Anal. Chem. 2001, 73, 5436-5440
Exact Mass Measurements for Confirmation of Pesticides and Herbicides Determined by Liquid Chromatography/Time-of-Flight Mass Spectrometry Mila Maizels
Oak Ridge Institute for Science and Education, 26 West Martin L. King Drive, Cincinnati, Ohio 45268 William L. Budde*
Office of Research and Development, National Exposure Research Laboratory, U.S. Environmental Protection Agency, 26 West Martin L. King Drive, Cincinnati, Ohio 45268
In a recent publication,1 microbore liquid chromatography (LC) and positive ion electrospray mass spectrometry (MS) with a linear quadrupole mass spectrometer was reported as an effective technique for the determination of carbamate, urea, and thiourea
pesticides and herbicides in water. Retention time precision, peak width precision, concentration measurement precision, mean recoveries, and instrument detection limits were determined for 16 analytes in reagent water with the aid of an internal standard. The analytes were also measured in fortified environmental water samples from a recreational lake, a groundwater well, a cistern, a farm pond, and a treated drinking water. These measurements were at 5 ng/mL for each analyte, which is within the range expected for environmental pesticide and herbicide contaminants. The analytes were separated from the environmental water matrixes with an on-line extraction and concentration to provide rapid sample analyses without an off-line liquid-liquid or liquidsolid-liquid extraction and extract concentration. Recoveries of 12 of the analytes from four environmental water samples were in the range of 75-124% with relative standard deviations in the range of 11-16%. However, the analytes were identified in the total ion chromatogram by only one or a few ions and their retention times that are subject to variation with experimental conditions. Electrospray is a soft ionization process, and only one or a few ions are usually observed and these are often adducts of the molecule with a proton, a sodium ion, or an ammonium ion. Fragment ions and the rich mass spectra typical of electron ionization (EI) are generally not produced in the electrospray ionization process. Because of the limited data usually available from electrospray LC/MS, the identity of a pesticide or herbicide is less certain than it is when volatile pesticides are determined with gas chromatography and EI mass spectrometry. Additional data are usually needed to confirm a suspect substance. One option used to obtain confirmatory data is collision-induced dissociation (CID) of a (M + 1)+ or an analogous ion in the collision cell of a multiple-analyzer instrument, an ion storagetype mass spectrometer, or an electrospray or an atmospheric pressure chemical ionization ion source.2 While this strategy is generally effective and usually produces one or more fragment ions from most analytes, the operational parameters, that is,
* Corresponding author: (phone) 513-569-7309; (fax) 513-569-7757; (e-mail)
[email protected]. (1) Wang, N.; Budde, W. L. Anal. Chem. 2001, 73, 997-1006.
(2) Budde, W. L. Analytical Mass Spectrometry: Strategies for Environmental and Related Applications; Oxford University Press: New York, 2001; Chapter 5.
The accuracy and precision of exact mass measurements are determined using positive ions formed in the electrospray of 10 nonvolatile or thermally unstable carbamate, urea, and thiourea pesticides and herbicides. Environmentally significant ∼7-ng quantities of the analytes were separated with microbore liquid chromatography, and the exact mass measurements were made in real time with a benchtop time-of-flight mass spectrometer. The positive ion electrospray mass spectra of the analytes generally consist of one or a few ions which are usually adducts of the molecule with a proton, a sodium ion, or an ammonium ion. Fragment ions and the rich mass spectra typical of electron ionization (EI) are generally not produced in the soft electrospray ionization process. Confirmation of the identity of a nonvolatile pesticide or herbicide depends largely on the masses of the few ions formed and the retention time, which can vary with chromatography conditions. Identifications of these analytes in environmental or other samples are less certain than identifications of volatile pesticides determinated by gas chromatography and EI mass spectrometry. The benchtop time-of-flight mass spectrometer was equipped with an electrostatic mirror, and resolving powers of 3500-5000 were routinely obtained and used for these exact mass measurements. This type of mass spectrometer is significantly less costly and complex than other types of mass spectrometers with exact mass measurement capabilities. The mean errors from three replicate exact mass measurements of the 10 test analytes were in the range of 0-5.4 parts-per-million. Potential interferences from substances with similar exact masses were evaluated.
