2

Vol. 34, No. 2, 2001

1 Editorial: Basic Research 16 Modeling a Fusion Plasma at ORNL Heating Process and Stellarator 2 ORNL’s Search for Rare 22 Isotopes 17 Neutron Sources and Nanoscale Science 2 ORNL Theorists and the Nuclear Shell Model 18 Quantum-Dot Arrays for Computation 5 Beam Technologies Enable HRIBF Experiments 20 Carbon Nanotubes and Nanofibers: The 7 Neutrons, “Stripes,” and Self-Assembly Challenge Superconductivity 22 Incredible Shrinking Labs: 12 Weighing a Move to the ORNL’s Neutron Sources 8 Nanoscale and Nuclear Astrophysics 24 Basic Geochemical 10 Modeling Magnetic Mate- Research Supports Energy rials for Electronic Devices Industries

12 In Quest of a Quark: 26 Fermi Award Winner ORNL’s Role in the Opened New Fields in PHENIX Particle Detector 20 Atomic Physics 14 New Hope for the Blind 28 Improving the Internet’s from a Spinach Protein Quality of Service

15 Human Susceptibility and 29 QOS for Wireless Mouse Biology Communication

This issue provides highlights of basic research at ORNL, including research on producing, characterizing, and finding applications for carbon nanotubes produced at ORNL by several techniques. The cover shows single-wall carbon nanotubes created by laser vaporization (note pink laser plume), which were then chemically purified and spray-deposited onto a substrate. The image above the cover image shows a carbon nanotube on electrodes, a possible nanoscale replacement for today’s microscale silicon transistors. See pp. 20-21 for a description of this and other nanoscale research. Images by Jason Fowlkes, Derek Austin, and Henrik Schittenhelm. Basic Research at ORNL

Basic Research at ORNL: A Distinguished Past, An Exciting Future RNL has a time-honored tradition of basic research accomplish- ments in physics, chemistry, biology, and the materials sciences. In 1951 ORNL researchers were the first to confirm that a neu- Editor and writer—Carolyn Krause O tron decays into a proton, electron, and electron antineutrino. Assistant editor—Deborah Barnes ORNL biologists discovered the function of messenger RNA and the chromo- Editorial Board—Lee Riedinger (chair), somal basis for sex determination in mammals. This year the Department of Energy Fred Bertrand, Eli Greenbaum, recognized three ORNL discoveries as among the top 23 of DOE national labora- tories’ scientific accomplishments. They are nickel and iron aluminides for high- Russ Knapp, Reinhold Mann, temperature applications, ion beam techniques for making longer-lasting artificial Stan Milora, Thomas Zacharia hips and knees, and the Z-contrast electron microscopy technique, which produced Designer and illustrator—Jane Parrott Lee Riedinger the most detailed image of a crystal structure ever recorded in a microscope. Photographer—Curtis Boles In December 2000, an ORNL physical chemist, the late Sheldon Datz, received DOE’s prestigious Enrico Fermi Award for pioneering the use of crossed molecular beams to study the details of chemical reactions and for demonstrating ion channeling, The Oak Ridge National Laboratory in which charged atoms pass between a thin crystal’s rows of atoms. (See p. 26.) Review is published three times a year Our basic research tradition continues today, thanks largely to support from and distributed to employees and others DOE’s Office of Science. Much of the research, described in this issue, deals in a associated with or interested in ORNL. big way with very small features of matter ranging from quarks, electrons, photons, neutrons, and heavy ions to DNA and carbon nanotubes. By studying the properties of short-lived nuclei and searching for rare iso- Jim Roberto topes, using the Holifield Radioactive Ion Beam Facility, ORNL nuclear physicists Editorial office address: are gaining a better understanding of the origin of the elements and the mechanism Building 4500-North, M.S. 6266, of energy generation in stars. ORNL physicists and electronics experts have played a major role in devel- Oak Ridge, TN 37831-2008 oping electron and photon detectors and electronic components, to help search for evidence that the universe’s Telephone: (865) 574-7183; beginning has been mimicked in DOE’s Relativistic Heavy Ion Collider. FAX: (865) 241-6776; ORNL is well positioned to become the world’s leading center in neutron science with the comple- Electronic mail: [email protected] tion of our upgraded High Flux Isotope Reactor (HFIR) in 2003 and the Spallation Neutron Source (SNS) ORNL Review Web address: in 2006. Using neutron scattering at HFIR and elsewhere, ORNL researchers recently found evidence to www.ornl.gov/ORNLReview/ support a leading theory for explaining high-temperature superconductivity. Neutron data from another ORNL neutron source, the Oak Ridge Electron Linear Accelerator, is helping computational astrophysi- cists improve their understanding of nuclear processes by which isotopes are synthesized in stars. The SNS may also be used for this purpose. The Review is printed in the United The SNS will enable advances in characterizing nanoscale materials, such as triblock copolymers. States of America and is also available In other nanoscience efforts at ORNL, we are working toward fabricating nanofluidic lab-on-a-chip from the National Technical Information devices, conducting experiments with them, and predicting fluid behavior in them, as well. We are Service, U.S. Department of Commerce, devising innovative ways to trick nanoscale bits of matter into assembling themselves into useful products, 5285 Port Royal Road, Springfield, VA such as artificial membranes and electronic components. An ORNL team is designing a quantum-dot array 22161 that can be operated at room temperature to carry out innovative computations. As disk drives and transistors are further downsized, today’s materials will eventually reach limits in performance. We are addressing these issues through experimentation and computer simulation. Oak Ridge National Laboratory is DOE’s Office of Basic Energy Sciences has recommended managed by UT-Battelle, LLC, for the construction funding in fiscal-year 2003 for a proposed Center for Department of Energy under contract Nanophase Materials Sciences at the SNS. If Congress approves, the new center will integrate nanoscience research with our unique com- DE-AC05-00OR22725 bination of capabilities in neutron scattering, synthesis of new mate- rials, and theory and simulation, using supercomputers at DOE’s Cen- ter for Computational Sciences at ORNL. ISSN 0048-1262 Among its many uses, the Lab’s IBM supercomputer is being used to simulate a fusion heating and plasma control method involv-

ing radio waves, as well as to model a future fusion device that may Rendering by John Jordan Oak Ridge National Laboratory is a be built here in a few years. This basic research is essential to devel- Artist’s conception of the proposed multiprogram, multipurpose laboratory oping fusion as a future energy source. To help make fossil fuel, Nanophase Materials Science Center. that conducts research in energy nuclear, and geothermal energy sources more economical for pro- production and end-use technologies; ducing power, our geochemists are performing important basic research. biological and environmental science In computer science, we perform basic research to enable computational researchers to get better results faster. Recently, an ORNL algorithm was shown to reduce and predict delays in data delivery over the Internet. and technology; advanced materials In the biological area, ORNL received a 2001 R&D 100 Award for its protein structure prediction synthesis, processing, and tool. ORNL basic research involving mice and their DNA may shed light on why some humans are more characterization; and the physical susceptible than others to getting cancer after exposure to low doses of radiation or environmental chemi- sciences, including neutron-based cals. Research being conducted at ORNL and the University of Southern California suggests that a spinach science and technology. protein might someday restore sight to the legally blind. Basic research has been a pillar of our work. As ORNL gets new facilities this decade for neutron science, computer science, nanoscience, and biology, an even brighter future is in store for fundamental research at the Laboratory. Ð Deputy Director for Science and Technology Ð Associate Laboratory Director for Physical Sciences

Number Two, 2001 1 Basic Research at ORNL ORNL’sORNL’s SearchSearch forfor RareRare IsotopesIsotopes ORNL researchers are taking advantage of HRIBF’s special ability to make neutron- rich beams to allow studies of highly unstable nuclei not found on the earth. These studies may allow them to better understand the nature of nucleonic matter and the origin of elements in the cosmos.

ur earth is home to 81 stable decay. Isotope half- elements, including slightly lives range from less fewer than 300 stable isotopes. than a thousandth of Several other elements (e.g., a second to billions thoriumO and uranium) and more than 130 unstable of years. Short-lived HRIBF has the largest tandem isotopes also exist on the earth, but all of them isotopes cannot be electrostatic accelerator in the eventually decay. That is, the atomic nuclei of these found naturally on world, which is located in isotopes eventually capture or emit electrons, the earth because ORNL’s landmark tower. positrons, or alpha particles, or they undergo spon- they have long since Jim Richmond taneous fission, making the isotopes radioactive. decayed in that our planet was formed about 4 bil- The properties of most rare isotopes are un- Many radioactive isotopes are quite useful in the lion years ago. Yet, thousands of short-lived iso- known and can be only inferred, with considerable diagnosis or treatment of diseases, biological and topes are continually created in the cosmos. Their uncertainty, from theoretical calculations. Never- environmental studies, archeology, national secu- existence may be fleeting, but they play a crucial theless, at several laboratories around the world, in- rity, and energy generation. role in the ongoing formation of the elements in the cluding the Department of Energy’s Holifield Ra- Each unstable isotope is characterized by its universe. In fact, synthesis of heavy elements in- dioactive Ion Beam Facility (HRIBF) at ORNL, it half-life—the time it takes for half of the sample to volves unstable nuclei at every stage. is possible to produce beams of some rapidly de-

ORNL Theorists and the Nuclear Shell Model Bruce French (University of Rochester) and by Cheuk-Yin Wong, Joe McGrory, and Edith Halbert (all from ORNL), among others—was seminal in providing In 1997 ORNL physicist David Dean received the Presidential Early Career an understanding of nuclei with up to 30 nucleons (neutrons and protons). Award for Scientists and Engineers for his leadership role in developing the With advancing computational technology, a second generation of shell model Quantum Shell Model Monte Carlo technique for solving the nuclear many- codes was developed in the late 1970s and 1980s, using ideas of the U.K. body problem. He performed this work with Steve Koonin and collaborators theorist John Whitehead. at the California Institute of Technology and has continued developing the Currently, a Joint Institute for Heavy Ion Research postdoctoral fellow, Andrius technique during his time at ORNL. The techniques are particularly well suited Juodagalvis (from Lithuania), is working with Dean to develop a parallel imple- to investigating thermal properties of nuclei such as phase transitions and mentation of the Whitehead scheme on modern parallel supercomputers. critical temperature phenomena that can be induced by bombarding a nucleus with projectile ions and other particles. Quantum Monte Carlo methods (pursued by Dean and collaborators) solve this same problem with path-integral techniques and are used to investigate The idea of independent neutrons and protons that make up a nucleus both the ground-state and thermal properties of nuclei. These methods are moving in a common potential, an attractive force created by the self-inter- based on a quantum Monte Carlo integration technique. This technique al- action of the nucleons that binds them into a nucleus, is central to the lows for the integration of very large, multidimensional integrals that naturally description of atoms, metals, and hadrons. It is also realized in nuclei and arise in the nuclear structure problem. The methods are also applied to other was first put on a firm theoretical basis in 1949 by Maria Goeppert-Mayer areas of science where the particles involved are quantum-mechanically strongly and Hans Jensen, who later shared a Nobel Prize for their work. correlated. The differing approaches are complementary and, thus, allow a In contrast to other quantal systems cited above, the residual interaction more complete picture of the nucleus to emerge. between the protons and neutrons in a nucleus is strong and severely per- Shell model diagonalization has recently been used to understand the struc- turbs the naive picture of the nucleus as a collection of independent par- ture of nickel-56 (56Ni). This nucleus has 28 neutrons and 28 protons, so it is a ticles. This interaction mixes together many different independent-particle doubly magic nucleus. In 2000 an ORNL/UT experimental group led by Cyrus configurations, generating phenomena such as superfluidity (pairing), de- Baktash—along with Dean, Witek Nazarewicz, and collaborators—published formation (non-spherical ground-state shapes), and collective rotations and a scientific paper on 56Ni entitled “Rotational Bands in the Doubly Magic vibrations (where many neutrons and protons are involved in the motion) Nucleus 56Ni” in the 82nd volume of Physical Review Letters (1999). The paper that are common properties of nuclei. won a Technical Achievement Award from UT-Battelle in 2000. In the 1960s scientists developed one of the first shell model codes to solve According to Dean, the nuclear shell model indicated both a spherical ground- the nuclear structure problem by diagonalization. This work—carried out by state band and a highly deformed rotational band at higher excitation energies,

2 Oak Ridge National Laboratory REVIEW 2 Oak Ridge National Laboratory REVIEW Basic Research at ORNL

The binding energy that results from the combina- tion of these forces determines the stability of nu- clei—that is, which combinations of protons and neu- trons are stable, which are not, and how unstable nu- clei decay. One of the greatest challenges in nuclear physics is to understand the boundaries of nuclear existence, which are commonly depicted on a nuclear chart. (See nuclear chart and its caption on p. 4.) One of the most recognized nuclear phys- ics theorists in the United States is Witek Nazarewicz, a University of Tennessee physics professor and the deputy director for science at HRIBF. According to Nazarewicz, neutron-rich nuclei pose many fascinating questions. They have a “neutron skin” consisting of numerous neutrons with a greater radius than that of bound neutrons deep within the nucleus. The question of how many neutrons are too many for a given isotope is one of great interest to nuclear physicists. They talk about the “neutron drip line” beyond which the nucleus cannot exist as a bound system. If one neu- tron too many is added to a nucleus, a neutron will “drip out.” It’s similar to what happens when water is added with a medicine dropper to a full cup of water; one drop will cause the water to spill

Illustration enhanced by LeJean Hardin The Holifield Radioactive Ion Beam Facility (HRIBF) is an international user facility that produces beams of over, or drip out of, the cup. short-lived, unstable nuclei for research in nuclear structure physics and nuclear astrophysics. A “Our challenge is to find out where the drip production beam (e.g., hydrogen nuclei) from the Oak Ridge Isochronous Cyclotron (ORIC) strikes target line is for each neutron-rich nuclide,” says material. Reaction products diffuse out of the target and into the positive ion source where they are ionized. Nazarewicz. “Knowing the drip line will help us The positively charged ions may be several different isotopes and atomic species, so they are separated in better understand how well neutrons are bound to the mass/isotope separator. The ions are then converted to negatively charged ions in the charge exchange protons in a nuclear cluster—that is, the limits of cell so they can be accelerated . The mass separator filters out the desired isotope for the radioactive ion beam, which is accelerated by the 25-MV tandem accelerator in the Holifield tower for use in experiments. nuclear existence for a given element. Helium-5 doesn’t exist. If you add a neutron to helium-4, it caying, rare-isotope nuclei that can be used to study are composed of protons and neutrons, together flies away, but you can make helium-8. We know properties of nuclei that are difficult to access. called nucleons. The attractive force between nucle- the limits of nuclear existence up to oxygen. It is NATURE OF THE NUCLEUS ons is responsible for holding them together to form possible to make oxygen-24, but not oxygen-25 a nucleus. However, this force is to some extent or oxygen-26. Many more experiments are needed The nucleus is the core of the atom, contain- counterbalanced by the electrostatic repulsion be- to find the limits of nuclear existence for heavier ing more than 99.9% of the atom’s mass. Nuclei tween the protons, which are positively charged. elements. It won’t be easy.” and experimental and theoretical results agreed. At the time, these calculations were among the largest performed using the shell model diagonalization approach. HRIBF offers physicists a unique capability for understanding both nuclear struc- ture and certain astrophysical phenomena, using similar experimental tech- niques (see main story and other sidebar). “We are using our quantum Monte Carlo techniques developed for nuclear structure studies to understand the microphysics of supernova explosions and other astrophysical phenomena,” Dean says. For example, electron capture on iron-group nuclei plays a key role in deter- mining energy release and subsequent element production in supernova ex- plosions. Understanding the underlying nuclear structure turns out to be quite important in obtaining reliable electron capture rates. “Through electron capture by a nucleus, a proton can become a neutron, producing a more neutron-rich nucleus,” Dean explains. “What happens is that an electron is absorbed onto a nucleus where it reacts with a proton to produce a neutron and neutrino. Emitted neutrinos can drive the supernova

explosion and aid in the synthesis of elements. With our current shell-model Illustration enhanced by LeJean Hardin technologies, we are modeling various aspects of these nuclear reactions.” Understanding the nucleus and its interactions with other particles requires Dean and his colleagues run the Quantum Shell Model Monte Carlo codes at significant interplay between experimental and theoretical efforts. Shown are results of Shell Model Monte Carlo calculations of Gamow-Teller the Department of Energy’s National Energy Research Supercomputing Center excitations within various iron-group nuclei compared to experiment (a). in Berkeley, California, and on the IBM supercomputer at DOE’s Center for Low-energy electrons cause this excitation through their weak interaction Computational Sciences at ORNL. They recently received funding from DOE’s with nucleons in the nucleus. The quality of these results required a Scientific Discovery through Advanced Computing (SciDAC) program to con- significant investment in computational algorithm development and tinue to develop the new parallel shell model diagonalization codes and quan- computer time. They have influenced our understanding of element tum Monte Carlo algorithms. production in supernovae. The bright dot at the top of the galaxy M98 is the supernova SN1997bu (b). Many of the tools of nuclear theory are common These researchers have contributed to the theoretical understanding of the to other quantum-mechanical, many-body systems (clusters, quantum dots, structure of the atomic nucleus and the microphysics of exploding stars. atoms) in which the interacting particles are highly correlated (c).

