Introductory lecture: atmospheric chemistry in the Anthropocene

The term “ Anthropocene ” was coined by Professor Paul Crutzen in 2000 to describe an unprecedented era in which anthropogenic activities are impacting planet Earth on a global scale. Greatly increased emissions into the atmosphere, re ﬂ ecting the advent of the Industrial Revolution, have caused signi ﬁ cant changes in both the lower and upper atmosphere. Atmospheric reactions of the anthropogenic emissions and of those with biogenic compounds have signi ﬁ cant impacts on human health, visibility, climate and weather. Two activities that have had particularly large impacts on the troposphere are fossil fuel combustion and agriculture, both associated with a burgeoning population. Emissions are also changing due to alterations in land use. This paper describes some of the tropospheric chemistry associated with the Anthropocene, with emphasis on areas having large uncertainties. These include heterogeneous chemistry such as those of oxides of nitrogen and the neonicotinoid pesticides, reactions at liquid interfaces, organic oxidations and particle formation, the role of sulfur compounds in the Anthropocene and biogenic – anthropogenic interactions. A clear and quantitative understanding of the connections between emissions, reactions, deposition and atmospheric composition is central to developing appropriate cost-e ﬀ ective strategies for minimizing the impacts of anthropogenic activities. The evolving nature of emissions in the Anthropocene places atmospheric chemistry at the fulcrum of determining human health and welfare in the future.


Introduction
The term "Anthropocene", proposed in 2000 by atmospheric scientist and Nobel Laureate Paul Crutzen, 1 describes the era when humans began to have a global impact on our environment. Providing the basic human needs of food, water, and shelter requires both agriculture and energy, the latter provided to date primarily by fossil fuel combustion. While emissions from such human activities have been known to impact air quality for at least 1000 years, 2 the recognition of global impacts is a relatively recent phenomenon.
The Anthropocene is generally accepted to begin with the Industrial Revolution around 1750-1800 (ref. [3][4][5][6][7][8][9][10] (although some arguments have been made that it started with agricultural activities around the beginning of the Holocene 11 600 years ago, 11 with others arguing for much later times, the late 1940's and early 1950's 12 ). Fig. 1 shows some indicators of this period of industrialization. 13,14 While population and fossil fuel use increased at the beginning of the Industrial Revolution, there was a clear acceleration in all of the indicators aer World War II. This recent period has been dubbed the "Great Acceleration". 8 The atmosphere is intimately linked to the biosphere through exchanges involving both emissions from, and deposition to, the land and oceans (Fig. 2). 15 Terrestrial emissions include both anthropogenic as well as biogenic sources which are closely intertwined through changes in land use. 16,17 Ocean-atmosphere interactions are also very important as oceans not only act as a source (e.g., of halogens and organics) but also as a sink for gases and particles. [18][19][20] Thus, a key part of the atmosphere-biosphere interaction is deposition. This is central to ecosystem functioning and health as well as productivity that supports human existence through the production of food and other ecosystem services. 21 However, deposition can also lead to deleterious effects on impacted media. One example is uptake of carbon dioxide (CO 2 ) by the oceans, which has already resulted in a 26% increase in the hydrogen ion concentration. 22 Another example is increased nitrogen deposition which can alter the uptake and emissions of other gases such as CO 2 and nitrous oxide (N 2 O), and impact biological diversity and processes both on land and in the oceans. [23][24][25] Combustion of coal, oil and natural gas have provided most of the energy needed for essentially all facets of life as we know it today. Associated with this are emissions of highly reactive NO x (]NO + NO 2 ), sulfur dioxide (SO 2 ) and organic compounds, as well as particles and toxic metals such as mercury. 26 Reactions of the direct (primary) emissions lead to formation of secondary pollutants, yielding a very complex mixture of gases and particles in the atmosphere. Indeed, the  13 World energy consumption by source from https://ourfiniteworld.com/2012/03/12/worldenergy-consumption-since-1820-in-charts/ based on ref. 14. troposphere would have a drastically different composition if not for the (photo) chemical conversion of primary emissions to secondary pollutants. Fig. 3 summarizes some overall reaction sequences beginning with the conversion of nitric oxide (NO) to nitrogen dioxide (NO 2 ) by alkylperoxy free radicals (RO 2 ) arising from the oxidation of volatile organic compounds (VOC) in air. [27][28][29][30] Species that initiate the oxidation of VOC include the hydroxyl (OH) and nitrate (NO 3 ) free radicals as well as ozone (O 3 ). There is increasing evidence for a contribution from chlorine atoms 31 and from bromine atoms reacting with  some species such as aldehydes. 18,19,32 The formation and photolysis of NO 2 serves as the sole signicant source of anthropogenically produced O 3 in the troposphere. Ozone data (Fig. 4) from non-urban areas in Europe from 1876 to 1983 (ref. 33) and more recent global data 34 show increases in many locations in the Northern Hemisphere by as much as a factor of ve from pre-industrial times to the present, which has been attributed to increased NO x emissions. 33 This rise in O 3 is a dramatic demonstration of the impact of photo-and secondary chemistry on global atmospheric composition.
Emissions of NO x and SO 2 from fossil fuel combustion lead to nitric (HNO 3 ) and sulfuric (H 2 O 4 ) acids in air, which are signicant contributors to atmospheric particles (Fig. 5). 27,35 Organics are also a major component, and their source is largely oxidation of gas phase organic precursors to form low volatility products, giving rise to secondary organic aerosol (SOA). 36 As discussed in more detail below, there are both natural and anthropogenic sources of organic precursors to SOA. This is also the case for ammonia (NH 3 ) and amines that play important roles in particle formation. Thus, anthropogenic and natural processes are inextricably intertwined.
