Ascertaining Hydrogen-Abstraction Reaction Efficiencies of Halogenated Organic Compounds in Electron Impact Ionization Processes by Gas Chromatography-High-Resolution Mass Spectrometry.

H-Abstraction reactions occurring during electron impact ionization processes in electron ionization mass spectrometry (EI-MS) are a long-standing and crucial topic in MS research. Yet, some critical relevant mechanisms are controversial and ambiguous, and information about the EI-induced H-abstraction reactions of halogenated organic compounds (HOCs) is completely in the dark. This study provides a systematic investigation of H-abstraction reactions of HOCs taking place in the EI source using 13C6-hexachlorobenzene (13C6-HCB) and 13C6-hexabromobenzene (13C6-HBB) as exemplary compounds by gas chromatography (GC)-high-resolution mass spectrometry (GC-HRMS). The H-abstraction efficiencies were evaluated with the MS signal intensity ratios of ions with H-abstraction relative to the corresponding original ions (without H-abstraction). Ion source temperatures, EI energies, and numbers of heavy isotope atoms (37Cl or 81Br) of isotopologues were investigated in terms of their effects on the H-abstraction efficiencies. The H-abstraction efficiencies of individual isotopologues generally decreased from the first to the last isotopologues of respective ions, and those of individual ions were different from each other, with the highest values of 0.017 and 0.444 for 13C6-HCB and 13C6-HBB, respectively. The overall H-abstraction efficiencies involving all measured ions of 13C6-HCB and 13C6-HBB were 0.004 and 0.128, respectively. With increasing ion source temperatures, the H-abstraction efficiencies first increased to a summit and then began to linearly decrease. EI energies and emission currents could impact the H-abstraction efficiencies but showed no certain tendency. The H-abstraction reactions were inferred to belong to ion-molecule reactions, and the siloxanes bleeding from the GC column might be a hydrogen source. Some strategies were proposed for eliminating or alleviating the interference triggered by the H-abstraction reactions in EI-MS in identification of halogenated organic pollutants (HOPs). Our findings provide a better understanding of the EI-induced H-abstraction reactions of HOCs and may benefit the identification of HOPs in environmental analysis, especially for novel HOPs.

Introduction

n class="Chemical">Hydrogen abstraction (H-abstraction) reactions (or intermolecular pan> class="Chemical">hydrogen transfer reactions), which may have important significance for mass spectrometry (MS) analysis, can commonly take place in various MS ionization sources, such as electron ionization (EI),[1,2] chemical ionization (CI),[3] electrospray ionization (ESI),[4−8] atmospheric pressure chemical ionization (APCI),[9,10] matrix-assisted laser desorption ionization,[11] atmospheric pressure photoionization (APPI),[12−14] and low-pressure photoionization (LPPI) sources.[15−18] The mechanisms of H-abstraction reactions on CI, ESI, APCI, and photoionization (PI) sources have been well studied.[3,5,6,9,13] However, the real mechanisms of H-abstraction reactions in EI-MS remain far from being completely explicit although the H-abstraction phenomenon has been observed and investigated for a very long time (2–5 decades).[1,2,19−24]

It was reported that the generation of molecular ions via H-abstraction in the EI source, such as [M + H]+, [M + 2H]+, and [M + 3H]+, might be related to the structures of analytes and the ion source temperatures.[23,24] Some studies concluded that the transferred hydrogen atoms in H-abstraction reactions in the EI source were derived from H2O in the ion source and/or samples.[1,2,21,25,26] However, a study revealed that the hydrogen atoms did not stem from residual water in the ion source, nor residual H2.[23] The author inferred that the hydrogen source was related to thermal initiation and EI ionization.[23] Most of the previous studies indicated that the H-abstraction reactions in the EI source belonged to ion–molecule reactions,[1,2,19,22−27] which can be expressed as

This reaction equation is similar to that of some H-abstraction reactions occurring in the PI source.[9,28−31] The initial ionization process of PI is[9,29,30]and that of EI is

It can be seen from eqs and 3 that the initial ionization processes of PI and EI are to some extent similar. As a result, the revealed mechanisms behind the H-abstraction reactions occurring in the PI source may provide illuminating insights into the H-abstraction fundamentals in the EI source. In the PI source, the n class="Chemical">hydrogen source for H-abstraction reactions is protic solvents used for injection, such as pan> class="Chemical">toluene,[29] water,[32] methanol,[32] ethanol and hexane,[28] whereas aprotic solvents such as tetrachloromethane and aprotic gases including N2, O2, and CO2 could compete with analytes to abstract hydrogen atoms, thus impairing the MS signal intensities of [M + H]+.[28] In addition to solvents, the temperature, ionizer voltage, and reaction length of the PI source also play critical roles in the H-abstraction reactions,[28,29,33,34] showing positive correlations with the H-abstraction efficiencies. Furthermore, the proton affinities of analytes and the reaction enthalpy changes (ΔH) are also important factors affecting the H-abstraction reactions in the PI source, with a positive correlation with the H-abstraction efficiencies.[28,34]

