Article pubs.acs.org/ac
Compound Specific Stable Chlorine Isotopic Analysis of Volatile Aliphatic Compounds Using Gas Chromatography Hyphenated with Multiple Collector Inductively Coupled Plasma Mass Spectrometry Axel Horst,* Julian Renpenning, Hans-Hermann Richnow, and Matthias Gehre Department of Isotope Biogeochemistry, Helmholtz Centre for Environmental Research − UFZ, Permoserstr. 15, 04318 Leipzig, Germany S Supporting Information *
ABSTRACT: Stable chlorine isotope analysis is increasingly used to characterize sources, transformation pathways, and sinks of organic aliphatic compounds, many of them being priority pollutants in groundwater and the atmosphere. A wider use of chlorine isotopes in environmental studies is still inhibited by limitations of the different analytical techniques such as high sample needs, offline preparation, confinement to few compounds and mediocre precision, respectively. Here we present a method for the δ37Cl determination in volatile aliphatic compounds using gas chromatography coupled with multiple-collector inductively coupled plasma mass spectrometry (GC-MC-ICPMS), which overcomes these limitations. The method was evaluated by using a suite of five previously offline characterized in-house standards and eight chlorinated methanes, ethanes, and ethenes. Other than in previous approaches using ICP methods for chlorine isotopes, isobaric interference of the 36ArH dimer with 37Cl was minimized by employing dry plasma conditions. Samples containing 2−3 nmol Cl injected on-column were sufficient to achieve a precision (σ) of 0.1 mUr (1 milliurey = 0.001 = 1‰) or better. Longterm reproducibility and accuracy was always better than 0.3 mUr if organics were analyzed in compound mixtures. Standardization is carried out by using a two-point calibration approach. Drift, even though very small in this study, is corrected by referencing versus an internal standard. The presented method offers a direct, universal, and compound-specific procedure to measure the δ37Cl of a wide array of organic compounds overcoming limitations of previous techniques with the benefits of high sensitivity and accuracy comparable to the best existing approaches.
C
minerals and rocks due to small fractionation and insufficient instrumental precision of >1 mUr.9 Significant differences in natural inorganic samples were published by Kaufmann et al. in 1984 who presented a method with a precision of better than 0.24 mUr using dual inlet gas source isotope ratio mass spectrometry (DI-IRMS).10 Fractionation of chlorine isotopes was first reported in organic compounds due to the larger isotope effects.11,12 Isotope ratio measurement of chlorinated organics improved with the establishment of p-TIMS methods (positive ion thermal ionization mass spectrometry)13−15 and offline conversion of organics to CH3Cl measured by DIIRMS.16,17 Later, online techniques for volatile chlorinated organics were developed measuring fragments of molecules of compounds such as trichloroethene, dichloroethene, and vinyl chloride using GC-IRMS.18,19 Other recent approaches included, for instance, the application of gas chromatography quadrupole mass spectrometry (GCqMS)20,21 and online conversion to HCl for isotope ratio determination via GCIRMS.22,23
hlorinated organic compounds are ubiquitously distributed in the environment. Many of these chemicals, produced and emitted in large amounts in the past, are considered priority pollutants due to their toxicity, persistence, and adverse effects on climate and the stratospheric ozone layer.1,2 Apart from industrially manufactured organohalogens, thousands of naturally produced compounds are released into the environment.3 Compound-specific stable isotope analysis (CSIA) has become an indispensable tool in characterizing sources, transformation paths, and sinks of organic compounds, overcoming the limitation of solely quantifying concentration levels.4 Especially in the field of contaminant science, CSIA is widely used to monitor degradation of chlorinated solvents, BTEX (benzene, toluene, ethylbenzene, xylene) and other pollutants in soils and groundwater.5 To date the overarching majority of studies applied stable carbon and, to a lesser extent, hydrogen isotope analyses to study fate and transformation processes of these substances. Publications presenting δ37Cl measurements are still few in comparison to C and H isotope applications6 even though first chlorine isotope ratio determinations were carried out already in the twenties and thirties of the last century.7,8 Early approaches, however, failed to determine isotopic differences in © XXXX American Chemical Society
Received: May 17, 2017 Accepted: August 7, 2017
A
DOI: 10.1021/acs.analchem.7b01875 Anal. Chem. XXXX, XXX, XXX−XXX
Article
Analytical Chemistry Table 1. Results for Measurements of Chlorinated Compounds Measured Throughout This Studya compounds Methanes methyl chloride (MC) dichloromethane (DCM) trichloromethane (CF) carbon tetrachloride (CT) Ethenes vinyl chloride cis-dichloroethene (cDCE) trichloroethene no. 2 (TCE 2) trichloroethene no. 6 (TCE 6) tetrachloroethene no. 1 (PCE 1) tetrachloroethene no. 5 (PCE 5) Ethanes 1,2-dichloroethane (DCA) 1,1,1-trichloroethane (TCA) 1,1,2,2tetrachloroethane (TeCA)
supplier
purity (%)
formula
δ37Cl off-line DI-IRMS converted to CH3Cl (mUr)
δ37Cl online GC-ICPMS single compound (mUr)
δ37Cl online GC-ICPMS compound mixture (mUr)
Linde AG
99.8
CH3Cl
+6.02 ± 0.03
+6.02 ± 0.15 (37)*
+6.02 ± 0.10 (25)*
ChemSolute
99.9
CH2Cl2
n.a.
