Anal. Chem. 2010, 82, 420–426
Direct Compound-Specific Stable Chlorine Isotope Analysis of Organic Compounds with Quadrupole GC/MS Using Standard Isotope Bracketing ¨ rjan Gustafsson† Christoph Aeppli,† Henry Holmstrand,*,† Per Andersson,‡ and O Department of Applied Environmental Science (ITM), Stockholm University, Stockholm, Sweden, and Laboratory for Isotope Geology, Swedish Museum of Natural History, Stockholm, Sweden A method has been developed for the direct determination of the stable chlorine isotope composition (δ37Cl) of organochlorines that eliminates sample preparation, achieves precision comparable to earlier techniques while improving the sensitivity, and makes use of benchtop gas chromatography-quadrupole mass spectrometry instruments (GCqMS). The method is based on the use of multiple injections (n ) 8-10) of the sample, bracketed by a molecularly identical isotopic standard with known δ37Cl, determined using off-line thermal ionization mass spectrometry (TIMS). Mass traces of two isotopologues differing by one chlorine isotope were used to calculate δ37Cl values. Optimization of mass spectrometry and peak integration parameters as well as method validation was achieved using tetrachloroethene (PCE), p,p′-dichlorodiphenyltrichloroethane (DDT), and pentachlorophenol (PCP), spanning a δ37Cl range of -5.5 to +3.2‰ vs SMOC. Injecting 1.6-1100 pmol resulted in standard deviations (1σ) of 0.6-1.3‰, and the δ37Cl results agreed with values independently measured with TIMS. The method was tested by determining the Rayleigh fractionation during evaporation of pure liquid PCE, resulting in a chlorine isotopic enrichment factor of εCl ) -1.1 ( 0.4‰. Furthermore, position-specific δ37Cl analysis based on analysis of DDT mass fragments was evaluated. The GCqMS-δ37Cl method offers a simplified yet sensitive approach for compound-specific chlorine isotope analysis. Chlorinated solvents (e.g., tetrachloroethene, PCE) or pesticides (e.g., p,p′-dichlorodiphenyltrichloroethane, DDT, or pentachlorophenol, PCP) were produced and emitted in large amounts during the past century and are now found ubiquitously distributed throughout the environment. The persistence of these and other anthropogenic organochlorines has led to widespread contamination of soils and groundwater. In Europe, there is currently an estimated legacy of nearly three million contaminated sites that * Corresponding author. Phone: +46 8 674 7229. E-mail: henry.holmstrand@ itm.su.se. † Stockholm University. ‡ Swedish Museum of Natural History.
420
Analytical Chemistry, Vol. 82, No. 1, January 1, 2010
need to be remediated.1 Compound-specific stable isotope analysis (CSIA) is receiving increasing attention in environmental science, because it can be used to identify sources of contaminants as well as to assist in assessing their naturally occurring or artificially stimulated biotransformation as a means to remediate such contaminated sites. In contrast to other isotopic systems (e.g., 13C/12C or D/H), chlorine CSIA has just started to be used to identify and quantify natural degradation in contaminated environments,2,3 assuming a Rayleigh distillation type process.4 These studies are based on the observation in laboratory studies that dechlorination, which often leads to detoxifiction of chlorinated solvents and pesticides, induces shifts in the δ37Cl of the compounds because of kinetic isotope effects.5-9 The interest in using δ37Cl is growing because multiple physical and/or (bio)chemical processes affecting contaminants in the environment may be distinguished using two-dimensional isotopic analysis of chlorine together