Direct and immediate consequences of fossil fuel use include deleterious impacts on human health. Children are particularly susceptible 37-39 due to a combination of factors, including time spent outdoors and more rapid breathing; their metabolic, immune, and lung systems are also not as well developed. Health effects associated with the criteria pollutants such as O 3 , NO 2 , carbon monoxide (CO) and SO 2 are well documented, [40][41][42][43][44] and include a variety of pulmonary and cardiovascular effects. Ozone is especially toxic to humans and stringent air quality standards are set for this secondary pollutant; for example, the World Health Organization guideline for O 3 is 50 ppb (100 mg m À3 for 8 h mean exposure time). 45 Ozone also plays a major role in particle formation in air, and is a greenhouse gas, one of the short-lived climate forcers (SLCF) 16,46-49 whose lifetimes are less than that of CO 2 that yet still make signicant contributions to climate change. 50 Relatively less is known about the health impacts of particulate matter (PM), especially ultrane particles (<100 nm), although there is increasing evidence that they have a wide range of systemic effects which are summarized in Fig. 6. 51 Overall, particulate matter has been linked to an increased risk of asthma, cardiopulmonary disease and pulmonary effects including lung cancer, as well as  increased mortality and reduced life expectancy. [52][53][54][55][56][57][58][59][60][61][62][63][64][65][66][67] Still in their infancy are studies of the potential effects of PM 2.5 (particulate matter less than 2.5 mm in diameter) on diseases associated with the reproductive system (premature births, low birth weight, fetal growth and sperm quality) 37,68-73 and with the brain (neurodegenerative diseases such as Alzheimer's and Parkinson's disease, [74][75][76][77][78] strokes, 79,80 autism, 81,82 anxiety, 83 and dementia 84 ). A mechanistic basis for the effects of PM may be translocation of particles to the brain. 75,[84][85][86] Interestingly, cognitive decits due to air pollution were suggested around the year 1200, when the philosopher Maimonides attributed "dullness of understanding, failure of intelligence and defect of memory" to air pollution. 2 A recent study 87 linked outdoor air pollution, primarily particles, to 3.3 million premature deaths worldwide per year.
Fossil fuel combustion also intimately links air quality problems to climate change, 49,[88][89][90][91] and is largely responsible for the dramatic rise in atmospheric CO 2 to >400 ppm today. 92 Other human activities have led to increases in climateactive methane (CH 4 ), N 2 O, chlorouorocarbons (CFCs) and stratospheric water vapor, 93 particles and SLCF such as O 3 and hydrouorocarbons (HCFCs). 16,34,[46][47][48][49] Particles play a signicant role in climate change, but in much more complex ways because of their diverse and changing compositions. All particles scatter light, and a subset act as nuclei for the formation of both liquid and ice clouds. 34 The ability of particles to scatter light and participate in cloud formation has partly masked the warming effects of greenhouse gases. 34 However, soot/black carbon and some organics known as "brown carbon" 94 are strong light absorbers and contribute to warming.
In addition to fossil fuel combustion, the Anthropocene is characterized by increasing agricultural activities. 95,96 This impacts land use and involves application of chemicals such as fertilizers and pesticides, causing changes in emissions which inuence both air quality and climate. 95,96 The interrelationships can be quite complex. For example, the conversion of forests to croplands and grasslands changes biological volatile organic emissions, which inuences the formation of O 3 , particles and the oxidation capacity of the atmosphere 16,17 and hence the lifetime of gases such as CH 4 ; these impacts of land-use change on emissions and atmospheric chemistry can have impacts on climate that are similar in magnitude to the well-recognized associated changes in surface albedo and carbon release. 17 Conversion to croplands can also result in loss of soil carbon which would normally be sequestered. For example, Fig. 7 shows model-predicted percentage change in soil organic carbon per acre for U.S. croplands over a 30 year period; 97 about three quarters of the cropland in the U.S. is estimated to have suffered loss of soil organic carbon.
The costs associated with impacts of air pollution and climate change are difficult to accurately assess, but are clearly substantial. [98][99][100][101][102][103][104][105] For example, in addition to harming human health and impacting climate change, air pollutants may also alter weather, for example, through increased probabilities of heat waves, droughts, changes in rainfall and impacts on clouds and cloud dynamics that affect damaging storms and extreme weather events such as hurricanes and oods. [106][107][108][109][110][111][112][113][114][115] Air pollutants damage crops and forests and diminish gross primary productivity and crop yields. [116][117][118][119] It is noteworthy that changes in diffuse radiation from light scattering by particles can change gross primary productivity (GPP) of forests and of croplands and grasslands, and impact emissions and uptake as well. [120][121][122] In short, the changes in the atmosphere and its chemical processes during the Anthropocene are enormous. Thus the topics treated here will of necessity be selective, and the citations representative rather than comprehensive. For example, some key areas such as stratospheric ozone and its relationship to climate change are not addressed; the author apologizes in advance for omissions that have resulted. To narrow the scope of this article, the focus is on the atmospheric chemistry primarily associated with fossil fuel combustion and agriculture, and their impacts on air quality and climate. Atmospheric chemistry is at the core of understanding these linkages and thus provides the basis for the development of effective control strategies directed to protecting human health and welfare.
Tropospheric NO x chemistry A distinctive feature of the Anthropocene is increased emissions of NO x which forms O 3 as discussed earlier. However, NO x is also associated with the formation of other highly reactive compounds in the troposphere. An important example is the formation of nitrous acid (HONO) which is commonly found to be the major source of OH in the early morning hours, and oen the dominant source when averaged over 24 h: [123][124][125][126][127][128] HONO + hn (l < 400 nm) / OH + NO (1) Chemistry involving active oxides of nitrogen, including HONO, [130][131][132][133] has also been documented in the Arctic and Antarctic snowpack, [134][135][136] and more recently in mid-latitude snowpacks. 137 HONO is also believed to play a role in indoor air chemistry. 129,[138][139][140][141][142][143][144] Despite this, the relative importance of sources of atmospheric HONO are not yet well dened or quantied. 145 There are direct emissions of HONO from combustion, [146][147][148][149][150][151][152][153] and the reaction of OH with NO also generates a small steadystate HONO concentration during the day (since HONO photolyzes rapidly back to OH + NO).