+1.99 ± 0.15 (10)
+1.95 ± 0.18 (10)
Roth
99.9
CHCl3
n.a.
−1.95 ± 0.15 (10)
−2.04 ± 0.11 (10)
Fluka
99.5
CCl4
n.a.
+1.13 ± 0.13 (10)
+0.99 ± 0.22 (10)
Linde AG Indiana University Brandt PPG
99.9 >99
C2H3Cl C2H2Cl2
n.a. n.a.
+0.28 ± 0.14 (8) +0.21 ± 0.10 (8)
+0.30 ± 0.27 (7) +0.03 ± 0.15 (9)
>99
C2HCl3
−1.19 ± 0.01
−1.19 ± 0.12 (34)*
−1.19 ± 0.10 (20)*
Merck
99.5
C2HCl3
+2.17 ± 0.20
+2.04 ± 0.13 (29)
+2.04 ± 0.12 (8)
Brandt PPG
>99
C2Cl4
−0.55 ± 0.12
−0.71 ± 0.12 (9)
−0.47 ± 0.21 (4)
Merck
99.9
C2Cl4
+1.03 ± 0.08
+1.14 ± 0.13 (9)
+1.23 ± 0.11 (10)
Sigma-Aldrich
99.8
C2H4Cl2
n.a.
−0.82 ± 0.11 (10)
−0.82 ± 0.11 (10)
Merck
99.9
C2H3Cl3
n.a.
−3.46 ± 0.14 (9)
−3.48 ± 0.09 (10)
Sigma-Aldrich
>98
C2H2Cl4
n.a.
+1.00 ± 0.15 (10)
+1.35 ± 0.23 (10)
The δ37Cl were determined for various chlorinated organics and evaluated using offline values (DI-IRMS) published in a previous study23 and online GC-MC-ICPMS. All compounds were measured online as single compound and as compound mixture. All δ37Cl values have been normalized to the SMOC scale using two-point calibration and are expressed in the SI unit urey (mUr, equivalent to ‰). Replicates of analyses are given by the number in parentheses (n). GC-compatible offline characterized reference material used for 2-point calibration to adjust to the SMOC isotopic scale; n.a., not available. a
and S.27−29 ICP methods for chlorine isotopes, however, are very rare because of isobaric interference of the 36ArH dimer with 37Cl.30−33 In these previous studies interferences were overcome by operating the MC-ICPMS in high mass resolution mode (m/Δm > 10000) for separation of 36ArH and 37Cl. Due to the resulting lower sensitivity, 200−1300 nmol of Cl were required per analysis reducing the applicability of this method especially for environmental samples of organic pollutants. Hence, the objective of this study was to develop an improved method for the δ37Cl measurement of chlorinated aliphatic organics using GC-MC-ICPMS with the special focus on establishing a procedure for routine laboratory application. The most crucial task in this study was to largely avoid isobaric interference of 36ArH on mass 37, that is, sources of hydrogen were minimized by operating a dry plasma. A suite of previously offline-characterized (DI-IRMS) in-house standards were used to evaluate and optimize the performance of the method and to establish a robust protocol for online measurements of δ37Cl.
