Skip to main content
NCBI home page
ACS AuthorChoice logoLink to ACS AuthorChoice
. 2024 Feb 10;58(7):3437–3448. doi: 10.1021/acs.est.3c09341

Fluorinated Pharmaceutical and Pesticide Photolysis: Investigating Reactivity and Identifying Fluorinated Products by Combining Computational Chemistry,19F NMR, and Mass Spectrometry

Akash P Bhat †, William C K Pomerantz §, William A Arnold †,*
PMCID: PMC10883306  PMID: 38340057

Abstract

graphic file with name es3c09341_0006.jpg

Fluorinated breakdown products from photolysis of pharmaceuticals and pesticides are of environmental concern due to their potential persistence and toxicity. While mass spectrometry workflows have been shown to be useful in identifying products, they fall short for fluorinated products and may miss up to 90% of products. Studies have shown that 19F NMR measurements assist in identifying and quantifying reaction products, but this protocol can be further developed by incorporating computations. Density functional theory was used to compute 19F NMR shifts for parent and product structures in photolysis reactions. Computations predicted NMR spectra of compounds with an R2 of 0.98. Computed shifts for several isolated product structures from LC-HRMS matched the experimental shifts with <0.7 ppm error. Multiple products including products that share the same shift that were not previously reported were identified and quantified using computational shifts, including aliphatic products in the range of −80 to −88 ppm. Thus, photolysis of fluorinated pharmaceuticals and pesticides can result in compounds that are polyfluorinated alkyl substances (PFAS), including aliphatic-CF3 or vinyl-CF2 products derived from heteroaromatic-CF3 groups. C–F bond-breaking enthalpies and electron densities around the fluorine motifs agreed well with the experimentally observed defluorination of CF3 groups. Combining experimental-computational 19F NMR allows quantification of products identified via LC-HRMS without the need for authentic standards. These results have applications for studies of environmental fate and analysis of fluorinated pharmaceuticals and pesticides in development.

Keywords: direct photolysis, photoproducts, 19F NMR, PFAS, density functional theory

Short abstract

Fluorinated byproducts from pesticide and pharmaceutical degradation may be more persistent than the parent compounds. Combining computations with analytical chemistry approaches allows identification and quantification of these fluorinated byproducts.

Introduction

The addition of fluorine to pharmaceutical and pesticide structures has become common due to improved biological function of the molecule upon addition of one or more fluorinated functional groups.1,2 With the carbon–fluorine bond being one of the strongest in organic chemistry,3 these molecules also have increased resistance to thermal and oxidative stresses.4−6 The common fluorine functional groups incorporated in organic compounds are aromatic fluorine (Ar–F), a fluorine substituted heteroaromatic ring (“heteroaromatic fluorine”; Het-F), benzylic-trifluoromethyl (Ar-CF3), trifluoromethyl-substituted heteroaromatic (Het-CF3), aliphatic fluorines (Aliphatic-CFx), O-CF3, and S-CF3.1,4 The heteroaromatic structures encompass a variety of 5- and 6-membered rings, such as pyridine in sulfoxaflor, pyrimidine in saflufenacil, and triazole in sitagliptin. Photolysis is an important process in the fate of pesticides in pharmaceuticals in both natural and engineered aquatic systems.7,8 Product distributions from photolytic reactions can be complex, and persistent byproducts may be of continued concern. For example, several of the fluorinated functional groups in pesticides and pharmaceuticals remain intact during photolysis and oxidative treatment processes, forming multiple fluorinated persistent byproducts that could be long-lived in surface and drinking waters.9 More than 100 different fluorinated photoproducts have been identified from fluorinated pesticides and pharmaceuticals in sunlit surface waters or during photolysis-based degradation processes.9−15 Photoproducts may arise from small structural changes with the parent compound retaining the fluorinated motif. More extensive degradation leads to formation of smaller compounds like trifluoracetic acid (TFA)16 or fluoride (defluorination).12 Two major challenges are the identification of all fluorinated byproducts and understanding the variations in the extent of defluorination for different fluorinated motifs.

The most common method used for identification of fluorinated degradation products of pesticides and pharmaceuticals is liquid chromatography with nontargeted high resolution mass spectrometry (LC-HRMS).17 It was reported that using only LC-HRMS to quantify fluorinated compounds can result in fluorine mass balances falling short, with up to 90% of fluorine being unaccounted for.17−20 Recent studies on fluorinated product formation and identification during photolysis and advanced treatment processes have reported more complete product identification by complementing LC-HRMS with quantitative 19F-nuclear magnetic resonance spectroscopy (NMR) analysis.9−11,21,22 An NMR spectrum contains all fluorinated product peaks, and the chemical shifts provide insight into the identity of the functional group and its chemical environment. With the broad range of fluorine NMR shifts (∼400 ppm) and the 100% abundance of 19F isotope of fluorine in the environment, 19F NMR has been reported to be a highly suitable method to detect fluorinated compounds and products.23 Also, by increasing the number of scans four times, the signal-to-noise ratio is doubled, allowing minor products to be detected. Using 19F NMR to quantify the products formed assists in evaluation of chemical structures identified using the semiquantitative (at best) LC-HRMS data.10 For example, the photolysis of fluoxetine led to formation of norfluoxetine, TFA, and fluoride.10 On the other hand, the 19F NMR spectra after photolysis of sulfoxaflor showed that all product peaks formed were similar in shift to the parent peak, indicating that all products were closely related to each other and the parent structurally, making evaluation of product data obtained via LC-HRMS simpler, faster, and more precise.9 In cases such as this, however, where multiple 19F NMR peaks are within several ppm the parent peak, it is difficult to assign these NMR peaks to the specific structures identified by LC-HRMS to have modifications remote to the fluorinated functional group of interest, which hinders quantifying the relative amounts of different products.24

The 19F NMR shifts for common products like fluoride (F–), trifluoroacetate (TFA), and difluoroacetate (DFA) are known and are easily identified and quantified. For new fluorinated product structures formed via degradation of pesticides and pharmaceuticals, it is difficult to predict the upfield or downfield movement of 19F NMR shifts with respect to the parent compounds, thereby potentially making accurate matches of products identified using LC-HRMS with the NMR data challenging and complicated. Major products may go unidentified, unless these products are matched to their 19F NMR shifts. With the unavailability of mass-labeled standards for quantitative mass spectrometry for these newly identified fluorinated products and the difficulty in separating products using preparative scale chromatography, computational calculations for 19F NMR shifts could pave the way for precise and accurate identification of products leading to matches between experimental 19F NMR spectra and LC-HRMS product structures.

Computational methods can be used to obtain 19F NMR shifts for fluorinated organic compounds, even in the absence of standards. Previous studies have used computational methods, such as density functional theory (DFT), to predict 19F NMR spectra for smaller model fluorinated organic compounds and phenols as well as per- and polyfluoroalkyl substances (PFAS) with high prediction accuracy.25,26 Studies have also computationally predicted proton and carbon NMR shifts for organic compounds.27,28 Computationally predicting shifts for not only the parent, but also all the (potential) product structures, is the most practical next step for product identification and detection.29

It is also important to understand the driving factors that lead to the formation of stable fluorinated products versus defluorination. Experimental studies have shown differences in the extent of defluorination from different fluorine motifs.9 While Aryl or Het-F groups are generally not a concern with respect to stable fluorinated byproducts via photolysis, CF3 groups have been shown to form TFA along with persistent products retaining the CF3 groups.9−11 Previous studies on photolysis have reported that Het-CF3 and Aliphatic-CF2/3 motifs are retained in the product structures.22,9−11 An increased application of heteroaromatic fluorine motifs in compounds due to their improved metabolic stability coupled with the environmental stability (and retention) of Het-CF3 motifs in surface waters, drinking waters, and wastewaters upon photolysis and advanced treatment conditions could ultimately lead to an unparalleled increase in persistent and stable fluorinated byproducts in the environment.4,30

With several new fluorinated compounds being developed with CF3 groups, experimental determination of the stability of these fluorine groups in products can be difficult. Computational analyses, like electron density mapping, have been used to study active sites in the molecule for bond breaking or radical attack.31−33 Graphical computational analysis and Mulliken charges show electron-deficient and electron-rich sites in the molecule, which have been demonstrated to be active sites for nucleophilic and electrophilic attack, respectively.34,35 Additionally, bond dissociation enthalpies have often been reported for C–C and C–O bonds in organic contaminants to compare and predict products.36,37 Enthalpy comparison for different bonds have assisted in understanding the sites for bond cleavage and mechanisms during the rate-determining steps.36,38,39

The goal of this work is to develop a combined computational-experimental protocol for the identification, prediction, and quantification of fluorinated products from photolysis of fluorinated pesticides and pharmaceuticals. Computational calculations were performed for fluorinated compounds to obtain the isotropic shielding values (shifts), and the best level of theory and solvation effects were evaluated by comparing them with experimental 19F NMR shift values. The calculation method was further used to calculate shifts for multiple experimentally observed as well as hypothesized products to build a combined product identification protocol with experimental 19F NMR and LC-HRMS measurements and computational 19F NMR calculations. A total of 9 fluorinated pharmaceuticals or pesticides were chosen from previous studies to determine the accuracy of experimental product identification using computational methods. Additionally, bond breaking enthalpies and electron densities were calculated for several compounds to compare defluorination trends with experimental results.

