Review of the bulk and surface chemistry of iron in atmospherically relevant systems containing humic-like substances

As the fourth most abundant element by mass in the Earth's crust, iron is ubiquitous and its chemistry is rich and interdisciplinary in nature. This review synthesizes the current state of knowledge of iron chemistry in multicomponent atmospheric aerosols. This knowledge is also applicable to other atmospherically relevant systems that include iron-containing anthropogenic nanodust, ocean surfaces and buildings. Because of the abundance of humic-like substances in these systems, this review focuses on the chemistry of these substances with iron compounds. Findings from field measurements and laboratory studies are summarized to highlight the major themes in the chemical reactivity of iron, which varies depending on the solubility, the redox conditions, the absence and presence of UV-visible light and reactive oxygen species, the pH and the temperature. This review also highlights the key differences between the bulk and surface chemistry of iron-containing materials, which varies considerably because of the structure of the interfacial water and the solvent cage effect. Additional laboratory, field and modelling studies are needed to better understand the contributions of transition metal chemistry to the formation of secondary organic aerosols and also the chemistry, uptake and release of trace gas phase species. This information will improve the predictive power of models that incorporate aerosol chemistry and physics.


Introduction
3][4][5] The Fih Assessment Report of the Intergovernmental Panel on Climate Change showed that the representation of aerosols in climate models is still inadequate as a result of their highly complex physicochemical properties that change over time. 6This large uncertainty associated with aerosols is a result of the relatively low level of scientic understanding of their indirect effects on climate, i.e. their role in acting as condensation nuclei for clouds and ice. 73][14] The uptake of water by aerosols is controlled by their chemical composition and interfacial properties and is sensitive to changes in meteorological parameters such as the relative humidity (RH). 8,15[18][19] Hind A. Al-Abadleh is currently an Associate Professor of physical chemistry at Wilfrid Laurier University.Her research interests include heterogeneous photochemistry in atmospheric aerosols driven by transition metals and the surface chemistry of arsenic and phosphorus compounds in model soil systems.She has received an Early Researcher Award from the Ontario Ministry of Research and Innovation and the Petro-Canada Young Innovator Award, among other awards.Al-Abadleh completed her BSc (Hons) degree in chemistry at the United Arab Emirates University (1999), followed by a PhD at the University of Iowa with Professor Vicki Grassian  (2003) and postdoctoral training with Professor Franz Geiger at Northwestern University (2005).
One important class of organic matter in atmospheric aerosols is humic-like substances (HULIS). 20,21As a result of their complex chemical nature, HULIS in aerosols have been studied less than the other classes of organic compounds identied in atmospheric aerosols.As noted by Zetzsch and co-workers, 22,23 the lack of information about the true composition of HULIS is underestimated.Atmospheric HULIS are emitted from primary sources such as wind-blown marine sediments, soils and the burning of biomass, or are formed in the atmosphere through condensation and polymerization. 20,21,24They comprise 15-60% of the watersoluble organic carbon (WSOC) in aerosols.Duarte et al. 24 found from infrared (IR) and solid-state 13 C-NMR spectroscopy that atmospheric HULIS contain conjugated carbonyl groups, ethers, aromatic phenols, carboxyl groups, alcohols, oxygenated aliphatic carbon, branched alkyl chains and a high level of aromatic and aliphatic content.As a result, most laboratory studies so far have used surrogates for atmospheric HULIS, such as humic and fulvic acids from terrestrial and aquatic sources, 25,26 tannic acid (1,2,3,4,6-pentagalloyl-O-glucose, C 76 H 52 O 46 ), [26][27][28] gallic acid (3,4,5-trihydroxybenzoic acid), 29 catechol (1,2dihydroxybenzene), 23,30,31 guaiacol (2-methoxyphenol) 23,32 and shikimic acid. 33The latter three compounds are semi-volatile phenolic compounds emitted from the burning of biomass and are known aromatic SOA precursors. 23,33he hygroscopic growth of these model compounds has been studied using a number of techniques, including electrodynamic balance and hygroscopic tandem differential mobility analysis in both their pure form and when mixed with inorganic salts, 34,35 and insoluble mineral aerosol, 35 IR spectroscopy 36 and X-ray techniques. 26These studies found that surrogates for HULIS undergo growth and evaporation as a function of the RH, which changes their size.These model systems are also efficient cloud condensation nuclei, can lower or enhance water uptake when mixed with soluble salts 34 and can enhance the adsorption of water when coating insoluble calcite particles. 35Experiments on the hygroscopic properties of HULIS in aerosols showed continuous water uptake as a function of the RH, with the exception of fulvic acid, which showed phase separation. 26,34,37,38he structure of water on some of these systems resembled that of water at the interface with polar organic solvents. 36The amount of water uptake was found to vary depending on the carbon functional groups in the fulvic acid samples aer phase separation. 26This means that, depending on the size and water content of the particles, preferential partitioning of HULIS to the surface could become important in their overall surface reactivity.Although bulk photochemical reactions involving humic substances in atmospheric aerosols and aquatic systems have received much attention, 13,14,[39][40][41] fewer studies have reported the heterogeneous photochemistry of atmospheric HULIS.Examples include: (a) the fast photosensitized formation of HONO from the photoreaction of NO 2 with humic, 42,43 tannic and gentisic acid lms 28 under dry and humid conditions; (b) the photooxidation, formation and characterization of secondary particles from catechol and guaiacol; 23 and (c) changes to the functional groups characteristic of solid tannic acid and the formation of new carbonyl groups characteristic of aryl aldehydes and/or quinone under humid conditions. 368][49][50][51] This latter pathway could be driven by the electronically excited states of some organic chromophores such as HULIS or by soluble and insoluble transition metals such as iron. 14n addition to inuencing the oxidative power of the atmosphere and the global sulfate budget, 41,52 particulate iron can catalyse reactions that lead to oxidative stress in living cells as a result of the production of reactive oxygen species (ROS). 53Dark and photochemical processes that lead to the release of soluble iron into the aqueous phase have consequences for the availability of this essential element for phytoplankton productivity, the extent of algal blooms and the uptake of atmospheric CO 2 . 54,55he objective of this review is to synthesize the current state of knowledge on the role of iron in the chemical aging of atmospheric aerosols containing organic matter, particularly those that model HULIS.The review is organized into ve main sections and starts with a synthesis of recent results from eld measurements and modelling studies, followed by a summary of the highlights of iron chemistry under dark conditions and the photochemical reactions driven by iron obtained from bulk and surface-sensitive measurements.Because of the molecular level differences between bulk and surface water, a concise summary of the literature on the latter topic is also provided.The review concludes with a summary and directions for future research.