with carbon isotopes, as has been recently predicted by Hunkeler and co-workers.10 Beside that, Reddy et al. showed that enzymatic chlorination of aromatic compounds yielded larger depletion of δ37Cl (-12‰) compared to abiotic chlorination (-3.5‰).11 This finding suggests that δ37Cl can be used for efficient differentiation between various anthropogenic and natural sources of organochlorine compounds; the latter are apparently also quite numerous.12 This (1) EEA Europe’s environment. The fourth assessment; European Environmental Agency: Copenhagen, 2007. (2) Sturchio, N. C.; Clausen, J. L.; Heraty, L. J.; Huang, L.; Holt, B. D.; Abrajano, T. A. Environ. Sci. Technol. 1998, 32, 3037–3042. (3) Holmstrand, H.; Mandalakis, M.; Zencak, Z.; Andersson, P.; Gustafsson, ¨ . Chemosphere 2007, 69, 1533–1539. O (4) Elsner, M.; Hunkeler, D. Anal. Chem. 2008, 80, 4731–4740. (5) Heraty, L. J.; Fuller, M. E.; Huang, L.; Abrajano, T.; Sturchio, N. C. Org. Geochem. 1999, 30, 793–799. (6) Numata, M.; Nakamura, N.; Koshikawa, H.; Terashima, Y. Environ. Sci. Technol. 2002, 36, 4389–4394. (7) Reddy, C. M.; Drenzek, N. J.; Eglinton, T. I.; Heraty, L. J.; Sturchio, N. C.; Shiner, V. J. Environ. Sci. Pollut. Res. 2002, 9, 183–186. (8) Hofstetter, T. B.; Reddy, C. M.; Heraty, L. J.; Berg, M.; Sturchio, N. C. Environ. Sci. Technol. 2007, 41, 4662–4668. (9) Abe, Y.; Aravena, R.; Zopfi, J.; Shouakar-Stash, O.; Cox, E.; Roberts, J. D.; Hunkeler, D. Environ. Sci. Technol. 2009, 43, 101–107. (10) Hunkeler, D.; Van Breukelen, B. M.; Elsner, M. Environ. Sci. Technol. 2009, 43, 6750–6756. (11) Reddy, C. M.; Xu, L.; Drenzek, N. J.; Sturchio, N. C.; Heraty, L. J.; Kimblin, C.; Butler, A. J. Am. Chem. Soc. 2002, 124, 14526–14527. (12) Gribble, G. W. Acc. Chem. Res. 1998, 31, 141–152. 10.1021/ac902445f 2010 American Chemical Society Published on Web 12/11/2009
concept of δ37Cl-based source apportionment has been applied toward polychlorinated diobenzo-p-dioxins in ball clays.13 The substantial potential in applying chlorine stable isotopes in environmental sciences and other fields is, however, hampered by analytical limitations. Current methods frequently involve laborintensive steps where the target organochlorine substances has to be separated off-line, enriched and converted into a measurable species such as CH3Cl for dual inlet isotope ratio mass spectrometry (DI-IRMS),14 AgCl for fast atom bombardment IRMS (FAB-IRMS),15 or CsCl for thermal ionization mass spectrometry (TIMS).16 Encouragingly, there are recent attempts toward online measurements of δ37Cl by connecting a GC to either an inductively coupled plasma mass spectrometer (ICP-MS),17 an IRMS,18 or a benchtop quadrupole mass spectrometer (qMS),19 but presently all these online methods have limitations that make them unlikely candidates for broadscale applications. The GC-ICP-MS suffers from the low ionization yield of chlorine and the high mass resolution (>104) needed to separate 36Ar-H from 37Cl. The direct GC-IRMS method proved to produce reliable results in a laboratory study.9 However, the Faraday-cup configuration in the IRMS has to be (manually) adjusted for each compound, in practice limiting this method to a narrow set of compounds. In contrast, the GCqMS method allows for scanning a wide range of masses within milliseconds. Although the single detector of qMS is expected to negatively affect the reproducibility and precision of this method, Sakaguchi-So ¨der and co-workers demonstrated in a first test that the precision of GCqMS was sufficient to analytically resolve 37Cl enrichment during abiotic reduction of PCE and trichloroethene (TCE).19 However, that