Acceptance Criteria for Ultratrace HPLC−Tandem Mass Spectrometry

A Shimadzu LC-10AD pump was used to deliver eluent through a Keystone ..... In these cases, one must either resort to monitoring just those ions obser...
0 downloads 0 Views 556KB Size
Download PDF
Anal. Chem. 1996, 68, 3397-3404

Acceptance Criteria for Ultratrace HPLC-Tandem Mass Spectrometry: Quantitative and Qualitative Determination of Sulfonylurea Herbicides in Soil Lily Y. T. Li, Dale A. Campbell, Patrick K. Bennett, and Jack Henion*

Advanced BioAnalytical Services, Inc., 15 Catherwood Road, Ithaca, New York 14850

Eight commonly used sulfonylureas (SUs: nicosulfuron, thifensulfuron methyl, metsulfuron methyl, sulfometuron methyl, chlorsulfuron, bensulfuron methyl, tribenuron methyl, and chlorimuron methyl) and deuterium-labeled nicosulfuron (nicosulfuron-d6), used as an internal standard, were isolated from soil by solvent extraction and identified under quantitative and qualitative ion spray LC/ MS/MS conditions using the selected reaction monitoring (SRM) mode of acquisition. The lower level of quantitation for these SUs in soil was determined at the 0.05 ppb level using a TurboIonSpray adapted LC/MS interface without a precolumn split and optimizing MS/MS tuning conditions for individual SUs. The eight SUs were qualitatively identified and quantitatively determined in soil. The standard curve for each SU was linear from 0.05 to 10 ppb. This SRM LC/MS method demonstrates high sensitivity and high specificity for these SUs in soil and shows at least a 400-fold improvement in sensitivity over previous reports. Acceptance criteria for forensically valid data are suggested for qualitative SRM LC/MS experiments. These include HPLC retention time reproducibility ((2%), at least two and preferably three precursorproduct ions selected, and relative abundance criteria for selected ions ((20% absolute). The trace determination of organic residues in environmental samples presents a challenging analytical problem. In particular, the isolation, identification, and quantitation of polar, labile analytes such as sulfonylureas (SUs) in soil has been shown to be difficult due to a variety of problems.1-5 Although gas chromatography with conventional detectors has been shown to be applicable in certain instances,5,6 more recently it has been shown that combined HPLC and mass spectrometry (LC/MS) is a more suitable approach.1,3,4,7 Recent applications involving the determination of sulfonylurea herbicides have included thermospray LC/MS techniques which have shown a lower level of quantitation (1) Reiser, R. W.; Barefoot, A. C.; Dietrich, R. F.; Fogiel, A. J.; Johnson, W. R.; Scott, M. T. J. Chromatogr. 1991, 554, 91-101. (2) Wittenbach, V. A.; Koeppe, M. K.; Lichtner, F. T.; Zimmerman, W. T.; Reiser, R. W. Pestic. Biochem. Physiol. 1994, 49, 72-81. (3) Reiser, R. W.; Fogiel, A. J. Rapid Commun. Mass Spectrom. 1994, 8, 252257. (4) Dietrich, R. F.; Reiser, R. W.; Stieglitz, B. J. Agric. Food Chem. 1995, 43, 531-536. (5) Cotterill, E. G. Pestic. Sci. 1992, 34, 291-295. (6) Meyer, M. T.; Mills, M. S.; Thurman, E. M. J. Chromatogr. 1993, 629, 5559. (7) Shalaby, L. M.; Reiser, R. W. In Mass Spectrometry of Biological Materials; McEwen, C. N., Larsen, B. S., Eds.; Practical Spectroscopy Series 8; Marcel Dekker, Inc: New York, 1990; pp 379-402. S0003-2700(96)00375-7 CCC: $12.00

