboratory coolers are incorporated at the inlets of both columns for testing and large pilot scale. This collaboration progressed precise control over operating conditions. The stripping but was ultimately postponed due to unresolved contrac- column has a reboiler for desorption of CO2 from the sol- tual differences. However, a collaborative relationship was vent as well as a condenser at the top of the stripping col- developed between PNNL and the Energy and Environmen- umn to remove solvent and moisture from the CO2 stream tal Research Center (EERC), affiliated with the University of prior to exhaust. The solvent cart can operate with a feed North Dakota with a coal combustion facility and pilot flow of between 5 and 20 slpm using 2 liters of solvent. The scale CO2 capture capabilities for both solvents and sor- solvent flow rate is varied between 50 mL/min and 200 mL/ bents. EERC can benefit from the thermodynamic and min, and the columns contained 3.2 cm-diameter Sulzer kinetic capabilities developed as part of this project; in EX packing with a maximum packing length of 83 cm in turn, PNNL can benefit from EERC’s larger-scale testing. each column. As with the solid cart, performance of the EERC worked with GreenCentre Canada to ship PNNL sol- material and process are monitored via mass spectrometer vents for testing using the wetted wall column and PTX cell. analysis of the product gas streams from both columns. These solvents have not been developed sufficiently to war- The solvent cart was completed in FY 2011. rant the larger-scale testing available at EERC. These tests are underway, and results are pending. Our project modified the solvent cart to a flow-through configuration to measure specific reboiler duty for a In addition to EERC and NCCC, PNNL has collaborated with

CO2-loaded potassium carbonate solution. We measured Aspen Aerogels (AA) in testing the hydrophobic aerogel reboiler duties at 0.5, 1, and 2 bar pressures and 75% CO2 materials that AA has developed. PNNL agreed to measure loading. The reboiler temperature was varied and a range single gas species isotherms and thermal conductivity for of lean loadings were produced. Before testing could be two candidate materials that they have developed. AA and performed, this project improved the capability of the sol- PNNL will team to prepare a journal article from the results vent cart, including improved flow measurement and con- of these analyses, which should provide an unbiased analy- trol for the reboiler oil loop, the addition of three sample sis of the AA materials and publicity to PNNL to showcase locations axially along the stripper to collect samples and capabilities to the carbon capture community. These tests measure gas flow, improved column liquid level sensing to are also underway, with the results pending. prevent flooding, and the ability to operate the column at sub-atmospheric conditions. In addition, issues related to pumping the high temperature stripper product were addressed with the purchase of a new, high temperature pump. PN10010/2252

63 Chemistry 64 PN11014/2336 biomass-derived lowvalue oxygenates.biomass-derived to produce higherhydrocarbons from awiderange of processes the knowledgebasethat canbeusedto develop ing higherhydrocarbons from alcoholsbutalsogenerate ful outcomesfrom leadtoproduc thisproject- willnotonly including alcohols,carboxylic acids,andketones. Success- oxygenates, biomass-derived fuel from awiderangeof one-potsynthesis toproduce full-performance jet achieve tal knowledgegenerated to willbeusedtodesigncatalysts to meetthestringentjetfuelcomposition.Thefundamen - towards theproduct andselectivity slatethe targetactivity andstructure withcontrolled morphology lysts toachieve We anddesignthecata catalysts willprobe- thenature of (MTG) byMobil. methanol-to-gasoline process developed usingasimilarconcepttothe conversion nol-to-gasoline stepsinvolvedintheetha- We are studyingtheelementary number tested. C8betweenthecatalysts product fractiongreater organic than thecarbon Comparison of relies onotherresources for itsaromatic content. jetfueland contentof theparaffinic cesses produce only unfavorable pro cost,theabove- mentionedoilandalgae point,andhighvolumetricdensity.flash Inaddition to stringent specifications suchaslowfreezing point,high (straightandbranched)hydrocarbon dueto C16 paraffinic C8to aromatic content,alongwith~60wt%of andcyclic resources.ural oilandalgae Jet fuelrequires ~40wt%of aprocess toproduce jetfuelfrom isdeveloping nateywell - inrenewable fuelresources,With ahighinterest level Hon- Currently, allaviation from fossil fuelisderived sources. Karthikeyan K.Ramasamy Conversion ofBiomasstoJetFuels Chemistry of biomass-derivedlowvalueoxygenates. full-performance jetfuelsfromawiderange generated todesigncatalystsproduce bons, utilizingthefundamentalknowledge sion ofbioethanoltojetfuelrangehydrocar We seektounderstandthecatalyticconver - - 360°C and300psigwithWHSVbetween4.77.9h (M1, M2,andM3).Theseexperimentswere conductedat fresh (F1,F2,andF3)modifiedcatalysts catalyst chain lengthC8generated from thecommercially acquired liquidproduct fractiongreater organic thancarbon son of nomenon canbereplicated. Thefigure showsthecompari- zeolitematerialwith inert (silicalite) themovingbedphe- bedlengthandmixingthecatalyst modifying thecatalyst Our secondapproach wasbasedonthehypothesisthat for probing thephenomenonexplainedearlier. vated whichmadethisapproach thecatalysts, unsuitable compounds reacted withtheinternalacidsitesanddeacti- was cracked ontheexternalacidsites,and fragmented However,pore sizeopening. results showedthat theTEOS andmodifythe thecatalyst deposit ontheouterlayer of porecatalyst diameter(~0.55nm).TEOSwasexpectedto largercompared(~0.96nm) issubstantially tothezeolite agent thissilylating icate (TEOS),asthekinetic diameterof was treated- agentsuchastetraethylorthosil withsilylating the modifiedpore. To prove thistheory, thezeolitecatalyst branchedcompoundsthat canfitthroughstraight, lightly the allowsonly life,along thecatalyst whichselectively the pore mouthsizemodification duetocarbondeposition hypothesiswasbasedon stand thephenomenon.Thefirst these results, tounder twohypothesiswere developed life. thecatalyst From compounds inthelater stageof lifecompared thecatalyst tohigherinstraightchain of tion changesfrom higharomatic contentinthebeginning From earlier experiments,weidentifiedproduct distribu- erated from ethanoltofitintothejetfuelcomposition. compoundssuchasaromaticscyclic andnaphthalenes gen- and olefin (straightchain)selectivity reduceunsaturated During FY2012,ourfocus and wastoimprove theparaffin lene (unsaturated compounds. cyclic) fuelrange)dominated byaromatic andnaphtha - > C8(jet carboninethanolendsupthechainlength of Atoptimizedconditions,over60wt% hydrocarbon gases. remaining rangefractionandlight carboninthegasoline in ethanolgoes toward jetfuelrangefraction,withthe thecarbonpresentefficiency. Resultsshowedthat ~50%of highjetfuelfractionwith100percent carbon to achieve reaction mechanisms andprimary thecatalysts nature of this work, the experiments were designedtounderstand thejetfuelfraction.In contentof matic andthe cyclic During FY2011,ourfocus ethanoltoaro wasconverting - the biomassresources. renewable jetfuelfrom from ethanolandultimately From thisresearch, weare one step closertoachieve when compared totherespective fresh, modifiedcatalyst. Clearly, there andolefincontent isincreased paraffin -1 - . Development and Deployment of Reactive Surface Area Measurement Capabilities Using Nuclear Magnetic Resonance Spectroscopy Karl T. Mueller measurements of unique species via the isotropic chemical A molecular-level understanding of the spec- shift. The results of this research will provide a clearer understanding of chemical reactivity at complex surfaces ificity, speed, and energetics of solid material by describing the sites on the surface that are reactive and surface reactions opens new possibilities for what the structures of reactive molecules are when they are energy production control, pollution mitiga- found at surfaces. By acquiring data at different tempera- tion, and a fundamental process under- tures, we are also able to describe the energetics of forma- standing that can be harnessed to produce tion and removal of these reactive species. chemicals more efficiently in cost and energy. In the past year, we built and deployed a set of dosing stations to load NMR-active probe molecules onto surfaces for analysis utilizing the NMR instrumentation at EMSL. Two The surfaces of solid materials are primary reaction sites gas-phase dosing stations and an in situ loading cart were for a wide range of important chemical processes, whether developed for this work, which allowed adsorption of these surfaces are found in the environment or exploited 13C-enriched ethanol and acetic acid as probes of surface for advanced industrial applications. Previously, reaction sites. The probe molecules were labeled at molecular posi- rates have been measured and mechanisms predicted for a tions that correspond to the carbons that are part of the wide range of important chemical processes, but reactions alcohol and acid groups, respectively, allowing for facile at surfaces are especially difficult to characterize at the 13C NMR studies of bonding environments. molecular level when the systems are either heterogeneous Primary experiments focused on 1H/13C cross-polarization or when they occur in natural environments such as soils magic-angle spinning (CPMAS) NMR measurements and were or sediments. carried out at magnetic fields of 14.1 T, 17.6 T, and 20.0 T The objective of this project is to deploy a range of new corresponding to 1H resonance frequencies of 600, 750, experimental tools for measuring reactive sites and reac- and 850 MHz, respectively. At these high magnetic field tion rates on the surfaces of materials. We are using solid- strengths, spectra were accumulated with superior signal- state nuclear magnetic resonance (NMR) spectroscopy to to-noise ratios and sensitivity, and maximal spectral probe surfaces and molecules adsorbed to these surfaces. resolution was achieved. Solid-state NMR is an isotopically selective tool that pro- Results were obtained detailing the bonding environments vides information on local electronic and magnetic envi- of the probe molecules ethanol and acetic acid on polycrys- ronments of NMR-active detected isotopes. For spin-1/2 13 talline alumina (the gamma phase) and titanium dioxide nuclei such as C, the primary information comes from (the anatase phase). In particular, the range of binding sites for acetic acid on anatase was investigated. One- and two- dimensional NMR spectra demonstrated the resolution possible at very high magnetic field strengths and up to three, possibly four unique binding sited have been identified. By stretching the 13C spectrum into a second spectral dimen- sion, connectivities of the carbon atoms to unique 1H environments have been established. Additional studies included the temperature dependence and pre-dosing processing conditions for the adsorption of ethanol on the surface of a g-alumina sample. In this instance, primarily physisorbed species were detected, and it appears that the chemical shift range for ethanol on aluminas may be too small to detect differences between physisorbed and chemisorbed species. Overall, the discovery of multiple binding sites as well as the experimental protocols for detecting these sites Solid-state 1H/13C CPMAS NMR offers a superior characterization of via solid-state NMR have added greatly to our knowledge the adsorption of 1-13C acetic acid to the surface of anatase. Multiple of reactive surface structures on complex materials such binding sites are clearly resolved in the one-dimensional NMR as titania. spectrum, and our assignments are illustrated with molecular models. PN12020/2421 65 Chemistry 66 PN11024/2346 stability of Lisalts. stability of DFTcalculations on the electrodes f) electrolyte; indifferent cycled solvents andLisalts;c)-e)XPSof a)-b) Cycling stability of gen-rich electrochemical conditionsare critical. thatto dothis,nonaqueouselectrolytes are stable inoxy- inairelectrodes areand oxygen needed.Inorder evolution bifunctionalelectrocatalysts forAdvanced oxygen reduction which istheprerequisite development. for Li-airbattery oxygen reactionis carriedouttounderstand mechanism intheairelectrode. Thefundamentalstudy gen evolution oxygen reductiondue tothesluggishreactions andoxy of - power capability, efficiency, lifetime andshort lowenergy Its practicalapplication iscurrently limitedbyitspoor development. stageof early isat avery The Li-airbattery Yuyan Shao for RechargeableLithium-AirBatteries Development ofBifunctionalElectrocatalysts for theadvancementofthistechnology. such asbifunctionalelectrocatalystsiscritical Developing advancedairelectrodematerials comparable togasoline-poweredvehicles. enable drivingrangesofelectricvehicles tional energystoragetechnologythatcan The lithium(Li)-airbatteryisatransforma- Chemistry

is themostcriticalissuein community. research, forelectrolytes rechargeable which Li-airbattery This project leadstoaresearch platform andidentifystable stability trend. tions were carried toilluminate mechanismsbehindthe (DFT)calcula- studyanddensityfunctional theory chemistry Li triflate (LiSO Li triflate 4) acetonitrileandDMSOare themoststable. Lisalts, For moreare stable and thancarbonates even andethers; more stable thancarbonates; 3)nitrile-basedelectrolytes bonates; 2)like ether-based variousglyme, are electrolytes linear carbonates aremore car slightly stable thancyclic bonate-based areunstable, electrolyte usually amongwhich We have learnedthat for nonaqueoussolvents1)alkylcar development. tant for practicalLi-airbattery that is,therefore, abattery impor cathode, of electrolyte) thethree keycomponents(anode, in electrolyte, oneof becausethiswork needstobedone development catalyst ical for bothoxygen reaction mechanismstudyandelectro- nonaqueous solventsandLisalts.Astable iscrit- electrolyte methods. We focused onstable study, electrolytes including compared withdays hours orweeksintraditionaltest few oneisa electrocatalysts thetesttimeof andelectrolytes: The platform accelerates thescreening alsosignificantly of oxygen reaction mechanisminsimulated Li-airconditions. platform simplifiesthestudyof Thisnew development. form for mechanismstudyandelectrocatalyst/electrolyte plat- technology optimizedthenew In FY2012,wefurther electrocatalysts materials andstable electrolytes. of system andguidancefor thedesignanddevelopment energy thisnew of provide amechanistic understanding conditionsto electrolyte tion reaction inLi-airbattery oxygen reduction reaction- mechanismsandoxygen evolu nient operation. We willperform fundamentalstudiesof lowcost,highthroughput, simplified,andconve- tages of platformelectrode; that a technology - willhave theadvan characterization platforma new based onarotating disk systems.We storageandconversion aimtodevelop energy performance electrolytes/electrodes for transformational electrode reaction mechanismsandhigh- of understanding The research underthisproject isproviding afundamental rate (LiClO Imide (LITFSI)are themoststable, followed by Liperchlo- 4 ) andLihexafluorophosphate (LiPF 3 CF 3 ) andLiBis(Trifluoromethane-sulfonyl) 6 ). The surface ). Thesurface - - - Development of Inorganic Water Oxidation Electrocatalysts Aaron M. Appel and to match the energetics of catalytic intermediates to This project will develop new classes of homo- approach thermodynamic equilibrium for each step. Multi- geneous inorganic water oxidation electrocat- electron and multi-proton chemistry is facilitated in nature by controlling the proton inventory at enzymatic active sites alysts that improve on the efficiency, rate, using amino acids. Our synthetic systems incorporate proton stability, and cost of known catalysts. With acceptors in the secondary coordination sphere to perform the use of highly modular ligands, this sys- these functions. tematic approach toward catalyst discovery We attack the challenging four-electron process to oxidize is expected to generate new families of water by studying the catalytic cycle in two stages. The first effective water oxidation catalysts. stage will cover the two-electron oxidation of a water molecule bound to a complex to form a metal oxo species, the first intermediate in oxygen bond formation. This can be studied spec- The development of a non-fossil fuel energy infrastructure is troscopically and chemically probed by oxygen atom transfer one of the major challenges facing our country for environ- reactions. The second stage will couple the metal oxo interme- mental, economic, and national security reasons. An answer diate with another metal oxo or hydroxide to initiate the to this challenge is the development of an artificial photo- evolution of oxygen and close the catalytic cycle. synthetic system. In nature, plants use sunlight to generate chemical fuels by reducing carbon dioxide and oxidizing In FY 2011, two ligand systems were explored that fit the water to evolve oxygen. A practical synthetic device would above requirements. The first is a modified porphyrin with require an efficient, fast, and stable catalyst capable of oxi- pyridines incorporated in the meso position as proton relays. dizing water under ambient conditions. Despite several The iron complex did not appear to be active for water decades of research, there are only a few examples of het- oxidation. However, the reverse reaction (oxygen reduction) erogeneous catalysts for water oxidation and most require explored by our University of Washington collaborators is an the use of precious metals. Aside from expensive noble met- active electrocatalyst that demonstrates high selectivity to the als, current catalysts do not have the efficiency, speed, or reduction of O2 to water. The second ligand system studied in stability to function on a practical scale. However, we know FY 2011 explored the use of five-coordinate pyridine ligands 2+ 2+ from nature that water oxidation can be performed under (Py5). The complexes [(Py5)M(H2O)] and [(Py5)M(OH)] all of these conditions using cost-effective first row (M= Fe, Co) led to a rare example of an FeIII—OH complex. transition metals. The proton-coupled electron transfer from the aquo complex oxidation and thermo-chemical properties were studied, Our approach to catalyst design is to mimic and incorporate which led to the following year’s research direction. the essential features of natural systems into synthetic systems. To avoid high or low energy intermediates in the cata- In FY 2012, our efforts focused on incorporating proton relays lytic cycle, we seek to utilize effective proton management into the second coordination sphere of four-coordinate FeII aminopyridine complexes. Recently, a family of complexes that

shared a cis-[Fe(N4)X2] (X = Cl or Br) formulation was reported to catalytically oxidize water to O2. The synthetic possibilities for modification of the ligand environment to incorporate hydrogen bonding groups were well-established, so we set out to examine the structures and catalytic water oxidation behavior of a series of novel FeII complexes. These complexes are indeed water oxidation catalysts, with initial rates higher than the corresponding complexes which lack the proton relays. The study of these complexes has identified promising features for the design of new catalysts, as submitted for publication and presented at two technical conferences. The results of this project will influence the design of additional catalysts, has resulted in interactions with currently funded projects, and will be included in other proposals.

Future work should focus on complexes that contain a pKa Outline of the catalytic cycle: Stage 1 shows the formation of the metal matched proton relay that will work with chemical oxidants oxo probed by reaction of an easily oxidized substrate; Stage 2 initiates or directly as electrocatalysts. the oxygen-oxygen bond formation with the reactive metal oxos. PN10025/2267 67 Chemistry 68 PN11028/2350 the biofilm. An integrated the biofilm.approach Anintegrated that combinescomple- essentialfor metalreductioncells andmineralsurfaces by redox moleculesthat electron facilitate transferbetween extracellular several theidentityof revealed at interfaces generated chemical gradients MS imagingexperimentsof plates. onagar grown microbial betweendifferent communities interfaces or betweenbiofilmsat andmineralsurfaces interfaces generated chemicalgradients detailed characterization of material inbiofilms. Thesetoolshave beenused for the extracellular for comprehensive MScharacterization of approaches thisproject, wehave developed of the course verts CO verts lightcon- photosynthetic bacteriumthat inthepresence of unknown.Synechococcussp.is stilllargely isa PCC7002 biofilms inShewanella port responsible for metal reduction However, theextracellular electron trans- amechanismof uraniumandtoxic metals. potential usefor cleanup of its studiedbecauseof hasbeenextensively Shewanella CO metalreduction and of understanding ing molecularlevel biofilms asamodelsystems for obtain- and cyanobacteria microbial biofilms. of We are oneidensis usingShewanella thesenoveltoolsforscale andapply thecharacterization MStothe nano- and unsurpassed chemicalspecificityof istobringsoftionization, sensitivity,ods. Theobjective tion ascompared toopticalandelectron microscopy meth- ions,andlimitedspatial resolu analyte - fragmentation of include samplepre-treatment unwanted priortoanalysis, current state-of-the-art MSimagingtechniques tions of speed, andunprecedented chemicalspecificity. Thelimita- complexsystemsincludinghighsensitivity,acterization of for char uniqueadvantages anumber of systems. Itoffers chemicalandbiological of understanding molecular-level MS imagingisapowerfultechniquefor obtaininga Julia Laskin of ComplexChemicalandBiologicalSystems Spectrometry Approaches forSpatialImaging Development ofNewSoftIonizationMass Chemistry edented spatialandmassresolution. produced bybiologicalsystemswithunprec- enable mappingofchemicalcompounds etry (MS)imagingtechniquesthatwill complementary cutting-edgemassspectrom- rity, andhealthsciences.We aredeveloping diation, carbonsequestration,nationalsecu- interest forenvironmentalcleanup,bioreme- to understandbiochemicalprocessesof Novel experimentalapproachesareessential 2 sequestration bymicrobial communities. 2 into sugars and other organic molecules.During andotherorganic intosugars

- of carbondioxide. of andonbiologicalsequestration transfer tomineral surfaces unique insightsonmolecules responsible for electron andSynechococcus biofilms, providing Shewanella of Future research willinvolvemultimodalcharacterization between microbial communities. toolfor studyingmetabolic exchangeents auniquenew metabolites basedontheMS/MSdata presclassification- of microbial communitiesusingnanoDESIandefficient living California, metabolites SanDiego. from detectionof Facile of by our colleaguesat the University matics tools developed were combined withuniquemolecularnetworking bioinfor Thesenovelexperimentaltools films andmineralsurfaces. microbialbetween different communitiesandbetweenbio- interactions of is essentialfor detailedunderstanding by microbial communitiesinresponse totheirenvironment preparation. chemical signals generated Spatial profiling of directly from mediumwithoutanysample thegrowing are grownexperiments, bacterialbiofilms onagar analyzed thisMSimagingtechnique. Inthese a uniquecapability of bacterialcolonies, living for samplinganddirect of analysis nanoDESI In FY2012,wehave demonstrated theutilityof intact lipidspecieswithgood sensitivity. increased andmassrange-enabling detectionof sensitivity cifically, of iontransferopticssignificantly thesimplification improved platform drastically upgrades performance; spe- present insuchcomplexsystems.Recentinstrumental logical samplesiskeytocharacterizingchemicalgradients bio- bilities. Theability toresolve isobaricpeaksinSIMSof The secondsystemcombinestheC traditional ambientpressure ionization techniques. surface olution >10µm,10×betterthanthat obtainedusingmore ambientnanoDESIimagingwithspatial res sensitive - highly state. biologicalsamplesintheirnative Weof demonstrated resolved spatially chemicalcharacterization sensitive highly for Thismethodprovides uniqueadvantages (nanoDESI) MS. ambient nanospray desorption electrospray ionization platformfor imagingbiological samples.Thefirst utilizes To twonovelplatforms thesegoals, wedeveloped achieve U, Cr, Tc, andPu. toxicand participate metalssuchas inthetransformation of metal-richacidminewastestreams, acidificationcatalyze of andMn, metalssuchasFe cycle howmicroorganisms sively electron microscopy holdsgreat conclu- promise toanswer MS-basedimagingapproaches withopticaland mentary apparatus. Theone-of-a-kind C transform resonance ioncyclotron MSimaging (FT-ICR) (SIMS)systemwithahighresolution Fourier spectrometry resolving power(m/Δm (<1ppm),highmass characterized byhighmassaccuracy 50% >200,000),andtandemMScapa- 60 SIMSFT-ICR MSapparatus is 60 secondary ionmass secondary - Development of Preparative Mass Spectrometry for the Creation of Novel Catalyst Materials

Grant E. Johnson

microscopy (TEM). Atomically monodisperse gold clusters This project will employ soft landing of mass- containing exactly 11 gold atoms were prepared on surfaces selected ions and characterization with in situ and characterized using scanning TEM. The soft landing approach offers unprecedented sample cleanliness as techniques to prepare precisely-defined cata- molecular precursors, reactive intermediates, neutral lytic materials and to establish structure- clusters, and solvent molecules are removed and only reactivity relationships that will enable the clusters of a selected size are deposited on the surface. rational design of improved catalysts. The approach has widespread potential application for TEM sample preparation, as it may be extended easily to other materials. These results are the subject of a The size, structure, charge state, and composition of publication in Analytical Chemistry. nanoparticles, clusters, and complex molecules supported on surfaces significantly impacts their chemical and physi- A new position-sensitive detector referred to as IonCCD™ cal properties. The existing approach for preparing metal has been examined using different charged particles. It is a nanoparticles and clusters involves the reduction of metal modified light-sensitive charge-coupled device (CCD) engi- precursors in solutions containing organic capping ligands. neered for direct charged-particle detection. Combining the While this method may produce nanoparticles with con- IonCCD™ with a double-focusing sector field mass spectrom- trolled size distributions and morphologies, it is unable to eter, we demonstrate fast data acquisition. Detection of access the subnanometer size regime with truly monodis- hyperthermal biomolecular ions produced using an electro- perse “atom-by-atom” precision. Soft landing of mass- spray ionization source is also presented. In addition, we selected ions is a powerful approach for the preparation demonstrate the possibility of simultaneous separation and modification of surfaces and enables unprecedented and micro-array deposition of material using a miniature control over the composition of the deposited materials. MH-MS. This work was published in and featured on a cover For instance, through soft landing of mass-selected ions, of the Journal of the American Society for Mass Spectrometry. metal clusters with specific sizes (8–10 atoms) exhibited pro- The IonCCD™ was employed to characterize the ion optics nounced catalytic activity toward reactions such as the oxi- and ion beam focusing in a custom built mass spectrometer dation of CO to CO 2 and the dehydrogenation of propane. designed for soft and reactive landing of mass-selected ions Further, electrospray ionization of molecular precursors onto surfaces. The IonCCD™ was placed at several stages enables the production of high intensities of catalytically along the path of the ion beam to determine the focusing active clusters for subsequent soft landing onto substrates. capabilities of the various ion optics, which include an elec- In addition, physical synthesis techniques such as laser trodynamic ion funnel, two radiofrequency (rf)-only colli- vaporization and magnetron sputtering combined with sion quadrupoles, a mass resolving quadrupole, quadrupole gas aggregation may prepare unique materials that cannot bender, and two einzel lens assemblies. Combined with be synthesized using traditional approaches. SIMION simulations, we demonstrated that the IonCCD™ The size and charge state of metal clusters is known to can identify minor errors in the alignment of charged-parti- exhibit a dramatic influence on their optical, electronic, cle optics that result in erratic trajectories and significant and reactive properties. Using both in situ time of flight deflections of the ion beam. This information may be used (TOF) and Fourier transform ion cyclotron resonance (FT-ICR) to facilitate the design, assembly, and maintenance of cus- secondary ion mass spectrometry (SIMS), triply charged gold tom-built mass spectrometry instrumentation used for soft clusters were shown to retain their charge when deposited landing of mass selected ions. These results were published onto fluorinated monolayers. In contrast, partial and com- in the Journal of the American Society for Mass Spectrometry. plete neutralization of charge was observed when the same Continuing areas of inquiry will include the following: clusters were soft landed onto carboxylic acid terminated 1) preparation of atomically monodisperse nanoparticles and alkyl thiol monolayers, respectively. These results dem- and clusters on surfaces and examination of their size and onstrate a significant step forward in the ability to control shape-dependent reactivity; 2) investigation of the charge the size and charge state of metal clusters on surfaces and retention and neutralization behavior of multiply charged are the subject of a publication in ACS Nano. anionic polyoxometalate clusters deposited onto surfaces; Characterization of size-dependent properties of nanoparti- and 3) structural and thermodynamic characterization of cles and clusters relies on the availability of clean homoge- subnanometer gold clusters in the gas phase.

neous samples for analysis by transmission electron PN10026/2268 69 Chemistry 70 PN12027/2428 side of thechamber.side of Afterthereaction, and theparticles duced byoxidizing α α of lysis aerosol generated particles organic from the ozono- tion of evapora- experiments,westudiedthe seriesof In thefirst climate forcing intheselarger-scale models. of istoimprove thisresearch theaccuracy strategy goal of inlarger-scale models.Theultimate lifecycle particle organic ber data willleadtoamore accurate representation of theprocess from modeltocham- theapplicationgained of modelassumption.Insights process modelwhilevarying attempted toreproduce experimentaldata usingadetailed their sizeandchemicalcompositioninreal time. We then in acontinuous-flow environmental chamberandmeasured We undercontrolled particles generated conditions organic condensation. particle asasolventfor organic further serve as semi-solidsandtherefore anddonot slowly evaporate exist particles andthat anthropogenic organic gases, ence of formationparticle is altered from biogenicgases inthepres- described isaimedat testingthehypothesesthat organic totheaforementioned The work regards fieldobservations. in particularly lifecycle, particle theorganic standing of modeling studiesaimedat improving thescientificunder experimentsand This project focused oncoupledlaboratory aerosol onclimate forcing. of reducing associated theuncertainties withtheeffect consequencesfor properties inmodelshasimportant these assumed inmostmodels.Accurate reproduction of are solidsorsemi-solidsrather thanliquidsascurrently have particles suggestedthat organic Second, observations forcing mechanismthat isnotcurrently captured inmodels. significantanthropogenic aclimatically radiative implying anthropogenic emissionsbysomeunknownmechanism, of tobeenhancedinthepresence genic compoundsappears formation particle organic from bio- First, lifecycle. particle theorganic inmodelrepresentations gaps two important of the atmosphere. Recentfieldmeasurements have elucidated in observed particles organic andconcentrationserties of atmospheric in modelshavereproducing difficulty theprop- thetotalatmospheric aerosol mass.However,20 to90%of Measurements aerosol showthat comprise particles organic John E.Shilling actions ofAnthropogenic andBiogenicVOCs Environmental ChamberStudiesoftheInter- detailed processmodeling. lifecycle usinglaboratoryexperimentsand understanding oftheorganicaerosolparticle of thisprojectistoimprovethescientific challenge foratmosphericmodels.Thefocus aerosol particles,onclimateremainsa particles, whicharealargefractionofall Accurately representingtheeffect oforganic Chemistry -pinene, were abiogenicemission.Particles pro- -pinene with 300 ppbv ozone in one ozone inone -pinene with300 ppbv

low-NO dation of isoprene in the presence of different NO different isoprene inthepresence of dation of aerosol formed particles organic from the photooxi- yield of Technology tomeasure thevolatilityGeorgia Instituteof and experiments,wecollaborated withthe In thesecondseriesof atmospheric models. more thanwouldbepredicted slowly from traditional may particles existinthesemi-solidstate andevaporate experiments. Together, results aerosol confirmthat organic the of theaerosol massoverthecourse ~20%of oration of - slowevap time upto24h.Intheseexperiments,weobserve sizedistributionswere7:1, andparticle monitored inreal tion chamber. Thedilutionratio intheseexperimentswas transferredchamber wasrapidly inasinglebatch tothedilu- formed experimentsinwhichaerosol intheproduction To overlongertimescales,weper evaporation investigate nomeasurable aerosol evaporation. particle we observed conditions, of thechamber3–5h.Regardless tion sideof demonstrate acomplexdependenceonNO produced underhighNO particles nuder at 100°Ccompared of to<40%evaporation inthethermode- massevaporated isoprene particle organic yield andvolatility byNO smallperturbations inparticle biogenic emission,soeven trations. Isoprene isthemostabundant non-methane tions ismore low-NO volatile. Resultsshowthat ~60%of ever, it isfound that aerosol formed underlow-NO increased were asthe particles photochemicalaged.How- oxidized, highly low-volatility species the formation of decreasesparticles afterpeakaerosol indicating growth, that organic Ingeneral,thevolatility of that weare stillanalyzing. oxidized underhigh-NO that specieswithlowvolatility are formed whenisoprene is 5–15, loading12–150μg/m In theseexperiments,wevariedtheaerosol dilutionratio filled carbon. withactivated less steellineora15Lreservoir tothedilutionchamberthrough eitheraplainstain- uously (RH <2%)intheproduction chamberandtransferred contin- conditions wereParticles generated continuously underdry variables rate evaporation on the particle were explored. experimental different respectively. several of Theeffects (SMPS) andahigh-resolution massspectrometer (AMS), sured inreal sizer timewithascanningmobilityparticle sizeandchemicalcompositionismea- Particle evaporation. chamber, where are they dilutedwithpure airtoinitiate the resultant were gases transferred tothesecondsideof NO NO weretion. Particles generated inourchamberat different aerosol produc on localandglobalorganic - have aneffect compared to full evaporation of low-NO compared of tofullevaporation ticle volatility. H thermodenuder at temperatures different par toinvestigate x x conditions, 10% of aerosol mass remains even at 200°C, at aerosol 200°C, massremains even conditions,10%of concentrations at lowRH(<5%)andpassedthrough a x andhigh-NO 2 O 2 andHONOwere for usedasOHprecursor x experiments,respectively. Theyields x conditions.Our results show that x x conditions.Further, underhigh- (ananthropogenic emission)may 3 , andresidence timeinthedilu- x particles, suggesting suggesting particles, x concentration concentration x x concen- condi-

- aerosol produced from mixtures of biogenic and anthropo- before reaching final conclusions. Nevertheless, we devel- genic emissions may be transported over longer distances oped a unique measurement-modeling strategy that can be in the atmosphere than aerosol produced in the absence of applied in the future under programmatic funding to probe anthropogenic emissions. These results will be presented at organic aerosol formation mechanisms in a novel fashion. the American Geophysical Union fall meeting and are being Finally, in another series of experiments, we generated parti- prepared for publication. cles in the chamber, collected them of Teflon filters, and To reproduce the full aerosol lifecycle, both evaporation and sent the particle-laden filters to University of British Colum- growth of particles must be studied. To understand the bia (UCB) and University of California, San Diego (UCSD) organic particle growth mechanism, data from the chamber collaborators for analysis. Two types of particles were gener- studies were interpreted with the Model for Simulating Aero- ated. First, we generated particles from the photooxidation sol Interactions and Chemistry (MOSAIC), a comprehensive 1,2,4-trimethylbenzene under low NOx conditions in the con- aerosol box model developed at PNNL. Organic particles tinuous-flow chamber. 1,2,4-trimethylbenzene was injected were generated either from the ozonolysis of α-pinene or into 24 sLpm of pure air flow to produce concentrations of the photooxidation of isoprene in the presence of NOx (see 4.8 ppmv in the chamber prior to reaction. An additional the preceding paragraphs for experimental details). Particle 1 sLpm flow of pure air was bubbled through a 50% solution size distributions and mass loadings were measured in real- of hydrogen peroxide in water and added to the chamber. time using the SMPS. Organic particle formation in MOSAIC Photooxidation was initiated by turning on 105 UV (Q-labs was parameterized using 4-bin volatility basis set (VBS) fits of UVA-340) lights surrounding the chamber, which generate the time-dependent SMPS data. Gas-particle mass transfer of OH radicals from hydrogen peroxide photolysis. The resul- these four organic species was performed dynamically to tant particles were collected on filters and sent to UCB, size-distributed aerosols, as opposed to bulk equilibrium where they were extracted with water, combined with treatment used in previous modeling studies. Thus, in addi- ammonium sulphate, and analyzed using optical and fluo- tion to simulating the organic particle mass formed, MOSAIC rescence microscopy. As particle humidity increases, we also simulated the evolution of aerosol size distribution dur- observe the particles separate into two non-crystalline ing particle growth. This added feature can be used to probe phases in particles, one organic rich phase and one aqueous the underlying mechanism(s) of organic particle formation phase. These results reveal that atmospheric particles can such as reactive uptake vs. absorptive partitioning. In the undergo liquid-liquid phase separations that will affect the reactive uptake formation mechanism, the condensing gas-particle partitioning of atmospheric semi-volatile organic organic vapors are assumed to be effectively nonvolatile (i.e., compounds, the scattering and absorption of solar radiation, they are assumed to instantaneously react upon condensing, and the uptake of reactive gas species on atmospheric all of so that their vapor pressures are essentially zero), in which which have important implications for climate predictions. case the growth of particles would be controlled by aerosol These results were published in the Proceedings of the surface area. In contrast, an absorptive gas-particle partition- National Academy of Sciences. ing mechanism based on Raoult’s law assumes that condens- We generated particles from the photooxidation 1,2,4-tri- ing organic vapors are absorbed into the pre-existing organic methylbenzene under high-NOx conditions using similar aerosol mass while maintaining non-zero partial pressures methods. However, in these experiments, OH was generated over the particles after absorption. In this case, the growth from HONO photolysis rather than hydrogen peroxide, proof particles would be controlled primarily by aerosol vol- ducing particles that have different chemical composition. ume. The time-dependent evolution of the size distributions Particles were collected silicon nitride windows and sent to resulting from reactive uptake and UCSD collaborators who performed absorptive partitioning may be organic functional group analysis by quite different depending on the scanning transmission x-ray micros- actual vapor pressures of the copy with near edge x-ray absorp- condensing species, pre-existing tion fine structure (STXM-NEXAFS) aerosol size distribution and com- spectroscopy. Analysis showed high position. In general, reactive uptake concentrations of carboxylic car- tends to narrow the size distribu- bonyl, alkyl, and hydroxyl groups in tion while absorptive partitioning the non-spherical particles, imply- tends to broaden the size distribu- ing that they were liquid when tion. Model analysis of the chamber impacted on the window. These data for the isoprene experiments laboratory-generated particles had showed that the observed evolution similar spectral features with those of size distribution resembled that collected in field measurement produced from a reactive uptake Example of comparison of observed and modeled campaigns in Bakersfield, CA; Whis- mechanism in the early stages of (reactive uptake) size distributions at increasing tler, BC; the Scripps Pier, CA; Mexico growth while gradually shifting to reaction times. Excellent agreement at 50-min interval suggests reactive uptake mechanism is sufficient to City; and Chile. These results have an absorptive-partitioning type been submitted for publication to mechanism in the latter stages. explain the growth, but broadening in the observed size distribution compared to model at 110-min Aerosol Science and Technology. These results are still preliminary

interval suggests that absorptive mechanism has PN12027/2428 and need to be fully analyzed begun to play a role. 71 Chemistry 72 PN12029/2430 Specifically, the kinetic andequilibrium properties collected promising kinetic adsorption-based separation processes. separationwith different develop techniques andultimately modelsprovidement anddifference ameanstoexperiment COMSOL multiphysicsandAspenadsorption. Thesefiniteele- forequilibrium andkinetic process parameters modelsusing assembled for thisproject. Data collectedincludedboth andadsorption breakthroughtion analyzer instrument Initially, wecollecteddata usingahidden dynamic adsorp- theactualseparations process. plete physicalpicture of that providelation amore methodswillbedeveloped com- simu- advanced air collectionandpurification systems.New conceptsfor next-generationexperiments toadvance whole This project combinesnovelfinite-elementsimulations with requirements), benefitingfielddeployable further systems. operate at ambienttemperatures (noexternalcooling smaller thancurrent designs.Additionally, R-PSAschemes and lessoncapacity; therefore, future systemscanbemuch rely more smaller adsorption bedsbecausethey ontiming separations allowformind. Kinetic-basedgas theuseof with kinetic properties specifically in materials developed application-specific adsorbent gas separation needs for new adsorption acceptancefor orR-PSA)are gaining rapidly rapid-pressure-swing-Kinetic separations techniques(e.g., data arenotavailable. generally tooxygen-nitrogenrelevant separations, however, kinetic found limiteddata intheliterature. Withtheexception of nitrogen. Equilibriumdata for someadsorbentscanbe ments, andremove theneedfor consumables suchasliquid have require potentialtoreduce equipmentsize,- energy theadsorption process techniques that exploitthekinetics of adsorbent.Separation or theneedtoheat largeamountsof cryogens duetotheuseof intensive process andare energy theadsorption traps, relyontheequilibriumproperties of liquidnitrogen cooledorlargeambienttemperature of from theatmosphere. Current techniques,includingtheuse techniquesfor separatingsive andconcentrating tracegases inten- lessenergy moreThere efficient, isaneedtodevelop Richard M.Williams in GasSeparations Exploitation ofKineticProcesses currently used. fewer consumablescomparedtothemethods ration techniquesthatrequirelessenergyand is theidentificationofsorbentsandgassepa- toring. Asuccessfuloutcomeofthisproject applications suchasnuclearexplosionmoni- rations processesforuseinnationalsecurity We areinvestigatingkinetic-basedgassepa- Chemistry

adsorption kinetics. Arduetoslow Hyden data showingnon-equilibriumuptake of competition withN However,promising. acareful examination accountingfor single componentuptake data for XeonChabazite appeared assessing proposed instance, separation the techniques.For promising for argon separation. Data have beenvaluable in andappearingshowing thelargestkinetic themost effects accomplished. CaChabazite standsoutasthematerial Our process modeling toolpreparation hasalsobeen usingCOMSOLmultiphysics. are beingdeveloped that include macropore convectionandmicropore diffusion to describeadsorption kinetics, andmore complexmodels forceterized. Alineardriving approximation hasbeenused MS 13X,ETS10,CaChabazite, andMOF5have beencharac- charcoal,ing activated Carboxen 1000,Carboxen 569,MS5A, included N combinations have adsorbent-gas for largenumber of related toArseparation. sioned that pressure thiswillbearapid process cycle swing for aconceptpapers wholeairseparation. Itisenvi- develop willalsobeusedto tiphysics. Thedata andmodelingeffort kinetic inCOMSOLmul- process simulation beingdeveloped is plannedusingtheChabazite data andthemore complex shift from Apaper data collectiontoprocess development. ited, andmore data collectedwhenrequired. Thefocus will bents. Basedonmodelingresults,- adsorbentswillberevis In FY2013,wewillcontinue tomeasure promising adsor adsorption analyzer. adsorbent/adsorbate systemsusingtheHiden relevant asignificantfractionof measurement campaignof and otherkinetic properties andcompleting aninitial techniquestoextractdiffusivities tion-based data analysis carbon-based adsorbents.Milestoneshave includedsimula- the atmosphere wouldbelowerthanthat seenwithmany 2 , O 2 , CO 2 2 , Ar, Kr, Xe, andCH andO

2 indicates that Xeuptake from 4 . Adsorbentsinclud-

Fundamentals of Carbonate Formation: Interactions of Carbon Dioxide with Supported Metal Oxide Clusters Xiao Lin

of strongly bound CO2 on top of Obs. Strong quadrupole- The goal of this project is to obtain a quadrupole interactions were observed between Ti5c-bound CO and O -bound CO . In conclusion, this combined exper- detailed, integrated understanding of how 2 b 2 imental and theoretical study provides insight into CO2 to control catalytic reactions of CO2 (and rel- binding configurations on surface sites on TiO2(110), dem- evant molecules) on different oxide catalysts. onstrates their highly dynamic nature, and shows that quadrupole-quadrupole interactions play an important role in determining detailed surface structures at different As industries are relying on fossil fuels for energy and the coverages. global warming problem mounts, there is an immediate We also studied CO adsorption on TiO (110) with pre-dosed need for improving the fundamental understanding of CO 2 2 2 oxygen. CO molecules were found to sit preferentially on activation, which is instrumental for carbon sequestration 2 Ti sites next to the oxygen adatoms (O s) to form CO /O and CO conversion to fuels. Using TiO (110) and other well- 5c a 2 a 2 2 complexes. Our results show that at 50 K, CO is bound to characterized oxides surfaces as the models provides us 2 O only via weak dispersion forces. Further temperature with an opportunity to understand the fundamental a dependent studies revealed that the CO binding energy aspects of elemental processes such as CO adsorption, 2 2 next to O s is about 20 meV smaller than that on V ’s. diffusion, and reaction. a O Future studies will extend to other oxides, e.g., RuO (110), In FY 2010 and FY 2011, we studied the adsorption and 2 which will be grown epitaxially on the Ru(0001) surface dissociation of propane-1,3-diol and CO2 molecules on that has the same structure as rutile TiO2(110) but is metal- TiO2(110) surfaces. In FY 2012, we focused on the CO2 layer structure on TiO surfaces and CO interaction with the lic and significantly more active. It has been shown to acti- 2 2 vate CO readily to chemisorbed CO and form negatively oxygen adatoms on TiO by means of scanning tunneling 2 2 2 charged dimerized species, CO -CO . Further H can be che- microscopy, molecular beam techniques, temperature 2 2 2 programmed desorption, infrared reflection adsorption misorbed on Ru-cus sites and dissociate to form two bridg- spectroscopy, and theoretical calculations by dispersion- ing hydroxyl groups. While H transfer from Ru sites to corrected density functional theory and ab initio lattice oxygen is observed on bare surfaces, we will explore possibilities of H transfer to chemisorbed CO – leading to molecular dynamics. 2 C-H bond formation. Comparison of results from different Detailed level of understanding about CO adsorption, 2 oxides (TiO2, RuO2) with identical structural motifs will pro- site specific binding, translational and rotational diffusion vide valuable information about structure-activity relation- dynamics, and desorption kinetics on TiO2(110) surface ships on such model systems. have been obtained. Using a combined experiment and theory approach, we obtained a comprehensive understanding of the coverage dependent of CO2 behavior on rutile TiO2(110). We found that CO2 binds through one of the O atoms on top of Ti5c site and is tilted towards one of the neighboring Ob rows. At low coverages, this special configuration provides a low diffusion barrier that proceeds via rotation and tumbling mechanism. At 2/3ML coverage, the

CO2 diffusion is observed to cease. Pairs of CO2 molecules bound on neighboring Ti5c sites are observed with a free Ti5c site separating the pairs. As the coverage increases to 1 ml, all Ti5c sites become occupied and rotation motion also ceases leading to zigzag arrangement of tilted CO2 molecules, which is a result of fine balance between attractive quadrupole-quadrupole interactions and steric repulsion. Above 1 ML, CO can bind on O rows with a lower desorp- 2 b STM images from CO overlayers adsorbed on TiO (110) without tion energy compared with the Ti row, but the amount is 2 2 5c oxygen vacancy at 50 K. Panels (d-f) show high magnification images limited as 0.5 ML, a result of an out-of-phase zigzag from (a-c). The images are overlaid with the time average of the arrangement of CO2 on neighboring Ti5c rows that, while configurations sampled by CO during STM imaging. The grid dotted 2 PN10038/2280 blocking half of the Ob sites, allows for 0.5 ML adsorption lines correspond with the positions of Ti5c ions. 73 Chemistry 74 PN11040/2362 scCO at reactions PNNLtoinvestigate betweenmineralsand suite of thisproject apowerful are tointegrate goals of The chief (scCO tion depths,carbondioxide existsasasupercritical fluid basalt formations orexhaustedoilwells.Attargetedinjec- face CO face simulations reactivetransport tailored for subsur ment of ine forsterite andthe serpentineine forsterite antigorite, andtoinitiate plete experimentsthat were already- underway ontheoliv progress inFY2012.Ourfocus duringFY2011wastocom- FY2011setthestage forperformed for theremainder of While fundingfor mid-year, thisproject arrived thework water scCO mineraltransformation relevant processescally inlow geologi- of success willleadtoanincreased understanding microscopy (XRD).Thisproject’s (AFM),andx-ray diffraction spectroscopy,magnetic resonance (NMR) atomic force spectroscopy,red (IR) magicanglesamplespinningnuclear and caprock materials. Theseinsituprobes includeinfra- host relevant geologically these capabilities onarangeof saturated scCO to water- bedominated bydry caprocks willeventually formation water, andthepore spaceadjacenttooverlying whereby CO the atmosphere isgeologiccarbonsequestration (GCS), greenhousegases into One approach toslowthefluxof John S.Loring tions betweenSupercritical CO In SituMolecular-Scale InvestigationsofReac- whereas anhydrous scCO carbonating andhydrating minerals, capable certain of Relevant toGeologicalCarbonStorage between wetscCO thereactions conditions that at weunderstand reservoir seal. To itisvital GCS, predict theviability andrisksof thecaprock of thelong-termintegrity potential toaffect volume andthusporosity andpermeability, have the they solid minerals. Becausethesegeochemicalprocesses affect Chemistry carbon dioxide(scCO sorption/desorption studiesinsupercritical and hydration/dehydrationtoinsitu mineral dissolution/nucleation/precipitation sequestration, includingprocessesfrom cesses involvedincarboncaptureand for molecularmineraltransformationpro- This projectwillresultinnew, uniquedata 2 2 containing variable amounts of water, containingvariable amountsof andtofocus ). Withrepeated injections,thescCO 2 experimental capabilities recently developed in situ experimental capabilities recently developed reservoirs. 2 - , whichwillhelptoinform andguidedevelop 2 isstored suchasinsaline deepunderground, 2 . Water-bearing scCO 2 andhostcaprock minerals. 2 candehydrate water-containing 2 ). 2 is highly reactive and ishighly 2 willdisplace

- high surface areahigh surface calciteandporous amorphous SiO spacing andwater contentat anyscCO accumulated that canbeusedtopredict theinterlayer conditions.Resultshave been under geologicreservoir clays, chlorites, feldspars, and/ornaturalclays, rock chlorites,feldspars, samples. mineralstoincludenon-expandable to otherrelevant andfayalite; forsterite and3)broadenolivines ourstudies thickness ondissolvedionconcentrations, focusing on the adsorbedwater-film detailed studiesonthedependence of in three targetedpublications; 2)initiate andcomplete montmorilloniteclays, whichwillresult tion studiesof approachinvestigative to1)completehydration/dehydra- capabilityIn FY2013,wewillcontinue ourintegrated manuscripts, and11conference presentations. FY 2012resulted infour twosubmitted publishedarticles, content inthescCO increased withincreasing temperature, pressure, water and antigorite carbonation inwetscCO andinterpretanalyze IRspectroscopic data onforsterite There were three wasto mainfocuses inFY2012.Thefirst theproject. of eral studiesinthelast2years were min- purchased usingFY2011fundsinpreparation of montmorillonite. Additionally, chemicalsandequipment theexpandable clay,studies onhydration/dehydration of lonite, exposedtowetscCO theexpandable clay,hydration/dehydration of montmoril- Finally, ourwork duringFY2012wasthe thethird focus of intermediate wasalsodetected. bonation products, andanamorphous calciumcarbonate tons that adsorbmore water from thescCO areasilicate skele highsurface polymerized - formation of dissolution mechanismfor antigorite, whichleadstothe bicarbonate). Thiswork alsouncovered anincongruent ashigh3M dissolvedions(e.g., high concentrations of adsorbedwater,in thinfilms of andthese films contain scCO pyroxene-like mineral,wollastonite, exposedtowet IR,NMR,andXRDcarbonationgrated studyonthe The secondfocus wastoinitiate andcompleteaninte- metal carbonate precipitation. revealed that mineralsexposedtowetscCO revealed intercalated CO molecular-scale detailsaboutalso reveal thestructure of tonite (amontmorilloniteclay). Thespectroscopic results concentration for Na-,Ca-,andMg-saturated Wyoming ben- evidence that CO evidence timedirectXRD studyshowedfor spectroscopic thefirst 2 . Thisdetailedresearch showed that carbonation rates 2 andwater. Insummary, thework in 2 intercalates andexpandstheseclays 2 2 . Water films were showntoadsorbon and Minerals andMinerals 2 . This integrated IR,NMR,and . Thisintegrated 2 2 water saturation . Theseresults 2 2 andhinder are covered 2 car -

Increased Sensitivity and Improved Quanti- fication of TH and U in Particles by SIMS

Albert J. Fahey

lower statistical variability. The second issue will focus on This research will lead to a basic understand- understanding the basic mechanism of oxide and ion forma- ing of how oxide and elemental ions form on tion. Other compounds will be characterized on the various various substrates and under different condi- substrates used. We will attempt to understand and quantify these results, specifically relating them to basic chemical tions through elemental oxide measurements properties of the element and oxide ions formed. and dioxide signals from single particles under prescribed conditions. This may yield to an By the end of FY 2011, initial steps were taken to obtain sam- over-determined system amenable to least ple and substrate material and to bring the laboratory up to operational status. Substrate materials were procured to squares analysis, providing redundancy and mount uranium oxide particles with the appropriate doped consistency checks on the validity of the trace elements. A cursory method was developed to produce resultant data. SIMS substrates for silicon and gold materials. Purchased pol- ished graphite substrates were also used. Finder-grids were A specific high-value method involving the measurement of evaporated on silicon and carbon substrates through an thorium and uranium was developed during the late 1990s. etched Mo mask to facilitate relocation of particles once In advancing the method from its current state, the ioniza- dispersions were made, and initial results were obtained. tion efficiency of elemental and oxide ions will be explored In FY 2012, research was conducted to verify that U isotopes for various substrate materials and under conditions of measurements can be made as the oxide species on various oxygen flooding. The primary focus of this research is to substrates (Si and Au) without isobaric interferences. A table increase the ion signal intensity of the oxide and dioxide showing isotopic ratios and deviations from certificate values ions of thorium and uranium in secondary ion mass spec- of U isotopes measured as the dioxide species on a Si sub- trometry (SIMS). Once the magnitude of the effect is mea- strate is shown. The data, taken on the PNNL Cameca ims-4f, sured and confirmed on the PNNL IMS-4f, samples will be show that deviations are close to zero, within the uncertain- investigated on an IMS-1280. Data and methods pertaining ties. There is a small mass dependent bias, and the 236U to the application of method and fundamental mechanisms abundance is slightly high. This may be due to an improper of sputtering ionization will be obtained and analyzed. interference correction due to the 17O and 18O. Variations are likely due to instrumen- Ratios 234U/238U 235U/238U 236U/238U 235U (cps) tal instabilities that have R002 0.08967 ± 0.00010 10.40244 ± 0.00805 0.03955 ± 0.00022 502505.92 ± 38294.94 been remedied since data were acquired. In R004 0.08959 ± 0.00009 10.39986 ± 0.00464 0.03936 ± 0.00018 689961.36 ± 14933.44 addition, sample R006 0.08935 ± 0.00007 10.34925 ± 0.00542 0.03843 ± 0.00010 1367055.02 ± 34107.99 mounts have been made R008 0.0894 ± 0.00009 10.38017 ± 0.00617 0.03873 ± 0.00049 898314.62 ± 39148.11 to test the veracity of Delta 234U/238U 235U/238U 236U/238U measuring the oxide species on gold versus R002 0.23% ± 0.12% 0.26% ± 0.08% 3.33% ± 0.58% the elemental ion spe- R004 0.14% ± 0.10% 0.23% ± 0.04% 2.84% ± 0.47% cies on carbon. Detailed R006 -0.12% ± 0.08% -0.25% ± 0.05% 0.41% ± 0.27% results and their potential to customer require- R008 -0.07% ± 0.10% 0.04% ± 0.06% 1.21% ± 1.28%

ments will be discussed Isotopic ratios and deviations from the certificate values of U isotopes measured as the dioxide elsewhere. species on an Si substrate. In the coming year, production prototype This research will address two related issues that arise from substrates will be produced and U isotope measurements our observations: the first is directly applicable to improv- with the dioxide will be repeated. Data from the University ing measurements for nuclear non-proliferation and foren- of California-Los Angeles will be analyzed and reports written sics. We will determine optimal protocol for uranium, on the results of the measurements. A small round-robin test focusing on ionization the efficiency, identification, and may be initiated to test the method in FY 2013. reduction of interferences, increased signal stability, and PN11041/2363

75 Chemistry 76 PN10050/2292 impacts electron emittance. The observed roughening, roughening, impacts electron emittance. Theobserved cathode asurprising surface, result that dramatically tion procedure thephoto- leadstodramatic roughening of (HIM) images(seefigure) that- photocathode activa revealed at JLab.with staff Inparticular, ourheliumionmicroscopy tions. TheFY2011work prepared for collaboration further cleancondi- duced duringsamplehandling underrelatively contamination, consistentwiththatand lightsurface intro- (TOF-SIMS) resultsspectrometry showednoionimplantation photoinjector chamber. ionmass Time-of-flight secondary bulkdamageduetoionbackscattering inthe cation of roughening withnoindi- despitesurface crystalline highly (RBS) results showedthephotocathode lattice as trometry detailedRutherfordFY 2012.Highly backscattering spec- SpecialTopicsPhysical andBeamsin Review –Accelerators photocathodes that produced a manuscript published in expired of analysis In FY2011,wecompletedpost-mortem byionbackbombardment sputtering. surface Cslayera modelinwhichthesurface isremoved from the under operating photo-injectorconditionsconsistentwith mined that thephotocathode emissionyielddegrades anddeter (JLab) National Laboratory Accelerator Facility used(expired) photocathodes from theThomasJefferson of failure mechanisms.We analysis conducted post-mortem andmodelingphotocathode byunderstanding longevity Ourgoala collaborative istoincrease effort. photocathode ode designstoimprove current andfuture light sources in testing photocathode materials andenhancingphotocath- capabilities for new making This project and willdevelop photocathode damage. the mechanismsof ever, noconsensusexistswithintheusercommunityfor addressing thisissue;how- ment andisininitialstagesof - more photocathode scientific develop approach tonew tional lightsource communityrecognizes theneedfor a fying downstream accelerator orFELdesign.Theinterna- twoormore)simpli- bysignificantly of (afactor enormously reducecould potentially lightsource construction costs and emittancecharacteristics.Novelphotocathode designs or x-ray sources are dependentonphotocathode brightness ultraviolet new designs, outputcharacteristicsandcostof Whether basedonfree electron laser(FEL)orstoragering Wayne P. Hess and Development Light Source PhotocathodePerformance distribution). intensity), andemittance(electronangular energy dependentelectronyield(bunch properties, includingworkfunction,photon als anddesignsmeasuringphotophysical ties fortestingexistingphotocathodemateri- This projectdevelopsnewresearchcapabili- Chemistry

1000 and7000Csemission,respectively. afterthewaferhasgenerated morphology panels: changestosurface adding aCsmonolayer, whichresults roughening; lower insurface heating thewaferand conditioningstepof wafer; upperright:first anunused and undergoes of operation. Upperleft:smoothsurface conditioned astheGaAswaferisprogressively photocathode surfaces heliumionmicroscope imagesof view The panelsare 1µmfield of test facilities. to accelerator warranted,transfer thetechnology and, if Finally, onthermal emittance wewillmeasure theeffect usingtwo microscopies. morphology emittance andsurface rougheningavoid andcharacterize photoelectron surface metal photocathode withouthightemperature treatment to tance intheemissionmechanism. We ahybrid willactivate photocathodes andannealsamplestotestF-centerimpor compositions toenhancequantumyieldfrom CuandAg various thinfilms of In FY2013,wewilltesttheutilityof states.characterize mid-gap also produced ahybridmetalphotocathode andplanto to UPA We surface. andARPESfrom thecleansingle-crystal testsample. We comparedon apolycrystalline theseresults work functionreduction tonear 3.0eVmeasured previously sion spectroscopy (ARPES).TheUPSresults confirmedthe electron spectroscopy (UPS)andangle-resolved photoemis- MgOandperforming ultraviolet photo- two monolayers of with sition chamberbycoating acleanAg(100)singlecrystal progress. We testedthemulti-source further thinfilm depo- Tools toFY2012 inFY2011contributedgreatly developed will testinFY2013. that capped we GaAswafers arsenic vation makes useof thetypeusedat theJLab.GaAs photocathodes of Thisacti- has leadustopropose anovelmethodfor of activation

- Molecular Structure and Interaction at Aqueous, Non-Aqueous Liquid Interfaces and Catalytic Solid Surfaces Hongfei Wang For the past two decades, scientists have been using this Using cutting-edge instrumentation, this tool to understand and characterize the details of the project will lead to new discoveries in sur- structure and interactions at various molecular surfaces and interfaces. This has been a growing field and is ready face science and have broad applications on for exploring applications related to material and biologi- the characterization of structure and interac- cal surfaces and interfaces. PNNL played leading roles in tions at various molecular interfaces that are developing the experimental methodology and theoretical critical but elusive in many scientific and treatment to push forward the advancement in this field engineering fields. from the simplest air/liquid interface to more complex liquid/solid, liquid/liquid, liquid/membrane, and nanoparticle interfaces. The interface selectivity and submonolayer sensitivity of the second order nonlinear optical spectroscopy make it a Specifically, this project aims to employ the recent unique tool to study the structure and interaction of mole- development of ultrafast laser technology to push the sur- cules at various surfaces and interfaces relevant to energy face nonlinear spectroscopy and dynamics unprecedent- as well as environmental and biological processes. In the edly to higher spectral resolution, time resolution, and second order optical method, two photons with the same detection sensitivity to solve problems with complex frequency (second harmonic generation) or different fre- molecular surfaces and interfaces. It is a unique capability quencies (sum-frequency generation) interact simultane- and a game changer that will impact in understanding ously with the same set of atoms or molecules to generate molecular surface science, especially to complex surfaces a new photon at the sum of the frequencies. The symmetry that ubiquitously exist in energy, environmental, and bio- requirement for these processes forbids the processes from logical systems that cannot be accessed or studied in situ the bulk liquid or amorphous solid with centrosymmetry and at ambient or extreme conditions using other x-ray, but allows them to be generated from the surface or inter- electron, or ion probe tools and instrumentations. faces that always have asymmetry. The new photon gener- We commenced work in 2010 to support experimental ated is also background free from the large number of and theoretical studies applicable to various aqueous, non- incoming photons, so it allows sensitive detection such as aqueous liquid interfaces, and catalytic solid surfaces. The photon counting for submonolayer level vibrational and key to successful completion rests on the successful design, electronic spectral measurement. construction, and testing of the complete laser system; the

(a) (b) (c) The HR-BR-SFG spectrometer enables measurement with unprecedentedly accurate line shapes in different polarization combinations, and the coherent vibrational dynamics can be accurately predicted using the spectral data. (a) and (b) For the air/dimethyl sulfoxide (DMSO) interface SFG spectra, line shapes in different polarization combinations enable resolving the peaks of the two methyl groups of the interfacial DMSO molecule, which appear to be one peak in each spectrum in the ssp and ppp polarization combinations, respectively. The separation between peaks is 2.8 cm-1, compared with full width at half maximum ~8-9 cm-1. In addition, the phases and oscillator strengths of the ssp and ppp spectra can accurately determine the tilt angle of the two methyl groups from the interface normal. (c) The coherent vibrational dynamics as probed with ultrafast SFG free-induction decay (SFG-FID) is quantitatively reproduced with the high-resolution spectral data. This demonstrated that the HR-SFG is not only a tool for spectroscopic and structural studies, but a powerful tool for dynamics studies. Such information has not been expected, and this development is expected to change the related fields significantly. PN10058/2300

77 Chemistry 78 PN10058/2300 water spectraat theaqueous/SiO acterization; measured thehydrogen bonded andanalyzed dure char for surface usingSFG-VSinsitunanoparticle of heavy conditions. elementsunderdifferent of inliquid-liquidextraction acetylacetone, amodelligand andmeasured keto-enolfaces; the surface structures of Spectral resolution is~0.6cm trometer (HR-BR-SFG-VS). generation vibrational sum-frequency spec- band surface high resolution broad builttheworld’s first - successfully whichwasnotattainablemolecular interface, before. We SFG-VS enables resolving spectralandstructural detailsof the highresolution andexcellent lineshape inHR-BB- theyear.tested bytheendof InFY2011,weshowedthat All designandconstruction were underway, setup, and measurements.optical cellstobeusedfor varioussurface nonlinear opticalexperimentalsetup;andthevarious on CeO from aceticacid reaction catalytic product of we observed using highresolution andscanningSFG-VSspectrometer, non-aqueousliquidand solid/liquidinterfaces ical tool.For shallmake- the molecularinterface SFG-VSauniqueanalyt SFG-VS toresolve detailedchemicalshiftsandstructure of SFG intensityspectrameasurement. Theability for surface through accurateand detailedSFGspectralphaseanalysis spectral lineshape, allowingspectrallineshape analysis SFG thesurface spectra allowaccurate measurement of than current Highresolution BR-SFGspectrometers. SFG-VS structural analysis forstructural SFG onmetalsurfaces. analysis excellent signal/noiseratio, polarization allowingquantitative and thesspspectrallineshapeWith thehighresolution iswith SFG, ppp polarization, unclear. andthesspspectrallineshape isusually dominatedmolecule. by isusually SFGspectrum onmetalsurface theODT theconformationrelationships of allowdetermination of canidentify8peaks,andtheirintensityphase gold surfaces assembledmonolayer on ODT self in bothsspandpppspectraof the peaks.(b)Uniquefitting can alsoallowuniqueassignmentof resolved from thecongestedspectra. Thepolarization dependence air/water More interface. than10overlapping peakscanbewell cholesterol at and structure (a)TheHR-BR-SFGspectralof analysis. spectral toolfor complexsurface HR-BB-SFG-VS asgame-changing Chemistry 2 nanoparticle surface, withtheestablished surface, proce nanoparticle - (a) -1 , about 10to20timesbetter 2 andaqueous/CaF (b) 2 sur - - (C demonstrated onthelong-chainnonadecanenitrile overlappingin closely vibrational peaks.Examplesare captureclearly spectralinterference andphaserelationship layer surfaces.We showedthat high-resolution SFG-VScan relativelydetails of simpleliquidandLangmuir mono- Pushing thelimitforresolving spectralandstructure focusing onthefollowingmolecular interfaces aspects: In FY2012,wehave continued theresearch onvarious in surface/interface studies. in surface/interface opportunities SFG isapowerfultoolthat hasopenednew importantly, theseresults have shownthat high-resolution Most themolecularinterface. of mental understanding linear spectroscopy forward, leadingtoadeeperfunda- non- of surface movedthefield These studiessignificantly interactions. lar surface molecu- of dynamicsanalysis broadengreatly thescopeof overlappedtion for moleculeswithclosely peaks.Thisshall SFG-VS measurement canbeusedfor obtainsuchinforma- domainhigh-resolutionvibrational peaks,thefrequency dynamics measurement that islimitedtowell-separated neous broadening andinteractions.Unlike thedirect molecules,includinghomogeneousandinhomoge- surface predict thecoherentquantitatively vibrational dynamicsof obtained from high-resolution SFG-VSlineshape can demonstrated that thepeakintensity, positionandwidth thehigh-resolutionin thelineshapeof SFG-VS.We surfacemolecules of Exploring thedynamicinformation vibrational fingerprint the new region. These pointtothepossibilityuseC-Hvibrations as identify spectraldetailsincomplexvibrational spectra. as theexampletoshowthat high-resolution SFGcanclearly Wecules onmetalsurfaces. alsousedcholesterol molecule mole- to perform onSFGspectraof polarization analysis octadecanethiol (ODT)/gold anddemonstrated surface, how figure at left)intheC-Hstretching vibration for the region ple, 3or4intheliterature; weresolved 8peaks(versus see exam- For surfaces. aqueous/oxides andcatalyst interfaces, surfaces, surfaces andinterfaces,suchasnanoparticle and structure complex andinhomogeneous detailsof tostudythespectral Applying whathave beenlearned monolayers. 18 CN) and the 4-pentyl-40-cyanoterphenyl (5CT)Langmuir CN) andthe4-pentyl-40-cyanoterphenyl

Multiscale Synthetic Studies Targeted Towards the Development of Nanostructured

Heterogeneous CO2 Reduction Catalysts János Szanyi

CH4 can be varied systematically by changing the metal dis- The aim of this work is to synthesize, charac- persion on alumina support. Palladium atomically dispersed on the oxide support catalyzed CO reduction to CO terize, and test two families of nanostruc- 2 with above 90% selectivity, while CH4 was produced on large tured materials with catalytic activity toward metal clusters with more than 90% selectivity. Catalytic test- CO2 reduction. ing of carbon nanotube-supported palladium catalysts

revealed the critical role oxides play in CO2 reduction. Our work this year focused on alumina-supported Ru catalysts, Transition metals have been shown to be active components known to be active methanation catalysts. Based on Pd/ of homogeneous and heterogeneous catalysts in the conver- Al2O3 system results, we examined the effect the Ru active sion of CO2 to either methanol or methane. Understanding phase dispersion on the activity and selectivity of Ru/Al2O3 the requirements that govern the activities and selectivities catalysts. Measurements revealed that indeed, product of the catalytic centers in the reduction of CO2 will enable selectivity toward CO and CH4 varied significantly with Ru us to design and synthesize nanostructured, bioinspired particle size: small (“atomically” dispersed) Ru centers pro- heterogenous catalysts operating at moderate temperatures duced CO with high selectivity, while CH4 was the primary with the desired selectivities. With its high thermodynamic product on nm-sized Ru clusters. Apparent activation stability, CO is the most underutilized carbon source in the 2 energy determined from kinetic measurements at low con- production of high energy density organic molecules. To version levels systematically showed a ~20kJ/mol difference minimize the CO2 chemical conversion energy requirement, for CO and CH4 production, suggesting different reaction heterogeneous catalysts with high production effeiciencies mechanisms for formation of these two products. and selectivities are required. Thus, understanding the elementary steps of the reduction process and the nature We continued the synthesis of single metal ion sites sup- of the active catalytic centers is necessary. ported on and incorporated into polysiloxane frameworks. We synthesized metallo POSS frameworks containing Ru, Transition metals alone have proven ineffective for the Pd and Ni species and developed synthetic methods and activation of CO , leading to the development of bi/multi- 2 schemes to synthesize metallo POSS frameworks with bridg- functional catalytic systems. In enzyme catalysis, Lewis acid- ing metals. Structural characterization of these compounds base pairs are utilized for the activation CO that ultimately 2 using solution NMR (29Si, 13C and 1H) confirmed the existence leads to its reduction at ambient temperatures. The activa- of monomeric and dimeric compounds formed during the tion of CO needs to be accompanied by a reduction function 2 synthesis. We also continued evaluation of phosphine based of the catalyst, requiring the presence of a transition metal RuPOSS complexes and studied the conversion of CO2 to for the activation of molecular hydrogen. These functional- formate and monitored the conversion and kinetics by ities can conveniently be built into heterogeneous catalysts in-operando NMR spectroscopy. The catalyst synthesized containing both metals and oxides as active components. -1 reduced CO2 at moderate rates with nominal TOF of 1-3 h . We are designing and constructing active sites for CO2 reduc- This provided an evidence for proof-of-concept to demon- tion from two directions: well-defined metal clusters on strate the catalytic nature of these compounds. Catalytic oxides and multifunctional nano-scale clusters. These mate- evaluation of the RuPOSS framework on heterogeneous rials enable understanding of the metal oxidation state supports was achieved by a two-step process: the RuPOSS effects, the role of metal and oxide support interaction, and complex synthesized was deposited into a MCM-41 frame- the importance of the oxide itself in the overall CO2 reduc- work by incipient wetness techniques, and the solvent was tion activity and product selectivity. Multi-functional cata- removed. We studied the conversion of CO2 to CH4 and CO lysts utilizing polyhedral siloxane building blocks will be under reaction conditions and compared it to Ru on Al2O3 designed synthesized and tested. These supramolecular at two different loadings. architectures allow fine tuning and optimization of function- Detailed kinetic studies on both oxide-supported heteroge- alities required for CO reduction. The primary project goals 2 neous and POSS-based metal catalyst systems, together with for FY 2012 were the synthesis, characterization, and cata- further spectroscopy characterization in and ex situ will be lytic testing of selected materials, and the expansion of the focus of our research efforts in FY 2013. These studies detailed kinetic studies to a number of catalysts. should yield insight the mechanisms of CO2 reduction of The results of initial testing of Pd-based catalysts clearly these catalysts. showed that the selectivity toward the formation of CO and PN11055/2377 79 Chemistry 80 PN10062/2304 Lutidine•BCl hydrogen activation. insolution,andkinetics of chemistry structure andreaction paths,ing of reactivity andthermo- we obtainedresults inthree thrust areas: theoretical model- structures reaction paths andenergetics.DuringFY2011, computing complex. We acid-basepairs, screened several forming atermolecularreactive and entropy penaltiesof base are present inthesamemolecule toovercome energy acidandLewis bifunctionalsystemsinwhich Lewis tages of - project theadvan initiationFor inFY2010,weanalyzed and chemicals. infuels storingenergy asameansof thischemistry uses of tionships. Inaddition, wehave beenexploringthepotential simple,information predictiverela- willbeusedtodevelop andmolecularorbital).Thiscollective sity functionaltheory tion withcomputational electronic structure methods(den- spectroscopy, calorimetry, incombina- andx-ray analyses) experimentalmethods(nuclear magneticresonance suite of tohydrogenphase environments usinga activation relevant basepairproperties insolutionandsolid- acid-Lewis Lewis resolve thisproblem, wehave beenworking tocharacterize storageapplications. for catalysts To potentialenergy sive and inexpen- new of thedevelopment ing inturnhinders - understand Thislackof less commonandlittleunderstood. withnon-metalsis practiced,activation iswidely catalysts hydrogen usingtransitionmetal of While theactivation Donald M.Camaioni Storage inChemicalBonds Non-Metal ActivationofHydrogen forEnergy B(C reacting nitrogen baseswithtris(pentafluorophenyl)borane, mal reaction calorimetry. Ourresults were obtainedby theheat flow measured by isother the timedependenceof reaction rate constantsfor hydrogen byFLPsfrom activation capability toextractthe anew FY2012, wedeveloped For resonance spectroscopy. and nuclear magnetic (FLPs)usingcalorimetry pairs Lewis processes involvedinthehydrogen byfrustrated activation andkinetics for the methods tomeasure thermochemistry and lutidiniumBCl4-.Simultaneously, wehaddeveloped above 85°C, we observed the formation of lutidine•BHCl theformation of weobserved above 85°C, used incatalyticreductionofsmallmolecules. of hydrogenbynonmetalLewispairstobe mechanistic understandingoftheactivation renewable fuels.Thisprojectaimstodevelop nitrogen iscriticalforthedevelopmentof hydrogen, carbondioxide,methane,and The activationofsmallmoleculessuchas Chemistry 6 F 5 ) 3 in aprotic non-polar solvents (e.g., toluene and toluene and inaprotic non-polar solvents(e.g., 3 does not activate H doesnotactivate 2 at room temperature but

- insight intoH This approach detailedexperimental provided thefirst when designingrelated H systemstoactivate componentsthatimportant shouldbetaken intoaccount lutidine (Lut)—in the presence of H lutidine (Lut)—inthepresence of bases—tetramethylpiperdine (TMP),N-methyl-TMP(MeTMP)and We H alsoperformed acomputational studyof Time-dependent conversion data for reaction of B(C data forTime-dependent conversion reaction of the dative complex, in regions relevant toH complex,inregionsrelevant the dative away surface from possible toexplore thepotentialenergy strong bonds,electronic dative structure theoriesmake it instrument timeconstant. the reaction mechanism. AlthoughtheNH and A Physical Chemistry that appeared of inboththeJournal and results from ourproject work thisyear produced papers bromobenzene) under100-psiH the interaction of H the interactionof force of complex.Ourdetailedanalysis intheactivated charge-transfer interactionsare thedominantstabilizing complex plays asignificant andthatrole intheactivation FLPs. We theprecursor findthat structural of reorganization acid/base pairs NH acid/base pairs dissociation and heterolytic promoted byprototype Lewis activation parameters for step: theactivation parameters activation variable temperatures provided bothrate constantsand H2withlutidine/B(C6F5)3intolueneat for thereaction of themodeltodata squares Application regression. of tofittheheat flow model wasdeveloped data byleast luted from theinstrument response. Amechanistickinetic hydrogen stepcouldinsomecasesbedeconvo- activation heat flow tracesshowedthat therate information for the Dalton Transactions publications. C 7 H 9 N +H 2 activation by FLPs. byFLPs. activation 2 +B(C 3 /BX 2 withtheFLPprovides insightintothe 3 (X=H,F, andCl)tounderstand 6 F 5 ) 3

g 2 , corrected for thecalorimeter [C 2 gas. Our analysis of the the of Ouranalysis gas. 7 H 9 NH + ][HB(C 3 /BX 2 activation in in activation 2 6 . Ourprogress F 3 5 2 pairs form pairs ) 6 activation activation 3 F with Lewis withLewis 5 ) 3 – ]

OnSpec Advanced Condition Monitoring and Fundamental Research of Aggregation Precursors in Nuclear Materials Reprocessing Tatiana G. Levitskaia increase in the conductivity of the organic phase can serve as This project evaluates the potential of spec- indicators of the third-phase onset. troscopic monitoring to detect organic phase In another series of experiments, the mechanism of third aggregation in separation systems and define phase formation using proton-ionizable bis(2-ethylhexyl) the analytical capabilities of an aggregation phosphoric acid (HDEHP) extractant in an alkane solvent was monitoring system (referred to as OnSpec). investigated. FTIR measurements indicated that high loading with trivalent f-metal ions (M3+) results in the appearance of the bridged M3+-HDEHP aggregates of progressively increased This research contributes to the fundamental understanding size, which serve as precursors to the third-phase formation. of the formation of pernicious aggregation precursors that The observation that these precursors can be readily detected lead to third-phase formation, extractant oligomerization, by FTIR spectroscopy opens a new opportunity of their online and interfacial crud formation in liquid-liquid extraction real-time monitoring. Raman measurements indicated processes, along with monitoring these precursors online changes in water bands in the organic phase upon formation (OnSpec). In liquid-liquid extraction separations processes, of M3+-HDEHP aggregates that can also be utilized for the third-phase or interfacial crud pose significant operational monitoring purposed. and safety challenges and are associated with nuclear criti- The effect of the phosphate versus phosphonate extractant cality concerns due to the potential accumulation of fissile functional complexing group on the formation of interfacial material in equipment not designed to be critically safe. crud upon binding and high loading with M3+ was compared. While previous research has thoroughly defined PUREX sys- The structural aspects of the formed cruds have been exam- tem parameters that encourage detrimental third-phase for- ined by scanning electron microscopy (SEM). In this study, the mation, a significant knowledge gap exists regarding similar neutral and proton-ionizable phosphate and phosphonate processes in other separation systems and the progressive ligands were evaluated. It was found that regardless of the assembly of the aggregated species as precursors to the M3+ extraction mechanism, phosphate extractants form highly third-phase formation in general. To date, no efforts have ordered gel cruds, which adopt rod-like organization and been made to develop online spectroscopic monitoring readily retain water. In contrast, phosphonic ligands formed technology for the prediction of the organic third phase. less organized gels with partially opened water channels. This If developed, an OnSpec aggregation monitoring system will result speculatively suggests that formation of the crud in the significantly enhance our understanding of mechanisms of extraction systems can be manipulated by varying basicity of the third-phase formation and pathways to maintain homo- the functional group. geneity of the extraction organic phase, eliminating associ- This project ultimately resulted in identifying the spectro- ated criticality concerns and expanding the working operation scopic signatures of the third-phase precursors that can be range of the extraction processes relevant to the used nuclear utilized for the development of OnSpec aggregation monitor- fuel reprocessing. To achieve these goals, we have conducted ing system to enhance safety and operation control of the studies under various conditions to examine the spectro- used fuel reprocessing flowsheets. scopic signatures of the aggregation precursors for both neutral and proton-ionizable organic ligands, which facilitate the transport of metal ion across liquid-liquid interfaces via ion- pair and ion exchange extraction mechanisms, respectively. For neutral tri-butyl phosphate (TBP) extractant in n-dodecane solvent, the third phase formation upon contact with aqueous nitric acid (HNO3) or plutonium-IV Pu(IV)/HNO3 solutions was investigated by vibrational Raman/Fourier transform infrared (FTIR) spectroscopy or visible absorbance spectroscopy, respectively. For the first time, the spectroscopic signature of Pu(IV) in the heavy third phase was documented and reliably measured. This discovery can be used to develop online monitoring of plutonium speciation in reprocessing streams to prevent critical conditions. SEM images demonstrate that phosphate and phosphonate etractants It was also observed that the increase in water concentration form gels of different organization structure suggesting new pathways

to maintain homogeneous organic extraction phase even at high PN12060/2461 as measured by Raman spectroscopy with the associated loading conditions. 81 Chemistry 82 PN09057/2225 selected region intersects a fluidically generatedgradient. a fluidically oxygen selected region intersects withoptodesineitherPDMS(A)orpolystyrene(B) polymer.devices The positionshowingpore network Fluorescence lifetimeasafunctionof andarerescent compatible fiberopticpointsensors with mentation, current work influo - optodesextendprevious inopticalimagingtechniques and digitalinstru- advances fluorescent sensitive dyes.materials With withchemically polymer and across planarareas. Theoptodesconsistof chemicalimagingat specificlocations optodes capable of withinmicroenvironmentsgradients focuses onfluorescent studyingchemicalconcentrations and methodof A new Jay W. Grate Microfluidic Microbial Models Oxygen OptodeforChemicalImagingin contain microorganisms. spatially structuredmicrofluidicmodelsthat oxygen concentrationsandgradientswithin fluorescent imaging(optodes)toobserve is developinganalyticalmethodsbasedon redox stateandgradients.Thisproject experience spatialandtemporalchangesin members ofmicrobialcommunitieswhich the earthdonotactalone;theyperformas nutrient, geochemical,andenergycycleson Microorganisms thatplaycrucialrolesin Chemistry

replace the dry etchedsilicon pore network. replace thedry oped asilicon-on-glassmicrofabrication approach to guishable from thecontinuous fluorescent film.devel- We thepore network thestructure wasnotdistin- of imaging, cover plate andpore network; usingfluorescent microscopy continuous fluorescent film was present betweenthe Consequently,prior tobondingthesilicondevice. a design byspincoating thefilm ontheglasscoverplate plate side. Theoptodefilm wasincorporated intothis the pore from network thecover canbeimaged only containing etchedsilicondevice cover plate overadry a glass micromodelsnetworks. consistingof Previous microfabrication methodfor anew pore developing in twodimensionsovermicrofluidic pore networks and Research inFY2012 focused onimagingoxygen gradients alongalineacross theporegradients network width. pores in individual aswelldetectfluidically-generated fluids inthemicromodel cansense oxygen concentration incontactwith optode film ontheglasscoverunderside structures andborosilicate glasscoverplates. Thepolymer mental micromodels withetchedsiliconpore network experi- fabricationoptode films of intothe andassembly for incorporating amethodology planarpolymer developed fluorophore andits oxygen permeability. isdispersed, We inwhichthe onthepolymer usedependinpart range of centration detection,where thefluorescent intensityand materials oxygen forrin containingpolymer sensitive con- As theproject progressed, wefocused onplatinum porphy- oxygen-sensitivefluorophores. rescence lifetimeimagingof inin-housemeasurement capabilitydeveloped for thefluo- phores knowntorespond tooxygen concentrations, and fluoro intheproject,Early - weconsidered anumber of filmshave wehave synthesizedandpolymer formulated. nanospheres microspherespolymer aswell we included acommercial sol-gelmaterial andcommercial fluorescent-sensing materials and formats.of Thesehave measurement capability at anumber PNNL,weinvestigated microhabitats thisnew for Indeveloping microorganisms. inpore-network as across gradients micromodels serving This project is focused onsensingoxygen withinpores and thus respond tochangesinthesensedparameter. and areThese fluorescent designedtobe sensors reversible interest. luminescence isdependentontheparameterof using fluorescence measurements, where optodematerial whichcanbesensedoptically H ionconcentrations, eachof Key for chemicalmapping parameters includeO structured spatially heterogeneous environments. models of microfluidic as flow cells recent indeveloping advances 2 , CO

, and , and

This new method provides the precision advantage of dry coated with a PDMS overlayer for bonding. In each case, 2D etched silicon while creating a structure that is transparent fluorescence lifetime images were recorded while fluidically across all microfluidic channels and pores, and can be creating a dissolved oxygen gradient from zero oxygen to imaged from either side. A silicon layer is bonded to an 100% oxygen saturation. Representative images using our underlying borosilicate glass substrate and thinned to the in-house scanning techniques are shown in the figure. The intended height of the microfluidic channels and pores presence of opaque pillars between the observation direc- to be created. The silicon is then patterned and through- tion and continuous fluorophore film yields images that etched by deep reactive ion etching (DRIE), with the retain defined spatial structure in the sensor image. This underlying glass serving as an etch stop. This glass base is work has recently been published in Lab Chip. transparent, while the nonetched silicon is opaque. A glass The scanning method for fluorescent lifetime imaging is rel- cover plate with the optode film can be bonded in the cus- atively slow because it measures fluorescence lifetime one tomary way. This device can be imaged from the glass base pixel at a time. PNNL has purchased a Lambert LIFA-P fluo- side, so that the fluorescent oxygen-sensing film on the glass rescence lifetime imaging system that uses a modulated LED cover slip can only be observed through the microfluidic and a CCD camera with a modulated intensifier that deter- channels where the silicon has been removed by the mines lifetime at every pixel in the image field simultane- etching process. Thus, the fluorescent signal encodes ously. With this capital investment, it will be possible to the oxygen concentration in the channel, and the presence image the entire micromodel rapidly using the automated of a fluorescence signal reveals the microfluidic structure stage or follow dynamic processes at a selected image field. simultaneously. This system can measure the fluorescence lifetimes of long- Optodes were incorporated in the silicon-on-glass micro- lived oxygen sensing fluorophores, where conventional fluo- models using a Pt porphyrin fluorophore in a polydimethyl- rescent lifetime imaging systems on microscopes measure siloxane (PDMS) film serving as oxygen-sensing layer and only the lifetimes of short-lived fluorophores. bonding surface or with Pt porphyrin a polystyrene film PN09057/2225

83 Chemistry 84 PN12062/2463 we demonstrated that shiftsinspectral features criticalto conceptsuponwhichthe projectunderlying relies. First, als synthesis,andmodeling required totestandverifythe tasksinspectroscopy,three preliminary intertwined materi- In the4monthssincethisproject wecompleted started, relaxstickers more slowly, overmanyminutes). typically seconds;glow-in-the-dark (millionths tobillionthsof ation timesonascalecomparable tomoleculemovements spectralfeatures (color) and changesinrelax- energies of light. We measure shiftsin changesinspectra,particularly molecules that approach sitethat acatalyst absorb oremit arelated emittedlighttostudythe motionandkinds of of bedroom. achild’s We areing of usingcoloranddecay rate anyone totheceil- whohas pastedglow-in-the-dark “stars” desired chemical reaction to rate). Lightemissionisfamiliar that wouldotherwiseslow barriers ucts, loweringenergy site environment (atoms that bindtoreactants andprod- and emittedfrom tostudythat catalysts) immediate catalyst We lightabsorbed adapted aspectroscopy (measurement of into fuels. thebiomass of conversion or beinvolvedin,thecatalyzed moleculesthatmation canaffect, about theorientation of sites. Thisproject willyieldthosecompositionsandinfor can learnhowtoguidemoleculesandfrom thecatalytic aspossible, preferablyas locally rightat sowe thecatalyst products, itisessentialtomeasure thoseconcentrations reactants and reactions dependontheconcentrations of chemical their immediate environment. Becauserates of selves, muchlessisknownabout theirinteractionwith research them- thecatalysts intothemolecularstructure of canbeimproved. Whiletherethey hasbeenconsiderable work howthecatalysts andidentifyways inwhich aspect of Theresearchcatalyst. willtherefore one helpusunderstand liquidreactants andproductsstructure incontactwitha of quantify thedetailed(molecularscale)compositionand Thisproject providesline ordieselprecursors. away to reactions that leadfrom- theoriginalplantmaterial togaso toaccelerate catalysts anddirect thechemical the useof biomassintoliquidfuelrequires converting The process of Robert S.Weber Pores ofSupportedMetal-oxoCatalysts Polarizable ReactionMixtures Insidethe Probing CompositionandStructure of Chemistry tants andproductsofbiomassconversion. catalysts andtheirinteractionwiththereac- capabilities todescribeandunderstandthose ment, computational,andchemicalsynthesis We arecombiningPNNLadvancedmeasure- - information that will help quantify each type of interaction. information that willhelpquantifyeachtypeof site, thefluidnearcatalyst inthestructuredifferences of research willtestwhetherwecandescribe (right). Thenextstageof asolidpore (middle), andwhenthesiteisattached tothewallof in free solution(left),whenthesiteisattached toaplanarsupport confinement: fluidmoleculesinthreedifferent typesof different This representation site(*)interactingwithtwo showsacatalyst of amicroscopic channelorpore. of andwhenfixed tothewall surface; when mountedonaflat amongreactant isfree andproductlyst todiffuse molecules; molecularscaleconfinement: whenthecata- of ent degrees reactants andproducts inthree interactwithcatalysts differ samplestolearnhow catalyst more complex,realistic setof results toa thesepreliminary In FY2013,wewilladvance materials synthesis,andcatalysis. disciplines:spectroscopy,different quantumcalculations, We notethat further theproject combinedfour hasrapidly thosestagesprofited from priorworkAll three of at PNNL. ubility asaway topurifyandseparate themodelcompounds. in nonpolarsolvents.Goingforward, wecanexploit theinsol- ourmodelcompounds someof lowsolubilityof found avery to fluidpolarizability. sensitivities range of We unexpectedly tions suggestedmodelcompoundsthat shouldexhibita sitesare onwhichthecatalytic ports attached; ourcalcula- by polarcompounds.Resultscanbeusedtohelpdesignsup- bored bynonpolarhydrocarbons andwhenitissurrounded able ispredominately todistinguishwhenthecatalyst neigh- fluids,weare confidentthat wewillbe duces bothtypesof proalcohols have- high.Becausebiomasstofuelconversion fuel,hydrocarbons have lowpolarizability; waternent of and can becorrelated tofluidpolarizabilty. Theprincipalcompo- sitesandthatcomposition nearthecatalyst thespectralshifts Results showthat thespectraare tofluid indeedsensitive anditsenvironment. thecatalyst of spectral features tothemolecularstructure andcomposition Finally, weperformed calculations that relate state-of-the-art materials toactualcatalysts. that are equivalent functionally be modifiedinways that willpermitustocreate modelsof whose detailedmolecularstructure isbothknownandcan families four sample Next, weprepared inaseriesof thefirst project successare readilymeasured withequipmentinhand.