While NO 2 does not react at a signicant rate in the gas phase with water, it is known to react on surfaces to form HONO and HNO 3 [154][155][156][157][158][159][160][161][162][163] However, the mechanism and kinetics on different surfaces remain uncertain, which precludes accurate representation in atmospheric models. One possible mechanism involves the asymmetric form of the NO 2 dimer, ONONO 2 , which can autoionize in the presence of water to form NO + NO 3 À ; subsequent reaction of the ion pair with surface-adsorbed water generates HONO and HNO 3 . [164][165][166][167][168][169] In the troposphere there are widely varying amounts of water vapor, which typically form different structures on surfaces, ranging from islands at low relative humidity (RH) to multi-layer lms at high RH. 170 It is likely that the details of the water structure affect its interactions with gases such as NO 2 , or the dimer N 2 O 4 , as well as the nature of the surface-bound species. For example, theoretical treatment of N 2 O 4 on a thin water lm carried out using ab initio molecular dynamics simulations show that a water lm can stabilize the asymmetric form of the dimer, ONONO 2 so it is available for reaction. This also activates it towards nucleophilic attack (e.g., by chloride ions, see the section on biogenic-anthropogenic interactions below). 171 The presence of inorganic ions 172 or organic surfactants 173 has been shown to affect the kinetics of conversion of NO 2 to HONO at the interface of aqueous microjets, and presumably may also do so on surface-bound water.
Other potential sources of HONO in the atmosphere include NO 2 dark reactions with soot [174][175][176][177][178][179][180][181][182][183][184][185][186] and with organics, [187][188][189][190] photoenhanced interactions of NO 2 on both inorganic and organic surfaces, [191][192][193][194][195][196][197][198][199][200][201][202][203][204][205][206][207] and photolysis of o-nitrophenols. 201,208 Photocatalysis by TiO 2 and components of mineral dust and building materials has also been shown to cause the conversion of NO 2 to HONO. [209][210][211][212][213][214][215][216][217][218] Despite this large body of research on the formation of HONO, signicant discrepancies remain between the results of laboratory studies and many atmospheric measurements of HONO, 128,219-224 leading to the possibility that there are still unidentied sources. If nitrication processes are responsible, increased fertilizer use since about 1950 ( Fig. 1) is expected to lead to increased HONO emissions and hence increased oxidative capacity of the atmosphere. Recent studies 225 conrm an earlier report 226 that there is a source in soils. Soil-associated sources include ammonia-oxidizing bacteria 227,228 and direct emissions from biological soil crusts which contain lichens, mosses, algae, cyanobacteria, heterotrophic bacteria, fungi and archaea. 229 Nitrication converts ammonium to nitrite (NO 2 À ) and then nitrate (NO 3 À ). Under sufficiently acidic conditions, NO 2 À is released as HONO to the gas phase 225,230,231 as depicted in Fig. 8a. However, it has been shown that the nature of the surface, and particularly surface charge, plays an important role in the uptake and release of HONO from soils. 232 For example, as shown in Fig. 8b, soil surface components such as aluminum and iron oxides/hydroxides are present in a number of different forms such as MO À , M-OH and MOH 2 + (M ¼ metal) which interact with oxides of nitrogen, changing the pH at which HONO is released to more basic conditions compared to pure water by as much as several pH units. 232 The relative importance of these HONO sources will depend on location, the nature and chemical composition of the surfaces, and atmospheric conditions such as solar intensity and the amount of water as indicated by the relative humidity (RH). Although HONO is ultimately derived from NO x , it does not always trend with measured NO 2 concentrations. 152,233,234 While model-predicted contributions of various sources to HONO suggest that processes involving NO 2 conversion on the ground (both dark and photoenhanced) are important, 152,220,235 the identity and reactivity of the key surface intermediates that generate HONO aer the exposure of surfaces to oxides of nitrogen, water vapor and/or light remain controversial. 160,196,236 As described above, theory predicts the formation and chemistry of intermediates such as the asymmetric N 2 O 4 dimer and NO + NO 3 À . [164][165][166][167][168][169] It is likely that similar intermediates are involved at ice interfaces. 167,206 Other surface-bound species may include nitric acid or nitrate. 194,195,237,238 The absorption cross sections for HNO 3  reported to be larger than in the gas phase, 239 which may reect interactions with other, as yet unidentied, species on the surface. This is consistent with the much higher than expected photolysis rates of surface-bound nitrate species in "urban grime". [240][241][242] For example, while one might expect the surface HNO 3 to be dissociated in the presence of water, there is evidence that a large fraction is actually in the form of nitric acid hydrates and even the nitric acid dimer. 243 There is also indirect evidence for a rich and reactive surface-bound "soup" of different species in concentrated nitric acid, 243,244 which might bear some similarity to the surface-bound acid concentrated in urban surface lms. Finally, reactions between NO x surface species can occur, such as that of adsorbed HNO 3 with NO to generate NO 2 and some HONO, which was hypothesized to react further with surface HNO 3 . 245,246 These heterogeneous reactions not only demonstrate the importance of deposition, but also show that a mechanism is available for NO x to re-enter the gas phase and contribute to further O 3 formation and other secondary chemistry.

on surfaces have been
Along with HONO, a small yield of N 2 O from the surface hydrolysis of NO 2 has been reported. 159,160,247,248 Nitrous oxide is a strong greenhouse gas 34 and is a signicant source of nitrogen oxides to the stratosphere 249 as originally proposed by Crutzen in his Nobel prize-winning work. 250 Despite the knowledge that N 2 O is generated by the NO 2 heterogeneous reaction with water on surfaces, the mechanisms involved remain obscure.
There are also new sources which have been identied recently, which add to direct emissions of N 2 O. 249,251 For example, N 2 O can be generated from the photolysis of ammonium nitrate (NH 4 NO 3 ) on surfaces at room temperature 252 as well as at low temperatures. 253 The ambient temperature photolysis was estimated to generate 9.3 Gg of N 2 O per year over North America. 252 The Anthropocene not only ushered in greater use of fertilizers such as NH 4 NO 3 , which will contribute to N 2 O through biological processes in soils and photolysis as just described, but it has also given rise to increased use of pesticides. One category of pesticides is the neonicotinoids (NNs) shown in Fig. 9. 254 These NNs came into use starting in 1991 and have largely supplanted the use of organophosphates and carbamates. 255,256 Fig. 10 shows the increasing use of NNs and another insecticide, pronil, as a function of year for several countries and the state of California in the U.S. 256 A signicant concern that has arisen with their use is the impact on pollinators such as bees. There are a number of factors 257 potentially involved in the observed bee colony collapse disorder besides NNs, including parasites and pathogens, loss of habitat and its diversity, and climate change. In any event, restrictions have been placed on NN use in parts of Europe.
