中国科技论文在线 http://www.paper.edu.cn Environ. Sci. Technol. 2010, 44, 263–268

In general, the monomeric ferric complex FeIII(OH)2+ is Photochemical Cycling of Iron the most photoactive species in the absence of organic ligands in clouds and fog (3). The photolysis of FeIII(OH)2+ under UV Mediated by Dicarboxylates: Special • Effect of Malonate irradiation leads to generation of Fe(II) and OH(eq 1). + hv • FeIII(OH)2 98 Fe(II) + OH (1) ZHAOHUI WANG, XI CHEN, HONGWEI JI, WANHONG MA, CHUNCHENG CHEN, AND • DOM + OH f oxidized products (2) JINCAI ZHAO* Beijing National Laboratory for Molecular Sciences, Key Extensive field measurements have shown that dissolved Laboratory of Photochemistry, Institute of Chemistry, The organic matters (DOM) are very ubiquitous in atmospheric Chinese Academy of Sciences, Beijing 100190, China water droplets where the dissolved iron coexists at a comparable concentration (1). DOM that have relatively poor Received July 1, 2009. Revised manuscript received affinity to Fe(III), such as most monocarboxylates, may November 1, 2009. Accepted November 24, 2009. enhance the production of Fe(II) by scavenging •OH radical (eq 2), which decreases the rate of reoxidation of newly generated Fe(II) (9). Our recent work also revealed that influx of various DOM (10) or inorganic chromium species (11) can Photochemical redox cycling of iron coupled with oxidation change Fe(II)/Fe(total) ratio in different ways. of malonate (Mal) ligand has been investigated under conditions Among these Fe(III)-DOM species, Fe(III)-oxalato com- that are representative of atmospheric waters. Malonate plexes are highlighted for their considerably high photoac- exhibited significantly different characteristics from oxalate tivity under sunlight irradiation. The photolysis of ferrioxalate and other dicarboxylates (or monocarboxylates). Both strong complex may proceed as follows (eqs 3-8) (4, 5):

chelating ability with Fe(III) and strong molar absorptivities, but hv III 3- 98 III -+ •- much low efficiency of Fe(II) formation (ΦFe(II) ) 0.0022 ( [Fe (C2O4)3] [Fe (C2O4)2] 2CO2 (3a) 0.0009, 300-366 nm) were observed for Fe(III)-Mal complexes (FMCs). Fe(III) speciation calculation indicated that Mal is hv III 3- 98 II 2- + •- capable of mediating the proportion between two photoactive [Fe (C2O4)3] [Fe (C2O4)2] C2O4 (3b) species of Fe(III)-OH complexes and FMCs by changing •- - - - the Mal concentration. Spin-trapping electron spin resonance CO + [FeIII(C O ) ]3 f [FeII(C O ) ]2 + CO + C O2 · 2 2 4 3 2 4 2 2 2 4 (ESR) experiments proved the formation of both the CH2COOH · (4) and OH radicals at lower total Mal concentration ([Mal]T), but · •- + f + •- only CH2COOH at higher concentrations of malonate, providing CO2 O2 CO2 O2 (5) strong evidence for competition between malonate and OH- and •- + + T • subsequent different photoreaction pathways. Once FMCs O2 H HO2 (6) • dominate the Fe(III) speciation, both photoproduction and 2HO f H O + O (7) photocatalyzed oxidation of Fe(II) will be greatly decelerated. 2 2 2 2 + f + • + - There exists an induction period for both formation and decay Fe(II) H2O2 Fe(III) OH OH (8) of Fe(II) until FeIII(OH)2+ species become the prevailing Fe(III) forms over FMCs as Mal ligand is depleted. A quenching Ferrioxalate complex absorbs a photon and undergoes mechanism of Mal in the Fe(II) photoproduction is proposed. photodissociation without electron transfer from the oxalate to iron (eq 3a) (12) or with a ligand-to-metal charge transfer The present study is meaningful to advance our understanding (LMCT) process, yielding Fe(II) and oxalate radical anion of iron cycling in acidified carbon-rich atmospheric waters. •- •- 0 (C2O4 ) (eq 3b) (5, 13). CO2 is a strong reducing agent (E )-1.8 V (NHE)) and can react with another ferrioxalate •- Introduction molecule or can instead reduce O2 to superoxide anion (O2 ) at near-diffusion-controlled rate (k ) 2.4 × 109 M-1 s-1)(14). Iron species have been identified as a ubiquitous component The photochemically generated Fe(II) can be reoxidized by •- • • in atmospheric water droplets (i.e., cloud, rain, or fog droplets) O2,O2 /HO2, OH, H2O2, and other oxidants. in field measurements (1). The maximum concentration of Malonic and succinic acids are also dominant dicarboxylic dissolved iron species was up to 200 µM in fogwater (2). acids only inferior to oxalic acid in carbon-rich atmospheric Photochemistry of atmospheric iron has a significant effect waters (7, 15). Since oxalate and malonate have the strongest on numerous chemical processes in atmospheric waters, chelating capacity with Fe(III) among all dicarboxylates and especially in redox and radical chain reactions such as natural monocarboxylates (see Table S1), the photochemistry of fluctuation of reactive oxygen species (ROS) (3-5), oxidation Fe-malonate complexes is expected to be important in iron of dissolved sulfur dioxide (SO2)(6) and organic substances cycling of atmospheric waters. However, there is little (7), and redox cycling of other trace metals (e.g., Cu, Mn) (8). information available so far regarding the photochemistry Furthermore, photolysis of iron species could contribute to of Fe(III)-dicarboxylate complexes except for that of ferri- iron input to open-ocean surface water via atmospheric oxalate (4). deposition, thereby increasing the bioavailability of iron to The objective of this study is to investigate the photo- aquatic biota (6). chemical behaviors of Fe(III)-malonate complexes (FMCs) comparing with those of Fe(III) complexes of oxalate and * Corresponding author fax: +86-10-8261-6495; e-mail: jczhao@ other carboxylates. The effect of malonate ligand on iron iccas.ac.cn. cycling and the photochemical reaction mechanism are also