Number Two, 2001 3 Number Two, 2001 3 Basic Research at ORNL

Are the ba- Map of bound nuclear systems as a function of the proton sic properties of number Z (vertical axis) and the neutron number N very neutron-rich (horizontal axis). This nuclear landscape forms nuclei much dif- the territory of radioactive nuclear beam ferent from those physics. The black squares show of nuclei closer to stable, or nonradioactive, nuclei, which form the the valley of stabil- “valley of stability.” ity that have been

Illustration enhanced by LeJean Hardin studied? Accord- Illustration enhanced by LeJean Hardin ing to Nazarewicz and his collabora- tors, who have performed some of the most impor- tant calculations in The yellow color indicates nuclei that have been this field, the an- produced in laboratories and that swer is, “Yes.” live a short time. Adding either protons Nuclei close to the or neutrons (called nucleons) to a The cornerstone of nuclear structure for over half drip line provide nucleus can move it away from the valley a century has been the shell model, in which each us with a new form of stability, allowing it to reach the “drip nucleon (neutron or proton) is assumed to move of nuclear mat- line,” where nuclear binding ends because the forces in average potential generated by its interactions ter—something between neutrons and protons are no longer strong enough to with all of the other nucleons. This potential leads hold these particles together. The nuclei beyond the drip lines are unbound to to the prediction that the quantum levels in a that resembles a dilute neutron gas nucleon emission. Many thousands of exotic radioactive nuclei with very small nucleus form shells within which nucleons reside. or very large neutron-proton (N/Z) ratios are yet to be explored. In the (Z,N) This picture of nucleon motion explains a host of with many new landscape, they form the “terra incognita” indicated in green. The lines of phenomena, such as the existence of particularly properties. One of astrophysics r and rp processes, which are responsible for the production of stable “magic” nuclei corresponding to the most tantaliz- completely filled shells (corresponding to particle heavy elements in stars, are indicated. The red lines show the magic numbers numbers 2, 8, 20, 28, 50, 82, and 126). The left- ing predictions is known around the valley of stability. However, because the structure of nuclei is hand-side diagram shows the shell structure that the shell-like expected to change significantly as drip lines are approached, it is not known characteristic of nuclei close to the valley of structure of atomic how the nuclear shell structure evolves at the extreme N/Z ratios. The doubly stability. The right-hand-side diagram shows nuclei, introduced magic radioactive nuclei—neutron-poor tin-100 (50 protons and 50 neutrons) schematically the shell structure predicted in drip- more than 50 years and neutron-rich tin-132 (50 protons and 82 neutrons)—are indicated by the red line nuclei, which corresponds to a more uniform ago by Maria dots. Research at HRIBF addresses these unknowns. distribution of energy levels and the quenching of magic gaps. Radioactive ion beam facilities such Goeppert-Mayer as HRIBF offer unparalleled access to exotic and Hans Jensen, may alter dramatically as we the Physics Division, led experiments at the nuclei where such predictions can be tested. approach the neutron drip line. That is, the magic Holifield Heavy Ion Research Facility in which numbers* that by and large determine nuclear uranium targets were bombarded with protons Nazarewicz cites the difficulty of finding properties as we know them may be washed out and alpha particles. Through these early Coulomb the limit of nuclear existence for polonium, which or rearranged. (See sidebar on nuclear shell struc- excitation experiments, they inferred that the ura- has several neutron-rich isotopes (e.g., polonium- ture on pp. 2-3.) However, these predictions are nium-238 nucleus is shaped like a diamond. 210 has 84 protons and 126 neutrons and has a based on experimental information obtained from Nearly 20 years later, ORNL researchers began lifetime of 138 days). “Polonium-218 was stud- nuclei that are not too far from stability. They need producing neutron-rich radioactive ion beams by ied by Ernest Rutherford back in 1904,” he says. to be validated by studying unstable nuclei that bombarding a uranium-238 (238U) target with “It took 94 years for scientists to find a way to add are located much farther out. Understanding of 40-million-electron-volt (MeV) protons from the one neutron to polonium-218 to get polonium-219 these far-from-stability nuclei is also the key to Oak Ridge Isochronous Cyclotron (ORIC) and and then 2 neutrons to make polonium-220. The understanding nuclear synthesis in the cosmos and then extracting the fission-fragment nuclei out work was done in 1998 in Darmstadt, Germany.” origin of the elements. of the ion source to make beams. (See sidebar on These experiments, how- p. 5 for more details.) This technique is referred ever, are not easy; many of the to as isotope separation on-line (ISOL). most interesting nuclei cannot be “We get more than 100 different RIBs from produced from reactions involv- a three-dimensional matrix of carbon fibers with ing beams of stable nuclei. There- a thin layer of uranium carbide on the fiber sur- fore, several nuclear physics labo- face,” says Jim Beene, director of HRIBF. “These ratories around the world have beams include atomic species ranging from gallium developed, or are developing, to lanthanum. The trick is to capture the fission frag- new capabilities to produce ments diffusing out of the target and to make them beams of unstable nuclei that into an accelerated beam (which may be as weak would extend their reach. HRIBF as 1000 nuclei per second) before they decay.” is the pioneering U.S. facility for producing and accelerating radio- (continued on p. 6) active ion beams (RIBs) for nuclear physics and nuclear as- trophysics studies. * In the nuclear shell model, the constituent nucleons (protons and neutrons) move in nuclear-energy THE BEAM IS THE levels, or shells, that are filled, or closed, when the TARGET number of protons or neutrons equals 2, 8, 20, 28, 50, 82, or 126. These are the “magic numbers”

Curtis Boles In the 1980s, the late because they indicate especially stable nuclei (e.g., David Radford adjusts a gamma-ray detector at HRIBF. Paul Stelson, former director of helium-4 and tin-132).

4 Oak Ridge National Laboratory REVIEW Basic Research at ORNL

Beam Technologies Enable dissociates the molecules and negatively range of fission fragments. More than 132 ionizes 17F efficiently as required for direct radioactive isotopes from 28 elements have HRIBF Experiments acceleration with the tandem accelerator. been extracted from this target so far. The energy of the fission products aids their dif- The principal developer of beam-production “The key to solving the target problems is the fusion from the target to the ion source. There technologies that have brought praise and use of highly permeable fibrous or thin-layered an electron beam plasma ion source is used awards to experimentalists at ORNL’s Holifield composite materials with dimensions chosen to knock off electrons from the fragments, Radioactive Ion Beam Facility (HRIBF) is Gerald so that the species of interest can diffuse from making them positively charged. These posi- Alton, leader of the Advanced Concept R&D the material within its lifetime,” Alton says. “We tive ions are run through a cesium cell where Group in the HRIBF Section of ORNL’s Physics were the first to propose the use of custom- they pick up a pair of electrons each, making Division. He and current group members Yuan dimensioned target materials to optimize re- a negative ion beam of 200 keV. The tandem Liu, Sid Murray, Charles Reed, and Cecil Will- lease from the target. We select refractory accelerator can boost the energy of this beam iams have been the unsung heroes in enabling chemical compounds by performing thermal to 350 or 400 MeV. world-class experiments in nuclear astrophys- and chemical analyses and then format them ics and nuclear structure physics at ORNL. Dan with the dimensions appropriate for the fast These target and ion source developments Bardayan received the Publication of the Year release of short-lived species.” have enabled experiments to be done with, Award from UT-Battelle and the American for example, tellurium-132 at intensity levels Physical Society’s Dissertation of the Year More recently, Alton’s group developed a tar- of up to 2 × 107 particles per second. The fi- Award for 2000 for one of these experiments. get for producing beams of neutron-rich brous targets in combination with the kinetic HRIBF operations staff members, including Dan nuclei for nuclear structure studies (see main ejection ion source produced beam intensities Stracener and Paul Mueller, were vital allies in article). This target consists of a stack of 12 to required to complete pioneering studies of implementing and validating the ideas com- 14 carbon-fiber disks that are 2 millimeters reactions important in understanding element ing out of Alton’s group. thick and 15 millimeters in diameter. These formation in stellar explosions. carbon-fiber disks are coated with ~ 10 microns The Advanced R&D Group’s responsibilities in- of depleted-uranium carbide, using a process “We have tried to move the state of the art in the technologies of both target and ion-source clude selection of target materials, design of developed for this purpose by McDermott design from ‘black art’ toward science,” Alton high-release-efficiency targets and develop- Technology, Inc. ment of efficient sources for their ionization. says. “These developments will have a strong Alton came up with the ideas that led to the When a 40-MeV proton beam from ORIC impact on present and future radioactive ion production of the fluorine-17 (17F) beam used bombards the uranium carbide coating of the beam facilities, based on the isotope separa- by Bardayan, group leader Michael Smith, and disks, the uranium-238 fissions, forming a wide tor on-line technique used at the HRIBF.” their colleagues. Many radioactive beam sci- entists believed that a 17F beam could never be produced at the needed intensity at ORNL, but the developments by Alton’s group helped to produce this “impossible beam” and prove them wrong. To make an 17F beam, Alton had the idea of using highly permeable targets made from thin fibers of refractory oxides to allow fast diffu- sion of 17F ions out of the target and fast trans- port to the ion source before they decay. The oxygen-16 in the oxide target is transmuted to 17F by deuteron beams from the Oak Ridge Isochronous Cyclotron (ORIC). Oxide materials were selected by assessing their chemical equi- librium compositions and determining the high- est temperature to which they could be heated. The best candidates proved to be aluminum,

zirconium, and hafnium oxides (Al2O3, ZrO2, and HfO2). The preferred target material is Al2O3, because the release product is aluminum fluo-

ride. On the other hand, both ZrO2 and HfO2 are more refractory than Al2O3 , meaning that they can stand more beam-deposited heat. The speed of release strongly increases with tem- perature, so a hotter target can be important. Dan Stracener and his HRIBF operations col-

leagues found that a combination of HfO2 + Al2O3 target material gave the best results. The HfO2 is the production target and absorbs the beam energy while the Al O provides the Al 2 3 Gerald Alton’s group developed target materials for producing radioactive ion beams. The for forming the aluminum fluoride molecules. above two scanning electron micrographs show hafnium oxide fibers for producing These molecules are rapidly transported to a fluorine-17 beams. The bottom two micrographs show uncoated carbon fibers (left) and a special ion source conceived by Alton, called carbon-fiber disk coated with depleted-uranium carbide, using a process developed for the kinetic ejection negative ion source, which this purpose by McDermott Technology, Inc.

Number Two, 2001 5 Basic Research at ORNL

HRIBF is the first ISOL facility in the is the weakest beam we have worked with. It has collided with the carbon target. This information United States; it specializes in low-energy nuclear 20,000 particles per second, making it a million allows us to determine how long it took the physics and nuclear astrophysics research. How- times weaker than the stable tellurium-128 beams excited-beam nucleus to decay back to its ground ever, unlike other ISOL facilities, it can acceler- we use for experiments.” state which, in turn, tells us about the nature of the ate rare radioactive ions to energies sufficiently David Radford and his colleagues in the excited state.” high to produce nuclear reactions. Physics Division have used these experimental tools “In addition to being a unique facility now,” “The Holifield facility is the only place in and the neutron-rich beams in several pioneering says Nazarewicz, “the HRIBF will be a bridge to the United States where we can do preliminary experiments to study nuclei close to the “doubly the DOE’s proposed Rare Isotope Accelerator experiments with accelerated ISOL beams,” says magic” tin-132. In the past, these nuclei could be (RIA), which will be built somewhere in the United Beene. “We have developed radioactive ion beams studied only in a limited way via beta decays of States in the next 10 years. Given the compelling of fluorine-17 for nuclear astrophysics by making their parents. The availability of accelerated physics that can be addressed with radioactive ion a novel target and ion source. We can now make neutron-rich beams at the HRIBF opens up many beams, the nuclear physics community in the neutron-rich nuclei by causing a uranium-238 new and exciting possibilities. It is not easy to United States has given construction of this facil- target to fission after bombarding it with protons. make an intense beam of tin-132, but a beam of ity its highest priority.” Uranium-238 is itself neutron rich, having 92 pro- tellurium-134 (134Te) or tellurium-136 (136Te) “Nazarewicz and his collaborators in tons and 146 neutrons, so its fission fragments are nuclei can be produced from the 238U ion source at nuclear physics theory group have played a lead- also rich in neutrons.” the HRIBF. “Because tellurium-134 is only two pro- ing role in predicting the properties of nuclides “Our special niche in this field,” says Cyrus tons away from tin-132,” Baktash says, “it teaches far from stability,” Baktash says. “They have also Baktash, head of the Radioactive Ion Beam Phys- us a great deal about the magic proton number 50.” developed a compelling physics case for RIA.” The ics Section of ORNL’s Physics Division, “is our “David Radford and his colleagues have proposed RIA will make it possible to produce ability to make neutron-rich beams to allow studies used reactions that range from ‘gentle’ Coulomb and study more than 1000 new rare isotopes in the of nuclei far from stability. In addition, ORNL can excitation to more energetic ones—like nucleon laboratory. The primary accelerator at RIA would accelerate beams of neutron-rich nuclei to high transfer and fusion. When energy is deposited in a be capable of delivering intense beams of many energies, whereas at CERN near Geneva, Switzer- nucleus by either bombarding it with a beam or elements from hydrogen to uranium, with beam land, these nuclei can be produced only at low en- scattering it off a target, the nucleus becomes ex- power in excess of 100 kilowatts and beam ener- ergies. At Holifield, we can accelerate these nuclei cited. This excitation can take different forms. For against a target to study nuclear excitations. We are gies per nucleon up to 400 MeV. example, the nucleus can vibrate, rotate, or sim- RIA will combine the two major techniques now in a position to break the barrier to creating ply rearrange the orbital motion of a few of its the next level of neutron-rich nuclides. This opens of rare isotope production: fragmentation or fis- nucleons. This excitation, however, is short-lived. sion of the primary beam with in-flight separation up a new area of experimental investigation.” The nucleus gives up its excess energy by emit- “Besides being the only place where neu- of the reaction products, and target spallation or ting particles or gamma rays.” tron-rich beams can be produced and accelerated fission ISOL, followed by acceleration of the iso- for use in experimental physics studies, ORNL “The Coulomb force, which causes the in- tope of interest. The integration of both produc- now has the best suite of detectors in the world for coming beam to scatter off a target nucleus, such tion techniques into a single, advanced rare- radioactive ion beams,” Baktash continues. as carbon-12, also excites the projectile to a higher isotope research facility will allow use of the full “ORNL staff members have developed sophisti- energy level,” Beene says. “Upon de-excitation, arsenal of experimental techniques being devel- cated gamma-ray detectors and charged-particle the nucleus emits gamma rays. The angles and oped worldwide. In the meantime, HRIBF will detectors that can detect emission of particles and energies of these gamma rays are recorded by our serve as an important facility where new physical radiation following nuclear reactions, even though sensitive gamma-ray detectors. We simultaneously phenomena will be explored using radioactive ion the background radioactivity is high and beam in- detect carbon nuclei knocked out of the target to beams, the necessary teams of researchers will be tensities are 10,000 to a million times weaker than verify that the gamma rays we are detecting are trained, and new research tools and experimental stable beams. For example, a tellurium-136 beam the consequence of excitation in nuclei that have techniques may be developed.

Illustration enhanced by LeJean Hardin

Coulomb excitation of neutron-rich nuclei is shown in this schematic in which a beam of tellurium-132 (132Te) nuclei collides with a target containing carbon-12 (12C) nuclei. Information about the shape of and charge distributions within a nucleus may be obtained from its electromagnetic properties. One of the best techniques for gleaning this information is to force an accelerated nucleus to collide with an appropriate target at energies just below the Coulomb barrier. Here, a 132Te nucleus in a beam is excited to higher energy levels through collision with a 12C nucleus. The excited 132Te nucleus then releases this additional energy by emission of gamma rays. Simultaneous detection of charged particles (the scattered target 12C nuclei) and gamma rays allows experimentalists to determine electromagnetic properties. In pioneering experiments at the HRIBF, this technique was used to investigate several radioactive isotopes of tin and tellurium for the first time.

6 Oak Ridge National Laboratory REVIEW Basic Research at ORNL Neutrons, “Stripes,” and Superconductivity Using neutron scattering at HFIR and elsewhere, ORNL researchers have found evidence to support a leading theory that explains high-temperature superconductivity.

ow electrons behave in high- trons can split in two parts temperature copper oxide with one of the parts carry- superconductors is a mystery, ing the charge and the other but progress is being made in the spin. In this case, the resolvingH that mystery on both theoretical and ex- spins can form supercon- perimental fronts. One of the leading theories pos- ducting pairs without their tulates that electronic matter organizes itself into charges trying to keep them fluctuating regions where the charge (holes) and apart. The stripes provide the the magnetism (spins) are separated in space in one- one-dimensional regions dimensional regions called stripes. These striped that make it possible for the phases were predicted in theories developed by electrons’ spins to separate physicists at the Leiden Institute in the Netherlands, themselves from their DOE’s Brookhaven National Laboratory, and the charges.” University of California at Los Angeles. The neutron scattering Herbert A. Mook and Pencheng Dai, both results on stripes have been of ORNL’s Solid State Division, and colleagues published in three papers in from the University of Washington in Seattle have the prestigious journal found evidence for this stripe theory in experi- Nature. The experiments led ments on high-temperature superconducting by Mook and Dai were per-

materials such as yttrium-barium-copper oxide formed between 1998 and Curtis Boles (YBCO). The results were obtained in neutron- 2000 at the Rutherford Aerial view of the High Flux Isotope Reactor, where some of the neutron scattering experiments on superconducting Appleton Laboratory’s spal- evidence was obtained to support a leading theory for explaining high- temperature superconductivity. samples prepared by Rodney Hunt of ORNL’s lation neutron source at the Chemical Technology Division. ISIS Facility in the United Kingdom and at A third experiment, also conducted at Superconductivity in standard low- ORNL’s High Flux Isotope Reactor (HFIR). Their HFIR, demonstrated that the spin distribution temperature materials is understood in terms of first experiment at the ISIS spallation neutron observed earlier was really one dimensional in the theory developed by John Bardeen, Leon source showed that the electron spins separated nature. “This was a key issue in the stripes argu- Cooper, and Robert Schrieffer (known as the BCS into distinct regions in the high-temperature, ment,” Mook says. “We initially submitted the theory), who found a way to make superconduc- superconducting material. paper to Nature as a letter, but the head physical tivity work by pairing up electrons of opposite “In our experiment at ISIS,” Mook says, sciences editor wanted it expanded into a full article.” spins. Normally, electrons bump into each other, “we determined the spatial distribution of the It will probably be some time before a con- impeding conductivity. However, the pairs glide electrons’ spins, based on the neutron scattering sensus is reached on the mechanism behind high- through the superconductor like couples waltz- pattern.” These measurements provided the first temperature superconductivity. However, because ing across a ballroom dance floor, completely direct evidence that striped phases occur in the of the neutron scattering experiments, the “stripe” unhindered by the other electron couples on the YBCO superconductors. theory is regarded as one of the leading theories for floor. The distance between the electrons in each A second set of experiments at the HFIR explaining high-temperature superconductivity. pair is called the superconducting coherence showed that the charge length. Both electrons in each pair have a nega- distribution was also con- tive charge so that they repel each other, but the sistent with stripes. “At HFIR coherence length is long in the standard materi- we saw how the charges were als, so the electron partners are quite far apart distributed in crystals by mea- and, in fact, have many other electron pairs danc- suring their effect on the lat- ing between them. tice vibrations, which are In the copper oxide superconductors, how- called phonons,” Mook says. ever, the coherence length is very short, so the “The changes in the phonons paired electrons are dancing close together. This allowed us to see a periodic- makes it very hard to find a pairing interaction ity in the pattern of these vi- Arrangement of the spins and charges in a striped phase. The circles represent the copper atoms in the copper-oxygen planes of the strong enough to provide the glue to keep the brations, which demonstrated superconductor. The spin stripes are represented by the arrows while pairs together. that the charge part matched the open circles are the hole stripes. The shaded circle shows that the “This is where stripes come in,” Mook up with the spin part of a holes are not uniformly distributed on the hole stripe and may move says. “Rather amazingly in one dimension, elec- striped phase.” along the stripe.