A wider use of stable chlorine isotope measurements is, however, prevented by the various analytical limitations of these techniques. Offline methods yield a precision of often better than 0.1 mUr but they demand relatively large sample amounts of 50−85 nmol Cl for TIMS and more than 10 μmol for DIIRMS in addition to time-consuming offline conversion techniques.15,17,24 GC-IRMS and GCqMS are more sensitive and allow for the direct measurement of the 37Cl/35Cl ratio in molecular ions and fragments. GC-IRMS, however, is limited to a narrow range of compounds depending on the fixed Faraday cup configuration. GCqMS proved to be more versatile but the instrumental precision of >0.5 mUr may not always be sufficient to resolve isotopic differences in environmental samples. Another challenge is the strong dependence of isotope ratios from the sample amount complicating the referencing approach for GCqMS. Online conversion of organics to HCl was a promising approach to create a truly universal CSIA method for chlorine isotope measurements but the technique suffered from memory effects preventing a routine application of this approach.23 Multiple-collector inductively coupled plasma mass spectrometry (MC-ICPMS) is a sensitive method for the determination of isotope ratios but it is mainly used for inorganic heavier elements such as lead, neodymium, hafnium or strontium.25 Hyphenation of gas chromatography (GC) with MC-ICPMS was first demonstrated by Krupp et al. in 2001 for isotope analysis of organic lead species26 and has also been applied for organic species of other elements such as Br, Hg,
■
EXPERIMENTAL SECTION Chemicals and Standards. Chlorinated methanes, ethanes, and ethenes were purchased from different producers. Details about suppliers and purity of organics are listed in Table 1. Compounds used as standards were characterized offline in a former study by DI-IRMS.23 Five different standards (methyl chloride (MC), 2× trichloroethene (TCE), 2× tetrachloroethene (PCE)) with a known δ37Cl signature were used throughout this study to evaluate the performance of the B
DOI: 10.1021/acs.analchem.7b01875 Anal. Chem. XXXX, XXX, XXX−XXX
Article
Analytical Chemistry method. Additionally, eight compounds with a previously unknown δ37Cl were analyzed: dichloromethane (DCM), trichloromethane (CF), tetrachloromethane (CT), dichloroethane (DCA), trichloroethane (TCA), tetrachloroethane (TeCA), vinyl chloride (VC), and cis-dichloroethene (cisDCE; Table 1). Preparation of Test Mixtures. All organics were analyzed as single compounds and as compound mixtures in the gas phase. Septum bottles (250 mL) were flushed with helium and crimp sealed (Teflon coated Wheaton stoppers). For liquids, 2.5 μL of the pure organic and for methyl chloride 2.5 mL of gas were injected through the stopper with a gastight syringe. Three different compound mixtures were prepared: mix 1 (MC, VC, cisDCE, TCE, and PCE); mix 2 (MC, DCM, CF, and CT); and mix 3 (MC, DCA, TCA, and TeCA). Instrumental Setup, Tuning, and Data Acquisition. For separation of organics, a Thermo Scientific Trace 1310 gas chromatographic system was used, equipped with a flame ionization detector (FID). Samples were injected with a gastight syringe into a split/splitless injector which was kept at 250 °C. A split ratio of 1:10 and a constant carrier gas flow of 2 mL/min were applied. A Zebron ZB-1 capillary column (60 m × 0.32 mm i.d., 1 μm film thickness; Phenomenex Inc.) was operated isothermally (150 °C) for single compound analyses. For compound mixtures a temperature program was applied (hold 40 °C for 2 min, 10 °C/min to 150 °C, hold 10 min). Once separated, the compounds entered the plasma via a Thermo Elemental Transferline AE2080 (Aquitaine Electronique, France) heated to 250 °C to avoid condensation. We used an auxiliary helium flow of 5 mL/min as additional carrier through the transferline to optimize chromatographic peak shape (Figure 1).
Table 2. MC-ICPMS Parameters Used Throughout This Study parameter
value
RF power cooling gas flow rate (Ar) intermediate gas flow rate (Ar) sample gas flow rate (Ar) mass resolving power (m/Δm) extraction cones dwell time Faraday detectors
1200 W 16 L/min 0.7 L/min 1.15 L/min 400 (low) standard nickel 0.262 s C (mass 35), H2 (mass 37)
Supporting Information, Figure S1). The signals of the two Faraday cups were collected with a dwell time of 0.262 s and exported from the chart recorder of the Multicollector Software 3.2 as ASCII files for further data treatment. Isotope Ratio Calculation and Two-Point Normalization. The software for acquiring isotopic ratios from MCICPMS instruments is not optimized for recording transient signals; that is, analytes are usually in solution and introduced continuously into the plasma producing a constant signal. Isotopic ratios are computed directly from these signals recorded at the corresponding detectors (point-by-point calculation). In contrast, gas chromatography generates transient signals (“peaks”), which have to be transformed into isotopic ratios. In this study, isotopic ratios were calculated in two ways: (A) via regression analysis and (B) via conventional integration of the signal area (Figure 2). For regression analysis (A), the intensity recorded for mass 37 was plotted versus mass 35 at every time point of the transient signal. The slope of the best linear fit of these data points represents the isotopic ratio.28,31 A threshold value of 0.05 V for mass 35 was applied to determine start and end of the transient signal. For conventional integration (B), the areas of the transient signals of mass 35 and mass 37 were calculated and a slopebased approach was applied to determine the start and end of the chromatographic peak.34 We used a starting slope of 0.002 and an end slope of −0.004. The baseline was determined by averaging 10 points before the chromatographic peak (Figure 2). The 37Cl/35Cl ratio is computed by dividing the area of the mass 37 signal by the area calculated for mass 35. The delta value expressed as the deviation from standard mean ocean chloride (SMOC) was determined in two steps. First, the sample was referenced against one standard with a known δ37Cl. The difference between sample and standard is given in Ur (urey) according to recent IUPAC recommendations.35
Figure 1. Schematic of the instrumental setup for the GC-MC-ICPMS method.