Sources and chemical characterization
The source of the iron added to ocean surfaces is mainly mineral dust, 2,56,57 with contributions from anthropogenic emissions 58 and the burning of biomass. 59,60Aer the deposition of dust, iron undergoes dissolution and complexation with organic matter.2][63] Other sources of the dissolved iron added to the oceans include hydrothermal vents and reductive and non-reductive release from oceanic sediments.Conway and John 57 analysed samples along a section of the North Atlantic Ocean to determine the source-sensitive dissolved stable iron isotope ratios (d 56 Fe) and the iron concentration [Fe].The results showed that 71-87% of dissolved iron in the North Atlantic Ocean originated from Saharan dust aerosols, 10-19% from the non-reductive release of iron from oxygenated sediments, 1-4% from the reductive dissolution of sediments and 2-6% from hydrothermal venting.
Iron is a limiting nutrient for phytoplankton in about 40% of the ocean. 64,65The estimated annual amount of total dissolved iron deposited on ocean surfaces from mineral dust is 24 Â 10 9 mol per year, 66 which is unevenly distributed and tends to be concentrated in the tropical Atlantic Ocean, the Indian Ocean, the Mediterranean Sea and the Arctic. 67In a related study, von der Heyden et al. 68 analysed the size, iron oxidation state, and the composition and degree of crystallinity of iron-containing particles in the waters of the Southern Ocean euphotic zone from South Africa to Antarctica.High-resolution images and synchrotron-based X-ray spectra showed that the particles ranged in size from 20 to 700 nm, with a variety of phases that included Fe(III) and Fe(II) (oxyhydr)oxides, in addition to mixed phases of both elements.The relative concentration of each iron-rich particle phase varied depending on the sample location.For example, the samples closest to the African continent showed the largest degree of chemical heterogeneity and consisted mostly of Fe(III)-rich particles.This was in contrast with the samples from south of the southern boundary, where the samples were mostly rich in Fe(II).The ratio Al : Fe was also found to vary with the depth and the distance from land sources to the open ocean.References to aluminium are made as this element is considered to be a solubility modier of Fe(III) minerals and an indicator of the source of particles.High Al : Fe ratios (0-0.47) are usually found in weathered minerals inland as a result of the substitution of aluminium with iron and are less soluble than the unsubstituted minerals.In this study, 68 higher Al : Fe values were found in deeper waters than in surface waters (up to 0.17) and nearer to the land (up to 0.2); these ratios decreased towards the open ocean.The study also compared the average summer chlorophyll a concentration with the abundance and distribution of labile forms of iron, reecting the effect of iron speciation on biological systems.
][71][72][73][74][75][76][77][78][79][80] Schroth et al. 63 demonstrated that iron speciation (the oxidation state and bonding environment) varies with the source of the aerosol.For example, soils in arid regions are dominated by Fe(III) (oxyhydr)oxides, glacially weathered particles by Fe(II) silicates, and oil y ash from fossil fuel combustion by Fe(III) sulfates.As a result, variations in iron solubility were observed among these aerosol types, which increased in this order: arid soils (<1% of iron was soluble), glacial products (2-3% iron soluble) and oil y ash (77-81% of iron soluble).Fig. 1 shows the iron signature associated with biological material in a representative positive ion mass spectrum of individual sea spray particles analysed by aerosol time-of-ight mass spectrometry (ATOFMS). 76When the pH of the droplet was acidic, the organic material was concentrated on the surface as an outer layer. 81The chemical and physical properties of marine primary organic aerosols and their impact on the Earth's climate have been reviewed elsewhere. 82][85] The speciation of iron has been determined from singleparticle analysis in a number of urban and rural sites in the USA, including Michigan, 74 Georgia, 69 Ohio 75 and Los Angeles. 77sing a particle-into-liquid sample coupled with a liquid waveguide capillary cell and UV-visible spectrophotometry, Oakes et al. 74 measured ne particle (PM2.5)water-soluble ferrous iron [WS_Fe(II)] in Dearborn, Michigan and Atlanta, Georgia.The concentrations ranged from 4.6 to 400 ng m À3 , the highest concentrations of which were found to be associated with sulfate plumes and those with highest apparent aerosol acidity.The temporal trends in WS_Fe(II) were found to be linked with industrial emissions or atmospheric processing of these emissions leading to the formation of WS_Fe(II).In another study by the same group, 69 X-ray absorption near-edge structure spectroscopy and microscopic X-ray uorescence were used to measure the mineralogy and oxidation state of iron in single particles collected from urban and rural sites in Georgia.These measurements were complemented by experiments to determine the fractional solubility of iron.The X-ray absorption near-edge structure spectroscopy measurements showed that iron was present as a mixture of Fe(II) and Fe(III), with the majority of the particles (74%) characterized as Al-substituted Fe oxides followed by Fe aluminosilicates (12%).The spatial distribution of iron in coarse-mode particles was studied using X-ray elemental mapping and computercontrolled scanning electron microscopy. 75Fig. 2 shows the scanning electron microscopy images and energy-dispersive X-ray analysis elemental maps of representative ironcontaining particles from Cleveland, Ohio, which were classi-ed as y ash, mineral dust, NaCl-containing and Ca-Scontaining agglomerates (top to bottom).The concentration of iron-containing particles peaked in the 3-6 mm diameter range to about 5 mg m À3 .Fly ash particles are characteristic of industrial emissions (e.g.steel production) 86 and their concentration was found to decrease strongly with the distance from the source emissions. 75As noted by the authors, 75 the Cleveland study highlights the importance of accounting for anthropogenic sources of iron-containing particles to explain soluble iron (Fe S ) concentrations that exceed estimates based solely on mineral dust aerosols. 87sing single-particle aerosol time-of-ight mass spectrometry, the dynamic size and chemical composition of aerosols were measured near Los Angeles and in the Long Beach Port region. 77Particles in the 0.1-1 mm size range showed the characteristic signatures of soot, i.e. the transition metals iron, vanadium and nickel associated with sulfate and nitrate.The sources of these particles were attributed to primary emissions from oil combustion in ships, reneries and traffic in the port region, in addition to secondary processing during transport.The particle concentration ranged from 14 to 44 mg m À3 over the collection time.As stated by the authors, 77 the signicance of this study lies in the fact it highlighted that the aforementioned primary sources need to be regulated to improve the air quality in California, as for car and truck emissions.Deboudt et al. 79 analysed thousands of individual particles collected during a campaign off the Atlantic coast of West Africa during the African monsoon, where the sampling sites were located on the path of the Saharan dust plumes.The study aimed to analyse the particle size, morphology, chemical composition and mixing states using scanning electron microscopy-energy-dispersive X-ray spectrometry, transmission electron microscopy-electron energy loss spectroscopy and Raman microspectrometry.The three classes of aerosolsmineral dust, carbonaceous compounds and marine compoundswere predominantly externally mixed.About 10.5-46.5% of the analysed particles were found to be internally mixed binary systems, i.e. dust/carbonaceous, carbonaceous/ marine and dust/marine.The mineral content included iron oxides, aluminosilicates and Mg aluminosilicates, calcium carbonate, gypsum, titanium oxide and silica.The carbonaceous content originated from the burning of biomass (tar balls) and anthropogenic (soot) emissions and was found to be amorphous in structure rather than crystallized carbon.The results also revealed that marine and carbonaceous compounds generally formed a coating of mineral dust particles.
Moreover, because the oxidation state and morphology of iron-containing aerosols can modify the chemical reactivity, Takahama et al. 80 performed a comprehensive analysis of 63 iron-containing particles from ve eld campaigns using nearedge X-ray absorption ne structure L-edge spectroscopy coupled with scanning transmission X-ray microscopy.Data generated from this method detected heterogeneities in the distribution of iron and the redox state over individual particles.These particles ranged in size from 0.2 to 4.5 mm with ratios of  Fe(II) to total iron (Fe T ) from 0 to 0.73.The X-ray images showed many different morphologies; in some cases there was a clear barrier between the iron and the carbonaceous regions.The variability in the Fe(II) fraction was analysed as a function of the distance from the surface.The results showed higher Fe(II) fractions near the surface than towards the interior, which was consistent with the surface reduction mechanisms for iron.
The complexation of iron and other transition metals such as copper and manganese to organic ligands in size-segregated ambient aerosol particles was investigated by Scheinhardt et al. 88 Samples were collected from nine sites in Germany that covered urban, rural and coastal areas.Coupled with aerosol thermodynamic modelling results, the analysis showed that pH, [Ca 2+ ], [Mg 2+ ] and the formation of insoluble Ca oxalate governed the availability of oxalate as the main strong organic ligand that preferentially bound Fe(III).Other factors, such as the season, the origin of the air mass, the sampling site and the particle size, were also found to affect the complexation of transition metals to organic ligands.In summary, the relatively high loading of atmospheric iron-containing dust and the processing pathways with and without organic ligands demand adequate representation in atmospheric models; this representation starts with the mineralogy of the particles.

Modelling the cycling of iron
Nickovic et al. 89 developed a high-resolution dataset (GMINER30) of the mineral composition of potentially dust-producing soils that occupy the majority of the arid regions on a global scale.Fig. 3 shows the global distribution of the effective percentage of minerals in soils around the world.The processes that lead to the generation of atmospheric dust particles by wind action from soils have been reviewed elsewhere and include suspension, creep, saltation and sandblasting. 90In another study by Nickovic et al., 91 a regional atmospheric dust-iron model (DREAM) was developed to numerically simulate the atmospheric route of iron from deserts to sinks in the ocean.This model included the parameterization of the transformation of iron to a soluble form caused by the mineralogy of the dust, cloud processes and solar radiation.The modelling results were compared with observations collected from several Atlantic Ocean cruise routes dominated by dust aerosols.This model underestimated Fe T and Fe S for reasons that included small iron emissions at the source origins and neglecting the inuence of anthropogenic sources and the burning of biomass. 91Using the aerosol chemistry version of Integrated Massively Parallel Atmospheric Chemical Transport (IMPACT), which includes three classes of ironcontaining minerals, Ito and Xu 92 investigated the deposition of lterable (i.e.soluble) Fe S and its response to changes in the anthropogenic emissions of combustion aerosols and precursor gases.They found that the scavenging efficiency of dust particles depended on the surface coating of these aerosols by sulfate, nitrate and ammonium species and that the release of iron from minerals was a function of the acidity of the aerosol.Although this release process could happen readily under highly acidic conditions (pH < 2), the results showed that iron released slowly at higher pH values during long-range transport appeared to be important.This model also projected a decrease in the deposition of lterable iron to the Western North Pacic from ironcontaining mineral dust as a result of less acidication in dust from Asia as a result of air quality improvements, which have reduced the emission of nitrogen oxides (NO x ).In another modelling study, Ito 60 aimed to estimate the atmospheric sources of bioavailable iron; an explicit scheme for the dissolution of iron in combustion aerosols as a result of photochemical reactions with inorganic and organic acids in solution was implemented in an atmospheric chemistry transport model.The results showed that 40% of Fe T over major portions of the open ocean in the Southern Hemisphere originated from the deposition of Fe S from combustion sources.
Sholkovitz et al. 93 provided a synthesis of a global-and regional-scale dataset on the fractional solubility of aerosol iron (%Fe S ) relative to Fe T from over 1000 samples collected from a number of sites, including open and coastal ocean sites and some continental sites.The trend that emerged was described by a simple two-component mixing model, where %Fe S reected the mixing of mineral dust (high Fe T and low %Fe S ) and aerosols from combustion sources (low Fe T and high %Fe S ). 93ore sophisticated global chemical transport models were used for the same purpose, such as the GEOS-Chem 94 and IMPACT 95 models.For example, the former model was used to analyse the magnitude and spatial distribution of mineral dust and Fe S deposition to the South Atlantic Ocean. 94The model predicted a <1% dissolved iron fraction of mineral dust over the South Atlantic Ocean as a result of the low ambient concentrations of acidic trace gases available for mixing with dust plumes.Sensitivity studies showed that the initial amount of Fe S in the dust source regions to a large extent controls the amount of Fe S deposited to the South Atlantic Ocean. 94In another study, the effect of aerosol emissions from ship plumes on the solubility of iron in particles from combustion sources was modelled over the high-latitude North Atlantic and North Pacic Oceans using IMPACT. 95The results showed that combustion particles with a low iron loading (1-110 ng m À3 ) contributed more than 10% of the Fe S .This model predicted that, in the year 2100, Fe S from ships could contribute 30-60% of the total Fe S deposition.Before deposition and while suspended in the atmosphere, Guo et al. 96 found from eld measurements in Hong Kong and modelling results that aqueous phase reactions involving transition metal ions such as iron, copper and manganese in deliquescent aerosols signicantly affected the mixing ratios of H 2 O 2 in the gas phase by acting as net sources or sinks depending on the metal content of the aerosols.In summary, these eld studies were aimed at collecting, analysing and tracking the fate of iron in atmospheric aerosols from different sources.The studies summarized here clearly show the impact that the mineralogy and the atmospheric and surface ocean processing of eld-collected aerosols have on the availability of soluble iron and its association with organic matter.Such results provide a framework for improving the representation of iron-containing particles in models.They also highlight the importance of understanding the chemical reactivity of iron at the molecular level under various conditions, which is the subject of the following sections.

Speciation of bulk iron in the dark
Iron chemistry in the dark provides a potentially important abiotic pathway for the night-time oxidation of organic compounds that could compete with nitrate radical chemistry. 97n the bulk aqueous phase, the chemistry of iron depends on which species is present at a given pH.The bulk speciation curves of Fe(III) in solution show that the pH affects the concentration of the hydrated species [Fe(H [98][99][100] In the presence of halides, such as chloride (Cl À ) ions, species such as [Fe(H 2 O) 5 Cl] 2+ and [Fe(H 2 O) 4 Cl 2 ] + also exist (Fig. 4).Curves shown in Fig. 4 are generated by the following reactions and equilibrium constants at pH # 3 and 25 C: 99 In addition to Cl À ions, the presence of other inorganic and organic species affects the speciation of Fe(III).Wittmer et al. 101 have listed equilibrium constants for reactions with bromide, sulfate, oxalate and catechol, which could be used in constructing curves such as those in Fig. 4. The following sections highlight a number of examples of chemical reactions driven by these iron species.