© 1996 American Chemical Society

(LLQ) in the range of 20 parts per billion (ppb).7 Continuousflow fast atom bombardment (CFAB) LC/MS techniques using packed capillaries have also been used.1,3 A recent review8 describes and compares the utility of thermospray, particle beam, and atmospheric pressure ionization (API) techniques for environmental applications for polar pesticides and related compounds. The report offers the rather obvious conclusion that LC/MS techniques extend the range of applications beyond those amenable to GC/MS techniques for environmental applications. It also provides examples where tandem mass spectrometry (MS/MS) techniques provide additional structural information as well as higher specificity than single mass analyzers equipped with mild ionization ion sources. Chen et al.9 described the use of immunoabsorbents for the selective trace enrichment of phenylurea and triazine herbicides in environmental waters. This use of molecular recognition for recovering and traceenriching targeted compounds from complex samples is shown to have considerable potential in future sample preparation strategies. In an alternative analytical approach, Garcia and Henion10 described the use of capillary electrophoresis coupled with electrospray mass spectrometry (CE/MS) and CE/MS/MS for the rapid on-line separation and characterization of sulfonylureas as synthetic mixtures as well as fortified soil extracts. These polar, thermally labile compounds were shown to be very amenable to this analytical strategy, although ultratrace detection limits appear to be precluded due to the very small sample volumes that were loaded into the CE capillary. Although the above analytical approaches may be suitable for certain analyses, it is desirable to reach lower levels of detection while still maintaining sufficiently high specificity to unequivocally identify and quantify selected target analytes in complex matrices. For this reason, we have chosen the combined techniques of HPLC and ion spray tandem mass spectrometry (LC/MS/MS)11-13 to detect eight targeted sulfonylurea herbicides isolated from soil samples. The experiments were carried out on-line with HPLC in the selected reaction monitoring (SRM) mode and are referred to in this work as the SRM LC/MS technique (SRM LC/MS/MS would be redundant). The structures of nicosulfuron, thifensul(8) Slobodnik, J.; van Baar, B. L. M.; Brinkman, U. A. Th. J. Chromatogr. 1995, 703, 81-121. (9) Chen, L.; Hennion, M.-C.; Daniel, R.; Martel, A.; Le Goffic, F.; Abian, J.; Barcelo, D. Anal. Chem. 1995, 67, 2451-2454. (10) Garcia, F.; Henion, J. J. Chromatogr. 1992, 606, 237-247. (11) Bruins, A. P.; Covey, T. R.; Henion, J. D. Anal. Chem. 1987, 59, 26422646. (12) Covey, T. R.; Lee, E. D.; Bruins, A. P.; Henion, J. D. Anal. Chem. 1986, 1451A-1461A. (13) Huang, E. C.; Conboy, J. J.; Wachs, T.; Henion, J. D. Anal. Chem. 1990, 62, 713A-725A.

Analytical Chemistry, Vol. 68, No. 19, October 1, 1996 3397

of the target compounds in unknown samples. This report details an analytical approach for the ultratrace qualitative and quantitative determination of the eight selected SUs in selected soil samples and suggests acceptance criteria for these determinations under SRM LC/MS conditions.

Figure 1. Structures for nicosulfuron, thifensulfuron methyl, metsulfuron methyl, sulfometuron methyl, chlorsulfuron, bensulfluron methyl, tribenuron methyl, and chlorimuron methyl, and nicosulfurond6.

furon methyl, metsulfuron methyl, sulfometuron methyl, chlorsulfuron, bensulfuron methyl, tribenuron methyl, and chlorimuron methyl as well as deuterium-labeled nicosulfuron (nicosulfurond6), used as an internal standard, are shown in Figure 1. We have employed conventional sample extraction techniques using liquid/ liquid extraction for the selective isolation of the eight targeted sulfonylureas from soil.7 This approach is parallel in principle to other environmental and related analyses reported previously.14-16 The combined attributes of liquid/liquid extraction coupled with ion spray LC/MS/MS techniques under SRM conditions provide a unique combination of analytical capability14 which minimizes adsorptive and degradative losses of the fragile sulfonylurea compounds while providing very high sensitivity and high specificity for these compounds in environmental extracts. The LC/MS/MS approach described here provides a lower level of detection (or LLQ) of 0.05 ppb for the SUs. The combination of HPLC retention time with selected unique product ions from the corresponding precursor ion of each of the target compounds provides information that can confirm the presence or absence of a target compound.17 When the extracts of actual unknown soil samples are compared with control soil samples known not to contain the target compound as well as spiked soil samples with known low levels of the target compounds, one is able to report with confidence the “presence” or “absence” of trace levels (14) Edlund, P. O.; Lee, E. D.; Budde, W. L.; Henion, J. D. Biomed. Environ. Mass Spectrom. 1989, 18, 233-240. (15) Rule, G.; Mordehai, A.; Henion, J. D. Anal. Chem. 1994, 66, 230-235. (16) Rule, G.; Henion, J. D. J. Chromatogr. 1992, 582, 103-112 . (17) Weidolf, L. O. G.; Chichila, T. M. P.; Henion, J. D. J. Chromatogr. 1988, 433, 9-21.

