Comprehensive nontargeted analysis of fluorosurfactant byproducts and reaction products in wastewater from semiconductor manufacturing

Semiconductor manufacturing employs per‑ and polyfluoroalkyl substances (PFAS) as fluoroasurfactants to enhance the quality of photolithography lines. Our research, employing a fragment‑based approach to investigate nontarget PFAS, overcomes conventional homologous series limitations. In a mixture of PFAS standards used as a quality control sample, 92% (36 out of 39 compounds spanning 11 compound classes) were detectable through the fragment‑based nontarget procedure. This indicates the effectiveness of this approach in identifying the hydrophobic and hydro‑ philic characteristics of various fluorosurfactants. Eighty‑three PFAS were detected in wastewater and effluent samples from semiconductor industry, including 29 newly discovered compounds, categorized into three groups. First, besides detecting perfluorobutane sulfonamido ethanol (FBSE), various fluoroalkyl chain structures of FBSE derivatives were initially identified in wastewater. These include perfluorobutyl ether sulfonamido ethanol, unsatu‑ rated FBSE, and hydrogen‑substituted FBSE. These derivatives were also detected in trace amounts in commercial authentic standards of FBSE. To quantify their presence, we analyzed the FBSE derivatives from the authentic stand‑ ard, and the relative proportions of these derivatives contribute to approximately 0.5%. This suggests that the FBSE derivative series detected in wastewater may arise from byproducts of chemical formulations used in the manufac‑ turing of semiconductors. Second, transformation products from FBSE during oxidation, including the first identified intermediate transformation compounds, perfluorobutane sulfonamido acetaldehyde and its hydrate, were discov‑ ered. Third, diverse reaction products are generated from the intricate processes of semiconductor manufacturing, which utilize strong acids, bases, and solvents under UV light or heated conditions. These processes include the for‑ mation of PFAS‑related compounds through hydration, sulfonation, oxidation, and nitrification. This study revealed 25 isomeric PFAS, encompassing headgroup isomers and functional tail group isomers. These findings underscore the importance of comprehending diverse reactions and the overall emission compositions of PFAS in semiconductor wastewater, highlighting its complexity and presenting challenges for subsequent wastewater treatment.

PFAS have aroused concern owing to their health [13] and environmental impact [14].Under the Toxic Substances Control Act Sect.8(a) (7), the United States Environmental Protection Agency mandates reporting regulations for PFAS in 2023, requiring manufacturers and processors to submit detailed information on PFAS substance usage from 2011 to 2022, covering identification, applications, quantities, environmental impacts, and health effects [15].Perfluorohexanesulfonic acid (PFHxS) including its salts and over one hundred PFHxS-related compounds, defined as structures containing C 6 F 13 capable of degrading into PFHxS [16], have been listed in the Stockholm Convention to restrict and eliminate their use since 2022.A significant portion of PFHxS-related compounds primarily comprises sulfonamides and sulfonamido substances, posing challenges in analysis due to the lack of authentic standards.Despite the convention's recommendation to use total oxidable precursors [17] and extractable organic fluorine [18] to assess the presence of precursors, methods for identifying individual compounds are currently unavailable [19].
To identify nontarget PFAS, high-resolution tandem mass spectrometry has been used widely in contaminated soil [20], surface water [21], wastewater from a fluorochemical manufacturing [22], and municipal wastewater [23].Mass defect and homologous patterns are potent tools for uncovering PFAS within a series of homologous substances.Chemicals utilized in industrial processes display diverse and precise functionalizations, and the complex blend in wastewater prompt various reactions like hydration, oxidation, and deamination [24,25].Given that these reactions occur mostly in the polar functional section, each transformation yields distinct mass defect values, making the utilization of homologous patterns for screening impractical.Consequently, our research employed a fragment-based approach to investigate PFAS-related substances by screening the polar functional groups and nonpolar fluoroalkyl chains of fluorosurfactants.Ten wastewater samples from five semiconductor plants and three effluents from their associated wastewater treatment plants (WWTPs) were investigated for comprehensive identifications of PFASrelated byproducts of chemical formulations, transformation products generated during the oxidation process, and diverse reaction products generated from the intricate processes of semiconductor manufacturing.

Sampling
In November 2020 and January 2021, ten wastewater samples (no.184-187, 190, 191, 194-197) were collected from five semiconductor plants (A, B, C, D, E) situated in three different Science Parks, along with three effluent samples (WWTP I, WWTP II, WWTP III) from their respective wastewater treatment plants, as depicted in Fig. S1 of Supplementary Materials (SM).We collected wastewater downstream from each factory, prior to its entry into the main wastewater pipe.Hence, the wastewater is specifically attributed to the discharge of each factory.In addition, effluents were gathered from discharge points of wastewater treatment plants, all employing the activated sludge process.The samples were kept in 1 L polypropylene bottles and sent to the laboratory at 4 °C for analysis.

