OH-INITIATED HETEROGENEOUS OXIDATION OF ATMOSPHERIC ORGANIC AEROSOLS. Ingrid Jennifer George. A thesis submitted in conformity with the requirements

Transcription

1 OH-INITIATED HETEROGENEOUS OXIDATION OF ATMOSPHERIC ORGANIC AEROSOLS by Ingrid Jennifer George A thesis submitted in conformity with the requirements for the degree of Doctor of Philosophy Graduate Department of Chemistry University of Toronto Copyright Ingrid Jennifer George (2009)

2 OH-INITIATED HETEROGENEOUS OXIDATION OF ATMOSPHERIC ORGANIC AEROSOLS Doctor of Philosophy 2009 Ingrid George Department of Chemistry University of Toronto ABSTRACT The chemical aging of organic aerosols by OH-initiated heterogeneous oxidation was investigated using both model organic and ambient aerosol particles. Organic aerosol particles were exposed to OH radicals in an aerosol flow tube and the modification of their chemical composition and particle properties was studied. Overall, this work has shown that OH-initiated heterogeneous oxidation enhanced the degree of oxidation and the Cloud Condensation Nucleus (CCN) activity of organic aerosol particles for equivalent OH exposure timescales of a few days to a week. Aerosol Mass Spectrometer (AMS) measurements showed that the heterogeneous uptake kinetics of OH radicals onto model primary organic aerosols was efficient. The heterogeneous reaction of organic aerosols with OH led to the ii

3 production of high molecular weight particle-phase species with the addition of multiple oxygenated functional groups. These results were consistent with the observed increase in particle density with OH exposure. With the exception of solid organic aerosols, the particle volume and mass of organic particles were reduced by less than 20% from OH oxidation at high OH exposures due to volatilization of particle-phase reaction products. The degree of oxidation of the organic fraction of urban ambient aerosols was significantly enhanced for an equivalent atmospheric OH exposure time of 4 days for a daily average atmospheric OH concentration of cm -3. Ambient aerosol particles sampled from a sparsely populated, forested region were initially more oxygenated than the urban aerosol particles and did not become more oxidized from reaction with OH radicals. The modification of the hygroscopicity of model primary and secondary organic aerosols from chemical aging was investigated by measuring the CCN activity of organic aerosols exposed to OH radicals. Primary organic aerosols, initially CCN inactive, became as CCN active as secondary organic aerosols due to heterogeneous reaction, where surface tension reduction played a major role. The CCN activity for model secondary organic aerosols was also enhanced due to OH oxidation, but changes were less dramatic than for the model primary organic aerosols. iii

4 ACKNOWLEDGEMENTS There are many people I need to thank, who have helped me throughout the process of completing this thesis work. First and foremost, I would like to thank my supervisor Professor Jonathan Abbatt for his guidance and support throughout my entire graduate career. I truly could not have asked for a more thoughtful and encouraging supervisor. Jon always made time from his ever busier schedule to talk with me about my research work. I am grateful that Jon helped me to see the bigger picture with respect to my research work and in terms of my future career. I am thankful that my committee chair Prof. Jamie Donaldson and committee member Prof. Frank Wania gave me many constructive suggestions regarding the progression of thesis work throughout the years. I want to thank Prof. Myrna Simpson for being an excellent professor of one of my graduate courses and for her advice during my Ph.D. qualifying exam and Paul Ziemann for being my thesis defense external examiner. I am grateful that my undergraduate supervisor at University of Calfornia at Davis, Prof. Cort Anastasio first gave me the opportunity to conduct research in atmospheric chemistry, when I first gained a passion for this field. I want to thank the past and current members of the Abbatt group for their support. I could not have done the CCN experimental work without help from Keith Broekhuizen and Rachel Chang. I want to thank Jay Slowik for demystifying the AMS and patiently answering all my AMS-related questions. I am grateful to have had Nana Kwamena and Dan Aubin as graduate student mentors. I want to thank Yue Zhao and Vladimir Danov for being great undergraduate students to mentor. The ESIiv

5 MS work was not possible without the assistance of Alex Young in the AIMS laboratory. The members of Chemclub helped to remind me that there are also wonderful people outside of the Environmental Chemistry Programme in the Department. I had a blast at all the Chemclub events, and I was given possibly the only opportunity in my life to manage a store. I am so fortunate to have met two wonderful women in Toronto, Benedicte Fontaine-Bisson and Cynthia Aoki, who were my roommates when I moved here and who have become my dearest friends since. I want to thank my boyfriend Kushan Fernando for being my foundation my throughout my time in Toronto and for always helping me to see the brighter side of life. I am so grateful that his parents George Fernando and Siromi Fernando and his sister Karen Fallon treated me like part of their family and made me feel at home in Canada. I want to thank my best friend Grant Eshoo, who has always been there for me for the past 13 years regardless of our distance apart. Finally, I could not have gotten this far without the love and encouragement from my family: my father Sanky George, my mother Ludiya Benjamin and my sister Bernadette George. Although they didn t want me to move farther away for school, they were nonetheless very supportive of my educational goals, and for that I will always be thankful. v

6 TABLE OF CONTENTS Page ABSTRACT ii ACKNOWLEDGEMENTS.. iv TABLE OF CONTENTS..vi LIST OF TABLES. x LIST OF FIGURES...xi LIST OF APPENDICES xvi PREFACE...xvii CHAPTER ONE: INTRODUCTION: Chemical Aging of Atmospheric Organic Aerosols MOTIVATION SOURCES AND SINKS OF ORGANIC AEROSOLS Organic Aerosol Sources Organic Aerosol Removal Pathways CHEMICAL ANALYSIS OF ORGANIC AEROSOLS Offline Techniques Online Techniques ATMOSPHERIC AGING OF ORGANIC AEROSOLS Physical Aging Processes Particle-phase and Multiphase Chemical Aging Heterogeneous Oxidation of Organic Aerosols A KINETIC FRAMEWORK FOR HETEROGENEOUS OXIDATION LABORATORY STUDIES OF HETEROGENEOUS OXIDATION Experimental Methods O 3 Oxidation Radical-Initiated Oxidation Evidence for Heterogeneous Oxidation in the Field Emerging Research Issues Regarding Heterogeneous Oxidation 32 vi

7 1.7. RESEARCH OBJECTIVES REFERENCES CHAPTER TWO: Heterogeneous Oxidation of Saturated Organic Aerosols by Hydroxyl Radicals: Uptake Kinetics, Condensed-Phase Products, and Particle Size Change ABSTRACT INTRODUCTION EXPERIMENTAL Aerosol Reactor Flow Tube Setup OH Quantification Kinetic Experimental Methods Product Characterization Methods RESULTS AND DISCUSSION Kinetic Studies Particle Size and Density Modification Product Identification Reaction Mechanism ATMOSPHERIC IMPLICATIONS AND CONCLUSIONS REFERENCES CHAPTER THREE: Modification of Cloud Condensation Nucleus Activity of Organic Aerosols by Hydroxyl Radical Heterogeneous Oxidation ABSTRACT INTRODUCTION EXPERIMENTAL METHODS Particle Oxidation Setup Particle Analysis Techniques Surface Tension Measurements. 104 vii

8 3.4. RESULTS AND DISCUSSION Modification of CCN Activity Particle Size Change Surface Tension Measurements CONCLUSIONS SUPPLEMENTARY MATERIAL Appendix A: Stearic Acid Particle Mass Change Calculation Appendix B: Köhler Calculation For Oxidized BES Particle REFERENCES CHAPTER FOUR: PART ONE: Chemical Aging of Ambient Organic Aerosol from Heterogeneous Reaction with Hydroxyl Radicals ABSTRACT INTRODUCTION EXPERIMENTAL RESULTS AND DISCUSSION REFERENCES CHAPTER FOUR: PART TWO: OH-Initiated Heterogeneous Oxidation of Ambient Organic Aerosol in a Biogenically-Impacted Region INTRODUCTION EXPERIMENTAL RESULTS AND DISCUSSION CONCLUSIONS REFERENCES CHAPTER FIVE: Chemical Evolution of Laboratory Secondary Organic Aerosol from OH-Initiated Heterogeneous Oxidation INTRODUCTION. 165 viii

9 5.2. EXPERIMENTAL METHODS Aerosol Generation Oxidation Experiment Particle Characterization RESULTS AND DISCUSSION AMS Organic Mass Spectra Mass, Volume, Density and Hygroscopicity ESI-MS Analysis CCN Activation ATMOSPHERIC IMPLICATIONS AND CONCLUSIONS REFERENCES CHAPTER SIX: CONCLUSIONS SUMMARY Kinetics, Products and Reaction Mechanism Evolution of Particle Properties Modification of Hygroscopicity Heterogeneous Oxidation of SOA and Ambient Particles ATMOSPHERIC RELEVANCE OF HETEROGENEOUS OXIDATION FUTURE RESEARCH REFERENCES APPENDIX A: Photochemical Model for OH Quantification. 222 ix

10 LIST OF TABLES Page Table 1.1. Major component classes observed in organic aerosol (adapted from Pöschl) Table 1.2. Limiting cases for kinetics for loss of particle-phase species (P).. 24 Table 2.1. Table 4.1. Condensed-phase product masses from high resolution ESI-MS analysis of reacted BES particles and proposed chemical formulas...76 Average ToF-AMS particle mass concentrations.138 Table 4.2. AMS aerosol composition. 155 Table 5.1. Table 5.2. Table A.1 SOA precursor mixing ratios and particle properties R 2 values for linear plots of PMF factors with reference mass spectra and organic PMF factors from ambient AMS data in literature Reactions for photochemical model x

11 LIST OF FIGURES Page Figure 1.1. Figure 1.2. An overview of sources and sinks of atmospheric organic aerosols (adapted from Pöschl) Conversion time of hydrophobic particles at initial concentration of 1000 cm -3 and diameter of 50nm to CCN at S = 0.3% for different conditions (adapted from Kanakidou et al.) Figure 1.3. Resistor model Figure 1.4. Figure 2.1. Figure 2.2. Figure 2.3. Figure 2.4. Figure 2.5. Figure 2.6. Proposed reaction mechanism for radical initiated oxidation of alkane (adapted from Molina et al.) Chemical structure of Bis(2-ethylhexyl) sebacate (BES)...52 Aerosol flow reactor setup for the study of the heterogeneous oxidation of organic aerosols..54 ToF-AMS mass spectra of BES particles normalized to same particle number concentration: a) unreacted particles, b) reacted particles at OH exposure of atm-s, c) difference spectra of reacted spectrum minus unreacted spectrum, with positive values shown in green and negative values shown in red.60 Change in m/z 297 signal intensity as a function of OH exposure. Experimental data are shown as square symbols, and solid lines show calculated slopes using Eq. (1) with γ 0 values of 0.1 (green dotted), 0.5 (blue dash-dotted), 1.0 (red dashed), and 2.0 (pink solid). Inset is a magnification of the plot showing the data used for calculation of γ Size distributions for unreacted (black solid line) and reacted (red dashed line, OH exposure = atm-s) BES particles normalized to the same particle concentrations: a) DMA volume-weighted particle concentration as a function of mobility diameter. b) ToF-AMS nitrateequivalent mass concentration as a function of vacuum aerodynamic diameter.. 65 Particle beam width profile for unreacted (black squares) and reacted (red circles, OH exposure = atm-s) BES particles xi

12 Figure 2.7. Figure 2.8. Figure 2.9. Relative change in volume (black squares) and density (red circles) of BES particles as a function of OH exposure. Error bars show one standard deviation of averaged data. Lines are fits to guide the eye. 68 Kinetic plot for product fragments m/z 44 (black squares), 127 (red circles), 153 (green triangles), and 181 (blue triangles). Error bars show one standard deviation of averaged data. Solid lines are fits to data with the function y = 1 + a(1-exp -bx ).. 70 ESI-MS mass spectra of BES particles: a) unreacted particles, b) reacted particles (OH exposure = atm-s), c) magnification of panel b for mass 450 to 550 amu 72 Figure Generalized reaction mechanism for OH-initiated oxidation of a hydrocarbon 78 Figure Relative particle mass change as a function of OH exposure. Black squares are calculated mass values from the product of density and volume measurements, red dashed and blue dotted lines show theoretical contribution of carbonyl + alcohol and decomposition reaction pathways to mass change, respectively, assuming a VOC/OH ratio of 0.1 and average VOC molecular mass of 58 g mol -1, and pink line shows theoretical overall mass change (See text) 83 Figure 3.1. Experimental setup for OH oxidation of organic aerosols Figure 3.2. Figure 3.3. Figure 3.4. Figure 3.5. Figure 3.6. CCN activation curves of 145 nm mobility diameter BES particles at several OH exposures. All lines are sigmoidal fits to the curve Critical supersaturation of 145 nm mobility diameter BES particles and 150 nm mobility diameter SA particles. Open symbols represent data where particles were exposed to OH radicals and solid symbols show data from control experiments in the absence of OH Kappa values corresponding to the data in Figure Volume change normalized to particle number concentration of 100 nm mobility diameter (squares) and polydisperse (circles) stearic acid particles as a function of OH exposure. Error bars represent one standard deviation of averaged data. 111 Surface tension (left axis) and WSOM mass (right axis) as a function of OH exposure time (bottom) and OH to BES ratio (top) for oxidized BES film extracts. Error bars represent one standard deviation of averaged data 115 xii

13 Figure 3.7. Figure 3.8. Figure 3.9. Surface tension as a function of WSOM concentration for oxidized BES film extracts (1 ml). Dashed line shows the exponential fit to the BES data. Dotted line represents water surface tension (σ w ). Star and diamond symbols represent surface tension of saturated azelaic acid and nonanoic acid solutions, respectively ESI-MS spectra for A) oxidized BES film water extract with reaction time of 2 hours and B) oxidized BES particle methanol extract adapted from George et al. (2007). Peaks marked with asterisks represent masses that were included in the calculation of F WSOM. Boxes encompass peaks with asterisks that were chosen as WSOM products 117 Calculated Köhler curves for 145 nm diameter reacted BES particles Figure Relative mass, density, and volume changes as a function of OH exposure Figure 4.1. Figure 4.2. Figure 4.3. ToF-AMS organic mass spectra taken on 08 October 2007: (a) BGF (bars) and ODF (crosses) mass spectra in the presence of OH with an exposure of atm-s normalized to the sum of the signals over all masses. (b) Difference spectrum of ODF and BGF from top panel. Typical fragments that arise from several organic classes are marked with filled symbols ToF-AMS time series of aerosol species during 04 and 08 October, Solid symbols represent data in the ODF in the presence of OH. OH exposures for 04 and 08 October ranged from to atm-s and to atm-s, respectively. (a) and (b) organic (ORG), sulfate (SO4), ammonium (NH4), and nitrate (NO3) concentrations. (c) and (d) m/z 44 and 57 fractions on the left and right axes, respectively. (e) and (f) Relative change in m/z 44 and 57 fractions. (g) Mass size distributions smoothed by 5-point adjacent averaging for particles in the BGF (ORG, SO4, NH4, NO3) and for particles exposed to OH (Oxidized ORG) in the ODF with an OH exposure of atm-s during 08 October Change in organic mass (ORG) and relative fractions of m/z 44 and m/z 57 in the ODF normalized to BGF values as a function of OH exposure. Data represent averages of five 2-minute measurements taken on one day and error bars are one standard deviation of the average. Values at 0 atm-s OH exposure are measurements taken under dark conditions in the absence of O 3 averaged over the entire study.144 xiii

14 Figure 4.4. Figure 4.5. Figure 5.1. Figure 5.2. Figure 5.3. Figure 5.4. Figure 5.5. Figure 5.6. Figure 5.7. Figure 5.8. Normalized organic mass (top panel), Frac44 and Frac57 values (bottom panel) for Experiment 2. Dashed lines represent average values for each dataset and solid lines represent 1 standard deviation of the averages. Solid symbols represent ODF data and open symbols are BGF data.. Organic mass contribution at m/z 29 was not included 158 ODF to BGF ratios for Frac57, Frac44 and normalized organic mass (Org) for Experiment 2. Horizontal lines follow same convention as. Figure 4.4. Vertical dashed lines divided data points based on OH exposures expressed as atmospheric oxidation timescale in days assuming [OH] = cm Factor profiles for the 2-factor solution for data set A (A-B, F-G) and for the 3-factor solution for data set B (C-E, H-J) modeled for Fpeak values of Fpeak = 0 (A-E) and Fpeak = -0.5 (F-J) AMS organic mass spectra normalized to total organic mass for HO/HA-, LO/HA- and LO/LA-type SOA and α-pinene + O 3 chamber SOA from Bahreini et al Correlation of normalized mass spectra for LO/LA and LO/HA versus HO/HA. 180 Difference HO/HA-type SOA organic mass spectra for control conditions Hg Lamp and Dark O 3 subtracted from Dark mass spectra. Mass spectra for SOA exposed to OH were subtracted from Hg Lamp mass spectra..181 Relative changes in F44 and F55 with OH exposure. Error bars represent one standard deviation of average values.183 Relative changes in time series mass contributions of PMF factors (Fpeak= 0) for data set A (left) and data set B (right) with OH exposure with bottom x-axis in experimental exposure, top x-axis in atmospheric exposure time assuming [OH] = cm -3. Fits to the data are shown to guide the eye Relative changes in volume and density for HO/HA, LO/HA and LO/LA SOA particles Relative changes in SOA organic mass. Solid symbols are measured values and open diamonds are calculated from the data shown in Figure xiv

15 Figure 5.9. ESI-MS mass spectra for HO/HA SOA under Hg Lamp and OH exposure. Inset shows magnification over a smaller m/z range Figure Average κ values for various flow tube conditions (left). Activation diameter (bars) and corresponding κ values (squares) of LO/HA and HO/HA SOA for control and for OH exposure (right) 193 Figure Relative change in F44 for one week of atmospheric OH exposure with initial F Figure A.1. Measured and modeled OH concentrations as a function of O 3 concentrations Figure A.2. Measured and modeled OH concentrations as a function of RH 227 xv

16 LIST OF APPENDICES Page APPENDIX A: Photochemical Model for OH Quantification. 222 xvi

17 PREFACE The thesis is arranged in a series of chapters that are based on several manuscripts that are in preparation for submission to, have been submitted for publication to, or have been published in peer-reviewed scientific journals. Therefore, there will be some repetition in the introductory material in each chapter of this thesis. An appendix at the end of the thesis is comprised of supplementary material on the experimental methods, which was excluded from the manuscripts. All manuscripts were written by Ingrid Jennifer George, who is listed as the primary author of these publications, with critical comments provided by Jonathan P. D. Abbatt and the other coauthors. The contributions of the other authors to the work outlined in each manuscript are described below. The following also provides the bibliographic reference information regarding the manuscripts, upon whose material each chapter is based. CHAPTER TWO: George, I. J., Vlasenko, A., Slowik, J. G., Broekhuizen, K., Abbatt, J. P. D Heterogeneous oxidation of saturated organic aerosols by hydroxyl radicals: Uptake kinetics, condensed phase products, and particle size change. Atmospheric Chemistry and Physics 7, Contributions: The experimental setup was developed by Ingrid J. George with assistance from Keith Broekhuizen. Experiments were performed by Ingrid J. George with assistance Jay G. Slowik and Alexander Vlasenko. Data analysis and interpretation was performed by Ingrid J. George. CHAPTER THREE: xvii

18 George, I. J., Danov, V., Vlasenko, A., Chang, R., Abbatt, J. P. D Modification of cloud condensation nucleus activity of organic aerosols by hydroxyl radical heterogeneous oxidation. Submitted for publication to Atmospheric Environment. Contributions: CCN experiments were performed and analyzed by Ingrid J. George with assistance from Rachel Chang. Vladimir Danov carried out and analyzed the surface tension measurements. Coated-wall flow tube CIMS experiments were performed and analyzed by Vladimir Danov with assistance from Alexander Vlasenko. Data interpretation was undertaken by Ingrid George. CHAPTER FOUR: PART ONE: George, I., Slowik, J., Abbatt, J. P. D Chemical aging of ambient organic aerosol from heterogeneous reaction with hydroxyl radicals. Geophysical Research Letters 35, L13811, doi: /2008gl Contributions: Ingrid George performed the experiments and analyzed and interpreted data with assistance from Jay Slowik. CHAPTER FOUR PART TWO and CHAPTER FIVE: George, I. J., Abbatt, J. P. D Chemical evolution of laboratory secondary organic aerosol from OH-initiated heterogeneous oxidation. In preparation for submission. Contributions: Ingrid J. George performed the experiments and analyzed and interpreted the data with exception of the PMF analysis of the AMS data collected during the Whistler study, which was performed by Jay Slowik. xviii

19 1 Chapter One INTRODUCTION: Chemical Aging of Atmospheric Organic Aerosols

20 2 1.1 MOTIVATION An aerosol is broadly defined as a suspension of fine liquid or solid particles in gas. Atmospheric aerosol compositions and concentrations are highly spatially and temporally variable with particle sizes typically spanning a range of 10-9 m up to 10-5 m in diameter and with concentrations ranging from 10 up to 10 5 cm Atmospheric aerosols are generally comprised of inorganic salts (i.e. sulfate, nitrate, ammonium and sea salt), carbonaceous matter (organic matter and elemental/black carbon), crustal material and water. Organic matter comprises a significant mass fraction, between 20 to 90%, of atmospheric aerosols. 2,3 In this text, atmospheric organic aerosols specifically refer to the organic component of the particles in atmospheric aerosols consisting of a highly complex and variable mixture of hundreds of organic compounds, a sizable fraction of which remain heretofore uncharacterized. 4 The role of atmospheric aerosols in the environment has been a key research theme in atmospheric science over the past few decades. Atmospheric organic aerosols reduce visibility and impact climate directly by scattering or absorbing solar radiation, altering Earth s radiative balance. 5 Aerosols can also influence climate indirectly by acting as cloud condensation nuclei (CCN) or ice nuclei (IN), thereby altering cloud properties. The aerosol indirect effect represents one of the largest uncertainties in current climate models and our understanding of climate change. 5 Atmospheric particulate matter has been implicated for causing adverse health effects. 6 Particles influence atmospheric chemistry by providing surfaces, on which low-volatility gases can condense and trace reactive gas-phase species are lost or

21 3 produced. Major examples of such heterogeneous reactions include N 2 O 5 heterogeneous hydrolysis and the reactions of gaseous reservoir chlorine species on polar stratospheric clouds leading to O 3 depletion. The influence of atmospheric organic aerosols on these environmental processes is ultimately controlled by their particle properties (i.e. chemical composition, size, optical properties, morphology and hygroscopicity). Therefore, it is vital not only to characterize the sources and sinks of atmospheric organic aerosols, but also to understand the aging mechanisms that lead to the chemical transformation of atmospheric organic aerosols SOURCES AND SINKS OF ORGANIC AEROSOLS Figure 1.1 summarizes the processes that control the production and loss of atmospheric organic aerosols. Atmospheric organic aerosols emitted directly into the atmosphere are termed primary organic aerosol (POA). Secondary organic aerosol (SOA) particles are formed in situ from the gas-phase oxidation of volatile organic compounds (VOCs) that can either form new particles through nucleation or condense onto pre-existing aerosols. Sources of both POA and SOA may arise from either natural or anthropogenic processes. In the atmosphere, aerosols undergo a multitude of physical and chemical transformations that alter particle chemical composition and physical properties. Atmospheric aging mechanisms include coagulation, condensation of low-volatility inorganic and organic gases, cloud processing and heterogeneous reactions. These atmospheric aging processes will be discussed in greater detail in the next section. The two removal processes for atmospheric organic aerosols are dry deposition to Earth s surface and wet deposition

