Optimizing electrochemical conditions to allow high voltage

To conduct electrochemical investigations on the two LiB additives, it was first necessary to optimize the experimental conditions. The optimization parameters included pH, solvents, electrolytes, electrodes, and modifiers. The ideal parameter values are shown in Table 1. A complete enumeration of all EC parameters utilized for the transformation of TPFPB and TPFPP is provided in the Supporting Information (SI-2). BDD was selected for its extensive potential range and minimal background interference [63].

Table 1 EC parameters developed for transforming TPFPB and TPFPP, using the μ-PrepCell equipped with BDD electrode

One of the primary challenges encountered was the limited solubility of TPFPP in conventional EC solvents, including ACN, MeOH, and H2O, requiring the use of non-polar solvents and solubilizing solvents to prevent precipitation. Another considerable obstacle was the low and absent transformation (TPFPB and TPFPP, respectively) within the BDD potential range (–2.5 V to +3.5 V). This range is based on water electrolysis. Consequently, the level of water was constrained to mitigate the risk of electrolysis and to expand the potential window while maintaining environmental conditions. The resulting potential ranges for TPFPB were 0.0 to 4.1 V and −4.0 to 0.0 V, while for TPFPP the ranges were 3.5 to 4.1 V and −4.1 to −2.5 V. The potentials were ramped at a velocity of 20 mV/s and the experiments were conducted offline to account for the different detection techniques required for each substance.

In conclusion, these previously unreported conditions allow for extended potentials while maintaining environmental conditions. The need for high potentials was expected, given the known voltages (up to 4.2 V) that LiB electrolytes are subjected to in practice.

Identification and annotation of 49 TPFPB-derived TPs from several degradation pathways

Various laboratory simulation methods were utilized to mimic the environmental degradation of the LiB additive TPFPB. As an outcome of the investigation, a total of 49 different TPs were identified, of which 29 were conclusively identified by their sum formulae (Table 2, SI-3) and 28 were identified for the first time.

Table 2 Compilation of the results of the different simulation methods employed on TPFPB. The analysis was performed using LC-QTOF (negative mode). Compounds are considered confirmed if they were detected in at least two replicates of a simulation set. The proposed chemical formulae, intensities, and confirmation level (CL) are displayed for each compound. Adducts are often formed including formic acid (FA, HCOOH) or acetonitrile (ACN, CH3CN). The intensities are indicated relative to the strongest signal (vs: very strong) > 60%, (s: strong): 40–60%, (m: moderate): 20–40%, (w: weak) 10–20%, (vw: very weak) < 10%). Abbreviations of the simulation methods: EC (oxidative and reductive conditions); ECox (EC oxidative conditions); ECred (EC reductive conditions); UV (UV-C irradiation); Fe (Fenton reaction); H (hydrolysis with basic and acidic conditions); Hb (hydrolysis alkaline conditions); Ha (hydrolysis acidic conditions)

The confidence in identifying unknown compounds through HRMS techniques was evaluated using the categories suggested by Schymanski et al. [61]. Here, annotation confidence is based on analytical information including exact mass, isotope pattern, availability and conclusiveness of MS/MS data, library matches, rt comparison, and reference standards. Level 1 is the most reliable, as it requires confirmation by an analytical standard [64]. However, as no analytical standards are available for novel compounds, level 2b was the highest possible assignment for the identified TPs in this study. At level 2b and above, no structural isomers are possible. The annotation levels were developed for the non-targeted analysis of complex mixtures. Hence, their requirements can be considered as too strict in this work, as the possible chemical space for de novo annotations is highly restricted by the chemical elements of the precursor molecules. Nevertheless, we apply these levels to allow for insights into the possibility of structural isomers and to assess the reliability of the MS and MS/MS data.

Further validation could be obtained from the observation that numerous TPs were generated by multiple simulation methods. TP 19 emerged as the most prevalent product across all simulations, while certain methods showed near-complete conversion. The presence of other TPs was detected at low levels. The TOP assay simulation method yielded no results, as no TPs were detected following the simulation. This indicates that the TOP assay resulted in complete decomposition of the substance into metabolites not detectable by LC and GC–MS. Figure 2 depicts the proposed structures of the TPs and the transformation pathways. For illustration, one likely constitutional isomer was selected, as MS/MS spectra did not always provide distinctive formulas for derived TPs as implied by the CL (Table 2).