Number Two, 2001 7 Basic Research at ORNL ORNL’s Neutron Sources and Nuclear Astrophysics Neutron data from ORELA and the SNS will help computational astrophysicists improve their modeling of the nuclear processes by which isotopes are synthesized in stars.

he Oak Ridge Electron Linear inner regions of the star out to its atmosphere, In the red giant stardust model, micro- Accelerator (ORELA) at where they are visible to astronomical observa- scopic grains of refractory materials such as sili- ORNL is the only U.S. facility tions. In fact, the observation of the radioactive con carbide (SiC) form in the cooler outer re- T that can provide much of the element technetium in a red giant star’s atmo- gions of red giant stars, trapping within them neutron data needed by computational astrophysi- sphere provided the first direct proof that nucleo- trace amounts of the heavy elements made in the cists modeling the nuclear processes by which synthesis occurs in stars.” s process. Strong stellar winds from red giants isotopes are synthesized in red giant stars and Intimately linked to the mixing process in disperse the products of the s process through- supernovae. Measurements from ORELA are a red giant (which our sun will become in a few out the galaxy. Some of these grains have reached being used to improve computer models designed billion years) is the synthesis of elements heavier the earth aboard meteorites. The relative abun- to enhance our understanding of the life and death than iron through a chain of nuclear reactions dance of isotopes of many of the trace elements of stars, the chemical evolution of our galaxy, known as the slow neutron capture, or s, process. trapped within these stardust grains can be mea- and the formation of our solar system. In the s process, free neutrons are produced as a sured with exquisite precision by analyzing the Almost all stable isotopes of the elements by-product of nuclear reactions that generate the content of stardust recovered from meteorites. found on the earth and even some radioactive iso- energy that powers the star. Heavier nuclei are These detailed, isotopic signatures provide a rich topes not naturally present on our planet were synthesized when lighter nuclei capture one of set of observational data with which to test as- formed in stars as they burned fuel or exploded. these free neutrons. However, a typical stable trophysical models of stars and of the chemical “Almost all elements heavier than iron were made nucleus can capture only a few neutrons before evolution of our galaxy. in stellar environments where neutrons play an it becomes unstable. Through a process known The first, or “classical,” model of the important, if not dominant, role,” says Paul as beta decay (whereby one of the neutrons in s process was proposed in 1957. In this simpli- Koehler, a nuclear physicist in ORNL’s Physics the nucleus disintegrates into a proton, an elec- fied model, it is assumed that the temperature, Division. “Roughly half the isotopes of elements tron, and an antineutrino), the nucleus is trans- neutron density, and matter density are constant heavier than iron were synthesized in red giant formed into the next heavier element. Starting during the helium-burning pulses in which the stars. Through a complicated mixing process, from iron “seed” nuclei, this chain of neu- s process was thought to occur. In contrast, the newly synthesized elements were carried from tron captures and beta decays continues all s-process environment in real stars is thought to the way to elements as be very dynamic, with large changes in all these heavy as lead. properties over relatively short periods of time. Nevertheless, the classical model has been very successful in reproducing the observed abun- Illustration enhanceddances by of the s-process isotopes in our solar sys- tem. However, as the neutron capture data Jane Parrott have become more precise, cracks in the classical model have begun to show. In the 1990s, precise new neutron-capture- reaction-rate measure- ments from ORELA and elsewhere showed that the classical model predicted incorrect abundances of several key s-process isotopes of barium, neodymium, and tin. Precise, new neutron-capture data have allowed astrophysicists to move beyond the sim- plistic classical model and test more realistic models of the s process that are closely tied to models of actual stars. The first precise test of the new stellar models of the s process, includ- A cutaway drawing of the Oak Ridge Electron Linear Accelerator facility showing the accelerator (center) for bombarding a tantalum target with electrons, causing the production of neutrons that travel ing the red giant stardust model, was made pos- through various beam lines for experiments. Researchers use these neutrons to bombard various targets sible recently by work at ORELA. There, ORNL to determine the probability that these neutrons will be absorbed by nuclei in the targets. physicists made the first precise measurements

8 Oak Ridge National Laboratory REVIEW Basic Research at ORNL

of the neutron-capture cross sections for two iso- neutron-capture cross sections from ORELA and ticle, we will be able to supply the best constraints topes of neodymium, 142Nd and 144Nd. A cross other facilities, as well as new stardust and other for nuclear statistical models.” section is a measure of the probability that a astronomical data, are helping theorists unravel the The neutron alpha measurements are still nuclear reaction occurs, such as the capture of a mysteries of complicated mixing processes.” very difficult because of a background known as neutron by a nucleus. the gamma flash. At ORELA neutrons are pro- Stellar s-process model calculations made duced in secondary reactions that occur when an using previously accepted cross sections for these electron beam strikes a tantalum target. The elec- isotopes were in serious disagreement with trons first produce a very intense beam of pho- stardust data. The problem was that these calcu- tons that can blind detectors. lations relied on neutron cross-section measure- According to Koehler, “With the help of ments made over too limited a range of energies, seed money and scientists from Russia and so they had to be extrapolated down to the lower Poland, we were able to scale up a detector that temperatures predicted by the new stellar mod- had been pioneered at ORELA—called a com- els. The new ORELA measurements, which were pensated ionization chamber—that makes it pos- made with an improved apparatus and over the sible to count the alpha particles in the presence entire energy range needed by the new stellar of the very intense gamma-flash background. models, showed that the old data were in error. Surprisingly, we found that our first proof-of- With the new ORELA data, the agreement be- principle neutron-alpha data, on samarium-147 tween the stellar model predictions and the mea- and neodymium-143, are in reasonably good surements of neodymium isotopic abundances in agreement with the older model from Caltech, stardust are excellent. Planetary nebulae from the death of a red giant star. but that the rates predicted by two very recent Five more measurements of this type have The strong stellar winds responsible for planetary models are roughly a factor of three different from been made at ORELA through a collaboration nebulae help to disperse elements synthesized (via our data, but in opposite directions.” led by Koehler, consisting of scientists from neutron-capture reactions during the s process) by red The high neutron flux at DOE’s Spalla- ORNL’s Physics and Computational Physics and giant stars throughout the galaxy. tion Neutron Source (SNS), which will be oper- Engineering divisions and scientists from ating in 2006 at ORNL, will allow astrophysics Koehler and his colleagues recently re- Denison University and Lawrence Livermore measurements to be made using samples 10,000 ceived ORNL seed money to study the forma- National Laboratory. In four out of five cases times smaller than those currently studied at tion of proton-rich isotopes of elements heavier studied so far, Koehler says the precise ORELA ORELA. The SNS should be especially useful than iron during the p process. Some scientists data have demonstrated that extrapolations from for measurements of neutron-capture rates of believe that proton-rich isotopes, such as previous measurements are in error by 2 to 3 radioactive isotopes, because only tiny amounts molybdenum-92 and ruthenium-96, are formed times the estimated uncertainties. will be needed. The SNS could also be used to when high-energy photons produced late in the “We are measuring neutron cross study very rare stable isotopes that are so expen- sections of isotopes thought to be produced solely lives of massive stars or in supernova explosions sive that only very small isotopically separated by the s process (s-only isotopes) because they knock out neutrons from nuclei. High-energy samples are affordable for research. are the most important calibration points for test- photons can knock out not only neutrons but also “The excellent time-of-flight resolution ing the stellar models,” says Koehler. “If an alpha particles, in gamma alpha reactions. at ORELA makes it the only facility in the element’s probability of capturing a neutron is “Cross sections for gamma alpha reactions United States capable of measuring small, very small, it is unlikely that it will be transmuted are next to impossible to measure for these resonance-dominated neutron cross sections,” to another element, so its abundance will be high. heavier nuclei, but the nuclear statistical model Koehler says. “ORELA would also be an excel- If its cross section is very large, it more likely should be able to predict them reliably enough lent facility for developing detectors for experi- will be destroyed, so its abundance will be low. for p-process calculations,” Koehler says. “How- ments at the SNS. ORELA and spallation sources There are about 30 s-only isotopes. Cross sections ever, current models do a poor job of reproduc- such as the SNS are complementary facilities, at low energies have been measured for only 4 of ing the few measurements that have been made. both of which are essential to cover the wide these isotopes, using ORELA. Precise neutron- It turns out that by measuring the rate of neutron capture rates for the 26 other s-only isotopes alpha reactions at ORELA, in which each nucleus range of measurements needed for nuclear could also be measured at this neutron source. captures a neutron and then ejects an alpha par- astrophysics.” “This information also helps us better un- derstand how the s process synthesizes elements, which of the heavier elements were formed first, and how the abundance of elements evolved over the lifetime of the galaxy,” Koehler adds. “It al- lows us to predict how well materials in stars are mixed during the convection process, when heat from the star drives fluid flow. Convection and mixing are thought to play an important role in many astrophysical environments, such as in Illustration enhanced by LeJean Hardin supernova explosions. But, because of their rela- tive simplicity and because most of the important nuclear physics information can be measured in A very small portion of recent neutron-capture data from ORELA for isotopes of platinum. The the lab on the earth, red giant stars offer perhaps entire data set spans the range from 0.02 to 500 kilo electron volts (keV). The excellent resolution the best hope for understanding convection and of the ORELA facility makes it possible to discern the many peaks in the cross sections and thus mixing in astrophysical environments. Precise new determine the astrophysical rates for these neutron-capture reactions to high accuracy.

Number Two, 2001 9 0

Basic Research at ORNL

As disk drives and transistors are further Modeling Magnetic downsized, currently used materials will eventually reach Materials for fundamental limits in performance. ORNL researchers are Electronic Devices addressing these issues. igital cameras, computer- disk drives work. Most of these drives read the alloy. A current is passed between the pinned and game-playing stations, laptop stored data using giant magnetoresistive (GMR) free magnetic layers. When the free layer senses computers, and portable de- read heads. a magnetic moment signifying “1,” it stays or vices that record music and “The read head is like a jet flying at su- becomes parallel to the pinned layer; when it D personic speeds a meter above a pasture and play back television programs use magnetic disk senses a “0,” it adopts a position perpendicular drives as small as a fat credit card—taking up much counting the grass blades pointed in two differ- to the pinned layer. In one position, the electri- less space than the shoebox-sized drives in desktop ent directions,” Schulthess says. “Think of each cal resistance is high between the magnetic lay- computers. The number of bits of information (1’s grass blade as a magnetic particle, or region of ers (because the magnetic moments of the free and 0’s) per square inch is doubling every year as magnetization, pointing in one direction—or the and pinned layers are not parallel), so the cur- magnetic disk drives shrink yet store and access opposite direction—like a compass needle. The rent passing through them is low; in the other more data in less time. But in its quest to make disk directions, or magnetic moments, represent bits position, the reverse is true. The shifting strengths drives that are extremely fast and small, the mag- of information, either a 1 or 0.” of an electrical signal as a result of changes in netic recording industry will soon be hitting a wall. The reader is a spin-valve device consist- resistivity allow the read head to copy stored data Currently used materials will eventually reach ing partly of layers of nickel-iron and cobalt to a computer. fundamental limits in performance as disk drives (separated by copper spacers). These layers are are further downsized. What are the problems that will result free to rotate, or spin, in response to an applied from downsizing disk drives made of currently ORNL and the Department of Energy’s magnetic field. The total thickness of these free Brookhaven National Laboratory (BNL) are col- used materials? According to Butler, “As the disk magnetic layers may be only 10 nanometers. laborating with industry to address the fundamen- drive density gets higher, the magnetic particles Based on its orientation, the small magnetic field tal scientific problems confronted by the magnetic representing each bit get smaller. So the read head of each magnetic particle on the disk affects the recording industry. The industrial partners in this must be shrunk to several hundred angstroms. Laboratory Technology Research (LTR) project, electrical resistance of the read head. Because Seagate wants the fixed AFM magnet funded by DOE’s Office of Science, are Seagate to be much thinner, the material structure may Recording Heads, Inc., the world’s largest manu- have to be changed to make it work effectively. facturer of computer hard disk drives; the Almaden “Secondly, the smaller read head must Research Laboratory of IBM, the inventor of the

Jane Parrott Jane have a higher sensitivity to detect smaller bits. magnetic hard disk drive (in 1956) as a direct ac- As the density of the regions of magnetization cess storage device; and Imago Scientific, Inc., a become smaller, the magnetic fields that must startup company in Madison, Wisconsin, that can be sensed to read the data get weaker. We need analyze magnetic materials using its patented local to obtain the largest possible change in resis- electrode atom probe (LEAP) technology. The

Illustration enhanced by by enhanced Illustration tance for relatively small changes in the mag- ORNL researchers involved in this LTR project are netic field. One solution may be to grow Bill Butler, Malcolm Stocks, Mike Miller, and smooth films with the proper interfacial Balazs Ujfalussy, all of the Metals and Ceramics structure to optimize the magnetic per- (M&C) Division; Thomas Schulthess of the formance of the device.” Computer Science and Mathematics Divi- The third problem is that, as bits sion; and Xiaoguang Zhang of the Compu- get smaller, the superparamagnetic limit tational Physics and Engineering Division. is reached. “When magnetic particles be- “Designers of the next generation Schematic diagram of a GMR read sensor used in a disk drive. of magnetic readers and magnetic media The sensor responds to the changing magnetic fields on the disk as come small enough, their magnetism it spins beneath it. The blocks indicate different layers of magnetic are working with badly blurred vision,” may be affected by changes in tempera- and nonmagnetic material, which may be only a few nanometers ture,” Butler says. “These thermal fluc- Butler says. “They are unable to determine in thickness. A resistance change occurs when the magnetic fields the structure of magnetic material at the from the bits on the disk cause a change in the relative orientation tuations can upset a tiny magnetic necessary atomic scale. We plan to use of the magnetic moments in the two magnetic layers. particle’s magnetization, causing loss of atomic probe technology and computa- valuable data. To solve this problem, a tional simulation to better understand the limits In operation, the magnetic moments of one magnetic disk could be made out of a magneti- of currently used magnetic disk drive materials of the “pinned” magnetic layers are held fixed cally hard material, such as neodymium iron and to identify new material structures and depo- by an adjacent magnet, which has an internal boride. But that would create a new problem: sition processes that may overcome these limits.” magnetic field that does not destroy data on the You could not write new information on such a To understand the problems, first it is nec- magnetic media; it is made of an antiferromag- disk without using higher magnetic fields. To essary to understand how today’s small magnetic netic (AFM) material, such as a manganese generate these higher magnetic fields, the write