5436 Analytical Chemistry, Vol. 73, No. 22, November 15, 2001
10.1021/ac010601o CCC: $20.00
© 2001 American Chemical Society Published on Web 10/03/2001
collision energy and interface conditions, for optimum sensitivity and fragmentation are often analyte dependent. Therefore, a multiple-analyte LC/MS method requires real-time optimization of these parameters, which may not be available, and requires considerable time to define the appropriate operational parameters for each analyte. An alternative confirmation strategy is exact mass measurements of the ions produced by electrospray. Exact mass measurements with an accuracy of 10 parts-per-million (ppm) or better have been used for more than 40 years to determine ion compositions.2,3 Until about the mid-1990s, nearly all exact mass measurements were made with double-focusing instruments consisting of electrostatic and magnetic sectors. These spectrometers are often equipped with heated reservoir or direct probe sample introduction systems that provide a steady pressure of analyte vapor for 30 s or more for the relatively slow-scanning electromagnets or slow peak-matching procedures. Chromatographic inlet systems, particularly high-resolution gas chromatography (GC), require fast data acquisition. Exact mass measurements in GC/MS with magnetic sector instruments are usually made with selected ion monitoring (SIM), which is limited to measurements of ions from target analytes.2-4 The Fourier transform mass spectrometer (FTMS) is a very important instrument capable of exact mass measurements with several sample introduction systems including high-resolution LC and electrospray.5 A significant disadvantage of both the double-focusing sector instruments and the FTMS is that these instruments are usually priced in the $500 000-$1 500 000 range which is well beyond the means of most pesticide residue laboratories. In recent years, some benchtop time-of-flight (TOF) mass spectrometers equipped with electrostatic mirrors have become commercially available.6 These spectrometers are capable of resolving powers (m/∆m) of 3500 and greater and exact mass measurements. This type of instrument is also far smaller, less complex, and significantly less costly than double-focusing magnetic sector and FTMS spectrometers. Another advantage of the TOF spectrometer is that all ions in the mass spectrum are measured with true mass peak profiles. There are no requirements to specify target analyte ions for SIM, and potential deficiencies from peak top SIM with magnetic sector instruments are precluded.4,6 The purpose of this research was to determine whether exact mass measurements with a relatively low-cost TOF-MS during microbore LC with environmentally significant concentrations of analytes are of sufficient accuracy and precision to confirm identifications of analytes when only one or a few ions are normally present in their mass spectra. This evaluation was made using 10 test analytes that were included in the earlier research1 and a simple external mass calibration without any real-time optimization for individual analytes. EXPERIMENTAL SECTION Materials. The pesticides, herbicides, reagents, solvents, and analyte solutions were the same as those described previously.1 (3) Russell, D. H.; Edmondson, R. D. J. Mass Spectrom. 1997, 32, 263-276. (4) Gross, M. L.; Sun, T.; Lyon, P. A.; Wojinski, S. F.; Hilker, D. R.; Dupuy, A. E.; Heath, R. G. Anal. Chem. 1981, 53, 1902-1906. (5) Shen, Y.; Tolic, N.; Zhao, R.; Pasa-Tolic, L.; Li, L.; Berger, S. J.; Harkewicz, R.; Anderson, G. A.; Belov, M. E.; Smith, R. D. Anal. Chem. 2001, 73, 30113021. (6) Cotter, R. J. Anal. Chem. 1999, 71, 445A-451A.
Azobenzene, morpholine, and 3,3′-dimethylbenzidine were obtained as commercial fine chemicals and were used without purification. Liquid Chromatography. Separations were conducted with a Micro-Tech Scientific (Sunnyvale, CA) Ultra-Plus II liquid chromatograph and a Phenomenex (Torrance, CA) 1.0 mm (i.d.) × 150 mm, 3-µm C18 LUNA column. The mobile-phase gradient was hold at 65% water/35% acetonitrile for 2 min and then a linear program to 98% acetonitrile in 17 min. Both the solvents contained 0.001% (v/v) acetic acid. The injection volume was 10 µL containing ∼7 ng of each analyte, and the flow rate was 35 µL/min. At the conclusion of an LC/MS analysis, the column was flushed with 65% water/35% acetonitrile for 15 min before the next injection. Mass Spectrometer. The TOF mass spectrometer was an Applied Biosystems (Framingham, MA) Mariner benchtop instrument equipped with an electrospray ion source. The electrospray design is similar to that described by Bruins et al.7 A high voltage (HV) is applied to an