Experimental Section

Chemicals

The fluorinated compounds chosen for the computations (19F NMR parent compounds as well as product identification) were saflufenacil, sulfoxaflor, penoxsulam, fluoxetine, 4-nitro-3-trifluoromethylphenol (TFM), flecainide, florasulam, voriconazole, and fluroxypyr. Additional compounds chosen for parent compound 19F NMR computations were hexafluorobenzene (HFB), trifluoroacetic acid (TFA), favipiravir, 2-fluorophenol (2-FP), 3-FP, 4-FP, 2,6-difluorophenol (2,6-DFP), 3,5-DFP, enrofloxacin, celecoxib, sitagliptin, broflanilide, flucarbazone, and trifluralin. Except for saflufenacil, celecoxib, and sitagliptin, all experimental shifts were obtained from previous studies.9−11,29 Sources and purities of all the chemicals used in this study are given in the Supporting Information (SI).

Photolysis, Experimental 19F NMR, and LC-HRMS

Experimental results from photolysis of sulfoxaflor, penoxsulam, fluoxetine, TFM, flecainide, florasulam, voriconazole, and fluroxypyr, including 19F NMR spectra and LC-HRMS product identification, were used from our previous studies.9−11,22 The photolysis in our prior studies was performed with a mercury vapor lamp, solar simulator, and LEDs with 255, 275, 308, 365, and 405 nm peak wavelengths. Detailed methodology for photolysis is available in the prior works. For this study, photolysis of sitagliptin (10 μM) and celecoxib (15 μM) was performed using a 275 nm peak wavelength narrow bandwidth UV-LED reactor, and saflufenacil (10 μM) was photolyzed under 308 nm irradiation (which resulted in the largest number of products in our prior study).11 The solutions were photolyzed until the parent compound concentration was reduced by four natural-log (ln) units for saflufenacil and celecoxib and three ln units for sitagliptin. Details of the experimental procedures and the high pressure liquid chromatography method used to monitor reaction progress are in SI Section 2. 19F NMR analysis was performed on the initial and photolyzed solutions to determine the extent of defluorination and the number of products formed; intermediate products formed during the photolysis were not analyzed.

To determine the accuracy of computations for product prediction, experimental products from the 275 nm UV-LED photolysis for saflufenacil were separated using a preparative Supelco Discovery RP-Amide C16 column, 10 cm × 21.2 mm, 5 μm particle size column with a mobile phase of 20:80 ACN:unbuffered H2O at a flow rate of 3.0 mL/min with an injection volume of 100 μL and detection wavelength of 240 nm. Eight fractions of approximately 2 min duration were collected at least three times and concentrated to concentrations >0.1 μM using a vacufuge (Eppendorf-5301) from 25 to 30 mL to approximately 800 μL for 19F NMR and LC-HRMS analysis. The 2 min collections were sufficient to isolate peaks observed on the HPLC chromatogram.

A 600 MHz Avance Neo NMR for 19F NMR spectral acquisition was used to analyze and quantify the fluorinated molecules in the samples with concentrations >0.1 μM and a resolution of 0.1 ppm. The method was the same as our previous studies.9 Hexafluorobenzene (HFB) with a reference chemical shift (−164.9 ppm) was used as an internal standard for quantification. Detailed explanations of the 19F NMR methodology, acquisition parameters, and analyses are provided in SI Section 3.

A Velos HRAM LC MS Orbitrap system equipped with a Luna C18 nanocolumn was used for LC-HRMS (tandem MS2) analysis. The detection range was from 50 to 800 m/z. Compound Discoverer (Thermo Fisher Scientific) was used to analyze the spectra, and analysis of formulas were largely restricted to those containing fluorine. Detailed procedures, acquisition parameters, and product identification protocols are discussed in SI Section 4. The results from our previous experiments were used in the current study, but the data were reanalyzed using Compound Discoverer to identify additional features and products and additional fragment matches and reevaluate product structures using the m/z values for compounds without fragment matches.

19F NMR Computations

All quantum chemical calculations were performed using Gaussian 16 b.0140 software at the Minnesota Supercomputing Institute. Geometry optimization for 19F NMR calculations was performed with the Becke three parameter Lee–Yang–Parr (B3LYP) functional41−44 and 6-311+G(d,p) basis set with the SMD solvation model45 using water as the solvent, based on previous studies.25,26 Isotropic shielding constants (σ) for fluorine atoms in the molecule (lowest energy conformer, if applicable) were calculated using the Gauge-Independent Atomic Orbital (GIAO) method with the same functional and basis set in the gas phase.21 The effect of the basis set (testing included 6-31+G(d,p), 6-31++G(d,p), 6-311+G(d,p), and 6-311+G(2d,p) with double-ζ (6-31), triple-ζ (6-311), addition of a diffuse function (+), and addition of a d-orbital polarization function (2d)) and including solvation on the NMR calculations were tested. The computational shift (δ in ppm) was calculated using method previously used for polyfluorinated alkyl substances,26 using a hexafluorobenzene (HFB, −164.9 ppm) reference molecule (eq 1).

graphic file with name es3c09341_m001.jpg 1

First, computational shifts for all 33 fluorine motifs from 23 parent compounds with known experimental 19F NMR (also referenced to HFB) shifts were calculated to plot the experimental versus computational NMR shifts. A subset of the molecules is shown in Figure 1. Second, computational NMR shifts were calculated for LC-HRMS identified structures of fluorinated photolysis products previously predicted from data for penoxsulam, saflufenacil, florasulam, sulfoxaflor, voriconazole, flecainide, fluoxetine, fluroxypyr, and TFM.9,11 Additionally, more potential product structures were hypothesized to calculate the shift values, and an in-depth LC-HRMS structure search was performed for any calculated shifts that were in the range of observed experimental shifts. For products with addition of −OH to the ring, the position of −OH was varied throughout the molecule to calculate different possible shifts. These computed shifts were referenced to the 19F NMR shift of the parent pesticide or pharmaceutical (instead of HFB) for the respective fluorine motif (for structures containing 2 fluorine functional groups) if they were within ±10 ppm of the computed parent shift. Otherwise, they were referenced to HFB. The parent shift referencing was observed to give more accurate comparisons to experimental 19F NMR spectra with respect to upfield and downfield shifts of the products with a similar structure to that of the parent.

Figure 1.

Figure 1

Open in a new tab

Structures of the fluorinated pesticides and pharmaceuticals (noted with an asterisk) used in the current computational study covering key fluorinated motifs. This figure is nonexhaustive, and the structures of other fluorinated compounds are in SI Section 5.