22 4 that can occur in cloud or below cloud, whereby particles that act as CCN are lost through or are scavenged by precipitation. Figure 1.1: An overview of sources and sinks of atmospheric organic aerosols (adapted from Pöschl 7 ) Organic Aerosol Sources The relative importance of various POA sources and SOA formation processes, and hence the chemical composition of the organic aerosol, depends on regional emissions and atmospheric conditions. Major natural and anthropogenic POA sources include biomass burning (i.e. forest fires, domestic wood burning), fossil fuel combustion (i.e. motor vehicles, energy production) and cooking. Combustion POA typically consists of elemental carbon (EC) and organic material, such as polycyclic aromatic hydrocarbons, unburned fuel and lubricating oil. EC, also

23 5 termed graphitic or black carbon, is the refractory portion of combustion aerosols containing mostly carbon in a structure similar to graphite. 1 Fine-mode marine POA with diameters < 2.5 μm produced by bubble bursting containing dissolved organic matter and organic surfactants from the sea surface microlayer have been measured in the Northern Atlantic during periods of high biological activity. 8 Coarse-mode POA with diameters greater than 2.5 μm can be formed by several wind driven processes, including suspension of road dust, soil and plant/biological material (i.e. waxes, pollen, bacteria). Secondary organic aerosol mass is produced from the reaction of biogenic and anthropogenic VOCs with atmospheric oxidants, such as O 3, NO 3 and OH, resulting in more oxygenated, low volatility reaction products that can partition to the aerosol phase. Monoterpenes are thought to dominate SOA formation out of all biogenic VOC classes on a global scale, 9 while anthropogenic emissions of aromatics are considered to be important SOA precursors in urban regions. 10 Many laboratory chamber studies have investigated SOA formation process by measuring overall SOA aerosol yields and characterizing gas and aerosol-phase products for these standard SOA precursors. Thus far, SOA formation has been shown in these studies to be dependent on a large number of environmental factors including the initiating oxidant, NO x levels, temperature, particle acidity, relative humidity and pre-existing aerosol mass. 11 Several modeling studies of atmospheric OA concentrations based on chamber data have significantly underpredicted SOA mass, 12,13,14 highlighting the

24 6 discrepancies between laboratory and atmospheric SOA formation processes. Possible sources for these discrepancies may be due to large missing sources of SOA mass from non-traditional SOA precursors, poorly understood mechanisms for SOA formation or inaccurate emissions databases. Recent studies have highlighted the potential importance of semi-volatile compounds with a wide range of volatilities that can originate from POA particles near their emission sources and evaporate during atmospheric dilution as possible SOA precursors. 15,16 Heterogeneous oligomer formation in the particle phase and multiphase oxidation of volatile SOA precursors, i.e. oxidation of dissolved gases in the aqueous phase, represent two SOA formation mechanisms of potential importance requiring further study. 11,17, Organic Aerosol Removal Pathways Aerosol particles are removed from the atmosphere via dry and wet deposition. The size and density are the most important factors that determine the dry deposition velocity of a particle. 1 Ultrafine particles with diameters < 0.05 μm are transported efficiently to the Earth s surface by Brownian diffusion and in this way behave similarly to gas molecules. Coarse-mode particles also have high deposition velocities due to inertial impaction and gravitational settling. Deposition velocities reach a minimum for fine-mode particles with diameters between μm. Thus, the major removal process for this size fraction is wet deposition representing 70 to 85% of tropospheric removal for OA. 18 Organic aerosol lifetimes in the atmosphere can range from approximately a few days to a few weeks depending on latitude and altitude. 18

25 7 Wet deposition of atmospheric aerosols is generally controlled by various large scale and microscale processes involved in cloud and precipitation formation. The focus here is on the characteristics of atmospheric organic aerosols that control their activation to cloud droplets, which is the key process in their removal by wet deposition. CCN activation of atmospheric aerosols can be described by the Köhler equation (Equation 1.1) that relates the equilibrium water vapour pressure ratio (s) to the solution droplet diameter (D): 19 p 4M wσ s = S + 1 = = a w exp (1.1) p RTρ wd where S is the supersaturation (%), p is the water vapour pressure over the droplet (Pa), p o is the water vapour pressure over a flat surface of water (Pa), a w is the water activity. The exponential term in Equation 1.1 represents the Kelvin effect that accounts for the increase in the water vapour pressure over a curved surface compared to the value over a flat surface, where M w is the molecular weight of water (kg mol -1 ), σ is the surface tension of the droplet (N m -1 ), R is the ideal gas constant (J mol -1 K -1 ), T is the temperature (K) and ρ w is the density of water (kg m -3 ). The water activity term reflects a lowered water vapour pressure over a solution containing soluble material compared to the vapour pressure over pure water, known as the Raoult effect. The a w term can be expressed in terms of D as follows: a w 6nsM = 1 πρsol w 1 D 3 (1.2)

26 8 where n s is the number of solute moles that depends on the van t Hoff factor (ν) of the solute, i.e. the number of dissociated ions per molecule. Köhler theory predicts the minimum supersaturation required for a particle to spontaneously take up water and activate to form a cloud droplet. The critical supersaturation for a given particle depends on its size, density and chemical composition, including molecular weights, van t Hoff factors, solubilities and surface activities of its solute components. The organic fraction can influence the CCN activation of atmospheric aerosols in several potentially conflicting ways: 1) addition of soluble material, 2) addition of insoluble material that increases particle size, 3) droplet surface tension lowering and 4) influence on water uptake kinetics. Water soluble organic matter (WSOM) can make up an important fraction of the organic component in atmospheric aerosol. 2,20 WSOM can enhance the CCN activity of organic aerosols by enhancing the Raoult effect, but even a small fraction of inorganic salts in mixed organic/inorganic aerosols will likely dominate this effect. 21 Surface active organic species in WSOM from atmospheric aerosols reduce the Kelvin term in Equation 1.1, thereby enhancing CCN activation. 22,23 Recent work has shown that humic-like substances (HULIS), a complex mixture of macromolecular, high molecular weight compounds of unknown structure in WSOM, has exhibited surface active properties in aqueous atmospheric aerosol extracts Organic films on aerosols may limit water condensation and evaporation, thus slowing droplet growth rates. 22,27 Numerous laboratory studies have shown that Köhler theory can reasonably predict the CCN activation of single component particles composed of a small range

27 9 organics with variable solubilities, mostly dicarboxylic acids, and particles containing salt/organic mixtures. 28 However, the application of Köhler theory in an explicit manner to atmospheric organic aerosols is highly impractical, given the paucity of information on the characterization of the chemical composition and hygroscopicity of ambient aerosols. The Köhler equation has recently been parametrized, so that the collection of parameters describing the particle properties and composition are typically represented by one parameter for organic aerosols that can be readily used in climate models Recent laboratory studies have applied these parametrizations to characterize the CCN activation measurements of more complex organic aerosols, such as chamber SOA and WSOM extracted from ambient aerosols CHEMICAL ANALYSIS OF ORGANIC AEROSOLS Despite the wide range of analytical techniques available today, the full chemical speciation of atmospheric organic aerosols remains an elusive goal due to the chemical complexity of particulate organic matter. The chemical composition of atmospheric particles is contingent upon the aerosol sources and aging processes involved and can vary temporally, spatially, with particle size and even from particle to particle. These variabilities render complete identification of organic aerosols on a molecular basis practically impossible. Nevertheless, there have been major developments in aerosol measurement due to both real-time and off-line analytical instrumentation. The following is a brief overview of the current state of understanding of the chemical characterization of organic aerosols and analytical techniques used for this purpose.

28 Offline Techniques Traditional sampling and chemical speciation methods for atmospheric aerosols are offline techniques. 36,37 Particles can be separated using sizing techniques, i.e. cyclones, impactors or virtural impactors, and then collected on a solid substrate, such as filters. Bulk concentrations of organic carbon (OC) can be determined by difference between measurements of total carbon (TC) not including inorganic carbonates and elemental carbon (EC) using thermal-optical methods and evolved gas analysis. 38 Chemical separation is typically achieved through analysis of particle filter extracts by chromatography (gas (GC), ion (IC) or liquid chromatography (LC), etc.) and detected with spectroscopy (e.g. UV, VIS, IR, absorption, fluorescence, NMR, etc.) or mass spectrometry (MS; electron impact, chemical ionization, electrospray, etc.). Derivatization of targeted functional groups, e.g. conversion of carboxylic acids to more volatile esters, can allow for analysis of water soluble organic species using GC-MS. The application of these long established analytical techniques has allowed for the molecular-level identification of several major classes of compounds in organic aerosols listed in Table 1.1 along with likely sources for each chemical class. 7,38 Quantitative functional group analysis of organic aerosols has more recently been carried out with the application of NMR 39 and FTIR 40. Single particle analysis techniques for particles collected on a solid substrate give information on the mixing state of aerosols, i.e. how the aerosol constituents are distributed within the aerosol population. Elemental analysis and particle morphology on single particles can be

29 11 observed offline using electron microscopy requiring vacuum conditions that may lead to the volatilization of organic matter. 36 Single particle analysis using electron microscopy and X-ray spectroscopy has more recently been carried out under higher pressures. 41,42 A major disadvantage for offline sampling is its low temporal resolution, typically requiring hours to days of sampling time depending on aerosol mass loadings and instrument sensitivities. Significant positive and negative sampling artifacts are associated with aerosol filter sampling. Volatilization of semi-volatile aerosol components leads to negative artifacts, whereas adsorption of organic gases onto filters can be a significant positive artifact. Furthermore, particles sampled onto the filter may react with trace gases during sampling. These artifacts can be significantly reduced through the application of denuders in series with filter sampling.

30 12 Table 1.1. Major component classes observed in organic aerosol (adapted from Pöschl 7 ) Chemical Classes Sources 1 aliphatic hydrocarbons aliphatic alcohols and carbonyls fatty acids and other alkanoic acids aliphatic dicarboxylic acids multifunctional aliphatic and aromatic compounds polycyclic aromatic hydrocarbons (PAHs) nitro- and oxy PAHs levoglucosan proteins and other amino compounds cellulose and other carbohydrates secondary organic oligomers/polymers humic-like substances (HULIS) BM, FF BM, SOA/aging BM, SOA/aging SOA/aging SOA/aging, soil/dust FF, BB FF, BB, aging BB BM BM SOA/aging, soil/dust SOA/aging, soil/dust 1 BM=Biomass, FF = fossil fuel combustion, BB = biomass burning Online Techniques Online measurement techniques for particle size and other properties are now widely used in atmospheric aerosol measurements. 36 Recent online analytical techniques have been developed for the bulk chemical characterization and quantification of organic aerosols with high time resolution. The particle-into-liquid sampler (PILS) is an automated technique that grows sampled aerosols to droplets by condensation of water vapour. The droplets are collected on an impaction plate, and the solution is transported to an IC for quantification of anions and cations. 43 PILS has also been coupled to a carbon analyzer for water soluble organic carbon (WSOC) measurements, where WSOC is defined as the carbon content in WSOM. 44 As

31 13 mentioned previously, the importance of high molecular weight compounds measured in WSOM, i.e. humic-like substances (HULIS), has gained recent attention. 45 Similar to humic substances, the exact molecular structures of HULIS compounds and their formation processes remain difficult to elucidate. Mass spectrometers have been developed in the past decade for the real-time characterization of atmospheric aerosols. Aerosol mass spectrometers have emerged as powerful analytical tools for the quantification of inorganic and organic aerosol components with high sensitivities. Many techniques have been developed to provide single particle and particle size-resolved compositions Aerosol mass spectrometer techniques vaporize particles by either laser or thermal heating, and the vaporized components are ionized by lasers or by traditional mass spectrometric ionization sources, such as electron impact or chemical ionization. Two commercially available aerosol mass spectrometers have been developed by TSI and Aerodyne Inc. Because the work in this thesis is based on particle analysis from an Aerodyne instrument, the following will focus on the Aerodyne Inc. instrument only. The Aerodyne Aerosol Mass Spectrometer (AMS) employs thermal desorption followed by electron impact ionization coupled with a quadrupole or time-of-flight mass spectrometer. 47,51 The Aerodyne AMS quantifies the size-resolved non-refractory component (i.e. components that can vaporize at temperatures 600 C) of fine mode atmospheric aerosols. This instrument consists of three differentially pumped chambers: 1) a particle sampling chamber, 2) a particle sizing chamber and 3) a particle composition detection chamber. Particles are first sampled through the inlet

32 14 of the particle sampling chamber and are focused into a narrow particle beam by passing through an aerodynamic lens with 100% particle transmission efficiency for particles with diameters between 70 and 500 nm. After exiting the sampling chamber, particles then pass through a chopper with a duty cycle of a few percent sending bursts of particles into the particle time-of-flight sizing chamber. Particle flight times through the particle sizing chamber are measured to determine particle vacuum aerodynamic diameters. The non-refractory components of the particles are vaporized by impaction on a heated surface at ~600 C. Finally, the vapour is ionized by electron impact with 70 ev electrons, where ions are analyzed by either a quadrupole or a time-of-flight mass spectrometer. Aerodyne AMS instruments have been employed to characterize the chemical composition of submicron aerosols in various laboratory studies and field studies in different regions. Non-refractory inorganic components of atmospheric aerosols, such as sulfate and nitrate, fragment into a few peak masses, and therefore are easily identified and quantified with the Aerodyne AMS. However, mass spectra of the organic fraction of atmospheric aerosols are generally quite complex and are dominated by many mass fragments, rendering identification of individual particlephase organic compounds essentially impossible. Delta analysis has been applied to yield some chemical information from the AMS mass spectra of organic aerosols, where a series of mass peaks separated by 14 amu (i.e. CH 2 group) is attributed to a specific chemical class. 52 Recently, principle component analysis (PCA) 53 and positive matrix factorization (PMF) have been applied to deconvolve ambient AMS

33 15 organic mass spectra into organic components with differing chemical properties or into specific aerosol source profiles. 54,55 Zhang et al. applied PCA to field measurements from 37 sites in urban, rural and remote regions to classify the organic aerosols into fractions with varying degrees of oxidation, namely, an oxygenated organic aerosol (OOA) associated with SOA and aged aerosol and hydrocarbon-like aerosol (HOA) that is generally linked to POA. 56 This work indicated that OOA dominates HOA in almost all regions, where the fraction of POA tends to decrease away from urban areas, i.e. POA sources ATMOSPHERIC AGING OF ORGANIC AEROSOLS Atmospheric aerosol aging refers to a collection of physical and chemical processes that transforms the chemical composition of atmospheric aerosols generally from hydrophobic to hydrophilic nature. Because the ability of a particle to activate to cloud droplets strongly depends on its hydrophilicity, atmospheric aging of aerosol plays a vital role in controlling the wet deposition removal rates for atmospheric organic aerosols. Organic aerosols are generally more hydrophobic close to emission sources, such as urban regions. As they are transported away from their emission sources, organic aerosols tend to become more oxygenated and hydrophilic with increasing residence time in the atmosphere. This phenomenon has been observed in several field studies with AMS measurements. 12,57,58 Ultimately, it is expected that the particle composition of organic aerosols approaches that of highly aged background aerosols measured in remote regions. The atmospheric residence time required to convert aerosol particles from a hydrophobic to hydrophilic nature, typically termed

34 16 aging or turnover time, is an important parameter to include in climate models to accurately determine aerosol removal rates and lifetimes in the atmosphere. Generally, models have assumed a first order rate for the hydrophobic-to-hydrophilic conversion of aerosols with a turnover time of approximately 0.7 to 1.2 days. 18 This section provides a brief discussion of the physical and chemical aging processes of organic aerosol particles. In subsequent sections, the current understanding of heterogeneous oxidation of organic aerosols, the least understood of the aging mechanisms, will be reviewed in greater detail as it is the major subject of this thesis Physical Aging Processes Both coagulation and condensation are physical aging processes that can increase the amount of soluble material in the aerosol with a resulting increase in particle size. Condensation typically adds soluble material to pre-existing particles by adsorption of inorganic gases (e.g. H 2 SO 4 and HNO 3,) and SOA precursors onto aerosols, whereas coagulation can increase the soluble fraction of aerosols through scavenging of particles that are more hydrophilic (e.g. aged or background aerosols). The relative importance of these two physical aging processes on the hydrophobic-to-hydrophilic conversion of organic aerosols is dictated by precursor gas and oxidant concentrations (SO 2, VOCs, OH, etc.), particle sizes and particle number concentrations. Condensation is dominant under the conditions of high precursor gas concentrations and/or a low number of pre-existing aerosol particles. Kanakidou et al. evaluated aging times necessary for 63% of fresh hydrophobic particles with initial number concentration of N=1000 cm -3 to become

35 17 CCN due to condensation of H 2 SO 4 and coagulation with aged background aerosol. 18 Figure 1.2 summarizes these calculations for a range of background aerosol number concentrations (N acc ) and SO 2 levels that are relevant for different regions. Hydrophobic particles that are initially 50 nm diameter size require a soluble fraction of 68% to become CCN active at S = 0.3% according to Köhler theory. Figure 1.2 illustrates that the aging time in urban areas is on the order of hours, where coagulation is clearly the dominating physical aging process is due to the high particle number concentrations. The conversion times for particles under conditions typical for rural and remote marine regions were in the range of days and weeks, respectively. SOA formation was not considered in this calculation, which would enhance the importance of condensation, especially in urban regions with high VOC concentrations.

36 18 Figure 1.2. Conversion time of hydrophobic particles at initial concentration of 1000 cm -3 and diameter of 50nm to CCN at S = 0.3% for different conditions (adapted from Kanakidou et al.) Particle-phase and Multiphase Chemical Aging Atmospheric chemical aging involves heterogeneous, condensed-phase and multiphase chemical reactions of condensed phase organics in atmospheric aerosols resulting in a more hydrophilic and CCN active organic fraction. Association reactions are a set of non-oxidative, condensed-phase reactions that will increase the molecular weight and reduce the volatility of particle-phase reaction products. 11 Specific examples of association reactions, most of which are acid catalyzed, that form oligomeric species are aldol condensation reactions, hemiacetal formation reactions and reactions of Criegee intermediates. Enhanced SOA yields have been observed on acidic aerosols seed particles that are added to facilitate the condensation of SOA mass, 59 but oligomers can be formed in the absence of aerosol seed

37 19 particles. 17,60 Although these reactions do not lead to more oxygenated products, they may enhance SOA formation and increase particle size, which will influence the ability of particles to act as CCN. There is some evidence that oligomeric species are formed through these association reactions in ambient aerosols. 61 The formation mechanism of sulfuric acid in clouds through multiphase reactions is well understood. Analogous multiphase reactions of organic species have received little attention to date. Gas-phase aldehydes, such as glyoxal, may partition into aqueous phase and become oxidized to form organic acids (e.g. oxalic acid). This process will lead to an increase in organic mass and degree of oxidation of the organic aerosol particle. Our current understanding of multiphase chemical mechanisms of organic aerosol species remains poor Heterogeneous Oxidation of Organic Aerosols The major focus of this thesis work is the atmospheric aging of organic aerosols by heterogeneous oxidation. Organic aerosols can be chemically transformed during their residence time in the atmosphere by reaction with gas-phase oxidants, such as O 3, OH, NO 3 and halogen radicals. Heterogeneous oxidation reactions are expected to add oxygenated functional groups to the condensed-phase organic species, thereby increasing their oxidation state, hydrophilicity and CCN activity. Depending on the reaction pathway, heterogeneous oxidation may also lead to the decomposition and volatilization of reaction products, releasing oxygenated VOCs to the gas phase. The following provides a brief explanation of the kinetic processes involved in heterogeneous reaction with a liquid particle, followed by an overview of

38 20 the laboratory studies on chemical aging of organic aerosols with greater emphasis on radical-initiated heterogeneous oxidation, as it is the subject of this thesis A KINETIC FRAMEWORK FOR HETEROGENEOUS OXIDATION A series of interactions between the gas and particle-phase reactants controls the rate of heterogeneous reaction. 62,63 The kinetics for the heterogeneous reaction can be generally described by a reactive uptake coefficient (γ), which is the fraction of gas-particle collisions that lead to the reactive loss of the gas-phase reactant to the particle-phase. The observed reactive uptake coefficient (γ obs ) is the main kinetic parameter that is determined in laboratory studies, either from the observed loss of gas-phase or particle-phase reactants (assuming no particle-phase secondary chemistry). Here, the kinetics for the reactive loss of particle-phase species from heterogeneous reaction is the focus as it is germane to the work in this thesis. The flux (J in molecules cm -2 s -1 ) of the reactant gas into the condensed phase is expressed in terms of the uptake coefficient as follows: J [ G] cγ obs = (1.3) 4 where [G] is the bulk gas concentration (molecules cm -3 ) and c is the average gas molecular speed (cm s -1 ). The observed loss rate of particle-phase species concentration [P] can be used to calculate γ obs with Equation 1.4 for a given particle radius (a). 63 d[ P] [ G] cγ obs 3 = (1.4) dt 4 a

39 21 As shown in the diagram in Figure 1.3, the observed uptake coefficient reflects a number of coupled processes, such as gas diffusion, mass accommodation, bulk and surface chemical reaction and dissolution, which can be described by coupled differential equations. A resistance model for gas uptake and reaction on a liquid particle has been proposed as a reasonably good approximation that decouples these processes. 62 The net uptake coefficient can be expressed as a sum of resistances, where each decoupled process is represented by a resistance term (1/Γ) that is normalized to the gas collision rate: 1 γ net 1 = Γ diff 1 + S S α 1 + Sα Γ rxn + Γ s 1 + Γ b diff (1.5) The resistance terms in Equation 1.4 represent gas diffusion to the particle surface (Γ diff ), reaction in the bulk (Γ rxn ), reaction on the surface (Γ s ) and diffusion through the bulk of the particle (Γ b diff). The mass accommodation coefficient (α) is the fraction of gas molecular collisions that lead to the molecules being taken up by the particle bulk, which is represented in Equation 1.4 as two separate parameters: first is adsorption to the surface (S) and second is the dissolution into the liquid ((S-α)/Sα).

40 22 1 Γ s 1 1 Γ diff S S-α Sα 1 Γ rxn 1 Γ b diff Figure 1.3. Resistor model. The gas-phase diffusion resistance term has been formulated by Fuchs and Sutugin 64 in Equation 1.6 with Knudsen number (Kn) defined as Kn = λ/a, where the gas mean free path (λ) is defined in Equation Γ Kn = Kn(1 Kn) diff + (1.6) 3D g λ = (1.7) c The gas diffusion term becomes important for large particles and large observed uptake coefficients.

41 23 The resistance terms for bulk reaction (Γ rxn ) and surface reaction (Γ s ) are expressed in Equations 1.8 and 1.9, respectively: 4HRT 1 Γ rxn = [coth( a / l) ( l / a)] (1.8) II c D k [ P] g II 4k s H s RTK s[ P] Γ s = (1.9) c where H (M atm -1 ) and D g (cm 2 s -1 ) are the Henry s law constant and diffusion coefficient for the gas species dissolved in particle phase, respectively, the reactodiffusive length l is defined as l = (D g k II [P]) 1/2. The heterogeneous reaction kinetics will be controlled by dominating resistances terms in Equation 1.5 that will vary with the specific reaction system. Kinetic formulations for Equation 1.4 in terms of experimental parameters have been developed for several specific reactive uptake limiting cases as summarized in Table ,65 The limiting processes on the γ obs can be determined experimentally by measuring the loss of particle-phase species as a function of time for several particle sizes.

42 24 Table 1.2. Limiting cases for kinetics for loss of particle-phase species (P) 63,65 Limiting Cases [P] t /[P] t=0 = Case 1: Gas diffusion and/or accommodation limited (i.e. very fast bulk reaction) Case 2: Fast reaction near surface Case 3: Slow reaction throughout bulk of particle 3ng c 1 4a [ P] 0 1 Γ diff HRTng Dk 1 2 [ P] α II II exp( HRTn k t) g 1 3 t a t 2 3 II Case 4: Surface reaction exp( HRTng Ksks t) a 1.6. LABORATORY STUDIES OF HETEROGENEOUS OXIDATION In contrast to atmospheric VOC oxidation, 66 our current understanding of heterogeneous oxidation of organic aerosols is quite poor and still in its early stages. Most of what is known has been based on research over the past decade involving laboratory studies of relatively simple model systems, such as oxidation of organic surfaces, films and single-component aerosol systems. Early work on heterogeneous oxidation of organic aerosols has been reviewed by Rudich. 67 These studies focused on the kinetics for the reactions of gas-phase oxidants (mostly O 3 ) with organic surfaces and initial work on the identification of reaction products.