10.1021/es901956x  2010 American Chemical Society VOL. 44, NO. 1, 2010 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 9 263 Published on Web 12/09/2009 转载 中国科技论文在线 http://www.paper.edu.cn

discussed. One of the most important findings is that, in comparing with ferrioxalate/ferric hydroxo species, a high level of Fe(II)/Fe(t) can be maintained under UVA irradiation in the presence of excess Mal. This phenomenon is unique among the dicarboxylates (C2-C6) and monocarboxylates (C1-C3).

Experimental Section Chemicals. Iron(III) perchlorate hydrate, iron(II) perchlorate hydrate, and 2,2,6,6-tetramethylpiperidine-1-oxyl (TEMPO) were purchased from Aldrich. 5,5-Dimethyl-1-pyrroline-N- oxide (DMPO) was from Sigma Chemical Co. Iron (III) tripotassium oxalate trihydrate was from Alfa Aesar. Oxalic acid, malonic acid, sodium hydroxide, perchloric acid, and 1,10-phenanthroline were of reagent grade and used as supplied. Barnstead UltraPure water (18.3 MΩ cm) was used FIGURE 1. UV-vis absorption spectra for solutions containing for all experiments. 100 µM Fe(III) and Mal ligand with different concentrations. Experimental Procedures. A 100-W Hg lamp (Toshiba Inset: Speciation (molar fraction) of 100 µM Fe(III) in solutions SHL-100UVQ-2) was employed as the ultraviolet irradiation with total malonate concentration ([Mal]T) between 10 and 1000 source. It mainly emits at 312 and 366 nm (16), overlapping µM. pH, 3.0. the solar UVA spectrum (320-400 nm) (see Figure S1). All experiments were conducted in a 70-mL cylindrical Pyrex (20). Typical instrumental conditions were as follows: center vial (Corning, Inc.) under continuous magnetic stirring. The field, 3480 G; sweep width, 100 G; resolution, 1024 pts; Pyrex vial also served as a high-pass filter so that only light microwave frequency, ∼9.77 GHz; microwave power, 12.68 with wavelengths greater than 290 nm penetrated into the mW; modulation frequency, 100 kHz. To minimize experi- vessel. Unless otherwise specified, all experiments were mental errors, the same quartz capillary was used for all the performed under exposure to air at room temperature. An measurements. The simulations of ESR spectra were obtained exhaust fan was employed during all the reaction processes with the use of WinSim EPR simulation software. to maintain the temperature below 35 °C. Fe(III)-organic acid solutions were freshly prepared by dilution of stock Results and Discussion solutions of 0.01 M dicarboxylic acids (DAs), 5 mM Fe(III) at Fe(III) Speciation. Table S1 (see Supporting Information) pH 1.5 (HClO4). The initial pH was adjusted with dilute HClO4 summarizes Fe(III) complexes with different carboxylates or NaOH. Under our experimental conditions ([Fe(III)], 100 and the corresponding equilibrium constants. Malonate µM; pH 3.0), the polynuclear iron complexes are negligible shows strong complexation ability with Fe(III) and is capable and are not considered in the photochemistry of iron cycling of forming mono-, di-, and trimalonato complexes with (see discussion in Supporting Information). For deaerated Fe(III), although the equilibrium constant for each stoichi- experiments, the solutions in the cap-sealed Pyrex vial were ometry of Fe(III)-Mal species is somewhat lower than that bubbled with high-purity Ar (O2 e 0.001%) for at least 20 min of the corresponding Fe(III)-oxalate complex. Due to their prior to UV irradiation and continuously purged throughout poor complexation capacity with Fe(III) (21, 22), the well- the experiment. During each kinetic experiment, a 1-mL known