Photooxidation-Induced Changes in Optical, Electrochemical and Photochemical 1 Properties of Humic Substances

29 Three dissolved humic substances (HS), two aquatic fulvic acids and one soil humic acid 30 were irradiated to examine the resulting changes in HS redox and photochemical properties, the 31 relationship between these changes, and their relationship to changes in the optical properties. 32 For all HS, irradiation caused photooxidation as shown by decreasing electron donating 33 capacities. This was accompanied by decreases in specific UV absorbance and increases in the 34 E2/E3 ratio (254 nm absorbance divided by 365 nm). In contrast, photooxidation had little effect 35 on the samples’ electron accepting capacities. The coupled changes in optical and redox 36 properties for the different HS suggest that phenols are an important determinant of aquatic HS 37 optical properties and that quinones may play a more important role in soil HS. Apparent 38 quantum yields of H 2 O 2 ,  OH, and triplet HS decreased with photooxidation, thus demonstrating 39 selective destruction of HS photosensitizing chromophores. In contrast, singlet oxygen ( 1 O 2 ) 40 quantum yields increased, which is ascribed to either decreased 1 O 2 quenching within the HS 41 microenvironment or the presence of a pool of photostable sensitizers. The photochemical 42 properties show clear trends with SUVA and E2/E3, but the trends differ substantially between 43 aquatic and soil HS. Importantly, photooxidation produces a relationship between the 1 O 2 44 quantum yield and E2/E3 that differs distinctly from that observed with untreated HS. This 45 suggests that there may be watershed-specific correlations between HS chemical and optical 46 properties that reflect the dominant processes controlling the HS character. 47


Introduction
Dissolved organic matter (DOM) is a ubiquitous component of natural surface waters 50 produced by transformation of plant and plankton-derived precursor molecules. It comprises moderately hydrophobic aromatic polyelectrolytes of variable molecular weight (100's to 1000's 52 of g/mol) (1,2) and plays an important role in the biogeochemistry of aquatic environments. For 53 example, microorganisms use DOM as a source of C and N and as an electron shuttle in 54 anaerobic respiration (3). DOM also plays important roles in pollutant dynamics, for instance by 55 sorbing organic contaminants and chelating trace and heavy metals (4,5). Absorption of sunlight 56 by aromatic chromophores in DOM (1,6) leads to formation of reactive oxygen species (ROS) 57 including singlet oxygen ( 1 O2), hydrogen peroxide (H2O2), and hydroxyl radical (OH) (7,8). 58 DOM triplet states ( 3 DOM * ) are both important precursor species for many of these ROS and 59 strong oxidants in DOM-sensitized photoreactions (7)(8)(9). Together, these photooxidants play a 60 critical role in the redox speciation of trace metals (10)(11)(12), transformation rates of organic 61 contaminants (9,(13)(14)(15), and solar inactivation of pathogens (16). 62 Absorption of sunlight also leads to DOM photobleaching (destruction of chromophores), 63 photooxidation, and production of low molecular weight organic compounds and inorganic 64 species such as CO and CO2 (17)(18)(19)(20)(21)(22)(23). These processes may involve the loss of specific 65 functional moieties and lead to changes in the physicochemical and optical properties of DOM. 66 For example, lignin phenols disappear rapidly in the early stages of photooxidation (24)(25)(26). 67 Other studies have used FT-ICR-MS and 13 C NMR spectroscopy to show that DOM loses 68 aromatic groups during photooxidation (27,28). Concomitant changes in DOM optical 69 properties are consistent with a loss of DOM aromaticity (% aromatic C by 13 C NMR), including 70 decreases in specific UV absorbance (SUVA, absorbance per mg-C) and fluorescence intensity, 71 and increases in spectral slope and the E2/E3 ratio (ratio of the absorbance at 254 to 365 nm) 72 (29)(30)(31)(32)(33)(34)(35). 73 fulvic acids (FAs) and, for contrast, a soil humic acid (HA). Changes in the absorption spectra 97 and apparent quantum yields for the photooxidants 1 O2, H2O2, OH, and triplet HS ( 3 HS * ) were 98 measured as a function of irradiation time. The extent of photooxidation was quantified by 99 monitoring changes in EDC and EAC. Spectroscopic data were also used to examine whether 100 correlations between optical and photochemical properties for photooxidized HS are consistent 101 with reported correlations for native DOM isolates (36,37 Solution volumes (determined gravimetrically) and pH were measured at the beginning 128 and end of each irradiation period. After each period, small amounts of water were added to 129 replace evaporative losses (always < 0.2 mL) and small volumes of 1 M NaOH were added to re-130 adjust the pH to 7.0 (the pH never fell below 6.5). At selected intervals, aliquots of solution 131 were removed for analysis and experimentation. To allow for intra-HS redox equilibration, these  Optical Properties and Absorbed Energy. Absorbance spectra were collected in 1 cm 138 quartz cuvettes on a Cary 100 spectrophotometer (Varian) using 1 nm slits and phosphate buffer 139 as a blank. Prior to measuring absorbance spectra, HS solutions were diluted in 5 mM phosphate 140 buffer (pH 7.0) by a factor of three (SRFA and NAFA) or ten (ESHA) to ensure that 141 measurements fell into the linear range of the instrument. Optical parameters, including the E2/E3 ratio and specific UV absorbance at 280 nm (SUVA280) were calculated from the 143 measured spectra as detailed in the Supporting Information. 144 Absorption and lamp emission spectra were used to determine the energy absorbed 145 between 300 and 500 nm during irradiation (details in Supporting Information). This wavelength 146 range was chosen for its importance to HS photochemistry (20,34,43,47,(51)(52)(53)(54)(55)(56). The conclusions 147 drawn in this study are based on relative changes and change little by setting the long wavelength 148 cutoff to 400 or 450 nm.