Pretreatment
A solid phase extraction cartridge (Oasis WAX, Waters) features a weak anion-exchange and reversed-phased sorbent was applied to condense water samples.The capture strategy involves performing the load and wash procedures at pH < 3. The elution is performed at a neutral pH for neutral PFAS and at a pH > 8 for acidic PFAS.Detailed pretreatment procedures are included in the SM.

Analysis
Table S1 in SM lists authentic reference standards used to establish the patterns for scanning fragmentation and neutral loss values in PFAS screening.Table S2 includes lists of authentic reference standards employed to confirm the identified PFAS.The injection of a 50-100 μL extract was carried out using a UPHLC binary pump system and interfaced with a negative electrospray ionization source coupled to a Q-Exactive hybrid mass spectrometer.The mass spectrometric analysis was conducted in full MS and 20 data-dependent MS2 scans, involving a full scan analysis ranging from m/z 77-1155 with applied fragmentation energy (NCE 15 and 50).Mass calibration before each analysis ensured accurate results, and the inclusion of acetic acid in the calibration solution is crucial to enhance mass accuracy, especially for low mass measurements.Specific details regarding the LC gradient program and mass spectrometry settings can be found in Table S3.
The data analysis workflows for nontarget analysis of PFAS are illustrated in Fig. 1.Thermo Scientific Compound Discoverer 3.3 software was employed for compound detection and identification (Fig. S2).In particular, the Compound Class Scoring node (MS2) and the Search Neutral Losses node (MS2) were utilized for searching fragments and neutral losses, respectively, while the Search for ChemSpider node screened databases based on formulas within a 5 ppm mass tolerance (MS1).The detailed specifications of the workflow nodes are outlined in Table S4.The identification confidence levels (CLs) were adapted from Charbonnet's classification [26], as illustrated in Table S5.In databases equipped with an MS2 library (e.g.mz cloud), the matching is classified as Fig. 1 The workflow of nontarget analysis CL2a.In databases that provide only formula information without MS2 libraries, identification relying on formula matching (isotope pattern matching, MS1) is classified as CL4.Moreover, CL is determined by the quantity of fragments, with a single fragment corresponding to CL3, while two or more fragments are classified as CL2 or above.Based on the scope of identification, we divided our analysis into two aspects: CL1-identified PFAS and tentatively identified PFAS as shown in Table S5.

Quality control and quality assurance (QA/QC)
To verify the process of the nontarget procedure in Fig. 1, a PFAS standard mixture containing 39 compounds from 11 PFAS categories listed in Table S1 were prepared as a QC sample.The result showed that 92% (36 out of 39 compounds) were detectable through the processing of fragment-based nontarget analysis as detailed in Table S6.The specifics regarding the detection of mass errors, fragments, neutral losses of the 39 compounds are provided in Table S7 and Table S8.

Nontarget analysis of PFAS and identified classes
Screening functional polar heads is crucial for identifying nontargeted PFAS without apparent fluorocarbon fragments and discovering various reaction products formed in the polar functional groups during the semiconductor manufacturing.Hence, we have streamlined the approach by summarizing the identified surfactant PFAS search through fragment and neutral loss screening of polar functional groups for the 12 classes with distinct polar functional groups, as shown in Table 1, serving as a valuable reference for nontarget analysis of PFAS.For miscellaneous perfluoroalkane sulfonamides (FASAs) and perfluoroalkane sulfonamido ethanols (FASEs; subclass 1d-1i and subclass 2 g-2 h of Table 2 2. The following is a discussion of the identified results for the 12 classes with distinct polar functional groups.