Imidacloprid (IMD) is the major NN in use on a worldwide basis, for example as a seed coating. Relatively little is known about its photochemistry and atmospheric reactions on surfaces such as seeds or soil. However, recent work in the author's laboratory has shown that photolysis of IMD generates N 2 O with a yield of $50% relative to the loss of the parent compound. 258 Based on the estimated 20 000 tonnes worldwide production of IMD in 2010, 256 $2 Gg of N 2 O could be produced from photolysis of IMD alone. While this is not signicant compared to the estimated global increase of 20 Tg emitted per year as N 2 O (equivalent to 13 Tg N) from the terrestrial biosphere, 251 it could contribute to measurements of concentrations and uxes made over soils that contain NNs.
The solid phase products identied from IMD photolysis are the urea and desnitro derivatives 258 shown in Fig. 11. Neither of these absorbs in the actinic region 258 so once formed, they will be stable with respect to photolysis in the  troposphere. Previous toxicology studies 259 have shown that the desnitro derivative is more toxic than the parent imidacloprid. As is the case with malathion, 260 this is an example of the importance of considering not only the toxicity of the parent compound, but also that of its products from chemistry and photochemistry in the atmosphere.
In short, activities associated with the Anthropocene have led to increased emissions of oxides of nitrogen and ammonia, as well as the use of newly developed chemicals such as pesticides that can also contribute to oxides of nitrogen chemistry in the atmosphere. While a great deal is known about the gas phase reactions, heterogeneous chemistry is much less understood. New experimental and theoretical approaches would be very helpful in probing the composition, chemistry and photochemistry of these surface reactions.

Reactions at liquid interfaces
While uptake and reactions on solid surfaces in the atmosphere have been recognized for many decadesbut not well understoodthere is now rm evidence that the composition and chemistry at liquid surfaces in the atmosphere is oen different from the bulk, 261-264 with potentially important implications for tropospheric chemistry. [265][266][267][268] In the Anthropocene, the importance of reactions at liquid interfaces can be impacted as the distribution, lifetimes and properties of clouds, fogs and particles that provide an interfacial medium are changing.

Inorganics
An example of unique chemistry at interfaces is an initially surprising result from a combination of experiments, atmospheric modelling and molecular dynamics simulations which suggested that chloride ions are present at interfaces of aqueous salt solutions as models for sea salt particles. 269 This enhanced surface availability of chloride ions resulted in oxidation by incoming gas phase OH that was different in terms of both kinetics and mechanisms from the bulk, generating gas phase Cl 2 much more efficiently than expected from bulk phase chemistry. 269 Subsequent experimental [270][271][272][273][274] and theoretical [275][276][277] work rmly established that larger and more polarizable halide ions are enhanced at the interface relative to the bulk. This means, for example, that bromide ion chemistry at interfaces is relatively more important than chloride ion chemistry, which may contribute in Fig. 11 Structures of the two major products formed on photolysis of a thin film of imidacloprid on a surface. 258 part to the importance of bromine chemistry in the Arctic boundary layer at polar sunrise despite low bromide concentrations. 19,[278][279][280][281][282][283] Iodide ions are even more enhanced at the interface; 275,284,285 while present in much smaller concentrations than chloride or bromide, iodine from various sources including organoiodine compounds can play a signicant role in particle formation in coastal areas. 18,286,287 The relative importance of interface chemistry may be altered in the Anthropocene by changes in OH or by altered sea salt particle and cloud/fog concentrations due to changes in meteorology and wave action. 288 Photochemistry at interfaces may also be enhanced due to a reduced solvent cage, [289][290][291][292][293] and this effect can be altered by the presence of other species. [294][295][296][297][298][299] For example, nitrate ion photolysis at interfaces is more efficient than in the bulk, and is inuenced by the presence of halide ions. [294][295][296][297][298][299] However, it should be noted that whether nitrate ions are enhanced at the interface in water is somewhat controversial, [300][301][302] and could depend on cluster size, for example. 303 Organics Interfacial organic surfactants are well known in laboratory and atmospheric systems. [304][305][306] In the atmosphere, they can alter the exchange between the gas phase and clouds, fogs and particles, 307-313 change the kinetics of reactions such as the hydrolysis of NO 2 (ref. 173) and the oxidation of organics, 268,314,315 and alter ion composition and chemistry at the interface. 316,317 There is also evidence that mechanisms and kinetics of reactions of organics can be altered at the interface compared to the bulk. For example, organic surfactants can form quite densely packed monolayers at the air-water interface. In the event of free radical formation, e.g., by direct photolysis, OH reaction or Htransfer to a triplet photosensitizer, radical-radical reactions can compete with scavenging of the radical by O 2 . [318][319][320][321][322][323] This leads to the formation of oligomers as well as a variety of reactive products. Fig. 12, for example, shows some proposed pathways for the reaction of nonanoic acid initiated by H-atom transfer from the triplet state of the photosensitizer 4-benzoylbenzoic acid, 4-BBA. There are a variety of potential photosensitizers in atmospheric particles, including humic acids and imidazoles formed from the ammonia and amine reactions involved in the "browning" of SOA. Interestingly, there is evidence for HO 2 radical production from the imidazole photosensitized reaction of citric acid. 324 Reactions of organics at interfaces may also have biological relevance; for example, the formation of peptide bonds at the air-water interface in the presence of Cu 2+ ions has been observed. 325 There remain a number of outstanding questions with respect to differences in organic reactions at interfaces under atmospheric conditions where, for example, the surface packing of a complex mixture of organics and the proximity of photosensitizers is ill-dened.