inner stainless steel tube containing the mobile phase, and nitrogen nebulizing gas flows through an outer concentric tube to provide pneumatic assistance. The HV was +3500 and the nebulizing gas flow rate was 0.4 L/min. A flow of heated nitrogen was also maintained at 0.6 L/min countercurrent to the aerosol spray to promote evaporation of the solvents from the charged droplets and desolvation of the positive ions. The tip of the spray needle was located a few millimeters from an orifice leading to the mass spectrometer and was positioned for optimum ion detection and spray stability. Ions are injected orthogonal to the flight tube and push and pull electrodes accelerate the ions in rapid pulses into the flight tube. The field free flight tube is 0.6 m followed by a second-order electrostatic mirror and a return path to a microchannel ion detector, giving an effective path length of 1.3 m for the TOF analyzer. Resolving powers of 3500-5000 based on the full peak width at half peak height definition were routinely obtained and used in these experiments. The analyzer mass range was set at 10-10 000 Da, and individual spectra were accumulated over a period of 2 s each. The mass spectrometer was calibrated for exact mass measurements by flow injection using the spectrometer’s internal syringe pump and a solution containing 2 ppm of morpholine, azobenzene, and 3,3′-dimethylbenzidine (Chart 1) in 50% (v/v) acetonitrile/water. The injection flow rate was 35 µL/min, and the other conditions were the same as in the LC/MS experiments. No acid was used in the calibration solution because the compounds gave abundant (M + 1)+ ions without the addition of acid. RESULTS AND DISCUSSION Chart 1 shows the compound names, structures, and calculated exact masses of the ions used to calibrate the mass spectrometer for the exact mass measurements. These compounds were selected because they give abundant (M + 1)+ ions in the mass range expected for ions from the analytes. A solution of the three substances was infused into the electrospray source, and standard instrument vendor software was used to calibrate the instrument and store the calibration parameters for subsequent use. No (7) Bruins, A. P.; Covey, T. R.; Henion, J. D. Anal. Chem. 1987, 59, 26422646.
Analytical Chemistry, Vol. 73, No. 22, November 15, 2001
5437
Chart 1
internal calibration standards were used during the LC/MS separation and measurements of the analyte ions because of potential coelutions of the calibration compounds with the analytes and to avoid lengthening the analysis time to accommodate the elution of the calibration compounds. External calibration is simpler, and one of the objectives of this research was to determine the accuracy and precision of exact mass measurements using the simplest and lowest cost techniques. Some instruments have been designed with dual electrospray ion sources to allow intermittent introduction of calibration standards and to avoid internal calibration standards.8 While this design has some clear advantages, it also is a more costly approach that may require more intense maintenance. All spectrometer operating parameters were the same, as close as possible, during calibration and during the LC/MS measurements. All interface heaters and all other instrument heat sources were activated for several hours in a reasonably constant temperature laboratory environment to ensure instrument thermal equilibrium before calibration and before any exact mass measurements were made. The accuracy of exact mass measurements with the TOF spectrometer was dependent on comparable instrument temperatures during calibration and analyte measurements. The effect of temperature on exact mass measurements with a TOF mass spectrometer is well known.6 The compound names, abbreviations, structures, and calculated exact masses of the 10 analyte ions measured in these experiments are shown in Charts 2 and 3. All the analyte ions except one are protonated molecules with the proton shown associated with an atom that is a reasonable choice. The sodium adduct ion of aldicarb was measured and evaluated as the confirmation ion for this analyte. These ions were usually the most abundant ions in the spectra of the analytes and sometimes the only significant ion.1 A 10-µL aliquot of a solution of the analytes at a concentration of ∼0.7 ng/µL of each was injected into the LC column and the total ion chromatogram obtained is in Figure 1. Assuming a typical concentration factor of 1000 during sample preparation (1 L of sample to 1 mL of extract), the amount of each analyte injected (8) Wolff, J.-C.; Eckers, C.; Sage, A. B.; Giles, K.; Bateman, R. Anal. Chem. 2001, 73, 2605-2612.