Generation of reactive oxygen species and further reaction with organic compounds
Fenton reactions, driven mainly by the addition of H 2 O 2 to bulk aqueous solutions containing Fe(II), are efficient in degrading soluble organic matter as a result of the formation of hydroxyl radicals, as shown in reaction (5): [102][103][104][105][106][107] Fe(II) 9][110][111][112][113][114] As shown in the following photochemistry sections, light accelerates the recycling of Fe(III) to Fe(II) in a system containing H 2 O 2 according to reaction (6): 107 Organic species such as quinones and carboxylate ligands were shown to have a similar effect. 107,115Even in chemically large compounds such as HULIS, the high density of the carboxylate and quinoid units have been shown to enhance the degradation of pyrene and phenolused as model organic pollutants released by anthropogenic sourcesby promoting dark Fenton reaction (5) in the aqueous phase. 116 number of studies have identied the degradation products of gallic acid and catechol by Fenton's reagent, H 2 O 2 and Fe(II).Duesterberg and Waite 107 reported a kinetic model based on experimental data that included reaction rate constants for the degradation of gallic acid by Fe(III).On complexation with Fe(III), a gallic acid-semiquinone species formed, which, in the presence of Fe(III)/Fe(II) resulted in the formation of a gallic acidquinone compound.An attack by cOH radicals on the latter species resulted in ring opening and then further reactions until mineralization was complete.In a study by Zazo et al., 104 catechol was formed rst from the oxidation of phenol, which then underwent ring opening.Aliphatic C2-C4 organic acids were detected at later stages, such as maleic acid, which was identied as the primary product from ring cleavage, and fumaric acid (C4).Further oxidation to oxalic (C2) and formic (C1) acids was observed when high Fe(II) and H 2 O 2 concentrations were used.Similar results were reported by M'hemdi et al. 117 on the role of Fenton and photo-Fenton processes in the complete mineralization of catechol.Using attenuated total internal reectance Fourier transform infrared (ATR-FTIR) spectroscopy, Arana et al. 118 studied the photo-Fenton degradation of phenol, which resulted in the formation of catechol and hydroquinone as intermediates.It was suggested from their results that pyrogallol formed as a result of the further degradation of catechol.These latter two compounds complexed with Fe(III), lowering its free concentration, and, subsequently, the reaction progress.Fenton-like reactions driven by the combination of Fe(III) and H 2 O 2 in the presence of Cl À ions were reported as key ingredients in the formation of short-lived volatile carbon suboxide (C 3 O 2 ) 119 and volatile trichloromethane (CHCl 3 ) from catechol 120 over long time frames exceeding 24 h.The formation of oxalic acid in solution was reported as a primary process aer 1 h from the dark oxidation of catechol and other hydroxylated benzenes, a process that was also observed in soils rich in organic matter with spiked Fe(III) and H 2 O 2 . 121Fig. 5 summarizes these results and shows a suggested degradation pathway based on these ndings and other studies.
In addition, Fig. 5b shows that the complexation of Fe with an organic ligand with the catechol moiety results in thermal oxidation in the bulk aqueous phase to form quinone.The spontaneous oxidation of polyphenols to the corresponding quinone, according to reaction (7), proceeds with an oxidation potential of about À0.7 V: 122,123 Polyphenol / o-quinone For example, the values for the oxidation of gallic acid, catechol and pyrogallol to the corresponding o-quinone are À0.799,À0.792 and À0.713 V, respectively. 122Despite being a spontaneous reaction for these chemicals, this is a kinetically controlled reaction and is very slow in the presence of dissolved oxygen alone. 122The addition of Fe(III) speeds up this oxidation because the reduction potential to Fe(II) is 0.749 V.The quinone species was found to be relatively unstable in solutions containing gallic acid under acidic conditions. 124Catechol oxidation produces semiquinone and quinone species through a twoelectron process (i.e.n ¼ 2 in reaction ( 7)). 123,125In these studies, the quinone was mainly identied using UV-visible spectrophotometry with its signature broad and relatively weak absorption around 400 nm.This wavelength is very close to that calculated from the HOMO-LUMO gap (3.23 eV) of o-quinone with tert-butyl substituents. 126Fig. 6 shows the UV-visible spectra of gallic acid and catechol on mixing with Fe(III) in the dark under acidic conditions. 31This formation of quinone species is accompanied by the reduction of Fe(III) to Fe(II), which, in oxygenated solution, undergoes autoxidation back to Fe(III) stabilized by complex formation with catechol-containing molecules. 122,123][133] In the case of catechol, an intense green colour forms in the pH range 2-4 on mixing with Fe(III) as a result of the formation of a bidentate mononuclear catechol-Fe(II) complex with a ligand-to-metal charge transfer (LMCT) band around 700 nm. 122,134Complex formation between gallic acid and Fe(III) under acidic conditions (pH < 3) results in the formation of a blue colour with an LMCT around 660 nm and assigned to the bidentate mononuclear gallic acid-Fe(II) complex. 124Using electrospray ionization mass spectrometry, Ross et al. 135 reported complex ion formation with the addition of Fe(III) to tannic ligands that included gallic acid.They determined that the formula [L-3H + Fe(II)] À could explain the observed masses.A more detailed experimental investigation using electrospray ionization mass spectrometry by Lutui et al. 136 coupled with ab initio calculations also conrmed the oxidation state of iron in these complexes and proposed a number of different structures that could explain the results.Heyden et al. 137 used X-ray based iron L-edge and carbon K-edge spectromicroscopy to analyse particles from oxic marine and freshwater sites for iron speciation and associations with organic matter.Fig. 7 summarizes the results of their paper by showing the types of organic functional groups and their distribution among the different iron-containing samples.The ubiquity of Fe(II) in these colloidal samples sheds some light onto the changes in the kinetics of redox cycling as a result of complexation to organic ligands.
The presence of the -OCH 3 group as a substituent on the benzene ring, as in the case of guaiacol, favours the formation of oligomers at the expense of stable guaiacol-Fe complexes. 138he addition of iron results in the development of soluble amber-coloured (about 412 and 470 nm) dimers and trimers of guaiacol.These have also been observed by Hwang et al. 139 from the biochemical oxidation of guaiacol by manganese peroxidase in the presence of H 2 O 2 .The presence of oxidants results in the formation of phenoxy radicals that initiate C-C coupling reactions to form dimers. Analysis of the UV-visible spectra of mixtures of solutions showed a decrease in the intensity of the bands in the 400-500 nm range on the overnight storage of solutions. 139,140Schmalzl et al. 138 assigned the 470 nm peak to an unstable 4,4 0 -diphenoquinone intermediate to explain these observations.These molecular level results provide insights into the increase in optical properties observed experimentally from the interactions of Fe(III) 141 and other metals 142 with dissolved organic matter.In summary, the bulk aqueous phase chemistry of iron is dependent on pH, with different soluble species capable of chelating organic compounds with different affinities.Under dark conditions and with certain redox-active organic compounds, the cycling of Fe(II) and Fe(III) and the role of dissolved oxygen and other oxidants become important factors in determining the fate of organic compounds.