⎛ 37Cl ⎞ ⎜ 35Cl sample ⎟ δ Cl = ⎜ 37 − 1⎟ ⎜ Cl standard ⎟ ⎝ 35Cl ⎠
The MC-ICPMS instrument, a Neptune (Thermo Fisher Scientific, Germany), was operated in low resolution mode (m/ Δm = 400). Main instrumental parameters are summarized in Table 2. Instrument tuning and optimization was carried out daily in preparation of the measurements. A CH3Cl/He gas mixture (300 μL/L) was aspirated via the nebulizer following a 60 min warm-up period after lighting the plasma (Figure 1). Two Faraday detectors were adjusted to collect the signals for mass 35 and mass 37 respectively. Torch position (x,y), ion extraction lenses and focus were adjusted daily for maximum sensitivity. Sample gas flow and the lateral torch position (z) were optimized once for best sensitivity and minimum background and not altered throughout this study (see also
37
(1)
If expressed in mUr (milliurey), this SI unit is interchangeable with the common permil (‰) scale; that is, 1 mUr = 1‰. The raw δ37Cl values of the samples were obtained by referencing all compounds versus methyl chloride which was either injected before each sequence of single compounds or as an internal standard in compound mixtures. This procedure also corrected for instrumental drift if necessary. In a second step, raw δ37Cl values as obtained from eq 1 were linked to the international reference scale for chlorine (SMOC) by applying C
DOI: 10.1021/acs.analchem.7b01875 Anal. Chem. XXXX, XXX, XXX−XXX
Article
Analytical Chemistry
Figure 2. Comparison of 37Cl/35Cl isotope ratio calculation methods. In (A) the ratio is given by the slope of the linear regression of the signals recorded for mass 37 and mass 35 at each time point (0.262 s) of the chromatographic peak.28,31 In (B) parameters for integration of the transient signals are shown for calculation of isotope ratios from chromatographic peak areas adopted from Ricci et al. .34
a two-point linear normalization approach.23,36,37 Three offline characterized in-house standards (DI-IRMS) were used for this procedure: MC (+6.02 mUr vs SMOC), TCE2 (−1.19 mUr vs SMOC), and TCE6 (+2.17 mUr vs SMOC). Before the measurement of samples started, a sequence of 3−5 MC and TCE2 were injected daily to determine the scale difference compared to DI-IRMS which is considered to present the SMOC scale: δ 37Cl SMOC = a ·δ 37ClRAW + b
(2)
The slope a (scale expansion factor) and intercept b (additive correction factor) represent the linear regression of the measured δ37ClRAW of CH3Cl and TCE2 plotted versus the offline characterized δ37ClSMOC (“true values”) of these compounds (Table 1). A third standard (TCE6) was used to evaluate the accuracy of this correction procedure; that is, the two-point calibrated δ37ClSMOC of TCE6 had to agree with the offline value of +2.17 within 0.2 mUr.
■
Figure 3. Injection of three aliquotes (0.5, 0.3, and 1 mL) of a 20000 ppm of CH4/He gas mix corresponding to 179, 107, and 357 nmol of H injected on-column. The effect on the mass 37 baseline is an increase of at most 0.006 V.
RESULTS AND DISCUSSIONS Background and Isobaric Interference. In the current study measurements were carried out under dry plasma conditions; that is, the addition of hydrogen in the form of water or acids into the plasma was avoided to reduce the formation of the 36ArH dimer which interferes with 37Cl. In dry plasma the main sources of hydrogen are the analyzed organics. The influence of these small amounts of hydrogen on the signal of mass 37 was rigorously tested and examined in different ways. First, the lateral torch position (z) and sample gas flow were adjusted for optimum sensitivity and low background during aspiration of a CH3Cl/He gas mix (300 μL/L) via the nebulizer. Next, peak scans were carried out which did not show the typical shape of isobaric interferences as reported previously, that is, no signs of additional plateaus in the mass scan peaks were detected.30 Background levels were usually below 0.03 V for mass 37 and less than 0.01 V for mass 35 if only He from the GC entered the plasma (Supporting Information, Figures S1−S3). In order to evaluate the effect of hydrogen on 36ArH formation in the plasma, methane was injected into the GC as a hydrogen donor. The addition of up to 360 nmol of H (contained in CH4) to the plasma yielded a signal increase of at most 0.005 V on mass 37 (Figure 3). For comparison, a sample of methyl chloride containing 360 nmol of hydrogen would yield a theoretic signal amplitude of about 190 V for mass 35 (70 V for mass 37) far exceeding both
faraday cups’ capacities. Regular samples would be measured with at most 30 V on mass 35 (see below) and the contribution of 36ArH to the signal of mass 37 would be less than 0.001 V in this case. Hence, we concluded that isobaric interferences of the 36 ArH dimer were negligible in our system and that 37Cl/35Cl measurements would not be significantly affected. As a consequence, the MC-ICPMS was operated in low resolution mode throughout this study which increased the sensitivity significantly compared to medium and high resolution. Data Acquisition and Isotope Ratio Determination. Calculation of the 37Cl/35Cl ratios from the recorded transient signals is a crucial step in isotope measurements influencing the quality of final δ37Cl values. The most widely used technique is the integration of the areas of the transient signals after Ricci et al.34 For continuous flow IRMS this calculation is carried out automatically by programs such as ISODAT 3.0 (Thermo Fisher Scientific). For MC-ICPMS methods no automated approach has been available yet and integration is commonly computed manually using programs such as Microsoft Excel or Matlab.30,32,38 We compared this peak area integration method with the regression technique that was first demonstrated by D
DOI: 10.1021/acs.analchem.7b01875 Anal. Chem. XXXX, XXX, XXX−XXX
Article
Analytical Chemistry
Figure 4. (A) Test for independence of the measured isotope ratio from sample size (linearity) tested for CH3Cl for signal sizes up to 27 V. The dashed line indicates the standard deviation of 0.1 mUr which was achieved for measurements with signal sizes of more than 3 V. (B) Determination of the isotopic limit of detection (LOD) which was the minimum amount of Cl required to produce a chromatographic signal of 3 V.