43 Experimental Methods Recent laboratory studies involving heterogeneous oxidation of condensedphase organics have typically involved three types of experimental systems: coated wall flow tubes, aerosol flow tubes and aerosol chambers. 67 In coated-wall flow tube studies, a liquid or solid organic monolayer or film coating a flow tube is exposed to the oxidant. The kinetics of the reaction is determined by observing the reactive loss of the oxidant, typically monitored with mass spectrometry as a function of exposure time with the organic coating. The effect of heterogeneous oxidation on organic coatings has also been studied using a Quartz Crystal Microbalance 68 and surface characterization techniques, such as contact angle measurements, FTIR, laser-induced fluorescence and X-ray photoelectron spectroscopy (XPS) Recent developments in online aerosol mass spectrometric techniques have allowed for the measurement of the reactive loss of particle-phase organic species in aerosol particles due to heterogeneous oxidation in aerosol flow tube experiments. AMS techniques applied to the study of the O 3 + oleic acid aerosol system for example include Aerodyne AMS, 72 aerosol chemical ionization mass spectrometry 73 and thermal desorption particle beam mass spectrometry, 74 single particle mass spectrometry 65 and photoelectron resonance capture ionization (PERCI) mass spectrometry 75. Although the model aerosols in these studies are a step closer to approximating atmospheric aerosols compared to films and monolayers, aerosol flow tube experiments often require higher oxidant concentrations for short reaction times compared to atmospheric conditions.

44 26 Environmental or smog chambers have been used extensively for the study of SOA formation, 11 but have been applied more recently to study heterogeneous oxidation of organic aerosols using similar detection methods as for aerosol flow tube studies. The major advantage of using chambers is that the large reaction volume allows for longer reaction times at lower oxidant concentrations, more closely mimicking atmospheric conditions. Given the low oxidant levels, typical residence times of atmospheric aerosols of days to a week and their chemical complexities, the ability for a laboratory aerosol experiment to fully match atmospheric conditions remains a considerable challenge O 3 Oxidation The O 3 + oleic acid aerosol reaction system has emerged in recent years as a benchmark for heterogeneous oxidation of organic aerosols. 76 This reaction system has several desirable characteristics, such as the ease of O 3 production and measurement as compared to radical oxidants. Further, the specificity of the O 3 + alkene reaction reduces complexity of the reaction mechanism compared to radical oxidation that is much less specific. Laboratory studies have attempted to fully characterize the heterogeneous reaction of O 3 with oleic acid films and particles. We now have a better understanding of this specific reaction, including the kinetics, reaction mechanisms, gas- and condensed-phase reaction products, as well as how the reaction modifies particle properties. The major findings from these laboratory studies have been reviewed by Rudich for the early work, and by Zahardis and Petrucci and Rudich et al. for recent studies. 67,76,77

45 27 The reactive uptake coefficient has been measured for pure oleic acid films and particles (γ ~ ), as well as mixed unsaturated/saturated models, indicating that mixed phases can reduce the uptake coefficient by over an order of magnitude compared to liquid unsaturated aerosols. Case 2 from Table 1.2, i.e. reaction within a thin shell near the surface, was concluded to be the most likely kinetic process for ozonolysis of unsaturated liquid organics, but there are some inconsistencies in the dependence on particle size 76. Gas- and condensed-phase products have been identified to elucidate detailed reaction mechanisms, including small molecular weight aldehydes and acids, and high molecular weight organic peroxides from particle-phase secondary chemistry. Oleic acid particle and film oxidation studies indicate that the physicochemical properties of unsaturated particles can be significantly modified due to ozonolysis. The heterogeneous ozonolysis reaction leads to the reduction in particle size 78 due to formation of volatile aldehyde products (e.g. nonanal for oleic acid ozonolysis) and an increase in particle densities for oleic acid coated particles 79. An enhancement in the hygroscopicity of oleic acid films and particles has been observed from ozonolysis for high O 3 concentrations for subsaturated water uptake, 68 and with respect to CCN activation. 80, Radical-Initiated Oxidation Organic matter in atmospheric aerosols contains a significant fraction of saturated organic species that are unreactive to O 3 but should react efficiently with atmospheric radicals such as OH, NO 3 and halogen radicals. 82 There have been

46 28 relatively few laboratory studies on the radical-initiated oxidation of condensedorganics at the time of the start of this thesis work. This is partly due to the experimental challenges involved in laboratory studies of highly reactive radical species that are more daunting than studies with more stable molecules such as O 3. The following is an overview of the research on heterogeneous oxidation reactions with organic surfaces, focusing on the studies published at the start of this thesis work. Chapter 6 will reflect the major developments in our current understanding of heterogeneous oxidation. Radical uptake on organic surfaces is extremely efficient with high measured uptake coefficients. Cooper and Abbatt reported an OH uptake coefficient of γ OH > 0.2 for 1-hexanol adsorbed to ammonium sulfate coatings in a coated-wall flow tube, where OH was monitored using resonance fluorescence. 83 Bertram et al. expanded on this work using a coated-wall flow tube with CIMS measuring γ OH for various surface types including self assembled organic monolayers, halocarbon wax, stearic/palmitic acid mixture, pyrene, paraffin, soot and alumina coatings. 84 The reactive loss of OH on organic coatings was much more efficient (γ OH > 0.1) than on halocarbon wax (γ OH < ). Cl uptake measured by Moise and Rudich was as efficient as OH on alkane and alkene monolayer surfaces (γ Cl > 0.1), whereas Br was less efficient (γ Br = ). 70 The NO 3 radical is less reactive for a range of alkane and alkene surfaces than these other radicals (γ NO3 < 0.015), with general trends in reactivity that are consistent with analogous gas-phase reactions. 85,86

47 29 Surface analytical techniques have been applied to observe the modification of organic surface properties due to the radical oxidation. 70,71,84,86 Contact angle measurements indicated that organic monolayers became more hydrophilic from exposure to OH, Cl and Br radicals. 70,71,84 Molina et al. 71 and Eliason et al. 87 presented the first studies to measure gas-phase and condensed-phase products, respectively, from the radical oxidation of organic films. Molina measured CO, CO 2 and small molecular weight oxygenated VOCs, such as HCOOH and CH 3 OH, from the OH oxidation of paraffin films. 71 Several surface characterization techniques, including XPS and FTIR, showed that carbon was efficiently volatilized from OH oxidation of an alkane monolayer from the reaction, with 2-3 OH needed to volatilize a C18 monolayer. Volatilization was much less efficient for the aromatic monolayer. Eliason et al. identified condensed-phase oxidation products by derivatization of carbonyls followed by GC-MS, including ketones and small aldehydes, from the OH oxidation of hexadecane and 1-dodecene under NO x -free conditions. 87 A reaction mechanism (Figure 1.4) for the radical-initiated oxidation of an alkane (RH) was proposed by Molina et al. that is analogous to gas-phase oxidation chemistry. 71 Although the study focused on OH oxidation, the reaction mechanism is expected to proceed in a similar manner for other atmospheric radical oxidants. The initial step is hydrogen abstraction by the oxidant (e.g. OH in Figure 1.4), followed by rapid reaction with molecular oxygen to the alkyl radical (R) to form an alkyl peroxy radical (RO 2 ). The RO 2 radical reacts with NO or another RO 2 to an form alkoxy radical (RO) or RONO 2, from which Molina et al. found self-reaction to be

48 30 dominant. 71 Peroxides may form from the reaction of RO 2 with HO 2. The RO 2 selfreaction proceeds via the Russell mechanism forming a carbonyl and alcohol, 88 which appears to be the dominant reaction pathway in organic liquids but not in gas phase. 89 The RO radical may follow one of several reaction pathways to form more stable products including reaction with NO x forming RONO x, reaction with O 2 producing a carbonyl and HO 2, decomposition, or isomerization. The decomposition reaction is the only pathway that forms smaller molecular weight products through C-C scission, yielding an aldehyde and alkyl radical, whereas the other pathways produce higher molecular weight oxygenated products compared to the original alkane. The isomerization pathway leads to a 1,5 H-shift on the carbon chain forming an alcohol group on the alkyl radical. The work by Molina et al. indicated that the decomposition pathway is dominant. Recent studies on radical-initiated oxidation of organic aerosols have observed less significant volatilization, suggesting the other pathways may be important in aerosols. 70,86,89 Therefore, the relative importance of each pathway in the reaction mechanism for organic aerosols is still unclear.

49 31 RH OH ROOH HO 2 R RONO x NO O 2 RO 2 RO 2 RO 2, NO ROH + R R C=O (Russell mechanism) NO x RO O 2 ROH R R C=O + HO 2 (Isomerization) R CHO + R (Decomposition) Figure 1.4. Proposed reaction mechanism for radical-initiated oxidation of alkane (adapted from Molina et al. 71 ) Evidence for Heterogeneous Oxidation in the Field The most direct evidence for heterogeneous oxidation in the atmosphere has been reported from the analysis of measurements of pairs of reactive chemical markers in OA normalized to unreactive aerosol components as reference species. Robinson et al. found that a pair of hopanes, chemical markers for gasoline that should be reactive to OH, normalized to EC, were significantly depleted in particle filter samples taken during the summer, i.e. during times of high photochemical activity, compared to winter samples. 90 This work is consistent with efficient

50 32 heterogeneous oxidation. The comparison of oleic and palmitoleic acids normalized to stearic acid as a pair of unsaturated markers for evidence of aging was less convincing. 90 Recent AMS field studies have shown that organic aerosols become more aged and oxygenated with atmospheric residence time, determined by the observation of an increase in the organic mass fraction of m/z 44 (CO + 2 ) that is a marker in AMS organic mass spectra for oxygenated organics. 12,58,91,92 Because these studies have observed an increase in organic mass concomitant with photochemical aging time, SOA formation has generally been implicated as the dominant chemical aging process responsible for observed overall oxidation changes. Furthermore, Zhang et al. have suggested that heterogeneous oxidation resulting in the conversion of hydrocarbon-like organic aerosol (HOA) mass to oxygenated organic aerosol (OOA) mass cannot explain the high OOA fraction in aged aerosols due to the low HOA fractions in ambient organic aerosols observed in most regions. 56 Clearly, both heterogeneous oxidation and SOA formation can occur during periods of high photochemical activity. Decoupling these two processes and their effect on the bulk properties of organic aerosols with aging time remains difficult due to our lack of understanding of heterogeneous oxidation Emerging Research Issues Regarding Heterogeneous Oxidation Given that the current state of research regarding heterogeneous oxidation of atmospheric aerosols, especially by atmospheric radicals, is still in its infancy, many questions remain to be answered in this area. At the start of this thesis, there had been

51 33 essentially no published works on the radical-initiated oxidation of organic particles. These reactions have only recently been studied for laboratory organic aerosols and considerable work remains to gain a full picture of the impact of heterogeneous oxidation as an atmospheric aerosol aging mechanism. The kinetics and reaction mechanism for radical oxidation of organic aerosols are not well understood. As discussed previously, reactive uptake coefficients on organic aerosols have been quantified for O 3 uptake onto unsaturated organics and radical uptake onto a few pure organic surfaces. At the start of this thesis work, the uptake coefficient for radicals on saturated organic aerosol particles had not been quantified. The kinetic mechanism for radical uptake on organic aerosols had not been established, but it most likely proceeds through a surface reaction (Table 1.2 Case 4) due to high OH reactivity with organics surfaces. Several studies have shown that particle phase can significantly affect reaction kinetics of heterogeneous ozonolysis of unsaturated organic aerosols. 74,93-95 The kinetics and reaction mechanism for radical oxidation of organic aerosols may also be affected by particlephase. The reaction mechanism for the specific heterogeneous oxidation system of O 3 + oleic acid has been elucidated in detail including secondary chemistry of particle-phase reaction intermediates. 74 For radical heterogeneous oxidation, there has been limited evidence that the reaction mechanism may proceed in a similar fashion to gas-phase oxidation of VOCs, 71,87 but the relative importance of the reaction pathways in Figure 1.4 has not been determined and major oxidation products have

52 34 not been identified for organic particles. For example, there has been considerable disagreement on the importance of the decomposition pathway from observed volatilization of smaller molecular weight reaction products. If volatilization is the major pathway for OH oxidation with yields suggested by Molina et al. 71, OH oxidation would reduce particle size and may be a significant sink for organic aerosols, resulting in a potentially important source of oxygenated VOCs in the atmosphere. 96 Discrepancies may be due to the initial oxidant and experimental conditions. The role of particle-phase secondary chemistry of oxidation intermediates may be an important deviation to gas-phase chemistry that alters the expected particle composition. The effect of heterogeneous oxidation on particle properties is a key area of research. The modification of particle size, density, composition and hygroscopic properties from radical oxidation of organic aerosols may alter the role of organic aerosols in climate. This work has been initiated for ozonolysis of unsaturated organics, 79,80 but the effect of radical reactions on properties of saturated organic aerosols had not been studied at the start of this thesis. Particularly, the modification of hygroscopic properties from heterogeneous reaction will determine the extent to which this chemical aging mechanism may alter the wet scavenging rates and lifetime of organic aerosols. In sum, it has not been determined whether radical-initiated heterogeneous oxidation will influence the physico-chemical properties and lifetime of organic aerosols for atmospherically relevant timescales.

53 RESEARCH OBJECTIVES The major goal of this thesis is to investigate the importance of heterogeneous oxidation as a mechanism of chemical aging of organic aerosols. In order to address the gaps in knowledge discussed in Section 1.6.5, laboratory studies were conducted to investigate OH-initiated heterogeneous oxidation. In particular, this work aimed to determine the extent to which this reaction modifies the chemical composition and physical properties of organic aerosol particles. This research attempted to answer the following specific questions: 1. How efficient is OH heterogeneous uptake kinetics on organic aerosols and what are the major oxidation products? 2. How does the reaction alter the particle properties and does the reaction lead to significant volatilization of organic aerosol mass? 3. Does the reaction lead to a modification in the particle hygroscopicity? 4. Does the reaction enhance the degree of oxidation of complex and oxygenated aerosols? Online particle analytical techniques including an Aerodyne Time-of-Flight Aerosol Mass Spectrometer (ToF-AMS) were employed to monitor the modification of particle composition and physical properties of organic aerosols during heterogeneous oxidation. The effect of chemical aging on hygroscopicity of organic aerosol particles

54 36 was determined by the measurement of the CCN activation of organic aerosols exposed to OH radicals. The kinetics and reaction mechanism for OH-initiated heterogeneous oxidation of model POA was studied (Chapter 2). The kinetics from the loss of condensed-phase reactions was measured with the ToF-AMS and the reaction mechanism proposed was based on offline analysis of condensed phase products with electrospray mass spectrometry. The modification of the CCN activity of model POA from heterogeneous oxidation was observed (Chapter 3) and the role of surface tension in the hygroscopicity changes was evaluated. Chapter 4 summarizes results from OH-oxidation of ambient aerosols in two locations, where observed changes in chemical composition varied with initial aerosol composition. Chapter 5 presents work on the OH-oxidation of model SOA produced from the reaction of α-pinene with O 3 in an attempt to gauge the ability of chemical aging to alter the chemical composition of SOA. Results from these chapters are summarized in the context of recent work in Chapter 6 and outstanding research issues regarding chemical aging of organic aerosols are discussed.

55 REFERENCES (1) Seinfeld, J. H.; Pandis, S. N. Atmospheric chemistry and physics: from air pollution to climate change; John Wiley and Sons, Inc.: New York, (2) Saxena, P.; Hildemann, L. M. Journal of Atmospheric Chemistry 1996, 24, 57. (3) Andreae, M. O.; Crutzen, P. J. Science 1997, 276, (4) Goldstein, A. H.; Galbally, I. E. Environmental Science & Technology 2007, 41, (5) IPCC, 2007: Summary for Policymakers. In: Climate Change 2007: The Physical Science Basis. Contribution of Working Group I to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change [Solomon, S., D. Qin, M. Manning, Z. Chen, M. Marquis, K.B. Averyt, M.Tignor and H.L. Miller (eds.)]. Cambridge University Press, Cambridge, United Kingdom and New York, NY, USA.,, (6) Bernstein, J. A.; Alexis, N.; Barnes, C.; Bernstein, I. L.; Bernstein, J. A.; Nel, A.; Peden, D.; Diaz-Sanchez, D.; Tarlo, S. M.; Williams, P. B. Journal of Allergy and Clinical Immunology 2004, 114, (7) Poschl, U. Angewandte Chemie-International Edition 2005, 44, (8) O'Dowd, C. D.; Facchini, M. C.; Cavalli, F.; Ceburnis, D.; Mircea, M.; Decesari, S.; Fuzzi, S.; Yoon, Y. J.; Putaud, J. P. Nature 2004, 431, 676. (9) Tsigaridis, K.; Kanakidou, M. Atmospheric Chemistry and Physics 2003, 3, 1849.

56 38 (10) Henze, D. K.; Seinfeld, J. H.; Ng, N. L.; Kroll, J. H.; Fu, T. M.; Jacob, D. J.; Heald, C. L. Atmospheric Chemistry and Physics 2008, 8, (11) Kroll, J. H.; Seinfeld, J. H. Atmospheric Environment 2008, 42, (12) de Gouw, J. A.; Middlebrook, A. M.; Warneke, C.; Goldan, P. D.; Kuster, W. C.; Roberts, J. M.; Fehsenfeld, F. C.; Worsnop, D. R.; Canagaratna, M. R.; Pszenny, A. A. P.; Keene, W. C.; Marchewka, M.; Bertman, S. B.; Bates, T. S. Journal of Geophysical Research-Atmospheres 2005, 110, D (13) Heald, C. L.; Jacob, D. J.; Park, R. J.; Russell, L. M.; Huebert, B. J.; Seinfeld, J. H.; Liao, H.; Weber, R. J. Geophysical Research Letters 2005, 32, L (14) Volkamer, R.; Jimenez, J. L.; San Martini, F.; Dzepina, K.; Zhang, Q.; Salcedo, D.; Molina, L. T.; Worsnop, D. R.; Molina, M. J. Geophysical Research Letters 2006, 33, L (15) Robinson, A. L.; Donahue, N. M.; Shrivastava, M. K.; Weitkamp, E. A.; Sage, A. M.; Grieshop, A. P.; Lane, T. E.; Pierce, J. R.; Pandis, S. N. Science 2007, 315, (16) Lim, Y. B.; Ziemann, P. J. Environmental Science & Technology 2005, 39, (17) Gao, S.; Ng, N. L.; Keywood, M.; Varutbangkul, V.; Bahreini, R.; Nenes, A.; He, J. W.; Yoo, K. Y.; Beauchamp, J. L.; Hodyss, R. P.; Flagan, R. C.; Seinfeld, J. H. Environmental Science & Technology 2004, 38, (18) Kanakidou, M.; Seinfeld, J. H.; Pandis, S. N.; Barnes, I.; Dentener, F. J.; Facchini, M. C.; Van Dingenen, R.; Ervens, B.; Nenes, A.; Nielsen, C. J.;

57 39 Swietlicki, E.; Putaud, J. P.; Balkanski, Y.; Fuzzi, S.; Horth, J.; Moortgat, G. K.; Winterhalter, R.; Myhre, C. E. L.; Tsigaridis, K.; Vignati, E.; Stephanou, E. G.; Wilson, J. Atmospheric Chemistry and Physics 2005, 5, (19) Pruppacher, H., R.; Klett, J. D. Microphysics of clouds and precipitation; D. Reidel: Dordrecht, (20) Zappoli, S.; Andracchio, A.; Fuzzi, S.; Facchini, M. C.; Gelencser, A.; Kiss, G.; Krivacsy, Z.; Molnar, A.; Meszaros, E.; Hansson, H. C.; Rosman, K.; Zebuhr, Y. Atmospheric Environment 1999, 33, (21) Broekhuizen, K.; Kumar, P. P.; Abbatt, J. P. D. Geophysical Research Letters 2004, 31, L (22) Shulman, M. L.; Jacobson, M. C.; Carlson, R. J.; Synoveca, R. E.; Young, T. E. Geophysical Research Letters 1996, 23, 277. (23) Facchini, M. C.; Mircea, M.; Fuzzi, S.; Charlson, R. J. Nature 1999, 401, 257. (24) Kiss, G.; Tombacz, E.; Hansson, H. C. Journal of Atmospheric Chemistry 2005, 50, 279. (25) Taraniuk, I.; Graber, E. R.; Kostinski, A.; Rudich, Y. Geophysical Research Letters 2007, 34, L (26) Salma, I.; Ocskay, R.; Varga, I.; Maenhaut, W. Journal of Geophysical Research-Atmospheres 2006, 111, D (27) Shantz, N. C.; Leaitch, W. R.; Caffrey, P. F. Journal of Geophysical Research-Atmospheres 2003, 108, (28) Sun, J. M.; Ariya, P. A. Atmospheric Environment 2006, 40, 795.

58 40 (29) Petters, M. D.; Kreidenweis, S. M. Atmospheric Chemistry and Physics 2007, 7, (30) Rissler, J.; Vestin, A.; Swietlicki, E.; Fisch, G.; Zhou, J.; Artaxo, P.; Andreae, M. O. Atmospheric Chemistry and Physics 2006, 6, 471. (31) Wex, H.; Hennig, T.; Salma, I.; Ocskay, R.; Kiselev, A.; Henning, S.; Massling, A.; Wiedensohler, A.; Stratmann, F. Geophysical Research Letters 2007, 34, L (32) Padro, L. T.; Asa-Awuku, A.; Morrison, R.; Nenes, A. Atmospheric Chemistry and Physics 2007, 7, (33) Asa-Awuku, A.; Sullivan, A. P.; Hennigan, C. J.; Weber, R. J.; Nenes, A. Atmospheric Chemistry and Physics 2008, 8, 799. (34) Engelhart, G. J.; Asa-Awuku, A.; Nenes, A.; Pandis, S. N. Atmospheric Chemistry and Physics 2008, 8, (35) Huff Hartz, K. E.; Rosenørn, T.; Ferchak, S. R.; Raymond, T. M.; Bilde, M.; Donahue, N. M.; Pandis, S. N. Journal of Geophysical Research 2005, 110, D (36) McMurry, P. H. Atmospheric Environment 2000, 34, (37) Jacobson, M. C.; Hansson, H. C.; Noone, K. J.; Charlson, R. J. Reviews of Geophysics 2000, 38, 267. (38) Gelencser, A. Carbonaceous aerosol; Springer: Dordrecht, (39) Decesari, S.; Facchini, M. C.; Fuzzi, S.; Tagliavini, E. Journal of Geophysical Research-Atmospheres 2000, 105, 1481.

59 41 (40) Russell, L. M. Environmental Science & Technology 2003, 37, (41) Semeniuk, T. A.; Wise, M. E.; Martin, S. T.; Russell, L. M.; Buseck, P. R. Journal of Atmospheric Chemistry 2007, 56, 259. (42) Takahama, S.; Gilardoni, S.; Russell, L. M.; Kilcoyne, A. L. D. Atmospheric Environment 2007, 41, (43) Weber, R. J.; Orsini, D.; Daun, Y.; Lee, Y. N.; Klotz, P. J.; Brechtel, F. Aerosol Science and Technology 2001, 35, 718. (44) Sullivan, A. P.; Weber, R. J.; Clements, A. L.; Turner, J. R.; Bae, M. S.; Schauer, J. J. Geophysical Research Letters 2004, 31, L (45) Graber, E. R.; Rudich, Y. Atmospheric Chemistry and Physics 2006, 6, 729. (46) Tobias, H. J.; Kooiman, P. M.; Docherty, K. S.; Ziemann, P. J. Aerosol Science and Technology 2000, 33, 170. (47) Canagaratna, M. R.; Jayne, J. T.; Jimenez, J. L.; Allan, J. D.; Alfarra, M. R.; Zhang, Q.; Onasch, T. B.; Drewnick, F.; Coe, H.; Middlebrook, A.; Delia, A.; Williams, L. R.; Trimborn, A. M.; Northway, M. J.; DeCarlo, P. F.; Kolb, C. E.; Davidovits, P.; Worsnop, D. R. Mass Spectrometry Reviews 2007, 26, 185. (48) Prather, K. A.; Hatch, C. D.; Grassian, V. H. Annual Review of Analytical Chemistry 2008, 1, 485. (49) Hoffmann, T.; Bandur, R.; Hoffmann, S.; Warscheid, B. Spectrochimica Acta Part B 2002, 57, (50) Murphy, D. M.; Thomson, D. S. Aerosol Science and Technology 1995, 22, 237.