Nature of surface water at different interfaces
Studies that highlight the molecular level differences between bulk and interfacial environments in atmospherically relevant systems have been the subject of thematic special issues of the Journal of Physical Chemistry C 143 and Physical Chemistry Chemical Physics. 144In addition, the nature of surface water and its role in the heterogeneous reactions of sea salt and mineral dust particles with gas phase species such as cOH, O 3 , SO 2 , cNO 2 , HNO 3 (to name a few) have been reviewed by Finlayson-Pitts 17 and Grassian, 145 respectively.The studies highlighted in these reviews showed that: (a) the liquid layer on the surface of sea salt is enhanced with ions (Cl À , Br À and I À ) even at low RH, which directly affects the reaction probabilities of gases and interfacial acidity; (b) the surface versus bulk partitioning of ions such as nitrate in the presence of water depends on the amount of water; (c) surface water enhances ionic mobility, which leads to the regeneration of surface sites; (d) surface water on airborne dust and organic monolayers 146 is "structured" compared with bulk liquid water, which affects their ability to act as efficient condensation nuclei; and (e) the photolysis of chromophores such as deliquesced surface nitrate is more efficient in producing gas phase O( 3 P), cOH and cNO 2 than photolysis in solution.Grassian and co-workers 147,148 reviewed studies on the heterogeneous chemistry of metal oxide and carbonate interfaces and how surface water plays a role in: (a) altering reaction pathways and surface speciation; (b) enhancing surface reactivity as a result of an enhancement in ionic dissociation and mobility; (c) solvating ions from the adsorbed phase; (d) inhibiting surface reactivity due to site blocking; and (e) hydrolysing reactants, intermediates and products.
To investigate the role of organic compounds, the adsorption of water from the gas phase on at surfaces terminated with organic functional groups was studied using IR spectroscopy, as reviewed by Moussa et al. 146 and Asay et al. 149 Analysis of the  stretching mode of OH groups, v(OH), provides further insight into the nature of the hydrogen bonding network for surface water in different samples.For example, the v(OH) region in the IR spectrum of liquid water is characterized by a broad feature at about 3400 cm À1 , which red shis to about 3200 cm À1 in ice as a result of stronger and more ordered hydrogen bonds. 150,151t was found from experimental data and molecular dynamic (MD) simulations of water clusters on hydrophobic organic monolayers that the relative intensities of the 3200/3400 cm À1 components correlated with the number of hydrogen bonds among water molecules. 146Fig. 8 shows the adsorbed water as a function of RH for different surfaces with and without organic species.Water molecules in contact with organic monolayers at low RH have fewer hydrogen bonds and give rise to spectral components at 3200 cm À1 , whereas molecules in the interior of water clusters have three and four hydrogen bonds similar to bulk water.In a related study, Nichols et al. 152 studied the reaction of a benzophenone-catechol mixture with NO 2 under dark conditions at 20% RH using ATR-FTIR and noticed an increase in the absorption of a broad feature between 2000 and 3000 cm À1 with the reaction time.This structured feature, with resolved components at 3100 and 2900 cm À1 , was assigned to the v(OH) of strongly hydrogen bonded adsorbed water on the solid phase of more hygroscopic reaction products (mostly 4-nitrobenzene-1,2-diol).
Using diffuse reectance infrared Fourier transform spectroscopy (DRIFTS), Cowen and Al-Abadleh 36 recorded the IR spectra of water adsorbed on tannic acid powder as a function of RH before and aer irradiation.Fig. 9 shows the difference absorbance spectra obtained using dry tannic acid as the reference.The spectral region between 1800-1000 cm À1 contains fundamental vibrations of organic functional groups and the bending mode of water, d(H 2 O).Changes in this region due to water adsorption on organic surfaces are rarely analyzed (see references in Moussa et al. 146 ).The trends observed in the spectral region below 1800 cm À1 suggest changes to the phase of the starting material with increasing RH as a result of the dissolution of tannic acid at the interface or the formation of tannic acid hydrates.Detailed analysis 36 of the v(OH) region suggested the formation of a strong hydrogen bonding network in tannic acid hydrates arising from tannic acid-water or waterwater interactions at the interface, which dominated at low and high RH, respectively.In addition, changes to the functional groups of tannic acid aer irradiation seemed to inuence the hydrogen bonding network in adsorbed water, as evidenced by  the red shi of the broad feature maximum by about 39 cm À1 and the increase in the intensity of a lower frequency component at 3039 cm À1 .
Using non-linear sum frequency generation (SFG), the spectra of water molecules at the vapour-water and organic-water interfaces were also studied. 153,154The SFG spectra of water at these interfaces show broad features extending from 3500 to 3000 cm À1 , with a maximum around 3480 cm À1 for the vapourwater interface that blue shis to about 3500 for the CCl 4 -water interface.This broad feature was assigned to water molecules involved in strong hydrogen bonding with nearby water molecules at the interface. 153High-frequency features have been observed in the SFG spectra of interfacial water molecules near 3700 cm À1 , which are referred to as the "free OH" mode of highly oriented water molecules. 153This feature is narrow and centred at 3705 cm À1 for vapour-water that is red shied to 3674 cm À1 for alkanes-water, 3669 cm À1 for CCl 4 -water and 3650 cm À1 for CDCl 3 -water interfaces.This sharp feature was used as an indicator for the presence of water-organic species interactions and the relative strength of this interaction: the lower the frequency of "free" v(OH), the stronger the water-organic species interaction.Hence the aforementioned trend observed in SFG studies is indicative of a weaker alkane-water interaction than the CCl 4 -water interaction and the strongest interaction is observed for water at the CDCl 3 -water interface as a result of its higher polarity.It was concluded from these studies, which were complemented by MD calculations, that these weak interactions drive a molecular ordering behaviour that extends well into the organic phase. 153e have studied the uptake of water on FeCl 3 particles as a function of the RH using DRIFTS. 31This is a hygroscopic salt that has been reported to be completely deliquesced at 77% RH for suspended micron-sized particles, 155 which is a much higher value than that estimated from the bulk solubility (45%) by Cohen et al. 155 Our IR data showed that at RH values < 40% the v(OH) band had maxima at 3572 and 3070 cm À1 , which became enveloped by features at 3375 and 3228 cm À1 with increasing RH.The presence of OH groups with fewer hydrogen bonds than those found in the bulk phases was manifested by spectral features with frequencies between 3700 and 3500 cm À1 .The data suggested that water-FeCl 3 interactions at low RH resulted in the formation of only a few, but strong, hydrogen bonds and that, with increasing RH, the water-water interactions became more pronounced and gave rise to a more "liquid-like" hydrogen bonding network.Interestingly, the uptake of water on FeCl 3 particles in our studies did not exhibit a sharp or well-dened increase at deliquescence.This behaviour was observed previously in electrodynamic balance experiments 155 and was explained by the formation of polymeric ferric hydroxide species with increasing water uptake according to: where m, n ¼ 1, 2, 3, .. The release of protons in reaction (8)  clearly suggests that the "quasi liquid" phase at the air-solid FeCl 3 interface is acidic in nature.The change in interfacial acidity as a result of hydration is not oen discussed in water uptake experiments on iron (oxyhydr)oxides.][158][159][160] Donaldson et al. 161 demonstrated experimentally that the surface acidity of soils controlled by amphoteric aluminium and iron (oxyhydr)oxides determines the uptake of HONO and the efficiency of desorption.
We also investigated the effect of irradiation on the hydrogen bonding network of water adsorbed on FeCl 3 at 30% RH. 30 Fig. 10 shows the changes in d(H 2 O) as a function of irradiation time from difference spectra with growing features at 1643 and 1601 cm À1 , suggesting that the iron species that formed with irradiation were [Fe(H 2 O) 6 ] 2+ , where a sixth water molecule bound directly to Fe(II) to complete its rst hydration shell to six ligands. 162,163The v(OH) region appeared to be narrower than that observed in the dark, with additional narrow peaks on the low frequency shoulder at 3221 and 3182 cm À1 .This was a manifestation of the disturbances to the hydrogen bonding network in the rst and subsequent shells around the Fe(III) and Fe(II) centres as a result of the absorption of light.
The structure of water adsorbed on g-Fe 2 O 3 , 156 nano Fe 2 O 3 (ref.157) and goethite particles [158][159][160] was studied using infrared spectroscopy as a function of RH.Spectra collected by Goodman et al. 156  groups were seen at 2712, 2697 and 2510 cm À1 .This study also examined the adsorption of CO 2 (g) on Fe 2 O 3 particles in the presence of adsorbed water and provided data for the formation of surface bicarbonate (FeOCO 2 H), which, in the presence FeOH groups, formed FeCO 3 À and adsorbed H 3 O + , thus acidifying the surface.A related study was conducted by Song and Boily 159 to identify the types of hydroxyl groups on goethite nano-rods under dry conditions and with increasing RH.Fig. 11 summarizes their ndings on dry particles where the experimental IR spectrum was correlated with a calculated spectrum using MD calculations.On the adsorption of water on this hydroxylated surface of goethite, the high-frequency absorption band decreased in intensity and new features around 3582, 3551 and 3510 cm À1 appeared below 10 Torr water vapour (<50% RH).These were enveloped by the most intense band around 3340 cm À1 , leading to the interpretation that liquid-like water thin lms formed at the goethite surface.Wijenayaka et al. 158 investigated the effect of particle size on the adsorption of water on goethite nano-and micro-rods and found that the total amounts of water normalized to the surface area were similar.A size effect was observed in the adsorption of nitric acid to these materials, in which the micro-rods took up more irreversibly bonded HNO 3 than the nano-rods.This was explained by surface structural changes to the hydroxyl groups; a decrease in the total reactive hydroxyl group density per unit area was observed in going from larger to smaller goethite particles. 158o summarize this section, gas phase water interactions with solid or liquid organic surfaces, iron salts and iron (oxyhydr) oxides lead to the formation of an adsorbed phase with a different hydrogen bonding network to that of water in the bulk liquid and solid phases.Depending on the hygroscopicity of the solid material, the surface water can either increase the ionic mobility with impacts on the surface acidity or increase the degree of hydroxylation of the surface sites.Each case presents a unique chemical environment for reactions with organic compounds.