Enthalpy of Bond Breaking

Optimized geometries and vibrational frequencies for all molecules for enthalpy calculations were computed using the local M06-L functional46 and the 6-311+G(2df, 2p) basis set with an “ultrafine” integration grid.47 Subsequent single point energy calculations were computed using the M06-2X functional for higher accuracy of thermochemical quantities.47 Water was used as the solvent with the SMD solvation model. First, the parent compound was optimized, and subsequently each positively charged cation without a single fluorine in the molecule was optimized. The fluoride ion was optimized separately. Energy values (in Hartree) were obtained as electronic energy with thermal enthalpy correction and converted to kcal mol–1. Enthalpy of heterolytic bond cleavage was calculated as the difference between enthalpies of products and reactants, where products are the sum of enthalpies for the optimized cation molecule and the fluoride ion, and reactant is the optimized parent molecule (eq 2), where EE is the sum of electronic energies and Hcorr is the thermal enthalpy correction to energy.

graphic file with name es3c09341_m002.jpg 2

The goal of this calculation was to determine the relative energies of the reactions for different molecules. We note that these computations do not consider responses of explicit water molecules to change in charge, and to get robust values, a training set of molecules with known values would be needed. Given that fluoride is produced in all the reactions and cations are of similar size, these errors should approximately cancel, allowing a comparison of enthalpy values on a relative scale.

In addition to pesticides and pharmaceuticals, model benzylic or heteroaromatic ring compound structures with a CF3 group and some common substituents (e.g., −OH, =O, −NH2) were used for enthalpy calculations, and these structures are shown in SI Section 5.

Electron Density Maps

Molecular electrostatic potentials (MEPs) from electron density calculations were visualized using the surface visualization tool in the Gaussian 16 program. Parent fluorinated compounds optimized using the B3LYP functional and 6-311+G(d,p) basis set were used for electron density visualization.

Results and Discussion

Photolysis and Photoproduct Identification

Photolysis of two Het-CF3 containing pharmaceuticals, sitagliptin (containing a triazole) and celecoxib (containing a pyrazole), was carried out to support previous results and to assist in the computational analysis of fluorinated compounds. The pseudo-first-order rate constants were 0.069 ± 0.0011 and 0.0167 ± 0.0006 min–1 for sitagliptin and celecoxib, respectively, under 275 nm narrow bandwidth UV-LED irradiation of 7.18 mW cm–2. Quantum yields were not calculated and considered out of scope for this study. 19F NMR analysis of the sitagliptin after photolysis (only 34% defluorination) agrees with the previously found results on the stability of other Het-CF3 containing compounds with triazole, pyridine, and pyrimidine rings.11 Approximately 63% of total fluorine was a product retaining the CF3 motif from the initial 30 μM CF3 group. This indicates that the Het-CF3 motif was stable during photolysis. Hence, photolysis of CF3 groups only produced 40% of total fluoride, while the other 60% fluoride came from the defluorination of the Aryl-F groups. The results for the defluorination of celecoxib, also a Het-CF3 containing compound, were contrary to the previous findings on Het-CF3 groups. Celecoxib formed approximately 52 μM (87%) fluoride from the initial 60 μM parent Het-CF3 group. These results indicate that there are Het-CF3 containing pesticides and pharmaceuticals that could be defluorinated upon photolysis, which was not observed in previous studies. This susceptibility to defluorination, however, depends on the structure of the molecule, electron density distribution around the fluorine functional group, and excitation of the molecule upon the absorption of photons, which is discussed in a subsequent section.

Saflufenacil was photolyzed using the 308 nm UV-LED to approximately 4 ln reduction in parent concentration and yielded 12 different 19F NMR product shifts after photolysis (see SI for NMR spectra). The 19F NMR analysis on the parent compound and products isolated using preparative HPLC is also shown in SI Section 5. Saflufenacil had two peaks: −67.9 ppm for the Het-CF3 (pyrimidine) and −119.9 ppm for the Aryl-F group. The first fraction produced a peak at −84.02 ppm. The second fraction was determined to contain trifluoroacetate (TFA) at −77.3 ppm. The third and the eighth fractions produced peaks at −67.8 ppm, indicating two different structures close to the parent structure without the Aryl-fluorine. These products show that the Aryl-F is easily defluorinated compared to Het-CF3, which is retained during the degradation pathway. The fourth and the seventh fractions each had two peaks at −67.9 and −119.9, and −67.8 and −119.9 ppm, respectively. LC-HRMS analysis on these fractions showed that fraction 4 was the parent saflufenacil. Potential LC-HRMS structures for all the compounds in the fractions were determined for further computational prediction of NMR shifts for these products.

Computed 19F NMR Shifts

The accuracy of computed shifts was evaluated based on solvation models and various basis sets. To evaluate the effect of solvation on the GIAO computed 19F NMR isotropic shielding constants, comparisons were made for calculations in the gas phase and using the SMD universal solvation model with water as the solvent. The chemical shifts of saflufenacil, sulfoxaflor, penoxsulam, fluoxetine, TFM, florasulam, voriconazole, and fluroxypyr were calculated (SI). The linear fitting of computed and experimental NMR shifts with water as the solvent resulted in an R2 of 0.92 as compared to 0.98 with the gas phase calculations. The mean absolute error (MAE) with solvent calculations for these 8 compounds was 2.2 ppm, while the MAE with gas phase calculations was 1.7 ppm. Gas phase calculations for fluorinated pharmaceuticals and pesticides are more suitable for 19F NMR calculations as they are more accurate than the solvation models. Previous studies with 13C and 1H NMR reported higher accuracy with implicit solvation models, and previous research on PFAS and fluorinated aromatic compound 19F NMR computations using similar functionals and basis sets reported no significant impact or increased error with the solvation models.26,27 Our results show, however, that calculation in the gas phase rather than in the solvent gives better results for 19F NMR shifts, possibly due to the high electronegativity and low polarizability of fluorine atoms.26 These results indicate that different functionals and polarization models will lead to varied accuracies for different groups of compounds with fluorine functional groups. It is also necessary to understand how different basis sets, i.e., different sets of polarization functions and equations, impact the chemical shifts (see SI). Triple-zeta basis sets were consistently more accurate with lower absolute errors compared to double-ζ basis sets. Adding a diffuse function had limited impact on the chemical shifts, except increasing the error by ∼0.2 ppm for two compounds. Adding an additional d-orbital polarization function slightly decreased the absolute error (<1 ppm) for Aryl-F and Het-F groups while increased the error by >3 ppm for all CF3 groups. This is likely due to the electronegativity of the fluorine atoms. With more fluorine atoms close to each other, the addition of the polarization function allows for the electron density around the group to polarize. Because of the low polarizability of aliphatic fluorine atoms, no benefit from additional d-orbital polarization for CF3 groups was observed, and this is consistent with the higher errors observed by others.26 Considering the high error due to polarization, the most appropriate basis set for shift calculation was determined to be 6-311+G(d,p) and was used in all further shift calculations.

Errors associated with various fluorine motifs were then evaluated by using the optimal basis sets and gas phase calculations. Linear fitting of computational vs experimental shifts resulted in an R2 of 0.98 and MAE of 1.74 ppm (Figure 2), indicating that NMR computations reproduce experimental shifts for most compounds. The MAE was also calculated for individual fluorine functional groups with 15 compounds containing Aryl-F, 5 each containing Het-F, benzylic-CF3, and Het-CF3, respectively, and 4 containing aliphatic groups (including TFA). The MAEs were 2.18, 4.04, 0.83, 0.95, and 1.04 ppm for Aryl-F, Het-F, benzylic-CF3, Het-CF3, and aliphatic groups, respectively. Eight compounds containing CF3 groups had absolute errors of <0.6 ppm with gas phase NMR computations. Absolute errors >2.7 ppm were only observed for shifts upfield of −120 ppm (in the Aryl-F and Het-F range).

Figure 2.

Figure 2

Open in a new tab

Computed 19F NMR shifts for 33 fluorinated motifs from 23 fluorinated pharmaceuticals and agrochemicals and the corresponding experimental 19F NMR shift with a linear regression R2 value of 0.98. The mean absolute error (MAE) was 1.74 ppm. Experimental 19F NMR shifts were either measured in the current study, measured in our previous studies using hexafluorobenzene (HFB) as the shift reference standard, or taken from the literature with CFCl3 as the shift reference standard. More information on the compound names, shifts, and literature values is in SI Section 6. The structures with the two highest errors are shown in the figure.