60 42 (51) Jayne, J. T.; Leard, D. C.; Zhang, X. F.; Davidovits, P.; Smith, K. A.; Kolb, C. E.; Worsnop, D. R. Aerosol Science and Technology 2000, 33, 49. (52) McLafferty, F. W.; Turecek, F. Interpretation of mass spectra; University Science Books: Mill Valley, (53) Zhang, Q.; Alfarra, M. R.; Worsnop, D. R.; Allan, J. D.; Coe, H.; Canagaratna, M. R.; Jimenez, J. L. Environmental Science & Technology 2005, 39, (54) Lanz, V. A.; Alfarra, M. R.; Baltensperger, U.; Buchmann, B.; Hueglin, C.; Prevot, A. S. H. Atmospheric Chemistry and Physics 2007, 7, (55) Paatero, P.; Tapper, U. Environmetrics 1994, 5, 111. (56) Zhang, Q.; Jimenez, J. L.; Canagaratna, M. R.; Allan, J. D.; Coe, H.; Ulbrich, I.; Alfarra, M. R.; Takami, A.; Middlebrook, A. M.; Sun, Y. L.; Dzepina, K.; Dunlea, E.; Docherty, K.; DeCarlo, P. F.; Salcedo, D.; Onasch, T.; Jayne, J. T.; Miyoshi, T.; Shimono, A.; Hatakeyama, S.; Takegawa, N.; Kondo, Y.; Schneider, J.; Drewnick, F.; Borrmann, S.; Weimer, S.; Demerjian, K.; Williams, P.; Bower, K.; Bahreini, R.; Cottrell, L.; Griffin, R. J.; Rautiainen, J.; Sun, J. Y.; Zhang, Y. M.; Worsnop, D. R. Geophysical Research Letters 2007, 34, L (57) Takegawa, N.; Miyakawa, T.; Kondo, Y.; Blake, D. R.; Kanaya, Y.; Koike, M.; Fukuda, M.; Komazaki, Y.; Miyazaki, Y.; Shimono, A.; Takeuchi, T. Geophysical Research Letters 2006, 33, L15814.

61 43 (58) McFiggans, G.; Alfarra, M. R.; Allan, J.; Bower, K.; Coe, H.; Cubison, M.; Topping, D.; Williams, P.; Decesari, S.; Facchini, C.; Fuzzi, S. Faraday Discussions 2005, 130, 341. (59) Jang, M. S.; Czoschke, N. M.; Lee, S.; Kamens, R. M. Science 2002, 298, 814. (60) Kalberer, M.; Paulsen, D.; Sax, M.; Steinbacher, M.; Dommen, J.; Prevot, A. S. H.; Fisseha, R.; Weingartner, E.; Frankevich, V.; Zenobi, R.; Baltensperger, U. Science 2004, 303, (61) Kalberer, M.; Sax, M.; Samburova, V. Environmental Science & Technology 2006, 40, (62) Kolb, C. E.; Davidovits, P.; Jayne, J. T.; Shi, Q.; Worsnop, D. R. Progress in Reaction Kinetics and Mechanism 2002, 27, 1. (63) Worsnop, D. R.; Morris, J. W.; Shi, Q.; Davidovits, P.; Kolb, C. E. Geophysical Research Letters 2002, 29, (64) Fuchs, N. A.; Sutugin, A. G. Highly dispersed aerosols; Ann Arbor Science Publishers: Ann Arbor, (65) Smith, G. D.; Woods, E.; DeForest, C. L.; Baer, T.; Miller, R. E. Journal of Physical Chemistry A 2002, 106, (66) Atkinson, R.; Arey, J. Chemical Reviews 2003, 103, (67) Rudich, Y. Chemical Reviews 2003, 103, (68) Asad, A.; Mmereki, B. T.; Donaldson, D. J. Atmospheric Chemistry and Physics 2004, 4, 2083.

62 44 (69) Mmereki, B. T.; Donaldson, D. J. Physical Chemistry Chemical Physics 2002, 4, (70) Moise, T.; Rudich, Y. Geophysical Research Letters 2001, 28, (71) Molina, M. J.; Ivanov, A. V.; Trakhtenberg, S.; Molina, L. T. Geophysical Research Letters 2004, 31, L (72) Katrib, Y.; Martin, S. T.; Hung, H. M.; Rudich, Y.; Zhang, H. Z.; Slowik, J. G.; Davidovits, P.; Jayne, J. T.; Worsnop, D. R. Journal of Physical Chemistry A 2004, 108, (73) Hearn, J. D.; Smith, G. D. Journal of Physical Chemistry A 2004, 108, (74) Ziemann, P. J. Faraday Discussions 2005, 130, 469. (75) Zahardis, J.; LaFranchi, B. W.; Petrucci, G. A. Atmospheric Environment 2006, 40, (76) Zahardis, J.; Petrucci, G. A. Atmospheric Chemistry and Physics 2007, 7, (77) Rudich, Y.; Donahue, N. M.; Mentel, T. F. Annual Review of Physical Chemistry 2007, 58, 321. (78) Thornberry, T.; Abbatt, J. P. D. Physical Chemistry Chemical Physics 2004, 6, 84. (79) Katrib, Y.; Martin, S. T.; Rudich, Y.; Davidovits, P.; Jayne, J. T.; Worsnop, D. R. Atmospheric Chemistry and Physics 2005, 5, 275. (80) Broekhuizen, K. E.; Thornberry, T.; Kumar, P. P.; Abbatt, J. P. D. Journal of Geophysical Research-Atmospheres 2004, 109, D24206.

63 45 (81) Shilling, J. E.; King, S. M.; Mochida, M.; Martin, S. T. Journal of Physical Chemistry A 2007, 111, (82) Rogge, W. F.; Mazurek, M. A.; Hildemann, L. M.; Cass, G. R.; Simoneit, B. R. T. Atmospheric Environment Part a-general Topics 1993, 27, (83) Cooper, P. L.; Abbatt, J. P. D. Journal of Physical Chemistry 1996, 100, (84) Bertram, A. K.; Ivanov, A. V.; Hunter, M.; Molina, L. T.; Molina, M. J. Journal of Physical Chemistry A 2001, 105, (85) Moise, T.; Talukdar, R. K.; Frost, G. J.; Fox, R. W.; Rudich, Y. Journal of Geophysical Research-Atmospheres 2002, 107, (86) Knopf, D. A.; Mak, J.; Gross, S.; Bertram, A. K. Geophysical Research Letters 2006, 33, L (87) Eliason, T. L.; Gilman, J. B.; Vaida, V. Atmospheric Environment 2004, 38, (88) Russell, G. A. Journal of the American Chemical Society 1957, 79, (89) Docherty, K. S.; Ziemann, P. J. Journal of Physical Chemistry A 2006, 110, (90) Robinson, A. L.; Donahue, N. M.; Rogge, W. F. Journal of Geophysical Research-Atmospheres 2006, 111, D (91) Takegawa, N.; Miyakawa, T.; Kawamura, K.; Kondo, Y. Aerosol Science and Technology 2007, 41, 418.

64 46 (92) Weber, R. J.; Sullivan, A. P.; Peltier, R. E.; Russell, A.; Yan, B.; Zheng, M.; de Gouw, J.; Warneke, C.; Brock, C.; Holloway, J. S.; Atlas, E. L.; Edgerton, E. Journal of Geophysical Research-Atmospheres 2007, 112, D (93) Hearn, J. D.; Smith, G. D. Physical Chemistry Chemical Physics 2005, 7, (94) Katrib, Y.; Biskos, G.; Buseck, P. R.; Davidovits, P.; Jayne, J. T.; Mochida, M.; Wise, M. E.; Worsnop, D. R.; Martin, S. T. Journal of Physical Chemistry A 2005, 109, (95) Knopf, D. A.; Anthony, L. M.; Bertram, A. K. Journal of Physical Chemistry A 2005, 109, (96) Kwan, A. J.; Crounse, J. D.; Clarke, A. D.; Shinozuka, Y.; Anderson, B. E.; Crawford, J. H.; Avery, M. A.; McNaughton, C. S.; Brune, W. H.; Singh, H. B.; Wennberg, P. O. Geophysical Research Letters 2006, 33, L15815.

65 47 Chapter Two Heterogeneous Oxidation of Saturated Organic Aerosols by Hydroxyl Radicals: Uptake Kinetics, Condensed- Phase Products, and Particle Size Change I. J. George, A. Vlasenko, J. G. Slowik, K. Broekhuizen, and J. P. D. Abbatt, Atmos. Chem. Phys., 2007, 7, Reproduced with permission from the Atmospheric Chemistry and Physics Copyright Copernicus 2007.

66 ABSTRACT The kinetics and reaction mechanism for the heterogeneous oxidation of saturated organic aerosols by gas-phase OH radicals were investigated under NO x - free conditions. The reaction of 150 nm diameter Bis(2-ethylhexyl) sebacate (BES) particles with OH was studied as a proxy for chemical aging of atmospheric aerosols containing saturated organic matter. An aerosol reactor flow tube combined with an Aerodyne time-of-flight aerosol mass spectrometer (ToF-AMS) and scanning mobility particle sizer (SMPS) was used to study this system. Hydroxyl radicals were produced by 254 nm photolysis of O 3 in the presence of water vapour. The kinetics of the heterogeneous oxidation of the BES particles was studied by monitoring the loss of a mass fragment of BES with the ToF-AMS as a function of OH exposure. We measured an initial OH uptake coefficient of γ 0 = 1.3(±0.4), confirming that this reaction is highly efficient. The density of BES particles increased by up to 20% of the original BES particle density at the highest OH exposure studied, consistent with the particle becoming more oxidized. Electrospray ionization mass spectrometry analysis showed that the major particle-phase reaction products are multifunctional carbonyls and alcohols with higher molecular weights than the starting material. Volatilization of oxidation products accounted for a maximum of 17% decrease of the particle volume at the highest OH exposure studied. Tropospheric organic aerosols will become more oxidized from heterogeneous photochemical oxidation, which may affect not only their physical and chemical properties, but also their hygroscopicity and cloud nucleation activity.

67 INTRODUCTION Atmospheric aerosols play an important role in atmospheric chemistry, climate, visibility, and human health. 1-3 Knowing that organic matter can make up a significant fraction of tropospheric aerosol, 4,5 the question arises as to whether atmospheric organic particles can be chemically transformed by heterogeneous reactions with gas-phase oxidants such as O 3, OH, Cl and NO 3. Chemical aging of organic particles from heterogeneous oxidation modifies particle physico-chemical properties such as size, morphology, composition, hygroscopicity, and ability to act as cloud nuclei. Therefore, aging may significantly impact the role of organic aerosols on climate, atmospheric chemistry and other relevant environmental processes. Despite the importance of chemical aging of atmospheric organic aerosols, our current understanding of this process is limited. Recently, there has been a focus of research on understanding the chemical transformation of atmospheric organic aerosols. Laboratory research has concentrated on the reaction of O 3 with condensed-phase unsaturated organic compounds, e.g., oleic acid, as a proxy for chemical aging of organic aerosols Fewer studies have been conducted to investigate the chemical aging of condensedphase saturated organics by atmospheric radicals OH is the most efficient atmospheric oxidant in the troposphere, and therefore, it is expected to play an important role in chemical aging of atmospheric condensed organic matter. Several studies have shown that the reaction of organic surfaces with OH is highly efficient compared to reaction with other radicals such as

68 50 NO 20,23 3 and Cl 16, with OH reactive uptake probabilities ranging from 0.2 to 1. 15,18,24 A few attempts have been made to elucidate the reaction mechanism for the heterogeneous reaction of OH with organic surfaces by observing reaction products. Molina et al. observed complete volatilization of alkane monolayers from reaction with OH, indicating that heterogeneous oxidation of organic aerosol by OH leads to the release of oxygenated volatile organic products into the troposphere. 18 Molina et al. further suggest that volatilization from chemical aging by reaction with OH may be as important as wet deposition as a removal pathway for organic aerosols in the lower atmosphere. 18 However, other studies suggest that the reaction pathway leading to release of volatile products from the oxidation of organic surfaces by OH plays a minor role in the reaction mechanism. A more recent study by Knopf et al. found that oxidation of alkane monolayers by atmospheric exposures of NO 3 leads to volatilization of no more than 10% of the surface. 20 Moise and Rudich measured 20% carbon loss from an alkane monolayer due to reaction with halogen radicals. 16 The results from these studies contrast with the Molina et al. findings, even though the oxidation of a saturated organic surface by Cl or NO 3 is expected to proceed through a similar reaction mechanism as oxidation by OH. Therefore, the importance of volatilization resulting from the oxidation of atmospheric saturated organic matter by OH is currently still unclear. The study of condensed-phase products from heterogeneous oxidation of saturated organic surfaces has been largely restricted to X-ray photoelectron

69 51 spectroscopy (XPS) analysis of oxidized alkane monolayer surfaces, which provides information on the degree of oxidation and on the presence of oxidized functional groups. 16,18,20 Eliason et al. reported the formation of condensed-phase ketones and short-chained aldehydes from the reaction of OH with hexadecane film using GC-MS analysis. 17 Although these studies suggest that atmospheric organic aerosols should become more oxidized from chemical aging, this has not yet been confirmed in the laboratory with saturated organic aerosols. Laboratory studies involving the oxidation of organic surfaces by OH have typically used organic monolayers or films as proxies for organic particles, even though the greater physical and chemical complexity of particles may influence the aging process. Further, there have been no efforts to simultaneously characterize both gas- and particle-phase products from the heterogeneous reaction of OH with organic surfaces or particles. Therefore, because of the uncertainties mentioned above, the reaction mechanism for this process is not fully understood. To address the gaps in knowledge of the aging process of organic particles, we investigate the heterogeneous reaction of organic aerosols with OH under NO x -free conditions using online particle analysis techniques. The experimental NO x -free conditions in this study were chosen as an initial simplified case, which simulates atmospheric conditions in remote regions. Bis(2-ethylhexyl) sebacate (C 26 H 50 O 4, BES) aerosols were selected as a proxy for atmospheric aerosol containing saturated organic matter. The chemical structure of BES is shown in Figure 2.1. Because of its

70 52 structure, BES has several properties that make BES particles a good proxy for organic aerosols in this laboratory study. BES has a very low vapour pressure ( Pa at 25ºC) to minimize gas-phase reactions with OH. It also contains no unsaturated moieties and, therefore, will not react with O 3. Further, BES particles are liquid at room temperature eliminating complications of particle shape on particle analysis. The use of liquid organic aerosols allows for the study of heterogeneous kinetics including surface and bulk particle processes more closely mimicking aging processes occurring in organic aerosols unlike the study of organic films. Figure 2.1. Chemical structure of Bis(2-ethylhexyl) sebacate (BES). The heterogeneous reaction of BES particles with OH was studied in an aerosol reactor flow tube coupled to an Aerodyne time-of-flight aerosol mass spectrometer (ToF-AMS). There are four major objectives of this study: 1) measure the kinetics of the heterogeneous reaction of OH with BES particles by observing the reactive loss of particle-phase BES, 2) determine the effect of the reaction on the particle physical properties (particle size, density, and extent of volatilization), 3) characterize the particle-phase products, and 4) elucidate the reaction mechanism.

71 53 This study focuses on the identification of condensed-phase products, while a future paper 25 will discuss the characterization of volatilized products EXPERIMENTAL Aerosol Reactor Flow Tube Setup The experimental system used to study the heterogeneous oxidation of BES particles is shown in Figure 2.2. BES particles were generated by homogeneous nucleation by passing 0.3 slpm flow of N 2 (BOC, %) through a Pyrex tube containing liquid BES (Fluka, > 97%), that is heated to approximately 110ºC. The aerosol flow passed through a neutralizer (TSI model 3077), and the particles were then size-selected to monodisperse aerosol with a mode mobility diameter (D m ) of D m = 150 nm by a differential mobility analyzer (DMA) column (TSI model 3081). The ratio of sheath to sample flow rates in the DMA was held at a constant value of 10. The monodisperse aerosol then passed through an activated carbon trap to remove volatile organics from the flow.

72 54 Figure 2.2. Aerosol flow reactor setup for the study of the heterogeneous oxidation of organic aerosols. OH radicals were produced by the photolysis of O 3 in the presence of water vapour as follows. O 3 was generated by passing a mixture of N 2 and O 2 (BOC, %) through an O 3 generator (Jelight model 1000). O 2 concentrations in the flow tube ranged from approximately [O 2 ] = molecules cm -3. In general, O 2 concentrations positively correlated with O 3 concentrations and OH exposures. The influence of O 2 concentration on the reaction was examined by adjusting the O 2 concentrations at fixed OH exposures by the addition of a pure O 2 flow to the ozone flow. O 3 concentrations were monitored before entering the aerosol reactor flow tube by measuring absorbance at 254 nm in a 14 cm long UV absorption

73 55 cell. The O 3 flow was mixed with a humidified N 2 flow that had passed through a water bubbler. The humidified O 3 and aerosol flows having a total volumetric flow rate of approximately 0.7 slpm were passed into a mixing flow tube 0.8 L in volume. The mixed humidified O 3 /aerosol flow was then introduced into the reactor flow tube that was at 1 atm and room temperature of 24ºC. The mixture was illuminated in the flow tube by a 22.9 cm O 3 -free Hg Pen-ray lamp (UVP) inside the reactor flow tube. The UV lamp had a primary energy output at 254 nm and was surrounded by a Quartz sheath tube that was purged with compressed air to cool the lamp. The UV lamp housing was treated to filter out 185 nm light to prevent O 3 production by the lamp. The average reaction time of OH with the BES particles was calculated from the illuminated volume (V = 0.87 L) and the flow rate. An O 3 denuder was placed downstream of the reactor flow tube to destroy the O 3 in the flow. After passing through the O 3 denuder, the reacted particles were analyzed (see sections and 2.2.4) OH Quantification The steady-state OH concentrations were calculated in each experiment with a photochemical model using the Acuchem program. 26 Experimental inputs into the model were O 3 and H 2 O concentrations in the reactor flow tube. The O 3 concentrations in the reactor flow tube, with the range of [O 3 ] = molecules cm -3, were determined from the O 3 concentrations at the O 3 detector (see Figure 2.2) and the volumetric flow rates of the O 3 flow and total mixed flow. The relative

74 56 humidity of the flow exiting the reactor flow tube was measured with a hygrometer (VWR, ±1%) and varied between approximately 30 and 60%. The Acuchem model included gas-phase reactions involved in the production of OH from O 3 and water vapour and major OH loss reactions with kinetic rate constants associated with these reactions taken from DeMore et al. 27 The model was validated in our laboratory by comparing modelled OH concentrations to measured OH concentrations under varying experimental conditions. Steady-state OH concentrations were measured during the model validation experiments by reacting OH with SO 2 (Matheson, 99.98%) in the reactor flow tube. The decay of SO 2 from its reaction with OH was measured using a chemical ionization mass spectrometer (CIMS) in negative ion mode using SF - 6 as the reagent ion. The CIMS setup has been previously described in detail by Thornberry and Abbatt. 9 The extent of decay of SO 2 from its reaction with OH was determined by monitoring intensity of the m/z 102 ion peak corresponding to F 2 SO - 2 ion, a product of the reaction of SF - 6 with SO 2. We measured the change in SO 2 signal due to the illumination of the humidified O 3 /SO 2 /aerosol flow compared to dark reactor flow tube conditions. The OH concentration was determined from the decay of SO 2 signal intensity using a rate constant for the reaction OH + SO 2 calculated from DeMore et al. 27. Control experiments showed that SO 2 levels were not affected individually by H 2 O, O 3, or UV lamp; SO 2 levels only declined when all were present. Two model parameters were adjusted to fit the measured OH concentrations: the photolysis rate constant for O 3 (J O3 ) and an OH wall-loss rate constant (k wall ). The model predictions

75 57 deviated from the measured values by less than ±24% after these adjustments are made. Note that SO 2 was present only for the experiments to validate the OH model; it was absent when the aerosol oxidation experiments were conducted Kinetic Experimental Methods The kinetics of the heterogeneous reaction of OH with BES particles in the aerosol flow reactor were investigated by measuring the loss of particle-phase BES as a function of OH exposure ranging from 0 to atm-s. As shown in Figure 2.2, the reacted particle flow was split after the O 3 denuder and the particles were analyzed by a scanning mobility particle sizer (SMPS) and an Aerodyne time-offlight aerosol mass spectrometer (ToF-AMS). The SMPS, including electrostatic classifier (TSI model 3080), neutralizer (TSI model 3076), DMA column (TSI 3081) and condensation particle counter (TSI model 3025), measured the reacted particle size distribution over the mobility diameter range D m = 50 to 300 nm with a scan time of 2.5 min. The ToF-AMS provided size-resolved chemical composition of the reacted particles and particle mass distributions in terms of vacuum aerodynamic size diameter. The ToF-AMS instrument has been described in detail elsewhere. 28 For the kinetic studies, 10 minute averaged measurements of the unreacted particles and reacted particles were compared. All measurements were normalized to SMPS particle number concentration to compensate for variations in particle generation stability. Under unreacted particle conditions (referred to as I 0 ), the aerosols were entrained in a flow of N 2, O 2, and water vapour with the UV lamp on continuously, but O 3 was absent. The conditions under which particles are reacted (referred to as I)

76 58 differed from I 0 only in that the O 3 generator was turned on, producing OH in the reactor flow tube. Control experiments showed that exposing BES particles to O 3 or the UV lamp individually did not affect the particle composition or size within experimental error Product Characterization Methods Several techniques were used to analyze the composition of the condensedphase and gas-phase reaction products. The ToF-AMS provided the mass spectrum of the reacted particles, and the major mass fragments yield structural information about the condensed-phase reaction products. However, due to the nature of the ToF- AMS ion source (electron impact at 70 ev), fragmentation of the reaction products was too extensive to permit detection of the molecular ion. To obtain the molecular ions, the particle-phase products were also analyzed by a Micromass QTOF Ultima mass spectrometer (ESI-MS). This instrument utilizes an electrospray ionization (ESI) source that is sufficiently soft to preserve the molecular ion peaks of the product. To prepare samples for ESI-MS analysis, polydisperse BES aerosol with a mode diameter of 146 nm (σ g = 1.4) and number concentration of cm -3 was oxidized and collected for approximately 3 hours onto a glass fiber filter (GF/A 47 mm, Whatmann). A control sample was prepared by collecting BES particles under unreacted particle conditions for 1 hour. Filters were extracted with approximately 10 ml of methanol (Fisher, HPLC grade) in amber vials and sonicated for 30 minutes. The extracts were transferred to clean vials and concentrated to 1 ml under a flow of N 2.

