Julie A. Korak

Critical analysis of commonly used fluorescence metrics to characterize dissolved organic matter.

The use of fluorescence spectroscopy for the analysis and characterization of dissolved organic matter (DOM) has gained widespread interest over the past decade, in part because of its ease of use and ability to provide bulk DOM chemical characteristics. However, the lack of standard approaches for analysis and data evaluation has complicated its use. This study utilized comparative statistics to systematically evaluate commonly used fluorescence metrics for DOM characterization to provide insight into the implications for data analysis and interpretation such as peak picking methods, carbon-normalized metrics and the fluorescence index (FI). The uncertainty associated with peak picking methods was evaluated, including the reporting of peak intensity and peak position. The linear relationship between fluorescence intensity and dissolved organic carbon (DOC) concentration was found to deviate from linearity at environmentally relevant concentrations and simultaneously across all peak regions. Comparative analysis suggests that the loss of linearity is composition specific and likely due to non-ideal intermolecular interactions of the DOM rather than the inner filter effects. For some DOM sources, Peak A deviated from linearity at optical densities a factor of 2 higher than that of Peak C. For carbon-normalized fluorescence intensities, the error associated with DOC measurements significantly decreases the ability to distinguish compositional differences. An in-depth analysis of FI determined that the metric is mostly driven by peak emission wavelength and less by emission spectra slope. This study also demonstrates that fluorescence intensity follows property balance principles, but the fluorescence index does not.

Molecular and Spectroscopic Characterization of Water Extractable Organic Matter from Thermally Altered Soils Reveal Insight into Disinfection Byproduct Precursors

To characterize the effects of thermal-alteration on water extractable organic matter (WEOM), soil samples were heated in a laboratory at 225, 350, and 500 °C. Next, heated and unheated soils were leached, filtered, and analyzed for dissolved organic carbon (DOC) concentration, optical properties, molecular size distribution, molecular composition, and disinfection byproduct (DBP) formation following the addition of chlorine. The soils heated to 225 °C leached the greatest DOC and had the highest C- and N-DBP precursor reactivity per unit carbon compared to the unheated material or soils heated to 350 or 500 °C. The molecular weight of the soluble compounds decreased with increasing heating temperature. Compared to the unheated soil leachates, all DBP yields were higher for the leachates of soils heated to 225 °C. However, only haloacetonitrile yields (μg/mgC) were higher for leachates of the soils heated to 350 °C, whereas trihalomethane, haloacetic acid and chloropicrin yields were lower compared to unheated soil leachates. Soluble N-containing compounds comprised a high number of molecular formulas for leachates of heated soils, which may explain the higher yield of haloacetonitriles for heated soil leachates. Overall, heating soils altered the quantity, quality, and reactivity of the WEOM pool. These results may be useful for inferring how thermal alteration of soil by wildfire can affect water quality.

The Case Against Charge Transfer Interactions in Dissolved Organic Matter Photophysics

The optical properties of dissolved organic matter influence chemical and biological processes in all aquatic ecosystems. Dissolved organic matter optical properties have been attributed to a charge-transfer model in which donor-acceptor complexes play a primary role. This model was evaluated by measuring the absorbance and fluorescence response of organic matter isolates to changes in solvent temperature, viscosity, and polarity, which affect the position and intensity of spectra for known donor-acceptor complexes of organic molecules. Absorbance and fluorescence spectral shape were largely unaffected by these changes, indicating that the distribution of absorbing and emitting species was unchanged. Overall, these results call into question the wide applicability of the charge-transfer model for explaining organic matter optical properties and suggest that future research should explore other models for dissolved organic matter photophysics.