workflow, it would be desirable if MC-ICPMS software packages offered the possibility of automated peak integration and/or slope calculation for regression analysis to avoid the manual data treatment. Linearity and Limit of Detection for Isotope Analysis. Linearity is an important parameter for isotopic measurement techniques describing whether the measured isotope ratios are independent from the quantity of the analyte or not. Linearity was tested by injecting different amounts of CH3Cl resulting in chromatographic signals of 0.4 V up to 27 V (Figure 4a). Statistically, no significant relationship exists between isotope ratios and signal sizes on the 95% confidence interval if peaks of 3−27 V are considered (R2 = 0.06, p-value = 0.07). A similar test was carried out for PCE and no dependence of isotope ratios and signal sizes could be detected (Supporting Information, Figure S6). The limit of detection for isotope analysis (LOD) is defined as the minimum amount of chlorine which is necessary to achieve a certain precision. For single compound analyses of CH3Cl we achieved a precision of 0.1 mUr or better if the transient signal was larger than 3 V. This minimum acceptable signal size was produced by 2 nmol of chlorine injected oncolumn as CH3Cl (Figure 4b). For PCE, 2.4 nmol of Cl were required to generate a chromatographic peak of 3 V. Overall, these LOD are on a par with gas source IRMS methods and one to 2 orders of magnitude more sensitive than previous MCICPMS or TIMS methods for Cl isotope measurement. Instrumental Drift. Isotope ratios obtained from MCICPMS methods may be subject to instrumental drift; that is, isotope ratios change over time.25 For transient signals, drift is usually corrected by injecting standards with a known isotopic composition before and after the sample. This approach known as standard-sample-bracketing has been used widely for GCMC-ICPMS applications such as Hg, Pb, and Br isotopes.26,28,38 Generally, the reason for drift is not well understood yet. Krupp and Donard investigated systematically the reasons for drift in different GC-MC-ICPMS setups for the measurement of Hg and Pb isotopes and suggested chromatographic issues as the main reason for drift, which, however, could not be proved further.39 In the current study, instrumental drift was generally low. Multiple injections of a CH3Cl/He gas mix did not reveal a systematic trend in the isotopic ratios over the course of more
Fietzke et al. for laser ablation ICPMS and further evaluated by Epov et al. for GC-MC-ICPMS systems.28,31 The performance of both techniques was generally comparable yielding a similar precision for both, single and multicompound analyses. For a sequence of 10 samples of CH3Cl instrumental precision for regression analysis was 0.1 mUr and 0.15 mUr for the integration approach. The δ37Cl of compound mixtures agreed within the analytical error with single compound analyses (Supporting Information, Table S2). The regression method may, however, be the more robust approach because of its insensitivity to variations and noise in the background signal.28 To evaluate this method further, isotope ratios of a transient signal with considerable tailing (TeCA) were calculated with changing parameters and resulting isotope ratios were compared. Specifically, the threshold value of 0.05 V defining the peak end was stepwise increased. For the front part of the chromatographic peak the threshold of 0.05 V was kept constant. This procedure did not change the isotope ratio by more than ±0.15 mUr even though only half of the signal was used in the most extreme case demonstrating the advantage of this method (Supporting Information, Figure S4, Table S1). It should be mentioned at this point that the application of the regression technique requires careful evaluation of the setup of the mass spectrometer. Specifically, amplifier/faraday detector combinations can be changed for certain instruments such as the Neptune and this procedure resulted in a lower R2 and, consequently, a lower precision of the isotope ratios obtained for certain pairings of amplifiers and detectors. A sequence of eight methyl chloride injections, for instance, yielded a standard deviation of only 0.7 mUr compared to the 0.1 or better if higher R2 are achieved. The lower R2 of 0.999377 was caused by a slight hysteresis shape of the data compared to the almost ideal regression (R2 > 0.999997) obtained for the initial amplifier/detector combination (Supporting Information, Figure S5). The reason for this effect is not quite clear, but it might be explained with different time lags in data transmission if amplifiers are rotated. Overall, and after comparing both isotope ratio calculation techniques for some of our data, we decided to use the regression technique throughout this study because of the robustness against background noise and easier handling when calculated manually. For the future, and to further simplify the E
DOI: 10.1021/acs.analchem.7b01875 Anal. Chem. XXXX, XXX, XXX−XXX
Article
Analytical Chemistry
to the procedure outlined above; that is, compounds were first referenced against the internal standard CH3Cl and then a twopoint calibration was applied to relate raw δ37Cl to the SMOC scale. Chromatographic separation of compounds was usually very good as demonstrated in Figure 5. Even if retention times