77 RESULTS AND DISCUSSION Kinetic Studies We studied the kinetics of the heterogeneous oxidation of BES particles by gas-phase OH radicals in a NO x free environment by monitoring the intensity of a mass fragment characteristic to BES with the ToF-AMS. Figure 2.3a shows the ToF- AMS mass spectrum of unreacted BES particles. Due to extensive fragmentation of organic molecules in the ToF-AMS, the molecular ion for BES at m/z 426 was not observed in the ToF-AMS mass spectrum, even when the electron energy was lowered from 70 ev to 20 ev. This is consistent with results reported earlier by Alfarra 29. Instead, several mass fragments characteristic of BES were monitored including the marked peaks in Fig. 2.3a. In Fig. 2.3b, the mass spectrum of the reacted particles clearly shows that these BES fragments have decreased in intensity. Note that the mass intensities shown in Fig. 2.3a and 2.3b have been normalized to the same particle number concentration. We found that the mass fragment m/z 297 had the fastest decay rate as a function of OH exposure, indicating that it is least likely to be a fragment of the particle-phase products. Furthermore, the rate of loss of m/z 297 had yielded a comparable rate constant value to that of the formation of several primary product peaks (see section 2.3.3), validating our assumption that mass 297 has similar decay kinetics to BES. Therefore, the decay of the m/z 297 fragment was monitored as a function of OH as a proxy for the reactive loss of particle-phase BES. In Fig. 2.4, the relative change in signal at m/z 297 is plotted as a function of OH exposure (atm-s). The experimental conditions for I and I 0 were described earlier

78 60 in the Experimental section. We note that the precision of the m/z 297 signal was 5% for a ten minute timescale in the absence of OH. Figure 2.3. ToF-AMS mass spectra of BES particles normalized to same particle number concentration: a) unreacted particles, b) reacted particles at OH exposure of atm-s, c) difference spectra of reacted spectrum minus unreacted spectrum, with positive values shown in green and negative values shown in red. The measured reactive uptake coefficient (γ) for the reaction of OH with BES particles is defined here as the fraction of OH collisions with the particle surface leading to reactive loss of particle-phase BES. We make an assumption that the

79 61 reactive loss rate of particle-phase BES is equal to the rate of OH reaction with BES. The validity of this assumption will be discussed below. As shown in Fig. 2.4, the rate of decay of m/z 297 signal decreases at high OH exposures. The reactive uptake coefficient will decrease at high OH concentrations as the surface concentrations of condensed-phase BES decrease and OH increasingly reacts with particle products. Therefore, we will focus on the determination of the initial reactive uptake coefficient (γ 0 ), where γ 0 is defined as γ extrapolated to zero OH exposure, which can be considered an upper limit for γ. Figure 2.4. Change in m/z 297 signal intensity as a function of OH exposure. Experimental data are shown as square symbols, and solid lines show calculated slopes using Eq. (1) with γ 0 values of 0.1 (green dotted), 0.5 (blue dash-dotted), 1.0 (red dashed), and 2.0 (pink solid). Inset is a magnification of the plot showing the data used for calculation of γ 0.

80 62 The initial reactive uptake coefficient (γ 0 ) of OH with BES particles was calculated using the formulation in Katrib et al.: 11 ([ I 297 ] [ I 297 ] ) 0 d( [ OH ] t) d V 4RT γ 0 = [ BES] 0 (2.1) A c Here, the first term is the slope of the linear fit of the experimental data in Fig. 4, V/A is the particle volume to surface area ratio (particle diameter/6 for the spherical particles used in this study), c is the mean speed of gas-phase OH molecules, R is the gas constant, T is the temperature and [BES] 0 is the initial condensed-phase BES concentration. As mentioned before, the signal at m/z 297 is used as a proxy for [BES]. To calculate γ 0 from Equation 2.1, the slope of a subset of the experimental data in Fig. 2.4 with OH exposures less than atm-s was used. The experimental data used to calculate γ 0 are shown in the inset of Fig This cut-off point was chosen because the inclusion of further data at higher OH exposure significantly reduced the γ 0 value and the correlation coefficient of the linear fit of the experimental data in Fig. 2.4, from which the slope was determined. The value of γ 0 calculated from experimental measurements using Eq. 2.1 was corrected for gas-phase diffusion by applying an empirical formulation by Fuchs and Sutugin. 30,31 The gas-phase diffusion coefficient of OH (D OH ) needed for this calculation has not been measured in N 2, O 2, or H 2 O. Therefore, we used gas-phase diffusion values of H 2 O in N 2, O 2, and H 2 O as an approximation for OH, 32 which was calculated to be D OH = cm 2 s -1 for our flow conditions. This value is close to the experimental D OH value measured in dry air. 33

81 63 The lines in Fig. 2.4 represent slopes calculated from Eq. 2.1 using diffusioncorrected γ 0 values ranging from 0.1 to 2. It is clear from the inset in Fig. 2.4 that a diffusion-corrected γ 0 value close to 1 best fits the initial data used to calculate the experimental γ 0 value. The diffusion-corrected initial reactive uptake coefficient value was calculated to be γ 0 = 1.3(±0.4). The reported error is one standard deviation of the overall experimental error (±30%). Due to the high uptake coefficient, the diffusion correction was significant, with the adjustment being approximately 35%. Our γ 0 value is similar to those obtained in studies of the reactive uptake of OH on organic films and monolayers, where γ values ranging from 0.2 to 1 have been measured. 15,18,24 It should be noted that these studies determined γ by measuring the reactive loss of OH in the gas-phase as opposed to measuring the loss of particle-phase species as in our study. In the first case, γ must be by definition less than or equal to unity. In contrast, the latter case may lead to γ values that are greater than unity when secondary condensed-phase chemistry is an important loss mechanism for condensed-phase species. For example, Hearn and Smith 19 measured γ 0 = 2.0 for OH oxidation of BES particles, which is somewhat larger than our value. From their results, Hearn and Smith 19 suggested that OH-initiated secondary chemistry lead to additional loss of BES. Within our experimental uncertainties, we do not see a strong indication of such secondary chemistry, but we cannot rule out the possibility either. In a recent smog chamber study, Lambe et al. studied the reaction of hexacosane particles with OH using lower OH concentrations ([OH] =

82 64 molecules cm -3 ) and calculated a γ value of γ = 1.04, which is in agreement with our γ 0 value Particle Size and Density Modification The SMPS and ToF-AMS were used in tandem in order to gain information on the particle size and density changes arising from the oxidation of organic aerosol by OH. These changes may indicate whether this reaction leads to release of oxygenated volatile organics from the particles. As suggested by Molina et al. 18, volatilization from heterogeneous oxidation may be a significant atmospheric sink for particle-phase organic matter. Figure 2.5a shows the SMPS volume distributions as a function of mobility diameter for unreacted and reacted (OH exposure ~ atms) BES particles normalized to the same particle number concentration. Both volume distributions have approximately the same mode mobility diameter (D m = 151 nm) and both are relatively monodisperse. The reacted particle size distribution shows a slight decrease in concentration of particles with D m = 150 nm and corresponding increase in smaller particles with D m = nm. The mean number-weighted mobility diameter decreased from 148 to 142 nm. For an OH exposure of atm-s, total particle volume normalized for particle concentration decreased by 7%, even though the signal intensity at m/z 297 decreased by 70% at this exposure.

83 65 Figure 2.5. Size distributions for unreacted (black solid line) and reacted (red dashed line, OH exposure = atm-s) BES particles normalized to the same particle concentrations: a) DMA volume-weighted particle concentration as a function of mobility diameter. b) ToF-AMS nitrate-equivalent mass concentration as a function of vacuum aerodynamic diameter. In Fig. 2.5b, the ToF-AMS total organic mass distributions for the unreacted and reacted aerosol populations are displayed as a function of vacuum aerodynamic diameter (D va ). The mass particle distribution shifted from a mean D va = 133 nm to 148 nm at an OH exposure of atm-s. The reaction-induced increase in vacuum aerodynamic diameter and corresponding decrease in mobility diameter suggests that the particle density increases as a result of OH exposure. Typically, oxygenated organic species have higher material densities than hydrocarbon species. It is therefore likely that the density of reacted particles would increase with oxidation if the oxidized products remain in the condensed phase. The particle density of reacted BES was determined using the following equation: 34

84 66 ρ p D D va = ρ 0 (2.2) m where ρ p is the particle density and ρ 0 is the standard density (ρ 0 = 1.0 g cm -3 ). Equation 2.2 is valid only for spherical particles with no voids, which is the case for the liquid BES particles used in this study. We confirmed that the particles remained spherical even after reaction by measurement of the divergence of the particle beam in the ToF-AMS as described by Katrib et al. 8 The particle beam divergence for spherical particles is smaller than for non-spherical particles of the same D va. The particle beam shape is measured by translating a 0.5 mm wire into the particle beam approximately 10 cm from the vaporizer and measuring the resulting decrease in particle transmission as a function of wire location. Figure 2.6 shows the change in transmission as a function of wire shadow on the vaporizer for reacted (OH exposure = atm-s) and unreacted BES particles. Because the reaction has no effect on the beam width, we conclude that the BES particles remain spherical after reaction.

85 67 Figure 2.6. Particle beam width profile for unreacted (black squares) and reacted (red circles, OH exposure = atm-s) BES particles. The relative changes in particle density and particle volume of reacted BES particles are displayed as a function of OH exposure in Fig The average density of unreacted BES particles calculated from Eq. (2.2) is ρ p = 0.91 (± 0.02) g cm -3, i.e., within one standard deviation of the material density of liquid BES (ρ = g cm - 3 ). Figure 2.7 indicates that the particle densities of reacted BES increased linearly with OH exposure. In contrast, the particle volumes increased slightly at lower OH exposures (V/V 0 = 1.04), then decreased. At the highest OH exposures studied (~ atm-s), particle density increased by 20% and particle volume decreased by 17%. An increase in particle density was also found for the oxidation of particles coated with oleic acid reacted with O 3 using similar analytical techniques. 8 These results suggest that the particles are becoming more oxidized with increasing

86 68 exposure to OH. Furthermore, volatilization causes a nonlinear decrease in the particle volume, where measurable volume loss is not apparent until BES particles are exposed to atm-s. The effect of O 2 concentration on particle volatilization was investigated by repeating oxidation experiments at OH exposures of and atm-s with the addition of an O 2 flow to increase the O 2 concentration to near atmospheric levels ([O 2 ] = molecules cm -3 ). These experiments showed no difference in volume change relative to experiments carried out under similar OH exposures, suggesting that our results are relevant under conditions with atmospheric O 2 concentrations. Figure 2.7. Relative change in volume (black squares) and density (red circles) of BES particles as a function of OH exposure. Error bars show one standard deviation of averaged data. Lines are fits to guide the eye.

87 Product Identification Several analytical methods were employed to characterize the oxidation products arising from the reaction of OH with BES particles in order to elucidate the reaction mechanism for this heterogeneous process. We have studied the release of gas-phase products from OH reaction with a BES film using proton-transfer reaction mass spectrometry (PTR-MS), which will be presented in a forthcoming paper. Here, we focus on characterization of particle-phase oxidation products. The ToF-AMS provides size-resolved on-line chemical composition information of the particle-phase species. Typical mass spectra of the unreacted and reacted particles taken by ToF- AMS are shown in Fig. 2.3a and Fig. 2.3b, respectively. The characteristic mass fragments that are marked in Fig. 2.3a have clearly decreased in intensity in the reacted particle spectrum in Fig. 2.3b, whereas several fragments increase in intensity, corresponding to the formation of condensed-products. The change in the mass spectrum due to OH exposure is shown more clearly in the subtraction spectrum in Fig. 2.3c. The peaks highlighted in red represent a decrease in the characteristic BES fragments, whereas the green fragments represent an increase in mass intensities, e.g., m/z 44, 127, 153, and 181. The kinetic production of these product fragments was measured as a function of OH exposure as shown in Fig The kinetic data were fitted to exponential rise to maximum fits. The average kinetic rate constant (k avg ) for the product fits for m/z 127, 153, and 181 (k avg = 7.6(±1.6) 10 7 atm -1 s -1 ) is close to the overall rate constant of decay of m/z 297 (k 297 = 6.1(±0.4) 10 7 atm -1 s -1 ), which indicates that these masses are primary product

88 70 fragments. However, we are not able to determine chemical structures of the product fragments 127, 153, and 181 solely from the ToF-AMS data. The kinetic rate + constant for the production of m/z 44 (k 44 ), a mass fragment corresponding to CO 2 ion signifying presence of carboxylate groups, was significantly slower (k 44 =1.5(±0.2) 10 7 atm -1 s -1 ) than the values for the other product peaks. This suggests that the oxidized products containing carboxylic groups arise from secondary chemistry and thus form more slowly than primary products. Indeed, there is some indication that the intensities of product masses 127, 153, and 181 decreased at very high OH exposures (greater than atm-s) as a result of secondary chemistry. Figure 2.8. Kinetic plot for product fragments m/z 44 (black squares), 127 (red circles), 153 (green triangles), and 181 (blue triangles). Error bars show one standard deviation of averaged data. Solid lines are fits to data with the function y = 1 + a(1- exp -bx ).

89 71 High molecular weight organic species tend to fragment extensively in the ToF-AMS, and molecular ion peaks are frequently not present in the mass spectrum. Electrospray ionization mass spectrometry (ESI-MS), which uses a softer ionization technique than electron impact, was employed in order to observe the molecular ion peaks of the products and thereby determine the molecular mass of these species. Figure 2.9 displays the ESI-MS mass spectra of unreacted (Fig. 2.9a) and reacted (Fig. 2.9b) BES particles. The two major peaks in Fig. 2.9a at m/z of 427 (MH + ) and 449 (MNa + ) both correspond to BES. The stronger signal intensity at MNa + compared to MH + is due to the fact that esters are readily sodiated during electrospray ionization (Alex Young, private communication), where the source of Na + ions may have been from leaching of the glass container holding the extract. The intensities of these BES peaks visibly decreased in the reacted spectrum, while a series of peaks appeared at masses greater than 449 amu in the mass range of amu. This area is magnified in Fig. 2.9c.

90 72 Figure 2.9. ESI-MS mass spectra of BES particles: a) unreacted particles, b) reacted particles (OH exposure = atm-s), c) magnification of panel b for mass 450 to 550 amu. We obtained precise masses of the product peaks from the high-resolution ESI-MS measurements, from which we were able to accurately identify chemical compositions of the product peaks. The ESI-MS software (Analyst QS 1.1) was used to identify the most likely chemical composition of each product peak, where the calculated mass of the proposed chemical formula was within a specified error range from the measured peak mass. For example, there were only two proposed chemical formulas for m/z 449 within 10 ppm of the measured mass, C 26 H 50 O 4 Na + (0.37 ppm)

91 73 and C 28 H 49 O 4 (5.0 ppm). The first formula corresponds to sodiated BES as expected, which has the lower mass error. Thus, from analysis of the precise product masses we found that the most likely chemical formulas of the product peaks corresponded to the multiple additions of oxygen atoms in the form of carbonyl or hydroxyl groups, or combinations thereof, in place of CH 2 groups in the BES molecule. For example, the most likely chemical formulas for masses 463 and 465 are C 26 H 48 O 5 Na + and C 26 H 50 O 5 Na +, respectively, which is consistent with the addition of a carbonyl and alcohol, respectively. Furthermore, each set of peaks shown in Fig. 2.9c moving from lower to higher masses represents the addition of a successive number of oxygen atoms, ranging from 1 to 6 oxygens, in all possible combinations of carbonyls and alcohol groups. Because the BES particles collected for the ESI-MS analysis were exposed to atm-s with a corresponding OH to BES ratio of 3, it is not unreasonable that the BES molecules could have reacted with a number of OH radicals to gain more than one oxygenated functional group. Table 2.1 summarizes the assigned chemical formulas for the product masses and associated error in calculated mass. We note that we did not quantify the relative ionization efficiencies for BES and particle-phase products. Thus, we cannot accurately compare product yields based on the relative intensities. All assigned formulas, with the exception of peaks with 6 added oxygens, had the lowest or second lowest error of the proposed formulas, where in the second case the formula with the lowest error had several more carbons than BES and thus was deemed an unlikely

92 74 chemical composition. All errors were within 20 ppm except for the peaks with 6 added oxygens, whose intensities were relatively weak and whose errors were within 100 ppm. Therefore, it is clear from these results that the major condensed phase products are high molecular weight molecules containing carbonyl and alcohol groups. Several product fragments of reacted BES were observed in both the ToF- AMS and ESI-MS mass spectrum, such as m/z 127, 153, and 181. Tandem mass spectrometry (MS/MS) analysis confirmed that the fragments were indeed produced from fragmentation of the protonated product peaks, as the sodiated peaks did not fragment in the mass spectrometer. For example, the major peaks in the MS/MS fragmentation spectrum of m/z 441, the mass corresponding to the protonated form of a major product C 26 H 48 O 5, included mass fragments 127, 153, and 181. The most likely chemical formulas for product fragments m/z 127, 153, and 181 as determined by precise masses are C 8 H 15 O +, C 9 H 13 O + 2, and C 10 H 13 O + 3, respectively. Further MS/MS analysis of these fragment peaks revealed that m/z 153 and 181 had similar fragmentation patterns, whereas m/z 127 had a distinctly different fragmentation pattern. Therefore, m/z 127 likely originates from a different moiety of the product molecule than m/z 153 and 181. The most likely structure for m/z 127 would be a fragment consisting of the 2-ethylhexyl moiety with a carbonyl group on the α- position carbon (i.e., CH 3 (CH 2 ) 3 CH(CH 2 CH 3 )C(O) + ). Preferential OH attack of carbon in the α-position of the alkoxy group has been observed by Picquet-Varrault et al. 35 from gas-phase OH oxidation of isopropyl, isobutyl and tert-butyl acetates. The

93 75 structures of m/z 153 and 181 are less clear, but it is apparent that m/z 153 results from a loss of CO from the fragment represented by m/z 181. It appears from the MS/MS spectra that these fragments are most likely from the sebacate moiety of the molecule (i.e., C(O)(CH 2 ) 8 C(O)) with an addition of an oxygenated functional group to the alkyl chain.

94 76 Table 2.1. Condensed-phase product masses from high resolution ESI-MS analysis of reacted BES particles and proposed chemical formulas. Relative intensities are scaled to the most intense peak at m/z 463. Error in ppm is the fraction of error between calculated and measured masses. Number of added oxygen atoms is expressed as number of added functional groups as carbonyls (C) or alcohols (A). m/z Relative Chemical Number of Error Intensity Composition Added Oxygen (ppm) C 26 H 48 O 5 Na + 1C C 26 H 50 O 5 Na + 1A C 26 H 46 O 6 Na + 2C C 26 H 48 O 6 Na + 1C+1A C 26 H 50 O 6 Na + 2A C 26 H 44 O 7 Na + 3C C 26 H 46 O 7 Na + 2C + 1A C 26 H 48 O 7 Na + 1C + 2A C 26 H 50 O 7 Na + 3A C 26 H 42 O 8 Na + 4C C 26 H 44 O 8 Na + 3C + 1A C 26 H 46 O 8 Na + 2C + 2A C 26 H 48 O 8 Na + 1C + 3A C 26 H 50 O 8 Na + 4A C 26 H 40 O 9 Na + 5C C 26 H 42 O 9 Na + 4C + 1A C 26 H 44 O 9 Na + 3C + 2A C 26 H 46 O 9 Na + 2C +3A C 26 H 48 O 9 Na + 1C + 4A C 26 H 50 O 9 Na + 5A C 26 H 38 O 10 Na + 6C C 26 H 40 O 10 Na + 5C + 1A C 26 H 42 O 10 Na + 4C +2A C 26 H 44 O 10 Na + 3C + 3A C 26 H 46 O 10 Na + 2C + 4A C 26 H 48 O 10 Na + 1C +5A C 26 H 50 O 10 Na + 6A 61.2

95 77 In separate experiments, we have characterized the volatile organic compounds (VOC) that evolved from OH oxidation of a BES film by coated-wall flow tube-cims coupled to a PTR-MS, from which the results will only briefly be touched upon here and will be presented in a future publication. The major gas-phase products from heterogeneous reaction of BES with OH are consistent with the formation of aldehydes and acids ranging in size from 1 to 8 carbons, with a VOC to OH product yield of approximately 10% Reaction Mechanism The reaction mechanism for the heterogeneous oxidation of saturated organic species by OH has been assumed to proceed in a similar manner to the gas-phase reaction mechanism. 36 A generalized reaction scheme for the reaction of OH with a saturated organic compound has been proposed by other researchers 17,18 and is reproduced in Fig under NO x -free conditions. The reaction mechanism for gasphase OH reaction with alkanes has been reviewed in detail by Atkinson 37 and is summarized below. Initially, OH abstracts a hydrogen atom from the alkyl chain, producing H 2 O and an alkyl radical (R). The alkyl radical R quickly reacts with O 2 in the atmosphere to form RO 2, an alkyl peroxy radical. Alkyl peroxy radicals can self-react through two reaction pathways, leading to the formation of a carbonyl and alcohol (pathway (2)) or to two alkoxy radicals (RO). Alkyl peroxy radicals may also react with HO 2 to form an organic peroxide ROOH (pathway (1)). The branching ratio for pathway (2) leading to formation of a carbonyl + alcohol is approximately in the gas-

96 78 by three different pathways: isomerization (pathway (3)), decomposition (pathway (4)), and reaction with O 2 (pathway (5)). RO can isomerize through a 6-membered ring transition-state leading to 1,5 H-shift (pathway (3)) or can react with O 2 to form a carbonyl and HO 2 radical (pathway (5)). Alternatively, the RO radical can decompose through C-C bond scission leading potentially to volatilization of small- chained products (pathway (4)). phase f or primary and secondary RO 2 radicals and not accessible for tertiary RO 2 radicals. 37 The alkoxy radical (RO) formed from the self-reaction of RO 2 can react Figure Generalized reaction mechanism for OH-initiated oxidation of a hydrocarbon.

97 79 We determined from the ESI-MS analysis that OH reaction with BES particles leads to the formation of more oxidized condensed-phase products. These results are supported by the observed increase in particle density with OH exposure. The ESI- MS mass spectra showed that the condensed-phase products consisted of high molecular weight carbonyls and alcohols. It is possible that organic peroxides have formed from the reaction of alkyl peroxy radical with HO 2 (1), but it is unlikely that organic peroxides could have been observed during ESI-MS analysis due to their highly unstable nature. The slow increase of the m/z 44 in the ToF-AMS data suggests that carboxylic acids were also being formed, possibly through secondary reactions, e.g., OH-initiated oxidation of aldehydes. It is unlikely that OH would attack the BES terminal carbon leading to formation of a terminal acid with a higher mass than BES. Therefore, the rise in m/z 44 is likely due to shorter-chained acids, which were not observed in the ESI-MS product spectra. The slow kinetics of acid formation and absence of acids in the ESI-MS spectra suggest that acids are not the major condensed-phase products. This is the first study of its kind to detect both carbonyl and alcohol groups from heterogeneous oxidation of condensed-phase saturated organic species by OH. These results indicate that the channel (2) may be an important pathway. We observed higher signal intensity from the carbonyl product than the alcohol, suggesting that another pathway may be contributing to the formation of the carbonyl product. Because the relative product intensities measured by the ESI-MS are uncalibrated, one should take caution in estimating relative yields from them. High

98 80 molecular weight carbonyls may also be formed by channel (5). Furthermore, alcohols could be formed from the isomerization of RO (4) and subsequent reaction of the radical to form a hydroxy-substituted carbonyl. This reaction may explain the presence of products containing an addition of both alcohols and carbonyls. Because different reaction pathways lead to carbonyl and alcohol products, the relative importance of each pathway is not clear. Other studies have also observed oxidized condensed-phase products from the oxidation of condensed-phase saturated organics. For example, Docherty and Ziemann 38 observed the formation of carbonyl nitrates and alcohol nitrates as oxidation products from the heterogeneous reaction of NO 3 with oleic acid particles. Although the reaction mechanism differs from this study in that the initial step involves addition of NO 3 to the double bond of an unsaturated organic, it leads nonetheless to the formation of products containing carbonyl and alcohol groups similar to this study. Furthermore, Knopf et al. 20 observed the formation of alcohol, carbonyl and carboxylic acid groups from XPS measurements of a saturated monolayer oxidized by NO 3 radicals, concluding that pathway (2) may potentially be a major pathway in their system. In contrast, Eliason et al. 17 measured the production of carbonyls but not alcohols from OH oxidation of hexadecane film, and therefore concluded that pathways (4) and (5) were most important. In a recent study, Hearn et al. 7 investigated the reaction of BES particles with Cl, whereby they estimated product yields of condensed-phase products from this reaction. They also found high molecular weight carbonyls and alcohols to be the major reaction products, which is

99 81 in agreement with this study. Furthermore, they found that the ratio of the product yields of carbonyls to alcohols to be a function of O 2 concentrations. Their study indicates that O 2 concentrations will not only affect the reaction rate by controlling the formation rate of RO 2, but will also influence the RO + O 2 channel as well. We observed volatilization of oxidation products indirectly from the decrease in particle volume with OH exposure. The relative importance of the RO decomposition pathway to the pathways leading to the formation of high molecular weight carbonyl and alcohols was examined by comparing observed particle mass changes to predicted mass change from the formation of condensed and gaseous oxidation products. For this calculation, we assumed a product yield for the formation of volatile compounds of 10% with an average product molecular weight of 58 amu. These assumptions are based on unpublished work by our group (now published as Vlasenko et al. 25 ), in which the masses and product yields of volatile organics from OH oxidation of BES film were determined by use of coated-wall flow tube coupled to a PTR-MS. We also assumed that carbonyls and alcohols were formed in equal amounts such as in pathway (2) as a simplified case and that the particle density did not change from volatilization. Therefore, 1 of every 10 OH collisions with a BES particle will lead to the loss of 58 amu from a BES molecule. We assumed that the other 9 collisions out of 10 (γ = 1) lead to the formation of alcohol or carbonyl groups with average mass addition of 15 amu to BES molecule for each collision leading to this pathway.