3398 Analytical Chemistry, Vol. 68, No. 19, October 1, 1996

EXPERIMENTAL SECTION Materials. Salt-free nicosulfuron, thifensulfuron methyl, metsulfuron methyl, sulfometuron methyl, chlorsulfuron, bensulfuron methyl, tribenuron methyl, and chlorimuron methyl, and nicosulfuron-d6 were supplied by E. I. du Pont Nemours & Co. (Wilmington, DE). Solvents used for soil extraction were acetonitrile (ACN), methanol (Catalog No. 015-4, Catalog No. 230-4, Baxter/ Scientific Products), and chloroform (Catalog No. 9174-03, J. T. Baker). Solutions of 90:10 and 10:90 ACN/5 mM ammonium acetate (pH 3) were prepared for HPLC mobile phases. Ammonium acetate was purchased from J. T. Baker, and formic acid (Suprapur grade, Catalog No. 11670-1) was used to adjust solvent pH. Control (blank) soil samples, known to have been free from any crop protection chemical treatment for at least the past 12 years, were obtained from the private property of one of us (J.D.H.). Preparation of SU Standards. Each SU was dissolved in 100% ACN to produce a concentration of 5 µg/mL. Then, 100 µL of each SU solution (5 µg/mL) was mixed and diluted with ACN to a volume of 1 mL to give a solution containing 0.5 µg/mL of each SU. Appropriate volumes of this solution (0.5 µg/mL) were used to spike soil to produce standard curves from 0.05 to 10 ppb. Preparation of Soil Samples. Ten grams of room temperature, air-dried soil ((0.3 g) was weighed into a 50 mL polypropylene conical tube. One milliliter of internal standard (0.1 µg/ mL nicosulfuron-d6) in methanol was added to 10 g of soil to provide 10 ppb of nicosulfuron-d6 in each soil sample, except for the control blank soil, which received 1 mL of blank methanol. An additional 3 mL of methanol was added to each soil sample, and the samples were vortexed to produce a slurry. Standards were then spiked into the slurried soil samples. After mixing on a vortex mixer, soil samples were dried under N2 in a TurboVap (Zymark, Hopkinton, MA) at 45 °C for 1 h to evaporate the methanol. Liquid/Liquid Extraction of Sulfonylureas from Soil. Ten milliliters of 90:10 ACN/H2O was added to air-dried soil samples. The samples were mixed on a vortex mixer at high speed. After centrifugation of the solution at 2000 rpm (DuPont Sorvall, 6000D centrifuge) for 5 min, the upper liquid layer was transferred and filtered through Whatman Autovials (AV125IMAO) equipped with 0.2 µm Nylon-66 membrane. The filtrate was evaporated under N2 at 45 °C for 1 h, leaving ∼1 mL of solution. An additional 3 mL of chloroform was added to each sample, which was then capped and rotated for 5 min on a hematological mixer. The upper aqueous layer was removed with vacuum suction, and the remaining organic layer was evaporated under N2 to dryness at 45 °C. The residue was reconstituted in 100 µL of 90:10 ACN/ H2O in 5 mM NH4OAc (pH 3.5) and filtered through 0.2 µm TitanMSF nylon microsample filter (SRI 62215-NN) by centrifugation at 2000 rpm for 2 min. The filtrate was then transferred to a polystyrene autosampler vial (SRI 78743-TM) for analysis by SRM LC/MS under positive ion TurboIonSpray conditions. Instrumental Conditions. A Shimadzu LC-10AD pump was used to deliver eluent through a Keystone Betasil C18 column (2 mm × 150 mm, 5 µm, 100 Å), at a flow of 200 µL/mL. Two

Table 1. Summary of MS/MS Mass Spectra for SU Precursor Ions and Product Ions Shown in Figure 3a m/z

b

sulfonylurea

(M + H)+

product ionsb

nicosulfuron thifensulfuron methyl metsulfuron methyl sulfometuron methyl chlorsulfuron bensulfuron methyl tribenuron methyl chlorimuron ethyl nicosulfuron-d6

411 388 382 365 358 411 396 415 417

182*, 213, 366 141, 167*, 205 141, 167*, 199 150*, 199 141, 167* 149, 182*, 213 155*, 181, 199, 363 186*, 213 188*, 213, 371

a These data were obtained by continuous infusion experiments. Asterisk indicates predominant product ion.

solvents [solvent A: 90:10 ACN/5mM NH4OAc (formic acid, pH 3); solvent B: 10:90 ACN/5 mM NH4OAc (formic acid, pH 3)] were employed to run a gradient condition from 78% B to 50% B in 10 min, to 30% B in 2 min, and to 20% B in 3 min. Undiluted formic acid (Suprapur grade) was used to adjust the eluent pH to 3, and the injection volume was 10 µL. The initial eluent conditions of 78% B were reestablished in 0.1 min, and the column was equilibrated for another 5 min. A PE Sciex API IIIplus triple-quadrupole mass spectrometer was operated in the positive ion mode using an ion spray LC/MS interface with the TurboIonSpray option (PE Sciex, Concord, Ontario, Canada) under SRM conditions. Preliminary tuning and mass axis calibration were carried out with continuous introduction of a dilute solution of poly(propylene glycols).12 Poly(propylene glycol) 425, 1000, and 2000 (Catalog No. 20230-4, 20232-0, 20233-9, Aldrich Chemical Co.) in 50:50 methanol/H2O (2 mM NH4OAc, 0.1% formic acid) were used for tuning and mass axis calibration of each mass-resolving quadrupole Q1 and Q3. Unit mass resolution was established and maintained in each massresolving quadrupole by maintaining a full width at half-maximum of between 0.6 and 0.7 Da. Data acquisition was divided into four periods (Table 1), where individual ion optics and MS/MS tuning parameters were optimized to provide optimal sensitivity for the individual SUs. Nebulization gas (liquid N2 dewar) was set at 60 psi; curtain gas (UHP N2) was maintained at 1.2 L/min; collision gas (UHP argon) was set at an approximate collision thickness of 250 × 1012 atoms/cm2. The API III standard software (v2.4, Rev.D) from PE Sciex was used for data acquisition and presentation on a Macintosh Quadra 800. RESULTS AND DISCUSSION Recovery experiments were carried out at the 1 and 10 ppb levels of spiked soils. The control soil sample is representative of an organic soil. Five replicates of each SU were made at each of the indicated levels. The recovery of each SU from the soil at each of the two levels ranged from 30 to 60%, and the precision for the 1 and 10 ppb levels for all of the SUs was within 10%. Recovery studies as a function of pH, percent organic matter, extended storage time, etc. were not undertaken. All eight components in the synthetic mixture of SUs were initially separated by HPLC within 17 min, with the exception of bensulfuron methyl and tribenuron methyl, using just one tuning period (Figure 2). The HPLC separation was conducted on a 2 mm i.d. × 150 mm Betasil C18 column, which was used for its

Figure 2. Total SRM LC/MS chromatogram for a synthetic mixture containing eight SUs (500 pg of each). The chromatogram was obtained using one period for the entire analysis at a collision energy of 15 eV and a dwell time of 300 ms for each selected ion transition.