- Probing Structural Dynamics with High Spatial and Temporal Resolution

Nigel Browning

re-install. Thus far, the instrument has performed up to speci- We are developing a fundamental under- fication as a field emission microscope and is undergoing standing of materials dynamics (from micro- acceptance tests. The next stage of the installation is to build the two laser systems that will be used to drive the sample seconds to nanoseconds) in systems where and create the pulsed electron beam (expected completion the required combination of spatial and in November 2012). temporal resolution can be reached only In addition to the move and equipment installation, work by the dynamic transmission electron commenced to develop new technologies for implementa- microscope (DTEM). tion within the instrument framework. The first of these technologies involved collaboration with the National Center for This research will develop the DTEM capability to perform Microscopy and Imaging Research at the University of Califor- single-shot in situ measurements in liquid and gas stages nia, San Diego. The goal was to determine the ability of direct with a combined spatio-temporal resolution at least four electron detectors to provide increase in the signal-to-noise orders of magnitude faster than any other competing tech- over conventional CCDs and thereby increase the overall spa- nique (including aberration corrected STEM/TEM). In this tio-temporal resolution of the DTEM. Work here is expected temporal regime, the DTEM is expected to have atomic spa- to be completed by the end of September 2012 and show tial resolution, providing an in-situ TEM capable of studying that a simple frame shifting device plus detector should be nanoscale dynamic phenomena with several orders of mag- able to generate a frame interval of ~50 µs. nitude time resolution advantage over any existing in situ The second area of technology development involves build- TEM. In addition to providing unique insights into the inor- ing a complete model of the electron optical system to model ganic and organic systems studied, this research will test the the propagation of the large electron pulse within the elec- limits of temporal resolution and define electron optics and tron microscope. Here again, the overall aim is to generate alignment for the next generation of truly ultrafast TEM (i.e., the understanding that will enable an increase in the spatio- sub-ns). For nanoclusters in fluid around an electrode, inter- temporal resolution. In the first year of this project, a collab- actions between the particles leading to the nucleation and oration was established with the University of Michigan and growth of thin films will be examined for the first time. CEOS Ltd., the latter which is the manufacturer of the aberra- The ability to tion correctors within the DTEM. The agreement to provide observe “live” the key proprietary instrument parameters for the model has biological systems been accomplished and the expectation is that a complete will provide unique model will be available by the time the microscope comes insights into bio- online in 2013. logical systems. The third area of research being addressed in this project The primary goal involves the development of pulse compressors to be incor- in FY 2012 was to porated into a future microscope. Pulse compression is move the DTEM achieved through an RF cavity that effectively acts like a from the University temporal lens. Collaborating with McGill University (which of California-Davis recently showed that a pulse of 107 electrons could be com- (UC-Davis) to PNNL pressed to 80 femtoseconds), test parameters consistent with and install the DTEM use were defined and will be used to determine the microscope within limits of spatio-temporal resolution. the EMSL Q-wing. For all of these projects, the first year of the program was This involved com- design/initiation, where key experimental parameters for the pleting negotiations DTEM were defined. After the successful move and micro- for the sale of the scope install, test parameters and understanding propagation instrument to PNNL, of the electron pulse are underway and expected to coincide organizing the with the DTEM coming online in mid FY 2013. In addition, move, and setting The dynamic transmission electron key staff members were employed on the project to develop capabilities for the microscope (DTEM) the DTEM capabilities and operate the instrument. PN12063/2464

85 Chemistry 86 PN12067/2468 occupancies between structures. inatomic thedifferences observing to multiplesystems bysimply planecanbeextended structural boundary unitmodelfor thegrain structures.consistent withthoseseeninother fccderived The structural units boundariesshowaset of [001]tiltgrain Images of pathways for structural instability. Suchmeasurements environmental conditionsandidentifylikely function of a tobequantifiedas interfaces/nanoparticles stability of stageandenvironmental TEMwillallowthe aninsitugas of micro tomacro inreal scaleimportant systems.Application structure, thereby tothe extendingatomic scaleanalyses have onthe they rities/vacancies present andtheeffect structure, impu- aparticular thenumberthe presence of of and control. Ourapproach willprovide statistical validityto understanding of level and quantified,leadingtoanew composition controlling theseproperties canbeobserved imaging technologies,criticalfluctuations instructure and orsystem.Usingnew adevice the overallproperties of chemical processes that take placeinmaterials aswell and thestructural, electronic, control manyof Interfaces Nigel Browning Chemistry ChangesatInterfaces Quantitative ImagingofAtomicScale Chemistry occurs atinterfacesandinnanoparticles. the atomicstructureandchemistrythat methods necessarytoquantifychangesin This researchwilldevelopthestatistical

ture (i.e., double unit cell of SrTiO ture (i.e., double unitcellof structures- byasingle2×superstruc all three crystal terms, presenting similaritiesbetween In crystallographic theunit cell. O ionsandadd anextraSrioninthecenterof thesehard spheresdensest packing, mustrepresent Tiand to those observed for SrTiO to thoseobserved units (dislocation cores) that showaremarkable similarity 3D spacebutdonotpenetrate. TheCeO thehighestpossiblepacking densityin touch, achieve “hard spheres,” identicalmodeling spheres packed sothey metalscorrespondsFCC tothecubicdensestpacking of dislocations willlooksimilar).Theatomic arrangementin boundary same symmetriesinbulkcompounds,theirgrain boundary structures inSrTiO boundary as amodelsystem.Specifically, grain westudiedarangeof boundariesinoxides focused ongrain In FY2012,analysis and compositionwithlargerscaleproperty measurements. structure interface compare atomic-scale determination of for imageinterpretation framework tive way to istheonly principlessimulations.imental results Aquantita- andfirst has notbeenusedfor detailedcomparisonsbetweenexper ad-hocand chemicalimaginghasbeenappliedtitative only gas conditions).Quan- ambient underdifferent (specifically structuresnanoparticle underenvironmental conditions interface/ chemicalimagingof quantitative for theframework ditions. Thespecificaimistodevelop materials underreal working thefunctionalityof con- ing of - improveapplied ourunderstand together willsignificantly ies inCeO Experimental results boundar obtainedfrom [001]tiltgrain properties tobepredicted from thebulkstructure. rules willallowthedefect rapid rate –thecrystallographic materials for specific applications, toproceed at amore defectproperties, new andthereforeing of thesynthesisof rules- graphic that for willallowtheunderstand interfaces - crystallo haspotentialtocreateSuch analysis asetof structures. boundary rules grain for theformationset of of metalstructuresto face-centered-cubic(FCC) todeterminea Ce ions. For theSrTiO Ce ions.For thispacking with aunitcellof original four hard spheres of all eighttetrahedralvoidswithOionsandidentifyingthe densestpacking bythefilling from of thisformderived of structures metalsSrTiO (i.e., becauseFCC theparent thedensestpacking crystal structures of ing of boundariescanberelated- grain directly toanunderstand theseunitsincomparable coincidentsitelattice (CSL) of metals(seefigure).structural unitsseeninFCC Similarities 2 show a systematic arrangement of structural showasystematic arrangementof 3 structure from derivation thecubic 3 grain boundaries and also the boundariesandalsothe grain 3 andCeO 3 inallthree dimensions) 2 andcompared them 2 structure canbe 3 andCeO 2 have the

- - means that there must be klassengleiche subgroup-super- interfaces means that future experiments that will be group relationships (the full structure interaction is performed on individual grain boundaries can be related described by Bärninghausen trees). This can be represented back to a more extensive set of systems based on the pictorially (see figure), where crystal structures are plotted similarities between the structures and chemistries of in only the upper-right front octant of the “superstructure parent compounds. cell.” This octant also represents the whole unit cell of The goal for FY 2013 is to investigate the effect of cation the entire SrTiO structure. Unoccupied ion positions in 3 doping on the grain boundary plane structure. Cation segre- CeO and FCC metals are represented in the table by the 2 gation at the interface is often used as a means to control “Schottky void symbol.” Obviously, there are no unoccupied the overall properties of the interface. We will investigate ion positions for SrTiO . For a representation of the CeO 3 2 whether cations change the overall crystallography inter- structure, one needs to keep only those ion positions of the face, whether it forms a completely different structure or superstructure labeled as Ce in the accompanying table and is merely a chemical modification of the existing structure put O ions into the centers of all octants. One then obtains (similar to doping in the bulk system). Initial results from the unit cell of CeO . The FCC structure is analogously 2 SrTiO grain boundaries prepared to be intentionally doped obtained by the removal of all ions except those 3 with Pr and Eu suggest that dopants affect more than just arranged in a face-centered cubic Bravais lattice. chemistry. The presence of Pr in the structure seems to limit Having derived the crystallographic origin of the similarities the grain boundary plane to one particular structural vari- in structure, we can now use the understsnding to make ant, a symmetric structure with no rigid body shift. This structural predictions for other crystals. Also shown in the means that composition fluctuations in the nominally figure are the expectations for Mg2Sn, MgO, and ReO3. On undoped grain boundary plane cause two equal energy the basis of their corresponding occupied ionic positions structures to form while the intentional doping of Pr on the and Schottky voids in this table, one can predict structural other hand can shift the energy balance to the symmetric units in high-angle grain boundaries. For identical grain boundary plane. The Eu doping has the opposite effect boundary orientation parameters, the Mg2Sn structural and pushes structural stability in the other direction. In units will look “very similar” to the CeO2 units. Besides pos- addition, the structure is much more variable along the sible differences in distortions caused by the presence of boundary plane. point defects in the boundary plane (something outside Work is underway to evaluate the statistical prevalence of the crystallographic prediction but is the overall aim of this structures and correlate with energy changes at the bound- work), the main differences in units is due to ion replace- ary plane. These models for the single system will be ment. Simply, this will be because the structures of the two extended to analyze effects on multiple similarly structured compounds are those of anti-fluorite and fluorite structural systems. Once these are complete, we will perform in-situ prototype. Structural units of MgO will also possess similar reduction/oxidation reactions to investigate the stability of “overall shape” as FCC structural units but will show addi- this range of structures to the presence of oxygen vacancies. tional O ions if the transmission electron microscope is suf- Such analyses will provide fundamental atomic scale ficiently sensitive. Analogously, ReO structural units will for 3 insights into the mechanisms of vacancy/dopant incorpora- identical grain boundary orientation parameters look simi- tion in structural components involved in energy technolo- lar to their SrTiO counterparts with all Sr ions removed 3 gies such as battery electrodes and their effect on lifetime and possible differences in distortions. during cycling. The main result from FY 2012 has been identifying a classifi- In addition to the scientific results, there were several cation scheme for grain boundaries in metals and ceramics efforts underway to improve the ability to obtain high reso- based on the crystallography of parent compounds. This lution images by electron microscopy. This included the finding is important for future analysis of grain boundary move of a JEOL 3000SFF from the University of California- properties as it allows the “base structure” of all boundaries Davis to PNNL. This microscope was successfully shipped to be predicted without the need for extensive experiments and will be installed in FY 2013. Additionally, stages were and simulations. In the materials used in energy technolo- purchased that will permit materials to be studied by gies, these types of interfaces are key components in the tomography and while in-situ deformation takes place. overall properties. The properties themselves are typically All of these acquisitions will be employed during FY 2013. characterized by ionic/atomic diffusion into and out of the interface structure. This characterization framework for PN12067/2468

87 Chemistry 88 PN11075/2397 ture around molecularHNO theremarkablerevealed inthesolvation differences struc - In FY2012,ajointexperimentalandtheoretical study research.are area anactive of ionsandproton phenomena regarding important transfer the where asamodelinterface muchof serves interface experiments.Tothe interpretation of thisend,theair-water systemsand provideimportant sorely neededdata toaidin these of ourunderstanding quantum mechanicstoadvance basedin molecularsimulation methodology and develop Thus,there touse isauniqueopportunity observation. ments require highqualitysimulations inorder tointerpret and current andnext-generation x-ray absorption experi- toward controlling matter. spectroscopy sensitive Surface step ionsinsolutionisanimportant of understanding enhancement relative tothebulk.Thus,afundamental butlargepolarizableinterface, anionsshowinterfacial chargedionsarehard and/ormultiply excluded from the wherein theaqueousinterface, have presented apicture of In recent theoretical years, andexperimentalstudies interactionsareunderstood. important stillnotfully propensity tobepresent andthe at aninterface research; however,energy anion’s fundamentalsof solvation canchangehowweconductbasic at interfaces selectivity, andactivity. ion thenature Understanding of predictionsquantitative about reaction thermodynamics, environments isacrucialfacial componenttomaking tion protocols toprobe betweenbulkandinter differences novelsimula- of Thedevelopment bonds (i.e., chemistry). polarization, andtheability tomake andbreak chemical with interactionpotentialsthat containchargetransfer, molecularsimulation Thisrequiresinterfaces. theuseof that takes placeat and characterizingthenovelchemistry sciencesisunderstanding A significantissueinbasicenergy Marcel D.Baer in BulkandInterfacialEnvironments The StatisticalMechanicsofComplexProcess was postulated molecularHNO for thestabilization of molecular form at lowconcentration. Thesamemechanism and water stabilize moleculesat the thesolutionsurface the B and Physical Chemistry of were publishedintheJournal in bulk solutionat highconcentration.occurs Theseresults solvation, suggestingthat hydrogen bondsbetweenHNO Chemistry and homogeneousenvironments. and structureincomplexheterogeneous capabilities forsimulationofreactivity This projectwilldevelopnewresearch Journal of Physical Chemistry C(withcover). Physical Chemistry of Journal 3 under bulk and interfacial underbulkandinterfacial

3 that

- that emerges from ourstudyisthat IO charge. Thephysicalpictureatoms have negative formally region behaves like ananionwherein theiodate oxygen chargewhiletheother iodine atom positive has aformally and interfacial environments. and interfacial and thiocyanate. iodate, iodide,Schematic thesolvation representations of of acid (HCl,HBr, andHNO inconcentrated characterizing experimental observables interaction potentialsandsamplingmethodstoward of This includesbutisnotlimited tocontinued development usingmolecularsimulation. at interfaces complex chemistry We of abetterunderstanding willcontinue to pursue the propensity. Theseresults are acceptedfor publication in polarizablility surface thatlying isastrong indicator of thehydration structureterization rather of thantheunder conditions. Ourresults suggestthat itisindeedthecharac- their hydration structure underbothbulkandinterfacial propensities. surface different nificantly We have studied withsimilarpolarizabilities butsig- iodate andthiocyante) propensity. To three thisend,weinvestigated ions(iodide, solvation shellindetermining/predictingthe first interfacial We expandthisconceptandconsidertherole canfurther of . Letters Physical Chemistry of in theJournal kosmotrope (structure-maker). Theseresults were published alarge, weakly-hydratedostensibly anion,actsrather asa that one region of IO that oneregion of absorption spectroscopy We (EXAFS)wasinvestigated. found under bulkconditionsasmeasured byextendedx-ray structural rational for counterintuitive behavior of whyIO structural rational behavior for of counterintuitive theair-watermost notably andprovides thefirst interface cations inits ability toreside at hydrophobic interfaces, understanding of the hydration structure of IO thehydration structure of of understanding hydrated asazwitterion.Thisrevised and completely tion shellstructure about thepolyoxyanion, iodate (IO In ajointexperimentalandtheoretical study, solva- thefirst Farday Discussions. 3 - behaves like cation wherein the 3 ) and electrolyte solutions in bulk solutionsinbulk ) andelectrolyte 3 - is indeed strongly isindeedstrongly 3 - hasimpli-

3 - ), ),

3 - - , Earth and Space Sciences Computational Framework for Diagnostics, Validation and Intercomparison of Numerical Simulators for Geologic Sequestration Mark D. White Publications of code validations against laboratory or field This project is directed at developing a com- experiments, simulator verification workshops and code putational framework that makes validation comparison publications are all static snapshots of the state of numerical simulation science for greenhouse gas and benchmarking a dynamic process, where sequestration. numerical solutions evolve with modeling capability advances and benchmark problem For the past two years, our project was involved with developing the dynamic framework and assimilating internationally complexity evolves with growth in available recognized problems and numerical simulators for geologic field and experimental data. sequestration into the framework. We completed a prototype framework that provided access to problems/case studies Numerical simulation capabilities for modeling sequestra- through a geographic or process map, the interface of which tion processes associated with the fate of greenhouse gases enabled users to locate specific problems/case studies via geo- injected into geologic reservoirs has evolved rapidly during logic location or source, with the process map locating these in the past decade. Ten years ago, the accepted state of the a grid of coupled-process squares with increasing complexity. science was numerical simulators that could address dissolu- The principal objective for progress in FY 2012 was to establish

tion, structural, and mineralization trapping, isothermal the PNNL-developed STOMP-CO2 software as a world-class conditions, aqueous-based geochemistry, and decoupled numerical simulator for CCUS applications. We approached geomechanics for idealized conceptualizations of the sub- this objective by 1) developing new innovative, integrated and

surface. The standard for geologic sequestration modeling open-source capabilities into STOMP-CO2; 2) creating a suite of is continually advancing but now includes capabilities for benchmark problems that were documented and incorporated predicting hydraulic trapping, non-isothermal conditions, into the Geologic Sequestration Software Suite (GS3) collabora- transitions to subcritical conditions, ground-surface interactive computational framework; 3) growing the STOMP user

tions, injection wells, co-sequestration, supercritical-CO2- community through short courses; and 4) acquiring funding based geochemistry, coupled hydrology-geochemistry- for new code development cycles. From our efforts, four new

geomechanics, heterogeneous basin-scale domains, and capabilities were implemented into STOMP-CO2. The underlying wettability transitions. grid structure was altered to allow for faulted domains and internal boundaries. A trajectory-based fully coupled well Licensing requirements for CO2 injection wells in the United States include numerical simulation to forecast the fate of model was developed that automatically transitions between the injected greenhouse gases and displaced host reservoir flow and pressure controlled injections. An open-source rigid- fluids. PNNL has developed a suite of subsurface flow and body spring method based geomechanics solver was fully integrated into STOMP-CO without requiring an independent grid. transport simulators, named STOMP, which include capabili- 2 ties for modeling the coupled multifluid flow and transport, The vertical-equilibrium modeling approach was incorporated heat transfer, geochemical, and geomechanical process of into the simulator in a manner that provided full or hybrid geologic sequestration (THMC). The STOMP-CO code provides implementations for code users. The mathematical formula- 2 tion, application use, and benchmarking of STOMP-CO were the carbon-capture-utilization-sequestration (CCUS) commu- 2 nity with an open-source simulator with capabilities for deep documented in a PNNL report and released to the public. Project staff from PNNL held three STOMP-CO short courses saline reservoirs with fully integrated THMC capabilities. 2 during this fiscal year: in UAE, Korea, and at the U.S. Environ- When properly applied, numerical simulators and their ana- mental Protection Agency in Chicago, IL. lytical tools provide a valuable understanding of complex geologic sequestration processes and have been invaluable In summary, this research developed new critical simulator in determining the suitability of sequestration projects. His- capabilities, created a collection of benchmark problems for torically, simulators are verified against analytical solutions, geologic sequestration in deep saline reservoirs, educated benchmarked against alternative implementations, or vali- users in the software application, and documented the soft- dated against laboratory or field experiments. Generally, ware and benchmark problems, the latter of which have all three checks are used during the simulator development been incorporated into a new computational framework process, with benchmarking and validation dominating, as for collaborative projects in geologic sequestration. PN10014/2256 the code becomes mature, through extensive application. 90 Earth and Space Sciences Decision Support Research for Integrated Regional Earth System Modeling Richard H. Moss sions. Previous case study research with stakeholders has This project is researching what uncertainties identified specific information needs, but further work is matter most in different decisions, what required to categorize these needs more generally, relate them to decision making frameworks, and identify stan- information users want regional models dardized model couplings and work flows to meet informa- to provide, and how information can be tion requirements. presented in the context of uncertainty. Specifically, our project is continuing research on stakeholder information needs in the Midwest pilot region, using Regional decision makers trying to make sound climate these data to create a typology of decision criteria and deci- change mitigation and adaptation decisions face a wide and sion making processes. We are reviewing the project’s deci- complex array of uncertainties that span human and envision support methods, including the use of model-based ronmental earth systems. Because most of these uncertain- scenarios of future conditions. The results will be used to ties will not be reduced or resolved until after key decisions plan a pilot decision support process for iRESM and contrib- are made, regional planners face the conundrum of how to ute to the literature on scenarios in adaptation and mitiga- proceed with imperfect information. Decision makers need tion decision making. In addition, the project has analytical tools and information to help navigate decision developed a knowledge base of the breadth and depth of complexity and information uncertainty. This project is Midwest mitigation and adaptation decisions, with the developing a prototype strategy for relating model results results published in the journal Mitigation and Adaptation and information about uncertainty from integrated regional Strategies for Global Change. earth system models (iRESM) to different decision-making Key concerns for the Midwest region decision making processes, including risk communication methods (graphi- include renewable portfolio standards, biofuels production, cal, visualizations, mock deliberative processes, etc.) that hydrology in the Great Lakes Basin, water availability, convey uncertainties relevant to specific questions or deci- extreme precipitation/temperature events, and changes in long-term averages. Adaptation alternatives are being evaluated for water resources, urban infrastructure, transportation, agriculture, recreation, ecosystems management, and forestry. Mitigation decisions concentrate on the electricity sector, buildings, transportation, agriculture, and forestry. Based on a review of both peer-reviewed and grey literature, as well as interviews conducted with over 100 individuals spanning these interests, the team focused on energy and water, and developed and coded relevant variables: types of decisions, use of a formal framework such as cost- benefit analysis, criteria for decision making, and decision making processes. In addition, the team investigated decision support systems, including decision theaters and their effectiveness in presenting model results in decision making processes. During FY 2013, the decision typology will be extended to other regions and issues and then linked to standardized iRESM model configurations to address the most common decision variables and interests identified in the literature and interviews. A white paper on decision support options for regional adaptation and mitigation decision making will be prepared. In addition, results will be used to plan initial Relating the knowledge base regarding adaptation and mitigation numerical experiments with the iRESM framework as well as decisions to model development. Results will identify frequently used decision criteria and guide experiments and identification of a decision support experiment using iRESM outputs.

standardized workflows. PN12019/2420

91 Earth and Space Sciences 92 PN11025/2347 reservoir for safeCO reservoir thegeological theoverallsuitability of is criticaltoevaluate geomechanicalsimulation, not have which thecapability of multiphase flow, code(STOMP) andtransport reactive does atmosphere. nonisothermal, at PNNL,subsurface Developed opening of pre-existing weakening fractures, rock fracturing, opening of lead toCO through thecapleakage rock andoverburden, whichcan as displacements at the ground surface after 13.5 years of injection. of after13.5years as displacements at surface theground locations, andfluidpressure intheDEM simulation domain,aswell displacement,fracture horizontaldisplacement, vertical of Contours CO Geologic sequestration of Yilin Fang Sequestration and GeomechanicalCapabilityforCarbon Development ofCoupledFlow, Thermal mation processes toaddress keyresearch issuesinCO couples nonisothermalfluid flow andgeomechanicaldefor asimulationIn thisproject, toolthat weare developing containment. face fracture networks above targetCO caprock and thelong-termsealingability of used toevaluate be simulation thatlarge-scale subsurface willeventually a logical sequestration. interest Aparticular istodevelop major riskassociated withCO and largestress changesonnatural fracturesA andfaults. pore fluids tion phasecanresult inlargedisplacementsof use. However,encing energy highpressure duringtheinjec- influ- emissionswithoutadversely reducing gas greenhouse CO Earth and SpaceSciences ties inlarge-scalesubsurfacesimulation. tion. ThisprojectwilldevelopPNNLcapabili- research issuesinCO geomechanical capabilitiestoaddresskey This researchisdevelopingcriticallyneeded 2 injectionandstorage. Geomechanicalriskssuchas 2 release orthe intoshallowpotable aquifers 2 injectionandCO 2 2 geologicalsequestra- is an attractive optionfor isanattractive 2 sequestration ispotentialCO 2 storage reservoirs during during storagereservoirs 2 ’s long-termsubsur 2 geo-

- - 2

eSTOMP-CO couplingwithgeomechanicscodein eventual to facilitate explored usingOpen Source FiniteElementSoftware Elmer model. CoupledDEMandcontinuum scalemodelwas lution of effective stresses duringCO inrocks andfaults effective lution of reservoirs asscCO reservoirs (eSTOMP-CO nonisothermal, multiphaseflow code andtransport reactive has beenincorporated intotheexistingparallelsubsurface ElementMethod(RIEM),which ics code, Rigid-bodyInterface In FY2012,wemadesomeimprovement onthegeomechan- to couplemodelsat themesoscaleandcontinuum scale. Additionalfor methodswillbedeveloped modelintegration. at largerscales,andprovide reliable androbust submodels nating micro-mechanisms that treated cannotbeaccurately fractures,and closetoindividual capture domi- therelevant are usedtoexaminethestructure-property relationship in Enhanced discrete elementmethods(DEM)at themesoscale phase flow simulation capabilities tools. for thesuiteof faulted domainduringCO faulted geomechanics. Afracture propagation problem withina Discrete ElementMethodtosimulate coupledflow and of late deformations. ground We initiated modeldevelopment sitetosimu- was applied toanAquiferStorageandRecovery CO modeltosolvegeomechanicalmodelingof on theanalytical research hasproduced andGeosciences apaper inComputers scripts onthisresearch have beensubmitted.To date, our oncaprockthermal effect stability usingElmer. Two manu- FutureGen 2.0andWallula basalt pilotsite. will beapplied topotential fieldsites:storagesite for implement fracture andplasticmodelinRIEM.Themodels methodtocouple DEMwithRIEM.We developing will In FY2013,wewillcontinue testingtheDEM modeland contributed anotheronthe application totheASRsite. RIEM,andwe manuscript describingtheimplementation of geomechanical response andfluid flow. a Wealsosubmitted andaccountsfor thetwo-way effects couplingbetweenthe along withpressure duringCO increases inthereservoir andinducedseismicityincrease, faults, thestrength of of ture CO thegaseous emittedbycoal-fired gas powerplantsistocapgreenhouse - for our approachinjection. Onemethodology toreducing injection. Potentially hundreds of supercritical CO hundredsinjection. Potentially of term injections and the aggregation of scCO of term injectionsandtheaggregation theselong- impactof todeterminethecollective necessary willbe wells willoperate intheIllinoisBasin,andanalysis (eSTOMP-CO isothermal, multiphaseflow code and transport reactive non- intotheexistingparallelsubsurface linear elastictheory stepsincorporatedOur first ageomechanicalmodelfrom the caprock onstructural integrity. 2 geological sequestration, which considers theporoelastic geologicalsequestration, whichconsiders 2 2 2 . We alsoexplored caprock permeability and ) toprovide coupledgeomechanicalandmulti- ) for linearandnonlinearproblems. Themodel 2 - relyonpredicting. Initialanalyses theevo 2 effluent and inject it into deep geologic andinjectitintodeepgeologic effluent 2 injection was tested using this injectionwastestedusingthis 2 beneath the 2 injection injection

2 2

Development of Prototype Integrated Earth System and Environmental System Models Lai-Yung (Ruby) Leung managed and unmanaged ecosystems, and biogeochemical This research will provide a new modeling processes in a single modeling framework. Based on the Com- munity Land Model (CLM), this model will be scalable to the capability to represent interactions between extent supported by data to enable coupling with the integra- the atmosphere, ocean, terrestrial ecosys- ted human system models to capture human and earth system tems, sea ice, and biogeochemical processes interactions at a wide range of scales. For scalability, a distrib- that allow us to investigate climate change uted extension of CLM (DCLM) has been developed to incorpo- impacts at the regional scale.. rate more detailed water management in DCLM that takes advantage of the distributed framework, and couple DCLM and a crop model for more detailed representations of the man- Global climate models (GCMs) are the primary tools used to aged ecosystems and land management. Below, we summarize predict the evolution of the climate system. Through decades the progress made in each area of model development. of development, GCMs have demonstrated the useful skill of Regional Earth System Model (RESM). In FY 2012, we contin- simulating the global mean temperature response to various ued our work from last year, progressing in several areas. On natural and anthropogenic forcings of the last century. More developing a prototype RESM, the Weather Research and Fore- recently, global earth system models (ESMs) that integrate casting (WRF) model was coupled with CLM through a flux cou- multiple climactic processes have emerged to enable the pler. After testing the model over the western United States in global carbon cycle to be included for projections of global FY 2011, we shifted the testing over a North American domain climate change at the century time scale. Further, some prog- at 20 km grid resolution. A simulation is being performed with ress has been made in demonstrating potential skill in cli- boundary conditions from the NCEP/DOE global reanalysis for mate predictions at the decadal time scales. Despite these 1979–2005. The same model configuration was used to per- efforts, our ability to understand and predict regional climate form simulations driven by large scale conditions from the change and its impacts on the adaptive capacities and vulner- NCAR Community Climate System Model (CCSM4) for current abilities of environmental and human systems is still limited (1975–2005) and future (2005–2100) climate. Preliminary anal- because current GCMs/ESMs do not represent the myriads of ysis shows that the WRF-CLM simulation compares quite well human-earth system interactions that influence global and with observations and is similar to the WRF-Noah simulation regional climate change. In addition, important human-earth except for a somewhat larger warm bias. system interactions that manifest at the local to regional scales cannot be explicitly represented in global models Land Surface Modeling. CLM has been selected to provide a that have limited grid resolutions because of the hydrostatic consistent modeling framework with for modeling land surface formulation and computational constraints. and carbon cycle processes across the global model (CCSM), regional model (WRF), and spatially explicit models that will be This project aims to develop the prototype models to repre- coupled with energy infrastructure and operation models. To sent regional earth system processes and the environmental enhance the CLM capability for modeling across scales, the soil systems. This project includes the development of two major hydrology component of CLM has been replaced with that of components: the Regional Earth System Model (RESM) and the widely tested macroscale hydrological unit: the Variable the environmental system models of the integrated Regional Infiltration Capacity (VIC) model. Comparison of CLM-VIC with Earth System Model (iRESM). The RESM will provide down- CLM and observations over flux tower sites and the American scaled climate simulations to assess climate impacts and River watershed shows that CLM-VIC generally performs better adaptation options. This component includes coupling than CLM for vegetation with short roots. Also, CLM-VIC pro- state-of-the-science community models of the regional duced more reasonable simulations of subsurface flow, while atmosphere, ocean, land, and sea ice through a flux coupler the simulated subsurface flow produced by CLM often shows to represent regional earth system processes and their inter- high peaks and short recession that are un-characteristic of the actions. Representations of land use and water use are to be slow-varying base flow processes. These results were published added to interact with the natural water cycle to characterize in the Journal of Geophysical Research. regional human-earth system interactions that are critical for projecting regional climate change. For a more comprehensive evaluation of CLM-VIC, two types of numerical experiments have been designed. First, CLM-VIC has The second component includes developing a fully integra- been configured to run over the North American Land Data ted environmental system model for iRESM to assess local/ Assimilation System (NLDAS-2) domain, which covers the con- regional mitigation and adaptation strategies. Motivated by terminous United States and part of Canada and Mexico at the need to reconcile the ubiquitous redundancy and incon- 1/8-degree resolution. Simulations have been performed for sistency that exists across hydrology and ecosystem models 1979–2008 for comparison with observations and simulations in their representations of water and biogeochemical cycles, by other land surface models that are part of NLDAS-2. Our PN10027/2269 a model is being developed to represent hydrology, soil, results showed that CLM and CLM-VIC generally outperform 93 Earth and Space Sciences 94 PN10027/2269 streamflow over theCRB resolution. at 1/2-and1/16-degree for CLMhave beenusedtosimulate Model (RTM)developed Transport routing modelandRiver river inputs, thenew Washington withcalibration as of performed at University simulated bytheVICmodel Usingrealistic runoff routing. routing andthevariableriver storage methodfor channel subnetwork and hillslope methodtosimulate wave kinematic networks tracing methodtoconstruct aswellthe river Transport (MOSART), usedahierarchical River river tive routing model,Modelfor river ScaleAdap- model. Thenew implementations andfor couplingwithawater management andsubbasin-based CLM routing grid- modeltosupport providing amore accurateFY 2011withthegoal river of Basinduring andtestedovertheColumbiaRiver developed routing modelwas river River RoutingModel.Anew the modelsandMODISare at shownbythenumbers thebottomright. MOPEX basinintheUnitedStates. Theaverage latent heat fluxes for (NLDAS-2). Eachpointcorresponds at one withobservation/simulation AmericanLandData theNorth Assimilation System that are of part andMOSAIC) CLM andCLM-VICthree other models(NOAH, VIC, simulated latent heat modelsincluding fluxfrom fivelandsurface latent heat from theMODISsatellite observed andthe Comparison of sites across theUnitedStates. fluxtower heatin simulating fluxes surface ina number of inCLMleadtothelargestuncertainty runoff with subsurface to. Theresults suggestthat associated thethree parameters that heat thesimulatedters fluxes surface are mostsensitive toidentify parame- quantificationuncertainty framework key inthemodel,astudyhasbeenperformed an toapply ters CLMtohydrologic parame- To of thesensitivity understand NLDAS-2gridded domainfor thesameregion. andthesimulation overtheregularly ison withobservations Simulations have beenperformed for 1979–2008for compar andtheMidwest. Basin(CRB)inthePacificNorthwest River VIC usingthesubbasin basedapproach overtheColumbia experimentsinvolvetestingCLMandCLM- The secondtypeof . Hydrometeorology results of for submissiontotheJournal Apaper isinpreparationobservations. tosummarizethe betweenCLM-VICwith basin, showingtheclosestagreement LHat eachMOPEX simulated andobserved comparison of basins.Thefigure froma theMOPEXriver belowshows off compared LHandrun totheMODISsatellite- retrieved runoff other NLDAS-2 modelsinsimulating latent heat (LH)fluxand Earth and SpaceSciences

- CLM-MOSART overtheNLDAS-2 domain. seasonalmean streamflow simulated bythecoupled An exampleof model. theirrigation compared withdata toassesstherealism of domain. Thesimulated water irrigation amountis being been testedbyperforming simulations overtheNLDAS-2 modelingcapability irrigation inCLMhas tion, sothenew streamflow.- water Animportant usein many isirriga regions water managementon of MOSART tomodeltheeffects coupledwith The water managementmodelhasbeenfully prepared todescribethewater managementmodel. themodel.Inaddition, apaper willbe the performance of andimplementedtoimprove ments are being developed streamflow.flow for comparisonwithobserved More refine- Basintosimulateapplied theregulated totheColumbiaRiver unpounded byanupstream reservoir. Themodelhasbeen itis ponds andfrom themainstemat themodelunit if torepresent retention irrigation runoff from hillslopesurface water extractionmoduleextractswaterface at themodelunit usesandenvironmental constraints.Thelocalsur sumptive releases for andothercon- reservoir water irrigation supply of hydropower, partitioning andconstrainextractionbydaily rules for flood control, environmental flow, and irrigation on water storage, regulation that operating follows monthly The regulation moduleextractwater based at thereservoir includes twomainmodules:regulation andlocalextraction. operations onstreamflow. Thewater managementmodel reservoir of inFY2012tosimulate theeffects been developed Water Management Model.Awater managementmodelhas water management. CLMmodelwith routingfor intheoffline modeling river and RESMframework integrated can beusedinafully As noted,MOSARThasbeencoupledtoCLMsothat it streamflow. monthly simulation of basinswhere routing for isimportant river correct yielding similarresultssuperiorthantheRTMfor larger far andsubbasin-basedworks approach, wellwith boththegrid- resultssubbasin approach. showthat themodel Preliminary routing themodelistestingusing model at regulargrids, river Inadditionundergoing thenew revision. toapplying andis Hydrometeorology of been submittedtotheJournal modelanditsperformance paper has that describesthenew pares muchbetterwithnaturalized streamflowA thanRTM. The simulated streamflow modelcom- produced bythenew

Development of Regional-Global Climate Assessment Model Leon E. Clarke tion of a new method of modeling agriculture and land use This project creates a new, unique modeling that is particularly amenable to subregional analysis. The capability to explore climate change and its code for this material has been completed and has been responses, resulting in a regional integrated successfully tested with preliminary data. The final data set model to enable analysis at sub-regional and production of an operating version in GCAM were completed. The second critical prerequisite is the ability to run scales (such as states or provinces) in an GCAM at arbitrary time steps. Prior to FY 2010, GCAM model integrated framework explicitly linked to ran on 15-year time steps. This temporal resolution was global-scale phenomena. based on the long timeframe generally used for integrated assessment analysis. In addition, the RGCAM model will This research is part of a larger effort to perform climate require linkage to a range of other modeling tools (e.g., change integrated assessment at a regional scale. Inte- detailed land use, regional climate, and electricity models), grated assessment modeling in the context of climate all of which operate at temporal scales well below 15 years. change is a form of analysis in which all relevant human The code and a range of model input developments were and natural processes are brought into a single computa- completed, with the new capability included in the GCAM tional framework; i.e., they combine humans systems such core version. In FY 2011, a fully subregional capability for as economic systems, energy systems, and agricultural sys- modeling building energy was completed, along with an tems in a single platform with key natural systems such as exploratory version of a 50-state version of the model. The the climate and carbon cycle. Although integrated assess- code developed to create the buildings facilitated development modeling has a history at the global scale, there is ment of additional subregional sector representations. no regional level integrated assessment model or modeling In FY 2012, the prototype version of the RGCAM, 50-state projects. This project will build such a regional scale inte- buildings model created in FY 2011, was made a fully func- grated assessment model within the context of a global tioning model. This included bringing in a new representa- modeling framework. We are maximizing the fact that tion of energy service demands that explicitly allowed for PNNL uses the Global Change Assessment Model (GCAM), the consideration of changing heating and cooling degree an integrated assessment model that combines human days over time. The model was then utilized for other systems and natural systems into a single platform for experiments, including uncertainty analysis and sensitivity consideration of climate change and climate change analysis exploring the implications of different technology response strategies. assumptions, long-term climate goals, and climate model The implementation of this project is based on the comple- outputs. Beyond the buildings model, there were further tion of two prerequisites, both of which were completed developments to elements of the 50-state regional capabil- during FY 2010. One critical prerequisite was the comple- ity in RGCAM, including industrial energy demands, transportation demands, and electricity generation. A r-based data system was developed to take data from a variety of data sources and seamlessly balance it to create 50-state assessments of these systems for use in GCAM. In addition to work on GCAM, the BEND model was completed, which provides explicit modeling of adaptation and mitigation in fine detail within buildings. BEND permits a user to develop a population of representative buildings for a geographic region, calibrate those buildings to actual historical weather and energy consumption, and accurately project energy consumption under a variety of different climate and technology scenarios. Finally, data protocols were developed for calibrating RGCAM and BEND for use in coupled experiments in other portions of this project. Heating and cooling energy demands late in the century, under the assumption of no efforts to constrain carbon emissions. PN10030/2272

95 Earth and Space Sciences 96 PN12031/2432 reflects an eventual declineinannual emissionsafter2050. reflects an annual carbondioxide emissionsworldwide, whereas theB1scenario scenarios intheCMIP3archive. TheA2scenarioreflects increasing without climate influence(“fixed”) CCSM3A2andB1 and forthe use(2005,2050,and2095)inscenarios Heating andcoolingenergy increas- designedtoconductaseriesof multi-year effort since2010,theiRESMencompassesa under development one-of-a-kind multi-scale, multi-domainmodelingsystem inherentmodeling uncertainties inthesimulation. A nexus,whilealsoaccountingforland-water-energy the adaptation decisions,andtheassociated impactsonthe lating regional climate change, and regionalmitigation numericalcapability experimentssimu- through aseriesof Wealternatives. willdemonstrate theiRESMframework oradaptation mitigation certain bility orinfeasibilityof gistic orantagonistic and,assuch,may indicate thefeasi- Theseinteractionsmay besyner regional scalemodeling. resources through integrated, berevealed that canonly between humansystemsandland,water, andenergy actions taken inresponse willproduce critical interactions change, climate policy, andadaptation andthemitigation The iRESMispremised onthehypothesisthat climate Jennie S.Rice logy PenetrationintheMidwestPilotRegion System InteractionsduetoRenewableTechno- Exploration ofHumanandEnvironmental Earth and SpaceSciences characterization research. stakeholder decisionsupportanduncertainty ual modelingefforts withtheresultsof (iRESM), bringingtogetherresultsofindivid- integrated RegionalEarthSystemModel We aredemonstratingthecapabilityof - duced themost significant results. Itdemonstrated not assessmentmodelpro- demand intheregional integrated buildingenergy climate tostate-level the couplingof thevarious Phase1experimentsconductedinFY2012, Of and emissionsscenario. parameterization, modelchoice, downscaling method, from uncertainty model theexplorationfacilitating of with additional climate simulations insimilar formats, ments. Thecouplingsare repeatable andflexible for use couplings andtestedtheinPhase1experi - variable, spatial transformations toaccomplishthemodel codes andscriptsthat perform integration necessary of from theclimate datasets. Thisproject produced aset matchscales that what donotnecessarily isavailable climatemodels hasparticular data needsat particular thedownstream spatial andtemporalscales,whileeachof variables andusedifferent sets provide setsof different models are climate challengingbecausethedifferent data- The modelcouplingsbetweenclimate andotheriRESM the iRESMframework. asakey modelcomponentof system modeldeveloped RCP simulations are produced withtheregional earth assessment.The (RCP) beingusedintheupcomingfifth results Representative Concentration for Pathways thenew downscaledsimulationment report anddynamically Intergovernmental assess- PanelonClimate Change(IPCC) Model Intercomparison Project (CMIP3)usedinthefourth theCoupled downscaledclimate datacally from Phase3of characterizationuncertainty process, we are usingstatisti- Tocould begin. climate futures represent inthe arangeof stepbefore first theinitialexperiments were anecessary model, andthecrop model.Thesecouplings productivity assessment demandmodel,theregional integrated energy between regional climate modelingandthebuilding couplingstrategies of project thisyear wasdevelopment the accomplishmentof terization (UC)process. Aprimary experiment todemonstrate charac- theiRESMuncertainty Theproject Phase2 alsoinitiatedframework. itsfirst the regional climate modeltovariousmodelswithinthe economy,energy andoncrop bycoupling productivity demand,itssubsequentimpactontheoverall ing energy that signalonbuild- pilot region andexplore theimpactof experiments toidentifytheclimate signalinthemidwest Phase1 During FY2012,theproject conductedaseriesof couplings withintheframework. complex experimentstoexerciseingly variousmodel only that climate has a significant impact on building project identify a relatively small set of relevant modeling energy demand at the national scale, but also that this uncertainties for testing with the fractional factorial sensi- impact exhibits large spatial heterogeneity across states. tivity analysis approach. The uncertainties selected were In the experiment, the climate change impact on heating population growth, rate of growth of economic output and cooling demand in the 21st century is measured (GDP) per worker, future carbon dioxide emissions, and through estimating heating and cooling degree-days (HDD/ choice of general circulation (i.e., global climate) model CDDs) derived from gridded climate data for two future (GCM). The analysis utilized three scenarios (low, medium, greenhouse gas emissions scenarios and comparing the and high) for population growth and for GDP per worker, results to HDD/CDD calculated without climate drivers two future emissions scenarios (A2 and B1), and three and based only on the influence of socioeconomic develop- GCMs. A full factorial analysis would require 162 cases, ment, technological change, and energy service expansion whereas the fractional factorial approach focuses on the on historical levels. The figure indicates the spatial pattern five first-order cases (includes the policy variable) and the of building energy use by state for these three cases. The ten second order pairwise interactions. scale impact in building energy use modeling emphasizes The experiment compared the performance of the frac- the importance of developing a building energy model tional factorial analysis to the full factorial with analyses that represents socioeconomic development, energy service of variance and determined that the fractional factorial expansion, and their interaction with climate change in a explained over 99.9% of mean squared error, demonstrat- spatially explicit manner. ing the efficiency of the fractional factorial approach for In the Phase 2, UC experiment begun during FY 2012 this example. In addition, the results indicated that just expanded on the Phase 1 experiment described above two of the four modeling uncertainties (future carbon (involving climate-driven building energy demand in the emissions and rate of GDP growth) explain over 98% of the regional assessment model) to demonstrate the application variance. This result shows that only these two uncertain- of the stakeholder-driven UC process to a policy decision ties should be carried through into the uncertainty propa- regarding future building standards. For FY 2012, the focus gation phase. Moreover, the overall reduction in the was demonstrating how the identification of a particular dimensionality of the UC was significant for this experi- stakeholder decision and associated decision criteria, along ment: an order of magnitude reduction in the number of with the use of fractional factorial sensitivity analysis, pro- sensitivity cases to be run (162 vs. 15) and the number of vide an efficient way to identify key modeling uncertainties variables for which probability distributions would need to to pursue with uncertainty propagation. Without such an be developed was reduced by half. approach, analyses conducted in another project have In FY 2013, the continuing UC experiment will develop shown that the dimensionality of the uncertainty charac- and apply uncertainty propagation methods for the iRESM terization across the integrated framework becomes too framework. Additional Phase 1 experiments are planned unwieldy and results in prohibitively long run times, comparing the impacts of the RCP and CMIP3 simulation even in a high performance computing environment. results. Also, the project plans a series of Phase 2 model The decision context for the UC experiment involved three coupling experiments. These will couple the regional inte- alternatives for future building energy codes and appliance grated assessment model with each of three sector models and equipment standards: no change, standards resulting (building energy demand, crop productivity, and energy in a 50% reduction in energy use by 2050, and standards infrastructure), all also coupled with climate. These experi- resulting in an 80% reduction by 2050. The initial decision ments will address particular stakeholder interests from criterion for comparing the alternatives was residential and the Midwest pilot region, investigate alternative coupling commercial building services costs per square meter of strategies under uncertainty, and demonstrate the floorspace. While broader sets of uncertainties (and deci- multi-scale, multi-domain modeling capabilities of sion criteria) will be addressed in later experiments, the the iRESM framework. demonstration nature of this experiment dictated that the PN12031/2432

97 Earth and Space Sciences 98 PN10048/2290 systems andenergy, water, andlanduse. testbeds willhighlighttheinteractions betweenthenatural metrics. Theprototype iRESM evaluation and developing thetestbeddatabase,regions andtimeperiods,developing selectinggeographic for modeltestingandevaluation, addressed thescientific approach bytheiRESM,developing sciencequestionsorhypothesestobe a focused setof iRESM modelanditscomponents.Thisincludesdeveloping the to selecttheiRESMregionaltestbedthat willevaluate andexecute asystematicThis project process willdevelop behaviors andsensitivities: behaviors model andunderstand oped andapplied toevaluate scientific - approaches have beendevel typesof ments. Four andtoprovideuncertainties insightsfor modelimprove- modelbiasesand key andquantification of understanding tance toensure canimprove that the themodelevaluation vitalimpor testbedsorpilotstudiesare of the selectionof process through careful andsystematic specification and orvalidation. Establishingis evaluation arobust evaluation anymodelingactivity processes. componentof important biogeochemistry, socioeconomic andglobalimplications of backs andinteractionsbetweenenergy, hydrology, climate, regional feed- of the regional scaleenable theevaluation hydrology, at land systems,socio-economics,andenergy climate andtoolsthat with integrate modeling frameworks systemsthrough new andtestingof The development Lai-Yung (Ruby)Leung Specification andSelection (iRESM) Prototype RegionalTestbed Integrated RegionalEarthSystemModel • • • • Earth and SpaceSciences system models(iRESMs). bilities oftheintegratedregionalearth demonstrate andevaluatetheuniquecapa- approach todevelopingtestbedsstudies This projectaimstoexecuteasystematic conditions. simulation underacommonsetof and uncertainties Model intercomparison, toassessmodelbiases externalperturbations of processes, ortheeffects modelparameterizations, modelfeedback of theimpacts Sensitivity experiments, tounderstand modelerrors to attribute thesources of modelsimulations, Analysis anddiagnosticsof and theassociated spatial andtemporalvariability on theirskill inreproducing thehistoricalconditions Hindcast experiments, where modelsare evaluated

- percentile precipitation duringsummer, showingwith andWRF-CLMsimulated 95th andCCSM observed son of current climate (1975–2005).Thefigure shows acompari- from for theCommunityClimate the SystemModel(CCSM4) bylarge-scaleconditions ocean)driven than interactive (i.e., RESMwithprescribedtemperature seasurface rather RESM,asimulation hasbeenperformedFor withWRF-CLM highlights somekeyresults. and efforts modelevaluation summarizes examplesof evaluation. This report phasemodel teams toward thefirst In FY2012,progress modeling hasbeenmadebydifferent two-way coupledmodelswillbeevaluated. InPhase3,some one-way couplingwillbeevaluated. conditions. InPhase2,somemodelswith observed by ated inahindcastmodewitheachmodeldriven - modelsare phases.InPhase1,individual evalu different numericalperforming simulations andevaluating inthree includes Theframework the iRESMmodelswasdeveloped. During FY2011,acoordinated for evaluating framework characterization.uncertainty and for modelevaluation tomap astrategy we began infrastructure Finally,for andhydrologic energy modeling. thepilotregion, includingdata modelingof supported energy, andwater. adatabase that Next,wedeveloped sents significantinteractionsamongclimate, land use, pilotstudybecauseit choiceforrepre thefirst - region of emergedasthe andSoutheast.TheMidwest Northeast, states Southwest,Great intoNorthwest, Plains,Midwest, the U.S. theconterminous National Assessment,dividing In FY2010,theteamadoptedregion definitionusedin June-July-August averaged over1975–2004. (upperleft)andWRF-CLM (upper right)inmm/day forCCSM4 the (bottom)andsimulated 95thpercentile precipitationObserved from increased grid resolution, WRF-CLM simulated larger intense precipitation than CCSM and compared better with observations. The Community Land Model (CLM) has been applied to the Midwest pilot study region at 1/8-degree resolution using both grid based and subbasin based configuration using NLDAS-2 atmospheric forcing. Extensive analyses are being performed to evaluate the simulations using observations and compare the grid based and subbasin based approach to understand their pros and cons at different resolutions and river basins with different climate/hydrology regimes. The new river routing model, Model for Scale Adaptive River Transport (MOSART), has been applied to the Midwest pilot study region. The figure shows a comparison of the simulated streamflow at different stream gauge stations Observed (black) and simulated (red) monthly streamflow by MOSART when MOSART was driven by runoff generated by NLDAS-2 at three stream gauge stations in the Missouri (upper), Mississippi models. The results show reasonable agreement with (middle), and Ohio (bottom) river basins. observed streamflow in basins that are not strongly influenced by human activities. For evaluation of the energy consumption model (BEND) and electric consumption model (MELD), the calibrated models were used to simulate historical energy and electricity consumption. For BEND, historical weather data and historical building characteristics for 1990–2005 were used as inputs to simulate electricity consumption for selected utility regions and natural gas consumption for selected states. A similar approach is used for MELD. The above figure shows a comparison of estimated and actual electricity consumption for commercial, residential, and industrial regions. These results appeared in the publication the Journal of Great Lakes Research.

Observed (black) and simulated (red) monthly streamflow by MOSART at three stream gauge stations in the Missouri (upper), Mississippi (middle), and Ohio (bottom) river basins. PN10048/2290

99 Earth and Space Sciences 100 PN10056/2298 and scCO caprock theoverlying of rock. Theintegrity the reservoir intotheatmospheretransport andthestoragecapacity of To improve our understanding of subsurface scCO To subsurface of improve ourunderstanding various wettability conditionsislimited. Geological storage of CO Geological storageof Martinus Oostrom Related toCO Sealing Efficiency and Trapping Mechanisms Micromodel Pore-Scale StudiesofCaprock- needs tobeexceeded before scCO pressure that atcapillary the caprock: interface reservoir the isameasure of needed. Thecaprock-sealing efficiency andtrappingrelated are tocaprock-sealing efficiency processes pore-scale experimental andnumerical studiesof by scCO brine displacementof tions are scarce, andknowledgeof ring at thepore-scale. Data related interac- tointerfacial wettability) occur porous medium-fluidinteractions(e.g., density,properties (e.g., viscosity, tension),and interfacial pore geometry,(e.g., sizeanddistribution),fluid grain large extentdeterminedbyporous mediumproperties than mineraland dissolutiontrapping. trapping processes may occuronmuchsmallertimescales scale, hydrodynamicnisms at andcapillary thereservoir these trapping mecha - through geochemicalreactions. Of dissolutioninformation brines,andmineral trappingping, trap- free-phase through hydrodynamic gas andcapillary rock. Themajortrapping mechanismsincludestorageof of scCO of sequestration projectany subsurface are potentialleakage fluid displacementinporous media. Two mainissues for fer, tension, andwettability) influence greatly interfacial (suchascapillarity,and fluid-mineralinterfaces masstrans- twooptions,interactions at fluid-fluid thefirst beds. For andunminable coal reservoirs, depletedoilandgas fers, sealing efficiency andfree-phase scCO sealing efficiency interactions.Bothcaprock- byinterfacial are affected voirs Earth and SpaceSciences upscaling ofscCO scale isanessentialfirststeptofacilitate efficiency andtrapping.Researchatthepore studies ofprocessesrelatedtocaprock-sealing ducted pore-scaleexperimentalandnumerical and caprocksealingefficiency, wehavecon- processes tofieldapplications. supercritical carbondioxide(scCO To improveourunderstandingofsubsurface 2 2 out of the reservoir intothecaprock, thereservoir followed by outof (main drainage) and vice versa (imbibition)under (maindrainage)andviceversa 2 storage capacity in deep saline aquifers orreser storagecapacity indeepsalineaquifers 2 2 -brine displacement isconsidered indeepsalineaqui- 2 canmoveintothecap- 2 2 Sequestration trapping are toa 2 ) storage

2 storage, - - CO throats. We alsodemonstrated theability toinvestigate thin films andas bridgesacross around thepore pillars porous the wetting phaseas structure; andretention of thewettingphasewithin phases; wettingbehavior of betweenthetwoimmiscibleliquid tions; theinterfaces phaseidentityat specificloca- visualizationfacilitates of NileRedwithimagingmethods chromic dye. Theuseof between competingfluids at thepore scaleusingsolvato- anovelmethodtodistinguish In FY2010,wedeveloped conditions. reservoir high pressures andtemperatures found under typically Displacement experimentshave beenconductedunder displacement visualization usingsolvatochromic dyes. method tomodifymicromodel wettability andimprove angle data at supercritical conditions.We a developed and acapability tensionandcontact toobtaininterfacial two high-pressure systemsfor micromodel experimentation To address thesechallenges,wedesignedandconstructed experiments showedthat mineralprecipitation alongCO technique was publishedinWater ResourcesResearch. thisnovel scale underambientconditions. Adescriptionof dyes todistinguishbetweencompetingfluids at the pore in FY2010usingthesetypesof novel methoddeveloped for thispurpose wasbasedona phases. Thediscovery between twoimmiscibleliquid locations andinterfaces phasesaturations at visualizationfacilitates specific of chromic dye for useundersupercritical conditionsas it During FY2011,weidentifiedCumarin153asasolvato- injectionandsealing efficiency.plume marginsmay affect the following scientificchallenges: model experimentsat supercritical conditionsaddressing micro- In thisproject, wecompletedalargenumber of • • • 2 -related chemistry issuesat themicromodel scale.-related Our chemistry processes tolargerscales,suchasthefieldscale. identify methodsfor upscalingpore-scale displacement during pore-scale displacementprocesses. pressure, fluidsaturation, and permeabilityrelative therelationships betweencapillary of understanding studyandimprove andnumerically experimentally sis) play whenscCO during hydrodynamic trapping, ume), andwettability (includingcontactanglehystere- properties, displacementcharacteristics(rate andvol- identify roles that porous mediumproperties, fluid brine displacesscCO trapping when displaces brine, andduringcapillary 2 2

The micromodel experiments at high-pressure injection Also in FY 2011, we developed and tested new methods to conditions conducted in FY 2011 were conducted with change the wettability of initial water-wet micromodels to newly designed and constructed high-pressure cells oil wettability using treatment with silanes. The work pro- equipped with high quality syringe pumps for accurate and vides a benchmark evaluation of surface wettability altera- precise fluid injections at constant pressure or at constant tion by the silanization method, demonstrates a linear injection rate. Full control of either boundary pressure and correlation between the resulting measured air-water and injection rate is needed to understand observed phenom- oil-water contact angles, and was featured on the cover of ena and for modeling purposes. The design of the new the journal Langmuir. This methodology can be applied to pressure cell allows for fluid displacement visualization studies of immiscible fluid displacement across scales with using epifluorescent microscopy equipped with charge-cou- applications to important systems, including geologic car- pled device camera and/or hyperspectral cameras. The bon dioxide sequestration, enhanced oil recovery, and non- temperature of the system is controlled within 0.1°C up to aqueous phase liquid migration in the subsurface. 60°C. During FY 2012, displacement experiments were conducted We used the new high-pressure apparatus in FY 2011 to in intermediate and oil wet micromodels. Compared to dis- conduct CO2-water displacements in a pore network micro- placement of brine in water-wet models, it was clearly model with high and low permeability zones. Published in demonstrated that imbibing scCO2 was able to remove Energy & Fluids, results showed two distinctive fingering brine at much lower capillary pressures. This observation mechanisms: viscous and capillary. Other results published means that standard constitutive models for fluid retention in Environmental Science & Technology reveal that due to and movement, as typically used in continuum simulators the low viscosity ratio, unstable displacement occurred at for water-wet sediments, are not appropriate. Instead, all injection rates over two orders of magnitude. CO2 dis- specified relations accounting for wettability changes are placed water only in the high permeability zone at low needed for accurate predictions of multifluid flow for obtained for low injection rates. However, the continuum- porous media that are not water wet. scale model does not account for the unstable fingering In addition, a method was proposed to improve predictions processes observed at higher rates and hence tends to over- of scCO imbibition in water wet model. The new method, estimate CO saturations when viscous fingering occurs. 2 2 using decoupled capillary pressure – saturation and relative permeability – saturation relations, as opposed to the typically employed fully coupled method, considerable improved predictions of displacement behavior in homogeneous and heterogeneous micromodels. The new method was published in the Vadose Zone Journal. Finally, experiments were completed to study pore-scale

entrapment of scCO2 during displacement of brine. It was found that scale of entrapment occurred over multiple pores resulting in relative large disconnected blobs. Over time, the entrapped blobs shrink due to the relatively large

dissolved CO2 concentrations in brine.

scCO2 in a micromodel as a function of injection velocity. At low velocities, capillary fingers form. At very high velocities (e.g., near a well), displacement is in the form of viscous fingers. PN10056/2298

101 Earth and Space Sciences 102 PN12048/2449 chemical reactions that occurintheenvironment. micro-pores in nature of an understanding andfacilitate in provide fluidmovement data for thequantification of successful,ourproject could composition. If with varying phases crystal and acomplex,microscale assemblageof bytargeting natural sampleswithintactporeto begin space inmid-year FY2012.OurapproachThis project began was chemical data. solidsamplesthat include to provide a detailed 3D image of information) withx-ray images, tomography crystal-phase microscope images(whichdoincludecompositionaland electron tion. Thisproject typesof willcombinedifferent specific compositionsthat are responsible for the absorp- through thesamplebutdonotincludeinformation onthe eitherasrotatableimages canbeviewed solidsorslices dimensions usingmultiplex-ray collections.Theresultant absorptiontional computational reconstruction inthree of which maps absorption intwodimensions,withtheaddi- substances toabsorb x-rays analogous toamedicalx-ray, Tomography physical relies ability onthedifferential of scans) hasemergedasasignificanttool. produce (3Dimagesanalogous x-ray toCAT tomograms a natural ormanufactured material. Recently, theability to voidspacewithin within thesample;anddistributionof properties crystalline different regions; thedistributionof 2Dspatial relationships orsub- amongsampleparts tion of sample, includingvariations incomposition;thevisualiza- the thecompositionof includesrepresentations of imagery eye, for example. More afly’s specialized electron imageof ushave seenthescanning scientificsamples;manyof of scanning electron micrographs showingtheinvisibledetails most commonimagesare photomicrographs, particularly Thebestknownand andnewspapers. such asmagazines and scientificimagesare commoninthepopularpress domains. Microscale imagingisubiquitousinthesciences, microscale imagesacross multiplephysicalandchemical thisresearch istocombineandexamine of The objective James P. McKinley and ScanningElectron Microscopy Substrates UsingCombined X-rayTomography Microscale ReconstructionofBiogeochemical Earth and SpaceSciences valuable thanitsindividualparts. represent ascientifictoolthatismore techniques andacrossimagingscaleswould ability tocombineimagesacrossimaging complimentary imagesofsolidsamples.The scientific imagingtoolsisavailabletocollect scientific investigations,andafamilyof imagery isapowerfultoolusedinmany The collectionofdataintheformsample

intact sediment imageof tomography Phase contrastx-ray projecting chemicaldata into3Dx-ray space. exploring thecomputational methodsand limitations to scanning electron microscopy results andwillbegin of intotheassimilation In FY2013,wewillextendour efforts methods. usingconventionaltomographic observed compositional that areregimes not quantificationreadily of ing (seefigure); thismethodwillaidinthedetectionand usingphase-contrast imag- collected micro-scale tomograms datagraphic across collectionplatforms. We successfully tomo- system andare theincorporation investigating of theEMSLtomography optimizedresolution of achievable cial andpurpose-written software todeterminethe lution. We theresults usingcommer evaluating have begun using conventionalandphasecontrastmethodsat highreso - dissectedsamples of portions imagesof collect tomographic PhotonSourceAdvanced at to Argonne National Laboratory To data investigate collectionacross scales,weusedthe algorithms. using project-developed tional data willbepropagated solid intothetomographic viaelectron microscopyanalyzed methods;andthecomposi- were sectionedand collected;thesampleswillbephysically lengthscales.Tomographic thesamples different imagesof substancesat analogues usingmetal,mineral,andorganic synthesizing artificial sediment andsoilsamples,began We experience andabilities assembledintact evolve. initially multi-scaledata asour progress tomore complexvolumetric, late andadapt; thedata andto usingsoftware wedevelop Our approach istocollectx-ray andchemicaldata; toassimi-

- Non-destructive Challenge Measurements of Environmental Radiotracers Robert Runkle needed to detect new phenomenon such as dark matter. This project defined challenge measurements Additionally, the isotope 22Na has the potential for age that have the potential to answer key ques- dating of young ground waters. Despite PNNL’s CASCADES tions in environmental science that range array that presents a near-ideal tool for measuring 22 from the flow of water in ocean systems to signatures from Na, there are apparently no ongoing measurements of this isotope due to complexities of the stability of carbon in soils. The identified analyzing young groundwater. measurements guide investments into novel, state-of-the-art radiation detection Dual measurements of the radioisotopes 14C and 3H have technology. significant potential to inform the stability of carbon in soils. Deploying age-dating using these two isotopes provides two independent age dates that could result in an Environmental radioisotopes are commonly used as tracers approach to measure the stability of different fractions of of environmental processes. The first and most renowned organic matter extracted from different soils. Successful example is 14C, which was employed for dating the age of measurements will provide unparalleled insights into the organic matter. Since the late 1940s when this method was stability of different soil organic matter and contribute to developed, an expanding suite of environmental radioiso- new foundations for modeling soil carbon cycling processes. topes has been used as tracers of various processes, includ- We recommend ing groundwater movement (injection of water into aquifers developing the and the flow of water through aquifers), mixing of ocean necessary sam- water, and sediment deposition. More generally, radiotrac- ple preparation ers can be used to describe the role of environmental radio- and execute a isotopes, whose key value is that they are already present in demonstration the environment at useful concentrations and therefore do measurement. not affect the environmental processes under investigation. To be a valuable tool, the half-life of a radioisotope must be The emissions reasonably matched to the time horizon of the environmen- from the Fuku- tal process under study. shima nuclear power plant in Recent investments in new technology produced measure- Japan offer the Schematic of a detector concept capable of ment capabilities in PNNL’s Shallow Underground Labora- exploiting a suite of signatures, often emitted potential to simultaneously, from radioisotopes released tory with potential to impact environmental science measure the during the Fukushima event. Harvesting significantly through measuring environmental radioiso- migration of various signatures is key to increasing topes at extremely sensitive levels. The purpose of this study various radio- detection system sensitivity. was to define a set of challenge measurements to be pur- nuclides through environments, particularly through aquatic sued under PNNL’s ultra-sensitive nuclear measurements systems. Considerable interest will exist in coming years to research program. We initially examined a large set of measure trace concentrations of emissions that possess half- radioisotopes naturally present in the environment but lives in a range of under 10 years. Currently, technology is quickly focused on a narrow set of opportunities, including not sensitive enough to access some of these isotopes. We radioisotopes in the hydrosphere, age dating of young recommend pursuing novel detection technologies with groundwater, measurements of carbon stability in potential to exploit compound signatures, such as coinci- biomass, and radionuclides released from the dent emissions of various plutonium isotopes that allow Fukushima event in Japan. separation from other elements like americium. The isotope 39Ar presents the opportunity for age dating This project defined a set of key challenge measurements of subsurface water in the broad time frame from 50 to that can be targeted by novel instrumentation. Future proj- 1000 years. This analysis demonstrated that PNNL’s special- ects will develop the necessary chemical preparation tech- ized proportional counters offer the capability to measure nology and precision instrumentation required to execute reasonable sample sizes for ocean waters but not ice cores. the measurements with state-of-the-art technology in In addition, there is considerable demand in the fundamen- PNNL’s Shallow Underground Laboratory.

tal science community for assaying “ancient” argon that is PN12055/2456

103 Earth and Space Sciences 104 PN12059/2460 susceptible to large numerical errors. susceptible tolargenumerical errors. thishierarchy aswellpotentially of at levels different processes are twosituations that are bothubiquitous hand, strong compensation andcompetitionbetween complexinteractionspathways. Ontheother number of sub-components–leadstoalarge of levels contain several majorcomponentsthat again amodelinto afew of division GCMs–namely,one hand,thehierarchical structure of the climate research institutions.Itwasdeterminedthat onthe model codes from various of search combined withreview withaliterature started During FY2012,ourinvestigations in climate predictions. processes sensitivities aswellconstrainingartificial betweendifferent numerical andphysicalconsistency of ahigherlevel couplingmethods,weaimat achieving new of coupling inclimate models.Through thedevelopment reduce thenumerical associated errors withcomponent thisproject isto of anthropogenic Theobjective forcing. existinthesimulateduncertainties climate responses to current climate models,whichimpliesthat great structural amajorweaknessin details. Thesymptomshave revealed modelbehavior tonumerical of worrisome sensitivities studieshave shownstrong and By contrast,anumber of pling technique. orstandard the cou- practiceregarding established theory themodel,there isnotyet an other andwiththerest of ods. Althoughtheseprocesses witheach interactstrongly ral scalesthat represented are explicitly bynumerical meth- spatio-tempo- broadenswhich dramatically thespectrum of microphysics have beenincorporated intoglobalmodels, Complexprocesses andcloud suchasaerosol lifecycle ing. thecurrentmotivated emphasisonhigh-resolution model- regional climatefor changeallhave detailedinformation of computingpower, of growth need andthedecisionmakers’ mate research. Recentprogress inprocess studies,thefast used for more asafundamentaltoolincli- than50years atmospheric motionhave been equations of the evolution General circulation models(GCMs)that solve numerically Hui Wan Model Components Through Improved Interactionbetween Numerically RobustClimateSimulation Earth and SpaceSciences climate change. tools forunderstandingandpredicting will helptodevelopmorerobustandreliable spheric processesinnumericalmodels.It representing theinteractionsbetweenatmo- The projectwillprovidenewmethodsfor

studies carried out with this widely usedclimate model. studies carried outwiththiswidely feedback andaerosol-precipitation-climate interaction therefore expected tohave a majorimpactonthecloud are aerosol inCAM5.They indirect effect magnitude of essential information that determinethe about factors plannedfor FY2013willprovide theactivities cal artifacts, reducing numeri from Apart - thegeneralgoalevaluated. of andprecipitationlated cloudlifecycle formation willbe experimentstorepresenttivity theirinteractions.Thesimu- methods willbeusedinsensi- Different the budgetanalysis. Compensating andcompetingprocesses willbeidentifiedin thecloud/rain water budget inCAM5. of view quantitative meteorologicaldifferent conditions toprovide adetailed withthesinglecolumnmodelunder tions anddata analysis numerical simula- outaseriesof FY2013,wewillcarry For be usedasthemaintoolinnextsteps. whichwill andvalidatedbeen ported onPNNLcomputers, themodelhas of nical progress, asinglecolumnversion tech- 5.Intermsof munity Atmosphere Model(CAM),version used inthecloudmicrophysics parameterization intheCom- methods both thecouplingtechniqueandtimeintegration Throughout FY2012,weperformed atheoreticalof analysis incurrent uncertainty climate simulations. major source of GCMs, whichhave alsobeenshownbynumerous studiesasa componentsincurrent Theseare global thenewest to hours. ics andmicrophysics that happen at timescalesfrom seconds on competingandcompensating processes inclouddynam- thisproject. Thefollowing research willfocus of activities ogy model have andmethodol- enabled ustorefine thestrategy andresults theliterature from thetoy review Outcomes of andefficiency. between accuracy a posteriorierror control canprovide excellent trade-offs determinedsub-stepsizein combination with dynamically reduce Alternatively, theerrors. explicitmethodsusing equation cansignificantly theevolution for of linearpart implicitschemes system inthemodelandapplication of solutiontoasub- analytical numerical solution.Theuseof biasesinthe in globalclimate models,leadstosevere explicit timesteppingandlongstepsize, acommonchoice show that the sequentialoperator splittingcombinedwith couplingmethods.Results different performed to evaluate pete for available Numericalconvergencetestsare gas. formation particle cesses, whilecondensation com- andnew compensating strongly proand condensation- form apairof Inthistoy particles. problem, production new formation of phase viacondensation onpre-existing aswell particles by chemicalreactions phase,toliquid andconverted ingas speciesproduced agas equation of describes theevolution As thesecondstep, atoy problem isestablished that