5438
Analytical Chemistry, Vol. 73, No. 22, November 15, 2001
Chart 2
Chart 3
corresponds to a sample concentration of ∼1 µg/L, which is in the general concentration range where pesticides and herbicides are often found in aqueous environmental samples.1 Actual detection limits are in the 0.1-1 µg/L range since most of the
Figure 1. Total ion chromatogram showing the separation of the test analytes. Table 1. Largest Errors from Single Measurements of Exact Masses and Mean Errors and Relative Standard Deviations (RSDs) from Three Replicate Determinations in ∼96 min and Eight Replicate Determinations in ∼256 min 3 replicates in ∼96 min
8 replicates in ∼256 min
analytes
largest errors from single meas (ppm)
mean errors (ppm)
RSD (ppm)
mean errors (ppm)
RSD (ppm)
ADC ASX ASN MTH CBF MXC DIU LIN SID ANT
19 21 18 38 11 14 16 8.8 13 6.9
0 4.8 4.9 1.2 5.4 0.45 3.0 0.40 0 2.5
11 16 12 35 6.1 12 12 8.1 12 1.6
5.0 8.9 1.1 7.4 1.5 3.6 5.1 4.1 3.6 3.6
12 14 12 23 7.2 11 9.8 5.6 11 2.4
mean
17
2.3
13
4.4
11
peaks in Figure 1 have a signal/noise of >3. As previously reported, two isomers of SID were separated but only the first eluting was measured in this study. The second isomer elutes immediately after the first and before MXC but is not identified in Figure 1. All analytes eluted in ∼15 min. After calibration of the instrument, eight LC injections of the 10 analytes were made over two consecutive work days to ensure that the exact mass measurements would reflect normal day-today variations in experimental conditions. The amounts injected were the same as those used for Figure 1. The exact masses of the ions were measured in spectra taken near the apexes of the peaks in the total ion chromatogram. The mass measurement error is the difference between a measured exact mass and a calculated exact mass (Charts 2 and 3) expressed in ppm. Table 1 shows the largest error observed for each analyte in any single measurement from the eight analyses of the mixture of 10 analytes. These errors range from 6.9 to 38 ppm, and the mean error is 17 ppm. While many of the single measurements were very accurate, often to within 1 or 2 ppm, and two of the largest single measurement errors are 16 ppm, and this was probably caused by previously reported larger errors at the low end of the calibrated mass range.9 The analyte MTH has the smallest mass of the measured analyte ions. For comparison, the mean errors and RSDs from all eight replicate injections made in a little over 4 h over two consecutive days are also shown in Table 1. While it is probably generally impractical to acquire data from eight replicate injections during routine sample analyses, it was of interest to determine whether additional measurements would greatly improve the accuracy and precision of the results. However, there is relatively little improvement in either accuracy or precision in the eight replicate analyses compared to the three analyses. The mean errors from the eight injections range from 1.1 to 8.9 ppm and the grand mean is 4.4 ppm (Table 1). Over the longer time required to make eight injections, variations in instrumental and other conditions, for example, instrument temperature, increase measurement variabilities. The mean RSD was 11 ppm, which is just a slight improvement over the 13 ppm obtained with three measurements. Therefore, three consecutive injections made over a relatively short time are recommended to achieve acceptably accurate exact mass measurements with the LC/MS system used in this work. To determine whether this level of accuracy and precision is adequate to confirm the presence of the analytes, a standard computer program was used to find all chemically reasonable compositions whose exact masses are within an acceptance range of (10 ppm of the calculated exact masses of the analyte ions in Charts 2 and 3. The range of (10 ppm was selected because it is approximately twice the largest mean error observed with three replicate analyses (Table 1). For most analytes, the program was permitted to use up to C16H20N5O5S2, which has a calculated mass of 426 or about twice the masses of the analyte ions. Since the Na+ adduct of ADC was measured, the program was allowed to use the exact mass of a Na+ for this search. For SID the maximums were expanded to C20H30N5O5S5, and for DIU and LIN the SID maximums were used except the S5 was replaced by a maximum of five 35Cl. The results of these calculations are shown in Table 2. For the analytes CBF, MXC, SID, and ANT, the only compositions found that have exact masses within the acceptance range are the compositions of the protonated analyte molecules (Table 2). Therefore, if an ion is detected with an exact mass within the acceptance range of one of these analytes, and the retention time corresponds to that analyte, it is virtually certain that the signal is caused by that analyte. Two possible compositions were found that have exact masses within the acceptance ranges of the analyte ions of ASX, ASN, and MTH (Table 2). One of these is the correct (9) Debre´, O.; Budde, W. L.; Song, X. J. Am. Soc. Mass Spectrom. 2000, 11, 809-821.
