UC San Diego Electronic Theses and Dissertations

UC San Diego UC San Diego Electronic Theses and Dissertations

Title Heterogeneous and multiphase chemistry of reactive nitrogen oxides in the marine boundary layer

Permalink https://escholarship.org/uc/item/5zd857fz

Author Ryder, Olivia Sharmelle

Publication Date 2015

Peer reviewed|Thesis/dissertation

eScholarship.org Powered by the California Digital Library University of California UNIVERSITY OF CALIFORNIA, SAN DIEGO

Heterogeneous and multiphase chemistry of reactive nitrogen oxides in the marine boundary layer

A dissertation submitted in partial satisfaction of the requirements for the degree Doctor of Philosophy

Chemistry

Olivia Sharmelle Ryder

Committee in charge:

Professor Timothy Bertram, Chair Professor Lihini Aluwihare Professor Meinrat Andreae Professor Patricia Jennings Professor Francesco Paesani Professor Michael Tauber

2015

Olivia Sharmelle Ryder, 2015

The Dissertation of Olivia Sharmelle Ryder is approved, and is acceptable in quality and form for publication on microfilm and electronically:

______

Chair

University of California, San Diego

2015

iii

DEDICATION

To my parents- thank you for your love, sacrifices,

and for always believing in me.

EPIGRAPH

If research was easy, everyone would be doing it -B. J. Finlayson-Pitts

TABLE OF CONTENTS

Signature Page...... iii

Dedication...... iv

Epigraph...... v

Table of Contents...... vi

List of Figures...... ix

List of Tables...... xii

Acknowledgements...... xiii

Vita...... xvi

Abstract of The Dissertation...... xviii

Chapter 1 Introduction ...... 1

1.1 The Role of N2O5 in the Atmosphere ...... 2

1.2 N2O5 Hydrolysis ...... 3

1.3 The Reactive Uptake Probability, (N2O5) ...... 4

1.4 Particle Chemical Composition and N2O5 Reactive Uptake ...... 5 1.4.1 The Effect of Chemical Composition of Ambient Particles: Field Measurements of (N2O5) ...... 5 1.4.2 Uptake to Inorganic Salts: Laboratory Studies ...... 6 1.4.3 The Impact of Particulate Organics: Laboratory Studies ...... 6 1.4.4 Laboratory and Field Measurement Agreement ...... 7

1.5 Product Formation Following N2O5 Reaction on Chloride-Containing Surfaces ...... 8

1.5.1 Ambient Measurements of ClNO2 and (ClNO2) ...... 9 1.6 Summary and Remaining Areas to Address ...... 10 1.7 Synopsis of Chapters ...... 12 1.8 References ...... 13

Chapter 2 On The Role of Particle Inorganic Mixing State in The Reactive Uptake of N2O5 to Ambient Aerosol Particles ...... 17 2.1 Abstract ...... 17 2.2 Introduction ...... 17

2.2.1 Factors Controlling the Reactive Uptake of N2O5 to Aerosol Particles ...... 18

2.2.2 Single Particle Chloride-to-Nitrate Mixing State in The Polluted Marine Boundary Layer ...... 19 2.3 Materials and Methods ...... 21 2.4 Results and Discussion ...... 22

2.4.1 Impact of Particulate Chloride-to-Nitrate Mixing State on (N2O5) ...... 22 2.4.2 Observation-Based Constraints on the Impact of Chloride-to-Nitrate Particle Mixing State on (N2O5) ...... 26 2.5 Insight from Single Particle Measurements of Reacted Sea Spray Aerosol ...... 31 2.6 Atmospheric Implications ...... 34 2.7 Acknowledgements...... 35 2.8 Supporting Information ...... 35 2.8.1 Sampling Locations and Details ...... 35 2.8.2 Measurements ...... 37 2.8.3 Calculations and Interpretations ...... 40 2.9 Acknowledgments ...... 48 2.10 References ...... 48

Chapter 3 On the Role of Organic Surfactants and Coatings in Regulating N2O5 Reactive Uptake to Sea Spray Aerosol ...... 53 3.1 Abstract ...... 53 3.2 Introduction ...... 54 3.3 Materials and Methods ...... 57

3.3.1 N2O5 Generation and Detection ...... 57 3.3.2 Sea Spray Aerosol Generation and Duty Cycle ...... 58 3.3.3 Molecular Mimics for a Synthetic Phytoplankton Bloom ...... 60 3.4 Wave Channel Generated Sea Spray ...... 63 3.5 Seawater and Sea Spray Aerosol Characterization ...... 64 3.5.1 Aerosol Particle Characterization ...... 64 3.5.2 Water Characterization ...... 66

3.6 (N2O5) Determinations ...... 66 3.7 Results and Discussion ...... 70 3.7.1 Aerosol Organic Volume Fraction ...... 70

3.7.2 Organic Addition ESffect on (N2O5) ...... 72 3.8 Conclusions ...... 76 3.9 Acknowledgements...... 77 3.10 References ...... 77

Chapter 4 On the Role of Organics in Regulating ClNO2 Production at the Air-Sea Interface 83

vii

4.1 Abstract ...... 83 4.2 Introduction ...... 83 4.3 Materials and Methods ...... 86

4.3.1 N2O5 Generation and Sample Exposure ...... 86 4.3.2 Ambient Sample Collection ...... 87 4.3.3 Laboratory Organic Mimic Preparation ...... 88 4.3.4 Surface Tension Measurements ...... 89 4.3.5 Box Model ...... 89 4.4 Results and Discussion ...... 90 - 4.4.1 (ClNO2) Dependence on Cl ...... 90

4.4.2 ClNO2 Yields from Ambient Seawater Samples ...... 92

4.4.3 (ClNO2) from Laboratory Mimics ...... 95 4.5 Linking Laboratory and Field Studies...... 101 4.6 Summary and Conclusions ...... 103 4.7 Acknowledgements...... 104 4.8 References ...... 105 Chapter 5 The potential role of divalent cations at the air-sea interface in regulating the production rate of ClNO2 ...... 109 5.1 Abstract ...... 109 5.2 Introduction ...... 109 5.3 Materials and Methods ...... 111 5.3.1 Solution Preparation ...... 111 5.3.2 Real and Proxy Ocean Water Solutions ...... 111

5.3.3 N2O5 and ClNO2 detection ...... 112

5.3.4 N2O5 generation and delivery ...... 112 5.3.5 Ion Chromatography Analysis ...... 113 5.3.6 Solution pH and Surface Tension Measurements ...... 113 5.3.7 Inductively Coupled Plasma-Mass Spectrometry (ICP-MS) Analysis ...... 114 5.4 Results and Discussion ...... 114 5.5 Future work/ challenges ...... 124 5.6 Acknowledgments ...... 124 5.7 References ...... 125

viii

LIST OF FIGURES

Figure 1.1 The overlap between the NOx and O3 cycles in the troposphere...... 2

Figure 1.2 Box model results showing the fraction of NOx lost at night due to N2O5 and NO3 reaction...... 3

- Figure 1.3 ClNO2 yield following N2O5 reaction as a function of [Cl ] in solution from which particles were atomized. The results shown are from 3 separate studies...... 9

Figure 1.4 Model calculations illustrating the dependence of NOx loss (left y axis) and chlorine radical production following photolysis of ClNO2 (right y axis) on the magnitude of (N2O5). This plot has been adapted from Bertram et al. 2009.20 ...... 11

Figure 2.1 A) Model parameterization of (N2O5), where nascent SSA are emitted at high - - - nH2O/nNO3 (aq) and nCl /nNO3 , and exhibit large (N2O5) in the absence of organic films. As - - nascent SSA is chemically processed in the atmosphere, nCl /nNO3 decreases and the rate of N2O5 uptake slows...... 24

Figure 2.4 Probability distribution functions (PDF) of (A) fROA (calculated using the representative organic ions, m/Q (37 + 43) / total ion peak area) and (B) the contribution of m/Q 43 to the organic signal (m/Q 43 / m/Q (37 + 43) ) for the more-polluted (red) and less-polluted (blue) pier data sets...... 30

- - Figure 2.5 Probability distribution function (PDF) as a function of nCl /nNO3 for: A) more- polluted pier data, B) less-polluted pier data, and C) nascent SSA data sets. D) Single particle - - nCl /nNO3 as a function of particulate organic fraction (fROA) for nascent SSA...... 32

- - Figure 2.6 Probability distribution curves of particulate nCl /nNO3 calculated for four cases: ambient aerosol from Jeju Island, Korea (grey) and Long Beach, CA (black), nascent sea spray aerosol generated in the SIO wave channel (blue), and the same nascent sea spray after reaction with HNO3 (red)...... 42

Figure 2.7 Assessment of variation in the cumulative distribution function arising from sources - - unrelated to particle-to-particle variability. Panel A shows the raw variability in nCl /nNO3 over 24 hours of sampling at Jeju Island (light grey markers). The running median over 500s of sampling is shown in blue ...... 45

Figure 3.2 Plunging duty cycle flag as a function of time (Panel A) where a value of 1 indicates plunging is turned on for SSA production mode, and a value of zero indicates plunging is turned off. Panel B shows the response of the total (red) and sub-micrometer (green) surface area to changes in the plunging duty cycle as a function of time...... 61

Figure 3.4 Panel A shows a representative time trace of total particle surface area (grey) and super-micrometer surface area (green) during a typical experimental cycle, where time=0 corresponds to plunging turned off, thus switching from SSA production mode to SSA decay mode...... 68

Figure 3.5 Representative plot of instantaneous kobs versus the total surface area in the flow reactor. Data was filtered such that over the length of one period of residence time in the flow reactor, the surface area was not changing by more than 20 % (see text for details)...... 70

Figure 3.6 AFM derived organic volume fraction of 0.33-0.56 m particles (measured as 0.21 m at 21 % RH by AFM) collected using a MOUDI, imaged for each water matrix vs. the fraction of occurrence...... 71

Figure 3.7 (N2O5) as a function of water side analysis metrics including carbon concentration added to the water matrix (A), water pH (B), surface tension (C). All points are colored by the amount of carbon present in the water matrix, and the real ocean sample is colored green...... 73

Figure 3.8 (N2O5) as a function of the organic mass fraction from this work (green, color coded by TOC water content) as compared to values obtained from a real mesocosm bloom generated in the wave channel facility at Scripps Institution of Oceanography in 2011 (yellow) ...... 76

Figure 4.1 Signal intensities for N2O5 (top panel) and ClNO2 (bottom panel) during N2O5 delivery to CI-QMS inlet (I), after flowing over a saturated NaCl solution (6.1 M NaCl) (II), and back to the CI-QMS inlet (III)...... 91

Figure 3.2 ClNO2 yield as a function of chloride concentration as determined in this work (red squares) and fit values from previous determinations, Roberts et al., 2009 (circles), Bertram and Thornton, 2009 (inverted triangles), and Behnke et al., 1997 (diamonds)...... 92

Figure 4.3 Dependence of the measured ClNO2 yield on chlorophyll-a, a proxy for biological activity, for ambient sea-water samples collected from coastal and open ocean water and from a large scale mesocosm bloom experiment...... 96

Figure 4.6 ClNO2 yield as a function of carbon concentration, assuming homogenous mixing, following sequential additions of cholesterol (blue squares) and humic acid (black circles) into a 500 mM NaCl solution...... 103

Figure 5.2 ClNO2 yield following MgCl2 anhydrous addition to a 500 mM NaCl solution (circles, panel A). The solid line indicates the typical concentration of Mg2+ in the ocean. Panel B shows 2+ the ClNO2 yield following EDTA addition to a solution of 53 mM Mg in 500 mM NaCl solution...... 116

Figure 5.3 ClNO2 yield as a function of EDTA molar equivalent additions to 53 mM MgCl2 anhydrous (blue) and artificial reef salt (green). The dashed lines are fits to serve as a guide. ... 117

2+ Figure 5.4 ClNO2 yield as a function of [Mg ] added to a 500 mM NaCl solution for both MgCl2.6H2O (grey squares) and MgCl2 anhydrous (red squares). The oceanic concentration of magnesium is denoted by the blue line at 53 mM Mg2+...... 119

Figure 5.5 ICP-MS analysis of coastal ocean water (A), and artificial reef salt (B), MgCl2.6H2O, Sigma Aldrich (C), MgCl2.6H2O, Macron (D), MgCl2 anhydrous (E)...... 122

Figure 5.6 A closer view of Figure 5.5, showing ions present in lower concentrations. Inset Panel F shows a closer view of the 57Fe peak signal...... 123

LIST OF TABLES

Table 2.1N2O5 values calculated using the assumption of an externally mixed particle population, an internally mixed population, and measured from SIO pier, along average RH and particle properties for sampling Periods 1 and 2...... 27

Table 2.2 The fraction of particles removed before CDF and/or N2O5 calculation...... 40

Table 3.1 Synthetic seawater components and their respective properties...... 63

Table 4.1 Bulk carbon concentrations following sequential addition of organic molecular mimics to 500 mM NaCl solutions...... 89

Table 4.2 Ambient water samples, corresponding ClNO2 yields, and co-located measurements of chlorophyll-a, colored dissolved organic matter (CDOM), dissolved organic carbon (DOC), and surface tension where available...... 94

Table 5.1 Sample solutions and their associated measured pH, [Mg2+] as measured by IC, and surface tension...... 118

xii

ACKNOWLEDGEMENTS

I would like to thank Timothy Bertram, my thesis advisor, for taking me on as his first graduate student and for his training, encouragement, support and for affording me many invaluable research opportunities.

