Photolysis of Dissolved Organic Matter over Hematite Nanoplatelets

Solar photoexcitation of chromophoric groups in dissolved organic matter (DOM), when coupled to photoreduction of ubiquitous Fe(III)-oxide nanoparticles, can significantly accelerate DOM degradation in near-surface terrestrial systems, but the mechanisms of these reactions remain elusive. We examined the photolysis of chromophoric soil DOM coated onto hematite nanoplatelets featuring (001) exposed facets using a combination of molecular spectroscopies and density functional theory (DFT) computations. Reactive oxygen species (ROS) probed by electron paramagnetic resonance (EPR) spectroscopy revealed that both singlet oxygen and superoxide are the predominant ROS responsible for DOM degradation. DFT calculations confirmed that Fe(II) on the hematite (001) surface, created by interfacial electron transfer from photoexcited chromophores in DOM, can reduce dioxygen molecules to superoxide radicals (•O2–) through a one-electron transfer process. 1H nuclear magnetic resonance (NMR) and electrospray ionization Fourier-transform ion cyclotron resonance mass spectrometry (ESI-FTICR-MS) spectroscopies show that the association of DOM with hematite enhances the cleavage of aromatic groups during photodegradation. The findings point to a pivotal role for organic matter at the interface that guides specific ROS generation and the subsequent photodegradation process, as well as the prospect of using ROS signatures as a forensic tool to help interpret more complicated field-relevant systems.