Predicting the Feasibility of Geologic

Co-Sequestration of CO2, SOx and NOx Under a Broad Range of Conditions Diana H. Bacon To determine the geologic scenarios where co sequestra- Using mineralogy of deep saline systems to tion is practical, we began in FY 2010 by calculating the carbon sequestration and co-sequestration potential of co sequester NOx and SOx with CO2 could significantly lower the cost of retrofitting 150 minerals to use as a guide for future simulations. Geo- chemical simulations indicated that co-injection of CO and coal-fired power plants for carbon sequestra- 2 SO resulted in SO mineral sequestration. The amounts tion. We are developing a simulator that will 2 2 improve predictions of the impact of geologic sequestered were greater and more persistent in dolomite and basalt than in glauconitic sandstone. After 5,000 years, co sequestration of CO2, SOx, and NOx on 87 to 0 percent of the SO2 remains sequestered in the glau- target formation and caprock hydraulic prop- conitic sandstone. In all cases, co-injection of 1 percent SO erties under a broad range of mineralogical 2 with CO2 did not appreciably reduce the amount of CO2 and phase conditions. sequestered nor did it induce a measureable change in

porosity versus injection of CO2 alone. Due to limitations of previous simulators in handling the injection of multi-component and multi-phase mixtures, previous modeling studies of co-injection of SO2 with supercritical (sc) CO2 have either simplified hydrogeological systems or have already dissolved in an aqueous phase, rather than acted as constituents of the non-aqueous phase. The opposite (and equally unrealistic) extreme case is a scenario in which SO2 is limited by diffusion through a stationary scCO2 phase. A better method is to simulate the injection of sc/gas mixtures, but a combined component/ phase mixture approach is lacking, along with few modeling studies on the impacts of NOx co-sequestration. This project is improving predictions of the impact of geologic co-sequestration of CO2, SOx, and NOx on target formation and caprock hydraulic properties under a broad range of mineralogical and phase conditions. Numerical simulation of CO2, SOx, and NOx co-sequestration into deep geologic reservoirs requires modeling complex, coupled hydrologic, chemical, and thermal processes, including multi-component, multi-fluid flow and transport, and chemical interactions with aqueous fluids and rock miner- Co-injection of CO and H S in a basalt formation. H S is consumed by als. Our work will develop a three-phase variable compo- 2 2 2 mineral reactions more rapidly than CO . nent nonisothermal simulator capable of mimicking the 2 geological sequestration of a mixture of gases in deep In FY 2011, we focused on building numerical simulation saline and depleted oil and natural gas reservoirs. The capabilities for geologically sequestering a mixture of gases new simulator that incorporates these complex processes in deep saline and natural gas reservoirs. We developed the will be used for a survey of the effects of co-sequestration new two-phase, nonisothermal, multi-component opera- under a wide range of reservoir hydrogeochemical conditional mode for the PNNL-developed STOMP simulator. The tions. The results of the modeling survey will help identify new operational mode has been founded on the premise the range of reservoir conditions under which co-seques- that each phase will have variable compositions and com- tration is feasible and will guide future studies involving ponent numbers. Phase and component flexibility has micromodel pore-scale experiments with mineral sub- been handled numerically through implementation of the strates and mixtures of CO , SO , and NO . 2 x x Peng Robinson cubic equation of state with a two-phase PN10067/2309

105 Earth and Space Sciences 106 PN10067/2309 of hematite dissolutionexperimentsinCO of been extensively tested to ensure that calculations of inter testedtoensure that calculationsbeen extensively of simulator has models andexperimentaldata. Thenew and brineshave beencompared tohigher-order solubility inpure thesegases water for eachof the solubilityof dards andTesting database REFPROP9.Predictions of Stan- state intheNational for Instituteof allthesegases with theresults predicted byhigherorder equations of have beencomparedenthalpy andthermalconductivity state inSTOMPtion of for componentdensity, viscosity, and SO include 10 gases: CO include 10gases: In FY2012,thecomponentdatabase wasexpandedto applicable toco-sequestration inredbeds. in collaboration withtheNational CarbonInstituteinSpain NO co-contaminants may bespecifiedandisnotlimitedto state hasbeenwrittensothat ageneralnumber of of equation equilibria.Thenew defined through fugacity equilibriummodel,whereflash phasecompositionis such asO gases Hgandotherco-contaminantsaswellflue of Earth and SpaceSciences x andSO 2 . The predictions of the new Peng-Robinson equa- thenew . Thepredictions of 2 . We akinetic geochemicalmodel alsodeveloped x . This allows the simulation of co-sequestration . Thisallowsthesimulation of 2 , O 2 , N 2 , Ar, CH 4 , C 2 H 6 , C 2 , SO 3 H 8 2 , C andbrine 4 H 10 , H 2 S - of the code to co-sequestration of CO thecodetoco-sequestration of of andapplicationpared onthedevelopment ajournalarticle capacity due tochangesinporosity. Inaddition, wepre- component tosimulate theimpactonformation storage geochemical reactions withformation mineralstoeach to simulate well injection.We added thecapability tolink Mass andvolumetricsource termshave beenimplemented two-phase conditionsaswellduringphasechanges. mole fraction,andsalinityunderbothsingle-phase non-aqueous pressure, component non-condensable gas variables, includingtemperature, aqueousand primary fluxes are all correct for perturbationsnal andboundary of liquid phase. formulation, butwilladditionallyincludeanonaqueous state willalsobefounded onacubic equation of new of CO of be added toSTOMP for problems involvingtheinjection dolomite, basaltandredbeds. Additional capabilities will co-sequestration invariousformations suchassandstone, modeltofieldscalesimulations of thenew application of using globalarrays. We willprepare on ajournalarticle FY2013,wewillimplementaparallelsolutionscheme For 2 into partially depleted petroleum reservoirs. The depletedpetroleum reservoirs. intopartially 2 andH 2 S inbasalt. Quantitative Framework Development to Support Effective Monitoring for Metal and Radionuclide Contamination Michael J. Truex tion at a compliance location. Key findings of the simulations Many methods for subsurface remediation are summarized below. mitigate exposure risk but leave contami- • The ability to monitor a perturbation in moisture content nants in place. In these cases, especially for caused by an aqueous waste discharge is a function of the contaminants such as metals that do not relative size of the waste discharge and the vadose zone degrade and radionuclides that decay slowly, thickness. The quantitative results from these simulations have important implications in relation to the resolution effective monitoring approaches are neces- and sensitivity of vadose zone moisture monitoring. sary to verify continued remedy performance • Over time, recharge rate becomes the controlling factor in and human health/environment protection. contaminated moisture migration in the vadose zone. In most cases, except for large disposal volumes relative to Most traditional monitoring approaches focus on point-based the vadose zone size, the majority of contaminant flux to compliance monitoring and yield contaminant concentrations groundwater occurs under recharge-driven moisture flux at selected locations at snapshots in time. These data alone conditions due to dissipation of elevated moisture content may be insufficient to evaluate remedy performance. and dispersion of contaminants caused by unsaturated Recently, the need for expanded approaches to monitoring flow processes. Thus, monitoring of recharge conditions is design has been recognized to support effective inferences critical to evaluating contaminant flux to groundwater. about contaminant plume conditions and potential future • The impact of low permeability layers on water and contaminant movement. The use of numerical simulation contaminant migration is a function of moisture content analyses in conjunction with a conceptual model provides the magnitude change induced by waste discharge. For rela- ability to quantify the role of specific features and processes tively low moisture content change, low permeability in controlling contaminant fate and transport and in meeting layers have little impact on vertical migration; for higher, remediation goals. low permeability layers cause lateral spreading and/or The quantitative framework developed in this project perching conditions, where the contaminant migration addresses a current challenge in the monitoring technique rate to groundwater can be significantly controlled by development and selection: understanding how monitoring layers until moisture reverts to recharge-driven conditions. data provided by a candidate technique (e.g., type, sensitivity, Thus, specific monitoring of low-permeability layers is and resolution) can be used for interpreting contaminant fate primarily diagnostic under high, not low, waste-discharge and transport. The quantitative framework for monitoring conditions. design and interpretation is intended to support advance- Future efforts will focus on completing a technical report ment of lines-of-evidence systems-based monitoring describing implications of the quantitative simulation results approaches that provide a more cost-effective approach on vadose zone monitoring and related simulations of to long-term site management than conventional groundwater plume dynamics. Additional simulations will monitoring approaches. be conducted, including an assessment of groundwater- Numerical models were configured for conceptual surface water interfaces. model scenarios, including several types of imposed initial contamination conditions representative of key DOE issues. The focus in FY 2012 was on contaminants residing in the vadose zone that have potential to migrate downward and contaminate groundwater. These contaminants pose a potential long- term liability because vadose zone migration is slow and may generate groundwater issues for years after initial waste disposal. The goal of numeric simulations was to identify monitoring targets in the vadose zone with appropriate resolution for use in interpreting or Simulated water and solute flux from the vadose zone to groundwater. Even non-sorbing predicting contaminant mass discharge to contaminants migrate slower than the added water pulse and, as discharge volume

decreases, more of the contaminant discharge to groundwater occurs under recharge- PN12066/2467 groundwater and/or contaminant concentra- driven conditions. 107 Earth and Space Sciences 108 PN12071/2472 are bathed continuously withwater, dualdating of that soils Given bothfactors. protection, oracombination of bon recalcitrance iscontrolled bychemicalstructure, physical whethercar carboninsoil:itisuncertain organic of the fate to thisissue, however, of remain gaps inourunderstanding Despiterecenting theoverallglobalcarboncycle. attention exchange betweensoilandatmosphere crucial toconstrain- thecarbon trial soilcarbonpool,make understanding theterres- carbon content,combinedwiththesheersizeof global climate change. Heightenedconcernoveratmospheric to predict the long-term disposition of carbon insoils. to predict thelong-termdisposition of in turnimprove andourability current modelingefforts carbon stability insoilsand of our current understanding carbon pools.Thisinformation enhance wouldpowerfully persistent used toidentifypoolturnover rates inseemingly dent onimpactfrom location inthesoil).Tcouldalso be thechemicalform andmore depen- more independent of thechemicalstructure controlled bythenaturepletely or of and whetherthisisphysicalorchemicalprotection (i.e., com- protected carbonis T couldidentifyhowdynamicphysically co-measurement of co-measurement of simultaneous A primeexamplehighlightingthepowerof its native, complexformgas toasuitableform. simplified from converted asamplecanbequantitatively whenever instance, CO Studies showthat anincreasing atmospheric concentration of ous analysis of both14-carbon( of ous analysis Typical samplepreparation techniquespreclude simultane- James J.Moran Simultaneous A uniquecapability existsfor simultaneous withlowelectronegativity). expectedfrom gases sensitivity that (withhigher toavolatile asamplebeconverted only gas dual isotopemeasurements onsinglecompounds,requires andhousedat PNNLandprovides acapabilitydeveloped for Proportional Counter(ULBPC),whichis Low-Background separately.lyzed bytheUltra- analysis IncontrasttoAMS, graphite, forcing onthesamesample tobeana- Tanalysis requires conversion of an organic sampletoeitherCO anorganic requires of conversion Using SoilOrganicMatter Earth and SpaceSciences be storedinsoils. different carbon-containingmoleculescan global carboncyclingandtowhatextent better understandhowsoilcontributesto ing thesemethodstosoilsamples,wewill for age-datingorganiccompounds.Apply- We aredevelopinghigh-resolutionmethods 2 and other C- and N-based greenhouse gases are driving are gases driving andotherC-N-basedgreenhouse 14 C analysis by accelerator mass spectrometry (AMS) (AMS) byaccelerator massspectrometry C analysis 14 C and T is within soil organic carbon. carbon. C andTiswithinsoilorganic 14 C) and tritium (T). For C) andtritium(T).For 14 C and T Dating: A Case Study C andTDating:ACaseStudy 14 C and T analysis C andTanalysis 14 C and C and 2 or or

- method for making simultaneous moleculestabilitya insoilsis organic of scientific objective sequestration iscritical.Inaddition, centraltothe form of compoundsareorganic stable insoilandrepresent alasting which understanding (soil, atmosphere, aqueous,etc.), carbon storageandallocation systems betweendifferent capacity carbon.Withheightenedconcernover for organic atremendous storage Throughout theworld, soiloffers within thesoil. components shuttledbetweenthedifferent been actively in astable oldcarbonthat chemical molecule, has versus oldcarbonlocked and weshouldbeable todistinguishvery ing at specific bond-breaking molecules, withintheseorganic tope method,wewillhave enhancedtheresolution for look- through microbial orabiogenic activity. Withthisdual-iso- in thesoilandmay notcapture molecularprocessing thecarbon theageof reveals only materials, but thisbyitself where combustedtoCO samples are first proportional counter. We settledonatwo-phasedapproach, form that isappropriate for isotopemeasurement usinga material methods for intoachemical soilorganic converting research team’s mainfocus toidentifythemostsuitable Specifically, ourwork inFY2012commenced withour the initialisotopecontent. molecule suitable for dualisotopedating whilenotaltering compounds intoa asystemtoprocess organic to develop Wehigher sampleamountsandadditional processing. seek preparative stepsandmeasurements for eachdate, requiring molecules.Traditionalorganic approaches require separate these compoundsusing soilorganic measuringtheageof ously We amoremethodfor simultane- are sensitive developing with soil organic carbonassessedthe with soilorganic within thesoil. thecompounds thecompoundsor the location of ture of soilcarbonbasedeitheronthechemicalstruc- ent classesof differ wehopetodemonstrateshort, thevariable stability of richsoil. In compounds fromdifferent alocal, organically stepstoextract,identify, aseriesof will begin andseparate (as measuredbyanorthogonal method).Next,we specifically material thestarting themethaneproduct matches that of of above-described methodtoensure that the isotopecontent onassessing the During FY2013,weintendtofocus first carbon. ous dualisotopedates onsoilorganic making simultane- goal of method iscrucial tothisproject’s chemicalconversion Refiningthis kind of an isotopeanalysis. methane. Themethaneispurifiedandthendirectly used for these products thenusedasafeedstockfor synthesizing 14 C andTdating approaches. work Asnoted, previous 14 C andTagedates on 14 C age of soil organic soilorganic C ageof 2 andwater, with - Tank Residual Waste Stabilization to Reduce Contaminant Release

Kirk J. Cantrell

retrieval and closure operations. With more than 140 sin- We are developing and demonstrating a sci- gle-shell tanks waiting to be retrieved at Hanford alone, ence-based approach for stabilizing contam- total cost savings could be in excess of $1 billion. inant release from tank residual waste that With the aid of a thermodynamic solubility release model can be used to determine waste retrieval developed to predict future uranium release from tank endpoints that are risk-based rather than residuals, a simple and inexpensive method was designed based on an arbitrary volume goal. to render uranium insoluble within the residual waste through the addition of a small layer of lime (Ca[OH]2) on top of the tank residuals prior to closure. In addition, it is Recent work at PNNL has demonstrated that the develop- possible to immobilize reduced Tc(IV) and Cr(III) in insolu- ment of the scientifically based mechanistic release model ble iron oxide mineral phases through a co-precipitation is critical to understanding contaminant release from post- process that isolates reduced Tc(IV) and Cr(III) from the pore - 2- retrieved tank waste residuals. For example, studies about water and prevents their reoxidation to TcO4 and CrO4 . the uranium release from Hanford’s C-200 series tanks A third approach was to form stable phosphate bonded have shown that in spite of the 99% retrieval goal being ceramics to sequester contaminants. This type of process achieved, uranium release from the residual waste is still can be used as part of a risk-based closure strategy to dem- expected to be over three orders of magnitude above the onstrate that a greater volume of waste can be left in the maximum contaminant level (MCL). The focus of this proj- tanks while reducing actual overall risk levels. ect is to demonstrate that chemical additives can be used Three treatment methods were tested: a Ca(OH)2 addition, to stabilize the residual tank wastes post-retrieval, greatly a ferrous iron/goethite treatment method, and an in situ reducing contaminant release (in particular, technetium ceramicrete waste form based on chemically bonded phos- [Tc], chromium [Cr], and U). This innovative approach has phate ceramics. These approaches rely on chemical reduc- the potential to revolutionize Hanford’s tank retrieval pro- tion and the formation of highly insoluble forms of the cess by allowing larger volumes of residual waste to be left contaminants of concern. Results have demonstrated that in tanks, while providing an acceptably low level of risk for all three methods, the release of the three most signifi- with respect to contaminant release that is protective of cant contaminants of concern from tank residual wastes the environment as well as human health. Such an can be dramatically reduced, compared to contact with approach could enable DOE to realize significant cost simulated grout porewater. For uranium specifically, all savings ($10 million or more per tank) through streamlined three treatments methods reduced the leachable uranium concentrations well over three orders of magnitude (refer to the figure). In the case of uranium and technetium, the release concentrations were well below their respective MCLs for the wastes tested. For two of the three wastes tested, chromium release concentrations were also below the MCL. For the third waste sample, chromium release was above the MCL, but it was considerably reduced relative to the untreated waste. Work will be ongoing throughout the next fiscal year to identify the specific mechanisms responsible for reducing these contaminant releases through various characterization techniques and geochemical modeling that will be performed after testing is complete. Results of this work Experimental results for uranium, demonstrating the effectiveness will be disseminated in a peer-reviewed journal article of three tank waste stabilization methods relative to grout only in FY 2013. (simulated, with no solid Ca[OH]2 added). PN12073/2474

109 Earth and Space Sciences 110 PN12077/2478 to assessmicrobial response potentialasa robust, rapid, foam-based amendment components concentrations of We cultures oneidensis tograded exposed Shewanella value toremediation andmanagementdecisions. rapid metricsof low-cost,relevant, This work developed remediationsubsurface technologiesunderdevelopment. sediments toassesspotentialmicrobial ramifications of communitiesfrom andnative deepvadose zone organisms proof-of-concept, weusedatiered approach withmodel assess microbial contributionstoremediation outcomes.As toincorporateing framework moleculartechnologiesand atiered wederived test- development, phaseof In thefirst for microbial-remediation interactions. Microbial-remediation andassessments tools,strategies, assessmentframework: framework. agreed-upon efficient, tools willneedtobestandardized andutilizedwithinan molecular remedialmatures, activities developed poses of microbial communitiesforlation thepur- orcontrol of (DNA-RNA-Protein-System).molecular level Asthemanipu- andcontrolling microbialunderstanding process at the investmentshave gone towardgovernment agency at manyDOEwastesites.Consequently, DOEandother nized andstudiedfor itspotentialusefulnessorrelevance recog- hasbecomewidely Microbial lifeinthesubsurface Fred A.Tilton Applied Tools forRemediationandMonitoring Translating Microbial EcologyAdvancesinto Earth and SpaceSciences risk quantitatively. tive, defensiblebenchmarkstoassess addressed withtoolsthatprovidepredic- monitoring ofcontaminatedsitescouldbe costly “blind-spot”inmanagementand face remedialactivities.Thispotential communities thatcouldarisefromsubsur or adverseoutcomesdrivenbymicrobial Few regulatorystandardsassessbeneficial

- approaches into a standardized regulatory framework. approaches intoastandardized regulatory to incorporate noveltools(suchasPCRarrays) and microbiology. bywhich Thisstudyprovides aframework and in toxicological riskassessment,geochemical testing, concepts radionuclide stabilization of asanintegration wide applicability remediation beyondsubsurface and microbial communities. Thistestinghas subsurface of amendment pre-testing to“build-in”controlled responses tousemicrobial-amendmentopportunity metricsin phase, thisprojecttions. Initsfirst demonstrated anovel microbialscreen, andremediation andevaluate interac- toprioritize, without aframework andineffective costly in response toCAPBinourTier1screen wasnotobserved ~200uL/hr.columns at flow rates of Themarginaltoxic theamendment componentsin tions containingoneof depths (withandwithoutmicrobial stimulation) tosolu- we exposedsedimentsfrom the200Area deepvadosezone conditions mimicking remediation Consequently, activities. identified in2010)wouldelicitsimilar responses under microbialdeep vadosenative communities(DNA isolated/ We wantedtoaddressTier 2testing. whetherthe200Area risks, wemovedSDSandcocamidopropyl betaine(CAPB)to With significantmicrobial responses andpredicted relative that isadaptable tostate-of-the-art moleculartools. we established arapid, robust bench-top screening system remediation withwelldescribedandmalleable genetics, significancetometal/radionuclide of model organism foralbeit at fielddeployment. unrealistic Usinga levels constituents showedsignificantbactericidalproperties, concentrations found inproposed foam blends.Other growth rates at asignificantstimulation of tial, given (SDS)presented thegreatest poten- sodium dodecylsulfate response (death/growth) relative tocontrols; however, low-cost Tier1screen. Mostresulted inasignificant Site-specific remediation decisions willprove munities inthedeepvadosezone. malleable active, microbial com- presence of be informed bythesedata, whichhighlightthe remediation success.Future formulations will microbial responsesrelevant that couldimpact that manyamendmentcomponentsinduce whichdemonstrated evidence, of a congruency sedimentsbutatnative lowerrates. We observed fromgrowth SDSexposure wasreplicated in “sterilizing” deepvadosezonesediments.The tericidal properties may have marginalimpactin 200Areanative sediments.Consequently, itsbac- Uncertainty Quantification and Risk Assessment Pipeline for Carbon Sequestration

Guang Lin

coupling input uncertainty characterization, sampling gen- Motivated by the challenges of uncertainty eration, ensemble sampling simulation runs at supercom- in geological carbon sequestration modeling puter centers, analyzing ensemble results and visualizing the statistical results together. We developed a demonstra- and simulations, this project aims to tion package that can automatically generate ensembles develop an uncertainty quantification and given uncertainty distribution, conduct parallel ensemble risk assessment pipeline for carbon seques- runs at supercomputer centers, transfer simulation output tration and to improve the predictive mod- files back, and visualize uncertainty associated with carbon eling capability for geological carbon sequestration to illustrate the capability of our Velo-based sequestration simulators. workflow for the uncertainty quantification of carbon sequestration simulations. The developed workflow prototype laid a solid cornerstone for developing a UQ and risk With worldwide energy efficiency improvements underway, assessment pipeline for the carbon sequestration simula- fossil fuel usage – a major source of atmospheric emissions tions for testing performed during this current FY. of CO2 – will continue to provide the dominant portion of For a UQ study on the impact of uncertainty and heteroge- total energy in the world. CO2 capture and storage (CCS) in geological formations has become a promising option to neity of spatial permeability field on the CO2 plume radius achieve the goal of stabilization of atmospheric concentra- propagation, we employed the quasi-Monte Carlo and tions of greenhouse gases set at the United Nations Frame- probabilistic collocation methods in FY 2012 to generate work Convention on Climate Change. The key issue in the permeability sampling inputs for eSTOMP and conducted global deployment of geological CCS technology is gaining the ensemble runs using eSTOMP on large-scale supercom- the acceptance of regulators and the general public, a pro- puter centers. Additionally, we developed a response sur- cess that requires scientific risk assessment and cost esti- face method that can adaptively guide the sampling mation to understand the full implications. Predictive location and greatly reduce the computational cost in modeling of multi-scale and multi-physics subsurface sys- ensemble runs to study the relationship between the uncertainty and heterogeneity of spatial permeability field tems for CO2 geological sequestration requires accurate, data-driven characterization of the input uncertainties on the CO2 plume radius. These results were recorded and and understanding how they propagate across scales and submitted to the journal Mathematical Geosciences. Fur- alter the final solution. ther, we explored effects of data quality, data worth, and

redundancy of CO2 saturation data on injection reservoir In this project, we will leverage existing state-of-the-art characterization through PEST inversion, reporting our tools as much as possible to provide uncertainty quantifica- results to Greenhouse Gases: Science and Technology. Fur- tion (UQ) capabilities for the suite of simulation tools and ther outcomes from this past year included the creation of integrate them together as a UQ pipeline. A quantification research quality codes for building adaptive response sur- toolkit and workflow management software involve indi- face and developing efficient sampling method to reduce vidual uncertainty quantification tools, such as sampling the computational demand for uncertainty quantification. methods, statistical analysis methods, numerical integra- We also worked with another project, where we conducted tion routines, response surface evaluation, reduced-order the model comparison and uncertainty quantification of model development, parameter calibration/optimization, the SIMSEQ data sets. We presented these results collec- and workflow management capabilities. The toolkit and tively at both the American Geophysical Union Fall Meeting software will provide a reliable means of quantitatively and the 11th Annual Conference on Carbon Capture Utiliza- predicting the uncertainty of and assessing the risk to CO 2 tion & Sequestration. geological sequestration, information that is of significant value to DOE and to our nation. For FY 2013, we will perform the uncertainty analyses of

CO2 plume expansion in reservoirs and wellbore leakage During FY 2011, we developed a preliminary uncertainty in aquifers. We will continue to develop the UQ and risk quantification framework for studying the effect of reser- assessment workflow pipeline and integrate more UQ voir spatial heterogeneity on CO sequestration. Continuing 2 components into the pipeline. Finally, we will develop into FY 2012, we developed a prototype UQ and risk assess- efficient UQ and model calibration methods and software ment workflow pipeline based on Velo (A Knowledge Man- for results in top journal publications. agement Framework for Modeling and Simulation) by PN11077/2399 111 Earth and Space Sciences 112 PN10078/2320 iRESM component model, testing the importance of those of iRESM component model,testingtheimportance ineach uncertainty sourcesdetermines therelevant of arewhich alternatives compared) andthensystematically holder decisionsanddecision criteria(i.e., thebasesupon intractable. TheUCprocess identifies stake for- iRESMfirst theUCproblem becomes quickly the dimensionalityof onkeyvariables,context tofocus analysis theuncertainty inavacuum.Inaddition,important withoutadecision are uncertainties ular uncertainty—all potentially - anypartic of sion, itisimpossibletojudgetheimportance this process isthat withouttheframingprovided byadeci- decision.Thekeyinsightof aparticular in thecontextof characterization (UC)process predicated onconductingUC a10-stepuncertainty In FY2011,thisproject developed methods. manner usingavarietyof feasible,in acomputationally transparent, andpracticable vance, inmodelresults uncertainty must becharacterized the twinrequirements rele- for scientificfidelity andpolicy oradaptmitigate tosuchchangewithinregions. To meet policies to climate changeaswelltheconsequencesof theregionalconsequencesof understand decision makers insightsinaregionalcontextandtohelp policy-relevant the samesystems.Inaddition, iRESMisintended toprovide regionsnestedinnationalgraphic andglobalchangesin sub-national geo - occur withinandbetweenindividual coolingwaterture availability). affecting Theseinteractions electricityproduction, streamflow(e.g., quantity, tempera- systems modeling interactionsbetweenhumanandearth change andadaptation optionsby impactsandmitigation aimstoproject climateThe iRESMmodelingframework Richard H.Moss of Uncertainties Understanding theSources andConsequences Earth and SpaceSciences decision makers. uncertainty andcommunicatingresultsto experiments, andmethodsforanalyzing development, thedesignofnumerical Research resultsareinformingiRESMmodel Earth SystemModel(iRESM)framework. propagation) intheintegratedRegional (including uncertaintyquantificationand ble methodsforuncertaintycharacterization ing anddemonstratingrigoroustracta- mitigation decisions.Thisprojectisdevelop- results iscrucialforrobustadaptationand quences ofuncertaintyinregionalmodels Understanding thesourcesandconse- costs per square meter to the four sources of uncertainty uncertainty costs persquare metertothefour sources of emissions,andtheoperating gas sumption, greenhouse con- energy of thesensitivity RGCAM, theteamanalyzed Using iRESM’s assessment model, regional integrated sions scenario). inregional climateuncertainty response, anemis- given cases, andthree generalcirculation models(incorporating mental PanelonClimate A2andB1emissions Change’s ered infuture uncertainty emissionswiththeIntergovern- step undermodelskill andcompleteness,theteamconsid- economicoutputperworker. of growth rate of Asafirst tification, theteamconsideredgrowth population andthe model skill/model completenessissues.Underinputquan- inputquantification and of tion focused ontheeffects characteriza- charges, andcarbonprice).Theuncertainty square meter(includingequipmentdepreciation, energy emissions,andoperating costsper gas tion, greenhouse consump- the demonstration, theteamfocused onenergy decisioncriteria;forregion haddifferent thepurposes of stakeholderby 2050.Different inthe14-state groups use use by2050,andan80%reduction inbuilding energy (nochange),a50%reduction“frozen inenergy technology” equipment standards. Theteamconsidered three cases: codes andresidential andcommercial appliance and states buildingenergy mightimplementmore aggressive change. Theteamtookasitsmodeldecisionat what rate asameansfor adapting climate toormitigating ciency - effi codesandequipmentenergy buildingenergy roles of wereAmong otherissues,stakeholders interested inthe states. andfocusedsion makers on14upperMidwest and adaptation deci- questionsasked bysubnational-level intheliteraturefrom ananalysis about climate mitigation context for thedemonstration. information Theteamdrew to provide adecision needs inthepilotregion(USMidwest) iRESMresearchresults into stakeholder of decisionsupport theUCprocess.Theproject utilizedthe Demonstration of summarized below. are documentedand theseefforts Resultsof framework. propagation uncertainty forrequirements theiRESM of approachesproject andanalyzing toaddress run time inFY2011byanother theUCprocess developed stration of In FY2012,project research focused oninitiating ademon- these keyuncertainties. and theprocess propagation concludeswithuncertainty of withprobability distributions), are characterized(e.g., fully inthesekey sources theuncertainties Next,only analysis. sourcesuncertainty for thedecisioncriteriawithsensitivity described above. The team generated RGCAM results for all squared error in costs of residential and commercial 50 states from 2005 to 2095 following a 162 combination building services in Wisconsin in 2050 and 2095 for the full factorial design: 3 regional standards scenarios × 2 162 combinations (see table). Only 3 model terms are nec- national/emissions pathways × 3 population scenarios × essary to explain over 98 percent. Parallel ANOVA analyses 3 GDP growth rates × 3 climate models. The team analyzed for the energy consumption and carbon emissions criteria, this dataset with both full factorial and second-order frac- and for other states, show that this explanation of impact is tional factorial models, the latter featuring the 5 main unique to the policy experiment, decision criterion, region, effects and 10 second-order pairwise interactions. The frac- and time step but is easily replicable for other conditions tional factorial analyses show how the amount of total (models, experiments, and decision criteria) and can variance in each decision criterion (energy, carbon emis- quickly identify the sensitive inputs in each case. The ANO- sions, and cost) is explained by the various sources of VAs also show that though a specific RGCAM output can be uncertainty and thus indicate the important input uncer- summarized by a parsimonious factorial model, distinct tainties to analyze in an uncertainty propagation. For RGCAM outputs required distinct factorial models, with the example, the analyses of variance (ANOVA) of the fractional implication that more runs would be required to character- factorial models show that the 5 main effects and 10 sec- ize RGCAM uncertainty than one is lead to believe by the fit ond-order interactions explain over 99.9 percent of the of one fractional factorial model to one output. Run-time analysis. From the large number of simulations Wisconsin Wisconsin Residential Commercial required for a sufficient sample size, it is possible that runtime issues may arise in propagating uncertainties across FractFact Model Term 2050 2095 2050 2095 the coupled models of the framework. Initial work done Standards 7.0 31.4 18.5 40.9 during FY 2011 suggests that run times may be extremely Conc Pathway 7.6 33.4 12.2 25.7 long (running to months on a parallel processing platform with ~2000 cores available) for fully integrated runs of the Climate Model 0.0 0.0 0.0 0.0 complete iRESM suite. During FY 2012, the team investi- GDP 84.6 33.8 68.3 31.9 gated several response surface approximation methods to Population 0.0 0.0 0.1 0.1 create a mathematical model of the propagation of uncer- Standards: tainty. The team examined two powerful non-linear multi- 0.3 0.4 0.2 0.3 Conc Pathway variate regression approaches to optimize the response: Standards: Climate 0.0 0.0 0.0 0.0 artificial neural networks (ANN) and multivariate adaptive Model regression splines (MARS). With only a few training datasets, Standards: GDP 0.4 0.9 0.4 1.0 the methods establish the response relationships between Standards: Population 0.0 0.0 0.0 0.0 the dependent variables of interest (outputs) and multiple Conc Pathway: independent variables (inputs) and can be used in predict- 0.0 0.0 0.0 0.0 Climate Model ing outputs for expensive high-fidelity simulations or pre- Conc Pathway: GDP 0.1 0.0 0.1 0.1 dicting future events for a time-evolved system. Conc Pathway: 0.0 0.0 0.1 0.0 This project concludes in FY 2012. Work that builds on the Population results of this project will continue in FY 2013 other proj- Climate Model: GDP 0.0 0.0 0.0 0.0 ects, which will include work to broaden the sensitivity Climate 0.0 0.0 0.0 0.0 analysis to other decision criteria and experiments with Model:Population RGCAM, use sensitivity analysis to identify critical model GDP: Population 0.0 0.0 0.0 0.0 couplings with the other iRESM models, implement strate- All 15 Terms 100.0 99.9 99.9 100.0 gies for quantifying the uncertainties identified in the sen- Highlighted Terms 99.2 98.6 99.0 99.5 sitivity analyses, and develop the tools and methods needed to address uncertainty propagation across the Percent mean square error of residential building services iRESM framework. costs per m2 of floorspace explained by a fractional factorial model of predicted costs for Wisconsin. PN10078/2320

113 Earth and Space Sciences Energy Supply and Use A Multi-Layer Data-Driven Advanced Reasoning Tool for Smart Grid Integrated Information Systems Ning Lu ses to maintain relevance and effectiveness. With the added The objective of this research is to develop problem of stove-piping of data, the resultant increased a decision support tool – Multi-layer, Data- volume and complexity of data will only exacerbate human driven Advanced Reasoning Tool (M-DART) performance difficulties and lead to information overload – for improved power system operation. and degraded decision making. In view of these challenges, M-DART will investigate existing reasoning tools and apply two best performed tools to per- Today’s information management in power grid operation form higher-level reasoning (cross-correlate and examine tools mainly follow a stovepipe process. Each sensor net- different data sources to assess anomalies, infer root causes, work produces its own domain-specific data (e.g., phasor and anneal data into actionable information) to identify measurement unit [PMU], supervisory control and data highest priority problems for the human decision maker to acquisition [SCADA], and smart meter measurements), and address. By summarizing results obtained from different each data source is processed by domain-specific tool sets. reasoning methodologies, M-DART focuses on more accu- The outputs are then presented to grid operators responsi- rate early detection of emerging power system events and ble for making decisions. Isolated data processing mecha- identifies highest priority actions for the human decision nisms focusing on individual data domains may not be able maker to address. M-DART represents a significant advance- to reveal the root cause of failures or accidents and assess ment over today’s grid monitoring technologies that apply their impact to plan for action in a timely manner. As the offline analyses to derive model-based guidelines for online number of different data sources increases, the possible real-time operations and use isolated data processing action options increase exponentially; therefore, it is a mechanisms focusing on individual data domains. significant cognitive challenge to the human operator who is stressed to analyze these relationships and/or In FY 2012, the efforts were focused on the development of draw on relevant past experience. M-DART. Two prototypes were implemented, one of which characterized the PMU data quality. Another M-DART tool is the Columnar Hierarchical Auto-associative Memory Pro- cessing in Ontological Networks (CHAMPION), a semantic- graph-based domain-independent reasoning system. CHAMPION can integrate multiple data types and sources. The hierarchy of reasoning components divides the analysis into portions of a potentially large semantic graph to facilitate scalability. To illustrate the reasoning process, assume we have monitor data from network measurements from PMU, SCADA, or smart meter data records. CHAMPION process the input and deposit appropriately formatted data into Working Memory, which is implemented as a semantic graph. The CHAMPION processors comprise a hierarchy of reasoners that is configured by ontologies that determine the processors’ analysis of input data. Through this analysis, the reasoners construct and enlarge the semantic graph. Conceptual illustration of the M-DART architecture Also during FY 2012, we successfully collected 48-hour data from our industry partner, ISO-New England. The The current practice in the power grid operations commu- data analyzed includes 31 PMU channels within 6 PMU nity is to run massive offline analyses to derive model- groups, SCADA data and state estimator. One example of based guidelines for online real-time operations. However, analysis result is shown in the figure, which indicates the because power system components and configurations quality performance of PMU installed in different areas change in real time, it is hard for offline model-based analy- can vary significantly. Collectively, the results of our work PN12001/2402

115 Energy Supply and Use 116 PN12001/2402 coming duringthenextandfuture fiscal years.. isalsoforth- partners Additional coordination withindustry the required system. for interfaces/displays theintegrated exercise thereasoning systemandselecttoolscreate in-depththeappropriate further develop testscenariosto with selectedPNNLtools.Inaddition, wewillcontinue to reasoned advanced system andproceed withintegrating theCHAMPIONreasoning knowledgetodrive senting expert ontologiesrepre FY2013,wewillcontinue todevelop - For operations.PMU data quality for thegrid’s Future work willcontinue toexplore theimplication of during thisyear were presented at twoIEEEconferences. Energy Supply andUse

corruption ratio for regions. different PMUdata quality. resultAnalysis of Eachcolorrepresents PMUdata A Statistical State Prediction Methodology to Improve Reliability and Efficiency of Power System Operation Ning Zhou

Motivated by challenges of increasing uncertainty and variation brought by the high penetration of renewable generation to power system operations, we are developing a short-term prediction and uncertainty quantification methodology at grid level to enable well-informed proactive operations, which will in turn improve power grid operation efficiency and reliability.

A state estimator is an essential tool for power system operation. The estimated states reflect power system status and are the foundation of many essential operation decisions. Yet due to the delays from communications and computa- The comparison of prediction accuracy of NIS between our tions in the range of 0.5–5 minutes, current state estimators prediction method and the submitted values. can only provide power grid status in the past. Operation (see figure). We also explored several ensemble approaches decisions based on these past states can lower the reliability (including bagging and stacking, incremental learning) to and efficiency of power grid operation, especially when combine the prediction results from several prediction quick changes and large uncertainty are brought in by the method to improve the accuracy of the prediction. high penetration of renewable generation. Also in FY 2011, we developed a novel efficient uncertainty The objective of this study is to enable operators to make quantification algorithm (probabilistic collocation) and per- well-informed proactive operation decisions by providing formed an initial study on a sensitivity based adaptive sam- forecasted states. To achieve this goal, we are developing a pling method to propagate the uncertainty of using an IEEE short-term prediction and uncertainty quantification meth- 30-bus system. In FY 2012, the sensitivity based adaptive odology at grid level from three perspectives: 1) the predic- sampling method was refined and compared with Monte- tion method, which forecasts the values of power grid Carlo, linearization, and collocation methods based on the variables; 2) the prediction error (or uncertainty) quantifica- IEEE 30-bus model. To achieve same level of accuracy, the tion method, which gives the confidence interval of the fore- sensitivity based adaptive sampling method used 61 sam- cast; and 3) the uncertainty propagation method, which pling points, while the collocation method needs 301 sam- calculates the prediction errors associated with the derived pling points and the Monte Carlo method requires 10,000 variables or states. It is expected that the study results will sampling points. Errors using the linearization method were improve the situational awareness of an operator by provid- large compared to the other methods. The study shows the ing a comprehensive look-ahead view of power system, significant advantage of the sensitivity-based adaptive sam- which in turn allows time for proactive operations to pling method for reducing the computational complexity improve power grid operation efficiency and reliability. required to achieve a certain level of accuracy. In FY 2011, we developed five forecasting methods and The goal of FY 2013 work is to position PNNL at the leading applied them for predicting Net Interchange Schedule (NIS) edge of power system prediction through improving the per- from a set of field measurement data. In FY 2012, a new formance of the prediction methods and building demo set of field data measurement data of about one year were cases. Based on the work in FY 2012, we will enhance the obtained to test the applicability and generalization of the performance of prediction methods through improving the prediction methods. The prediction methods are applied ensemble approach for prediction. We also plan to integrate to the new data set and additional auxiliary inputs (e.g., the error quantification methods with uncertainty propaga- season, time of day, day of week, trends) are included to tion methods. A demonstration case will be built with inte- improve their performance. The comparison shows grated prediction and uncertainty quantification methods to improved prediction accuracy over the submitted NIS

show the prediction results using real-world data. PN11004/2326

117 Energy Supply and Use 118 PN12017/2418 energy storage. Modelpredictivecontrol (MPC)isusedto energy and generators windwhileconsideringequipmentlifeof of lem are tominimizefuelcostsandmaximizetheutilization operation theoptimization performance. prob- Thegoals of etration whilemaintainingsystemeconomicsandnormal tomaximizewindpen- storageanddieselgenerators energy Centralized control algorithms are proposed tocoordinate publications. four peer-reviewed total of Duringthe pastyear,microgrid. thiswork resulted ina thedistributionnetwork oranisolated ate of aportion managementsystems(DMS)inorder tooper bution energy distri- willalsobeat of theheart DERs.They penetration of operating apowersystemwithhigh of the technicalbarriers inthisproject arealgorithms aimedat developed removing applications for testingtheproposed Thecontrol strategies. R-DERsandC-DERsare usedasexample high penetration of configurationsMicrogrid grid-connected) with (isolated and duringstressed conditions. help thegrid coordinating DERsto decentralizedimmediate response of control algorithmsThe project for alsodevelops armingand resourcesfrom (R-DERs). localrenewable distributedenergy forstrategies andvariability C-DERstodealwithuncertainty optimalcontrol algorithms andcoordinationby developing renewable penetration. Thisproject addresses thisproblem additional variability from anduncertainty increasedof coordination andcontrol are strategies challengingbecause understressed conditions.However,also helpthegrid their large-scalerenewable generation. TheDERscan integrating tributed generation have thepotentialtocontributewhen storage, electricvehicles,anddis- demand response, energy resourcesControllable (C-DERs)suchas distributedenergy Marcelo A.Elizondo Resources forGridSupport Cooperative Control ofDistributedEnergy Energy Supply andUse optimally coordinatetheactionsofDERs. tralized controlstrategiesaredesignedto stressed conditions.Centralizedanddecen- objectives forgridsupportundernormaland resources (DER)toachievelocalandglobal dination strategiesfordistributedenergy This projectdevelopsoptimalcontrolcoor-

- conditions andneeds. resources toadapt theirresponse tocurrent system/resource the istoarmandmodifysettingsof Theobjective support. for contingency distributed storage)wasalsodeveloped for C-DERs(demandresponse, and distributed generators, ment. Aproof-of-concept decentralizedcontrol strategy computation, andadaptability toachangingenvironof - Q-learning approaches leadtoabettersolutioninterms MPCand Itwasshownthat thecombination of MPC. to approximate theoptimalterminalpenaltyfunctionfor improved. AparameterizedQ-learningalgorithm isdevised wasfurther theMPCbaseddispatch strategy performance of Illinoisat Urbana-Champaign,the of Florida andUniversity In anongoing of collaboration withboththeUniversity demand response. constraintsfor determine thereal-time powerandenergy isneeded,however, investigation the system.Further to andincreaseconventional generators thewindutilization in the reducewith theotherDERscanfurther thefuelcostsof formed todemonstrate DR that optimalcoordination of inthedispatchresponse problem. (DR) Studieswere per extendedbyincluding demand wasfurther control strategy windprofilesover different days.for different Theproposed and windpenetration levels wind utilization) undervarying a muchbetteroverallperformance (lowercostandhigher The results had indicated that theclosedloopMPCstrategy Simulation studieswere performed onthistestplatform. turbines andPVs),thedistributionnetwork equipment. R-DERs(wind age, demand response dieselgenerators, etc.), the IEEE34nodedistributionfeeder, includingC-DERs(stor DIgSILENT.simulation package using Itwasbuiltspecifically inacommercial powersystem operation was developed operating amicrogrid inislanded A simulation modelof resources. renewable energy high penetration of pensate for additional andvariability uncertainty causedby MPC isconsidered closedloopandhastheability tocom- using apre-computed, control law. off-line, Inthismanner, control action at eachstepiscomputedon-linerather than problem issolved usingfeedbackfrom thesystem.The step,approach afinitehorizonoptimalcontrol andat every solve alook-aheaddispatch optimization problem. Inthis - - Decision Support for Future Power Grid Organizations Gariann M. Gelston and-play evaluation of a variety of potential decision support This project examines human and organiza- visualization technologies. The framework equips operator tional influencers on distributed decision organizations with multiple communication options and records the information exchanged among decision makers making among power grid operators. After across organizations. The recorded communication data can investigating causes of communication errors, be analyzed to provide insight into how communication we designed a prototype decision-support relates to grid contingency response effectiveness. assessment tool as an analytical method to We presented our research at NERC meetings and conducted improve distributed decision performance of multiple site visits to major grid organizations. A critical individual operators, with contributions from feedback from industry experts highlighted a lack of existing a variety of grid organizations. economical training technologies and absence of a targeted tool to analyze operator communications. To address these issues, we will incorporate new features (such as voice chat) Rapid and effective response to unknown, undefined, or into the multi-organization decision-support and training unplanned system events is a core challenge for the grid framework in FY 2013 to improve operational capability. The operator community. This project analyzes how inter-organi- team also received segments of real operational communica- zational communication and coordination among grid enti- tion data from grid organizations that will be the basis of a ties is currently practiced, and examines distributed decision systematic communication analysis scheduled for FY 2013. making under uncertainty. To that end, we developed a pro- We will continue collecting and analyzing communication totype decision support assessment system to investigate data from grid organizations and share research findings rapid, adaptive decision making to target enhancements with industry partners. for improved grid reliability. While previous human factors research focused on improving individual operators’ situa- Our project links across expertise in visual analysis, human tion awareness and performance, we examined operational factors, organization behavior, and emergency response to communication and coordination to address grid contingen- develop an organization-driven view of distributed decision cies on an organizational level. support. This project will lead to an in-depth understanding of the key decisions, decision timescales, and challenges in Our goals included developing a role-based communication organizational communication and coordination; help framework to represent grid system decision makers and enable information sharing and better coordination; inform inter-organizational relationships; engaging the North the design of innovative decision support technology; and American Reliability Corporation (NERC) and grid organiza- develop decision support solutions to communication tions to advise tool development; developing an analytical challenges for the current and future grid. approach to understanding operator decision making; integrating with other projects by sharing industry contact and lessons learned; and developing publications to document technology development effort. Important project findings from FY 2011 suggest that for unknown, undefined, and unplanned contingencies, automation cannot fully replace operators’ decision making; the type, nature, and frequency of communication for distributed industry decision making is highly variable and complex, making a one-size-fits-all approach unlikely to represent this operational reality; and both systematic modeling and in-depth organizational studies are needed to enhance the empirical validity and generalizability of project outcomes. With these findings, the project team developed a prototype in FY 2012 for an interactive multi-organization decision-support assessment framework. The interface has the capacity to display images of a grid event rendered by a The multi-organization decision-support assessment framework, dispatcher training simulator and is designed to allow plug- powered by CORETM, has an uncomplicated user interface to facilitate

an interactive power grid distributed decision-making exercise. PN11020/2342

119 Energy Supply and Use 120 PN10022/2264 of modelswillbedeveloped: of thisproject, thefollowing set of Throughout thecourse methodological gap. processThis project tofill initiates this adevelopment uncertainties. climate changeandclimate policy context of detail for investmentdecisionsinthe of level a sufficient infrastructure andoperating new at expanding, siting, of way toaddress thetechnicalandeconomic complexities Current andintegrated toolsetsdonotprovide anefficient climate change. conditionsimposed orcausedby to sustainnew structure that willberesilient, affordable, andflexible infra- regional infrastructure todesignenergy planners willaddress framework foranalytical aknowledgegap This socioeconomicprocesses tobeevaluated. tions of hydrology, climate, biogeochemistry, andglobalimplica- regional feedbacksandinteractionsbetweenenergy, at theregional scaletoenable economics, andenergy climatethat withhydrology, integrate landsystems,socio- andtools systemsmodeling frameworks and testingnew isdeveloping The iRESMresearch activity anddevelopment Michael C.W. Kintner-Meyer Infrastructure SystemsFramework Development ofaRegionalEnergyand • • Energy Supply andUse System Model(iRESM). of thelargerintegratedRegionalEarth infrastructure. REIFisacriticalcomponent and adaptationstrategiesfortheenergy change impactsandassociatedmitigation of toolsandmodelstoanalyzeclimate Infrastructure Framework(REIF)consisting This projectisdevelopingaRegionalEnergy and economic growth andweatherand economicgrowth dependency, and andtheirresponse topopulation sectors, ufacturing building stock,electricvehicle transportation, theman- A modelfor electricitydemand(MELD)representing the delivery. structure service for fuelinputandenergy infra- transportation infrastructures aswellenergy resource availability, resources, and renewable energy land-use, resolved waterhighly geospatial data setsof Thismodelwillutilize fuel orthetransmissiongrid. infrastructures for zones anddistancestonecessary ability, andspatial considerations suchasexclusion locational marginalpricesfor energy, land-useavail- A powerplantsitingmodel(calledSITE)that considers

and natural resources, such as body of waterand natural orrivers. resources, suchasbodyof infrastructuretions from anylocation toexistingenergy complex criteria.Further,of itcomputesdistancesfunc- aset landgiven large-scale queriestoidentify suitability of and testedtheopen-source toolstoperform GISanalysis share withresearch community. Additionally, weenhanced Alldata canbemadepublicto use modelingactivities. infrastructure,the energy water management,andland- 1 kmand30×mresolution. ThisGISdatabase supports resources, land-use, population, andotherdata witha1× infrastructure, energy naturalries across 137data layers of vant GISdata intooneconsistentformat that allowsque- iRESMreleSITE model.We- completedtheconsolidation of each model: In FY2012,ourproject accomplishedthefollowing with other data. infrastructure,energy natural resources, landuse, population, and GISdata showingqueriesacrossiRESM relevant multiplelayers of • impact ontheinfrastructure canbeanalyzed. weathersuch that conditionsandtheirfull severe model willberesponsive toweather inducedloads, available powersystems.The for normaloperation of water assess whetherornotsufficient resources are model willbelinked toiRESM’s hydrological modelto loads.The powerplantstomeethourly dispatch of thetransmissionsystemtoperform aneconomic tics of powerplantsandthecharacteris - individual nomics of theeco- Itconsiders theelectricgrid. the operation of An electricityoperations model(EOM)that simulates fority demandshourly 1year. technologies.Thismodelwillcalculate theelectricgrid demandresponse enabledadvanced strategies bysmart We started and completed an evaluation of the SITE model EOM model. We initiated a new model development that based on 11 years (2000–2010) of power plant siting data. represents the operations of power plant dispatch of power The evaluation refined the complex exclusion and power plants within a bulk-power transmission network. Previ- plant siting suitability criteria based on actual power plant ously, we used a utility-grade production cost model called siting documentations. The results of the evaluation are PROMOD to perform this functionality for the iRESM effort. documented in a report and will be published as a peer- The new model development of the EOM is an open-source reviewed journal paper. model. We compared the cost-optimal dispatch of the new EOM model with that of PROMOD and evaluated the differ- MELD model. We completed a full evaluation of the MELD ences. We also analyzed the impacts of severe weather in model for the part of the 14-state Great-Lake pilot region. the summer time on the grid in the Great Lakes region to This included a parameter estimation that seeks the test the sensitivity of the electric grid to severe weather weights for 5 residential and 11 commercial building types conditions. The preliminary results suggest a high sensitiv- at 6 different sizes, 7 vintages, and 3 different HVAC sys- ity of the electric grid with respect to hot and humid condi- tems to match the combined load simulated by a building tions over longer periods of time. simulation program to the system load in that region. The outcome of the evaluation is a calibrated model of the During FY 2013, we will work toward accomplishing the building stock that matches both hourly electric loads as following tasks: well as total electric energy consumption in a region. This • evaluate and test coupling/interactions with other model represents the building stock as well as the load iRESM models behavior to with respect to weather. • integrate EOM with SITE and BEND/MELD, and complete testing. PN10022/2264

121 Energy Supply and Use 122 PN10028/2270 (which is desired for reversible Li-O (which isdesired for reversible Li-air orLi-oxygen (O Jason Zhang Development ofRechargeableLi/airBatteries only glymes could lead to significant formation of Li couldlead tosignificant formation glymes of only The results indicated that amongthestudied solvents, nuclear spectroscopy. magneticresonance (NMR) electrolytes, Li-O electrolytes, oxide radicalanion anditsrelated products. Usingthese which make super thesolventmoleculesunstable against inthemolecularmoietyascarbonategroup solvents, themhave –C(=O)– were notstudiedbecauseallof esters with carbonate solvents.Ketones, lactones,andcarboxylate are stable withLimetal)andionicliquids,compared sulfoxides, phosphates, nitrileswithlongchains(which aprotic solventswere orglymes, selectedfrom ethers mance anddischargeproducts inLi-O aprotic solventsonthedischargeperfor organic types of sixdifferent of studiedtheeffects During FY2012,wefirst storagesystems. themostpromising future energy of applications isone inwhichtherechargeable Li-airbattery EV thenextgeneration of for thelong-rangeoperation of EVs, higher energy density technologiesare required battery cansatisfy needof theshort-term technology Li ionbattery (PHEVs) andpure electricalvehicles(EVs).Whilethecurrent for worldwide plug-inhybridelectricalvehicles devices storage environmental friendlinessare criticalenergy density,energy/power good reliability, affordability, and surrounding theirstability. batteries withhigh Advanced there are stillmanyquestionsabout anduncertainties have reported significant rechargeability inLi-airbatteries, authors conventional Liionbatteries. Althoughseveral whichismore thanfivetimeshigher 2,790 Wh/kg, of exhibitatheoreticalbatteries. They specificenergy and rechargeable thanmost available primary cific energy spectrometry, andsolid-state tron chromatography-mass spectroscopy (XPS),gas (XRD),x-rayniques, includingx-ray photoelec- diffraction process usingvarioustech- were analyzed systematically on thecarbon-basedairelectrodes during thedischarge under similarconditions.Thereaction products formed Energy Supply andUse mechanism. batteries andunderstandtheirfundamental gate therechargeabilityoflithium(Li)-air The maingoalofthisprojectistoinvesti- 2 batteries were assembledand discharged 2 ) batteries have amuchhigherspe- 13 C-magic anglespinning 2 orLi-airbatteries) 2 batteries. The 2 O

- - glyme intherechargeableglyme Li-O studied, including the glymes. LessLi studied, includingtheglymes. Li amount of although asmall discharge capacity were found inLi-O work toward reversible andstablework Li-O toward reversible substrates that can (including solventandsalt)new systems electrolyte are required new efforts todevelop carbonbasedairelectrolyte. Therefore, of moreeffect ies. Thisinstability may alsoberelated tothecatalytic ever, Li sulfoxide. How- ionic liquid,and phate, nitrile, ated from phos- was alsogener of Li-O of charged products ent inthedis- LiF were alsopres- pares the cycling stability of Li-O stabilitypares of thecycling Thefigure investigated. been systematically belowcom- more or less unstable during the operation of Li-O more orlessunstable during theoperation of solventandLisaltareIt isnotedthat allcommonorganic best amongthestudiedLisalts. andLiTFSIwere the salts.LiTf theelectrolyte the rest of Journal of Power. Sources of Functional Materialsandthe Journal thisproject researchresults intwojournals,Advanced of ous solvents on the rechargeability of Li-O ous solventsontherechargeability of vari- of solvent.Theeffects (BDG)astheelectrolyte diglyme decomposition of LiBF decomposition of effect on the performance of Li-O ontheperformance of effect copy. saltswere Theelectrolyte found tohave adrastic products were byXRD, analyzed andNMRspectros- XPS, ucts inLi-O anions onthedischargeperformance anddischargeprod- Lisaltswithdifferent of We theeffects investigated further stability.best cycling vents, dimethylsulfoxide andacetonitrilesolventshowthe electrode sol- invarioussolvents.Amongtheinvestigated the Li-O charge process of during thedis- Li wasconfirmedbyXRDandXPS. borates respectively 2 O 2 wasidentifiedasthemajordischargeproduct for 2 2 batteries using all the six types of solventsystems batteries using allthesixtypesof CO 2 batteries, 3 2 and batteries. Seven Lisaltswere testedinTetra batteries. Seven - 2 O - 2

Comparison of the cycling stability of Li-O stability of thecycling Comparison of electrolyte withvarioussolvents. electrolyte reaction onglasscarbonelectrode inthe 4 andLiBOBtoform LiFandLi 2 batteries. Thedischarge 2 reaction onglasscarbon 2 batteries. Significant 2 2 CO batteries usingbutyl 2 batteries. The 3 andmuchhigher 2 batteries have 2 batter 2

- Development of Regional Agriculture-Land Use Models Allison M. Thomson To combine the relative strengths of these modeling sys- Our work developed an advanced capability tems, we are applying the agricultural model Environmen- for simulating climate impacts on agriculture tal Policy Integrated Climate (EPIC) in a geographic data framework for regional analyses to provide input to the and forestry to allow investigation of adap- Regional Global Change Assessment Model (RGCAM). The tation needs and mitigation potential of land initial phase of our approach focuses on a pilot region of management. With this information, decision the U.S. Midwest. We have developed the necessary data makers can use this information to pose and modeling capability to apply EPIC across the domain at questions about food security and green- 56 m spatial resolution and use the results, in combination house gas mitigation from regional-scale with detailed U.S. Department of Agriculture (USDA) data agriculture of interest. and remote sensing products, to inform GCAM. This process will enable integrated assessment to account for crop yield, bioenergy production, soil carbon, greenhouse gas emis- Climate change and food security are global issues that sions, and economic costs of different agriculture manage- are increasingly linked through decision making that takes ment practices. With this information, GCAM can simulate place across wide scales from on-farm management to the optimum mix of land management to achieve specific international climate negotiations. While still supplying climate mitigation or adaptation targets. sufficient food to feed the global population, understanding how agricultural systems respond to climate change In FY 2012, we completed model development required to through mitigation or adaptation requires a multi-sector initialize RGCAM with EPIC model results for the selected tool in a global economic framework. Integrated assess- crops, management systems and subregions in the Midwest ment models are one such tool; however, they are typically pilot area. The input data were assembled for the 2005 driven by historical aggregate statistics of production in base year from a combination of USDA inventory informa- combination with assumptions of future trends in agricul- tion and EPIC model simulations. EPIC was parameterized tural productivity. They are also incapable of exploring to simulate major crops under traditional management as agricultural management practices as adaptation or mitiga- well as with residue removal for bioenergy and no tillage tion strategies. Yet, there are agricultural models capable practices for soil carbon sequestration. In addition, dedi- of detailed biophysical modeling of farm management and cated bioenergy crops for cellulosic biofuels were also sim- climate impacts on crop yield, soil erosion, and carbon and ulated. The resulting GCAM-USMidwest-AgLU model has greenhouse gas emissions, although these are applied at been run for the 21st century under both reference case point scales incompatible with coarse resolution integrated and global climate mitigation policy assumptions. In addi- assessment modeling. tion, a downscaling procedure was developed to allow the resulting subregional GCAM output to be re-mapped to the EPIC 56 m resolution crop systems. The resulting findings indicate several important insights about the role of U.S. agriculture in global climate mitigation. Overall production in the region of interest increases, which reflects the high relative productivity of food and energy crops and the opportunity for crop residue removal for additional bioenergy resources. This capability will be expanded to incorporate climate impacts on crop yields using the EPIC model driven by climate projections to inform future production in GCAM. The work will also be expanded to the U.S. Gulf Coast region under a newly funded project to explore climate adaptation opportunities in agriculture, in addition to the Total crop production in the U.S. Midwest: no policy (left side bar mitigation considered in this current work. chart) and with the Climate Mitigation Policy (right chart). PN10029/2271

123 Energy Supply and Use 124 PN12034/2435 vent-based CO 550 MWsupercritical coalpowerplantequippedwithasol- forward osmosis(FO)water desalination systemwithina requirements a that wouldresult of from theintegration potentialcostsavings andchangesin energy assessment of thisproject wastoobtainapreliminary The objective with thewater intothefuture. willneedtogrow that they technologies that provide drought-stricken communities tion toanappreciation for itthat desalination canfacilitate from consump - aperception thatlic, itis ataxonenergy bythepub- could profoundly isviewed changetheway CCS to which the cost of CO to whichthecostof toestimate thedegree technical andeconomicparameters included gas. Theanalysis combustionflue from theplant’s commercial technologiesfor CO currentRecent studiesindicate that thedeployment of James R.Collett Desalination andCarbonDioxideSeparation Forward OsmosisSystemforIntegrated CO that canproduce valuable fresh water asabyproduct of system capture theprocess withthenew byintegrating solution. ThisCO gas is absorbed intoaconcentratedflue aqueousammonia developed this novel approach to reduce the cost of CO thisnovelapproach toreducedeveloped thecostof water desalination usingaforward osmosissystem.We integrated approachintegrated for post-combustionCO an thesechallengesbydeveloping propose tomeetbothof use present core scientificand technologicalchallenges. We emissionsmanagementandsustainable waterhouse gas increase limitfor- suchsystemsis35%.Thus,bothgreen produce bymore than80%,where DOE’s targetedcost theelectricitythey power plantswillincrease thecostof directed intoasolventspray tower, wherein theCO In ourenvisionedapproach,gas is combustionflue technology. cleanenergy anew and deployment of apath systems,bycharting for thedemonstration energy This work addressed acore PNNLmission,transforming fresh water madefrom brackish orseawater. groundwater producingplant were andsellinghighvalue also capable of draw solution for the separation of puredraw water solutionfor from theseparation brack- of asahigh-osmolarity FOmembrane modulestoserve of Energy Supply andUse fossil fuelpowerplants. seawater withCO desalination ofbrackishgroundwateror new forwardosmosistechnologiesforthe bon captureandstorage(CCS)byintegrating We aretransformingtheeconomicsofcar 2 capture. thisapproach Thesuccessfuldemonstration of 2 capture theCO systemthat removes 90%of 2 -rich solutionisthenrouted intoabank 2 capture could be mitigated if the capture if couldbemitigated 2 separationsystemsat 2 CCS at coalcombustion CCS 2 capture and 2 inthe

- 2

based CO for thesameplantequippedwithaconventional,amine- 684 MJ/sec estimate of whichissimilartoourderived sec, posed technology at anactualpowerplant. posed technology the foundation thepro- for pilot-scaledemonstration of tolay inthelaboratory experimentaldevelopment further ogy. desalination approach CCS/FO Theintegrated merits nomic penaltyimposedbycurrent carboncapture technol- the DOE’s reducing for theeco- achieving goal of strategy provides apromising framework desalination withinaCCS FO Our initialassessmentindicates of that theintegration ammonia-based CO required seawater for and simultaneousFOdesalination of to $32/ton.We alsoestimated that thecombinedheat duty water for each ton of capturedwater CO for eachtonof desalinated water perday fromto 26tonsof 2.5%TDSsea- couldproduce up equipped withourproposed technology Our findingsindicate that a550MWcoal-fired powerplant capture anddesalination. both asasorbentfor CO tem isthat thehighmolarityaqueousammoniacanserve tion unitoperations withinasolvent-basedCO FOdesalina- of theintegration of use. Thekeyadvantage orindustrial the plantorsoldfor municipal,agricultural, distillation that reboilers may beusedasmakeup water at pure watergas fromabsorber; and3)astreamflue the of site; 2)anammoniavapor stream that isreturned tothe pression carbonstorage toanunderground andtransport water per ton of CO water pertonof fresh lower salinity(0.5%TDS)couldgenerate 48tonsof Moreover, desalinating brackish withamuch groundwater perday (MGD). 91Mgallons fresh water outputstream of tion, thereby reducing capital and energy costs of CO tion, thereby costsof reducing capital andenergy high osmolaritysoluteinadraw solutionfor FOdesalina- CO avoided by upto57%,thereby loweringthecostpertonof at theplant CCS theexpenseof nue couldbeusedtooffset upto$220Mperyear. stream revenue a gross of - Thisreve desalinated fresh water for sellitsoutputof competitively could power plantequippedwithourproposed technology Atthisprice,recently estimated a tobe$3.59/1000gallons. at anew, osmosisplantinTexas 100MGDreverse was rent totalunitcostfor desalinating brackish groundwater cess that generates 1)a90%pure CO solution isthenrouted distillation intoaninnovative pro- into thehigherosmolaritydraw solution.Thediluteddraw themembrane moves from side of the salinefeedwater ules, therichsolventisdilutedfive-fold, aspure water orseawater.ish groundwater Withinthemembranemod- 2 emissions from the current baseline of $75/tondown emissionsfrom thecurrent baseline of 2 capture systeminstead. 2 2 , for yield of upto167MGD., for yieldof Thecur capture at theplantwouldbe736MJ/ 2 from flue gas andasavolatile, from flue 2 , thereby generating a 2 gas stream gas for com- 2 capture sys- 2

- Future Power Grid Control Paradigm Karan Kalsi

At the area level, This project proposes a novel control para- automatic genera- digm that enhances reliability and efficiency tion control is of the future power grid by exploiting the modified and value of new monitoring and information coordinated with decentralized technologies coupled with new business robust controllers needs and policy changes. This will lead to to restore tie line creating a more flexible and reliable U.S. power to their energy infrastructure. scheduled values. The efficacy of the proposed distrib- With the premise that transmission system loading will con- uted hierarchical tinue to increase and the trend will be toward maximum coordination con- utilization of transmission system close to thermal capacity, trol architecture is security and reliability will are crucial issues for the future demonstrated by power grid. The problem of power system security and reli- testing it on the ability is evident considering past events such as the 2003 IEEE 16-machine Northeast Blackout, which swept out a large geographical 68-bus test system area in 9 seconds, affected 50 million people, and caused and on the Enhanced transient stability of proposed ~ $10 billion dollar losses. With the advent of the smart robust control over current controls. 17-machine simpli- grid, the current power grid requires a new look at monitor- fied WECC system model. We demonstrated that proposed ing and operation capabilities and a new from-the-ground- control achieves enhanced transient stability over the con- up design of the control architecture. ventional control methodology under different types of con- Despite technological advancements, the monitoring and tingencies and system disturbances. control capabilities of the power grid are still based on the Another goal for FY 2012 was to study the interactions of conceptual design of the 1970s and 80s; i.e., current control typical remedial action schemes (RAS) with the distributed strategies rely on offline planning studies and lack good hierarchical controller and to identify a path for optimal real-time spatial and temporal coordination. While many coordination of these two control actions. We demonstrated ideas address specific control needs, this work reevaluates that the proposed robust control strategy can reduce the the power system control paradigm based on availability of burden on the RAS, which in turn reduces the system stress new technologies and proposes applicable solutions that due to tripping of generators. Finally, we proposed and will lead to a resilient, self-healing future power grid. Spe- tested a novel control framework to address the voltage sta- cifically, we offer a conceptual control architecture that bility issue, one of root causes of the 2003 blackout. The emphasizes real-time coordination of control schemes tem- principle for this new scheme coordinated different voltage porally and spatially to maximize operation margins or effi- resources (e.g., flexible load consumption, generator reac- ciency. This novel control paradigm minimizes impact and tive power, and capacitor switching) in an optimal, more exploits the potential of smart grid technologies, leading to efficient way using model predictive control. high penetration of clean, environmentally friendly renewable energy resources and non-conventional loads. In FY 2013, distributed dynamic state estimators will be designed for feedback in the distributed hierarchical robust In FY 2012, we proposed a novel modeling framework that controller designed in FY 2012 to drive the overall closed- represents multi-machine power systems as large-scale loop system. Also, the work in FY 2013 seeks to coordinate interconnected systems. A subsystem is defined as a collec- the voltage control schemes with robust controller and tion of substations (consisting of none, one, or more genera- remedial action schemes to provide multiple defense plans tor groups, transmission lines, and loads). Next, a for the future power system. The robustness of the distrib- distributed hierarchical controller design was proposed that uted hierarchical output feedback based coordinated con- facilitates frequency restoration and power balancing func- trol will be tested with a high penetration of wind and solar tions to be decoupled and implemented at different levels. in the system. The proposed distributed hierarchical output At the local level, robust decentralized generator controllers feedback based coordinated control will be validated on the are designed to improve transient stability and restore sys- IEEE Standard test systems and on a simplified WECC model. tem frequency after faults and disturbances in the system. PN12035/2436

125 Energy Supply and Use 126 PN10039/2281 shows top-level model definition categories of GS modeldefinition categories of shows top-level (GRDECL file) with keywords for material distribution.Inset form grid generated in fitted Petrel/Eclipse boundary Visualization of well intothe21 production energy fuelswillremain themainstayFossil of Mark D.Williams Numerical ModelDevelopment Geological SequestrationSoftware Suite: behavior of CO behavior of DOE’sgoal sequestration of research the istounderstand The unmineable coalseams,anddeepsalinereservoirs. in geologicformations, reservoirs, includingoilandgas development prototype. development ever, CO increased concentrations of theUnitedStates. How- for theprosperity andsecurityof isessential these fuelstoprovide clean,affordable energy plishing thisisthrough carbonsequestration: storingCO carbon emissionstotheatmosphere. accom- Oneway of systemsreduce the sions are expectedunlessenergy Energy Supply andUse natural environment. approach, giventheuncertaintiesof ation oftheefficacy ofthesequestration performance computingwillfacilitateevalu- models andoperationalcomplexityinhigh scale simulationswithmultipleconceptual sites. Theabilitytodevelopandrunmulti- for greenhousegassequestrationatspecific hensive, rapiddevelopmentofsimulations and datastructurestoprovideforcompre- ware systemwithaconsistentuserinterface (GS This GeologicalSequestrationSoftwareSuite 3 ) projectwilldevelopanintegratedsoft- 2 whenstored ingeologicformations. st century. Theability andavailability of 2 from carbonemis- 3 numerical model

developed focused theCO onsupporting developed toolwithlimitedfunctionalitywas model development sented totheproject team,andaprototype numerical andpre- During FY2010,adesignconceptwasdeveloped and distributions). material propertyresulting values reassignment of modifications and grid process (e.g., the development thenumerical modelspecification4) revising during theconceptual model;and response of torevisions models; 3)easeinregenerating numerical modelsin 2) buildingnumerical modelsfrom multipleconceptual scale numerical modelsfrom acommonconceptualmodel; cation independently. 1)buildingmulti - Thiswillfacilitate: the application managesdata associated witheachspecifi- and itstranslation toanumerical modelingframework, To thedistinctionbetween theconceptualmodel preserve properties, geochemicalreactions, andgeomechanics. andinitialconditions,source terms,material boundary fications grids, geology,for numerical modelsinclude collaborative geologicsequestration studies.Userspeci- of and temporal)numerical modelsthat avariety support and provides thecapability togenerate multiscale(spatial conceptual modelstoanumerical modelingframework conceptual models.Thisapplication willaidintranslating tools for building multiscalenumerical modelsfrom site algorithms and thisproject istodevelop The focus of acceptable.mentally formation andthat CO anunderground of will notimpairthegeologicintegrity This information iskeytomaking sure that sequestration or Petrel), a workflow the for exporting wasdeveloped Vision Earth models(e.g., stratigraphy basedonsolidearth complexhydroFor supporting - complex specification. toexternalfiles geometric objects,andlinkage for more selected andproperties specifiedbasedonsimpleindexing, themodeldomaincanbe the NetbeansIDE.Portions of are generated automatically and theJAXB within parsers XML schema.Also, theXMLformat supported, iswidely for reading andwritingthefiles basedonapre-defined selected for storingthemodeldefinitionwith JAXB parsers TheXMLformat was design,anddevelopment. typing, it provides for form rapid, proto userinterface graphical - Environment platform for Java because asadevelopment Development team adoptedtheNetbeansIntegrated operational mode of theSTOMP-COoperational modeof can belaunchedviaJava from webstart theGS platform independent(Linux, Windows,OSX)andprograms languagewasselectedbecausethecodeis programming 2 storageissecure andenviron- 2 simulator. TheJava 2 injectionbased 3 wiki. The numerical grid, assigning properties to the grid within nition XML file is initiated throughout the GUI on the client Petrel or EarthVision, and importing the zonation back machine and the resulting file is checked into the GS3 into the model definition process. CMS. A STOMP simulation using this input file can then be launched on a remote workstation with the simulation Progress during FY 2011 included completing the initial results, along with graphical output, stored back on the GS3 forms and user interface, XML schema/file structure, and CMS. The remote job launching functionality is part of the code for generating STOMP simulator input files for CO 2 GS3 platform-developed software. simulations with IJK-type zone specifications. These capabilities enabled creation and generation of the two STOMP Progress during FY 2012 included supporting additional short course CO2 test problems and a Mt. Simon Injection features of the STOMP-CO2 simulator, using boundary fitted case with a radially-symmetric grid. Other major accom- grids generated by Petrel/Eclipse (GRDECL files), enhance- plishments included linkage to an OpenGL visualization ments to the grid visualization, and accessing grid files toolkit (replacing the Java3D library implemented last year) stored on the Alfresco CMS. The additional STOMP-CO2 for viewing grids, model hydrostratigraphy, and selected supported features are the coupled well model and more nodes/elements; linkage to the updated GS3 Alfresco Con- options for solution controls, material properties, and out- tent Management System for storing numerical model defi- put variables. The GRDECL files are used to define a grid nition XML files and generated STOMP input files; added in Petrel/Eclipse, along with using optional keywords for functionality for Solid Earth Models (e.g., EarthVision and defining the spatial distribution of materials in the grid Petrel) for mapping material codes in files to simulation and spatial distributions of a specific property material properties; selection logic for portions of the finite (e.g., permeability or porosity). difference grid using polygons and lines; and tracking In collaboration with the GS3 platform project, prototype changes made to the model definition XML file based on development started in FY 2012 on a STOMP Input Advisor. user-supplied input. This tool will help guide conceptual/numerical model A software demonstration was developed that integrated development and collect data for building numerical the entire prototype GS3 system, including the features models or for conversion from another simulator. developed in this numerical model building project. The functionality of the GS3 numerical model development Related to this project, the demonstration showed the prototype tool was demonstrated by reproducing the process of creating/revising a numerical model definition benchmark problems published in the FY 2012 STOMP-CO and building STOMP simulator input files from which simu- 2 manual (except for the geochemical reaction network case lations were launched on remote workstations. This process which is not currently being supported by this GS3 tool). An involved in revising a numerical model definition XML file uncertainty quantification simulation software demonstra- stored on the GS3 platform included launching the Java tion was developed across the GS3 projects in FY 2012 that numerical model development software from the GS3 wiki applied the numerical model definition tool for building using Java webstart, checking out the model definition XML the initial simulation base case. This base case was then file from the platform CMS, making revisions to the model used for selecting model parameters to vary and launch- definition locally on the client machine through the GUI, ing/monitoring/visualizing the simulations using tools and then checking the revised XML file back into the CMS. developed by other GS3 projects. Building a STOMP simulator input file from the model defi- PN10039/2281