I would also like to thank all members of the Bertram group, from when we were at our strongest (Timia, James, Michelle, Nicole, Steve, Katy, Matt), for great times spent together in the lab/office and during field studies, for your support during hard times, and for constant comic relief! There was never a dull moment, and you all helped make my graduate school experience a great one- thank you so much!

Additionally, Kim Prather and everyone involved in CAICE since its beginning have had a profound impact on my development as a researcher and in my ability to collaborate, and for this I would also like to thank you all.

Last but not least, my family and friends who have stuck by me all throughout my education, and especially over the last 6 years - thank you so much. Without you all backing me, cheering me on, and supporting me the whole way through good times and bad, I would not have been able to achieve this.

Chapter 2, in full, is a reprint of the material as it appears in Environmental Science and

Technology. Olivia S. Ryder, Andrew P. Ault, John F. Cahill, Timothy L. Guasco, Theran P.

Riedel, Luis A. Cuadra-Rodriguez, Cassandra J. Gaston, Elizabeth Fitzgerald, Christopher Lee,

Kimberly A. Prather, and Timothy H. Bertram. (2014), On the Role of Particle Inorganic Mixing

State in the Reactive Uptake of N2O5 to Ambient Aerosol Particles, Environ. Sci. Technol., 2014,

48 (3), pp 1618–1627, doi: 10.1021/es4042622. The dissertation author was the primary investigator and author of this paper.

xiii

The authors would like to thank Joel Thornton for the use of the CIMS and flow tube apparatus used in the ambient portion of this study. This research was supported by the Office of

Science (Office of Biological and Environmental Research), U.S. Department of Energy (Grant

No. DE-SC0006431) and the National Science Foundation (Grant No. CHE1038028). O.S.R is grateful for a Graduate Research Fellowship from the National Science Foundation (2011-2014).

Chapter 3, in part, is currently being prepared for submission of the material.

Olivia S. Ryder, Nicole R. Campbell, Holly Morris, Matthew J. Ruppel, Alexei Tivanski, and

Timothy H. Bertram. The dissertation author was the primary investigator and author of this material.

This research was supported by the National Science Foundation via the Center for

Aerosol Impact on Climate and the Environment, a Center for Chemical Innovation (NSF

CHE1305427). The authors gratefully acknowledge Dr. Chris Cappa for use of SMPS and growth factor data used in this study and for helpful discussions throughout the experiments. The authors also thank Steven Schill and Sara Forestieri for participating in the collaborative effort of running these experiments. O.S.R gratefully acknowledges a Graduate Research Fellowship from the

National Science Foundation (2011-2014).

Chapter 4, in full, is currently under review in The Journal of Physical Chemistry A.

O.S. Ryder, N.R. Campbell, M. Shaloski, H. Al-Mashat, G. M. Nathanson, and T.H. Bertram.

(2014), On the Role of Organics in Regulating ClNO2 Production at the Air-Sea Interface. The dissertation author was the primary investigator and author of this material.

This research was supported by the National Science Foundation via the Center for

Aerosol Impact on Climate and the Environment, a Center for Chemical Innovation (NSF

CHE1305427). The authors gratefully acknowledge Dr. Grant Deane and Dale Stokes for use of the tensiometer used in this work. Dr. Matthew Zoerb (UCSD) for the collection of Western

Atlantic Cruise II 2014 water samples, Bob Vaillancourt, Jeremiah Stone, and Evan Ntonados,

xiv

(all from Millersville University of Pennsylvania) for providing WACS2 chlorophyll-a concentrations, and the entire CAICE IMPACTS team and the staff of the SIO Hydraulics

Laboratory. O.S.R gratefully acknowledges a Graduate Research Fellowship from the National

Science Foundation (2011-2014).

Chapter 5, in part, is currently being prepared for submission of the material.

Olivia S. Ryder and Timothy H. Bertram. The dissertation author was the primary investigator and author of this material.

This work was supported by the National Science Foundation via the Center for Aerosol

Impact on Climate and the Environment, a Center for Chemical Innovation (NSF CHE1305427).

The authors would like to gratefully thank Scott Wilson and Dr. Skip Pomeroy for IC analysis,

Dr. James Day for ICP-MS analysis, Dr. Dale Stokes and Dr. Grant Deane for use of the surface tensiometer used in this work, Dr. Kim Prather, Dr. Michael Tauber and Dr. Andreae Meinrat for helpful discussions. O.S.R gratefully acknowledges a Graduate Research Fellowship from the

National Science Foundation (2011-2014).

VITA

2006-2008 Undergraduate Research Assistant, Department of Chemistry University of California, Irvine

2008 Bachelor of Science, Chemistry (Honors), Magna Cum Laude, University of California, Irvine

2009-2010 Teaching Assistant, University of California, San Diego

2009-2015 Research Assistant, Department of Chemistry University of California, San Diego

2011-2014 National Science Foundation Graduate Research Fellowship Program Fellow

2011 Master of Science, University of California, San Diego

2015 Doctor of Philosophy, University of California, San Diego

PUBLICATIONS

Ryder, O. S.; Ault, A.; Cahill, J. F.; Guasco, T. L.; Cuadra-Rodriguez, L.; Thornton, J. A.; Gaston, C.; Fitzgerald, E.; Lee, C.; Prather, K. A.; Bertram, T. H. On the Role of Particle Mixing State in the Reactive Uptake of N2O5 to Ambient Aerosol Particles. Environ. Sci. Technol., 2014, 48(3), 1618-27

Ault, A. P.; Guasco, T.L. ; Baltrusaitis, J.; Ryder, O. S.; Trueblood, J.; Collins, D. B.; Ruppel, M. J.; Cuadra-Rodriguez, L. A.; Prather, K. A.; Grassian, V. H. Heterogeneous Reactivity of Nitric Acid with Nascent Sea Spray Aerosol: Large Differences Observed Between and Within Individual Particles. J. Phys. Chem. Lett., 2014, 5, 2493-2500

Ault, A. P.; Guasco, T. L.; Ryder, O. S.; Baltrusaitis, J.; Cuadra-Rodriguz, L. A.; Collins, D. B.; Ruppel, M. J.; Bertram, T. H.; Prather, K. A.; Grassian, V. H. Inside versus Outside: Ion Redistribution in Nitric Acid Reacted Sea Spray Aerosol Particles as Determined by Single Particle Analysis. J. Am. Chem. Soc. - Communication, 2013, 135(39), 14528-14531.

Ebben, C.; Ault, A.; Ruppel, M.; Ryder, O.; Bertram, T.; Grassian, V.; Prather, K.; Geiger, F. Size-Resoslved Sea Spray Aerosol Particles Studied by Vibrational Sum Frequency Generation. J. Phys. Chem. A, 2013, 117 (30), 6589–6601

Prather, K. A.; Bertram, T. H.; Grassian, V. H.; Deane, G. B.; Stokes, M. D.; DeMott, P. J.; Aluwihare, L. I.; Palenik,B.; Azam, F.; Seinfeld, J. H.; Moffet, R. C.; Molina, M. J.; Cappa, C.

xvi

D.; Geiger, F. M.; Roberts, G. C.; Russell, L. M.; Ault, A. P.; Baltrusaitis, J.; Collins, D. B.; Corrigan, C. E.; Cuadra-Rodriguez, L. A.; Ebben, C. J.; Forestieri S. D.; Guasco, T. L.; Hersey, S. P.; Kim, M. J.; Lambert, W.; Modini, R. L.; Mui, W.; Pedler, B. E.; Ruppel, M. J.; Ryder, O. S.; Schoepp, N.; Sullivan, R. C.; Zhaoa, D. Bringing the Ocean into the Laboratory: Impacts of Chemical Complexity of Sea Spray Aerosol on Climate Properties. Proc. Nat. Acad. Sci. USA, 2013, 110(19), 7550-7555

Stokes, M. D.; Deane, G. B.; Prather, K.; Bertram, T. H.; Ruppel, M. J.; Ryder, O. S.; Brady, J. M.; and Zhao, D. A Marine Aerosol Reference Tank System as a Breaking Wave Analogue. Atmos. Meas. Tech., 2013, 6, 1085-1094

Riedel, T. P.; Bertram, T. H.; Ryder, O. S.; Liu, S.; Day, D. A.; Russell, L. M.; Gaston, C. J.; Prather, K. A.; and Thornton, J. A. Direct N2O5 Reactivity Measurements at a Polluted Coastal Site. Atmos. Chem. Phys., 2012, 12, 2959-2968

Bertram, T. H., Kimmel, J. R., Crisp, T. A., Ryder, O. S., Yatavelli, R. L. N., Thornton, J. A., Cubison, M. J., Gonin, M., and Worsnop, D. R. A Field-Deployable, Chemical Ionization Time- of-Flight Mass Spectrometer. Atmos. Meas. Tech., 2011, 4, 1471-1479

FIELDS OF STUDY

Major Field of Study: Chemistry (Atmospheric & Analytical Chemistry)

xvii

ABSTRACT OF THE DISSERTATION

Heterogeneous and multiphase chemistry of reactive nitrogen oxides in the marine boundary layer

Olivia Sharmelle Ryder

Doctor of Philosophy in Chemistry

University of California, San Diego, 2015

Professor Timothy Bertram, Chair

Dinitrogen pentoxide (N2O5), a primary nocturnal reservoir for NOx (NOx ≡ NO + NO2), readily undergoes heterogeneous reaction on aqueous interfaces. These reactions potentially alter the chemical, physical, and hygroscopic properties of ambient particles, and ultimately their climate impact. Upon reacting with aqueous chloride-containing surfaces, such as sea spray,

xviii

nitryl chloride (ClNO2) is produced. ClNO2 is capable of extending the lifetime of NOx and inducing ozone production in high NOx environments via chlorine radical creation. The extent to which N2O5 reactivity impacts atmospheric oxidant loadings depends on the heterogeneous reaction rate of N2O5 with a given surface. This, in turn, is directly influenced by interfacial chemical composition, morphology, and phase state.

In this dissertation, I describe advances in the understanding of how organic and inorganic constituents at atmospheric interfaces regulate N2O5 reactivity.

First, I developed an observation-based framework for assessing the aerosol particle mixing state role on N2O5 heterogeneous reaction kinetics using simultaneous measurements of the N2O5 reactive uptake coefficient to ambient aerosol with direct measurements of aerosol mixing state before and after reaction. Second, as part of the NSF Center for Aerosol Impacts on

Climate and the Environment, I extended the use of a novel Marine Aerosol Reference Tank

(MART) capable of generating nascent sea-spray aerosol, to determine heterogeneous reaction kinetics. Here, the MART was doped with ocean organic mimics to assess how organics present in the sea surface microlayer (SSML), and subsequently ejected into sea-spray, impact N2O5 reaction kinetics. Third, I conducted laboratory studies to investigate suppressed ClNO2 production rates at the ocean surface, using ambient seawater from various locations and under varied biological conditions, along with model sea salt solutions with proxy organics. This study was designed to determine the role of unsaturated organics and surface-enhanced inorganic ions in the SSML in altering ClNO2 production rates.

xix

Chapter 1 Introduction

The tropospheric atmosphere is comprised of two main classes of components: aerosol particles (or simply aerosol) and gases. Aerosol are defined as liquid or solid particles suspended within a gas.1 It is critical to understand aerosol and their chemical and physical changes throughout their lifetime in the atmosphere, as they have the ability to impact visibility, human health, air quality, and climate.1 More specifically, aerosol can impact climate by acting as cloud condensation nuclei (CCN), by which they serve as a surface upon which water vapor can condense to form a cloud droplet.2 However, the size and composition of aerosol particles will dictate their ability to act as CCN, and as such it is important to understand these complex components of the atmosphere. Aerosol vary in size and composition, both of which are initially linked to the production mechanism and source of the particle.1,2 However, throughout their lifetime, these particles undergo atmospheric processing, including adsorbing and desorbing water, coagulation, or reaction with gas-phase molecules, capable of altering their chemical composition. The latter process is referred to as heterogeneous chemistry and is the main focus of this dissertation.