Organic oxidations and particle formation in the Anthropocene
Organics are a ubiquitous component of atmospheric particles (Fig. 5). A large fraction of SOA typically originates in the oxidation of biogenic VOCs (BVOCs), which results in a "modern" carbon 14 C isotope signature. [326][327][328] Biogenic emissions are impacted by anthropogenic activities, for example by drought associated with climate change. 329 However, accurate assessment of biogenic emissions is challenging due to their dependence on a large number of factors such as plant type, temperature, sunlight intensity, soil characteristics, and stressors such as attack by insects. 330,331 Recent modelling efforts that include 147 individual species suggest that a relatively small number of compounds (about a dozen) are responsible for $80% of the BVOC emissions, with isoprene and the monoterpenes dominating. 332

Ozone-alkene reactions
The reactions of biogenic alkenes with the increasing amounts of O 3 (Fig. 4) are particularly important in SOA formation. The rst steps in alkene ozonolysis are well-known, involving the formation of a primary ozonide which decomposes to an aldehyde or ketone and a carbonyl oxide, known as a Criegee intermediate. Criegee intermediates (CI), were rst proposed about 1950 by Rudolph Criegee, 333 and can be formed in the reactions of O 3 not only with C]C, but also with C]N and C]P groups. 334 The reaction is exothermic, and the resulting CI has excess energy through which isomerization (e.g. to a vinyl hydroperoxide, VHP) and decomposition reactions can occur, generating OH radicals 335-337 which then attack the alkene. Depending on the nature of the CI and pressure, collisional deactivation can lead to the formation of stabilized Criegee intermediates (SCI). Reactions of SCI with atmospheric species such as water vapor dimer are believed to be important in air, [338][339][340][341][342][343][344] and theoretical studies 345 suggest that interface reactions of the CI with water could also occur.
However, SCI also react with organics to generate SOA, whose composition is sensitive to the structure of the SCI. For example, in the trans-3-hexene reaction with O 3 , the SOA composition reects sequential addition of Criegee intermediates to RO 2 radicals, 346,347 which is also predicted theoretically for the CH 2 OO Criegee intermediate. 348 However, for larger alkenes of more complex structure such as a-pinene or a-cedrene, 349-354 the chemistry of Criegee intermediates is clearly much more varied, resulting in SOA with very complex composition. 355 A major advance in understanding this chemistry has been the development of techniques to directly detect SCI in the gas phase 356 and to generate them in a relatively clean manner, for example from the reaction of O 2 with a-iodoalkyl radicals CH 2 I and CH 3 CHI to form CH 2 OO and both the syn-and anti-forms of the CH 3 COO Criegee intermediate, respectively. 338,357 While these are sufficiently stabilized to undergo bimolecular reactions, they may still contain some excess energy. 358 The ability to generate SCI has facilitated studies of the spectroscopic properties [359][360][361] and reactions of SCI with many different potential atmospheric species, and there has been an explosion of papers in this area (e.g., see recent reviews 339,340,362 ).
Alkyl peroxy radicals (RO 2 ) are formed in ozone-alkene reactions, either through the reactions of the Criegee intermediate or through the generation of OH radicals which then attack the alkene. There is now experimental evidence that some larger RO 2 radicals can undergo intramolecular isomerization and become increasingly oxidized through an autooxidation mechanism, 363-371 as known for many years to occur in the condensed phase and predicted earlier for the gas phase. 372 Structure plays a key role in determining the importance of this pathway, with endocyclic alkenes particularly exemplifying this chemistry. 368,371 For example, Fig. 13 shows the initial steps in the ozonolysis of 1-methylcyclohexene, forming an RO 2 radical through the VHP channel of the Criegee intermediate; quantum chemical calculations support the occurrence of a 1,6hydrogen transfer with an energy barrier of $21 kcal mol À1 and a rate constant of 0.27 s À1 to form a hydroperoxide and a new alkyl radical. 371 The autooxidation mechanism then continues to form highly oxygenated, extremely low volatility organic compounds (ELVOC). Measurements of highly oxidized RO 2 radicals in the ozonolysis of a-cedrene are consistent with an autooxidation mechanism in that case as well. 373 It is interesting, however, that calculations for intermediates in the ozonolysis of a-pinene suggest analogous mechanisms are not as energetically favorable, despite the presence of a similar methylcyclohexene structure. 371 Alternative mechanisms that still involve autooxidation as initial steps, have been proposed to explain high molecular weight particle phase products, such as the formation and uptake of diacyl peroxides into particles followed by their decomposition. 374

SOA phase
Given the thousands of potential SOA precursors in air and the range of oxidation products they form, 36,375 there are a wide range of structures and volatilities of species that could contribute to SOA. 376,377 As seen in Fig. 14, 378,379 these are oen lumped into bins designated VOC, IVOC, SVOC, LVOC and ELVOC according to their saturation vapor pressures (I ¼ intermediate, S ¼ semi-, L ¼ low, EL ¼ extremely low). Given their very low vapor pressures, ELVOC are believed to be important in the earliest stages of particle formation and growth in air, while somewhat higher volatility compounds contribute more to their subsequent growth. 353,363,366,367,369,374,[380][381][382][383] However, analysis of SOA composition in some eld studies suggests that more volatile products can be found in SOA than expected based on their volatility. 384 This may be related to the particle phase and growth mechanisms.  Understanding the phase of organic particles is important for quantifying exchange with the gas phase, understanding water uptake, describing the chemistry in the bulk and on the surface, and being able to predict the growth and composition of SOA in air. Phase also affects photochemistry in SOA. 385 It had been assumed until relatively recently that SOA particles were oily liquids. Given particle diameters of the order of a few hundred nanometers and typical diffusion constants in liquids, diffusion of gases in and out of the particles should be quite rapid, on the order of ms for 100 nm particles; on the other hand, it can be as much as a year for high viscosity particles (Fig. 15). [386][387][388][389][390][391] In this case, exchange with the gas phase is not sufficiently fast that quasi-equilibrium can be assumed (as has been the case in most atmospheric models) and a kinetically limited, condensation type mechanism must be considered. 27,28 In the latter case, molecules that adsorb on the surface with a sufficient residence time become incorporated into the particle, and once this happens, it is essentially irreversible. Of course, there will be intermediate cases where diffusion is sufficiently slow that the quasi-equilibrium assumption is not valid, yet exchange with the gas phase can occur on similar timescales as removal of particles from the atmosphere.