127 Energy Supply and Use 128 PN11037/2359 accomplish a battery withsuperiorcapacity,accomplish abattery wefocused capacity, wecanusehigh-capacity metal hydrides.To air cathode. Because thecathode inaMHABdoesnotlimit nickel hydroxide cathode with alightweightbi-functional efficiency. Our approach isto replace the capacity-limiting lowers capacity andeffectively limitstheenergy severely nickel oxide cathode. Thisheavy metaloxide cathode hasametalhydride anodeanda hydride (NiMH)battery Within theresearch context,thecommercial nickel metal -life technology. MHAB, wewillobtainahighercapacity, robust, longer the Byadvancing storageare limitingfactors. energy existinghigh-capacity tions, thecostandperformance of applica- grid-scale storage. For large-scaleenergy cost of common separator materials that reduce willgreatly the while incorporating with electrolyte anaqueousalkaline bi-functionalaircathode,materials withanefficient combines novel,highcapacity metalhydrideanode availablecally materials andprocesses. Ourapproach MHAB that canbemanufactured usinglow-cost,domesti- long-life,We high-capacity areversible, are developing transfer duringcycling. hydride anodematerials mass that enable highrates of high-capacity metal istodevelop to realize thistechnology storageapplications.and costtargetsfor grid-scale Thekey storageperformance significant impactonmeetingenergy ogy. wouldhave a technology Thisbreakthrough inbattery (MHAB),ahighercapacity,battery robust, long-life technol- ametalhydrideair of One proposal istheadvancement efficiency. lowers capacity andeffectively limits theenergy oxide cathode. Thisheavy metaloxide cathode severely hasametalhydrideanodeandnickel hydride battery storage isalimitingfactor. Thecommercial nickel metal existinghigh-capacity energy tions, theperformance of applica- grid-scale For thepowergrid. cations tosupply inlarge-scale appli- more efficiently a needtostore energy With theincreasing demandsonrenewable energy, there is Ewa C.E.Rönnebro Air Battery High-Capacity ReversibleMetalHydride Energy Supply andUse market. for thefuturerenewableenergystorage energy storagewithefficient performance approach tolow-cost,large-scalestationary project aimsatprovidinganewhigh-impact Envisioning asustainabletechnology, this to exceed 400mAh/g. themetalhydrideairbattery obtain superiorcapacity of coatings withthetunedmetalhydridematerials to new with nocapacity loss.Inthefuture, wewillcombineour hydride anodematerials, obtaininggreater than40cycles metal quasicrystal Missouri-St. Louis,weexplored new Additionally, of incollaboration withtheUniversity to uncoated materials. ourmetalhydridesmaterials compared higher capacity of approximately andrelatively 50cycles lifeof longer cycle layer-by-layer silicabasedcoating approach that enabled selected. Basedonasol-gelsynthesisroute, weinventeda life),andpromising candidatesand cycle were down- through withrespect totheirperformance (i.e., capacity anodeconceptswerecapacity materials screened andnew cathode, similartoaNiMHbattery. DuringFY2012,high- are basedonametalhydrideanodeandnickel hydroxide for high-capacity metalhydridematerials. Thesetestcells screen nineelectrochemical testcellstoeffectively up of and manufactured ananodescreening capability made mentionedapproaches.ously InFY2011,wedesigned capacity- metalhydridematerials byusingthetwoprevi high- advanced demonstrating thesuperiorperformance of therefore, aswell onanodematerials development, water-based electrolyte. Aspecific focus inFY2012was, the high-capacity metalhydridematerial corrodes inthe isthat The mainchallengetorealize theMHABtechnology candidates withgreater than400mAh/g. identifying somepromising high-capacity metalhydride and metal hydrideanodematerial obtaining50cycles having acoated demonstrated improved performance of concepts. DuringFY2012,ourkey accomplishmentswere anode materials MHAB; andtodown-selectadvanced the asmall-scaleprototype of to showproof-of-concept of tify promising high-performing metalhydridecandidates; were todesignandfabricate ascreening capability toiden- When theproject commencedinFY2011,ourmaingoals liferelative tothestate-of-the-art. and cycle anovel anodematerial withsuperiorcapacitydeliver with corrosion-resistant metals.Theultimate goal isto thealloy coating onthemetalhydrideandtuningof tive sidered twomainapproaches: aprotec - the application of corrosion-resistant anodematerial, duringwhichwecon- the through candidates. Thekeytosuccessisdeveloping screening anodematerials andefficiently on developing Modeling of Distributed Energy Resources in the Smart Grid Shuai Lu response This project will develop aggregated models devices). The for distributed energy resources (DER) in the trained DBN will smart grid, particularly demand response. help to forecast the availability of resources and To build a more reliable, efficient, sustainable power estimate the system, the transformation into the smart grid must distribution of emphasize distributed resources, featured by local device parame- controls and responses at each device to real-time price ters. In the sec- signal, system frequency, and system voltage or their ond approach, derivatives. Distributed energy resources will reduce the ST model aggre- Result of aggregated state transition model and peak capacity requirement on generation and transmission gates the TCL detailed simulation (GridLAB-D) of 10,000 home assets, improve survivability of subsystems during severe devices being models: A/C load response under sequential set disturbances at the interconnection level, and solve the modeled into a point changes (T set = 75 -> 74 -> 75°F). intermittency issue associated with high penetration of few states of temperature. The transition of devices wind and solar energy. between these states caused by either environmental parameters or control actions is simulated and transition This project is developing aggregated models that will rates are calculated. Based on the number of devices in represent the dynamic or steady-state characteristics of each state, the total power consumption of the devices can a large population of DERs in the design, planning, and be estimated. Simulation results of the aggregated models operational studies of the smart grid. The work is very were compared with that from the detailed model (Grid- challenging in that it requires multi-disciplinary knowledge LAB-D) for accuracy. of various areas, including power systems, power electronics, control theory, and computational and statistical tech- In FY 2012, we also performed a week of experiment on niques. Developing DER aggregated models and making an electric water heater in a model home on the PNNL them available to scientists and engineers will remove campus, in which data were collected to validate the indi- technical barriers, reduce the repetitive work for each DER- vidual water heater model (the basis of the aggregated related study, and speed up design and implementation. model). Sensitivity studies and uncertainty quantification on HVAC model parameters were also performed to deter- In FY 2012, this project made significant progress in the mine the uncertainty range of simulation results and following activities: aggregated HVAC and electric water identify important parameters for the construction of heater modeling, electric water heater field experiment the model. To assist the design of demand response for model validation, sensitivity study and uncertainty controllers, properties of HVAC response to control signals quantification on the parameters of the HVAC model, were characterized. Response speed and magnitude can characterization of HVAC demand response performance, be derived from the study results to determine suitable dynamic Bayesian network (DBN) model for HVAC load power grid services that can be provided by HVAC demand forecast, and multi-scale modeling of transmission response. Also, the methodology for multi-time scale simu- systems with demand response. lations was developed to study the transmission system DBN and state transition (ST) model were the two steady-state and dynamic behavior with demand response approaches initiated in FY 2011 to derive the aggregated in the same simulator. models for thermostatically controlled loads (TCL), such as For FY 2013, calibration techniques will be applied to HVAC units and water heaters. In the first approach, DBN aggregated HVAC and electric water heater models to used training data obtained from a simulation of detailed obtain the correct set of parameter values for a population models from GridLAB-D to obtain appropriate model struc- of devices. Methodology for developing aggregated ture and parameters. The DBN model is then adjusted demand response models will be integrated, which using expert knowledge and test data. In real-world appli- includes the DBN and ST models, uncertainty quantifica- cations, the training data are expected to derive from his- tion, and model calibration. torical data accumulated from operations (e.g., a load aggregation company operating a large amount of demand PN11053/2375 129 Energy Supply and Use 130 PN10063/2305 gate the effect of thisCO of theeffect gate tocarboncapture andsequestrationbeen devoted tomiti- methods of O methods of alternate aimedat determiningif studies andanalyses unit operation. Finally, thisproject smallpaper supported based powerplantswhere CO carbondioxide,for from thecapture coal- primarily of portfolio novel,of cost-effective, material-based approaches a thisproject istodevelop selection. Theoverallgoal of house gas CO house gas impact theworld- green climate through generation of large coal-basedplants,hasthepotentialtoadversely fossil fuelsfor with powergeneration, especially The useof David L.King Novel CarbonCapture Materials membrane for pre-combustion CO vided support forvided support aCO initialwork aimedat developing oping numerous We other opportunities. were alsopro- - coalviableidentified, making anddevel oxycombustion of the cost of CO the costof (as opposedtostorage),andfor thisreason minimizing driven pH swing CO pHswing driven via ionicliquid-typesolvents(COBOLs),electrochemically removal of CO removal of identifying possiblelargecostsaving through theco- The project produced asignificantpaper studyaimed at In FY 2012, we focused development efforts for CO efforts In FY2012,wefocused development carbon capture andsequestration relative toMEA. mate goal withinDOEisa70%reduction inthecostof mono-ethanolamine (MEA)solventtechnology. Theulti- standard iscomparedtheindustry against technology towardmately commercialization). comparison,each For ulti- (advancing tothenextlevel toadvance technology technical requirements that mustbemetin order for a identifies Thisanalysis analysis. (TRL)gate Readiness Level aso-calledTechnical framework, supplied toananalysis capabilitieslytical togenerate measurements whichcanbe ana- of Within this,anadditional goal isthedevelopment aspect is the actual selective capture and release of CO captureaspect istheactualselective andrelease of increase thediscrimination for CO materials modifiedtomakehydrophobic to theirsurfaces Energy Supply andUse mass gasification. and pre-combustionsuchfromcoal-orbio- combustion inpulverizedcoalpowerplants, from post-combustionprocessessuchascoal for efficient, costeffective captureofCO ing candidatematerials(solidsandliquids) This projectsupportsinvestigationsoflead- 2 2 2 . Major efforts invariouslaboratories. Majorefforts have 2 /N recovery is a significant driver in technology intechnology isasignificantdriver recovery and other gaseous impuritiesinasingle andothergaseous 2 separations (besides cryogenic) couldbe separations (besidescryogenic) 2 capture, framework andmetalorganic 2 generation. The most costly generation. Themostcostly 2 ispresentgas. intheflue 2 2 adsorption overwater. capture. 2 capture 2

gations fromgations 2011(CO twoinvesti- for 2012,wecontinuedSpecifically of support in preferential H absorption of thatappears thismaterial there ispotentialapplication of for someadditional physicaladsorption.is evidence It CO rized below. projects. short Additionaltiated results several are summa- CO (NETL)for Technology from Laboratory theNational Energy material, specifically material, CO for specifically application in removal of tion). Additional were opportunities seenfor thistypeof include: conclusionsfrom thestudy ethanolamine (MEA).High-level CO Attention hasfocused onethyldiethanolamine(EDEA).For is at atmospheric pressure inanoxidizing environment. reducing environment, whereas post-combustioncapture tions areapplied pressures at generally elevated andina isthelatter difference applica - coal orbiomass.Aprimary 1 equivalent H 1 equivalent and pH swing CO and pHswing Integrated Emissions.ImplementingCO Integrated betterthantheindustrialstandard Selexol. far H toward active separations,natural andEDEAishighly gas toward methaneorhigherhydrocarbons, for important and physicalabsorption. nouptake MDEAshowsvirtually chemical overall data are consistentwithacombination of uptake pressure. onpartial strong dependenceof However, the toactlikevent. Itappears aphysicalsolventintermsof asastand-alonesol- solution buthasnotbeenevaluated H CO technologies for NO thebestavailable control baseline systemconsistingof a concepts that couldimprove thecost-performance of control togenerate andassessnew This involvedefforts couldbemoreemissions simultaneously cost-effective. whether technologiesthat couldcontrol multipleregulated holistic lookat allemissionscontrol forms todetermine electricity.cost of Thissub-project waschargedtotake a adds significantcapital andoperating burdens ontothe pulverizedcoalpowerplants emissions control trainof 2 2 S, either from natural gas sources or pre-combustion of eitherfrom sources natural gas orpre-combustionS, of S. Evidencefor strong interactionisseenatS. upto 2 2 2 2 , EDEAhasbeentestedfor CO capture combus- from coal-basedpowerplants(post solventabsorption/stripping systembasedonmono - . Beginning in FY 2012, this project gained funding inFY2012,thisproject gained BOLs. Beginning 1. Equipment andoperating costsfor CO cost effectively as possible as the primary driver. aspossible astheprimary cost effectively low-cost application willprobably control CO coststocontrol othercontaminants. The dwarf 2 S perEDA, there whileabove that level 2 capture inaqueousmedium),andweini- x , SO 2 -binding organic liquids[CO -binding organic x , andparticulate control, plusa 2 S overCO 2 capture asanaqueous 2 2 capture ontothe factor of ~10), of (afactor 2 control 2 BOLs] 2 2 and as Control of other emissions will be opportunistic capacity. Specifically, this approach incorporates hydropho-

to an efficient CO2 system. Keeping individual bic moieties on the external surface of the MOFs via physi-

controls for NOx, SOx, and particulate may cal adsorption, with the aim of increasing surface remain the best option. hydrophobicity in the porous materials without the need 2. The energy demands of the baseline MEA process of introducing bulky functionalities inside the pore. Sup- are so large that implementation will reduce the port for this work began late in FY 2012. Preliminary work electrical output nearly 30%. The plant will have with MgDOBDC and the surfactant Pluronic P123 indicate to be significantly expanded or de-rated. Imple- that a reduction of H2O adsorption relative CO2 adsorption is promising. mentation costs of CO2 are effectively the process equipment for the capture system and capital Oxygen-Nitrogen Separation Using Perfluorinated Sol- and operating costs to expand the plant. vents. Two short studies examined new ideas for oxygen-

3. Solvent based CO2 absorption and capture employ nitrogen separation. The specific target would apply this relatively standard equipment (i.e., towers with separation to coal combustion (oxy-combustion), in which

engineered internals), but size pushes the limits case the primary combustion products CO2 and H2O would of equipment designs. New approaches to “mega- not be diluted by N2 gas introduced from air. sized” equipment or process packages need to be The first project examined using a perfluorocarbon such as developed to reduce costs. perfluorodecalin as a solvent to extract 2O from air. Such Electrochemically Driven pH Swing. An electrochemical materials have been applied in artificial blood due to their aqueous-based method for capturing CO2 from flue gas capacity for O2. However, using thermodynamic data, based streams was developed and tested for proof-of-principle on a surprisingly low preference of the fluorocarbon sol- at the laboratory scale at the end of FY 2010. For FY 2011, vents for oxygen over nitrogen and the preferential solubil- a preliminary analysis of the concept was carried out as ity of CO2 over oxygen, the concept as proposed would applied to a large scale power plant. The feasibility of using suffer both from ingress of nitrogen to the combustion an electrochemically generated pH gradient to capture CO2 loop through the solvent and from loss of CO2 through the from a power plant was demonstrated. Through an engi- solvent to the atmosphere. Based on data, the fluorinated neering analysis, the process could compete favorably with solvents would not be suitable, and a solvent with suitable the current MEA technology, provided that a reasonable properties is not known to exist. It was decided that no fur- cell over voltage could be attained. Contributions to the ther investigation was warranted. overpotential included the IR drop in solution, flooding The second project explored the use of magnetic fields to of gas diffusion electrodes, and kinetics of H2 oxidation in neutral solution. Ultimately, the latter proved to be a limit- separate O2 from air, because O2 is paramagnetic and N2 is ing factor, but electrode flooding may still occur. Prelimi- not. Calculations were made assuming linear field gradi- nary studies suggest that new catalyst materials may ents. The derived equation shows that the density differ- require development to overcome this problem. Most ence is enhanced by larger gradients, wider containers, work in this area reported in the literature focused on lower temperatures, and higher fields. These results can either acidic or basic (but not on near neutral) solutions. be applied to “real world systems” to determine feasibility. We conclude that future work should focus on the CO2 Membrane Development. This project investigates the mechanism of oxidation in neutral solutions, with feasibility of developing a CO2 membrane at warm temper- careful examination of the catalyst type. atures (300–500°C) for its removal from syngas at elevated

Development of Hydrophobic Molecular Organic Frame- pressure. The specific composition is a Na2Mg(CO3)2 double work (MOF) Solid Adsorbents. MOFs are high surface area salt that shows reproducible uptake and release of CO2 at porous materials that combine organic and inorganic com- around 400°C and partial pressures above 4 bar using a ponents in a variety of ways to generate multiple struc- pressure swing operation. The physical properties of the materials are such that it has proven difficult to convert tures. For some time, MOFs have been of interest for CO2 capture because of their high surface area and porosity; to a free-standing membrane. The conclusion of this work however, they absorb water competitively, which decreases is that the double salt is best admixed with a high temperature polymer that can seal the cracks and fissures devel- their value in CO2 capture for post-combustion. Moreover, some MOF materials are unstable in a steam/water envi- oped by the solid absorber. The seed funds for this ronment. A novel approach suggested a method to modify project resulted in the acceptance of a separate project MOFs toward increasing the moisture stability and mini- for FY 2013. mizing water sorption without compromising CO2 sorption PN10063/2305

131 Energy Supply and Use 132 PN12057/2458 characterization methods to understand initialmaterialcharacterization methods to understand material advanced design studiesincludes:1)use of thatmaterials. arose Theframework from conceptual was performed, withthefocus onreactor structural material degradation based approach toinsitu RTMof During FY2012,theconceptual designfor thescience- damage at themicrostructural level. of methodsthat attemptanalysis toquantifythelevel inverse conditionsandevaluated behavior underharsh tomonitor materials hardenedenvironmentally sensors Monitoring andpredictive capabilities.We assessed damage. stages of measurement methodsat early of sensitivity gauge electrical) bymeso-scalemodelsandexperimentsto magnetic, thematerial (acoustic, physical properties of mechanismsandtheresulting changesin evolution Behavior assessment.We thedegradation evaluated neutron irradiation, chemicalattack, andotherstressors. temperature,nisms inmaterials undertheinfluenceof mecha- prototypical degradation onset andprogression of Microstructural science.We worked the tounderstand three maintechnicalthrust areas:This project consistedof material structure. when externalinterrogating fieldsare applied toprobe thematerialto detectable variations intheresponse of microstructuralchanges intheunderlying characteristics isascience-basedapproachthis objective that relates margin tocomponentfailure. Acriticalstepinachieving the orunderstand thedegradation progressionmitigate of are about thematerial gained state that canbeusedto mechanisms,betterinsights material degradation stages of insitu.Bydetectingtheearly material degradation RTM of structural materials andtoward examination using of tive inspectionbyperiodicnondestruc- in-service approach of istomovebeyondthecurrent thiseffort The focus of Robert O.Montgomery Structural andFuelMaterials Environmental DegradationofReactor Novel In-SituMonitoringFrameworkfor Energy Supply andUse with materialsdegradation. the mountingissuesthatoperatorsface a capabilityrepresentstimelysolutionto power production.Thedevelopmentofsuch structural andfuelmaterialsusedinnuclear monitoring (RTM) ofearlydegradationin We areadvancingS&Ttosupportreal-time

Science-based framework for RTM of material degradation. for RTMof Science-based framework (impacts interpretation of themeasurement).(impacts interpretation of damagefor heterogeneous materialspredict detectionof the ability togenerate sensormodelswhichstatistically the RTMtechnique),and of ties (impactsthedevelopment with heterogeneous electromagnetic andacoustic proper microstructuresnique), theability togenerate modelsof theRTMtech- (impactsthefieldability of crack precursors enough volumestofindrare features suchasdefectsand theabilityfor tosamplelarge thesemethods,mainly gaps technology changes. Theassessmentidentifiedseveral magnetic oracousticbehavior canbefrom microstructural experimental work hasshownthat changesinmaterial microstructure features onphysicalproperties. Recent response force microscopy) tocharacterizetheimpactof acoustic microscopy, magneticforce microscopy, andpiezo- atomuse scanningprobe force microscopy methods(e.g., Assessments were thepotentialto performed toevaluate damage state reconstruction. quantification anduncertainty methods for inversion signal environments;ments for and5)useof useinharsh sensormaterials andinstru advanced - of 4) development heterogeneous electromagnetic andacousticproperties; andthat cansimulateconductivity) sensorresponse to properties (coercivity, elasticconstants,andelectrical microstructural changesonphysical late theimpactof meso-scalemodelingmethodologiesto simu- struction of microstructure changesonkey material properties; 3)con- lower-length scale techniques that candetecttheimpactof microscopic physicalproperty imaging of 2) development dislocation defectdensity, andresidual stress conditions); sizeandorientation, texture, state conditions(grain - Quantify the “State of Health” of Nuclear Structure/Materials with an Inverse Resonance Inspection Algorithm Canhai (Kevin) Lai

Our research builds a unique suite of experimentally validated computational algorithms to correlate inversely the measured, real- time structural acoustic spectra to the possible degraded state of nuclear structure/ material.

Nuclear energy is an important element for increasing the Accuracy comparison of three inversion algorithms nation’s clean energy portfolio for environmental friendly sustainable growth. Currently, existing U.S. nuclear reactors were summarized in a paper accepted to the Journal of account for at least 70 percent of low greenhouse gas-emit- Sound and Vibration. On algorithm development, we ting domestic electricity production. Though the existing adopted two more sophisticated correlation functions, U.S. nuclear fleet has a remarkable safety and performance both providing needed improvements on the accuracy of record, many reactors are approaching their designed life the inverse algorithm. The effectiveness of three correla- spans. However, because DOE’s R&D Report to Congress tion models has been evaluated and compared. Frequency identifies the development of “technologies and other shifts for a stainless steel dog-bone with local stiffness solutions that can improve the reliability, sustain the safety, reduction at different locations are predicted with ABAQUS and extend the life of current reactors,” non-destructive simulations and are used as numerical examples. We also structure condition monitoring techniques must be devel- developed a prototype software suite that establishes vir- oped to quantify the “state of health” of reactor structure tual flaw database from the finite element method (FEM) materials and predict their residual safe operating life with simulation results and predicts the locations and nominal the in situ quantified material properties. sizes of up to three flaws. The objective of this research is to build a unique suite In the software suite, a maximum correlation-based algo- of experimentally validated computational algorithms rithm for resonance inversion is numerically examined inversely to correlate measured, real-time structural acous- and validated, using three correlation functions mentioned tic spectra to the possible degraded state of nuclear struc- above. The results demonstrated that the inverse reso- ture/material. We will create a unique capability addressing nance inspection approach can quantify accurately the the above needs by developing an integrated experimen- locations of up to three unknown dispersed flaws. How- tal/modeling approach using inverse resonant ultrasound ever, the prediction accuracy starts to decrease when more spectroscopy or resonance inspection (RI). Widely used in flaws exist and when the flaws are close to each other. production, current RI technologies lack the specific means We also identified several important links that need to to identify the location and size of flaws. This project be established before this inverse resonance inspection involves developing inverse algorithms, simulating finite methodology can be put into engineering practice. The elements, sample preparation, and experimental RI data results of the inverse algorithms study were summarized collection. More accurate correlation algorithms will be in another manuscript (in press) for the ASME Journal of developed to improve inverse resonance inspection to Vibration and Acoustics. pinpoint structural flaws and build a prototype inverse RI algorithm software suite. In FY 2012, we met our new reprioritized goals with a reduced project budget. We further improved the inverse We achieved several important technical milestones in this resonance inspection algorithm and turned the prototype 2-year project. In FY 2011, we developed a computationally software suite developed in FY 2011 into a mature, efficient method based on linear perturbation theory to advanced Inverse-Resonance Inspection Suite (I-RIS) for predict resonance frequency shift of a flawed part based which we designed the sensitivity matrix and computa- on the finite element analyses of a perfect part. The results tional algorithm utilizing the database. The matrix consists PN11063/2385

133 Energy Supply and Use 134 PN11063/2385 predictions location onbothflaw andseverity. Because algorithm canprovide reasonably reliable andaccurate Numerical results showedthat RI theproposed inverse databases. flaw thevirtual fromcan bedifferent that of degradation) of (sizeand degree from severity whichflaw dation) andtheirvariouscombinations were considered, - degra structural (cracks,voids,andlocalstiffness flaws of have beenrun totestthealgorithm. Three types different numerous numerical I-RIS, simulations With thehelpof factors. anddegradation geometry andcracks)withuser-defined (voids, degradation, flawsinput file various andallowsausertoinsert FEM figure snapshot). Thesoftware reads anABAQUS for thispurpose (referModifier” hasbeendeveloped to ciently, asoftware suitenamed“FEMModelAnalyzer/ changes intheFEMinputfile. - To perform thetaskeffi necessary orinserting intotheoriginalpart virtually flaws This simulation taskrequired way tointroduce anefficient butat locations. different same sizeandseverity) the of (i.e., degradation thesamecategory of many flaws databases, eachcontainingFEMsimulation results for A specificFY2012 goal wastocreateflaw manyvirtual types, sizes,andshapes at locations. different works threeflaws for withdifferent identifyingthefirst simulations were run, whichproved that I-RIStechnology thesoftware suite.figure Numerous showsasnapshot of tofindthebestfitineachiterationautomatically step. The databaseflaw for thesimulation orallowthealgorithm choosewhich canmanually users part, an unknownflawed flaw virtual database.specific) imported flawsin Tofind iteration stepsusingall(ora shifts in theendof quency cient at eachiteration stepandthesmallestresidual fre- thehighestcorrelation- coeffi can seekacombination of computationalbases, theiterative algorithm implemented data flaw virtual - of largenumbers the I-RIScanimport alsoperformed. As analysis various sizes,withsensitivity radations, voids, andcracks)at variouslocations andat databases flaw virtual - flaw types(deg for various several of Inverse resonance inspectionsuite(I-RIS) Inverse Energy Supply andUse ASME Journal of VibrationandAcoustics. of submitted totheASMEJournal thesimulation work usingI-RISwas The summation of flaws. tobeinducedbysmallvirtual mode switches quency structure inthenumerical modelsandnosubsequentfre- theundamaged thematerial properties/geometries of of Inturn,thissituationparts. requires correct representation shiftsbetweengood andanomaly frequency of accuracy algorithm demonstrated caseshingesonthe onvirtual thecurrent of It shouldbenotedthat prediction accuracy theengineeringstructures. health of theactualstate of estimationthe quantitative of tobehelpfulinestablishing findings are believed size dependent onflaw for internalvoids.These whilelinearly size and flaw for modulus-degradation, modulusloss found tobeproportional totheproduct of into themixed Thepredicted flaws. damageweightwas getinvolvedandwheninternalvoidsareflaws mingled shows that whenmore the prediction andlarger deviates coexist.It flaws able accuracy, whenavarietyof even canidentifyuptothree with unknownflaws reason- ogy databases. Itisdemonstrated that theproposed methodol- internal voids,andcracks)usingsimulated flaw virtual (modulusdegradation, flaws virtual with varioustypesof numerical studies onastainlesssteeldog-bonesample by isevaluated thismethodology of The effectiveness is withunknownflaws part available. ananomaly trum of spec- whenthefrequency structural quantitatively flaws of been presented toidentifythelocation andseverity algorithm has forinverse resonance inspectiontechnology In thisproject, amaximumcorrelation-based iterative engineeringsystems. in-service active of monitoringthestructural health useful inquantitatively methods, thistechniqueisexpectedtobeparticularly directregions for physically visualinspectionorother identificationsuch flaw helpsnarrow potentialdamaged FEM inputfile analyzer/modifier Validation through experimental data is crucial for I-RIS Accurate numerical representation of flaws in real life is technology to be applicable to industry. For experimental another crucial issue to this inverse RI approach. For exam- tasks in FY 2011, we fabricated 30 structural parts and ple and as elaborated in the previous section, actual cracks completed RI tests on them. The goal was to establish usually introduce stiffness reduction around the crack consistency between FEM predictions and experimental wake, and model the virtual cracks accurately requires RI measurements for perfect and flawed populations. additional characterization of the size of the flaw-affected However, from challenges and factors such as RUS mea- region and severity of the degradation. In the end, because surement accuracy, possible material property imperfec- the algorithm developed is a general approach, more com- tions (heat treated vs. non heat treated), and geometry plicated flaw cases or complex flaw combinations can also non-consistencies, the spectrum result comparison be considered by establishing additional virtual flaw data- between FEM simulation results and experiment measure- bases accordingly. ments was not as consistent as expected. With the reduced budget in FY 2012, work on the experimental side has shifted to future strategic planning. Specifically, we submitted two proposals, identified technology gaps, and attempted to establish collaboration with Vibrant Corporation, the industry leader for precise RUS instruments and measurements. Although the collaboration was not funded, our efforts provided a solid path toward bringing I-RIS from technology readiness level 3 (TRL3) to TRL5. PN11063/2385

135 Energy Supply and Use 136 PN12068/2469 under the Microsoft .NET code development platform.under theMicrosoft The .NETcodedevelopment written userinterface asamoderngraphical developed To was FEDSuserinterface anew meetproject objectives, buildings. retrofitscapabilities toexisting toidentifycost-effective fits. Thecompletedproduct willalso provide modeling ability toidentifyenergy, cost,andemissionssaving retro- morewill beasignificantly robust FEDS modelandthe thiseffort capabilities.tionality andanalysis Theresult of to thenextgeneration func- for implementingadvanced FEDS investmentinmigrating stageof asthefirst interface FEDSuser wastointroduceThe project anew objective functionalityandfeatures. the nextgeneration of tool-of-choice,continue itneedstoadopt astheanalysis platformrent canprovide. FEDSto codedevelopment For and capabilities that surpass thecapabilities that thecur significantly,has alsoadvanced features mandating new desktopcomputingtechnology years, FEDS development FEDShasincreased inkind.sis capabilities Duringthe of emissions reductions grow, significantinterest- intheanaly Asaffordableopportunities. energy, security, energy and investment efficiency introducing atoolfor buildingenergy asamodestinvestmentin1990withthegoal of began existing buildingstock.To combat thesechallenges,FEDS and carbonemissionissues,then,onemustlookat the electricity. To security, address energy security/stability, grid thenation’snow) willconsumemore thantwo-thirds of one couldsurmisethat existingbuildings(thoseinplace available electricity;thus, buildings willconsume77%of projections Further indicate that by2030, existing building. theaverage that of buildingswillbeabout half new sity of useinten- and designpractices,itisassumedthat energy development, inbuildingcodes,technology From advances James A.Dirks and ExpandingMarkets:GUIUpdate Restructuring FEDSforContinuedGrowth Energy Supply andUse for existingbuildings. mance anddeterminingtheoptimalretrofits focused onmodelingbuildingsystemperfor Decision System(FEDS)software–whichis management tool–theFacilityEnergy This projectaddscapabilitiestotheenergy

- - capabilities andenhancements. willallowmovingforward face more readily withnew All project goals userinter were accomplished.Thenew demands. that theFEDSuserinterface detail of capabilitycific FEDSanalysis butwithoutthelevel tier” hasalready beenestablished bytoolsthat require spe- layer. codefunctionalitytoa“middle of Thismigration added middle-tier screens code interface andintoanewly technologieswasremovedming interface from theuser used byfuture webapplications orapplication- program More specifically, anyfunctionalitythat couldbeeffectively werewhere activities implemented. non-userinterface alsoincludedreorganizinghowand This project effort project “inferenceand anew process” interface. requiring minimumset input, a detailedinputwithoutfirst input tips,more comprehensive qualityassurancefeatures, screen anenhanceduserguidanceviatooland resizing, between project inputscreens, afunctionalFEDSinput ments, whichincludeda“caseexplorer” toolfor navigating functionality. Thesecondtaskadded software enhance- platform tomaximizeamorebilities inthenew modern capa - form. of Included inthisprocess wasanevaluation plat- andverification,screens, tothenew inputchecking, existingcapabilities, pieceported including The first output. graphical of allocation, andintegration memory use, tionality toenhancetheuserexperience, easeof ing VB6codetoVB.net, andupdating andincreasing func- approach focused ontwomainareas: theexist- converting Screenshot of the new “inference process” thenew FEDSinterface Screenshot of - Stabilized Li Metal Anode for Li Batteries Jason Zhang

salt resulted in wire-like lithium deposition morphology, Rechargeable lithium metal batteries are con- which is quite different from other lithium salts. The salts

sidered “the holy grail” of energy storage of LiPF6, LiClO4, and LiTFSI also showed higher cycling systems. Unfortunately, uncontrollable den- efficiency and better cycling stability than other lithium dritic lithium growth inherent in these bat- salts. Therefore, LiPF6 was chosen as the standard salt for further studies. teries (upon repeated charge/discharge) has prevented their practical application over the When testing the effects of electrolyte additives on the past 40 years. A novel approach has been morphologies of lithium deposition films, we identified developed in this work to solve this problem. several salt additives that led to significant improvements on the smoothness of lithium metal films deposited in a

copper substrate. For instance, when CsPF6 was added to Lithium metal has an extremely high specific capacity the electrolyte of 1 mole/liter (M) LiPF6 in propylene car- (3820 mAh/g) and low potential (–3.04 V vs. standard bonate (PC), the dendritic morphology of deposited lithium hydrogen electrode), which make it an ideal anode for high film was changed to dendrite-free morphology at 0.01 M energy lithium batteries. However, dendrite growth and CsPF6 addition. At 0.05 M concentration of CsPF6 in this Li/electrolyte side reactions are the two main barriers that electrolyte, the lithium deposition film had a metallic mir- prevent its practical applications. After 40 years of limited ror-like shining, suggesting the smoothness of the lithium work in this field, the urgent need to double the energy film. This Cs-additive could effectively smooth the already density of batteries used for electric vehicles and for devel- formed lithium dendritic morphology in the continuous opment in nanomaterials drive a renewed effort around deposition. The in situ optical microscopic observation of the world to improve the stability of the lithium anode. lithium deposition in electrolytes with and without the CsPF additive demonstrated lithium smooth growth in In FY 2011, we performed a fundamental analysis of the 6 lithium dendrite nucleation and growth processes using an the Cs+-containing electrolyte but dendritic growth in approximate quantum mechanics method and systemati- the control electrolyte. In addition, the Cs-additive could cally investigated the effects of various electrolyte solvents improve the cycling efficiency and long-term cycling stabil- on lithium dendrite growth at the surface of the copper ity of lithium electrode for more than 400 cycles. substrate during the electrochemical deposition process. Based on our results, we proposed a novel self-healing We also developed special membranes to minimize lithium electrostatic shield (SHES) mechanism to explain the effec- dendrite growth and found that nearly no dendrite formed tive prevention of lithium dendrite growth during lithium at the surface when treated with hydrothermal crystalliza- deposition and long-term charge/discharge cycles. It is tion; therefore, a zeolite membrane was determined to be under the assumption that an additive cation has an effec- a good protection layer to suppress dendrite formation. tive reduction potential lower than lithium cation achiev- In FY2012, we studied the effects of lithium salts with dif- able by the indication of the Nernst equation. During the ferent anions in a cyclic carbonate solvent on morphologies deposition, lithium and additive cations are attracted to and cycling efficiencies of lithium deposition films. We the existing lithium tip, but the additive cations are not reduced. This forms a positively charged electrostatic shield found that the salts of LiPF6, LiAsF6, LiTFSI, LiBOB, LiDFOB, and LiI could form a compact film to cover the copper sub- around the lithium tip that repels other positively charged strate, but the deposited lithium films from electrolytes cations so that more will be preferentially deposited in the with LiBF , LiCF SO , and LiClO were not good. The LiBOB adjacent region of the existing tip. This process continues 4 3 3 4 until all substrate valleys are filled with deposited lithium, indicating the repeated self-healing process. Computational

simulations on lithium deposition in LiPF6-PC electrolytes

with and without the CsPF6 additive are consistent with the proposed SHES mechanism. SHES additives will not be consumed during operation so they can be used for long-term cycling. These promising results indicate that lithium metal can be used as the anode in rechargeable lithium metal batteries to achieve high energy density lithium batteries SEM images of Li metal film grown on Cu substrates without additive for electric vehicle applications. (left) and with additive (right). PN11072/2394

137 Energy Supply and Use 138 PN12072/2473 these lithium salts. from films thedecompositionof onAlsurface organic and denseinorganic protection isduetotheformation of protection couldbeoccurred anodicscan. The at thefirst the saltadditivecontentincreased to3%and more, the could protect Alfrom scan.When corrosion afterthefirst LiBOB (bisoxalatoborate), with2wt.%intheelectrolytes that twolithiumsalts,LiDFOB(difluorooxalatoborate) and additives.Itwasfound withtwelvedifferent electrolytes scanningvoltageinLiTFSI-based trode withthechangeof theAlelec- usedtotestthecurrent was first responses of (CV)method voltammetry sion bytheTFSIanion,cyclic protecting theAlsubstrate from corro of - As for theeffort addressed byadding appropriate additives. interphaseelectrolyte graphite films anodes)are on imide (LiTFSI)andlithiumtetrafluoroborate (LiBF more stable salts:lithiumbis(trifluoromethylsulfonyl)- withthermalandhighvoltagestabilities byusing lytes non-aqueouselectro- thisproject istodevelop The goal of batteries for large-scaletransportation applications. able toimprove thesafety, Li ion andcalendarlifeof cycle, stable andelectrochemically that electrolytes thermally are vents. Therefore, there isanurgentneedtodevelop sol - electrolyte of materials, anddegradation ode active cath- electrode-surface-protection-films, dissolutionof of cause seriousproblems tothebatteries, suchascorrosion The formed byproducts are acidsthat strong acidandLewis hexafluorophosphate (LiPF tions. TheSOA basedonlithium non-aqueouselectrolytes applica for- EVandgrid storagedevices and safeenergy inSOA LiionbatteriesAdvances are criticaltoensure high Wu Xu for LithiumIonBatteries Stable andHighVoltage Electrolytes strate) andLiBF problems withLiTFSI(corrosion toAl,thecathode sub- LiPF thelithium salt thebatteries, especially reduced lifetimeof tothesafetyproblems and vents contributesignificantly Energy Supply andUse teries, leadingtomorereliableproducts. the lifetimeandenergydensityofLiionbat- voltage stabilitywillsignificantlyincrease of newelectrolyteswithhighthermal,high- lithium (Li)ionbatteries.Thedevelopment endar lifethanthestate-of-the-art(SOA) higher energydensity, andlongercycle/cal- tems tohavehighersafety, lowercost, (EVs) demandsadvancedenergystoragesys- The commercializationofelectricvehicles 6 duetoitsthermalinstability andmoisture sensitivity. 4 (poor capability of forming good capability solid (poor of 6 ) and organic carbonate sol- ) andorganic 4 ). The

such asLiNi tion onhighervoltage, highercapacity cathode materials stable for electrolytes thesethermally gate theirapplica- LiNi tive test of theLiPF testof tive high current rates thantheoneswithLiPF havewith LiTFSIelectrolytes muchhighercapacities at The electrochemical that characterizations batteries reveal In the effort of replacing LiPF of In theeffort LiBF latter hasamuchhigherionicconductivity. electrolyte The the LiPF stability similarto both showsimilarlong-termcycling (withandwithouttheLiBOBadditive) LiTFSI electrolytes higherrate performance.to even Inaddition, thetwo ity. leads 3%LiBOBintheLiTFSIelectrolyte Theaddition of better capacity retention thantheLiPF Li|LiFePO on therate capability stability of andlong-term cycling LiBOBasadditiveinLiTFSI-basedelectrolyte of The effect cells, nearly all of theLiBF allof cells, nearly use inLiFePO with theappropriate additives at amountsfor certain the 2 wt.%concentration inLiBF cycling stability of theLiBF stability of cycling Theseadditivescanalsoenhancethelong-term lyte. Co LiPF havemeaning LiTFSIelectrolytes better rate capability than Li ionbatteries for EVapplications. toimprove safety,developed andcalendarlifeof cycle, of LiBF of cells, allstudiedadditivescanimprove therate capability stability.cycling Ontheotherhand,inLi|Graphite ate are thebestinbothrate capability andlong-term the studiedadditives, LiBOB, LiDFOB, andvinylenecarbon- on LiTFSIandLiBF stable andelectrochemically based electrolytes thermally It hasbeendemonstrated from theabove results that the investigated. to befurther Theoptimaladditivecontentneeds in theelectrolytes. 2%) additives(only consumptionof due totheeventual electrolyte formulationselectrolyte basedonLiTFSIandLiBF andoptimized.Feasible trations willalsobeevaluated the LiPF ent additives)have enhancedrate agreatly capability over is stillnotasgood astheLiPF Li|LiFePO 1/3 0.5 6 4 O2 (4.4V),Li-richcompositecathode (4.6 V),and electrolyte,- althoughthelatter hashigherconductiv electrolytes alsoshowhigherdischargecapacity electrolytes and Mn 4 -based electrolytes overtheLiPF -based electrolytes 6 6 electrolyte. electrolyte especially at highrates, especially electrolyte althoughthe 1.5 4 4 andLi|Graphite cells,alongwiththecompara- inCR2032coincells. batteries was evaluated O 0.80 4 (4.9V).More additives andadditiveconcen- 4 battery system. InFY2013,wewillinvesti- battery Co 0.15 4 are able toreplace theunstable LiPF 6 Al control electrolyte. IntheLi|LiFePO 0.05 O 2 (chargedto4.1V),LiNi 4 4 -based electrolytes (withdiffer -based electrolytes electrolyte, buttheirstability 4 electrolytes were electrolytes testedin 6 6 basedelectrolyte, likely withLiBF 6 electrolyte. Among 6 4 control electro- , 10additivesat 6 electrolyte, 1/3 4 willbe Mn 1/3 6

- Engineering and Manufacturing Processes Development of Functional Membranes for Improved Selectivity David W. Gotthold varying water content to a test cell and permeate extrac- We are developing new methods for making tion using either a vacuum or carrier gas to a mass separation membranes to enable higher spectrometer for analysis. selectivity, improved transport, and extended Bias membrane. For this work, the focus was on determin- lifetimes. The initial focus is on improved ing the relative change in water transport with the applica- dehumidification membranes for application tion of electrical bias. The tested film was a commercial to metal-air batteries, building climate control, DuPont Teflon® PFA 125 µm thick membrane already carbon capture, and natural gas processing. under investigation for water permeation. Both sides were coated with sputtered Al (0.5 µm thick) after a 5-minute

exposure to O2/Ar plasma required for proper adhesion. To date, most membrane work has focused on specific The Al was then patterned using a pulsed excimer laser to material systems, polymers, metals, or ceramics. Hybrid produce a grid of 5 µm diameter holes on 20 µm centers. system work has been primarily about the use of structural The film was tested using 50% RH zero air with biases up to support as opposed to combining functional materials. The 500 V; however, no changes in permeation were observed. objective of this project is to achieve higher performance Graphene oxide membrane. This work was based on pub- by combining different materials into a single membrane lished results that showed GO has significant potential as a structure and developing processes necessary for scaling to highly selective membrane for water vapor with selectivi- manufacturing. To achieve this goal, we focused on two ties >1010. GO was prepared following the Tour group’s different approaches to improved selectivity membranes. published improvement over the commonly used Hummer The first was to improve the selectivity of standard mem- method. Improved graphene oxide (IGO) is produced from branes using applied electric fields and integrated porous graphite flake through oxidation with potassium perman- thin film electrodes on standard polymer membranes to ganate (KMnO ), sulfuric acid (H SO ), and ortho-phosphoric affect the relative transport of polar and non-polar mole- 4 2 4 acid (o-H PO ). The IGO is recovered through multiple cules. The second approach was to evaluate the possibili- 3 4 rinsing steps with distilled water and ethanol and re- ties of a newly discovered graphene oxide (GO)-based suspended in fresh water and lyophilized to enable membrane that demonstrated remarkable water vapor it to be re-dispersed. selectivity. Specific procedures and membranes used are highlighted below. GO membranes were prepared by filtration and consisted of several µm thick multilayer stacks of IGO molecules. Dry Capabilities. To test the membranes, a new evaluation sys- IGO was suspended at a 10 mg IGO/80 ml water using bath tem was assembled that enabled a range of gasses with sonication followed by stirring. Aqueous suspensions of IGO were drawn through 47 mm diameter membranes using a filtration apparatus with mild vacuum. Free-standing IGO membranes were prepared on anodized alumina membranes with 0.02 to 0.2 µm pore size. The air-dried IGO membrane was easily peeled from the alumina, provided it was at least 4 µm thick and demonstrated selectivities

for H2O over N2 and O2 at or near the measurement limit 4 of 10 with very high H2O vapor transport rates (see figure). To enable manufacturable membranes, support and protection layers were added. Supported IGO membranes were prepared by drawing the IGO suspension through standard filter paper and both polycarbonate and polypropylene membranes. The filter paper roughness was too high and disrupted the IGO membrane, but both polymer-supported membranes were functional and observed as susceptible to GO membrane with selectivity for H2O over N2 and O2 above measurement limit (>104). temporary disruption due to liquid water accumulation on PN12022/2423

140 Engineering and Manufacturing Processes their surface. This led to penetration by both N2 and O2 In summary, we demonstrated that GO based membranes permeation, though the high selectivity returned once the could be successfully integrated into multilayer structures to liquid water was eliminated. To provide a protected film provide high selectivity, high permeability and the potential for manufacturing operation, IGO membranes were pre- for manufacturable processing as long as the support struc- pared with permeable surface coating. This membrane ture was sufficiently smooth. The addition of a surface coat- showed acceptable transport rates and did not exhibit ing improved the robustness of the device operation in liquid water disruptions. conditions where liquid condensation could occur. PN12022/2423

141 Engineering and Manufacturing Processes Materials Science and Technology Advanced Sorptive and Signature Indicating Materials for Ultra-Trace Detection Jay W. Grate and the desorption was completely reversible. The uptake The collection and detection of signatures of xenon by MOF-5 at 1 bar is only 26 wt%, about half that of undeclared nuclear activity is an essential of NiDOBDC. The uptake values for other atmospheric gases such as N (1 wt%) were very small at 1 bar. Further, technical capability for monitoring treaty 2 in addition to providing excellent xenon uptake, the MOF compliance. materials provide more facile release of gases than activated charcoal. This feature can be advantageous in tech- Signatures of nuclear activities can occur in the gas phase nological applications for gas capture, concentration, and as gaseous molecules or aerosol particles, in the liquid release. The energetics of the xenon interaction with phase as dissolved ions, or as residues or components of NiDOBDC were examined in terms of the isosteric heat of solid samples. This project will develop three different adsorption, and the interaction of the guest xenon mole- but complementary technologies for concentrating and cules with the MOF host surfaces. This work has resulted in detecting signatures. Under the first task, we select and a peer reviewed journal article featured on the front cover synthesize appropriate metal organic framework (MOF) of Chemical Communications. candidates that are relevant to xenon gas sorption, collection, and separation. The second task is about enhancing signatures of a wide variety of dissolved or suspended nuclear materials, fission products, and related process chemicals in water or other liquids. The third task involves the design of uranyl ion binding TCC-cpGFP proteins for engineering sensors capable of uranium detection for treaty monitoring applications. These tasks and their specific project progress are described below. Metal Organic Frameworks (MOFs) for Noble Gas Sorp- tion. With regard to the gas phase, the collection of xenon gas is an important aspect of treaty monitoring and detection of undeclared nuclear explosions. Sorptive materials play a key role in this collection. This project addresses a new type of sorbent, MOFs, with superlative properties compared with current carbons and zeolites for gas capture and separations. We are developing and investigating MOF materials for gas collection and separation of xenon. The development of these materials will lead to more effective collection of xenon in the face of variable backgrounds. We have investigated a number of MOF candidates, starting with the well-known MOF-5, focusing on capture at nonc- ryogenic temperatures. We have found a MOF structure, NiDOBDC, that captures xenon gas with twice the capacity of MOF-5, and which has superior structural stability with uptake and release of guest molecules. The uptake can be Sorption of xenon by an MOF as suggested schematically and featured further affected by the metal atoms that comprise the MOF. on the cover of Chemical Communications. The uptake of xenon by NiDOBDC has been compared with that of activated charcoal, which is a conventional sorbent Magnetically Active Sorbents for Proliferation Signature evaluated for comparison. Neither NiDOBDC nor charcoal Collection and Detection. Capture of low concentration has significant uptake for 2N . The uptake of Xe by NiDOBDC signatures, such as radionuclides, heavy metals, and is comparable to that of carbon overall, with superior organic compounds from complex matrices and/or large uptake from 0.25 to 1 bar. The calculated weight percent- volume samples is challenging. This effort focused upon age of xenon at 1 bar, on NiDOBDC was found to be 55 wt% the development of magnetic nanoparticles for nuclear PN09008/2176

143 Materials Science and Technology proliferation signature collection and detection. By modify- detection. Design of the sensor material was based on ing the surface chemistry of these high-surface area mate- metal binding trimeric coiled coil peptides fused to circu- rials extremely high affinity and capacity sorbent materials larly permuted variants of fluorescent proteins. Our can be created. Sorbent performance of the modified research used the green fluorescent protein (GFP) because nanoparticles far exceeds that of commercially available its structural and fluorescent properties have been well sorbents for the collection of trace levels of radionuclide. characterized. However, the results can also be extended Selectivity and affinity of the nanoparticles can be adjusted to the design of other novel metal binding fluorescent with installation of the desired surface chemistry and new proteins. materials with extremely high affinities for fission products Genes encoding the cpGFP-TCC peptides and controls have and f-block elements have been created and demonstrated. been synthesized and incorporated into both yeast and The small particles disperse easily in solution enabling bacterial surface display vectors. Metal binding on the sur- rapid collection and concentration. After capture of the tar- face of both yeast and bacteria displaying the proteins get radionuclides the magnetic properties of the nanoparti- have been demonstrated. Initial rounds of directed evolu- cles allows them to be rapidly separate the nanoparticles tion to alter metal binding specificity have been accom- from the bulk solution and magnetically direct them into a plished and continue to be characterized using iterative geometry ideally suited for analysis. We have explored and rounds of mutagenic PCR. In vitro expression of the con- optimized a range of methods for collection of or trace structs to produce soluble forms of the sensor proteins radionuclides signatures from large volumes. We have were accomplished but the yield of soluble protein was demonstrated accelerated assay of the nanoparticles with relatively low. Therefore we developed an additional gamma detectors, beta detectors, alpha detectors count- approach to creating metal-specific sensor proteins by ing, proportional counters, liquid scintillation and ICP-MS. starting with a naturally occurring metal-binding protein Efforts from this project have resulted in 5 peer reviewed that we fused to cpGFP to create a protein that fluoresces manuscripts (to date) and were featured on two journal only in the presence of the metal to which it naturally covers (Health Physics Journal and ChemSusChem). Patent binds. Eight directed mutations were conducted on the applications and additional papers are being assembled wild-type scaffold to tailor the protein specifically to bind and submitted based upon this work. to uranyl ions. This mutated protein was fused to cpGFP to Protein Based Uranium Detection. This task enhances create a protein that fluoresces only in the presence uranyl the capability for stand-off detection of upstream uranium ions. These sensor proteins are expressed very well in E. coli processing using novel protein-based fluorescent sensors and shown to be active by fluorescence in vitro. The fluo- that bind to uranium species. These proteins were designed rescent sensor protein for uranyl ion detection has not to fluoresce only in the presence of labile uranyl species been described in the literature and is thus novel. and are being functionally tested for high affinity/specificity to demonstrate effective and enhanced fluorescence PN09008/2176

144 Materials Science and Technology interaction between graphene and IL. Subsequently,interaction between graphene This formation enabled ustostudythemolecularlevel ILmonolayer ensures surface. theformation ongraphene of ILinwater in thissynthesisandtheextreme solubility of GtoIL(1:1.5) used based method.Thelowmolar ratio of synthesized (hereafter termedG-IL)usingaqueous solution trifluoromethanesulfonate (TfO IL, a monolayer of 1-butyl-3-methyl-imidazolium(BMIM IL, amonolayer of To themolecularinteractionbetweenGand understand and ionicliquids. regionbetweengraphene insight intotheseinterfacial molecularlevel mental andtheoreticaltogain analyses regions.Wecomplex interfacial performed combinedexperi - these themolecularstructure anddynamicsof standing of region,itisessentialtohave acomprehensiveunder facial between theILsandgraphene. To create thisoptimalinter layer spacing(i.e., ILthickness)andmore favorable affinity layers.graphene Suchaprocess requires controlled inter calating theILsindesired thicknessbetweenindividual sheetrestackingture graphene ispreventing toallowinter aG-ILnano-hybridstruc- The majorchallengeindeveloping lithiumionbatteries. that of densitycloserto withenergy ultracapacitors (ILs) toachieve exfoliated (G)andionicliquids graphene region consistingof ionic liquids.Hence, weare designinganoptimalinterfacial moreand anelectrochemically stable electrolyte, suchas electrode area material withahighersurface (i.e., graphene) densitycanbeincreased byemploying electrolyte, an energy the theelectrode andtheelectrochemical stabilityarea of of isdependentonthesurface stored energy the amountof Because optimalinterfaces. density duetotheabsence of ever, thesituation alsoinflictsaseriouslimitation inenergy andhighercharge/dischargerates. How- life (>10,000cycles) properties,leads tomanyattractive includinglongercycle- and ultracapacitors (EDLC) formation. of Thisistheheart electrodes toform anelectricdoublelayer capacitance charged canformelectrolyte electric-dipolewithoppositely region, ionsfrom the Inthisinterfacial ultracapacitors. of tothefunctioning ispivotal and ionicsolutionanelectrolyte) (anelectrode) The interactionbetweenachargedsurface Vijay Murugesan Hybrid MaterialforUltracapacitors Development ofGraphene/IonicLiquid ultracapacitors. faces toachievehighenergydensity graphene/ionic liquid-baseddesignerinter regions. Thisprojectfocusesondeveloping storage devicesneedoptimalinterfacial The compactandhighperformingenergy - ) on graphene surfaces was was surfaces ) ongraphene

+ - )

- - - designer-interface materials.designer-interface withoptimal densityultracapacitors lead tohighenergy willultimately Theseefforts during charge/dischargecycles. in situNMRandXPSspectroscopy totesttheseG-ILmaterials uniqueprobe systemstoperform highresolutiondevelop designprocess. Further,G-IL optimalinterfacial wewill properties andprovide inputfor cycling tailoringILsinthe tions inG-ILnano-hybridmaterial for betterelectrochemical As anextstep, wewillfocus ontestingvariousILscomposi- anddesiredinterfaces spatial separation (<10nm). and Ionicliquidintoanano-hybridstructure withoptimal graphene of Thisleadstosuccessfulintegration the surface. byincorporatingis prevented epoxy andhydroxyl on groups Based onthisinformation, sheets graphene restacking of region, andtheresults canbesummarizedasfollows: This combinedapproach shedslightontotheG-ILinterfacial correction (DFT-D3) basedmethods. withempiricaldispersion density functionaltheory lated withmolecular models,whichwereusing derived rial were results carriedout.Theseanalytical are thencorre- onthisG-ILmate spectroscopyand infrared- analyses (FTIR) nuclear x-ray magnetic resonance photoelectron (NMR), (XPS), on graphene surfaces. on graphene liquid moleculesandoxygen containingfunctional defectgroups interaction betweenionic of through fundamentalunderstanding G-IL hybridnanostructure withoptimalinterlayer spacingisachieved • • layer capped byTfOanionlayer. subsequently p BMIM cations layer orientparalleltothegraphene from oxygen (i.e., containingdefectgroups hydroxyl andepoxy) regions,whereasgraphene TfOanionsare repelled by interactwithdefect-free justasthey surface graphene BMIM cations onthe caninteractwithfunctionalgroups - p stacking interaction to form a primary interfacial interfacial stacking interactiontoform aprimary Materials Scienceand Technology

145

PN12023/2424 146 PN12024/2425 tigating theelec- tigating inves- capable of a modularsystem we willconstruct spectrometers, ment-specific XES low-cost,ele- of Through theuse mal effects. order andther by structural dis- less influenced character andis ties, andbond magnetic proper spinstate, ting, as orbitalsplit- information such tronic structure detailed elec- moregives interactions, soit electron-electron the probe” of identity.ligand However, XESisa“more unlike local XAS, oxidation state,ment of bondlength,andneighboring troscopy (XAS)inthat itprovides element-specificmeasure- tox-ray absorption spec- emission (XES)iscomplementary storageandmaterials. X-ray inenergy particularly ing, andmaterials engineer chemistry mental tomanyfieldsof ionsisfunda- multivalent detailed electronic structure of structure withlowerspatial the resolution. Understanding surements electronic andlessdirect measurements of materials relies onexsitumea- work onbattery previous mations and changingelectronic the structure. Muchof transforelucidating mechanisms that leadtoirreversible lifetimeandperformance canbeimproved by Battery Nancy J.Hess Spectroscopy Nanoprobe Development ofHard X-RayEmission Materials Scienceand Technology that impactbatteryperformance. knowledge ofirreversibletransformations during operationtogainfundamental electronic structureofbatterymaterials We aredevelopinganinsituprobeofthe - - Ni at 8352eVfor comparison. 8347 to8352eV. theCl to Eshowstheratio of XANES for NiandNiCl NiCl Na/ A andBare NiandClfluorescence maps of 2 battery, respectively. Cshowsnormalized 2 . Dmaps theNiratio at

- - NiCl poration intotheminiXSspectrometer. transmission XASexperimentsandiswell-suitedfor incor research isdesigned for team.TheSSRLthinfilm battery from theStanford Synchrotron (SSRL) Research Laboratory maps of Ni and Cl alone, identifying small grains of NiCl of NiandClalone, identifyingsmallgrains maps of Ni andNiCl theXANESspectra characteristicof sponding toregionsof combining maps taken at incidentenergiescorre different - date isNa Na-NiCl a asampleenvironment for theinsituoperation of of The nextstepinourresearch project istheconstruction beamlineswithahigher resolution. cept at thex-ray fluxof thecon- 9000 counts/s.Thisdemonstrated thesuitability of sample at sector 20-IDwithapeakintensityover battery designs.NiK-betaspectrawere recorded fromillary theNi structed inbothminiXSandpolycap NiXESspectrometers - Washington, wecon- of In collaboration withtheUniversity withNi. and theirinterface dards weanticipated Fe (Fe, thestan- anddidnotmatch different anyof significantly weremetal, butthespectrafrom thehighcurrent battery at lowcurrent cycled were predominantly iron the battery XANESspectraof mance undertheseconditions.TheFe and therefore may berelated tothediminishedperfor at highcurrent, cycled more pronounced inthebattery were approximately was co-located). Thissegregation thesebatteries andNi (inthechargedstate Fe states of from Niinthedischarged Fe of segregation observed PhotonSource.(XANES) at sector 20theAdvanced We currents usingx-ray absorption nearedgespectroscopy at charged anddischargedstates highandlow cycled fully Na/NiCl(ZEBRA)batteriesWe in obtainedchemicalmaps of the nanometerscale. ple environment andwithspatial resolution at ultimately light actinideelementswithinavariable temperature sam- thetransition metal,lanthanide,tronic and structure of NiCl The NiXANESspectrawere allconsistentwithNimetalor spectra for thiscompound. molten saltelectrolyte, andweare seeking reference 2 2 . Thiswasmuchmore informative thantheelemental chemistry. By withtheexpectedbattery , inagreement 2 BASEbattery. We planto modifyanexistingdesign 6 FeCl 2 , wewere able toproduce achemicalmap of 8 that wouldindicate areaction withthe 2 O 3 , FeCl 3 , FeS). Ourbestcandi- , FeS). - 2

- built an ECQCL with a continuous mode operating frequency built anECQCL withacontinuous modeoperating frequency quantumcascade laserinstrumentation, wecustom ability of Group andtheavailBased ondiscussionswiththe Raschke - usage), lowcost,andruggedness. wide-spread use(permitting incorporation isitseaseof quantumcascadelaser of age. The secondadvantage point across thesample, allowingrapid spectraldata cover Hzwillallowinfrared spectratobetaken point-by- 100s of vibrational bands.ECQCL’s rapid scanning capabilities inthe manyfundamental fingerprint (8–12µm),therangeof region work extremely wellinthemolecular chain. Thesedevices infrared lightsource tothefemtosecondOPO complementary (ECQCL)willbecoupledtothisinstrument asa cade lasers reactions External cavity at quantumcas- theatomic level. speciesinvolvedinsurface-mediated tion about individual microscope (LT STM)that willprovide spectroscopic informa- capability onEMSL’s low-temperature scanningtunneling establish aninelasticelectron tunnelingspectroscopy (IETS) We noblemetalnanoparticles. are working of facets to probe crystal moleculesondifferent adsorption of heat of bondingconfigurations and thedistributionof to investigate capability inscattering infrared near-field opticalmicroscopy atomic force anovel,state-of-the art microscope todevelop aquantumcascadelaserwithan This project willintegrate nor hasresolution that approaches 10nm. tunablenone isbroadly overalargespectroscopic range nano-imaging at highresolution, butwithoneexception, instruments that worldwide canprovide near-field infrared building blockswithinthe10–100nmrange. There are themolecular associated withthesizeandinteractionsof simultaneous spatial resolution lengthscale ontherelevant specificspectroscopicprovide identification achemically with techniques that routinely tal challengeduetothelackof als’ structure andfunctionhasremained amajorexperimen- themateri- of understanding nm. Gainingamolecularlevel 10sof electron correlations onmesoscopiclengthscalesof tionalities from intra-andintermolecularinteraction other natural theiruniquefunc- andsyntheticmaterials gain molecular assemblies,proteins, correlated systems,andmany nanostructures, andsupra- Molecular andinorganic polymer A. ScottLea Scattering Near-Field OpticalMicroscope particles UsinganEnhancedScatteringInfrared Facet SpecificChemistryofNobleMetalNano- that occurinnanostructuredmaterials. sized noblemetalsandthephysicalprocesses tive sitesonvariousfacesofcrystallinenano- equipment thatallowsinvestigationofreac- We areaddinganewcapabilitytoexisting

- between 1280cm nm- toµ-sizespatial variability initsinfrared absorbance weidentifiedanexperimentalsystemwith the integration, resonance of thenitro (bat group 1270cm resonance of whentheIR lightistunedtothe in the2ΩSNOMsignalisobserved contrast tetranitrate the300nmhighcrystals, depositedongold. For pentaerythritol of for crystals 2Ω scattering SNOMsignals(b, d,andf) ande) thecorresponding AFM tapping (a,c, modetopography range of 1260cm range of off-resonance (dandf). off-resonance TiO single-molecule vibrational spectra of O single-molecule vibrational spectraof recording ing thesystemfor initialexperimentscomprised of IETS spectra.We configur alsobegan signal-to-noise ratio of greatlya improve to and lock-inamplifierwillbe developed samplebiasvoltagemodulation technique onthebasisof troscopy intotheLT adetection STMandhave developed We have incorporated theinelasticelectron tunnelingspec- conducts COadsorption experimentsonPdnanoparticles. We alsohadanenvironmental chambermachinedthat gold usingIRwavelength tunedtotheoscillator. of surface nitro containingmoleculesonthe group tial distributionof tionally, toimagethespa- weutilizedECQCLspectroscopically etched inagold film depositedonpotassiumbromide. Addi - harmonic (2Ω)scattering SNOMsignalimagingslotantennas thesecond and have demonstrated fieldenhancementof itintothes-IRSNOMinstrumentation integrated successfully length range. We purchased a5.02–5.52µmECQCLandhave it iscriticaltohave lasertunability inthe5.0–5.4µmwave- C-O adsorption experimentsonnoblemetalnanoparticles, the tion withthefemtosecondlasersystemandECQCL.For toPNNL,andre-assembledported andqualified for opera- In FY2012,thes-IRSNOMsystemwasdisassembled,trans- oxide (PE/PEO)that phaseseparate at room temperature. andpolyethylene polyethylene co-blends composedof 2 (110). -1 -1 to1400cm and1380cm Materials Scienceand Technology -1 inFY2011.To demonstrate -1 and selected polymer andselectedpolymer -1 ) butnotat wavelengths 2 , OH,andH 2 O on O on

- 147

PN11035/2357 148 PN12039/2440 materials by solution grown self template consists of forming forming template consistsof materials self bysolutiongrown aerogels process. Roomtemperature carbon preparation of pathway,growth hydrothermal carbonization, and modified als usingthree self-template strategies: different bysolution In FY2012,wesynthesizedandcharacterizedcarbonmateri - (oil)imports on foreign energy for improved performance reducing Americandependence fundamentalmechanisms of ourunderstanding advance synthesizedmaterials will chemical measurements of Characterizationtrode/electrolyte andelectro interfaces. - electrode that materialsstrategies offer withenhancedelec- for improved synthesis performance enables modification of standing fundamentalmechanismsandproperties required toelectrodes. electrolytes Under als ispooraccessibilityof key scientificbarrier for lowcapacitances withfibrous materi- aspotentialalternative. The pathwaysion diffusion serves porous carbonfiber-based electrode materials withefficient tionalization meso- limitingtheiruse. strategies Inexpensive CNTs, performance, highcostof affecting andlimitedfunc- andstackinmultilayersgate from vanderwall interactions than metalhydrideandLi-ionbatteries.- GNtendstoaggre magnitudelower isanorderstate-of-the-art capacitors of candidate existing material. However, densityof theenergy area, thermalconductivity, andstability have beenusedasa materials withoutstandingproperties includinghighsurface mance. nanosheets(GN) Carbonnanotube(CNT)andgraphene densitieswhileproviding highpowerperforincrease energy to aspotentialalternatives bination with batteries serve densitiesincom- Supercapacitors withhighpowerandenergy Satish K.Nune High EnergyUltracapacitors Hybrid Electrodes forNextGeneration Synthesis of Cu-aminoacidfibers Synthesis of Materials Scienceand Technology electrolyte interface. properties fromanenhancedelectrode/ tion forimprovedmass/chargetransfer surface areasandcontrolledporedistribu- for carbonarchitecturesynthesiswithlarge We aredevelopingsolution-basedstrategies diameters in aqueous inaqueous diameters with100–200nm bers glutamic acidnanofi- thesized Cu-aspartitic/ syn- also successfully hard templates. We require anyexternal method doesnot materials. Our resulting infibrous ization/leaching, glycolate andcarbon-

- tion followed bycarbonization. carbon-basedmaterial reduc orpermanganate - presence of metaloxides inthe will useelelectrochemical depositionof onperformance. highvoltage electrolytes Inaddition, we of electrode materials,micro/mesoporosity andthechemistry of metaloxideWe oncapacitance, therole will investigate of surements areas. on synthesizedmaterials withhighsurface well-positioned inFY2013toperform electrochemical mea- nel potentiostat. We are and acquired 8-chan - area with highsurface thesized carbonfibers We syn- successfully pore distribution. (e.g., 2028m (e.g., areas to synthesizecarbonmaterials withlargesurface approach anew developed In anothertask,wesuccessfully solvent molecules. coordinated bythelossof framework resulted collapsing of not retained. We fibers theorizethat thethermaltreatment of thesematerials is indicatingclearly that of fibrous framework resulted incarbonarchitectures areas, withlowsurface Cu-aspartate andglytamate cess. Eventhecarbonization of tate actsasan insitutemplate during thecarbonization pro- conditions. We formed hypothesizethat Cu-aspar theinitially 118 m areas (1125to surface reduction of resulted insignificant coating metaloxides aerogels. However, ified process used for single stepusingmod- carbon materials in coat metaloxides on approach to oped new - method, wedevel obtained after pyrolysis atobtained afterpyrolysis 900°Cis379m thesolid areadration BETsurface of andpolymerization. followedand glucosealdehydegroups, bysubsequentdehy- amine spheres cysteine were obtainedbythereaction of asstructural Micro constructors. - and aminoacids(cysteine) vated temperatures (180°C).We usedcarbohydrates (glucose) mal carbonization route toobtaincarbonmaterials at ele- densities.Inthesecondapproach, weusedhydrotherenergy enhanced electrode/electrolyte interactionresulting improved materials withcontrolsurface overpore sizedistributionfor synthesizehigh wecouldpotentially in theseconstructors, carbonchainlengthandstericbulk esized that byvarying or 1,2hexadecanediolasstructural We constructors. hypoth- 2 /g) and /g) and 2 /g) using ethylene glycol or neopentyl glycol /g) usingethyleneglycolorneopentyl

Carbon fibers obtainedusingglycols Carbon fibers 2 g -1 . In the third . Inthethird

- - The designhastwouniquefeatures: reactants are mixed. notfully the exactconcentrations duringthereaction whilethe reactants at thereaction temperature isnot knowing mixingthe reaction temperature. of The disadvantage for synthesismixthereactants nanoparticles at the microfluidic reactor designs intheliteraturereported synthesiswasdesignedandfabricated.particles Most A novelmicrofluidic silicon/pyrex reactor for Pdnano- aberration-corrected transmissionelectron microscopy. synchrotron techniquesandexsituatomic-resolution ation withmillisecondtimeresolution andgrowth using nucle nanoparticle - techniquesfor thestudyof analysis data collectionand of microfluidic cellanddevelopment electron microscopy. Theproject willresult inansitu and atomic-resolution aberration-corrected transmission combining microfluidics, x-ray absorption spectroscopy, insolutionby nanoparticles nucleation of andgrowth insitutoolsfor imagingthe new Our goal istodevelop insitucharacterization tools. andpaucityof understanding afundamental rely ontrialanderror from thelackof processes.and growth Consequently, mostsynthesisefforts the nucleation observing experimentally of to thedifficulty due tunable methodbutisnotwellunderstood promising, insolutionisthemost nanoparticles The synthesisof Ayman M.Karim of NanoparticlesinSolution Imaging theNucleationandGrowth • • energy storageandconversionapplications. neered nanostructuresforawiderangeof to makematerialswithspecificallyengi- therefore acceleratesynthesisdevelopment understanding ofsynthesismechanismsand visualization wouldallowafundamental nanoparticles insolution.Directrealtime zation toolsforimagingthesynthesisof This projectisdevelopinginsitucharacteri- the inlet,middle two, and outletchannels. serpentine channelwhere thex-ray windowcovers along An isothermalheated zoneconsistingof and temperature at eachpointintheinletchannel. and, more importantly, provides aknownconcentration mixed. This allowsforuntil fully controlled heating low temperature (where thereaction rate isnegligible) A heat exchanger at theinletkeeps thereactants at others), whichrequiresothers), thinnerwindows. that require Pt,Au,and lowerx-ray energiesthanPd(e.g., other metals be modifiedtoallowthecharacterization of ThemicrofluidicPd nanoparticles. reactor designwillalso ping agentsonthenucleation mechanismsof andgrowth temperature, reducing of andcap- theeffect to understand The work inFY2013willbefocused onsynchrotron studies ACS Catalysis. submitted toaspecialissueof Thework growth. hasbeen leading toanomalousparticle atomic betweenparticles migration of a direct evidence nisms byinsituTEM.Theresults time, showed,for thefirst model systemtostudythestability mecha- andgrowth The narrow provided size distributionPdnanoparticles a withnarrowsmaller Pdnanoparticles sizedistribution. perature. Increasing thesynthesistemperature resulted in onthenucleation kineticseffect andgrowth at room tem- thatthe microfluidic visiblelighthasastrong cell revealed in Pdnanoparticles results for thesynthesisof Preliminary scale inthemiddle andoutletchannels. at thenanoparticles minutes time of as wellthegrowth ation at millisecondtimeresolution intheinletchannel nucle thenanoparticles - This designallowsustoanalyze temperature zonesandawell-definedt at bytheserpentine theinletandheaters channelallowfor two synthesis.Theheat nanoparticles exchanger spectroscopy studiesof Microfluidic reactor for millisecondtime-resolved x-ray absorption resolution duringthegrowth. time resolution duringthenucleation anduptominutes time inlet, twomiddle, andoutletchannels,whichallowsfor millisecond Materials Scienceand Technology 0 . The x-ray window covers the the . Thex-ray windowcovers 149

PN12041/2442 150 PN11044/2366 sample. A strategy on using data reduction of statistical onusing data reduction of sample. Astrategy adual-phasehigh-strength stainlesssteel of integration code wasimplementedand applied tochemical image cal correlation functionwasusedinreconstruction. Source images togenerate alargehigh-resolution image. Astatisti- tion, high-magnification scanningelectronic microscopy cation opticalmicroscopy imageandmultiplehigh-resolu - withalow-resolution,In FY2011,westarted low-magnifi- remediation, andpharmaceutical innovation. production, environmentaltures withvalue toenergy complex,heterogeneous struc- of complete, detailedview asampleprovide a olution imagesthat coverlargeareas of nanometer scaletometerscale. Collectively, thesehigh-res- sible for multiphysics,multiscalesimulation rangingfrom modelwouldmake itpos- thisnew on exascalecomputing, Drawing mechanical phenomenaare likely tobeoccurring. identify the“hotspots”where complexchemical or images. Theresulting imagewouldhelpresearchers revised small,high-resolution andseveral based onasinglecoarse oping theability tosimulate alarge, high-resolution image To attempt making- the impossiblepossible, weare devel sources onmultiplelengthscales. chemical imaginginformation obtainedfrom different light sources, onechallengeraisedishowtointegrate andpowerof chemicalimagingfacility improvement of atom probe tomography, andthelike. Withtherapid ionmassspectrometry,tion massspectrometry, secondary suchashighresolu- development, in-house imagingfacility imaging instrumentation. inour Thistrend isalsoobserved set generated at highspeedtypicalinlightsource chemical largedata- of tomeetthedevelopment must beadvanced for chemicalimaginginstrumentationImage integration Dongsheng Li Resolution Merge Instrumentation byDataReductionand Integrating MultimodalChemicalImaging Materials Scienceand Technology geneous structures. complete, detailedviewofcomplex,hetero- cover largeareasofasampleandprovide oped willcreatehigh-resolutionimagesthat tion andlengthscales.Themethodsdevel- imaging instrumentationatdifferent resolu- bilities forintegratingmultimodalchemical This projectwilldevelopnewresearchcapa- nents of large-scaledata managementandanalysis. nents of techniqueswithothercompo- thedeveloped will integrate Later, pipelineonimagingintegration. tional analysis we imaging instrumentation opera- results afully todeliver insituandexchemical andvisualization of analysis capabilities tothe thedeveloped In FY2013,wewillapply method. nanoprism, andnanoricewasdeterminedusingthis nanorods, ulated heterogeneous systemcomposedof demonstrated inthefigure, asim- componentstructure of high throughput fluctuation x-ray scattering technique. As throughin theheterogeneous the systemcanberetrieved component particles theindividual correlation functionsof heterogeneous system.Usingthisrelation, theringangular autocorrelation functionmultipliedbymolarratios ina single-particle functionisalinearcombination of retrieved Applied onheterogeneous disordered ensembles,the singleparticles. converged toautocorrelation functionsof celed cross-correlations and particles betweendifferent functions calculated from patterns x-ray can- diffraction ringangularcorrelation ing overalargenumber of istic feature structure. determinedbytheparticle Averag- We usedtheringangularcorrelation function,acharacter interesting statistical information todeterminestructure. In FY2012,wefocused onimagereduction toretrieve ing asinputtoreconstruct thehigh-resolution image. withinformationgated, - from low-resolution imagesserv correlation functiontoexpeditecomputation wasinvesti- ensembles usingthefluctuation x-ray scattering technique. structural single-particle determination from heterogeneous Scheme of - bination of electrons in CdSe with holes in P3HT as shown in electrons inCdSewithholesP3HT asshownin bination of P3HT, from therecom- emissionpeakisobserved andanew themainPLemissionfrom bothCdSeand in quenchingof CdSequantum dotsinP3HTresults (PL). Incorporation of and CdSequantumdotsbyphotoluminescence spectroscopy chargetransferbetweenP3HT of spectroscopic evidences enhances thisphasetransformation. Inaddition, wefound ing onSi.Further, decreasing theCdSequantum dotssize structure following dropmation cast- tozincblendecrystal structure, whichundergoes aphasetransfor crystal wurtzite intolueneexhibitpredominantly oxide orhexadecylamine that theCdSequantumdotscapped withtrioctylphosphine dataCdSe quantumdots.Theinsitumicro show diffraction properties of environment ontheelectronic andcrystalline We studiedthestructure-property relationship of andeffect hybridsolarcells. from current examination of butarefrom thequantumdotstoP3HTpolymer unavailable alignmentsthat govern charge-transferdynamics level energy the provide criticalinsightintothefundamentalaspectsof tools.Resultsfrom thisstudywill developed ing andnewly exist- byapplying at themolecularlevel 5 nm)andpolymers sizebetween2and between CdSequantumdots(varying geometrical, chemical,andelectronic structure interaction induced chargetransfer. Thisincludescharacterization of howthesevariations influencethephoto- and understand theCdSequantum dotsinP3HT dependent distributionof systematic approach tostudytheconcentration andsize- matrix. Therefore,interact withthepolymer weadopteda theway inwhichthequantumdots of little understanding for solarcellapplications,studied extensively there exists hybridsolarcell system hasbeen quantum dots-polymer structures solarcells.Although semiconductor for efficient todesignoptimumthinfilm assistefforts immeasurably matrix will polymer (3-hexylthiophene) (P3HT)conductive ed interactionbetweenCdSequantumdotsandthepoly thephotoinduc- efficiency. of Afundamentalunderstanding highpowerconversion matrix isakeyelementinachieving the chemicalinteractionbetweenquantumdotsandpolymer hybridsolarcells, In semiconductorquantumdotspolymer Dongsheng Li Hybrid SolarCells in SemiconductorQuantumDots-Polymer to Probe thePhotoinducedChargeTransfer Integration ofMolecularImagingTechniques transfer inhybridsolarcells. understanding ofthephoto-inducedcharge bilities thatwilladvancethefundamental molecular imagingandspectroscopycapa- We aredevelopinganovelintegrationof

their molecular level interactionsinthe hybrid solarcells. their molecular level CdSe QDsinP3HTand neous distributionof thehomoge- understand strate thenecessityto QDs inP3HTanddemon- CdSe agglomeration of of vious observation images confirmourpre - Both STXMandHIM samples are inprogress. thenew mentation of existing data andexperi- the of ther analyses bon from theP3HT. Fur excess car presence of P3HT suggestingthe region for CdSeQDswith intensity inthepre-edge tra showanenhanced K-edge absorption spec- the CdSeQDs.TheC HDA capped on ligands hydrophobic ence of from theprespossibly - the P3HTandCdSeQDs interactions between pared CdSeQDsandwithP3HTsuggestingweak betweentheas-prespectra shownosignificantdifference - ing toobtainin-depthinformation. TheSeL-edgeabsorption adopting bettersamplepreparation methodsfor STXMimag- QDs waslimitedbySTXMresolution. We are intheprocess of intheas-prepared nanoclusters CdSe individual Imaging of CdSe QDsinP3HTcompared totheas-prepared CdSeQDs. the of STXM imagesdepictedtheagglomeration tendency theCdSeQDsandfollowing interactionwithP3HT. of analyses tion, wecollaborated withCanadianLightSource for STXM vidual quantumdotsat extremely smalllengthscales.Inaddi- indi- these techniquesandchallengesinchemicalimagingof matrix. Results from thesetechniqueshighlightthestrength of tum dotspriortoandafterincorporating theseinP3HT CdSequan- individual lution TEMandheliumionimagingof charge transferbetweenthem.We alsoperformed highreso- varioussizestoestablish thesizedependent tum dotsof P3HTwithCdSequan- We established thebandalignmentof QDs. of Thisalsoresults agglomeration insevere diagram. level energy - - Materials Scienceand Technology b) aftermixingwithP3HT. spectrum from QDsa)before and b) STXMmaps andabsorption chargetransfer forcethe driving of CdSeQDsandP3HTdepicting of varioussizes a) Bandalignmentof