than 2 h (Supporting Information, Figure S7). When compound mixtures were analyzed, a slight shift of at most 0.3 mUr was observed in the standards TCE2 and CH3Cl over an entire measurement day (8 h). Drift was corrected by referencing analytes against CH3Cl which was either part of the compound mixtures or injected prior to each sequence. These runs lasted for at most 20 min and drift could be considered negligible over this short time range. Consequently, we assume that, due to the small magnitude and our referencing approach, drift does not affect final δ37Cl values. Memory. Artifacts of measurements in the instrumental setup may influence the isotopic composition of subsequently analyzed samples. This memory or carry-over effect was very pronounced for other chlorine isotope measurement techniques such as high temperature conversion of chlorinated organics to HCl.23 We tested our method for memory effects by alternating sequences (n = 5) of CH3Cl and TCE2. Despite the difference of more than 7 mUr between these compounds, no memory effect could be observed (Supporting Information, Figure S8). Results of both CH3Cl and TCE2 agreed within standard deviation (0.15 mUr) of these 15 samples respectively and no significant trend toward the isotopic value of the previous compound was observed. Hence we suggest that successively eluting chromatographic peaks will not influence the δ37Cl of each other if multiple organics are measured in a compound-specific approach. Precision, Reproducibility, and Accuracy of Single Compound Analyses. To evaluate the instrumental precision of δ37Cl measurements, the standard deviations (σ) of numerous single compound analyses were compared. Measurements were carried out within a single sequence and standard deviations were usually better than 0.1 mUr if chromatographic signals smaller than 3 V were omitted. For instance, the multiple injections of CH3Cl for linearity testing yielded a σ of ±0.1 mUr even though signals of 3−27 V were included (Figure 4a). If only chromatographic peaks of similar size were considered, a standard deviation of better than 0.1 mUr could be achieved (e.g., TCE, n = 5, signal size 3 V, σ = 0.03 mUr). This instrumental precision may, however, not be achieved in day-to-day laboratory routine because it does not involve variations due to sample preparation, tuning of the instrument, and referencing. Long-term reproducibility and accuracy was tested by injecting in-house working standards (CH3Cl, TCE2, TCE6, PCE1, PCE5) on three different days preparing new samples for each of these sequences. Additionally, a two-point calibration was carried out every day to relate results to the SMOC scale. Despite these procedures, the δ37Cl of single compound analyses did not differ by more than 0.2 mUr from offline characterized values (Table 1), which is comparable to or better than previously reported for continuous flow techniques such as MC-ICPMS, GC-IRMS, GCqMS, and GC-HTC-IRMS.18,21,23,30 CSIA of Chlorinated Organics. Once the GC-MC-ICPMS method was rigorously evaluated using offline characterized inhouse standards, eight additional chlorinated methanes, ethanes, and ethenes were purchased to prepare compound mixtures. Environmental samples as extracted from contaminated aquifers, for example, may have a similar composition as these multicompound mixtures and experiments were carried out to demonstrate the potential for compound-specific applications. Compound mixtures were analyzed on at least three different days with newly prepared samples for each measurement campaign. The δ37Cl was determined according
Figure 5. Chromatogram for compound-specific chlorine isotope analysis of a chlorinated ethene mixture, including CH3Cl (MC) for referencing, as exported from the chart recorder (Multicollector Software 3.2, Thermo Fisher Scientific).
were very close as for 1,2-DCA and 1,1,1-TCA for example, isotope ratios were not affected (see Supporting Information, Figure S-9, for further chromatograms). Results for these compound-specific analyses ranged from −3.48 mUr versus SMOC for TCA to +1.95 mUr versus SMOC for DCA (Table 1). Generally, δ37Cl from multicompound analyses did not differ, within analytical uncertainty, from results obtained for single compounds. Working standards, if included in these mixtures (TCE6, PCE1, PCE5), were accurate and agreed with the offline characterized value within ±0.2 mUr (Table 1). Multicompound analyses of the additionally analyzed chlorinated compounds agreed within standard deviation (reproducibility) with single compound measurements with a deviation of at most 0.27 mUr for vinyl chloride. Experiments with mixtures of chlorinated organics demonstrate the application of this compound-specific method for samples containing multiple different pollutants as they may be found, for example, in environmental samples from contaminated aquifers. The standard deviations of always better than 0.3 mUr represent the reproducibility and accuracy and may be achieved in day-to-day laboratory routine measurements for compounds that are present as gases or for contaminants that were extracted from water samples.