pioneering work fell short of a thorough evaluation of critical mass spectrometry and peak integration parameters as well as of effects of referencing to an isotopic standard that is different than the target molecule. Beside that, the novel GCqMS study was unable to report the Cl isotope ratios relative to standard mean ocean chloride, SMOC, preventing a cross-validation with established δ37Cl methods. The objective of this study was to build on the pioneering work by Sakaguchi-So¨der et al. and to establish an improved measurement strategy and data evaluation scheme for compound-specific δ37Cl analysis of organochlorines using GCqMS. Therefore, we used isotope bracketing with externally isotopically constrained standards that are molecularly identical to the target molecules. This allowed derivation of direct δ37Cl values relative to SMOC and an intermethod comparison with TIMS using PCE, DDT, and PCP as a test set of organochlorine compounds. (13) Holmstrand, H.; Gadomski, D.; Mandalakis, M.; Tysklind, M.; Irvine, R.; ¨ . Environ. Sci. Technol. 2006, 40, 3730–3735. Andersson, P.; Gustafsson, O (14) Holt, B. D.; Sturchio, N. C.; Abrajano, T. A.; Heraty, L. J. Anal. Chem. 1997, 69, 2727–2733. (15) Westaway, K.; Koerner, T.; Fang, Y.; Rudzin ˜ski, J.; Paneth, P. Anal. Chem. 1998, 70, 3548–3552. ¨ . Anal. Chem. 2004, 76, 2336– (16) Holmstrand, H.; Andersson, P.; Gustafsson, O 2342. (17) Van Acker, M. R. M. D.; Shahar, A.; Young, E. D.; Coleman, M. L. Anal. Chem. 2006, 78, 4663–4667. (18) Shouakar-Stash, O.; Drimmie, R. J.; Zhang, M.; Frape, S. K. Appl. Geochem. 2006, 21, 766–781. (19) Sakaguchi-So ¨der, K.; Jager, J.; Grund, H.; Mattha¨us, F.; Schu ¨ th, C. Rapid Commun. Mass Spectrom. 2007, 21, 3077–3084.
EXPERIMENTAL SECTION Chemicals and Preparation of Standard Solutions. PCE (g99.5%), PCP Supelco (98.9%), PCP Aldrich (98%), and cyclopentane were purchased from Sigma-Aldrich (Bellefonte, PA), n-hexane (p.a) was obtained from Merck (Darmstadt, Germany), and an in-house DDT standard was from BDH. A sample of DDT Aldrich was provided by C.M. Reddy (Woods Hole Oceanographic Institution, MA) and a PCE sample from PPG (Pittsburgh, PA) was provided by H.H. Richnow (UFZ Leipzig, Germany). Standard stock solutions of the organochlorines (1-6 g/L) were prepared in cyclopentane (PCE) or n-hexane (DDT, PCP) and subsequently diluted to final concentrations of 0.15-1100 µmol L-1 using volumetric glass flasks and gastight glass syringes. PCP was quantitatively acetylated according to literature.20 GC/MS Analysis and Peak Integration. Volumes of 1 µL of PCE (0.16-16 µmol L-1 in cyclopentane), DDT (1100 µmol L-1 in n-hexane), or acetylated PCP (40 µmol L-1 in n-hexane) were injected on-column in a GC system (GC 8000, Fisons, Manchester, U.K.) equipped with a Supelco Equity-5 capillary column (30 m, 0.25 mm i.d., 0.25 µm film thickness, Sigma-Aldrich) using an AS 2000 autosampler (Fisons). Helium was used as carrier gas with a constant column head pressure of 60 kPa. The temperature programs were for PCE: 40 °C (7 min), ramped to 130 at 30 °C/min (held 1 min); for DDT: 60 °C (1 min), ramped to 210 at 30 °C/min and to 280 at 10 °C/min (held 3 min); for acetylated PCP: 70 °C (1 min), ramped to 230 at 10 °C/min and to 280 at 30 °C/min (held 3 min). For each sample, two masses of molecular ions from isotopologues containing zero and one 37Cl atom (PCE: m/z 164 and 166; DDT: m/z 352 and 354; PCP: m/z 264 and 266) were recorded in the single ion monitoring (SIM) mode using positive electron impact ionization (EI+ at 70 eV) on a MD 800 benchtop quadrupole mass spectrometer (Fisons). An optimal scan time