151

PN12044/2445 152 PN10051/2293 aqueous inkscontainingCu-Zn-Sn(OH) has produced uniform films. developed InFY2012,wefurther vapor pressure solventalsoallowedfor spray depositionthat duringthinfilm deposition.Usingahigh evenly evaporates saltsusingahighvapor pressurecost inorganic solvent that copper, of metal nanoparticles from low indium,andgallium FY 2011aprocess tosynthesizeinkscontainingencapsulated . Wecomposition, thickness, anduniformity reported in niques thatproducethinfilmCZThydroxide withtarget Develop metalhydroxide anddeposition tech- nano-inks performance targets. aggressive thisproject requires discoveriestomeet The goal new of configuration asuperstrate based CZTSandapplying intocell. non-vacuum solution synthesisof green cost are developing costly. Two approaches attractive todecreasing production and thedepositionprocess complex,timeconsuming, ders great promise,Despite offering CZTS’s complicated nature ren- CIGS. small cellconfiguration, whichisclosertothe record of (CIGS). Recently, inthe CZTSdemonstrated 11%efficiency thin film solar technologies,copper-indium-gallium-selenide for theleadingcontenders potential replacement for oneof abundant earth materials have emergedasa composed of PV cellsbasedonthinfilm copper-zinc-tin-selenide (CZTS) John P. Lemmon Photovoltaic Manufacturing Materials andMethodsforLowCost to minimize the use of organic solventandreduce organic experiment to minimizetheuseof cles from alow-cost,low-temperature hydrothermal process confirm the composition of Cu,Zn,andSn. confirm thecomposition of droplets. c)showsEDX of contact angledifferences show PEImodifiedZnSsubstrates, andinsets reveal inkdeposited onZnS(a).b)andd) CZT particles Materials Scienceand Technology carbon emissions. dence onpetroleumimportswhilereducing significant todecreasingournation’s depen- vs. currentcostlyhighvacuumprocessesis photovoltaic (PV)manufacturingmethods Developing newsolution-basedthinfilm 6 (CZT)submicron- parti CZT inks. from aqueous form thinfilm crack-free, uni- we produced substrates, modifying by surface ink. However, the aqueous strates using ing onthesub- uniform coat- was achieving A key challenge requirements.

and costimpacts. mance testedinPVcellstodetermine overallefficiencies CZTS microstructure optimizedandperfor willbefurther movesintoPhaseII,where In FY2013,theprogram the and accurate communication withpotentialPVclients. capability performance data willallowfornew publication of outdated irreproducible systemthat yieldedhighly results. The measurement system waspurchased toreplace aninaccurate, product formation. Inaddition, aturnkeyPVperformance reactions changesinreaction for new oremissivity surfaces perature inchemical differences changesfrom free energy tomonitorreal tem- thermal imagingcamerawasdeveloped monitoring chemicalreactions andprocesses witha3–5µm An large format cells. andtodevelop efficiencies focused onoptimizingthinfilm processing toyieldhigher that hinderelectron isnow transportation. Theprogram size andvoidsare generated infilm duringselenization be explainedbythefilm microstructure, grain where small CZTS PVcellsfrom Thehighresistivity could superstrates. is high,theserepresent attempts toproduce ourfirst full mance testshave beenconducted.Althoughtheresistance ated from PhaseImaterials. Fullcellfabrication andperfor PVcellsgener fullsuperstrate of and performance evaluation approach, significantprogressfabrication wasmadetoward depositonthesubstrate. successfully Basedonthesimilar window layer andhighperformance TCO layer have been results frombased CZTSmethods.Pervious FY2011,CdS cellusingsolution- Build asmall-scaleCZTSsuperstrate amenable PVcellworkup. to further FY 2012ledtodenseabsorber films with~5µmthickness issue andoptimizeexperimentalconditions.Selenization in applied Sevapor totheprocess toaddress onsuperstrates this performance.with metalcontactsanddeteriorate We device dled usingappropriate equipment)butalsoreacts protective PV cellwithgold electrodes. middle:Left: asdepositCZTsuperstrate; afterselenization; right:full However, H vapor andselenization growth wasused topromoterate. grain . Previously,films withoutinfluencingthesuperstrate H Optimize selenizationconditiontoachieve desired CZTS in situopticalreactor using asapphire windowfor 2 Se vapor is not only highly toxic highly (andmustbehan- Se vapor isnotonly

Se Se

- window (left)toa meta-material coated window (right). daylighting from astandard Schematic comparisonof nant dipintransmissionasexpectedfor singlewavelength structures were fabricated that demonstrated astrong reso- from fabricated structures. AlsoinFY2011,preliminary measure boththetransmissionandfullscattering response addition, anoptical spectrometer wasemployed that could thesenanostructures.formed topredict In theresponse of numerical modelingsimulations were per Complementary tofabricate theappropriatedeveloped nanostructures. In FY2011,nanofabrication toolsandexperiencewere rapid results. todeliver efforts modelingandfabrication a synergisticcombination of challenge overthefullvisiblespectrum. Theapproach used structured coatings toredirect light,andtoaccomplishthis project nano- are todemonstrate theproof-of-concept of this of usage. Thus,theoverall objectives for buildingenergy savings. Thiswouldbeaproverbial game-changer energy low-cost nanostructured coatings couldalmostdoublethe intoabuildingto30feetwithpotentially further light even energy.potential tosave 1quadof Theability toredirect 15feethave the daylighting schemeswithpenetration of the commercial buildingsectorat about 3.5quads.Standard electricityconsumptionin accounts for of thelargestportion technology.ing usingpassive Interiorlightingcurrently consumption for buildinglight- significantly reduce energy low-costnanostructuredtive meta-material coatings would lightthroughThe ability innova- to tailortheredirection of Kyle J.Alvine Windows Meta-Material CoatingsforDaylighting over standarddaylightingwindowsystems. nearly 11milliontonsofcarbonperyear 0.66 quadrillionBTUsofenergyandoffset new technologyisprojectedtosaveup ficial lightingenergyuse.Ifsuccessful,this nificantly furtherintobuildingstooffset arti- window glasscoatingstobringdaylightsig- This projectisdevelopinglowcostadvanced structures. nano- of these types ble from was possi- redirection cant light that signifi- showed results Modeling structures. - novel technology is very promising. is very novel technology light redirection performance. Overall,theoutlookfor this numerical improve modelingusedasaguidetorapidly packing density. Future work willaddress theseissues with andimproved metalliccoating, improved conductionof ment toelicitimproved response: defectreduction, areasstructures. have beenidentified Several for improve- thefabricated modeling andthecurrent performance of the predicted redirection performance from numerical Additional future work between isneededtobridgethegap visible lightspectrum. shouldallowforthat lightredirection they overthefull thestructures describedabove andshowed the designof were alsousedtoguide modelingefforts Complementary that broad-band response ispossiblewiththesestructures. the spectrum, thisnovelresult nonethelessdemonstrates performance andtoshifttheresponse totheblueregion of refinementsfurther are neededtoimprove light redirection thenearinfrared lightspectrum aswell.While of portion thevisiblelightandasubstantial 70%of roughly covers from approximately 500nmupto1100nm.Thisresponse optical scattering setupandshowedabroad-band response These testsampleswere measured subsequently withthe improvements inlateral packing anddefectreduction. improve performance. There isstillroom for substantial thestructuresnique withahighlateral packing to ratio of fabrication this tech- wasputintodeveloping ficant effort process. Signi- self-assembly were fabricated usingthisnew Multiple testsamplesbetween1 inFY2012. modelingefforts and guidedbycomplementary structures inFY2011 were basedondesignsdeveloped These assembled nanotemplating andmetalsputtering. 750 nmwere self- fabricated through acombination of wavelengths from 400nmupto tion overabroad rangeof structures that were light redirec predicted tobecapable- of To thisgoal experimentally, nano- achieve typesof new setup for improved collectionperformance. toward improving theopticalscattering measurement duringthecurrent fiscal some effort year was redistributed light spectrum for accurate colorrendition. Inaddition, structures that redirect lightoverthefullvisibleorwhite- response iscriticaltooptimalperformance bycreating across thefull visiblelightspectrum ispossible. The nanostructures abroad-band spectrum capable response of both withmodelingandexperimentthat fabrication of In FY2012,theproject wasfocused ondemonstrating Materials Scienceand Technology ” ×1 ” and1 ” diameter 153

PN11051/2373 154 PN10060/2302 containing different microstructuralcontaining different interest in features of ties for generating 3Dandmeshingsyntheticmicrostructures During FY2011,weexpanded automated software capabili- representations. using3Dgenericmicrostructure began Mg alloys grains for elastic-vosciplastic materialparameters modeldescribing tomaterial sensitivity microstructures. Aparametricstudyof algorithm for generating 2Drealizations basedonMgalloy ior. Atthe microstructure apreliminary scale, wedeveloped thermo-mechanical properties aswellMg-Alalloy behav- pointdefectand descriptionof tial provides asatisfactory Based onthemodifiedembedded atom method,thepoten- for Mgalloys andchoseapotentialfor oursimulations. Previously, theavailable interatomic potentials wesurveyed needed capability at PNNLinmaterials research. add creating valuetomaterial modeling efforts, amuch- thismethodwould of Development scale material modeling. priate at themicrostructure scale, anddemonstrating multi- material modelsappro developing computational- analysis, capability toincludebuildingsyntheticmicrostructures for rules that canguidealloy develop selectionandcreate anew ulations, data from literature, andexperimental results to tocombineinformationwe willstrive from multiphysicssim- beat approach However, andcumbersome. that isexpensive Traditionally, hasfollowed aheat-and- alloy development the elements. appropriate temperature characteristicswithoutuseof Mgalloy compositionsthatit isdesirable have todevelop foreign insensitive material markets, dependsonsuppliers rare-earth elements.Becauseobtainingthis the addition of magnesium (Mg)alloys are promising butwouldrequire itsweight. Lightweight tution isintheenginebecauseof through lightweightalloy substi- improving vehicleefficiency be justifiedbythehighercost.Themostpromising area for lightweightalloys, with expensive theweightsavings cannot cars mance. Whileonecouldreplace thebodypanelsof perfor weight reduction relative tosteelwithoutlossof 50% gies; for targetof example, DOEhassetanautomotive lightweightalloys for transportationdeveloping technolo- material science problems. There interest isgrowing in Multiscale, multiphysicsmodelsareapplicable widely to Erin I.Barker and ApplicationtoMagnesiumAlloys Multiphysics CapabilityDevelopment Materials Scienceand Technology and lightweightalloysfortransportation. cially inadvancednuclearenergysystems els relevanttoenergytechnologies,espe- and demonstratemultiphysicsmaterialmod- The primaryaimofthisprojectistodevelop -

analysis of results. of analysis canleadtoabetter Visko withscientifictoolpackages tools were incorporated intoVisko andhowcombining theprocess. Apaper isinpreparation of discussinghowthe available toalargercommunityandimprove automation concept project toshowhowthetoolscan bemadereadily were incorporated intheVisko asaproof-of- webframework samples The toolsusedtogenerate andmeshthepolycrystal and apaper discussingtheseresults isinpreparation. were presented at theInternational SymposiumonPlasticity, mismatch dominates theoverallresponse. These results in thestress-strain response, indicating that localtexture were utilized.Stronger textures resulted inmore variation specifictextures butindividually equivalent statistically of thesample. Aseries tion intexture to theglobalresponse of varia- texture wascompleted that quantified theimpactof theimpactof and outputfile formats. Aparametricstudyof additional material models,andmodifications totheinput includingperformance enhancements, All-Wheel Drive), FY2012,updates wereFor madetoFEAWD(FiniteElement asellipsoids. sented thesamegeometrically Inclusionsandpores are repredecohesion andsliding. - elementsfor modeling interface allowing for of theinsertion berespected, interface) andpore/inclusion-grain as grain meshingisthat internalboundaries (such A requirement of thesamples. element meshesare generated of for analysis has alsobeenapplied tosinglematerial samples.Also, finite interest. pores Therefore,also anarea of of theinsertion grain structure where adefined isnotpresent castMg, is of porosity ontheductility Theinfluenceof within thegrains. sions are represented placed asellipsoidsandrandomly sized pores orinclusionswasadded thisyear. Pores orinclu- sub-grain samples are beinggenerated. of Theinsertion 2Dor3Dpolycrystal equivalent tion distribution,statistically andanorienta- grains sizeornumber average of grain Given these features improves orreduces response. features influencematerial response andhowvariation in studiestoisolate howindividual parameter andsensitivity syntheticmicrostructure samples,wecanconduct suite of size, andorientationand grain distribution)andcreating a microstructural features (suchasbulkporosity, average pore experimentation. Byextractingstatistical information about ments, weare supplementingthephysicalwithdigital attempt toreproduce experi- results from alimitedsetof than self-limitingtechnique. Rather it isatime-consuming, simulation results asingletest, andmaterial modelsagainst microstructures observed canbeusefulfor validating tally While reconstructing experimen- digitalrepresentations of Mg alloys that canbeutilizedfor finiteelementanalysis.

and thermo-mechanics. For energy storage devices, several several storage devices, energy and thermo-mechanics. For problems influiddynamics typesof for several developed neous multiscalemethods.Such techniqueshave been addressing intoheteroge suchmultiscaleproblems- fall of behavior. capable Thecomputational methods potentially transport theircollective computational prediction of ing uniquedemandsontheirtheoretical descriptionand plac- anisotropy andspace chargeeffects, crystallographic solid-state are electron subject tostrong andiondiffusion trons intransitionmetaloxides are polaronic innature, and elec- On atomistic lengthandtimescales,charge-carrying lengthandtimescales. range of across awide thedemandtoincludeeffects because of for thissystemischallenging Design principlesdevelopment network properties such asdensityandporosity.grain properties, and sizesandboundary metal oxides, grain properties of chargetransport turn are linked todiffusive distances,whichin contactareas,electrolytes, anddiffusion electrode materials and and acomplexinterdependence of onelectrode-electrolyte interfaces nanosizingeffects ing of - major outstandingunknowns,includingpoorunderstand Improvedstrategies. predictability remains complicated by currently notheoretical toguideoptimization framework andthere improved is overallperformance isnotuniversal, distancesfor lithiumionsandelectrons.diffusion However, andelectrodebetween electrolyte solid-state andshorter contactareasto bulkmaterials duetohigherinterfacial performance andcyclability relativeimproved conductivity batteries. Itisthoughtthat nanostructured electrodes yield conversion substrates and energy storage devices suchasLi storagedevices substrates andenergy conversion yet appear toholdgreatmental level promise asenergy erogeneous materials that toprobe are difficult at afunda- Nanostructured transitionmetaloxides are disordered het - Maria L.Sushko Storage Materials Multiscale ModelingofEnergy significantly enhancedperformance. guide novelelectrodematerialdesignwith for ionandelectronconductivitywould trodes. Adeepunderstandingofthebasis transport inpolycrystallinemetaloxideelec- micron scalesforcoupledionandelectron charge transportdynamicsfromatomicto performing first-eversimulationsofcollective transport inenergystoragematerialsand open sourcesoftwareformodelingcharge simulation techniqueandnovelefficient We aredevelopingahierarchicalmultiscale +

rials suchasLiNi and electron inselectmetaloxide transport electrode mate- coupled ion a micron scale;andstudyingthemechanismof informed byPNP/cDFTfor simulations chargetransport on aPhaseFieldmodel computing environment; developing (PNP/cDFT)software for highperformance tional Theory ware basedonPoisson-Nernst-Planck/classical DensityFunc- simulation multiscalechargetransport soft- performance of focused onthreeOur FY2012activities areas: optimizingthe storagematerials. energy hierarchical multiscalecodefor theoretical modelingof thefirst scales. Theprojectgap aimstofill anddevelop this phenomenafromcharge transport atomistic tomicron there are multiscalemodelsdescribing currently notruly multiscale modelslinking smallerlengthscales.However, andfor novelelectrode materials.devices Thelatter requires storage thesemodelsasinsilicodesigntoolsfor energy of obtained experimentally. thepredictivepower Thishinders field modelsthatfor electrodes onparameters heavily rely models suchasfinitedifference, finiteelement,andphase Thesecodesare basedoncontinuum chemical devices. electro electrodes- withmacroscopic performance of ties of models existthat linkmicroscopic proper chargetransport used. Analysis shows that only boundary values need to be valuesneedtobe boundary showsthat only used. Analysis redundant computation, andaparallelJacobimethodis ing equations, modulization stepsare employed toremove approximation successive solv- Picard isused.For methodof removed through of modulization, andaparallelversion rated modules. Redundant computation intodifferent was chemicalpotentials,computational stepswereciently sepa- method.To- equations byNewton-Raphson computeeffi chargedspecies)usingthecDFTandsolvingPNP tials of chemicalpoten- of cal multiscalemodeling(i.e., evaluation There are twomajorcomputations involvedinthehierarchi - via cDFTcoupledwithPNPformalism for dynamiceffects. interactions are introduced on themesoscale(10–500nm) long-rangeelectrostaticThe collective andexcluded volume coupledionandelectron diffusion. simulations of energy andstate-of-the-art moleculardynamicsfreetransport charge elementary quantum mechanicalsimulations of embedded cluster This modelmaximizeshighaccuracy finerscalemodels. of effects interactions withshort-range long-range collective weintroduce noveltypesof upscaling, formalphysics approach consecutive where insteadof storagematerials.structured Themodelusesamulti- energy for modelingcoupledionandelectron innano- transport cDFT methodfor hierarchical multiscalemodelingtechnique We aparallelprocessing approach developed basedonPNP/ 0.5 Mn 1.5 O Materials Scienceand Technology 4 cathodes andTiO 2 anodes. -

155

PN12051/2452 156 PN12051/2452 cies on the structure of LiNi cies onthestructure of oxygen vacan- of theeffect mechanics simulations toreveal this end,weemployed molecularmechanicsandquantum importance.To particular isof on charge/dischargecurves vacancies onLiNi oxygen properties inthebulkmaterial.port Therole of knowledge of the elementary coupledLi theelementary knowledge of LiNi themultiscaleapproach tonanostructuredApplication of density andcapacity duringdischargingandcharging. ties, chargerates, current andstresses onvoltageversus neous, andanisotropic thermodynamicandkinetic proper microstructure, of inhomoge- andtheeffect and electrolyte double layer at betweensolid theelectrochemical interface predict chargeseparation a associated with theformation of neous medium.Theresults demonstrate that themodelcan inhomoge- andelectrically and electricfieldsinanelastically in themodel.Iteration methodsare usedtofindtheelastic deformation are duetoLi-ioninsertion/extraction considered andstresses chargedparticles, associated withlattice ties of mogeneous andanisotropic dielectricconstantsandmobili- taken asamodelsystemtotestthephase-fieldmodel.Inho- porous microstructure nanoparticle filled is withelectrolyte mobilespecies. A tions, andtheotherisconcentrations of size,describing nanoparticle orientation, andspatial distribu- microstructure. theorder parameters Oneiscomprisedof fieldvariableselectrodes. are Two usedtodescribethe setsof innanocomposite transport conditions onchargedparticle andkinetic propertiesture, andcharging thermodynamic, of microstrucphase-field - approach theinfluence investigates scalephasefieldmodel.Thedeveloped inform acoarser are usedto onthePNP/cDFTlevel free energiesevaluated To the accesslargerlengthscalesandincludeelasticeffects, anisotropic dielectricconstantsanddiffusivities. with nanoparticles neous environments suchasnetworks of functionalitiesthat allowsimulations inheterogewith new - the HPCenvironment. The3DPNP/cDFTcodewasenhanced computingnodesin and nearlinearscalability for number of onasingleprocessor workstation similaraccuracy results of 20-fold speedup compared totheJacobimethodachieve unknowns. Resultsalsoindicate amore than number of has closetolinearcomputation complexitywithrespect to method Numerical results showtheimplementedmultigrid equations differential tosolvePNPequations.elliptic partial for solving methods,themostefficient we applied multigrid To improve PNP/cDFTsoftware computingperformance, high performance computingenvironment wasachieved. themethodwithrespect in toprocessors linear scalingof method.Thealmost thedeveloped modeling capability of parallel PNP/cDFTmethoddemonstrated themultiscale storagematerialsnanostructured usingthedeveloped energy Numerical results oncoupledionandelectron in transport overrelaxationcessive scheme)were alsoinvestigated. suc- parallelJacobimethod(e.g., of speeding upstrategies implementation suitable for solvingthe problem. Possible communicated loops,whichmakes betweendifferent Jacobi Materials Scienceand Technology 0.5 Mn 1.5 O 4 spinelcathode materials requires adetailed 0.5 Mn 1.5 O 4 ion and electron conductivity and and ionandelectron conductivity 0.5 Mn 1.5 O 4 spinel.We showed that + andelectron trans-

- process (i.e., Li gen vacanciesinLiNi Ab initiosimulations neutraloxy suggest that- formation of with experimentalobservations. oxygen vacanciesare introduced. Theseresults well agree to electrostatic couplingbetweenLi under aseparate project from FY2010.Usingthismode, due approach basedontheKineticMonteCarlo modelinitiated We continued work multiscalesimulation onanalternative and electron conductivity. and Ni sitefrom 4+to~3+ Mnionsadjacenttothevacancy state of voltage degradation offers away toimprove LiNi offers voltage degradation Understanding duringelectrochemical cycling. voltage fade lowvoltageregions,whichexplains increases thewidthof vacancies Theconcentration of voltage-capacity curves. arereduced. responsible Theseeffects for lowvoltagetailon electricfield at theLi Ni andMncations, theeffective siteonto twoelectrons from thevacancy to redistribution of higher degree of symmetry (P symmetry of higher degree configuration has for stoichiometric spinel,thelowestenergy ity inrutile TiO storage devices with high-power and -energy densities. withhigh-powerand-energy storage devices oxide materials for next-generation electrochemical energy nanostructured electrodes withotherpromisingdesign of significant importance for the lation results couldbealsoof nanostructured electrode material. of Thesimu- conductivity correlated flux,whichisbeneficial forhigher capacity and strongly beforefuse wellinsidethenanoparticle theonsetof localized carboncoating allowsfor ionsandelectrons todif- trons, whichlimitsoverallmaterial capacity. Conversely, siteswithionsandelec- surface leads tothesaturation of theelectrostatic fullcarboncoating, coupling For coating. contact area andcarbon between thetitaniananoparticle withthe chargetransport for changingthemechanismof cies significantly reduces local barriers for Li cies significantly reduceslocalbarriers oxygenduring cathode vacan- delithiation. Thepresence of ordering instoichiometric LiNi of Ni/Mn aspecifictype significant). Thus,dominanceof (i.e., allpossibleNi–Mnarrangementsbecomestatistically and disordered range configurations occupythesameenergy room temperature. Incontrast,inreduced spinelordered ing that stoichiometric spinelwillhave aP thananydisordered configuration,lower inenergy suggest- 16d sites.Moreover, theordered configuration was0.4eV ordered ordering over 4b and concentration of oxygen vacanciesonmesoscopicLi and concentration of nanostructuring spinel are therole underway of toreveal addition, multiscalePNP/cDFTsimulations for LiNi thesepromising cathodecommercialization materials. In of along-standingproblem that effect, hinders voltage fade vacancies. Theseresults provide insightintotheoriginof that for increase electron barriers extractionnearoxygen stability through nano-structuring orbyintroducing dopants cathode material performance through controlling crystal 2+ ionstoNi Fd3 – m spinels with Ni and Mn randomly occupying occupying spinelswithNiandMnrandomly + 2 diffusion). The coupling effect isresponsible Thecouplingeffect diffusion). nanoparticles is determined by the slowest isdetermined bytheslowest nanoparticles sites,whileallotherconfigurations are dis- + state, electron whichfacilitate extraction 0.5 Mn 1.5 O 4 4 leads to a change of oxidation leadstoachangeof 0.5 3 32 space group) due to Ni duetoNi 32 spacegroup) Mn + 1.5 ande O 4 is lifted once neutral isliftedonceneutral 4 - - , totalconductiv 3 + 32 structure at extraction: due extraction:due 0.5 Mn 0.5 Mn + 1.5 site is siteis O 1.5 + 4

num oxide templates. The length and width of columnsiscontrollable.num oxide templates. The lengthandwidthof using anodizedalumi- electrochemically Silver nanocolumns grown materialsbandgap coated onmetals;andsecond, ites, usingnovelsemiconductor, mixed-metal, andwide- nanostructures (seefigure), andcluster-based nano-compos- photocathode materials pillar anddesigns,suchasvertical innovative develop Theproject willfirst ode degradation. photocath- themechanismsof durability byunderstanding UTEM. Specifically, we willdirectlyaddress photocathode monic designstoenable fourth-generation lightsources and semiconductor,inventive mixed-metal, composite, andplas- photocathode sources using new expect todesignseveral intense electron yieldsandultralowthermalemittance. We robust photocathodeThis project sources willdevelop with contribution tothisessentialactivity. science canandshouldmake animmediate andtimely materials science, solid-state photochemistry, andsurface addressing thisissue. Thefieldsof in theinitialstagesof andis photocathodeentific development approach tonew light source communityrecognizes theneedfor amore sci- downstream accelerator orFELdesign.Theinternational simplifying construction bysignificantly costsdramatically reducephotocathode designscouldpotentially lightsource cathode brightnessandemittancecharacteristics.Novel compact lightsources, are dependentuponphoto- heavily next-generation ultraviolet (UV)orx-ray sources, including ring designs,theultimate outputcharacteristicsandcostof UTEM. Whetherbasedonfree electron laser(FEL)orstorage fourth-generation lightsources andtodevelop needs of tomeetthe mustbe advanced Photocathode technology Wayne P. Hess Generation LightSources Photocathode DevelopmentforNext- electron microscopy(UTEM). development ofultrafasttransmission significantly reducedcostandenablethe tion offourthgenerationlightsourcesat photocathode designstoenabletheconstruc- The goalofthisprojectistoproducenew

the XPSinstrument intothe10 getterpumptoreduce basepressuresnon-evaporative of lampexcitation sources(broadband andinstalled mercury) We laser(3.8eVheliumcadmium)andUV interfaced for nano-columngrowth. the wetchemicalmethodology multi-source thinfilm depositionchamber, anddeveloped a example, wecompletedconstructionFY 2012.For of in2011ledtoconsiderable progress in Developments generation light sources. emittance, andoperational durability required bynext- demonstrate highelectron bunchintensity, lowthermal of intraband defects. of and photoemissionmeasurements todetermine therole hybridmaterials. Finally,the utilityof wewillusemodeling wewillmeasure thermalemittancetodemonstrate UPS, performed onotherhybridmaterials. Usingangle-resolved validategrowth ourthinfilm results, experimentswere also thesecandidate photocathode systems.Toproperties of electronic themodeling of of will benchmark theaccuracy thetheoretical andexperimentalresults Comparison of Bratoms desorbedfollowing UVexcitation. maximum of resolved DFT. canpredict thekinetic energy The theory (DFT)andtime- embedded cluster densityfunctionaltheory thehybridsystemsare tobecalculated using energies of Ag:MgO thinfilm structures. Theelectronic properties and Cu:CsBrand In FY2012,weinitiated ab initiomodelingof quantum yield. The UPSresults confirmedthepredicted enhancementin largeelectron yieldswere produced. over oneeVandvery The work cleanCu,by functionwasreduced from that of Cu(100) is4.5eV. crystalline Theworkactivation. functionof thehybridsystempriortophoto- sure thework functionof lamp(<4.1eV)wasusedtomea- 24 h.Afiltered mercury after photoemissionsignalisobserved 20.Nolossof tor of wasenhancedbymorefac- thana The quantumefficiency thehybridphotocathode couldbeenhanced. of efficiency In late FY2012,wetestedthehypothesisthat thequantum troscopy constitutedaproject milestone. byUPSandXPSangleresolvedgation photoemissionspec- initial and weconductedXPSUPSonthehybridsamples.Our angular distribution).Ahybridphotocathode wascreated, electron yield(bunchintensity)andemittance(electron candidate photocathodes, measure including properties of Next,wewereoxide able matrix to viaOstwaldripening. plasmonic silvernanoparaticles were formed withinthe ionbombardment. Further,a MgOmatrix usinghigh-energy within andimplantedsilvernanoparticles column growth results, includingthesilvernano- plishments ledtonew growth of coated photocathodes andinterro- of in situgrowth Materials Scienceand Technology -10 torrrange. Theseaccom-

157

PN11061/2383 158 PN11064/2386 Nine experiments were carried outwithfour compositions. mix thehard andthesoftphaseswithout decomposition. theprocess canhomogeneous 1 μmto~100nm,and4)if grain from theprocess > canrefinetexture, thecoarse 3)if reach 100%density, theprocess cancreate crystalline 2)if theprocessed materialscal for can magnetfabrication: 1)if have four focusedOur efforts questionscriti- onanswering DOE support. evidence andseek sufficient istogather oped. Ourstrategy - devel system, itisimpossibletohave themethodfully andadependenceonmaterials processing parameters of x-ray, EBSD, SEM,andmagnetometry. Withthecomplexity process andperforming routine characterizations suchas ysis. PNNL’s thefrictionconsolidation role isdeveloping to provide raweffort materials andperform EELS-TEM anal- Corporationtron (EEC)are Energy twocollaborators inthe andElec- toprove thisconcept.TheAmesLaboratory effort thisproject of nanocomposite magnet.Itistheobjective exchange-coupling potential tobeusedfor fabrication of texture. Ithasthe of grains andhighdegree duce fine The frictionstirweldingprocess hasbeenshowntopro- precursors. nanoparticle ment of processing have significantchallengesduetothe require- sucharchitecture.develop Current conceptsfor solid-state ble pathway usingtheknownsolidification processes to magnets. However, thereaccessi- isnothermodynamically theart thestate product of more thantwicethatenergy of specific nanostructured architecture hasatheoretical aligned andwithauniform soft/hard This phaseinterface. to produce fully a100%densealloy withnanoscalegrains areliable andeconomicalprocesschallenge istodevelop magnetic properties withouttheneedfor criticalREE.The superior composite magnets,itmay bepossibletoachieve nanostructured Through creationtromagnetic devices. of usedfor product elec- (58MGOe)andiswidely high-energy state-of-the-art Nd isdirectlyproportional tomagnetstrength.conversion The energy of are needed.Knowingthis,theefficiency urgently product orwith reduce rare elements(REE)content– earth and Dy, magnets–eitherwithhigherenergy alternative criticalelementsNd coupled withlimitedavailability of inelectricvehiclesandwindenergy With expectedgrowth Jun Cui Rare Earth-lessPermanentMagnet Materials Scienceand Technology security andtechnologyleadership. benefit theUnitedStatesinbothenergy tion ofnextgenerationmagnetmaterialto dation methodthatwillenablethefabrica- We aredevelopinganewsolid-stateconsoli- 2 Fe 14 B compound exhibits exceptionally B compoundexhibitsexceptionally materials. include permanentmagnet, softmagnet,andstructural nanocomposite fabrication. Ourtargetapplications of Our final goal istoestablish PNNLastheleaderin field method. thefrictionconsolidationhub proposal for developing we proposed anindependenttaskinthecriticalmaterials Researchto Advanced Projects Agency-Energy. Moreover, demonstration program weproposed atechnology the EEC, net. BasedonthesefindingsandtogetherwithAmes required tofabricate nextgeneration nanocomposite mag- tion, themethodisproven, holdingallkeycharacteristics idation method.Thoughstillneedingsignificantoptimiza- demonstrated textured nanostructure usingfriction consol- for thisproject inFY2012isthatThe key achievement we can pushthecoercivity to benchmark. improvedfurther to>80%.Consequently, we webelieve thetexture canbe ferricsteelplates, webelieve welding of isaligned.Basedonourexperiencewithstirfriction grain the showsaboutthe goal 41.3%of (12kOe).EBSDanalysis of short eightruns butwhichstillfalls over theprevious method. Itscoercivity is6kOe, asignificantimprovement benchmark prepared usingtheconventionalsintering properties isfrom Run9.Itsmagnetization iscloseto the heavier texture. To date, thesamplewithbestmagnetic andlongerprocessingparticles, time(>240s)produces size(50~200μm)ispreferredlarger particle oversmaller irregular shape withsharp edgeispreferred oversphere, size/shapeparticle andprocessing time. We found that and shape. twoparameters: InFY2012,weinvestigated speed, processing time, ambientenvironment, powdersize, theresulted samples:plungeloads,rotationproperties of on microstructure magnetic affecting andconsequently were processing found parameters havingSeveral effects results are summarizedinthefollowing table. tents are characterizedusingVSM,SEM,andX-ray. The The magneticproperties, microstructure, andphasecon- of-the-art aberration-correctedof-the-art S/TEMandother micros- byextendingthe state-applications. Thiswillbeachieved materials andenvironment for nextgeneration energy under insitu/operandoconditionandcreate much-needed andenvironment materials energy of scopic analysis capabilities imagingandspectro- atomic-level that drive chemicalimaging thisproject istodevelop The goal of information. complementary instrumentations,platforms thereby providing of atparticularly thesamesitewhileacross different complex systemsunderdynamicoperating conditions, ties that insitumeasurements enable of atomic-level toolsandcapabiliestablished.- Itisthuscriticaltodevelop sis underdynamicoperating conditionsremains tobe - imagingand spectroscopic analy copy [SPM]),atomic-level atom probe [APT],andscanning probe tomography micros- rected scanning/transmissionelectron microscopy [S/TEM], (suchasaberration-cor beusedroutinely that cantypically toalevel high vacuumconditionshave beendeveloped Though structural andspectroscopic information under its operation, process acatalytic duringsystemfunction. during materials inabattery active of structural evolution nisms underdynamicoperating conditions,suchasthe materials istocapture mecha- andvisualizeatomic-level human-made systems.To energy/environment understand andability tocontrollimit ourunderstanding natural- and chemical andbiologicalreactive sitestructure andfunction tools andcapabilities tomeasure moleculardetailsof control/manipulate material properties. However, existing interest theability asthis,inturn,renders to materials of controlled byprecise, characterization of atomic-level inscientificinnovationsprocesses. isoften Advancement chemicalandbiological mental issuesinvolvecontrol of mental scientificproblems andenviron - related toenergy mankind andare centraltoDOE’s mission.Manyfunda- andtheenvironment are keyissueschallenging Energy Dongsheng Li Multi-Instrument Approach and ExSituChemicalImagingBasedon and/or OperandoConditionsUsingInSitu Structure-Property RelationshipUnderDynamic Site SpecificAtomicResolutionProbing of for theenergysecurityandefficiency. advances anddiscoveryofnewmaterials for chemicalimagingthatwillleadtothe This projectisdevelopinginsitucapabilities - and discharging. charging followingmining whylithiumbatteries cyclic fail community,the lithiumionbattery whichinclude deter forus auniqueopportunity probing intochallenges facing dynamic process. Essentially, this capability willprovide andtheelectrochemicalin liquid,solid-liquidinterface, nucleationphenomenon suchasnanoparticle andgrowth extends ourimagingcapability toallowprobing into under liquidenvironment). Thewetelectrochemical cell sample sample) andbiasingliquidcell(enables biasingof liquid in twomodes:plainliquidcell(allowsimagingof We wetelectrochemical cellthat canbeused adeveloped high temperature. environment materialscatalytic at underagaseous process underrealistic conditions, suchasimagingof thechemical This uniquecapability of enables observation with pressure to20mbarandtemperature to1200°C. resolution underoxidizing/reducing environment gases ronmental TEM(ETEM),whichprovides anAngstrom level capability,we established anew aberration corrected envi- resolution. Additionally, with0.3eVenergy atomic level materials tostudyenergy-related atthe opportunity the high spatial resolution. TheEELS andSTEMimagingoffer dinate chemicalenvironment around aspecific atom with chemicalidentity,ing of electronic structure, andthecoor spectroscopy (EELS) overthelastyear that enables theprob- loss electron energy thecapability of wedeveloped effort, S/TEMandAPT.of S/TEM continuing ourprevious First, capabilities across theplatform of andintegration catalysis, storagematerials,ried outfrom three aspects:energy andtheenvironment. Ourwork wascar energy lenges of these capabilities toprobe thefundamentalscientificchal- chemicalimagingcapabilitiesadvancing inPNNL.We used keyaccomplishmentsfor During FY2012,wemadeseveral microscopy (FIB/SEM). S/TEM, APT, andfocused ionbeamandscanningelectron information obtainedfrom varioustechniques, including site-specific atomic resolution structural andchemical of integration complementary tosupport be developed capabilitiesunder dynamicoperating will conditions.New atomic-resolution imagingandspectroscopic analysis enabling level, copy andimagingcapabilities toanew Materials Scienceand Technology

- - - 159

PN11070/2392 160 PN11070/2392 such as GM and Applied Sciences to understand thefunda- such asGMandAppliedSciencestounderstand collaborating research withleadingindustries activities thelithiumionbattery.safe operation of We alsoinitiated carbon toallowhighcapacity, life, andthe longcycle materialsmation basedonsiliconand for tailoringnew profile journalpublications, whichprovide essentialinfor carbon nanocomposite. Thesestudiesledtothree high silicon/siliconand we studiedthelithiation behavior of UsingourinsituTEMcapability, lifecycle. battery short con, thelatter willexpand~300%,whichcausethe However,tronic devices. lithiumintosili- wheninserting which iscurrently usedinrechargeable batteries for elec- graphite, ~15timesof silicon hasastoragecapacity of silicon.Intheory, storagecharacteristicsof ied theenergy storagedirection, westud- For energy scientific questions. capabilities, weprobedWith thesenew intothefollowing across instrument different platforms. information tion conditionsandcapturing complementary capabilities materials enable underreac probing- energy information. Thesenew platform andyieldcomplementary asampleacross instrument different a specifiedsiteof FIB,platform APT, of andS/TEM,whichare keyfor imaging the capability asampleacross the that enables imagingof materials. Atthesametime, weenhanced energy studies of insitu thefieldof whichadvances the globe, theresult of used byothernational across laboratories anduniversities materials. Our insituconcepthasbeenimplantedand nificant impactonthe research community for energy inthisprojectcept andcapabilities have asig- developed research. Thefundamentalinsitucon- and sulfurbattery lithium metal,anessentialcomponentfor lithiumion,air, samples.Thiscapability air-sensitive allowsustohandle of environment for gas We aninert handling alsodeveloped reconstruction of g-Al reconstruction of structure interaction.The surface andPtsurface Tomography y-Al andhigh-resolution studyof Materials Scienceand Technology 2 O 3 isdecorated withPt. 2 O 3

tion, surface tion, surface transforma- of phase of mental issues study funda- in situto ex situand lished estab- newly we usedthe materials, catalytic For teries. vehicle bat- for electric developing arethey materials of erties mental prop- - g-Al well-defined andsynthesizedamorphologically developed erogeneous system.To catalyst perform thisstudy, we Pt nanoparticles supported byg-Al supported Pt nanoparticles the mechanismof thecoarsening cence, revealing have beenapplied coales- tostudytheatomic nature of underinsituheatingresolution conditions observations catalytic phenomena. catalytic of aphysical-basedunderstanding toward developing atexperiments (performed ANL)andab initiomodeling current TEMimagingwithx-ray integrating spectroscopy of under insituandexconditionsare intheprocess clusters catalytic capabilitytomography for imagingof conditions. We discrete are alsoattempting todevelop underoxidizing nanoparticles catalytic andrecuing of situ imagingmethodsandprocedures tostudythestability Looking toward FY2013,wewillfocus onestablishing in project. (ANL)for thatNational battery Laboratory posal andare subcontractedfor withArgonne 3years Today 2012Innovation Award, weapplied toaDOEpro- PNNL seminarandworkshop. Inaddition toaMicroscopy a MeetinginBoston,andweorganized MRS 2012Fall Imaging across theElectrons, Photons,andIons”for the Chemical for asymposiumtitled“Frontierorganizer of talks, and15conference presentations. We are theleading journalpublications, three invited peerreviewed seven Throughout FY2012,ourwork inthisproject hasledto . in Nanotechnology intheAPT.evaporation Thework hasbeenpublished fundamental data for identifyingthelaser-induced atom bytheAPTdata. Thiswork providesas evidenced the andO, Au,Mg, that are thenanoparticles composedof andMgO. betweenAu-nanoparticle observed We found coherent, wereon similartips.Highly sharp interfaces performed high resolution STEMimagingandAPTanalysis embedded inMgOand Aunano-clusters the systemof STEM,FIB/SEM,andAPT, of theintegration westudied For structure, interactionfor ag-Al andmetal-surface structures of transitionAl structures of thesurface timeadetailed forreconstruction thefirst of surface areasurface g-Al high relevant, industrially representativebut ishighly of 2 O 3 catalytic support that support catalytic represents amodelsystem 2 O 3 . Atomic-level TEM studied has revealed TEMstudiedhasrevealed . Atomic-level 2 O 3 surfaces. TheS/TEMhigh- surfaces. 2 O 3 . 2 O 3 het- hybrid anode. sulfur loading)ascathode andgraphite-Li mesoporous carbon-composite(22nm,50% with Li-Sbattery of Charge-discharge curve storage. energy for breakthrough large-scale conceptsininexpensive, storagetechnologiesbeyondbatteries and for energy and redox flow batteries. ThisLi-Sinvestmentiscritical strong inLi-ionbatteries, programs Na-betabatteries, storageandhasdeveloped significant investmentinenergy interactions.PNNL hasmadea nent andtheirinterfacial eachnanomaterial compo- theroles of of understanding desirable 3Dapplications andfundamental duction of approaches proand integration - that enable cost-effective economicsynthesis,assembly,Our project isdeveloping composite nanomaterials for Li-Sbatteries. 3Dstructured todevelop electrolyte. Itisthusimportant channelsforalso voidspacetoprovide the diffusion good electricalandmechanicalintegrity. Theseconditions toprovide framework stable conductive and mechanically material electroactive nanomaterials (sulfur) willintegrate Three-dimensional structureddischarge cycles. composite anionsduringcharge/ polysulfide of and shuttleeffect from extremely poorstabilityfers duetothedissolution with hightheoretical specificcapacity. Unfortunately, itsuf- isapromising technology The lithiumsulfur(Li-S)battery Xiaolin Li Nanomaterials forEnergyStorage Three-Dimensional Structured Composite of lithiumbatteries. structures andassistwiththeadvancement bring revolutiontoelectrodematerialsand 3D structuredcompositenanomaterialswill portation powersupplies.Developingnew devices toaccomplishDOE’s targetsfortrans- ogies fornextgenerationenergystorage good cyclestabilityarekeyenablingtechnol- Lithium batterieswithhighcapacityand leading toareason- alleviated, tively - posite waseffec from thenanocom- anion polysulfide film, theshuttleof cation exchange of with athinlayer Li-S battery. Coated materials for the ite ascathode sulfur nanocompos- tured graphene- synthesized astruc- In FY2010,we fundamental study suchasinsitucharacterizations. cathode materials for Li-Sbatteries butalsoascargo for asnew canbeusednotonly electrospin method.They low carbonnanospheres have beensynthesizedusingthe hol- adding pore-generation of reagents. Thecarbonfibers and precursor thepolymer adjusting theconcentration of canbecontrolled by thesefibers diameter andporosity of cathodes. The for Li-Sbattery or compositecarbonfibers We alsobuiltanelectrospin equipmenttosynthesizepure theshuttleeffect. alleviate anionsinsidethehollowspherethe polysulfide and channels,thisuniquestructure diffusion helpsconfine lyte Asmesopores at thecarbonwallprovideloading. electro- the hollowcarbonspheres increase willgreatly the sulfur carbon precursor. Thelargevoidspaceinthemiddle of the conjugated spheres becauseof have highconductivity 100 nmusingthehard template method.Thesecarbon 300nmor low carbonspheres withtunable pore sizeof Inanotherapproach,over 300cycles. wesynthesizedhol- ~800mAh/g for stabilitygood withthecapacity cycling of Thematerials produced sulfur wasalleviated. of tle effect and goodat thecathode confinementeffect side, theshut- anodedesign hybrid anodefor Li-Sbatteries. Withthenew (22 nm,50%sulfurloading)ascathode andgraphite-Li In FY2012,weapplied themesoporous carbon-composite lithiumions. of supply anionsandproducing asteady polysulfide of tion/diffusion thereby reducing dissolu- contact betweensulfurandMC, performance bypromoting theelectrical improved battery that modification sulfurfilling partial ledto andsurface were obtained.We ~600mAh/gover50cycles found of ~1100mAh/gandretention batteries. An initialcapacity of as aconductingmatrix toincorporate sulfurintotheLi-S pore volume, theirapplication investigating systematically rous carbon (MC) materials withtunable pore sizesand mesopo- FY2011,wesynthesizedaseriesof For devices. were asmaterials alsoinvestigated for Li-basedstorage Transition metalsulfidebasednanocomposites 100 cycles. 75%capacity retention over stabilityable of cycling TEM image of hollowcarbon spheres with~300nmdiameter.TEM imageof Materials Scienceand Technology 161

PN10074/2316 Mathematics and Computing Sciences A Distributed Systems Architecture for the Power Grid Jenny Yan Liu conditions to improve reliability, maintain causality, and This project demonstrates the efficiency of a guarantee responsiveness. The middleware is equipped with distributed system architecture for decentral- a modeling tool to estimate how the distributed power model is affected by factors of data communication, geographical izing power models in a large-scale power locations, and computing resource demands. As a result, this grid. The architecture eliminates the bottle- research facilitates the design of cost-effective situational neck of a centralized location for overloaded awareness applications within a large-scale power grid. data management, monolithic computation, The deliverables of this project form a testbed to configure which is increasingly a burden to fast-grow- decentralized power applications in an evolving distributed ing data rates and quantity. power gird for real-time analysis. Our major accomplishment from FY 2012 is a software demonstration deployed on the PowerNet Lab testbed that encompasses: The substantial growth of high quality sensors provides a comprehensive view of the entire connection of the power • scalable middleware that, beyond connecting distributed grid. With this sensor data and high performance computing code, coordinates the dynamic data communication (HPC), a revolution in the operations of the power grid, according to various synchronization criteria namely real-time operations, becomes possible. The key to • an orchestration framework that monitors a step-by-step leverage these benefits is keeping up with the pace of high data exchange frequency samples generated from these sensors. Fast distributed monitoring and analysis applications are increas- • a prototype of distributed state estimation with ingly important to power system operation and control. PMU data A distributed systems architecture testbed for evaluating • a testbed environment leveraging the laboratory power system distributed applications is critical to plan the deployment of HPC platforms to connect their decen- • rigorous simulation/analysis tool on end-to-end delay tralized electric power applications and form the guideline estimation. of distributing power applications to meet real-time In addition, the architecture demonstrated the ability to pro- requirements of critical operations and control in the vide open interfaces with plug-and-play power system appli- electric power system. cations to a data communication platform, offered support to Our previous implementation has the peer-to-peer data connect distributed applications in any language and proto- exchange through TCP sockets. Therefore, the data commu- cols, and provided optimized solutions for deploying distrib- nication logic has been hard-coded in the power application. uted power applications on available HPC clusters given This mechanism has limitations for validating distributed resources and real-time latency constrains. power models (such as the state estimation algorithm), as In FY 2013, we will integrate this distributed systems architec- the study of algorithm behavior requires that different parti- ture with other power grid applications and infrastructures. tions of the same power grid network be tested. The parti- We are targeting integration efforts, including developing tion may change how the sub-areas are connected, hence PMU failure detection using Volttron™ platform, which has peer-to-peer interactions. Hardcoding the data communi- the capability to designate agents to substations and launch cation in the state estimation code is not sufficiently the agents to collect PMUs working status. Integrating our flexible to represent the procedural steps of the architecture with Volttron™ agents, allows the middleware to distributed state estimation. investigate whether a PMU is working in the normal state. In Our research project further enhances the distributed sys- addition, we will integrate with GridOPTICS™, a plug-and-play tems architecture we developed in FY 2012 and introduces a platform to bridge various power applications to data storage, service-oriented layer that contains a workflow environment management and access. The PMU data access in distributed for composing distributed procedural steps, observing data state estimation is one application to be supported by GridOP- exchanges individually, automatically launching remote TICS™, and the architecture will communicate with GridOP- jobs running on HPCs. This layer also links to the coordina- TICS™ API. The success of this integration will lead to quick tion and fault-tolerant middleware, as coordination logic deployment of distributed systems to support power grid is developed in the middleware. The architecture is imple- applications. The solution enables this and similar projects mented as a middleware platform that coordinates distrib- to make informed choices when balancing real-time delays

uted procedural behavior and adapts to runtime computing and resource capacities. PN11002/2324

163 Mathematics and Computing Sciences 164 PN11001/2323 we aim not only toextend existingcapabilitieswe aimnotonly tosupport mentation, registration, andvisualization. Inthisproject, compression, data reduction, imagereconstruction,- seg semanticcharacterization, datamanagement andanalysis, tation, andstorage(ICAT andMeDICi), collaborative data capabilities forsuch asdeveloping data acquisition, anno- components, theproject isbuildingon existingefforts results across modalities.Initscore different exemplary toallowexperimentalsteeringandcorrelationanalysis of emphasisonreal-timemulti-modal imagingwithparticular tosupport isbeingdeveloped The PNNLframework chemical imaging. applications beyond have arangeof will potentially thisframework underlying desired goals, thetechnology to create thisuniquecapability. In addition toachieving thesefurther anddevelop experimentaldata, integrate of core technologiesfor thecapture, andintegration analysis concepts insemanticmapping andrules, workflows, and existing research willbringtogetherarangeof framework community.for example, isthecrystallography Thenew capturematically results; andanalyze at theforefront, toinstallsingleexperimentpipelines canauto- facilities data. Initialstepsatability anumber topetabytes of of chemicalimagingresults andprovides scal- multitude of that worldwide appropriatelydevelopment handlesthe There available iscurrentlynosuitable framework orunder management,andanalysis. data descriptors, preservation, thereforeintegration includescapabilities necessarily for and forThe proposed chemicalimageanalysis framework distributed,highdata ratewithin ahighly environment. and optimalimplementation integration tofacilitate experiment. Ourfocus istodefinethecorrect capabilities results through more completeinformation duringthe improvedimaging technologies,supporting scientific real-time capabilities analysis across chemical different imaginginformaticsPNNL aimstofurther bydeveloping Kerstin KleesevanDam for ChemicalImaging A Multi-ModalIntegrationFramework Mathematics andComputing Sciences for real-timeanalysis. and datacomplexityanincreasingneed growing datavolumes,acquisitionrates, against thebackgroundofexponentially results fromseveralimagingtechnologies parison, andintegrationofexperimental to allowthefastandpreciseanalysis,com- integration frameworkforchemicalimaging This projectwilldeliveranopenmulti-modal

research Science domain-specifictools canalso withothers. also allowsscientiststowork togetherandshareinterfaces and manipulate data. theuser Thecollaborative nature of their projects withouthaving find/invoke tomanually tools entists tomanagetheirresearch from withinthecontextof and afull-featured allowsci- desktopclient.Theinterfaces interacting withdata: alightweightWeb-based interface As shown,Velo for interfaces scientists twodifferent gives and environmental remediation. chemicalimaging, bon sequestration, climate modeling, ascar Itiscurrently deployed indomainsasdiverse ing. and largecomplexdatasets usedinscientificmodel- experiments, computations, helps scientiststoorganize form that canbetailored tospecificsciencedomains. Velo management technologiestocreate ascalable core plat- extends Web-based opensource collaborative anddata modeling andsimulation. and Velo integrates, leverages, ment, andknowledgemanagementinfrastructure for Velo isareusable, domainindependent,eScienceenviron- capabilities intotheframework. developed munity, community tointegrate theopportunity andoffers reconstruction, registration, tothecom- andsegmentation for data management,annotation, compression, reduction, that provideysis (REXAN)framework reusable components infrastructureanalysis (Velo) ExperimentalAnal- andRapid customizable collaborativea highly data managementand imaging modalities.Theproposed environment consistsof with singleinstruments and,more importantly, across processes theanalysis for theguidance andautomation of methods.Further,analysis weare exploringthepotential workflows,analysis thusmovingtoward more sophisticated tify core capabilities analysis of essentialtolargergroups real-time butweare imageanalysis, alsolooking toiden- Framework andVeloFramework infrastructure. Chemical imagingarchitecture combinestheMeDICiIntegration - be simply plugged into Velo and exposed to users through the Web or desktop tools. Velo includes integrated Wiki capabilities so that objects like simulations and input files can automatically be annotated by users through familiar wiki editors. The Wiki components additionally provide semantic markup, search, and query capabilities that can be exploited to support a wide range of knowledge management and collaboration requirements. Velo leverages a mature, graph-based knowledge management system, Alfresco, as its foundation. It provides data storage and a rules framework as well as key functionalities such as versioning, metadata management, and relationships between datasets, projects, and tools (models, scripts, applications, workflows). The CII architecture also integrates Velo with the MeDICi workflow engine for running REXAN framework and its mapping to specific use cases. complex pipelines to process and transform instrument data into visualizable datasets. MeDICi is a middleware began developing two test analysis capabilities. In FY 2012, platform to integrate data- and compute-intensive process- we analyzed the results of our initial analytical tools and ing components with data sources and visualization tools created a framework set of libraries of reusable compo- across multiple scientific domains. MeDICi is connected to nents for the flexible creation of analytical pipelines for Velo through the use of Velo’s Integrated Job Execution both single and multi-modal experiments. The integrated Framework (IJEF), which seamlessly connects MeDICi run- Velo/REXAN system was then used to create a range of ning on an HPC server to the knowledge management analytical pipelines and tools now used in production server and user interface. It automatically stages all by the scientists: required input files, submits the MeDICi workflow job, • PNNL High Resolution Mass Spectrometry – reduces monitors the running workflow (updating status informa- analysis times from hours to seconds and enables much tion in real time), and performs simulation-specific post- larger data samples (100 KB to 40 GB) at the same time processing operations such as filtering results or streaming results (or remote links to large data sets) back to the • ALS X-Ray tomography – reduces analysis times of knowledge management system for integration combined STXM and EM data collected at the ALS with analysis tools. from weeks to minutes, decreases manual work, and increases data volumes to be analyzed in a We contend that it is possible to describe most analytical single step pipelines through a limited set of conceptual workflows that have a range of recurring components that can be • Multi-modal nano-scale STXM and TEM data analysis – reused across a range of conceptual workflows. Therefore, provides a semi-automated process for particle a set of libraries containing highly optimized versions of detection. those reusable components would help with the acceler- In FY 2013, we will productize the analytical environment ated development of new analytical pipelines as well as for distribution to other partners, continue to expand its raise the overall speed, scalability and performance of the functionality and utilize the framework to develop more resulting pipelines. In FY 2011, the project identified the analytical pipelines. core components for such analytical pipelines in collaboration with the initiative researchers and stakeholders and PN11001/2323

165 Mathematics and Computing Sciences 166 PN10002/2244 without using the checkpoint/restart mechanism. andbeableprocessor torecover faults from them cally, weare adding capabilities toNWChem tohandle for high-endsystemsthat existtoday.developed Specifi- ment infrastructure, whichneedstobedesignedand tolerancemanage- erant GlobalArrays, ARMCI,andfault methodrequires- afault-tol The computational chemistry calculations. coupled clustermethodusedinhighaccuracy methods suchasthe tolerant computational chemistry highperformance andfault isalsotodevelop objective model layersby programming suchasGlobalArrays. Our tion runtime system.Theinfrastructure isalsobeingused (ARMCI),ahighperformance communica- Copy Interface across RemoteMemory systemsoftware used byaggregate being usedat tolerance multiplelayers for fault vertical tolerancemanagementinfrastructure Thisfault errors. is theinfrastructure forcost, andleveraging hard andsoft to usefulentitiesinthesystemreduce overallrecovery propagating thisinformation low overheadmechanismsof hardware mechanismsprovided bythemodernsystems, detectionusing These componentsincludescalable fault required tolerancetotheapplications. toprovide fault agement infrastructure components that embodiesseveral toleranceman- afault In thisproject, weare developing and innovation. to accomplishleadingedgescientificdiscovery application software toutilizethesesystemseffectively sustainedpeakperformance aswellenabling scalable of anyrealistic goalsware toachieving stackisimperative tolerancetothesystem soft- components. Providing fault varioussystem mean timebetweenfailures (MTBF)of the remaining majorchallengesishowtohandlereduced thesesystems,oneof peakperformance outof achieving processing elementsandcomponents.Besides millions of combining computing communityisenteringaneraof toexascalesystems,the high performance and theadvent petascalescalecomputingplatforms of With thearrival Abhinav Vishnu and ScientificApplications and AlgorithmswithProgramming Models A ScalableFaultTolerance Infrastructure Mathematics andComputing Sciences form thenation. scientific discoveryandinnovationtotrans- performance ofthesesystemstoaccomplish ing-edge scientificapplicationstoutilizethe elements andcomponentswillenablelead- puting platformswithmillionsofprocessing Delivering faulttoleranceonexascalecom- managers who visited the laboratory duringthisperiod. whovisitedthelaboratory managers meetingsandprogram ourresearch duringreview of ples wasimplemented.We demonstrations performed live thecoupledclustertri- structure, of tolerant version afault tolerant infrastructure totheapplication. Withthisinfra- was designedandimplementedtoprovide required fault globalarrays tolerant applications using(andontopof) withinNWChem, an Specifically fault approach tobuilding International Conference onHighPerformance Computing. communication infrastructure waspresented at the2010 munication benchmarks The withahard nodefault. com- resilient todemonstrate thefault execution of began and ARMCIastheresearch vehicle, we anddevelopment communication operations.collective Usingglobalarrays manager inadditionresilient, tofault non-data moving data tolerantprocess communication, andafault arbitrary resilient communication protocols, amessaginglayer for ment infrastructure,fault detection, providing efficient tolerancemanage- initialfault In FY2010,wedeveloped demonstrating the successful recovery from asingle-nodefailure. demonstrating thesuccessful recovery arun across on eightprocessors four nodes, The outputof In FY 2011, we developed fault tolerance infrastructure corruption are an artifact of low voltage systems and for Cray Seastar systems in addition to demonstrating fault increased density of integration. In the memory sub- tolerance capability of Global Arrays and NWChem on the system, they are likely to occur in SRAMs, frequently used Supercomputing Conference Floor at the PNNL booth. The for designing cache architecture. Conference is a premier conference that attracts 9000+ State-of-the-art approaches have primarily focused on attendees from across the world. Cray Seastar is a proprie- using iterative methods to circumvent this issue; however, tary network used on the Oak Ridge National Laboratory studies have shown that this is not always feasible and may Jaguar supercomputer system. There were many attributes not result in correct solution even after increased number to the design and development of infrastructure, with our of iterations. We were challenged to design efficient algo- design able to detect faults in the system using the primi- rithms in time and space, along with program invariants so tives provided by the Cray Seastar network. The design that we could have high accuracy with minimal data repli- choices included hardware notification from the network cation and recomputation on the fly. We designed an on failures or using software timeouts to detect a fault. The approach for Density Functional Theory in computational scale of the system also required to re-visit actual cost of chemistry focusing on different attributes of data in the recovery to be proportional to the degree of failure. To computation. We believe that our approach will have a meet this stringent requirement, we designed and devel- strong impact on the co-design of algorithms in computa- oped an adaptive method for fault detection to leverage tional chemistry and global address space models, making hardware mechanisms as much as possible while using PNNL a leader in designing fault tolerant applications and software timeouts when the hardware mechanisms cease programming models. to exist. To expedite recovery cost, we designed fault tolerant collective communication primitives to broadcast the In FY 2012, we made significant strides in designing and information efficiently with minimal performance degrada- developing fault tolerant solutions for computational tion. A combination of these capabilities ensured that the chemistry and global arrays, specially looking at soft errors. cost of recovery was proportional to the degree of failure. We first completed the remaining non-fault tolerant elements in computational chemistry, especially singles and A critical component of the infrastructure is the fault toler- doubles that are iterative in nature but require much more ant process manager. In the previous year, we designed one intricate understanding of the algorithm for fault toler- for InfiniBand based systems, but we did not have system ance. However, the primary focus in FY 2012 was soft software available in open source for Cray. For successful errors, especially due to growing concerns of these types completion of this project on Jaguar systems, Cray’s collab- of errors in upcoming large-scale systems. oration was critical. Observing the success of our project, Cray designed and implemented a fault tolerant process We performed an in-depth analysis of computational manager and provided us the software stack for demon- chemistry routine sensitivity to bit flips. The undertaken strating fault tolerance. At the same time, we have contin- approach is statistical for an insight on the effects. We exe- ued to design and develop scalable fault tolerant methods cuted a large number of simulation runs with bit-flips in in computational chemistry using NWChem. We increased many important data structures to understand flip sensitiv- the scalability of fault tolerant NWChem’s methods by ity, with the results divided into four quadrants: 1) the sim- designing methods for singles, doubles, 4-index transfor- ulation took same number of iterations and the result was mation in addition to triples energy correction. Our major correct; 2) the simulation took larger number of iterations result is that we can achieve fault tolerance with only an and the result was correct; 3) the simulation converged additional memory of 2–3%, insignificant time overhead in with incorrect results; 4) hang/abort due to mishandled the absence of failures and <10% overhead in the presence exception. About 46% fell in the first category, implying of failures. Compared with the state-of-the-art checkpoint/ that many of the bit flips can be handled successfully in restart systems that allow 85% of CPU cycles to be available iterative procedures without a need for extra fault toler- for computation, our proposed solution outperforms them ance strategy. Approximately 13% required extra iterations, by a significant margin. which leaves us with 41% of the runs that resulted in failure. We are currently using this study to define numerical Most of the research proposed above considers hard faults, assertions that can allow the handling of 80–90% of fail- those in which components become permanently unre- ures. We expect that it would be difficult to handle 100%; sponsive. While these failures are increasingly prevalent, however, we can handle a significantly higher number of there is another class that have seen tremendous incre- failures with simple, effective numerical assertions. ment during last couple of years. Soft errors, transient errors in the hardware logic, may lead to silent data PN10002/2244

167 Mathematics and Computing Sciences 168 PN12003/2404 to identifythe needfor external SA.Additionally, weattended utilities,balancingauthorities, andreliability coordinators) of (includingrepresentative with organizations cases interviews the masterspecification document. We conducted numerous During FY2012,significant progress wasmadeincompiling document. challenges identifiedwithin themasterspecification visualization applicationPerspectives that directlyaddresses theShared andthedesignimplementation of events; attempt toshare when they informationtors duringcritical opera- SA that identifiesuse-cases,challenges,andneedsof amasterspecification document creation for fostering of Project progress isbasedontwofundamentalobjectives: thepowergrid. of bothwithin andoutside events broader aSAof achieve that to willallowpowersystemoperators Perspectives, application,to thesefunctionalitiesthrough Shared anew stability. impactgrid can potentially We access willsupport that so that events have organizations abetterawareness of weather,the powerdomain(e.g., political/social,cyber, etc.) that strategies incorporateanalytic information outsideof boundaries. We visual willextendthis mechanismwithnew nizations toshare information securely across organizational visualcommunication- mechanismthat willalloworga a new This project theneedfor willsupport externalSAbydesigning monitoring andcollaborative problem-solving efforts. wide-area, tosupport between organizations high-level communication nal SAneedsbybroadening theconduitof visualization methodsthatfor exter canbettersupport new in order toincrease externalSA.Inthiscontext,there isneed that toseeacross canhelporganizations seams organizational successfulresearch of thereorganization), islittleevidence internalSAcapabilitiesport state (i.e., withinthe awareness of methodsthat sup- haveWhile manyorganizations developed ties toescalate andcascadeintolarge-scaleblackout events. there issignificantpotential for smalllocalpowerinstabili- Without real-time, system-wideawareness inthepowergrid, Luke Gosink For PowerGridSituationAwareness ToolsActionable Visualization Mathematics andComputing Sciences communication andcollaboration. gies thatincreaseinter-organizational the designofactionablevisualizationstrate- greater SAwithinthegriddomainthrough project addressesthefundamentalneedfor ness (SA)forallorganizationsinthegrid.This transparency andwide-areasituationaware- relies heavilyuponreal-timeinformation A sustainable,efficient powergridoperation

reduce time-to-resolution for various events. to Shared Perspectives of validation studyontheeffectiveness back. Afinaltask a for FY2013willbeto design andperform feed- industry larger use-casesandsolicita broader range of todemonstrate theapplicationthis task,wewillbegin on our research onthemasterspecification document. Toward tofurther engagement In FY2013,wewillcontinue industry use-cases. for industry customapplications components enables therapid designof visualization components.Thissetof collectionof interactive based onascalable webarchitecture that ahighly supports andfeedbackfromconversation is industry. Thetoolitself has beenconstructed andiscurrently beingusedtoinitiate tions through Aprototype information for thistool sharing. - betweenorganiza and securely buildcommon understanding applicationPerspectives that addresses theneedtoflexibly From ourfindings,weinitiated designwork for theShared tion tohelpbuildcommonunderstanding. informa- can“push”andshareorganizations varioustypesof where both audioandacommonvisualization perspective communication conduitthrough acommontoolthat supports challenge, ourresearch identifiestheneedtobroaden the that canresult indisastrous consequences.To address this alone creates significantopportunity for miscommunication relying onphonecommunication critical, complexevents, however,sufficient; evidence indicates thatgrowing during non-criticalsituations, thiscommunication For mediumis ing. - on telephonestocreate commonunderstand afoundation of communicating across oftendependexclusively organizations are lacking inmethodsthat externalSA.Operators support Based onourinitialresearch, we identifiedthat organizations used togenerate thisawareness. several workshops toidentifycurrent approaches andseminars Shared Perspectives application.Shared Perspectives Advanced Optimizations for Extreme-Scale Homogeneous Systems Darren Kerbyson IBM Blue Gene/Q systems. Our specific aims are to In this project, we combine techniques aimed 1) develop performance diagnostics applicable to Global at tackling performance, power, and reliabil- Arrays applications that allow both detailed and summary profiles to assist in bottleneck identification; 2) integrate ity on the latest supercomputers that contain the novel Energy Template approach for optimizing energy a massive number of homogeneous proces- use into the PNNL Molecular Dynamics application code; sor cores. 3) explore optimized Global Array implementations on the latest advanced homogeneous systems; and 4) develop Computing at extreme scales increasingly requires the data-centric approaches optimizations that can be combination of advanced techniques to achieve reliable, applied to PNNL’s coupled cluster (CC) molecular high performance at acceptable energy efficiencies. It is science applications. critical that all components are considered in combination With a mid-year start, our work on performance diagnostics on future supercomputers in the context of applications centered on the use and extension of the HPC toolkit to of interest to obtain high computational rates and achieve analyze applications that use the Global Arrays toolkit. the most science on the available systems. PNNL has Currently, there are no available tools that attribute perfor- enabled the exploration and development of techniques mance information to global data objects, but there is a for enabling performance analysis, fault tolerance, energy need to understanding data motion not only for perfor- efficiencies, and optimizations for advanced computational mance but also for energy consumption. We enabled HPC resources, including GPUs and multi-core processing nodes. toolkit to examine global data object with respect to time, For this project, we aim to go one step further and explore rank, and calling context. Use performance included met- the use of these techniques in combination, as appropri- rics such as “bytes accessed per non-local read” and “aver- ate, in the context of the largest systems. In particular, age latency per blocking write.” To collect profiles and we target systems that comprise homogeneous traces at scale with minimal overhead, we combined the processor-cores. sampling of remote reads and writes with that of program This project focuses on the development and integration of behavior in the calling context. The analysis of the PNNL capabilities in performance diagnostics, energy efficiency Molecular Dynamics ARGOS application was demonstrated optimizations and software fault tolerant techniques with and enables an analysis of its use of GA, in particular the a particular emphasis on data movement. The work will be code and amount of global data transferred. These findings applied to molecular science applications and targets sys- were presented at the Workshop on High Performance tems such as Cray XE6 and Cascade type systems as well as Infrastructure for Scalable Tools (WHIST ’12). The latest in the Blue Gene BG/Q was to explore and optimize Global Arrays for its use at extreme scales. Notably, BG/Q was at both Lawrence Livermore National Laboratory and Argonne National Laboratory this year in systems that have achieved unprecedented levels of performance and concurrency. Our work has enabled Global Arrays to maximize low-level communication primitives available on BG/Q and has overcome some architecture limitations, including the absence of atomic memory operations (AMOs). An in-depth performance evaluation of its implementation has been undertaken, which demonstrates its use in medium-scale (up to 4,096 processor-cores) runs of NWChem’s SCF operation. A report on this work is being prepared for publication.

Processor groups localize data required in the computation of a The general structure of the processor-group-based itera- coupled-cluster (singles, doubles) CCSD calculation. Each subset uses tive CCSD (CC with singles and doubles) algorithm has been local replication of input data, reducing communication across a large- developed and implemented. This has been achieved scale system. PN12006/2407

169 Mathematics and Computing Sciences 170 PN12006/2407 cessor-core basiswhencores have significantwaittimes. saving techniquesonaperproARGOS that- will useenergy template FY2013,wewillincorporate into For anenergy processor groups. number of arbitrary current algorithm enables theusertoexploitan communication requirements across asystem.Our localize thedata neededfor eachsubsetreducing the processor (PG),andhasenabled group ustoeffectively bital indices.Eachsubsetisformed onacorresponding configurations orbycorresponding anddoubly spinor gly equations whereCCSD eachsubsetischaracterizedbysin- the through intosubsetsthewholesetof the partitioning Mathematics andComputing Sciences - unrestricted case. Hartree-Fock four-index transformation withthenew integrated for the algorithm willbefully to acorrespondingThisnew PG. excited clusteramplitudeswillbecopied anddoubly grals two-electron inte- algorithms where subsetof therelevant on PGversion theCCSD addition, wewillextendtestingof thatsor-group localizesitsglobaldata version accesses.In theproces- of toquantifyeffectiveness and willuseCCSD within theHPCtoolkit aimedat costingdata movement applicationsWe of willalsoextendourdata centricanalysis Advanced Statistical Network Models for the Integration of Experimental and Open Source Textual Data for Bioforensic Analyses Bobbie-Jo M. Webb-Robertson In FY 2011, our work led to two major development This project is developing a statistical meth- creations, which are described below. odology from composite signatures to iden- Complex Biomedical Event Extraction Pipeline. We tify possible origins of biological samples. developed the event extraction pipeline that uses novel The results will reduce the time it takes to signatures composed of traditional linguistic features and identify the origin of samples, which can semantic knowledge to predict event triggers and their increase the speed of prosecution. Our candidate arguments. The developed pipeline was lever- research also presents a novel approach aged to search scientific publications for specific types of to extracting environmental factors or culture media. The event extraction pipeline was used to determinants from open-source literature. generate signatures of each publication in our open-source dataset in simple binary vectors based key terms.

The integration of disparate data into robust signatures is Methodology for composite signature generation. necessary for many problems in the intelligence commu- Composite signature generation was derived in a Bayesian nity. Some examples include bioforensic analyses to identify framework. The conditional independence of data streams the “location” of a sample origin, authenticating food ori- allowed information integration directly to a final probabil- gins to determine the cause of a contamination, and track- ity of interest: institution (I) given the experimental evi- ing explosives to their originating location. In all cases, dence and P(I|ED). The network linked experimental and there will typically be sensor-level information that yields textual data through a culture media node. The textual sig- clues to the content or composition of the subject of inter- nature developed from the event extraction pipeline was est. This information alone may be inadequate to deter- used to define the appropriate conditional probabilities. mine origin. Integrating with other non-experimental For FY 2012, we generated the full composite Bayesian net- sources of information into signatures would help narrow work and tested it on an annotated set of documents and the investigative field. data for Bacillus anthracis and Yersinia pestis. We illustrate This project developed a computational framework to that the Bayesian network model can assign a higher poste- quantify statistically entity identification (person, place, rior probability than expected by random chance to appro- etc.) associated with biological sample origin through apply- priate institutions when trained using only a small set of ing composite signatures. The first set of data includes tradi- manually analyzed documents. tional experimental forensic approaches, such as chemical The results of this project demonstrate that it is both feasi- composition to determine culture media, genomic informa- ble and advantageous to integrate laboratory measure- tion, and proteomic data. The second set is analytical about ments with automated text searches in a probabilistic capabilities required to grow the sample and includes manner for investigative microbial forensics. results of scientific literature and a list of scientists and other personnel associated with those publications.