Oversimplification of Dissolved Organic Matter Fluorescence Analysis: Potential Pitfalls of Current Methods

O the past 25 years, there has been a significant increase in the use of fluorescence spectroscopy to characterize the physicochemical properties of dissolved organic matter (DOM) in natural and engineered systems. Fluorescence intensity measured across a range of excitation and emission wavelengths produce thousands of data points per sample that are displayed using a three-dimensional contour plot, known as an excitation−emission matrix (EEM) (see Figure 1a). Fluorescence is an attractive analytical approach due to its ease of use, but the complexity associated with the interpretation of an EEM is often oversimplified. In many cases, the use of EEMs has replaced targeted analytical methods (e.g., fractionations) used to assess the chemical composition of DOM, at the expense of certainty regarding the chemical identity of specific constituents. The photophysics of DOM are complex. Only a subset of the molecules that comprise DOM contain functional groups capable of absorbing light in the ultraviolet−visible wavelengths greater than 200 nm (this subset is known as chromophoric DOM (CDOM)). In turn, only a subset of CDOM fluoresces (FDOM) upon relaxation from an excited singlet state. Different models could be invoked to describe the optical properties of DOM, including superposition of individual, noninteracting fluorophores or charge transfer (CT) interactions between electron donor−acceptor complexes. To simplify the analysis of DOM EEMs, several qualitative and quantitative approaches have been developed. One of the first approaches includes peak picking, which compares fluorescence intensity in four wavelength regions commonly observed in DOM samples (see Figure 1b): humic substances (Peaks A and C) and so-called protein peaks (tryptophan (Peak T) and tyrosine (Peak B)). Quantitative procedures to analyze EEMs include fluorescence regional integration (FRI) and parallel factor (PARAFAC) analysis. The FRI method assigns regions of an EEM to specific DOM fractions: humic acid-like, fulvic acid-like, soluble microbial products and aromatic proteins (see Figure 1c). It is important to note that the FRI method defines three regions at excitation wavelengths less than 250 nm, but many commercial fluorometers do not provide correction factors for excitation wavelengths less than 240 nm where lamp output is low, making signals in this area unreliable. PARAFAC decomposes EEMs into n independent components representing groups of fluorophores with similar spectra. Given the complexity of DOM fluorescence, these methods are very attractive to scientists and engineers as they offer the opportunity to relate fluorescence behavior to a reduced number of chemical entities. However, caution is recommended when using these methods for the analysis of DOM composition by f luorescence. Interpretation of fluorescence data needs to consider that FDOM is a very small subset of DOM. Based on reported fluorescence quantum yields for DOM of diverse origins, less than 3% of the photons absorbed (i.e., CDOM) are emitted as fluorescence. Although there are clear trends in behavior between FDOM and DOM as a whole, the limitations of drawing conclusions about an entire system based on f luorescence data alone have to be considered carefully. Therefore, when using fluorescence as a tool to characterize DOM, it has to be understood that the analysis only applies to FDOM and may not be representative of DOM as a whole. The association of EEM regions with operationally defined DOM fractions (i.e., humic acids (HA), fulvic acids (FA) and proteins) is an oversimplification that cannot be supported for aquatic systems. HA and FA are not distinct chemical subgroups in the sense that proteins and carbohydrates are separate groups with different chemistries. FA and HA are operational definitions based on their adsorption to XAD resins and solubility at low pH. Although we know that, in general terms, FA are smaller and more polar than HA, they can be seen as a continuum of similar chemical structures, albeit with different degrees of oxidation. Fluorescence analysis of FA and HA show that both fractions fluoresce in the same regions (Peaks A and C/Regions III and V) (see Figure 1b and c).