per mass of 50 ms with a delay time of 10 ms was determined (see Results and Discussion). The recorded mass span was set to 0.1 mass units. To avoid interference between isotopologues differing by one mass unit, the qMS was tuned for optimal isotopic resolution within the mass range of interest by optimizing the low and high mass resolution parameters using tris(perfluorobutyl)amine reference gas. The source and interface temperatures were kept at 250 °C, the electron current was set to 245 µA, and the photomultiplier voltage was set to 450 V. Data were recorded using the Masslab software (Fisons) and converted to text files, which were evaluated using an R script (The R Foundation for Statistical Computing, www.r-project.org) detailed in the Supporting Information (section S-3). Thereby, peak and background identification was performed by analyzing the slopes of the mass traces following established procedures from δ13C analysis.21 Critical peak detection and integration parameters are discussed in Results and Discussion. Calculation of Chlorine Isotopic Ratios. In a molecule containing n chlorine atoms, the probability Pz of an isotopologue containing z 37Cl and (n - z) 35Cl atoms is:10 (20) Regueiro, J.; Becerril, E.; Garcia-Jares, C.; Llompart, M. J. Chromatogr., A 2009, 1216, 4693–4702. (21) Ricci, M. P.; Merritt, D. A.; Freeman, K. H.; Hayes, J. M. Org. Geochem. 1994, 21, 561–571.
Analytical Chemistry, Vol. 82, No. 1, January 1, 2010
421
Pz )
n! (X )z(X35Cl)n-z z!(n - z)! 37Cl
(1)
where X37Cl and X35Cl are the relative abundances of 37Cl and 35 Cl, respectively. As Pz is proportional to the area Az under the mass trace of the corresponding isotopologue in the chromatogram, the ratio of areas from two isotopologues containing one and zero 37Cl atom, respectively, is:
Referencing Strategy. In order to report the results against the international standard SMOC, the chlorine isotopic signature of the isotopic standard (δ37Clstd) was measured against SMOC using the TIMS method described below. The chlorine isotopic signatures of the measured samples, δ37Cl, are therefore (in ‰ vs SMOC): δ37Cl )
A1 X37Cl )n× ) nRCl A0 X35Cl 37
(2)
m/z166 1 A1 × m/z164 4 A 0
(3)
DDT RCl )
m/z354 1 A1 × m/z352 5 A 0
(4)
PCP ) RCl
m/z266 1 A1 × m/z264 5 A 0
(5)
Correction for Isotopic Fractionation in the Ion Source and for 13C. As is common practice in stable isotope analysis, we corrected for possible isotopic fractionation during the electron impact ionization process by referring the measured RCl relative to the measured ratio RClstd of a suitable isotopic standard and reporting the result in ‰ vs the isotopic standard:
(
RCl std RCl
)
- 1 × 1000‰
(6)
Because isotopic fractionation in the ion source is very likely compound-specific, the isotopic standards were the same compound as the sample (PCE, DDT, and PCP), injected directly before or after the sample (i.e., standard isotope bracketing). Note that the mass traces of isotopologues having two 13C but zero 37Cl interfere with the monitored monoisotopic 12C compounds with one 37Cl. However, this effect can be corrected for based on the carbon isotopic compositions (RC ) 13C/12C) of the compound (see Supporting Information, section S-1 for derivation of eq 7): corr RCl ) RCl -
1 nC(nC - 1) × × RC2 n 2
(7)
where nC is the number of carbon atoms per molecule. This correction can be applied if the carbon isotopic compositions of the sample and isotopic standard are known, or the range of possible δ13C values can be estimated. 422
std RCl
)
- 1 × 1000‰ + δ37Clstd
≈ δ37Cl' + δ37Clstd
In analogy, RCl of DDT and PCP were calculated from the peak areas of isotopologues with zero 37Cl and with one 37Cl:
δ37Cl' )
RCl
RCl std RCl
(8)
35
where RCl is the chlorine isotopic ratio ( Cl/ Cl). Hence, the chlorine isotopic composition of PCE was calculated from the areas of m/z 166 peak (A1m/z166; isotopologue 12C235Cl337Cl) and of m/z 164 peak (A0m/z164; isotopologue 12C235Cl4):
PCE ) RCl
(
Analytical Chemistry, Vol. 82, No. 1, January 1, 2010
The approximation in eq 8 results in errors 100 scans per peak and yielding a standard deviation of 0.8‰ (1σ, Figure 2A), which is sufficient for many applications. Whereas shorter scan times of 10 ms resulted in higher standard deviations (>1.0‰) due to lower signal-to-noise ratios, scan times of 200 ms led to standard deviations >1.2‰, presumably due to too few scans per peak (40-50). A delay time between the two recorded masses of 10 ms proved to be necessary, because using a shorter delay time (2 ms) resulted in inaccurate results. The injected amount of sample and isotopic standard was also optimized (Figure 2B). The central value of δ37Cl and its precision were constant for injected amounts (sample and isotopic standard) in the range of 1.6 to 16 pmol PCE (corresponding to 0.2-2.2 ng chlorine mass). Too low PCE concentrations (injected amount 4‰. Large PCE injections of 160 pmol yielded chromatographic problems, detector overload, and carry-over effects. The lower detection limit of PCE for the presented method is therefore 1.6 pmol, which is at least 1 order of magnitude lower than the detection limits of PCE using CG-IRMS18 or the previously Analytical Chemistry, Vol. 82, No. 1, January 1, 2010
423
Figure 2. Optimizing the GCqMS-δ37Cl analytical precision and accuracy using a PCE sample (horizontal lines represent the TIMS δ37Cl value of 2.90 ( 0.58‰). (A) Dependence on scan times per mass. (B) Effect of injected amount of the sample and isotopic standard. (C) Influence of varying the sample concentration relative to the isotopic standard.
Figure 3. δ37Cl values of three PCE (circles), one DDT (square), and one PCP sample (triangle) determined by GCqMS compared to the result from the established TIMS method. See Table 1 for details. The results of the two methods are in good agreement (δ37ClGCqMS ) 1.02 × δ37ClTIMS - 0.14, R2 ) 0.97).
information about position-specific chlorine isotope composition (see below). The presented method proved to produce accurate δ37Cl values, as the results from GCqMS and TIMS were identical within the uncertainty for all compounds (Figure 3). This demonstrates that the developed GCqMS method yields accurate results over a broad δ37Cl range and is applicable for small as well as for larger organochlorine molecules. We did not make a correction for the case of two 13C atoms in one molecule according to eq 7 for the values reported in Table 1, because such a correction is small compared to the analytical uncertainty for PCE and PCP (2‰ from the TIMS δ37Cl value (Figure 2C). However, by keeping the concentrations of bracketing-injected sample and isotopic PCE probe within 20%, this effect became small compared to the overall uncertainty. Note that the necessity to achieve equal intensities for sample and standard has also been reported for δ13C-CSIA in continuous flow GC-combustion-IRMS.24 TIMS-GCqMS Intercomparison. To further evaluate the accuracy of the direct GCqMS method, different organochlorine samples (PCE, DDT, and PCP) were analyzed in parallel with a TIMS method (Figure 3, Table 1). Besides the different monitored mass traces, the qMS and data evaluation parameters were the same for DDT and PCP as for PCE. However, as the molecular peaks of DDT are relatively low (