Reactivity of surface iron with organic compounds: complexation and enhanced dissolution
Soluble and insoluble iron-containing materials are reactive towards organic compounds because iron centres are strong complexing agents and, in the presence of oxidants such as H 2 O 2 , they initiate Fenton and Fenton-like reactions.The complexation of organic compounds to iron could affect their degradation in the dark and their photochemical decay rates depending on the amount of surface water.This is because the hydrogen bonding network among water molecules at the gassolid interface is different from that in bulk water.In many cases, this chemical reactivity leads to the dissolution of iron (oxhydr)oxides, with consequences for the bioavailability of iron and the chemical and photochemical reactivity of the system.In this section, several examples related to this topic are highlighted.
The heterogeneous chemistry of soluble iron salts such as FeCl 3 is relevant to that of aged iron-containing mineral dust particles, whereas that using hematite mimics the chemistry of freshly emitted dust.The adsorption of short-chain volatile organic compounds (VOCs), including aliphatic alcohols, ketones and acids, on a number of minerals and metal oxides has been reviewed by Usher et al. 90 In addition, the catalytic activity of natural and synthetic iron (oxyhydr)oxides in Fenton and Fenton-like reactions has been reviewed by Pereira et al. 164 A less studied topic is the adsorption of aromatic VOCs on solid metal-containing salts.
We recently investigated the complexation of catechol to FeCl 3 particles under dry and humid conditions using DRIFTS. 31The spectra of surface species shown in Fig. 12A resemble those collected for aqueous phase catechol in the absence of Fe(III) in solution at acidic pH values.The spectral data suggest that, under dry conditions, catechol adsorbs molecularly and is fully protonated, showing no evidence for complexation with Fe(III).On increasing the RH until it reached 30%, the spectra collected as a function of time (Fig. 12B) showed clear changes to the shape of the bands and the intensity assigned to the functional groups in catechol, in addition to an increase in the amount of surface water (band at 1620 cm À1 ).The enhancement in Fe(III) mobility under these conditions led to the formation of stable catechol-Fe complexes at the gas-solid interface.These spectra resembled that collected for a catechol-Fe solution at pH 4, with a few peaks slightly shied as a result of the unique hydration environment at the gas-solid interface.These data also showed that the uptake of water by hygroscopic FeCl 3 particles changed the interfacial pH to acidic values, giving rise to a different complexation behaviour with organic compounds from that of coordinated Fe(III) in hematite.
5][166] To show the contrast of the described studies with hematite, we conducted similar experiments for catechol vapour uptake under dry and humid conditions.Because hematite is an insoluble metal oxide, catechol adsorption occurs via exchange with ligands on the Fe sites, as shown in reaction (9): The release of water caused little change to the acidity of the adsorbed water layer.
Fig. 13 shows the spectral features of surface catechol under dry and humid conditions.When compared with the spectra of aqueous catechol, 31 the data under the dry conditions suggested the formation of monodentate catechol-Fe complexes that were hydrogen bonded to neighbouring sites.For the spectra collected at 30% RH, the features assigned to the surface catechol closely resembled those observed for the adsorption of aqueous phase catechol on hematite 167 and goethite 168 particles at neutral to basic pH values.So, again, with this comparison of the aqueous phase spectra collected at known pH, we can obtain an idea about the interfacial pH in the absence of direct measurements.
0][171] Examples of these processes from a recent review of the geochemical literature are shown in Fig. 14. 172 In another review, Liu et al. 173 summarized studies of the adsorption of small molecular weight organic ligands and fulvic and humic acids on goethite surfaces.Cwiertny et al. 174 synthesized the literature on the specic dissolution processes of iron (oxyhydr)oxides that included proton-promoted, ligand (oxalate)-promoted, dark reductive via ascorbic acid, and  photochemical reductive dissolution in the presence of oxalate.The following section highlights selected studies on the role of organic molecules that mimic HULIS in the complexation and dissolution of iron (oxyhydr)oxides over a wide pH range.
Using surface-sensitive infrared spectroscopy, ATR-FTIR and batch dissolution experiments, Gulley-Stahl et al. 167 correlated the IR spectra of catechol complexes on high surface area Fe 2 O 3 particles and other metal oxides (Cr 2 O 3 , TiO 2 and MnO 2 ) with their dissolution behaviour (i.e.[metal] free ) in the pH range 3-10.The results on Fe 2 O 3 showed that dissolution aer 30 min of mixing had a higher dependency on pH than on the ionic strength.The presence of catechol promoted dissolution under neutral and basic conditions compared with the proton-driven dissolution that was signicant under acidic conditions (pH < 5).This was explained by the formation of inner sphere catechol complexes at pH > 5 with a higher degree than outer sphere complexes under acidic conditions.In the light of other studies that showed that bidentate binuclear complexes of catechol inhibit dissolution, these workers inferred that bidentate mononuclear complexes at pH > 5 were responsible for promoting dissolution.
In a later study, Yang et al. 168 studied the adsorption of catechol on goethite particles using ATR-FTIR and complemented these measurements with density functional theory (DFT) calculations of catechol complexes in the mononuclear monodentate (M-M), binuclear bidentate (B-B) and mononuclear bidentate (M-B) forms (Fig. 15).By examining the time prole of spectral features assigned to these complexes, they concluded that M-M and B-B complexes coexisted in the pH range 5-9, with the possibility of partial conversion of M-M to B-B via proton exchange with neighbouring surface sites under basic conditions and high surface coverage.The dissolution of goethite was studied in the presence and absence of catechol for 12 h in the dark under neutral to basic pH, where it was found that dissolved iron concentrations were below the detection limit of 5 mg L À1 .This nding was explained by the dominance of B-B complexes that inhibited dissolution under these conditions and the possibility of dissolution-readsorption as observed in oxalate-goethite systems.In another study by the same group, 175 the adsorption of salicylate was investigated in the presence and absence of catechol using ATR-FTIR and DFT calculations under neutral to basic conditions.The data revealed the formation of M-M complexes through the carboxylic acid group.The addition of catechol resulted in the formation of M-M and B-B complexes, which coexisted with salicylate complexes.It was also found that catechol replaced some of surface salicylate with no enhancement in goethite dissolution.
Redox-active organic compounds promote the reductive dissolution of iron (oxyhydr)oxides.Anschutz and Penn 176 investigated the reductive dissolution of crystalline ferrihydrite and goethite nanoparticles using hydroquinone as an electron donor.The production of p-benzoquinone and Fe(II) was monitored using HPLC and a ferrozine assay, respectively.The results showed that the surface area normalized rates were 100 times faster for ferrihydrite than for goethite particles, with faster rates observed for smaller sizes of nanoparticle.These results were explained by the lower degree of crystallinity and higher surface energy of ferrihydrite relative to the goethite nanoparticles.Because quinones play a key role in electron transport in minerals and microbes, Orsetti et al. 177 reported measurements of the reduction potential of reactive iron species at the goethite surface using a non-sorbing quinone species, namely anthraquinone-2,6-disulfonate.Fig. 16 shows a comparison of the reduction potential of this system in relation to other environmentally relevant redox couples at pH 7.  Shi et al. 178 studied the reductive dissolution of goethite and hematite nanoparticles by reduced avins, a class of organic compounds secreted by marine iron-reducing bacteria, Shewanella sp., with chemical structures that resemble HULIS.The results showed a higher reactivity for hematite than goethite in the pH range 4-7 under anaerobic conditions based on surface area normalized dissolution rates.When compared with poorly crystalline ferrihydrite, heterogeneous electron transfer from the reduced avins was orders of magnitude higher than that using crystalline goethite.Other factors that affected the extent of reaction and the initial rates included the structure and functional groups of the organic substrate, the redox potential of the reductants and the iron (oxyhydr)oxides, the aggregation state and the pH.
To investigate the dominant mechanism that leads to the dissolution of iron in dust, Shi et al. 179 collected iron-containing dust samples from two sites that represented sources of Saharan and Asian desert dust.In the laboratory, they simulated the effect of cycling between wet aerosols (i.e.acidic conditions) and cloud droplets (i.e. higher pH conditions) on the concentration of dissolved iron over a 3 h timeframe.The results showed that insoluble iron dissolved readily under the acidic conditions relevant to wet aerosols, whereas at higher pH values the dissolved iron precipitated as poorly crystalline nanoparticles.Therefore, in the long-range transport of mineral dust particles, the time spent as either a wet aerosol or in clouds will affect the amount of bioavailable iron on deposition.Acidic uptake within clouds can also enhance the dissolution of iron.As noted by DeMott et al., 180 cycling between the liquid, semisolid and solid phases of aerosols takes place as result of changes in the temperature and RH.Variable solvent to solute ratios also change the pH and the relative importance of surface versus bulk chemistry.
In a related study by Chen and Grassian, 181 the ligand-versus proton-promoted dark dissolution of iron by acids was investigated using coal y ash and Arizona test dust as representatives of anthropogenic and natural mineral dust, respectively.Although the pH of the solutions was maintained at 2, the relative capacities of the three acids investigated to dissolve iron were in the order: oxalic acid > sulfuric acid > acetic acid.The formation of bidentate complexes of surface iron with oxalate explained the great extent of iron dissolution compared with monodentate complexes with acetate.It was noted that, at low oxalate concentrations, the competition of dissolved iron and surface iron for oxalate complexation suppressed the dissolution rate of iron.This study highlighted the similar role that could be played by other organic ligands that are preferentially associated with atmospheric dust aerosols, such as C2-C12  dicarboxylic acids, C16-C30 fatty acids 182 and polycyclic aromatic hydrocarbons. 183he accelerated dissolution of iron (oxyhydr)oxides was observed when these materials were trapped in ice in the presence of organic electron donors in the dark. 184Fig. 17 shows the concentration of total dissolved iron, [Fe tot ], from the dissolution of goethite and maghemite in the bulk aqueous phase and in ice samples in the presence of organic ligands under acidic conditions.The authors found that most of the dissolved iron was in the ferric form, suggesting that this dark dissolution process was not reductive.The highest rates of dissolution were found with high surface area iron oxides and strong iron-binding ligands.The results were explained by a "freeze concentration effect" where solid particles, organic ligands and protons were concentrated in the liquid-like ice grain boundary region.Fig. 18 shows the location of the grain boundary region in ice crystals. 185There have been a number of reviews on the structure of the surface of snow 186 and on other types of chemical processes that take place there. 187o gain an insight into the mechanisms of the sizedependent dissolution rates of iron (oxyhydr)oxides, Lanzl et al. 174 studied the inuence of the primary particle size and solution pH (from acidic to approximately neutral) on 8 and 40 nm hematite nanoparticle aggregates.The smaller particles exhibited a 3-10 times enhancement in the mass-normalized reactivity from the pH dependent dissolution rates.This study also investigated the inuence of aggregation on the dissolution of hematite, which led to a loss of reactive surface area as previously observed in the dissolution of goethite nano-rods. 188ey found that the dissolution of aggregated 8 nm particles was uniform, whereas preferential etching at the edges and structural defects took place for 40 nm particles.Although this study and an earlier one by Cwiertny et al. 189 on the dissolution of goethite nano-and micro-rods were conducted using oxalic and ascorbic acids as the organic ligands, they served as blueprints for future dissolution rate experiments using organic ligands other than chelated iron.
This section has summarized key investigations on the dark complexation of organic ligands from the gas and liquid phases to surface iron in soluble and insoluble materials, how this process inuences the cycling between Fe(II) and Fe(III) species and the overall dissolution of iron oxides.These results provide the rst step in our mechanistic understanding of the extent of photochemical reactions driven by iron under various reaction conditions, as detailed in the following sections.

Bulk photochemistry of iron
][19]190 The kinetics and mechanisms of these photochemical reactions are pH dependent as a result of speciation. 100Fig. 19 shows the cross section of Fe(III) species a function of pH, which overlaps with the solar actinic ux at l > 290 nm. 1,190,191When compared with similar data for important chromophores in atmospheric aerosols (nitrate, nitrite and HULIS) 14 and also H 2 O 2 , 192 the cross section values in Fig. 19 are nearly 100 times larger than those of these chromophores.The complexation of Fe(III) with organic compounds results in a red shi in its UV absorption spectra to the visible region of the electromagnetic spectrum, with consequences for the overall photochemical reactivity of the system. 31,124,129,141,193he photochemistry of Fe(III) species in aqueous aerosols provides an important pathway for photodegrading WSOC. 16eactions involving aqueous Fe(III) can be described by eqn (10)  and (11): where I is the photon ux, f is the quantum yield and L is an organic ligand.The quantum yield for reaction (10) varies depending on the technique and wavelength range of the light source.Values ranging from 0.07 to 0.31 have been reported by Lim et al. 99 based on their literature review.The production of cOH radicals 100 in reaction (10) will also result in the formation of oxygenated products. 1948][199] Factors that affect the photolysis of the Fe(III)oxalate complex, in particular, such as the Fe(III) concentration, speciation and the wavelength and intensity of the excitation light were further investigated by Weller et al. 200 In another related study by the same group, Fe(II) quantum yields and reaction mechanisms were reported from the photolysis of Fe(III)-carboxylate complexes. 199The quantum yields for the photolysis of Fe(III)-gallocatechin complexes in aqueous solutions to yield solvated electrons were also reported. 201This class of molecules is relevant to atmospheric chemistry as it contains contain the catechol moiety.Quantum yields values were found to be around 0.26 at pH 11.5 and 0.13 at pH 7.5.The higher values under basic conditions were explained by the ease of oxidizing deprotonated metal complexes.The following section highlights some examples of the photochemistry driven by iron species in the bulk aqueous phase.