19F NMR computations have been previously shown to accurately compute (MAE < 2 ppm) upfield and downfield shifts in frequency based on small structural changes for aromatic compounds.25 Photolyzed samples have multiple 19F NMR product peaks, and previous studies have identified more than 100 product structures using LC-HRMS for the compounds chosen for this study.9−15 The next step is to assign these identified structures to the observed 19F NMR shifts by computing the shifts of the HRMS identified products. Experimental 19F NMR spectra for saflufenacil after photolysis show a product peak at −67.7 ppm, <0.2 ppm downfield from the parent shift at −69.9 ppm. The spectra also show more peaks at −71 ppm, at −77 ppm, and in the range of −81 to −88 ppm (Figure 3A). The products that were successfully isolated and the LC-HRMS structures identified in our study are shown with red crosses in Figure 3. Two isolated products had NMR shifts of −67.7 ppm (<0.2 ppm downfield of parent), which were identified to be formed from slight modifications to the parent compound and defluorination of the Aryl-F (Structures 1a and 1b in Figure 3). The computed NMR shifts for both these products were −67.9 ppm, i.e., <0.1 ppm error in computations. Isolated product 1c that was identified by LC-HRMS resulting from SO2 extrusion and rearrangement also had computed shifts within 0.2 ppm of the experimental shifts for the CF3 group and within 0.03 ppm for the Aryl-F group. Additionally the concentration of fluorine at −119.9 ppm (Aryl-F product 1c) was 4.1 μM, and because this product also contained the CF3 group the CF3 concentration must be 12.3 μM. The CF3 concentration observed was 15.1 μM indicating that the difference, i.e., 2.8 μM of fluorine, could be attributed to the nonaryl-F containing products 1a and 1b. The isolated product (1g, 4,4,4-trifluoro-3-hydroxyvaline, identified by LC-HRMS) at −84.02 ppm had a computational shift of −83.3 ppm (0.7 ppm error) and was consistent with an aliphatic product being formed via ring opening. Note that with a fully fluorinated methyl group, this compound satisfies the OECD definition of a PFAS.48 Overall, the computational 19F NMR shifts of the isolated products matched with their experimental shifts with an absolute error below 0.7 ppm, indicating that computationally predicting NMR shifts of products can be successfully used to assign the structures to the collected spectra. Additionally, with multiple products sharing the same shifts, combining computations and experimental data can assist in accurate quantification of each product.

Figure 3.

Figure 3

Open in a new tab

Experimental and computational 19F NMR shifts for the photolysis products of saflufenacil identified by LC-HRMS. (A) Experimental 19F NMR spectra for the CF3-containing products. (B) Experimental spectra for Aryl-F containing products and fluoride. The star symbol (★) represents the parent compound shift in both parts A and B. Computational shifts are represented as dashed (blue) lines below the experimental spectra with the corresponding identified LC-HRMS product structures. This figure is representative for only one wavelength and does not contain all the products formed in different photolysis conditions; all structures are in SI Section 7. The crosses (×) are experimentally measured shifts for structures that were isolated and concentrated from experimental photolysis solutions. The relative area of the experimental spectra shows the relative concentrations of the products, while the different lengths of the computational spectral lines are only to graphically fit the structures in the figure. Spectra reproduced with permission from ref (11). Copyright 2023 American Chemical Society.

Further, we computed NMR shifts for all the products of saflufenacil previously predicted using LC-HRMS and not isolated in this study due to the specific preparative column used.11 Most products except 1d, 1e, and trifluoroacetamide (1f) did not yield computational shifts that fell in the range of the experimentally observed peaks. These products, however, were identified with high resolution mass spectrometry and fragment matches, and they may still be present but at concentrations in nM concentrations, i.e., lower than the NMR detection limits of approximately 0.1 μM. None of the previously predicted products appeared in the range of −84 to −88.4 ppm, within which three products are observed in the experimental NMR spectra. These products are potentially aliphatic products from ring opening, similar to the isolated product 1g. We computed shifts for several hypothesized structures, and only a few predicted vinyl-CF2 products had computed shifts in the range of −88 ppm. Further, we reanalyzed all of the previous LC-HRMS features and found a single product that had a computational shift of −89.2 ppm. This product is potentially the vinyl-CF2 product 1h in Figure 3 (4,4-difluorobut-3-enamide), and this shift is consistent with previously hypothesized vinyl ring opening products.21 The NMR peak however shows a broad singlet with slight splitting while a vinyl-CF2 product is expected to have at least a doublet peak (due to a high F–F coupling constant) with similar heights.49 Because this product could not be isolated in the study, synthesis of this product is needed to confirm its presence due to the uncertainty of the doublet. Most products from saflufenacil had to be newly identified in this study to match with the experimental NMR spectra. Adding computational analysis to the protocol also gave multiple insights into new mechanisms of degradation including SO2 extrusion and rearrangement in saflufenacil and the formation of aliphatic CF3 and potential vinyl-CF2 products in the NMR range of −80 to −88 ppm. No studies have previously reported these product structures during saflufenacil degradation. Additionally, previously calculated mass balances with quantitative 19F NMR shifts could be updated to include quantification of specific product structures using computed 19F NMR shifts (see SI).

Photolysis of penoxsulam at different wavelengths in our previous study led to between 5 and 8 products peaks in the CF3 range (−55 to −65 ppm) and more than 6 products in the CF2 range (−126 to −136 ppm).11 We computationally predicted NMR shifts for all the products from our previous studies.9 Many of the suspected products of penoxsulam have calculated NMR shifts close to several experimentally observed NMR peaks (Figure 4). The location of −OH group addition is often not determinable on the ring using only LC-HRMS and MS2 analysis. With multiple possible locations around the ring for OH addition, computations were able to assist in the prediction of the exact products. For isomers with OH in different positions, computational analysis revealed that the location of the −OH group on the ring leads to shifts of up to 5 ppm. The positions of CF3 with respect to the OH group also lead to large shifts as determined by computations. For example, product 2c has a computed shift of −63.8 ppm, while its structural isomer 2c2 has a shift of −56.8 ppm. The small peaks at −65.8 and −66.2 ppm could not be identified. All of the computed aliphatic CF2 products were consistent with the experimental peaks. The large experimental peaks at −130 ppm could be attributed to multiple products with computed shifts within 1 ppm. None of the previous products predicted from HRMS, however, had a shift of −136 ppm. Computational aliphatic CF2 peaks for several hypothesized structures were not upfield at −131.5 ppm. We hypothesized multiple CF2 and CF containing products out of which only some vinyl-CF2 products lie in the range of −136 ppm. We reanalyzed all the LC-HRMS features from the previous studies and found three structures with a vinyl-CF2 motif (Products 2r2, 2s, and 2t in Figure 4). The proposed enol 2r, 2s, and 2t structures are tautomers of the 2j and 2k keto structures. The computational shifts for these enol products were between −136.05 and −136.2, in the range of the unidentified experimental shift, and may be more stable than their keto tautomers.50,51 Quantification of products using previously reported shift-based quantification8 and current shift-product matching is provided in SI Section 8. This analysis allows specific products that were structurally identified via MS to be quantified by using 19F NMR without authentic standards.

Figure 4.

Figure 4

Open in a new tab

Experimental9 and computational 19F NMR shifts for photolysis products of penoxsulam identified by LC-HRMS. (A) Experimental 19F NMR spectra for the CF3-containing products. The star symbol (★) represents the parent compound shift. (B) Experimental spectra of aliphatic-CF2-containing products. Computational shifts are represented as dashed (blue) lines below the experimental spectra with the LC-HRMS product structures. This figure is representative for only one wavelength and does not contain all the products formed in different photolysis conditions; all product structures are in SI Section 8. The relative area of the experimental spectra shows the relative concentrations of the products while the different lengths of the computational spectral lines are only to graphically fit the structures in the figure. Note that 2r, 2s, and 2t structures are a tautomers of 2j and 2k structures.