To date, reported studies concerning H-abstraction reactions in the EI source have mainly focused on n class="Chemical">quinones.[20,21] However, pan> class="Chemical">halogenated organic compounds (HOCs), e.g., organochlorines and organobromines, may also abstract hydrogen atoms in the EI source, in view of their lone electron pairs of halogen atoms (e.g., Cl and Br). HOCs are a large category of chemicals produced from both natural and anthropogenic activities[35−38] and have enormous influences on human beings and the environment, in either positive or negative ways.[39−43] Many HOCs are persistent, bioaccumulative, and toxic pollutants, such as polychlorinated dioxins/furans,[44−46] dichlorodiphenyltrichloroethanes (DDTs),[47−49] polychlorinated biphenyls,[45,50] polybrominated diphenyl ethers,[48,49] polybrominated biphenyls,[51] chlorinated paraffins,[52] perfluorinated compounds,[53] and etc., and have been posing serious pollution problems to the environment. In addition to known halogenated organic pollutants (HOPs), numerous previously unknown, novel, and emerging HOPs have been screened and identified in various matrices.[54−58] Most HOPs can be identified by gas chromatography coupled with EI-MS (GC-EI-MS), which is the mainstream approach for analysis of HOPs.[59] Recently, we successfully developed a quasi-targeted analysis method for identification of novel HOPs in environmental matrices using GC-EI-HRMS.[60] Additionally, some non-hyphenated MS techniques such as EI-FTICR-HRMS have been applied to the identification of HOPs.[61] Nevertheless, anticipatable H-abstraction reactions in EI-MS may introduce plausible interference peaks and could therefore compromise the identification of HOPs, especially non-targeted or quasi-targeted analysis of unknown and novel pollutants. Accordingly, the H-abstraction reactions of HOCs in EI-MS require systemic investigation to facilitate the screening and identification of these compounds.

Here, we conducted an in-depth study concerning the H-abstraction reactions of n class="Chemical">HOCs occurring during the electron impact ionization process in EI-MS using GC double-focusing magnetic-sector HRMS (GC-DFS-HRMS), with two stable isotope-labeled stanpan>dards, i.e., pan> class="Chemical">13C6-hexachlorobenzene (13C6-HCB) and 13C6-hexabromobenzene (13C6-HBB), used as model compounds. The H-abstraction efficiencies were calculated as the ratios of MS signal intensities between ions with H-abstraction and the corresponding original ions (without H-abstraction), and the effects of ion source temperatures, EI energies ,and numbers of heavy isotope atoms (37Cl or 81Br) of isotopologues on the H-abstraction efficiencies were investigated. This study reveals some details of the H-abstraction reactions of HOCs in EI-MS, provides new insights into the possible mechanisms of the EI-induced H-abstraction reactions, and may be conducive to non-targeted and quasi-targeted qualitative analyses of HOPs, particularly novel HOPs.

Experimental Section

Chemicals and Materials

Stock solutions of stable isotope-labeled standards, n class="Chemical">13C6-HCB (100 μg/mL in pan> class="Chemical">nonane; isotopic purity, ≥99%) and 13C6-HBB (100 μg/mL in toluene; isotopic purity, ≥99%), were purchased from Cambridge Isotope Laboratories Inc. (Andover, MA, USA). Structures and the CAS numbers of the standards are documented in Table S1. Chromatography-grade isooctane was purchased from CNW Technologies GmbH (Düsseldorf, Germany). Acetone and n-hexane used for washing the autosampler injection syringe were bought from Merck Corp. (Darmstadt, Germany). Perfluorotributylamine used as a reference standard for HRMS calibration was purchased from Sigma-Aldrich LLC. (St. Louis, MO, USA). The purchased stock standards of 13C6-HCB and 13C6-HBB were further diluted with isooctane to prepare working solutions at concentrations of 5 and 1 μg/mL, respectively. All the standard solutions were kept in a fridge at −20 °C prior to use.

GC–HRMS Measurements

The GC–HRMS system consisted of dual gas chromatographs (Trace-GC-Ultra) coupled to a double-focusing magnetic-sector HRMS (BE geometry) and an extra TriPlus autosampler (GC-DFS-HRMS, Thermo-Fisher Scientific, n class="Chemical">Bremen, Germany). The prepared working solutions were directly analyzed by the GC–HRMS. The chromatographic separation was conducted with a DB-5MS capillary column (60 m × 0.25 mm, 0.25 μm film thickness, J&W Scientific, Folsom, USA).

The GC temperature-ramping program is as follows: The temperature was held at 120 °C for 2 min, ramped to 220 °C at 20 °C/min, held for 16 min; then ramped to 235 °C at 5 °C/min, held for 7 min; then ramped to 260 °C at 5 °C/min; and finally ramped to 310 °C at 40 °C/min and held for 0.75 min. The total time of a GC run was 40 min. The solvent delay time was 6.5 min. Ultrahigh-purity n class="Chemical">helium was used as the carrier gas, and a constant flow rate of 1.0 mL/min was maintained. The GC injection port and transfer line were kept at 260 and 280 °C, respectively. Splitless injection mode was applied, and the injection volume was 1 μL.

The working settings and parameters of the DFS-HRMS are provided as follows: The EI source operating in positive mode was applied; EI energies were set at 35, 45, 55, and 65 eV; ion source temperatures were set at 160, 190, 220, 250, 280, and 300 °C; filament emission currents ranged from 0.463 to 0.989 mA with the corresponding EI energies from 35 to 65 eV; data acquisition was performed in multiple ion detection (MID) mode; the dwell time was 20 ms for each isotopologue; scan time segments were set at 8–20 min (for n class="Chemical">13C6-HCB) anpan>d 20–36 min (for pan> class="Chemical">13C6-HBB); mass resolution (5% peak definition) was ≥10,000 and the MS detection accuracy was set at ±1 mu. If not specified, all measurements were performed with an EI energy of 45 eV and ion source temperature of 250 °C. The HRMS real-time calibration was carried out with perfluorotributylamine during MID operation.