100 82 Figure 2.11 shows the observed and predicted particle mass changes as a function of OH exposure based on the assumptions explained above. The observed particle mass changes were determined from the density and volume measurements. The red and blue lines represent the relative contributions to mass change from the carbonyl + alcohol pathway (2) with a product yield of 90% and decomposition pathway (4) with product yield of 10%, respectively. The pink line represents the overall predicted mass change from the two contributions stated, which indicates that the particle mass should increase linearly with OH exposure up to 10%. The observed particle mass does not however change linearly with OH exposure. At OH exposures below atm-s, the particle mass increases to approximately 10% following the trend for the carbonyl + alcohol contribution. At higher OH exposures, the particle mass then decreases to no net mass change. This observation suggests that the volatilization contribution to the particle mass change is not linear. One possible explanation for nonlinear volume change is that at higher OH exposures the condensed-phase products gain more functional groups, so as the products react with OH they decompose more readily through channel (4) than BES. These results indicate that the particle mass change contributions from the formation of condensed- phase and volatile products cancel out to some degree, with 10% maximum change in particle mass.

101 83 Figure Relative particle mass change as a function of OH exposure. Black squares are calculated mass values from the product of density and volume measurements, red dashed and blue dotted lines show theoretical contribution of carbonyl + alcohol and decomposition reaction pathways to mass change, respectively, assuming a VOC/OH ratio of 0.1 and average VOC molecular mass of 58 g mol -1, and pink line shows theoretical overall mass change (See text). In this study, volatilization of products from pathway (4) was found to be a minor reaction pathway in the OH oxidation of BES particles under our experimental conditions. Our study is in contrast to the recent study by Molina et al. 18, who observed complete volatilization of an alkane monolayer for an OH exposure of 3 OH collisions per alkane molecule. At the same OH exposure (i.e., atm-s), we did not see any evidence of volatilization. It is possible that because BES has a longer chain and higher molecular weight than the alkane monolayer studied by Molina et al. 18 BES may require more OH collisions for decomposition products to

102 84 volatilize. Under our experimental conditions, particles would completely volatilize only if the VOC formation to OH loss product ratio was 0.7 or higher. The studies by Knopf et al. 20 and Moise and Rudich 16 are more in line with our results. Knopf et al. 20 found that NO 3 reaction with an alkane monolayer lead to a small amount (up to 10%) of volatilization of the surface carbon. Furthermore, Moise and Rudich 16 measured up to 20% loss of carbon from the oxidation of alkane monolayers by Cl and Br radicals. Although NO 3, OH and halogen radicals should react with organics by a mechanism similar to that shown in Fig. 2.10, these studies suggest that the importance of the pathways for RO may depend sensitively on specific experimental conditions. One major difference between these studies and ours is that we have studied the reaction using liquid aerosol particles as opposed to monolayers on a solid substrate or solid films. It has been suggested that the particle phase may influence the reaction pathways, 20,38 with channel (2) suggested as being the dominant reaction pathway for liquid organics Eliason et al. detected small-chained products from the reaction of a thin liquid hexadecane film with OH in the condensed-phase only, but did not detect products in the gas phase. Their results indicate that decomposition is an important pathway for oxidation liquid organic species by OH, but it is unclear to what extent compared to the other pathways. The results in our study clearly show that for liquid organic aerosols, the pathways leading to the production of carbonyls and alcohols, i.e., pathways (2), (3) and (5) are more important than the

103 85 decomposition pathway. The same conclusion may not hold, however, for solid organic particles, which will require further study. It is possible that the reaction mechanism for OH oxidation of organic aerosols may be influenced by experimental conditions. For example, recent work by Hearn et al. 21 confirmed that pathway (5) is affected by O 2 concentrations as expected. It is possible that as O 2 concentration is reduced, RO decomposition may be favored over reaction with O 2. Note that we did not see an enhancement of volatilization at O 2 concentrations two orders of magnitude lower than atmospheric levels. The chemical composition of the model organic compound, such as degree of oxidation, chain length and degree of branching, may also influence the reaction mechan ism. For example, the branching ratio for pathway (2) may be reduced during the oxidation of branched organic compounds such as BES compared to unbranched compounds as the mechanism requires an H-atom transfer. 39 Under the NO x -free conditions in this study, it is likely that pathway (2) is enhanced compared to the RO 2 self-reaction leading to RO formation. The presence of NO x may lead to increased yields of RO formation, thus reducing the yield of the carbonyl + alcohol (2) channel and perhaps leading to nitrated products. The influence of these experimental factors on the OH-initiated reaction mechanism should be systematically studied in the future ATMOSPHERIC IMPLICATIONS AND CONCLUSIONS In this work, we investigated the heterogeneous oxidation of BES particles by OH under NO x -free conditions. These conditions were chosen as a simplified case

104 86 simulating remote atmospheric conditions. The oxidation of organic aerosol by OH is highly efficient with a reactive coefficient of γ 1 for the reaction of OH with BES particles. This study serves as a simple model for the chemical aging of atmospheric aerosols in the troposphere containing saturated organic matter, such as primary organic aerosol, organic coatings on marine aerosol, biomass burning aerosol and secondary organic aerosol (SOA) formed from photochemical oxidation of volatile organics. 2 Although studies with more chemically complex particles are required, this system can be useful as an initial proxy for saturated organic aerosol in the troposphere. This study suggests that chemical aging of organic aerosol by OH has an important impact on the physical and chemical properties of tropospheric organic aerosol. We observed the accumulation of oxidized products in the condensed phase from the heterogeneous oxidation of BES particles. Several recent field studies have observed a greater degree of oxidation of ambient organic aerosols with an increase in photochemical age Moreover, Robinson et al. 43 discovered evidence for chemical oxidation of tropospheric organic aerosol during regional transport, especially during the summer. To determine the importance of heterogeneous oxidation of atmospheric organic aerosol by OH relative to the lifetime of atmospheric organic aerosols, we calculated the oxidation lifetime (τ) of a model tropospheric organic particle in a manner similar to that presented in Robinson et al. For this calculation, we assumed a 24-hour averaged OH concentration of [OH] = 10 cm and that γ = 1 for reactive

105 87 uptake of OH onto a model particle with a radius R = 50 nm containing saturated hydrocarbons with an average molecular weight of M = 300 g mol -1. The oxidation lifetime of the model organic aerosol is defined as the time needed for every organic molecule initially present in the aerosol to be oxidized by one OH radical 43 As with our kinetic calculations, we assume that the rate of reactive uptake of OH equals the rate of oxidation of the hydrocarbon molecules and that the hydrocarbons are equally mixed throughout the particle. The latter assumption is reasonable as the diffusion time of a hydrocarbon in a particle with a diameter of 100 nm with an assumed diffusion coefficient of D ~ 10-6 cm 2 s -1 was calculated to be 30 μs. Under the given assumptions, we calculated the oxidation lifetime of an organic particle as the number of initial hydrocarbon molecules in the particle (N) divided by the reactive flux of OH into the particle (J) using the equation given in Robinson et al.: ρn a πr N 3 M 4 ρn a τ = = = R (2.3) 2 J γπr c[ OH ] 3 γc[ OH ] M Here, c is the average thermal speed of OH and ρ is the particle density assumed to be ρ ~ 1 g cm -3. From Eq. (2.3) we calculated an oxidation lifetime of 2.6 days, which is consistent with the oxidation lifetime calculated by Robinson et al. 43 This simple calculation suggests that organic aerosol can be significantly oxidized within the typical lifetime of an atmospheric particle of approximately 5 to days. Therefore, OH oxidation may be an important mechanism for the chemical transformation of tropospheric organic aerosol in the time scale of regional transport

106 88 and less important for local urban organic aerosol sources. 45 The oxidation lifetime may strongly depend on the chemical composition and phase of the atmospheric particles and the ability of the organic molecules to diffuse to the particle surface. Furthermore, organic molecules that preferentially partition to the surface, e.g., organic coatings on aqueous aerosols, will be oxidized more rapidly than organics in the bulk phase. In this study, we observed the production of high molecular weight carbonyls and alcohols as the major particle-phase products. Recent smog chamber studies of SOA formation have indicated that heterogeneous oligomerization reactions may explain the observed increase in SOA mass compared to expected SOA formation Heterogeneous oxidation of organic particles by OH is another mechanism to increase organic aerosol mass in the atmosphere as well as increase the degree of oxidation. An increase in oxidized particle-phase species may influence the hygroscopic properties of the particles and their ability to act as cloud condensation nuclei. Heterogeneous oxidation of aerosols will result in an increase the density of the aerosols, as observed in our studies, as well as alter their optical properties. Although we found evidence for the formation of volatile oxidation products by observing the loss of particle volume up to 17%, the reaction pathway leading to release of volatile products was a minor one from a mass-weighted perspective. Kwan et al. 49 evaluated the importance of photochemical oxidation of organic aerosols as a source of oxidized VOC based on the work by Molina et al. 18 assuming each OH collision leads to the loss of one VOC consisting of 6 carbons. Our study

107 89 suggests that the calculated VOC flux from aerosol oxidation may not be as important if the VOC yield from liquid organic aerosols is lower than that for solid aerosols. It should be noted that extrapolation of laboratory results to ambient conditions may be problematic if the kinetics or mechanism of chemical aging of organic aerosol are influenced by the complexity of the chemical matrix that make up the aerosol particles or the concentration of oxidants. The particle phase may have a significant impact on the chemical transformation of organic aerosol. This has been shown to be true for the heterogeneous chemistry of O 3 with organic aerosol containing oleic acid. Several researchers have found that mixtures of liquid unsaturated oleic acid with solid saturated organic species, such as stearic acid, may slow the kinetics of chemical aging. 11,12,50 These results may explain the discrepancy in the short lifetime of oleic acid predicted by experimental data and the much longer lifetimes measured in ambient conditions. It is also important to validate that the same heterogeneous kinetics and reaction mechanism apply at lower oxidant concentrations that are more similar to ambient conditions. Thus, it is vital not only to study the particle morphology and phase of ambient organic aerosol, but also to focus future laboratory studies on investigating chemical aging of more complex aerosol systems with experimental conditions that more closely mimic atmospheric conditions.

108 90 ACKNOWLEDGEMENTS We thank the Advanced Instrumentation for Molecular Structure laboratory for use of the ESI-MS instrument and Alex Young for assistance with ESI-MS work. We greatly appreciate the insightful comments and discussion by J. Thornton, A. Ivanov, S. Trakhtenberg, P. Ziemann, G. Smith and anonymous referees. This research project was funded by NSERC. Infrastructure support to SOCAAR came from ORF, OIT and CFI.

109 REFERENCES (1) Seinfeld, J. H.; Pandis, S. N. Atmospheric chemistry and physics: from air pollution to climate change; John Wiley and Sons, Inc.: New York, (2) Finlayson-Pitts, B. J.; Pitts, J. N. Chemistry of the Upper and Lower Atmosphere: Theory, Experiments, and Applications; Academic Press, (3) Ramanathan, V.; Crutzen, P. J.; Kiehl, J. T.; Rosenfeld, D. Science 2001, 294, (4) Jacobson, M. C.; Hansson, H. C.; Noone, K. J.; Charlson, R. J. Reviews of Geophysics 2000, 38, 267. (5) Saxena, P.; Hildemann, L. M. Journal of Atmospheric Chemistry 1996, 24, 57. (6) Moise, T.; Rudich, Y. Journal of Physical Chemistry A 2002, 106, (7) Hearn, J. D.; Smith, G. D. Journal of Physical Chemistry A 2004, 108, (8) Katrib, Y.; Martin, S. T.; Hung, H. M.; Rudich, Y.; Zhang, H. Z.; Slowik, J. G.; Davidovits, P.; Jayne, J. T.; Worsnop, D. R. Journal of Physical Chemistry A 2004, 108, (9) Thornberry, T.; Abbatt, J. P. D. Physical Chemistry Chemical Physics 2004, 6, 84. (10) Hearn, J. D.; Lovett, A. J.; Smith, G. D. Physical Chemistry Chemical Physics 2005, 7, 501. (11) Katrib, Y.; Biskos, G.; Buseck, P. R.; Davidovits, P.; Jayne, J. T.; Mochida, M.; Wise, M. E.; Worsnop, D. R.; Martin, S. T. Journal of Physical Chemistry A 2005, 109,

110 92 (12) Knopf, D. A.; Anthony, L. M.; Bertram, A. K. Journal of Physical Chemistry A 2005, 109, (13) Ziemann, P. J. Faraday Discussions 2005, 130, 469. (14) Morris, J. W.; Davidovits, P.; Jayne, J. T.; Jimenez, J. L.; Shi, Q.; Kolb, C. E.; Worsnop, D. R.; Barney, W. S.; Cass, G. Geophysical Research Letters 2002, 29, (15) Bertram, A. K.; Ivanov, A. V.; Hunter, M.; Molina, L. T.; Molina, M. J. Journal of Physical Chemistry A 2001, 105, (16) Moise, T.; Rudich, Y. Geophysical Research Letters 2001, 28, (17) Eliason, T. L.; Gilman, J. B.; Vaida, V. Atmospheric Environment 2004, 38, (18) Molina, M. J.; Ivanov, A. V.; Trakhtenberg, S.; Molina, L. T. Geophysical Research Letters 2004, 31, L (19) Hearn, J. D.; Smith, G. D. Geophysical Research Letters 2006, 33, L (20) Knopf, D. A.; Mak, J.; Gross, S.; Bertram, A. K. Geophysical Research Letters 2006, 33, L (21) Hearn, J. D.; Renbaum, L. H.; Wang, X.; Smith, G. D. Physical Chemistry Chemical Physics 2007, 9, (22) Lambe, A. T.; Zhang, J. Y.; Sage, A. M.; Donahue, N. M. Environmental Science & Technology 2007, 41, (23) Moise, T.; Talukdar, R. K.; Frost, G. J.; Fox, R. W.; Rudich, Y. Journal of Geophysical Research-Atmospheres 2002, 107, 4014.

111 93 (24) Cooper, P. L.; Abbatt, J. P. D. Journal of Physical Chemistry 1996, 100, (25) Vlasenko, A.; George, I. J.; Abbatt, J. P. D. Journal of Physical Chemistry A 2008, 112, (26) Braun, W.; Herron, J. T.; Kahaner, D. K. International Journal of Chemical Kinetics 1988, 20, 51. (27) DeMore, W. B.; Sander, S. P.; Golden, D. M.; Hampson, R. F.; Kurylo, M. J.; Howard, C. J.; Ravishankara, A. R.; Kolb, C. E.; Molina, M. J. Evaluation No. 12, JPL Publication (28) Drewnick, F.; Hings, S. S.; DeCarlo, P.; Jayne, J. T.; Gonin, M.; Fuhrer, K.; Weimer, S.; Jimenez, J. L.; Demerjian, K. L.; Borrmann, S.; Worsnop, D. R. Aerosol Science and Technology 2005, 39, 637. (29) Alfarra, M. R. Insights into Atmospheric Organic Aerosols Using an Aerosol Mass Spectrometer, University of Manchester, (30) Fuchs, N. A.; Sutugin, A. G. Highly dispersed aerosols; Ann Arbor Science Publishers: Ann Arbor, (31) Worsnop, D. R.; Morris, J. W.; Shi, Q.; Davidovits, P.; Kolb, C. E. Geophysical Research Letters 2002, 29, (32) Massman, W. J. Atmospheric Environment 1998, 32, (33) Ivanov, A. V.; Trakhtenberg, S.; Bertram, A. K.; Gershenzon, Y. M.; Molina, M. J. Journal of Physical Chemistry A 2007, 111, (34) DeCarlo, P. F.; Slowik, J. G.; Worsnop, D. R.; Davidovits, P.; Jimenez, J. L. Aerosol Science and Technology 2004, 38, 1185.

112 94 (35) Picquet-Varrault, B.; Doussin, J. F.; Durand-Jolibois, R.; Carlier, P. Journal of Physical Chemistry A 2002, 106, (36) Atkinson, R. Journal of Physical and Chemical Reference Data 1994, 26, 1. (37) Atkinson, R. Journal of Physical and Chemical Reference Data 1997, 26, 215. (38) Docherty, K. S.; Ziemann, P. J. Journal of Physical Chemistry A 2006, 110, (39) Russell, G. A. Journal of the American Chemical Society 1957, 79, (40) de Gouw, J. A.; Middlebrook, A. M.; Warneke, C.; Goldan, P. D.; Kuster, W. C.; Roberts, J. M.; Fehsenfeld, F. C.; Worsnop, D. R.; Canagaratna, M. R.; Pszenny, A. A. P.; Keene, W. C.; Marchewka, M.; Bertman, S. B.; Bates, T. S. Journal of Geophysical Research-Atmospheres 2005, 110, D (41) McFiggans, G.; Alfarra, M. R.; Allan, J.; Bower, K.; Coe, H.; Cubison, M.; Topping, D.; Williams, P.; Decesari, S.; Facchini, C.; Fuzzi, S. Faraday Discussions 2005, 130, 341. (42) Takegawa, N.; Miyakawa, T.; Kondo, Y.; Blake, D. R.; Kanaya, Y.; Koike, M.; Fukuda, M.; Komazaki, Y.; Miyazaki, Y.; Shimono, A.; Takeuchi, T. Geophysical Research Letters 2006, 33, L (43) Robinson, A. L.; Donahue, N. M.; Rogge, W. F. Journal of Geophysical Research-Atmospheres 2006, 111, D (44) Kanakidou, M.; Seinfeld, J. H.; Pandis, S. N.; Barnes, I.; Dentener, F. J.; Facchini, M. C.; Van Dingenen, R.; Ervens, B.; Nenes, A.; Nielsen, C. J.; Swietlicki, E.; Putaud, J. P.; Balkanski, Y.; Fuzzi, S.; Horth, J.; Moortgat, G.

113 95 K.; Winterhalter, R.; Myhre, C. E. L.; Tsigaridis, K.; Vignati, E.; Stephanou, E. G.; Wilson, J. Atmospheric Chemistry and Physics 2005, 5, (45) Zhang, Q.; Worsnop, D. R.; Canagaratna, M. R.; Jimenez, J. L. Atmospheric Chemistry and Physics 2005, 5, (46) Jang, M. S.; Czoschke, N. M.; Lee, S.; Kamens, R. M. Science 2002, 298, 814. (47) Kalberer, M.; Paulsen, D.; Sax, M.; Steinbacher, M.; Dommen, J.; Prevot, A. S. H.; Fisseha, R.; Weingartner, E.; Frankevich, V.; Zenobi, R.; Baltensperger, U. Science 2004, 303, (48) Baltensperger, U.; Kalberer, M.; Dommen, J.; Paulsen, D.; Alfarra, M. R.; Coe, H.; Fisseha, R.; Gascho, A.; Gysel, M.; Nyeki, S.; Sax, M.; Steinbacher, M.; Prevot, A. S. H.; Sjogren, S.; Weingartner, E.; Zenobi, R. Faraday Discussions 2005, 130, 265. (49) Kwan, A. J.; Crounse, J. D.; Clarke, A. D.; Shinozuka, Y.; Anderson, B. E.; Crawford, J. H.; Avery, M. A.; McNaughton, C. S.; Brune, W. H.; Singh, H. B.; Wennberg, P. O. Geophysical Research Letters 2006, 33, L (50) Hearn, J. D.; Smith, G. D. Physical Chemistry Chemical Physics 2005, 7, 2549.

114 96 Chapter Three Modification of Cloud Condensation Nucleus Activity of Organic Aerosols by Hydroxyl Radical Heterogeneous Oxidation I. J. George, R. Y.-W. Chang, V. Danov, A. Vlasenko, and J. P. D. Abbatt, Atmospheric Environment, In Press. Reproduced with permission from Atmospheric Environment Copyright Elsevier 2009.

115 ABSTRACT The modification of cloud condensation nucleus (CCN) activity of saturated organic particles resulting from heterogeneous oxidation by OH radicals was studied. Submicron Bis-2-ethylhexyl sebacate (BES) and stearic acid particles were exposed to OH radicals in a reactor flow tube and CCN activity was monitored. Monodisperse stearic acid and BES particles were found to be converted from CCN inactive to moderately CCN active particles with a hygroscopicity parameter of up to κ = 0.08 as a result of OH exposures equivalent to atmospheric exposure timescales of several days to a week. The oxidation of stearic acid particles led to a 50% reduction in particle volume at high OH exposures, indicating an enhanced degree of volatilization of oxidation products compared to oxidized BES particles, along with possible shape/phase change. Surface tension measurements of water extracts of oxidized BES films showed a significant reduction in surface tension due to oxidation. Köhler calculations modeling the CCN measurements suggest that the surface active oxidation products play an important role.