Many of the experimental NMR product peaks for sulfoxaflor are between −68.9 and −69.8 ppm, near the parent peak.9,11,52 With the product shifts so close to each other, it is evident that the structures must be very similar to the parent structure with the retention of the heteroaromatic ring. The computational shifts for products identified in two previous studies were all within ±0.8 ppm of the parent peak, indicating accurate prediction (products shown in SI Section 9). Two products were upfield of the parent, while others were downfield, agreeing with the experimental results. We were able to assign peaks to products based on the shift and the relative peak areas in the NMR and LC-HRMS results (see SI). Solar photolysis of sulfoxaflor also yielded peaks at −88 ppm,9 which are possibly vinyl-CF2 peaks (5,5-difluoropent-4-enamide, 3f, and a 2-(difluoromethylene)-2,5-dihydropyridine based structure, 3i) using a combination of computations and reanalysis of LC-HRMS features. The vinyl products from sulfoxaflor agree well with similar products observed from saflufenacil. Another aliphatic product peak (4,4,4-trifluoro-3,3-dimethylbutan-1-ol as 3h) was computed to be −83.9 ppm, where small experimental peaks were observed for sulfoxaflor photolysis in the solar simulator (see SI for structures). Product 3h is a PFAS accrording to OECD definitions.48

Other compounds like florasulam, fluoxetine, voriconazole, and fluroxypyr have a lesser number of products than saflufenacil, penoxsulam, and sulfoxaflor due the absence of stable groups like Het-CF3 and aliphatic fluorines. The major product of photolysis for these compounds is fluoride. Computational analysis was performed for the products previously observed for these compounds and most of the products were accurately predicted. We were able to assign these products to individual peaks, giving information about the concentration of each product that could not be determined in any previous study (SI Sections 7–15).

Computational Insights into Functional Group Reactivity

Compelling trends have been observed for benzylic and Het-CF3 groups in several studies where Het-CF3 groups have been reported to be stable and retained in products after photolysis. Nevertheless, we observed that photolysis of celecoxib resulted in fluoride being a major product despite being a Het-CF3 compound. This is potentially due to structural differences and the distribution of electron density around the molecule leading to different susceptibility to C–F bond breaking via photolysis. The heterolytic bond dissociation enthalpies for different CF3-containing compounds are shown in Figure 5. The absolute enthalpy values may have inherent errors (see Experimental Section), and here, we are interested in the differences in values. The C–F bond breaking enthalpies are consistently lower for benzylic-CF3 groups (computed average 52.6 kcal mol–1) as compared to the Het-CF3 groups (computed average 66.9 kcal mol–1;Figure 5A). The relative enthalpy values agree with the previously reported results regarding the experimental stability of these groups. The enthalpy of bond cleavage, however, depends on the structure of the heteroaromatic group. The most common heteroaromatic group in the pesticide or pharmaceutical structure consists of a pyridine ring and has a comparatively high computed enthalpy value between 63.5 and 70.2 kcal mol–1. We studied several model heteroaromatic groups with pyridine, pyrimidine, pyrazole, and triazole rings and found that the enthalpies are consistently higher than those of model benzylic groups (see SI for structures). The only heteroaromatic compounds that had a lower enthalpy value in the range of benzylic-CF3 groups were model compound #8 (pyrazole) and celecoxib (Points 5 and 6 in Figure 5A). Our experimental results on celecoxib agree with the predicted higher defluorination of this CF3 group. The results indicate that the extent of defluorination is dependent not only on having a benzylic or heteroaromatic group but also on the structure of each heteroaromatic ring and its effect on the energy of bond cleavage.

Figure 5.

Figure 5

Open in a new tab

(A) Heterolytic C–F bond breaking enthalpies (in kcal mol–1) from the CF3 group for fluorinated compounds containing benzylic or heteroaromatic CF3 motifs. The enthalpies are the average of three values for the three fluorines, and standard deviation bars are smaller than the marker. The values have not been validated with a training set, and the key result is the relative differences among the different structural classes. Structures for model compounds 1–8 are available in SI Section 16. The horizontal lines in the benzylic-CF3 and Het-CF3 sections respectively represent the average heterolytic bond breaking enthalpy values (52.6 kcal mol–1 and 66.9 kcal mol–1, respectively). Points 5 and 6 are not included in the heteroaromatic average. Compounds corresponding to markers numbered 1–5 are shown in (B) and (C) and marker 6 has a lower enthalpy because it is the model structure corresponding to celecoxib (5). (B) Molecular electrostatic potential (MEP) from electron density calculations for compounds 1–5. The common color map legend is shown on the top right with the positive–negative value range (electrons Å–3) in parentheses next to each compound name. Red denotes negative values, and blue denotes positive. CF3 groups can be seen at the bottom right of the compound in (1) and right side of the compound in (2)–(5). (C) Fluorine mass balances for compounds 1–5 calculated using 19F NMR before (initial) and after (photolyzed) photolysis using 275 nm UV-LED with approximately 4-log degradation (except sitagliptin with 3-log). The mass balances for penoxsulam, fluoxetine, and saflufenacil were taken from our previous study. Reproduced with permission from ref (11). Copyright 2023 American Chemical Society. Experiments for sitagliptin and celecoxib were performed in the current study. Each differently colored/shaded bar is a different type of fluorine in the solution where the corresponding legends show the fluorinated product or the type of fluorine peak observed for each product along with the frequency shifts, i.e., singlets (S), doublets (D), doublets of triplets (DT), trifluoroacetic acid (TFA), and fluoride (F–). Parent compounds are marked with P in the legends. Errors bars, which are not presented in the bar graphs, were less than 7%.

Different (hetero)aromatic rings have different electron density distributions around the ring and the CF3 group (Figure 5B). The electron density around the CF3 group in penoxsulam and fluoxetine is negative, making it a potential site for electrophilic attacks or substitution of fluorine.34,35 It was observed that the molecular electrostatic potential around the CF3 group in saflufenacil and sitagliptin is generally positive, consistent with low experimental defluorination. On the other hand, celecoxib has a strongly negative electron density around the CF3 group, agreeing with the enthalpy values and the extent of defluorination observed. Because the defluorination step is often reported to be a rate-determining step,53−57 the value of the enthalpy and the electron density around the fluorine groups appear to be good indicators of the extent of defluorination.

Structural dependence on defluorination has been previously reported for various degradation processes where electron densities around specific sites affect defluorination of fluorine motifs in organic contaminants.31−33,58 The bar graphs from 275 nm UV-LED photolysis show the quantities of fluoride and fluorinated products formed (Figure 5C) for 5 representative compounds (saflufenacil, sitagliptin, and celecoxib measured as part of this work as described above; the other two compounds are from previous studies). Penoxsulam has two fluorine motifs: benzlyic-CF3 and aliphatic-CF2, out of which the benzylic-CF3 is defluorinated to form approximately 29 μM fluoride from the initial 30 μM and the aliphatic-CF2 forms stable products all retaining the CF2 motif. Photolysis of fluoxetine, also a benzylic-CF3-containing compound, leads to formation of 27 μM fluoride. The extent of defluorination from Het-CF3 groups, however, depends on the enthalpy change during fluoride removal. While saflufenacil and sitagliptin form stable products due to the electron density distribution, celecoxib formed fluoride. With multiple new fluorinated compounds continuously being introduced to the market, computational analysis coupled with rapid degradation experiments could give crucial information on the photochemical stability of these new fluorinated compounds.

Implications

The combined approach based on LC-HRMS and experimental and computational 19F NMR provides crucial structural information and allows the quantification of photolysis products that may be often missed by using individual approaches. The approach can be improved upon the development of a spectral database for products25 or the addition of machine learning into the identification workflows.59 Because the fluorine nucleus is highly sensitive to its structural environment during NMR analysis with a wide shift range of >200 ppm, computing NMR shifts can assist in accurately assigning structures to experimental peaks with an accuracy of less than 2 ppm. With multiple experimental peaks that remained unidentified in previous studies, hypothesizing a group of structures with computations that lie in the unidentified experimental shift range assisted in re-evaluating LC-HRMS features. Additional experiments with 1H and 13C NMR with isolated products could be used to provide additional structural information, but such analyses would be challenging for the mixtures obtained in the degradation studies.

The highest accuracy in computational shifts was observed for CF3 groups, with average absolute error of less than 0.9 ppm, while the errors increased for shifts above 120 ppm to approximately 2–4 ppm for Het-F, aryl-F, and aliphatic-CF2 groups. We observed that it is generally harder to accurately predict products for Het-CF3 and aliphatic compounds due to the stability of these motifs leading to multiple products. The position of the −OH group on the ring is often unidentified during LC-MS analysis but computations can assist in determination of the OH position because of differences in shifts with the varied positions. Additionally, with products containing different fluorine motifs, multiple shifts, products sharing the same shifts, or newly identified PFAS products computations assist in identifying products and match concentrations for accurate mass balances. Incorporating NMR prediction in solvents and negative electrospray ionization in MS could improve the protocol for other solution matrices.