Chemical structures of the investigated compounds along with their dechlorination/den class="Chemical">bromination radicals were depicted with ChemDraw (Ultra 7.0, Campan> class="Chemical">bridgesoft), and the exact masses of the chlorine/bromine isotopologues were calculated with a mass accuracy of 0.00001 u. Since the carbon of 13C6-HCB and 13C6-HBB comprises only the 13C isotope, only the chlorine/bromine isotopologues were studied. Hydrogen (1H) was the only involved hydrogen isotope since the others (deuterium and tritium) have extremely lower abundances in nature relative to hydrogen. For a molecule or radical containing n Cl/Br atoms, all the isotopologues (n + 1) were chosen. The exact mass-to-charge ratios (m/z) of ions were calculated by subtracting the exact mass of one electron from that of individual isotopologues. The m/z values of ions with H-abstraction were obtained by adding the exact mass of a hydrogen atom to that of each original ion, and only the H-abstraction reactions transferring merely one hydrogen atom were taken into consideration. These m/z values were imported into the MID module for detection. The detailed information including chemical formulas, exact masses, and exact m/z values of the isotopologues of the investigated compounds is listed in Table S2, and the representative chromatograms along with high-resolution mass spectra of the compounds are shown in Figure .

Figure 1

Representative chromatograms and high-resolution mass spectra of 13C6-hexachlorobenzene (13C6-HCB) and 13C6-hexabromobenzene (13C6-HBB). Representative chromatograms of (a) 13C6-HCB and (b) 13C6-HBB; representative mass spectra of (c) 13C6-HCB and (d) 13C6-HBB. P-Cl: dechlorination product ion possessing n Cl atom(s); P-Br: debromination product ion possessing n Br atom(s); NL: nominal level; m/z: mass to charge ratio.

Data Processing

The H-abstraction efficiency of an isotopologue is expressed as the MS signal intensity ratio of the isotopologue with H-abstraction relative to the original isotopologue (R) and was calculated usingwhere i is the number of n class="Chemical">37Cl or pan> class="Chemical">81Br atom(s) in the isotopologue, I′ is the MS signal intensity of isotopologue i with H-abstraction, I is the MS signal intensity of the original isotopologue i. The whole H-abstraction efficiency of an ion (R) was calculated withwhere n is the number of Cl/Br atoms of the ion. In addition, the overall H-abstraction efficiency involving all the detected ions of a compound (Rall) was calculated with

Prior to exporting MS signal intensities, background subtraction was carried out by deducting the baseline signal intensities close to both ends of individual chromatographic peaks. If n class="Species">not specified, the number of injection replicates was six, anpan>d the experimental data derived from these replicated measurements were applied to calculation of average H-abstraction efficiencies anpan>d stanpan>dard deviations (1σ). Inpan> order to describe this work more concisely anpan>d clearly, we use IST-n to refer to anpan> isotopologue with n – 1 pan> class="Chemical">37Cl or 81Br atom(s), apply P-Cl to denote a dechlorination product ion having n Cl atom(s), and utilize P-Br to represent a debromination product ion containing n Br atom(s). Because IST-1 of P-Cl1 with H-abstraction of 13C6-HCB ([13C6H35Cl]+, m/z: 113.99626) was seriously interfered by the [C2NF4]+ ion (m/z: 113.99614) of perfluorotributylamine, the H-abstraction efficiency of IST-1 was not calculated.

Method Performances

As documented in Tables S3– S5, the precisions (standard deviations) of the H-abstraction efficiencies of individual isotopologues, the whole H-abstraction efficiencies of individual ions, and the overall H-abstraction efficiencies of individual compounds are within the ranges of 0.00002–0.01608, 0.00011–0.00871, and 0.00015–0.00511, respectively. These results demonstrate that the analysis method could meet the requirements for investigating the H-abstraction efficiencies of 13C6-HCB and 13C6-HBB in EI-MS.

Statistical Analysis

Statistical analysis was conducted with SPSS Statistics 19.0 (IBM Inc., Armonk, USA) and Origin 9 (OriginLab Corp., Northampton, USA). The paired-sample T test and independent-sample T test (performed with SPSS) were applied to determine the p values (two-tailed) with an α of 0.01 as the threshold value for significance. The differences of the H-abstraction efficiencies and MS signal intensities among n class="Chemical">isotopologues were examined with the paired-sample T test, anpan>d those of the whole H-abstraction efficiencies among ions anpan>d the overall H-abstraction efficiencies between the two compounpan>ds were evaluated with the independent-sample T test. If a p value is less thanpan> 0.01, the null hypothesis (e.g., no difference between two groups of H-abstraction efficiencies) is rejected, indicating that there is indeed a significanpan>t difference. Linear anpan>d non-linear regressions (conducted with Origin) were employed to reveal the relationships between experimental data anpan>d parameters/factors of interest.

Results and Discussion

H-Abstraction Efficiency Measurement

Isotope-labeled n class="Chemical">13C6-HCB anpan>d pan> class="Chemical">13C6-HBB instead of native HCB and HBB were chosen to perform this study owing to the fact that 13C6-HCB and 13C6-HBB contain only 13C and Cl/Br atoms and therefore the interference caused by carbon isotopes is nonexistent. In addition, as the isotopic purities of the two reference standards were higher than 99%, potential impurities were thus deduced to contribute negligible interferences to measured data. Based on the detected MS signal intensities of the original ions and those with H-abstraction, we calculated three hierarchical H-abstraction efficiencies, i.e., H-abstraction efficiencies of individual isotopologues, whole H-abstraction efficiencies of individual ions, and overall H-abstraction efficiencies of individual compounds. Thus, H-abstraction efficiencies of the investigated compounds were evaluated at three levels, namely, isotopologues, ions, and compounds.