Class 1 Sulfonamides
By searching for [SO 2 N] -fragment searching and excluding substances lacking fluorine based on isotopic formulas based on the workflow in Fig. 1, we identified the predominant FASAs and various FASA series chemicals, resulting in the detection of 9 subclasses (subclass 1a-1i of Table 2).Among them, 3 subclasses (1a-1c) displayed variations in the fluoroalkyl chain but had a polar head of -SO 2 NH 2 .This class exhibits a unique characteristic fragment, [SO 2 N] -(77.9644), which is observed in FBSA, hydro-substituted FBSA (H-FBSA), and ether FBSA (E-FBSA).Notably, no obvious fragmentations from the fluoroalkyl tail were detected during screening, suggesting that [SO 2 N] -is a feasible screening tool for identifying these subclasses.In Fig. S3, due to the lack of fragments to determine the position of the hydrogen, the CL for H-FBSA remained at 3a.  H] - (64.9697) were observed, along with the weak fragment [SO 2 F] -(82.9603).N-Methyl FBSA (MeFBSA) was identified in the effluents of WWTP I and WWTP II (Fig. S6a).It was subsequently confirmed by an authentic standard of MeFBSA, matching both in RT and MS2 spectra, resulting in setting the CL to 1a (Fig. S6b).To the best of our knowledge, 9 identified PFAS (including 2 E-FBSA isomers) of these subclasses have been reported for the first time, as shown in Table 2. Remarkably, while previous studies illustrated that fluorotelomer sulfonamides detected with the [H 2 NSO 2 ] -fragment [20,27], and the hydrogens of the [H 2 NSO 2 ] -fragment is attributed to the substantial presence of hydrogens in the fluorotelomer.
After obtaining the authentic standard of MeFBSE, we found that while the RT and MS1 spectra matched, the MS2 spectrum did not resemble that of the sample (Fig. S11b).As a result, MeFBSE is classified as CL 3a due to the presence of only one fragment corresponding to the headgroup, consistent with the identification criteria outlined in Table S5.However, due to lacked information for the identification of MeFBSE, further study is needed.Remarkably, fluorotelomer sulfonamido ethanol, such as 6:2 FTSAm-EtOH, which was identified by other researchers are detected, contains the [SO 2 NC 2 H 4 O] − (121.9912)fragment [27].This fragment, elucidated as [SO 2 NC 2 H 4 O] − (121.9912),remains unaffected by the functional fluoroalkyl chain, including fluorotelemer, serving as a unique indicator of the class of sulfonamido ethanol.

Class 4 Sulfonamido carboxylic acid
This subclass features a structure with sulfonamido linked to a carboxylic acid group, releasing the carboxylic acid group during collision and producing characteristic fragments such as [SO 2 N] − (77.9656) and [SO 2 F] − (82.9610).A precursor ion at 355.9654, with a neutral loss of CH 2 CO 2 (8.31 ppm), was identified as perfluorobutane sulfonamido acetic acid (FBSAA) (Fig. S14a) and confirmed by its authentic standard at CL1a (Fig. S14b).The features observed at 337.9743 in Fig. S15, which include the detection of neutral loss of CH 2 CO 2 , fragments such as [SO 2 N] − and [C 4 F 7 ] − , and the formula inferred through isotopic pattern matching as C 6 H 5 F 8 O 4 SN (mass tolerance 1.26 ppm, within 5 ppm), were further suggested to represent hydrogen-substituted FBSAA (H-FBSAA).Additionally, the features observed at 369.9805 in Fig. S16a, exhibiting neutral loss of CH 2 CO 2 , along with fragments such as [SO 2 F] − and [C 4 F 9 ] − , and the formula inferred through isotopic pattern matching, were proposed to represent MeFBSAA.This identification was further supported by matching the RT and MS2 spectra of the authentic standards (Fig. S16b).Furthermore, features as 371.9594 are proposed as perfluorobutyl ether sulfonamido acetic acid (E-FBSAA), which has two structural isomers, was separated with the column at RT of 18.7 and 19.7 min.Fragments of fluoroether [CF 3 O] − and [C 2 F 5 O] − were observed and providing the position of oxygen for the CL at 2b.These findings have been summarized in Table 1 to serve as a reference for assisting in the identification of nontarget PFAS.Notably, a structural isomer of MeFB-SAA, was identified for the first time as perfluorobutane sulfonamido propanoic acid (FBSPrA) (Fig. S17).The neutral loss of C 2 H 4 CO 2 released from collision indicates propanoic acid.The presence of only [SO 2 F] − (82.9603) fragments, not [SO 2 N] − (77.9644), serves as a distinguishing rule between perfluoroalkane sulfonamido acetic acids (FASAAs) and N-alkyl FASAAs.Notably, the distinct neutral loss of CH 2 CO 2 might be affected by the significant presence of hydrogens in the fluorotelomer, resulting in the neutral loss of HF instead [27].However, the fluorotelomer series was not detected in this study; additional information for exclusion purposes is provided.