For relatively simple systems, viscosity (h, Pa s) and the diffusion coefficient (D, cm 2 s À1 ) are inversely related through the Stokes-Einstein (S-E) equation developed to describe the diffusion of large spherical molecules through a continuum of solvent that provides frictional resistance. 392 Although the S-E equation provides a rst-order approach to the relationship between viscosity and diffusion coefficients, its assumptions may not always be met for diffusion of species in SOA. For example, diffusion of water in sucrose solutions of well-dened viscosity 393,394 and in SOA 395 can differ from the S-E relationship by orders of magnitude under some conditions. A number of interesting experimental approaches to determining viscosity of SOA have been developed and applied recently. [396][397][398][399][400] One of the rst indications that some SOA may not be liquid came from studies of particle bounce in impactors, 401-404 a phenomenon recognized for many decades during atmospheric sampling. [405][406][407][408][409][410][411][412][413][414] Fig. 16, for example, shows impaction patterns on a germanium attenuated total reectance (ATR) crystal for dry and wet inorganic salts as well as for SOA formed from a-pinene ozonolysis. 404 Liquid sodium sulfate (Na 2 SO 4 ) particles form a series of "spots" immediately below the holes in the impactor, as expected for liquids which do not bounce, while dry solid Na 2 SO 4 or ammonium sulfate ((NH 4 ) 2 SO 4 ) form "clouds" due to bounce either along or perpendicular to the air ow. When grease is applied to the crystal, the bounce stops (Fig. 16b). The pattern for SOA is similar to that of the dry salts, indicating it does not behave like a liquid.
Since the initial discovery of semi-solid SOA particles, there have been many different studies and approaches to understanding the phase and related properties of these particles under different conditions of formation and of subsequent changes in exposure (e.g., to water vapor). 404,[414][415][416][417][418][419][420][421] Li et al. 416 summarized published studies of viscosity, diffusion coefficients and related phenomena such as evaporation of components from SOA, as well as chemical reactivity. In brief, there is not yet a coherent picture that provides a basis for predicting phase, viscosity and effects on reactivity etc. over a broad range of conditions. Variables include the nature and concentrations of the SOA precursors and the experimental conditions, e.g., temperature, whether water is present during or aer SOA formation, the sampling collection and handling procedure, whether an OH scavenger is present in ozone-alkene reactions, the presence or absence of NO x , etc.
Despite the breadth of approaches, systems and experimental conditions studied to date, there are some common trends. First, viscosity decreases and diffusivity increases when RH is increased either during SOA formation or if SOA Fig. 16 Impaction patterns from the impaction of particles of liquid Na 2 SO 4 , dry Na 2 SO 4 , dry (NH 4 ) 2 SO 4 and dry SOA from a-pinene ozonolysis. The ones marked "greased" have vacuum grease applied to the right half of the surface to prevent particle bounce; the impacted particles can be seen embedded in the center of the grease. The mechanisms leading to the patterns are shown schematically on the left. Adapted from ref. 404. is formed under dry conditions and is subsequently exposed to increasing water vapor concentrations. Second, diffusivity depends on the nature of the diffusing molecule as expected, with water diffusing faster (D $ 10 À9 cm 2 s À1 at 280 K) 395 than larger compounds like pyrene (D ¼ 2.5 Â 10 À17 cm 2 s À1 ), 415 or pinonaldehyde and acetic acid (D $ 3 Â 10 À14 cm 2 s À1 within a large, order of magnitude, error bar). 382 Third, SOA with smaller O : C ratios and those formed from larger precursors tend to remain semi-solid at higher RH. 416 Fig . 17 demonstrates the importance of having a full understanding of the phase of particles. These studies attempted to model particle formation and growth (so-called "banana plots") in several different locations assuming that the particles were liquid and that quasi-equilibrium with the gas phase applied, or that some fraction was solid, for which a condensation mechanism applied. The best t to the experimental observations required that at least 50% proceeded via a condensation mechanism characteristic of highly viscous SOA. 422

SOA structure and composition
Associated with the composition and phase of SOA is the issue of the 3-D structure of SOA particles. If particles are liquid, they should be well-mixed, while if they are semi-solids, this might not be the case. One illustrative example (albeit not directly atmospherically relevant), is the formation of organic particles on a surface from the ozonolysis of a terminal alkene covalently bound to a surface in the form of a self-assembled monolayer. 423 A free radical reaction occurred that detached the chain from the surface and formed large organic aggregates that were detected by atomic force microscopy (AFM), scanning electron microscopy (SEM) and Auger electron spectroscopy. Single particle FTIR indicated that the Fig. 17 Measured (a) and predicted particle formation and growth on April 15, 2007 in Hyytiälä, Finland using the assumption of equilibrium partitioning, (b) 100% equilibrium partitioning; (c) 50% equilibrium partitioning and 50% condensation mechanism, and (d) 100% condensation mechanism. 422 The more red the colors, the larger the particle number (N) concentration expressed as dN/dlog D p where D p is the particle diameter. Adapted from ref. 422. aggregates were highly oxidized (Fig. 18a). However, depth proling using secondary ion mass spectrometry (nanoSIMS) showed that the O : C ratio was not uniform throughout, but rather increased from the surface into the interior (Fig. 18b). This suggests that during formation the aggregates self-assembled with oxygenated polar groups buried inside a hydrophobic shell (Fig. 18c). In this case, despite what is presumably a high oxygen content on average, one would not expect the aggregates to be efficient in taking up water. If this mechanism holds for some particles, they may not have the ability to serve as cloud condensation nuclei (CCN) and a correlation between water uptake and CCN activity with degree of oxidation or O : C ratio would not necessarily be expected. This is, in fact, observed in some cases. 424,425 However, this is clearly an area that requires much more in-depth, molecular level understanding in order to dene the implications for ambient air, particularly in a changing climate.