H-Abstraction Efficiencies of Individual Isotopologues

As shown in Figure and Table S3, the first six n class="Chemical">chlorine pan> class="Chemical">isotopologues of the molecular ion of 13C6-HCB, i.e., IST-1 to IST-6, have H-abstraction efficiencies within the range from 0.00010 ± 0.00009 to 0.0026 ± 0.0003, exhibiting a gradual decline from the first to the last isotopologues (Figure a). For the first five chlorine isotopologues of P-Cl5 (IST-1 to IST-5), the H-abstraction efficiencies were from 0.016 ± 0.001 to 0.0184 ± 0.0003, showing a generally decreasing tendency from IST-1 to IST-5 (Figure b). The H-abstraction efficiencies of individual chlorine isotopologues of the other four dechlorination product ions (P-Cl1 to P-Cl4) ranged from 0.0020 ± 0.0001 to 0.010 ± 0.001 (Figure c–f, respectively), presenting no explicit variation trend along the isotopologue sequences with statistical significance (Figure c–e). The H-abstraction efficiencies of IST-1 of the molecular ion and P-Cl5 were significantly higher than those of other isotopologues of the respective ions (p ≤ 0.001). The H-abstraction efficiency difference between any two neighboring chlorine isotopologues of the molecular ion was statistically significant (p ≤ 0.008), and that between any two neighboring isotopologues of the first three isotopologues of P-Cl5 was also statistically significant (p ≤ 0.001).

Figure 2

H-abstraction efficiencies ([M + H]+/[M]·+ or [P + H]+/[P]+ MS signal intensity ratios) of chlorine/bromine isotopologues of molecular and dechlorination/debromination product ions of 13C6-HCB and 13C6-HBB during the electron impact ionization process in electron ionization high-resolution mass spectrometry (EI-HRMS). H-Abstraction efficiencies of chlorine isotopologues of the (a) molecular ion, (b) P-Cl5, (c) P-Cl4, (d) P-Cl3, (e) P-Cl2, and (f) P-Cl1 of 13C6-HCB; H-abstraction efficiencies of bromine isotopologues of the (g) molecular ion, (h) P-Br5, (i) P-Br4, (j) P-Br3, (k) P-Br2, and (l) P-Br1 of 13C6-HBB. [M]·+: molecular ion; [M + H]+: the corresponding molecular ion with hydrogen abstraction; [P]+: dehalogenation product ion; [P + H]+: the corresponding product ion with H-abstraction; IST-n: isotopologue of an ion possessing n – 1 37Cl or 81Br atom(s). Error bars denote the standard deviations (1σ, n = 6).

H-abstraction efficiencies ([M + H]+/[M]·+ or [P + H]+/[P]+ MS signal intensity ratios) of n class="Chemical">chlorine/pan> class="Chemical">bromine isotopologues of molecular and dechlorination/debromination product ions of 13C6-HCB and 13C6-HBB during the electron impact ionization process in electron ionization high-resolution mass spectrometry (EI-HRMS). H-Abstraction efficiencies of chlorine isotopologues of the (a) molecular ion, (b) P-Cl5, (c) P-Cl4, (d) P-Cl3, (e) P-Cl2, and (f) P-Cl1 of 13C6-HCB; H-abstraction efficiencies of bromine isotopologues of the (g) molecular ion, (h) P-Br5, (i) P-Br4, (j) P-Br3, (k) P-Br2, and (l) P-Br1 of 13C6-HBB. [M]·+: molecular ion; [M + H]+: the corresponding molecular ion with hydrogen abstraction; [P]+: dehalogenation product ion; [P + H]+: the corresponding product ion with H-abstraction; IST-n: isotopologue of an ion possessing n – 1 37Cl or 81Br atom(s). Error bars denote the standard deviations (1σ, n = 6).

The H-abstraction efficiencies of the n class="Chemical">bromine pan> class="Chemical">isotopologues of the molecular ion of 13C6-HBB decreased continuously from IST-1 (0.222 ± 0.009) to IST-7 (0.0005 ± 0.0005) (Figure g), and those of any neighboring isotopologues showed statistically significant differences (p ≤ 0.00001). Likewise, the product ions P-Br5, P-Br4, P-Br3, and P-Br2 of 13C6-HBB presented gradually descending H-abstraction efficiencies from the first isotopologues to the last ones, with H-abstraction efficiency ranges of 0.424 ± 0.016 to 0.569 ± 0.006, 0.111 ± 0.007 to 0.23 ± 0.01, 0.173 ± 0.003 to 0.195 ± 0.003, and 0.041 ± 0.001 to 0.059 ± 0.005, respectively (Figure h–k). For these product ions, the H-abstraction efficiencies of the bromine isotopologues of each ion (except IST-4, IST-5, and IST-6 of P-Br5) can be confidently differentiated from each other (p ≤ 0.002). In addition, there was a slight decline for the H-abstraction efficiencies from IST-1 to IST-2 of P-Br1 (p ≤ 0.004) (Figure l).

Whole H-Abstraction Efficiencies of Individual Ions

As shown in Figure a, the magnitudes of measured whole H-abstraction efficiencies of individual ions of n class="Chemical">13C6-HCB follow the order pan> class="Chemical">P-Cl5 > P-Cl1 > P-Cl4 > P-Cl3 > P-Cl2 > molecular ion, with the range from 0.0011 ± 0.0001 to 0.0172 ± 0.0002 (Table S4). These H-abstraction efficiencies are statistically distinguishable (p ≤ 0.00004). The scale of the whole H-abstraction efficiencies of individual ions of 13C6-HBB showed the following order: P-Br5 > P-Br3 > P-Br1 > P-Br4 > P-Br2 > molecular ion (Figure b), with the range from 0.037 ± 0.001 to 0.444 ± 0.009 (Table S4). These H-abstraction efficiencies can also be confidently differentiated (p ≤ 0.0003). The ion P-Br5 presented an extremely high H-abstraction efficiency (0.444 ± 0.009), suggesting that around 30.7% of the amount of the initial ion underwent the H-abstraction reaction. It is noteworthy that the patterns of the whole H-abstraction efficiencies of ions of the two compounds are somewhat similar, probably implying similar mechanisms of H-abstraction reactions for the detected ions between the two compounds in EI-MS.

Figure 3

Whole H-abstraction efficiencies of individual ions of (a) 13C6-HCB and (b) 13C6-HBB, and overall H-abstraction efficiencies of (c) 13C6-HCB and 13C6-HBB in EI-HRMS.