Analytical Chemistry, Vol. 73, No. 22, November 15, 2001
5439
Table 2. Compositions of Ions Whose Exact Masses Are within (10 ppm of the Calculated Exact Masses of the Analyte Ions in Charts 2 and 3 compd
exact mass of measd ion
ADC
213.0668
ASX ASN MTH CBF MXC DIU
207.0798 223.0746 163.0535 222.1125 223.1441 233.0243
LIN
249.0192
SID ANT
233.1649 203.0638
composition of the protonated analyte molecule, and the other is either an unsaturated hydrocarbon or an oxygenated unsaturated hydrocarbon. Hydrocarbons are not ionized by electrospray, and oxygenated hydrocarbons such as alcohols, phenols, and ketones are usually not ionized or give very low abundance ions. Therefore, exact mass measurements in combination with a reasonable retention time provide a clear confirmation of the presence of ASX, ASN, and MTH in a sample. Four compositions were found whose exact masses were within the acceptance range of sodiated ADC (Table 2). One of these is the correct ion, and one is a sodiated hydrocarbon, which is easily rejected because hydrocarbons do not form Na+ adducts in electrospray spectra. However, a protonated molecule with the composition C12H9N2O2 corresponds to several compounds contained in the NIST/EPA/NIH database of electron ionization mass spectra.10 These compounds generally contain electron-withdrawing groups, for example, the four possible nitrocarbazoles, and they may not form abundant ions in positive ion electrospray, but potential interferences are indicated for sodiated ADC. The protonated molecule composition C4H13N4O4S does not correspond to any compounds in the 107 886 compound database and is an unlikely composition because it has too few carbons for the number of hydrogens and other atoms. Nevertheless, the sodiated ADC ion is not a good choice as an analyte ion because of potential interferences. The protonated ADC fragment at mass 116.0533, which is produced by the losses of a methylisocyanate and a water molecule from protonated ADC,1 would be a superior choice, even though it may be less abundant, because no other ions were found with masses within (10 ppm of its exact mass. Four and six compositions, respectively, were found whose exact masses were within the acceptance ranges of the exact masses of the protonated DIU and LIN molecules. However, all of these except the two correct ion compositions can be discarded because either they do not contain a Cl atom or they contain either one or three Cl atoms. The analytes DIU and LIN each contain two Cl atoms, which is clearly indicated in their electrospray spectra by ions at 3:2 abundance ratio at 233/235 and 249/251, respectively. Therefore, confirmation is ensured by the exact mass measurement, the retention time, and the abundance ratios of the ions caused by the presence of the naturally occurring isotopes of Cl. (10) NIST/EPA/NIH Mass Spectral Database; National Institute for Standards and Technology: Gaithersburg, MD, 1998.
5440 Analytical Chemistry, Vol. 73, No. 22, November 15, 2001
compositions within (10 ppm C7H14N2O2SNa (correct), C15H10 Na (3.2 ppm), C12H9N2O2 (4.4 ppm), C4H13N4O4S (7.5 ppm) C7H15N2O3S (correct) and C15H11O (2.9 ppm) C7H15N2O4S (correct) and C15H11O2 (3.6 ppm) C5H11N2O2S (correct) and C13H7 (4.3 ppm) C12H16NO3 (correct) C12H19N2O2 (correct) C9H11N2OCl2 (correct), C15H5O3 (4.3 ppm), C8H16OCl3 (7.9 ppm), C10H6N4OCl (7.9 ppm) C9H11N2O2Cl2 (correct), C16HN4 (1.6 ppm), C15H5O4 (4.0 ppm), C10H 6N4O2Cl (7.3 ppm), C8H16O2Cl3 (7.2 ppm), C15H6N2Cl (8.8 ppm) C14H21N2O (correct) C11H11N2S (correct)
Confirmation by exact mass measurement resolves ambiguities caused when an ion of the same nominal mass is used to identify two different analytes that could have similar or the same retention times.1 An ion of mass 223 is used to identify both ASN and MXC. While these analytes are separated by more than 11 min in the chromatogram in Figure 1, the retention time of MXC is highly variable depending on the pH of the mobile phase because MXC contains a basic amine function (Chart 2). However, the exact masses of the two analyte ions differ by 312 ppm while the measurement precision is less than 10 ppm, which allows the analytes to be easily distinguished even if they are just marginally separated in the chromatogram. Another potential source of error occurs with DIU and SID, which elute within a few minutes of each other, and ions with the same nominal mass 233 are used to distinguish these analytes. However, the exact masses of these ions differ by 609 ppm and these analytes are readily distinguished even when they are marginally separated in the chromatogram. CONCLUSION Exact mass measurements with a benchtop TOF mass spectrometer provide necessary confirmatory information for the LC/ MS determination of nonvolatile and thermally liable pesticides and herbicides that often give only one or a few ions in their electrospray spectra. These TOF measurements are relatively economical compared to alternative instruments with exact mass measuring capabilities and are an attractive alternative to fragmentation by CID, which requires optimization for each analyte in a complex sample mixture. ACKNOWLEDGMENT This research was supported in part by an appointment of M.M. to a postdoctoral research position administered by the Oak Ridge Institute for Science and Education through an interagency agreement between the U.S. Department of Energy and the U.S. Environmental Protection Agency. The authors express their appreciation to Phil Epstein of Applied Biosystems for his generous assistance in providing the Mariner system for evaluation and use in this research.
Received for review May 30, 2001. Accepted August 29, 2001. AC010601O