116 INTRODUCTION Atmospheric aerosols containing organic matter play a key role in climate by directly interacting with radiation and indirectly by acting as cloud condensation nuclei (CCN) that activate to form cloud droplets, thereby altering cloud properties. The role of organic aerosols (OA) in cloud formation is dependent on their wide ranging chemical composition and physical properties, and is therefore one of the major uncertainties in current climate modeling. 1,2 To address these uncertainties, current research has focused on understanding how the highly variable physical and chemical properties of OA can affect their ability to act as CCN to form cloud droplets. 2 OA can undergo chemical transformations in the atmosphere through several physical and chemical aging processes, including condensation of soluble inorganic and oxidized organic species, condensed-phase chemistry, cloud processing, and heterogeneous reaction with gas-phase oxidants such as O 3, OH, NO 3 and halogen atoms. 3 These aging processes can convert organic aerosols from hydrophobic into hydrophilic particles, thus altering their hygroscopic properties and CCN activity. Currently, heterogeneous oxidation is the least understood of these aging processes. Therefore, its importance in determining the timescale for hydrophobic-tohydrophilic conversion of OA relative to other aging processes cannot be accurately gauged and has been the subject of recent laboratory work. 3,4 Heterogeneous reaction of OA may significantly alter its CCN activity through two possible routes: 1) addition of water soluble material and 2) production

117 99 of surface active species. Köhler theory predicts that a reduction in solution surface tension from the presence of surface active species reduces the vapor pressure over the droplet and, consequently, the supersaturation required for particles to form cloud droplets. Surface tension depression from the presence of surface active organic species has been measured in cloud water, fog water, and aerosol extracts The role of surface active organic species on the CCN activity of chemically aged OA, especially primary OA (POA) containing saturated organic matter, is not known. To date, the work of Petters et al. 12 has been the only research to our knowledge that has focused on hygroscopicity changes of model saturated POA due to chemical aging processes involving heterogeneous reaction. Petters et al. 12 focused on a low OH exposure (atmospheric OH exposure ~ 0.2 days) relative to the typical atmospheric lifetime of OA. The impact of higher, atmospherically relevant OH exposures and the influence of surface active species on CCN activation of saturated OA remain to be explored. In this study, the modification of CCN activity of saturated organic aerosols from heterogeneous reaction with OH radicals was investigated. Laboratory generated Bis(2-ethylhexyl) sebacate (C 26 H 50 O 4, BES) and stearic acid (C 18 H 36 O 2, SA) particles, used as conceptual models for POA, were oxidized in a reactor flow tube from exposure to OH radicals and resulting changes in their CCN activity was measured. In addition to the CCN measurements, we provide two sets of complimentary measurements to shed further light on the nature of the CCN activation changes due to chemical aging. First, in continuation of our previous work

118 100 characterizing changes in particle properties of organic aerosols from chemical aging, 13 we studied the size changes of SA particles as a result of OH-initiated oxidation. Our previous study focused on the particle properties of liquid BES particles, whereas this work includes the oxidation of organic particles that are solid at room temperature. Thus, a comparison of the two systems may shed some light on the role of particle phase on the aging process and resulting modification of CCN activity. Second, surface tensions were measured of water soluble organic matter (WSOM) extracts of oxidized BES films used as the model heterogeneous oxidation system. These measurements were applied to the Köhler model to predict the measured CCN activities of chemically aged OA by evaluating the importance of water soluble oxidation products and surface active species. 3.3 EXPERIMENTAL METHODS Particle Oxidation Setup Figure 3.1 displays a diagram of the experimental setup used to expose organic particles to OH radicals. For a more detailed explanation of the reactor flow tube setup, refer to our previous work. 13 Organic particles were generated by homogeneous nucleation, where a flow of N 2 (BOC, %) was passed over BES (Fluka, > 97%) or SA (Fisher, >97%) in a Pyrex tube heated to ~110ºC. The aerosol flow passed through a neutralizer and was size-selected with a differential mobility analyzer (DMA, TSI model 3081). To minimize gas-phase reactions with OH, volatile organics were removed from the monodisperse aerosol flow with an activated carbon trap (Sigma-Aldrich, 4-14 mesh).

119 101 Figure 3.1. Experimental setup for OH oxidation of organic aerosols. The aerosol flow was then passed into a mixing chamber along with a humidified flow containing ozone. Ozone was produced by passing a mixture of N 2 and O 2 (BOC, %) with variable O 2 ratios through an O 3 generator (Jelight model 1000). O 3 concentrations during this study were varied from 1.3 to 50 ppm and relative humidity ranged from 13 to 40%. The mixed humidified O 3 /aerosol flow was introduced into the reactor flow tube, where particles were exposed to OH radicals. OH radicals were formed from the photolysis of O 3 in the presence of water vapor in the reactor flow tube by the illumination of the flow using a 22.9 cm long, O 3 -free Hg Pen-ray lamp (UVP).

120 102 Steady-state OH concentrations in the reactor flow tube were calculated using a photochemical model containing relevant gas-phase reactions with model inputs of O 3 and H 2 O concentrations dictated by experimental conditions. This model was validated in a separate set of experiments, where OH concentrations were measured by monitoring the loss of SO 2 (Matheson, 99.98%) upon reaction with OH in the reactor flow tube using a chemical ionization mass spectrometer (CIMS) in negativeion mode with SF - 6 as the reagent ion. We report an error in the calculated OH concentrations from measured OH concentrations of ±50% due to variability in flow tube conditions over time. This method for OH quantification has been used in our two previous publications on OH heterogeneous oxidation. 13, Particle Analysis Techniques Particles were passed through an O 3 denuder after exiting the reactor flow tube and were then introduced into the particle analysis instrumentation. For the tandem DMA (TDMA) experiments, particle size distributions were measured using a scanning mobility particle sizer (SMPS, TSI model 3080) including a DMA and a condensation particle counter (CPC, TSI Model 3010). In the CCN activation experiments, particles were first dried by passing the flow through a silica-gel diffusion drier (RH < 10% after dryer) and size-selected with a DMA. The aerosol flow was thereafter split to the CPC to measure the total particle concentration and to a continuous-flow thermal gradient diffusion chamber (TGDC) followed by an aerodynamic particle sizer (APS; TSI model 3320) to measure the number of activated droplets ranging in sizes of 0.5 to 20 microns. The CPC and APS each are

121 103 specified to measure particle concentration with accuracy of ±10%. A detailed description of the TGDC can be found in previous work. 15 The chamber supersaturations were calibrated by measuring the activation diameter of ammonium sulfate (AS) particles formed by atomizing (TSI 3076) a 0.03% by mass AS (Fluka, >99.5%) solution at a range of supersaturations. We applied the κ-köhler theory as outlined in Petters and Kreidenweis 16 to calculate effective supersaturation (S eff ) values for AS particle calibrations and to evaluate the changes in water uptake properties of the reacted particles. The κ-köhler theory is based on the equilibrium Köhler theory that relates the droplet diameter (D) and particle properties to the saturation ratio over the solution droplet s surface (s): s = a w 4M wσ sol exp (3.1) RTρ wd where a w is the water activity, M w is molecular weight of water, σ sol is the surface tension of the solution, R is the universal gas constant, T is temperature, ρ w is the density of water. Note that supersaturation S is related to s in Equation 3.1 as follows: S=(s-1) 100. In this approach, the a w term in the Köhler equation (3.1) is expressed as a function of κ, the sole parameter that describes hygroscopic properties of the particle: 3 3 D D0 a w = (3.2) 3 3 D D 0 ( 1 κ )

122 104 where D 0 is the dry particle diameter. Kappa values can range from κ = 0 corresponding to an insoluble, wettable particle to κ > 0.5 for highly soluble inorganic salts such as NaCl and AS. We used κ = to calculate S eff values for AS calibrations based on experimental activation diameters taken from the sigmoidal fits to the activation curves. The standard deviation in the effective supersaturation values is 0.04%. AS calibrations were performed several times throughout the study over an effective supersaturation range of 0.14 to 0.67% Surface Tension Measurements BES films were oxidized by OH radicals in a coated-wall flow tube reactor system, which has been explained in detail elsewhere. 17 Similar experiments involving the OH reaction with organic films have been undertaken in our group. 18 The organic films were prepared by coating the inner wall of a Pyrex tube of 5 cm length and 1.8 cm ID with 0.05 ml of BES. The coated Pyrex tube was placed into the flow tube and a carrier gas flow of 500 sccm (STP) of O 2 was passed through it. OH radicals were formed by the following reaction: H + NO2 OH + NO (3.3) H atoms were generated by passing a flow of argon (50 sccm) containing trace amounts of H 2 through a microwave discharge cavity. The H atoms were delivered to the coated-wall flow tube reactor by flowing through a 50 cm length of cm o.d. Teflon tube that runs through the movable injector. At the same time, a N 2 flow containing a small fraction of NO 2 is passed through the injector around the Teflon

123 105 tube. The moveable injector was pushed completely forward to monitor the OH concentration via NO - 2 signal in the CIMS that is connected to the outflow, or the injector is pulled back to oxidize the BES film with OH radicals. OH radical concentrations were approximately molecules cm -3 with an error of approximately ±30% and exposure times ranged from 1 to 8 hours corresponding to an average OH to BES ratio (i.e. the number of OH radicals reacted with film to the total number of BES molecules in the film) of up to 0.2 assuming each OH molecule produced is lost to the film surface. 18 Reacted BES films were washed off the coated wall with 5 ml of methanol. The solvent was subsequently evaporated, and 9 ml of Milli-Q water were added and allowed to sit overnight to extract water soluble products. Surface tension measurements of the water extracts of the BES films that were concentrated to 1 ml were performed using the capillary rise method. 19 The water extracts were subsequently evaporated to dryness under a N 2 stream to determine WSOM mass. The capillary rise method was validated by measurement of solutions of n-butanol, hexanoic acid and NaCl, where experimental values were consistent with literature values and deviations were within 5%. Surface tension measurements were taken under constant temperature of 25 C controlled with a water bath RESULTS AND DISCUSSION Modification of CCN Activity The influence of heterogeneous oxidation on the hygroscopicity of organic aerosols was probed by measuring CCN activity of unreacted and reacted BES and

124 106 SA particles in the TGDC as a function of OH exposure. In control experiments, we found that dry SA and BES particles did not activate in the absence of OH within the experimental supersaturation range of this study as corroborated by literature. 12,15 Figure 3.2 shows the activation curves for dry BES particles of 145 nm mobility diameter at several OH exposures. It is evident in Figure 3.2 that the critical supersaturations (S crit ), i.e. values at the point of inflection on the sigmoidal fit to the data, decrease with increasing OH exposures indicating increased CCN activity. These measurements were repeated over a range of OH exposures (0 to molecules s cm -3 ) for dry 145 nm diameter BES and 150 nm diameter SA particles as summarized in Figure 3.3. The critical supersaturations for the activation curves at low OH exposures and in the absence of OH were greater than the experimental range and were represented in Figure 3.3 as having S = 0.67% as a lower limit. Figure 3.3 shows that both SA and BES particles, initially CCN inactive, become CCN active after being oxidized by OH radicals. The plot includes the S crit value calculated from Petters et al. 12 for 145 nm BES particles under the OH exposure in their study with the assumption of OH uptake coefficient of unity. Although our work focused on higher OH exposures compared to the Petters et al. 12 study, our findings are consistent with their results.

125 107 Figure 3.2. CCN activation curves of 145 nm mobility diameter BES particles at several OH exposures. All lines are sigmoidal fits to the curve. The S crit of reacted particles appears to plateau to a minimum of 0.24% for 145 nm BES particles and 0.34% for 150 nm SA particles at OH exposures greater than approximately molecules s cm -3. Experimental OH exposures were converted to atmospheric OH exposure times as shown in the top x-axis in Figure 3.3 assuming the 24-hour average OH concentration is cm -3. Thus, the CCNinactive organic particles were converted to CCN active aerosols under equivalent atmospheric OH exposures between 0 and 5 days. It appears that slightly lower supersaturations were required to activate BES particles than SA particles in the plateau region, but the spread in the measurements suggests that this difference is likely not statistically significant. There may be several possible explanations for this

126 108 small difference in CCN activity between BES and SA particles, such as differences in particle phase, solubilities and molecular weights of the condensed-phase oxidation products, and their surface active properties. The S crit data shown in Figure 3.3 were used to calculate κ values from Equation 3.2 substituted into Equation 3.1. We initially made the assumption that σ sol = J/m 2, the surface tension of pure water at T = 298 K. Figure 3.4 summarizes the κ values as a function of OH exposure calculated from the data shown in Figure 3.3. The minimum κ values that can be inferred in this experimental work for particles with critical supersaturations greater than S = 0.67% were κ = for D 0 = 150 nm and for D 0 = 145 nm. Figure 3.4 shows that calculated κ values increased with OH exposures to plateau values of κ ~ 0.08 for BES and κ ~ 0.04 for SA. In comparison, Petters et al. 12 observed an increase in the κ value for BES particles from κ 0 to for an equivalent atmospheric OH exposure of 0.2 days. The κ values in this work are several times higher than the maximum suggested κ values for chemically-aged POA (κ 0.01) including OH-oxidized BES particles and are more consistent with particles containing sparingly soluble or high molecular weight multifunctional organic species, such as adipic acid, secondary organic aerosol and fulvic acid. 16 Note that the κ values reported here may be overestimated if surface tension depression from surface active species has a significant impact on CCN activation of the oxidized model OA, which will be explored later.

127 Figure 3.3. Critical supersaturation of 145 nm mobility diameter BES particles and 150 nm mobility diameter SA particles. Open symbols represent data where particles were exposed to OH radicals and solid symbols show data from control experiments in the absence of OH. 109

128 110 Figure 3.4. Kappa values corresponding to the data in Figure Particle Size Change In previous work, we have reported size change of oxidized BES particles. 13 Here we show the particle size change of monodisperse 100 nm diameter and polydisperse SA particles with a mean diameter of 150 nm normalized to total particle concentration as a function of OH exposure in Figure 3.5. The volume change (V/V 0 ) values shown in Figure 3.5 were calculated by comparison of the particle volumes of SA particles exposed to OH radicals (V; Hg lamp on, with O 3 ) to unreacted particles (V 0 ; Hg lamp on, no O 3 ). Values at OH exposure of 0 molecules s cm -3 represent conditions where the Hg lamp was on and O 3 was present in a dry flow

129 111 (RH = 0%). From these values we estimated an error in the volume change measurements of ~10%. Both monodisperse 100 nm and polydisperse SA particle volumes decrease as a function of OH exposure to a maximum decrease of 50% with no difference in size change trends between polydisperse and monodisperse aerosols. These results are in contrast to our previous work, where BES particles (D 0 =150 nm) were observed to decrease in volume only at relatively high OH exposures with a maximum observed loss of 17% at an OH exposure of ~ molecules s cm The most likely explanations for the measured particle volume loss due to oxidation are volatilization of reaction products or particle shape change. Figure 3.5. Volume change normalized to particle number concentration of 100 nm mobility diameter (squares) and polydisperse (circles) stearic acid particles as a function of OH exposure. Error bars represent one standard deviation of averaged data.

130 112 Recently, our group has characterized the gas-phase product yields ([VOC]/[OH]) from OH radical oxidation of BES and SA films in a coated wall flow tube setup using a Proton Transfer Reaction Mass Spectrometer and CIMS. 18 We applied the product yields from the OH oxidation of SA from Vlasenko et al. 18 to infer density changes by comparing expected particle mass changes from oxidation to the particle volume change measurements in this work. Particle mass change calculations are described in Supplementary Material in Section 3.7. The expected net particle mass change over the range of OH exposures is within 2% of unity, resulting in a calculated particle density increase of a factor of two at high OH exposures when the observed volume change is taken into account. Given that particle density changes due to heterogeneous oxidation have been experimentally observed to be no more than approximately +20% for ozonolysis of oleic acid particles 20,21 and for OH oxidation of BES particles, 13 the density change calculated here appears to be unreasonably high. This suggests that particle oxidation and volatilization may not completely account for the measured volume changes, and that, instead, a fraction of the DMA size decrease is likely due to shape or phase changes leading to more spherically-shaped particles. This hypothesis is supported by measurements with an Aerodyne Aerosol Mass Spectrometer (AMS). Mean SA particle vacuum aerodynamic diameter (D va ) from AMS measurements and SMPS mean mobility diameter (D m ) values were compared over a range of OH exposures, where the ratio D va /D m is equal to the particle density for spherical particles with no internal voids. 22 The D va /D m ratio was

131 113 observed to increase with OH exposure up to approximately +50%, whereas a density increase similar to what was observed for BES particles in the absence of a shape/phase change would only lead to a +20% increase in this ratio. The greater than expected enhancement in the D va /D m ratio can be explained by a morphological transition to a more spherically shaped particle. Therefore, these results provide further evidence that oxidation of SA particles leads to a shape/phase change to form more spherical particles. Particle phase may play an important role in the hygroscopicity and CCN activity of organic aerosols. In some cases, organic particles that are not completely dried or are in a liquid state may reduce the barrier for dissolution of the solute that needs to be overcome to activate particles to cloud droplets. Previous studies have observed greater CCN activity than expected for some sparingly soluble organics suggesting that these particles were in a metastable liquid state This is one reason that may explain why SA particles become more CCN active through oxidation, but not the whole reason given that BES particles also become more CCN active and were liquid phase before oxidation. 13 Overall, there are likely several factors that may drive the hygroscopicity changes observed including phase and solubility, but also surface tension depression, which we will explore below Surface Tension Measurements We investigated the role of surface active organic species on the CCN activity of aged organic aerosols by taking surface tension measurements of water extracts of BES films exposed to OH radicals in a coated-wall flow tube CIMS setup. The

132 114 surface tension measurements were then integrated into the Köhler model to better predict our CCN measurements for oxidized BES particles. These surface tension measurements are summarized in Figure 3.6. In control measurements, the surface tension of water extracts of unreacted BES films was σ = (±0.001) J/m 2, comparable to the surface tension of water (σ w = J/m 2 ) at 298 K. Figure 3.6 shows a clear surface tension reduction of WSOM extracts (1 ml) with increased OH exposure to a maximum 25% decrease compared to σ w value that corresponds to an increased production of WSOM mass. The surface tension measurements are displayed as a function of WSOM concentration in Figure 3.7. Note that we did not observe a plateau effect that is expected to occur at WSOM concentrations nearing the critical micelle concentration. The empirical fit to surface tension data falls between values of saturated azelaic and nonanoic acid solutions shown here as reference surface active species.

133 115 Figure 3.6. Surface tension (left axis) and WSOM mass (right axis) as a function of OH exposure time (bottom) and OH to BES ratio (top) for oxidized BES film extracts. Error bars represent one standard deviation of averaged data. Figure 3.7. Surface tension as a function of WSOM concentration for oxidized BES film extracts (1 ml). Dashed line shows the exponential fit to the BES data. Dotted line represents water surface tension (σ w ). Star and diamond symbols represent surface tension of saturated azelaic acid and nonanoic acid solutions, respectively.

134 116 Although the major products from the OH oxidation of BES particles have been previously identified, 13 their properties (surface tension, solute density, solubility) are unknown. And so, to model the CCN activity measurements with Köhler theory using the measured surface tensions from the film extracts, we must make assumptions regarding the following parameters: 1) the WSOM mole fraction in the reacted BES particles (F WSOM ) and 2) the water solubility of the WSOM products (C sat ). Note that there are considerable uncertainties in these parameters; we approach these Köhler calculations in a mechanistic manner to see if reasonable assumptions can explain the observed data. Electrospray ionization mass spectrometry (ESI-MS) was applied to estimate the WSOM fraction of reacted BES particles. In particular, we compared ESI-MS mass spectra of water extracts of oxidized films to those of methanol extracts of oxidized particles under OH exposure of molecules s cm -3 from previous work 13 to identify the water soluble products that would most likely constitute the WSOM fraction of the oxidized BES particles. This comparison of mass spectra (Fig. 3.8) illustrates that the products formed from the addition of at least 4 oxygenated functional groups, i.e. alcohols and carbonyls, to the BES carbon chain (m/z range = ) or the addition of at least 1 oxygenated functional group to smaller molecular weight scission products (minor products in BES particle oxidation, m/z range = ) were the most soluble in the 1 ml water extracts. The observed number of added functionalities per BES molecule in the water extracts of oxidized BES film may seem unreasonably high compared to the average

135 117 OH to BES ratios, which were less than unity for the film oxidation experiments. However, this can be somewhat attributed to an inhomogeneity in the BES film thickness that was caused by the liquid slowly pooling to the bottom of the coated wall tube during the experiments. Therefore, BES molecules in the thin part of the film would have experienced a greater number of OH collisions per BES molecule compared to the thicker film region allowing for the addition of oxygenated groups per BES molecule greater than expected from the OH to BES ratios. Figure 3.8. ESI-MS spectra for A) oxidized BES film water extract with reaction time of 2 hours and B) oxidized BES particle methanol extract adapted from George et al. (2007). Peaks marked with asterisks represent masses that were included in the calculation of F WSOM. Boxes encompass peaks with asterisks that were chosen as WSOM products.

136 118 To estimate the WSOM fraction in the oxidized BES particles, we assumed that the products whose peaks are marked in Fig. 3.8b with asterisks inside the boxes comprise the WSOM fraction. The unreacted BES and other product peaks marked by asterisks outside the boxes compose the water insoluble fraction (1-F WSOM ). By summing the peak height intensities for both soluble and insoluble products shown in Fig.3.8b, we calculated the WSOM mole fraction of 21% for oxidized BES particles. We applied a modified form of the Köhler theory as outlined in Broekhuizen et al. 23 to predict the measured CCN activity of oxidized BES particles using WSOM concentration dependent droplet surface tensions (see Supplementary Material). The reacted BES particle is modeled as a simplified case containing 2 components: 1) an insoluble fraction of unreacted BES and oxidation products and 2) a WSOM fraction of sparingly soluble oxidation products. The insoluble component and undissolved WSOM fraction form an insoluble core in the droplet. The amount of undissolved WSOM material varies with droplet diameter and its water solubility (C sat ). To determine whether the measured enhancement in CCN activation from heterogeneous oxidation can be attributed solely to solubility changes, the solubility of the WSOM fraction of the 145 nm diameter BES particle was varied from being insoluble (C sat = 0 kg/kg) to fully soluble (C sat = 5 kg/kg) in the Köhler model (Figure 3.9). Solution surface tensions in the droplet were represented initially as σ w. The supersaturation needed to activate a 145 nm diameter particle assuming that it is fully insoluble (S crit = 1.4%) is qualitatively consistent with our measurements of CCN activity of unreacted BES particles. In the opposite case, a BES particle with 21% of the particle

137 119 mass that is fully water soluble has a critical supersaturation of S crit = 0.66%. Because the observed plateau value of critical supersaturation for oxidized BES particles is S crit = 0.24 %, it is clear from these two extreme cases that solubility changes alone cannot explain the observed enhancement of CCN activation of the model OA due to heterogeneous oxidation. Therefore, we conclude that we must consider surface tension lowering effects and solubility changes simultaneously in the Köhler model. In the following cases, solution surface tensions were modeled using the fit to the surface tension measurements of the reacted BES film extracts shown in Figure 3.7. As an initial approximation, the 21% WSOM component was calculated as being moderately soluble with a solubility ranging from C sat = to kg/kg, similar to saturated azelaic acid and nonanoic acid solutions, respectively. The two curves in Figure 3.9 representing C sat = and kg/kg show that an increase in the solubility can drastically reduce the S crit value, which is almost solely due to the surface tension lowering effect of the WSOM component. The critical supersaturation is reduced from S crit = 1.1% for C sat = kg/kg, corresponding to the point at which the particle begins to take up water vapor and dissolve, to S crit = 0.11% for C sat = kg/kg, where the particle is completely dissolved at activation. This range of S crit values widely encompasses the measured S crit values in this study suggesting that the WSOM solubility is between that of the reference solutions. A solubility of C sat = kg/kg was needed to match the S crit values measured here for oxidized BES particles at the plateau region, i.e., S crit = 0.24%.

138 120 This value represents the point at which the particle begins to grow and dissolve. The calculated surface tension at activation under these conditions was σ = J/m 2 corresponding to a region in the surface tension fit in Figure 3.7 that is not constrained by measurements. Furthermore, this surface tension value is significantly lower than typical values for saturated solutions containing surface active species as well as measurements of water extracts of organic material from ambient aerosols 5,9-11. This suggests that the surface tension fit we used may likely underestimate the droplet surface tensions at low droplet radii where WSOM concentrations are highest, leading to an underestimation of the water solubility of the WSOM component that is required to predict CCN activation measurements. Although the oxidation products are assumed to have limited solubility because of their high molecular weight, they may be present as a metastable liquid state as discussed previously. The CCN activity of particles in metastable liquid states or that become wettable can be modeled sufficiently using equilibrium Köhler theory assuming full solubility. Under the assumption that the WSOM mass fraction of the particle is infinitely soluble or supersaturated, the surface tension needed at activation for a particle would be σ = J/m 2. This surface tension value appears to be more reasonable than that calculated for the moderate solubility (C sat = ( ) 10-3 kg/kg) case in light of range of surface tension values measured in previous studies. In sum, the Köhler modeling described here illustrates that increased solubility, surface tension reduction, and possibly particle phase likely play a significant role in the enhanced CCN activation of oxidized BES particles.