Class 5 sulfonamido diacetic acid
The MS2 spectrum of sulfonamido diacetic acid shows two pairs of neutral losses of CH 2 CO 2 , indicating the presence of the diacetic acid structure.The main characteristic fragment was [SO 2 F] − (82.9603), while the intensity of [SO 2 N] − (77.9644) was notably weak.Perfluorobutane sulfonamido diacetic acid was identified with CL to 2b [10].

Class 6 N-hydroxyethyl sulfonamido acetic acid
The substance with the precursor ion 399.9903 was identified by the specific fragment of [O 2 SNC 2 H 4 O] − (121.9912).An obvious fragment 341.9852 was referred to as the FBSE fragment.The difference between 341.9852 and 399.9903 represents the neutral loss of CH 2 CO 2 .Consequently, we referred to this substance as N-(2-hydroxyethyl) FBSAA (FBSEE mono-ol monoacid) at CL 2b [10].

Class 7 Sulfonamido acealdehyde
The class with the structure where sulfonamido linked to acetaldehyde.A precursor ion [C 6 H 3 F 9 NO 3 S] -was observed with the RT of 13.1 min and the RT of 30.1 min, identified for the first time as perfluorobutane sulfonamido acetaldehyde in its enol form and keto form (Figs. S18a and S18b).The enol form illustrated plenty of fragments ([SO 2 N] -, [SO 3 ] -, [SO 2 F] -, [SO 2 NCH 2 CHO] -), while the keto form showed simple fragments ([SO 2 ] - and [SO 2 H] -).

Class 8 Sulfonamido acealdehyde hydrate/hemiacetal
[C 6 H 5 F 9 NO 4 S] -(357.9812,2.99 ppm) was observed with a neutral loss of H 2 O which was identified as perfluoroalkane sulfonamido acetaldehyde hydrate, as water adds to the carbonyl function of acetaldehydes (Fig. S19).Moreover, [C 7 H 7 F 9 NO 4 S] -(371.9954,-0.89 ppm) was detected with a neutral loss of CH 3 OH, which indicated the production of additional reaction of methanol to acetaldehyde.Hydrates and hemiacetals are the products of addition reactions of oxygen-based nucleophiles, such as water and methanol, to aldehydes, which have been reported for the first time.

Class 9 carboxylic acid
Integrating CO 2 neutral loss into the nontarget approach improves the identification of carboxylic acid-containing compounds, revealing their collision characteristics in the collision activated dissociation cell, including subclasses 9a-9e, encompassing perfluoroalkyl carboxylic acids (PFCAs), unsaturated PFCAs (U-PFCAs), perfluoroalkyl ether carboxylic acids (E-PFCAs), unsaturated-E-PFCAs (U-E-PFCAs), hydro-substituted PFCAs (H-PFCAs), and hydro-substituted E-PFCAs (H-E-PFCAs).The MS2 spectrum of U-E-PFCA(C6) is shown in Fig. S20.The detection of neutral loss as CO 2 alongside other fluorinated fragments ([ and confirmation through isotope pattern matching with a formula of C 6 F 9 O 3 H allowed us to propose the structure of U-E-PFCA (C6) as CL 3a, due to uncertainties regarding the position of the unsaturated bond.In addition, H-PFCAs, H-E-PFCAs and H n -E-PFCAs (subclasses 9d and 9e) produced neutral loss of CO 2 HF, which was derived through the combination of CO 2 and HF, with HF originating from the hydrogen-substituted fluoroalkyl chain, as their MS2 spectrum are shown in Figs.S21 and  S22 for H-PFCA(C4) and H 2 -E-PFCA(C4), respectively.Contrarily, achieving a neutral loss of CO 2 in the PFCA series typically requires the presence of [M-H] -as a precursor ion and [M-H-CO 2 ] -as a product ion.However, due to E-PFCA's inherent susceptibility to in-source collision-induced dissociation [21,28,29], CO 2 can be lost, resulting in the formation of [M-H-CO 2 ] -as the precursor ion.Subsequently, within the collision chamber, C-O cleavage generates [C x F y O z ] -fragments.Thus, there is no neutral loss of CO 2 between the precursor ion and the production ion.This occurrence was observed during our QC sample testing with HFPO-DA (GenX), where the precursor ion was identified as [M-H-CO 2 ] -(284.9779) in Table S7 and the product ion as [C 3 F 7 O] -(184.9842) in Table S8, precluding a neutral loss of 44.Therefore, E-PFAC detection relies on [C n F 2n+1 O] -fragments and isotope pattern matching rather than CO 2 screening.This aspect is also outlined in Table 1, highlighting the importance of simultaneously scanning the fluorinated tail and polar head to enhance the effectiveness of fragment-based nontarget analysis.