Figure 3. Full-scan ion spray MS/MS mass spectra of eight SUs (125 pg/µL of each in 100% acetonitrile). The mass spectra were obtained by continuous infusion at 10 µL/min, scanning at 0.1 amu with a dwell time of 2 ms for 10 summed scans.

combination of practical analytical ruggedness and optimum ion spray sensitivity. The synthetic mixture analyzed to produce the results shown in Figure 2 contained 500 pg each of the eight sulfonylurea compounds under study, and the SRM LC/MS ion current profile was obtained using just one tuning period for the entire analysis. The collision energy was 15 eV, and the dwell time for each transition monitored was 300 ms. The gradient was optimized to provide the maximum separation possible in a minimum time period. The eluent and gradient conditions used are detailed in the Experimental Section. The full-scan MS/MS mass spectra for each of the sulfonylurea compounds described in this study are shown in Figure 3. The major ions of interest observed in each full-scan collision-induced dissociation (CID) mass spectrum are listed in Table 1. Those ions marked by an asterisk in Table 1 were used as the selected product ions for the corresponding compounds listed in Table 1. Analytical Chemistry, Vol. 68, No. 19, October 1, 1996

3399

Table 2. Experimental Conditions for the SRM LC/MS Determination of SUs in Soil Extracts Using Four Different Individually Optimized Tuning Periods MS/MS conditions precursor ion (m/z) product ions (m/z)

1 (9 min) nicosulfuron, 411 f 182 nicosulfuron-d6 417 f 188

declustering potential (V) collision energy (eV) dwell time (ms)

30 15 1000

2 (4 min)

3 (3 min)

4 (4 min)

thifensulfuron methyl, 388 f 167 metsulfuron methyl, 382 f 167 sulfometuron methyl, 365 f 150 chlorsulfuron, 358 f 167 30 18 600

bensulfuron methyl, 411 f 149 tribenuron methyl, 396 f 155

chlorimuron ethyl, 415 f 186

30 16 1000

30 15 1500

The most abundant product ion transition for each compound was monitored to obtain the highest quantitative sensitivity possible. To further optimize the sensitivity and MS/MS conditions for the individual SU compounds, SRM LC/MS was performed using four different settings of the ion optics and MS/MS tuning conditions (Table 2) to achieve the lowest level of quantitation for the SUs in soil. The chromatographic run shown in Figure 2 was later divided into four time intervals to establish individually defined regions of optimized ion optics parameters and MS/MS tuning periods. This provides progressive optimized tuning to account for differences in the gradient eluent composition as well as differences in the analytes themselves. In contrast to the use of one tuning period for the entire SRM LC/MS analysis, the use of multiple periods provides individualized MS/MS tuning conditions for the SU compounds to optimize the ion current intensity for each analyte. Although it is possible to monitor a number of different compounds in one SRM LC/MS run using the same MS/MS conditions (e.g., collision gas thickness, rod offset voltages, settling times, etc.), improved sensitivity and performance may often be obtained by individually optimizing these tuning parameters for each compound. This results from varying collision cross sections of different analytes as well as their differing bond strengths, which affects the ease of bond fragmentation and the production of their product ions. Therefore, the individualized ion optics and MS/ MS tuning were empirically optimized for each time window shown in Figures 4-6. This individualized tuning optimization provided a ∼5-fold sensitivity improvement over the tuning conditions used for the single tuning period ion spray SRM LC/ MS results shown in Figure 2 (data not shown). In addition, the mass spectrometer sampling rate or dwell time can be compromised when all the selected ions (e.g., >15 ions in this work) are monitored simultaneously in one MS/MS tuning period. By dividing the chromatographic run time window into four periods, one need not monitor ions for all the analytes simultaneously throughout the chromatographic run. Therefore, unusually long dwell times ranging from 600 to 1500 ms may be used (see Table 2). These increased dwell times can provide some additional, although not significant, improvement in detection limits. The chromatograms shown in Figures 4-6 reveal the four distinct tuning periods, which begin at 0, 9.0, 13.0, and 15.3 min, respectively. (The vertical lines at these time intervals indicate a period change.) Quantitative Determination. The quantitative determination of the target compounds in soil was accomplished by measuring the chromatographic ion current peak area of the predominant (most abundant) precursor-product ion transition for each SU 3400 Analytical Chemistry, Vol. 68, No. 19, October 1, 1996

Figure 4. Total selected ion current chromatograms of blank soil extract. The chromatograms were obtained following optimization of MS/MS conditions for each SU compound using four different tuning periods. The arrows denote the expected retention time for each SU compound.