Heterogeneous chemistry involves chemical reactions that take place between species of different phases, such as between a gas molecule and an aerosol particle. These reactions are important not only in terms of aerosol processing, potentially making particles more or less hygroscopic, but also due to the reactive gases these reactions can remove from the atmosphere.

Further, these reactions are capable of producing reaction products, which may partition into the gas phase, and undergo further reaction. Here we will focus on the heterogeneous reaction that takes place between dinitrogen pentoxide (N2O5) and reactive surfaces available in the marine environment.

1.1 The Role of N2O5 in the Atmosphere

N2O5 is an important tropospheric nitrogen-containing nocturnal species, that is formed from the reaction of NO2 and NO3. Its formation and reaction are limited to the nighttime hours as

1 it is subject to thermal decomposition, and photodissociates between =200 - 400 nm. NO3 is also photolabile, dissociating within the visible range (410-670 nm), resulting in a lifetime of approximately 6 seconds. The diminished availability of NO3 greatly limits the subsequent generation of N2O5 during sunlight hours.

N2O5 is considered a nocturnal terminal sink for NOx (NOx=NO + NO2) species in the atmosphere as upon reaction with an aqueous surface, it produces HNO3, thus removing NOx

3,4 from the atmosphere. This in turn has a direct impact on the ozone (O3) and hydroxyl radical

(OH) budgets as shown in Figure 1.1.

Figure 1.1 The overlap between the NOx and O3 cycles in the troposphere.

At night, NO2 will react with O3 to produce NO3, a key oxidant in the nocturnal atmosphere, that establishes equilibrium with N2O5. Therefore, the fate of N2O5 overnight will dictate the oxidative capacity of the atmosphere during this period. Additionally, if N2O5 persists

at dawn, it will thermally decompose back into NO2 and NO3. However, if it has reacted during the night, a lower concentration of NO2 will be present during the morning hours, effecting subsequent O3 and OH generation. It is therefore of key importance to understand N2O5 reactions to accurately predict the ozone, OH and NOx budgets in the polluted marine atmosphere. Figure

1.2 shows a model calculation illustrating that a significant fraction (here, 60 %) of NOx present at sundown can be removed via N2O5 reaction at night.

Figure 1.2 Box model results showing the fraction of NOx lost at night due to N2O5 and NO3 reaction.

1.2 N2O5 Hydrolysis

E.g.5–12 N2O5 hydrolysis has been studied on a variety of surfaces and the accepted mechanism is shown in Reactions 1.1-1.6.7,13,14

N2O5 (g) ↔ N2O5 (aq) (R1.1)

+ - N2O5 (aq) + H2O(l) ↔ H2ONO2 (aq) + NO3 (aq) (R1.2)

+ + H2ONO2 (aq) + H2O (l) ↔ H3O (aq) + HNO3 (aq) (R1.3)

+ - H2ONO2 (aq) + X ↔ XNO2 (aq) +H2O (l) (R1.4)

XNO2 (aq) ↔ XNO2 (g) (R1.5)

After N2O5(g) accommodation to the particle surface, autoionization allows for

+ - formation of nitronium ion (NO2 ) and nitrate (NO3 ), shown in Reactions 1.1-1.2. If the particle

+ - + contains a high concentration of nitrate (where k(NO2 + NO3 ) > k(NO2 + H2O)) the ionization

13,14 reaction will be driven in the reverse direction. For example, the probability of N2O5 reaction

- with aerosol decreases by a factor of 90 when [NO3 ] aerosol loading increases from 13.4 to 158 mol/kg.13–15 The fate of the nitronium ion is determined by the composition of the particle.19,20

+ NO2 will either react with water, forming nitric acid (Reaction 1.3) or, if the particle contains a halide ion (X-), will react to form a nitryl halide species.8,16–18 The nitryl halide species, depending on its Henry's Constant value (KH), can partition out of the aqueous phase into the gas phase. It is important to note that Reaction 1.3 represents a terminal loss of NOx in the nocturnal atmosphere.

1.3 The Reactive Uptake Probability, (N2O5)

When N2O5 collides with a surface, it does not necessarily react. The ability of N2O5 to undergo hydrolysis within a particle depends strongly on the composition of the particle, as will be described in the next section. The metric used to determine the extent of N2O5 reaction with a surface is given by the reactive uptake coefficient, (N2O5), which can also be thought of as the probability of an N2O5 molecule colliding with a surface and reacting (Equation 1.1):

(E1.1)

khet is the pseudo first order rate coefficient for N2O5 loss to particles,  is the mean molecular velocity of N2O5, and SA is the surface area of the particle population with which N2O5 is reacting. Within this expression,  accounts for a number of physical and chemical processes that

5 are involved in the "uptake" of a gas to a surface, including diffusion and gas-surface collision frequency, the mass accommodation involved in the gas crossing the interface, solubilization and diffusion into the bulk, and reaction within the particle.2 Laboratory measurements suggest that

N2O5 hydrolyzes within the first 50 nm of the air-particle interface, and thus is considered more a

6,7 surface rather than bulk reaction. As such N2O5 reactivity is strongly affected not only by the chemical composition of particles,19,20 but is acutely sensitive to the composition at the surface layer, where the reaction takes place. The chemical composition of particles influencing (N2O5) is a primary focus of this dissertation.

1.4 Particle Chemical Composition and N2O5 Reactive Uptake

1.4.1 The Impact of Chemical Composition: Field Measurements of (N2O5)

Ambient measurements of (N2O5) have been reported from a number of areas within the

United States and show a wide range of variability in measured (N2O5). Ambient reactive uptake studies in a polluted coastal location showed significant day to-day (N2O5) variability (<0.001-

0.029), which the authors attribute to varying levels of particulate nitrate content.21 Wagner et al. observed a similar reliance on nitrate loadings during wintertime studies in a continental region,

22 where they measured (N2O5) range between 0.02-0.1. Bertram et al. made direct measurements of (N2O5) on ambient aerosol particles, correlating the values to the particulate organic matter

(POM).20 They discovered that during times of low POM, the ambient relative humidity and particulate nitrate concentration regulated (N2O5), while during periods of high relative humidity, (N2O5) was controlled by the relative amounts of POM:sulfate, with higher gamma values resulting from lower POM concentrations. In these studies, (N2O5) spanned from <0.01 to

>0.03. Similarly, Brown and coworkers observed a 10× decrease (<0.001- 0.017) in (N2O5) correlated with organic and sulfate loadings during studies in the Northeastern United States.19

This summary of recent studies illustrates the significant variability in ambient (N2O5) measurements, and indicates a correlation with a variety of compositional factors. Laboratory studies provide a more constrained measure of the impact of each of these components influencing (N2O5). The laboratory studies detailed below focus on the impact of salts and organics on N2O5 reactive uptake.

1.4.2 Uptake to Inorganic Salts: Laboratory Studies

(N2O5) has been measured on a variety of substrates in the laboratory, with NaCl and seawater providing the most relevant matrices for the marine environment and this dissertation.

Stewart et al. measured uptake to particles generated from both NaCl and artificial seawater at

-3 -2 varying relative humidities (RH), finding that (N2O5) spans a range from 1.5×10 to 1.4×10 for

RH 30-80 %.6 Similar measurements by McNeill and coworkers report a gamma range of

1.5×10-2 to 3.4×10-2 on NaCl and laboratory-generated natural seawater aerosol for RH of

51-71 %.5 These results are in good agreement, however do not account for the range of values measured during ambient studies.

1.4.3 The Impact of Particulate Organics: Laboratory Studies

The presence of organics in aerosol particles can create a physical barrier for N2O5, or hinder its diffusion through the organic phase to the aqueous core. Many groups have performed laboratory study investigations into the effect of a variety of organics on N2O5 reactive uptake

-5 -2 5,8,10,15,23–26 kinetics, finding (N2O5) ranging from <2×10 to >1.5×10 . Folkers et al. found

(N2O5) was reduced by a factor of 35 when sodium sulfate particles were coated with 93% volume fraction a-pinene oxidation products, and a factor of 45 reduction when the particles were comprised only of the oxidation product, indicating the reduction in (N2O5) is due to the presence of an organic film.10 Escorcia and coworkers coated ammonium bisulfate particles with

-pinene oxidation products, increasing the organic mass of the particles up to 79 %.15 Their

results show an order of magnitude decrease in (N2O5) when 9 % of the particle mass was organic, and smaller decreases following this, suggesting that the presence of an organic layer on the particle either reduces the mass accommodation coefficient or hinders the dissolution and diffusion of N2O5 into the particle bulk following accommodation. Cosman and coworkers looked into the effect of straight and branched surfactants, finding that upon coating H2SO4 particles with straight chain organics of increasing length, a 17-61× reduction in (N2O5) was observed as

9,24 compared to uptake to uncoated H2SO4 particles. However, when phytanic acid, a branched organic surfactant, was used to coat the particles, no suppression in (N2O5) was observed, implying that straight chain organics pack efficiently on particle surfaces, preventing N2O5 from reaching the sub-phase, whereas branched organics are unable to pack in a tight layer, thus

9,24 allowing for N2O5 diffusion.

1.4.4 Laboratory and Field Measurement Agreement

Despite these ambient and laboratory studies, for complex particles such as sea spray aerosol (SSA), the factors controlling (N2O5) remain largely unclear. Ocean particles contain a complex mix of inorganic and organic species stemming from ocean biology, however the identity and subsequent impact of organic species that transition from the sea surface to SSA, which can account for 17- 83 % of sub-micrometer aerosol mass,27–29 are not well constrained and remain an area of current research.

19,21 It is not understood whether the variability seen in coastal (N2O5) measurements is due to nascent SSA compositional variability or due to processing of particles post-production.

While ambient studies have investigated reactive uptake in coastal marine environments,19,21 it is impossible to isolate ocean-generated particles from other aerosol in the atmosphere such as dust, biomass burning particles etc. Laboratory studies often utilize NaCl or artificial sea salt matrices and coat the particles with organic proxies, such as sodium dodecyl sulfate,5 phytanic acid,9,24 and

8 oxidation products of biogenic gases10,15 however, without a baseline ambient measurement of pure SSA, it is challenging to elucidate the forces driving (N2O5) variability and magnitude.

Additionally, many of the laboratory proxies used are not representative of the natural organics present in ocean water.

1.5 Product Formation Following N2O5 Reaction on Chloride-Containing Surfaces

As mentioned above, when N2O5 dissociates within a particle containing a halide ion, a nitryl halide species is formed. In the marine environment, particles of oceanic origin contain Cl-,

8,16–18 and thus nitryl chloride (ClNO2) is produced following N2O5 uptake. ClNO2 is initially produced in the aqueous phase (R1.4), and with a Henry's law constant of 2.4 - 4.6×10-2 mol L-

1 -1 17,30 atm , efficiently partitions out into the gas phase. In the presence of sunlight, ClNO2 undergoes photolysis, with a lifetime of only a few hours after sunrise, resulting in the formation of chlorine radicals and recycling NOx back into the atmosphere (Reaction 1.6).31,32

ClNO2(g) + h  Cl(g) + NO2(g) (R1.6)

Unlike Reaction 3, by proceeding through this reaction pathway, N2O5 reactive uptake is no longer providing a terminal sink for NOx species, but rather regenerating NO2 in the atmosphere and creating highly oxidizing chlorine radicals.31 In fact, the estimated global contribution of Cl to the atmosphere from this reaction alone is 8-22 Tg yr-1.31

The amount of ClNO2 produced following reactive uptake of N2O5 to a particle surface is determined by calculating the yield of ClNO2, (ClNO2) given by Equation 1.2,

(E1.2)

where the yield of ClNO2 is given by the change in the amount of ClNO2 relative to the change in the amount of N2O5 present. Laboratory studies have shown (ClNO2) is directly related to the

12,17,33 amount of chloride present in a particle. Roberts et al. noted (ClNO2) ranges between 0.2-

0.8 for NaCl particles with [Cl-] = 0.02-0.5 M,12 while Bertram and Thornton measured the yield

- 33 from mixed NaNO3/NaCl particles as approaching 1 for [Cl ] >1 M. Figure 1.3 summarizes a number of laboratory measurements of (ClNO2) on chloride-containing particles. It is important to note that these studies generally agree that a 0.5 M Cl- concentration (the concentration of chloride in the ocean) produces a yield of 0.8. Though these studies are in agreement, ambient particles are complex mixtures of chemical components, and thus show a wide variability in

(ClNO2) upon reaction with N2O5.

- Figure 1.3 ClNO2 yield following N2O5 reaction as a function of [Cl ] in solution from which particles were atomized. The results shown are from 3 separate studies.