Class 10 dicarboxylic acid
Perfluoroalkyl dioic acids (PFdiCA, C3-C8) are comprised of the functional group of dicarboxylic acid.PFdiCA (C3-C5) exhibited a neutral loss of CO 2 ; PFdiCA (C6-C8) lacked the detection of neutral loss of CO 2 , presumably due to their low abundance in the samples.The fragment-based fluoroalkyl chain as C n F 2n-1 and the isotopic pattern of fluorine could be alternatives to identify this subclass.Fragments of [

Class 12 Sulfinic acid
Perfluorobutyl sulfinate (PFBSi) was identified with clear fragments of [SO 2 F] -as the specific feature of sulfinic acid.It was further confirmed with an authentic standard based on the RT and the MS2 spectrum.Perfluoroalkane sulfinic acids, arising from the degradation of commercial precursor compounds containing the C n F 2n+1 SO 2 N moiety, may act as degradation by-products of fluorosurfactants in 3 M foam [10,30].

Byproducts from chemical formulation
Two primary methods for PFAS production are the electrochemical fluorination (ECF) process, favored by 3 M [31], and the telomerization process, employed by DuPont [32].The ECF process generates byproducts, along with both shorter and longer PFAS, with a higher prevalence of branched PFAS, while the telomerization process primarily yields linear PFAS [33].In our previous study on semiconductor wastewater, FBSE was found at concentrations ranging from 0.883 to 482 µg L −1 , while perfluoropropane sulfonamido ethanol (FPrSE) and perfluoroethane sulfonamido ethanol (FEtSE) were detected at semi-concentrations of n.d. to 0.037 µg L −1 and n.d. to 0.014 µg L −1 , respectively [10].FBSE is associated with 3 M's electronic surfactant 4200 [7], which is added to buffered hydrofluoric acid for etching solutions in semiconductor manufacturing to enhance wetting properties and improving pattern quantity performance.In this study, we reported several FBSE derivatives with varying fluoroalkyl chain in wastewater, including those with backbones containing oxygen as ether (E-FBSE), those with backbones containing unsaturated bonds (U-FBSE), and those with hydrogen-substituted F as polyfluoroalkyl chains (H-FBSE).The varied RTs of these substances compared to that of FBSE are as follows: H-FBSE (15.8-19.1 min) < U-FBSE (16.9 min) < H-E-FBSE (21.1 min) < U-E-FBSE (23.3 min) < FBSE (29.9 min) < E-FBSE (33.8-34.9min).The use of a reversed-phase C18 column for analysis indicates that, in terms of polarity, only E-FBSE was less polar than FBSE, while the others were much more polar, resulting in faster elution.The aforementioned FBSE derivative series were found as impurities in the authentic standard of FBSE.Their relative area proportions, shown in Table 3 in descending order, were as follows: FPrSE (0.21%), E-FBSE (0.10%), H-FBSE (0.095%), FBSAA (0.036%), U-E-FBSE (0.033%), FEtSE (0.014%), U-FBSE (0.007%), headgroup isomer of FBSAA (0.002%), and H-E-FBSE (0.001%), contributing to a total of approximately 0.5%.In addition, H-FBSA (0.01%) and E-FBSA (0.06%) are included in the authentic standard of FBSA, while H-FBSEE diol (0.4%) is included in the authentic standard of FBSEE diol, as shown in Table 3.The presence of FBSE derivatives, FBSA derivatives, and FBSEE diol derivatives in the authentic standard suggests that these derivative series detected in wastewater may originate from the byproducts of chemical formulations, like 3 M's electronic surfactant 4200, which contains FBSE as a primary component in semiconductor production.However, these byproducts may arise from transformation processes during manufacturing and among the treatment of wastewater, including the elimination of HF to produce unsaturated PFAS [34,35] and defluorination leading to hydrogen-substituted PFAS under photoelectrochemical processes [36].Additionally, under biotic metabolism, alcohol oxidation can lead to the formation of carboxylic acids through hydration and oxidation processes [10,37], with further discussion below.This suggests that in wastewater treatment, using hydrophobic interaction separation methods such as activated carbon adsorption may lead to lower removal efficiency for these polar substances, decreasing susceptibility to adsorption removal [38].When discharged from wastewater treatment plants, the environmental distribution of these substances may differ significantly due to their distinct properties.Notably, the semiconductor industry encompasses various suppliers, including those for logic integrated circuits, dynamic random-access memory devices, solid-state drives, photodiodes, photo-imaging sensors, and more.Generally, plants A, B, C, and D primarily manufacture similar products, while plant E specializes in a different product line.We discovered This observation could serve as a distinguishing characteristic for identifying different production lines.