While aqueous phase chemistry of SO 2 and NO x is reasonably well-established, 27 less is known about the chemistry and photochemistry of organic compounds in the condensed phase. Atmospheric particles may be organic, aqueous, homogeneous or heterogeneous. For example, when particles contain signicant amounts of some organics as well as an aqueous phase with dissolved electrolytes, liquid-liquid phase separation [426][427][428][429][430][431] can occur, providing both organic and aqueous phases within one particle. The aqueous phase may be in the Fig. 18 (a) FTIR of two typical particles observed after ozonolysis of a C ¼ 8 terminal alkene self-assembled monolayer on a silicon substrate. The spectra use the unreacted SAM as the background to show only changes due to reaction. The peaks in the 3300 cm À1 and 1700 cm À1 regions reflect the formation of oxidized groups such as -COOH and -C] O; (b) nanoSIMS of a particle from this reaction. A cartoon is shown in (c) that depicts two possibilities, the one on the left where the polar functional groups which are clearly evident in the FTIR spectrum in (a) are buried inside a hydrophobic shell, and the one on the right with polar groups on the surface. The left graphic is consistent with the O : C ratio increasing as the particle is probed from the surface inwards as seen in the nanoSIMS data in (b). Note that the first high O : C at the particle surface is an artifact from adsorbed water. Adapted from ref. 423. form of clouds and fogs, or aerosol particles. These media differ in liquid water content, from $10 À6 to 0.1 cm 3 liquid water per m 3 of air from particles to clouds/ fogs, and in concentrations of dissolved species as reected, for example, in the ionic strength which can vary from 10 M or more to 10 À4 M from particles to clouds and fogs. 432 There is increasing evidence for rich chemistry and photochemistry of organic compounds in these condensed phases. 36,94,306,385,[432][433][434][435][436][437][438][439][440][441][442][443][444][445][446][447][448][449][450] Detailed treatment is beyond the scope of this paper, but there are a number of excellent discussions of this area 36,306,432,[434][435][436][437][438][439][440] since it was highlighted by Blando and Turpin. 433 Thermal reactions include accretion reactions such as aldol condensation reactions, acetal and hemiacetal formation and the formation of esters and oligomers, as well as reactions with non-photochemically derived species such as ozone and amines. For example, reactions of ammonia and amines, both associated in part with agricultural activities, 16,451,452 with components of SOA form nitrogen-containing organics such as imidazoles that absorb light in the visible region, dubbed "brown carbon". 94,444,446,447,[453][454][455][456][457][458][459][460][461][462][463][464][465] These compounds affect the optical properties of the particles and have also been proposed to act as photosensitizers to form more SOA. 466 Thermal reactions involving uptake of carbonyl compounds such as glyoxal, followed by hydration and then reaction with sulfate ions are known to generate organosulfates. 375,467 Reactions of epoxides, for example from isoprene oxidation in the atmosphere, 468 also generate signicant amounts of organosulfates via ring-opening reactions with sulfuric acid/ sulfate. 469,470 Photochemical reactions in the condensed phase include direct absorption and photochemistry by organics, photosensitized reactions induced by interactions with excited triplet states (e.g. ref. 306, 471 and 472) and reactions with species generated by photochemistry such as the OH radical and likely singlet oxygen, O 2 ( 1 D g ) as well.
Oxidation in the condensed phase has been increasingly recognized to be important not only for determining the chemical and physical properties of SOA but also potentially for their health effects, where oxidants can initiate inammatory responses. The term reactive oxygen species (ROS) is dened in practice as any species that oxidizes a selected non-uorescing organic to an oxidized, uorescing form. It includes free radicals such as OH, HO 2 and organic free radicals, superoxide anion, peroxynitrites and H 2 O 2 as well as organic peroxides. 473 Uptake of some of these from the gas phase is one source of the non-ionic species, and photochemical reactions involving transition metals such as iron is another. Reactive oxygen species have also been reported to be generated in SOA formed from oxidation of small carbonyls such as methyglyoxal 474 and from biogenic VOCs in solution 475 as well as from the interaction of SOA with water. 476 Understanding the role of ROS in health impacts is clearly and area of great interest. 473,[477][478][479][480][481][482][483][484][485] In short, much remains to be learned about the chemistry and photochemistry of organics in the condensed phase under atmospheric conditions. This is especially challenging due to the complex mixtures involved, the presence of both organic and aqueous phases with a range of viscosities, and the wide range of concentrations that are found. Advances in both experimental 486 and theoretical approaches 487,488 have proven invaluable and their continuing integration is important for advancing our understanding of multiphase processes in the atmosphere and their impacts on human health, visibility, climate and weather.

Sulfur compounds in the Anthropocene
Sulfur dioxide from fossil fuel combustion is oxidized in air to sulfuric acid through a combination of gas and aqueous phase processes, with some contribution from heterogeneous reactions on surfaces. 27 Sulfuric acid has been credited as the major source of new particle formation in air for many decades. 489 For example, small clusters containing up to four sulfuric acid molecules have been observed to be correlated to nucleation events. 490 However, the observed rate of new particle formation in air from water and sulfuric acid 491 is typically many orders of magnitude larger than predicted by classical nucleation theory. 489,492,493 A key discovery has been that amines, as well as ammonia, play a central role in new particle formation. 489,491,[494][495][496][497][498][499][500][501][502][503][504][505][506] There are many sources of these bases, 451,452 including livestock operations, 507 food processing and cooking, composting, sewage treatment, combustion and tobacco smoke. 451 One source of amines that may increase in the future is carbon capture and storage that uses amines to capture CO 2 . [508][509][510][511][512] Although much less is known about the role of organics in new particle formation, there is increasing evidence that under many atmospheric conditions they may play a key role in stabilizing and growing small sulfuric acid-amine clusters (Fig. 19) to detectable particle sizes 378,513-515 and ultimately to $100 nm where particles scatter light effectively and also serve as CCN. 516,517 In addition, there are direct interactions of some organics with sulfuric acid to form/grow particles. [518][519][520][521][522][523] There have been signicant reductions in anthropogenic SO 2 emissions since the 1970's (Fig. 20). 524 While this is expected to have a substantial inuence on particle formation in air, SO 2 will continue to be formed from oxidation of organosulfur compounds 525,526 such as dimethyl sulde (DMS). These compounds have a wide variety of biogenic oceanic and terrestrial 527-533 as well as anthropogenic sources including agricultural activities, landlls and municipal waste 95,534-544 and even human breath. 