1.5.1 Ambient Measurements of ClNO2 and (ClNO2)

In 2008, Osthoff and coworkers reported the first ambient observation of ClNO2, finding that in the Gulf of Mexico marine boundary layer, ClNO2 correlated with N2O5 at night, and was

10 produced in abundances up to 1 ppb.34 However, in an urban environment, Mielke et al. found peak ClNO2 levels only reached 0.25 ppb, indicating that reaction with particulate chloride was a

35 minor product of N2O5 uptake compared to heterogeneous hydrolysis. Additionally, ship-based measurements sailing up the coast of California resulted in a maximum [ClNO2] = 2.1 ppb when the air mass was influenced by Los Angeles outflow,36 yet ship-based measurements in the Long

Island Sound resulted in average ClNO2 measurements between 150-250 ppt, while hundreds of miles from the coast in the North Atlantic, levels were between 10-100 ppt.37 A recent study measured vertical profiles of ClNO2 and N2O5 along with particle composition analysis in a continental region. During two pollution plume events, they found ClNO2 levels reach 0.6 and 1.3 ppb and estimated (ClNO2) between 0.3-1, almost the entire allowable range, stating the reason for this large range is uncertain.

1.6 Summary and Remaining Areas to Address

Understanding the driving forces of (N2O5) variability and constraining (N2O5) in the atmosphere is important, as the value used in models to predict NOx removal at night is highly

20 dependent on the (N2O5) value used, as is shown in Figure 1.4 adapted from Bertram et al.

Here it is evident that the choice of (N2O5) not only impacts NOx removal from 0 to 50 % for gamma values ranging from 10-5 to 10-1 , but also has a profound influence on chlorine radical production following ClNO2 generation and photolysis. This has important implications for regional climate models that include heterogeneous chemical reactions in their predications of atmospheric oxidant loadings and air quality.

Figure 1.4 Model calculations illustrating the dependence of NOx loss (left y axis) and chlorine radical production following photolysis of ClNO2 (right y axis) on the magnitude of g(N2O5). This plot has been adapted from Bertram et al. 2009.20

As is illustrated above, laboratory and ambient measurements span almost the entire (N2O5) range shown in Figure 1.4, thus constraining our understanding particulate compositional drivers on (N2O5) remains necessary.

Despite the strives made in this area and the detailed investigations into (N2O5) and

(ClNO2), critical areas still remain to be addressed:

1) What is the role of particle mixing state on (N2O5)? Many model parameterizations and heterogeneous reaction kinetics calculations assume particles are homogeneous within a population, meaning they all exhibit some average chemical composition. However, in the ambient atmosphere, particles are not necessarily of an average composition, especially near large aerosol sources, such as the ocean. The impact of mixing state and the effect of average chemical composition assumptions has yet to be addressed.

2) What is (N2O5) on fresh nascent sea spray that is unperturbed by oxidants and other atmospheric influences? Further, how do biological conditions in the ocean impact (N2O5) by altering the chemical composition of SSA, and do the organics present in SSA act as efficient

surfactants, pack tightly, and reduce (N2O5)? To date the literature lacks direct studies of (N2O5) on nascent sea spray aerosol that allow for assessment of the role that organics on the surface of

SSA may play in N2O5 uptake kinetics.

3) How do organics present in the top layers of the ocean impact the ability of the ocean to act as an efficient source of ClNO2 following reaction with N2O5? Additionally, are there inorganic mechanisms that might influence (ClNO2) from the ocean surface?

1.7 Synopsis of Chapters

This dissertation addresses the disconnect in ambient and laboratory measurements of

(N2O5) and (ClNO2), utilizing ambient field data, laboratory measurements, real ocean water systems and representative sea spray generation mechanisms. Specifically this dissertation seeks to investigate the three unaddressed areas listed above.

Chapter 2 details investigations into the role of aerosol particle mixing in the marine boundary layer on (N2O5) by correlating (N2O5) measurements made in a polluted coastal environment with chloride-to-nitrate particulate loadings in the first simultaneous (N2O5) ambient observations with co-located single particle chemical composition. This work compares model predictions of (N2O5) in both polluted and clean coastal conditions, suggests controlling factors, and provides a framework for assessment of the role of mixing state in N2O5 reactive uptake to complex ambient aerosol.

In Chapter 3, the role of biological ocean conditions is investigated in a laboratory study, specifically probing the effect of organics during a controlled synthetic mesocosm and the impact they have on SSA and (N2O5). The organic volume fractions of the particles generated from the synthetic mesocosm are measured, and the data is correlated to the observed (N2O5) values.

Additionally, real ocean water is used to generate nascent SSA and the associated (N2O5) values are used for comparison with synthetic mesocosm-derived gammas.

Chapters 4 and 5 provide laboratory measurements of laboratory proxy and natural seawater from multiple locations and during a phytoplankton bloom to constrain the role of organics (chapter 4) and inorganics (chapter 5) in controlling the production of ClNO2 from the ocean surface, which was found in previous literature, to be minimal. The organic investigation utilizes the ratio of reactive organic species to chloride at the air-water interface to illustrate how

+ organic enhancement at the ocean surface could be responsible for titrating NO2 , resulting in low to minimal (ClNO2). Chapter 5 similarly investigates the role of inorganic ions in the reduction

2+ of ClNO2 production from the ocean surface, focusing on the Mg effects.

1.8 References

(1) Seinfeld, J. H.; Pandis, S. N. Atmospheric Chemistry and Physics: From Air Pollution to Climate Change, 2nd Edition; Wiley-Interscience, 2006.

(2) Finlayson-Pitts, B. J.; Jr, J. N. P. Chemistry of the Upper and Lower Atmosphere: Theory, Experiments, and Applications; Academic Press, 1999.

(3) Mozurkewich, M.; Calvert, J. G. Reaction Probability of N2O5 on Aqueous Aerosols. J. Geophys. Res. Atmospheres 1988, 93 (D12), 15889–15896.

(4) Morris, E. D.; Niki, H. Reaction of Dinitrogen Pentoxide with Water. J. Phys. Chem. 1973, 77 (16), 1929–1932.

(5) McNeill, V. F.; Patterson, J.; Wolfe, G. M.; Thornton, J. A. The Effect of Varying Levels of Surfactant on the Reactive Uptake of N2O5 to Aqueous Aerosol. Atmos Chem Phys 2006, 6 (6), 1635–1644.

(6) Stewart, D. J.; Griffiths, P. T.; Cox, R. A. Reactive Uptake Coefficients for Heterogeneous Reaction of N2O5 with Submicron Aerosols of NaCl and Natural Sea Salt. Atmos Chem Phys 2004, 4 (5), 1381–1388.

(7) Thornton, J. A.; Braban, C. F.; Abbatt, J. P. D. N2O5 Hydrolysis on Sub-Micron Organic Aerosols: The Effect of Relative Humidity, Particle Phase, and Particle Size. Phys. Chem. Chem. Phys. 2003, 5 (20), 4593–4603.

(8) Thornton, J. A.; Abbatt, J. P. D. N2O5 Reaction on Submicron Sea Salt Aerosol: Kinetics, Products, and the Effect of Surface Active Organics. J. Phys. Chem. A 2005, 109 (44), 10004–10012.

(9) Cosman, L. M.; Bertram, A. K. Reactive Uptake of N2O5 on Aqueous H2SO4 Solutions Coated with 1-Component and 2-Component Monolayers. J. Phys. Chem. A 2008, 112 (20), 4625–4635.

(10) Folkers, M.; Mentel, T. F.; Wahner, A. Influence of an Organic Coating on the Reactivity of Aqueous Aerosols Probed by the Heterogeneous Hydrolysis of N2O5. Geophys. Res. Lett. 2003, 30 (12), 1644.

(11) Griffiths, P. T.; Badger, C. L.; Cox, R. A.; Folkers, M.; Henk, H. H.; Mentel, T. F. Reactive Uptake of N2O5 by Aerosols Containing Dicarboxylic Acids. Effect of Particle Phase, Composition, and Nitrate Content. J. Phys. Chem. A 2009, 113 (17), 5082–5090.

(12) Roberts, J. M.; Osthoff, H. D.; Brown, S. S.; Ravishankara, A. R.; Coffman, D.; Quinn, P.; Bates, T. Laboratory Studies of Products of N2O5 Uptake on Cl− Containing Substrates. Geophys. Res. Lett. 2009, 36 (20), L20808.

(13) Wahner, A.; Mentel, T. F.; Sohn, M.; Stier, J. Heterogeneous Reaction of N2O5 on Sodium Nitrate Aerosol. J. Geophys. Res. Atmospheres 1998, 103 (D23), 31103–31112.

(14) Mentel, T. F.; Sohn, M.; Wahner, A. Nitrate Effect in the Heterogeneous Hydrolysis of Dinitrogen Pentoxide on Aqueous Aerosols. Phys. Chem. Chem. Phys. 1999, 1 (24), 5451– 5457.

(15) Escorcia, E. N.; Sjostedt, S. J.; Abbatt, J. P. D. Kinetics of N2O5 Hydrolysis on Secondary Organic Aerosol and Mixed Ammonium Bisulfate−Secondary Organic Aerosol Particles. J. Phys. Chem. A 2010, 114 (50), 13113–13121.

(16) Brown, S. S.; Ryerson, T. B.; Wollny, A. G.; Brock, C. A.; Peltier, R.; Sullivan, A. P.; Weber, R. J.; Dubé, W. P.; Trainer, M.; Meagher, J. F.; Fehsenfeld, F. C.; Ravishankara, A. R. Variability in Nocturnal Nitrogen Oxide Processing and Its Role in Regional Air Quality. Science 2006, 311 (5757), 67–70.

(17) Bertram, T. H.; Thornton, J. A.; Riedel, T. P.; Middlebrook, A. M.; Bahreini, R.; Bates, T. S.; Quinn, P. K.; Coffman, D. J. Direct Observations of N2O5 Reactivity on Ambient Aerosol Particles. Geophys. Res. Lett. 2009, 36 (19), L19803.

(18) Finlayson-Pitts, B. J.; Ezell, M. J.; Pitts, J. N. Formation of Chemically Active Chlorine Compounds by Reactions of Atmospheric NaCl Particles with Gaseous N2O5 and ClONO2. Nature 1989, 337 (6204), 241–244.

(19) Behnke, W.; George, C.; Scheer, V.; Zetzsch, C. Production and Decay of ClNO2 from the Reaction of Gaseous N2O5 with NaCl Solution: Bulk and Aerosol Experiments. J. Geophys. Res. Atmospheres 1997, 102 (D3), 3795–3804. (20) Francis Schweitzer, P. M. Multiphase Chemistry of N2O5, ClNO2, and BrNO2. J. Phys. Chem. A 1998, 102 (22).

(21) Riedel, T. P.; Bertram, T. H.; Ryder, O. S.; Liu, S.; Day, D. A.; Russell, L. M.; Gaston, C. J.; Prather, K. A.; Thornton, J. A. Direct N2O5 Reactivity Measurements at a Polluted Coastal Site. Atmos Chem Phys 2012, 12 (6), 2959–2968.

(22) Wagner, N. L.; Riedel, T. P.; Young, C. J.; Bahreini, R.; Brock, C. A.; Dubé, W. P.; Kim, S.; Middlebrook, A. M.; Öztürk, F.; Roberts, J. M.; Russo, R.; Sive, B.; Swarthout, R.; Thornton, J. A.; VandenBoer, T. C.; Zhou, Y.; Brown, S. S. N2O5 Uptake Coefficients and Nocturnal NO2 Removal Rates Determined from Ambient Wintertime Measurements. J. Geophys. Res. Atmospheres 2013, 118 (16), 9331–9350.

(23) Anttila, T.; Kiendler-Scharr, A.; Tillmann, R.; Mentel, T. F. On the Reactive Uptake of Gaseous Compounds by Organic-Coated Aqueous Aerosols: Theoretical Analysis and Application to the Heterogeneous Hydrolysis of N2O5. J. Phys. Chem. A 2006, 110 (35), 10435–10443.

(24) Cosman, L. M.; Knopf, D. A.; Bertram, A. K. N2O5 Reactive Uptake on Aqueous Sulfuric Acid Solutions Coated with Branched and Straight-Chain Insoluble Organic Surfactants. J. Phys. Chem. A 2008, 112 (11), 2386–2396.

(25) Knopf, D. A.; Cosman, L. M.; Mousavi, P.; Mokamati, S.; Bertram, A. K. A Novel Flow Reactor for Studying Reactions on Liquid Surfaces Coated by Organic Monolayers: Methods, Validation, and Initial Results. J. Phys. Chem. A 2007, 111 (43), 11021–11032.

(26) Park, S.-C.; Burden, D. K.; Nathanson, G. M. The Inhibition of N2O5 Hydrolysis in Sulfuric Acid by 1-Butanol and 1-Hexanol Surfactant Coatings. J. Phys. Chem. A 2007, 111 (15), 2921–2929.