photochemistry of Fe(III)-OH complex is predomi- aliquot was sampled with a new syringe each time and nant in the presence of other mono- or dicarboxylates. immediately disposed for the consequent analysis. Figure 1 shows the speciation of Fe(III) as a function of Methods and Analysis. The concentration of Fe(II) was total malonate concentration ([Mal]T) at pH 3.0. The con- measured spectrophotometrically by a modified phenan- centrations of all the hexaaquo and hydroxylated Fe(III) throline method (10, 17). Briefly, 0.5 mL of 1,10-phenan- complexes decreased significantly with increasing [Mal]T, throline solution (5.0 mM), 1 mL of sodium acetate/acetic whereas FMCs predominated the speciation of Fe(III) at high III + III - acid buffer (pH 5.5), and 0.5 mL of ammonium fluoride [Mal]T.Fe (Mal) and Fe (Mal)2 are the major species of III 3- solution (0.1 M) were premixed, followed by addition of 1 FMCs at pH 3.0 whereas Fe (Mal)3 seems negligible under mL of the sample solution. The absorption of the resulting the present experimental conditions. solution was read at 510 nm using a 1-cm quartz cell on a Molar Absorptivities. Table S3 shows the values of molar Hitachi U-3100 spectrophotometer. We eliminated the pos- extinction coefficients of individual complexes of FMCs sibility that malonate would interfere with Fe(II) measure- within the range of 300-366 nm, which were obtained by ments by blank experiments (Figure S2). A DX-120 ion multivariate linear regression method based on the known chromatograph (Dionex Co.) with conductivity detection was molar absorptivities of Fe(III)-OH complexes (Figure 1). Both + - applied to quantitively determine organic acids. Five mM Fe(Mal) and Fe(Mal)2 exhibited strong light-absorbing + - NaOH solution was chosen as an eluent. Six mM Ar-purged abilities, for example, εFe(Mal) and εFe(Mal)2 at 300 nm are 3512 -1 -1 potassium ferrioxalate (0.05 M H2SO4) was used as a chemical and 3044 M cm , respectively, larger than ferric hydroxyl actinometer to measure quantum yield (ΦFe(II)) for Fe(II) chromophores. However, despite stronger light absorption formation. An average actinometer quantum yield of 1.14 of FMCs, their Fe(II) quantum yields were rather lower than was used. All light at the photolysis wavelength (300-366 that of monomeric Fe(OH)2+ (see discussion below). nm) was approximately absorbed by 6 mM ferrioxalate (18). Photocatalyzed Oxidation of Fe(II). Dissolved Fe(II) is Molar extinction coefficient of species involved in this study the predominant oxidation state in some atmospheric liquid was calculated by spectral analysis method reported by Hug waters and aerosol particles, accounting for 20-90% of the et al. (19). total Fe in fogwater samples from Zu¨rich (2), 40-72% of the Electron spin resonance (ESR) spectra of spin-trapping dissolved Fe in cloudwater samples from Kleiner Feldberg radicals by DMPO were recorded at room temperature on (23), and 85 ( 13% of total dissolved Fe in the sunlit aerosol a Bruker EPR ELEXSYS 500 spectrometer equipped with an particles from Okinawa (24). Therefore, an initial ratio of in situ irradiation source (a Quanta-Ray ND:YAG laser system Fe(II)/Fe(t) ) 63% was chosen to examine the photocatalyzed λ ) 355 nm). TEMPO (g ) 2.0051) was chosen as a standard oxidation of Fe(II) in the presence of different carboxylates. for determination of g factors as recommended elsewhere The photocatalyzed oxidation of Fe(II) proceeded rather

264 9 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 44, NO. 1, 2010 中国科技论文在线 http://www.paper.edu.cn