Reaction products
Due to the intricate nature of semiconductor processes, among the 12 prominent semiconductor industries in South Korea, 11 utilize 135 chemical constituents [39].These include sulfuric acid, chromic acid, tetramethyl ammonium hydroxide, ethylene oxide, potassium dichromate, isopropanol, and formaldehyde, 33% (range: 16-56%) of the chemical compositions remain undisclosed due to commercial confidentiality.Notably, the undisclosed ingredients are predominantly employed in the photolithography process.The complex chemical condition involved in these semiconductor processes include strong acids, alkalis, and potent oxidants, and UV light is used in the photolithography process.Among the intricate blends of industrial chemicals in wastewater, various reactions may take place, including hydration, oxidation, sulfonation, amide formation, and nitration.Additionally, during the biological treatment in wastewater treatment plants, reactions such as oxidation, deamination, desulfonation, and dicarboxylation occur [10,37].Due to the exceptional stability of the fluoroalkyl chain, reactions in the polar functional section led to distinct mass defect values for each transformation.This renders the use of homologous patterns for screening impractical.Our fragment-based approach overcomes the limitations of conventional homologous series.In total, we have identified 83 PFAS from 43 subclasses, with 29 substances reported for the first time.(Table 2).
Oxidation of FBSEE diol yields FBSEE diacid, while FBSE oxidation produces tentatively identified FBSAA, both with tentative identifications at CL 2b due to the lack of standards [10].A key distinction in current study is the inclusion of a standard for FBSAA confirmation.Through RT alignment and consistent MS2 spectra, (Fig. S14a) we verified the presence of FBSAA at 15.5 min with CL1a.Notably, an isomer of FBSAA with higher polarity presented at RT 9.9 min, showcasing distinct MS2 spectra (Fig. S14c) that suggest structural differences.Considering the potential isomerization between acid and diol forms [40], fragment analysis led to the proposal of a diol structure for the isomer (Fig. S14c).In addition, we introduce intermediate products of FBSE oxidation: FBSAcAL (RT 30.0 min, Fig. S18b) and its hydrated form as FBSAcAL hydrate (RT 30.0-32.6 min, Fig. S19), detected for the first time via mass spectrometry.Another structural isomer of FBSAcAL (7a-1) at RT 13.0 min was identified with higher polarity as its MS2 spectrum shown in Fig. S18a.Here, we propose an oxidation pathway, including the first publication of intermediate transformations of aldehyde (enol and keto form), aldehyde hydrate, and two isomers of FBSAA in Fig. 3.

Structural isomers from headgroup
FBSA-PrNO 2 and FBSE-Am are isomers with distinct polar head groups-sulfonamide and sulfonamido ethanol, respectively (Fig. 2).FBSA-PrNO 2 exhibit [SO 2 N] - fragment along with an additional signal for the neutral loss of C 3 H 5 NO 2 , categorizing it as miscellaneous FASAs (Subclass 1 h).In contrast, FBSE-Am demonstrates a specific signal at 121.9917, confirming its classification as FASEs, and its neutral loss of CONH categorizes it within the miscellaneous FASEs (Subclass 2 h).The RT of FBSA-PrNO 2 was 8.8 min, while that of FBSE-Am was 29.9 min, indicating significant differences in polarity.Thus, the headgroup isomers exhibit unique reactivity and physicochemical properties, may impact the variations in sludge metabolism during wastewater treatment in WWTPs [38].Furthermore, E-FBSEs and FBSAcAL hydrate (Subclass 8a) are differentiated by their respective polar head groups: sulfonamido ethanol and sulfonamido acetaldehyde hydrate (Fig. 4).Specifically, fragment 121.9917 is Fig. 3 The oxidation pathway of FBSE to FBSAA involves intermediate compounds, including aldehyde and aldehyde hydration.Two isomers of FBSAcAL (keto and enol form) and two isomers of FBSAA were detected unique to sulfonamido ethanol, while the hydrate can be identified by the neutral loss of H 2 O. Finally, MeFBSAA and FBSPrA, which are polar head isomers, exhibit differences in the number of carbons in the sulfonamido carboxylic acid (Fig. S24).This clearly results in distinct neutral losses-one for acetic acid (CH 2 CO 2 ) and the other for propanoic acid (C 2 H 4 CO 2 ).Additionally, MeFASAA showed no [SO 2 N] -fragment signal, indicating solely the presence of only the [SO 2 F] -fragment signal due to N-methyl substitution.Conversely, FBSPrA exhibited a distinct [SO 2 N] -fragment, signifying the lack of additional N-substitution.The specific fragmentation patterns have been classified in Table 1.This investigation unveiled 25 isomeric PFAS, including isomers with different headgroups and functional tail groups.
These results emphasize the significance of understanding varied reactions and the overall composition of PFAS emissions in semiconductor wastewater, highlighting its complexity and posing challenges for subsequent wastewater treatment.