(27) Facchini, M. C.; Rinaldi, M.; Decesari, S.; Carbone, C.; Finessi, E.; Mircea, M.; Fuzzi, S.; Ceburnis, D.; Flanagan, R.; Nilsson, E. D.; de Leeuw, G.; Martino, M.; Woeltjen, J.; O’Dowd, C. D. Primary Submicron Marine Aerosol Dominated by Insoluble Organic Colloids and Aggregates. Geophys. Res. Lett. 2008, 35 (17), L17814.

(28) O’Dowd, C. D.; Facchini, M. C.; Cavalli, F.; Ceburnis, D.; Mircea, M.; Decesari, S.; Fuzzi, S.; Yoon, Y. J.; Putaud, J.-P. Biogenically Driven Organic Contribution to Marine Aerosol. Nature 2004, 431 (7009), 676–680.

(29) Prather, K. A.; Bertram, T. H.; Grassian, V. H.; Deane, G. B.; Stokes, M. D.; DeMott, P. J.; Aluwihare, L. I.; Palenik, B. P.; Azam, F.; Seinfeld, J. H.; Moffet, R. C.; Molina, M. J.; Cappa, C. D.; Geiger, F. M.; Roberts, G. C.; Russell, L. M.; Ault, A. P.; Baltrusaitis, J.; Collins, D. B.; Corrigan, C. E.; Cuadra-Rodriguez, L. A.; Ebben, C. J.; Forestieri, S. D.; Guasco, T. L.; Hersey, S. P.; Kim, M. J.; Lambert, W. F.; Modini, R. L.; Mui, W.; Pedler, B. E.; Ruppel, M. J.; Ryder, O. S.; Schoepp, N. G.; Sullivan, R. C.; Zhao, D. Bringing the Ocean into the Laboratory to Probe the Chemical Complexity of Sea Spray Aerosol. Proc. Natl. Acad. Sci. 2013, 110 (19), 7550–7555.

(30) Frenzel, A.; Scheer, V.; Sikorski, R.; George, C.; Behnke, W.; Zetzsch, C. Heterogeneous Interconversion Reactions of BrNO2, ClNO2, Br2, and Cl2. J. Phys. Chem. A 1998, 102 (8), 1329–1337.

(31) Thornton, J. A.; Kercher, J. P.; Riedel, T. P.; Wagner, N. L.; Cozic, J.; Holloway, J. S.; Dubé, W. P.; Wolfe, G. M.; Quinn, P. K.; Middlebrook, A. M.; Alexander, B.; Brown, S.

S. A Large Atomic Chlorine Source Inferred from Mid-Continental Reactive Nitrogen Chemistry. Nature 2010, 464 (7286), 271–274.

(32) Nelson, H. H.; Johnston, H. S. Kinetics of the Reaction of Chlorine Atoms with Nitrosyl Chloride and Nitryl Chloride and the Photochemistry of Nitryl Chloride. J. Phys. Chem. 1981, 85 (25), 3891–3896.

(33) Bertram, T. H.; Thornton, J. A. Toward a General Parameterization of N2O5 Reactivity on Aqueous Particles: The Competing Effects of Particle Liquid Water, Nitrate and Chloride. Atmos Chem Phys 2009, 9 (21), 8351–8363.

(34) Osthoff, H. D.; Roberts, J. M.; Ravishankara, A. R.; Williams, E. J.; Lerner, B. M.; Sommariva, R.; Bates, T. S.; Coffman, D.; Quinn, P. K.; Dibb, J. E.; Stark, H.; Burkholder, J. B.; Talukdar, R. K.; Meagher, J.; Fehsenfeld, F. C.; Brown, S. S. High Levels of Nitryl Chloride in the Polluted Subtropical Marine Boundary Layer. Nat. Geosci. 2008, 1 (5), 324–328.

(35) Mielke, L. H.; Furgeson, A.; Osthoff, H. D. Observation of ClNO2 in a Mid-Continental Urban Environment. Environ. Sci. Technol. 2011, 45 (20), 8889–8896.

(36) Riedel, T. P.; Bertram, T. H.; Crisp, T. A.; Williams, E. J.; Lerner, B. M.; Vlasenko, A.; Li, S.-M.; Gilman, J.; de Gouw, J.; Bon, D. M.; Wagner, N. L.; Brown, S. S.; Thornton, J. A. Nitryl Chloride and Molecular Chlorine in the Coastal Marine Boundary Layer. Environ. Sci. Technol. 2012, 46 (19), 10463–10470.

(37) Kercher, J. P.; Riedel, T. P.; Thornton, J. A. Chlorine Activation by N2O5: Simultaneous, in Situ Detection of ClNO2 and N2O5 by Chemical Ionization Mass Spectrometry. Atmos Meas Tech 2009, 2 (1), 193–204.

Chapter 2 On The Role of Particle Inorganic Mixing State in The Reactive Uptake of N2O5 to Ambient Aerosol Particles

2.1 Abstract

The rates of heterogeneous reactions of trace gases with aerosol particles are complex functions of particle chemical composition, morphology, and phase state. Currently, the majority of model parameterizations of heterogeneous reaction kinetics focus on the population average of aerosol particle mass, assuming individual particles have the same chemical composition as the average state. Here, we assess the impact of particle mixing state on heterogeneous reaction kinetics using the N2O5 reactive uptake coefficient, (N2O5), dependence on the particulate

- - chloride-to-nitrate ratio (nCl /nNO3 ). We describe the first simultaneous ambient observations of single particle chemical composition and in situ determinations of (N2O5). When accounting for

- - particulate nCl /nNO3 mixing state, model parameterizations of (N2O5) continue to over predict

(N2O5) by more than a factor of two in polluted coastal regions, suggesting that chemical composition and physical phase state of particulate organics likely control (N2O5) in these air masses. In contrast, direct measurement of (N2O5) in air masses of marine origin are well captured by model parameterizations and reveal limited suppression of (N2O5), indicating the organic mass fraction of fresh sea spray aerosol at this location does not suppress (N2O5). We provide an observation-based framework for assessing the impact of particle mixing state on gas- particle interactions.

2.2 Introduction

The production rates of tropospheric ozone (O3) and secondary organic aerosol (SOA) are dependent on the abundance of nitrogen oxides (NOx), the concentration and speciation of volatile organic compounds (VOC), and oxidant loadings (e.g., OH and Cl). Accurate assessment

of the dependence of O3 and SOA production on NOx emission rates requires a detailed description of NOx removal mechanisms and the coupling of these processes with other oxidants.

Laboratory measurements, confirmed by field observations, have shown that the heterogeneous reaction of dinitrogen pentoxide (N2O5) with aerosol particles serves as an efficient loss mechanism for NOx, where the fraction of emitted NOx that is removed from the atmosphere via this pathway is estimated to be as large as 50% by current chemical transport models.1 Further, the reaction of N2O5 on chloride containing particles results in the production of nitryl chloride

(ClNO2), that has been estimated to account for as much as 60% of the total chlorine radical

2 production rate in polluted coastal regions. Despite both the importance of N2O5 chemistry to oxidant loadings and an abundance of detailed mechanistic studies of the reaction mechanism on single-component particle surrogates in the laboratory, accurate parameterization of N2O5 reactive uptake in the atmosphere remains challenging.3

2.2.1 Factors Controlling the Reactive Uptake of N2O5 to Aerosol Particles

1,4 Direct measurement of the reactive uptake coefficient, (N2O5), to ambient aerosol, and determinations of (N2O5) through steady-state analysis indicate that parameterizations do not

3,5 capture the observed magnitude and variability in (N2O5). Model-measurement disagreement is most pronounced in regions characterized by large chemical heterogeneity within the particle

3 population, where parameterizations routinely overestimate (N2O5). This disagreement arises from the dependence of (N2O5) on single particle chemical composition, physical phase state, morphology, and liquid water content.

Several key components of the N2O5 reaction mechanism have been defined: 1) the presence of organic films/coatings6,7 and monolayer surfactants8–11 has been shown to suppress

12 + the rate of N2O5 uptake and 2) the competition for the N2O5 dissociation product NO2 , between

- - strong nucleophiles (e.g., H2O and Cl ) and the nitrate ion (NO3 ), can act to accelerate or suppress

13–15 (N2O5), respectively. Both competition processes highlight how the distribution of chemicals within a single particle or particle population can impact reaction mechanisms, as the reaction rate depends on the single particle composition, not the ensemble average concentrations.

2.2.2 Single Particle Chloride-to-Nitrate Mixing State in The Polluted Marine Boundary

Layer

The chemical compositions of individual particles within a population of particles have varying degrees of chemical similarity to the population average. In one extreme, all particles are chemically identical to the population average (internally mixed), and in the other extreme, individual particles, or size regimes of particles, have unique chemical signatures, which differ from the mean state (externally mixed).16 The extent to which particles are internally or externally mixed within any given air mass is a function of particle source and chemical and physical processing (i.e. heterogeneous reactions, condensation, etc.) post-emission or formation.

Typically, particle populations are externally mixed near source regions, where primary aerosol particles are emitted to the atmosphere with characteristic size and chemical composition unique to the source mechanism.17,18 Within a Lagrangian framework, as the air mass evolves in time away from the source region, it is thought that the particle population will tend toward an internal mixture as: 1) the semi-volatile components of the primary particles redistribute themselves among the entire population,19 2) low-volatility products of gas-phase reactions condense onto existing surfaces20 and 3) coagulation and deposition act to consolidate particles into the

18,21 accumulation mode (0.08 µm < particle diameter (dp) < 2.5 µm). While a trajectory toward an internally mixed state is consistent with a particle population dominated by semi-volatile organic compounds, particle morphology and physical phase state could act to slow the approach toward an internally mixed state, and the presence of organic and inorganic acids and bases can result in pH differences between particles of different sizes and limit the formation of an internally mixed

20 population.22,23 Further, a particle population within a stagnant fixed coordinate system, where particles of specific composition are continuously added, such as in the marine boundary layer

(MBL), can remain as an external mixture when there is a kinetic or diffusion limitation to gas- particle exchange, as particles within the population are of different ages.

Heterogeneous and multiphase reactions occurring on or within sea spray aerosol are of particular interest in atmospheric chemistry as they can serve as halogen activation mechanisms.

Here, we discuss the activation of particulate chloride to ClNO2 which rapidly photolyzes to produce Cl radicals. Nascent sea spray aerosol (SSA) is heavily enriched in Cl- with respect to

– - - 4 16 NO3 due to the high molar chloride-to-nitrate ratio (nCl /nNO3 ) of seawater ( > 3×10 ). The chemical processing of SSA by nitric acid (HNO3) and N2O5 in the polluted MBL results in rapid

- - - - depletion of particulate Cl and accumulation of NO3 until nCl /nNO3 achieves equilibrium with gas-phase HNO3 and HCl (R1-R2).

  Cl aq HNO3 gHCl g  NO3 aq (R1)

  Cl aq  N2O5 g ClNO2 g  NO3 aq (R2)

Given the volatility of both HCl and HNO3, and that each particle is exposed to the same

- - HCl/HNO3 gas-phase ratio, it is expected that nCl /nNO3 would be internally mixed within the particle population and in equilibrium with the gas-phase. However, concurrent field observations of particle and gas-phase nitrate and chloride in the polluted boundary layer have shown nCl-

- /nNO3 is often larger on super-micrometer particles as compared with sub-micrometer particles and the super-micrometer component is often not in equilibrium with the gas-phase. 22,24–26 This result suggests that either: 1) inter-particle variability in pH or liquid water content acts to sustain

- - particle-particle variability in nCl /nNO3 , 2) the approach to equilibrium is slowed due to the

21 presence of organic films, 3) fast production and size-dependent deposition compete on the time- scale required to establish equilibrium, or 4) HNO3 reactions on or within super-micrometer particles are diffusion-limited.27

- - In what follows, we use direct measurements of single particle nCl /nNO3 and ambient

- - determinations of (N2O5) to assess the impact of nCl /nNO3 mixing state on (N2O5) and the potential use for single particle measurements to better constrain parameterizations of (N2O5).

The results obtained here provide an upper limit on the role of inorganic mixing state in determining (N2O5), where any remaining differences between observation and parameterization likely result from the direct impact of organic coatings and/or their control over particle liquid

H2O.

2.3 Materials and Methods

Ambient measurements of single particle chemical composition and in situ

1 determinations of N2O5 reactivity were conducted simultaneously between October 4th-6th 2009 from the Scripps Institution of Oceanography (SIO) Pier. The flow reactor apparatus and

28 modulation technique implemented for determining (N2O5) have been described previously.

Briefly, ambient air was sampled through a 0.25" ID stainless steel inlet either directly into the flow reactor, or via a Teflon filter, providing 99% particle removal from the air stream, into a 5.9"

ID stainless steel tube, 35" in length. N2O5 was produced in situ via the reaction of excess NO2 with O3, created via photolysis of O2, in a secondary mixing cell. N2O5 was then injected into the reactor, allowing for reaction of the gas with ambient aerosol. The mixing ratio of N2O5 at the top of the flow reactor was ca. 1 ppbv, and was monitored using a chemical ionization mass spectrometer. In this application, an Aerosol Time of Flight Mass Spectrometer (ATOFMS)29 sampled both ambient aerosol as well as ambient aerosol that had been reacted with N2O5 permitting analysis of the extent of chloride depletion on a single particle level.