Contrary to previous reports, Het-CF3 motifs are not always stable, and we found that celecoxib is one such compound that defluorinates under 275 nm UV irradiation. Defluorination depends on the enthalpy of C–F bond cleavage, molecular structure, and electron density, where negative electrostatic potentials around the motif assist in defluorination. Here these tools are applied to direct photolysis reactions, but they should be applicable to any degradation process for fluorinated compounds. These results can be used for future studies for the product identification and analysis of newly synthesized fluorinated pharmaceuticals and pesticides in the market.

Acknowledgments

Funding for this project was provided by the graduate fellowship to A.P.B. from the University of Minnesota College of Science and Engineering, the Doctoral Dissertation Fellowship to A.P.B. by the University of Minnesota Graduate School, the Minnesota Environment and Natural Resources Trust Fund as recommended by the Legislative-Citizen Commission on Minnesota Resources (LCCMR), and the National Science Foundation (CHE #2304963). Thanks to Yesenia Vega for help with the computations. We thank the Minnesota NMR Center for access to instrumentation. Funding for NMR instrumentation was provided by the Office of the Vice President for Research, the Medical School, the College of Biological Science, NIH, NSF, and the Minnesota Medical Foundation. Thanks to Peter Villalta and Yingchun Zhao at the analytical biochemistry mass spectrometry services shared resource at the Masonic Cancer Center, University of Minnesota, for the help with LC-HRMS instrumentation. The authors acknowledge the Minnesota Supercomputing Institute (MSI) at the University of Minnesota for providing resources that contributed to the research results.

Supporting Information Available

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acs.est.3c09341.

  • Additional experimental information including chemical sources and purities, photolysis conditions, and 19F NMR and HRMS analytical details and additional results including NMR spectra and computed NMR shifts of parent compounds and photoproduct structures, fluorine mass balances, and additional computational data (PDF)

The authors declare no competing financial interest.

Supplementary Material

es3c09341_si_001.pdf (4.7MB, pdf)