Overall H-Abstraction Efficiencies of Individual Compounds

As shown in Figure c and Table S5, the overall H-abstraction efficiency of 13C6-HBB (0.128 ± 0.005) was more than one order of magnitude higher than that of 13C6-HCB (0.0044 ± 0.0001), with the p value <0.00001. This result indicates that 13C6-HBB has a significantly higher H-abstraction capability than 13C6-HCB in EI-MS.

Distribution Patterns of Measured MS Signal Intensities of Molecular Isotopologues

As shown in in Figure and Table S6, it is very surprising that the distribution patterns of the measured MS signal intensities of n class="Chemical">isotopologues between the molecular ion ([M]·+) anpan>d the molecular ion with H-abstraction ([M + H]+) of each compounpan>d are apparently different. For the [M]·+ of pan> class="Chemical">13C6-HCB, the measured MS signal intensities were in the following order: IST-2 > IST-3 > IST-1 > IST-4 > IST-5 > IST-6 > IST-7 (Figure a), which is identical to the theoretical relative abundance order. Meanwhile, the detected isotopologue abundances of the [M + H]+ of 13C6-HCB presented the order IST-1 > IST-2 > IST-3 > IST-4 > IST-5 > IST-6 > IST-7 (Figure b). The detected abundance of IST-1 of [M + H]+ was significantly higher than those of IST-2 and IST-3 (p ≤ 0.00005), which contradicts the scenario of [M]·+ and therefore the order of theoretical relative abundances of isotopologues.

Figure 4

Distribution patterns of MS signal intensities of chlorine/bromine isotopologues of the molecular ions ([M]·+) and the molecular ions with hydrogen abstraction ([M + H]+) of 13C6-HCB and 13C6-HBB on GC-EI-HRMS. Intensities of chlorine isotopologues of (a) [M]·+ and (b) [M + H]+ of 13C6-HCB; intensities of bromine isotopologues of (c) [M]·+ and (d) [M + H]+ of 13C6-HBB.

In addition to 13C6-HCB, 13C6-HBB also showed different measured isotopologue abundance orders between [M]•+ and [M + H]+. For [M]•+, the order of the measured isotopologue abundances was IST-4 > IST-3 > IST-5 > IST-2 > IST-6 > IST-1 > IST-7 (Figure c), which is well consistent with the normal distribution and the theoretical relative abundance order. However, the measured isotopologue abundances of [M + H]+ exhibited the following order: IST-3 > IST-2 > IST-4 > IST-1 > IST-5 > IST-6 > IST-7 (Figure d), which is completely different from that of [M]•+ and accordingly the theoretical relative abundance order. These findings also point to the significant differences of H-abstraction efficiencies among different chlorine/bromine isotopologues in EI-MS.

Parameters and Factors Affecting H-Abstraction Efficiency Variation

To investigate whether some parameters can influence the H-abstraction efficiencies in EI-MS, we measured the H-abstraction efficiencies of the molecular ion of n class="Chemical">13C6-HCB with varied ion source temperatures anpan>d EI energies. Some previous studies have proved that the ion source temperature could dramatically impact the extents of H-abstraction reactions in EI-MS, positively correlating with the reaction extents.[23,24] This observation could be attributable to the fact that the average tranpan>slational energies of the pan> class="Chemical">analyte ions and hydrogen donor molecules increase with rising temperature and therefore enhance the collision rate, accordingly raising the H-abstraction reaction extents. In the present study, however, we observed a different variation mode of H-abstraction efficiencies with ion source temperatures than those found in previous studies. As shown in Figure a and Table S7, the measured H-abstraction efficiencies first increased from 0.0071 ± 0.0008 to 0.0088 ± 0.0003 with the ion source temperatures changing from 160 to 190 °C and then almost linearly decreased to 0.0008 ± 0.0001 with the temperatures varying from 190 to 300 °C. As Ahmed et al. reported that the relationship between the H-abstraction reaction extents and the APPI source temperatures follows an Arrhenius-type curve,[29] we accordingly plotted the measured H-abstraction efficiencies versus the reciprocal Kelvin temperatures. However, we did not get the Arrhenius-type temperature-dependent curve describing the relationship between the measured H-abstraction efficiencies and the ion source temperatures. Instead, the H-abstraction efficiencies first linearly increased from the lowest value to a summit with the 1/K values varying from 0.00174 to 0.00216 K–1 and then began to decrease as the 1/K increased to 0.00231 K–1 (Figure b). This observation also demonstrates that potential isotopic impurities (such as HCB) in the reference standard of 13C6-HCB could not interfere with the measured H-abstraction efficiencies.

Figure 5

Plots of the whole H-abstraction efficiencies of the molecular ion ([M]·+) of 13C6-HCB versus (a) the ion source temperatures, (b) the reciprocal Kelvin temperatures of ion source, (c) the EI energies, and (d) the emission currents of EI filament, as well as the correlations of (e) the H-abstraction efficiencies versus the varied MS signal intensities caused by different ion source temperatures and (f) the H-abstraction efficiencies versus the varied MS signal intensities trigged by different EI energies. The number of injection replicates was three.

Plots of the whole H-abstraction efficiencies of the molecular ion ([M]·+) of n class="Chemical">13C6-HCB versus (a) the ion source temperatures, (b) the reciprocal Kelvin temperatures of ion source, (c) the EI energies, and (d) the emission currents of EI filament, as well as the correlations of (e) the H-abstraction efficiencies versus the varied MS signal intensities caused by different ion source temperatures and (f) the H-abstraction efficiencies versus the varied MS signal intensities trigged by different EI energies. The number of injection replicates was three.