139 121 Supersaturation (%) Fully Insoluble, σ w 20% Fully Soluble, σ w C sat = 2.4x10-4 kg/kg, measured σ C sat = 1.5x10-3 kg/kg, measured σ C sat = 2.4x10-3 kg/kg, measured σ Range of CCN measurements Droplet Diameter (μm) Figure 3.9. Calculated Köhler curves for 145 nm diameter reacted BES particles CONCLUSIONS The work outlined in this paper adds to a growing body of research that aims to elucidate the chemical aging process of organic aerosols through heterogeneous oxidation, starting with a simple single-component model of oleic acid ozonolysis then moving to more complex systems, such as multi-component and ambient aerosols. The ultimate goal of these studies is to integrate our knowledge of the hydrophobic-to-hydrophilic conversion process of OA into climate models to more accurately evaluate the impact of organic particles on global climate change. The specific objective of this laboratory study was to determine whether OH-initiated

140 122 heterogeneous oxidation can alter the hygroscopic properties of model POA under relevant atmospheric timescales. Size change experiments of oxidized stearic acid particles indicate that volatilization and density change may not fully explain the observations, but that there was likely also a shape/phase change, which may influence the hygroscopicity of OA. Furthermore, surface tension measurements of oxidized BES films were incorporated into the Köhler calculations used to model the CCN activation measurements. These model calculations indicated that solubility alone cannot explain the enhanced CCN activity due to heterogeneous oxidation, and that surface active products likely play a critical role in the CCN activity enhancement observed in this work. Asa-Awuku et al. 5 found that the hydrophobic organic fraction was mostly responsible for the observed surface tension depression of water extracts of biomass burning aerosol. Given that POA can consist of long chain hydrocarbon species, it is entirely conceivable that chemical aging of the water insoluble fraction of OA could lead to more soluble, oxidized products with surfactant properties. In laboratory experiments involving the oxidation of stearic acid and BES particles, particles were converted from being CCN inactive (κ = 0) to being moderately CCN active (κ = ) when exposed to the equivalent OH atmospheric timescale of several days to a week. In contrast, the conversion of pure oleic acid particles to CCN active particles by ozonolysis reaction required much higher laboratory ozone exposure times than atmospheric ozone exposures. 21 Petters et al. 12 conducted a cloud parcel model study to gauge whether oxidation of POA can

141 123 increase cloud droplet activation rates. They found that an increase in the POA hygroscopicity from chemical aging to κ > 0.01 was needed for significant enhancement of POA wet scavenging rates. Given that we have observed greater changes in the hygroscopicity of model POA from OH oxidation than was observed by Petters et al. 12, we conclude that heterogeneous oxidation has the potential to be an important pathway for the hydrophobic-to-hydrophilic conversion of POA and may significantly influence the wet scavenging rates under certain conditions (i.e., high updraft velocities, low POA number concentrations). Its atmospheric importance relative to other OA aging mechanisms, such as coagulation and condensation of soluble material, should be evaluated in climate models. This paper focused on the hygroscopicity changes of single-component model organic particles exposed to relatively high OH concentrations under NO x -free conditions. Given the chemical complexity of atmospheric OA, atmospheric particles containing POA may not necessarily follow similar behavior that model POA displayed in this work upon oxidation. Therefore, this research should be expanded to include chemical aging of more complex aerosols, e.g. secondary organic aerosol and POA mixed with water soluble components. Particle aging experiments should also be conducted under lower OH concentrations and NO x levels more relevant to polluted regions. ACKNOWLEDGEMENTS Funding has been provided by NSERC and structural funding to SOCAAR was provided by CFI/OIT.

142 SUPPLEMENTARY MATERIAL Appendix A: Stearic Acid Particle Mass Change Calculation The method for calculation of particle mass change for a stearic acid particle undergoing oxidation with OH radicals is summarized as follows. First, we make the assumption that the volatile product composition and product yields from our work with OH oxidation of stearic acid films are directly applicable to the particle oxidation experiments. 18 One reason this assumption may not be legitimate is due to the presence of NO x in the film oxidation experiment from the OH source and its absence in particle experiments, which could result in differences in the product composition and product volatilization rates of oxidized organic films and particles. Research by Molina et al. involving OH-initiated oxidation of organic monolayers and films showed to the contrary that the observed volatilization rates were not altered by the presence of NO. 28 Therefore, we believe that our assumption is reasonable. From the OH oxidation of stearic acid films, we observed the production of short chain aldehydes/carbonyls and carboxylic acids with an average chain length between 3 to 5 atoms depending on the carrier gas (3 in O 2 ; 5 in He) with volatile product yield per OH collision ([VOC]/[OH]) = 0.34). 18 The mass change (Δm) of stearic acid particles due to OH radical oxidation was calculated assuming that 34% of the OH collisions with the SA particle would lead to the formation of the volatile product propanal and a C15 aldehydic acid as the nonvolatile product (net Δm = -28 amu). 18 The other 66% of the OH collisions will likely lead to the formation of a carbonyl moiety on the carbon chain (net Δm =+14

143 125 amu). The number of reactive OH collisions (N OH ) per particle was calculated as follows: N OH = γtvπr 2 [ OH ] (3.3) where γ is the OH reactive uptake coefficient taken to be unity, 13,29 t is reaction time in s, [OH] is OH concentration in molecules/cm 3, v is OH gas velocity in cm/s, and R is particle radius. Figure 3.10 summarizes the calculated relative mass change (m/m 0 ) and density change (ρ/ρ 0 ) as a function of OH exposure that results from comparing the calculated mass change with volume change measurements in Fig 3.5 using the exponential fit to the data (V/V 0 = exp(-([oh] t)/ ).

144 Mass Volume Density Relative Change OH Exposure (x10 12 molecules s cm -3 ) Figure Relative mass, density, and volume changes as a function of OH exposure Appendix B: Köhler Calculation For Oxidized BES Particle We outline the modified version of Köhler theory adapted from Broekhuizen et al. 23 that was used to calculate the Köhler curves in Fig. 3.8 for oxidized BES particles. This model accounts for a particle containing sparingly soluble species that forms an insoluble core until fully dissolved. For a 2-component model with a WSOM fraction and an insoluble fraction, the insoluble core volume (V insol = πx(d insol ) 3 /6) is calculated as:

145 127 V insol = ( 1 F ) WSOM ρ n insol dry M insol + F WSOM n dry π ( D D ) 0 ρwc M sat WSOM M ρ WSOM WSOM (3.4) where F WSOM is WSOM mole fraction, n dry is the total moles of solute in the dry particle, and C sat (kg/kg) is the water solubility of the WSOM fraction. Furthermore, the following parameters were used in the Köhler calculations in (3.4): an ESI-MS signal intensity weighted mean molecular weight of soluble fraction of M sol = 0.42 kg/mol and insoluble fraction of M insol = 0.45 kg/mol, assumed soluble fraction density of ρ sol = 1100 kg/cm 3 and water insoluble component of ρ inols = 915 kg/m 3. The first and second terms of Equation 3.4 represent the volumes of the insoluble and the undissolved WSOM fractions, respectively. Equation 3.4 is valid initially until the droplet size is reached where the WSOM fraction is completely dissolved and the second term equals 0. For larger droplet sizes the equation is reduced to the first term only. The water activity a w in the Köhler equation (3.1) is expressed as follows: a w 6ν n M = (3.5) ) ρ s w 1 π ( D 3 3 Dinsol w where ν is the van t Hoff factor, n s is the moles of dissolved solute, and ρ w is the density of water. The term n s in (3.5) was calculated as a function of D by converting V insol (3.4) to moles and subtracting this value from total dry moles of solute (n dry ).

146 REFERENCES (1) IPCC, 2007: Summary for Policymakers. In: Climate Change 2007: The Physical Science Basis. Contribution of Working Group I to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change [Solomon, S., D. Qin, M. Manning, Z. Chen, M. Marquis, K.B. Averyt, M.Tignor and H.L. Miller (eds.)]. Cambridge University Press, Cambridge, United Kingdom and New York, NY, USA (2) Sun, J. M.; Ariya, P. A. Atmospheric Environment 2006, 40, 795. (3) Rudich, Y.; Donahue, N. M.; Mentel, T. F. Annual Review of Physical Chemistry 2007, 58, 321. (4) Rudich, Y. Chemical Reviews 2003, 103, (5) Asa-Awuku, A.; Sullivan, A. P.; Hennigan, C. J.; Weber, R. J.; Nenes, A. Atmospheric Chemistry and Physics 2008, 8, 799. (6) Dinar, E.; Mentel, T. F.; Rudich, Y. Atmospheric Chemistry and Physics 2006, 6, (7) Facchini, M. C.; Mircea, M.; Fuzzi, S.; Charlson, R. J. Nature 1999, 401, 257. (8) Facchini, M. C.; Decesari, S.; Mircea, M.; Fuzzi, S.; Loglio, G. Atmospheric Environment 2000, 34, (9) Kiss, G.; Tombacz, E.; Hansson, H. C. Journal of Atmospheric Chemistry 2005, 50, 279. (10) Salma, I.; Ocskay, R.; Varga, I.; Maenhaut, W. Journal of Geophysical Research-Atmospheres 2006, 111, D23205.

147 129 (11) Taraniuk, I.; Graber, E. R.; Kostinski, A.; Rudich, Y. Geophysical Research Letters 2007, 34, L (12) Petters, M. D.; Prenni, A. J.; Kreidenweis, S. M.; DeMott, P. J.; Matsunaga, A.; Lim, Y. B.; Ziemann, P. J. Geophysical Research Letters 2006, 33, L (13) George, I. J.; Vlasenko, A.; Slowik, J. G.; Broekhuizen, K.; Abbatt, J. P. D. Atmospheric Chemistry and Physics 2007, 7, (14) George, I. J.; Slowik, J.; Abbatt, J. P. D. Geophysical Research Letters 2008, 35, L (15) Kumar, P. P.; Broekhuizen, K.; Abbatt, J. P. D. Atmospheric Chemistry and Physics 2003, 3, 509. (16) Petters, M. D.; Kreidenweis, S. M. Atmospheric Chemistry and Physics 2007, 7, (17) Thornberry, T.; Abbatt, J. P. D. Physical Chemistry Chemical Physics 2004, 6, 84. (18) Vlasenko, A.; George, I. J.; Abbatt, J. P. D. Journal of Physical Chemistry A 2008, 112, (19) Shoemaker, D. P., Garland, C.W., Nibler, J.W. Experiments in Physical Chemistry, 5th Ed. ed.; McGraw-Hill, (20) Katrib, Y.; Martin, S. T.; Rudich, Y.; Davidovits, P.; Jayne, J. T.; Worsnop, D. R. Atmospheric Chemistry and Physics 2005, 5, 275.

148 130 (21) Broekhuizen, K. E.; Thornberry, T.; Kumar, P. P.; Abbatt, J. P. D. Journal of Geophysical Research-Atmospheres 2004, 109, D (22) DeCarlo, P. F.; Slowik, J. G.; Worsnop, D. R.; Davidovits, P.; Jimenez, J. L. Aerosol Science and Technology 2004, 38, (23) Broekhuizen, K.; Kumar, P. P.; Abbatt, J. P. D. Geophysical Research Letters 2004, 31, L (24) Bilde, M.; Svenningsson, B. Tellus Series B-Chemical and Physical Meteorology 2004, 56, 128. (25) Chan, M. N.; Kreidenweis, S. M.; Chan, C. K. Environmental Science & Technology 2008, 42, (26) Hartz, K. E. H.; Tischuk, J. E.; Chan, M. N.; Chan, C. K.; Donahue, N. M.; Pandis, S. N. Atmospheric Environment 2006, 40, 605. (27) Hori, M.; Ohta, S.; Murao, N.; Yamagata, S. Journal of Aerosol Science 2003, 34, 419. (28) Molina, M. J.; Ivanov, A. V.; Trakhtenberg, S.; Molina, L. T. Geophysical Research Letters 2004, 31, L (29) Bertram, A. K.; Ivanov, A. V.; Hunter, M.; Molina, L. T.; Molina, M. J. Journal of Physical Chemistry A 2001, 105, 9415.

149 131 Chapter Four: Part One Chemical Aging of Ambient Organic Aerosol from Heterogeneous Reaction with Hydroxyl Radicals I. J. George, J. Slowik, J. P. D. Abbatt, Geophysical Research Letters 35, L13811, doi: /2008gl Reproduced with permission from the Geophysical Research Letters Copyright American Geophysical Union 2008.

150 ABSTRACT The chemical transformation of ambient particles due to heterogeneous oxidation was investigated. Ambient aerosol particles sampled in Toronto, Ontario during September and October of 2007 were exposed to OH radicals in a flow tube reactor to simulate relevant atmospheric OH exposures. An Aerodyne Aerosol Mass Spectrometer measured changes in chemical composition and degree of oxidation of reacted ambient particles. Mass spectral changes of ambient particles exposed to OH revealed a loss in the hydrocarbon-like organic fraction along with an increase in the oxygenated organic fraction. Furthermore, a minor fraction of the organic mass was lost through volatilization at high OH exposure. These observations indicate that an atmospheric OH exposure time of no more than 4 days for a typical OH concentration of cm -3 causes significantly enhanced oxidation of the organic matter in ambient particles. Therefore, heterogeneous oxidation may be an important chemical transformation pathway for ambient particles during regional transport.

151 INTRODUCTION Atmospheric processing of ambient particles significantly transforms the chemical composition and particle properties of organic aerosol (OA), thereby altering its role in climate and environmental processes. One important mechanism of atmospheric aging is the heterogeneous reaction of OA with atmospheric oxidants such O 3, OH, and NO 3. 1 Heterogeneous oxidation of OA is not well characterized and thus has been the focus of many recent laboratory studies, which seek to elucidate the kinetics, reaction mechanisms, and modification of particle properties (e.g., size, density, hygroscopicity) from the oxidation of both single and multicomponent organic aerosols and surfaces. 2 Many of these studies have employed the condensedphase oleic acid + O 3 reaction as a model to study heterogeneous oxidation of ambient particles in a laboratory setting. Fewer studies have investigated OH oxidation of condensed saturated organics (George et al. 3 and references therein). Robinson et al. 4 gave strong evidence that atmospheric OA is oxidized by OH radicals through heterogeneous reaction during regional transport. In that study, the concentrations of two hopanes (condensed-phase organic tracers for motor vehicle exhaust that are reactive toward OH), normalized to elemental carbon (EC) concentration, were compared in a large ambient composition data set. The hopane to EC ratios were significantly depleted in summertime measurements compared to those in wintertime and to emission sources, suggesting that oxidation by OH is the explanation for these observations. Therefore, heterogeneous oxidation may be an important mechanism for chemical processing of OA.

152 134 An alternate method by which the degree of oxidation of OA can increase is through the condensation of secondary organic aerosol (SOA) mass that formed from the vapor-phase oxidation of volatile organic compounds (VOCs). In recent field studies, increases in the degree of oxidation of OA with photochemical age have been attributed solely to the production of SOA mass. 5-9 Similar conclusions were reached in a survey of 37 field campaigns utilizing an Aerodyne Aerosol Mass Spectrometer (AMS). 10 Although it is clear that the formation of oxidized organic mass with photochemical age is likely from SOA formation, the possibility that heterogeneous oxidation of OA contributes to the increased degree of oxidation of OA has not been fully considered. Because SOA production and heterogeneous oxidation occur during times of high photochemical activity, the relative importance of each process as a mechanism for chemical aging of OA still remains unclear. In this work, laboratory techniques previously used in a heterogeneous OH oxidation study 3 of model OA were applied to ambient particles to investigate whether ambient OA could be oxidized. An Aerodyne AMS was used to quantify changes in chemical composition. Although higher OH concentrations are used relative to atmospheric levels (as typical with many laboratory oxidation experiments), ambient particles are studied, thus bringing laboratory-based research on chemical aging closer to atmospherically relevant conditions EXPERIMENTAL Ambient particle sampling took place from the 3rd floor of Lash Miller Laboratories (80 St. George Street) facing Willcocks Street, a light traffic area on the

153 135 St. George Campus of the University of Toronto in downtown Toronto, Canada. Experiments were conducted during daytime over 5 days in late September to early October of Ambient particles were exposed to OH radicals in a heterogeneous oxidation flow tube setup described previously. 3 Because preliminary experiments showed that the ambient aerosol composition varied on a faster timescale than a typical oxidation experiment, the reactor flow tube setup was slightly modified, as described below. Before entering the reactor flow tube, the ambient particles entered a 3 L glass bulb to enhance particle mixing and reduce temporal variability in particle composition. The particle flow was then passed through an active charcoal denuder that removes volatile organic species such as small aromatics with high efficiency. The ambient flow was then split into two flows: (1) the background flow (BGF) and (2) the oxidation flow (ODF). The BGF passed undiluted through a separate flow tube, serving as the reference state. The ODF was diluted by a N 2 /O 2 flow containing O 3 that was passed through the reactor flow tube system. Both BGF and ODF were each held at a flow rate of 0.4 lpm. The O 3 flow rate ranged from 0.05 to 0.2 lpm in the ODF, and the total ambient flow rate into the experimental system, ranging from 0.6 to 0.75 lpm, was controlled by pumps connected to the system. The setup for the ODF consisted of the following components: (1) a mixing flow tube (0.8 L) aiding the mixing of the O 3 and particle flows, (2) the reactor flow tube (residence time in OH production region: 132 s), and (3) an O 3 denuder. The relative humidity for the

154 136 ODF in the reactor flow tube during the entire study spanned from 23 to 47%, and O 3 concentrations were in the range of 3.3 to 47 ppm. An Aerodyne Time-of-Flight Aerosol Mass Spectrometer (ToF-AMS) was used here to measure the size-resolved chemical composition of the non-refractory component of ambient particles. 11 Particle sampling alternated between BGF and ODF sampling modes with 2-min sampling time for each mode using a 4-way valve, where each port was connected to the following: 1) ODF outflow, 2) BGF outflow, 3) ToF-AMS, and 4) a pump. Because of a 2 minute particle residence time lag in the ODF, the ODF measurements were directly compared to the previously sampled BDF measurements. During the course of each experiment, the ODF was typically exposed to the following conditions: 1) dark (Hg lamp off), without O 3, 2) Hg lamp on without O 3, 3) dark with O 3 and 4) Hg lamp on with O 3 (i.e., with OH). Results from the first three conditions were analyzed to assess changes in chemical composition in the absence of OH, while results from the fourth condition showed the effect of OH exposure. The differences between measurements of organic mass, m/z 44, and m/z 57 taken under conditions (2) and (3) compared to (1) in the ODF normalized to the BGF values were less than the variability in the measurements under condition (1), confirming that the Hg lamp on (2) and O 3 (3) do not systematically alter particle composition significantly. The variability in organic mass, m/z 44, and m/z 57 values in ODF under condition (1) relative to those in BGF, on a given experiment day is 9%, 11%, and 11%, respectively.

155 137 OH radicals were formed in the reactor flow tube from the reaction of water vapor in the ambient flow with O( 1 D), which is produced from the photolysis of O 3 with an Hg pen-ray lamp. OH concentrations were quantified by monitoring SO 2 loss as explained previously. 3 For this study, the SO 2 experiments were carried out with the addition of an ambient flow to more accurately quantify OH during ambient experiments. The results from these experiments were used to optimize the photochemical model used to determine OH concentrations during the ambient experiments. Due to the possible daily variability in gas-phase ambient constituents and potential impacts on experimental OH concentrations, we estimate that the overall error of the OH concentrations should be increased to approximately a factor of two for these experiments, i.e., above the error estimate in George et al RESULTS AND DISCUSSION Table 4.1 summarizes the ToF-AMS non-refractory mass concentration averaged over each experiment day. Overall particle mass consisted primarily of organics, ammonium and sulfate. Chloride and nitrate concentrations were below 0.02 μg m -3 and 0.13 μg m -3, respectively, at all times (not shown). The reported mass concentrations are lower than ambient concentrations, likely due to particle losses in the experimental setup. Table 4.1 also reports the ratio of the ToF-AMS signal massto-charge ratio (m/z) 44 and 57 to the total ToF-AMS organics. These ratios provide an indication of the degree of oxidation of OA. The m/z 44 (CO + 2 ) fragment is indicative of carboxylic acids in oxygenated organic aerosol (OOA) such as aged aerosol and SOA, whereas m/z 57 has been used as a marker for hydrocarbon-like

156 138 organic aerosol (HOA) such as primary combustion and diesel exhaust particles. 12 The average relative contribution of m/z 44 ranged from approximately 6 to 14%, while m/z 57 represented 1 to 3% of total background organic mass. Particles sampled on 27 September 2008 were the least oxidized compared to those sampled on all other dates. Table 4.1. Average ToF-AMS particle mass concentrations Date Organics a Sulfate a Ammonium a Frac44 b Frac57 b 27-Sep Sep Oct Oct Oct a AMS mass concentrations in μg m -3. b Percent contribution of m/z 44 and m/z 57 to total organic mass, in percent. The degree of oxidation of the ambient OA was observed by monitoring ToF- AMS mass spectral changes between the BGF and ODF measurements, particularly at m/z 44 and m/z 57. Figure 4.1 is a comparison of 10 minute averaged BGF and ODF organic mass spectra at an OH exposure of atm-s. The mass spectrum in the bottom panel displays the difference between the ODF and BGF spectra. Exposure to OH enriches the m/z 44 fragment contribution by approximately 30%. Further, the fragments typically associated with hydrophobic organic compound classes such as alkanes, alkenes, cycloalkanes, and aromatic hydrocarbons decrease during OH

157 139 exposure. The results suggest that the hydrocarbon-like fraction of OA is being oxidized by OH and converted to more oxygenated components. Figure 4.1. ToF-AMS organic mass spectra taken on 08 October 2007: (a) BGF (bars) and ODF (crosses) mass spectra in the presence of OH with an exposure of atm-s normalized to the sum of the signals over all masses. (b) Difference spectrum of ODF and BGF from top panel. Typical fragments that arise from several organic classes are marked with filled symbols. Figure 4.2 displays the chemical evolution of the ambient aerosol particles sampled during two experiment days, where sampling alternated between the BGF and ODF modes. For direct comparison of the data from BGF and ODF sampling modes, the ODF data have been corrected for flow dilution. The dilution factor used to normalize the ODF data on a given experiment day was calculated by normalizing

158 140 the BGF to ODF ratio of the organic mass signal in the absence of OH (i.e., conditions (1), (2), and (3)) to unity. The ToF-AMS time series data (Fig. 4.2a, 4.2b) show low temporal variation in particle composition during the time period of the experiment. Data sampled during OH exposure show no clear indication of changes in major particle constituents, with the exception of a reduction in the organic mass. In Fig. 4.2c and 4.2d, the m/z 44 fraction of the total organic mass in the ODF consistently becomes enriched during OH exposure compared to BGF levels, whereas m/z 57 becomes depleted under the same conditions. The relative changes in the degree of oxidation of the ambient particles are shown more clearly in Fig. 4.2e and 4.2f, where the relative change is defined here as a measurement under ODF conditions normalized to the preceding measurement under BGF conditions. In the absence of OH, we observed no differences in the m/z 44 and 57 fractions when comparing ODF and BGF measurements. The m/z 44 fraction is enhanced under exposure to OH, while the m/z 57 fraction is reduced, indicating reaction of hydrocarbon-like species and production of more oxidized organics in the sampled OA during OH exposure.

159 Figure 4.2. ToF-AMS time series of aerosol species during 04 and 08 October, Solid symbols represent data in the ODF in the presence of OH. OH exposures for 04 and 08 October ranged from to atm-s and to atms, respectively. (a) and (b) organic (ORG), sulfate (SO4), ammonium (NH4), and nitrate (NO3) concentrations. (c) and (d) m/z 44 and 57 fractions on the left and right axes, respectively. (e) and (f) Relative change in m/z 44 and 57 fractions. (g) Mass size distributions smoothed by 5-point adjacent averaging for particles in the BGF (ORG, SO4, NH4, NO3) and for particles exposed to OH (Oxidized ORG) in the ODF with an OH exposure of atm-s during 08 October