FIGURE 4. Rates of Fe(II) quantum yield (ΦFe(II)) and molar FIGURE 2. Photocatalyzed oxidation of Fe(II) in the presence of fraction of Fe(III) species as a function of [Mal]T. [Fe(III)], 100 malonate and oxalate. Ultraviolet irradiation source: 100-W Hg µM; pH 3.0. Irradiation source: 100 W Hg lamp; irradiation time: -2 - lamp (Toshiba SHL-100UVQ-2, average 3.6 W m at 320 400 8 min. Six mM Ar-purged potassium ferrioxalate (0.05 M H2SO4) nm). Fe(II):Fe(III) ) 63:37; [Fe(t)], 100 µM; dicarboxylate, 500 µM; was used as a chemical actinometer. pH, 3.0. aerobic Mal-containing solutions. Higher initial [Mal]T led to the longer induction period. As more Fe(III) was introduced to the Fe(III)-Mal-Fe(II) solution at the beginning of the reaction, the induction period was shortened considerably (Figure S3). Our recent study proved that the duration of induction period for the Fe(II) oxidation is closely related to the content of FeIII(OH)2+ species in Fe(III)-Fe(II) systems (10), that is, the photocatalyzed oxidation of Fe(II) should be initiated and then accelerated by photolysis of FeIII(OH)2+. In the present Fe(III)/Mal systems, Mal can substitute OH- ligand of FeIII(OH)2+ and further influence the Fe(III)- photocatalyzed oxidation of Fe(II), which has been verified by the following experiments. Fe(II) Photoproduction. According to Fe(III) speciation calculation, it is expected that Mal is able to control the proportion between Fe(III)-OH complexes and FMCs, which should further affect the photochemical reactions in irradi- FIGURE 3. Effect of [Mal] (marked beside the lines) on T ated solutions. Figure S4 shows the photoproduction of Fe(II) photocatalyzed oxidation of Fe(II). [Fe(t)], 100 µM; Fe(III):Fe(II) ) from Fe(III) at different [Mal] (0-1000 µM). The addition of 37:63; pH 3.0. T Mal led to an induction period for the Fe(II) production, slowly in the Mal system during the time scale of the which increased with [Mal]T. However, there was no induction experiment (Figure 2), whereas Fe(II) was rapidly oxidized period observed for Mal degradation (Figure S5), implying to Fe(III) in the presence of oxalate within 15 min of that slow Fe(II) production should be mainly attributed to irradiation. However, oxalate might be quickly mineralized the secondary (photo) chemical reactions involving reoxi- (11) and then Fe(II) would revive from Fe(III) and further dation of Fe(II) but not primary photolytic reaction of FMCs. approach the photosteady state equilibrium of Fe(III)/Fe(II) Quantum Yield (ΦFe(II)). The average ΦFe(II) at different (10). In contrast, for acetate and other di- or monocarboxylate concentrations of Mal was calculated according to the systems (data not shown), photocatalyzed oxidation of Fe(II) literature (eqs 9-11) (19). Where fi is the photon fraction showed behaviors similar to that without any carboxylates. absorbed by a specific species i; ci and εi are concentration The unique photoreaction behaviors of Ox and Mal among and molar extinction coefficient of species i, respectively; Kai di- or monocarboxylates should be attributed to the different is the rate of light absorption by species i; I0 is the incident photoreaction mechanisms of their ferric complexes. Ad- photon flux from the Hg lamp; l is optical path length; Φi is dition of Ox renders the acceleration of photocatalyzed individual quantum yield of species i for Fe(II) formation; oxidation of Fe(II), because photogenerated superoxide/ x(λ)isx (x ) fi or εi or I0) at certain wavelength λ. hydroperoxide radicals (eqs 5-6) may act as the oxidants ε (λ)c sink of Fe(II) species in acidic solution (5). However, the ) i i fi(λ) (9) reason photocatalyzed oxidation of Fe(II) was greatly retarded ∑ εi(λ)ci in the Fe/Mal system has not been reported. i Therefore, it is of interest to examine the factors controlling - the slow photocatalyzed oxidation of Fe(II) in the presence ) - ∑ εi(λ)cil Kai ∑ fi(λ)I0(λ)[1 10 ]∆λ (10) i of Mal ligand. Figure 3 shows the photocatalyzed oxidation λ of Fe(II) at different [Mal]T. The control experiments (curves a, b) indicated that both light exposure and presence of O2 ∑ ΦiKai are indispensable for the Fe(II) oxidation even in the presence ) d[Fe(II)]/dt ) i ΦFe(II) (11) of Mal. In addition, [Mal]T in solutions should be closely ∑ K ∑ K associated with the photocatalyzed oxidation of Fe(II). Fe(II) ai ai i i was rapidly oxidized after 20 min of irradiation in the absence of Mal, whereas quite a long induction period for the Figure 4 plots ΦFe(II) and molar fraction of Fe(III)-OH photocatalyzed oxidation of Fe(II) was observed in irradiated species against total malonate concentration (0-1000 µM).