10 Oak Ridge National Laboratory REVIEW Basic Research at ORNL

metallic multilayers transistors are shrunk to increase a chip’s com- with atomic-scale reso- puting power, the use of silicon dioxide to block lution. Dave Larson, or allow electron flow will limit transistor per- formerly of the M&C formance as an on-off switch, or memory unit, Division and now with for storing a bit (1 or 0). When silicon dioxide Seagate, has already becomes too thin, electrons will leak from it by done atomic probe quantum tunneling, making it useless as a gate studies of a structure electrode material. containing layers of Rodney McKee of the M&C Division and nickel, cobalt, and cop- his colleagues have shown that silicon dioxide can per. His images show be replaced by certain crystalline oxides whose some diffusion of superior electrical properties will allow reduction nickel atoms into the in transistor size without loss of performance; cobalt layer and diffu- more research is being done on this project by sion of cobalt atoms ORNL in collaboration with Motorola and DOE’s into the nickel layer. Such interdiffusion Pacific Northwest Laboratory under a coopera- could make differences tive research and development agreement. Stocks in resistivity impos- and his colleagues are using the IBM supercomputer to model the interfaces between Atom probe image of a magnetic multilayer. The individual atomic planes sible to detect. can be clearly discerned. “Atom probe in- silicon and silicon dioxide and between layers on formation may allow us silicon. These layers consist of strontium head must be made of material that has a higher to compare deposition processes to determine disilicide, strontium oxide, titanium dioxide, and saturation magnetization.” which ones make thin-film structures that work strontium titanate (to make the gate electrode fer- Another LTR objective is to create a struc- best as read and write heads in very small disk roelectric). tured material that has a higher number of un- drives,” Butler says. “We are simulating the electronic structure paired electron spins per unit volume, or a higher Using computer simulation at ORNL, of strontium disilicide and trying to determine saturation (density) of magnetization. Iron has a Stocks and Schulthess recently obtained insights how many layers of strontium oxide are needed magnetic moment of 2, or 2 unpaired spins per about antiferromag- atom; manganese has a moment of 5 and some netism in iron manga- rare earths, 7, but their spins are often in the op- nese (FeMn), an alloy posite rather than the same direction. Using com- used in the fixed AFM putational simulation at the IBM supercomputer magnet of spin-valve at DOE’s Center for Computational Sciences at GMR read heads. ORNL, Butler, Stocks, Schulthess, Ujfalussy, and They studied the Zhang hope to design artificial, layered structures alloy’s non-collinear for smaller read heads. These structures may con- magnetic structure, in tain iron, manganese, and rare-earth thin films which the magnetic that offer a higher magnetization saturation, as moments point at well as magnetic softness and corrosion resis- angles to each other tance. Such structures may also be used for write rather than up or heads for magnetic recording. down. The LTR project also has an objective to “We did calcu- use computational simulation to help enable lations using con- atomic probe field ion microscope tools (such as strained density func- those employed by Mike Miller of the M&C tional theory and spin Division, as well as the LEAP technology devel- dynamics,” Stocks oped by Imago Scientific) to better image says. “Our goal was to understand the alloy’s Detailed magnetic structure of an iron manganese alloy showing small noncollinear antiferro- relaxations of the magnetic moment directions relative to the 3Q structure. magnetic 3Q magnetic structure, which has the lowest known energy level to make it a good insulator,” Stocks says. “Some- for a system of atoms and electrons in a crystal, times the experimental observations do not agree compared with the 1Q and 2Q structures. We pre- with our calculations because we find electrons dicted there is a more relaxed, even lower-energy- diffusing between layers, while the observations level magnetic state that we call 3QR in which R suggest that the interfaces are clean.” The simu- stands for relaxed. In this state, the magnetic lations could aid the experimenters in improving moment orientations, shown as arrows, are their deposition processes to make a more effec- pointed in a slightly different direction, accord- tive thin-film structure as an advanced transistor ing to our simulation. This insight could lead to a gate material. design of improved GMR devices.” When a material’s performance in an elec- Stocks and his colleagues have also been tronic device is about to hit a wall as the device Plot of scattering intensity caused by cobalt involved in another problem in which a material gets very small, some ORNL materials scientists impurities in copper. currently being used in a basic electronic device focus on designing advanced materials to allow the will hit a wall as the device is downsized. As electronic industry to leap to the other side.

Number Two, 2001 11 Basic Research at ORNL

ORNL physicists and electronics experts have In Quest of a Quark: played a major role in developing detectors and ORNL’s Role in the electronic components to search for evidence that the PHENIX beginning of the universe has been mimicked in Particle DOE’s Relativistic Heavy Ion Collider. Courtesy of Brookhaven National Laboratory Detector ust as violent wars have liberated roam freely in a volume much larger than that of $600-million Relativistic Heavy Ion Collider oppressed peoples, scientists are a nucleus, forming an extremely hot soup called (RHIC) at the Department of Energy’s hoping that violent collisions be- a quark-gluon plasma. Such a soup is thought to Brookhaven National Laboratory. RHIC’s first tween gold nuclei racing around a have existed during the latter part of the first 10 gold-gold (Au-Au) collisions were observed on circularJ track in opposite directions will free the microseconds of the Big Bang, the explosion of June 13, 2000. In January 2001, at a conference building blocks of some protons and neutrons that an extremely dense blob of energy that formed at the State University of New York at Stony make up the atomic nucleus. These building our universe. Brook, physicists involved with RHIC announced blocks are quarks, which combine in groups of Terry Awes, Vince Cianciolo, Yuri that Au-Au collisions had produced the densest three to form individual protons or neutrons (both Efremenko, Frank Plasil, Ken Read, Soren matter ever created in a laboratory—the first step of which are called nucleons). Sorensen, Paul Stankus, Glenn Young, and other toward making a quark-gluon plasma. Such a Quarks are held together to form nucle- ORNL physicists have been searching for free particle soup in a very small volume would have ons by particles called gluons, the carriers of the quarks since 1986. That’s when they first conducted a temperature of 2 trillion degrees Kelvin, strong nuclear force. No one has succeeded in experiments using oxygen beams at the Super Pro- 100,000 times hotter than the sun. knocking a quark loose and seeing it in isolation ton Synchrotron at the European Laboratory for The ORNL physicists from the Physics as a free quark. It is believed that if the nucleons Particle Physics (CERN) near Geneva, Switzerland; Division and researchers from ORNL’s Instru- of heavy nuclei are compressed and excited later, also at CERN, they performed experiments mentation and Controls Division are involved in enough by colliding nuclear beams, the quarks with sulfur (1987-1992) and lead (1994-1996). In the 420-person PHENIX collaboration at RHIC. and gluons may be liberated temporarily from 2000 many of the scientists conducting experiments The $100-million PHENIX is one of two large their nucleon prisons. In this process, called at CERN claimed they had observed deconfinement. arrays of particle detectors at RHIC; the other deconfinement, these quarks and gluons would But other scientists called this announcement large detector is STAR. premature. Ac- On April 16, 2001, scientists in the cording to Young, PHENIX collaboration—including Awes, “We felt the an- Cianciolo, Efremenko, Plasil, Read, Sorensen, nouncement was Stankus, and Young—published a paper entitled premature and did “Centrality Dependence of Charged Particle Mul- not address sev- tiplicity in Au-Au Collisions,” in the Volume 86, eral expected sig- Number 16 issue of Physical Review Letters. Ac- nals of de- cording to Young, the paper provided evidence that confinement.” the PHENIX detector works as planned and that Now, the 6000 particles are emitted per collision as pre- ORNL physi- dicted. “Most of these particles are produced by cists and scien- conversion of energy into matter during the colli- tists all over the sion,” Young says. “These particles spraying in all world are pin- directions are the result of smashing the 394 ning their hopes nucleons contained by two gold nuclei together at of freeing quarks a total energy of 25 trillion electron volts.” and gluons on ORNL physicists and engineers have de- gold beams col- veloped two detectors and also electronic compo-

Courtesy of Brookhaven National Laboratory liding at nearly nents for many of the 500,000 particle-detection View of all three PHENIX spectrometer magnets installed in the collision region. the speed of light channels of PHENIX. The goals of this work are The long axis of the photograph spans some 50 feet. One of the central arm (100 billion elec- to sort through the data on a selected subset of the carriages has been retracted; the other can be glimpsed behind the central tron volts per 6000 particles emitted per Au-Au collision and to magnet. The octagonal magnets carry the muon spectrometers. nucleon) at the select the most meaningful collision events that

12 Oak Ridge National Laboratory REVIEW Basic Research at ORNL

are of greatest interest to the physicists. The STAR, plifies the charges in- on the other hand, is a gigantic digital camera, duced by the pulse on which has electronics for detecting and analyzing the wire, signaling the all 6000 particles emitted per collision. presence of a muon. “PHENIX is used to detect all the particles This development in striking it, but the electronics decide which par- low-noise electronics ticle events should be stored in memory and ana- is essential to the lyzed in detail later by a parallel computer,” detector’s success in Young says.“Gold-nuclei collisions occur 1000 finding muons. times a second. About 90% of the time the colli- To detect pho- sions are not interesting but 1 to 5% of the time tons and electrons they are very interesting. emitted by the Au-Au “Of the 6000 subnuclear particles emitted collisions at RHIC, a after each collision event, which lasts only calorimeter was built 10-22 second, about 90% are pions, which origi- for PHENIX by a nally were thought to be the particles that hold the team of researchers nucleus together. Our ORNL team is interested in from ORNL, Russia’s studying mainly photons, electrons, and muons, a Kurchatov Institute in heavier partner of the electron.” Pions are the light- Moscow, and the Uni- est particles to feel the nuclear force and are thus versity of Münster in most easily produced during the collision. Germany. This calo- ORNL researchers have focused on the de- rimeter measures the tector and electronics for two types of PHENIX energy of the photons detectors—a muon identifier and a calorimeter. The and electrons coming first detector detects muons and the second one into PHENIX. It is detects electrons and photons. ORNL also produced made of leaded glass electronics used in four other PHENIX detectors. similar to what is found in cut glass. Muons are 210 times as massive as electrons. Courtesy of Brookhaven National Laboratory First discovered in cosmic rays by Nobel Laureate The glass is 55% lead oxide. Because of the View of the two PHENIX central arm carriages with their detector and electronics Carl Anderson, a muon is unstable; it decays in 2.2 packages installed. The far one is in the process of being rolled into position. The microseconds into an electron plus two neutrinos index of refraction of cylindrical beam pipe threading the large central magnet (in green) is some 15 feet (almost massless particles that can penetrate the the glass, when a par- above floor level and carries the two counter-rotating beams of high-energy earth). Of the 6000 particles emitted in a collision ticle enters it, Ceren- nuclei, be they protons, gold nuclei, or intermediate species. between two gold nuclei, only about 10 are muons. kov light is emitted Even rarer is a muon pair (a positive muon paired and picked up by one of 25,000 photomultiplier be saved for later analysis. Similarly, if the 7000 with a negative muon), but if one is detected, it will tubes. These tubes amplify the light flash and con- elements of the muon identifier spot the pattern be of interest to physicists. vert it to 6 billion electrons, which form a brief (15 characteristic of a muon passing through, the flag This muon identifier is designed to spot a nanoseconds) pulse of electrical current caught by is again raised. the electronics. The pattern of the Cerenkov light muon by screening out other particles, including “We keep up to 10% of the raw informa- and its energy and time of arrival indicate whether pions, that cannot cleanly penetrate the detector’s tion on collision events that comes through,” says the particle is an electron or photon on the one 600 tons of steel walls as well as muons can. Young. “Our electronics take 4 millionths of a sec- hand, or a charged pion, kaon (another type of Instead, pions usually strike iron nuclei in the ond to decide whether to store a particle event in meson), or proton on the other. An ORNL I&C memory for later analysis or to not store it. A glo- steel, creating reactions that stop the pions or team led by Alan Wintenberg and including Mike bal triggered decision is made by electronics built events that destroy the particles. A muon loses Cutshaw, Mike Emery, Shane Frank, Don Hurst, by Iowa State University/DOE’s Ames Laboratory. just a bit of energy as it flies through these walls Gentry Jackson, Mark Musrock, Mike Simpson, If the decision is to keep the 500,000 bytes of in- and strikes proportional counters interleaved with David Smith, and Jim Walker designed custom cir- formation from one collision, the event is digi- steel plates containing an electrically conduct- cuits, chips, firmware, and circuit boards to read tized, sent along an optical fiber to data collection ing wire and a mixture of two gases—carbon di- out data from these photomultipliers. Physics Di- electronics prepared by Columbia University, and oxide (CO ) and isobutane (C H ) The muons 2 4 10 vision staffers providing assistance were Stankus, thence along another fiber to a magnetic tape drive knock electrons loose from the gas mixture, ion- Awes, Efremenko, Plasil, Young, and postdoctoral for storage. Later it is analyzed by a powerful par- izing it. These free electrons are attracted in an scientist Sergei Belikov. allel computer provided by BNL.” avalanche to the electric field of the wire where ORNL also designed and built custom elec- About 35 ORNL researchers and techni- they cause a pulse of detectable current that in- tronics for a novel proportional chamber readout cians and some 20 University of Tennessee gradu- dicates a muon got through. A team consisting with small “pixels,” a project handled by Bill ate students were involved, especially between of Vince Cianciolo and I&C members Bobby Ray Bryan, Usha Jagadish, and Melissa Smith. Alan 1995 and 1998, in designing and testing 2200 Whitus, Tim Gee, Steve Hicks, and Miljko Wintenberg and Shane Frank developed electron- electronic-circuit boards for PHENIX. Other Bobrek created custom electronics boards, logic ics, similar to those for the calorimeter, to read researchers in the electronic development chips, and firmware to detect these pulses and out a gas-filled Cerenkov counter used in PHENIX collaboration are from Iowa State University; record them for later analysis. to tag electrons. Chuck Britton, Tony Moore, and Columbia University; SUNY at Stony Brook; A set of precision proportional chambers, Nance Ericson built a system used to detect the Brookhaven and Los Alamos national laborato- which measure position to 100 microns, is placed tiny (4 femtocoulomb) signals induced on 100- ries; Lund University in Sweden; and the KEK just before the muon identifier and inside a large micron-wide strips of silicon diode, a detector used Laboratory, plus the universities of Tokyo, electromagnet. These chambers, which are used to to count the particles hitting PHENIX. Waseda, Hiroshima, and Nagasaki, all in Japan. measure the muons’ momenta, were built by a con- ORNL has developed “trigger electronics” Overall, some 200 persons worked on the elec- sortium headed by DOE’s Los Alamos National to decide what information to keep. If the 25,000 tronics for PHENIX. Laboratory. Chuck Britton and Mike Emery (and photomultiplier tubes of the calorimeter detect “Without such a diverse collaboration and earlier, Mark Musrock) of the I&C Division devel- an interesting amount of energy, the electronics an amazing collider,” Young says, “we could never oped a preamplifier that senses, adds up, and am- raise a red flag and ask for this collision event to hope to mimic the beginning of the universe.”

Number Two, 2001 13 NewNew HopeHope forfor thethe BlindBlind fromfrom aa SpinachSpinach ProteinProtein Research at ORNL and the University of Southern California suggests that a spinach protein might someday restore sight to the legally blind.

in the eyes of legally blind patients, many of these single photosynthetic reaction centers. This work people can perceive image patterns that mimic was published in 2000 in an issue of the Journal the effects of stimulation by light. Greenbaum of Physical Chemistry B. The measured suggested that it might be possible to use PSI photovoltage values, typically 1 volt or more, are proteins to restore photoreceptor activity. ORNL sufficiently large to trigger a neural response. experiments showed that PSI proteins can cap- “We are proposing the insertion of puri- ture photon energy and generate electric voltages fied PSI reaction centers into retinal cells to de- (about 1 volt). The question is, can these volt- termine whether they will restore photoreceptor ages trigger neural events, allowing the brain to function in persons who have AMD or RP. Once interpret images? we demonstrate this is possible, USC research- Currently, in the United States, degeneration ers will test the technique in the laboratory, and of the retina—the light-sensitive layer of tissue at if feasible, later in humans in clinical trials.” Liposomes are tiny spheres that consist of a the back of the eye—has left 20,000 people totally In recent research, the collaborators showed bilayer membrane and an inner aqueous blind and 500,000 people visually impaired. RP is that PSI reaction centers could be incorporated into compartment. Photosystem I (PSI) reaction an inherited condition of the retina in which specific a liposome, an artificial membrane made of lipids centers have been inserted into the bilayer photoreceptor cells, called rods, degenerate. The loss that mimics the composition of a membrane of a membrane to create PSI-proteoliposomes. The liposomes will be used as delivery vehicles for of function of these rod cells diminishes a person’s living cell. They also demonstrated that the PSI insertion of the PSI reaction centers into retina ability to see in dim light and gradually can reduce can be functional inside a liposome—that is, it pro- membranes, illustrated here. peripheral vision, as well. duces the experimental equivalent of a voltage AMD is a disease that affects the center when light is shone on it. A liposome will likely be used to deliver PSI to a retinal cell. pinach may make Popeye the of vision; people rarely go blind from the dis- ease but may have great difficulty reading, driv- Also, in work recently published in Pho- Sailor Man strong, but a protein ing, and performing other activities that require tochemistry and Photobiology (June 2001), the from spinach may someday fine, sharp, straight-ahead vision. AMD affects collaborators have shown that isolated PSI reac- strengthen the vision of people tion centers can photo-evolve hydrogen, indicat- who canS barely see. Researchers at ORNL and the macula, the center of the retina. When light ing that PSI maintains its voltage-generating the University of Southern California (USC) are is focused onto the macula, millions of cells investigating whether this chlorophyll-contain- change the light into an electrical current for the properties under conditions of current flow. ing protein might be useful in restoring sight to benefit of the neural wiring that tells the brain Greenbaum has long envisioned that his the legally blind. The protein could replace a key, what the eye is seeing. group’s research in photosynthesis could have light-receiving part of the human eye that has lost Using internal funds from the Laboratory important impacts on humans in terms of energy its ability to function. People who suffer from Directed Research and Development Program at production and biomolecular electronics. Now, age-related macular degeneration (AMD) or re- ORNL, Greenbaum, Tanya Kuritz, and James W. he is especially excited that it also could lead to tinitis pigmentosa (RP), diseases that are the lead- Lee, all of CTD; Frank W. Larimer of ORNL’s Life restoration of vision to the blind. ing causes of blindness worldwide, may find hope Sciences Division; Ida Lee and in this research. Barry D. Bruce, both of the Although the neural wiring from eye to University of Tennessee; and brain is intact in patients with these diseases, their Humayun and his team at the eyes lack photoreceptor activity. Eli Greenbaum Doheny Retina Institute at USC and his colleagues in ORNL’s Chemical Technol- are seeking a cure for AMD and ogy Division (CTD) propose replacing these in- RP. “We have assembled an out- active photoreceptors with a spinach protein that standing interdisciplinary team of gives off a small electrical voltage after capturing scientists, vitreo-retinal surgeons, the energy of incoming photons of light. Called ophthalmologists, and biomedical Photosystem I, or PSI (pronounced PS One), the engineers, to attack this important main function of this “photosynthetic reaction problem,” Greenbaum says. center” protein is to perform photosynthesis, us- This project, he adds, is ing the energy of the sun to make plant tissue. based on recent original discov- Greenbaum made this proposal after meet- eries in CTD. “Using the tech- The PSI-proteoliposomes will fuse with deficient neural cells of the eye. If functional PSI reaction centers are inserted into a neural cell, ing with Mark Humayun of the Doheny Retina nique of Kelvin force micros- sodium and potassium ions may be set in motion by the voltage Institute at USC. Humayun and his research team copy, we have performed the first generated by PSI reaction centers when stimulated by light. This ion showed that if retinal tissue is stimulated electri- measurements of voltages in- motion can lead to the propagation of a voltage impulse along a neuron cally using pinhead-sized electrodes implanted duced by photons of light from that transmits information to the brain. Figure by Tanya Kuritz.