545,546 Thus, even in the absence of the combustion of sulfur in fossil fuels, some SO 2 , and hence sulfuric acid and sulfate particles, will be formed. Fig. 19 Schematic of the role of organics in new particle formation and growth from sulfuric acid and amines. Adapted from ref. 513. Another product of the oxidation of organosulfur compounds is methanesulfonic acid, CH 3 S(O)(O)OH (MSA). While initial studies suggested that MSA would not contribute to new particle formation in air due to its higher volatility compared to sulfuric acid, 547,548 subsequent studies have shown that like H 2 SO 4 , MSA does form particles in the presence of amines and water vapor. [549][550][551][552] However, the mechanism of MSA formation has not been entirely clear. As shown for DMS oxidation in Fig. 21, MSA can be formed via the generation of the free radical CH 3 S(O)(O)Oc, followed by hydrogen abstraction. While abstraction from organics had been suggested, there was little direct experimental evidence for this. This is again where quantum calculations prove very helpful. 553 As seen in Fig. 22, reaction of CH 3 S(O)(O)O with HCHO is predicted to proceed via a submerged transition state in a barrierless reaction. While the radical reaction with CH 4 does have a barrier, other hydrocarbons such as those with weaker    553 However, note that in these studies NO x emissions were kept constant and as seen in Fig. 21, NO x plays a role in forming MSA, something which has been observed in laboratory studies. 525 Supporting the potential role of NO x in the mechanism are observations in some eld campaigns that MSA is correlated with NO x . 554

Biogenic-anthropogenic synergies in the Anthropocene
There are increasing numbers of examples of synergistic interactions between anthropogenic and natural emissions. These include the effect of NO x on the formation of MSA during the oxidation of DMS, and the role of biogenic oxidations by anthropogenically derived oxidants in SOA formation discussed above. A third example is the interaction of oxides of nitrogen with sea salt particles. Dinitrogen pentoxide (N 2 O 5 ) formed in the gas phase from the reaction of the NO 3 radical with NO 2 (ref. 27) undergoes rapid hydrolysis on surfaces and in particles to form HNO 3 . [555][556][557][558][559] This reaction has been shown in both laboratory and eld studies to be impeded by the presence of nitrate ions. 555,560 Organic coatings also appear to decrease N 2 O 5 uptake on particles, although this may be sensitive to the relative amounts of water present (and hence relative humidity) as well as the viscosity of the organic coating. 312,313,556,[561][562][563][564][565][566][567] If chloride ions are present, there is a competing reaction to form nitryl chloride, ClNO 2 . [568][569][570][571][572][573] It is interesting that the chloride ion reaction occurs at the interface, while that with water occurs deeper into the bulk of the particle. 574 Field campaigns have documented the formation of ClNO 2 in coastal urban regions and other areas inuenced by both sea salt particles and oxides of nitrogen, [575][576][577][578][579][580][581] with ClNO 2 correlated to gas phase N 2 O 5 as predicted from laboratory studies. A surprising observation is the formation of ClNO 2 at inland continental locations removed from sources of sea salt (Fig. 23). [582][583][584] This has usually been attributed to the reaction of N 2 O 5 with particulate chloride, with the source of the latter possibly being uptake of HCl from the gas phase. Another potential source in cold climates is road salt.
The possible mechanism harks back to reactions of oxides of nitrogen on surfaces discussed earlier. For example, exposing silica surfaces to N 2 O 5 and gas phase HCl generates ClNO 2 , 585 which was proposed to occur via ionic forms of oxides of nitrogen (and possibly the HCl) stabilized on surfaces:  Indeed, NO 2 + was observed on an ATR ZnSe crystal by infrared spectroscopy, and as seen in Fig. 24a, the peak disappeared on exposure to gas phase HCl. 585 The yields of ClNO 2 increased with water vapor concentration up to the predicted maxima (DClNO 2 /DN 2 O 5 ¼ 1), which was surprising since water was expected to compete with the HCl reaction, forming HNO 3 instead: However, theory 171,585-588 predicts that even one water molecule promotes the autoionization of N 2 O 5 to NO 2 + , and the subsequent reaction with HCl is predicted to be barrierless in the presence of two water molecules (Fig. 24b). 585 In addition, ab initio molecular dynamics simulations show that for water lms on surfaces, N 2 O 5 can be stabilized at the air-water interface in such a way that reaction with Cl À via an S N 2 reaction can occur. 588 Baergen et al. 242 proposed that similar chemistry could occur in "urban grime" which contains both oxides of nitrogen and chloride ions, as well as a complex mixture of organics. [589][590][591][592][593][594][595] A fourth example involves SOA formation in areas that have large biogenic emissions. It has been noted that SOA concentrations in air are signicantly larger in the southeastern U.S. than in the Amazon, despite the much larger sources and concentrations of biogenic precursors in the latter location. One mechanism suggested for SOA production is the reaction of smaller, soluble organics in the aqueous phase of particles to generate larger, low volatility products that remain in the condensed phase when the water evaporates. 306,432,433,435,437,438,[596][597][598][599][600] In this case, SOA will be impacted by the availability of water in particles 601 and the composition of the aqueous phase, especially the pH, which is largely determined by the presence of hygroscopic anthropogenically derived species such as sulfate and nitrate. [601][602][603] This is an example of how the interplay between biogenic and anthropogenic species can introduce new, unexpected chemical mechanisms that would occur only in the Anthropocene era.

Summary
Research over the last decades has led to a vastly increased understanding of anthropogenic emissions, their chemistry, their interactions with natural systems and the ultimate impacts of these processes. The spatial and temporal scales involved in different aspects of the problem are very challenging, and will require additional new approaches for laboratory and theoretical studies as well as eld measurements and modelling. Although not addressed here in detail, there is a need to develop analytical techniques for a number of different purposes such as probing the molecular composition of particles layer-by-layer, elucidating the nature of surface-bound oxides of nitrogen, and measuring both gases and particles at scales appropriate to the problem at hand. For example, while relatively long-lived gases such as CH 4 do not generally need to be measured on small spatial scales, species that impact human health need data ideally at or close to the point of inhalation. The development of small, low-cost, robust analytical techniques that can be reliably calibrated as well as the widespread use of instrument platforms such as unmanned aerial vehicles (UAVs, "drones") will be needed to address these problems.
Clearly a number of research issues remain that need to be addressed in order to fully understand atmospheric chemistry in the Anthropocene, which in turn will provide a quantitative foundation for control strategies that minimize impacts on human health and welfare.