151
Germany). Anoxic buffer solutions were used as described previously (44,45,57). HS solutions with the amount of energy absorbed (Fig. 2). In contrast, much less photobleaching occurred for 176 ESHA even though it absorbed approximately twice the 300 to 500 nm energy ( Fig. 1), and the 177 changes in SUVA280 and E2/E3 were much smaller and non-linear. These data suggest that 178 ESHA was more resistant than the aquatic FAs to photooxidation. However, it is possible that 179 this difference in resistance was exaggerated by inner-filtering in the ESHA solution. A 180 substantially higher fraction of the energy absorbed by ESHA lay in the visible (ca. 90%) than 181 for the aquatic FAs (ca. 70%), and visible irradiation is known to cause much less efficient 182 photobleaching (34,35,64). Thus, the difference between the photobleaching efficiencies for 183 ESHA and the aquatic FAs is probably not as large as implied by these data.

184
The difference in magnitude of the optical changes notwithstanding, all samples displayed  (71)). The EDC of DOM also varies with Eh and pH in 203 ways comparable to low molecular weight phenols (44). Finally, the EAC of DOM correlates 204 well with its % aromaticity (44), and DOM accepts electrons over a range of reduction potentials 205 consistent with quinones as major electron acceptors (45). for the aquatic FAs than for soil HA are consistent with prior reports that aromatic-C and EDC 226 are well-correlated for diverse aquatic HS samples but not for different soil-derived HS (44).

227
Given that phenolic moieties constitute much more of the aromatic-C in the aquatic FAs than 228 ESHA (71,72) (Supporting Information), it appears likely that phenols are a major determinant 229 of both EDC and SUVA280 in aquatic HS. In contrast, SUVA280 appears to vary more with EAC 230 for ESHA (Fig. 4), which could imply that quinones are notable contributors to SUVA280 in soil 231 HA. However, this conclusion is tempered by the small and irregular changes in EAC with 232 photooxidation.
To examine the relationship of E2/E3 to EDC and EAC, a larger data set was analyzed 234 that included various untreated aquatic and soil HS (44) in addition to the photooxidized 235 samples; missing E2/E3 ratios were measured. For aquatic HS, E2/E3 shows an inverse 236 relationship to EDC, whereas for the soil HA these data are more scattered (Fig. 4). 237 Furthermore, E2/E3 appears to be more sensitive to changes in EDC (steeper slope) for the 238 photooxidized aquatic FAs than for untreated HS isolates. These observations suggest that (i) for 239 aquatic HS, E2/E3 depends strongly on phenol content and (ii) that this dependence is 240 pronounced in samples undergoing photooxidation. For soil HS, on the other hand, E2/E3 seems 241 to be more dependent on EAC (Fig. 4), whereas this relationship is a bit more scattered for 242 aquatic HS, and there is no obvious relationship for the photooxidized FAs. These differences 243 suggest that quinones contribute more to CT absorbance in soil HS than in aquatic HS.