14 Oak Ridge National Laboratory REVIEW Basic Research at ORNL Human Susceptibility and Mouse Biology ORNL researchers are studying gene expression in mice to help find out why some persons are more likely than others to get certain diseases, even though their environmental exposures are similar.

hy does one person get treated with a chemical mutagen, Rinchik hopes capillary electrophoresis) can compare DNA from cancer while other people to identify individuals that differ in their cellular the mutant mouse with the draft sequence of the do not, even though all response to such low radiation doses. He and his mouse genome, to identify mutant genes respon- have been exposed to ap- colleagues can then study whether such variation sible for the differential response to radiation. W ORNL Corporate Fellow Liane Russell of proximately the same environment? Does genetic is significant in the animals’ susceptibility to vari- variability play a role in the susceptibility of each ous diseases, including cancer. LSD, in studying the risk of damage to a mouse’s individual’s cells to radiation damage, toxic Using microarrays, they health from parental exposure to radiation or chemi- chemical exposure, and possible cancer induction? are looking at which genes are cals, has found that the developmental stage at which ORNL researchers are conducting stud- overexpressed the parental reproductive cells are exposed is of para- ies with mice, which are genetically similar to or under- mount importance not only to the frequency but humans, to help address these questions about expressed also to the nature of the transmitted genetic dam- human susceptibility. For example, the Depart- ages. Thus, the interval between exposure of an in- ment of Energy wants to know which parts of dividual to a given mutagenic agent and the the human and mouse genomes cause individu- conception of offspring has a major influ-

s ence on the likelihood that genetic dam- als to differ in their responses to radiation. Which le o genes may be overexpressed (overactive in age will be inherited. While different en-

rtis B u vironmental agents had been found to ex- producing proteins) or underexpressed C (turned down or off) in individuals ert their maximum effects at various de- whose DNA is more susceptible to ra- velopmental stages, ranging from stem diation damage than others’? cells to mature sperm or ova, no chemi- Ed Michaud and his colleagues in cal had been discovered to be especially ORNL’s Life Sciences Division (LSD) active during stages critical for chromo- are studying ORNL mouse mutants that ORNL Corporate Fellow Liane Russell, shown here with ORNL Director some pairing and segregation. Recently, are more susceptible to skin cancer. He Bill Madia, has found that the developmental stage at which parental Russell found that the topoisomerase-II is using gene microarrays to identify the reproductive cells are exposed is important to both the frequency and inhibitor “etoposide” not only induces nature of transmitted genetic damages. networks of genes that are altered in mice mutations during that interval but also that are more susceptible to skin cancer com- in exposed versus nonexposed tissues in these alters the frequency with which genes on homolo- pared with normally resistant mice. “We are in- different strains of mice. By looking at gene- gous chromosomes recombine. terested in determining which genes are involved expression ratios, they can identify variants in the LSD’s Bem Culiat is conducting a study response. Future genetic mapping and mutation- in the normal development of the skin and its to determine why some mice are more suscep- finding techniques (such as temperature-gradient function as a barrier to protect our bodies tible than others to DNA damage from toxic from various environmental agents,” chemicals. Some 20 years ago retired bi- Michaud says. “By determining the net- Award-winning Biology Feat ologist Walderico Generoso found evidence suggesting that eggs from some mouse works of genes that specify healthy skin, strains can correct damage in sperm from and how subtle changes in these genes male mice exposed to a toxic chemical, make mice more susceptible to skin can- thereby reducing the percentage of embryo cer following exposure to environmental deaths. By using a combination of genetic, stresses and toxins, we hope to gain a cytological, and gene expression studies in better understanding of the molecular newly fertilized eggs, Culiat hopes to de- mechanisms underlying human susceptibil- termine whether the egg repairs damaged ity to cancer and other diseases of the skin.” DNA (Generoso’s hypothesis) or whether LSD’s Gene Rinchik and his col- the egg’s extracellular coat can screen for leagues are beginning to study differences normal sperm and keep damaged sperm out in the genetic responses of the skin, lym- (an alternative Culiat hypothesizes). Early phoid, and reproductive tissues of mice results have neither ruled out nor substan- from different strains as a result of expo- Curtis Boles tiated either hypothesis. sure to low-level X rays. By studying in Ying Xu and Dong Xu, computational biologists at It thus appears that susceptibility to dis- specific tissues the responses of many ORNL, received an R&D 100 Award from R&D maga- eases or to exposures depends at least partly on zine in 2001 for the product of their basic research genes to low-dose radiation in different —a protein structure prediction tool called PROSPECT. multiple, genetically programmed internal re- strains of mice, or in descendants of mice sponses of the body to external influences.

Number Two, 2001 15 Basic Research at ORNL

the motion of the plasma particles as they orbit around magnetic field lines. Unfortunately, be- cause the orbiting plasma particles move at nearly Modeling a the speed of light, it has been impossible to calcu- late how they will respond to radio waves and how much electric current they will produce.” To address this problem, Fred Jaeger and Lee Berry of FED, working with Ed D’Azevedo of Fusion Plasma ORNL’s Computer Science and Mathematics Division, developed a computer program for the IBM supercomputer of the Department of Energy’s Center for Computational Sciences at ORNL to compute plasma waves across the entire Heating Process cross section of a fusion plasma. The program solves an enormous set of equations, providing the first two-dimensional (2D), high-definition picture of radio waves injected from an antenna into the plasma of a doughnut-shaped tokamak. Using 576 and Stellarator processors at speeds of up to 650 billion operations per second, the program shows that, at certain loca- ORNL researchers are using a supercomputer to simulate tions, the waves shift from a long-wavelength to a short-wavelength structure (mode conversion) and a fusion heating method and future fusion device. become rapidly absorbed by the plasma. The group has recently created a 3D code for this modeling he wave of the future for con- that could lead to a method of fine tuning the injec- trolling fusion energy—the pro- tion of the waves to maximize control of the plasma. cess that powers the sun—may Another area in which FED scientists are include more effective harnessing using supercomputers to advance fusion research of a fusionT device heating process, using electro- is in the analysis of very complex, nonsymmetric magnetic, or radio, waves. An understanding of magnetic systems for plasma containment, called how to get “radio-controlled fusion” may come stellarators. These are shaped like a cruller from calculations done using supercomputers wrapped with twisting magnetic coils. The ORNL at ORNL. supercomputers were used in the analysis and Controlled fusion energy, when achieved, design of a new type of magnetic fusion device could be a favored energy source someday. It called the Quasi-Poloidal Stellarator (QPS). QPS would draw upon an unlimited source of fuel— will use a much smaller plasma current and rely hydrogen isotopes from seawater—and would more heavily on external coils to provide the produce no greenhouse gases that could adversely needed magnetic fields for plasma confinement. affect climate. This device may result in a much smaller and The first key to achieving fusion energy is more economically attractive fusion reactor than existing stellarators and would eliminate the to heat charged particles (nuclei of hydrogen iso- Above: Top view of the Quasi-Poloidal Stellarator topes) to very high temperatures such that the (QPS), which is an optimized, low-aspect-ratio potentially damaging plasma disruptions that electrical repulsion of the nuclei is overcome fusion stellarator device. The plasma shape (colors plague conventional research tokamaks. It is during collisions, allowing nuclear fusion reac- indicate magnetic field strength) and filamentary hoped that QPS will be built at ORNL, using tions to occur. At these temperatures, the elec- magnetic coils (light blue) are shown. Below: DOE funds, starting in 2003. trons are completely stripped from the atomic Side view of the QPS optimized low-aspect-ratio “We are employing a Levenberg- nuclei, yielding an electrically conducting gas stellarator device. Visualizations here and on the Marquardt algorithm on the IBM supercomputer called plasma. The second key is to hold, or con- back cover by Don Spong. to calculate how to modify the plasma shape to fine, the particles and their energy with magnetic optimize energy transport and plasma stability,” fields long enough for many collisions and reac- says Don Spong of FED. “Once we have deter- tions to occur. Such fusion reactions would re- mined the best shape, then we will infer the lease enormous amounts of energy that can be design of external magnetic coils that can be converted to electricity. engineered cost effectively to achieve that shape.” In the quest for controlled fusion, scien- “We need supercomputers to model as many tists have attained several important milestones. as 40 variables that interact with each other to de- They have achieved plasma temperatures as high scribe the plasma,” Batchelor says. “We are twid- as 520 million degrees, more than 20 times the dling 40 knobs at the same time computationally to temperature at the center of the sun. More than get six or more competing physical properties 16 million watts of fusion power have been pro- simultaneously as good as they can be.” duced in the laboratory. The unsolved problem The beauty of the new technique devel- is how to control fusion plasmas to get sustained flow,” says Don Batchelor, head of the Plasma oped to study waves is that it can be extended to fusion reactions. The goal is to prevent the loss Theory Group in ORNL’s Fusion Energy Divi- 3D plasmas such as those in stellarators. These of heat from the plasma center to the edge as a sion (FED). “These waves have even been seen are significantly more complicated in shape than result of irregular fluctuations in plasma veloc- to improve the ability of the applied magnetic the tokamak, the present state of the art for plasma ity and pressure (turbulence) brought on by the field to hold the energetic particles and plasma wave computations. plasma current and other causes. energy inside the device.” With the help of ORNL’s supercomputers “Besides heating the plasma in the way “Radio waves give us the best ‘knob’ for and new funding from DOE’s Scientific Discovery that a microwave oven heats food, experiments precision control of the plasma,” says Mark Carter through Advanced Computation (SciDAC) initia- show that radio waves can drive electric currents of FED. “With radio waves we can control where tive, fusion researchers at ORNL are likely to make through the plasma and force the plasma fluid to the power goes, because these waves resonate with waves in this important energy research field.

16 Oak Ridge National Laboratory REVIEW Schematic of ORNL’s 40-m Small Angle Neutron Scattering (SANS) Facility, which will go on-line in 2003 at the High Flux Isotope Reactor.

hen ORNL’s new neutron materials formed by self-assembly of triblock co- continuous channels wind sources go on-line in the polymers. Participants in the study, which is sup- through a matrix. next few years, researchers ported by internal funding from the Laboratory “Previously we will have valuable tools for Directed Research and Development Program at had used SANS to study exploringW the features of matter as small as a few ORNL, include Tony Habenschuss of ORNL’s diblock copolymers billionths of a meter. That’s one reason why many Chemical and Analytical Sciences Division made of polystyrene and researchers involved in nanoscience and (CASD) and polymer chemist Frank Bates and his polybutadiene,” Wignall nanotechnology are excited by two ongoing ini- colleagues at the University of Minnesota. Triblock says. “In the LDRD tiatives at ORNL. One initiative is the upgrade of copolymers synthesized by Bates have been stud- study we investigated the High Flux Isotope Reactor (HFIR), which will ied by Wignall and Habenschuss using small-angle triblock copolymers of offer in 2002 the world’s highest thermal neutron neutron scattering (SANS), X-ray scattering, polyisoprene, polysty- Core-shell gyroid structure. intensities and in 2003 cold neutron fluxes com- atomic force microscopy, and electron microscopy. rene, and polydimethylsiloxane.” parable to the world’s best. The other initiative is SANS will be especially useful for studying these Self-assembled triblock copolymers could the construction of the Spallation Neutron Source fascinating materials at ORNL when the HFIR create hundreds of new morphologies. Such ma- (SNS), which by 2006 will provide 10 times the upgrade is complete. terials could result in improved power-producing flux of any other pulsed (spallation) neutron source In 2003, the cold neutron flux at HFIR will fuel cells and gas-separating membranes. in the world. be comparable to that of the Institut Laue Langevin “A large fraction of potential users of the “Reactor and spallation neutron sources in Grenoble, France, which is the best research reac- upgraded HFIR and the SNS have expressed in- are well matched to the nanoscale,” states the tor in the world for neutron scattering research. “The terest in studying soft matter, such as proteins and 1999 Nanotechnology Initiative Report of the De- upgrade will boost HFIR’s cold neutron flux and polymers, using SANS, reflectometry, and other partment of Energy’s Office of Basic Energy Sci- detection efficiency,” says Wignall. “Because our instruments,” Wignall says. As noted in the BES ences (BES). The report points out that neutron detector will be two-and-a-half times larger Nanotechnology Initiative Report, “the potential than what we have now, we will be able to make neutrons, for example, “can be used not only to impact of neutron sources for the elucidation of simultaneous measurements over a much wider range study nanoparticles themselves but also to these nanostructures is tremendous.” of angles and perform time-resolved experiments. examine protective coatings that are placed on According to Wignall, “We can expect stud- For example, we will be able to make real-time the nanoparticles to prevent oxidation. Neutrons ies of very small features in soft materials to get a studies of materials as they are processed and ob- can also be used to study dispersants that dis- serve the self-assembly of triblock copolymers.” big boost when ORNL becomes the world’s lead- perse nanoparticles in a solvent or host medium.” Triblock copolymers are made by joining ing center for neutron scattering research in the Because nuclei of hydrogen and deuterium three chemically distinct polymer blocks (large middle of this decade.” (heavy isotope of hydrogen) scatter neutrons dif- molecules), each a linear series of identical ferently, neutron scientists often selectively label monomers (small molecules). organic material with deuterium or use deuterated “Because the three blocks linked together solvents so they can highlight what they want to in a linear arrangement may be thermodynami- see and block out the rest. “Using this method,” cally incompatible, they will try to separate from says George Wignall, a neutron scientist in each other,” Habenschuss says. “But since they ORNL’s Solid State Division (SSD), “we can op- are joined together, they can only form separate timize the contrast between the nanoparticle and domains on a scale of the individual block sizes— coating so we can study the structure and behav- that is, on a nanometer scale for typical block ior of the organic material surrounding the lengths. These separate domains self-assemble into nanoparticle.” complex ordered nanostructures.” The use of organic materials, self- Examples of these structures are rods or assembled organic structures, and organic hybrid spheres of one material regularly arranged in the materials is emerging as a route to producing func- matrix of another. A particularly interesting mor- A new beryllium reflector (to redirect outflowing tional nanoscale structures. Wignall is a principal phology studied at ORNL is the continuous core- neutrons back into the reactor core) was investigator in a study of nanoscale-structured shell gyroid structure in which three-dimensionally recently installed as part of the HFIR upgrade.

Number Two, 2001 17 Basic Research at ORNL Quantum-Dot

Arrays for Illustration by LeJean Hardin

Two-dimensional quantum-dot array showing currents as input channels Iin and output channels Iout. A radiofrequency (RF) field is introduced to modulate the currents, providing the Computation necessary degrees of freedom to perform pattern recognition. An ORNL team is designing a quantum-dot array that will be used to carry out innovative computations at room temperature.

n recent years, considerable progress particles on a DNA template, which may eventu- the class of the observed pattern, to allow appro- has been made in the development ally prove the feasibility of this concept. priate actions to be taken. of advanced sensors, capable of de- Specifically, the ORNL team is designing “Our goal is to demonstrate via a proof- tecting increasingly complex signal a quantum-dot array that can be operated at room of-principle experiment how to produce a I temperature to carry out innovative computations. patterns. Hyperspectral imagers, for instance, that nanoscale information-processing device for a acquire data resolved not only in space but also The construction and operation of this array, with sensor,” says Jacob Barhen, director of ORNL’s spectrally (i.e., at many wavelengths), enable the the help of special algorithms, will constitute the Center for Engineering Science Advanced Re- extraction of unique characteristics, unobtainable world’s first successful use of a nanoscale device search (CESAR) and head of this project. Barhen, by other means. For example, looking at a ball to solve nontrivial computational problems such Yehuda Braiman, Vladimir Protopopescu, and containing blueberries and grapes, one can, with as signal discrimination. The advantage of embed- Nageswara Rao, all from CESAR and ORNL’s the same sensor, not only distinguish each indi- ding such a nanoscale computer into a micro- or Computer Science and Mathematics Division vidual fruit, but also estimate the freshness of each nanosensor is that it avoids the need to send the (CSMD), are developing the methodology and signal from the sensor to a conventional computer or whether it has humidity on its surface. The on- algorithms needed to implement neuromorphic a long “distance” away (both literally and figura- computations, which will allow the quantum-dot going trend toward sensor miniaturization (down tively in terms of scale). Information can be pro- to nanoscale sensors) provides a strong incentive cessed at the site of the sensor. An array of spe- array to learn and retrieve information. to develop computational capabilities for process- cially coated and closely placed gold nanoparticles, The practical goal of the ORNL team is to ing the rich signal information at the sensor’s scale. called quantum dots, may enable the operation of build a device that emulates a neural network. A major step toward developing computational such “smart sensors” at room temperature. Instead of the neurons and connecting synapses “brain” power to speed up the processing of sig- ORNL’s first quantum-dot array, which found in the brain, the nanoscale computer will nal patterns is being taken by a multidisciplinary will straddle gold electrodes, will receive a vec- depend on electrically charged gold quantum dots nanotechnology project at ORNL, supported partly tor of input currents from a sensor and produce a connected by electrons that tunnel between them by internal funding from the Laboratory Directed vector of output currents. It will be “trained” to at different rates. Research and Development Program. Research- classify patterns into categories that character- In a neural network, the learning process ers at the Laboratory are fabricating a nanoscale ize specific classes of properties to be detected modifies the interconnection strengths (synapses) pattern-recognition device, using gold nano- by the sensor. The output currents will indicate between neurons. In a quantum-dot circuit, the

(a) Sketch of a DNA-nanoparticle assembly. Functionalized gold clusters are attached to a DNA chain via peptide bonds. Atomic species are color- coded according to the table on the right. (b) An optimized structure of

the Au38(SCH3)24 cluster, obtained by quantum molecular dynamics computed on the IBM supercomputer at ORNL. See (a) for color coding.