■
CONCLUSIONS In this study we present a universal, direct and compoundspecific approach for δ37Cl measurement of volatile hydrocarbons using GC-MC-ICPMS. The method shows also high potential for the δ37Cl determination of other compound classes such as aromatic and cyclic hydrocarbons. Analysis of these often semivolatile chemicals may be within reach if condensation in the transferline and torch can be avoided. A previous study investigated brominated flame retardants via GC-MC-ICPMS finding diminished precision for halogen isotopes due to condensation and concluding that a continuously heated transfer to the plasma has to be guaranteed.40 For the tested volatile compounds in the current F
DOI: 10.1021/acs.analchem.7b01875 Anal. Chem. XXXX, XXX, XXX−XXX
Article
Analytical Chemistry
No. 55, World Meteorological Organization: Geneva, Switzerland, 2014. (3) Gribble, G. W. Chemosphere 2003, 52, 289−297. (4) Elsner, M.; Jochmann, M. A.; Hofstetter, T. B.; Hunkeler, D.; Bernstein, A.; Schmidt, T. C.; Schimmelmann, A. Anal. Bioanal. Chem. 2012, 403, 2471−2491. (5) Hunkeler, D.; Meckenstock, R.; Sherwood Lollar, B.; Schmidt, T. C.; Wilson, J. T., A Guide for Assessing Biodegradation and Source Identification of Organic Ground Water Contaminants Using Compound Specific Isotope Analysis (CSIA); U.S. Environmental Protection Agency: Washington, DC, 2008; Vol. EPA/600/R-08/148, EPA2008. (6) Eggenkamp, H. G. M., The Geochemistry of Stable Chlorine and Bromine Isotopes; Springer-Verlag: Berlin Heidelberg, 2014; p 172. (7) Curie, I. CR Hebd Acad. Sci. 1921, 172, 1025−1028. (8) Nier, A. O.; Hanson, E. E. Phys. Rev. 1936, 50, 722−726. (9) Hoering, T. C.; Parker, P. L. Geochim. Cosmochim. Acta 1961, 23, 186−199. (10) Kaufmann, R.; Long, A.; Bentley, H.; Davis, S. Nature 1984, 309, 338−340. (11) Baertschi, P.; Kuhn, W.; Kuhn, H. Nature 1953, 171, 1018− 1020. (12) Bartholomew, R. M.; Brown, F.; Lounsbury, M. Can. J. Chem. 1954, 32, 979−983. (13) Xiao, Y. K.; Zhang, C. G. Int. J. Mass Spectrom. Ion Processes 1992, 116, 183−192. (14) Magenheim, A. J.; Spivack, A. J.; Volpe, C.; Ransom, B. Geochim. Cosmochim. Acta 1994, 58, 3117−3121. (15) Holmstrand, H.; Andersson, P.; Gustafsson, O. Anal. Chem. 2004, 76, 2336−2342. (16) van Warmerdam, E. M.; Frape, S. K.; Aravena, R.; Drimmie, R. J.; Flatt, H.; Cherry, J. A. Appl. Geochem. 1995, 10, 547−552. (17) Holt, B. D.; Sturchio, N. C.; Abrajano, T. A.; Heraty, L. J. Anal. Chem. 1997, 69, 2727−2733. (18) Shouakar-Stash, O.; Drimmie, R. J.; Zhang, M.; Frape, S. K. Appl. Geochem. 2006, 21, 766−781. (19) Shouakar-Stash, O.; Frape, S. K.; Aravena, R.; Gargini, A.; Pasini, M.; Drimmie, R. J. Environ. Forensics 2009, 10, 299−306. (20) Sakaguchi-Soder, K.; Jager, J.; Grund, H.; Matthaus, F.; Schuth, C. Rapid Commun. Mass Spectrom. 2007, 21, 3077−3084. (21) Aeppli, C.; Holmstrand, H.; Andersson, P.; Gustafsson, O. Anal. Chem. 2010, 82, 420−426. (22) Hitzfeld, K. L.; Gehre, M.; Richnow, H. H. Rapid Commun. Mass Spectrom. 2011, 25, 3114−3122. (23) Renpenning, J.; Hitzfeld, K. L.; Gilevska, T.; Nijenhuis, I.; Gehre, M.; Richnow, H. H. Anal. Chem. 2015, 87, 2832−2839. (24) Numata, M.; Nakamura, N.; Gamo, T. Geochem. J. 2001, 35, 89−100. (25) Becker, J. S. Inorganic Mass Spectrometry; John Wiley & Sons Ltd, 2007; p 514. (26) Krupp, E. M.; Pecheyran, C.; Meffan-Main, S.; Donard, O. F. Fresenius' J. Anal. Chem. 2001, 370, 573−580. (27) Sylva, S. P.; Ball, L.; Nelson, R. K.; Reddy, C. M. Rapid Commun. Mass Spectrom. 2007, 21, 3301−3305. (28) Epov, V. N.; Rodriguez-Gonzalez, P.; Sonke, J. E.; Tessier, E.; Amouroux, D.; Bourgoin, L. M.; Donard, O. F. Anal. Chem. 2008, 80, 3530−3538. (29) Amrani, A.; Sessions, A. L.; Adkins, J. F. Anal. Chem. 2009, 81, 9027−9034. (30) Van Acker, M. R.; Shahar, A.; Young, E. D.; Coleman, M. L. Anal. Chem. 2006, 78, 4663−4667. (31) Fietzke, J.; Frische, M.; Hansteen, T. H.; Eisenhauer, A. J. Anal. At. Spectrom. 2008, 23, 769. (32) Zakon, Y.; Halicz, L.; Gelman, F. Anal. Chem. 2014, 86, 6495− 6500. (33) de Gois, J. S.; Costas-Rodríguez, M.; Vallelonga, P.; Borges, D. L. G.; Vanhaecke, F. J. Anal. At. Spectrom. 2016, 31, 537−542. (34) Ricci, M. P.; Merritt, D. A.; Freeman, K. H.; Hayes, J. M. Org. Geochem. 1994, 21, 561−571.