S3
photodegradation process, the temperature of the solution was maintained at room temperature using jacketed glassware and a circulating water-cooling system.The samples were taken out at predetermined time intervals from the vessel with a syringe and filtered with a 0.22 μm PVDF membrane filter.
Text S4.EPR Spectroscopy.A capillary with ID 0.8 mm and OD 1 mm was used to hold the solution in the EPR cavity with both ends sealed by Critoseal.Kinetic measurements were performed by recording EPR spectrum continually before, during and after illumination with a sweep time of 20.97 s and 16 scans (5.6 min per spectrum) at microwave power of 20 mW.The typical settings for the instrument were: microwave frequency = 9.32 GHz, sweep width = 150.0G, field modulation amplitude = 1.0 G, time constant = 40.96ms.All EPR simulations were performed using EasySpin 5.2. 2 To detect transient radicals, these short-loved species were detected using spin trapping.5-(Diethoxyphosphoryl)-5-methyl-1-pyrroline-N-oxide (DEPMPO)-trapped EPR spectra were obtained because this trap is capable of detecting a variety of oxygen-centered and/or carbon-centered free radicals, such as hydroperoxyl radical (•OOH), hydroxyl radicals (•OH), alkyl radical (•R), alkoxyl radical (•OR), and peroxyl radical (•OOR).Standard control spectra for each of these transient species were collected as reference spectra.
Text S5.XPS Methods.To prepare samples for XPS analyses, DOM-coated HNPs were suspended in deionized water and drop casted on to Si wafers containing a 300 nm layer of thermally grown SiO 2 .
They were then introduced into the load lock chamber and pumped down to ~1 × 10 -8 Torr, then transferred into the analysis chamber operating at a base pressure of 2E-9 Torr.Data acquisition was carried out on an area of 700 m × 300 m at normal take-off angles.High-resolution scans were collected at a pass energy, PE, of 40 eV (Full Width at Half Maximum, FWHM, of Au 4f7/2 of metallic gold at 40 eV PE was 0.8 eV), while survey scans were collected at 160 eV.For the solution-based samples, the supernatant solution was drop casted onto Au-coated Si wafers.The wafers were then transferred into the analysis chamber and analyzed in a similar fashion as described above.All data processing was carried out using CasaXPS software.Gaussian-Lorenztian product forms of lineshapes were used for peak deconvolution, and the FWHM of the C 1s components were constrained to a narrow range of ±0.2 eV.For the solution-based drop-casted DOM extracts, a 'U 3 Tougaard' type background approximation was chosen to model the rising nature of the baseline, while for the DOMcoated HNP samples an iterated Shirley type background approximation was used.
For each of the two DOM samples, high resolution C 1s XPS scans were collected for DOM before and after photolysis, and for DOM adsorbed on HNPs before and after photolysis.For the two DOM samples prior to adsorption on HNPs and prior to photolysis, peak deconvolution of the C 1s region shows that at least six different types of carbon chemical environments, or functional groups, are present.The main reference peak at 285 eV can be attributed to a C-C/C-H chemical state.The feature at ~0.5 eV lower binding energy indicate alkene (C=C) type species.The features at ~1.4 eV and ~2.8 eV higher binding energy were assigned to C-O/C-N and C=O type bonds, respectively, possibly originating from alcohol/amine and carbonyl type species.The feature at ~3.7 eV above the main peak is consistent with ester/carboxylic acid groups (O-C=O), such as from fatty acid or proteinlike structures, and the feature at ~4.5 eV above the main peak is consistent with carbonate (CO 3 ) species on the surface.Text S6.Nuclear Magnetic Resonance Spectroscopy.Centrifugation and filtration (0.22 µm polyvinylidene fluoride (PVDF) membrane, 25 mm syringe filter) were used to pellet/separate the HNPs from the samples.A volume of 7.0 mL of each sample's supernatant was aliquoted, lyophilized, and reconstituted in 0.550 mL of D 2 O (12.7 × concentrated) with 2 % (v/v) of Chenomx internal S5 standard (IS) solution (5.02 mM sodium trimethylsilylpropanesulfonate (DSS-d6) and 0.2% NaN 3 in 100% D 2 O) prior to transferring to 5mm Wilmad glass NMR tubes for measurement.Measurements were conducted at a regulated temperature of 25 °C using a Bruker Avance III spectrometer operating at a field strength of 17.6 T ( 1 H ν 0 of 750.24 MHz) and equipped with a 5mm Bruker TCI/CP HCN (inverse) cryoprobe with Z-gradient.The W5 WATERGATE ('zggpw5') water suppression pulse sequence 3 was used to acquire all 1D 1 H spectra. Experimental parameters employed included a spectral window of 19.0 ppm, a calibrated 90° pulse width, an acquisition time of 1 s (32k total points), a 45 s relaxation delay, and a total of 128 transients were coadded for each spectrum.The binomial water suppression delay was set to 91 µs thus placing the first nulls outside the signal region at 11.8 and -2.4 ppm.Post-acquisition processing included zero-filling to 64k points, exponential multiplication (1 -5 Hz line-broadening), and semi-automatic, multipoint smooth segments baseline correction using MestReNova version 14.0.1. 1H spectra were referenced to 0.0 ppm from the methyl signal in the internal DSS-d6 and limited spectral assignments were made using NMRSuite 8.5 Professional (Chenomx) as well as reference to the HMDB and BMRB spectral databases. 4,5 eNova 14.0.1 was used for deconvolution of select peaks for quantitation (acetate and formate) and integration of regions corresponding to key substructures/functional group types based on the chemical shift limits used in Hertkorn et al. 6 1D 1 H NMR spectroscopy was used to monitor changes in DOM composition by integrating and comparing five specific signal regions representing key organic substructural features using slightly With the addition of HNPs, the in situ EPR also showed that the intensity increased in the first 11 minutes and then reached steady-state for DOM WS (Figure 1D) and DOM MS (Figure 1E).For DOM WS in the presence of HNPs, spin adduct DEPMPO-OR dominated at equilibrium and DEPMPO-R was negligible (Figure S4C in the SI).When SOD was added to the system, DEPMPO-OOR was observed with a much low intensity.The •OR species was not observed with the addition of SOD, indicating that •OOH was the primary free radical while •OR was a secondary free radical (•OOH + R → •OR).As for DOM MS with HNPs (Figure 2E), the intensity increased in the first 17 minutes before plateauing, with DEPMPO-OOR and DEPMPO-OOH dominant at equilibrium and a small amount of DEPMPO-R (Figure S4D in the SI).Therefore, we conclude that hematite nanoplates promoted the generation of •OOH during the photolysis of DOM.     a a H is the hyperfine coupling constants for proton.
b a N is the hyperfine coupling constants for nitrogen.
c a P is the hyperfine coupling constants for phosphorus.
d G is Gauss, the unit of hyperfine constant.
e lw is a peak-to-peak (PP) linewidths in mT.
f k exchange is the exchange rate constant.