2.4 Results and Discussion

2.4.1 Impact of Particulate Chloride-to-Nitrate Mixing State on (N2O5)

The impact of particle mixing state on heterogeneous and multi-phase reactions is most pronounced when the property of interest is a nonlinear function of the ratio of two chemical components, resulting in different ensemble average kinetics for the same average chemical

30 composition. Recently, Bertram and Thornton suggested that (N2O5), in the absence of organic coatings, could be parameterized as:

   

'  1   N O  Ak2 f 1  (E1) 2 5  k H O(l)  k Cl      3 2   1  4             k2b NO3    k2b NO3  

+ - where the aqueous phase reactions are described by k2f for N2O5 + H2O, k2b for NO2 + NO3 , k3

+ + for NO2 + H2O, and k4 for a halide with NO2 . Additionally, A is an empirical pre-factor that includes V (the total particle volume concentration), Sa (the total particle surface area concentration), the mean molecular speed of  and KH (the dimensionless Henry’s law coefficient (KH ≡[N2O5]aq /[N2O5]g )). The parameterization highlights the dependence of (N2O5)

- - - on both particle nH2O/nNO3 and nCl /nNO3 , and suggests that (N2O5) will respond non-linearly

- - - - to changes in nCl /nNO3 and that individual particles with high nCl /nNO3 will be more reactive.

A hypothetical graphical representation of the (N2O5) parameterization is shown in Fig.

- - - 2.1A, where nascent SSA are emitted at high nH2O/nNO3 and nCl /nNO3 (red area/upper right), and exhibit large (N2O5) in the absence of organic films. As nascent SSA is chemically

- - processed in the atmosphere, nCl /nNO3 decreases and the rate of N2O5 uptake slows (lower left).

The impact of the particle mixing state assumption on the parameterized (N2O5) is shown in Fig.

2.1B, using the probability density function (PDF) for two hypothetical distributions. The PDF is employed as a means to examine both the extent of particle aging between populations and the degree of external mixing within a specific particle population. Calculation of (N2O5) for the particle population shown in Fig. 2.1B, can be conducted either: 1) on a single particle basis where E1 is applied to each individual particle in the population and the ensemble average

(N2O5) then calculated as the average of the single particle determinations (referred to here as

- - the externally mixed model) or 2) using the population ensemble average nCl /nNO3 (referred to here as the internally mixed model) to calculate the (N2O5). Comparison of these two model parameterizations to measured ambient values exposes the possible error that can result from assuming a particle population is internally mixed.

Calculation of (N2O5) for the externally mixed population shown in Fig. 2.1B, yields a

- population averaged reactive uptake coefficient of 0.025 (for nH2O/nNO3 = 3). In contrast,

- - - - assuming complete internal mixing of nCl /nNO3 (with same ensemble average nCl /nNO3 = 0.1) results in a 32% overestimation of (N2O5). The schematic representation highlights the potential importance for resolving particle mixing state, as gas-particle interactions depend on the surface and bulk chemical composition of single particles, which can be different from the ensemble average.

- Figure 2.1 A) Model parameterization of (N2O5), where nascent SSA are emitted at high nH2O/nNO3 (aq) - - and nCl /nNO3 , and exhibit large (N2O5) in the absence of organic films. As nascent SSA is chemically - - processed in the atmosphere, nCl /nNO3 decreases and the rate of N2O5 uptake slows. B) The impact of the particle mixing state assumption on the parameterized (N2O5) for two hypothetical distributions with the - - same ensemble average nCl /nNO3 . In this example, the internal mixing assumption results in a 32% overestimation of (N2O5)

In what follows, we use ambient measurements of single particle chemical composition retrieved from the ATOFMS as an input for the aforementioned (N2O5) parameterization. This permits calculation of (N2O5) under both internal and external mixing assumptions for

- comparison to the measured ambient value. The value of nH2O/nNO3 used in this analysis was determined for each case via the Aerosol Inorganics Model31, using measurements of ambient RH

- and ATOFMS derived chemical composition as input parameters. Values for nH2O/nNO3 ranged from 3.5 to greater than 6000. The parameterization of N2O5) under these conditions is

- relatively insensitive to the absolute value of nH2O/nNO3 . Error analysis is discussed in detail in

25 the supporting online information, however two important details regarding the application of the

ATOFMS data and the (N2O5) parameterization are as follows: 1) the parameterization limits (N2O5) to ca. 0.04, thus bounding calculation of the upper limit uncertainty, and 2) the

- - - - ATOFMS-measured single particle nCl /nNO3 using standard solutions of known Cl /NO3 is biased low, resulting in a reduced gamma value by at most 7% for the reported ambient pier measurements.

Figure 2.2 A) Backward air trajectories calculated for 24 hours prior to sampling for Period 1 and Period 2. B) Particle number concentration (N, left axis; see supporting information for sizing measurement details) - - and NOx (right axis) and C) Median nCl /nNO3 as determined by ATOFMS over the course of the two sampling periods (grey) with interquartile range (grey shaded region).

2.4.2 Observation-Based Constraints on the Impact of Chloride-to-Nitrate Particle

Mixing State on (N2O5)

We focus our ambient analysis on two specific air masses sampled from the Scripps Pier site during Fall 2009. To better understand the history of the air masses, HYSPLIT backwards trajectories were calculated for 24 hours prior to the sampling period at 500m above sea level using GDAS archived meteorological data.32,33 As shown in Fig. 2.2A, back trajectories calculated for Period 1 (October 3rd) were characterized by stagnant winds, where particles sampled from the SIO pier had significant interaction with coastal pollution. During this period, average relative humidity, measured at an adjacent coastal site, was 71%, NOx mixing ratios at the site averaged 10 ± 8 ppb, while sub-micrometer particle number concentration averaged 2477

± 23 particles cm-3 (Fig. 2.2B), consistent with previous literature values for a polluted marine boundary layer at this location.34 As expected, the median single particle measurement of nCl-

- /nNO3 (Fig. 2.2C) was 0.15 (interquartile range: 0.046 - 0.47), confirming that particles sampled during this period had undergone significant chemical processing. In contrast, 24 hour back trajectory analysis indicates strong westerly winds for the second sampling period (Period 2;

October 4th). As such, Period 2 was characterized by an average relative humidity of 64%, significantly lower NOx mixing ratios, 0.92 ± 0.5 ppb, particle concentrations, 864 ± 4 particles

-3 - - cm , and higher particulate nCl /nNO3 (0.63, interquartile range: 0.21 - 3.0) indicative of a less- polluted marine source for the air sampled. These values are summarized in Table 2.1. As

- expected, coincident in situ measurements of N2O5) reflect the average differences in nCl

- /nNO3 , where N2O5) was measured to be 0.012 ± 0.007 (1) for Period 1 (polluted air) and

0.037 ± 0.009 (1) for Period 2 (less-polluted marine air).

For the less-polluted marine conditions (Period 2), calculation of N2O5) using the

Bertram and Thornton30 parameterization (E1), assuming internal mixing, agrees well with the

‡ observed N2O5) (N2O5)internal mixing parameterization = 0.036 , as compared with

N2O5: N2O5: N2O5: Sub- Calculated Calculated Measurement Average micrometer Average Median Average m/Q Location Model; Model; NO Particle f RH x Cl/NO ROA 37/(37+43) Externally Internally (ppb) Conc. 3 3 Mixed Mixed (#/cm )

SIO Pier; 0.15, (iqr= 0.031 + 0.012 0.012 ± 2.69×10-3 more- 0.040‡ 71% 10 ± 8 2477 ± 23 0.046 - 0.448 ± 0.42 / - 0.008* 0.007 ±0.01 polluted 0.47)

SIO Pier; 0.63, (iqr 0.033 + 0.005 0.037 ± 0.92 ± 1.64×10-3 less- 0.036‡ 64% 864 ± 4 = 0.21 - 0.485 ± 0.42 / -0.006* 0.009 0.5 ±0.004 polluted 3.0)

- - *These error values account for both the accuracy and narrow distribution of nCl /nNO3 ‡Calculation of error values for these values results in error bounds that are not statistically different from the average value reported. This is due to the nature of the internally mixed calculation, and narrow - - distribution of nCl /nNO3 within this calculation. The absence of error bounds here however, does not represent the absolute accuracy of the value.

N2O5)meas = 0.037). Model-measurement agreement and N2O5) values consistent with

35 deliquesced NaCl (N2O5) = 0.03 ) suggest a more internally mixed aerosol population and, indirectly indicate, an insignificant role for organic films in suppressing heterogeneous reactions on nascent SSA, under the conditions sampled here. This result suggests that nascent SSA (at the surface area maximum) sampled at the SIO pier do not contain a significant organic film or surfactant coating that suppresses N2O5). While this is counter to previous laboratory suggestions conducted using straight chain organic model systems, 8,35 it is consistent with results obtained on branched organic chain systems. (Note: ‡ These error values indicate the narrow

- - distribution of nCl /nNO3 within the population, however the error does not represent the absolute accuracy of the value).

However, for polluted conditions, the internal mixing parameterization overestimates the

‡ observed N2O5) by a factor of 3 (N2O5); internal mixing parameterization = 0.040 , as

compared with N2O5)meas = 0.012). Overestimation of N2O5) for polluted conditions (Period 1)

- - can be attributed to either: 1) external mixing of Cl and NO3 or 2) a significant role for particulate organics in regulating particle liquid water concentration or access to it through the

1 - - formation of an organic film. Here, we use single particle measurements of Cl and NO3 to constrain the role of inorganic mixing state.

As discussed above, N2O5) can be calculated on a single particle basis, and then averaged for comparison with the measured N2O5). Using this approach, we calculate an average N2O5) of 0.031 (+0.012, -0.008), a 17% reduction in N2O5) as compared to the analogous calculation assuming internal mixing. This result suggests that inorganic mixing state is not the major controller in the observed suppression of N2O5), but rather that the organic component is acting to suppress N2O5) due to the amount, type, or phase state of the organic present.

Figure 2.3A illustrates the fraction of particles sampled during Periods 1 and 2 which,

- - based on their nCl /nNO3 values, exhibit a specific range of N2O5) when calculated using the

Bertram and Thornton parameterization under the external mixing assumption. It is evident here, that the fraction of particles producing the given calculated N2O5) ranges for both Period 1

(more-polluted) and Period 2 (less-polluted) are similar, with the least variation between the two for N2O5) > 0.03. This highlights the small differences in reactivity of fractions of the populations during these two sampling periods, as calculated using the parameterization assuming externally mixed particles.

Figure 2.3 A) Histogram of calculated single particle (N2O5), based on ATOFMS single particle - - measurements of nCl /nNO3 at the SIO pier, for sampling period 1 (more-polluted; pink) and period 2 - - (less-polluted; blue). B,C) PDF of ATOFMS nCl /nNO3 as measured for sub-micrometer (light line) and super-micrometer (dark line) particles for sampling period 1 and 2, respectively.

The above N2O5) calculation does not account for the correlation between particle

- - 22,24–26 surface area and nCl /nNO3 as has been seen in previous studies. Fig. 2.3B and C depict

- - the PDF of nCl /nNO3 for sub- and super-micrometer aerosol separately, indicating slightly

- - higher nCl /nNO3 for super-micrometer aerosol in the case of Period 2. We calculate N2O5), under the externally mixed population assumption, during Period 1 to be 0.030 (+0.013, -0.008)

30 and 0.031 (+0.012, -0.008), while for Period 2, the values are 0.029 (+0.010, -0.008) and 0.033

(+0.004, -0.005) for sub- and super-micrometer particles, respectively. Given that the super- micrometer surface area accounts for less than 20% of the total surface area under all conditions sampled at the SIO pier, inclusion of surface area weighting does not explain the differences between field observations and model parameterizations at this location.