Iron as a photosensitizer in the degradation of WSOC
Interest in the iron-driven aqueous phase photosensitized degradation of organic compounds is driven mainly by the desire to treat wastewater contaminated by soluble natural organic matter and phenolic compounds produced from the decay of plants or the food and pigment industries. 102,103,105This chemistry can also be used to explain homogeneous reactions in cloud droplets 41 and atmospheric organic aerosols. 106In this photochemistry, Fenton reagents, such as Fe(II) and H 2 O 2 , are most commonly used as efficient producers of cOH radicals (reaction ( 5)). 107For this reaction to be catalytic, the recycling of Fe(III) to Fe(II) according to reaction ( 6) is the rate-limiting step in the presence of H 2 O 2 . 107The kinetics of reaction ( 6) can be enhanced by light (reaction (10)) and in the presence of quinones and carboxylate ligands. 107,115The following section summarizes the key ndings from photo-Fenton studies on models for HULIS that include gallic acid, catechol and guaiacol.
Quici and Litter 202 used a 15 W UV lamp with a maximum output at 366 nm to quantify the kinetics of gallic acid photodegradation on adding Fe(III) to a gallic acid-H 2 O 2 mixture at pH 3.They reported a value for k UV of 0.049 min À1 compared with 0.0022 min À1 in the absence of H 2 O 2 .In another study, Quici et al. 203 showed that the electrophilic addition of photoproduced cOH species to the aromatic ring in the presence of oxygen resulted in the release of CO 2 and further oxidation of the gallic acid ring, which eventually resulted in ring opening and the formation of oxygenated organic compounds until the complete mineralization of gallic acid.Benitez et al. 204 reported the photodegradation of gallic acid using a 15 W Hg lamp with a maximum output at 254 nm in the absence and presence of H 2 O 2 and Fe(II) under neutral to acidic conditions.The highest k UV value (0.07 min À1 ) was reported when Fe(II) was added to an aqueous gallic acid-H 2 O 2 mixture at pH 3.
Lofrano et al. 205 investigated the conditions under which complete mineralization of catechol took place using Fenton reagents [H 2 O 2 and Fe(II)] at pH 3 and irradiation with light.Using a 125 W uorescent lamp with a maximum output at 350 nm and analysis by gas chromatography (GC) and UV-visible spectrophotometry, they found that 600 : 500 w/w of H 2 O 2 : Fe(II) and 30 min irradiation resulted in the signicant removal of aromaticity, even when starting with relatively high catechol concentrations of up to 200 ppm.Another study reported that only oxalic acid was detected as a product by ionexclusion HPLC from the photo-Fenton degradation of catechol, whereas a number of intermediates were identied by GC and ion-exclusion HPLC from the dark Fenton process.Using guaiacol, Samet et al. 206 also reported the fastest degradation rates from photo-Fenton processes using the sun as the light source and 40% less H 2 O 2 compared with the dark Fenton reaction.They noted that the photooxidation process initiated by cOH radicals formed a mixture of o-and p-quinone intermediates that absorbed light in the visible region.Further attack by cOH radicals resulted in ring opening and the formation of dicarboxylic acids and eventually CO 2 and water.
In summary, the photochemical generation of ROS from hydroxylated iron species is pH dependent and efficient in transforming WSOC to variable degrees depending on their structure and the relative concentration of the oxidants.In the presence of light, the cycling between Fe(II) and Fe(III) becomes catalytic in nature to a certain point in the process.This photoreactivity could be complicated in the presence of other iron species with different ligands such as halogens, discussed in the following section.

Photochlorination of WSOC in the presence of iron and the production of halogens
The absorption of light by solvated Fe(III) in the presence of halogens such as Cl À ions results in the formation of chlorine radicals according to reaction (12): The quantum yields for reaction (12) vary depending on the experimental method.Lim et al. 99 listed values between 0.093 and 0.13 from steady-state techniques and 0.46-0.57from laser kinetic spectroscopy measurements.In addition, the photochemistry of hydroxylated Fe(III) species in the presence of Cl À ions yields Cl/Cl 2 c À and ClOHc À radicals, which, on formation, can undergo a number of bimolecular reactions with inorganic and organic species in solution.Table 1 lists the rate constants for reactions involving Cl and cOH radical species, adapted from Wittmer et al. 101 The scavenging of cOH radicals by Cl À species competes with the formation of H 2 O 2 , explaining the overall inhibition of photo-Fenton reaction at high chloride concentrations (about 10 mM) under acidic conditions. 207,2080][211][212][213] For example, the bulk photochlorination of phenol 210 and organic matter in forest soils was shown to proceed through these pathways and result in the formation of volatile organochlorine compounds. 210,214eppler et al. 215 reported the formation of C1-C4 organochlorine compounds in their studies on catechol degradation in the presence of Fe(III) and Cl À .The photoproduction of Cl 2 gas from Fe(III) and Cl À solutions was also reported by Lim et al. 99 using UV light at 365 nm.
Homogeneous and heterogeneous pathways that affect the formation of Cl 2 gas are of great importance in atmospheric chemistry. 216Field studies have measured molecular halogens in coastal urban air at parts per trillion levels, which were mainly explained by known gas phase chemistry. 217,218Organic matter such as chlorophyll and aromatic ketones were shown to promote the heterogeneous photooxidation of halide ions at the air-salt water interface. 219,220In a related study, Ofner et al. 221 showed that reactive chlorine and bromine species react with SOA particles derived from a-pinene, catechol and guaiacol, which, in turn, changed their chemical (i.e.structure) and physical (i.e.size and optical) properties.Recently, Wittmer et al. 101 reported an order of magnitude increase in Cl 2 gas release from an Fe(III)-doped NaCl salt pan relative to plain NaCl samples on irradiation.The same group also observed the inhibition in Cl 2 gas release from the former samples on adding sulfate, oxalate and catechol, which was explained by the strong complexation of these ligands with Fe(III).An inhibition in the release of brominated and chlorinated gaseous species was observed in the presence of the SOA precursors catechol and guaiacol (no iron) on irradiation of a simulated salt pan. 221To summarize, in multicomponent atmospheric aerosols containing Fe(III), Cl À and organics, the aforementioned bulk chemistry inuences the photooxidation of halide ions and the nature of the organic phase.Although bulk phase chemistry is relevant for the micron-size droplets common in clouds, surface contributions become more important as the particle size decreases and, as detailed in the following section, are explained by different mechanisms from the bulk processes.

Surface photochemistry of iron
Current cloud chemistry models contain thermodynamic and kinetic parameters from known photochemical reactions in the bulk phase. 222These reactions do not necessarily and accurately represent photochemical reactions occurring at the surface of aerosols or on atmospherically relevant surfaces. 17Questions still remain about the relative efficiency of interfacial Fe(III) photochemistry in degrading organic matter in the presence of a few layers of adsorbed water compared with that in bulk liquid water under photon uxes that simulate the solar ux.Answering these questions demands molecular level information on the nature of the surface water (see Section 4.1) and the photochemistry of multicomponent surfaces containing transition metals such as Fe(III) in their soluble and insoluble forms.The following sections highlight recent studies in these areas.

Solvent cage effect and surface iron
In small droplets or at the interface of systems containing chromophores such as iron, an observed overall increase in quantum yield is expected for radical formation. 223This increase is explained by the enhancement in the surface concentration of reactants, increases in the light intensity due to morphology dependent resonances and/or refraction, and/or decreased solvent cage effects. 223Nissenson et al. 192,223 demonstrated through experiments and modelling that the third factor is the major contributing factor aer accounting for the surface enrichment of the organic species and the distribution of light throughout the droplet.One experiment was the photolysis of molybdenum hexacarbonyl, Mo(CO) 6 , in 1-decene, either as liquid droplets or in bulk liquid solutions, where the results showed that the rates in the aerosols were faster by at least three orders of magnitude than in the bulk liquids. 223The modelling study was carried out on cloud droplets containing benzene and [Fe(OH)] 2+ as a photosensitizer that produced cOH radicals on irradiation. 192The average angle and wavelength-relative light intensities were calculated as a function of distance from the droplet centre for three droplet radii of 1-3 mm.The results showed that the calculated light intensity decreased at the surface.The overall rate equation for the decay of benzene as a result of this chemistry included the rate equations for all possible reactions, such as reactions with scavengers in cloud droplets.The distribution of the reaction rate within the droplet volume was calculated from the integration of this equation over the radius range.The results of this exercise showed a sharp increase in the decay rate of benzene at the interface, where the fraction of the total reaction in the surface layer was calculated to be 35 and 15% for droplets with radii of 1 and 3 mm, respectively.Aer accounting for the surface enhancement of benzene and the distribution of light throughout the droplet, the increase in the surface reaction rate was explained by the increase in the photolysis quantum yield due to the decreased solvent cage.Some research groups in North America and Europe have studied the efficiency of surface-photosensitized reactions in metal-free organic-containing aerosols.This photochemistry can lead to the oxidation of surface organics, an enhancement in uptake and the production of gas phase species, and the formation of SOA. 19Examples include (but are not limited to) the following: Karagulian et al. 224 investigated the "bottom-up" photooxidation of a fatty acid adsorbed on NaCl mixed with solid NO 2 À and reported the formation of organic products consistent with either an O À or cOH radical attack depending on the amount of water.Cowen and Al-Abadleh 36 proposed mechanism for the formation of surface aryl aldehydes by irradiating solid tannic acid under humid conditions using DRIFTS.Kleffmann and co-workers measured the fast photosensitized formation of HONO from the reaction of NO 2 on humic acids 43 and mineral dust 225 under humid conditions.Donaldson and co-workers reported the photoenhanced uptake of NO 2 and O 3 and the oxidation of halides by chlorophyll at the salt water-air interface. 219,226,227In addition, a new photoinduced pathway for particle growth was reported by Monge et al. 228 as a result of the photoenhanced uptake of terpenes (limonene and isoprene) on seed particles containing humic acid, succinic acid and ammonium nitrate (1 : 10 : 1 in weight).In these experiments, humic acid acted as the main photosensitizer and the formation of the triplet state produced radicals at the surface of aerosols, which enhanced the uptake of non-condensable terpenes.Very little experimental work has explored the extent to which the known Fenton and photo-Fenton chemistry takes place at the gas-solid interface in the presence of organic matter and a few layers of adsorbed water.The following section highlights related studies that aimed to ll this knowledge gap.
6.2.Heterogeneous photochemistry of insoluble iron in the presence of organic species Iron (oxyhydr)oxides have band gaps in the visible region of the electromagnetic spectrum.For example, the band gap for hematite is around 2.2 eV (565 nm) 229 and that of goethite ranges from 2.1 to 2.5 eV (592-497 nm); 166 these band gaps have been found to increase with decreasing particle size. 230,231herefore band gap excitation using UV photons with energies >2.1 eV leads to the creation of electron-hole pairs followed by the formation of Fe(II) species.The photoreactivity of iron (oxyhydr)oxides towards inorganic gases such as nitrogen and sulfur oxides have been shown to be important for the renox-ication and oxidation of sulfur dioxide. 232The absorption of light by adsorbed organic species initiates LMCT to surface ^Fe(III), leading to the formation of reduced iron sites, ^Fe(II), which then desorb to the bulk phase (i.e.photoreductive dissolution).The formation of electron-hole pairs in the presence of organic species can also initiate redox chemistry, leading to the transformation of these materials. 13he photoreductive dissolution of iron (oxyhydr)oxides in natural water and seawater has received much attention and was reviewed by Zwiener and Frimmel 233 and Baker and Croot. 234Similar studies were conducted with iron-containing coal y ash and Arizona test dust as representatives of anthropogenic and natural mineral dusts. 181Briey, the studies showed the direct photoreductive formation of Fe(II) or organiciron complexes in photoreductive experiments, with the highest rates observed at low pH values.Fe(II) was subsequently oxidized to Fe(III), which reforms smaller iron colloids or binds to organic chelators.The presence of organic ligands enhances this process, particularly at low pH values.The indirect reduction (and oxidation) of iron species also takes place as a result of reactions with ROS such as superoxide (O 2