As indicated in Figure ,d, the H-abstraction efficiencies can be significantly influenced by EI energies and emission currents of the ion source filament. At an EI energy of 45 eV, the H-abstraction efficiency reached the highest (0.0044 ± 0.0007), which was significantly higher than the rest (p ≤ 0.003). Meanwhile, the H-abstraction efficiency at an EI energy of 35 eV was the lowest (0.0007 ± 0.0002), which was significantly lower than those at other EI energies (p ≤ 0.01). The H-abstraction efficiencies at EI energies of 55 and 65 eV were relatively comparable (0.0014 ± 0.0002 and 0.00168 ± 0.00001) and cannot be differentiated with statistical significance (p = 0.2). We deduce that internal energies of n class="Chemical">analyte ions increase as EI energies increase anpan>d thus the rate of collision between pan> class="Chemical">analyte ions and hydrogen donor molecules is enhanced, thus improving H-abstraction efficiencies. Meanwhile, the increasing internal energies of [M]·+ ions further enhance internal energies of [M + H]+ ions, accordingly impairing the stability of the [M + H]+ ions. As a result, the H-abstraction efficiencies are inferred to increase with EI energies at a lower energy region and decrease in a higher-energy region, reaching a summit at medium EI energy, which is consistent with the observation in this study (Figure c). Since the emission currents generally increased with the increase of EI energies, the order of H-abstraction efficiencies observed at different emission currents were the same as that of the H-abstraction efficiencies at the corresponding EI energies. However, the pattern of H-abstraction efficiencies plotted versus EI energies is apparently different from that of H-abstraction efficiencies versus emission currents (Figure c,d). This observation might be ascribed to the fact that the emission currents varied with the variation of EI energies more drastically in a lower-energy region than in a higher-energy region.

As shown in Figure e, when the MS signal intensity variation was caused by the variation of ion source temperatures, a strong positive linear correlation between H-abstraction efficiencies and MS signal intensities could be observed (R2 = 0.969). Nevertheless, when the MS signal intensity variation was triggered by the variation of EI energies, only a weak positive linear correlation between H-abstraction efficiencies and MS signal intensities was found (R2 = 0.274, Figure f). It is noteworthy that the highest MS signal intensity and the highest H-abstraction efficiency were all obtained at an EI energy of 45 eV, whereas at other EI energies (35, 55, and 65 eV), the MS signal intensities and the H-abstraction efficiencies were relatively close (Table S7). These results demonstrate that the MS signal intensity may be a parameter affecting the H-abstraction efficiencies of n class="Chemical">organochlorines on EI-MS.

As illustrated in Figure , the H-abstraction efficiencies of n class="Chemical">isotopologues generally decreased with the increase of heavy pan> class="Disease">isotope atoms (37Cl or 81Br) of the isotopologues. The relationships between the H-abstraction efficiencies and the numbers of 37Cl or 81Br atom(s) were well fitted with exponential curves for the molecular ions and the P-Cl5/P-Br5 of 13C6-HCB and 13C6-HBB, with the R2 values ranging from 0.975 to 0.998. In addition to the ions involved in Figure , the ions P-Br4, P-Br3 and P-Br2 also exhibited exponentially descending H-abstraction efficiencies of bromine isotopologues with the increase of 81Br atoms of the isotopologues, with R2 ≥ 0.982 (Figure S1). These findings suggest that chlorine/bromine isotopic effects may play a crucial role in the variation of H-abstraction efficiencies among chlorine/bromine isotopologues of HOCs on EI-MS.

Figure 6

Correlations between H-abstraction efficiencies and the 37Cl or 81Br atom numbers of chlorine/bromine isotopologues. (a) [M + H]+/[M]·+ intensity ratios of the chlorine isotopologues of the molecular ion of 13C6-HCB; (b) [P + H]+/[P]+ intensity ratios of the chlorine isotopologues of the P-Cl5 of 13C6-HCB; (c) [M + H]+/[M]·+ intensity ratios of the bromine isotopologues of the molecular ion of 13C6-HBB; (d) [P + H]+/[P]+ intensity ratios of the bromine isotopologues of the P-Br5 of 13C6-HBB. Solid curves refer to exponential regressions and shaded areas represent the corresponding 95% confidence intervals.

Correlations between H-abstraction efficiencies and the n class="Chemical">37Cl or pan> class="Chemical">81Br atom numbers of chlorine/bromine isotopologues. (a) [M + H]+/[M]·+ intensity ratios of the chlorine isotopologues of the molecular ion of 13C6-HCB; (b) [P + H]+/[P]+ intensity ratios of the chlorine isotopologues of the P-Cl5 of 13C6-HCB; (c) [M + H]+/[M]·+ intensity ratios of the bromine isotopologues of the molecular ion of 13C6-HBB; (d) [P + H]+/[P]+ intensity ratios of the bromine isotopologues of the P-Br5 of 13C6-HBB. Solid curves refer to exponential regressions and shaded areas represent the corresponding 95% confidence intervals.

Tentative Mechanistic Interpretation

Ion–Molecule Reactions and Hydrogen Source

Previous studies have concluded that the H-abstraction reactions in the EI source are ion–molecule reactions,[1,2,19,22−27] which can be expressed as eq . Prior to the H-abstraction reactions, the molecules of n class="Chemical">analytes are ionized as shown in eq , giving rise to molecular and fragmental ions, which are electron-deficient radical ions. In light of the similar ionization processes of PI and EI as well as the reported outcomes,[28,29] we infer the most predominant ion–molecule reactions are that as shown by eq , instead of others such as

On the other hand, eq is energetically unfavorable compared to eq , while eqs and 9 are low-probability events relative to eq due to the production of H· and H+ is not energetically favorable. In addition, we found that not only the molecular ions but also the den class="Chemical">halogenation fragmental ions participated in the H-abstraction reactions in the EI source, anpan>d the fragmental ions exhibited evidently higher H-abstraction efficiencies thanpan> the molecular ions (Figure ). This result manpan>ifests that the “molecules” in the ion–molecule reactions were not the molecules of pan> class="Chemical">analytes, and the “ions” in the reactions were the molecular and fragmental ions of the analytes, which is consistent with eq . Then, the subsequent questions are what are the real “molecules” in the ion–molecule reactions and where do the hydrogen atoms stem from.