244
Furthermore, in the aquatic FAs undergoing photooxidation, the extent of CT absorbance seems 245 to be independent of quinones. This conclusion is consistent with recent evidence that quinones 246 are not major determinants of CT absorbance in aquatic HS (46,70). Φ is a more fundamental parameter with broader general applicability to photochemical 254 modeling. Figure 5 shows how 1O2, ΦH2O2, ΦOH and fTMP vary with irradiation. For all HS, 255 the quantum yields of H2O2, OH, and 3 HS * decreased, but, in stark contrast, Φ1O2 increased.

256
Attributing physical causes to these trends requires recognition that the quantum yields 257 are "apparent" because they may be affected by a variety of secondary phenomena. For instance, 258 ΦH2O2 is a function of the primary quantum yield of O2and the relative rates of uncatalyzed 259 versus HS-catalyzed O2dismutation (36). Because uncatalyzed dismutation is apparently the  In the case of ΦOH, the TPA probe detects both free OH and other, lower energy 267 hydroxylating species (59). Here, we do not distinguish between these hydroxylating species.  The parameter used to assess 3 HS * formation, fTMP, depends on both the primary quantum 277 yield of 3 HS * and the rate constant for reaction between 3 HS * and TMP (61). Notably, HS do not 278 inhibit TMP oxidation (58). Decreases in fTMP thus reflect either a general decrease in 3 HS * precursor chromophores or a decrease in a specific 3 HS * pool that reacts rapidly with TMP. In 280 either case, the observed decreases in fTMP indicate that photooxidant production efficiency 281 decreases with irradiation, in agreement with the ΦOH results. For TMP oxidation, the 3 HS * 282 precursors are believed to be mainly aromatic ketones and possibly quinones (46,61). The 283 present data provide divergent evidence for the role of quinones. For instance, the aquatic FAs 284 have much lower EAC and fTMP than ESHA, consistent with excited state quinones as oxidants of 285 TMP. However, this conclusion does not seem to be compatible with the fact that decreases in 286 fTMP were not paralleled by comparable decreases in EAC.

287
Distinct from the other quantum yields, 1O2 increased with irradiation. Note that the 288 FFA probe detects 1 O2 in the bulk aqueous phase that has escaped the HS microenvironment 289 without being quenched therein (75). It is generally accepted that 1 O2 is produced by energy 290 transfer from 3 HS * to O2. Reports by Halladja et al. (76) and Sharpless (47) indicate that a high 291 degree of overlap exists between the 3 HS * pools that produce 1 O2 and those that oxidize TMP. If 292 true, a simple way to reconcile the increases in 1O2 with the decreases in fTMP is to hypothesize 293 that irradiation decreases the efficiency of 1 O2 quenching by HS. Thus, even if the yield of 3 HS * 294 decreases, higher 1 O2 yields could be observed if 1 O2 is less effectively quenched within the 295 DOM microenvironment. The loss of EDC is consistent with this hypothesis because quenching 296 of 1 O2 by HS probably occurs by an electron transfer mechanism (77), which is expected to 297 become less effective as electron donors in HS are destroyed. Also consistent with this view is a 298 recent report that wastewater DOM 1O2 increases with chemical oxidation by both ozone and 299 chlorine, which destroy electron rich groups selectively and non-selectively, respectively (41).   (36), and Peterson et al., who studied whole water from Lake Superior and its tributaries (37). 326 Peterson et al.'s results are shown as the reported linear trend because the data include many 327 samples with E2/E3 values well above the range used in Figure 6. For the aquatic FAs 328 undergoing photooxidation, the directions of the trends in H2O2 and 1O2 with E2/E3 agree with 329 previous reports (36,37). The H2O2 results display good quantitative agreement with those of 330 Dalrymple et al. (36), suggesting that E2/E3 may be a robust predictor of H2O2. However, 1O2 331 is less sensitive to changes in E2/E3 for the photooxidized FAs than for untreated DOM isolates.

332
Notably, the relationship between 1O2 and E2/E3 for photooxidized FAs resembles that of the 333 Lake Superior samples much more than that of the untreated DOM. This suggests that 334 photobleaching is a major control on DOM optical properties and photochemistry in Lake  With EDC, a weak decreasing trend in 1O2 is observed, while H2O2 increases sharply, 339 albeit with different slopes for aquatic HS and soil HS (Fig. 6). The trends can be explained in 340 terms of photophysical concepts expounded previously (36,47,74