18 Oak Ridge National Laboratory REVIEW Basic Research at ORNL

particles arranged every 10 or 11 DNA bases on a DNA scaffold will conduct electricity in the same way that water drips from a faucet rather than as a steady water flow. If the voltage is high enough, the electrons flow by single-electron tunneling, hopping between the weakly coupled nanoparticles and producing a very nonlinear relationship between the current and the voltage. A range of low voltages could produce no cur- rent, representing a “0,” and higher voltages could produce a current spike that represents a “1” for use in information processing. Employing a functional density code obtained from the IBM Research Division in Zurich, Wells, in collaboration with Wanda Andreoni of IBM, has modeled a bare cluster of 38 gold atoms, calculated their electronic struc- ture (the locations of the electrons in shells and between outer shells of the atoms), and related the size of a cluster to the number of its electron charges. He has also simulated the energy effects of adding or removing an electron. “This information is sufficient to predict

Curtis Boles the electrical capacitance of the nanocluster and to allow us to quantitatively analyze structures Leon Maya (left) makes gold nanoparticles and coats them with alkylthial and carboxyl molecules. Karen Stevenson (right) will attach the coated particles to a DNA template. in nanoscale circuits,” Wells says. “Such infor- mation is difficult to obtain from direct experi- inverse of a generalized capacitance matrix plays vated,” by sulfur-containing molecules attached mental observation.” the role of the synaptic matrix. Unfortunately, to gold molecules that connect into the DNA tem- Wells modeled a passivated 38-atom gold once fabricated, the elements of this capacitance plate. Specifically, the coating is a monolayer of cluster bound with 24 methylthiol groups (sulfur matrix are essentially fixed. Braiman suggested alkylthiol organic molecules terminated with a group plus a methyl group, or SCH3). “I found that exposing the array to an excitatory electro- carboxyl group (containing carbon, oxygen, and the binding with the chemical group causes the magnetic field would result in a modification of hydrogen) that binds to amino groups, attached reorganization or rearrangement of the gold the tunneling rates of single electrons hopping to the DNA at specified locations. (Amino atoms in the cluster, compared to the idealized between the quantum dots. By modifying the groups, which contain nitrogen and hydrogen, case of the unpassivated cluster,” he says. These tunneling rate, the device will produce different used in the coupling to gold, are external to the calculations being done to find new properties in output current patterns for the given input sig- DNA, in contrast to amines involved in the pair- very small features require a large amount of nals into the array. Using this phenomenon of ing of bases between two strands of DNA to make computing capacity—about one-third of the IBM photon-assisted tunneling, Barhen designed a it double stranded.) supercomputer’s nodes. neuromorphic learning algorithm, in which the The material’s fabrication and assembly The early research success in attaching amplitudes and frequencies of the polychromatic are being performed by CESAR’s Leon Maya, a gold nanoparticles to a DNA template has been excitatory field are used as control parameters to chemist in the Chemical and Analytical Sciences submitted for publication. The researchers are achieve maximal discrimination between various Division (he makes the gold nanoparticles and now taking on bigger challenges as they attempt classes of signals. Braiman and Protopopescu are coats them with the alkylthiol and carboxyl mol- to build a world-class nanoscale device to solve ecules), as well as CESAR researchers Karen focusing on issues related to the stability of the global problems. nonlinear dynamics underlying the device’s Stevenson, Muralidharan Govindarajan, and Tho- operation, while Rao is exploring more advanced mas Thundat, all of ORNL’s Life Sciences Divi- architectures for digital signal processing using sion, who attach the coated gold particles to the nanodevices. DNA template. So far, some 20 gold nanoparticles have been attached along a DNA “The microwave field provides degrees of template. The plan is to attach the gold freedom and the ability to change electron path- nanoparticles in a grid on a DNA scaffold, which ways and rates of electrical conduction,” Barhen will be fitted between the electrodes. If neces- says. “In this way, this device will mimic neu- sary, the DNA bases between the gold particles rons in the brain. The pathways that the electrons will be destroyed using ozone or ultraviolet light, take to minimize the discrepancy between a de- leaving the gold particles on a substrate span- sired pattern class signature and one produced ning the space between the electrodes. by the array under excitatory field illumination A gold nanoparticle is a cluster of gold at- will enable the solution of the pattern classifica- oms bound together by mutual attraction. In a clus- tion problem. The array will consist of gold ter, most electrons circulate around the gold nuclei, nanoparticles 1.5 nanometers (nm) in diameter but some hop back and forth between the outer shells that self-assemble by attaching to pre-selected of the gold atoms. Using the IBM supercomputer locations about 3.5 nm apart on a specially engi- at DOE’s Center for Computational Sciences at neered DNA template about 70 nm long. The ORNL, CESAR’s Jack Wells (of CSMD), working nanoscale size of the particles and their regular with David Dean and Mike Strayer (both of the placement in close proximity to one another is Physics Division), is simulating the electronic den- necessary for the array to function as a nanoscale sity of clusters of gold atoms in support of the de- computer at room temperature. sign of the quantum-dot array. Because of gold’s affinity to sulfur, the Understanding electron density in these Atomic force microscope image of a gold gold particles are coated and stabilized, or “passi- clusters is important because the gold nano- nanoparticle attached to DNA.

Number Two, 2001 19 Basic Research at ORNL Carbon Nanotubes and Nanofibers: The Self- Assembly Challenge ORNL researchers are devising innovative methods that will trick nanoscale bits of matter into assembling themselves into useful products, such as artificial membranes, electronic components, and structural materials.

hen Dave Geohegan and his hundreds of microns long. Their tensile strength current by raising the resistance in the channel. In colleagues in ORNL’s is the highest in the world, some 100 times stron- this way, a transistor can operate as an on-off switch Solid State Division (SSD) ger than steel, with only one-sixth the weight. or store a bit of information (1 or 0). This modula- W produce carbon nanotubes These materials may someday be linked together tion of conductivity is possible because the silicon by laser ablation, they get a tangled mass of car- or incorporated as fibers in a polymer composite transistor is doped in a controlled manner with im- bon that looks like black, fluffy felt to the naked to form structural materials for aircraft, space- purities that serve as charge carriers. eye and cooked spaghetti and meatballs in trans- craft, and suspension bridges. ORNL researchers have shown that a car- mission electron microscopy images. By putting Carbon nanotubes also conduct electric- bon nanotube supported on two electrodes on a the material into a solvent and shaking it up us- ity in different ways. In ORNL’s Instrumentation silicon dioxide insulating layer can replace that ing ultrasonic agitation, the carbon blobs disap- and Controls (I&C) Division, Mike Simpson, channel and conduct electricity when a potential pear and the spaghetti-like tubes become less Michael Guillorn, Derek Austin (University of is placed on the silicon base. The conductivity of tangled. Some carbon nanotubes are pulled out Tennessee graduate student), and others have the nanotube is controlled by whether the elec- of solution into a pipette and deposited on a spin- shown that the electronic properties of carbon tric field is turned on or off. Austin measured the ning silica substrate where they naturally stick, nanotubes make them suitable to serve as minia- change in current flow in a carbon nanotube that partly because of static electricity. The problem turized rectifiers (components that allow one-way resulted when the voltage on a silicon gate elec- with this manual technique is that it is slow, and flow of electrons) and replacements for silicon trode was changed. only a small concentration of nanotubes is at- channels in field-effect transistors (FETs). “To get the right number of charge carri- tached to the surface. “How the carbon sheet is rolled up into a ers, a certain concentration of dopants is needed A dream shared by some researchers is to tube affects its electronic properties,” Guillorn says. in a silicon transistor,” Guillorn says. “If the size get nanotubes to assemble themselves into long “Rolled one way, a carbon nanotube is a good elec- of the transistor is reduced considerably, it could rods, somewhat like uncooked spaghetti but far trical conductor and could be used as nanowires; have virtually no dopants or charge carriers. more flexible. Carbon nanotubes, which resemble rolled another way, it is a good semiconductor.” In That’s why a device approaching the nanoscale rolled-up chicken wire because their carbon an FET, electrons flow through a silicon channel may require a semiconducting carbon nanotube atoms are arranged in a hexagonal configuration, unless a gate electrode applies a voltage to a silicon to replace the silicon channel.” are only a few nanometers in diameter and up to dioxide film whose electric field pinches off the Can individual carbon nanotubes a few microns long be inter- connected chemically using molecular handles, incorporated as reinforc- ing fibers in a polymer matrix, or threaded through the helices of polymer molecules (e.g., polyphenylene)? Can a sheet of carbon nanotubes be grown using laser ab- lation? A major challenge in nanoscience is to trick tiny features of matter 50,000 times smaller than the period at the end of this sentence to assemble themselves into useful products, ranging from An atomic force microscope (AFM) image of an ORNL- An AFM image of single-walled carbon nanotubes deposited logic gates for nanoscale produced carbon nanotube bundle containing an impurity on a palladium electrode structure. These nanotubes were transistors, to structural on a gold-palladium (AuPd) electrode structure. produced by laser ablation at ORNL. materials for aircraft, to

20 Oak Ridge National Laboratory REVIEW Basic Research at ORNL

(Opposite page) An AFM image of a circular carbon gas, the source of carbon that grows as crystal- nanotube bundle on a AuPd electrode structure. line fibers on nickel, which catalyzes fiber syn- thesis. One carbon nanofiber grows on each long cables for power transmission and suspen- nickel dot. Plasma-enhanced CVD is used be- sion bridges. cause the plasma creates an electric field perpen- One approach being taken at ORNL is to dicular to the surface, causing the fibers to grow link carbon nanotubes using strands of DNA. In upwards on it. The patterned growth of vertically double-stranded DNA, the chemical bases of ei- aligned carbon nanofibers using PECVD is an ther strand are paired with their complementary example of controlled self-assembly of a mate- bases (A with T, C with G) in the other strand by rial at the nanoscale. hydrogen bonding. Because DNA can be termi- A carpet of closely spaced nanofibers nally modified with primary amines, which can could be used to create an artificial membrane selectively react with carboxylic acids under cer- that allows tiny molecules or small particles to tain reaction conditions, it may be possible to self- diffuse through while keeping out unwanted large assemble carbon nanotubes by first linking them molecules or big particles. Such a creation might to DNA strands. The DNA strands could bring mimic the membrane covering a living cell that the nanotubes together; the DNA could be re- allows signal proteins to enter its pores and acti- moved later by exposing it to high temperatures. vate cellular machinery. Carbon nanofibers could ORNL research showing early indications ORNL researchers have shown that a carbon also be used to make electron probes that char- nanotube supported on two electrodes could that carbon nanotubes can be connected chemi- replace a silicon channel in a field-effect transistor. acterize the electrochemical behavior of living cally using DNA is part of the “Biomolecule- cells and monitor intracellular phenomena. Assisted Self-Assembly of Three-Dimensional acids bond selectively with the terminal amines Guillorn, Simpson, and others are experi- Carbon Nanotube Nanostructures” project at that are added to DNA strands.” menting with using carbon nanofibers as mas- ORNL, which is supported by internal funding Britt notes that ORNL researchers cannot sively parallel electron guns—sources of precise from the Laboratory Directed Research and De- confirm whether they have succeeded in attach- electron beams that could create patterns of cir- velopment (LDRD) Program. The project’s lead ing nanotubes to DNA through the formation of cuits that range from 10 to 100 nm across. If investigator is Mitch Doktycz of ORNL’s Life chemical bonds without using advanced charac- nanocircuits could be produced by electron beam Sciences Division. Other participants are Doug terization tools such as Raman spectroscopy, lithography using computer-controlled, carbon- Lowndes, Vladimir Merkulov, and David Fourier transform infrared spectroscopy, and nanotube-tipped field emitters, the resulting chips Geohegan, all of SSD; Simpson and Guillorn, near-infrared spectroscopy. “We also need to would be closer to the semiconductor industry’s both of the I&C Division; Phil Britt of ORNL’s characterize ORNL-made nanotubes to determine 2004 goal: production chips with circuits 8 times Chemical and Analytical Sciences Division, and the nature of their defects,” Britt says. “We may denser and 16 times faster than chips of the same Anatoli Melechko, Lan Zhang, and Jim Fleming, even find that the defects are good because they size currently being etched by optical lithogra- all of the University of Tennessee. may provide handles for attaching the tubes phy. The electron beam approach is needed to “Our goal is to find ways to enable self- chemically to another material.” reach this goal because the wavelength of an elec- assembly of carbon nanotubes into nanoscale An earlier success in self-assembly at tron is much shorter than that of a photon of light. structures, systems, and devices by interfacing ORNL is the growth of a patterned carpet of car- Recent studies have shown that electrons natural materials with artificial materials,” bon nanofibers on a silicon base. Using plasma- are emitted from carbon nanotubes at relatively Doktycz says. “Self-assembly of nanosystems is enhanced chemical vapor deposition (PECVD), low electric field intensities of a few volts per recognized as one of the grand challenges facing ORNL researchers Merkulov, Anatoli Meleshko, micrometer. Thus, nanotube electron emitters nanotechnology. We believe a solution to many Lowndes, Yayi Wei, and Gyula Eres, all of SSD, should operate at the low voltages needed for of these challenges lies in the use of natural sys- have demonstrated that they can grow vertically nanoscale circuitry. tems such as DNA.” aligned carbon nanofibers in a desired pattern on “We demonstrated that a carbon nanofiber The objectives of this LDRD project are the substrate. They can choose the diameters of inside a well can emit electrons,” Guillorn says. to fabricate nanotubes and ordered arrays of the fibers, their positions, and the spacing be- “When we placed a field on the fiber, an electron nanofibers, attach biomolecules to carbon tween individual fibers. was emitted. The field allows electrons to escape nanotubes and nanofibers, develop nanoscale Fiber sizes and locations are determined from the tip.” characterization techniques, and design and build by depositing nickel dots and lines on a substrate, A carbon nanofiber could be used to make molecular assemblies. Nanoscale systems that using electron beam lithography. The substrate a higher-resolution electron microscope because could result from self-assembly could be is then placed in a furnace containing acetylene it would focus the microscope’s electron beam on a smaller spot. nanoelectronic components and devices, sensors Research is under way to develop less ex- and biosensors, nanomachines, and materials that pensive, more reliable, and energy-efficient flat- mimic the membranes enveloping living cells. panel displays to replace bulky, power-hungry “Before attaching nanotubes to DNA, a computer monitors. Because carbon nanofibers raw mass of nanotubes must first be purified,” are inert, they could be more stable and practical Britt says. “Single-walled carbon nanotubes pro- than metallic (silicon) and inorganic electron duced by laser ablation at ORNL by Dave sources for stimulating the emission of colored Geohegan and Alex Puretzky are digested in ni- light in flat-panel displays. “All types of electron tric acid to oxidize the carbonaceous impurities emitters work under ultrahigh vacuum,” Guillorn to make them soluble in water so they can be says. “But at the higher pressures that may occur washed away. The material is then heated in air in real-world devices, residual gases are ionized, at 550°C to oxidize the remaining carbonaceous and the resulting ions bombard the tip of a metallic impurities. Hydrochloric acid is introduced to or inorganic emitter, sputtering it off so it is less remove the remaining nickel-cobalt catalyst par- effective as an electron emitter. Carbon nanofibers ticles added during laser ablation to induce the “Forest” of randomly placed, vertically aligned resist this type of damage, so they should last growth of carbon nanotubes. During these puri- carbon nanofibers grown by ORNL’s Vladimir much longer in advanced flat-panel displays.” fication steps, carboxylic acids are formed at the Merkulov and Doug Lowndes, using plasma- Carbon nanotubes and nanofibers are in- nanotube ends, giving us a chemical handle for enhanced chemical vapor deposition. Note the nickel- credibly small, but their potential for creating attaching carbon nanotubes to DNA. Carboxylic catalyst nanoparticles at the tips of the nanofibers. advanced materials is huge.