We inferred that the reaction molecules might be n class="Chemical">siloxanes that ranpan> off from the packing materials (95% pan> class="Chemical">polydimethylsiloxane and 5% phenyl group) of the GC column and/or emigrated from the silicone inlet septum of the GC.[62] Therefore, the H-abstraction reactions can be described aswhere R represents a replaceable siloxane group. It has been reported that the methyl of toluene prefers to contribute hydrogen atoms for H-abstraction reactions in a PI source in contrast with the phenyl group.[13] Accordingly, it can be anticipated that the methyl of siloxanes is an excellent hydrogen donor for the H-abstraction reactions in the EI source. In conclusion, the siloxanes bleeding from the GC column and/or injection port pad might participate in the H-abstraction reactions in the EI source and could be a potential hydrogen source.

Another evidence for the above conclusion is the difference of overall H-abstraction efficiencies between n class="Chemical">13C6-HCB anpan>d pan> class="Chemical">13C6-HBB; i.e., 13C6-HCB presented significantly lower H-abstraction efficiency compared to 13C6-HBB (Figure c). The electronegativity of chlorine is higher than that of bromine; thus, the radical ions of 13C6-HCB are more competent to abstract hydrogen atoms from hydrogen donors. However, in the present study, we found that the overall H-abstraction efficiency of 13C6-HCB is significantly lower than that of 13C6-HBB. This implies that the densities of gaseous molecules of the hydrogen donors during the H-abstraction reactions were different for 13C6-HCB and 13C6-HBB. If not in this case, the H-abstraction efficiency of 13C6-HCB ought to be higher than that of 13C6-HBB. It can be inferred that the densities of gaseous molecules of the hydrogen donors for H-abstraction reactions of 13C6-HCB were lower than those for H-abstraction reactions of 13C6-HBB. This inference is in line with the reality that 13C6-HCB eluted from the GC column at a relatively lower temperature than 13C6-HBB and the higher oven temperature could cause more siloxanes to bleed from the column.

Temperature-Dependent Reactions

In previous studies, the H-abstraction efficiencies in the EI source were found to exponentially correlate with the ion source temperatures,[23,24] which was also observed in another study concerning the H-abstraction reactions in the PI source.[29] Nevertheless, in our study, the H-abstraction efficiencies rose first and then fell with the increasing temperatures of the ion source (Figure a). From 160 to 190 °C, the H-abstraction efficiencies rose to a summit, which is consistent with the findings of the previous reports.[23,24,29] Meanwhile, the H-abstraction efficiencies gradually decreased to the lowest value from 190 to 300 °C, which contradicts the outcomes of previous studies.[23,24,29] We deduce that the decline of H-abstraction efficiencies with the increase of ion source temperature might be attributable to the thermostability of [M + H]+. The pyrolysis rate of [M + H]+ rises as the temperature increases. On the other hand, the rate of H-abstraction reactions generating [M + H]+ may also increase along with the increase of ion source temperature. When the pyrolysis rate and generation rate of [M + H]+ are equivalent, the H-abstraction efficiency reaches the highest.

Chlorine and Bromine Isotopic Effects in H-Abstraction Reactions

The ion–molecule H-abstraction reactions of 13C6-HCB and 13C6-HBB in the EI source may be analogous to the H-abstraction reactions of chlorine and bromine radicals. We hypothesize a diatomic hydride HA (A is an imaginary element), and the H-abstraction reaction between HA and Cl· in the gas phase is

The transition-state complex is A--H--Cl. Meanwhile, taking the chlorine isotopes into account, the tranpan>sition-state complexes become pan> class="Chemical">A--H--35Cl and A--H--37Cl. Due to zero-point energies (ZPEs), the bond H–35Cl is slightly weaker than the bond H–37Cl. As a consequence, the energy barrier of A--H--35Cl is slightly lower than that of A--H--37Cl. Then, the formation of A--H--35Cl is a little faster than that of A--H--37Cl in the reaction as indicated in eq , thereby generating more H35Cl in comparison with H37Cl. Extending this inference to the ion–molecule H-abstraction reactions of 13C6-HCB in the EI source, we conclude that the production of [M + H]+ or [P + H]+ of the compound was slightly higher for the lighter chlorine isotopologues than for the heavier ones. Analogously, it can be deduced that the yields of ions with H-abstraction in the EI source of 13C6-HBB were a little higher for the lighter bromine isotopologues than for the heavier ones. These theoretical inferences are in agreement with the experimental outcomes of this study (Figure and Figure S1).

In addition to the aforementioned activation energy differences derived from the ZPE differences among n class="Chemical">chlorine/pan> class="Chemical">bromine isotopologues, some other factors might also contribute to the H-abstraction efficiency differences among isotopologues, including different average velocities of isotopologues and on-column chlorine/bromine isotope fractionation. In the mixture outflowing from the GC column, lighter isotopologues have higher average velocities than heavier isotopologues since all the isotopologues are at the same temperature condition. After electron impact at the same EI energy, the lighter isotopologues are anticipated to get more translational energy in contrast to the heavier isotopologues, thus having higher average velocities. Additionally, after traversing the electric field with the same accelerating voltage, the lighter isotopologues get higher average velocities than the heavier ones. The higher average velocities can lead to higher collisional rate constants during the collision reactions between analyte ions and siloxanes, resulting in higher H-abstraction efficiencies of the lighter isotopologues relative to the heavier. Furthermore, as revealed in our previous study, HCB and HBB presented large inverse chlorine/bromine isotope fractionation on GC columns; viz., the lighter chlorine/bromine isotopologues outflowed from GC columns later than the heavier isotopologues.[63,64] On the other hand, the density of gas-phase siloxanes in the ion source might be increasing from the beginning to the end of a GC run, leading to more collision chances for the lighter isotopologues than the heavier ones. As a consequence, the H-abstraction efficiencies of chlorine/bromine isotopologues in the EI source decrease with the increase of molecular weights of the isotopologues. In summary, all the above factors can cause higher H-abstraction efficiencies for the lighter chlorine/bromine isotopologues than the heavier ones, which coincides with the observation in the current study (Figure and Figure S1). We deduce that these factors collaboratively resulted in the significant discrepancies of H-abstraction efficiencies among chlorine/bromine isotopologues.