References

  1. Fujiwara T.; O’Hagan D. Successful Fluorine-Containing Herbicide Agrochemicals. J. Fluor. Chem. 2014, 167, 16–29. 10.1016/j.jfluchem.2014.06.014. [DOI] [Google Scholar]
  2. Inoue M.; Sumii Y.; Shibata N. Contribution of Organofluorine Compounds to Pharmaceuticals. ACS Omega 2020, 5 (19), 10633–10640. 10.1021/acsomega.0c00830. [DOI] [PMC free article] [PubMed] [Google Scholar]
  3. O’Hagan D. Understanding Organofluorine Chemistry. An Introduction to the C–F Bond. Chem. Soc. Rev. 2008, 37 (2), 308–319. 10.1039/B711844A. [DOI] [PubMed] [Google Scholar]
  4. Ogawa Y.; Tokunaga E.; Kobayashi O.; Hirai K.; Shibata N. Current Contributions of Organofluorine Compounds to the Agrochemical Industry. iScience 2020, 23 (9), 101467 10.1016/j.isci.2020.101467. [DOI] [PMC free article] [PubMed] [Google Scholar]
  5. Inoue M.; Sumii Y.; Shibata N. Contribution of Organofluorine Compounds to Pharmaceuticals. ACS Omega 2020, 5 (19), 10633–10640. 10.1021/acsomega.0c00830. [DOI] [PMC free article] [PubMed] [Google Scholar]
  6. Bégué J. P.; Bonnet-Delpon D. Recent Advances (1995–2005) in Fluorinated Pharmaceuticals Based on Natural Products. J. Fluor. Chem. 2006, 127 (8), 992–1012. 10.1016/j.jfluchem.2006.05.006. [DOI] [Google Scholar]
  7. Boreen A. L.; Arnold W. A.; McNeill K. Photodegradation of Pharmaceuticals in the Aquatic Environment: A Review. Aquat. Sci. 2003, 65 (4), 320–341. 10.1007/s00027-003-0672-7. [DOI] [Google Scholar]
  8. Remucal C. K. The Role of Indirect Photochemical Degradation in the Environmental Fate of Pesticides: A Review. Environ. Sci. Process. Impacts 2014, 16 (4), 628–653. 10.1039/c3em00549f. [DOI] [PubMed] [Google Scholar]
  9. Bhat A. P.; Pomerantz W. C. K.; Arnold W. A. Finding Fluorine: Photoproduct Formation during the Photolysis of Fluorinated Pesticides. Environ. Sci. Technol. 2022, 56 (17), 12336–12346. 10.1021/acs.est.2c04242. [DOI] [PMC free article] [PubMed] [Google Scholar]
  10. Bhat A.; Mundhenke T.; Whiting Q.; Peterson A.; Pomerantz W.; Arnold W. A. Tracking Fluorine during Aqueous Photolysis and Advanced UV Treatment of Fluorinated Phenols and Pharmaceuticals Using a Combined 19F NMR, Chromatography, and Mass Spectrometry Approach. ACS Environ. Au. 2022, 2 (3), 242–252. 10.1021/acsenvironau.1c00057. [DOI] [PMC free article] [PubMed] [Google Scholar]
  11. Bhat A.; Pomerantz W.; Arnold W. A. Wavelength-Dependent UV-LED Photolysis of Fluorinated Pesticides and Pharmaceuticals. Environ. Sci. Technol. 2023, 57 (13), 5327–5336. 10.1021/acs.est.3c00627. [DOI] [PubMed] [Google Scholar]
  12. Gauthier J. R.; Mabury S. A. The Environmental Degradation and Distribution of Saflufenacil, a Fluorinated Protoporphyrinogen IX Oxidase–Inhibiting Herbicide, on a Canadian Winter Wheat Field. Environ. Toxicol. Chem. 2020, 39 (10), 1918–1928. 10.1002/etc.4820. [DOI] [PubMed] [Google Scholar]
  13. Chen G.; Qiao Y.; Zhang X.; Liu F.; Liao H.; Zhang R.; Dong J.; Tao B. Identification and Characterization of Herbicide Penoxsulam Transformation Products in Aqueous Media by UPLC-QTOF-MS. Bull. Environ. Contam. Toxicol 2019, 102 (6), 854–860. 10.1007/s00128-019-02612-2. [DOI] [PubMed] [Google Scholar]
  14. Krieger M. S.; Yoder R. N.; Gibson R. Photolytic Degradation of Florasulam on Soil and in Water. J. Agric. Food Chem. 2000, 48 (8), 3710–3717. 10.1021/jf991290+. [DOI] [PubMed] [Google Scholar]
  15. Yang Y.; Ran L.; Pan T.; Yuan F.; Hu D.; Lu P. Degradation of Sulfoxaflor in Water and Soil: Kinetics, Degradation Pathways, Transformation Product Identification, and Toxicity. J. Agric. Food Chem. 2022, 70 (11), 3400–3408. 10.1021/acs.jafc.1c07362. [DOI] [PubMed] [Google Scholar]
  16. Joudan S.; De Silva A. O.; Young C. J. Insufficient Evidence for the Existence of Natural Trifluoroacetic Acid. Environ. Sci. Process. Impacts 2021, 23 (11), 1641–1649. 10.1039/D1EM00306B. [DOI] [PubMed] [Google Scholar]
  17. Gauthier J. R.; Mabury S. A. Noise-Reduced Quantitative Fluorine NMR Spectroscopy Reveals the Presence of Additional Per- and Polyfluorinated Alkyl Substances in Environmental and Biological Samples When Compared with Routine Mass Spectrometry Methods. Anal. Chem. 2022, 94 (7), 3278–3286. 10.1021/acs.analchem.1c05107. [DOI] [PubMed] [Google Scholar]
  18. Moody C. A.; Kwan W. C.; Martin J. W.; Muir D. C. G.; Mabury S. A. Determination of Perfluorinated Surfactants in Surface Water Samples by Two Independent Analytical Techniques: Liquid Chromatography/Tandem Mass Spectrometry and 19F NMR. Anal. Chem. 2001, 73 (10), 2200–2206. 10.1021/ac0100648. [DOI] [PubMed] [Google Scholar]
  19. D’Agostino L. A.; Mabury S. A. Identification of Novel Fluorinated Surfactants in Aqueous Film Forming Foams and Commercial Surfactant Concentrates. Environ. Sci. Technol. 2014, 48 (1), 121–129. 10.1021/es403729e. [DOI] [PubMed] [Google Scholar]
  20. Moody C. A.; Martin J. W.; Kwan W. C.; Muir D. C.; Mabury S. A. Monitoring Perfluorinated Surfactants in Biota and Surface Water Samples Following an Accidental Release of Fire-Fighting Foam into Etobicoke Creek. Environ. Sci. Technol. 2002, 36 (4), 545–551. 10.1021/es011001+. [DOI] [PubMed] [Google Scholar]
  21. Li J.; Mundhenke T. F.; Smith T. G.; Arnold W. A.; Pomerantz W. C. K. Fluorous Liquids for Magnetic Resonance-Based Thermometry with Enhanced Responsiveness and Environmental Degradation. Anal. Chem. 2023, 95 (14), 6071–6079. 10.1021/acs.analchem.3c00172. [DOI] [PubMed] [Google Scholar]
  22. Mundhenke T. F.; Bhat A. P.; Pomerantz W. C. K.; Arnold W. A. Photolysis Products of Fluorinated Pharmaceuticals: A Combined 19F-NMR and Mass Spectrometry Approach. Environ. Toxicol. Chem. 2023, 10.1002/etc.5773. [DOI] [PubMed] [Google Scholar]
  23. Mattes A. O.; Russell D.; Tishchenko E.; Liu Y.; Cichewicz R. H.; Robinson S. J. Application of 19F Quantitative NMR to Pharmaceutical Analysis. Concepts Magn. Reson. Part A 2016, 45A (5), e21422 10.1002/cmr.a.21422. [DOI] [Google Scholar]
  24. Buchholz C. R.; Pomerantz W. C. K. 19F NMR Viewed through Two Different Lenses: Ligand-Observed and Protein-Observed 19F NMR Applications for Fragment-Based Drug Discovery. RSC Chem. Biol. 2021, 2 (5), 1312–1330. 10.1039/D1CB00085C. [DOI] [PMC free article] [PubMed] [Google Scholar]
  25. Saunders C.; Khaled M. B.; Weaver J. D.; Tantillo D. J. Prediction of 19F NMR Chemical Shifts for Fluorinated Aromatic Compounds. J. Org. Chem. 2018, 83 (6), 3220–3225. 10.1021/acs.joc.8b00104. [DOI] [PMC free article] [PubMed] [Google Scholar]
  26. Mifkovic M.; Pauling J.; Vyas S. Computational Protocol for Predicting 19F NMR Chemical Shifts for PFAS and Connection to PFAS Structure. J. Comput. Chem. 2022, 43 (20), 1355–1361. 10.1002/jcc.26939. [DOI] [PubMed] [Google Scholar]
  27. Lodewyk M. W.; Siebert M. R.; Tantillo D. J. Computational Prediction of 1H and 13C Chemical Shifts: A Useful Tool for Natural Product, Mechanistic, and Synthetic Organic Chemistry. Chem. Rev. 2012, 112 (3), 1839–1862. 10.1021/cr200106v. [DOI] [PubMed] [Google Scholar]
  28. Lodewyk M. W.; Tantillo D. J. Prediction of the Structure of Nobilisitine A Using Computed NMR Chemical Shifts. J. Nat. Prod. 2011, 74 (5), 1339–1343. 10.1021/np2000446. [DOI] [PubMed] [Google Scholar]
  29. Gauthier J. R.; Mabury S. A. Identifying Unknown Fluorine-Containing Compounds in Environmental Samples Using 19F NMR and Spectral Database Matching. Environ. Sci. Technol. 2023, 57 (23), 8760–8767. 10.1021/acs.est.3c01220. [DOI] [PubMed] [Google Scholar]
  30. Theodoridis G. Chapter 4 Fluorine-Containing Agrochemicals: An Overview of Recent Developments. Adv. Fluor. Sci. 2006, 2 (C), 121–175. 10.1016/S1872-0358(06)02004-5. [DOI] [Google Scholar]
  31. Ji Y.; Yang Y.; Zhou L.; Wang L.; Lu J.; Ferronato C.; Chovelon J.-M. Photodegradation of Sulfasalazine and Its Human Metabolites in Water by UV and UV/Peroxydisulfate Processes. Water Res. 2018, 133, 299–309. 10.1016/j.watres.2018.01.047. [DOI] [PubMed] [Google Scholar]
  32. Yong L.; Zhanqi G.; Yuefei J.; Xiaobin H.; Cheng S.; Shaogui Y.; Lianhong W.; Qingeng W.; Die F. Photodegradation of Malachite Green under Simulated and Natural Irradiation: Kinetics, Products, and Pathways. J. Hazard. Mater. 2015, 285, 127–136. 10.1016/j.jhazmat.2014.11.041. [DOI] [PubMed] [Google Scholar]