Conclusions
For the nontarget PFAS approach, mass defect and homologous patterns are potent tools for uncovering PFAS within a series of homologous substances.However, industrial chemicals exhibit diverse and unique functional groups, making homologues less likely to occur in industrial manufacturing.Moreover, the intricate blend of industrial chemicals in wastewater may undergo diverse reactions, such as hydrolysis, oxidation Fig. 4 MS2 spectra of E-FBSE and FBSAcAL hydrate as polar head isomers and deamination.As these reactions unfold in the polar functional section, each transformation leads to distinct mass defect values, rendering the use of homologous patterns for screening.Consequently, our research employs a fragment-based approach to investigate nontarget PFAS, particularly focusing on hydrophilic head groups, overcoming the limitations of conventional homologous series.Utilizing this approach, we successfully identified 83 PFAS in wastewater and effluent samples from semiconductor plants, including 29 newly discovered compounds, which are categorized into three groups (1) The presence of FBSE derivatives, FBSA derivatives, and FBSEE diol derivatives in the authentic standard implies that these derivative series detected in wastewater may stem from chemical formulation byproducts.They constitute less than 0.1-0.5% of the total area compared to main components like FBSE, FBSA, and FBSEE diol.These newly identified byproducts, displaying variations in fluoroalkyl chains such as hydrogen-substituted, unsaturated, and ether structures, were first identified in authentic standards, indicating their origin as byproducts of the chemical manufacturing process.(2) Compounds that undergo intermediate transformation compounds during the oxidation of FBSE, specifically perfluorobutane sulfonamido acetaldehyde and its hydrate, are newly identified.(3) Diverse reaction products generated from the intricate processes of semiconductor manufacturing, which utilize strong acids, bases, and solvents under UV light or heat conditions, include novel PFAS-related reaction compounds generated through hydrolysis, sulfonation and nitrification.We found that FBSA and FBSE reacted with chemicals used in the procedures of semiconductor, and the hydrogen of FBSA on N reacted with sulfonic acid, leading to the detection of several neutral losses during neutral loss screening.This underscores the significance of understanding PFAS-related reactions in semiconductor manufacturing.
In brief, these discoveries underscore the efficacy of the fragment-based approach in identifying unique industrial chemicals and its value in understanding diverse reactions and the actual emission compositions of PFAS in semiconductor wastewater.Moreover, under the Stockholm Convention's regulation for PFAS precursor elimination, complexity arises due to the absence of standards and standardized identification methods for diverse PFAS precursors, which undergo degradation to form terminal PFAS through the amendment of polar functional groups.This poses challenges in recognizing and quantifying the levels of products, complicating market surveillance.Our streamlined fragment-based approach addresses these hurdles and provides a strategic concept for identifying PFAS precursors, particularly those exhibiting versatility in polar functional groups.
displays the MS2 spectrum and proposed structure of H-FBSE.The CL for H-FBSE remains at 3a due to uncertainties in identifying the position of hydrogen.Next, three isomers of E-FBSE were identified and three characteristic peaks from fragments [CF 3 O] − , [C 3 F 7 O] − , and [C 2 F 5 O] − were observed by chromatography, and their structures were proposed (Figs.S8a-S8c) with a CL of 2b.The longer RT (35.1 min) of the E-FBSE isomer indicates increased symmetry in the alkyl structure, consistent with the proposed structure.The presence of H-E-FBSE with the specific fluoroalkyl ether fragment [C 2 F 3 O] − (96.9907, 0.72 ppm) was indicative of the loss of HF from [C 2 F 4 HO] − (116.9971,1.54 ppm).U-FBSE with the fragment [C 4 F 7 ] − (180.9896,0.99 ppm) was detected.We proposed a CL of 3a due to the lack of further information to determine the position of the double bond.U-E-FBSE with the specific fragments [C 3 F 5 ] − (130.9928,1.91 ppm) and [C 4 F 7 O] − (196.9850,3.79 ppm) were detected, which suggested the position of the oxygen atom (Fig.