The basic framework of Bayesian composite signature generation for the purpose of bioforensic attribution. PN11007/2329 171 Mathematics and Computing Sciences 172 PN11013/2335 of oursoftware additional toolbysupplying algorithm of mance. During FY2012,weaugmentedthesecapabilities algorithm that wouldoptimizesignature detectionperfor parameter configuration touse for thecompressive sensing testcasestodeterminetheoptimal a largenumber of experiments. Theseexperimentscreated andevaluated structure torun “degradation/reconstruction/signature” In FY2011,wedesignedandimplementedasoftware infra- designing sensorandsignature threat detectionsystems. scenario andhowbesttoallocate specific resources when noise andinterference inanoperational canbemitigated ally, howefficiently ourwork willprovide knowledgeof data capture of for signature detection. Addition- efficiency signatures aswellincreasedment intherobustness of thiswork willprovide animprove- A successfuloutcomeof signature detection.We are focusing ontwoareas: thesereconstruction techniquesinthecontextof ness of signature- detection.Ourresearch quantifiestheeffective sensor applications and inthearea of leveraged been fully and hasprovided somegroundbreaking results, ithasnot research active While compressive sensingisahighly area cated whenthe data mathematical meetcertain conditions. from data streams that orobfus- are incomplete, degraded, methods for reconstructing ormeasurements observations algorithms that provides Compressive sensingisclassof Luke Gosink Compressive SensingforThreat Detection • • Mathematics andComputing Sciences are purposefullyobscured. environment orduringinstanceswhenthey of signaturesthatindicatethreatsinanoisy ing toimprovetheefficiency androbustness compressive sensingalgorithms,wearework- and predictnationalsecuritythreats.Byusing This researchimprovestheabilitytodetect rate andstilldetectthesignature. we canlowerthedata acquisitionordata transmission in normaloperation. We towhich willquantifythelevel the sensoreithertocapture ortransmitlessdata than canbeincreased byallowing environments. Efficiency Detecting signatures intime-anddata-constrained that thethreat signature canstillbedetected. features are inthe identifiable reconstructed data so theneeded andwedetermineif compressive sensing, measurement isreconstructedThe degraded using noise, interference,in thepresence orobfuscation. of Detecting signatures data set,suchas from adegraded - interferent chemicals. at multiplethreat chemicalscombinedwithmultiple theIRdata set,expandingtheproblem tolook of analysis dataparticular problem. Second,wewillcompletean configuration compressivesensingtoa touse apply testcasesneededtofindtheoptimal parameter number of by adding anintelligentsearch algorithm toreduce the oursoftware tool wewillimprove theusabilityFirst, of project toitssuccessfulcompletionandhastwopieces. Our research planfor FY2013isdesignedtoguidethe causedbyunknowninterferentpositives chemicals. false decreasingperformance bysignificantly the number of proof-of-concept study, ourtoolsimproved sensor theuseof airbornechemicalthreats.detect signatures Inaninitial of data setwas an infrared sensordesignedto spectroscopy (IR) FY2012target imageandspectraldata. Ourprimary types of data typesincludingmultiple observational a varietyof We have applied andexperimentedwithourmethodsusing the initiative. by capabilitiesin conjunctionwithotheranalysis developed toallowtheiruse framework them intoaspecificanalytical high performance computingenvironment andintegrated could beperformed. We alsoadapted ourtoolstorun ina and increased thespeedat whichasignature detection capabilities toimprove signature detectionperformance car’s license plate) are enhancedand more readable. interest representingthe features the of thesignature (i.e., thedigitsof compressivesensingtoreconstruct anoisyimagesothat The useof Computational Optimization and Predictive Simulation for Synthetic Biology Bill Cannon cells and their surrounding environment. The ability to We are developing a multi-scale predictive model cellular metabolism accurately and predict out- simulation capability for biology similar in comes would change the way that we conduct biological research and would lead to advances in biotechnology spirit to multi-physics codes based on physical that could revolutionize material production in industry. principles that will couple metabolic reactions of microorganism with their environment and Simulation code for state simulations of metabolism has allow for dynamic optimization of metabolic been written. The code is written in ANSI C and will run on any computer that has a C compiler. Development of processes to the available nutrients and the code required us to define probabilistically the rela- environment. tionship between chemical kinetics and thermodynamics for coupled reactions from the perspective of statistical The goal of this project is to develop a multi-scale predic- mechanics, which we were able to do. Understanding the tive simulation capability for biology that is predictive and probabilistic relationship between the two qualities was reliable. Currently, flux-based simulation approaches are essential for elucidating the assumptions in the state-based the method of choice for modeling metabolism because simulation approach. they do not require the use of rate constants. Flux-based A model system consisting of 10 coupled reactions has approaches also do not use the law of mass action – the been simulated both kinetically and using the new state- basis of thermodynamics – at all. Instead, the methods are based approach. We are currently working on scaling up based on fitting rates without regard for reactant or prod- the methods to apply the simulation technology to an uct concentrations. However, this convenience also limits independent model of the metabolism of bacteria. How- the predictive power of the methods in that metabolite lev- ever, our developments have gone beyond technology to els are difficult to predict from flux values. Further, flux- model biological systems based on their states. We have based methods do not provide information about energy additionally made progress towards including kinetic simu- costs of pathways, sets of pathways, or organisms. Ideally, lations of different types. Various biological models require the modeling of metabolisms would use kinetic simula- unique simulation approaches, which actually have signifi- tions; however, these studies would require knowledge cant overlap in their computational routines. of thousands of rate constants involved in the reactions. To achieve this flexibility and the computational perfor- In addition, measuring rate constants is labor intensive mance that the methods will require, we have developed a and thus are unavailable for most enzymatic reactions. high-performance acceleration framework for multicellular An alternative to both of the above approaches is to model biological system simulation, Biocellion. This simulation metabolisms using simulations of states rather than simu- allows its model developers to provide the specifics of their lations of reactions, in which the state is defined as the set biological system and then accelerates the common com- of all metabolite counts or concentrations. While kinetic putational routines using a high-performance parallel simulations model reactions using the law of mass action, computer. To this end, we tested the framework on moder- states are modeled based on likelihood ratios of mass ate-sized clusters (500+ cores) to simulate biological sys- action. Both approaches provide information about the tems up to hundreds of millions of cells. The remaining energy requirements of metabolic reactions and pathways. task is to integrate the state-based thermodynamics cell However, modeling states rather than reactions have the model to the framework and to scale up the framework advantage that the parameters need (standard free ener- to run on leadership class supercomputers to simulate gies of reaction) are much easier to determine than the billions to trillions of cells. parameters needed to model reactions (rate constants). For FY 2013, we plan to develop further the state-based It is thus necessary to include discussion about recent simulation code and couple this code to the HPC frame- results, assumptions, and issues in using simulations work using Biocellion’s interface. The state-based methods of state to model metabolism. will be developed to understand more fully the assump- Achieving the vision of realistic microbe model com- tions in the models. The methods for accelerating the munities requires high performance computing. The construction of models will also continue, as model computations have two major components: simulating building is often the research bottleneck. both individual cell behavior and interactions among PN12016/2417 173 Mathematics and Computing Sciences 174 PN12018/2419 ory appraisal (hypothetico-deductive). ory andthe- (abductive), development theory ation (inductive), gener approach:constructing theoriesinalife-cycle theory ising scientific approach. We extrapolated from ameansof ledtoasmallerteam,whichidentifiedprom- These efforts security. tocyber the socialandphysicalsciencesrelevant in aresearch paper that identifiedscientific approaches in securityscience. Ourdiscussionculminated applied tocyber scientific approaches from leveraging otherfieldstobe of meetingstodiscusstheconcepts task, wehostedaseriesof securityfield. researchers inthecyber To accomplishthis of team We thisyear started byidentifying aninterdisciplinary a testbedcapability that tosupport research. to perform more scientific research aswellto research thefield tocatalyze securitysciencemethodology a cyber duce generalizable results. Ourultimate goal istogenerate securityandproconduct fundamentalresearch- in cyber to Thisprojectground. attempts tobuildamethodology are andattackers, losing butthedefenders defenders betweenthe discipline iscaughtinacat-and-mouse game securityfieldhas realized andvocalizedthat our the cyber securityscience. Asmallcore within group ing soundcyber perform- ordiscover methodsof istodevelop Our objective metricsforparative determiningthesolutioneffectiveness. solutionstogenerate representative com- characteristics of ronments. Usingthesedatasets, the wewillinvestigate datasets that coverresearch andfinancialcomputingenvi - We propose togenerate tousecapabilities andtechnology potentialliabilities. becauseof securityevents actual cyber community, tocollectdata from anditisoftendifficult realisticexpansive, datasets for thecyber-related research togenerateand exploitation. Thereefforts have beenfew vent, deter, threats, andrespond vulnerabilities, tocyber scientific hypothesessurrounding methodstodetect,pre- to enable research experimentstotest,validate, andanswer security.entists incyber Specifically, datasets are necessary Workable datasets andmetricsaresoughtafterbysci- highly David Manz and Repeatability Cyber SecurityExperimentation Mathematics andComputing Sciences decisions. state butalsomoreinformedcybersecurity standing oftheoperationalcybersecurity which canleadtonotonlyabetterunder rigor intothecybersecurityresearchfield, This projectattemptstointroducescientific - - next several years. next several for underthat the willbesupported effort thiseffort all of Cyber Securityapproach or willbeimplemented,andparts AsymmetricResilient FY2013, thePNNLlaboratory-wide For results.analyzing scientist tospendmore timedesigningexperimentsand configuring machinesandwiringracks,whichenables the Our work reduces thetimea researcher needstospend approach that resources. focuses onconsolidating laboratory alaboratory-wide PNNLinstitutional computing, of is apart securitytestbedthat is amulti-purpose, multi-usercyber scientificexperimentation.testbed insupporting CyberSec we have thePNNLCyberSec improved thecapabilities of cyber securityscience, todefinea In addition toour efforts fromand asmallgroup PNNLattended theevent. askedtance, topic, they ustochairapanelonthisimportant workshop.ment Results(LASER) Inaddition totheaccep- accepted totheLearningfrom SecurityExperi- Authoritative were detailedinapaper efforts that was Our survey tific rigor current intothefield’s research environment. physical andthesocialsciences.Thisapproach injectsscien- which isafeedbackloopthat joinsmethodsfrom boththe cyber security science for defined ahybridmethodology toasocialsciencesstudyphilosophy.itself Therefore, we securitylends cyber naturethe unique, of oftenadversarial physicalsciences,but well withtheexperimentalnature of tific methodology. fit Themechanicsaspectsorcomputers knowledgerooted inscien- securitybodyof acyber tion of have ledtoafledgling efforts thegenera- approach tobegin itself. beconstrued avalidscienceinandof Our never andscientificprediction can theoretical understanding, formal study,make of ascience. Experimentation devoid crucial stepinthescientificprocess, butitalonedoesnot Cyber securitysciencehasfocused onexperimentation, a Cyber securitysciencemethodology. Cyber/Physical Security Vulnerability Assessment Integration Douglas G. MacDonald of each into a single VA capability. The second step was This project evaluated physical and cyber vul- bringing together cyber and physical world SMEs to nerability analysis (VA) techniques and pro- develop a “common language” by clarifying the various meanings of key terms shared between the two domains. vided a strategic approach to integrating the With similar verbiage and a better understanding of sys- interdependent relationships of each into a tem-level interactions, the SMEs applied these creations single capability that enabled increased iden- into a blended system framework for analysis and concep- tification and assessment of the true overall tualized how each element was affected by neighboring risk during a vulnerability assessment. elements in the overall framework. This includes how protection elements in one domain can be degraded by attacking a safeguard in the other. Understanding these Existing VA processes and software tools exist and are heav- component-level dependencies ultimately provided a ily utilized in determining predicted vulnerabilities within more accurate prediction of vulnerabilities associated the physical and cyber security domains. The problem is with the composite system. that the majority of the determinations is performed independently and interacts only on a superficial level. Both The four potential attack modes evaluated consist of the physical and cyber security domains offer solutions for the physical only, cyber only, physically enabled cyber, and discovery of vulnerabilities, but neither method fully repre- cyber enabled physical. A local critical infrastructure entity sents the true potential security risk to a site, facility, or volunteered to host this blended physical/cyber security asset nor comprehensively assesses the overall security assessment on a proof-of-concept basis, and the lessons posture. This project produced a new methodology to learned have been integrated into this effort. blend the independent VA techniques, cross-trained subject The prototype software tool was designed to capture the matter experts (SMEs) who understand the potential effects new methodology and both accurately model and quickly each domain can have on the other, and transformation assess each potential pathway to the target, identifying of the existing suite of software tools currently utilized each sub-system vulnerability and how it could affect the in the physical protection world to identify and assess overall protection of the asset. The new Physical and Cyber more accurately the risk associated with a blended cyber/ Risk Analysis Tool (PACRAT) was created to model each physical attack scenario. This new approach will allow potential pathway in a given scenario and demonstrate the the security analyst to determine a more accurate degradation a cyber attack could have on the physical pro- measure of risk than is currently accounted for in tection system that would reveal physical threats to an most vulnerability assessments. electronic communication system. This new tool is the first of its kind to model multi-directional traffic in an attack scenario, allowing the adversary to transverse areas in either direction (forward or backward). This flexibility allows the analyst to model the effect of degrading a protection element in one domain (given a successful chain of certain events) and then “backtrack” to exploit this weakened state in the other domain. This tool analyzes every potential pathway from off site to the target, accounting for sub-system and component-level inter- dependencies. The tool’s output is a visual and statistical representation of each potential pathway intended to highlight elements commonly exploited in successful attacks. The results of this tool can be used to guide security invest- PACRAT output summary showing the attack pathways used in both ments by improving the system effectiveness on the most the physical and cyber layers in a given scenario. vulnerable paths and improve the overall security posture by running “what if” scenarios. For the first time, the VA Our first task was to evaluate each of the physical and process will be able to account for and quantify the previ- cyber security VA techniques and provide a single strategic ously undeterminable risk associated with the inability to approach to integrating the interdependent relationships predict component-level interactions. PN11016/2338 175 Mathematics and Computing Sciences 176 PN11019/2341 legacy source code) by using biological theory tocapture source code)byusingbiologicaltheory legacy string-based information (suchasnetwork transactionsor interest in Thisallowsustodiscoverpatterns of events. sequencesof to geneandprotein sequences,toarbitrary expect bychance. We applied extendthisconcept,originally textthat are more similarthanone would ing sequencesof from bioinformatics-basedderived techniquesfor discover- themethodis identifications.true The andfalse root of todiscriminate between isnecessary orbehaviors events security. Thismethodismostimpactfulwhenasequenceof andnetwork signatures incyber of andanalysis the discovery This project capabilities isdirected at toenable developing exploitmethodologies. evolving constantly of the face knowwhat tolook must first for, whichisachallengein rule-based detectionschemesisthat we One limitation of transactionsare benign. environment thevastmajorityof infiltration andexploitation. inthis To make worse, matters mation toidentify increasinglysophisticated attempts at digitalinfor bodyof anenormous, ever-growing of analysis network transactions todefendtheDOEcomplexrequires of scale andcomplexity. theintent example, understanding For securitychallengesresult from acombinationMany cyber of Christopher S.Oehmen Inspired SignalDetection Data-Intensive AlgorithmsforBioinformatics- two sequences. to align the a mismatch (right),andcircles necessary showinsertions exactmatchesin thetwoentitiesabove. (left), Boxes showblocksof regions(bold) conserved Bottom: Alignmentbetweenthe two highly Mathematics andComputing Sciences network securitychallenges. signatures foruseinsolutionstocyberand method fordiscoveringsequence-based This projectisdevelopingageneralpurpose

signature drift. under whichsequence- andprofile-based modelscapture experiment suiteinFY2013 to uncovertheconditions We will follow this pilotexperimentwithamore rigorous somesoftware families. of based modelsperformed variants wellonrecognizing new thosemodels.We found thatof bothsequence- andprofile- andassessedtheperformance afunctionalfamily models of representation, webuiltsequence-basedandprofile-based toourstring software binary anevolving of versions several signaturesmodels toanalyze that drift.Afterconverting usingprofile-based ments todeterminethefeasibilityof build profile-based models. We thenconductedpilotexperi- to generated byourmethodandrecognized tobeinafamily thefamily. of InFY2012,weusedsequences ized) members uncharacter (previously used tomore flexibly recognizenew cule subsequences.Theseprofile-based modelscanalsobe used torepresent complexdependencies betweenbiomole- Profile-based models.Historically, thesemodels have been similarity.of sizeasourmeasure graph the upperandlowerboundsof andusedaGaussianmeanof implemented thestargraph true distance. graph We upperandlowerboundof mate of imations, areasonable wefound thestarmethodtogive esti- distanceapprox graph several to beinitiated.- Afterevaluating with increasing size, graph soanapproximation methodhad edit distanceisacomplexcalculation that doesnotscalewell binaries.However, of changebetweenrevisions signatures of tance provides thebestcomparisonandameanstoconstruct executable binariesare related. We determinedthat editdis- distance metricstoidentifywhichone(s)bestrecognize when graph several thecode. Withthis,weevaluated of the parts relationships bler that betweenall yieldsagraph-like setof disassem- were compared for similarityusinganoff-the-shelf representations.pile binariesintograph Executable binaries mechanismstocomparedifferent andways graphs todecom- entities.WeGraphical representation investigated cyber of andprofile-baseding graphs models. interest includ- thesignaturescomplex representations of of security andotherdomains.InFY2012,weexplored more potentialapplications incyber be applied to awiderspaceof algorithmic andcomputational infrastructure sothat itcan security.approaches tocyber Theintentistoextend existing in atheoretical toaugmentthecurrent framework rule-based pattern-based, grounded identification data-driven schemes Success wouldhave significantoperational impactbyenabling thosepatterns. of principles toextractpatterns withoutaprioriknowledge associated with digitalinformation andusemachine-learning relationships betweentextstrings the inexactandevolving

- Developing Functionality and Performance Enhancements to the Global Array Toolkit Bruce J. Palmer Semi-Empirical Electronic Structure Code. Coulomb inter- We are determining what performance action contribution to the Kohn-Sham matrix in the CP2K and functionality bottlenecks need to be electronic structure code is being converted from its original replicated data implementation using MPI to a fully addressed in global arrays (GA) to support distributed algorithm using GA. The original algorithm has applications that are expected to run on been modified so that the calculation is decomposed into exascale platforms. Our project will help pair interactions between spatial blocks to reduce latency implement proposed solutions in target costs in communication. Additionally, the algorithm has applications and evaluate their effectiveness. been verified to give numerical correctness and competi- tive performance with the original MPI code. Currently, we are working on scaling the algorithm to large core counts. GA is the premier data-centric programming model developed in conjunction with NWChem at PNNL, has been used Reactive Chemistry and Particle Tracking. The particle on most of the large-scale supercomputers deployed to tracking algorithm has been parallelized and incorporated date, and is under active development at PNNL. The GA’s into the TETHYS continuum fluid dynamics code. Efforts are approach is focused on providing simple one-sided access currently underway to combine this capability with the to arrays of data that are globally visible from any process bacterial communities micromodel. This model combines in the system. Despite GA being successful for the applica- continuum modeling of reactions using standard reaction- tions that use it, the application base itself is still relatively diffusion equations with reactions associated with discrete small. Existing applications include PNNL’s NWChem Chem- particles modeled using Lagrangian methods. istry code, the STOMP subsurface code, bioinformatics Lattice Boltzmann Algorithm. The PARAFLOW lattice (Scalablast), and conventional computational fluid dynam- Boltzmann simulation code was converted to use GA for ics (i.e., TETHYS). This limited application base is a concern the ghost cell updates that are a core part of the algorithm. because further development of GA would benefit from a For FY 2012, we added a non-blocking ghost cell updates broader range of requirements. Thus, this project incorpo- operation to GA and combined it with a two-phase update rates GAs into several applications that represent a large, of the LB cells so that we could attempt to overlap compu- diverse group of computing problems with the goal of tation and communication in the lattice Boltzmann algo- using these implementations to understand the scaling rithm. Splitting the LB updates into an inner and outer part behavior of GA within multiple scenarios. resulted in a loss of performance that outweighed the com- Target areas in this project include sparse matrix multipli- munication gain but we are currently looking at improving cations, continuum reactions, and discrete particle tracking performance of this kernel to increase overall code perfor- combined with unstructured continuum fluid flow solvers, mance and scalability. large-scale semi-empirical electronic calculations, and lat- The bulk of the work in FY 2012 has been focused on tice Boltzmann simulations of continuum flow using struc- implementing new algorithms in the target codes and has tured grids. These problems will help us to define a new positioned us to finish our tasks in FY 2013. Most of the functionality needed by application developers to incorpo- effort in FY 2013 will be oriented towards evaluating and rate GA into their codes. In FY 2012, progress was made in improving the algorithms developed thus far. These all four application areas, as detailed below. include the non-blocking ghost cell updates in the lattice Sparse Matrix Multiplication. A code has been developed Boltzmann model, the completed microbial communities that implements a sparse matrix-vector multiplication model in TETHYS, the distributed electrostatic contribution using the GA library and compares it with a comparable to the Kohn-Sham calculation in CP2K and an evaluation of operation using the MPI-based Hypre solver library. We one-sided versus message passing based communication in have modified the previous implementation so that the GA- sparse matrix calculations. We will also continue our out- based algorithm approximates the data layout of the Hypre reach and educational activities to promote further use of implementation and reduces communication. Both block- Global Arrays in HPC applications and to determine what ing and non-blocking implementations have been devel- new functionality and performance enhancements need to oped. Hypre implementation is currently faster than GA, be incorporated into the GA toolkit. based on scaling studies on PIC. We are currently attempting a more in-depth analysis to understand the origin of performance losses relative to Hypre. PN11021/2343 177 Mathematics and Computing Sciences 178 PN11030/2352 for studying a comprehensive set of chemical systems. for studyingacomprehensive setof capabilities versatile extremely scalable andscientifically parallelism.Ultimately,achievable theproject willdeliver types willbeconsidered inrelation totheirimpactonthe different expressed. Also, couplingsof of theimportance configurations, asitwilldictate howtheparallelism canbe coupling termsbetweenequations for reference thedifferent the methods. Ourresearch of involvesathorough analysis inherent inthe multi-reference methodsfor avarietyof thisproject istoexpress theparallelism of The objective configurations canbetreated concurrently. that from allreference thefact derives second level tion that wehave showncanbemadetoscalewell.The from equations associated withasinglereference configura- parallelismcomes of parallelism.Onelevel of two levels following theequations from that thestructure allowfor of promising candidates for parallelization, methods are very scalability.aspects insteadof Interestingly, multi-reference proof-of-concept haveis that beenfocused they ontheory multi-reference coupledclustermethods implementations of current methods for challengingproblems. Alimitation of multi-referenceapproaches have shownthehighvalueof ence problem andemphasizingvariouscharacteristics.These following philosophiesaddressing different themulti-refer In practice, manymulti-reference approaches are developed multi-reference coupledcluster methods. all them asreference configurations. Thisisthe basisof dominantconfigurations anduse allof is toselectasetof reference configuration. solution Insuchasituation, theonly guration that standsoutasasuitable candidate for the In thosecases,therenosingleelectronic confi- isessentially chemical systemexhibitsnear-degenerate electronic states. way. However, single-reference the methodscandecrease if scalable that thesemethodscanbeimplementedinahighly tronic structure theory. projects have Previous demonstrated useful for describingcorrelationscientifically inelec- effects Single-reference coupledclustermethodshave proven tobe Hubertus J.J.vanDam Coupled ClusterCapabilities Development ofParallelMulti-Reference Mathematics andComputing Sciences methods areessential. energy surfaces,forwhichmulti-reference catalysis andchemicallyreactivepotential problems inareassuchastransitionmetal ing implementationswilladdressimportant reference coupledclustermethods.Theresult- of scalabilityavailableinprinciplemulti- This projectwillbringoutextremelevels

these new capabilities. these new perform demonstration classstudiestoshowthestrength of coupled clustertheory. Thefinalaim for thenext year isto applied tosingle reference andsuccessfully we previously for which GPUdevelopments this,wewillrevisit method. For are tobehadfrom more applying computepowertothe CPU-bound approach; hence,performance further benefits remainsability multi-reference, a coupled clustertheory requirements. Evenwith improvedcation scal- andmemory limited inoverallscalability andrequires restricted communi- transformation preceding well, theintegral ithasbecome mentation. Nowthat thecoupledclustercomponentscales In FY2013,wewilladdress bottlenecksinthecurrent imple- andhavepapers, submittedanother paper to ajournal. presented ourwork at twoconferences, publishedthree materials. We fordomain important thecharacterization of application multi-reference anew coupledcluster that offers approach excitations that allowscore-level tobetreated with Wigner formulation results. Finally, weimplementedan showed that thiscorrectionimproves Brillouin- significantly tion for application tomulti-reference formulations and correc- We auniversal size-extensive. is notstrictly developed forimportant theBrillouin-Wignerformulation which byitself corrections, size-extensivity we considered anumber of we demonstrated scalability to80,000cores onTitan.Third, As expected,thetriplescorrection improved markedly, and the infrastructure created for thereference parallelism. level tions tothesingle-doublecode. Inthisapproach, webuilton an approach. Second,weadded triplescorrec perturbative - multi-reference coupledcluster scalewellusingsuch tions of found that boththeBrillouin-WignerandMukherjee formula- figurations are distributedacrossgroups. theprocessor We usingprocessor andthereferenceis partitioned groups con- processors thispurpose, parallelism.For thefullsetof level inFY2011wasmodifiedtoexploitthe reference developed In FY2012,weaddressed thecode three mainissues.First, scalability.achievable and tion termswasperformed focusing onbothaccuracy correc perturbative - the multi-reference of case. Ananalysis in accurate results, highly are even for important achieving reported inliterature. We alsofound that triplescorrections magnitudemore thanpreviously processors least anorder of We codescalestoat can achieve. at found thislevel, that even pools, andwedemonstrated thebasicparallelismthismethod task files andtheuseof improvements tohandlingintegral reference coupledclusterapproach. We implemented coupling termsbuthasthebasicinfrastructure for amulti- simple ence coupledclustermethod,whichusesparticularly In FY2011,wefocused ontheBrillouin-Wignermulti-refer

Enabling Hypothesis Driven Research and Discovery in Extreme Data Kerstin Kleese van Dam type research environment that will enable greater knowl- This project uncovers the significant limiting edge from data collections by researchers. We will establish factors preventing today’s scientists from quantitatively and qualitatively boundaries to exploit extreme data collections by the communities they serve making better use of the existing archives and develop strategies to overcome these boundaries by using a wider range of investigative meth- based on methodologies from the aspect of various odologies. We will explore strategies from difference sciences. the aspect of several different types of In FY 2010, the project broadly evaluated extreme-scale sciences to overcome previously identified data intensive science requirements of experimental and limiting factors and develop a prototype observational facilities and users with a more detailed eval- research environment. uation of the needs of a variety science groups working in PNNL’s major focus areas from systems biology and chemical imaging to atmospheric science and high energy physics. Advances in experimental and computational technologies Based on this requirements analysis, an initial architecture have led to an exponential increase in scientific data vol- concept was developed. As part of the work, we identified a umes and their complexity. This data has resulted in the gap in PNNL’s provision for extreme-scale data: a data trans- creation of a fast growing number of data collections, at fer node that would allow offering large-scale PNNL data times of extreme size. Unlocking the inherent potential in sets to its customer base as well as enabling PNNL scientists these data collections by enabling scientists to carry out to access their own (and other people’s) scientific data else- hypothesis-driven research and discovery through effec- where. For FY 2011, the project team was extended, and we tively synthesizing and probing data as a community and developed an initial prototype research environment that across boundaries will provide extraordinary scientific supports the collaborative, hypothesis-driven research in advantages. However, to achieve the translation of the extreme-scale data. Involvement with other research in collected data into exploitable knowledge requires a para- chemical imaging and BELLE II provided access to extreme- digm shift both in terms of tools and human behavior. scale data repositories and additional user groups. In The advent of extreme data has led to the development of collaboration with the University of Utah, requirements hypothesis-neutral data mining and social data analysis as a for a visual data exploration and analysis environment means to interact with the data. A new generation of tools, were explored as part of the collaborative discovery taking into account the human behavior, could play a piv- environment CAT. otal role in enabling the necessary paradigm shift to support hypothesis-driven scientific research and discovery. In FY 2012, we further addressed the extreme scale data issues by adopting and bootstraping use of PNNL’s Visual- Studies show that researchers will assign and display ization Knowledge (VisKo) approach, which mitigates the human emotions toward any technology with which they variance in service interface descriptions by providing uni- interact; therefore, behavioral and social concepts that form manner of describing data analysis modules, including govern human-to-human interaction may partially govern visualization. VisKo vocabulary is built on the format, type, researcher interactions with technology and digital informa- and parameter concepts and tackles the problem created in tion. Therefore, any new user environment must account the typical use of an implicit, weak agreement that a data- for common practices in the research environment that are set tagged as a given format adheres to a schema or set of required to build necessary trust in data (and/or creator), rules used to encode the data into a file. With this rich its handling, and analysis. Additional research studied suc- description of service interfaces, VisKo demonstrated its cessful strategies in software environments to stimulate capability to disambiguate machine’s understanding of desired behavior, such as collaboration and data reuse. multiple data structure stored in a single format. Knowing In this project, our goal is to discover the significant limiting services’ format, type and parameter restrictions, machines factors that are preventing today’s scientists from making have adequate information to support automatic discovery, better use of the existing archives by using a wider range orchestration, and execution of services in collaborative of investigative methodologies. We commenced exploring environments. Collectively, work under this project for strategies from behavioral science, cognitive psychology, this year produced five conference papers at various business sciences, and information technology that will workshops and symposiums and has a pending journal enable us to overcome the previously identified limiting article publication. factors. By the end of this project, we will develop a proto- PN10033/2275 179 Mathematics and Computing Sciences 180 PN11032/2354 a rigorous data model wasrequired the that integrates results by theproject team andstakeholder community,of To information facilitate andusage exchange, evaluation, experiments toaddress scientific questions. therelevant andnumerical evaluation, including modelcoupling, development, modelframework the successfulintegrated andcommunication challengestoenable semantic, tactic, technical,syn- are significantscientific, Thus, thebarriers challenges, suchasthefollowing: computational, anddata key scientific, ents arangeof presmodels from disciplinesandbackgrounds different - flow, andmedicineare envisaged for thefuture. Coupling other sciencedomainssuchasbiogeochemistry, subsurface crop, inclusionof economic, systems,further andenergy climate, socio- focused onmodelinginteractionsof initially scientificandeconomicprocesses. different While range of human andenvironmental systems,combiningawide integrated making of on thebackground sion andpolicy deci- andconsequencesof thetradeoffs of critical analyses that enables aframework tem Model(iRESM)istodevelop Sys - Earth Regional theIntegrated The fundamentalgoal of Kerstin KleesevanDam Integrated RegionalEarthSystemsData Enabling theMeaningfulExploitationof • • • • • • Mathematics andComputing Sciences exploitation ofresults. necessary collaborativeanalysisand ization) ofthemodelingframeworkand and exercise(includinguncertaintycharacter ture toenablethecollaborativedevelopment implement asupportingresearchinfrastruc- simulation modelsaswelldesignand disciplinary integratedregionalearthsystems software frameworkformulti-scale,multi- This projectwilldevelopaflexiblecoupling issues betweentheuniquescientificdomainsinvolved. data structure,differing representation, andsemantic data formatsdiffering andvariable names requirements implementation approaches, andcomputational simulation languages, models,programming differing temporal scales dealing withsignificantvariability inspatial and between models exchanges andfeedback defining scientific relevant across models consistently overlappingconsolidating andintegrating processes - PNNL-developed VeloPNNL-developed software as itscore data andknowl- coupling planned for FY 2013.Theproject usedtheexisting workflows inpreparation for themulti-scale, multi-domain componentstocreate complexscientific the developed ture, extendedtheMeDICi workflow system,and applied In FY2012,theproject builtthecollaboration infrastruc- building onsomepriorwork inotherdomains. and implementation oncore hasstarted components, Wesis, andevaluation. designedaninitialinfrastructure, - execution, analy experimental planning, tools tosupport as wellproviding accesstodata, models,workflows, and experiments, results throughout theprocess andanalysis discussions,decisions, the concentrated documentation of allowing modelingframework, studies withtheintegrated the infrastructure outscientific willberequired tocarry Moreplanning toexecution importantly, andevaluation. overtimefrom evaluation modelandframework support ment scientificandtechnicalmodelcouplingissues, data, model,andtools,discussdocu- access versioned infrastructure acollaborative platform that willoffer to toprovide research asupporting lenges, itwillbenecessary chal- development theframework thecomplexityof Given project work. late intheFY,coupling began buildingontheprevious required. Initialimplementation work ontheloosemodel also provided flexibilitytoincludeand remove models,as runtime, computing platforms anddata representation. It inmodelimplementationbridge differences approaches, employed MeDICiworkflow system for loosecouplinghelps where possibleandaloosecouplingwhere required. The models coupling approach that closecouplingof facilitated aflexible, hybrid Theprojectsresult developed evaluation. uation, numerical experiments,andmodelingframework approach- grated to collaborative modelcouplingandeval Mid-year, twoprojects were combinedtoprovide aninte- bycombiningdata models. can bedefinedonly areasareas datathat specifictoparticular models,ornew lapping areas data models,distinct betweenthederived helpingtoclassifyover modeling,” areas suchas“reservoir subject groupingsoverlaps of alsohelpdefine exist.They that create betweenthe disciplineswhere “crosswalks” viaontologies as NetCDFCF),whichare thenintegrated for data andmetadatacific standards (such andvocabulary theexistingcommunityspe- edge modelthat canleverage In FY2011,ourproject designedadecentralizedknowl- to theappropriate usercommunities. form aspectscanbeflexibly represented inwhich different data, information, andknowledgeprovided byeachintoa - edge management, analysis, and sharing infrastructure. Because the provenance model is a part of the overall To serve iRESM needs, the following new components iRESM data model, any referenced datasets, tools, and were developed. machines can be associated with the provenance providing a powerful cross reference search capability, referred to by the provenance community as knowledge provenance. iRESM Job Execution Components. iRESM streamlined the job execution process for all of the different models and couplings (and upcoming uncertainty quantifications) by invoking all simulations as different MeDICi workflows; therefore, a MeDICi Job Handler was created that was associated with all workflow instances. The MeDICi Job Handler invokes, monitors, and post-processes all running workflows in a consistent manner. Workflows are initiated by remotely dropping a MeDICi workflow bundle file to a monitored folder that serves as a pipeline endpoint. The MeDICi Job Handler parses the MeDICi status log as it is running to update the workflow instance metadata with the current run status and creates separate folders under the workflow instance’s Output folder for each run. At the completion of each MeDICi pipeline, the Handler automatically copies the MeDICi status file both to Velo as well as small output files and images. The entire output file set is automatically registered and archived via the Publisher component described below, and a remote link is created inside the workflow outputs folder pointing to the remote iRESM BEND model set-up interface. data. Any additional post-processing steps used for analysis (such as combining the results of multiple files into a single iRESM Data Model. This content model defines table) are handled by the pipeline itself as a final step. iRESM-specific objects and their relationships as they iRESM Provenance Components. In addition to adding will be stored in the Velo knowledge base, which has the Registry object definitions and automatically capturing three primary sub-models: the project model, registry, provenance these new provenance relationships during and provenance model. workflow instance creation and execution, iRESM also Project Model (Numerical Experiments). Projects are developed an Archive component that resides on the data objects that help scientists to organize their work and computer server. The Archive is a protected area where encapsulate related data. iRESM created a specific project data (such as simulation outputs) are copied so that they subtype for numerical experiments that encapsulates all required components, such as workflows, referential data, and any results from simulation runs. Registry. The registry defines key data types that iRESM added to greatly enhance the provenance graph provided by Velo: datasets, tools, and machines. Each of these types can be versioned and are included in the provenance graph. Provenance Model. Provenance provides a historical record of what transpired in a numerical experiment and is implemented in iRESM by a combination of metadata saved to objects and a graph of relationships connecting them. When a workflow instance is created inside a numerical experiment, Velo relationships are added, connecting the workflow instance to the workflow definition. As workflows run, status information is conveyed in near real-time and stored as part of the workflow instance metadata. Relationships are added connecting all the inputs and outputs to the workflow instance. A relationship is also added to connect the workflow to the machine on which it ran. Velo Job Handler and iRESM BEND Job Launcher. PN11032/2354 181 Mathematics and Computing Sciences 182 PN11032/2354 BEND. Inthenumerical experimentscarriedoutbythe setup,easily execute, complexscenariosfor andanalyze the project created anenvironment where scientistscould Using theiRESMinfrastructure andworkflow systems, profileto create theenergy for alarger region. As aresult, scientistshave torun themodelmultipletimes statistics withtheweather pattern for that area. geographic demandstatistics calculated bycombiningusage energy within whicheachrepresented buildingtypereceives type portfolios are created for regions, smallgeographical buildings across area. ageographic Representative building different thousands of based onthecharacteristicsof residential andcommercial buildings and consumptionof structure. demands TheBENDmodelquantifiesenergy environment theinfra- todemonstrate thecapabilities of more complexhigh-throughput modelingandanalysis BEND,BEND, For theproject RGCAM,andEPIC. created a three basicworkflowsto develop for specificiRESMmodels: The Velo infrastructure andtheMeDICisystemwere used configuration file inthe filesystemendpoint folder. workflows bycreating anSSHtunnelandplacingaMeDICi Archive. ThePublisherisremotely invoked as other MeDICi invokes thePublishertotransferdataView tothe remote dataset new registered viatheiRESMRegistry Every copied inparalleltotheArchive, accelerating theprocess. flow sothat extremely largedatasets (suchasRESM)canbe Archive. ThePublisherisimplementedasaMeDICiwork - data Publisher, whichisusedtopublishdata tothe In addition toArchive, iRESMcontainsacorresponding management server. from having tobetransferred backtotheknowledge It alsoimproves largedata performance asitprevents simulation. agiven canfindthe antee that they results of resultspromotes canguar inthat reproducibility users of location TheArchive also toavoid danglingpointers. in theknowledgemanagementsystempointtoanArchive deleted.Allremote linkscreatedcannot beinadvertently Mathematics andComputing Sciences - coupling work of themodelingteams. coupling work of capabilities intotheenvironment theinitial andsupport quantification uncertainty In FY2013,wewillintegrate aforementioned models. climate data intosuitable inputfiles for the of team alsoapplied 60years thescriptstoconvert formations from iRESMtoBEND, The RGCAM,andEPIC. their information. where Thecore trans - scriptsdeveloped scale andmeasures required byothermodelsconsuming in workflows that translate climate data intothesuitable transformationthe project scriptstobeused developed In addition toworkflow andinfrastructure development, for problematic runs. procedures model results anderror verificationrecovery of gainsthrough consistent enabled significantefficiency themodelingruns, theinfrastructureexecution has of simplifyingthesetupand In addition tothebenefitsof ple steps,enabling 70,000jobsrun timeperiod. inashort BEND team,upto17,000jobswere submittedinonesim- BEND exampleworkflow setup. Encrypted CPU Instruction Stream Richard Griswold for this processor to validate our design, and we are cur- This project seeks to deny adversaries the rently implementing these techniques in hardware using asymmetric advantage of well-known system an FPGA prototype that will be used to conduct experiments to measure the performance, resiliency, and architecture in cyber attacks by using instruc- effectiveness of these techniques. tion stream and program counter encryption to create unique, dynamic system architec- In FY 2012, we selected Plasma, a MIPS-compatible soft- tures. core processor, as the target for our prototype, and successfully implemented both instruction stream and program counter encryption in a software emulator. We modified While instruction stream encryption is not new a concept the GNU tool chain (compiler, assembler, linker, and associ- with references dating over 20 years, it has not yet been ated utilities) to support creating binaries with encrypted implemented at the hardware layer. Previous studies in this instruction streams and to support additional instructions area has demonstrated the effectiveness of the technique, needed for program counter encryption. The software but the research did not progress beyond implementation emulator and modified tool chain have allowed us to vali- in software emulators or operating system components. date our designs by running test programs and an operat- These systems suffered from significantly reduced perfor- ing system, including a network stack and web server, with mance, and the software components themselves were the entire instruction stream encrypted and program coun- vulnerable to attacks. We initiated this project to research ter encryption enabled. They have also enabled us to test methods for implementing this technique at the hardware caching techniques for improving performance and layer to increase performance and reduce the attack sur- measure effectiveness against two different attack face. During the course of this project, we researched the vectors: code injection and data injection. use of program counter encryption as a complimentary These tests demonstrated that instruction stream and technique to provide extra protection against additional program counter encryption are successfully able to defeat attack vectors. code injection attacks that seek to introduce malicious This project has two primary objectives: first, to demon- code into the system and combat data injection attacks strate that instruction stream and program counter that can compromise system security by hijacking the encryption can effectively defeat several attack vectors, execution flow. Tests have shown that by using a decrypted and second, to implement these techniques at the hard- instruction cache, we can keep system performance impact ware layer. We focused our efforts on embedded systems low – less than 10% increase in test execution times – instead of general-purpose workstations, which gives us a without compromising security. Based on our emulator larger variety of processor architectures to work with and modifications, we began modifying the Plasma processor much greater hardware configurability. We selected a pro- itself to implement instruction stream and program cessor architecture that will allow us to make necessary counter encryption. modifications in hardware, we created a software emulator In FY 2013, we will finish alterations and load the modified processor into an FPGA for validation and further examination, including a full battery of performance, reliability, and security tests. We will research ways to apply these techniques to more complex processors and operating systems with support for virtual memory, shared libraries, and multiple encryption keys while still providing full instruction stream encryption completely implemented at the hardware layer. We will undertake a more detailed cost/benefit analysis and investigate refinements to instruction stream encryption to provide more predicable behavior in handling improperly encrypted code. Code injection attack attempt with instruction stream encryption enabled PN12026/2427

183 Mathematics and Computing Sciences 184 PN12028/2429 Bayesian integration, clustering of proteins basedonfunc- Bayesian clusteringof integration, based ontheirdiseasestate. Theprocess involvesuseof biosignatures classifyingsamples accurately are capable of disease. These chronic pulmonary obstructive model of identify biosignatures from proteomic data from amouse for knowledge to withexpert combiningstatistical analysis apipeline thisproject, wedeveloped year of In ourfirst process. development asapplicationsfor duringour theproblems serving toward identifyingmore processes robust signatures of processes.on statistical We integration willalsowork toolsfor classification feedback andexpert will develop knowledge. expert Additionally,on incorporation of we bettersignaturessources that based allowsgeneration of data area. of We apipelinefor integration willdevelop impactful andhighly inthisdifficult allow advancement researchthe process. area that Thisisanimportant will of andexplicitinputfrom during adomainexpert tion efforts fromknowledge, large-scalecommunityannota- derived statisticalOur project approaches integrates withexpert for theirapplication. accuracy pretable andmay toadomainexpert, notshowsufficient data, may notbeinter not generalizewelltolargersetsof taken. Thebiomarkers do identifiedfrom theseanalyses approach approach istakenexpert-driven is oralargely plagued bytheproblem that eitherapurely data-driven biomarker identification hasbeen agents. Thefieldof bioterror production, anddetectionof diseases, bioenergy problems important includingtreatmentsnumber for of withthepotentialtosolvealarge molecularanalyses sive complexdatasets basedoncomprehen- with aplethoraof hasprovided investigators inbiology The ’omicsrevolution Karin D.Rodland the BiomarkerProblem Experts InundatedwithData: Mathematics andComputing Sciences with expertknowledge. put databyintegratingstatisticalapproaches of signaturesgeneratedfromhigh-through- the accuracy, robustness,andinterpretability We aredevelopingapproachestoimprove - bilities of ourCognosticstool. bilities of tion. Finally, capa wewillcontinue- towork ondeveloping asagold standard for proteinFY 2012toserve quantifica- work onaproteomics dilutionseriesthat wasgenerated in series. Additionally,ships betweenevent wewillcomplete benign)from statistical relationversus andexpert-based - asignature (malign fordevelop classifyingsessionactivity securitydataset. Wethe biologicaldomaintoacyber will thepipelineoutside will demonstrate theapplication of fromfunded inpart theexistingCPTAC project. Thesecond large-scale patient data from ovariancancerandwillbe is ittotwoindependentproblems.then apply Thefirst We willcompleteourbiosignature generation pipeline . inPLoSComputationalBiology asanarticle appears this underreview. thiswork Inaddition, of aportion with non-randommissingvaluesandhave amanuscript on ondataoped amethodfor principalcomponentanalysis the gold-standard dilutionseriesdataset. Finally,- wedevel asthebasisfor ourproposed workThis work with serves ization, andprotein quantification usingtrend analysis. processing pipeline, includingqualityassessment,normal- Additionally, wecompletedwork onaproteomics data pre- theprotein quantification toevaluate problem.cally high-quality gold standardfirst dataset generated specifi- sis ontheresulting asthe dataset. Thisdataset willserve - proteomic samples,andcompletedaninitialqualityanaly a dilutionseriesfor mouseplasmaproteomics, ranthese In FY2012,weperformed data pre-processing bydesigning application, proteins). featuresing individual usedinclassification (inourinitial andfilter grouping, ordering, cation process byvisualizing, tointeractwiththesignaturewhich allowsexperts identifi- methods. We theData CognosticsTool, have developed proteindard quantification for rigorous of evaluation asagold stan- dilution seriesproteomics dataset toserve data withnon-randommissingvaluesandgenerated a of Additionally, methodsfor statistical analysis wedeveloped through data asophisticated exploration andintuitive tool. toparticipate intheprocessand allowingtheexpert communityknowledge), tional annotation (expert-derived - Extreme Scale Capabilities for Subsurface Science and Engineering Steven B. Yabusaki eSTOMP-WR. The isothermal single-phase eSTOMP-WR sim- This project is creating an enduring, high ulator ran the first full implementation of a comprehensive quality software that provides easy mainte- genome-scale metabolic model for Geobacter metalliredu- cens coupled to 3D variably saturated flow and multicom- nance of capabilities under evolving hard- ponent biogeochemical reactive transport in a physically ware and software environments, and geochemically heterogeneous aquifer undergoing extensibility of capabilities by domain engineered uranium bioremediation. The cell-scale physi- scientists, modular replacement of code ology represented by a physico-chemically constrained net- components, and portability to accessible work of 697 intracellular and 51 environmental exchange computing architectures. reactions was solved for every iteration of every timestep for every grid cell using linear programming optimized for maximum growth. The simulation was run on EMSL Chi- We are expanding eSTOMP application capabilities by pro- nook using 1024 cores for 15 hours. viding the supporting software infrastructure for extreme- scale computing. The new simulation capabilities will For many applications, including the one above, the reac- include geomechanics, fully integrated well models, tion network solver in eSTOMP consumes over 90% and genome-scale models of microbial metabolic pathways, sometimes over 99% of the computational effort. While the and unstructured mesh discretization. The large-scale and eSTOMP reaction solver has been engineered to scale near comprehensively detailed applications that are targeted linearly, the large computational effort was shown to be will require considerable software engineering to achieve amenable to further efficiency using GPUs. The first robustness, scalability, performance, and fault tolerance on eSTOMP prototype with CUDA programming for the reac- existing and future extreme scale computing architectures. tion solver has shown that the combined CPU-GPU resource Significant issues that will be addressed include solver per- can be used effectively to reduce the total computational formance, load balancing, scalable I/O, and memory man- time. eSTOMP-WR is one of three existing simulators being agement. A key component of the project will be algorithm used in a high-level DOE project that is developing the next development to exploit non-uniform memory access and generation subsurface simulation capability for environ- heterogeneous accelerator-based architectures (e.g., GPUs). mental management. eSTOMP-CO e. The nonisothermal, multiphase flow and The new capabilities and infrastructure will enable applica- 2 brine transport eSTOMP-CO e simulator for supercritical CO tions of unprecedented magnitude and process model 2 2 detail. Large-scale, nonisothermal, multiphase flow and behavior in deep geologic systems was updated to improve geomechanics simulations of deep geologic carbon seques- convergence using developed techniques. A geomechanics capability for eSTOMP-CO e has been successfully tested for tration will be used to assess the long-term impact of large 2 arrays of supercritical CO injection wells on CO migration, elastic behavior. Development and testing of the coupling 2 2 with plastic behavior is ongoing. eSTOMP-CO e is slated to porewater displacement, and caprock integrity. In conjunc- 2 tion with recent advancements in subsurface sampling that be used in the U.S. Environmental Protection Agency’s reg- can determine a wide range of metabolic process activities ulatory role for carbon sequestration. by their association with genes, proteins, and peptides eSTOMP-WAE. Development of the GA/grid component from specific microorganisms, the integration of in silico engineered non-isothermal, multiphase flow and transport cell-scale metabolic models into the eSTOMP coupled pro- operational mode, eSTOMP-WAE was completed and tested cess subsurface simulator will provide a mechanistically against the STOMP test suite. This was the third eSTOMP detailed representation of the intracellular metabolic operational mode to be created using the GA-Grid Compo- pathways and environmental exchange reactions under nent. eSTOMP-WAE will be used in the assessment of dessi- realistically complex subsurface conditions. cation approaches for attenuating the transport of Tc-99 This project was terminated in early FY 2012 upon re-prior- and other mobile contaminants in the Hanford deep itizing of laboratory-level scientific issues and reallocating vadose zone. It will also be instrumental in the modeling of investments. Prior to that end, the following progress was carbon and nitrogen fluxes via surface vegetation and made: hydrological processes coupled to biogeochemistry in variably saturated near-surface systems. PN12032/2433

185 Mathematics and Computing Sciences 186 PN11036/2358 throughout theproject lifecycle. data used ment missing data andassumptions, andtrackof initialdata, docu- largevolumeof need theability toparse example,lenges similartocarbonsequestration. many For Veloface chal- scientificdomainsasmany acrossof several cific customizationVelo. of This enabled the use refactoring ware Suite(GS themwiththeGeologicalSequestration Soft- will integrate model, andmake available abroad We toolsetfor modelers. provide theability totrackdata andassumptionsinagiven Thesecapabilitiesfor carbonsequestration will modeling. frameworks nance capture andtoolintegration andanalysis, In thisproject, weare designingandcreating novelprove- anderror prone.repeatability time-consuming, yet isadhoc, each stageareintegrity, criticaltoanalysis verifiability, and andtracking data at managing, for asinglesite. Organizing, simulations run iterative, leadingtotensandthousandsof tomanage. Theprocess isalso anddifficult tions isextensive condi- toolsapplied undervarying data setsandnumber of Thenumber for of experiencedmodelers. challenge even simulation codesisatremendous uation bythesetypesof and transforming/scaling- intothenumerical modelfor eval domainconceptualmodel, data, buildingthesubsurface However, managing andinterpreting raw theprocess of In FY 2010, we developed the initial version of GS of theinitialversion In FY2010,wedeveloped quantification uncertainty andvisualization. of port GS framework, and deployedframework, GS monitor CO that and themodelingprocess supports usedtoevaluate sequestering CO of thefeasibilityandlong-termeffects and evaluating cesses are expectedtoplay acrucial role inunderstanding coupling physical,mechanical,chemical,andbiologicalpro - multiphaseflow modelingcodes andtransport Advanced Ian Gorton Suite Framework Geological SequestrationSoftware GS plotting toolsandGooglemap interface. geologicwelllogsandpresent using themvisually parse andsearch data. to We metadatatag, extractors developed toannotate,storage alongwithaninterface semantically based SemanticMediawiki toprovide data robust versioned domain. We control systemwith web- combinedversion for thecarbonsequestrationware specifically framework Mathematics andComputing Sciences target sitesforcarbonsequestration. ities integraltoevaluateandmodelpotential a softwareframeworkwitheffective capabil- We aredesigning,developing,anddeploying 3 3 intoVelo, ageneric scientific knowledgemanagement through asoftware toolsintegrated framework isasetof 2 sequestration sites.InFY2011,werefactored 3 ) anddemonstrate theircapabilities insup- 2 in large-scale deep geologic reservoirs. inlarge-scaledeepgeologicreservoirs. 3 asacarbonsequestration spe- 3 asasoft- In FY2012,wecontinued toadd andimprove manyGS tools. Because FY 2013 is the final year of theGS tools. BecauseFY2013isthe final year of STOMPInputAdvisor,graph, andthejobmonitoring In FY2013,wewillcontinue toimprove provenance features. Inparticular, we: non-GPL extensions. it easiertoseparate core GPLcomponentsfrom possible help pages.We willcontinue tomodularizecode, making we willcreate documentations suchasuser manual and Velo architecture. • • • • • • updated to newer version of underlying GS underlying of version updated tonewer STOMP inputfile generation. documents andcodesnippetstoguidethemthrough withappropriatetool, providing users reference theSTOMPInputAdvisor created of aninitialversion to explore various captured relationships canwalktheprovenancethrough graph whichusers provenance interface graph interactive provided anew undertheUQproject runs developed and demonstrate quantification paralleluncertainty (UQ) enabling ustointegrate Institutional Computingclusters, added for support running taskparalleljobsonPNNL directory files intoanew files inscratchpad andcopyorcreate softlinkstothe added theability tolaunchtoolsonacollectionof editor for ASCIIfiles withinthewiki richtexteditorforinterface, userannotation, and file navigation, copy/pastefunctionalitythrough web improved manycore functionalitiessuchastree-based withexternalapplicationsintegration with thelatest securityupdates andimproved APIsfor (Mediawiki andSemanticMediawiki) that provides GS 3 technologies technologies 3 project,

GRADIENT: Graph Analytic Approach for Discovering Irregular Events – Nascent and Temporal John R. Johnson III

detection and analysis on extremely large, dynamic, stream- This research addresses the challenge of ing data sets across a broad range of domains. We will use detecting time evolving signatures in large large-scale graph analytic techniques to enable dynamic sig- volumes of data using one or a few passes on nature detection in massive data sets. The uniqueness in our the raw data. Our approach will be to reduce research is reducing the raw data by transforming the signature detection problem into a graph representation and using the data size by reformulating the problem in large-scale graph analytic techniques and specialized, high- a graph theoretic or sparse matrix framework performance computing platforms to tackle this problem. and then leverage novel high performance In FY 2012, we focused on detecting the signature of a specific computing platforms to mitigate the increase cyber attack against an enterprise called pass-the-hash, which in complexity. takes advantage of vulnerabilities in Microsoft Windows security. We identified, catalogued, and described ten distinct The past several years have seen a renaissance in the study PNNL data sets that could be used for this activity. The first of graphs – or networks – and the development of graph phase of the project focused on three of these data sets: Win- analytics spurred by increased interest in social networks and dows event logs (6TB), firewall data (28TB), and internal net- the growing importance of engineered networks. Innovative work flow data (9TB). From the event logs, we converted the new technologies such as the Cray XMT supercomputer have data into the DIMACS-9 graph format. In this process, the been designed explicitly to support graph analytic computa- data were reduced to 15% of their original size. tion and are effective on irregular problems found in graph We gathered statistics on the graph using variable size time analytics. In addition to naturally occurring graphs, graph windows and variable time steps to characterize its structure. approaches can solve challenging scientific problems in very Statistics included average out-degree, number of edges, different fields, ranging from graph-based 3D flow field visual number of vertices, diameter, maximum out-degree, and analysis to bi-simulation for discrete, continuous systems. average clustering coefficient. We also developed algorithms With the continuing explosion of data across the scientific using randomization techniques to perform matrix operations and analytic community, new approaches are required that on subsampled matrices of smaller size while minimizing can find appropriate, compact representations of underlying error. The approach we used was to coarsen the graph using data to support discovery and analysis. Graph Minors, which enabled us to preserve walks in the Our research will advance the current state-of-the-art graph, an essential feature to help detect the signature by using a graph analytic approach to support signature we wanted. This allowed additional data reduction without loss of fidelity. Results from this research were presented in an invited talk at International Supercomputing in Hamburg, a mini- symposium at the SIAM General meeting in Minneapolis, and have been submitted to the CSIIRW workshop at Oak Ridge. In FY 2013, we will focus on developing techniques for automatic recognition of evolving patterns, triggered by operations such as those identified in the pass-the-hash attack. These automatic techniques will be analyzed to characterize whether and under what conditions the recognition can occur fast enough with respect to the attack to provide an opportunity for mitigation. While the majority of the work in this research will be forensic in nature, a goal will be to understand what the gaps are in moving to a near-real time environment to potentially enable future predictive signature detection. We will extend the basic results to other problems within cyber security and will also extend to other domains.

Reachability graph for pass-the-hash attack PN12037/2438

187 Mathematics and Computing Sciences 188 PN12040/2441 which thepair communicate. While methodstodiscover therequestor,into termsof responder, andthechannelsby isthen defined anetwork. Aservice of ality tomembers provides defined function- Anetwork service work services. entities, wefocused on discoveringandidentifyingnet- of assetsencompassalargenumber Becausecyber it evolves. a continuous thenetwork process as that learnsthestate of is interrelationships betweenthem mapped. Assetdiscovery bediscoveredness processes, andthe theassetsmustfirst Before assetscanbeassociated withtheir respectivebusi- asset inrelation totheenterprise’s businessprocesses. each perform identification, andassessthecriticalityof assetsandtherelationships betweenthem, cover cyber dis- andrapidly methodsthatinvestigate willdynamically pen-and-paper thisproject exercises. wasto Thegoal of exist,buttheseapproachescriticality evaluation are largely assetidentification and tional value. Manymethodsof opera- themodel,itisoftentooabstract tobeof develop do they If objectives. totheirorganizational importance theirbusinessmodelthat identifiesandassigns develop to enterprises expendthetimeandresources necessary thebusinessprocess. Unfortunately, of importance few mally, from therelative theassetisderived thecriticalityof computation, andinformation. threads For facilities, of includestaffing, they andswitches; routers, as computers, are complete tasks.They notjustphysicalcomponentssuch Cyber assetsare allocated tobusinessprocesses to and indirect coststotheenterprise. resources,impacts theavailability causingbothdirect of what computerandnetwork critical operations considers criticaland whichassetsbusinessoperations considers of complicatethat theirroles. willonly Further, amismatch such asthoseproposed for movingtargetinfrastructures environments evolving “static” dynamic, tions intoday’s assessingsitua- currentlyhave andoperators difficulties ers, defend- analysts, these environments ischange;thus,cyber constantin assets.Theonly allocated tocyber ephemerally paradigm inwhichoperational tasksareand dynamically Large enterprise computingenvironments are shiftingtoa Tom Carroll Inferring Criticality Identifying CyberAssetsand Mathematics andComputing Sciences the importanceofassettoenterprise. dependencies betweenassets,andevaluate methods todeterminecyberassets,discover networks. Inthisproject,weinvestigated maintaining andsecuringlargeenterprise assets tobusinessprocessesiscriticalfor Understanding therelationshipofcyber

business process is used to derive the criticality of theasset. thecriticality of business process isusedto derive that business process. Thepriorityassignedtothe of port permits ustoidentifyallassets that are employed insup- discovery thedependency known assetswiththeoutput of assets that are beingemployed.known cyber Combining thebusinesstaskandnote operations assessthecriticalityof extension tobusinessprocess modelandnotation, business municated tocomputerandnetwork operations. Usingan businessprocess; thisinformation rarelycom- tance of theimpor assets.Businessoperations understand cyber Finally, weassociate thebusinessprocess withitssupporting network probing theservice. actively certainty,sure of andwecanimprove by ourcertainty withassociate mea- priate node, wecanidentifyaservice ontheapproBy appropriately- evidence setting thevirtual heuristicsintoaBayesian network. that combinesasetof and, therefore, brittle. We arobust investigated approach protocol information) toidentifyaservice summary (e.g., thesetechniquesrelyon channelproperties assets, mostof While there methods toestablish theidentityof are several ideal for thisalgorithm. by NetFlow, thedefactonetwork profiling protocol, are encompass thecommunication. Therecord setsgenerated searchingaccesses andspecifically for thosepatterns that accesses are what discovered causesthese byunderstanding current requests issolvedbylarge samplingsets.Rare cause problems for withcon- thealgorithm. Thedifficulty algorithms. Concurrentprevious requests andcachingdo approach isthat itislessdependentonbeing“tuned”than thisintuitive respond backtotheclient.Thebenefitof before itcancompleteitstaskand from upstream servers must initiate requests toandreceiveresponses tion: aserver - isbasedonthefollowing observa discovery dependency unable tocompleteitstaskwithout“B.” Thealgorithm for discovery. “A” Anetwork service “A” dependson“B”if is We network-based investigated approaches for dependency byinteractingwiththem. probes services discovery active communication orlisteningintocommunications, while of identifiesentitiesbyexaminingpatterns discovery Passive are discoveredthat andproperly mostservices identified. toensure approaches are necessary discovery andactive sive We network-based pas- determinedthat acombination of towork, deployment onallsystemsisimprobable.devices employees formance, bringingtheirown andinthiseraof on eachsystem.Agentsare andimpactsystemper intrusive preferred. Host-basedapproaches require running anagent subject matter that experts network-based approaches are based approaches, of wedeterminedwiththesupport intonetwork- grouped canbebroadly andhost- services

- Immense Social Media Stream Analytics Courtney D. Corley and entities. This computational advancement will aug- Today’s analysts need methods to identify ment the abilities of analysts by providing them with com- and monitor significant events efficiently and plementary capabilities for analyzing social media streams. In addition, this project leverages PNNL’s institutional com- measure their effects as expressed through puting platform (currently configured with over 600 nodes, social media. Knowing how social media is 32 cores and 64GB RAM per node, Lustre file system). To used during and in response to anticipated develop our algorithms and framework, we are utilizing and unanticipated events (such as natural over 10 billion social media posts including tweets, blogs, disasters, disease outbreaks, speeches, microblogs, mainstream news articles, forums and classi- elections, and crises) enable more accurate fied ads. From these, 60% are in English and the other measurement of the potential effect of those 40% are distributed among at least 60 languages. or similar events and informs planning and In FY 2012, we developed unique analytics methods – response decisions. 34 in total – that span traditionally social media analytics to novel advanced time-series and event forecasting methods. For example, the temporal analytic methods to Existing analytic methodologies do not accommodate the 1) transform semi-structured data to a time series; 2) mea- innate characteristics of social media – they are limited sure distance between time series; 3) interpolate missing by the inability of existing computational methods and data; and 4) decompose time series into seasonal, trend systems to process the immense volume of data in real- and other effects. The value of these methods is that they time and the inability to cope with intentional or uninten- now enable the discovery of signal attenuation and misdi- tional message attenuation such as spoofing, damping, rection in social media. To develop the analytic methods, and thematic and lexical drifts. These types of analysis the project team engineered and developed a Computa- methods on social media streams remain a capability gap. tional Sensor Platform to ingest source-agnostic streaming Our research will develop methods that identify and pro- multi-mode data (>2MB/s). Next, the team engineered and vide access to analytically informative features, such as developed a distributed database, which enables analytics trends and transitions, key terms and topics, and events on massive semi-structured data at interaction speed. We will research methodologies for cutting through the platform noise, resolving message attenuation, leveraging proven techniques, and collaborating with users to ensure we provide analytically useful insight in to immense social media streams. The development of these computational methods at scale will enable the identification and characterization of behaviors within social media and the measurement and impact of these behaviors within social media. Additionally, we will develop visual analytic methods to provide access to these computational methods and to support understanding the pheno- menology of social media. Ultimately, by enabling discovery and exploration of behaviors within immense social media streams this research will empower analysts to anticipate and monitor crises and assess their impacts to the security of affected people and infrastructures. 2012 social media statistics. PN12042/2443

189 Mathematics and Computing Sciences 190 PN12043/2444 advances that areadvances enabled bythiswork are multiple. We use expectations are withindesirable range. Ourresearch time, whethertheapplication’s performance andenergy mance basedonthisinformation; andverifies,also at run- powerandperfor behavior; generates adynamicmodelof that program collectsat runtime information of pertinent aperformance andpowersteeringcapabilitywill develop Specifically,will operate level. at themostefficient we systemmanagementsothat anexascalesystem tion of theautoma- This project focuses ononeessentialaspectof resources. use of toensure themostefficient consumption isnecessary tools toimprove system’s steerperformance andpower collects information at runtime andprovides actionable the exascalesystem.Anautomated systemthat accurately andoverlooks thebenefitsandcapabilitiesductive of detailsiscounterpro- withmanaginglow-level grammers However,or systemadministrators. burdening thepro- these problems are addressed byprogrammers manually form adauntingchallengefor exascalesystems.Currently, technology, poorreliability, difficulty andprogramming about consumption, uncertainty Limitations on energy Adolfy Hoisie Extreme ScaleSystems Integrating AdvancedOptimizationsfor The SMIandRTMsystems. Mathematics andComputing Sciences use ofresources. is necessarytoensurethemostefficient approach betweenhardwareandsoftware To thisextent,anautomatedandconcerted biology, chemistryandnationalsecurity. scientific progressismanyareas,including limited powerbudgetisessentialtoensure Achieving exascaleperformancewitha - exascale runtime systemsthrough co-design. of thedevelopment and tousethisknowledge drive applications’ characteristics,performance/power trade-offs, Finally, systemtoexplore weplantousethedeveloped at runtime. nents that canimprove thesystemefficiency runtime system.We actionable compo- alsoplantodevelop MPI dynamic modelwiththeRTM,SMI,anduser-level We the technique for plantointegrate dynamicmodeling. novelmethodologiesand In FY2013,wewilldevelop application inwork currentlyinprogress. runtime systemfor auser-level MPI implementation of runtimerange, system.The theRTMsignalsuser-level mance orpowerconsumptionare notwithintheexpected thatHPC applications. theapplications Intheevent perfor reduce theruntime overheadintroduce whilerunning SMI andtheRTMwere designedasakernel modulesto andlowoverhead. high frequency perform monitoringwith module designedtoeffectively a specific kernel dynamic model.TheRTMconsistsof consumption meettheexpectations provided bythe verify whethertheapplication’s performance andenergy theapplications andto adynamicmodelof to develop tions andcollectsinformation andstatistics that are used anapplica- theexecution of monitors The RTMconstantly fit, SMIensures portability across systems. different that are usedtoobtaintheinformation. Asanadded bene- theactualhardware performance counters detailof level systemwithoutexposingthelow- tion tothehigher-level example,someperformance the SMIexports informaFor - language)that usetheinformationgramming provided. components(runtime systemandpro- and thehigh-level ware andoperating system)that provides raw information system component(hard- betweenthelow-level interface monitoring system(RTM).TheSMIspecifiesacommon (SMI)andtheruntimelayers: thesystemmonitorinterface twobasicsystem In FY2012,wedesignedanddeveloped various performance andpowerknobs. an instrumented runtime systemthat allowsto“dial-in” to validate the results through theimplementation of exascale architecture. From thesetasks,wewillbe able system information withtheapplicability toarangeof ing systemthat generates therequisite application and inthisarea.the-art Additionally, wewilldesignamonitor thestate-of- advancing significantly performance modeling, for dynamicpowerand will designageneralmethodology - - Integrating Power and Performance Modeling for Exascale Systems Kevin J. Barker We developed methods for optimizing application power As large-scale parallel systems continue to consumption using tools available today. Optimizing power increase in size and complexity, power will and energy consumption is not simply a concern of the hardware; methods must be developed for maximizing effi- become a primary constraint in the design of ciency at all system levels from the hardware through the both machines and software. We are explor- system software and application. PNNL is uniquely posi- ing methods to quantify predictively the tioned in this regard, with expertise in various levels of power consumed during software execution the software and hardware stack. At the hardware level, on a given platform through analytic model- we explored techniques for modifying power consumption ing approaches as well as examine tech- characteristics of devices during runtime (e.g., dynamically niques to optimize power and energy altering the voltage and frequency (DVFS) of CPU processor consumption by attacking all levels of cores to manage more efficiently a finite power budget). the software and hardware stack. For communication software, we explored methods for optimizing energy efficiency of single-sided data transfer operations implemented in the ARMCI communication Supercomputer systems have been designed primarily library. Finally, at the application level, we developed with performance as the main criterion that must be met. Energy Templates, which encapsulate knowledge of per- However, as we move towards the exascale era, power core application runtime behavior in an analytic model consumption will increasingly provide the pressure that to identify periods during execution to conserve energy. will drive system design. Simply scaling up current technol- All these techniques work in concert; tools provided by ogy will not provide an acceptable solution; current esti- the hardware such as DVFS are exploited by higher levels mates for power consumption of such as system range in the software stack that contain a complete picture of from the high tens to hundreds of megawatts, placing ongoing application activity. them outside the possibility of implementation. New solu- In FY 2010, we procured a measurement harness consisting tions are required at all levels of the system architecture of a power analyzer to gather measurements of power con- from hardware through middleware tools and to algo- sumed by relevant workloads running on current and near- rithms and applications with an eye towards optimizing term future hardware. We performed an initial examination both power consumption and achievable workload perfor- of the energy consumption of single-sided communication mance. The past three years of this project (concluding at primitives, where we used an interrupt signaling mecha- the end of this FY) have aimed to address these concerns. nism and dynamic frequency scaling to down-clock proces- Over the course of this project, we approached the chal- sors. In both cases, as message sizes increase, there are lenge from two vantage points. The first was developing a improvements that show reductions in energy consump- set of tools to quantify and characterize power consumption. We also leveraged existing relationships with major tion of a workload executing on a hardware and software hardware designers (e.g., AMD and Intel) for access to next- platform. We developed a set of micro-benchmarks kernels generation hardware and gathering data to characterize that exercise single functions of computing systems (e.g., applications with different requirements (e.g., compute, floating-point computation, memory accesses through lev- memory, and network). els of the memory hierarchy, and inter-processor commu- At this same time, we explored GPU accelerator hardware nication). The second component was developing a tool from the perspective of power and energy efficiency. GPU suite that could characterize the execution of a workload hardware provides a clear performance advantage over in terms of these activities. Measuring the power consumed today’s multi-core CPU processors on a variety of compute- by benchmark kernels and correlating these activities with intensive kernels. Often, this compute performance comes application characterization, the goal was understanding at the cost of increased power consumption. What is less the power consumption patterns of executing workloads. understood is how GPU hardware fares compared with This is challenging work that requires the availability of multi-core CPUs in terms of energy efficiency. With these fine-grained measurement platforms, which have also challenges, we utilized a system in FY 2011 of two Intel encompassed part of this project. The work is ongoing, 6-core Xeon X5650 Westmere processors (with a clock speed forms components of newly funded projects, and will of 2.67GHz) and four Nvidia Fermi C2050 GPUs. Results continue to have impact. PN10049/2291

191 Mathematics and Computing Sciences 192 PN10049/2291 mesh refinement). processor asmay cores result (e.g., from adaptive orduetoanimbalanceintheworkprocessors) across for algorithmic data reasons dependenciesacross (e.g., incurred idletime. However, idleperiodsmay result either chronization pointssimultaneously, little resulting invery pattern at results globalsyn- arrive inwhichallprocessors aretions. Whenallprocessors loadbalanced,anexecution with iterations separated byglobalsynchronization opera- applications processing exhibitsacommoniterative flow these powersaving features. instance, alargeclassof For question becomeshowandwhere totake of advantage idleness,akey power consumptionduringperiodsof mechanismssuchasDFSthat reduceGiven processor mayperiods duringexecution beconserved. where energy tion runtime modeltoidentify behavior inananalytic Templates per-core that encapsulate applica- knowledgeof Finally, Energy at wedeveloped theapplication level, returns. we expecttoseemore substantialenergy significant reductions inassociated powerconsumption, powerstates are madeavailablevariety of withmore ing systemgenerated interrupts decrease andasawider ated withchangingtheprocessor’s- powerstates andservic 256KB. Astheoverheadsassoci- 10% at messagesizesof nearly are possibleusingDFSandhave identifiedsaving of savingsoperations inorder todeterminewhetherenergy We conductedexperimentsusing theARMCIPutandGet for data arrival. wastingpollingoperations tocheck periodic andenergy down-clocking theprocessor cores orthrough eliminating an inter-processor data transferoperation byeither in whichtheprocessor iswaitingfor thecompletionof consumptionduringtheperiod used toreduce energy Bothareor interruptmechanismtosignaldata arrival. signalingmechanism–usingeitherapolling data arrival ormaximizeperformance –andthe energy to conserve altering theprocessor core’s clockspeedduringruntime scaling(DFS)– during data transfer: Dynamicfrequency methods are consumed available toinfluencetheenergy theGlobalArrays current toolkit. Given technology,of two operations within theARMCIlibrary, whichforms thebasis single-sideddata transfercommunication of efficiency weoptimizedtheenergy At theruntime software level, beyond0.7GF/W. efficiency energy count from 1 to4doesnotresult inimproved (0.5GF/W vs.0.27GF/W).However, increasing GPU for theGPU efficiency results inimproved energy on largematrices). equalpowerconsumption Nearly than whenusingall12CPUcores (~275GF/svs.~120GF/s higher performance usingasingleGPUissubstantially using amatrix-matrix multiplication kernel indicate that Mathematics andComputing Sciences contained instrumented powerdistribution unitsable to nected withQDRInfiniBand. Thisstate-of-the-art system AMDInterlagos Bulldozer~20K cores con - processors of tional Computing(PIC)clusterwasstoodupcontaining FY2012,thePNNLInstitu- suitable of scales.Atthestart cated measurement resources notavailable typically at measuring powerat asuitable resolution requires dedi- Testbeds thiswork. Accurate form acriticalcomponentof states increase. application inpower assystemscaleandthedifferences Templateings occurusingtheEnergy approach for this sav- 23% onthetestsystem.Increased powerandenergy of the peakpossiblepowersavings predicted analytically 8%issignificantbutshouldbecontrasted with saving of corresponding power runtime software. Themagnitudeof Templatethe extraoverheadinducedbyEnergy and Templates isdueto iswithin4%;thissmalldifference The execution timemeasured withandwithoutEnergy here.greater thanwhat wehave observed powersavings isexpectedtobeeven cores, observed in powerstates available onmore current andfuture percore.11W powerdifference Withgreater differences to acore oridlewitharesulting beingeitheractive resolved. Two powerstates are available corresponding remain idlewhilewaitingfor data dependenciestobe tion inwhichalgorithmic to designcausesprocessors awave-front applica savings- of performance andenergy The power-instrumented NW-ICE machinemeasured the Energy Templates Energy include: Key principlesof application phasesare oncenew immediately entered. totransitionlowerpowerstatesplates allowprocessors Temated withtransitioning tolowpowerstates. Energy - once itcanbeidentified,increasing theoverheadassoci- idleness time software areable toreact only toperiodsof run - by theapplication; of withoutguidance, lowerlevels decisions.Suchknowledgeisheldonly energy-conserving can thenbeusedtoguidetheruntime systeminmaking identify aprioriperiodsinwhichidlenessisexpectedand application behavior canbeexploitedto Knowledge of • • • • power state. processorabout core’s whethertoalteranindividual enable theruntime systemtomake informed decisions monitoring applicationently activity that enableuse triggers state transitionsbytranspar application’s expectedparallelactivity state toanother usingamodelthat represents the contain rules associated withthetransitionfrom one and idlestates for eachprocessor core that may vary represent active anapplication-specific sequenceof - sample consumption at the power outlet level with 0.33Hz within the system (e.g., CPU sockets, memory) at a very resolution across the full cluster. This represents a sizable high frequency. By determining how power is consumed increase over previously available resources. We are cur- by different application phases, we will endeavor to build rently using the PIC cluster to expand our Energy Templates intelligence into runtime software to optimize power con- work by exploring more applications at larger scale. This sumption at larger scale. The PIC cluster will serve as a vali- work is targeting a publication for the end of FY 2012. dation platform in this work. This is ongoing work and will While the PIC cluster is an excellent resource for exploring continue on other projects funded by external agencies. power consumption at larger scales, a goal of this project This project has yielded important and tangible outcomes, was to characterize power at fine-grained temporal resolu- including the development of methodologies for character- tion. We have the goal of tying power consumption charac- izing power, communication libraries optimized for power teristics to application workload features, many of which consumption, and the Energy Templates mechanism for are at the millisecond or finer resolution. To address this, facilitating information exchange between application soft- we have acquired a system of the same architecture as the ware and underlying runtime software. Looking forward, “quad-nodes” which compose the PIC cluster together with these techniques and tools will be expanded through the a high-speed data acquisition device. Intrusive instrumen- use of testbed equipment and applications of interest to tation on the system architecture will allow for the mea- the Laboratory and external agencies. surement of power consumed by individual components PN10049/2291

193 Mathematics and Computing Sciences 194 PN11045/2367 software agent execution environments simultaneously. multiple andsupports a modernCPUwithMMUsupport) form are that itishardware architecture agnostic(requiring oursoftware plat- of strengths anddistinguishingfactors intheelectric powersystem.Thekey intelligent devices form Volttron™, whichcanbeused to enable distributed oursoftware plat- thedesignof In FY2011,wedeveloped distributedgeneration. of integration sensor cloudimprove systemreliability andsupport distributed control paradigmsenabled bythisintelligent andthe Theintelligentsensors are beingserved. tomers at thedistributionlayersystem especially where endcus- theelectricpower for distributedoperation andcontrol of Our research willlay theinformation network groundwork andislanding. micro-grids anintelligentsystemthat supports form thebasisof can sensors with distributedautonomouscontrollers, concerns about systemstability. Finally, whenpaired systemflexibilityand tion capabilities duetolackof today are unable toaccommodateof distributedgenera- distributedgeneration asthepowersystems of integration power systemtoenable two-wayfacilitate powerflow to theelectric information through thevariouslayers of sharing are capablesystem. Theseintelligentsensors of control systemsusedintheelectricpower security of cyber improve the multiple domainswillsignificantly andcorrelateability measurements toobserve from works andelectricpowersystemsimultaneously. The ments inmultipledimensionssuchasinformation net- performing measure- methods andinthefunctionalityof adjustingmeasurement inthecapability of ifests itself whichman- toadda framework intelligencetothesensors, lize themisnotkeeping pace. Ourresearch project creates magnitudeby2020,butourability touti- of orders several nected totheelectricpowersystemisexpectedincrease con- (includingsensors) intelligentdevices The number of Bora A.Akyol Electric PowerSystem of Autonomous,AdaptiveOperationsinthe Intelligent NetworkedSensorsCapable Mathematics andComputing Sciences control paradigm. basis ofanewdistributedoperationsand electric powersystemtodayandformsthe driven sensorsandactuatorsusedinthe to add“intelligence”command-control Our researchprojectcreatesaframework chased inFY2011wasassembledintoahardware testbed presented apaper at MASCOTS-2012. Thehardware pur theAAMAS2012conference; wealso workshop of aspart platform wasacceptedandpresented at theATES 2012 oursoftware FY 2011.Apaper highlightingthedesignof agent platform basedonthedesigndocumentfrom the of During FY2012,wecompletedtheinitialversion for location functionsandagents. services disrupted andprovides distributed,decentralizeddirectory is whennetworkprovides connectivity robust securityeven and terminate misbehaving agents.Oursoftware platform We alsohave theability tomonitorresource utilization Hardware testbedplatform. and monitoringintheelectricpowersystem. that allowssoftwareframework agentstobedeployed for control Platform providesThe IntelligentDevice asecure, robust, resilient - during FY 2012 and used to showcase demonstrations of source release. We plan to port the platform to another our software platform, including coordination of plug-in CPU architecture, such as an ARM processor, and will inte- electric vehicle charging and cooperative load scheduling. grate with algorithms and software developed in other As part of the cross-focus activities, we interacted exten- related project focus areas. Additionally, we will work with sively with other projects and chose three candidate the PNNL SmartHomes and Grid Friendly EV Charger pro- projects for integration in FY 2013. In addition, we will grams to investigate the feasibility of using our software complete and harden the implementation of our software platform. Finally, we will continue our publishing efforts platform for potential commercialization and/or open with the goal of presenting at another agents’ conference. PN11045/2367

195 Mathematics and Computing Sciences 196 PN11046/2368 resources). thesolver’s This stability isa measure ability of around timethanwhensolvedusingless computational more computational resources turn- faster usage of give largeproblems solvedbycarefulmance computing(e.g., utilize thecomputational resources onthehigh-perfor nels toperform theiroperational tasksandtheirability to around andmatrix-vector timefor ker thelinearsolvers ismeasuredfastturn- byboth Thisefficiency computing. applicationspower grid simulated onhigh-performance andcomputationalcient, stable kernels linearsolvers for - effi thisproject istodevelop kernels. Thus,the purpose of and matrix-vector linearsolvers often arisesininefficient thesesolutionprocedures The computational bottleneckof computing willbenefitthesimulation capabilities. solution procedures that usehigh-performance mathematical equations, carefully designedparallel models,toobtainthesolutionsthese advanced For commercial andondesktopcomputers. software packages rate models,simplerequations were oftensolvedusing thefuture. Inthepast,for lessaccu- modelsof power grid mathematical equations themore underlying complex the mance computingwillbeneededtosolveefficiently willbeanecessity.computers Inparticular, high-perfor highperformance complexity andsize, theutilization of systems. Asthesesystemsandtheirmodelsincrease in and applied research thefuture inadvancing powergrid simulation willplay amajorrole inthefundamental theseinvestmentsisthat modelingandcomputer of applications through computersimulations. Thepurpose science toadvance high-performance computingfacilities have investedin sources), offices andotherDOEprogram andrenewable energy sensors neering technologies(e.g., infrastructures thatelectric energy engi- incorporate new more flexible, reliable, of investigation andhighercapacity inthe Electricityhasinvestedheavily DOE’s of Office Barry Lee Grid Simulations Matrix-Vector forPower Kernels Linear AlgebraSolversandAssociated Mathematics andComputing Sciences utilize high-performancecomputing. for currentandfuturepowernetworksthat kernels willenhancethemodelingcapability simulations. Theseadvancedsolversand ing, throughhigh-performancecomputing kernels, keyenablersforpowergridmodel- advanced linearsolversandmatrix/vector This projectresearchesthedevelopmentof - - - scales (e.g., several hundred cores). several scales (e.g., problems for securityboundariesusinglargerHPC small-signalstability of complete ourinvestigation limitedbudget, wewill FY2013 andwithavery For loadchoices. different with several solve thegeneralizedeigenvalue problem that corresponds to in MPI.Thisparallelization allowsussimultaneously componentsparallelized written inc/fortran withdifferent taken, andthen itisadapted). Thesoftware usedwas transformation andshift(actually, anapriorishiftisfirst restarted Arnoldi(SI-IRA)iteration withanappropriate be found together. We implicitly usedashift-and-invert itself. Rather, approximate ahandfulof eigenvalues must system), thelargestreal eigenvaluecannotbefound by symmetric matrices. Sincenislarge(35,878for theWECC eigenvalue Ax=γBx andthetaskistofind approximate largest real ageneralizedeigenvalue problem requires thesolutionof small-signalstabilitynon-iterative approach. Thisapproach work have andwhere shownpromise, oursolvers isanew net- apowergrid securitymarginsof the determination of stabilityOne important application, in whichisrelevant generalized eigenvalueproblems). eigenvalue problems (i.e., thestandard andthe for thetwogeneraltypesof solvers of development theseeigenvalues.Thistaskentailsthe calculations of transformations for andpreconditioners theefficient withtheappropriateand implementeigenvaluesolvers thepowersystem.We developed theJacobiansof of This stability isdeterminedbyfindingtheeigenvalues under smalldisturbancesabout astable operational point. network apowergrid thestabilityproblems of analyze for small-signalstabilityers problems. Succinctly, these eigenvaluesolv- of include examination anddevelopment milestonesduring FY2012 Successful completionof applications.power grid andstable linearalgebrasoftwarequality efficient for thisproject istodevelop mation. Thus, thetargetgoal of able properties inthemathematical equations isstate esti- application powergrid thatrelevant leadstotheseundesir tomakeinvalid.Onesuch small errors thesolutionstotally amplify dramatically can manysolvers equations, the numerical process of nately, some duetothemathematical properties of to generate accurate solutionstotheequations. Unfortu- γ. Here, AandBare (n×n)real-valued butnon- - Mapping Molecular Dynamics Algorithmic Parallelism to Heterogeneous Architectures T.P. Straatsma Global Arrays feature put-notify, which provides a conve- Novel computer science and software engi- nient means of using a push-data model for communica- neering approaches are applied to develop a tion, instead of the traditional pull-data model. high-performance molecular dynamics (MD) The approach described above has been implemented in a modeling and simulation implementation new MD kernel that in FY 2012 has been extended to the for biomolecular modeling research. application to heterogeneous systems. Additional optimizations for intra-node parallelism and topology-aware assignment of cell-pairs results in a highly optimized simulation A principal and general challenge for biomolecular modelframework capable of achieving simulations of close to ing and simulation in support of environmental molecular 70 ns/day. This is accomplished by communicating data science involves the generation of extended molecular tra- needed by multiple processes on a node only once to that jectories. These trajectories use advanced simulation meth- node, using a polling mechanism for each of the processes odologies to capture those features central to the research to determine the availability of the data. This approach is questions being addressed. This applies to understanding possible as one of the design features of Global Arrays. important biological processes relevant to human health In the simulations, one core per node is reserved for the and environmental issues such as carbon cycling and Global Arrays/Aggregate Remote Memory Copy Interface energy security by creating fuel from biomass. The mole- server thread, which is necessary to effectively deal with cular-level processes to be understood through the appli- the communication requirements of these simulations. cation of computer simulation include protein folding, In addition, three processes handle control tasks, such as protein docking, complex enzymatic reactions, and the global accumulation of kinetic energy and virial for con- association and function of large protein and protein-DNA stant temperature and pressure simulations as well as with complexes. The common situation in all these cases is the I/O requirements such as trajectory recording. Compared need for sufficiently large ensembles of conformations to with the NWChem runs, the new kernel scales to an order capture the relevant events, which requires highly efficient of magnitude more cores and results in a wall clock time computer codes to be designed and applied on the largest per step an order of magnitude smaller. supercomputers. Achieving high scalability to remove Research plans for FY 2013 include the development of current bottlenecks in applying MD simulation codes multi-ensemble implementations, energy optimization on very large processor count computer architectures is using Energy Templates, and the implementation of free approached by formulating an implementation strategy energy methodologies. that leads to significant performance improvements and through an extensive analysis of the algorithms used. Based on a domain decomposition approach, we have designed a new data structure to exploit locality and reduce communication and memory requirements. Distri- bution of the workload taking into account the workload based on relative orientation within the physical decomposition allows a number of processes to be used of up to 42 times the number of cells in the system. The most important challenge for scalability of classical MD is the need for synchronization after the evaluation of forces and after updating the coordinates. Large process counts make the use of global synchronizations, which is acceptable on simulations using less than 1,000 processes, very inefficient. In the redesigned kernel, synchronization is made implicit by processes waiting for the number of expected neighbor contributions before continuing. This implicit local synchronization approach effectively “buf- Wall clock timings for a 39k atom solvated trypsin simulation using the cell-distributed NWChem MD module and the newly designed cell-cell fers” operating system jitter, resulting in much improved pair distributed kernel. scalability. This feature is accomplished in part by a new PN11049/2371 197 Mathematics and Computing Sciences 198 PN12050/2451 red–green–blue joint histograms. Oncethefeaturesred–green–blue jointhistograms. are and descriptors, gradient of pyramidhistogram descriptors, features, includingdensescaleinvariantfeature transform associated images,wecalculated withthetype. several For featuressimilar features. Eachmediatypehasavarietyof tures andthentrainingthesystemtorecognize imageswith are data. created Classifiers bydetectingfea- first cessing of thatvideo data classifiers byapplying run duringthepro- patterns withinimageand withtheidentification of began amultimediarelationship thecreation framework First, of image relationships withtextual relationships. avisualapproach for linking of and thedevelopment for synthesizingmultimediadata, aframework creation of references. We intwobroad areas: madeadvances the logs andthemultimediacontentit activity comprised of data cyber software andanalyzing system for integrating multimedia anew developing In FY2012,webegan learn therelationship patterns inmultimediacontent. toexploit,extract,and the features andentitiesnecessary dia data withrelationships intext.We have beenidentifying that cancombinerelationshipstimedia analysis inmultime- This research systemfor leadstoward mul- avisualanalytic at PNNLandcouldbeappliedlysts toabroader audience. technical research usetoana- andimplementation are of intelligence. The tocyber for multimediaanalysis applying approaches of This work isresulting inthedevelopment cakes, flour,search for imagesof eggs, andoil. canalso and oil.Withamultimediasystem,theanalyst baking acake andissearching for thetextwords flour, egg, mightbeinterested example,textual data. an analyst in For ering patterns that are overlooked justthe whenanalyzing multimediadata willaidinuncov- toreview allows analysts est across information asystemthat silos.Thecreation of inter patterns withtheability of toderive vide theanalyst data ormultimediacontentinisolation, whichdonotpro- toolswork witheitherlog analysis accessed. Contemporary themultimediacontent spreadsheets withoutinclusionof amachinedisplayed incomplex of summarizing theactivity applied tolarge structured orsemi-structured textdata threats hasbeen inlogdata. Thisanalysis of indicators intelligenceisperformed bylookingTraditional for cyber Deborah A.Payne Multimedia AnalysisofCyberData Mathematics andComputing Sciences interests betweenentities. web sitestoinferrelationshipsorcommon similarities fromcyberdataretrieved lytic environmentthatidentifiesmultimedia project isthecreationofamultimediaana- Our approachforcyberintelligenceinthis