Number Two, 2001 21 Basic Research at ORNL Incredible Shrinking Labs: Weighing a Move to the ORNL researchers are showing they can fabricate nanofluidic lab-on-a-chip devices, do experiments with Nanoscale them, and predict fluid behavior within them, as well. magine a device the size of a deck panel of citizens has named the lab on a chip one In a lab on a chip, several channels, each of cards that could help a physician of the top 23 technologies developed using De- the size of a hair in width and one-fifth of a hair in quickly determine why you are feel partment of Energy funding. depth, connect reservoirs containing chemicals. I ing so sick and tired. The channels and reservoirs are By rapid analysis of the DNA and carved, or etched, in a glass chip proteins in a drop of your blood, a (such as a microscope slide) using “nanofluidic lab on a chip” could microfabrication techniques. Liq- indicate to your doctor a million uids are “pumped” through each to a billion times faster than cur- channel, using electric fields or rent devices whether a bacterium, pressure differences, at a rate of one virus, or a cancerous tumor is mak- millionth of a drop per second. ing you ill. At least, that’s one ap- “In a nanofluidic device,” plication seen by scientists in- says Ramsey, “the typical channel volved in the new field of would be 1000 to 10,000 times nanofluidics: the active transport smaller than a hair. We expect that of material through a channel or the interactions between the ma- conduit whose diameter is less terial in the channel and the than 100 nanometers (nm). A na- channel walls—what we call nometer is a billionth of a meter. solid-liquid interactions—will be Basic research in nano- much more dominant in a fluidics is being conducted by nanofluidic device from those in a ORNL and University of Tennes- microfluidic device. We expect see (UT) researchers, using inter- fluids to change their behavior as nal funding from the Laboratory we shrink the sizes of the channels Directed Research and Develop- through which they flow.” ment (LDRD) Program at ORNL. Above: Scanning electron microscope image of four nanotrenches vertically In the first nanofluidic ex- connecting two microtrenches. The nanotrenches are only 100 x 100 nm2 in The LDRD project is headed by cross section but more than 40 microns long. Below: Perspective of a 900 x 80 periment conducted at ORNL by Mike Ramsey, an ORNL corporate nm2 trench cut in silicon for nanofluidic studies at ORNL. The trench was Ramsey, Steve Jacobson, and fellow in the Laboratory’s Chemi- created using a focused ion beam at Vanderbilt University and imaged with an Chris Culbertson, it was observed cal and Analytical Sciences Division atomic force microscope at ORNL. that electrokinetic transport is re- and the inventor of the lab on a chip. duced as dimensions are scaled The lab on a chip is a down from the micrometer range microfluidic structure first built by to less than 100 nm. These results Ramsey 10 years ago to demon- are the first experimental verifi- strate the separation of chemicals cation of statistical mechanical in very small volumes. Six years theories developed more than 30 ago, this technology was licensed years ago for electrically driven to Caliper Technologies, Inc. fluid transport through small Ramsey and his group in CASD channels.“We fabricated an 80- are conducting considerable re- nanometer channel that is 20 search on developing improved microns wide and 80 nanometers lab-on-a-chip technologies for bio- deep,” Ramsey says. “We forced logical, environmental, forensic, an electrolyte—a sodium and defense applications. The lab tetraboride buffer solution— on a chip has been honored by through it. We confirmed the theory R&D magazine as one of the 40 that the fluid velocity would be re- top innovations it has recognized duced in such a small channel.” since beginning its R&D 100 com- The reduction in fluid petition in 1963. Additionally, a velocity at these small scales is re-

22 Oak Ridge National Laboratory REVIEW Basic Research at ORNL

lated to the fact that the channel dimensions smaller ones as a result of hydrodynamic ef- are similar to the electrical double-layer fects,” Ramsey says. “The bigger molecules dimensions. The electrical double layer is the flowing along at the center of the fluid won’t region of fluid near the solid-liquid interface get as close to the channel wall as the little where positive and negative charges may be ones. Thus, the larger molecules experience, separated so that the solution is not electri- on average, a greater velocity and travel cally neutral at a given location, giving rise to through a molecular-size channel faster. This electrically driven fluid transport. As the is speculation, but we will do experiments to channel approaches the double-layer thick- determine whether this hypothesis is valid.” ness, the “pumping regions” begin to overlap Another experiment the ORNL group and become less efficient in imparting momen- hopes to do is to show that a two-dimensional tum to the fluid. nanoscale channel structure can sequence a “In our LDRD project,” Ramsey says, single strand of DNA by obtaining its electri- “we are trying to show that we have the abil- cal signature. Evidence obtained at Harvard ity to fabricate nanofluidics devices with University and theory done there by Dan nanoscale features, that we can do experi- Branton and colleagues suggest that each of ments with them, and that we can understand four types of DNA bases—adenine (A), cy- their fluid transport characteristics experi- Courtesy of Caliper Technologies, Inc. tosine (C), guanine (G), and thymine (T)—can mentally and theoretically, using computa- These commercialized versions of ORNL’s lab on a chip produce its own distinctive electrical conduc- tional simulation. are products of Caliper Technologies, Inc., and Agilent tivity through a nanoscale channel. “The ability to fabricate fluidic struc- Technologies, Inc. “If we could send a strand of DNA bases tures with dimensions at the molecular scale single file along a nanoscale channel with elec- will allow fundamental studies of fluid transport biomolecules such as proteins or DNA strands trodes at each end,” Ramsey says, “we might be at the smallest possible dimensions. In addition, can be separated by size, based on their mobil- able to interrogate the strand by measuring a cur- practical tools for the analysis of biopolymer mol- ity through the channel. “We believe that the rent. The current variations indicate the order of ecules, such as DNA and proteins, could well re- larger molecules will move faster than the bases in the DNA strand.” sult from nanofluidic studies.” To help guide and interpret the results Ramsey’s group is trying to learn of experiments by Ramsey’s group, Hank how to fabricate nanoscale channels by Cochran of the Chemical Technology shrinking the width and depth of chan- Division and Shengteng Cui of UT are nels etched in glass slides for using computational molecular simu- microfluidic lab-on-a-chip devices. In lation. They are modeling the move- ORNL’s Solid State Division, Dave ments of individual atoms and mol- Zehner, Tony Haynes, and Arthur ecules in fluids confined in nanoscale Baddorf are working with Len channels. They are also simulating the Feldman and his colleagues at behavior and transport of water contain- Vanderbilt University to apply thin- ing salt and DNA or proteins in ultra- film and ion-milling technologies to creating nanoscale channels for small channels (e.g., through a 4-nm nanofluidic devices. pore) subjected to electric fields. In em- Sorting proteins by size and se- ploying these models, they take into ac- quencing DNA a million to a billion count the electric-field and surface times faster than current technologies forces that extend through the liquid are possible, practical applications of tightly confined inside a nanoscale nanofluidic devices. But first, proof- channel. Modeling fluids in nanoscale of-principle experiments must be channels under experimental condi- done. Ramsey’s group plans to do tions is challenging because the num- some of these experiments. ber of atoms or molecules simulated One problem in using a lab on must be larger than is currently feasible a chip to separate DNA molecules and by molecular methods, yet the models proteins by size is that a sieving poly- from continuum fluid dynamics be- mer must be added. When DNA mol- come inaccurate at the nanoscale. Cui ecules and proteins flow through this and Cochran are developing a new chemical sieve, the smaller biopoly- methodology that uses continuum mers move faster than the larger ones, equations, with local fluid properties thus causing the separation. determined from molecular simulation. “But it’s a nuisance to get siev- This new approach should enable an ing polymer into a lab on a chip,” A snapshot from a molecular dynamics simulation of sodium understanding of the results of Ramsey says. “It takes time and adds chloride aqueous solution in a 4-nm cylindrical pore: red (oxygen); nanofluidic experiments and guide the to the cost of making the chip.” green (hydrogen); yellow (sodium); and purple (chloride). Some of design of nanofluidic devices. the sodium and chloride ions are hidden from view. ORNL and UT Ramsey’s group plans to do an researchers are simulating the behavior and transport of water In time, nanofluidics research experiment with a one-dimensional containing salt and DNA or proteins in ultra-small channels (e.g., could lead to very small devices that nanoscale channel to determine if through a 4-nm pore) subjected to electric fields. may have a very large impact.

Number Two, 2001 23 Basic Research at ORNL

Basic research by Basic Geochemical chemists and geochemists could help boost Research Supports the economics of energy Energy Industries production.

deposits, hydrothermal crystal growth and plugging up pipes. Corrosion can result in the failure materials synthesis, and high-temperature of pressurized pipes. Metal-oxide nanoparticles can electrochemical processes. Hot, aqueous spall off from the surface through contact with cir- solutions and mineral deposits can also re- culating water at high pressures and temperatures.” spectively corrode and clog pipes, leading According to Wesolowski, when a metal- to inefficient heat transfer in geothermal oxide surface reacts with water molecules and ac- power systems and reducing the operating companying ions, an electric charge develops on the capacity and potential lifetime of water surface. If the surface charge is positive and if the desalination plants, boilers, power plants, ions in solution are negatively charged, the ions will and nuclear reactor cooling systems. deposit on the metal surface to which they are Chemists and geochemists in attracted, causing the buildup of corrosion products. ORNL’s Chemical and Analytical Sciences “Particles can be adsorbed onto the metal- Division (CASD) and their collaborators oxide surface or can be released from the sur- have carried out pioneering studies of the face, depending on pH, temperature, and the consequences of water interaction with chemical composition and concentrations of ions both natural and industrial materials. These in the water,” Wesolowski says. “When water re- studies range from those that address prob- acts with the oxygen atoms bonded to atoms at lems of water-solid interactions at the mo- the surface, hydrogen and hydroxide ions from lecular level to more coarse-scaled studies the water cause the surface to be charged up. This of natural geological systems. surface charge is the principal driving force for For example, chemists and geo- both the adsorption of contaminant ions in sub- chemists in CASD are engaged in a major surface aquifers and the formation and transport interdisciplinary research project, funded of colloidal particles in steam generator systems. by the Department of Energy’s Office of It is this molecular behavior at surfaces we are Basic Energy Sciences under the Complex focusing on in this project, through the use of and Collective Phenomena initiative, to sophisticated tools such as neutron reflectometry, quantify behavior at the oxide-water standing-wave X rays (with partners at Argonne interface. What happens at the interface National Laboratory), molecular dynamic simu- when water and its contaminants interact lations, and high-temperature pH cells.” This schematic model depicts the electrical double layer (EDL) at the interface between a positively charged metal with metals and minerals is of great In fact, CASD researchers Wesolowski, oxide surface and an aqueous solution containing dissolved interest to the energy industry. Don Palmer, and Pascale Benezeth-Gisquet, in cations and anions. In this illustration, adapted from a “Most of our world—from industrial collaboration with a group of visiting research- drawing by Gordon Brown (Stanford University), negatively materials to minerals in the earth—consists ers, including Mike Machesky (Illinois State charged anions are adsorbed directly onto the positively of metal oxides,” says ORNL’s Dave Water Survey) and Moira Ridley (Texas Tech), charged metal oxide surface. Note the difference in Wesolowski, the project’s principal inves- have pioneered the study of the chemical prop- orientation of water molecules (blue) around anions (green) compared with cations (red). Illustration by LeJean Hardin. tigator. “When metal reacts with water or air, metal ater is the key ingredient oxides form. When water that promotes and facili- and the dissolved ions tates the transfer of mass it contains—such as W and heat from one reser- chlorine, calcium, voir to another. The interaction between water iron, lithium, po- and solid phases, whether those phases are natu- tassium, sodium, ral minerals or metals in pipes, can typically lead strontium, and to the release of metals and the formation of sec- zinc—come in ondary corrosion or alteration phases. The solu- contact with these tions can range from dilute solutions of a single oxides, consider- solute in water to complex systems containing A representative portion of a rutile titanium oxide (TiO2) surface is shown interacting able reaction takes 2+ high concentrations of more than 10 dissolved place at the inter- with a strontium cation (Sr ). In situ X-ray reflectivity is used to measure the atomic- scale positions of ions such as Sr2+, rubidium (Rb+) and bromine (Br- ) in aqueous species. These solutions play a crucial role in such face. Secondary solutions adjacent to the oxide surface. This work involving ORNL researchers is geochemical and technological processes as, for minerals can be conducted in collaboration with scientists at the Advanced Photon Source at DOE’s example, hydrothermal formation of mineral formed there, Argonne National Laboratory. Illustration by LeJean Hardin.

24 Oak Ridge National Laboratory REVIEW Basic Research at ORNL

erties of oxides and other minerals as they exist concluded that increasing pressure sometimes has spear points made of obsidian (volcanic glass). in states from ambient conditions (10 to 50oC) to more impact than increasing temperature on hy- Traditionally, these artifacts have been dated by the extreme temperatures and pressures encoun- drogen-isotope ratio shifts,” Cole says. observing through an optical microscope how far tered in geothermal systems and commercial Also in 1999, ORNL’s Horita and Michael water from air and soil has migrated into the glass. o power plants (350 C, 150 atmospheres). These re- Berndt (University of Minnesota) published a pa- Research using ORNL’s ion microprobe demon- searchers have—for the first timeÐdirectly mea- per in Science magazine that explored the under- strated that the true position of the hydration front sured the sorption of ions on the surfaces of min- ground formation of methane, the earth’s most (penetration of hydrogen atoms in the absorbed erals that form naturally in geological environ- abundant natural gas. Most methane is formed moisture) was not at all coincident with the opti- ments, as well as the corrosion products that form either by the digestion of organic compounds by cal front (refracted light thought to represent the in steam generators and other industrial settings, microorganisms or by the thermal decomposi- hydration front), as was originally assumed. at temperatures exceeding 100oC. tion of organic matter. However, there is also evi- The participants in this study of metal- Armed with experimental diffusion data on wa- oxide surface chemistry at high temperatures in- dence that some “abiogenic” methane is formed ter in glass, which determines the rate of water clude Wesolowski, Palmer, Ariel Chialvo, from inorganic matter in the earth’s crust. migration into the glass, as well as measurements Lawrence Anovitz and Bénézeth-Gisquet, all of In laboratory experiments, Horita and Berndt of water’s true distribution in the glass, a new, CASD; Baohua Gu and Liyuan Liang, both of found that abiogenic methane can be produced more reliable dating method was proposed that ORNL’s Environmental Sciences Division; Bill rapidly from capitalizes on the sensitivity Hamilton of ORNL’s Solid State Division; Peter dissolved CO2 of the ion probe. Cummings of the University of Tennessee at in the pres- Using a one-of- Knoxville (UTK); Jim Kubicki and Serguei Lvov, ence of a natu- a-kind, vibrating-tube both with Penn State University; Paul Fenter and rally occur- densimeter (VTD), Neil Sturchio (ANL); and Machesky and Ridley. ring nickel- CASD’s Jim Blencoe The Geochemistry Group in CASD’s iron alloy un- is investigating the Physical and Materials Chemistry Section has der hydrother- thermophysical made its mark in other areas, as well. One such mal condi- properties area is in the determination of the effects of tem- tions similar perature, pressure, and chemical composition on to those pre- the redistribution of ratios of stable isotopes (rare sent in a mid- over common) of light elements, such as oxygen ocean ridge. 18 16 13 12 ( O/ O), carbon ( C/ C), hydrogen (D/H), ni- “We found 15 14 34 32 trogen ( N/ N), and sulfur ( S/ S) in fluids and that abiogenic rocks. Water (H O) in fluids deep within the earth 2 methane can has a stable isotope signature that can be very different from that of water in a river, lake, or the be formed rap- ocean. This is also true for gases such as carbon idly under re- ducing condi- dioxide (CO2) and methane (CH4).

CASD’s David Cole, Juske Horita, and tions encoun- Curtis Boles, enhanced by LeJean Hardin Lee Riciputi are using sensitive techniques, in- tered in the Lee Riciputi shows an obsidian souvenir depicting a native Mexican god of thunder cluding gas source isotope ratio mass spectrom- earth’s crust and lightning and an obsidian sample that he analyzed using the ion microprobe etry and secondary ionization mass spectrometry and that the ra- (secondary ionization mass spectrometer). The analysis casts doubt on the validity (ion microprobe) to determine shifts in isotopic tio of carbon- of a traditional method for dating glassy artifacts. signatures (isotope redistribution) that occur dur- 13 and carbon- ing the interaction of water with minerals, as a 12 isotopes is not a clear-cut criterion for distinguish- (e.g., density and volume) of volatile species such function of changes in temperature, pressure, and ing biogenic from abiogenic methane,” Horita says. as CO and CH at subsurface conditions. The different concentrations of dissolved salts. Natu- 2 4 “An important implication is that abiogenic meth- VTD measures the density of an individual gas rally occurring isotopes of O, H, C, S, and N pro- or gas mixture (e.g. CO -CH , CO -H O) at un- vide built-in tracers for monitoring the interac- ane could be far more widespread in nature than cur- 2 4 2 2 rently thought, especially in crystalline rocks on land precedented precision and accuracy at a given tion of fluids and solids in both natural and in- o dustrial settings. and on ocean floors, including in gas hydrates.” temperature (up to 500 C) and pressure (up to In 1999, Juske Horita, postdoctoral fellow Isotopic distributions can be determined 100 MPa). This activity is supplying a new body Thomas Driesner (now at Eidgenossische in situ in minerals, using the microbeam capa- of the pressure-volume-temperature data for high Technische Hochschule in Zurich, Switzerland), bility of CASD’s Cameca 4f ion microprobe. A temperatures and high pressures that is crucial and Cole published a paper in Science magazine tiny beam of ions (such as Cs+ or O-) is delivered to the development of equations of state (EOS) that refuted one fundamental assumption of stable to the mineral surface, liberating a cloud of sec- for geologically and industrially important flu- isotope geochemistry. For decades it was believed ondary ions that is then accelerated down the ids. These EOS are used to predict fluid behav- that temperature, not pressure, had a dominant flight tube of a mass spectrometer. This technol- ior in a diverse range of environments, such as effect on the hydrogen-isotopic composition of ogy, expanded and refined by CASD’s Lee oil, gas, and geothermal reservoirs; volcanic sys- hydrous minerals (e.g., brucite, or magnesium hy- Riciputi, allows determination of the concentra- tems; pipelines; the high-temperature treatment droxide) in the presence of water. Theoretical cal- tions and isotopic distributions of most elements of wastes (supercritical oxidation processes); and culations performed by Driesner indicated that in the periodic table, both laterally and vertically CO disposal in geological formations, such as this might not be the case. So, the group set out 2 to investigate experimentally the effect of pres- (depth profiling), for spot sizes on the order of a depleted oil and gas reservoirs and coal beds. The sure on hydrogen-isotope redistribution (changed few microns, in phases such as oxides, sulfides, results Blencoe obtains are directly relevant to D/H ratios) between brucite and pure water at silicates, and glasses. In one application, Riciputi the use of CO2 to displace CH4 from unmineable elevated temperatures. They found that equilib- and Cole, along with Larry Anovitz and Mike coal seams, a value-added technology that is be- rium D/H partitioning between brucite and wa- Elam of UTK, debunked the long-standing in- ing seriously considered by DOE for both pro- ter systematically increases by 1.24% as pres- dustry practice used to date ancient glassy arti- ducing an energy-rich gas and sequestering a sure increases from 15 to 800 MPa at 380°C. “We facts, such as prehistoric arrowheads, knives, and greenhouse gas.