FBSEE
diol was identified and detected in the form of an acetate adduct [M + CH 3 COO] − , [M-H] − , a formate adduct [M + HCOO] − , and a dimer ion [2 M-H] − , in the order of their observed intensities (Fig.S12).This structure exhibited characteristic fragments, including[SO 2 N(C 2 H 4 O) 2 ] − , [SO 2 N(C 2 H 4 O)CH 2 ] − ,and [SO 2 NC 2 H 4 O] − .Additionally, the high-intensity [SO 3 ] − fragment serves as a confirming marker.FBSEE diol exhibited a weak characteristic fragment of the fluorobutyl fragment [C 4 F 9 ] − , highlighting the feasibility of the screening method based on its functional group, as shown in Figs.S13a and S13b.Hydrogen-substituted FBSEE diol (H-FBSEE diol) was detected with an acetate adduct and the unique fragment [O 2 SN(C 2 H 4 O) 2 ] − (166.0174,-2.16 ppm) for the first time.The CL was set to 3a due to insufficient data to determine the position of hydrogen.

Table 1
, Summary of fragment and neutral loss (NL) screening from the fluorinated tail and polar head, respectively, in the study.Normal font subclasses recognized from both tail and head, bold subclasses only from headgroup, italicized subclass from tail

Table 2
Identified PFAS categorized by polar head class, subclass, compound number theoretical precursor ion, retention time (RT), confidence level (CL), and first report

Table 2
(continued) Compound number is made up of the subclass and serial number.For each chemical structure, please refer to TableS9of SM with further information in TableS9), the prominent signals at 297.9590, representing [C 4 F 9 SO 2 N] -, and 341.9852, representing [C 4 F 9 SO 2 NC 2 H 4 OH] -, suggest that these ions likely retain the original structures of FBSA and FBSE during fragmentation.Hence, the neutral loss between [M-H] -and their associated fragment ([C 4 F 9 SO 2 NH] -for FBSA and [C 4 F 9 SO 2 NC 2 H 4 OH] -for FBSE), would be practical to identify these series of headgroups, despite potential variations at the fluorinated tail end.This approach uniquely emphasizes the neutral loss of segments such as SO 3 , CONH, CH 3 OH, and H 2 O for miscellaneous FASAs and FASEs, underscoring its novelty in identifying reaction products reported for the first time.The initial proposed structure belonging to miscellaneous FASAs and miscellaneous FASEs are provided in Fig.2.In addition, our study leads the way in developing neutral loss screens for CO 2 , HFCO 2 , C n H 2n CO a b The presence of structural isomers varied in the fluorinated chain c The presence of structural isomers varied in the polar functional head d The structures of these compounds have been firstly identified e The precursor ion was an acetate adduct [M + CH 3 COO] - 2 , etc., serving as indicators to distinguish classes such as carboxylic acid, dicarboxylic acid, and sulfonamido carboxylic acid, each characterized by terminal carboxylate structures.A list of the 83 identified PFAS and their formula, retention time (RT), precursor ion, and theoretical m/z are provided in Table [26]ersely, E-FBSA, observed with two isomers in this study, was identified by its distinct ether fragments [CF 3 O] -(84.9907) and [C 2 F 5 O] -(134.9875).Accordingly, their proposed structures were presented in Figs.S4a and S4b, classifying them to CL 2b based on Charbonnet's classification[26].The remaining 6 subclasses (1d-1i) featured a fixed fluoroalkyl chain as perfluorobutane but showed variations in the functional group as -SO 2 N(X)H.The miscellaneous FASAs with the structure C 4 F 9 SO 2 NHX (X: SO 3 H; CONH 2 ; CH 2 CONH 2 ; CH 2 NO 2 ; C 3 H 6 NO 2 ; CH 2 N 2 H, subclass 1d-1i) were identified with the fragment of [SO 2 N] -(77.9644).Figures2 and S5illustrate a neutral loss of C 3 H 5 NO 2 between [C 4 F 9 SO 2 N] -(297.9589)and [M-H] -, representing a novel screening approach for miscellaneous FASAs with various headgroup.Notably, in the presence of substances with N-alkyl substitutions, the dominant fragments [SO 2 ] -(63.9625) and [SO 2 SO 2 NC 2 H 4 O] − (121.9925),[SO 2 H] − (64.9703) and [CH 3 COO] − (59.0141) at RT 35.8 min (Fig.

Table 3
List of impurities in authentic standards of FBSE, FBSA, and FBSEE diol a Compound number is made up of the subclass and serial number.For each chemical structure, please refer to TableS9of SM b The relative peak area (%) = Peak area of compound / Sum of peak areas of all compounds