Fig. 2

Proposed structures of the TPs and transformation pathways from the different simulations of TPFPB. Numbering is according to masses (see Table 2). Literature references are given to substances with structural similarity. Red background: indicates toxicological relevance; gray background: indicates high occurrence

Modifications

Transformation of TPFPB resulted in the formation of a diverse array of products, including defluorination products, hydroxylation products, boronic acid esters, dioxin structures, diphenyls, diphenyl ethers, and oligomers. Many of these compounds were previously undocumented. To gain insight into the formation of some TPs, references to structurally similar compounds in Fig. 2 were employed. For example, Piers et al. and Jacobsen et al. reported adducts with nitriles [65, 66], while Di Saverio et al. described the coupling of B–O–B [67]. Furthermore, Britovsek et al. demonstrated the synthesis of borinic and boronic acid esters (B–O–Ar) [68].

Hydrolysis

Hydrolysis of TPFPB under alkaline conditions yielded a series of products lacking boron, as confirmed by EC under reductive conditions. The primary products are a fluorinated phenol, fluorinated diphenyl ethers (TP 6, TP 8), fluorinated diphenyls (TP 4, TP 5), and even a fluorinated dioxin formed with EC (TP 7). Chen et al. also observed the formation of C–C bonds with fluorinated phenols through electrophotocatalysis [69]. TP 19, the primary product of all simulations, is formed by hydrolysis. This process has been documented for TPFPB [65, 67]. The process commences with the reaction between boron and water, which ultimately results in the gradual release of C6F5H, in agreement with the findings of Piers et al. [65]. The released compound may subsequently react to TP 1 through hydroxylation and F release. This discrepancy could be attributed to either a rapid reaction or a limitation of the analytical detection method, potentially due to ionization or solubility constraints.

Defluorination

Dehalogenation reactions were observed in a number of TPs (e.g., TP 26, TP 20, TP 13, TP 11, TP 32). The formation of these TPs was predominantly achieved through UV-C irradiation. For instance, TP 20 was produced via the hydrodefluorination process, involving the photoheterolysis and hydrogen abstraction from a polar solvent [59, 70, 71], which is considered direct photolysis [72]. Indirect photolysis involves the generation of reactive oxygen species that react with the substrate, as observed with TP 26. This process was also documented by Arnold in 2020 [73]. Comparable products are generated through hydrolysis and the Fenton reaction (TP 45, TP 46, TP 48). Another potential reaction pathway which can be considered leads to the formation of compounds such as TP 11, TP 13, TP 32, and TP 18. This pathway involves the electrophilic addition of water to the double bond of the aryl group.

Boronic acid esters

A number of TPs indicate the formation of boronic acid esters. Similar compounds were reported by Britovsek in 2005 [68], but via a different synthesis from a different precursor. In this case, C–B bonds have to be hydrolyzed first, as observed for TP 10. As described by Brooks et al. [74, 75], the released aryl groups may form phenols, which subsequently react with boronic acids to form esters. Oligomerization may occur, as observed in previous studies, with the formation of [B–O–B] bonds within the oligomers, known as diboroxanes [67, 74, 76].

Oligomers

The formation of oligomers (TP 48, TP 45, TP 46), including structures analogous to those of dioxin, diphenylethers, and even diboroxanes (TP 48), has been observed in the course of a number of different simulations. In the environment, oligomers exhibit significantly altered chemical fate. They are reduced in mobility and water solubility, and potentially changed in leachability and biological activity. This immobilization prevents the substances from leaching into groundwater and instead incorporates them into soil organic matter. However, biotic alterations could alter the substances again and lead to modified substances that would have a delayed health risk, such as phytotoxic effects, impacts on soil organisms, or uptake by crops [77, 78].

Electrochemistry

The highest number of TPs was formed by EC, although confirmation of related structures was not consistently achieved by secondary simulation methods. Nevertheless, these findings make EC an interesting screening method for the simulation of several abiotic processes. Furthermore, the transformation of TPFPB produced TPs containing many of the previously described alterations (e.g., TP 41, TP 15, TP 48, TP 45). These TPs comprise dioxin structures, diphenyl structures, boronic acid esters, hydroxylation, and defluorination. However, due to the complex and novel nature of these substances, a toxicological or environmental assessment of these specific TPs is not feasible. Nevertheless, for certain structural equivalents, general extrapolations may apply.

Toxicologically relevant equivalents

The halogenated equivalents of the TPs formed by alkaline hydrolysis are highly toxic compounds such as polybrominated diphenyl ethers, polybrominated biphenyls, and dioxins (polychlorinated dibenzo-p-dioxins), which are regulated under Annexes A and C of the Stockholm Convention on Persistent Organic Pollutants [79]. The formation of dioxins or dioxin-like compounds is a major concern due to their high toxicity, persistence, environmental distribution, bioaccumulation in all living organisms, and carcinogenic potential. Compared to their brominated or chlorinated analogues, fluorinated dioxins show less toxicity but pose a greater risk to the immune system [80, 81]. Conversely, one study found fluorinated phenols to be more toxic than other halogenated analogs [82].