modified 1 H
chemical shift , δ H (ppm), ranges based on Hertkorn et al6 .The first of the five regions corresponded to pure aliphatic protons (CCCH) from 0.50 -1.85 ppm.The next region, between 1.85 -3.10 ppm, encompasses functionalized aliphatics (XCCH, X = O, N, S, or NCH of primary and secondary amines).The region from 3.1 -4.5 ppm includes oxygenated (OCH) resonances as in carbohydrates, phenolic methoxy protons, as well as HC α protons of most amino acids.The region between 4.5 and 5.3 ppm is not included in the integrations due to the suppressed residual proton signal in the deuterated solvent (HDO).The region from 5.3 -7.0 ppm contains anomeric proton resonances from carbohydrates (O 2 CH) at the lower end of the range and olefinic/unsaturated (HC=C) throughout as well as some aromatic resonances, however, the bulk of the aromatic (including heteroaromatics) region is assigned from 7.0 -10.0 ppm.Text S7.FTICR-MS DataAcquisition and Data Analysis.DOM samples were diluted in MeOH at a 1:2 ratio to improve ESI efficiency, and then introduced directly to the ESI source with a fused silica tube (30 μm i.d.) at a flow rate of 3.0 μL/min by an Agilent 1200 series pump (Agilent Technologies, Santa Clara CA).The ion accumulation time (IAT) was adjusted between samples to account for different carbon concentrations.In addition, the needle voltage at +4.4 kV; Q1 was set to 50 m/z; and the heated resistively coated glass capillary was operated at 180 °C.DataAnalysis software (BrukerDaltonics version 4.2) converted raw spectra to lists of m/z values, with S/N threshold set to 7, mass measurement error <1 ppm, and absolute intensity threshold to the default value of 100.Chemical formulas were assigned with S/N >7 and the presence of C, H, O, N, S and P and excluding other elements.Van Krevelen diagrams were constructed to assign compounds to the major biochemical classes (i.e., amino sugar-, lipid-, protein-, lignin-, carbohydrate-, tannin-, unsaturated hydrocarbon-, and condensed aromatic-like) based on the assigned chemical formulas.Boundaries of classes on the van Krevelen diagram were based on the ratios of H to C and O to C. 7Text S8.DEPMPO-trapped EPR spectra of DOM WS photolysis as a function of illumination time (Figure1B) showed that the signal increased until reaching a plateau at 11 minutes.A similar trend was also found for DOM MS photolysis (Figure1C), indicating that the generation of free radicals reached steady state within 11 min.When DOM WS was illuminated, the intensity of spin adducts of S7 DEPMPO-OOR and DEPMPO-OOH increased within the first 11 minutes and plateaued, while DEPMPO-R plateaued at 1/3 of the intensity of DEPMPO-OOR and DEPMPO-OOH (FigureS4Ain the SI).When the SOD was added to DOM WS under illumination, the equilibrium intensity of DEPMPO-OOR and DEPMPO-R both reduced by 2/3, reinforcing that •OOH is the dominant species (FigureS4Ain the SI).For DOM MS , the intensity of all spin adducts also plateaued after 11 minutes, with the intensity of DEPMPO-OOR significantly dropping after the addition of SOD (FigureS4Bin the SI), further confirming •OOH as the dominant species.

Figure S3 .
Figure S3.The optimized simulation spectra (black lines) compared to the experimental spectra of

Figure S4 .
Figure S4.(A) Intensity changes of corresponding ROS in DEPMPO-trapped EPR spectra in the

Figure S5 .
Figure S5.Peak deconvolution of C 1s XPS spectra of dissolved organic matter.(A) C 1s XPS spectra

Table S1 .
EPR parameters of different DEPMPO spin adducts estimated from the best fits to the experimental spectra.

Table S2 .
DOM MS Major substructure types, chemical shift regions integrated, % total integrated area of region, and the % change in the integrated area for regions calculated as 100(A cond1 -A cond2 )/A cond2 .

Table S3 .
DOM WS Major substructure types, chemical shift regions integrated, % total integrated area of region, and the % change in the integrated area for regions calculated as 100(A cond1 -A cond2 )/A cond2 .

Table S4 .
The number of molecules of each chemical class from both energetically favorable and unfavorable compound groups in Wisconsin DOM.The values are mean of three replicates and values in parenthesis are fractions of the number of molecules in each chemical class to the total number of molecules in a sample.

Table S5 .
The number of molecules of each chemical class from both energetically favorable and unfavorable compound groups in Michigan DOM.The values are mean of three replicates and values in parenthesis are fractions of the number of molecules in each chemical class to the total number of molecules in a sample.

Table S6 .
The relative fraction of carbon type of DOM adsorption and photodegradation onto hematite nanoplatelets.