VOL. 44, NO. 1, 2010 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 9 265 中国科技论文在线 http://www.paper.edu.cn

SCHEME 1. Proposed pathways for photodecomposition of FMCs. For simplification, water ligand is omitted from (H2O)4Fe(III)Mal complex

FIGURE 5. ESR spectra of DMPO radical-adducts formed during the photolysis of Fe(III)/Mal solution under argon-purged conditions. DMPO, 0.04 M; Fe(III), 100 µM; pH, 3.0. (1a) experimental data, Mal, 1000 µM; (1b) simulated data of 1a; (2a) experimental data, Mal, 200 µM; (2b) simulated data of 2a. (*) presumption that Mal can significantly decrease molar · · III 2+ DMPO- CH2COOH, (V) DMPO- OH. fraction of Fe (OH) by forming FMCs. Under this condition, · · OH radical should be hardly detected. DMPO- CH2COOH III + Since Fe (OH)2 was expected to have photoreduction adduct increased with the irradiation time, with the rate of reactivity roughly close to that of FeIII(OH)2+ due to their radical formation for curve a (Mal, 1000 µM) and b (Mal, 200 III + -1 similar nLpOH-dLMCT absorption (25), Fe (OH)2 was µM) in Figure S6, 35.2 and 5.9 au s , respectively. regarded equivalent to FeIII(OH)2+ in this study. In the absence Mechanism Discussion. FeIII(OH)2+ and Fe(III)-Ox com- of Mal, the average ΦFe(II) (300-366 nm) was 0.028 ( 0.0014, plexes have been considered as the predominant photoactive which is comparable to the reported values by Faust and species in DOM-free (3) and DOM-rich atmospheric liquids Hoigne´(3)(Φ313 ) 0.14, Φ360 ) 0.017). However, with the (4, 5), respectively, and could be switched mutually with the increase of [Mal]T, molar fraction of Fe(III)-OH complexes influx and depletion of organic ligands (10, 11). The involved and ΦFe(II) decayed rapidly. This suggests that contribution photochemical reactions have been extensively studied. Here of the photolysis of FMCs to Fe(II) generation was minor. we present another photochemical reaction pathway of iron Assuming that the calculated quantum yield for FeIII(OH)2+ redox cycling in the presence of malonate ligand (Scheme (0.028, 300-366 nm) is constant irrespective of the presence 1). of Mal, then the quantum yield for individual FMCs can be Photoproduction of Fe(II). Fe(II) is formed through the further calculated by eq 11. The obtained wavelength- Fe(III) photoreduction of major light-absorbing species in averaged ΦFe(II) of FMCs was 0.0022 ( 0.0009, about 1 order solutions: FMCs (process I) and Fe(III)-OH complexes of magnitude lower than that in Fe(III) control system (0.028 (process II), respectively. Similar to photolysis of FeIII(OH)2+ ( 0.0014). Since ΦFe(II) is the ratio of total Fe(II) formed to species, FMCs with higher molar absorptivities transfer to photons absorbed, all contributions from direct and indirect an electronically excited state upon UV irradiation, followed photochemical reactions are incorporated into the Fe(II) by two competitive reactions: (1) return to ground state; (2) quantum yield. There was a significant difference between LMCT process with forming Fe(II) and a carboxylate radical quantum yields for Fe(II) generation (0.0074 ( 0.0008) and in solvent cage (species a). The radical undergoes decar- · Mal degradation (0.036 ( 0.005) in anaerobic solution boxylation to yield CO2 and a Fe(II)/ CH2COOH radical pair · containing 100 µM Fe(III) and 500 µM Mal. This indicates (species b). In this study, CH2COOH radical has been that low yield of Fe(II) generation is not due to poor detected by ESR-trapping technology (Figure 5), proving that photoactivity of Fe(III)-Mal complexes but rapid reoxidation the common LMCT process of the irradiated FMCs actually of the photoproduced Fe(II). Under aerobic conditions, even takes place. However, it seems difficult for Fe(II) to diffuse lower quantum yield for Fe(II) formation (0.0016 ( 0.0007) out of the solvent cage since the reoxidation of Fe(II) by · was observed because of O2-involved reoxidation of Fe(II). CH2COOH readily occurs (28-30) especially in the presence Free Radicals Involved. Spin-trapping ESR technique was of uncomplexed Mal ligand. employed to identify the possible short-lived radicals involved This quenching mechanism of Mal in Fe(II) formation is in the reaction systems. It was carried out under anaerobic similar to that proposed in Cu(II)/Mal systems (31-34), which conditions to avoid the interference of quenching effect of is further evidenced by measuring acetate as an intermediate dioxygen on carbon-centered radicals. Our results indicated (Figure S5). We observed the Mal degradation (Figure S7), the generation of two major kinds of radicals during the providing a direct evidence that this quenching mechanism photoreaction of Fe(III)-Mal solutions at a lower concentra- operates even in the absence of dioxygen. Mal degradation tion of Mal (200 µM) (2a in Figure 5). They were assigned to coupled with the Fe(II) reoxidation can explain the quite low · be DMPO- CH2COOH (RH ) 23.01 G, RN ) 15.34 G) (26) and Fe(II) quantum yield of FMCs. Therefore, once FMCs control · DMPO- OH (RN )RH ) 14.8 G) (27), respectively. The current the Fe(III) speciation, few Fe(II) are accumulated in solutions. ESR measurements provided direct evidence for the genera- FeIII(OH)2+ species gradually prevail over FMCs as Mal ligand · tion of CH2COOH radical during the photodecarboxylation is depleted (Figure 1). Once the photolysis of Fe(III)-OH of FMCs. It is interesting to note that when 1000 µMofMal complexes revives, Fe(II) would be produced rapidly (Figure was added, the DMPO- · OH signal was not observed. · OH S4). radical is derived from the photolysis of FeIII(OH)2+,sothe Photocatalyzed Oxidation of Free Fe(II). Fe(II) is very stable ESR signal intensity of DMPO- · OH adducts can be regarded at pH 3 in the dark. The half-life for the Fe(II) oxidation by III 2+ as an indicative of concentration of photoactive Fe (OH) . O2 is at least longer than 285 days (35). The presence of Mal The generation of DMPO- · OH adduct in 2a (Mal, 200 µM) ligand did not obviously accelerate the thermal/dark oxida- but not in 1a of Figure 5 (Mal, 1000 µM) supported our tion of Fe(II). Both light irradiation and presence of O2 are