- of multimediacontent. of dimensionsbasedonsimilarity adding supplementary visualization andlocations thatwhile actors cangraph Thus for FY2013,wewillextendthisconcept intoanew therelationships amongthesedimensions. of analysis exploring additional that visualmetaphors willaid inthe tiple (oftenthree) dimensionsintheirdata. We have begun therelationshipsnovel toolsfor amongmul- exploration of oftenneed input received, wedetermined that the analysts the of modified thesystembasedonuserfeedback.Aspart We have deployed ourprototype for useat PNNLandhave visualization. search andthemultimediafaceted ship framework operational systemthata fully includesboththerelation- reference themultimediacontent.InFY2012,wecreated termsusedto from thosequery thelogdata, especially that columns anddifferent have matched classifiers certain search toolallowstheusertocorrelate between images search co-occurrence reveals relationships, andthefaceted faceted Themultimedia video clusteringandclassifiers. imageand tion for multimediathrough theinclusionof image andvideodata search providing visualiza - afaceted withtextdata. Weexclusively extendedthisconceptto tion. To date, search faceted visualizations have beenused We chosetocreate search amultimedia faceted visualiza- multidimensionalrelationshipsreveal within multimedia. wouldbenefitfrom anovelvisualizationusers that could gence at PNNL.From thisinquiry, wedeterminedthat the intelli- that work six analysts withcyber of analysis inquiry through auser avisualinterface thecreation of began intelligence. We willbeapplicableour advances tocyber We are working withsubjectmatter toensure experts that visual interface. information intherelationship andfedtothe framework stored log alongwithtextdata gathered from thecyber andtheclusterinformationfrom theclassifiers are then appropriate have not classifiers yet beentrained.Thescores tion totheclassification, whichcanbecomevaluable when theimagedata inaddi- ageneralcharacterization of offer ods afterthefeature identifications step. Inthisway, wecan within imageandvideodata clusteringmeth- byapplying In addition toclassification, wechosetoidentifypatterns washingmachines,andrefrigerators. diagrams, computer-aided designdrawings, circuit charts, images of todetect stration, we trained oursystemonclassifiers For demon- whether animagebelongstoaclassification. classifier toproduce acompositescore, whichdetermines each thesystemcombinesresultsthe exemplars, of exemplar images. Aftertrainingon pre-collected setof vector machines–onefor eachfeature type–usinga support identified, we“train”thesystemusingaseriesof

Multi-Resolution Data Model and Directed Data Reduction, Reconstruction and Aggregation Kerstin Kleese van Dam implies there will be no bandwidth issues in transporting This project is delivering a collection of event the data generated by the PMUs. We published a confer- and relationship descriptions and algorithms ence paper outlining our conclusions. for identifying these phenomena across an During the second year of our project, we focused primarily archive of multi-channel historical data. on developing a better understanding of the generated data, particularly various types of events and translated that understanding into algorithms for use in detecting The future power grid will require bidirectional, real-time these events. We obtained access to approximately data flow to identify and respond to changes in demand, 1.5 years’ worth of PMU data (2 TB) and identified several including quick reaction to extreme events, operational events of interest within this data set. Next, we performed monitoring, daily and medium-term operational planning, an interactive analysis of the entire data set by using the and long-term facility planning. Unlike the current grid R statistical package running over a Hadoop cluster, dynam- with a relatively small number of power sources, the future ically generated on the PNNL Institutional Computing (PIC) grid will have thousands of potential sources, including environment. By using over 415,000 hours of compute time hybrids, wind sources, and solar panels in addition to tradi- on PIC, we characterized the underlying data by interac- tional substations and power plants. Each of these sources tively testing hypothesis against the entire data set. This will operate independently and thus respond differently was critical because many of the events in which we were to fluctuations in price, resources, and operational circum- interested happen rarely, limiting the value of traditional stances. Consumption devices will also be smarter, using techniques such as data sampling. real-time pricing to make decisions about their level and timing of consumption (e.g., an air conditioner will cycle During FY 2012, we spent significant time developing sev- on depending on not only current temperature compared eral algorithms to identify erroneous data elements. These to the desired but also the current price of electricity). Each algorithms range from fairly straightforward data stream of these production and consumption units as well as the analysis (e.g., if the value remains constant for more than connecting power infrastructure will be equipped with 10 min, there is a problem with the sensor) to much more smart sensors that will report on status and deliver opera- complex analysis (e.g., comparing frequency trends across tional information and act as receptors of information PMUs to identify when a sensor is not reporting correctly). and instructions (i.e., reduce demand/production). Coded in R, these algorithms were applied against our entire data set and have been used to generate a “cleaned” The original focus of this project was developing a multi- data version in which all known PMU sensor errors were resolution model as a data transport framework for the eliminated. We are using this new data set as the basis for power grid that compressed identified features and priori- identifying other events of interest, beginning with genera- tized data for transfer based on the importance of the data tor trips. An initial version of our generator trip identifica- and the requirements of the receiver. The fundamental tion algorithm has been created and the results of applying assumption underlying this work was that the amount of this algorithm across the historical data set are being data generated by various sensors in the power grid (PMU reviewed by domain experts. data in particular) in the future would overwhelm communications systems, and this type of streaming re-prioritiza- During the course of this project, we also developed a tion of information would be required to ensure that the repository where we can record metadata about specific critical information was effectively communicated to the events for fast retrieval through either a SQL interface or power engineers. a custom API. We designed an ontology capable of representing events in general and have used it as the basis for During the first year of our project, we quantified the design of the relational database schema. This repository amount of information that can realistically be expected is populated with generator trip events identified by our from PMUs nationwide once the technology was fully algorithms. In addition, we created a custom Java API that deployed. Our best estimate is that there will be a maxi- allows applications to answer specific questions about mum of 50k PMUs distributed across the nation when fully generator trip events. While the standard SQL interface deployed. This leads to an aggregate national bandwidth provides a generic query capability, our API provides a requirement of 144MB/sec. Given current communications domain-specific query capability that eliminates the protocols (e.g., 10GigE, 802.11n) and trends, our analysis

need for the calling application to understand our event PN11054/2376

199 Mathematics and Computing Sciences 200 PN11054/2376 analysis framework. analysis ourdata allowing ustodemonstrate thegeneralityof distribution network, sensor stream basedonthegrid’s from theirdistributionsystemthat willprovide adistinct tionally, wewillwork meterdata withAEPtoobtain smart will incorporate thisinformation Addi intoouranalysis. - tions across overthelast thePacificNorthwest year and We have beencollectingweather loca- data from several what ishappening. of perspective different provides avery frommation PMUdata, isderived theresulting information sensordata. Whilethisinfor of erates aspectralanalysis working onobtainingaccesstotheWISPcode, whichgen- including frequency, phaseangle, andvoltage. We are set alonewillprovide multiplechannelsfor ouranalysis, PMUdata coveringthepastyear.tional 2TBof Thisdata update for ourPMUdata stream that willbringinanaddi- which wehave access.Currently, weare obtainingan data setsto stepistoincrease the Our first number of statistical analysis. morewe willuseitasthebasisfor significantly advanced Once wehave established thismulti-channeldata set, phase angleandexternalinformation suchasweather. suchas will includeotherPMUchannelsinouranalysis data streams (ormodalitiesorchannels).Inparticular, we wewillincorporate ways.in twoimportant First, additional data. Ourfinal year’s work willexpandonthat foundation PMUfrequency asingledata stream –that of of analysis Our initialalgorithms have focused on basicstatistical thus willnotbeanongoing taskfor ourcurrent work. project inFY2013and intoanew tioned andintegrated thisproject willbetransi- infrastructure componentof indicesappropriately.ing queriesusetheunderlying The ontology. performance byensur ThisAPIoptimizesquery Mathematics andComputing Sciences - - will belost. processed informationbecause information ineffectively the modelsmustbeenhancedtowork innear-real time information inorder toexecute.fied amountof Finally, niques) andrequires that aspeci- themodelwork ononly on streaming byemploying data windowing tech- (e.g., ited resources andthusmodelsmustbemodifiedtowork be removed from themodel.Further, haslim- eachdevice must about accesstoexternaldata oraglobalperspective through Asa thatresult, anyassumptions specificdevice. toinformationor PDC)that hasaccessonly that passes aPMU (e.g., model isdeployed asanagentonadevice challenging platform for InVolttron™, data analysis. a onstration deployment methodbecauseitisaparticularly trip detectionmodels.We selectedVolttron™ asourdem- Volttron™ teamtodeploy ourdata cleaningandgenerator limited computational resources byworking withthe transitioning ourmodelstoadistributedplatform with In FY2013,wewilldemonstrate thefeasibilityof platform andthemodelitself. thedeployment boththeconstraintsof standing of existing modelfor deployment involvesadeepunder- close collaboration isrequired becauseadapting an teams toensure ourmodelshave abroad reach. This by otherprojects, withthese weexpecttowork closely for deploying capabilities analysis are beingdeveloped nated for usebyotherresearchers. Whiletheplatforms a generator trip), analgorithm willneedtobedissemi- event (suchasbaddata or detectingaspecific capable of andvalidated astatisticalOnce wehave model developed Multiscale Models for Microbial Communities Haluk Resat that microbe-environment interactions must be considered This project is developing mathematical mod- when the interaction between different guilds (e.g., extra- els that incorporate the possible effects of a cellular enzyme producers and cheaters) are investigated. heterogeneous 3D structure, the functional Preferential flow pathways determine how soluble nutri- biodiversity of the microbial population, and ents would be carried around in the soil system. One open geochemical processes into a unified model question is the partitioning of the flow through the inter- to improve our understanding of the interac- and intra-aggregate pores. We used computational fluid dynamics simulations to determine the flow patterns in tive roles between co-existing organisms in a aggregate-of-aggregates structures. Stacking the aggregates microbial community. with varying distances between them allowed us to compute how the flow is partitioned between intra- and inter- Soil is a highly complex heterogeneous system in terms of aggregate flow pathways as a function of the spacing inhabitant microorganisms and spatial properties. Meta- between the aggregates. bolic activity of microbial organisms is strongly coupled to We found that the soluble factors are mainly transported the local environment conditions. At the microscale, varia- via flows through inter-aggregate pores, and that the flow tions in local environmental conditions cause microbes to capacity through the aggregates is less than 2% when the encounter highly heterogeneous living conditions. Our aim spacing between the aggregates is larger than 18 µ. Such is to develop mathematical models that incorporate the studies enable us to identify what kind of assumptions in possible effects of the heterogeneous 3D structure, the the models are realistic and which computational methods functional biodiversity of the microbial population, and are required to investigate the bacterial dynamics in terres- biochemical processes into a unified model to improve trial ecosystems. our understanding of the role of interactions between co-existing organisms in a microbial community. This proj- We will continue to employ the developed models in pre- ect contributes to predicting the impact of the microbial dicting the impact of microbe-facilitated carbon cycling community-based management practices in controlling processes in terrestrial systems. We will be upscaling the and regulating carbon flow in soils. constructed quantitative models to pore scales with the eventual goal of making them modules for earth system In the earlier phase of the project, we developed a meta- models developed by the wider scientific community. Such bolic network model for bacterial dynamics in soil microag- efforts are being actively pursued by the DOE Terrestrial gregates. We then used our model to predict which type of Ecosystem Science and other BER program offices. cellulose degradation mechanisms might give survival and growth advantage to bacterial organisms in the soil system. We observed that microbial types with different hydrolase enzyme expression strategies can have different functional niches in cellulose utilization. Our kinetic model made it possible to predict an emergent behavior in which co-existence of membrane-associated hydrolase and extracellular hydrolase releasing organisms led to higher cellulose utilization efficiency and reduced stochasticity. Recently, we extended these studies to investigate how the carbon capture and utilization efficiency in soil aggregates depend on extracellular hydrolase production rates by microbes, diffu- sivity of the hydrolases, pore structure and aggregate packing, and insoluble carbon substrate distribution. A pursued sensitivity analysis predicted that the properties of the surrounding environment (i.e., aggregate packing) impact car- Flow patterns through the inter- and intra-aggregate pores in an bon acquisition and utilization by the bacteria more than aggregate of a four soil micro-aggregate construction that represents a the hydrolase enzyme production levels as well as the small segment of the soil system. Spatial dependence of the flow rates enzyme’s biomolecular properties and diffusion constant. is shown with a color map, where red and blue correspond with high In the context of ecological systems, our results suggested and low flow rates, respectively. PN09053/2221

201 Mathematics and Computing Sciences 202 PN12053/2454 Bayesian network model of astate’s cooperationBayesian withinternational agreements. safeguards network model of on the experts’ knowledge.on theexperts’ formulatingprocess, andbegin variousmodelsbased assemble twoexampleswithdata tostudytheanalytic the mathematical for structured framework analysis, data. Ourgoals for FY2012were tocommencedeveloping nuclear non-proliferation modelsand in thecontextof Through thisproject, weare demonstrating thesignatures andmakesexpertise explicitvariousprocess models. modeling formalisms. Themodel-basedapproach codifies and knownrelationships orother asinfluencediagrams issues are addressed byrepresenting expertise analytic thedata usedtoassessnuclear intent.These maries of normalized,context-informed, sum- todevelop expertise project addresses boththeheterogeneities indata and capabilities doesnotreside This withinanysingleanalyst. requiredcrosses multipledisciplines,theentire setof typically Whileexpertise analysts. executed byteamsof information anddata. Currently,of is activity thisanalytic types diverse expertise,andrequires of understanding of nuclear weapons requires program sets inputfrom diverse a orfurther Detecting andidentifyingtheintenttodevelop Paul D.Whitney Multi-Source Signatures ofNuclearPrograms Mathematics andComputing Sciences matter experts. intuition ofnuclearproliferationsubject tures thatareconsistentwiththeinformed features andsupportanalysisofthosefea- will enableidentificationofnuclearization lytic challenges. If successful, these signatures signatures inavarietyofnuclearintentana- demonstrating contextual,model-baseddata This projectisdeveloping,validating,and attributes and facility experts identifiedbythe nuclear engineering broad categories: thosebasedonthelistof Bayesianspecifically fall intotwo networks. Themodels mathematical expert-based models, of development begin Models.Thefinal goal for FY2012wasto Expert-Based data according towhichattributes thedata wasrelevant. critical attributes, andcategorizedfacility eachpieceof relevant alistof alsodeveloped examples, theexperts assemblingthe of Inthecourse critical totheanalysis. data were truly themanypieces of determined whichof source typeswasused(imagery,Theanalysts of text,etc.). eachexampleamix where For theoutcomewasuncertain. confidence, andasecondcase of known withadegree ples were chosen:onecaseinwhichthe outcome was models byassemblingtwoexampleswithdata. Two exam- expert-based of thedevelopment FY 2012wastofacilitate Analytic Examples.Thesecondgoal for ourproject during tion, orcomparingfacilities. signatures outlierdetec- canthenbeusedfor clustering, can bemapped intothedomaincontext.Thesenumeric context, theresulting signatures basedonthosefeatures thedomain basis for features. Becausethemodelscarry features arerow selectedas“topical;”thesebecomethe of distributed compared withthetypicaldistributionacross a types for comparison. Modeloutputsthat are unusually across alldata data objects,thusnormalizingdifferent eachdata object.Themodeloutputsare comparable of thecontent against modelsareexpert-based evaluated data objects,the relevant possibly and acollectionof nuclearmatical facilities features models for evaluating of mathe expert-based - acollectionof Given nuclear facilities. themathematical usedfor assessing framework developed . InFY2012,thisprojectMathematical Framework facility signatures.facility models canthenbeusedto create the be completedandvalidated. Thefinalized models currently underconstruction will models. Inaddition, theexpert-based tion isconsidered inthemathematical order toensure informa that- allrelevant third examplewithassociated data in a In FY2013,thisproject willdevelop validation. stages of invarious models underdevelopment five engineeringandhumanbehavior nuclear technology. Currently, there are toastate’s decision toseek relevant focused onpoliticalandsocialfactors Thelatter modeling. and motivation those modelsbasedonhumanbehavior Nanoscale-Macroscale Three-Dimensional Integration Using High Performance Computing James P. Carson capability, generating 3D datasets from images at this reso- This work establishes the capability to lution and scale is not known to exist at any institution generate 3D datasets of macroscale data at worldwide. The primary impediment to recreating a volumetric image of histological data at subcellular resolution nanoscale resolution by collecting multiple is that the images are extremely large. Each image can be serial 2D images at nanoscale resolution and up to 100 GB in size, and an entire volumetric dataset can developing the high performance computing occupy over a petabyte of data. Fortunately, 3D reconstruc- software to reconstruct the data. The capabil- tion of deformed tissue sections is a highly parallelizable ity to associate nanoscale events rapidly into task that can make efficient use of high performance the context of the macroscale is essential computing resources. to achieving greater innovation multiple In collaboration with the University of Washington, we disciplines. commenced production of serial sectioned multicellular pulmonary datasets. Each image in these test sets is Systems biology assumes that a life form by interactions approximately 80000 × 80000 pixels, which is equivalent between genes, macromolecules, and cells, which do not to the image size produced by a 6400 megapixel camera. act alone. In fact, systems biology presupposes that the These datasets represent different embedding conditions, spatial organization of specialized cells working closely various sectioning distances, and two different organisms: with their neighbors determines the functional and struc- a mouse and a rat. By acquiring datasets under these dif- tural attributes of an organ. Technological advances in the ferent variables, the effects of these variables on computa- tools used in systems biology have led to an explosion of tional performance can be examined. Then, optimal information that has changed our approach to the study of experimental conditions can be determined for future disease. This includes the complete sequencing of genomes data collection efforts. for many different organisms; the development of sophisti- An image registration algorithm based on optical flow cal- cated tools for generating genomic, proteomic, and metab- culations was compiled and tested on the Cray XMT super- olomic data; and the development of interactome network computer for a pair of small test images to ensure proper maps. A significant limitation of these techniques is that functionality of the registration. During this time, the they present a static snapshot of cellular events with no source image successfully deformed to match the target spatial information. The capability to associate nanoscale image. Next, the registration function code was re-written events rapidly into the context of the macroscale is essen- to be suitable for parallelization and Cray-specific parallel tial to achieving greater innovation in research. command-calls were added. Good scalability on a large This research brings genes, macromolecules, and cells number of threads was observed; increasing the number imaged at sub-micron spatial resolution into a 3D organ of central processor unit threads to 3200 from 800 (4:1) system context. Using high resolution histological tech- decreased calculation time by 71% (3.5:1) for registration niques, a small organ, organism, or bacterial population of large images. Additional optimization efforts continued can be fixed, serial sectioned, treated with specific cellular to improve performance. and/or sub-cellular identifiers, and digitally imaged at Warp-smoothing is an essential step during 3D reconstruc- 230 nm. These images can be computationally stacked tion to retain the proper overall shape of the sample. To into a volume representation at subcellular resolution. begin implementation of this step, an initial 3D reconstruc- Development of tion of a low-resolution version of a test dataset was cre- this capability is ated by applying an initial iteration of the warp smoothing part of the emerg- approach. In addition, we tested the application of calcu- ing field of bioim- lated elastic registration warp grids to annotated locations age informatics. within the test dataset. This arrangement will allow fea- Although 2D regis- tures detected in the unaltered data to be properly tration for small translocated into the reconstructed dataset. images is a gener- In FY 2012, we applied the registration and warp-smooth- Processor count scalability on the Cray XMT ally available ing techniques developed in FY 2011 to the high resolution using parallelized image registration code. PN11056/2378

203 Mathematics and Computing Sciences 204 PN11056/2378 calculated by anoptimizertofullscaleimages.Analgo- tool for 2Drigidandelasticregistrations applying pre- capability,a new specifically, parallelsoftware amassively FY2013,thisproject willcontinueFor implementation of ing betweenacquired tissuesections. viadecreasing thespac- improved quality canbeachieved tion inshapes insections100µmapart. Thissuggests tissueandthesignificantvaria- pulmonary naturesparse of formed at thehighestresolutions. Thisislikelyduetothe ing thanthehepatic testdataset for imageregistration per data ismore challeng- 2011. Resultssuggestthepulmonary dataset that wasalsoacquired inFY mouse pulmonary Mathematics andComputing Sciences - computing resources. mens at nanoscale resolution usinghighperformance the capability macroscale for speci- 3Dreconstruction of Collectively, thesefuture project milestoneswillestablish warps from imageswillbedesignedandimplemented. of rithms that minimizes reliance calculation oniterative andaccuratecontinue. warp Efficient smoothingalgo - opticalflow methodwill registration implementation of sor threads available. parallel Inaddition, optimization of culations based onimagedata characteristicsandproces - forimage scaleandregistration thesepre-cal parameters - for calculatingrithm willbederived theoptimaluseof Next Generation Network Simulations for Power System Applications Oreste Villa time intervals, and handle any error messaging or re-tries We are developing a next-generation model- required during the simulation. ing tool that captures the interaction of com- A test case was developed that focused on a transactive munication technologies with the power grid. control strategy simulated representation of the real-time This enables power systems engineers to demand response program of the gridSMART® Demonstra- understand and address the challenges of tion project. It used two-way communication at 5-min communication quality of service on the intervals to communicate a device’s state information and desire to consume electricity to the centralized auction control, monitoring, and operations of device, and the actual value of electricity as determined future power systems. by the auction back to the devices. The process required a robust communication system supporting hundreds New smart grid technologies and concepts such as dynamic to thousands of devices, and significant latency in the pricing, demand response, dynamic state estimation, or communication system can cause power system wide area monitoring, protection, and control are expected performance problems. to require considerable communication resources. As the Balancing distributed simulations in NS-3 was also tested cost of retrofit is too high, future power grids will require in FY 2012. We found that the time to serialize a node (the the integration of high-speed, secure connections with leg- whole TCP/IP stack, routing table, and the applications) is a acy communication systems while still providing adequate significant portion of the run time and hampers the perfor- system control and security. Considerable work has been mance. After reviewing the code, it was realized that the performed previously to co-simulate the power domain problem was because of the strong dependencies between with load models and market operations; however, limited objects in NS-3. We need to investigate the dependencies work has been performed in integrating communications and convert some of the pointer references and remote directly into a power domain solver. references to optimize the de-/serialization operations. In We are developing a scalable, high performance simulation fact, we have seen an instance of this problem in GridLAB- infrastructure for data communication networks for power Dl. Dependencies will cause major problems when we are transmission and distribution power grids. Our goal is to porting parts of GridLAB-D and NS-3 to the new simulator develop a unique infrastructure that will position PNNL as engine and will be investigated in FY 2013. the leader in modeling and planning power grid data communication networks. This project will yield a general scalable high performance computing (HPC) communication network simulation, modeling, and planning infrastructure. We will investigate bottlenecks and limitations in state-of-the-art simulation infrastructures, proposing new approaches, parallelization methods, and high level algorithms to enhance performance. We will map and optimize existing simulations to state-of-the-art HPC computers, proposing new engines for integration with current simulation infrastructures. We will also interface existing power grid simulation infrastructures with the new proposed network simulation, identifying network properties and requirements representative of the power grid network domain. For FY 2012, initial progress for a co-simulation environment for power and communication systems was created and tested using GridLAB-D and NS-3. An open-source Mes- Integrating communications into a domain solver for the simulation of sage Passing Interface (MPI) system was used to create a set communication and power systems to form an interrelationship. Com- of glue codes to synchronize time between the two simula- munication system latency, cyber security, and topology all will affect tors, pass information between the two simulators at given the power system itself and the controls to operate the power system. PN12054/2455

205 Mathematics and Computing Sciences 206 PN12054/2455 file usedtoinitializethesimulation. ponents. Thisinformation isextractedfrom thetopology in theclusterandminimizing communication amongcom- workload nodes todifferent distributing equalamountof assigns simulated componentstocomputing elementsby file loaded timebythesimulationat start engine. NGNPS componentscanbespecifiedinaconfigurationdifferent (i.e., connectionsamongcomponents)andattributes for networks usedtocollectthisinformation. Systemtopology and electricvehiclesorhouses,andtherouters tion of consump- theenergy resenting thetime-series behavior of componentslibrariesrep different - of envision arepository work andinacommunication network. Inthiscontext,we net- entitiesinapowergrid Components canbedifferent source code. theirimplementation withoutreleasing the objects of componentstodistributeonly of This allowsdesigners andloadedat run-time bytheengine. a stand-alonelibrary canbecompiled as APIs,andthey with NGNPSinterfacing Specializedcomponentsmustbecompatible simulating. componentsfor thesysteminterested fortime behavior of theenginehastodefine state-machine. Theuserof typical componentupdates itsstate following states ina specificentitiesintime. The to implementthebehavior of Components are classesthat virtual needtobespecialized each nodethat composesthemachineusingpthreads. multiplenodes(cluster)byusingMPI andparallelwithin of messages. NGNPSisparallelacross amachinecomposed variable timestepandexchange update theirstatus every components that system that canbemodeledasasetof anycomplex works, agentbasedsystems,andgenerally munication net- networks, powergrid supercomputers, instance, NGNPScanbeusedtosimulate computerorcom- that update theirstate andcommunicate toeachother. For variouscomponents systemscomposedof time behavior of and distributedobject-orientedcodeusedtosimulate the tion Network andPower Simulator (NGNPS)isaparallel lation engine for network andsystems. TheNextGenera- anHPC-orientedsimu- Also duringFY2012,wedeveloped applied toreduce simulation models. timefor integrated whetheroptimisticschedulingcanbe will investigate roll back,which needsmodularsoftware. InFY2013,we simulation. Optimistic schedulingrelies onefficient heavily tothesametimeandrestartsbacks thesimulators the sage exchanges. Duringmessageexchange, it also rolls ensures that thetimesare duringmes- synchronized only tokeep timesinsyncandinstead this: itdoesnottry to theother. toachieve Optimisticschedulingisdeveloped chronization whenonesimulator needs tosendamessage However,global timeinsync. needtoensure weonly syn- tion: wekeep waitingfor simulators eachothertokeep the rently, weare usingpessimisticschedulingfor synchroniza- optimistic synchronization simulation. for Cur integrated andprovide bothsimulators increase toleranceof thefault cient roll backmechanism.Withsuchamechanism,wecan - Increasing modularitywillalsoallowustobuildaneffi Mathematics andComputing Sciences - requirements investments. grid withsmart business casescenariosand testingcommunication system for theintegrated of collaborative development dards andTechnology) have expressed interest inusingour Stan- Washington, National Instituteof tory, of University Lawrence (e.g., Berkeley NationalExternal users Labora- whenanappropriate occurs). sient-based analysis event from steady-state,switching totran- time-seriesanalysis reassign timescales duringanongoing simulation (i.e., tor vs.communication simulator); and3)dynamically ferently inapowersystem toolvs.apowersystemsimula- domain solutionmethodologies(i.e., timeishandleddif- ware; 2)synchronize simulation timeacross multiple available andmodifications cannotbemadetothesoft- mercial where software source package codemay notbe set toallowinformation accessandexchange withacom- gluecodes:1)create integration agluecode generic setof goals towards creatingThis willaccomplishseveral amore toform simulation asingle, environment.tors cohesive work, exchanging information betweenthe three simula- solver, PowerWorld, namely intotheco-simulation frame- atransmissionlevel FY 2013willfocus onintegrating theproject. year of software duringthefirst were developed Both piecesof C++codedistributedacross 15classes. ~2000 linesof accountsfor over 20classeswhiletheSer++framework codedistributed approximately 3000linesof posed of for components.NGNPSiswritteninC++anditcom- parently generate serialization andde-/serialization code relies onacompiler-based approach calledSer++totrans- tem for eachspecificentity. To simplifythisstep, NGNPS be implementedbytheuserthat designsthesimulated sys- for classcomponent,theactualcodeneedsto thevirtual While theNGNPSenginecancallde-/serialization functions are usedtoprovide: serialization anddeserialization functions such asNGNPS, In anobject-orienteddistributedsimulation infrastructure • • • • to maximizeoverallperformance. another, destroying andresurrecting themasnecessary components from onephysicalnodeintheclusterto a loadbalancingschemethat (at run-time) migrates on disk. thecomponents by saving and restoring thestate of ties tocheckpoint andrestart theoverallsimulation tolerancebyprovidinga mechanismfor capabili fault - in thecluster. components allocated physicalnodes ondifferent messagesexchanged between collects andaggregate communication infrastructurea low-level that thecluster.nodes of simulated componentsonthevariousphysical of simulated model across acluster, allocating aportion an infrastructure anddistributethe topartition Real-time High-Performance Computing Infrastructure for Next-Generation Power Grid Analysis Peter S.Y. Hui

We are developing a platform for real-time computation in a high-performance computing (HPC) environment with the context of supporting real-time future power grid analysis. Specifically, we aim to provide a real- time operating system for high performance computing machines along with real-time libraries to support key power grid analysis kernels. High-level overview of targeted infrastructure. Compute nodes run a local copy of an existing real-time operating system (e.g., Xenomai), while data is moved between nodes via a real-time Infiniband layer, Real-time computing has been traditionally considered in the developed as part of this project. context of single-processor and embedded systems. Indeed, goals in simulation-based modeling of a power grid network the terms real-time computing, embedded systems, and con- to assess conditions, such as network stability at a given time. trol systems are mentioned in closely related contexts. How- The application models the network using a system of mathe- ever, real-time computing in the context of multi-node matical equations for the states of the network, and then systems (specifically high-performance, cluster-computing solves these equations using an ensemble Kalmann filter. To systems) remains relatively unexplored, largely because until keep up with the rate of data generated by the Phasor Mea- recently, a need has not existed for such an environment. surement Units deployed on the power grid, the goal of the With the future power grid growing significantly, HPC plat- system is to be able to compute results within 30ms. forms are integral to analyzing the grid. This project investigates infrastructure necessary to support cluster computation We developed a method of formally verifying the code to run over large datasets in real-time. Our motivating example is within the stated time constraints, something that has not pre- an analytical framework to support the next generation viously been done in a parallel environment such as ours. One North American power grid, which is growing in size and of the challenges was capturing complex timing interactions complexity. With streaming sensor data in the grid potentially that occur in a parallel computing environment and guaran- reaching rates on the order of terabytes per day, analyzing teeing that with all of the various possible control flow paths this data subject to real-time guarantees is daunting and among the processes and with complex fork/join interactions will require an HPC capable of functioning under real-time between them, all timing requirements could still be met. We constraints. took the approach of first modeling the control flow of each of the parallel processes using timed finite automata, specifically We spent FY 2011 building the infrastructure foundation. We including points at which child processes were spawned and scoped the requirements for a small-cluster computer and joined. Next, we analyzed control flow automata to determine purchased one Infiniband switch and one Ethernet switch. the worst case execution time of all component processes. We We developed a traffic prioritization for the Infiniband hard- also examined the model for possible interactions between ware interconnect that would enable higher priority traffic to the processes. Combining this analysis with worst-case delays, be transmitted with higher precedence than low-priority traf- we determined that timing constraints on the system would fic. For a real-time operating system (RTOS) to be deployed on not be violated. By scheduling all processes with Xenomai, a each of the compute nodes, we evaluated several candidate Linux-based real-time operating system, we implemented a operating systems, which led us to choose Xenomai with its time-bounded matrix multiplication kernel which is capable support for hard real-time deadlines; its simple, clean, “user of provably processing 128 × 128 matrices in under 9ms/cycle, friendly” API; and its wide and active user community. Finally, well within the 30ms delay goal. we prepared to integrate the software stack and sample power grid kernels into the RTOS real-time kernel. We presented a paper and poster at an international symposium and have submitted two additional conference papers. For FY 2012, our focus was developing a real-time version of For these potential presentations, we described methods used a power grid analysis routine, specifically an application for to verify time-critical temporal constraints in a parallel envi- dynamically estimating the electromechanical states of the ronment and our efforts to develop a real-time dynamic state power grid. Briefly, the chosen application targets one of the estimation application for the power grid. PN11065/2387 207 Mathematics and Computing Sciences 208 PN11066/2388 The data fabric of the power grid software integration framework. software integration thepowergrid The data of fabric the efficiency and reliability of power grids. power and reliability of the efficiency applicationsanalysis toimprove that usethesesensors andcontrol and sensors of growth to handletheexplosive real-time data middleware for systems futureof powergrid and control applications. We gathered designrequirements analysis setof alargeanddiverse the data accessneedsof abstractions tomeet is aimedtoprovide ageneralsetof deployed systems.Ourmiddleware sensors grid insmart thedata generatedand deliver from alargenumber of middleware tocollect,select, store, replicate, preprocess, In thisproject, webuiltscalable, flexibledata management applications. accesstoalarge providenumber of efficient applications viable optionto istheonly large number of datadata andthe layer amountof tomediate themassive turn themintoinformation andknowledge. Anintegrated available control decisionsand tounderstand for analysis ever, thesedata arecanbemade notusefulunlessthey electricity. distributionandconsumptionof efficient How- that behaviors enable thehumananddevice understand data mustbecollectedto an unprecedented amountof independence.grid, To smart fulfill theloftypotentialof that willreduce carbonemissionsandenhanceenergy sources variable renewable energy essential tointegrating electricaldistributionandis the monitoringandcontrol of promises toenhanceandautomate technology grid Smart Jian Yin Scalable SensorDataManagementMiddleware Mathematics andComputing Sciences power systemreliability. improve operationefficiency andincrease applications thatcanusesensordatato an increasingnumberofinnovativeanalysis growth ofpowergridsystemsensorsand This projectismotivatedbytheexplosive and reliability of powergrid. and reliability of and visualization software toolstoimprove theefficiency modeling variousanalysis, enables plug-and-play of thefuture. Thisframework of tool for thepowergrid tools intoacoherent, powerfuloperations andplanning software acollectionof software for integrating framework anovel GridOPTICS, thePNNL-developed the core of part applications This data becomes infuture fabric grids. smart andalargenumber of sensors from alargenumber of data middleware tomediate theenormousamountof ascalable and flexibledata management the data of fabric fundamental buildingblock,wedesignedandprototyped parallelism andimprove performance. Buildingonthis oursystemthat canuse of mented adistributable version andimple- solid-state devices In FY2012,weleveraged data. power grid Additionally, ourwork ledtoclientfundingfor managing ourreal-time data middleware system. implementation of andpresentationspapers describingthedesignandinitial for real-time data ingestion.Thisproject produced several mance isalsopredictable andconsistent,whichiscrucial formance improvements overtraditionalsystems.Perfor magnitudeper of orders several tests, oursystemachieved performance. Inpreliminary toachieve puter diskdevices cacheandcachesoncom- Linux buffer system leverages ourreal-time middlewarefeatures system.Notethat our of remote RAMandSSDefficiently. We alsoprototyped major We accesstoutilize canalsoexploitremote direct memory that randomaccess. support memory pen inthetypesof andSSDtoenable mostmetadata lookupstohap- memory localandremote main Ourdesigncanaggregate retrieval. disk access,whichallowscustomizedindexes for data In oursystem,adata one istranslated insertion intoonly in mainmemory. minimize metadata kept tothepointthat itcanbeeasily that datathe fact are intemporalorder, ofteninserted we Byleveraging data andfree spaceinthestoragedevices. of We alsodesignedspecializeddata structures tokeep track inode block,variedindirect blocks,orthedata blockitself. access canleadtomultiplediskaccessesfor reading the suchasfile systems,inwhichone interfaces higher level ing highoverheadsandunpredictability associated with eliminat - interfaces, layer using blockdevice block device sensordata writtendirectly tothefilepower grid system specialized logstructure-inspired storageschematostore Specifically, sensors. a largenumber of wedesigneda high-speeddata from for rapid anddeletionof insertion meet real-time requirements, andthestorageisoptimized Our systemisdesignedfor scalability andpredictability to - - To bridge the data access for different control purposes, we cases. Empirical studies and experimental results shows provide a scalable, lightweight event processing layer that that our system 1) provides essential system building blocks hides the complexity of data collection, storage, delivery to enable future power grid applications; 2) provides the and management. The main challenge is the requirements essential real time data fabric for GridOPTICS; 3) achieves to access huge amount of real time data. We address this real-time predictable performance for data access; 4) per- challenge though scalable system architecture that bal- forms close to the hardware limit; 5) offers linear scalabil- ances system performance and ease of integration. While ity to multiple nodes; and 6) allows easy integration of a high-level data and communication software layers make large number of diverse power grid applications that integration easier, indirection causes high overhead and requires reliable, efficient, real-time access of power unpredictable performance. We designed our system to be grid data. as close to hardware as possible and still provide a thin and In FY 2013, we will build on the previous work and con- light weighted layer to allow easy integration of a diverse tinue to improve the performance of our systems. We will set of applications. also investigate the reliability and fault tolerance design The detailed system architecture is shown in the figure. We of our middleware. Finally, we will design and prototype applied our framework in integrating several power grid system mechanisms to improve system reliability through analytics and visualization applications and the initial pro- replication and adaptive reconfiguration. totype of GridOPTICS was demonstrated with several use PN11066/2388

209 Mathematics and Computing Sciences 210 PN12070/2471 ability to equip software services asAFworkflowability toequipsoftware compo - services forAFMiddleware. WeSemantic Services provided the similarcomponents. themostsemantically quantitatively potentialsequenceorprofile measuring detectors, ette of sequent goals inselectingfrom are apal- users tosupport sequencefiles asaBoolean match/mismatch. Sub- arbitrary executable byproviding programs semanticannotations of compatible withsequencesrepresentingcally entire specific natures discovered executables intestprogram are semanti- initial determination iswhethersequences representing sig- while remaining The generalizability tobiologicalanalysis. coherence whilefocusing onqualitative, symbolicdata (DIA) project withAFdemosandprovide that integrates Algorithms forsive Bioinformatics-Inspired SignalDetection withtheData-Inten - interactingclosely sequence analysis, cyber- semantic technologiesbyselectingthedomainof We soughttoinstantiate anddemonstrate thevalueof todemonstrate ment sufficient applicability andfeasibility. - develop domain-specific demonstration inthecontextof requires aphasedapproach. Initially, weprovide afocused, applicabilitying abroad like area signature of science semantictechnologiestarget- Selection.Developing Focus components that are notapparent observation. at first signature workflowsfor of re-use andtheir andintegration thecapabilitywe willsupport todiscoveropportunities flows withperhaps somemodification. Mostimportantly, existingcomponentsandwork- foropportunities re-use of andprovide meansfor thesetousers themtoexploit reveal totheAFprovideprovide capabilities software to services these similaritiesinworkflows and/ortheircomponents; sics orinsiderthreat discovery, We respectively). willidentify chemical workflowsforen usefulinanotherdomain(e.g., - workflows, analytic orentire analytic of ponents, portions com- may analytic have individual or biomarker discovery) ite signatures. onedomain (bioforensics Aworkflow toserve compos- workflows andidentifynew signature discovery of guide scientistsandresearchers toautomate construction methods,measures, andalgorithmsThis project to develops Cliff A.Joslyn Semantic Workflows forSignature Discovery Mathematics andComputing Sciences algorithms. dations aboutinteroperablemethodsand measurement ofsimilaritiesandrecommen- type andcontentofinformationflowsallows semantic typespecificationsrepresentingthe workflow componentsanddataitemswith to theAnalyticFramework(AF).Equipping across domainsbyprovidingasemanticlayer We seektogeneralizesignaturemethods Workflow relationships andprocesses. andontology concepts. for measuringoverlaps of betweenmultiplegroups measure aHausdorf Future work includesexplication of conceptsinsemantichierarchies. of by groups regions circumscribed tances betweenconceptsandsizesof cal methodsfor metricsinordered setsfor measuring dis- generalization andselection.We ourmathemati advanced - user-cases data for andcomponent recommendation of workflows, components,anddata items,inthecontextof methodstomeasure analytical similaritiesof testing of comparison requires implementation, and development, annotationComparing SemanticAnnotations.Quantitative and agenciesfor collaboration development. andfurther UBuffalo, Mitre). We are(UMBC, outreaching tosponsors acquired from academicandgovernment communities domainontologies willinteractwithcyber tions (UBuffalo) - for BiomedicalInvestiga andtheOntology ture Discovery ourSigna- Atthesignature-genericinter-domain level, level. at the andourownSemwork ontology (UBuffalo), Artifact top ontology, weare exploitingtheSequence, Information asa (UBuffalo) onBasicFormal tem communities.Relying ontologies from signature science, cyber, andknowledgesys- developed andexternally project usecaseswithinternally We termsfrom other (conceptual) development. integrated resources requires lexical(terminological) andontological neededsemantic anddeveloping integrating, Identifying, . Semantic ContentDevelopment forSignature Discovery ing at workflow construction andexecution times. similarity. type-match ThisallowsBooleanandquantitative - andmeasure componentports to register thesemanticsof andSDIKnowledgebase(semantictriplestore) the AFLibrary Component Annotator, andWorkflow Creator) interactwith through SAWSDLprotocol. Userroles (ComponentDeveloper, anddata componentports itemsintheAFLibrary tions of nents withsemanticmetadata, includingsemanticannota- Signature Discovery Analytic Framework Adam S. Wynne the analytic framework consisting of the following high- Next-generation signature detection capabili- level elements: ties require a signature development plat- Signature Discovery Workbench. This will be an innovative form that provides advanced tools and desktop application supporting both novice and experienced services to discover, verify, and exploit com- users that leads the user through the signature discovery pro- plex threat signatures in multiple heteroge- cess and recommends computational services and workflows to complete the user’s stated task. neous data sources. This project creates a software architecture that integrates novel Analytic Components. These are the algorithms that implement the various signature discovery tools, which are made and existing data analysis algorithms into a available to client applications via the Signature Analytics single, enterprise-strength system that leads API as web services. users through a methodology of systematic Signature Discovery Library. This is a database with an asso- signature discovery across multiple domains. ciated API that stores workflow templates, components, and data associated with signature discovery experiments. The other components draw on this information to measure Software tools and infrastructures for the discovery of the similarity between services, make connections between domain-specific signatures are in use and well known, but domains, and make recommendations to users. none has the flexibility and reusable methodology required to be extended to multiple domains. These tools are typically Knowledge and Recommendation Engines. These contain applied ad-hoc for the task at hand or are integrated using the signature discovery knowledge base and assist the user in workflow software, which provides a graphical interface to choosing and building workflows as the user proceeds integrate disparate software tools. In either case, methodol- through the signature discovery process. ogy is not made explicit in that it is tied to a particular scien- We implemented, delivered, and demonstrated a set of inte- tific domain or not documented systematically. This makes it grated tools enabled by algorithms developed on the initia- difficult for signature development tasks to be repeated and tive throughout the year. The demonstration and testing of applied in other contexts. Further, although the workflow these tools focused on scenarios that illustrate common sig- tools help to automate and integrate distributed computing nature discovery tasks. These capabilities consist of 10 web resources, they are often challenging to apply in practice and services and 5 workflows covering 11 separate tools delivered cannot individually take advantage of the high-performance by individual projects. The workflows retrieve input files computing resources required to solve many of these prob- from the library and store intermediate and final results lems. To build a system that enables users to employ a fun- back to the library. In addition, we extended the Taverna damentally new methodology, we are conducting a literature workflow environment by creating a plugin that allows a review on current signature development practices, and we user to upload files to (and download files from) the library are learning about the generic tasks that a signature analyst that are used as input to workflows. Specifically, the tools must perform and the algorithms that must be used regard- implemented were integrated into workflows to support the less of domain. To support the execution of these tasks, we following scenarios: are also conducting research on software architectures and model-driven software engineering techniques that can be • comparing the quality of an existing signature used for partially automated workflow creation. for detecting a threat chemical against a signature for the same phenomenon produced by a sparse The objective of this project is creating a novel software representation signature architecture that combines a rigorous signature development methodology (developed under another research task) • comparing the accuracy of signatures to detect a with highly flexible, robust software architecture and associ- set of land types using satellite imagery data ated tools. Thus, this project will result in a lasting software • using an alignment algorithm from bioinformatics toolset that can be used to develop signatures in a wide to identify related executable files of interest. variety of domains. This will help achieve the central goals of the initiative by reducing trial and error, allowing users In FY 2013, we will continue to work with other project to reuse existing tools and approaches, and providing a teams to encapsulate their algorithms and tools as compo- reproducible mechanism for constructing and evaluating nents that are available to other users. The work will be signatures. led by the individual projects and supported and guided by analytic framework team members. In FY 2012, we worked toward the above objective by creating a service-oriented design and comprehensive vision for PN11067/2389 211 Mathematics and Computing Sciences 212 PN11068/2390 address the following types of questions: address thefollowing typesof operational environment. SQMwillenable scientiststo deployment inan prototypes, andeventual of evaluation research construction and anddevelopment, stages of asignature movethrough the systemsasthey quality of withaholisticapproachdecision makers for assessingthe and SQMistoprovide researchers, developers, The goal of improve it.” to Lord you Kelvin: cannotmeasure it,you “If cannot summed upbythiswell-knowncommentattributed signature qualitymetrics(SQM)canbe significance of theirintendedpurposes.signatures Theneedand achieve signatures willhelpresearchers ensurethe qualityof that sequently, for assessing adomain-agnosticmethodology scientific disciplinesand purposes.for manydifferent Con- cised bymanyscientiststheworld over, working inmany and detectionareconfirmed that exer signature discovery literatureExtensive searches at PNNLhave unequivocally Landon H.Sego Signature QualityMetrics Components of theSQM methodology.Components of • • • Mathematics andComputing Sciences particular application. goal ofidentifyingoptimalsignaturesfora given setofsignatures—withtheultimate costs, risks,andbenefitsassociatedwitha a calculusforexaminingtheperformance, the qualityofthesesignaturesbydeveloping Our projectaddressestheneedtoassess scientists invirtuallyeveryfieldofstudy. ize aphenomenonofinterestareusedby Signatures thatdetect,predict,orcharacter poor signature? a What canwedotoimprove theperformance of creating one? anew even existing signature bymodifying that signature or Does itmake sense toattempt toimprove uponan Is onesignature betterthan another? - - own projects. tools that willguideresearchers SQMtotheir inapplying projects. Finally, thesoftware wewillcontinue todevelop application other areas, tosupport especially context of we willcontinue todemonstrate andrefine SQMinthe the state-of-the-art inassessingsignature quality. Second, we plantoconducttheoretical toadvance investigations Our research planfor FY2013hasthree components.First, proteomics. an essentialcomponentof Yersinia pestisdissemination; and3)peptideidentification, Bacillusanthracis or toring; 2)bioforensics of analyses forareas: radiation moni- PVTsensors 1)gamma-ray portal We SQMtothree have completedtheapplication nearly of detection process inanydomain. asignature used toassessandvisualizethequalityof metricsthat may be theSQMproject isasetof outcome of to compare usingdecisiontheory. variousalternatives The that canbeused metrics intoasingledecisionframework alltherelevant SQM alsoincludestechniquestoaggregate ity, humansafety, use, andthelike. easeof DSAincludetime-to-decision, systemportabil- Examples of that are notalready accountedfor byfidelity, cost,orrisk. which may distinguishonesignature systemfrom another Domain specificattributes(DSA)includeanyotherfactors negatives. and false positives between false eoff assessthetrad- detection system.Thishelpsinvestigators thatwith decisionerrors may result byemploying the tothelikelihoodRisk refers andconsequencesassociated associated labor costs. consumable reagents, and detectionsystems,thecostof of or utilizethesignature system.Examplesincludethecost toresourcesCost refers expendedtodevelop, deploy, and/ F-score, andthelike.curves, includes metricslike sensitivity, specificity, accuracy, ROC interest. This predicts, orcharacterizesthephenomenonof tohowwellthesignature refers Fidelity systemdetects, four components: We explainSQMintermsof efforts. actualsignature discovery inthecontextof odology suring signature quality. We are demonstrating themeth- asystematic approach wedeveloped for mea- perspectives, of ence, anddecisionscience. Drawing onthisdiversity chology, operations research, computersci- riskanalysis, includingstatistics, backgrounds, economics,psy- diverse with experts hasbeentoassembleateamof Our strategy Single Node Optimizations for Extreme Scale Systems Daniel G. Chavarría-Miranda incur a high cost in data replication due to private per- This project investigates the necessary design process address spaces; this will no longer be feasible in and implementation techniques to support the exascale era due to expected reductions in per-core available memory. efficient utilization of node resources for extreme scale scientific applications. We In FY 2012, we focused on three primary activities: devel- believe that the efficient use of node memory opment and performance analysis of multithreaded ker- and processing elements are principal factors nels for the force calculations in the ARGOS classical molecular dynamics application and the performance to achieve the levels of scaling and perfor- evaluation of schemes to reduce metadata overhead. mance required for exascale applications. We implemented a full set of multithreaded routines for the force calculation in the next-generation ARGOS classical Extreme scale systems provide an unprecedented opportu- molecular dynamics package. ARGOS has two different rou- nity for scientists to attack larger, more complex problems tines for the force calculation: self-self interaction (particles and reach solutions dramatically faster. These new archi- from the same cell) and self-other interaction. These rou- tectures promise to be qualitatively different from existing tines have three versions for different interactions between computers, and extracting their full capabilities requires molecules – solvent-solvent, solute-solute and solvent-sol- significant innovations in programming models and the ute – for a total of six multithreaded routines. Preliminary underlying primitives used to implement them. The next results show that OpenMP implementation is scaling fairly generation of computers will be characterized by larger numbers of nodes, each with hundreds to thousands of cores. Although the collective capabilities of each node will be large, the resources available to individual cores will be much smaller. Resources will need to be shared to a much greater extent between processes than before, much higher levels of parallelism will need to be exploited by applications, and sophisticated scheduling will be required to avoid very limited bandwidth. Project outcomes include development of new optimization techniques for computation on emerging extreme scale nodes; the integration of such capabilities with Global Arrays (GA) for system-wide scalability; and the demonstration at scale using a molecular dynamics driver application. Applications with communication-intensive execution patterns such as molecular dynamics can benefit even more when they can effectively exploit intra-node resources (shared memory, multicore processors, accelerators) to reduce their communication requirements. Careful assignment of logical tasks to multicore nodes to exploit data access locality is essential to reducing inter-node communication requirements for communication- intensive applications. Most HPC applications have been developed using the Single Program-Multiple Data (SPMD) paradigm, which typically is realized by starting up a number of sequential processes (0…p - 1) on individual “processors” (cores in modern systems). Parallelism is typically realized only through this mechanism. Thus, this model can only imple- Execution time per MD step and parallel efficiency for the multithreaded force calculation in the ARGOS MD package. ment explicit parallelism in a single level. SPMD models PN11069/2391

213 Mathematics and Computing Sciences 214 PN11069/2391 calculation of theSCFapplication describedlastyearcalculation and of thetwo-electronWe usedacommunication traceof systems). processes for exascale total processes inthejob(millionsof constant (O(1))rather thanproportional tothe number of dedicated tometadata usecanbeproportional toafixed ing ontheapplication’s needs.Inthismanner, thespace insizedepend - use aconfigurable LRUcachethat canvary infrastructure We support. enable computeprocesses to application’s usedin data structures andminimizememory resources for the tomaximizeavailablememory memory hand, thecomputeprocesses canbettermanagetheir metadata requests from computeprocesses. Ontheother andprocessing resources (cores)their memory toattend to application’s computation andwouldinsteaddedicate wouldnotparticipate inthe processes. Metadata servers metadata ondemandtothecompute storing andserving wouldbeprocesses dedicated whichessentially to ers,” - “metadata serv incorporatesOur framework thenotionof communication operation. every uted metadata, whichmustbe“remotely”queriedon distrib- process tofully metadata onevery replication of ate metadata different from managementstrategies full - We toevalu alsodesignedandimplemented aframework tithreading diminishesfor smallinputdeck. thisrelatively per process), performance boostthat isobtainedfrom mul- cell-cellinteractions cesses increases (smallernumber of GApro- that asthenumberpresented. of Itisobserved theOpenMPexecution isalso of as well.Parallelefficiency GAprocessesper GAprocess withanincreasing number of is presented OpenMPthreads for anincreasing number of 40,000atoms. Thetime molecules inwater hasatotalof Trypsin The execution timeperMDstepfor asystemof except for configurations withmore than four threads. othercore domain). OpenMPthreads are mapped onevery use allthecores attached tothat domain(eightcores per domainspernode)andenable itto NUMA domain(four cores. Onthissystem,wemap eachGAprocess toa of pair on thenode, aswelltheresources shared byevery system thememory duetothenon-uniformity of sensitive threads andprocesses tocoresperformance highly ping of which usesAMDInterlagos making processors, themap - ducted onthePNNLinstitutionalcomputingcluster(PIC), managingtheparallelism.Ourexperimentswere con- of parallel work is small compared to the overhead amount of cess ownsjustoneortwocells.Thisisexpected,asthe decks whenwegettoconfigurations inwhicheachGApro- overhead from theOpenMPparallelization for smallinput well asweincrease core However, numbers. there issome Mathematics andComputing Sciences activities: In FY2013,thisproject willfocus onthefollowing ametadata cachingmechanism. of cating theeffectiveness theremote network transfer, tothecostof only thusindi- cachemisses isproportional is quitesmallandthecostof cachehits from the plots,thecostof As canbeobserved it. processes to1/8of proportional tothetotalnumber of hit andcachemissfor cachesizesrangingfrom different plots present theaverage timeinmillisecondspercache cesses (16to128processes, oneprocess pernode).The computepro(4 to7)attend- anincreasing number of dedicated metadata servers number of uses alogarithmic GAinstancemetadata. Theexperiment mapped itsuseof Metadata query costwithLRUcacheanddistributedservers. Metadata query • • • • message-based PGASframework. active theGlobalFutures toanew port framework ARGOS moleculardynamicsapplications the inthe contextof thesedevelopments evaluate NUMAlocality that take of advantage design andprototype multithreaded execution models for models support PGAS-basedprogramming allocation NUMA-aware memory design anddevelop SoftShuffle: A Game Changer for Secure Software System Implementation Mohamed Y. Eltoweissy ing shuffling of different implementation variants. Specifi- This project designs and implements secure cally, SoftShuffle will provide a runtime environment that resilient software systems that exploits mul- is resilient to attacks targeting software implementation flaws with a software development platform design that tiple, preferably independently developed allows programmers to develop multiple alternate imple- implementations of system components to mentations of system components. introduce implicit behavioral non-determinism and employs automated recovery from With SoftShuffle, we perceive a software system comprised of interconnected components that act and interact to implementation flaws to enhance system achieve a desired functionality. Each component adheres resilience. SoftShuffle will induce a paradigm to certain specifications defining its explicit behavior. At shift from today’s software monoculture to the same time, each component exhibits an inherent attain security and resilience through diver- implicit behavior due to the potential individuality and sity, shuffling and automated recovery. variability across different execution environments and different implementations of the same specifications. Accordingly, we can classify the types of vulnerabilities Developing secure, resilient software systems is challenged of a software system into either specification or implemen- by implementation flaws that could lead to serious security tation. The latter vulnerability is much more difficult to vulnerabilities despite careful auditing and analysis of the detect because auditing the implementation requires software specifications. The implementation flaws are greater effort than the specifications and is usually usually difficult to detect and fix before the system is oper- infeasible, especially for commercial, off-the-shelf ational due to the prohibitive cost of auditing the imple- components. It is also implementation dependent mentation or testing all possible inputs and system states. and almost unavoidable. Anonymous intruders could exploit implementation flaws to gain partial or total control of the system, or cause the During both FY 2011 and FY 2012, we identified system to crash leading to undesirable downtime. To several gaps and have enhanced our design accordingly. address this class of vulnerably, a novel approach for Specifically, we pioneered a biologically inspired active implementing mission-critical software systems is pro- component architecture that is composable, online proposed. SoftShuffle can tolerate the existence of implemen- grammable, and intrinsically resilient. We developed the tation flaws while considerably enhancing the security bio-inspired cell-oriented architecture (COA), which is the and resilience of the system. SoftShuffle decouples soft- foundation of the SoftShuffle that is being incorporated in ware specifications from system implementations and our cooperative autonomic resilient defense platform for performs “hot” shuffle among a multiplicity of diverse cyber-physical systems, built a demo of the COA engine, implementations for the same specifications. and integrated it with our automated recovery subsystem in a unified architecture. We have also successfully eva- The primary objectives of SoftShuffle are to confuse the lauted behavior non-determinism and developed a novel attacker by non-determinism through hot shuffling of soft- approach to measuring the amount of confusion and diffu- ware system component implementations, to improve the sion in the system at any time. As a result of this project availability of the system by providing means to automati- during this past FY, we have presented and published four cally recover from runtime flaws, and to enhance the soft- peer-reviewed papers. ware system by survival of the fittest through trust-based component selection in an online software component By introducing non-determinism of implicit behavior into marketplace. SoftShuffle makes software systems moving the system and employing automated recovery, the effort targets, thus effectively transforming white boxes into of discovering and exploiting implementation vulnerabili- “hardened” black boxes at runtime. As an added benefit, ties becomes futile. Further, continual enhancement in we envision the development of a new marketplace para- components is achieved by fostering an electronic digm for software implementations. The goal is to produce marketplace for component implementations. software component architecture capable of maintaining To make sure that our system is fully prepared, we have the operational integrity of the whole system while allow- continually conducted rigorous evaluations. PN11071/2393

215 Mathematics and Computing Sciences 216 PN11074/2396 data and view results.data To andview address thisneed, wedeveloped experiments, having towaituntiltheconclusion toprocess lacked real-time feedbackduringmulti-hour technology imaging thisnew complex systems.Scientists developing for tial imagingusingsoftionization massspectrometry spa- thattwo otherprojects, developed includinganeffort to chemicalimagingscientists.We applied thiscapability to feedback toallowinteractive userinterfaces with graphical chemical imagingresearch. We created aprototype tool cific needsandskills infeature for recognition andanalysis process toidentify spe- In FY2011,weinitiated adiscovery alization/interpretation. image-to-imageregistration,segmentation, anddata visu- data processing stepssuchas tools toaugmentpositively acquisition tofinal results interpretation, whichwillallow spanningfrom datainto animagehandlingframework image acquisitionprocess. Thetoolkit willbeintegrated provide more direct feedbacktoresearchers duringthe possibleand acquisition protocols analysis tomake new andcustomizable tospecificchemicalimaging integrated atoolkit withfeatureto develop recognition algorithms for theirchemicalimaging-baseddata.analysis Thegoal is feature interactive recognitionallow scientist-driven and This project capabilities wasinitiated toestablish that new to nanometerscale. rials, andbiologicalmoleculesat resolutions rangingdown toimagechemicals,mate- techniques are beingdeveloped lection instruments andfeature types.Currently, toolsand thespecificdata col- approaches needsof totheparticular algorithms accessibletoscientistsandcustomizesuch erally, there isaneedtomake existingfeature recognition capabilities. Gen- suchnew stepinachieving an important informationand integrating from multiple data sources is instrumentation new duction andutilization. Developing pro national- interest, includingenergy multiple areas of chemicalprocesses iscriticaltofuture researchdetails of in The ability to seebothintimeandspacethemolecular James P. Carson and AnalysisforChemicalImagingData Synergistic IntegrationofFeature Recognition Mathematics andComputing Sciences production. of soilremediationandrenewableenergy facilitating knowledge-discoveryinthefields datasets. Theimmediateimpactincludes nition ofverylargespatialtime-varying to performimageanalysisandfeaturerecog- new anduniquecriticalresearchcapabilities This projectisbuildingthefoundationfor ing imagingexperiments. andmorefaster accurate feedback toresearchers conduct- capabilitiestions tochemicalimageanalysis byproviding future project milestones willmake significant contribu- for managingchemicalimagingdata.framework These intoacommonimage handling kit willbeintegrated the tool- cells.Theinputandoutputof withinliving effects detectingnanomaterial andnanomaterial to thetaskof In FY2013,thefeature recognition toolkit willbeadapted imagedata catalysis are unsuitable formined if analysis. based onfeatures stored inaqualitydatabase that deter atest Thiswork developed analysis. geneous catalysis hetero- inthefieldof objectives incorporatedtionally new and thecapability for chemicalimagingresearch. We addi- thetoolkitkit wasrefined toincrease boththeutilityof imaging projects. InFY2012,thefeature recognition tool- otherchemical enabling theirapplication toavarietyof Our feature recognition toolsare modularandexpansible, interestthat inthisspecificdata. highlightfeatures of tool currentlyincludescustomizedimageprocessing filters spatial featuresthe visualization involumetricdata. of This noted. We aprototype portable toolallowing developed adjustmentsare necessary is stillavailable if butonly iment canbeimplementedwhileaccesstotheinstrument feedback ondata qualityquickly. Adjustments totheexper instruments islimited,thusmaking toreceive itimportant tron microscopes andbeam-lines.Allocated timeonthese lected from multipleimagingmodalities,includingelec- and functiondata bycorrelating volumetricimagescol- microbial investigated biofilmThe secondeffort structure experimentpriortocompletion. or restart afailed adjust theinstrument duringtheexperimentinteractively bycreating instrument thepossibilityto technology tion of andapplica- experiment. Thisenhancesthedevelopment are asthey collectedduringthe visualized andanalyzed a toolthat data allowsspatial tobe andspectralfeatures of multimodal integration framework for chemicalimaginganalysis. framework multimodal integration feature recognition onthe role andimageanalysis of The interactive - - Targeting Extreme Scale Computational Challenges with Heterogeneous Systems Oreste Villa based on the type of input used. We have implemented Future high performance computing (HPC) runtime auto-tuning strategies based on heuristic and table systems are expected to integrate heteroge- to select the most efficient (in performance and power) options and methods for specific index range values and neous processing elements, such as GPU orders in contraction. We have demonstrated up to 8 times together with 86 conventional multicore performance improvement respect to use conventional processors. With this level of complexity, it approaches. becomes paramount to develop strategies Module designs. We have designed a generalized chemis- and approaches to target these platforms try reaction module to be included in subsurface codes, cli- efficiently. mate modeling codes and carbon sequestration codes. We designed a C code version of the chemistry module easily Under this project, we are studying techniques and meth- parallelizable with OpenMP. When integrated on STOMP, it odologies to support scientific applications on large-scale provides a 40% speedup (executed with the same amount heterogeneous system. Our objective is to develop of processing elements). Numerical accuracy is virtually approaches, techniques, and methodologies (in form unchanged. We are developing a CUDA version of the of software libraries or prototypes) to facilitate scientific chemistry module to run on NVIDIA GPUs. We are also application mapping on large-scale heterogeneous designing a hybrid version of it that will perform chemistry machines. Challenges associated with the use of these reactions where resources are available trying to maximize systems involve the refactoring of commodity applications utilizations. Early tests demonstrate up to 12 times of to support new hardware features. New hardware features potential performance improvement. are dictated by evolution in architecture design necessary Transformation development. We are developing com- to achieve higher performance (shorter simulation times piler- and auto tuning-based code transformations tech- or larger simulations) or reduce power consumption. This niques for GPU. Modern GPU perform less efficiently if project is an important cornerstone in understanding ben- blocks of 32 CUDA threads (warp) follow a different control efits and applicability of new computer architecture for flow in the code execution. Also, if memory access patterns applications of strategic importance such as molecular sci- for the warp are non-consecutive within 128 bytes, perfor- ence, classical and quantum chemistry. The methodologies mance is significantly reduced. This is usually referred as and techniques that are developed by this project will non-coalescing memory accesses and thread divergence. allow the next generation of molecular science, classic, Following this consideration, we are focusing on loop trans- and quantum chemistry applications to run efficiently on formations and iteration reordering to reduce memory next generation large-scale heterogeneous systems. We are access and threads execution irregularities. This is done studying for instance code generators to allow for the easy by a combination of profiling, input characterization, and porting of applications on new hardware (from high level sampling. Early experiments demonstrate a potential of description manageable by domain scientists), software lay- 35% of performance improvements when using loop trans- ers to support transparent maximization of performance, formation techniques. We envision a runtime layer that and a reduction of power consumption. automatically recognizes input sets, which will enable the A few highlights from project progress and accomplish- method to perform reordering of iterations based on ments achieved during FY 2012 are described below. machine-learning approaches. Domain-specific code. We developed and implemented a Finally, we are targeting the execution of the ParaFLOW domain specific code generator and auto-tuners for tensor application on GPUs by using a set of custom codes devel- contractions in NWChem to target a variety of architec- oped on CUDA as well as a library-based approach by using tures. The code generator is able to automatically generate Stencil++. code for GPUs with CUDA, OpenCL (AMD, NVIDIA), multi- For FY 2013, we anticipate the completion of our CUDA cores with OpenCL, OpenMP, standard C/Fortran, and chemistry modules and will continue working on runtime DGEMM + Sort. The code generator supports many options code transformation and iteration reordering for GPU. We (such as the use of fused indices, streaming to overlap with are also planning to use the tensor contraction engine for memory transfers, use CPU to support GPU, etc.). Different GPU on multi-reference coupled cluster problem. versions and devices have different performance tradeoffs PN12074/2475

217 Mathematics and Computing Sciences 218 PN12075/2476 is overdecomposed intotasks, enabling automatic rebalanc- Thework performed based loadbalancingandwork stealing. distinct approaches toaddress thischallenge:persistence- two iterations. We of thedesign and evaluation investigated incremental rebalancing toimprove load balanceacross calculations. Suchapplications evolving requireor slowly identical Applications execution ofteninvolveiterative of reacting toapplication andhardware actions. running concurrent and withthe application, observing computations. Suchcomputations actassub-programs concurrent observer of sages, enabling thedevelopment work flexibleprogress mes- supports guarantees for active accessprotocols. Themulti-threaded frame- remote memory transactions that couldinvolvemultipleround-trips using complex execution of enables efficient message framework execution onfuture tive extreme scalesystems.The active future systems, thread-safety prepares applications- for effec theanticipated increasesystem. Given incore countson dataThread-safety inamany-core enables natural sharingof task-basedapplications. of for thedesignanddevelopment We athread-safe message-basedframework active developed computing platforms withreasonable effort. tofuturesustainable porting super- software for effective algorithms andoptimizations. Theresult willbe choice of localizedchangestotheapplicationselective through careful applications for adaptation architecture tounderlying with nity, performance portable of enabling thedevelopment theTASCELfor usebythewiderscientific commu- of library methodsandbereleased aspart computational chemistry anticipated regularity. Algorithms willalsooptimizekey putation granularity, communication requirements, and handle varioususecasesfor data accesscharacteristics,com- ous metrics(power, algorithms will performance). Developed thecomputation andcommunication tooptimizevari- ing of extreme scaleplatforms. Theruntime managestheschedul- scalable algorithms suitablewith focus for onthedesignof runtime priorapproaches that incorporates keyelementsof thismodelhave beenstudiedpreviously,of wewilldesigna specificinstances systems.Whileseveral distributed memory task-basedalgorithmsThis project for focuses ondeveloping systems makes itchallengingtomanagethemeffectively. Increased scaleandvariability inextreme scalecomputing Sriram Krishnamoorthy Based Optimizations TASCEL: AnExecutionModel forTask- and communicationtooptimizevariousmetrics. automatically theschedulingofcomputation programmers witharuntimethatmanages model thatexposesahigherlevelinterfaceto We aredevelopingatask-basedexecution Mathematics andComputing Sciences

iterative CoupledCluster Singles andDoublesmethod. iterative thecomplete of version also implementapreliminary inter-task dependences. We will managementof efficient and NUMA.Theinfrastructure willbeextendedtosupport data data-placementsupport locality awareness intermsof In FY2013,wewillextendthe TASCEL infrastructure to Graphs. Massive and Software of for Analysis for publication inthe2012Workshop onParallelAlgorithms approach detectionhasbeenaccepted toscalable homology this reported intheliterature. Aninitialdescriptionof sequence alignmentscomputedpersecond,thehighest strong scalingandupto2.42×10 theHopperCray XE6supercomputer demonstrate cores of alignment computation taskseffectively. Resultson120,000 work stealingtoparallelizeoptimalpairwise uted memory supercomputing platforms. Ourapproach employs distrib- parallel detectiononmassively large-scale optimalhomology applications. toconduct We ascalable framework developed withnumerousproblem inlarge-scalesequenceanalysis in TASCEL detection,afundamental toaddress homology We employed thedynamicloadbalancinginfrastructure High Performance DistributedComputing. work ispublishedinthe2012International Symposiumon (163,840 cores), andupto128,000cores onOLCF Titan.This full NERSCHopper(146,400cores) andALCF Intrepid systems onthe demonstrated lowoverheadsandhighefficiencies machines.We applicationstive ondistributedmemory based retentivework stealingalgorithm optimizedfor itera- Weincremental rebalancing. alsopresent anactive-message- based rebalancing algorithm that performs localized ing bytheruntime. We ahierarchical persistence- developed Comparison of applications that improveComparison of loadbalance. 7 optimal pairwise optimalpairwise

Visualizing Uncertainty in Conceptual and Numerical Models for Geological Sequestration Luke J. Gosink viable strategies to address these uncertainties, and acquir- This research will help scientists characterize ing data for testing and validating our strategies. Initially, and quantify uncertainty in the conceptual we identified the varying sources of uncertainty present in each of the existing projects supporting GS3. We collabo- and numerical ensembles that are commonly rated and consulted with other projects and identified both used to assess the viability and long-term elements of uncertainty and potential strategies for detect- impacts of sequestering carbon dioxide at ing and characterizing these uncertainties. Ultimately, we select geological regions. identified uncertainty to address in two stages of GS3’s modeling framework: its conceptual and numerical Despite being actively researched for several decades, the model development stages. field of uncertainty quantification (UQ) is still rapidly evolv- In FY 2011, we completed the construction of our statistical ing and developing. Existing UQ research attempts to quan- analysis framework for performing UQ. This framework is tify the uncertainty in a given system based on the system’s based on a Bayesian approach called Bayesian Model reliance on instrumentation (used to obtain ground truth), Aggregation (BMA) that combines ground truth observa- conceptual modeling structures (the mathematical and tions (i.e., monitoring data) with the predictions obtained domain-specific models used to create a system that can through an ensemble of conceptual and numerical models. predict and simulate a real-life event), and numerical The framework facilitates UQ analysis for both the ensem- approximations (the methods used to approximate the ble and its constituents by providing a mechanism for operation of the system). A fundamental challenge to identifying and quantifying the spatial, parametric, and assessing uncertainty and accuracy in sequestration models temporal regions where uncertainty and model errors are arises from the fact that ground truth data – essential for greatest. We extended this statistical framework with a pre- most UQ and risk analysis tasks – is largely unavailable; in liminary suit of scientific visualization software tools. These many cases reliable, sufficient subsurface ground-truth is tools, which are working prototypes for this project’s even- too costly or infeasible to obtain. tual application deliverable, help to convey the informa- Our objective is to address the above challenge by develop- tion generated from our statistical framework in a more ing novel methods for characterizing uncertainty that can meaningful way by visually conveying (through isocontours be used in the presence of minimal ground truth measure- and surfaces, scatterplots, and time series displays) where ments. We combine the analytical strengths of statistics model uncertainties are greatest. with the interactive power of scientific visualization and In FY 2012, we completed validation studies on our statisti- visual analytic methods. This approach addresses uncer- cal framework. Our approach and the validation results tainty through the comparative analysis of varying scenar- were presented at several conferences. We also completed ios, considering all likely events predicted by numerous the development of our software application called systems (i.e., varying conceptual and numerical models) for Dynamic Real-time Ensemble Analysis and Monitoring a fixed sequestration site. Our integrated approach is novel (DREAM). This software allows modelers to launch and in the geological sequestration, visualization, and uncer- monitor an ensemble of numerical simulations. Addition- tainty quantification communities. As an integrated com- ally, while simulations are running, DREAM enables model- ponent of the Geological Sequestration Software Suite (GS3), ers to visualize and analyze the simulations in-flight. The this project supports risk analysis, management, and miti- integration of our statistical framework with DREAM effec- gation tasks for sequestration activities through a unique tively forms an uncertainty quantification pipeline. This combination of techniques derived from the fields of statis- tool is a core component in the GS3 environment and will tical analysis, scientific visualization, and visual analytics. therefore be widely accessible for all GS3 clients. Ultimately, this work will enable scientists to advance the During FY 2013, we will add new functionality to DREAM to understanding of sequestration modeling and more accu- extend its visualization capabilities and its analysis capabil- rately assess the performance of and long-term impacts to ities. Additionally, we will submit several journal articles on sequestration activities at potential reservoirs. both the analysis methods we have designed as well as the Our work during FY 2010 focused on three stages: identify- application we have built. ing aspects of uncertainty to address within GS3, identifying PN10079/2321

219 Mathematics and Computing Sciences Nuclear Science and Engineering A Virtual Testing Toolbox for Predicting the Properties and Behavior of Multiphase Materials in Disposal Environments Joseph V. Ryan

interfaces between phases, and diffusivities of assorted This project is developing a virtual testing molecules through various phases. toolbox with capabilities for identifying and Time, as “critical time” benchmarks progress toward a determining values for critical modeling totally dissolved end state, such as time when concentra- parameters needed for the long-term prediction of a radionuclide in the near field at a given distance tion of waste system performance that meets from the waste form surface reaches 50% of its asymptotic regulatory criteria. Novel experimental tech- value. niques will also be developed to advance the The objective of this work is to develop a Virtual Waste science of multiphase interactions. form Testing Toolbox (VWTT) to provide the technology needed to reduce the number of physical/laboratory tests required to produce constrained parameters quickly and The safe immobilization of nuclear waste is of critical efficiently assess product quality and long-term degrada- importance to all U.S. citizens, whether it is a legacy of tion mechanisms of multiphase waste forms and other weapon development or the result of peaceful power gen- multiphase materials in disposal environments. A general- eration. This project aims to produce a process whereby ized model evaluation tool is needed that allows proposed critical properties of the most complex (multiphase) waste models to be systematically and quantitatively compared forms can be systematically identified, determined, and to experimental data and, just as importantly, to each incorporated into modular algorithms that fit seamlessly other. The development of such a tool is described below. into continuum transport models. Achieving these objectives will increase confidence in performance predictions The key to the tool is the development of methods for four and naturally result in a large reduction in the cost powerful computational packages to exchange data rapidly required for disposal, inspection, and maintenance of and reliably (see figure). Modern engineering computa- multiphase waste forms. tional fluid dynamic (CFD) codes are used to handle mass and thermal transport in highly complex structures with Every waste form system is inherently multiphase. Even nominally homogeneous materials such as waste glasses become multiphase problems when incorporated in a system (canister, fill material) or when corroded by water (amorphous gel layer, crystalline corrosion products). Many existing prediction algorithms, however, assume a relatively physically homogeneous source term. The difficulty of modeling complex phenomena such as the evolution and degradation of a waste form system originates in the underconstrained nature of the interactions. The particular system of interest contains several classes of parameters: Geometries, including pores, surfaces, interfaces between materials/phases, and spatial arrangement of and relationship between various phases. Materials and phases such as glass evolve into a system composed of bulk, surface alteration layers, near-field soil environment, and associated interfaces, where most (if not all) chemistries and compounds change from the original state. Similar processes occur for nearly every waste form.