Number Two, 2001 25 Basic Research at ORNL Fermi Award Winner Opened New

Sheldon Datz was chair of the Local Organizing Committee of the International Conference on Photonic, Fields In Electronic, and Atomic Collisions (ICPEAC), held July 18-24, 2001, in Santa Fe, New Mexico. ORNL and Los Alamos National Laboratory physicists and support personnel organized and

Curtis Boles ran the conference. Atomic Sheldon Datz’s seminal research in colliding beam chemistry and ion channeling helped him win two prestigious prizes. Physics Editor’s note: This article is based on a June efforts to use this difficult technique, Datz in- theoretically predicted. The cause? Channeling.” 21, 2001, interview with Sheldon Datz, who died vented a new method, “differential surface ion- Upon his return, he immediately joined a small August 15, 2001. ization,” in which two thin, heated wires—one meeting at Chalk River Laboratory in Canada heldon Datz, ORNL’s recent of tungsten and one of platinum—could be ro- where others reported effects that could be ex- winner of the Department of tated in the reaction plane and electrically record plained by channeling. Energy’s Enrico Fermi Award the elastic and reactive scattered product. The “This remarkable effect had actually and a senior corporate fellow, method greatly simplified the experiment. been ‘discovered’ in computer ‘experiments’ at S This pioneering work opened up the new ORNL,” said Datz. Because of concern about was a strong believer in curiosity-driven science. Throughout his remarkable career, he planted field of molecular beam chemistry. Much of our neutron-induced radiation damage in nuclear re- seeds in different places and recruited collabora- detailed understanding of chemical reaction actors, in 1962 Mark Robinson and Dean Oen, tors to help the seeds grow and bear fruit, open- dynamics has emerged from this new field of two researchers in ORNL’s Solid State Division ing up several new fields in atomic physics. research. The ORNL work also laid the founda- (SSD), attempted to model the effects of an en- In 1951, after earning an advanced degree tion for the research recognized by the 1986 ergetic copper projectile ion on a copper crystal in physical chemistry from Columbia University, Nobel Prize in Chemistry. Dudley R. Herschbach lattice. “They wanted to know how far a copper Datz came to ORNL. He was hired by Ellison (Harvard University), Yuan T. Lee (University of ion goes before it stops,” Datz said. “They let Taylor, then director of ORNL’s Chemistry California at Berkeley), and John C. Polanyi their Monte Carlo computer program run for a Division and his former supervisor at Columbia’s (University of Toronto) received the 1986 Nobel long time, but they sometimes couldn’t find Manhattan Project labs. Prize “for their contributions concerning the where the particle went. They changed the code Datz and Taylor were interested in find- dynamics of chemical elementary processes.” and their simulation showed that the copper atom ing out what happens at the molecular level dur- After receiving a Fulbright Senior Re- ing a chemical reaction. The proposed approach search Fellowship, Datz went to the Netherlands was to make a beam of one type of gaseous mol- in 1962 and worked as a guest scientist at the ecule and shoot it at a beam of another type of FOM Institute for Atomic and Molecular Phys- molecule. It was believed that measurements ics in Amsterdam. There he and Cornelis Snoek made where the two beams cross would shed light discovered that, at low energies, an argon ion on the details of gaseous chemical reactions. In could be bounced off, or scattered from, a cop- 1955, Datz and Taylor first demonstrated the use per atom on a solid surface. In 1964 they reported of crossed molecular beams in studying the the discovery of collisions between an ion and a mechanisms of chemical reactions. They crossed single metal atom, a process known as ion- beams of potassium and hydrogen bromide, and surface scattering. Other groups soon used this the resulting reaction yielded potassium bromide method to analyze the elemental composition and and hydrogen. structure of surfaces. Asked why he chose to work in Oak Ridge, While in Amsterdam, Datz proposed Datz responded that because only extremely shooting the ion beam at a single crystal target at small amounts of product could be expected from a small angle, with respect to the rows of atoms, crossed-beam experiments, the Oak Ridge to reduce interference from ions backscattered In 1962 ion channeling was discovered on a Graphite Reactor could be used to perform neu- from atoms below the surface. “The experiment computer at ORNL. Sheldon Datz and his worked and, surprisingly,” said Datz, “the en- colleagues conducted the first energetic ion tron activation analysis on the deposited potas- channeling experiments using Laboratory sium bromide. After a few less-than-successful trance angle could be made much wider than accelerators.

26 Oak Ridge National Laboratory REVIEW Basic Research at ORNL

often came out the other side of the lattice.” Their (recombination). Datz 1963 modeling led to their prediction that ions and ORNL’s Pete can travel through a crystal in the space, or chan- Dittner pioneered the nel, between rows of atoms and planes in the lat- study of electron-ion tice—hence the term, ion channeling. recombination, using In 1964, then SSD director Doug Billington merged beams of elec- asked Datz to start a program on particle-solid trons and ions. They interactions with emphasis on channeling. “We were the first to make needed energetic heavy ions,” Datz said. “Charlie recombination mea- Moak in the Physics Division produced them on surements on multiply the EN Tandem Van de Graaff accelerator. We also charged ions. needed very thin crystals. Tom Noggle, an SSD An FED group expert in radiation damage in thin crystals, pro- led by Clarence duced perfect gold crystals 300 atoms thick. We Barnett merged with chose iodine-127 and bromine-79 ions at energies Datz’s Atomic and of 50 to 100 million electron volts because they Molecular Physics could be viewed as synthetic fission fragments Section in ORNL’s similar to what is produced in reactors.” Physics Division to In 1965 Datz, Moak, and Noggle demon- perform basic fusion strated experimentally the phenomenon of ener- energy research. In the getic ion channeling in solids. They found that mid-1980s, this group the ions moving along the channels of gold single (which is today led by crystals lost only about half the energy they would Fred Meyer) built the yield if they traveled through the same volume Electron Cyclotron Resonance Multi- On December 18, 2000, Sheldon Datz received DOE’s Enrico Fermi Award from in a random direction. The explanation: Penetrat- President Clinton. Other Fermi Award winners from ORNL are Liane Russell ing ions lose energy mainly through collisions charged Ion Source (1994), Richard Setlow (1988), Alexander Hollaender (1983), Alvin Weinberg with electrons, but the collective attractive (ECR), which is used (1980), William Russell (1976), and Nobel Laureate Eugene Wigner (1958). action of nuclei in the planes gently pushes the to obtain fundamental positively charged ions away from the high con- data on plasmaÐwall interactions of interest to using heavy-ion storage rings to study low- centration of electrons in the atomic cores. fusion energy researchers. The ECR produces posi- energy collisions between molecular ions and In 1978 Datz and his collaborators re- tively charged ions by using highly energetic elec- electrons. One result of these collisions is “dis- ported the discovery of resonant coherent exci- trons, heated by microwave radiation and confined sociative recombination,” in which molecular tation of channeled ions. In this phenomenon, an by magnetic fields, to strip electrons from atoms ions recombine and break up into neutral frag- of a specific element. Recently, DOE asked ion with only one electron is raised to an excited ments. This process is of great importance to low- researchers in the Atomic Physics Section to fo- state by traveling at the right velocity (and fre- temperature plasma studies, astrochemistry, and cus on research using ions from an upgraded ECR. quency) past the local electric fields of a “picket aeronomy. Considerable work in this area is now DOE’s investment in high-energy nuclear fence” of atoms, along a channel in a crystal. being conducted at a number of storage rings and particle physics is considerable. In the Large Work in this area is currently ongoing in Japan. around the world. Hadron Collider being built at CERN near In 1979, Datz and collaborators in Cali- Although Datz did not win the Nobel Prize Geneva, Switzerland, lead ions will collide in a fornia (Lawrence Livermore National Laboratory for pioneering the crossed molecular beam tech- storage ring at speeds very close to the speed of and Stanford University) published two papers nique, his successes in opening up new fields of light. It is expected that atomic physics processes on their discovery of channeling radiation. They study in atomic physics brought him two other will occur with much greater likelihood than the prestigious awards. In 1998 Datz received the observed experimentally that ultra-relativistic desired nuclear events, thus interfering with these American Physical Society’s Davisson-Germer electrons or positrons entering a channel oscil- experiments. A prime concern here is the loss of Prize in Atomic or Surface Physics for his re- late in response to the string of atoms’ local elec- stored beam. search into atomic interactions with ions, elec- tric fields, resulting in the release of strongly “In the collision region, a bare lead ion trons, and photons. On November 9, 2000, Presi- forward-directed, high-energy X-ray radiation. captures an electron that it has created as part of The study of atomic collisions in solids an electron-positron pair,” Datz explained. The dent Bill Clinton named Datz a co-winner of was initiated largely in response to problems re- results of theoretical predictions differed widely DOE’s Enrico Fermi Award, one of the nation’s lated to radiation damage. The physics learned in the early 1990s. Accordingly, Datz initiated highest prizes in science. On December 18, 2000, from these studies has numerous applications, a program to study atomic collision physics Datz received this award for his pioneering re- including development of ion implantation, a at ultra-relativistic energies, using 33-TeV lead search in atomic and chemical physics. technique widely used in the fabrication of com- ions from the Super Proton Synchrotron at Reflecting on DOE’s decision to focus puter chips. CERN. This multinational effort included Randy ORNL’s atomic physics research on the ECR in In the late 1970s John Clarke, then direc- Vane, Herb Krause, and Dittner from ORNL and light of his career, Datz said, “DOE thinks that tor of ORNL’s Fusion Energy Division (FED), sug- collaborators from Sweden, Denmark, we do too many things, that we should work on gested to Datz that his group should look into some Germany, Switzerland, and South Africa. “The one or two projects. That is hard for us because interesting atomic collision problems affecting results, aside from quantifying expected phenom- we have developed many interests and capabili- fusion plasmas. Datz chose dielectronic recombi- ena,” said Datz, “disclosed some new, totally un- ties. I contributed to a lot of different areas, all of nation, which was at that time a little known but anticipated findings, which, in addition to their which related in some way to DOE missions. To nonetheless important area, especially for highly intrinsic interest, can be of use in designing very be sure, the main efforts at ORNL should be di- charged ions. In a plasma, excited electrons may expensive particle physics experiments.” rected at larger projects, but I think it’s vital to break free of nuclei (ionization) and free electrons During his tenure as guest professor in the Laboratory’s health to also encourage more may recombine with ions that have lost electrons Stockholm in 1990, Datz devised a method for speculative, curiosity-oriented research.”

Number Two, 2001 27 Basic Research at ORNL

A delayed response can be costly for a lot of rea- sons. Beyond the example of the physician and the patient’s X ray, there are also examples in science. Say, Improving the for instance, a million-dollar supercomputer is waiting for the data it needs to perform certain calculations. If an unanticipated delay occurs in the transmission of the data, that million-dollar supercomputer could remain Internet’s idle. A delay in relaying a complete message could also cause a chaotic response in a robot being operated re- An ORNL motely over a network, say, to run an experiment. “Just algorithm has as a drunk driver may cause a traffic collision because been shown to of a delayed response, a robotic arm could behave wildly Quality of because of a delay in delivering the complete set of com- reduce and mands,” says Geist. To reduce and predict delays in data delivery, predict delays in data CSMD’s Nageswara Rao has developed a computer pro- delivery over the Internet. gram called NetLet that is currently being tested on a Service small subset of the Internet. “NetLet allows computers to talk with each other and determine whether a com- f a physician wants to send a critically ill patient’s X ray over the Internet to a plete message got there and what the delay is in getting radiologist who specializes in cancer diagnosis, should that X ray be delayed by the dropped packet to the receiver,” Rao says. “This al- transmissions of data between computer game players? Who should decide who gorithm enables the computers to measure connection gets needed data the fastest or who should get a message delivered first when speeds and the delays of alternate pathways and then iden- there isI considerable traffic on the network? Should only people willing to pay for service- tify the best combination of pathways to get the informa- delivery agreements receive messages faster than others who do not pay or who pay less? tion delivered efficiently in the time or rate guaranteed.” Whether the desired performance in delivering messages and huge data files can be Demonstrations of NetLet have shown that the provided is an issue facing people concerned about the “quality of service” (QOS) of net- algorithm has improved the speed of data delivery by works. QOS is all about being able to guarantee that data will be received by a certain time, at about 40%, without any additional support from Internet a certain rate, or with some other desired quality. “It is analogous to next-day-air delivery by routers. “Some of our data files used to take 10 seconds Federal Express versus delivery by U.S. mail,” says Al Geist, head of the High-Performance to get from our computer to a destination computer,” Computing Research Section in ORNL’s Computer Science and Mathematics Division Rao says. “Those same data files can now get there in 6 (CSMD). “If your Internet file absolutely has to be there by a certain time, then you need seconds. That means that a huge data file that takes 10 QOS. Unfortunately, the Internet does not have any delivery guarantees, but researchers at hours to arrive at a destination computer can now get CSMD are trying to change that.” there in 6 hours.” When a data file is sent over the Internet, it is sliced into pieces, or data packets. Com- NetLet could be useful for speeding up the puters called routers direct these packets along different paths over the network. The packets delivery of large data files from neutron scattering ex- are then reassembled at their destination. One problem is that with today’s Internet a router is periments performed at the Spallation Neutron Source allowed to throw away data packets in response to traffic conditions and priorities assigned by (SNS) when it becomes operational in 2006 at ORNL. the service providers. In response, the lost packet must be resent, and there is no guarantee “We need to find ways to rapidly transfer hundreds of that this packet will not also be lost. The net effect is that delivery of the complete data file can gigabytes or terabytes of data from the SNS across the be delayed for an unknown amount of time. country to scientists requesting the information,” Geist says. “And our climate scientists are needing this capability today.” It takes about 2 hours to send 500 megabytes of climate data (calculated results of simulations of future climate) from DOE’s Center for Computational Sciences at ORNL to the National Energy Research Supercomputing Center at DOE’s Lawrence Berkeley National Laboratory in California. These data files are sent across DOE’s new Energy Sciences Network (ESnet), a semiprivate part of the Internet. Today, if high-energy physicists wish to trans- fer hundreds of terabytes of data from DOE’s Stanford Linear Accelerator Center in California to the European Laboratory for Particle Physics (CERN) near Geneva, Switzerland, they don’t use ESnet, because it would take a month to transmit all the data within these huge files. Instead, they deliver the information in a week by copy-

Illustration enhanced by LeJean Hardin ing it onto tapes and transporting the tapes to CERN by Federal Express. The goal is to further develop NetLet and other approaches so that huge files of data from high-energy physics facilities and the SNS can be The ORNL-developed NetLet algorithm reduces by 40% end-to-end delays in transmitting data files over the network, with no support from Internet routers. In one set of tests in which the algorithm delivered in one to two days. was used part of the time, the data files were sent between ORNL and the University of Oklahoma If NetLet is eventually incorporated into Internet (OU) over the Internet. In another set of tests, the data files were transmitted from ORNL to OU routers, more users may feel they are sending and receiv- via computers at Old Dominion University (ODU) and Louisiana State University (LSU). ing their information by express.

28 Oak Ridge National Laboratory REVIEW New Biology: BasicCovering Research All The at BasesORNL

QOS for Wireless Communication According to Nageswara Rao (shown here), connectivity- A team of mobile robots carrying radiation through-time protocol enables robots to communicate detectors in a remote building or at a with each other without network infrastructure waste burial site has a mission: and dynamic connectivity. Autonomously build a map of radiation levels in the area suspected of contamination. Through wireless communication the robots coordinate their activities to ensure a complete radiation profile

Curtis Boles, enhanced by LeJean Hardin of the building or waste site. In such areas, the infrastructure for conventional wireless networks is lacking, and the wireless connectivity can be quite unpredictable. For mapped by robots not in direct wireless contact. As the robot in example, in a typical building, a robot mapping one room for the hall moved past the entrance to each room, it communi- radiation cannot communicate directly by radio with a robot cated, by radio, with the robot mapping the room, keeping track in another room because radio waves cannot pass through of its progress in generating the radiation profile and the location the thick walls separating the rooms. A possible solution to and amount of building space that remained to be mapped. this problem has been devised at ORNL. To foster robust com- The mail robot picked up a message from a robot in one room munication between robots and more generally mobile nodes, and delivered the message to a robot in the other room, thus ORNL’s Nageswara Rao has developed the “connectivity- allowing these robots to communicate with each other. Though through-time” protocol and implemented a system using never before possible, with this new system for improving NetLet software in which a robot serves as a “router.” quality of service in wireless communication, robots can ex- In a recent demonstration at ORNL, a “mail” robot mapped change messages under conditions of highly unpredictable, the hall connecting two rooms, each of which was being wireless connectivity.

The connectivity-through-time protocol developed at ORNL exploits the motion of one robot to deliver messages to and from robots separated by walls, allowing wireless communication among all robots in a dynamic network.

Number Two, 2001 29 OAK RIDGE NATIONAL LABORATORY REVIEW P.O. Box 2008, Oak Ridge, Tennessee 37831-6266 PRSRT STD U.S. Postage PAID Permit No. 3 Oak Ridge, TN

A fusion research device called the Quasi-Poloidal Stellarator may be built by 2006 at ORNL. It is being designed by ORNL fusion theorists with the help of the IBM supercomputer at DOE’s Center for Computational Sciences in Oak Ridge. The above visualization by ORNL’s Don Spong shows a QPS- optimized stellarator plasma surface with a magnetic field line (white) superimposed. (See article on p. 16.)