266 9 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 44, NO. 1, 2010 中国科技论文在线 http://www.paper.edu.cn

indispensable for the Fe(II) oxidation (Figure 3). Since Fe(II) Acknowledgments species at pH 3 do not absorb UV light (>300 nm) (3), it is This work was financially supported by 973 project Fe(III)-OH complex or FMCs that probably acts as a main (2007CB613306 and 2010CB933503), NSFC (20537010, chromophore to photocatalyze Fe(II) oxidation by O2. Our 20677062 and 20777076), and CAS/SAFEA. previous study has proved that the photocatalyzed oxidation of Fe(II) in aerated solutions is closely related to the content Supporting Information Available + of FeIII(OH)2 species in Fe(III)-Fe(II) systems (10). Although Additional experimental evidence. This material is available · photogenerated Fe(II)/ CH2COOH radical pair may also react free of charge via the Internet at http://pubs.acs.org. with O2 to form a relatively stable six-membered chelating ring (species c) that can further decompose or oxidize another Literature Cited ) ( × 6 -1 -1 proximate ferrous ion (k (5.0 1.0) 10 M s )(30), this (1) Deguillaume, L.; Leriche, M.; Desboeufs, K.; Mailhot, G.; George, process only give minor contribution to catalytic oxidation C.; Chaumerliac, N. Transition metals in atmospheric liquid of Fe(II) since the free Fe(II) in solutions was oxidized more phases: sources, reactivity and sensitive parameters. Chem. Rev. slowly when FMCs dominated the Fe(III) speciation (Figure 2005, 105, 3388–3431. 3). Therefore, it should be Fe(III)-OH complex but not FMCs (2) Behra, P.; Sigg, L. Evidence for redox cycling of iron in atmospheric water droplets. Nature 1990, 344, 419–421. that is pivotal for Fe(II) oxidation by O2. In conclusion, the (3) Faust, B. C.; Hoigne´, J. Photolysis of Fe(III)-hydroxy complexes presence of Mal greatly limits the formation of Fe(III)-OH as sources of OH radicals in clouds, fog and rain. Atmos. Environ. complex, thereby indirectly preventing Fe(II) against oxida- 1990, 24A, 79–89. tion. This phenomenon is unique among all di- or mono- (4) Faust, B. C.; Zepp, R. G. Photochemistry of aqueous iron(III)- polycarboxylate complexes: roles in the chemistry of atmo- carboxylates studied since photolytic reaction of ferrioxalate spheric and surface waters. Environ. Sci. Technol. 1993, 27, 2517– •- • - accelerates the formation of O2 /HO2 oxidants (eqs 5 6) and 2522. Fe(III)-OH complex is always the prevailing Fe(III) species (5) Zuo, Y. G.; Holgne´, J. Formation of hydrogen peroxide and when other di- or monocarboxylates besides oxalate and depletion of oxalic acid in atmospheric water by photolysis of malonate are present. iron (III)-oxalato complexes. Environ. Sci. Technol. 1992, 26, 1014–1022. Environmental Implications. Our study reveals that (6) Zhuang, G. S.; Yi, Z.; Duce, R. A.; Brown, P. R. Link between iron malonic acid plays a quite different role in the photoredox and sulphur cycles suggested by detection of Fe(II) in remote cycling of iron relative to oxalic acid and other di- or marine aerosols. Nature 1992, 355, 537–539. monocarboxylic acids. Both oxalate and malonate are pos- (7) Zuo, Y. G.; Holgne´, J. Photochemical decomposition of oxalic, glyoxalic and pyruvic acid catalyzed by iron in atmospheric sible subproducts from the chain reactions of dicarboxylic waters. Atmos. Environ. 1994, 28, 1231–1239. acids (35) and both can form stable Fe(III) complexes with (8) Cies´la, P.; Kocot, P.; Mytych, P.; Stasicka, Z. Homogeneous high molar absorptivities, but their fates are quite different. photocatalysis by transition metal complexes in the environ- Fe(III)-oxalato complexes are very photosensitive and can ment. J. Mol. Catal. A: Chem. 2004, 22, 17–33. •- (9) Mcknight, D. M.; Kimball, B. A.; Bencala, K. E. Iron photore- easily generate reducing CO2 radical upon irradiation, which - duction and oxidation in an acidic mountain stream. Science can further reduce another Fe(III) oxalato complex (eq 4) 1988, 240, 637–640. and thereby enhance the Fe(II) formation (5). However, FMCs (10) Song, W. J.; Ma, W. H.; Ma, J. H.; Chen, C. C.; Zhao, J. C.; Xu, have much lower Fe(II) quantum yield because an oxidizing Y. M.; Huang, Y. P. Photochemical oscillation of Fe(II)/Fe(III) · induced by periodic flux of dissolved organic matter. Environ. CH2COOH radical is generated and reoxidizes the photo- generated Fe(II) in the presence of excess Mal. Therefore, Sci. Technol. 