Evaluation of optical surrogates for the characterization of DOM removal by coagulation

Optical surrogates (i.e., absorbance and fluorescence) are of interest to monitor drinking water treatment processes due to their potential for implementation as online sensors. This study compares the use of different optical surrogates to model dissolved organic carbon (DOC) removal by coagulation with aluminum sulfate in the dose range of 5 to 120 mg L−1 in 22 source waters with a wide range of water qualities (specific UV absorbance (SUVA254) – 1.0 to 4.0 L mgC−1 m−1). Linear regressions were developed relating percent DOC removal to percent fluorescence decrease at discrete wavelength pairs in an excitation-emission matrix (2029 wavelength pairings). DOC removal was modeled with an average 95% prediction interval between 10.5–15% at all emission wavelengths greater than 375 nm, suggesting that wavelength selection for use in fluorescence monitoring is relatively unimportant. Fluorescence data without inner filter corrections led to a decrease in model performance, but prediction intervals were still between 12–15% at emission wavelengths greater than 400 nm. By comparison, the average prediction interval for an analogous model with UV absorbance at 254 nm was 10.5% (DOCr% = 0.7 UVr%, R2 = 0.91), performing the same as the best possible fluorescence wavelength combinations. Additional modeling found that tracking multiple optical surrogates in tandem does not improve model performance due to correlated independent variables. Raw water optical properties (SUVA254 and fluorescence ratios) modeled DOC removal at a single coagulant dose (~40 mg L−1), but fluorescence indicators did not significantly outperform SUVA254. A PARAFAC model built with 112 fluorescence samples with six validated components demonstrated the range of fluorescence behaviors in the dataset and guided the selection of fluorescence ratios to predict removal based on raw water characteristics. These results demonstrated that fluorescence wavelength is relatively unimportant for online monitoring approaches that measure both raw and clarified samples, but emission wavelength is a driving factor for predicting removal based solely on raw water characteristics.

Temperature Dependence of Dissolved Organic Matter Fluorescence

The temperature dependence of organic matter fluorescence apparent quantum yields (Φf) was measured for a diverse set of organic matter isolates (i.e., marine aquatic, microbial aquatic, terrestrial aquatic, and soil) in aqueous solution and for whole water samples to determine apparent activation energies ( Ea) for radiationless decay processes of the excited singlet state. Ea was calculated from temperature dependent Φf data obtained by steady-state methods using a simplified photophysical model and the Arrhenius equation. All aquatic-derived isolates, all whole water samples, and one soil-derived fulvic acid isolate exhibited temperature dependent Φf values, with Ea ranging from 5.4 to 8.4 kJ mol-1 at an excitation wavelength of 350 nm. Conversely, soil humic acid isolates exhibited little or no temperature dependence in Φf. Ea varied with excitation wavelength in most cases, typically exhibiting a decrease between 350 and 500 nm. The narrow range of Ea values observed for these samples when compared to literature Ea values for model fluorophores (∼5-30 kJ mol-1) points to a similar photophysical mechanism for singlet excited states nonradiative inactivation across organic matter isolates of diverse source and character. In addition, this approach to temperature dependent fluorescence analysis provides a fundamental, physical basis, in contrast to existing empirical relationships, for correcting online fluorescence sensors for temperature effects.

Response to Comment on The Case Against Charge Transfer Interactions in Dissolved Organic Matter Photophysics