À
) and hydrogen peroxide (H 2 O 2 ) produced from photochemical reactions involving dissolved organic matter in seawater and enzymatic action in bacteria and phytoplankton.The cycling between Fe(II) and Fe(III), as summarized earlier, explains the apparent extended lifetime of Fe(II) in iron enrichment experiments in the Southern Ocean.
Some reports have shown that the photoreductive dissolution of iron (oxyhydr)oxides was accelerated for particles trapped in ice.Kim et al. 185 measured the concentration of dissolved Fe(II) from the dissolution of hematite, goethite and maghemite in ice samples under UV and visible irradiation in the presence and absence of organic electron donors under acidic pH conditions and compared the results with measurements from dark experiments and similar experiments in bulk liquid water.
The results showed an enhancement in the photogeneration of dissolved Fe(II) in ice regardless of the type of iron material or the organic ligand, with higher rates using UV rather than visible light.This is in contrast with dark dissolution in ice, where Fe(III) was mainly formed and its formation rate depended on the type of organic ligand. 184The photochemical mechanism was explained with the following equations for the case of hematite, which can be generalized for other iron (oxyhydr)oxides: 185 ^Fe(III)-L + hn / ^Fe(II)-Lc + (photoinduced LMCT) (13) 5,235 Examples included the photoenhanced uptake of NO 2 and the formation of HONO on mineral dust, 225 the photoenhanced uptake of O 3 by dust, 236 the photoenhanced uptake of formaldehyde 237,238 and short-chain alcohols 239 on TiO 2 , Fe 2 O 3 , dust and volcanic ash.The mechanisms that explain these observations involve the generation of electron-hole pairs as a result of the absorption of light, which serves as the driver for charge transfer reactions with adsorbed water, NO 2 and O 3 .The main radical that forms from the reaction of adsorbed water with the electron-hole pairs generated on irradiation is the cOH radical, which seems to contribute to the photoenhanced uptake of VOCs.This photouptake was reported to be dependent on the RH, as in the case of formaldehyde, 238 where the uptake coefficient was at a maximum at 30% RH and then decreased at higher values as a result of site blockage by adsorbed water.Nitrite and chloride anions are also reduced by reaction with holes, which, in the presence of aromatic compounds, can lead to nitration and chlorination of the benzene ring. 13The following section shows the differences between these studies and those involving soluble iron.

Heterogeneous photochemistry of soluble iron in the presence of organic compounds
The mechanisms that explain the reactivity of coordinated Fe(III) in iron (oxyhydr)oxides do not apply to solvated Fe(III) cations in hygroscopic iron salts.In a set of experiments probing the gasaqueous interface, Riha Kameel et al. 240 reported the Fenton oxidation of gaseous isoprene using acidic FeCl 2 aqueous microjets.In these experiments, the gaseous reactant streams, isoprene and H 2 O 2 intersected the droplets containing iron for about 10 ms and the products were analysed in situ via online electrospray ionization mass spectrometry.The results were consistent with an oxidation process initiated by the addition of cOH radicals to protonated isoprene oligomer homologues, followed by fast reactions involving dissolved H 2 O 2 , HO 2 c and O 2 that led to the formation of polyols, carbonyls and, to a lesser extent, carboxylic acids in the condensed phase.When taking into account typical values for the uptake coefficients of the reactants, the concentrations and Henry's law constants, it was concluded that this iron-driven pathway for the oxidation of VOCs and the formation of SOA was potentially important during both the day and at night.
Tofan-Lazar and Al-Abadleh 30 examined the role played by UV-visible light in the uptake of gas phase catechol by samples containing FeCl 3 particles.Experiments were conducted such that the solid sample was irradiated before the introduction of catechol vapour under humid or dry conditions; this was followed by the introduction of catechol vapour under continuous irradiation.The solid lines in Fig. 20a show the spectra of surface catechol as a function of the irradiation time at 30% RH, whereas the dashed lines represent the spectra collected under dry conditions (<1% RH).These spectra are different from those shown for the uptake of catechol under dark conditions (humid and dry, Fig. 20b).Overlapping spectral features between 1600 and 1000 cm À1 were assigned previously 31 as evidence for the formation of catechol-Fe complexes under humid conditions based on comparison with the infrared spectra of these compounds in bulk solution recorded as a function of the pH value.The solid lines in Fig. 20a represent the net of two processes: the continuous uptake of catechol from the gas phase and photodegradation.The kinetic curves shown in Fig. 20c were best described using an empirical sigmoidal growth equation (see supporting information in Tofan-Lazar and Al-Abadleh 30 for equation, best-t parameters and the calculated growth factors).Within the uncertainty of these measurements, the growth factors calculated at 30% RH under light and dark conditions were very similar and were higher by nearly a factor of three than in dry conditions over a period of 60 min (light and dark).This suggests that light does not signicantly enhance the uptake of catechol vapour on FeCl 3 particles and that surface water in equilibrium with 30% RH (in the light and dark) increases the surface concentration of catechol.
The irradiation of solid FeCl 3 samples containing adsorbed catechol under a constant ow of humid or dry air only (i.e.no gas phase catechol) gave difference infrared spectra that showed the breakdown of the surface catechol-Fe complexes with increasing irradiation time (Fig. 21a). 30This photodecay can be explained by eqn ( 10)-( 12) because the enhanced ionic mobility in the presence of surface water at 30% RH explained the presence of the reactant Fe species.We believe that reaction (12), producing cCl radicals, is responsible for the decay of the catechol-Fe complexes for the following reasons: (a) the high acidity of the "quasi liquid" phase at 30% RH as a result of FeCl 3 dissolution drives a higher concentration of [Fe(H 2 O) 5 Cl] 2+ (about 1 M, pH 1) relative to [Fe(H 2 O) 5 (OH)] 2+ (about 0.1 M, pH 1), as calculated from the bulk solubility of FeCl 3 (92 g/100 mL in water at 25 C) 241 and the bulk aqueous phase speciation curve shown in Fig. 4b the higher quantum yields for the photodissociation of [Fe(H 2 O) 5 Cl] 2+ versus [Fe(H 2 O) 5 (OH)] 2+ in the bulk aqueous phase (i.e. with a solvent cage) (see Section 6.1)these values are predicted to be higher at the gas-solid interface in the presence of adsorbed water; and (c) given the relatively low gas phase concentration of catechol in these experiments (estimated at 32 ppb (ref.31)) and an equilibrium constant for catechol-Fe complex formation of 4.32 Â 10 À2 M, 242 the concentration of [Fe-cat] 2+ is around 4 mM, assuming a pH of 1.In addition, catechol-Fe complexes are more susceptible to the electrophilic additions of cCl radicals to the aromatic ring than uncomplexed catechol as a result of LMCT in the former.The stretching absorptions of C-Cl bonds that indicate the chlorination of catechol would occur below 800 cm À1 .These absorptions were not observed in our data because they were obstructed by the liberational modes of surface water.These modes were most intense below 1000 cm À1 for bulk liquid water 243 and ice 244 and are known to blue shi and increase in intensity and bandwidth with increasing strength of hydrogen bonding. 245The absence of  carbonyl stretching modes between 1800 and 1700 cm À1 in Fig. 21a also suggested that there was no formation of quinone, ketone or aldehyde species, conrming negligible oxidation by photogenerated cOH radicals.
Fig. 21b shows the normalized kinetic curves under humid conditions and control experiments for the photodecay of adsorbed catechol under dry conditions (RH < 1%) and with no FeCl 3 at 30% RH.The inset shows the natural logarithm of these data and the best-t slopes corresponding to the apparent photodecay constants as listed in the caption.This analysis clearly showed that surface water enhances the initial photodecay kinetics of catechol-Fe complexes at the solid-humid air interface by a factor of ten relative to control experiments with no FeCl 3 under humid conditions or with dry FeCl 3 .The slower rate observed under humid conditions at longer irradiation times suggested a contribution from another reaction that caused the photodecay of the catechol-Fe complexes.Hence the main two ingredients for the fast photodecay of organic compounds to take place were solvated Fe(III) and surface water in equilibrium with at least 30% RH.The lifetime of the organic compounds was estimated to be 38 min as a result of this heterogeneous pathway, as calculated in the supporting information of Tofan-Lazar and Al-Abadleh, 30 which is about 1000 times shorter than that calculated for the fastest homogeneous reaction of cCl radicals with organic compounds.
In summary, as observed in the dark, this section has illustrated the key mechanistic differences in the surface photochemical reactivity of iron towards the transformation of organic matter, which depends on the source of the iron (hygroscopic versus insoluble material) and the amount of surface water.