2005, 39, 3121–3127. (11) Wang, Z. H.; Ma, W. H.; Chen, C. C.; Zhao, J. C. Photochemical the different photoreaction pathways and nature of radicals coupling reactions between Fe(III)/Fe(II), Cr(VI)/Cr(III) and derived from oxalate and malonate result in their opposite polycarboxylates: inhibitory effect of Cr species. Environ. Sci. roles in iron cycle. Technol. 2008, 42, 7260–7266. Zuo et al. reported that the Fe(III)-Fe(II)-Fe(III) cycling (12) Chen, J.; Zhang, H.; Tomov, I. V.; Rentzepis, P. M. Electron - transfer mechanism and photochemistry of ferrioxalate induced time for Fe(III) Ox complex was on the order of only 100 s by excitation in the charge transfer band. Inorg. Chem. 2008, (7). The half-life of Fe(III)-aquo complexes was approxi- 47, 2024–2032. mately 9 min at pH 3 (36). Our study reveals that high level (13) Pozdnyakov, I. P.; Kel, O. V.; Plyusnin, V. F.; Grivin, V. P.; Bazhin, of Fe(II)/Fe(t) (Figure 3) can be maintained for a long time N. M. New insight into photochemistry of ferrioxalate. J. Phys. in the presence of excess Mal. Many field measurements Chem. A 2008, 112, 8316–8322. (14) Surdhar, P. S.; Mezyk, S. T.; Armstrong, D. A. Reduction potential indicated that Fe(II) accounts for most of the dissolved iron of the carboxyl radical anion in aqueous solutions. J. Phys. Chem. in rainwaters (23, 24, 37), but the mechanism involved for 1989, 93, 3360–3363. (photo)stability of Fe(II) is unclear (38). It is of interest to (15) Chebbi, A.; Carlier, P. Carboxylic acids in the troposphere, totally identify the dissolved organic matters in rainwaters, occurrence, sources, and sinks: a review. Atmos. Environ. 1996, 30, 4233–4249. particularly dicarboxylic acids, to understand the speciation (16) Garcia-Lopez, E.; Marci, G.; Serpone, N.; Hidaka, H. Photoas- of dissolved iron in carbon-rich troposphere, although the sisted oxidation of the recalcitrant cyanuric acid substrate in correlation of Mal with Fe(II) against (photo)oxidation in aqueous ZnO suspensions. J. Phys. Chem. C 2007, 111, 18025– rainwater remains obscure so far because rainwater may have 18032. high concentrations of other oxidants generated via non metal (17) Tamura, H.; Goto, K.; Yotsuyanagi, T.; Nagayama, M. Spectro- photometric determination of iron(II) with 1,10-phenanthroline redox pathways. in the presence of large amounts of iron(III). Talanta 1974, 21, Despite that the concentration of Fe(III) and Mal in this 314–318. study are higher than those found in most atmospheric (18) Hatchard, C. G.; Parker, C. A. A New sensitive chemical waters, which usually contain micromolar to nanomolar actinometer II. Potassium ferrioxalate as a standard chemical concentrations of Fe and Mal, the ratios of Fe to Mal in our actinometer. Proc. Royal Soc. London, Ser. A 1956, 235, 518– - 536. study (1:1 1:10) should be representative of those in (19) Hug, S. J.; Laubscher, H-U.; James, B. R. Iron(III) catalyzed rainwater, fog, snow, and cloud waters (see Table 3 in ref (1) photochemical reduction of chromium(VI) by oxalate and citrate and Table S4). Concentrations of dissolved iron up to 200 in aqueous solutions. Environ. Sci. Technol. 1997, 31, 160–170. µM (greater than these in the present study) have also been (20) Weil, J. A.; Bolton, J. R.; Wertz, J. E. Electron Paramagnetic Resonance: Elementary Theory and Practical Applications; Wiley: deemed environmentally relevant in some cases (2). There- New York, 1994; p 498. fore, our findings in this study would be of great significance (21) Jørgensen, C. H. R. Inorganic Complexes; Academic Press: inevaluatingphotochemicalcyclingofironinthetroposphere. London & New York, 1963; p 97.