precursor ion that is present in its corresponding full-scan MS/ MS mass spectrum (Figure 3). The peak area of each transition was divided by the m/z 417 f 188 precursor-product ion transition peak area in the nicosulfuron-d6 internal standard. The quotient of these values provided the peak area ratio, which was plotted versus the soil concentration in ppb. The absence of interference from the soil matrix for any of the eight SUs in control soil samples assures that this SRM LC/ MS method provides high selectivity and specificity for the target SU compounds. Figure 4 shows the individual selected ion current profiles from the SRM LC/MS analysis of a blank soil extract while monitoring the most predominant precursorproduct ion transition for each of the eight SUs, including the nicosulfuron-d6 internal standard. The arrows in Figure 4 depict the retention time region where each of the target SUs would be

Figure 5. Selected ion current chromatograms for 0.05 ppb levels of SUs in soil. The chromatograms were obtained after optimization of MS/MS conditions for each SU compound using four different tuning periods.

expected to elute under these experimental conditions if they were present in the blank soil. The absence of any chromatographic components at the expected retention time indicates there are no interferences from either the target SUs or interfering components in these soil samples (Figure 4). From our earlier preliminary results (data not shown), the detection limit for SUs in soil was previously determined to be 1 ppb. These results were obtained without the TurboIonSpray feature and without using the four different optimized tuning periods. The sensitivity of the previous SRM LC/MS experiments (Figure 2) was improved at least 20-fold using the TurboIonSpray option and optimized multiple ion optics tuning periods for individual SU compounds. The TurboIonSpray feature is a relatively recent hardware option produced by PE Sciex. This device provides additional heated, high-volume nitrogen gas, which is directed orthogonally onto the ion spray droplet plume. This additional mechanical dispersion of the ion spray droplets improves vaporization of solvent and provides improved sensitivity under higher HPLC eluent flow conditions. After these technical modifications, the LLQ for the target SUs in soil was lowered to 0.05 ppb (0.05 ng/g of soil, vide infra. Figure 5 shows the TurboIonSpray SRM LC/MS selected ion current chromatograms from an extract of control soil fortified with the eight SUs at the 0.05 ppb level. This represents a conservative LLQ for these compounds, as indicated by the high signal-to-noise ratios observed in Figure 5. All the SUs are chromatographically resolved from each other under these experimental conditions with the exception of bensulfuron and

Figure 6. Total selected ion current chromatogram (TIC) of 1 ppb levels of SUs in soil. The upper chromatogram (A) was obtained using one tuning period. and the lower chromatogram (B) was obtained using four different tuning periods for the entire SRM LC/MS analysis.

tribenuron methyl. Nicosulfuron-d6 has the same retention time as nicosulfuron, as expected for a stable isotope-enriched compound. Therefore, the chromatographic retention times for six of the eight SUs differ, so that chromatographic retention time may be used as an additional element of qualitative identification for these compounds. In the case of the two coeluting compounds, bensulfuron methyl and tribenuron methyl, both their molecular weights and their product ion mass spectra differ significantly (see Table 1). Because of the major differences in their mass spectral behavior under these ion spray SRM LC/MS conditions, these two compounds may be readily quantified under these experimental conditions, even though they coelute chromatographically. The rather conservative LLQ data shown in Figure 5 were obtained by monitoring the most abundant product ion for each test article. This will provide the optimum sensitivity and thus the lowest level of quantitation for quantitative purposes. Figure 6 shows a comparison and alternative format for the presentation of data shown in Figure 5. Figure 6A shows a total selected ion current profile for all SRM transitions monitored at the 1 ppb level of SUs in soil using just one tuning period throughout the SRM LC/MS run. Figure 6B shows the corresponding total selected ion current profile when four different optimized tuning periods are used for SRM LC/MS acquisition. In this presentation format, each chromatographic peak consists of the summed ion current for each selected transition monitored for that compound. For example, from Figure 3 and Table 3, it is shown that metsulfuron methyl was detected (Figure 6) by monitoring two different fragmentation transitions, e.g., m/z 382 f 167 and m/z 382 f 199. Therefore, chromatographic peak 4 in Figure 6 is composed of the sum of the ion current from these Analytical Chemistry, Vol. 68, No. 19, October 1, 1996

3401

Table 3. Reproducibility of Relative Abundances Observed for Each Precursor Ion-Product Ion Transition from 10 Replicate On-Line SRM LC/MS Analyses of SU Soil Extractsa

sulfonylurea nicosulfuron thifensulfuron methyl metsulfuron methyl sulfometuron methyl chlorsulfuron bensulfuron methyl tribenuron methyl chlorimuron ethyl

precursor ion (m/z) f product ion (m/z) transitions

abundances of product ion relative to the base peak (%)

% RSDb (or % CV)

411 f 182 411 f 213 411 f 366 388 f 167 388 f 141 382 f 167 382 f 199 365 f 150 365 f 199 358 f 141 358 f 167 411 f 149 411 f 182 411 f 213 396 f 155 396 f 181 415 f 186 415 f 213