Electrochemistry for biotic processes

This study has focused on the abiotic formation of TPs. Another significant mechanism, namely, biotic transformation in the environment, was not considered, but EC has also been demonstrated to simulate biotic processes in several studies [42, 83,84,85]. Existing literature indicates the potential for biotic degradation of fluorinated aromatics through enzymatic removal of a fluorine by hydroxylation, resulting in the formation of phenols, catechols, hydroquinones, and trihydroxyfluorobenzenes [58, 86, 87]. The degradation of these compounds ultimately results in the formation of two end products: maleic acid and muconolactones. It is noteworthy that our study additionally identified hydroxylation and dihydroxylation products (TP 26, TP 31, TP 14) for TPFPB, which may reflect biotic processes.

EC and UV-C treatment leads to several novel TPs of TPFPP

The environmental impact of the TPFPP additive utilized in LiBs was investigated through a series of laboratory simulations. The investigation identified a total of nine distinct TPs of TPFPP, eight of which were conclusively identified by their respective sum formulas (see Table 3 and SI-3). Of these, seven are reported here for the first time.

Table 3 List of TPs of TPFPP generated from simulation methods, including suggested chemical formulas, intensities, CL, and simulation technique; TP order is according to molecular masses. Analysis was performed by GC-LRMS and GC-QTOF measurement (70 eV and 18 eV). If detected in two replicates of a simulation set, a confirmation was assumed. Intensity values are derived from a ratio relative to the highest signal, which is typically the precursor (vs: very strong) > 60%, (s: strong): 40–60%, (m: moderate): 20–40%, (w: weak) 10–20%, (vw: very weak) < 10%). Simulation methods are indicated by abbreviations: EC: oxidative conditions ECox: NH4Ac; ECox2: FA; UV (UV-C irradiation); Fe (Fenton reaction); Ha (hydrolysis acidic conditions). * TP molecule ion: 567.9928 (at 18 eV) and 554.9994 (at 70 eV)

The data presented in Table 3 were obtained by conducting a GC-LRMS and GC-QTOF analysis. The objective of the GC-QTOF analysis was the detection of exact masses and the confirmation of the detection of the molecular ions of the TPs. To this end, a two-step process was undertaken. This involved the measurement with two different ionization energies: standard conditions (70 eV) and LE-EI (18 eV). LE-EI yielded lower fragmentation and confirmed the molecule ions of all TPs except for TP 9. The molecular ion of TP 9 (564.9928 m/z) could only be detected by LE-EI. Confidence in the screening results was increased when a TP was detected with multiple methods. One TP (TP 7) could be confirmed by a spectral library entry match. In addition, the CL provides further information on confidence and even isomeric possibilities.

The applied simulation methods led to a relatively small number of TPs. However, some TPs were identified with relatively high intensities, especially TP 7 and TP 1. Once more, the TOP assay simulation method did not yield any TPs, suggesting that the substance had been completely decomposed by the conditions. Moreover, additional methods, such as alkaline hydrolysis and EC under reductive conditions, did not result in any changes in TPFPP. Consequently, no data are presented. Conversely, varying EC oxidation conditions yielded disparate outcomes as indicated by ECox and ECox2. Figure 3 illustrates the proposed structures of the TPs and the proposed decomposition pathways. To exemplify the structures, a probable constitutional isomer was selected, given the fact that the EI spectra were not always able to provide unambiguous formulae for the derived TPs, as indicated by the CL.

Fig. 3

Proposed structures of the TPs from different simulation experiments of TPFPP, organized as possible transformation pathways (numbering refers to Table 3). Simulation method abbreviations: ECox: NH4Ac; ECox2: FA; gray background: indicates high occurrence

Modifications

The generated TPs demonstrate oxidation on the phosphorous (TP 7) and several modifications on the aryl groups, including defluorination (e.g., TP 5, TP 3, TP 2), hydroxylation with and without fluorine release (e.g., TP 1, TP 4, TP 6, T 9), and diphenyl as well as diphenyl ether formation and dioxin structures (TP 1). TP 1, TP 4, TP 6, and TP 9 exhibit a multitude of modifications, whereas TP 4, TP 6, and TP 9 even experience a loss of the aromatic structures, possibly due to the hydrogenation of arene groups and water addition. The choice of the constitutional isomers of the TPs in Fig. 3 is based on an intriguing discovery. In phosphines, phosphorus(III) exhibits a positive mesomeric effect due to the free electron pair, which would therefore direct a substitution in the meta position. Nevertheless, the experimental data from Hanna (1977) with TPFPP yielded only products with para substitution. It is postulated that the overlap of the p-orbitals of the aryl ring and the empty orbitals of phosphorus stabilizes the negative charge on the phosphorus-bound carbon, resulting in para substitution [88]. TP 7, a phosphine oxide, is the only TP previously described [88, 89], and hence, it is confirmed by a spectral library entry. Given that phosphines are known to possess an oxophilic character, the formation of TP 7 is highly probable. The resulting product is expected to exhibit greater stability than the precursor, due to its inability to degrade further using the applied methods. The data indicate that the phosphine oxide exhibits even reduced reactivity, potentially attributed to a reduction in Lewis basicity. This emphasizes the necessity of eliminating aryl-bound fluorine for effective degradation.