Kinetics, where reaction rate constants describing the tem- Schematic of the VTTB showing the relationship between different poral evolution of a material into a new phase, creation of computational modules. PN11005/2327

221 Nuclear Science and Engineering 222 PN11005/2327 of five differential equations fivedifferential that assume: of NuclearMaterials.TheGRAALmodelconsists of Journal paper before that isinfinal submissiontothe revision GRAAL modelusingtheVTTB, leadingtoajointPNNL/CEA mental data. To the thisend,wesetupandevaluated appropriateother workers against glassdissolutionexperi- thisproject modelsof istobeable toevaluate of objective Akey theexperimentalobservations. accounts for mostof modeltodatelab that CEAisperhaps themostdeveloped bytheFrench atomic energy The GRAALmodeldeveloped MatLab that alsohandleddata reporting andplotting. using Phreeqc wasinterfaced model.Theentire package experimental data byrepeatedly running theCOMSOL/ from artificial analysis) parameter values(withsensitivity wasusedtopredictparameter optimization package corrosion model.ThePESTGauss-Marquardt-Levenberg asimpleglass toevaluate available iPhreeqc library newly theCOMSOLmultiphysicsCFDcodewith integrated data, Usingartificial we computational packages. essary We recently demonstrated thenec- theability tointegrate complexshapes. meshing of SolidWorks) (e.g., programs allowsthedesignanddynamic as acontinuum structure. tomoderndrafting Aninterface whilestilllooking eachprogram at thesystem aspects of utilizingthebest of theadvantage calculations gives whileusingCFDcodesfor with theseprograms transport tion kinetics. Thus,handlingreaction kinetics calculations minerals alongwithreasonable algorithms for solvingreac - thermodynamicdata solutions,and forextensive gasses, suchasPhreeqcbases andprograms andEQ3/6provide cal kinetics with great facility, however. Geochemicaldata - donothandlecomplexchemi- Modern CFDcodesgenerally necessary.where designvalidation iscriticalandlegally inchemicalandmechanicalengineering, used extensively scales.Theseare robust, well-validated packages variety of irregular natural orsymmetricalman-madeshapes at a • • • • Nuclear ScienceandEngineering • silica asH to supersaturation. a silicamineralphaseforms at arate proportional gets thicker. layerthe protective slowerasit thickness grows to thesilicasupersaturation. glasserosion/dissolution isproportional the rate of removed bysolution flow. boron insolutioncomesfrom dissolutionbutis is removed bysolution flow andmineral formation. 3 SiO 4 insolutioncomesfrom dissolutionbut appreciable changeineffectiveness. wasteloadingswithno can immobilizeawiderangeof Thiswork showedthat thecaststonematerialloading. wastesolution simulated secondary asafunctionof ters ization wasperformed toobtainvaluesfor theseparame- and microstructure. Coordinated experimental character the system,includingpore geometry, phaseassemblage, parameterization deficiencieswere identified quickly for waste form problematic tocalculate accurately. Several the cements,making long-termperformance of tion of - materials evolu complicates wellunderstood therelatively wasteintocementitious secondary rial. Theaddition of andamorphous mate- crystalline mineralphasesof mix of aheterogeneous complex,consistingof and structurally caststone, saltstone)are These materials chemically (e.g., forms waswrittenupandsubmittedtoWaste Management. pastoncementitiousmultiphasewaste years a coupleof In addition tothemodelingtask,experimentalwork from modelingtechniques. existing modelswiththeinverse wasteglasscorrosion issuesandtoanalyze high-level for itfurther programmatic fundingtodevelop petitive com- ledtotheawarding of VTTB thepotentialof tion of workshops for glasscorrosion. Thesuccessfuldemonstra- The results were presented mid-year at twointernational dissolvedspecies. concentration of incorporate thisdata aswellthemore common provide invaluable can information andtheVTTB glassspecieswithinthealtered layers distribution of depthprofiling techniquesthatnew canmap thespatial models, thanwaspossibleinthisproject. We notethat new of includingdevelopment detailed investigation, experimentsrequiresnot reproduce muchmore asetof outwhyamodeldoesor with themodel.Ferreting theexperimental data andcomparison examination of theresults allowclose perspective, From thisproject’s representation, inobtaining reproducible ordifficulty data. inappropriate model,onethat ismissingessentialphysical dataparticular setexamined.Thisindicates eitheran The fittothedata waspoor at bestdependinguponthe account chemicalprocesses. byCEAthat directlytakemodels are into beingdeveloped themodel,andnew of major drawback tothisversion have tobeused inthisinstance. Thisisanacknowledged cal reactions andassuchtheiPhreeqc moduledoesnot includeanychemi- Note that the modeldoesnotexplicitly - Advanced Condition Monitoring of Microstruc- ture Evolution and Property Degradation in Current and Advanced Nuclear Reactor Alloys Shenyang Hu

magnetic hysteresis loops and magnetic domain evolution We are developing mesoscale phase-field in a BCC single crystal iron with a nonmagnetic particle. models for studying the effect of microstruc- In these models, a simulation cell of 256 nm × 512 nm × tures on material response to applied fields, 1nm with periodic boundary conditions is used. The simula- understanding signal physics in nondestruc- tion shows the following: 1) 360 degree domain walls are tive examination (NDE), and providing guid- stable in a single crystal under applied magnetic field in easy magnetic axis; 2) both 360 degree domain walls and ance for developing real-time nondestructive nonmagnetic particles are nucleation sites of a new anti- detecting/monitoring methods. direction domain during domain switching; 3) a new anti-direction domain nucleates by splitting a 360 degree Important scientific and technological breakthroughs are domain wall much easier than from a nonmagnetic parti- needed in characterizing, detecting, and understanding cle; 4) nonmagnetic particles can destroy the 360 degree the early phase of environmental degradation in materials domain wall, which causes magnetic hardening; and used in nuclear reactors. Environmental material degrada- 5) nonmagnetic particle can pin and bend a domain tion is one of the main concerns impacting the perfor- wall when the wall passes through it and thus inhibit mance, reliability, and safety of structural materials in these domain switching. units. In FeCr alloys and zirconium, the preferred candi- The main achievement in FY 2012 is that we developed dates for structural materials in fusion and advanced fission and demonstrated new modeling capability to investigate reactor components, irradiation-enhanced phase separation the interaction mechanisms between nonmagnetic particles (the formation of Cr rich precipitate, Carbides, and hydride and magnetic domain walls and explore the signal physics precipitates) causes materials embrittlement, the reduction in nondestructive magnetic techniques used for monitoring of corrosion resistance, and creep failure. Therefore, pre- microstructure evolution in irradiated materials. dicting the kinetics of microstructure evolution, understanding the degradation mechanisms, and developing Our overall goal is to establish PNNL as the leader in a real-time monitoring capability of property degradation material degradation detection of nuclear materials. are key milestones to managing existing nuclear power plants. The objective of this project is to develop advanced modeling tools for investigating the impact of initial microstructure and environment conditions on microstructure evolution kinetics; the effect of microstructure on material response to applied fields such as ultrasonic and magnetic fields; signal physics in nondestructive examination; and critical degradation mechanisms and key precursor states. During FY 2012, we developed a phase-field model to study the magnetic domain wall stability and interaction between magnetic domain walls and nonmagnetic particles. FeCr alloy is taken as the model system. The nonmagnetic particles/precipitates could be Cr rich precipitates, carbides and voids which are formed during irradiation-induced phase separation in FeCr alloys. The magnetization vector M(r,t) is used to describe the spatial and time distribution of magnetic domains. The magnetic domain evolution is simulated by solving the Landau–Lifshitz–Gilbert (LLG) Initial magnetic domain structure in (A), including 90 degree domain walls and a nonmagnetic particle, is created. The color represents the magnitude of the Mz equation. The figure summarizes the predictive component, and arrows represent in-plane projects of the magnetization vectors. PN12004/2405 223 Nuclear Science and Engineering 224 PN10004/2246 fuel andfresh mixed-oxide fuel.Thefoundational assay uraniumandplutoniumisotopes inspentnuclear ment of direct measure - This project expandsthesciencebaseof materials diversion. of the prevention and significant steptoward both real-time accountancy wouldbea plutoniummasswithhighaccuracy ment of (no operator-declared information), anddirect measure- minutes), independent that (10sof couldprovide timely mass intheassembly. assay A nondestructive technology coupled tocomputational predictions toinferplutonium signatures from non-plutoniumisotopes,whichare then estimating plutoniummassare basedonthedetectionof areprocessing plant).Current methods forthe headendof for verification (e.g., at interimorpermanentstorage at reactor fuelandfresh mixed-oxide increase fuelwillonly improved fuelaccountingtechniquesfor spentlight-water propagate around theglobe,fuel cycles theneedfor unmet challengeininternational Asnuclear safeguards. content inspentfuelassembliesisawell-documentedbut theplutonium Accurate, measurements non-destructive of Glen A.Warren for Safeguards Advanced NondestructiveAssay was vetoed for the first 20µs. was vetoedfor thefirst gold thegold andsilverfoils. foil For measurement, thescintillator Time spectrafrom (n,γ)measurements withnofoil (empty)and tional safeguards. to improvematerialaccountancyforinterna- spectroscopy (LSDS)willbestudiedasameans fresh mixed-oxidefuel.Leadslowing-down plutonium isotopesinspentnuclearfueland base ofdirectmeasurementuraniumand This projectisaimedatexpandingthescience Nuclear ScienceandEngineering isolated resonances at specific neutron energies.Such measurements usingselectedfoils that have strong, correlations can bemeasuredenergy byconducting(n,γ) theslowingtime. Thetime- the neutrons asafunctionof for thefluxof neutrons. Second,one needstounderstand correlation time-energy theobserved needs tounderstand There one are First, twoaspectstocharacterizingtheLSDS. to conductinitialtestswiththeLSDS. measurements, butitprovided anadequate neutron beam theneutron beamisless-than-optimal for theLSDSture of neutrons. Theavailableused asthesource timestruc of - existing ThermoScientificP385D-Tneutron generator was comingintocontactwiththecadmium.The possibility of thecadmiumtoreduce the aluminum isplacedoutsideof cadmiumisplacedaround thecube.thin layerA of thewalls,andreturning tothecube),a cube, bouncingoff minimize room thelead return (neutrons escaping outof Interrogation Laboratory. theRDNSActive of as part To theLSDS FY2011 wascompletedinearly The installation of neutron response. fast no obvious measurements usinganeutron sourcepreliminary showed demonstrated aresponserays, whenexposedtogamma thesystem.Whilesamples resolution of mizing theenergy highpurity,occurring andof for whichisimportant maxi- for costandmeasurement efficiency. Theleadisnaturally theleadcubewasoptimized instrumentation. Thesizeof neutron generator, ameasurement sample, anddetector lead1.5monasidewith ment were alargecubeof theinstru- Theprinciplecomponentsof theLSDS. tion of focus for FY2010wasthedesignandconstruc- Our primary spentfuel. assay non-destructive of to develop and totestinstrumentation anddata acquisitionmethods This systemwillbeusedtostudycriticalphysicsissues shielding. neutron self andgamma-ray complexities of methods that address andtestanalysis the and 4)develop spectroscopy; 3)perform benchmarking measurements; requiredcialized fissionchambers byleadslowing-down capabilities todesign,fabricate, andcharacterizethespe- andusethe down spectroscopy instrument; 2)develop tasks: 1)design,build,andcharacterizealeadslowing- that empirical validation, wewillperform thefollowing retical studycoupledtoempiricalvalidation. To conduct retical andsimulation studiesbutrequire additional theo- techniques have shownpromise inproof-of-principle theo- willalsobeexamined.Both and combinations thereof accountancy. delayed spectroscopy gamma High-energy will studythat application asameans toimprove material method willbeleadslowing-downspectroscopy, andwe measurements were completed using gold and silver foils. MCNP. A large variety of issues were studied including the The time-spectra are shown in the figure. The structure in impact of room return, optimization of shielding, variation the empty target run around 50 µs is due to resonances in of flux with location in vertical hole, predictions for the the germanium of the BGO detector. The peaks above 60 µs (n, γ) measurements, and precision modeling of the in the gold and silver measurements are due to resonances neutron flux. in those materials. The neutron flux in the LSDS was mea- Though the project was cancelled in early FY 2012, the sured by activating two groups of small foils, one covered most significant accomplishment was the acquisition of a in cadmium to reduce the influence of thermal neutrons deuterium-tritium neutron generator with the appropriate and one without cadmium. Gamma spectroscopy was time structure to conduct LSDS measurements. Project performed on those samples to determine the level of leadership has been exploring options to fund completion activation. Analysis to determine the flux is in progress. of spectrometer characterization as well as assessing the Significant effort went into understanding the performance use of project capability in support of other client-funded of the LSDS using the radiation transport modeling code efforts. PN10004/2246

225 Nuclear Science and Engineering 226 PN12009/2410 ray activity, Duetoitslengthylifeandgeneration269 years. bycosmic argon withahalf-lifeof isotopeof radioactive longest-lived ultra-sensitive ultra-sensitive counters andliquidargon scintillationcounters detectors. gas-proportional gas in usedasafillArgon iscommonly Paul H.Humble Argon-39 Measurement The ultra-sensitive The ultra-sensitive environmental aswellnational securityapplications. water. Thiscapability willbenefitargon measurements for in argon samplesincludingdissolvedgasses from smallgas separation andpurification process that enables extracting argon ahighrecovery months,wewilldevelop several processed andpurified for thispurpose. During thenext tomeasurewith Fermilab ancientargon have that they to construct dark andweare mater collaborating detectors, (Fermilab) National Acceleratoroped byFermi Laboratory wells.Thisresource- to beselectnatural gas isbeingdevel argon at lowactivity thistimeappears promising source of Themost andargon trappedfers reservoirs. innatural gas indeepancient aqui- ancient argon includedissolvedgas Possible sources counters. of use ingas-proportional bility with enough sensitivity tomeasurebility withenoughsensitivity on other ultra-sensitive initiative projects to develop low projects initiative todevelop on otherultra-sensitive andwork beingdone laboratory the shallowunderground sensitive sensitive this an projectultra- are todevelop of The major objectives inthetimeframebetween50and1000 years. gas foropportunity agedating water ortrapped atmospheric late ancient( areliable source of ancient argon depletedin Withthismethod,itisdesirable tohavetors. asource of orinscintillation detec- gas inproportionalfill counters nuclear measurementsity inultra-sensitive whenusedasa security applications. age dating,fundamentalscience,andnational capability thathasrelevancetoenvironmental an ultra-sensitiveargonnuclearmeasurement This projectisdevelopingandimplementing Nuclear ScienceandEngineering 39 Ar nuclear measurement capability andformu- 39 - activ background Ar canbeamajorsource of 39 Ar measurement wouldalsoprovide the 39 Ar measurement capability leverages 39 Ar andhave adetectioncapa- 39 Ar depleted)argon for 39 Ar activity. An 39 Ar isthe

measure the andfinally package, ananalysis devise ating parameters, oper develop proportional counters, set upthenecessary mid-year,began andthescopefor timewas to thisshort proportional counterwork. gas Thrproject ultra-sensitive of anargon source withlow of Ar nuclear measurement capability andtheidentification The expectedresults from thisproject are ademonstrated nuclear explosion monitoring. Ar isotopes,including other measurementsextended toenable ultra-sensitive of argon with a recovery greater than80%. argon witharecovery surements aswell asamethodtoseparate andpurify ourargon mea- of publications describingthesensitivity environmental samples.We expecttohave oneormore argon from or smallpurposes suchasdissolvedgasses significantbenefit forseparating nique that wouldbeof argon separation and purification tech- on ahighrecovery argon calibration sources.develop We work willalsostart toward making measurements, otherargon activity and sample. Weobtained from gas theFermilab willwork FY2013,we willpublishthe results of quarter In thefirst be confirmed. ancientargon, andthiswill should containlowactivity separatedgas andpurified gas well.This form anatural samplefrom that Fermilab a gas have containsargon they room air.extracted from laboratory We have also received mademeasurements Andhave successfully onargonters. Armeasurements proportional coun- ultra-sensitive ingas Also duringFY2012,wesetupthecapability tomake theyear.before theendof willbeaccomplished sample,lab andtheseobjectives gas laborators at Fermilab.- We theFermi are inpossessionof measurements of measurements of toprovide ultra-sensitive are beingdeveloped packages this project specifically, andanalysis operating parameters For materialsbackground andproportional counters. 39 Ar activity of a gas sampleprovided bycol- agas of Ar activity 39 Ar activity. Thiscapability isbeing 37 Ar, in whichhasimportance 39 Ar activity that canbeusedin Ar activity - Background Reduction for Nonmetallic Nuclear Detection Components Raymond Addleman background levels intrinsically low enough that they The world’s most sensitive radiation detection can be used directly. In most cases, however, sufficient and assay systems depend on ultra-low back- radiopurity levels will not be available, and the location (surface, bulk) and likely origins (monomer feedstock, ground (ULB) methods and materials to achieve polymer synthesis, material processing, handling) will their remarkable sensitivities. ULB systems have have to be determined before methods to clean the application to basic science efforts such as neu- material of interest can be formulated. Once the loca- trino, dark matter studies, and practical function and source of contamination is known, methods to tions in nuclear forensics and environmental avoid or remove contamination and achieve ULB levels science. can be developed rationally. • Development of methods to remove contamination. It is likely that contamination will be unavoidable. Ultra-low level radiation detection systems need better In these cases, purification methods will have to be supporting materials (structural, mounting, electrical) that developed for materials of interest. provide lower radiological backgrounds. We explored new methods to provide ultraclean nonmetallic components In FY 2012, we made progress on all key technical chal- that will enable ultralow-level measurement systems. lenges identified above. The majority of activities focused Radio-impurities currently used to construct ULB systems on development of accurate ULB assay methods, which is a both on the surface and in bulk material continue to limit necessary capability to support all other efforts. We found the lowest backgrounds achievable. Significant effort has traditional nuclear counting methods such as gamma ray gone into finding and processing ULB metals such as Pb, spectroscopy required prohibitively long count times for Cd, Cu, and steel that act as shielding and supporting mate- the ULB assay, even when underground detector arrays are rials as well as critical detector components. Much less available. Digestion of nonmetallic sample removes the effort has gone into finding or generating nonmetallic matric, concentrates contaminates, and enables traditional ULB materials that provide essential electrical and thermal ultra-trace chemical methods such as inductively coupled insulation for nuclear detection systems and can provide plasma mass spectrometry (ICPMS) that is capable of rapid physical support structures. Nonmetallic materials could be analysis and high sample volume throughput. Sample advantageously used if ULB levels could be achieved. In preparation and handling methods are critical for obtain- this important and promising area, improvements in radio- ing meaningful results. Productive ULB chemical process purity of three orders of magnitude (from mBq/kg to μBq/ methods have been developed and are successfully demon- kg) are needed and if achieved would be widely utilized in strated on a number of polymeric materials of interest, the ULB systems. The main contaminates of concern are with ICPMS detection limits below 10 ppq (for U and Th). isotopes K, U, Rn, and Th and their respective progeny. Ide- Additionally, the chemical preparation processes have ally, nonmetallic materials of interest could be found or applications to other analytical methods such as alpha provided at 10 μBq/kg. spectroscopy, an excellent technique for radon (and The focus of this effort and key technical challenges for daughter products) contamination detection. finding and providing nonmetallic ULB materials of suffi- Contamination sources and the means to remove and avoid cient radiopurity are as follows: the impurities were also explored. Adapting PNNL-developed • Accurate assay of contamination levels. The point of methods for ULB copper, we explored leaching U and Th, this effort is to identify and provide ULB materials to primary contaminates of concern, from polymers. Using the scientific community, which in turn requires accu- ultrapure nitric and hydrochloric acid, we found that we rately measuring inorganic contaminates in diverse could remove surface contamination (i.e., from mineral organic matrices at ULB levels, a nontrivial challenge dust) after 2–3 days of rinsing without damaging the poly- that must be addressed. Effective assays will enable mer. This cleaning method has application for the reduc- materials of interest to be screened effectively for con- tion of background in polymers ranging for powdered tamination and support development of purification feedstock to final cleanup of machined parts. methods (should clean sources of desired materials not To disseminate the results and information gathered to the be found). Various analytical methods and approaches greater scientific community, several publications – one need to be evaluated comparatively. focused on analytical methods and another on polymer cleaning techniques – will result from FY 2012 activities. • Identifying source(s) of contamination and means to avoid them. Some materials may be identified with PN12011/2412 227 Nuclear Science and Engineering 228 PN12012/2413 rently assemblingthedetector, whichweanticipate tobe We have components andare acquired cur BazookaSPECT FY2013. early list andanticipate arrival are provided separately, andweare currently onawaiting inFY2013.Thedetectorread-out2012 orearly electronics FY kit hasbeenordered at theendof andshouldarrive toPNNL.ACERNGEMdetector andGEMdetectors kaSPECT bringingBazoo- During FY2012,weinitiated theprocess of technologies. applications that are enabled/or improved uponbythe response toionizingradiation. We willalsoseektodevelop demonstrate, characterize, andimprove theirdetection these twodetectorcapabilities at PNNLandinvestigate, thisproject istoestablish applications. of Ourobjective for thermalneutronreplacement detection technology interested GEMsasapotentialHelium-3 ininvestigating be fabricated inlargeareas at lowcost.We are particularly spatial resolution, andhighcount-rate capabilities andcan GEMsprovidethat excellent timing, weare investigating. (GEMs)are anotherpromisingpliers detectortechnology efit from itshigh-resolution capability. Gaselectron multi- imaging detector, inapplications particularly that willben- istoresearch itscapability thetechnology asaneutron of for imagingapplications, gamma-ray anatural extension isbecominganext-generation detector As BazookaSPECT in national securityandnuclear scienceapplications. electrons) fission fragments,betas,andconversion for use neutrons, alphaparticles, ionizingradiation (e.g., forms of ing andcharacterizingitsusefor detectingadditional imaging applications, andweare nowinterested inexplor beenemployedtor hasprimarily inpre-clinical medical Thisdetec- for high-resolution sensors imaging. (CCD/CMOS) metaloxide semiconductor coupled device/complementary andcharge imageintensifiers, bines columnarscintillators, detector, BazookaSPECT, that com- hasbeendeveloped gamma-ray classof anew years, During thelastseveral Brian W. Miller BazookaSPECT Neutron Imager security applications. medical imaging,nuclearscience,andnational radiation detectortechnologiesforusein We seektodevelophigh-resolutionionizing Nuclear ScienceandEngineering

- - thermal neutrons. theCERNkit that canbeusedforcation detecting of GEMdetectorormodifi- ogy, anew wewillbeingdesignof thedetector.of withGEMtechnol- Aswebecomefamiliar resolution experiments todeterminethespatial andenergy x-ray imaging This willincludesomeinitiallow-energy sition software provided bytheCERNRD51collaboration. withthedetectorandGEMacqui- ourselves will familiarize Once theGEMdetectorandread-out electronics we arrive, small-animal planarscintigraphy. used toidentifyandseparate isotopes,spectrometry, and that particles/photons canbe different cidence detectionof potential applications that includecoin- wewillinvestigate materials willlikelybetestedduringthisprocess. Some andspatial resolution. Various energy scintillation terms of tion, andwewillmeasure thedetector’s performance in ionizingradia- todetecttheseadditional formskaSPECT of Bazoo- In FY2013,wewilldemonstrate thecapability of electrons. betas,andconversion fragments, alphaparticles, mentioned:fission ionizingradiation previously forms of role, BazookaSPECT, particularly toincludeadditional ing applications. Instead,weare currently expandingtheir for useinneutronand GEMdetectors detectionandimag- We proposed boththeBazookaSPECT investigating initially have thiscapability. ference. Thedetectorcurrently beingassembledwillalso the prototype detectorat theSPIEOptics+Photonicscon- FY 2012,wewere invitedtopresent theconcept/designof perform full-body, rodents. Late in of dynamicscintigraphy (>10cmindiameter),whichcanbeusedto area detectors large- of that hasledtothedevelopment BazookaSPECT Recently, amajorbreakthrough occurred inthedesignof ing capability. whoarewith groups interested inthesmall-animalimag- tions, andweare establishing collaborations withinPNNL for useinpre-clinical medicalimagingapplicadeveloped - detectorwasoriginally rate by2times. TheBazookaSPECT detection USB 3.0CMOScameras,whichincrease theevent released detector acquisitionsoftware thenewly tosupport FY2013. We functional inearly fully have modifiedthe Characterization of a 14 MeV Neutron Generator and Measurement of Fission Products Produced Erin C. Finn

The objective of this work is to make the generator capabil- The recent installation and characterization ity available to a diverse user group. Securing capability of a 14 MeV neutron generator at PNNL has requires that all appropriate procedures and permits are in place, along with an understanding of system operation enabled nonproliferation research and develop- and what results to expect from various operating schemes. ment (R&D) by supporting the production of To accomplish this, Monte Carlo N-Particle (MCNP) model- unique radionuclides for use with several ing is used to predict the results, which are then experi- nuclear advancements, particularly the mea- mentally tested and validated. surement and evaluation of possible new Effective experiment planning requires a baseline series of short-lived isotope signatures. quantitative experimental measurements of the neutron environment, flux, and spectrum available. Such measure- Historically, PNNL’s efforts in nuclear nonproliferation R&D ments are made using standard dosimetry techniques, have been thwarted by a lack of readily available, correct wherein high purity wires or foils with known neutron radioactive standards and realistic radioactive samples that activation cross-sections are irradiated under known are needed to accurately test new measurement and data conditions. The activation of the materials is measured evaluation methods and improve nuclear data. The 14-MeV by gamma spectrometry and used to calculate the number neutron generator was purchased and installed in the 318 of neutrons incident on the sample. Use of multiple mate- Building low-scatter facility using general research equip- rials concurrently allows measurement of the neutron ment funds. The facility housing the generator is unique spectrum in addition to the neutron fluence. and enables us to achieve a very pure 14 MeV neutron All applicable permits and procedures are in place, and the spectrum with minimal scatter. Before the new generator neutron generator was approved for use early in FY 2012. can be used to produce samples required for high accuracy MCNP modeling has been completed to provide prelimi- nuclear measurements, the instrument must be set up nary estimates of neutron flux and spectra at various posi- and characterized to ensure a consistent high flux of pure tions relative to the generator. These models show that the 14 MeV neutrons. neutron flux and neutron spectra change measurably with change in position relative to the central point of the generator head. In addition, experimental measurements are underway to validate the effects predicted by the MCNP models. We completed a short series of measurements at the central point for different operating parameters and will measure at different positions at known sets of operating conditions. At the central point on the neutron generator face, total neutron flux measures 1.53 × 109 n/cm2/s (±5%) at the maximum operating power. The neutron flux is broken down into the following energy bins: 4.43 × 104 n/cm2/s (±24%) thermal (< 0.5 eV) 9.00 × 106 n/cm2/s (±15%) epithermal (0.5 eV to 0.1 MeV) 2.51 × 108 n/cm2/s (±10%) fast (0.1 to 10 MeV) Modeled (MCNP) and experimental (STAYSL) neutron spectra at the 2 D-T 14 MeV neutron generator face housed in the 318 Building low- 1.27 × 109 n/cm /s (±5%) fast (>10 MeV) scatter facility. Results show good agreement at 14 MeV; however, Mid-year 2012, the neutron generator successfully pro- experimental measurements indicate an increase of 10–50% in the duced short-lived radioisotopes for one of the projects. number of scattered neutrons over the MCNP model. Scattering is expected to become more pronounced at greater distances from the Operation of the neutron generator caused dose in adja- generator face. cent spaces in the facility, negatively impacting existing PN12014/2415

229 Nuclear Science and Engineering 230 PN12014/2415 using theinitialcharacterization data. previously. Otherprojects are alsoplanningexperiments thanhadbeenpossible scientificadvancements ect’s theproj - ing more completetestinganddemonstrations of have notbeen available tothisproject- inthepast,facilitat impact milestone. Thegenerator produced isotopesthat itshighest theprojects toachieve generator enabled oneof projects. However, theneutron initialcharacterization of impactsonother futureproblem and prevent negative acquired through othersources this mitigate tofurther andadditional fundingwas generator space shielding, projects inthose spaces.We received somefundingfor the Nuclear ScienceandEngineering expected toyieldasecondpublication. epithermal, andthermalneutrons. Thisresearch is fissionsproduced fromfast, 14MeVneutrons, fraction of product samples, whichwillenable ustodeterminethe experimentsonfission spectrometry gamma mance of Plans for later inFY2013includethedesignandperfor thedata onthisuniquecapability.lowed bypublication of techniquesinFY2013,fol- completed usingdosimetry The remaining characterization work isexpectedtobe - Exploiting Correlated Radiographic and Passive Signatures for Threat Detection in Cargo Erin C. Miller

radiograph and evaluating the impact of this approxima- This project develops methods to improve tion; second, investigating the possibility of low-dose radi- radiological threat detection in cargo containers ography with poor spatial resolution to provide this kind of shielding geometry estimate; and finally, developing by systematically combining information from detection metrics for quantitative comparison with existing passive radiological sensors and radiography. algorithms. These pieces will allow us to evaluate the Both techniques are currently deployed for feasibility of our approach and its sensitivity to the quality cargo screening and are widely acknowledged of the geometry estimate, lay the groundwork for future to provide complimentary information. injection studies of the method, and provide the basis for scientific publications. We are developing an inverse algorithm for describing a Work in FY 2012 focused on a model test case (a cargo con- possible radioactive source in cargo using radiographic and tainer with pallets, some of which contain a material that passive portal monitoring data. Our method developed attenuates and scatters radiation). Detector response for a may provide a quantitative and cost-effective way to har- portal as the cargo is driven through was calculated; radio- ness this information. Currently, cargo radiography is used graphs for the cargo container under high- and low-dose to detect contraband but not radiological threats. Previous conditions were calculated for pixel sizes ranging from work on combining radiography and passive sensors for 3 cm (similar to a currently deployed MobileVACIS) to cargo has largely focused on high-level data fusion, where 20 cm. Adjoint calculations describing the contribution both pieces of information are evaluated separately. This of a source at any point in the container to a particular work builds off a previous project that developed a source detector were determined for multiple scenarios: perfect localization algorithm for combining multiple passive information about geometry, geometry information sensors, given some estimate of attenuating or scattering obtained from a low-dose radiograph, and treating the materials. We are investigating the use of radiography for cargo as a homogenous mass. Perfect geometry informa- providing information about the problem geometry, with tion with attenuation only accounted for (scattering an emphasis on cargo screening. ignored) and with ray-trace only (scattering and attenuation ignored). Preliminary results indicate that the ability Research on this project covers three key areas: first, devel- to locate a source is improved when knowledge of geome- oping a method for making a 3D geometry estimate from a try is improved, but only when signal-to-noise is sufficiently high. Several preliminary detection metrics have been developed that will be used to produce receiver- operator curves for comparison with current detection algorithms. This progress will also enable us to judge whether the new algorithm provides improvements and what kind of geometric detail and accuracy of transport physics is needed. Work in FY 2013 will focus on a more complex test case, with an attenuating object as a source to include background radiation from cargo in addition to a potential threat. Additionally, quantitative comparisons with current methods will be produced. These results should allow us to evaluate the feasibility of our algorithm and the low-dose radiography when used for the detection of radiological threats. Method: Portal monitor responses and radiography are combined in an inverse algorithm to describe source location and position. PN12030/2431

231 Nuclear Science and Engineering 232 PN09042/2210 a length of 0.5meter.a lengthof CANDU fuelassemblieswith accountability challenge. LWRs. fuelelements presents an Thesheernumber of bles rather thantraditionalpelletshoused infuelrods like duetoreactortional safeguards design.PBMRusesfuelpeb- aproliferation threat arePBMR reactors lessof butposeaddi- duction inbothreactor types. shouldprovide ameanstoquantifyplutoniumpro- curves fluence. The changes versus resulting data andcalibration isotoperatio of ter (SIMS)toestablish calibration curves ionization massspectrome- usingasecondary PNNL analysis Test andAdvanced tor (MURR) Reactor(ATR), followed by Research Reac- CANDU)intheMissouriUniversity metals (for and PBMR) (for graphite istoconductirradiation of devices andcalibratingOur approach thesemonitoring todeveloping measurements more practical. for andfuelassembliesthat bothreactors shouldmake techniques samples.Thus,weare developing of numbers ments, thiscanbetime-consumingandimpracticalfor large these measurements occurringimpurityele- usingnaturally demonstrated ing plutonium.Whiletheteamhaspreviously anyotherisotopeproduced, includ- calculate theamountof materials have beenexposed,providing for theability to elements for determiningthefluencetowhichex-reactor impurity certain relative abundance of measuring ratios of anddemonstrated techniquesfor PNNL hasdeveloped Mark R.Mitchell Production Verification Bed ModularReactor(PBMR)Plutonium Deuterium Uranium(CANDU)andPebble Isotopic RatioFluenceMonitorsforCanadian declared reactoroperation. for objectiveverificationofadherenceto ses. Thesetechnologieswillprovideamethod safeguard purposeswithoutdirectfuelanaly- quantification ofplutoniumwithaccuracyfor analyze existingreactormaterials,enabling vessels, in-corestructures,andtechniquesto ing devicesthatcanbeattachedtoreactor We aredevelopingaconcepttocreatemonitor Nuclear ScienceandEngineering to lightwater (LWRs). reactors plicate item accountingcompared to bereloaded in operation com- CANDUreactor and theability of duty cycle, bundles, theirshort CANDUfuel small physicalsizeof challenges.The and safeguards presents unique nonproliferation CANDUreactors The designof

- tube has been exposed after several years of operation. of years tube hasbeen exposedafterseveral measure high fluencetowhichthepressure therelatively whichmay beagoodisotope ratios, methodto thelatter of tionally, theteammeasured titanium aswellzirconium thescraping samples.Addi- isotoperatioregarding studiesof discussions have occurred for collaboration withChalkRiver CANDU work ismovingforward asaDOE-fundedproject, and 2012. at before leastonejournalarticle theendof We to neededrepairs. andpublish willcompletetheanalysis due mostly tofinishanalysis SIMShasnotbeensufficient 4f ated carbonasafluenceindicator. theCameca Availability of intheirradi- beryllium clusions about therelative valueof ratios. Whenmeasurements are complete, wecandraw con- and weanticipate beingable tomeasure titaniumisotope neutrons, transmutedfrom carbonbyhighenergy beryllium 10.We expecttofindsome 10 AMUconsistentwithberyllium tial measurements indicated anionpresent withamassof ments were made inlate 2011 andcontinued into2012.Ini- measure- Preliminary forremoved analyses. at twointervals In FY2011,theremaining samplesirradiated at MURRwere cator that AECLwouldbereluctant touse. LWRs, more apurpose-built practicalthaninvestigating indi- for investigated Zircaoly impurities,aswehave previously of sure tubes.Thesecircumstances combinedtomake analysis remove routinely metalsamplesfrom andthey reactors, pres- wires orsimilarneutron absorbing material intheCANDU theproject: AECLisnotinclinedtoputflux of the beginning whichwewere unaware at of facts twoimportant revealed at theATR. personnel Laboratory MeetingswithChalkRiver yses wasprohibitive. metalsirradiation Theteaminvestigated obtainingtherequired anal- for thesemetals,andthecostof MURR. However, onfile MURRdidnothave safetyanalyses Original planswere toirradiate at metalswithgraphite for analysis. shipped totheSIMSlaboratory Graphite were exposure samplesfrom package thisfirst The capsule washighinactivity, whichdelayed unpackaging. 2010, andthealuminum capsule wasopenedinaglovebox. MURRsamplewasshippedtoPNNLinlate work. Thefirst the and presentations of for irradiation at MURR samples aging graphite includedpack- efforts FY2010,primary For samples toMURR. was shippinggraphite plishment inFY2009 The mainaccom-

Construction of PBMRfuel. Construction of

Novel N/Gamma Flux Monitoring Materials for Safeguards and Proliferation Detection Mary Bliss

The PNNL-developed manufacturing process produces This project investigates the fundamental 6Li glass of exceptional optical properties, but it was detection properties of bulk thermal neutron- unknown if bulk pieces could be routinely produced without clouding from phase separation. Producing a sample sensitive glass for combined neutron and of bulk glass was an early goal for this project. A key part gamma-ray response for use in compact of the production effort is to keep cerium in the reduced detector systems. oxidation state to maximize light emission and minimize absorption. Recent advances in high-speed electronics make it possible Signal readout. An XIA PIXIE 500 was procured for this to separate signals from various radiation sources without project, along with pairs of blue and red enhanced Hama- relying on sample geometry to minimize gamma-ray matsu single photon-counting 2.5 cm diameter photomul- response so that neutrons can be detected with confidence. tiplier tubes. Prior to PIXIE, an older system was used to The neutron-to-gamma-ray ratio is useful in monitoring characterize the samples, which were mounted to blue- and identifying various types of special nuclear material enhanced PMT using optical coupling grease. The PMT and for Arms Control Treaty verification and has commercial sample are placed inside a light-tight box. Depending on uses in geologic well logging and nuclear energy. measurement, a 252Cf neutron source was combined with This project will develop, test, evaluate, and potentially polyethylene moderator to maximize thermal neutron flux deploy uniquely designed materials for neutron and and/or 2 inches of lead shielding was used to suppress gamma ray monitoring with potential applications in safe- gamma rays. Cadmium foil suppressed thermal neutrons. 137 guards and proliferation detection. Our work will require A Cs source was used for gamma ray energy calibration fundamental radiation detection material design and and response. development, followed by integration into a prototype Pulse-shape processing. Each light-generating event is detector system with follow-on hardware testing under recorded with the time windows on the order of 5 ns, controlled radiation fields. We will create new instruments slightly slower but comparable to PMT response. In addi- for use in locations impossible for 3He detectors, advanced tion to requiring high-speed electronics for data collection, formulations of enriched glass that simultaneously detect a large amount of computer memory is needed to store the thermal neutrons and gamma rays. Energy windowing will pulses. One issue is defining pulse shape to maximize dif- be applied to gain information from gamma ray signals. ferences between gamma rays and neutrons. Along with New signal processing electronics and algorithms will also pulse shape analysis, Raman spectroscopy was used to offer the opportunity to use this material in bulk. explain variations in scintillation light output with glass Neutron sensitive glass. composition. Analysis indicated that the glass samples We developed processes to contain up to three distinct glassy regions with only one produce enriched lithium correlating to scintillation brightness. Stacking two disks glass of very high optical of glass did not significantly change the appearance of the quality to manufacture into pulse-shape processing data. waveguides for large-area Further research is needed to understand more succinctly neutron detection. The the nature of the pulses that appear between the neutron driver was to enhance the response cloud and the gamma ray response. This may be neutron signal over gamma due to recombination of excited states (non-proportional- rays geometrically. The thin ity), thermal depopulation of long-lived excited states, or A cast piece of PNNL-produced waveguides had minimal the neutron capture by products escaping the material. neutron sensitive glass (4 × 4 × gamma ray response when These results show that bulk enriched lithium glass can 3 1.5 cm ) illuminated with UV light. the neutron capture reac- be used as a higher performing thermal neutron detection tion deposited its full energy in a single waveguide. The material than previously thought. Our work was published PNNL waveguides transmit usable signals over lengths in the Journal of Non-Crystalline Solids and presented at of 2 m. an international conference. PN11059/2381

233 Nuclear Science and Engineering 234 PN12076/2477 rent state-of-the-art), MCNPX/MCNP6(MonteCarlo trans - andthecur considered todate theory are 3DB(diffusion each. Thecomputercodes the strengths andlimitation of approachesMultiple analytical are beingstudiedtoidentify staff. new and trainingof field nuclear analysis requiresadvancing exerciseregular skill setintherapidly withacompetitive ing acore staff withbettercomputinghardware.ues toadvance Maintain- high-fidelity depletionproblems ischallengingandcontin- are wellestablished, themtocomplex,full-core, applying from theprocess. Thoughthecomputational approaches requests andtoestablishgram reasonable expectations codesrequiredtransport response toprovide timely topro- methods andcapabilities for thecurrent three-dimensional This work establishes baselinecomputational models, locations. of samplesfrom awidervariety reactor operation anduseof the allowsfor moreanalysis accurate assessmentsof thereactor.of theneutronics Increasing thefidelityof or plutonium production energy to calculate theintegrated and samplelocation, reactor numerical modelsare used tron fluence at thesamplelocations. Basedonthefluence fuel reactor componentsandrelating those ratios toneu - fission reactor bymeasuringstable isotoperatios innon- orplutoniumproduction ina which estimates theenergy techniques usedat PNNListheisotoperatio method(IRM), ration independently. thekeynuclear archeology Oneof facility operator’s toconfirma decla- past operating history todeterminehistoricalinformationnuclear facilities and isthemeasurement basedstudyof Nuclear archeology Christopher J.Gesh bilities forNuclearArcheology Applications Three DimensionalNeutronics AnalysisCapa- tional androbustnationallaboratorycapability. required inmaintainingacurrent,fullyopera- archeology methods,significantworkis While PNNListheleaderindevelopingnuclear and thecommercialnuclearpowerindustry. ing armscontrol,nonproliferation/verification, United Statesneedsformanymissions,includ- Nuclear archeologyisavitalcapabilitythatthe Nuclear ScienceandEngineering - sampling duringsummer2013. model anIRTreactor that will besubjectedtoIRM expressed thisproject to interest inusingthisresults of interest. Theclienthas tospecificproblems of developed be coordinated withanotherproject thetools toapply tronics methods)are plannedfor FY2013.Thework will IRMneu- of cross sectionimprovement andonesummary measured IRMdata nuclear tosupport involving theuseof son betweenthecodesandmethods.Two (one papers Work continues inthedocumentation andresult compari- fluence modeling. the FY2013scopethat involvedex-core, in-vesselthermal cross sectionsremains apriorityfor theIRMwork andfor suitable of tions for Attiladepletion.Thedevelopment casespecificcross sec- isbeingusedtodevelop package) theSCALEcode k-eigenvalue. Inaddition, of Triton (part totheresulting calculated inputparameters of sensitivity on theAttilaMagnox modelinorder todeterminethe expected.Aparametricstudy was conducted originally hasbeenfound tobemore problematic than library asuitable burnup crossassigned. Developing section Attila, where themeshhasbeengenerated andmaterials A MAGNOX into core imported modelwassuccessfully nates results. mation willbecompared withsubsequentdiscrete ordi- Thisinfor the MCNPX/CINDERandMCNP6-betamodeling. nup for B-Reactorwaslocated andwill beusedtovalidate bur calculation. Measured fuelisotopicsasafunctionof of inthistype willbeusefulinquantifyinguncertainties eters thepowerdistribution,andlike. Theseparam- gence of fissionproducts tracked, numberrunning time, of conver burnzones,temperatures,the results tothenumber of Hanford B-reactor were of usedtodocumentsensitivities aMAGNOX reactor andthe MCNPX andMCNP6modelsof ordinates andMonteCarlo results. and willbeusedtocompare withsubsequentdiscrete model for a generic MAGNOX reactor hasbeencompleted andAttila(discreteport) a3DB ordinates First, transport). - - - Ultra-Low-Background Gas Measurement: Building an Advanced Capability Allen Seifert

For this project, FY 2012 began with the development of We are developing an ultra-low-background hardware to enable high-sensitivity demonstration mea- (ULB) measurement capability for gases avail- surements. Specifically, construction began on a general- purpose gas-handling system designed to handle a broad able to basic science and emerging application range of sample types, with evolution of the gas processing areas, leveraging our new facilities and achiev- capabilities planned for each year of the project as new ing scientific competitiveness on an interna- demonstration measurements are identified. The first such tional level. This work builds capability for development for the gas-handling system happened late in diverse new applications, resolves unique mea- FY 2012, as a large-volume transfer system was added to surement challenges, and will result in high- enable 37Ar and 39Ar. A successful demonstration, also at the impact demonstration measurements. end of the FY, proved the ability to load custom blends of an argon-based count gas into the ULB detectors.

Extremely sensitive nuclear measurements are exploited in Five ULB detectors were constructed this year to enable a number of scientific disciplines. Measurements of low innovative new measurements. Significant effort was put levels of radionuclides in the environment allow their use into characterizing and optimizing performance of the new as tracers and for age dating. Examples include studies of detectors at pressures and gas gains higher than previously fate and transport of dust (7Be) in the atmosphere, deep- employed. These new operating parameters allow for the ocean mixing (39Ar), estuary and deep ocean sedimentation introduction of greater quantities of sample and thus pro- (32Si), soil erosion (7Be), and studies of the mechanisms and vide an increased sensitivity to low-level samples. The pathways for carbon fixation by soil biota and residence combination of these new settings and new gas-loading times for organic and inorganic carbon (3H, 14C). Low-level capabilities enabled sensitive measurements of samples measurements are also useful in the verification of nuclear of opportunity of 37Ar and 39Ar. 37 treaties; e.g., detecting Ar produced by an underground An observed gap between modeled and experimental nuclear test. Across all of these areas, the need for greater detection efficiency offers the opportunity for significant sensitivity is paramount. gains through modeling and design work. Thus, a modeling ULB proportional counters have been used for decades for effort was undertaken to understand factors in the current samples that can be prepared as gases and are particularly design contributing to fiducial volume losses. The detector sensitive to low-energy beta and electron emitting radionu- electric field has been modeled with Ansoft’s Maxwell3D clides. Recently, PNNL developed a design with improved finite-element software, with further work planned using radiopurity materials, and this project moves forward from the Garfield gas detector simulation code to investigate that foundation. High-sensitivity measurement methods pulse formation and detector end effects. will be developed for specific isotopes, and the detector Activities planned for FY 2013 will develop calibrated, design will be enhanced to allow greater sensitivity. As an ultra-sensitive methods for specific measurements. Optimi- 14 example, a measurement of combined tritium and C in zation of the operating parameters and performance of the methane will be developed, which will allow these radio- ULB proportional counters will allow key radiotracer mea- tracers to be used in the study of hydrocarbon compounds surements (37Ar, 39Ar, 3H/14C) in parallel projects at previ- in soil bacteria. Analogously, this project will develop the ously unattained sensitivities. Further development of the gas handling and quantification tools necessary to work gas-handling system will be aimed at precision quantifica- with a variety of input gases derived from other environ- tion of gas samples. The ULB proportional counter design mental samples. The suite of ULB proportional counters will be refined to allow operation at higher pressures, and necessary for these high-sensitivity demonstration mea- detector electronics capable of operation at the corre- surements will also be constructed, with design improve- spondingly higher voltages will be developed. Finally, ments incorporated as they are developed. The ultimate novel uses of the ULB proportional counter will be goal is the development of an enduring ULB capability for explored, such as utilizing 3He to enable low- the measurement of environmental samples at PNNL that background neutron detection. will position the Laboratory to compete with other recognized leaders around the world. PN12078/2479

235 Nuclear Science and Engineering 236 PN12080/2481 ing the necessary collaborations withOakRidgeNational ing thenecessary - Significant timeduringFY2012wasdedicated todevelop ment facility. theenrich- predicting thethroughput andenrichmentof variousarcheological techniquesfor accurately the ability of tionally, thisstudywill furnishascientificbasis for assessing material production at auraniumenrichmentfacility. Addi- tion isusingPNNL’s toquantify scienceexpertise surface enrichment plantoperations.- Thisapplied scienceinvestiga centrifuge toprovide auniquemethodfor estimating past challenges, thisproject ismaximizingasignature insidea Withthese piping. ment plantsthrough examiningfacility enrich- thethroughput andenrichmentof of understanding notedthat some itmay bepossibletogain been previously thenonproliferation verification challenge. of Ithasalso grail” operations isa“holy to assesspastenrichmentfacility nonproliferation communityhasrecognized that anability anuraniumcentrifugeenrichmentfacility. of The history nique for determiningthethroughput andoperational There doesnotcurrentlyexistarobust measurement tech- Bruce D.Reid Signature Exploitation Uranium EnrichmentFacility sputter time. X-ray photoelectron spectroscopy concentration asafunction of these facilities. the internationalconcernsassociatedwith could haveavaluableapplicationforresolving for validatingenrichmentproductionhistory ous nonproliferationcredentials.Atechnology Nuclear Non-ProliferationTreaty orhavedubi- ment technologyarenotsignatoriestothe A numberofcountrieswithuraniumenrich- Nuclear ScienceandEngineering

accurately the growth rate of theUF rate thegrowth accurately of walls. Additional modelingisunderway toestimate more depths for depositeddecay products withinthecentrifuge initial modelinghasbeencompleted,whichestimates that were withthistechnology. unfamiliar Inaddition, the for inaboutbackground staff uraniumenrichmentfacilities The literature searches are complete. Thisprovided detailed combination. expectedwiththis accuracy of to determinethelevel uranium.Additional testingisunderway and quantityof corrosion/passivation isotopicratios, elementalratios, (TOF–SIMS) provided ionmassspectrometry secondary photoelectron spectroscopy (XPS),andtime-of-flight– scanningelectron microscopy (SEM),x-ray combination of with promising for results. the Thecurrent methodology initialmeasurements have beencompleted The project’s mentioned institutions. expected tomaintaintherelationships withtheabove andadditionalment hasbeenpositive, limitedfundingis - piping andapumpfor FY2013.DOE/NNSAclientengage withtheproject expectingtoreceive tions are encouraging, thesecollabora - Theresults project analysis. of purpose of hardware forlished toobtainactualenrichmentfacility the andClimate Changewasalsoestab - Energy of Department future Coordination objectives. withtheUnitedKingdom Virginiatoaccomplish of andtheUniversity Laboratory enrichments of UF enrichments of cated toprototypic conditionsandexposedtovarious interest have- beenfabri variousmaterials of Samples of thecorrosion layer.thickness of theoperation bymeasuring the determine thelengthof Byestimating rate, thegrowth analysis. wecan surface ation layer rate growth the willhelptheinterpretation of interest.- thecorrosion/passiv Having agood prediction of thematerials of centrifugefor allof layer of onthesurface analyze components from anoperatinganalyze enrichmentfacility. ments anddata interpretation techniques,andreceive and a uraniumenrichmentfacility, continue refining measure- for During FY2013,wewillmaintain themodelingeffort FY2013. isplannedfor of thestart mass-spec analysis andTOF–SIMS. Chemicaletchingand with SEM,XPS, enrichment facility. Initialmeasurements have been made theuranium needed todeterminethecharacteristicsof techniquesprovide theinformation or combinations of the project todeterminewhichmeasurement technique that canbeapplied toactualhardware.ogy Thisallows These samplesprovide amethodol- theability for develop 6 for surface and chemical analysis. andchemicalanalysis. for surface 6 corrosion/passivation

Physics Absolute Gas Counting Measurement Techniques Richard M. Williams

focused on the same low-level measurements that we are This project aims to improve significantly the attempting at PNNL. uncertainty in radiation measurements of very Our efforts have been organized into a 3-year time frame low activity gas samples. With these enhance- that began in FY 2012. During this first year of the project, ments, scientific endeavors ranging from significant progress was made in developing an ultra-low radioisotope dating of environmental samples background, absolute-counting measurement capability. to nuclear explosion monitoring will stand Specifically, two different sets of proportional counters to benefit. (or detectors) were fabricated at PNNL and passed initial characterization tests. The method of length-compensated proportional counting Low-level radiation measurements are extremely difficult relies upon the use of a set of proportional counters that and complex due to the ever present background radia- are identical, except for their length. The final outcome of tion. To suppress this background, PNNL has developed a the measurement (i.e., the specific activity of the gas sam- unique method to produce ultra-pure copper to fabricate ple) is calculated by analyzing the differential count-rates detectors which possess almost no naturally occurring (normalized by the differential gas volumes) from this set radioactive elements. In addition, PNNL has recently con- of unequal length counters. For this project, two sets of structed a shallow underground counting facility, which counters were fabricated: the first is a set of three unequal naturally shields some of the cosmic ray background. This length detectors fabricated using PNNL’s ultra-low back- project leverages these two existing capabilities and adds a ground (ULB) copper. These ULB counters will be used to novel counting method to yield a unique, one-of-kind mea- make the lowest level measurements in the underground surement capability. counting facility. In addition, a second set of four detectors The objective of our project is to develop a technique to was fabricated using high-purity, commercial copper and measure the level of radioactivity of gas samples without PNNL’s clean assembly techniques. This set is intended for the need for calibration of the system with reference stan- measurements that can be done in one of PNNL’s standard, dards that are often not available, very difficult to obtain, above-ground laboratories. and cost prohibitive to produce. Only a handful of labora- A key background test has been performed with the ULB tories around the world employ the technique that we are counter set. The detectors were filled with standard P10 developing here – a process referred to as length-compen- counting gas (but no radioactive sample) and placed in the sated proportional counting – but no other facilities are shielded counting tomb in the underground laboratory. The background counting rates were on the order of 100 to 200 counts per day (over the full energy spectrum). This outcome sets the conditions for making very low-level measurements with counting times of days or even weeks in the underground facility. Such measurements would not be possible using conventional materials in standard above-ground laboratories, where background count-rates would be in the range of 10,000 to 100,000 counts per day. During the next year of the project, the measurement methodology will be investigated and the uncertainty of the absolute low-level measurements will be quantified. We will engage with the international radiation metrology community and seek collaborative measurement opportunities to validate the capability that PNNL is developing PNNL-fabricated set of unequal length proportional counters for for basic science applications as well as for nuclear non- use in making absolute specific activity measurements of low-level proliferation treaty monitoring research and development. radioactive gas samples. PN12002/2403

238 Physics Highly Efficient and Cost Effective Gamma Detection Arrays for FRIB Luke E. Erikson

resulted in a packed schedule of plenary talks and break- This project designed a new detector array out sessions. that would be used in the next generation of • We performed maintenance operations on the second nuclear astrophysics experiments to increase MARS cryostat. In its original configuration, this detector understanding of the formation and evolution array had 9 of 14 crystals operating. At this time, all 14 crystals are operational (although with three under of the universe. biased). Unfortunately, this process introduced electrical noise into half of the array, so additional work will be A new DOE user facility currently under construction at required before the system is ready for deployment. Michigan State University will allow scientists to probe new • Through community outreach, we arranged for the exist- and exciting topics in nuclear physics. The Facility for Rare ing MARS array to be included in a nuclear astrophysics Ion Beams (FRIB) is expected to cost approximately $600 mil- experiment at TRIUMF during FY 2013. This experiment lion and become fully operational around the year 2020. will study the 20Ne(p,g)21Na reaction rate, which is relevant The higher beam intensities and advanced beam targets will to the manner nuclei are formed during stellar nova and allow scientists to make several first-time measurements of is an excellent opportunity to demonstrate PNNL’s approach key nuclear reactions. In nearly every case, measurements to HPGe detector arrays and their use in this community. made at this facility will require the efficient detection of • As part of this project, PNNL is now a member of the gamma radiation. In fact, several stages of gamma detection JENSA (gas jet target), SECAR (mass separator), and RDS will be required for many experiments. (decay station) collaborations for FRIB. While developing solutions for projects related to national The primary technical accomplishment is the advanced state security, PNNL has created a unique infrastructure and of the detector array design. This process was greatly acceler- expertise in the construction of High-Purity Germanium ated by extensive reuse of design work for national security (HPGe) arrays for gamma detection. One of the approaches clients. For example, the crystal platter and mounts were used successfully on the MARS, RNLabs, and Majorana proj- taken directly from RNLabs, while the integrated crysostat ects is the collection of commercially produced HPGe detec- cooling system was inspired by the MARS design. tors together within a single cryostat. This is a technically The present design offers excellent integration into the gas challenging approach that has several advantages toward jet target array under construction for FRIB. By taking advan- enhancing detection efficiency, meeting unique geometry tage of the large vacuum chamber of the gas jet target, the requirements, and using cost-effective materials. Specifically, detector array provides excellent efficiency for detection this project leverages PNNL’s capability in the construction and high resolution to increase the signal to noise ratio. of HPGe arrays into a leadership role in the scientific The system is designed to be flexible and can translate to program that will be undertaken at the FRIB. accommodate a wide variety of other detectors such as Throughout FY 2012, our project has had the following charged particle arrays. significant accomplishments: • Completion of initial design for a new 14 crystal HPGe detector array for use at the FRIB and other accelerator facilities. This includes engineering work for mechanical design, crystal mounting, cooling, bias and readout, initial finite element analysis, a rack system for mounting the array at the beamline, initial performance simulations, and an engineering specification/requirements document. • In May 2012, PNNL hosted a workshop on the Detector Arrays for Nuclear Science. More than 50 leaders from the scientific community from 18 different institutions attended this 2-day event. The agenda focused on PNNL’s capabilities in detector research and development. This

Summary of JENSA gas jet target. PN12038/2439

239 Physics 240 PN12045/2446 ence can claim familiarity with only ~17% of the universe’s theuniverse’s ~17%of withonly ence canclaimfamiliarity physicalresearch, sci- of From thelastcentury the universe. thematter composing asmallfractionof and understand glaring realization that know scientistsandresearchers only Public andscientificinterest indark matter stemsfrom the physics. particle thestandard modelof interactions of erwise doesnotparticipate intheelectromagnetic orstrong whosegravitational influenceisubiquitous butoth- particle interactingmassive aweakly isthepresence of tional effects Theleadingtheoreticalgalaxies. candidate for- thesegravita andtherotation gravitational of lensing, tic clustering, cosmicmicrowave- radiation, background galac intensity of from appear uniform indeviations gravitational effects through rays. thevisiblespectrum The anduptogamma the electromagnetic spectrum from radiofrequencies despite beingentirelyinvisibletoourtelescopes,measuring throughout theuniverse observed its gravitational effects physicistsalike.cal particle Dark matter iswellknownby physics measurement soughtbycosmologistsandtheoreti- is,perhaps,particles thesinglemostpressing astro particle - cosmologicaldark matter Unambiguous direct detectionof John L.Orrell and Science Low EnergyThreshold GermaniumDetectors CoGeNT detector. expectedcontinuum forGEANT-4 backgrounds the simulation of induced bydarkmatterparticles. a signatureofradiation-interactionevents for adarkmatterexperimentsearching project alsogeneratedabackgroundmodel environments atundergroundlocations.This surements madeinultra-low-background tors foruseinultra-sensitiveradiationmea- low-background cryostatforgermaniumdetec- This projectisdevelopingalow-capacitance, Physics

capability of CoGeNT-like forcapability dark detectors matter of research. stepisdemonstrating modelisanimportant the ground thisback- Publicationknown andunknownbackgrounds. of both dark matter isfraughtwiththedangerof detection of CoGeNT data. tothe However, events assigningthisexcess of detector are expectedtohave anupturnasseeninthe nuclear recoil reactions from insucha dark matter events from thesoughtafterdarktially matter interactions.The 2 keVee orpoten- backgrounds isduetoknownradioactive of seenbelowanenergy events therisingexcess of if for for theCoGeNTexperimentisimportant understanding model thisbackground of CoGeNT data. Thedevelopment overlaid thissimulation withthe effort ure showsresults of the CoGeNTdetectorthrough GEANT-4 simulations. Thefig- Additionally, of theexpectedbackgrounds weinvestigated whichare shown. of assembly thedesignandinitial theCoGeNTdetectorcryostat, of tous U/Th-chainisotopesshouldbereduced relative to that due toubiqui- thebackground considerations; nevertheless, byfabrication driven material isprimarily The choiceof construction.electroformed copperisusedinthecryostat OFHCcopperandPNNL’sence. Amixture ultra-pure of surement instruments for applied research andbasicsci- radiationexperience buildingultra-lowbackground mea- andcross-arm isbasedonPNNL’stom coppercryostat long thecus- experiment. Themechanicalandthermaldesignof ray suchasisusedintheCoGeNTdark spectrometers matter threshold germaniumgamma- for ultra-low-energy cryostat bled anext-generation low-capacitance, low-background year,In theprevious wedesigned,fabricated, andassem- unknown toscience. butisotherwise tional imprintsthroughout theuniverse dark matter –that wecaninfermust existfrom- itsgravita remaining ~83%issomeyet tobemeasured material –the free The hydrogen andheliumproduced bytheBigBang. and heavy elements,neutrinos,stars, matter composed of testing in the PNNL shallow underground laboratory cleanroom. laboratory testing inthePNNLshallowunderground undergoinglow-capacitance vacuum mount.Right:Prototype cryostat theinterior, of designandcut-away view Left: C4cryostat showingthe Mu2e September Beam Test David M. Asner

This project prepares a PNNL germanium gamma-ray detector for measurements in a beam test at TRIUMF. The goal is to provide proof-of-concept measurements of muonic X-rays from the stopping target. The project will initiate the effort to simulate detector The Mu2e detector as currently envisioned. The stopping target monitor will be located at the end of the detector solenoid at response and model the overall particle flux far right. at the stopping target monitor location. This project includes development of the detector data acquisition system in advance of the beam test at TRIUMF The Mu2e collaboration proposed a new facility to study where data will be collected at the muon stopping target in charged lepton flavor violation using the existing DOE several configurations including muon beam on with and investment in the Fermilab accelerator complex. Specifi- without stopping target and background data while the cally, Mu2e proposes to measure the ratio of the rate of the muon beam is off. Once data are collected, they will be neutrinoless, coherent conversion of muons into electrons analyzed to identify X-rays from the stopping target and in the field of a nucleus relative to the rate of ordinary other potential sources from the beam and room back- muon capture on the nucleus. The conversion process is grounds. The resulting measurements of rate and resolu- an example of charged lepton flavor violation, a process tion will be used in further research and development of that has never been observed experimentally. the stopping target monitor and will be presented to the To determine the number of muons captured in the stop- greater Mu2e collaboration. As a parallel task at PNNL, ping target, the characteristic X-ray spectrum generated by integration of the germanium gamma-ray detector into the the formation of muonic atoms will be measured with a Mu2e detector simulation and the simulation of the parti- high-purity germanium crystal detector. Due to high cle flux in the vicinity of the stopping target monitor will incoming muon flux, the Mu2e environment will poten- be initiated. tially contain high rates of neutrons from atomic disassoci- Additional activities during FY 2012 included three PNNL ation in the target. These neutrons can generate defects in staff participating in a week-long software and simulation the germanium lattice and degrade detector performance. workshop at Fermilab. Both the effort to model the germa- In addition, electrons generated and transported with the nium gamma-ray detector in the Mu2e detector simulation muons will create bursts of photons as they interact with and modeling the particle flux near the stopping target the target. The large number of photons generated will have been initiated. A XIA PIXIE-based data acquisition sys- potentially blind the detector to the muonic capture X-ray tem utilizing PNNL-developed software with data output spectrum. Despite these challenges, high-purity germa- into the high energy physics standard ROOT format for sub- nium is still the detector of choice for high-resolution sequent data analysis with the ROOT toolkit was also devel- X-ray and gamma-ray spectroscopy measurements. oped. Along with PNNL, scientists from Osaka University, the University of Washington, Fermilab, and TRIUMF will operate additional detectors at the TRIUMF beam test. Planning for the beam test and coordination of the data acquisition for the various detector systems was performed in collaboration with the other participants. Due to technical problems with a superconducting solenoid on the TRIUMF M9B beamline, the September beam test was cancelled and rescheduled for November. Finally, a proposal has been submitted to DOE to support PNNL par- ticipation in the Mu2e experiment at Fermilab.

Lyman X-ray spectrum from muon capture by Al. PN12049/2450

241 Physics 242 PN12058/2459 ing how stars age. For many stars, the manystars, age. For ing howstars reaction rates- arePrecise keytounderstand knowledgeof Robert Runkle High-efficiency HPGeArrays Nuclear Astrophysics Measurements using knowledge of the knowledge of thisproject our istoimprove substantially The goal of thatgrounds make difficult. data analysis experiments have from suffered lowstatistics andback- decades, attempts duringthelastcoupleof Despite several the lowrate also makes tomeasure. thereaction difficult plays role animportant inthestar’s However, evolution. production andthereforea bottleneckthat limitsenergy Nuclear Laboratory (TUNL). Nuclear Laboratory Nuclear Astrophysics (LENA) at theTriangle Universities for Experimental conduct theexperimentisLaboratory locations intheworld equippedto thefew staff. Oneof requires accesstoasuitable andsupport accelerator facility thesereactioncollaborations. rates Themeasurement of andlaboratory Our project hingedonsuccessfuluniversity security domaintoaddress abasic research need. oursuccessesinthenational toleverage an opportunity thisproject thisreason, as weviewed For detectors. ratory overtypical labo- advantages theMARSsystemseveral gives detectionefficiency,segmentation, andhigh resolution PNNL’sthe capabilities MARSarray. of Thecombination of in July 2012. in July intothe LENAThe MARSdetectorarray beamlineat integrated TUNL the evolutionofnormalmatterinuniverse. reaction, whichisimportanttounderstanding the-art detectorarraytomeasureanuclear This projectusesaPNNL-constructedstate-of- Physics 14 N(p,g) 15 O reaction whiledemonstrating 14 N(p,g) 15 O reaction is detector whilemeasuringthe LENA toproduce data approximately from theMARS 6GBof scientistsworked at withstudentsandfaculty PNNL staff beamline for theexperiment.Duringfollowing week, thesysteminto transportation andintegrated ground relocated MARStoLENA by By midFY2012,PNNLstaff techniquessuchasanti-coincidence counting. analysis andalsoallowsfor alternative thesesummingeffects of tion. However, detectorreduces asegmented thelikelihood are usedand/orwhenplacedincloserproximity tothereac- Thisproblem increases difficult. aslargerdetectors gammas rays perreaction, individual whichcanmake recognition of UNC. Preliminary results showthat the Preliminary UNC. thedata isbeingconductedat bothPNNLand of Analysis anti-coincidence modetoreduce backgrounds. for examplewhenoperatedhas hadbeneficialeffects inthe result. Additionally, theMARSarray nature thesegmented of toproduce apublishable detected withenoughaccuracy 0.5 Coulombs of 0.5 Coulombsof protons onaTa targetimplantedwith 15 Coulombsof atexperiment, accelerators LENA deliveredapproximately During the theuniverse. of humankind’s understanding LENA andusedtomeasure anuclear reaction to important deployedThe MARSdetectorsystemwassuccessfully at tion rate measurement. thereac- arrays, toreport andajournalarticle theresults of thePNNLapproach tohighefficiency discussing thevalueof tions tobeproduced atechnicalpaper from thiseffort: thisreason, weexpecttwopublica- physical experiment.For array produced at PNNLhasbeenusedfor anuclear astro- HPGe This represents timethat ahigh-efficiency thefirst throughout theexperiment. system ransmoothly The 14highpuritygermanium(HPGe)detectors. consisting of design systemisthesegmented over atypicallaboratory theMARSarray thismeasurement, of oneadvantage For for anexperiment. began LENA targetstation duringsummer2012,andpreparations wasobtainedforlocal support operating MARSwithinthe Carolina North (UNC), of In collaboration withUniversity 14 N(p,g) 15 O reaction tendstoproduce multiplegamma 14 N ions. 14 N(p,g) 15 O reaction. TheMARS 14 N(p,g) 15 O was Search for New Physics at the Intensity Frontier David M. Asner

This project analyzes the world’s largest sample of heavy quark decays recorded by the Belle detector at the KEK Laboratory in Tsukuba, Japan. The goals are to discover and characterize the properties of exotic bound states of matter and anti-matter. This project will also lay the foundation for leadership in the physics program at the upgraded facility Belle II/ SuperKEKB. Analysis of the Belle data sample reveals six bound states of a

bottom quark and its anti-particle including the previously hb(1P) and hb(2P). Discovery and characterization of the properties of exotic states of matter recorded by the Belle detector are the • We encountered a first observation of radiative transi- objectives of this research project. The Belle detector, oper- tions from p-wave to the s-wave singlet states denoted ated at the asymmetric electron-positron collider KEKB as hb(1S) and hb(2S). This work includes the most precise from 1999 to 2010, accumulated the world’s largest data measurements of the properties hb(1S): the Bottomo- sample for the study of B mesons (subatomic particles con- nium ground state. These results were presented at the taining one heavy bottom quark and a light anti-matter 11th Conference on the Intersection of Particle and quark) and Bottomonia (bound states between a heavy Nuclear Physics and submitted for publication. bottom quark and a heavy anti-matter bottom quark). • We performed searches for new phenomena in D Bottomonia states were first observed about 30 years ago. mesons (bound states of a charm quark and a light However, there remain areas in which knowledge of these antimatter quark) and are a focus of planned FY 2013 systems needs improvement, including anticipated but work. Two publications – one update of an article in unobserved states of matter. Specifically, bottomonia spec- collaboration with Cornell University from Physical troscopy is generally regarded as a key source of informa- Review D and two mini-reviews in the biennial Review tion necessary for the development of the understanding of Particle Physics – during FY 2012 lay the groundwork of the quantum theory describing the strong interaction for that effort. that holds atomic nuclei together. This project has enabled PNNL to build a reputation for A goal of this project is to refine our understanding by scientific excellence through the discovery and elucidation making new and more precise measurements of the of bottomonium states in FY 2012. Next year, the focus will parameters of the bottomonia states, their decays, and shift to the search for new phenomena in B and D meson their transitions. The expected results of this endeavor decays. A 50-fold increase in data volume is anticipated at provide important benchmarks for a variety of theoretical the upgraded Belle II detector. This project will lay the approaches. This project is seeding the initial collaboration foundation for PNNL leadership in the Belle II physics between PNNL and the international Belle research team, program. which consists of a collaboration of 300 physicists across four continents. In FY 2013, several conference talks and three specific publications are planned: one report in collaboration At the conclusion of FY 2012, this project has had the with Luther College on the study of the c (3P) bottomo- following significant accomplishments. b nium state, a second study on matter anti-matter oscilla- • We made a discovery of two exotic states of matter tions with D mesons, and finally an article in collaboration collectively referred to as Bottomonium p-wave singlets with Yonsei University in South Korea on a search for rare

and named hb(1P) and hb(2P), respectively. These results and forbidden decays of B mesons. were presented at the prestigious international conference Rencontres de Moriond in La Thuile, Italy and published in Physical Review Letters. Thus far, this journal publication has been cited 50 times. PN12069/2470

243 Physics 244 PN12079/2480 demonstration willbemadewith spectrum withaprototype experimentat UW. Theprototype trappedfrom electrons individual andreconstruct anenergy demonstrate theability radiation todetectthecyclotron theneutrino.sure themassof Ourimmediate goal isto thetechniqueanduseittomea- Technology todevelop of California-Santa Barbara,andtheCalifornia Institute sity of Technology,Astrophysics, MassachusettsInstituteof Univer ington (UW)Centerfor ExperimentalNuclearPhysicsand Wash- collaboration of withcolleaguesfrom theUniversity at PNNLhaveScientists andengineers formed theProject 8 100 timeslighterstill. toweightheneutrino,sufficient whichcouldbeupto andmayultimate stillnotbe neutrinomasssensitivity ments basedonexistingtechniqueshave reached their experts areexperts andprecisely, able todothissafely and ect 8needsastrong source andourresident radiochemistry rial toreplenish that source inFY2013. early PNNL madeonesource inFY2012andhas obtained mate- prepare sources for useinultra-highvacuumenvironments. uniqueandrequiresour caseissofar original methodsto eral source production recipes tobefound intheliterature, used as acalibration standard.widely Whilethere- are sev PNNL manufactures the frominstrumental true effects spectralfeatures. much more simple spectrum that willallow ustodiscern that emitselectrons withsimilarenergiesastritiumbutina to neutrino mass by Einstein’s relation of E=mc to neutrinomassbyEinstein’s relation of E small reduction inelectron energy a the neutrinoinvolvedintritiumbetadecay byobserving existing techniques.Thegoal istomeasure themassm resolution superiorto sion, ourmethodpromises energy Because frequencies canbemeasured withextreme preci- theradiation. of electron isencodedinthefrequency atrapped of trapped electrons.by magnetically Theenergy radiation cyclotron emitted trons basedon thedetectionof (1–100keV) elec- medium-energy for thespectroscopy of technique basic nuclear anew science, weare developing itstopprioritiesin neutrinomassasoneof the measuringof large-scalestructures AsDOErates intheuniverse. tion of - physicstotheevolu particle beyond thestandard modelof Neutrino masshasfar-reaching implications from theories Brent A.VanDevender to Measure theNeutrinoMass Ultra-Precise Electron Spectroscopy search fortheabsolutemassofneutrino. spectroscopy techniquetobeappliedinthe This projectisdevelopingafundamentallynew Physics 83m Kr sources for theprototype. Proj- 83m that canbeattributed Kr, gas aradioactive 2 . Experi- 83m Kr is Kr is

of of

- of Nuclear Physics. of with tritiumwillbechosenand proposed totheDOEOffice much largerexperimentrequired tomeasure neutrinomass Ultimately, theapproach that mostreadilyscalesfor the wasalready identifiedinFY2012. feasible cavity geometry aprototype basedonresonant cavities.mentation A of ment inFY2013,PNNLwillwork onthedesignandimple- prototype thewaveguide experi- of In parallelwithsupport . Letters submitted toPhysics Review and observations electrons basedonalargernumber of will be a measured energy spectrum of spectrum of will beameasured energy ted toNuclearInstrumentsandMethodsA.Thenextresult aboutarticle thetechniqueand apparatus willbesubmit - will besubmittedfor publication. Inaddition, acompanion radiation theircyclotron trapped electrons bydetectionof individual of theobservation regarding FY 2013.Anarticle FY 2012andisexpectedtogenerate result during itsfirst operating prototypeThe waveguide inlate willbegin the data acquisitionsystem. throughwaveguide thereceiver electronics tocollectionin thefullsignalchainfrom propagation inthe of and analysis onadetailedPNNLsimulation be successfulisbasedlargely chosen for thecurrent prototype andconfidencethat itwill guides and/orresonant cavities. designwas Awaveguide These were showntobeinferiorconceptsbasedonwave- concepts for Project 8were basedonantennareceivers. forsupport microwave Initial receiverdesignandanalysis. electricalengineering range. PNNLprovides theprimary signalisinthemicrowaveThe expectedcyclotron frequency The plot is the expected frequency spectrum of signalandnoise. spectrum of The plotistheexpectedfrequency that prepare thesignalsfor collectionbyadata acquisitionsystem. ineitherdirection tolow-noiseamplifiers downthewaveguide pagates waveguide. Cyclotron radiation from resulting trapped electrons pro- The Project 8prototype apparatus withsource decays intheWR42 gas 83m Kr conversion Kr conversion

On the cover:

Using expertise in material nuclear forensics, PNNL scientists and engineers are developing cutting-edge signature discovery tools related to the nuclear fuel cycle. This image of thorium nitrate and uranyl sulfate crystals was obtained from a Nikon 400 POL polarized light microscope (PLM) with cross-polars. The PLM provides unique insight into materials and enables rapid screening of unknowns. The cross-polars image of uranyl sulfate shows large 0.5-millimeter crystals exhibiting a wealth of interference colors. Contrasts in the images are due to thickness variations across the crystal, much like the interference colors from an oil slick on water. This work was jointly funded by NA-20 and National Technical Nuclear Forensics Center (NTNFC), and supports the Laboratory’s major initiative in Signature Discovery and Exploitation for Threat Detection and Reduction. PNNL scientists Leah Arrigo and Crystal Rutherford contributed to the project, which is managed by Jon Schwantes. This image was provided by PNNL scientist Edgar Buck.

A metal-organic framework (MOF) composed of zinc nitrate and 2,5-dihydroxy terephthalic acid was imaged at 10,000-times magnification as part of a study to understand how nanoporous frameworks can be created and modified to improve gas capture properties. The MOF is part of a new series of materials developed by PNNL that have tiny “cages” that can be used to trap carbon dioxide. The crystals in the cages have a diameter of about 25 nanometers. The cages themselves consist of metal ions linked together with organic ligands forming a porous network that traps gas molecules. The unique molecules have two to three times the capacity for carbon dioxide uptake as conventional solvents and appear to have an extremely high preference for carbon dioxide over other gasses. This research was funded by BES, and supports the Laboratory’s major initiative in Coupling Earth and Energy Systems for Sustainability. This image (in false color) was captured using a focused ion beam scanning electron microscope at the Environmental Molecular Sciences Laboratory (EMSL), a DOE national scientific user facility. Image provided by EMSL’s Bruce Arey and PNNL’s Carlos A. Fernandez and Praveen Thallapally.

Microbe-mineral and -metal interactions represent a major intersection between the biosphere and geosphere but require high-resolution imaging and analytical tools for investigating microscale associations. Electron microscopy has been used extensively at PNNL for geo- microbial investigations such as the controlled exposure of model organism Shewanella to environmentally relevant metals and radionuclides to understand how microbes carry out complex biogeochemical processes. At EMSL, scientists are using the state-of-the-art instrumentation and expertise in electron microscopy imaging and analyses for research of the effects of the interactions of microbes with the soil environment. Shown here is a scanning electron micrograph of Shewanella putrefaciens CN-32 cells in certain conditions that induce cells flocculation. CN-32 is a metal-reducing bacteria important for stabilization of metals in the environment. This work was funded by BER, and supports the Laboratory’s major initiative in Accelerating Innovation and Discovery and Transforming the Conduct of Science in EMSL. This research was carried out in PNNL’s Microbiology group led by Jim Fredrickson, the image was provided by EMSL scientist Alice Dohnalkova.