100 25 5 100 8 100 9 100 7 100 86 97 97 8 100 43 100 18

0 13 35 0 17 0 16 0 14 0 9 4 6 11 0 17 0 16

a The results were obtained using one tuning period while measuring either two or three product ions. b %RSD (relative standard deviation) is calculated by dividing the standard deviation obtained from the 10 replicate analyses by the mean and converting to percent.

two different transitions. As expected, the chromatographic integrity for all eight components is maintained, and the retention time for each component remains consistent with the data shown in Figure 5. Reference to the ion counts shown on the upper righthand side of Figure 6 shows that the single tuning period produced 2300 counts (Figure 6A), whereas the use of the individual tuning period shown in Figure 6B produced 13 340 counts. This is a 5.8-fold increase in ion current response as a result of using the four individual tuning periods. The 20-fold sensitivity improvement noted above resulted from SRM LC/MS monitoring of only the most abundant product ion for each compound. A representative calibration curve for the range from 0.05 to 10 ppb levels of nicosulfuron in soil is shown in Figure 7. Each level was analyzed in duplicate, and the resulting standard curve is linear through this limited range. In each case, the transition m/z 417 f 188 for nicosulfuron-d6 was monitored for the deuterium labeled internal standard, while the peak area for this chromatographic component was measured relative to the peak area of the other SUs studied in this work. The duplicate points on the calibration curve shown in Figure 7 were obtained by analyzing the extracts from the first set of fortified soil standards in order of their increasing concentration at the beginning of the autosampler tray, followed by analyzing the second set of standards in the same way at the end of the autosampler tray. The calibration curve shown in Figure 7 represents the mean of these two determinations and demonstrates that there are no significant analytical problems occurring during the course of the tray analysis. Qualitative Determination. In some analyses, a reliable qualitative determination for compounds of interest is also very important.18 Mass spectrometry increasingly is the method of choice for such qualitative confirmation situations. Although a full-scan mass spectrum may be preferred, it is generally accepted 3402 Analytical Chemistry, Vol. 68, No. 19, October 1, 1996

Figure 7. Representative calibration curve for nicosulfuron in soil. The calibration curve was obtained by the determination of six levels in duplicate to produce a calibration curve that ranged from 0.05 to 10 ppb.

that a limited mass scan or even monitoring of selected ions from a compound’s mass spectrum may be sufficient to provide confirmatory evidence for the presence of a target compound.19 This strategy is well known in the field of GC/MS using electron or chemical ionization conditions.20 It is generally accepted that confirmation under selected ion monitoring (SIM) GC/MS conditions requires that one monitor at least two or preferably three noncontiguous ions and that their relative abundances be within (10-20% of the absolute relative abundance obtained for a reference standard of the compound analyzed under the same conditions.21 In addition, the GC retention time should be within a certain percentage of a reference standard of that compound analyzed under the same experimental conditions.22 There is less guidance or precedence to date for qualitative criteria to be used in the corresponding LC/MS/MS experiments as presented in this work. It would seem, however, that similar criteria should apply to LC/MS/MS and GC/MS techniques. In both instances, one has a chromatographic retention time and an ion representative of the molecular weight and its fragments. It should be noted, however, that capillary GC provides much higher separation efficiency than HPLC techniques and that electron ionization (EI) can often provide more fragment ions than MS/ MS techniques. In each case, however, the data appear to be of comparable significance and usefulness. Therefore, it seems reasonable to suggest that, in SRM LC/MS experiments that are carried out to provide confirmatory support for the forensic identification of targeted compounds, comparable criteria should be applied: e.g., HPLC chromatographic retention time should agree within 2% of the retention time for a reference standard of the same compound analyzed under the same conditions,23 and the MS/MS mass spectral data should agree with those of a reference standard of the same compound analyzed under the same conditions. Relative abundance differences of product ions (18) Doerge, D. R.; Miles, C. J. Anal. Chem. 1991, 63, 1999-2001. (19) Watson, J. T. Introduction to Mass Spectrometry, 2nd ed.; Raven Press: New York, 1985, p. 314-321. (20) Garland, W. A.; Powell, M. L. J. Chromatogr. Sci. 1981, 19, 392-434. (21) Sphon, J. A. J. Assoc. Off. Anal. Chem. 1978, 61, 1247-1253. (22) Covey, T.; Maylin, G.; Henion, J. J. Biomed. Mass Spectrom. 1985, 12, 274287. (23) Henion, J. D., Cornell University, Ithaca, NY, unpublished results. In general, we find that LC/MS retention times may be reproduced to (0.2 min for an analytical run of 10 min.