To assess whether notable changes in the organic mass fraction could be deduced from the ATOFMS, we look at the probability distribution functions for the ratio of two organic peak

+ + areas (sum of the 37 and 43 m/Q peaks that represent the C3H and CH3CO fragments) to the total positive ion peak area. Here forward, we refer to this ratio as fROA, or the representative organic mass fraction of single aerosol particles.36 As shown in Fig. 2.4A and summarized in

Table 2.1, there is no discernible difference in fROA for the two sampling periods. We further look

+ at the ratio of oxygenated fragment (43 m/Q, CH3CO ) to the sum of the two organic markers (37

+ 43 m/Q). As shown in Fig. 2.4B, there is again no discernible difference between the two sampling periods. This result suggests that either the molecular composition of the particulate

31 organic species changed between the two sampling periods, or a subtle change in the RH at the sampling site altered the physical phase state of the organics.37,38

2.5 Insight from Single Particle Measurements of Reacted Sea Spray Aerosol

- - In the preceding section, we used ambient measurements of single particle nCl /nNO3 to calculate single particle (N2O5) for comparison with the population average determination of

(N2O5). During the fall 2009 field deployment, we also made observations of single particle

- - chemical composition post-N2O5 reaction using ATOFMS. The PDF of nCl /nNO3 for ambient

- - particles is shown in Fig. 2.5A and 2.5B along with the PDF of nCl /nNO3 for particles having been exposed to N2O5 in the flow reactor for sampling periods 1 and 2, respectively. The PDF of

- - nCl /nNO3 for particles having been exposed to N2O5 during Period 1 (Fig. 2.5A; more-polluted conditions) is statistically indistinguishable from that measured on ambient particles. For

- - comparison, the PDF of nCl /nNO3 for particles having been exposed to N2O5 during Period 2

(Fig. 2.5B; less-polluted conditions) is also statistically indistinguishable from that measured on ambient particles. This indicates that under both conditions, ambient particles underwent very

- - little chemical change with respect to their nCl /nNO3 ratio in the flow reactor. This is due to the experimental conditions of the ambient N2O5 reactivity experiment, where the N2O5 mixing ratio is held to ca. 1 ppbv to limit nitrate accumulation, which would bias the measured reactive uptake coefficient. Under that experimental conditions used here (e.g., [N2O5] = 1 ppbv and average

- - residence time = 258 sec), and a (N2O5) = 0.03, we would expect the nCl /nNO3 ratio at 0.01

(peak of ambient PDF during Period 2) to decrease by <1%. This unfortunately is too small a perturbation to detect via ATOFMS.

- - Figure 2.5 Probability distribution function (PDF) as a function of nCl /nNO3 for: A) more-polluted pier - - data, B) less-polluted pier data, and C) nascent SSA data sets. D) Single particle nCl /nNO3 as a function of particulate organic fraction (fROA) for nascent SSA. fROA is calculated using the representative organic ions, m/Q (37 + 43) / total ion peak area. Small blue circles indicate unreacted particles, while small green squares indicate flow tube-reacted particles. Median values are shown in large symbols.

However, direct comparison of the single particle composition of reacted and unreacted particles conducted at higher gas-phase concentrations may provide insight into the properties of aerosol that act to limit reactive uptake (e.g., organic fraction, liquid water content) on a single particle level. To assess this, we explore the use of single particle aerosol measurements of

nascent SSA, and nascent SSA reacted with HNO3 at significantly high mixing ratios ([HNO3] =

155 ppbv) to assess the potential for using real-time single particle approaches to probe for the selective reactivity of specific classes of aerosol particles within a population. Details concerning the experimental set up can be found in the supporting online material. Briefly, nascent SSA was generated in a state of the art, 30 m long wave channel located at Scripps Institution of

Oceanography, which was filled with real ocean water pumped through coarse grain filters and subsequently doped with bacteria and phytoplankton cultures to regenerate surface active organic

39 material potentially lost during filtration. Here, nascent SSA was aged via reaction with HNO3 in an entrained aerosol flow reactor at an average exposure time of 114 sec at 155 ppb HNO3. The chemical composition of the particles post-reaction were measured at the base of the flow reactor via ATOFMS. HNO3 was used in place of N2O5 in these experiments due to the larger effective

27 reactive uptake coefficient on nascent SSA ((HNO3) > 0.1) thus enabling significant particle

- - reactions on the short time scale of a flow reactor. The PDF of nCl /nNO3 for nascent SSA

- - particles (blue) is shown in Fig. 2.5C alongside the PDF of nCl /nNO3 for nascent SSA particles having been exposed to HNO3 in the flow reactor (green). The 4-5 order of magnitude shift in the

- - peak nCl /nNO3 indicates that the nascent SSA react significantly in the flow reactor, providing the opportunity to assess the variability in reaction kinetics within the aerosol population.

- - Single particle measurements of the nascent SSA nCl /nNO3 ratio (y-axis) as a function

- - of fROA are shown as blue circles in Fig. 2.5D. As shown, single particle nCl /nNO3 decreases

-6 with increasing fROA, from 686.4 to 4.90 for fROA of 7.81×10 and 0.25, respectively. While the source of this dependence is unknown, it suggests that either: 1) nascent SSA generated in this study is externally mixed displaying more than one type of aerosol particle40 or that 2) ATOFMS is not ionizing the entire particle volume and radial gradients in chloride, nitrate, and organics

- - impact the measurement of nCl /nNO3 at high fROA. For comparison, single particle measurements of nascent SSA having been exposed to HNO3 are also shown, in green squares, in

Fig. 2.5D. If all SSA were to react at the same rate, independent of the organic content of the

- - particle, it is expected that nCl /nNO3 would decrease uniformly, independent of fROA. However,

- - in contrast to the nascent SSA, single particle nCl /nNO3 decreases with decreasing fROA, from

-2 -6 76.4 to 2.24×10 for fROA of 0.25 and 7.81×10 , respectively. At high organic content (fROA

- - >0.1), nCl /nNO3 is nearly equal for both the reacted and unreacted particles, indicating that

-6 HNO3 reactive uptake is suppressed. In contrast, at low organic content (fROA = 7.81×10 ),

- - median nCl /nNO3 is over 4 orders of magnitude smaller for the reacted particles, indicating that the particles have undergone significant reaction. This result demonstrates that the presence of organics in nascent SSA, while rare in this study, can act to limit the reactive uptake of HNO3.

Ongoing work is focused on relating fROA quantitatively to the single particle organic mass fraction and potentially an organic film thickness. These results suggest that the reactive uptake of trace gases to nascent SSA may be significantly slower under conditions where the aerosol organic mass fraction has been shown to be large.8,35

2.6 Atmospheric Implications

We present an analysis of the first direct measurements of the dependence of (N2O5) on single particle inorganic mixing state using simultaneous atmospheric observations of (N2O5) and single particle chemical composition. These observations indicate that: 1) when accounting

- - for particulate nCl /nNO3 mixing state, model parameterizations of (N2O5) continue to over predict (N2O5) by more than a factor of two in polluted coastal regions, indicating that the chemical composition and physical phase state of particulate organics likely controls (N2O5) in these air masses, and 2) direct determination of (N2O5) in air masses of marine origin are well captured by model parameterizations and reveal limited suppression of (N2O5) relative to its known value on NaCl particles, indicating that the organic mass fraction of fresh sea spray aerosol at this location does not suppress (N2O5). Surprisingly, simultaneous measurements of

35 single particle organic composition revealed little difference in the mass fraction of organics between the two sampling periods, suggesting that the molecular composition and resulting physical properties of the organic phase of the aerosol control reaction kinetics at this site.

2.7 Acknowledgements

The authors would like to thank Joel Thornton for the use of the CIMS and flow tube apparatus used in the ambient portion of this study. This research was supported by the Office of

Science (Office of Biological and Environmental Research), U.S. Department of Energy (Grant

No. DE-SC0006431) and the National Science Foundation (Grant No. CHE1038028). O.S.R is grateful for a Graduate Research Fellowship from the National Science Foundation (2011- present).

2.8 Supporting Information

2.8.1 Sampling Locations and Details

2.8.1.1 Ambient Study: Scripps Institution of Oceanography Pier

Sampling occurred at the end of Scripps Pier, 300m offshore, located at Scripps

Institution of Oceanography, La Jolla, CA. The simultaneous measurement of N2O5 and single particle data took place between October 4th-6th 2009. Ambient NOx and O3 were measured at the site continuously along with particle size distribution (see below). Relative humidity data was obtained from a meteorological station at a nearby coastal site (<11 miles away) due to absent RH data from the Scripps Pier meteorological station during this time period.

2.8.1.2 Laboratory Study: Nascent Sea Spray Wave Channel

Nascent, fresh sea spray particles were generated in a laboratory setting using a 30 m- long ocean-atmosphere facility, filled with ocean water pumped directly from Scripps Pier after coarse filtration. A hydraulic paddle was utilized to create breaking waves in a fixed location

36 once every second, which accurately reproduced the bubble size distribution created when a wave breaks in the real open ocean.39,41 Nascent SSA discussed here was generated in this fashion from ocean water doped with bacteria (Alteromonas and Pseudoalteromonas atlantica) and phytoplankton (Dunaliella tertiolecta) cultures to reproduce organics possibly lost during coarse filtration. This experimental setup was a large study, employing a suite of instruments, to better understand the factors causing variability in measured SSA properties from ambient field studies.39 Pertinent to this paper, an ATOFMS sampled from the headspace above the breaking waves to allow for measurement of the chemical composition of nascent sea spray. A flow reactor system was used to age the particles with HNO3 prior to measurement, allowing for comparison of pre-and post-reacted particle chemical composition. This flow reactor was a 3" ID Pyrex tube, coated with halocarbon wax to minimize gas and particle interactions with the walls. Particles were injected through a side port at the top of the reactor, and allowed to mix with HNO3 throughout the residence time in the tube. The chemical composition was measured from the base of the reactor. HNO3 was generated using an NO2 permeation tube (Kin-Tek Laboratories, Inc.) coupled to an O3 production cell.

2.8.1.3 Ambient Study: Jeju Island, South Korea

A second ambient data set used as a comparison of single particle chemical composition in this supporting document involves particles sampled from Jeju Island, South Korea.42 Single particle data was collected with an ATOFMS between April 12th-May 15th 2007, with a one week period of data presented in this discussion. The island is located in the East China Sea, where air masses can typically be influenced by outflow from mainland China, South Korea, or fresh marine sources. The sampling site was in a rural location, situated 20 m from the ocean on the west coast of the island to avoid local contamination.

2.8.1.4 Ambient Study: Long Beach, California

A third ambient data set used as an example of chemical composition in a highly polluted particle population was sampled in Long Beach, California between November 16-26, 2007.43

ATOFMS measurements were taken from a stationary platform located on Terminal Island in the

Port of Los Angeles, heavily used by ships, allowing for the sampling of both daytime onshore flows, and nighttime offshore flows. For further details on the logistics of this study, please see

Ault et al. 2010.

2.8.2 Measurements

2.8.2.1 N2O5 Reactivity Determinations:

Ambient air from Scripps Pier was sampled through a 0.25" ID stainless steel tube either directly into the flow reactor, or via a filter, providing 99% particle removal from the air stream.

The flow reactor has been described in detail previously.28 Briefly, it is a 5.9" ID stainless steel tube mounted vertically, 35" in length and coated with halocarbon wax on the interior to decrease

N2O5 losses to the reactor walls. N2O5 was produced in situ via the reaction of excess NO2 with ozone, created via photolysis of O2, in a mixing cell. N2O5 was delivered to the flow reactor via

1/8" OD Teflon tubing, and was injected via a side injection port to either the top or the base of the reactor. N2O5 was monitored as I•N2O5 (m/Q 234) using a quadrupole chemical ionization mass spectrometer (CIMS) running in negative ion mode, with methyl iodide as the reagent ion, as described in Kercher et al., 2009.44

The resulting flow emerging from the base of the flow reactor was split between the

CIMS, sampling 2 slpm, ATOFMS, sampling 1 slpm and a scanning mobility particle sizing instrument (SMPS), sampling 0.4 slpm. During an hour of each sampling day, the N2O5 source was directed to a waste line, and only ambient particles were allowed to pass through the flow reactor. This allowed the ATOFMS to probe unreacted ambient particles.

Data obtained during periods of NO ≥ 1ppb was excluded from the gamma calculation, as high NO concentrations can cause titration of NO3, resulting in a shift in the NO3-N2O5 equilibrium, and thus N2O5 signal artifacts.

2.8.2.2 Sizing Measurements:

During the Scripps Pier sampling period, sizing measurements between 0.01-0.6µm were made using a Scanning Mobility Particle Sizer (TSI Inc. 3010 CPC & TSI Inc. 3081 DMA), connected to the ATOFMS sampling line. Particle number distributions measured with the SMPS were then converted into surface area distributions, which were used in the calculation of gamma.

The SMPS measured wet size distributions, so no correction or growth factors were applied to the data.

Sub-micrometer surface area was calculated using the wet size distributions from the

SMPS data. The super-micrometer surface area for both the more and less polluted sampling periods of the Scripps Pier data set was estimated using a similar data set from a separate prior

(August 8th-September 1st) sampling period. During the prior study, both an Aerodynamic

Particle Sizer (APS) and SMPS were measuring size distributions from Scripps Pier. Periods of similar SMPS size distributions were found in the current and prior SMPS data sets. The APS surface area contribution (between 9-24%) for the corresponding time period in the prior study was obtained, and applied to the current data set as an estimate for the super-micrometer particle fraction. Estimation of the super-micrometer surface area reaching the flow reactor may be an overestimation, since the inlet was curved, likely leading to loss of the larger particles due to impact on the inlet wall. For this reason, the reported measured gamma values are likely to be the lower limit.