Interactions in Dissolved Organic Matter Photophysics I response to our recent study regarding the role of chargetransfer (CT) interactions in dissolved organic matter (DOM) photophysics, Blough and Del Vecchio raise concerns about the use of solvent polarity, viscosity, and temperature to test for the prevalence of CT interactions in DOM. This topic is of great significance to environmental chemistry and engineering due to the ubiquity and reactivity of DOM in these systems. Blough and Del Vecchio argue against the types of potential donor−acceptor complexes occurring in DOM proposed in Figure 1 of McKay et al. (2018), which includes independent, covalently tethered, and conjugated donor−acceptor moieties. The counterargument asserts that the types of donor− acceptor complexes we propose play a limited role, suggesting instead that the model for polydopamine, as proposed by Dreyer and co-workers, also applies to DOM. In this model, structural units composing DOM would form static complexes as a result of H-bonding, π-stacking, and charge-transfer interactions. It is hypothesized that such donor−acceptor complexes in DOM are hindered kinetically and thermodynamically from dissociating into their respective independent moieties. Blough and Del Vecchio state that the dynamics of these static complexes would not be affected by temperature and solvent polarity in the same way as dynamic donor− acceptor complexes, offering an alternative explanation for the lack of change in spectral shape presented in our study. Further, the absence of solvatochromism reported in our study could potentially be due to solvent-protected donor−acceptor complexes. This model hypothesizes that this hydrophobic core is further stabilized by charged outer groups. Although we welcome the criticism and further discussion, we disagree with the presented counterarguments. First, the model for the primary structure of polydopamine proposed by Dreyer et al. is controversial within the literature. Specifically, a subsequent study argued that polydopamine monomers are covalently linked, as opposed to being held together by noncovalent interactions (e.g., chargetransfer). The merits of these individual studies are beyond the scope of this response, but it is noteworthy that there is debate over the structure of a material that is synthesized from a known monomer unit (i.e., dopamine), whereas the structural units in DOM are much more heterogeneous and poorly characterized. In addition, the isolation of donor and acceptor moieties from solvent is inconsistent with other lines of evidence used to support a CT model. For example, putative acceptor and donor moieties in the CT model, aromatic ketones/aldehydes and phenols/polyphenols/alkoxy phenols, respectively, exhibit aqueous phase photochemistry. If such CT complexes are protected from the solvent, they would also be expected to be protected from participating in bimolecular reactions in the aqueous phase. Furthermore, according to the CT model, reduction of DOM with sodium borohydride would result in DOM with substantially lower molecular weight due to disruption of these donor−acceptor interactions. This hypothesis conflicts with high-pressure size exclusion chromatography measurements published recently for DOM treated with sodium borohydride. In addition, it is difficult to reconcile how an anionic reductant, such as borohydride, but not neutral solvent molecules, could access such complexes in a static, anion-encircled hydrophobic microenvironment. Another remaining question is how solvents of such different polarity could have no effect on DOM spectra, when pH so readily affects the optical properties of DOM, as reported by Blough and Del Vecchio. In the referenced pH experiments, the CT model holds that increased pH deprotonates phenolic donors, which leads to increased charge-transfer excitation. However, for donor−acceptor pairs to be pH-sensitive, the donors would have to be solvent accessible, contrary to the proposed CT model. These observations demonstrate that the evidence presented in support of a CT model (i.e., reactivity with borohydride and pH-sensitive spectra) also contradicts the provision that donor−acceptor complexes reside in a solventinaccessible microenvironment. A final argument against solvent-inaccessible chromophores and fluorophores is that fluorescence quantum yield was significantly changed by solvent polarity at all excitation wavelengths measured (Figure 3c in manuscript), extending well into the visible wavelength range. This change in fluorescence intensity with solvent polarity across the DOM absorption spectrum indicates that these fluorophores are solvent-accessible. Lastly, Blough and Del Vecchio assert that the formation of particulates for soil humic acids in tetrahydrofuran supports the hypothesis that donor−acceptor moieties are in a solvent inaccessible hydrophobic microenvironment. The data presented in Figure 4 of our study contradicts this view. A solvatochromic shift in fluorescence spectra of >20 nm was observed for soil humic acids in acetonitrile and a soil fulvic acid in both acetonitrile and tetrahydrofuran at a range of excitation wavelengths. We hypothesized that this solvatochromism in the fluorescence spectra could be due to exciplex formation, but not ground state donor−acceptor complexes due to the lack of solvatochromism in the corresponding absorbance spectra. This result gives us confidence that solvent polarity would affect the dynamics of donor−acceptor complexes present in the other isolates investigated, if they were present. Furthermore, in the apolar organic solvent systems investigated, carboxylates would be protonated, removing a speculated stabilizing force, and thus encouraging complex dissociation. Lastly, the view that the low solubility of soil humic acids in tetrahydrofuran is evidence for a hydrophobic microenvironment does not fit our observation that these materials dissolved f irst and then, after 24 h, showed evidence of precipitates (Supporting Information section 3.4). The hypothesis that the stabilization by charged moieties controls