study precision, accuracy, and LOD are on a par with other state of the art approaches, but the presented method offers a more straightforward referencing technique which does not require molecular identical standards for each of the analytes in the sample. Furthermore, time-consuming standard-sample bracketing procedures are avoided. This simple method is readily applicable for environmental and laboratory samples of volatile aliphatic compounds which may be measured via headspace and purge and trap methods and it can be easily adapted to existing MC-ICPMS instruments. A crucial requirement for future chlorine isotope studies is, however, the establishment of suitable reference material. Particularly organic GC-compatible international standards are currently not available. Moreover, it would be necessary to establish a second official consensus point to the δ37Cl scale because all current measurements base on a scale determined by a few IRMS instruments. A recent study presented new reference material for perchlorates which, due to the very large range of 88 mUr, may be an appropriate candidate and a first step toward such a consensus point.41
■
ASSOCIATED CONTENT
* Supporting Information S
The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.analchem.7b01875. Further details on background mass scans/peak scans, quality of isotope calculation techniques, mass bias drift, memory and chromatograms of chlorinated methanes and ethanes (PDF).
■
AUTHOR INFORMATION
Corresponding Author
*E-mail:
[email protected]. ORCID
Axel Horst: 0000-0002-3475-2425 Notes
The authors declare no competing financial interest.
■
ACKNOWLEDGMENTS We are grateful to the Centre for Chemical Microscopy (ProVIS) at the Helmholtz Centre for Environmental Research supported by European Regional Development Funds (EFRE − Europe funds Saxony) for using the MC-ICPMS at their analytical facilities. Further we acknowledge offline preparation of organic standards by Tetyana Gilevska and Natalja Ivdra and δ37Cl-DI-IRMS measurement of standards by Magali Bonifacie, IPGP, Paris, France. This project has received funding from the European Union’s Horizon 2020 research and innovation programme under the Marie Skłodowska-Curie Grant Agreement No. 701350.
■
REFERENCES
(1) European Environment Agency. Hazardous substances in Europe’s fresh and marine waters. EAA Technical Report No. 8/2011, EEA: Copenhagen, 2011. (2) Carpenter, L. J.; Reimann, S.; Burkholder, J. B.; Clerbaux, C.; Hall, B. D.; Hossaini, R.; Laube, J. C.; Yvon-Lewis, S. A., OzoneDepleting Substances (ODSs) and Other Gases of Interest to the Montreal Protocol, Chapter 1 in Scientific Assessment of Ozone Depletion: 2014. Global Ozone Research and Monitoring Project Report G
DOI: 10.1021/acs.analchem.7b01875 Anal. Chem. XXXX, XXX, XXX−XXX
Article
Analytical Chemistry (35) Brand, W. A.; Coplen, T. B. Isot. Environ. Health Stud. 2012, 48, 393−409. (36) Coplen, T. B. Chem. Geol. 1988, 72, 293−297. (37) Paul, D.; Skrzypek, G.; Forizs, I. Rapid Commun. Mass Spectrom. 2007, 21, 3006−3014. (38) Horst, A.; Holmstrand, H.; Andersson, P.; Andersson, A.; Carrizo, D.; Thornton, B. F.; Gustafsson, O. Rapid Commun. Mass Spectrom. 2011, 25, 2425−2432. (39) Krupp, E. M.; Donard, O. F. X. Int. J. Mass Spectrom. 2005, 242, 233−242. (40) Holmstrand, H.; Unger, M.; Carrizo, D.; Andersson, P.; Gustafsson, O. Rapid Commun. Mass Spectrom. 2010, 24, 2135−2142. (41) Bohlke, J. K.; Mroczkowski, S. J.; Sturchio, N. C.; Heraty, L. J.; Richman, K. W.; Sullivan, D. B.; Griffith, K. N.; Gu, B.; Hatzinger, P. B. Rapid Commun. Mass Spectrom. 2017, 31, 85−110.
H
DOI: 10.1021/acs.analchem.7b01875 Anal. Chem. XXXX, XXX, XXX−XXX