Defluorination

However, defluorination is an important step in the overall decomposition process. Here, we report three TPs with specific defluorination, TP 5, TP 3, and TP 2, formed by UV-C irradiation and electrochemical oxidation. The proposed underlying mechanism is hydrodefluorination. This reaction is initiated by photoheterolytic cleavage and hydrogen abstraction in a polar solvent [59, 70, 71]. In addition to its environmental relevance, specific defluorination is also of importance with regard to partially fluorinated substances. These compounds are of great interest as agrochemicals and pharmaceuticals. This is because the synthesis of these partially fluorinated substances remains a challenging process, often requiring a complete fluorination followed by selective defluorination steps to obtain the desired product. The irradiation and EC methods reported here are therefore of interest for the synthesis of fluoroarenes [69, 90]. Another defluorination mechanism is indirect photolysis. Only TP 1, with a dioxin structure similar to that of TPFPB, was identified as the product of indirect photolysis (TP 16, TP 38).

Electrochemistry

The EC process yielded the greatest number of different TPs but required the use of elevated voltages. In contrast to the reductive electrochemical conditions, which did not yield reproducible products, the oxidative EC oxidation process resulted in the generation of reduction products. The observed outcome could suggest that the electrode and/or the conditions utilized are unsuitable for the reductive conversion of the precursor. Nevertheless, numerous different combinations of conditions (electrolytes, pH, solvents) did not alter the observed outcome. One possible explanation is that the high electron density within the aromatic structures and on phosphorus(III) impairs reactions under purely reductive conditions with a different underlying mechanism. Notably, alterations to the conditions within the EC (see SI-2 for a comprehensive enumeration of all EC parameters) did not lead to transformation under reductive conditions. Instead, the oxidative mode resulted in the formation of TP 7.

Consequently, the utilization of FA/acidic conditions results in a modification of the reaction and the formation of an oxidation product, whereas neutral conditions employing NH4Ac lead to defluorination and the generation of additional products. Some of the additional products with multiple modifications (TP 4, TP 6, and T 9) are formed exclusively with EC, albeit in low yields. While the proposed structures, based on the spectral analysis, are relatively plausible, their certainty is limited (CL = 4). In the absence of recognizable transformation pathways or precursors and no literature verification, it is uncertain whether these compounds have environmental significance, and annotation requires further confirmation. Nevertheless, the confirmation of all other TPs formed using EC demonstrates that EC is an effective screening approach for investigating TPs, as it is both simple and rapid.

Our results indicate a remarkable stability of the P–Ar bond, as evidenced by a resistance to cleavage even under conditions of high energy input. It is noteworthy that the hydrogenation of the arene groups, rather than the conventional bond cleavage, appears to be the primary mechanism responsible for the structural modification of these compounds. Potential competing pathways, including dehalogenation, hydrogenation, and hydroxylation, may contribute to the observed changes in chemical composition. These changes were detected even though the hydrogenation of arenes requires stringent conditions, which usually include a catalytic system.

Electrochemistry for biotic processes

For phosphines, the biotic transformation appears to differ from that of other substances, with limited literature evidence suggesting that phosphorus is exclusively oxidized (TP 7) in these reactions. This oxidation reaction, which is identified in this study as a predominant transformation, may also reflect biotic metabolism [91, 92]. This further supports the assumption that EC results can also be considered as a simulation of biotic transformation and, therefore, adds another important aspect to this study.

Exceptional stability

The transformation simulation of TPFPP yielded an unexpected result: its exceptional stability, despite the fact that phosphanes are strong oxophilic. This exceptional stability is evidenced by both the strict conditions necessary for its decomposition using the employed simulation techniques and the low minimal number of TPs observed. This stability can be attributed, in part, to its limited solubility in protic solvents, coupled with the focus of these methods on oxidative decomposition on the fluorinated ring, which is important to prevent the formation of stable intermediates. This process may be less favorable for a Lewis base with its free electron pair and therefore positive mesomeric effect, typically acting as an electron donor. This effect impairs a nucleophilic attack usually done by oxygen species on the fluorinated ring, which is already reduced in reactivity. This might explain the superior stability, especially in comparison to TPFPB. As a result, we categorize TP 7 as a possible “end product” and forever chemical of tomorrow. Even in the case of PFAS, which are commonly referred to as “forever chemicals” and consist of at least one fully fluorinated carbon, it is not necessarily the parent compounds that are “end products,” but the TPs.