It is noteworthy that the correlations between the H-abstraction efficiencies and the 37Cl or 81Br atom numbers of chlorine/bromine isotopologues exhibit well-fitted Arrhenius-type curves (Figure and Figure S1). This observation implies that the H-abstraction reactions in the EI source were activation energy-dependent (kinetically controlled) and the chlorine and bromine isotope effects played a part in the reactions.

Implications for EI-MS Analysis of HOPs

Nowadays, non-targeted and quasi-targeted analyses as well as suspect screening have become feasible identification methods for known and unknown HOPs.[60] The characteristic isotopologue distributions are always applied to screening HOPs of interest.[60,65−67] The comparison between the theoretical isotopologue distributions and the detected mass spectra of n class="Chemical">isotopologues canpan> facilitate the identification of HOPs, particularly novel HOPs, thus promoting the reliability of identification results. Even so, it remains unpan>clear whether the H-abstraction reactions in the EI source canpan> affect the identification reliability of HOPs when isotopologue distribution comparison is applied. For instanpan>ce, the ion pan> class="Chemical">P-Cl5 with H-abstraction (P-Cl5 + H) of 13C6-HCB has an identical chemical formula to that of the molecular ion of 13C6-pentachlorobenzene, and thus they cannot be differentiated by HRMS even though isotopologue distribution comparison is employed (Table S8). Likewise, the ion P-Br5 + H of 13C6-HBB cannot be distinguished from the molecular ion of 13C6-pentabromobenzene by HRMS in combination with isotopologue distribution comparison. As a result, the dechlorination/debromination ions with H-abstraction of the HOCs containing more chlorine/bromine atoms could trigger interference in the analysis of those containing less chlorine/bromine atoms when HRMS and isotopologue distribution comparison are employed, possibly leading to false-positive identification results. Since the H-abstraction efficiencies of some dechlorination/debromination ions are relatively high (e.g., the P-Br5 of 13C6-HBB), the interference may be considerable. Therefore, other separation techniques such as GC and ion-mobility MS are necessary to couple with HRMS for identification of HOCs, thus eliminating the interferences and improving the validity of analysis results. However, in some scenarios, the GC retention time ranges of HOC congeners containing different Cl/Br atoms may partially overlap. Therefore, more data such as chlorine/bromine isotopologue distribution patterns and quantitative structure–retention relationships are warranted for getting rid of the possible interference caused by H-abstraction reactions in EI-MS.[60] In addition, when using EI-MS data to evaluate the relative abundances of carbon isotopologues and carbon isotope ratios of HOCs, analysts should take into account the contribution of signals from the ions with H-abstraction to the carbon isotopologues containing one 13C atom.

Conclusions

In this work, we systematically investigated the H-abstraction reactions of n class="Chemical">13C6-HCB anpan>d pan> class="Chemical">13C6-HBB in the EI source using GC–HRMS. The H-abstraction efficiencies were assessed with the MS signal intensity ratios of the ions with H-abstraction relative to the corresponding original ions. Three levels of H-abstraction efficiencies, that is, H-abstraction efficiency of each isotopologue, whole H-abstraction efficiency of each ion, and overall H-abstraction efficiency of each compound, were calculated. For most ions, the H-abstraction efficiencies of individual isotopologues declined from the first to the last isotopologues. The whole H-abstraction efficiencies of individual ions can be distinguished from each other for individual compounds, and the overall H-abstraction efficiency of 13C6-HCB was significantly lower than that of 13C6-HBB. In addition, the isotopologue distributions of measured MS signal intensities between [M]·+ and [M + H]+ for each compound were evidently different, and those of the latter contradicted the distributions of theoretical relative abundances of isotopologues. The impacts of several instrumental parameters and factors including temperatures of the ion source, EI energies, and numbers of 37Cl or 81Br atoms of isotopologues on the H-abstraction efficiencies were revealed. The increase of ion source temperatures first enhanced the H-abstraction efficiencies to a maximum and then began to linearly reduce the efficiencies. The H-abstraction efficiencies were significantly affected by the EI energies and emission currents and reached the highest at an EI energy of 45 eV and emission current of 0.9 mA. The mechanisms underlying the H-abstraction reactions of HOCs in the EI source are tentatively proposed in light of the previous studies and evidence found in the current study. The H-abstraction reactions might pertain to ion–molecule reactions, and the siloxanes running off from the GC column possibly participated in the reactions as a hydrogen source. Furthermore, the H-abstraction reactions might be kinetically controlled, and chlorine and bromine isotopic effects were observed in the reactions. This study sheds light on the potential mechanisms and magnitudes of the H-abstraction reactions of HOCs in the EI source and will facilitate the qualitative analysis of HOPs in environmental research, particularly for identification of novel HOPs. In the future, it will be worthwhile to apply computational quantum chemistry in unveiling the in-depth mechanisms behind the EI-induced H-abstraction reactions of HOCs. The details of the possible hydrogen donor and siloxanes bleeding from the GC column also merit further investigation.