Summary and future directions
This review has synthesized the current state of knowledge of iron chemistry in multicomponent atmospheric aerosols, particularly those containing HULIS and their model compounds.Field measurements have focused on the sources, characterization and fate of iron to the oceans and have led to the improvement of a number of models.Laboratory studies have highlighted the complexity of iron reactivity, which varies depending on the solubility, the redox conditions, the absence and presence of UV-visible light, the presence of organic species and ROS, the pH and temperature.Several examples have shown key differences between the bulk and surface chemistry of iron-containing materials, which varied considerably as a result of the solvent cage effect and the structure of water.
Elucidation of the mechanisms and the extraction of kinetic and thermodynamic parameters are necessary for photochemical reactions occurring at the surface and in the bulk of multicomponent aerosol systems containing transition metals, such as iron, and organic matter.These experiments need to be conducted under the atmospherically relevant conditions of temperature, RH, the wavelength and intensity of light, and over relatively short and long timeframes, taking into account the phase, size, water content, oxidant content and acidity of model aerosol systems.These studies need also to be complemented with analytical techniques or procedures capable of identifying and quantifying the gas phase species that are either consumed or released.Working with the different classes of organic functional groups identied in ambient aerosols will provide correlations with the rates and mechanisms that are hard to extract when working with large multifunctional macromolecules.Hence, when characterizing this latter class of molecule, either as model compounds or collected from eld campaigns, it is insightful to report the amount and type of organic functional groups in addition to the O : C ratios.Comparison of the results obtained from photochemistry with those obtained in dark conditions will complete the picture on the relative importance of reactions during the day and at night.
Insights from computational chemistry on iron-water clusters in the presence of other inorganic and organic solutes would be invaluable in understanding the interfacial regions of these systems.These results would need to be coupled with the more accurate quantication and characterization of soluble iron and organic matter in eld samples.Such an integrated approach would improve the predictive power of climate and cloud chemistry models with respect to the heterogeneous dark and photochemistry of aerosols and other atmospherically important iron-containing surfaces and their role in generating SOA, altering the balance of gas phase species and increasing the solubility of iron in different materials.

Fig. 2
Fig. 2 Scanning electron microscopy images and energy-dispersive X-ray analysis elemental maps of particles representing (a) fly ash, (b) mineral dust, (c) NaCl agglomerates and (d) Ca-S agglomerates.Reprinted with permission from ref. 75 (Copyright © 2012, American Chemical Society).

Fig. 1
Fig. 1 Association of transition metals with biological material in sea spray aerosols.(A) Negative (left) and positive (right) mass spectra of a representative individual silver-containing particle.(B) Relative contributions of the different particle types as measured by aerosol time-of-flight mass spectrometry (ATOFMS) for particles without silver (left) and silver-containing particles (right).Bio ¼ bioparticles, SS-OC ¼ sea salt with organic carbon, SS ¼ sea salt.Reprinted with permission from ref. 76 (Copyright © 2014, American Chemical Society).

Fig. 3
Fig. 3 Global distribution of the effective mineral content of soils (as a percentage) for: (a) quartz; (b) illite; (c) kaolinite; (d) smectite; (e) feldspar; (f) calcite; (g) hematite; (h) gypsum; and (i) phosphorus.The mineral fraction was weighted with the clay and silt content of the soil.For minerals that were present in both clays and silts, the weighted values were summed.Reproduced with permission from ref. 89.

Fig. 4
Fig. 4 Speciation curves generated for 1 mM FeCl 3 solution as a function of pH at 25 C, plotted on a log scale, using an in-house MATLAB program courtesy of Dr D. Scott Smith, Wilfrid Laurier University.

Fig. 6
Fig. 6 Left-hand panels: UV-visible absorbance spectra of solutions collected after mixing (a) gallic acid and (b) catechol with FeCl 3 solutions for 3 min as a function of increasing molar ratios of Fe.Righthand panels: resultant spectra of reaction mixtures after subtracting the spectra of the reactants.Initial concentrations of gallic acid and catechol were 0.1 and 0.4 mM, respectively, at pH 3 and 0.01 M KCl ionic strength.Ratios listed are mol : mol of organic ligand : Fe(III).Reproduced with permission from ref. 31 (Copyright © 2013, American Chemical Society).

Fig. 7
Fig. 7 (A) Frequency of each set of organic functional groups as found in association with Fe-rich particulates of varying chemistry and sampling location.(B) Corresponding analyses of organic functional group frequency for Fe-poor regions of each particle or sample window.Reproduced with permission from ref. 137 (Copyright © 2014, American Chemical Society).

Fig. 8
Fig. 8 Infrared spectra of adsorbed water in equilibrium with water vapour at different percentages of RH on (a) plasma-cleaned glass, (b) glass coated with a C8 SAM and (c) glass coated with a C18 SAM.The dotted line in (a) is the spectrum of bulk liquid water.The dashed lines in (b) and (c) represent the fitting of the 80% RH spectra into two peaks centred at 3200 and 3400 cm À1 , respectively.The different colours correspond to adsorbed water at different percentages of RH: green (20% RH), blue (40% RH) and red (80% RH).Reproduced with permission from ref. 146 (Copyright © 2009, American Chemical Society).
Fig. 10 Difference DRIFTS absorbance spectra collected at 30% RH (solid lines) and under dry conditions (dashed lines) as a function of irradiation for a sample containing 1% FeCl 3 (w/w) in diamond powder in the absence of adsorbed catechol (as a control experiment).Reprinted with permission from ref. 30 (Copyright © 2014, American Chemical Society).

Fig. 11 (
Fig. 11 (a) TEM image of goethite particles (scale bar 50 nm) and (b) idealized particle morphology.((c), top) FTIR band assignments of hydroxyl groups on goethite surfaces and ((c), bottom) the MD-derived power spectrum of the (110) plane.Schematic representation of the first-layer waters adsorbed at the (110) surface of goethite from the top view (d) and a side view (e) as obtained from a snapshot of an MD simulation.Black dashed lines represent original H-bonding patterns in water-free systems, whereas the pink lines represent new H-bonds in the presence of water.Reprinted with permission from ref. 159 (Copyright © 2013, American Chemical Society).

Fig. 12
Fig. 12 Representative time-dependent DRIFTS absorbance spectra of surface catechol on 1% FeCl 3 (s) as a result of (A) flowing gas phase catechol in the dark under dry conditions, RH < 1% and (B) increasing the RH to 30%, resulting in complexation to Fe(III) in the presence of adsorbed water.The spectra shown in set (a) are the same as shown in (A) at times of 10, 30 and 60 min while flowing gas phase catechol under dry conditions.The spectra shown in set (b) are for adsorbed catechol after stopping the flow of catechol vapour with a continuous flow of dry air for 10, 30 and 60 min.The spectra shown in set (c) are for adsorbed catechol as the RH of the air flow is increasing to 30% RH (no gas phase catechol) at times of 0.5, 1, 1.5, 2, 2.5, 3, 4, 5, 10, 30 and 60 min (from bottom).Reproduced with permission from ref. 31 (Copyright © 2013, American Chemical Society).

Fig. 13
Fig.13DRIFTS absorbance spectra of catechol vapour uptake on solid hematite nanoparticles (1% w/w in diamond powder) under dry (bottom, RH < 1%) and humid (top, 30% RH).The reference spectrum used is that collected for hematite nanoparticles prior to introducing catechol vapour.Original work from experiments in the author's laboratory.

Fig. 14
Fig. 14 Schematic description showing the diverse mechanisms for the interaction of dissolved organic matter on the surface of metal oxides.Reproduced with permission from ref. 172 (Copyright © 2014, American Chemical Society).

Fig. 16
Fig. 16 Comparison of the reduction potentials of some environmental relevant redox couples at pH 7 vs. the SHE (E 00 H ). Reprinted with permission from ref. 177 (Copyright © 2013, American Chemical Society).

Fig. 18
Fig. 18 Optical images of polycrystalline ice showing the highly aggregated iron oxide particles trapped in ice veins (grain boundary region) at À20 C. (a) Pure ice, (b) ice with maghemite (g-Fe 2 O 3 ) (1 g L À1 ), (c) ice with colloidal haematite (0.8 g L À1 ) and (d) schematic illustration of concentrated hematite particles (orange circles) and formic acids (ball-and-stick models) in ice grain boundary region.Reproduced with permission from ref. 85 (Copyright © 2010, American Chemical Society).

Fig. 19
Fig. 19 Cross section of Fe(III) species as a function of pH using 1 mM FeCl 3 solution.

Fig. 20
Fig. 20 Representative time-dependent DRIFTS absorbance spectra of surface catechol on 1% FeCl 3 (s) as a result of flowing gas phase catechol in the dark under (a) irradiation and (b) in the dark.Solid and dashed lines show the spectra collected under humid (at 30% RH) and dry conditions (RH < 1%), respectively.The gas phase concentration of catechol is estimated as 32 ppb.(c) Kinetic curves showing the increase in the integrated absorbance from 1558 to 1018 cm À1 assigned to surface catechol in the presence of 1% FeCl 3 (s) as a result of flowing gas phase catechol under dry and humid conditions, both in the dark and irradiation.Error bars represent AEs from averaging two experiments.abs ¼ absorbance in y-axis label.Reprinted with permission from ref. 30 (Copyright © 2014, American Chemical Society).

Fig. 21 (
Fig. 21 (a) Difference DRIFTS absorbance spectra showing the photoreactivity of surface catechol in the presence of 1% FeCl 3 (s) as a result of irradiation under humid (at 30% RH, solid lines) and dry (at RH < 1%, dashed lines) conditions.The FeCl 3 samples were first exposed to gas phase catechol in the dark for 60 min, then the catechol flow was turned off while maintaining a humid or dry air flow over the sample during the irradiation part of the experiment.(b) Kinetic curves showing the decrease in the normalized integrated absorbance from 1589 to 1000 cm À1 assigned to surface catechol in the absence and presence of 1% FeCl 3 (s) as a result of irradiation under humid and dry conditions.Solid markers represent experimental data collected under dry and humid conditions.Open markers represent data from a control experiment with no Fe in the sample.Lines represent least-squares fittings using double-exponential equations.The inset shows experimental data plotted in the linearized first-order kinetics form for the first 20 min of irradiation.The values of slopes represent the apparent photodecay constant (min À1 ): 0.21 AE 0.03 (line 1), 0.023 AE 0.003 (line 2), 0.016 AE 0.002 (line 3) and 0.009 AE 0.001 (line 4).Error bars represent AEs from the averaging of two experiments.Reprinted with permission from ref. 30 (Copyright © 2014, American Chemical Society).

Table 1
Rate and equilibrium constants for reactions involving Cl and cOH radical species at zero ionic strength and 298 K (adapted from Table 2 in ref. 101 with permission)