  33. Wang Y.; Jing B.; Wang F.; Wang S.; Liu X.; Ao Z.; Li C. Mechanism Insight into Enhanced Photodegradation of Pharmaceuticals and Personal Care Products in Natural Water Matrix over Crystalline Graphitic Carbon Nitrides. Water Res. 2020, 180, 115925 10.1016/j.watres.2020.115925. [DOI] [PubMed] [Google Scholar]
  34. Nataraj A.; Balachandran V.; Karthick T. Molecular Orbital Studies (Hardness, Chemical Potential, Electrophilicity, and First Electron Excitation), Vibrational Investigation and Theoretical NBO Analysis of 2-Hydroxy-5-Bromobenzaldehyde by Density Functional Method. J. Mol. Struct. 2013, 1031, 221–233. 10.1016/j.molstruc.2012.09.047. [DOI] [Google Scholar]
  35. Song W.; Zhou Y.; Wang Z.; Li J.; Zhang X.; Fu C.; Du X.; Wang Z.; Qiu W. Accelerate Sulfamethoxazole Degradation and Detoxification by Persulfate Mediated with Fe2+&dithionite: Experiments and DFT Calculation. J. Hazard. Mater. 2022, 436, 129254 10.1016/j.jhazmat.2022.129254. [DOI] [PubMed] [Google Scholar]
  36. Yang X.; Hu J.; Wu L.; Hou H.; Liang S.; Yang J. Cooperation of Multiple Active Species Generated in Hydrogen Peroxide Activation by Iron Porphyrin for Phenolic Pollutants Degradation. Environ. Pollut. 2022, 313, 120097 10.1016/j.envpol.2022.120097. [DOI] [PubMed] [Google Scholar]
  37. Cui D.; Mebel A. M.; Arroyo-Mora L. E.; Holness H.; Furton K. G.; O’Shea K. Kinetic, Product, and Computational Studies of the Ultrasonic Induced Degradation of 4-Methylcyclohexanemethanol (MCHM). Water Res. 2017, 126, 164–171. 10.1016/j.watres.2017.09.005. [DOI] [PubMed] [Google Scholar]
  38. Xia H.; Zhang W.; Yang Y.; Zhang W.; Purchase D.; Zhao C.; Song X.; Wang Y. Degradation Mechanism of Tris(2-Chloroethyl) Phosphate (TCEP) as an Emerging Contaminant in Advanced Oxidation Processes: A DFT Modelling Approach. Chemosphere 2021, 273, 129674 10.1016/j.chemosphere.2021.129674. [DOI] [PubMed] [Google Scholar]
  39. Zhang Y.; Liu J.; Ghoshal S.; Moores A. Density Functional Theory Calculations Decipher Complex Reaction Pathways of 6:2 Fluorotelomer Sulfonate to Perfluoroalkyl Carboxylates Initiated by Hydroxyl Radical. Environ. Sci. Technol. 2021, 55 (24), 16655–16664. 10.1021/acs.est.1c05549. [DOI] [PubMed] [Google Scholar]
  40. Frisch M. J.; Trucks G. W.; Schlegel H. B.; Scuseria G. E.; Robb M. A.; Cheeseman J. R.; Scalmani G.; Barone V.; Petersson G. A.; Nakatsuji H.; Li X.; Caricato M.; Marenich A. V.; Bloino J.; Janesko B. G.; Gomperts R.; Mennucci B.; Hratchian H. P.; Ortiz J. V.; Izmaylov A. F.; Sonnenberg J. L.; Williams-Young D.; Ding F.; Lipparini F.; Egidi F.; Goings J.; Peng B.; Petrone A.; Henderson T.; Ranasinghe D.; Zakrzewski V. G.; Gao J.; Rega N.; Zheng G.; Liang W.; Hada M.; Ehara M.; Toyota K.; Fukuda R.; Hasegawa J.; Ishida M.; Nakajima T.; Honda Y.; Kitao O.; Nakai H.; Vreven T.; Throssell K.; Montgomery J. A. Jr.; Peralta J. E.; Ogliaro F.; Bearpark M. J.; Heyd J. J.; Brothers E. N.; Kudin K. N.; Staroverov V. N.; Keith T. A.; Kobayashi R.; Normand J.; Raghavachari K.; Rendell A. P.; Burant J. C.; Iyengar S. S.; Tomasi J.; Cossi M.; Millam J. M.; Klene M.; Adamo C.; Cammi R.; Ochterski J. W.; Martin R. L.; Morokuma K.; Farkas O.; Foresman J. B.; Fox D. J.. Gaussian 16, Revision C.01; Gaussian, Inc.: Wallingford, CT, 2016.
  41. Becke A. D. Density-functional Thermochemistry. III. The Role of Exact Exchange. J. Chem. Phys. 1993, 98 (7), 5648–5652. 10.1063/1.464913. [DOI] [Google Scholar]
  42. Lee C.; Yang W.; Parr R. G. Development of the Colle-Salvetti Correlation-Energy Formula into a Functional of the Electron Density. Phys. Rev. B 1988, 37 (2), 785–789. 10.1103/PhysRevB.37.785. [DOI] [PubMed] [Google Scholar]
  43. Vosko S. H.; Wilk L.; Nusair M. Accurate Spin-Dependent Electron Liquid Correlation Energies for Local Spin Density Calculations: A Critical Analysis. Can. J. Phys. 1980, 58 (8), 1200–1211. 10.1139/p80-159. [DOI] [Google Scholar]
  44. Stephens P. J.; Devlin F. J.; Chabalowski C. F.; Frisch M. J. Ab Initio Calculation of Vibrational Absorption and Circular Dichroism Spectra Using Density Functional Force Fields. J. Phys. Chem. 1994, 98 (45), 11623–11627. 10.1021/j100096a001. [DOI] [Google Scholar]
  45. Marenich A. V.; Cramer C. J.; Truhlar D. G. Universal Solvation Model Based on Solute Electron Density and on a Continuum Model of the Solvent Defined by the Bulk Dielectric Constant and Atomic Surface Tensions. J. Phys. Chem. B 2009, 113 (18), 6378–6396. 10.1021/jp810292n. [DOI] [PubMed] [Google Scholar]
  46. Wang Y.; Jin X.; Yu H. S.; Truhlar D. G.; He X. Revised M06-L Functional for Improved Accuracy on Chemical Reaction Barrier Heights, Noncovalent Interactions, and Solid-State Physics. Proc. Natl. Acad. Sci. U. S. A. 2017, 114 (32), 8487–8492. 10.1073/pnas.1705670114. [DOI] [PMC free article] [PubMed] [Google Scholar]
  47. Arnold W. A. One Electron Oxidation Potential as a Predictor of Rate Constants of N-Containing Compounds with Carbonate Radical and Triplet Excited State Organic Matter. Environ. Sci. Process. Impacts 2014, 16 (4), 832–838. 10.1039/C3EM00479A. [DOI] [PubMed] [Google Scholar]
  48. Wang Z.; Buser A. M.; Cousins I. T.; Demattio S.; Drost W.; Johansson O.; Ohno K.; Patlewicz G.; Richard A. M.; Walker G. W.; White G. S.; Leinala E. A New OECD Definition for Per- and Polyfluoroalkyl Substances. Environ. Sci. Technol. 2021, 55 (23), 15575–15578. 10.1021/acs.est.1c06896. [DOI] [PubMed] [Google Scholar]
  49. Koley S.; Cayton K. T.; González-Montiel G. A.; Yadav M. R.; Orsi D. L.; Intelli A. J.; Cheong P. H.-Y.; Altman R. A. Cu(II)-Catalyzed Unsymmetrical Dioxidation of Gem-Difluoroalkenes to Generate α,α-Difluorinated-α-Phenoxyketones. J. Org. Chem. 2022, 87 (16), 10710–10725. 10.1021/acs.joc.2c00925. [DOI] [PMC free article] [PubMed] [Google Scholar]
  50. Lindner P. E.; Lemal D. M. Perfluorinated Cyclic and Acyclic Keto–Enol Systems: A Remarkable Contrast. J. Am. Chem. Soc. 1997, 119 (14), 3259–3266. 10.1021/ja963788n. [DOI] [Google Scholar]
  51. Lindner P. E.; Lemal D. M. Highly Fluorinated Cyclopentanones and Their Enols. J. Org. Chem. 1996, 61 (15), 5109–5115. 10.1021/jo9602940. [DOI] [Google Scholar]
  52. Gauthier J. R.; Mabury S. A. The Sulfoximine Insecticide Sulfoxaflor and Its Photodegradate Demonstrate Acute Toxicity to the Nontarget Invertebrate Species Daphnia magna. Environ. Toxicol. Chem. 2021, 40 (8), 2156–2164. 10.1002/etc.5117. [DOI] [PubMed] [Google Scholar]
  53. Li L.; Zhang X.; Ning Y.; Zhang X.; Liu B.; Zhang Z.; Sivaguru P.; Zanoni G.; Li S.; Anderson E. A.; Bi X. Carbodefluorination of Fluoroalkyl Ketones via a Carbene-Initiated Rearrangement Strategy. Nat. Commun. 2022, 13 (1), 4280. 10.1038/s41467-022-31976-z. [DOI] [PMC free article] [PubMed] [Google Scholar]
  54. Yu Y.-J.; Zhang F.-L.; Peng T.-Y.; Wang C.-L.; Cheng J.; Chen C.; Houk K. N.; Wang Y.-F. Sequential C–F Bond Functionalizations of Trifluoroacetamides and Acetates via Spin-Center Shifts. Science. 2021, 371 (6535), 1232–1240. 10.1126/science.abg0781. [DOI] [PubMed] [Google Scholar]
  55. Sidnell T.; Wood R. J.; Hurst J.; Lee J.; Bussemaker M. J. Sonolysis of Per- and Poly Fluoroalkyl Substances (PFAS): A Meta-Analysis. Ultrason. Sonochem 2022, 87, 105944 10.1016/j.ultsonch.2022.105944. [DOI] [PMC free article] [PubMed] [Google Scholar]
  56. Shende T.; Andaluri G.; Suri R. P. S. Kinetic Model for Sonolytic Degradation of Non-Volatile Surfactants: Perfluoroalkyl Substances. Ultrason. Sonochem. 2019, 51, 359–368. 10.1016/j.ultsonch.2018.08.028. [DOI] [PubMed] [Google Scholar]
  57. Vecitis C. D.; Park H.; Cheng J.; Mader B. T.; Hoffmann M. R. Enhancement of Perfluorooctanoate and Perfluorooctanesulfonate Activity at Acoustic Cavitation Bubble Interfaces. J. Phys. Chem. C 2008, 112 (43), 16850–16857. 10.1021/jp804050p. [DOI] [Google Scholar]
  58. Rao U.; Su Y.; Khor C. M.; Jung B.; Ma S.; Cwiertny D. M.; Wong B. M.; Jassby D. Structural Dependence of Reductive Defluorination of Linear PFAS Compounds in a UV/Electrochemical System. Environ. Sci. Technol. 2020, 54 (17), 10668–10677. 10.1021/acs.est.0c02773. [DOI] [PubMed] [Google Scholar]
  59. Penner P.; Vulpetti A.. QM Assisted ML for 19F NMR Chemical Shift Prediction. ChemRxiv, 2023, 10.26434/chemrxiv-2023-sd3vq-v2 (preprint). [DOI] [PubMed]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

es3c09341_si_001.pdf (4.7MB, pdf)

Articles from Environmental Science & Technology are provided here courtesy of American Chemical Society

RESOURCES