2.8.2.3 Single Particle Composition Measurements:

For all studies reported, aerosol single particle chemical composition was measured using an Aerosol Time of Flight Mass Spectrometer (ATOFMS) which has been described previously.29

Briefly, aerosol particles enter the instrument, where they then encounter a UV laser pulse

(266nm) and are consequently desorbed and ionized. Both positive and negative ions are then detected by a dual time-of-flight mass spectrometer. The ATOFMS detection sensitivity of Cl-

- 45 and NO3 was considered to be approximately equal based on their similar electron affinities.

This is discussed in more detail in the next section. The ATOFMS employed measured particles between 0.1-1.67 m in geometric diameter. Geometric physical diameters were calculated from aerodynamic diameter by assuming particles were spherical, and using the following equation

(ES1):

(ES1)

3 3 where ρeff is the effective density of particles, taken to be 1.8 g/cm , ρ0 is 1g/cm , the density of particles for which the ATOFMS is calibrated. The size range measured by the ATOFMS allows for measurement of the surface area maximum of ~350nm (more polluted), and ~300nm (less polluted).

Two criteria were employed for filtering particles for analysis. Only particles that

- - contained chloride were considered, and from the particles remaining, those with nCl /nNO3

4 - - between 1x10 and 1000 were divided into 500 nCl /nNO3 bins and used to calculate the cumulative distribution function for the data set. The percentage of the original particle population remaining after each filter criteria are shown in Table 2.2. It should be noted that the reason for a large portion of particles being removed before analysis in the nascent SSA case is

40 due to the majority of particles containing large amounts of chloride in comparison to nitrate,

- - causing the nCl /nNO3 to be much greater than 1000.

Table 2.2 The fraction of particles removed before CDF and/or N2O5 calculation.

- - Location Particles containing chloride 0.0001

SIO Wave Channel; nascent SSA 99.8% 26.2%

Jeju Island, Korea 95.6% 89.2%

SIO Pier; polluted 78.5% 76.5%

SIO Pier; less polluted 65.8% 58.1%

2.8.3 Calculations and Interpretations

2.8.3.1 Measurements and Interpretation of Particulate Chloride-to-Nitrate Mixing State:

- - The PDF of particulate nCl /nNO3 for particles sampled at two ambient locations (Jeju

Island, Korea and Long Beach*, California) and under two different laboratory conditions

(generated nascent sea spray, and chemically processed generated nascent sea spray) are shown in

Fig. 2.6. The PDF is employed as a means to examine both the extent of particle aging between populations and the degree of external mixing within a specific particle population. The PDF

- - obtained from particulate nCl /nNO3 measured on individual nascent SSA (blue line, described above) indicates that greater than 50% of particles exhibit a ratio greater than 100, reflecting the

- - nCl /nNO3 of seawater. As a proxy for aging of particles in the atmosphere, nascent SSA were passed through a flow reactor (red line), resulting in a HNO3 exposure of 155 ppb over an average reaction time of 114 s. The resultant PDF associated with HNO3 exposure shows a marked shift

- - towards lower values of nCl /nNO3 as compared with the nascent SSA PDF, illustrating the result of chemical aging on a population of nascent SSA. This decrease in the ensemble average

- - particulate nCl /nNO3 would result in a corresponding suppression of N2O5 uptake when calculated using E1. Ambient data, collected over 8 days, from both Jeju Island and Long Beach,

(black and grey curves, respectively), exhibit differing degrees of ambient aging based on the

- - location of the PDF curve with respect to nCl /nNO3 . In comparison to the fresh nascent SSA, the

- - Long Beach particle population is shifted to lower values of nCl /nNO3 , and thus is more aged.

Following the same reasoning, the Jeju Island particle population appears more aged than nascent sea spray, but less aged than the particles measured in Long Beach. This is consistent with the proximity of the sampling location to nascent SSA emissions in the absence of strong anthropogenic influence. In addition to providing information on particle age with respect to

- -, emission (mean value), the observed PDF of nCl /nNO3 when computed on time-scales that are short relative to atmospheric variability, provides an indication of the mixing state of the particle population.

Ambeint (Long Beach,CA) HNO3-Reacted SSA -3 12x10 Ambient (Jeju Island,Korea) Nascent SSA

8 PDF 6

0 0.0001 0.001 0.01 0.1 1 10 100 1000 - - nCl /nNO3

2.8.3.2 Calculation of k'2f:

The value of k'2f used in the parameterization to calculate N2O5 at the SIO pier was

1.25×106 s-1 for the more polluted period, and 1.13×106s-1 for the less polluted period. According to the equation given in Bertram et al. 2009,30 this value depends on the liquid water content of the aerosol particles. Relative humidity for sampling periods 1 and 2 of the Scripps Pier study was obtained from a meteorological station at a nearby coastal site due to absent RH data from the Scripps Pier meteorological station during these time periods. Using this data, the Aerosol

Inorganics Model31 was utilized to determine the liquid water content of particles at an average relative humidity of 71% for the more polluted period, and 64% for the less polluted period. k'2f

43 for the wave channel nascent SSA data was calculated in a similar way, also using a relative humidity of 64%.

- Calculation of N2O5 using the extreme values from the nH2O/nNO3 range had a larger impact on the calculated gamma for the more polluted condition, producing at most a 28% error.

A similar calculation for the less polluted condition lead to at most a 10% difference in N2O5.

These values are summarized in Table 2.1 of the main paper.

2.8.3.3 Probability Distribution Function:

The shape of the cumulative distribution function is a useful tool in the assessment of particle population age and the degree to which the sampled population is externally mixed.

However, the width of the PDF is a complex function of: 1) single particle mixing state, 2)

ATOFMS instrument precision, and 3) temporal variability on the time-scale of the observation.

- - As shown in Fig. 2.6, the width of the PDF curves are highly variable, where nCl /nNO3 as

- - measured in Long Beach is narrowly distributed and nCl /nNO3 measured at Jeju Island is broadly distributed between 1x10-2 and 1x102. In what follows, we assess the contribution of

ATOFMS accuracy and precision and temporal variability in the measurements to the width of

- - the nCl /nNO3 distributions, using the Jeju island samples as an example.

- - Figure 2.7A shows the single particle nCl /nNO3 for particles sampled from Jeju Island over a 24 hour period (light grey markers). The large variability in the single particle data can be deconvoluted, to some extent, via the analysis of standard solutions prepared in the lab and analyzed using an ATOFMS (Figure 2.7B). Standard I is aerosol produced via atomization of a

10:80:10 solution of NaCl:NaNO3:NaI in water. Standard II is aerosol produced via atomization of a 45:45:10 solution of NaCl:NaNO3:NaI in water. NaI is added to the water to increase particle absorption of the 266 nm ionization laser pulses. Analysis of these solutions shows a range of

- - nCl /nNO3 , with the distributions tending towards lower ratios than expected based on the ratio of

44 components in solution. This bias is likely due to matrix effects in the ATOFMS ionization

- 45 process, caused by the slightly differing electron affinities between Cl and NO3. To determine the significance of this bias, the interquartile range of the median of single particle data for

Standard II, 65%, was applied as a percentage to the running median of the Jeju Island data (Fig.

2.7A blue line) resulting in the solid dark grey region in Figure 2.7A. This analysis indicates the

ATOFMS bias determined from standard solutions is significantly smaller than the interquartile

- - variability seen in the data set (light solid gray region). Figure 2.7C shows particulate nCl /nNO3 as a function of particle size and frequency of occurrence, with the particulate surface area

- - maximum (Figure 2.7D) deriving from particles with nCl /nNO3 of 0.25. It should be noted that with an ATOFMS size detection limit of 0.1-1.67 m, the percentage of particle surface area

- - captured in this data set is 74%. An increase in particle nCl /nNO3 is seen with increasing size,

- - providing additional evidence of contribution to the variability in the nCl /nNO3 arising from particle-to-particle differences.

Figure 2.7 Assessment of variation in the cumulative distribution function arising from sources unrelated - - to particle-to-particle variability. Panel A shows the raw variability in nCl /nNO3 over 24 hours of sampling at Jeju Island (light grey markers). The running median over 500s of sampling is shown in blue. - - Applying the nCl /nNO3 interquartile range, observed in a standard solution, as a percentage to the running - - median results in the solid dark grey area. Panel B shows nCl /nNO3 PDF curves of two standard solutions (Standard I, 10:80:10 of NaCl:NaNO3:NaI in water; Standard II, 40:40:10 of NaCl:NaNO3:NaI in water), - - along with the PDF of ambient Jeju Island data. Panel C: particulate nCl /nNO3 as a function of particle size and frequency of occurrence. Panel D: surface area data for Jeju Island (blue line), with standard deviation shown in blue shading.

2.8.3.4 Internally Mixed (Ensemble Average Assumption) N2O5 Calculation:

The ensemble average N2O5 calculation is derived by handling the chloride and nitrate concentrations of particles as if they had been collected on a filter sample. The sum of total chloride and total nitrate is calculated for the entire particle population from the single particle

46 measurements. An average chloride and an average nitrate value is obtained, and then the average chloride value divided by the average nitrate value in order to determine an ensemble average

- - nCl /nNO3 for the population. This ensemble average value is then used in the N2O5 parameterization to calculate a particle population ensemble average gamma value.

The error for the ensemble average calculation was calculated by propagating the error in

- - the ensemble average nCl /nNO3 calculation. This propagated error was then used to calculate a

- - lower and upper limit to the nCl /nNO3 , and these values used in the parameterization to determine the limits on the N2O5 value.

2.8.3.5 Externally Mixed N2O5 Calculation:

- - The externally mixed N2O5 values are derived from the particle population nCl /nNO3

- - probability distribution function (PDF). For each of the 500 nCl /nNO3 bins used in the PDF calculation, a midpoint is established. This is then used to calculate the N2O5 value of each bin, and is multiplied by the population of particles in that respective bin. The externally mixed N2O5 average value is then calculated by summing over the N2O5 values determined for each bin. The upper and lower error values were calculated utilizing the Standard I and II atomized aerosol. A probability distribution function of each standard was made and fitted with a log normal fit curve.

- - The nCl /nNO3 corresponding to the peak value of the curve was determined, and the percent

- - deviation from the theoretical maximum nCl /nNO3 value ( 0.125 for standard I, and 1 for standard II) found. For standard I, the percent deviation was 68.6%, and was 56.4% for standard

- - - - II. The upper nCl /nNO3 value was considered to vary by +68.6%. Upper nCl /nNO3 values were input into the parameterization, resulting in an upper bound for the externally mixed gamma calculations.

Standards I and II were also used to calculate an additional upper error bound, which was added to the previously calculated value, and a lower bound. The full width half maximum

(FWHM) for each standard PDF curve was determined. The x-values corresponding to the

FWHM spread were used to calculate a sigma value ( = 0.5 × FWHM). Sigma was calculated as

- - a percentage of the FWHM nCl /nNO3 value, resulting in 79.6% for Standard I, and 80.6% for

- - standard II. Assuming each nCl /nNO3 bin could differ by as much as 80.6% in the externally

- - mixed gamma calculation, this value was applied to the nCl /nNO3 ratios, and used in the parameterization to calculate a lower, and an additional upper bound. The upper bound error from this calculation was combined with the value previously described.

- 2.8.3.6 Calculation of nH2O/nNO3

- 31 nH2O/nNO3 was determined using the Aerosol Inorganics Model, principal model III.

Inputs used with AIM were the mean RH determined for Periods 1 and 2, and moles of particulate

- - Cl and NO3 with Na as a counter ion. Organics were not used as inputs into the model for a number of reasons: 1) We do not have specific mass fractions of organics on each ambient particle to include in the model input, 2) the chemical identity of the specific organic species on the ambient particles is unknown , 3) the phase state and spatial distribution in/on the particles themselves are not known, and 4) this study was intended to focus on the inorganic mixing state and its influence on the discrepancy between measured and parameterized values of (N2O5). For

- - - - both RH conditions, the model was run for nCl /nNO3 equal to 0.0001 (the smallest nCl /nNO3 used in the parameterization calculation) and 1000 (the maximum used in the parameterization.

- - - From the AIM output, nH2O/nNO3 was determined for each nCl /nNO3 , and a plot of

- - - nH2O/nNO3 vs. nCl /nNO3 created for each of the two RH conditions and fit linearly. The

- equation from the fit for was used in the subsequent calculation of nH2O/nNO3 for each of the

- - - - 500 nCl /nNO3 bins (for externally mixed case) and for the average nCl /nNO3 (for the internally mixed case).

2.9 Acknowledgments