Table 4. Retention Time Reproducibility from the On-Line SRM LC/MS Analyses of the Eight Targeted SUs in Soil Using Four Different Tuning Periods for the Entire Analysis in 20 Replicates sulfonylurea

retention time (min)

% RSDa (or % CV)

nicosulfuron thifensulfuron methyl metsulfuron methyl sulfometuron methyl chlorsulfuron bensulfuron methyl tribenuron methyl chlorimuron ethyl

7.8 10.0 10.8 11.5 11.9 15.0 15.0 16.8

1.3 0.7 0.7 0.7 0.7 0.4 0.4 0.2

a %RSD (relative standard deviation) is calculated by dividing the standard deviation obtained from the 20 replicate analyses by the mean and converting to percent.

may also be expected to agree within 20% of the absolute abundance of the same ions in the reference standard analyzed under the same experimental conditions. Unfortunately, just as with GC/MS techniques, some analyses require sufficiently high sensitivity that satisfactory full-scan LC/ MS/MS mass spectra cannot be obtained. This is the case in the work presented here. Modern tandem quadrupole mass spectrometers are capable of reaching low ppb levels by monitoring selected transitions in the SRM mode.25 This mode of acquisition parallels the well-known SIM experiments often used with GC/MS systems.20 We suggest that modern triple-quadrupole as well as magnetic sector mass spectrometers are capable of providing the same precision, accuracy, and reproducibility afforded by single-quadrupole systems that are known to provide reliable SIM results. Therefore, we have chosen to demonstrate by the examples presented in this work that criteria for SRM LC/ MS experiments may parallel the criteria adopted for GC/MS. These criteria include (1) HPLC retention times for the target compound(s) with reproducibility within at least 2% and (2) when available, at least two and preferably three independent precursor-product ion transitions whose absolute relative abundances agree within (20%. It should be noted there may be some degree of difficulty in routinely obtaining high precision and accuracy when measuring weaker ion current signals often obtained on less abundant ions at very low levels of analyte. The above criteria are met by the analyses presented in this report. Table 4 shows the experimental retention times obtained from 20 replicate on-line SRM LC/MS analyses of the target SUs in soil extracts. As can be seen, the gradient HPLC retention times for each analyte are well within the (2% criteria and are often within (0.2 min retention time in this ∼17 min run. Typically, we see retention time variation well within (1.0%.23 It should be noted that isocratic HPLC analyses are likely to provide even better retention time precision, especially if the column temperature is maintained constant via a column heater accessory. Therefore, just as with capillary GC/MS criteria for confirmation of target compounds, SRM LC/MS analyses can provide reliable and reproducible retention time data to complement the qualitative mass spectrometric data. MS/MS mass spectra provide additional definitive support for the confirmation of a target compound when combined with (24) Mule´ S. J.; Casella, G. A. J. Anal. Toxicol. 1988, 12, 153-157. (25) Edlund, P. O.; Bowers, L.; Henion, J. J. Chromatogr. 1989, 497, 49-57.

comparisons of standards and their corresponding HPLC retention times. If a full-scan spectrum is not possible due to the very low levels of the target compounds, then SRM LC/MS experiments may be used. These criteria follow from the now well-accepted capillary SIM GC/MS policies practiced in forensic urine drug testing,24 bioanalytical analyses in the pharmaceutical industry,25 and environmental analyses.26 The product ions selected for each of the eight SUs were chosen from the full-scan MS/MS mass spectra shown in Figure 3. These data were obtained by infusion of individual solutions of each SU dissolved in 100% acetonitrile (125 ng/mL each). Alternatively, the same data could have been obtained from an on-line full-scan LC/MS/MS experiment. From inspection of Table 1 and Figure 3, it can be seen that one may select as many as four different product ions for tribenuron methyl, but only two suitable product ions are available from sulfometuron methyl, chlorsulfuron, and chlorimuron methyl. Sometimes the structure of a compound is such that its MS/MS mass spectrum is relatively simple. In these cases, one must either resort to monitoring just those ions observed or possibly prepare a derivative or select an alternative mode of ionization to obtain additional fragment ions. The weakest transition displayed by any compound and used for qualitative purposes will determine the lowest level of detection for that compound. Usually one requires a signal-to-noise ratio of at least 3:1 from a chromatographic peak for inclusion in a weak transition for qualitative purposes. From Table 1, it can be seen that at least two different precursorproduct ions for each of the target analytes may be monitored, thus allowing one to conform to the above-suggested criteria in this example. The results presented here were obtained from actual on-line gradient SRM LC/MS determinations of SUs in soil and suggest that all the above-stated relative abundance criteria may be met (Table 3). These data include the relative abundances for each measured precursor ion-product ion transition corresponding to each of the eight SU compounds. The third column in Table 3 shows the average relative abundances of 10 replicate SRM LC/ MS analyses of each SU product ion. The fourth column in Table 3 presents the relative standard deviations (%RSD or %CV) of the 10 replicate SRM LC/MS analyses of SUs measured in soil extracts. When all these criteria for qualitative determination have been met, one can have high confidence in the qualitative confirmation of a test article in an unknown sample. Figure 8 shows the individual SRM LC/MS ion current profiles for the m/z 382 f 167 and m/z 382 f 199 precursor ion-product ion transitions for metsulfuron methyl. These transitions are unique to metsulfuron methyl in this sample mixture of targeted sulfonylurea compounds. The relative abundance ratio of ∼10:1 observed between m/z 167 and 199 in Figure 3 is reflected by the ion counts of 947 and 87, respectively, for the corresponding transitions shown in Figure 8. In addition, the identical chromatographic retention times observed in Figure 8 for both of these transitions support the common origin of both product ions. These data provide the ultimate in high-sensitivity and highspecificity qualitative determination of targeted compounds in complex samples. Of course, if additional unique product ions are present in a MS/MS mass spectrum, such as observed in this work for tribenuron methyl (Figure 3), these additional product ions may be monitored by SRM LC/MS to provide further qualitative support for a compound’s identity. (26) Levsen, K. Org. Mass Spectrom. 1988, 23, 406-410.