VOL. 44, NO. 1, 2010 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 9 267 中国科技论文在线 http://www.paper.edu.cn

(22) Pandey, R. N.; Smith, W. M. Carboxylate complexing of iron(III). (30) Mansano-Weiss, C.; Cohen, H.; Meyerstein, D. Reactions of 2+ I. The formation of monoacetatoiron(III) in aqueous solution: peroxyl radicals with Fe(H2O)6 . J. Inorg. Biochem. 2002, 91, equilibria and kinetics. Can. J. Chem. 1972, 50, 194–200. 199–204. (23) Deutsch, F.; Hoffmann, P.; Ortner, H. M. Field experimental (31) Morimoto, J. Y.; Degraff, B. A. Photochemistry of copper investigations on the Fe(II)-and Fe(III)-content in cloudwater complexes. The copper(II) malonate system. J. Phys. Chem. 1975, samples. J. Atmos. Chem. 2001, 40, 87–105. 79, 326–331. (24) Okada, K.; Kuroki, Y.; Nakama, Y.; Arakaki, T.; Tanahara, A. (32) Wu, C-H.; Sun, L. Z.; Faust, B. C. Photochemical formation of Wavelength dependence of Fe(II) photoformation in the water- copper(I) from copper(II)-dicarboxylate complexes: effects of soluble fraction of aerosols collected in Okinawa, Japan. Environ. outer-sphere versus inner-sphere coordination and of quench- Sci. Technol. 2006, 40, 7790–7795. ing by malonate. J. Phys. Chem. A 2000, 104, 4989–4996. (25) Lopes, L.; de Laat, J.; Legube, B. Charge transfer of iron(III) (33) Sun, L. Z.; Wu, C-H.; Faust, B. C. Photochemical redox reactions monomeric and oligomeric aqua hydroxo complexes: semiem- of inner-sphere copper(II)-dicarboxylate complexes: effects of pirical investigation into photoactivity. Inorg. Chem. 2002, 41, the dicarboxylate ligand structure on copper(I) quantum yields. 2505–2517. J. Phys. Chem. A 1998, 102, 8664–8672. (34) Stumm, W.; Morgan, J. J. Aquatic Chemistry; Wiley-Interscience: (26) Gilbert, B. C.; Hodges, G. R.; Lindsay Smith, J. R.; MacFaul, P.; New York, 1981; p 780. Taylor, P. Photo-decarboxylation of substituted alkylcarboxylic (35) Franch, M. I.; Ayllo´n, J. A.; Peral, J.; Dome`nech, X. Fe(III) acids brought about by visible light and iron(III) tetra(2-N- photocatalyzed degradation of low chain carboxylic acids methylpyridyl) porphyrin in aqueous solution. J. Chem. Soc., implications of the iron salt. Appl. Catal. B: Environ. 2004, 50, Perkin Trans. 1996, 2, 519–524. 89–99. (27) Klebanoff, S. J.; Waltersdorph, A. M.; Michel, B. R.; Rose, H. (36) King, D. W.; Aldrich, R. A.; Charnecki, S. E. Photochemical redox Oxygen-based free radical generation by ferrous ions and cycling of iron in NaCl solutions. Mar. Chem. 1993, 44, 105–120. deferoxamine. J. Biol. Chem. 1989, 264, 19765–19771. (37) Kieber, R. J.; Peake, B.; Willey, J. D.; Jacobs, B. Iron speciation (28) Pignatello, J. J.; Oliveros, E.; MacKay, A. Advanced oxidation and hydrogen peroxide concentrations in New Zealand rain- processes for organic contaminant destruction based on the water. Atmos. Environ. 2001, 35, 6041–6048. Fenton reaction and related chemistry. Crit. Rev. Environ. Sci. (38) Willey, J. D.; Whitehead, R. F.; Kieber, R. J.; Hardison, D. R. Technol. 2006, 36, 1–84. Oxidation of Fe(II) in rainwater. Environ. Sci. Technol. 2005, 39, (29) Behar, D.; Samuni, A.; Fessenden, R. W. Reactions of iron(II) 2579–2585. and titanium(III) with organic radicals. J. Phys. Chem. 1973, 77, 2055–2059. ES901956X

268 9 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 44, NO. 1, 2010