Analytical Chemistry, Vol. 68, No. 19, October 1, 1996

3403

Table 5. Reproducibility of Relative Abundances Observed for Each Precursor Ion-Product Ion Transition from Five Replicate On-Line SRM LC/MS Analyses of SU Soil Extractsa

sulfonylurea

precursor ion (m/z) f product ions (m/z) transition

abundances of product ion relative to the base peak (%)

%RSDb

nicosulfuron thifensulfuron methyl metsulfuron methyl sulfometuron methyl chlorsulfuron bensulfuron methyl tribenuron methyl chlorimuron ethyl

411 f 182 388 f 167 382 f 167 365 f 150 358 f 141 411 f 149 396 f 155 415 f 186

6 29 44 100 8 41 36 27

7 7 4 0 10 5 12 12

a The results were obtained using four different, optimized individual tuning periods while measuring the most abundant precursor ionproduct ion transition. b Mean of five replicate SRM LC/MS analyses.

Figure 8. Selected ion current chromatograms for 1 ppb level of metsulfuron methyl in soil. The chromatograms were obtained by monitoring two different product ions using one tuning period for the entire analysis.

From these results, it is apparent that it is possible to produce SRM LC/MS relative abundance ratios consistent with the guidelines suggested above. The results shown in Tables 3 and 5 give an indication of the reproducibility of the precursorproduction ion relative abundances measured under SRM LC/ MS conditions. The results shown in Table 3 were obtained using one tuning period while measuring either two or three product ions. Ten replicate analyses were acquired, and the %RSD was calculated. The results shown in Table 5 were obtained using four different, optimized tuning periods while measuring the most abundant precursor-product ion transition. Five replicate analyses were acquired, and the %RSD was calculated. The SRM LC/MS analyses carried out using just one tuning period, which most analysts use (Table 3), indicate that relative abundances for the SUs that are greater than 50% show reproducibility within 10% (e.g., chlorsulfuron, m/z 358 f 167; bensulfuron methyl, m/z 411 f 149, and m/z 411 f 182). When the relative abundances for the SU product ions are between 7 and 50%, the reproducibility falls to within 20% (e.g., nicosulfuron, m/z 411f 213; thifensulfuron methyl, m/z 388 f 141; metsulfuron methyl, m/z 382 f 199; sulfometuron methyl, m/z 365 f 199; bensulfuron methyl, m/z 411 f 213; tribenuron methyl, m/z 396 f 181; chlorimuron methyl: m/z 415 f 213). The product ion from nicosulfuron (m/z 411f 366) displays a weak (5%) relative abundance ion, which predictably has the largest CV (35%) due to the inherent error in measuring such a weak ion abundance. It should be recognized that the lowest level of detection of an SRM LC/MS method will be dictated by the relative abundance of the weakest precursor-product ion transition monitored. Since a minimum S/N ratio of 3:1 is often considered important for reliable qualitative analysis criteria, one must ensure that a defined minimum S/N ratio is met for the weakest transition in a group of qualitatively relevant selected precursor-product ion transi-

3404 Analytical Chemistry, Vol. 68, No. 19, October 1, 1996

tions. These criteria will usually dictate how many different transitions may be monitored and/or the level of detection that may be achieved for confirmation of target test articles in a sample. We prefer individually optimized tuning periods for both qualitative and quantitative LC/MS/MS determinations because this approach can often provide the optimum sensitivity and reproducibility for both of these determinations. CONCLUSIONS Liquid-liquid extraction of soil coupled with SRM LC/MS analysis of the soil extracts provides an ultratrace method for definitive qualitative and quantitative determination of the eight selected SUs studied in this report. A conservative LLQ for the SUs using the described method is 0.05 ppb, and the demonstrated ULQ is 10 ppb using the described SRM LC/MS method. These results suggest the applicability of ion spray LC/MS/MS techniques for the ultratrace determination of these and presumably other environmentally important compounds. SRM LC/MS acceptance criteria are suggested for the qualitative identification of these and other compounds at ultratrace levels. It would appear from the results presented that generally accepted qualitative criteria for SIM GC/MS techniques can be met by SRM LC/MS techniques. These include gradient HPLC retention time reproducibility to within (2%, the selection of at least two and preferably three noncontiguous precursor ionproduct ion transitions, and reproduction of the selected precursor ion-product ion relative abundances to within (20% of absolute for each target analyte relative to a standard of that analyte analyzed under the same experimental conditions. We suggest that these qualitative analysis acceptance criteria are conservative and that a properly tuned and calibrated tandem quadrupole mass spectrometer system is capable of meeting these qualitative criteria in most instances. Received for review May 3, 1996. Accepted July 3, 1996.X AC960375W X

Abstract published in Advance ACS Abstracts, August 15, 1996.