Automated Screening of 600 Pesticides in Food by LC/TOF MS Using a Molecular-Feature Database Search Application Food Safety Authors E. Michael Thurman and Imma Ferrer Pesticide Residue Research Group University of Almería 04120 Almería, Spain Jerry A. Zweigenbaum Agilent Technologies, Inc. 2850 Centerville Road Wilmington, DE 19808-1610 USA Abstract Searching a database using a molecular feature (MF) algorithm was developed for the screening of 600 pesticides and degradates in extracts of food by liquid chromatography time-of-flight mass spectrometry in positive ion mode with full-scan accurate mass spectra. The database search works by compiling the accurate mass of the ions detected and identified as real compound chromatographic peaks without ion extraction and compares them with the monoisotopic exact masses of the compounds in the database. The screening criteria consisted of ± 5 ppm accurate mass window, ± 0.2 minute retention time window, and a minimum area count of 1,000 counts (signal-to-noise ratio of ~10:1). The limit of detection and retention time was determined for 100 of the 600 compounds and varied from <0.01 mg/kg for 34% of the compounds to <0.5 mg/kg for 95% of the compounds. Strengths of the MF algorithm include rapid screening of hundreds of compounds at sensitive levels in minutes compared to a manual approach of hours to days. Introduction Contamination of foodstuffs with pesticides always means a risk to the consumer. Fetuses, infants, and children, a particularly sensitive group, are currently protected by a limit of 0.01 mg/kg in baby food by European Union legislation [1]. Maximum residue limits (MRLs) should be set at the lowest possible level with the aim of setting products on the markets without any measurable residues. However, this is often difficult because of the use of fungicides for transport and storage of fruit and vegetables, as well as the use of insecticides for crop protection. The analysis of pesticides in baby food has involved both GC/MS and LC/MS methods. Classically, GC/MS has been used to look for volatile pesticides, especially the chlorinated insecticides and herbicides, and LC/MS has been used to monitor the more polar insecticides, herbicides, and fungicides. GC/MS methods rely on both full scan and selected ion monitoring. Recently the use of reverse search methods for GC/MS has made it possible to search large NIST pesticide libraries in minutes [2] and has made screening quite simple for pesticides amenable to GC/MS methods.
Unfortunately, similar reverse-search methods have not been available for LC/MS for two reasons. First, the single quadrupole and triple quadrupole methods do not operate in full scan mode for pesticide screening because of a lack of sensitivity. Second, standardization and reproducibility of CID fragmentation energy for broad usage has been difficult, making common spectral libraries unavailable. Recently, the use of LC/TOF MS has shown that it has the capability and sensitivity to obtain full scan spectra with accurate masses of less than 1 ppm [3] suitable for database searching [4]. Experimental Vegetable and Fruit Extraction (QuEChERS) QuEChERS is the acronym for the extractionmethod, which stands for quick, easy, cheap, effective, rugged, and safe. It is a method that is widely receiving acceptance for rapid extraction of pesticides in food [5, 6]. Weigh 15 g of a previously homogenized sample into a 40-mL Teflon centrifuge tube. Add 15 ml of acetonitrile (containing 1% acetic acid), add 6 g of anhydrous MgSO 4 and 2.5 g of NaAc63H 2 O (sodium acetate trihydrate) and shake the sample vigorously for 1 min by using a vortex mixer at maximum speed or hand shaking. Centrifuge for 3 minutes at 3,700 rpm. Take 5 ml of supernatant into a 15-mL tube, add 250 mg of PSA adsorbent and 750 mg of MgSO 4, and vortex and shake for 20 s. Then centrifuge again for 3 min at 3700 rpm. Transfer 1.0 ml into LC/MS vial. The 1.0 ml supernatant is then evaporated to dryness and brought back up in 8/92% methanol/water for LC/MSD TOF and ion trap analysis. Direct analysis of the fruit and vegetable extracts were analyzed by injecting 50 µl. Nonfortified samples were analyzed directly at this same point by LC/MS TOF. LC/MS TOF Methods LC pumps: Agilent 1100 binary pumps, injection volume 50 µl with standard Agilent 1100 ALS Column: ZORBAX Eclipse XDB 4.6 150 mm C-8, 5-micron (p/n 993967-906) Mobile Phase A = 0.1% formic acid in water, and B = acetonitrile, gradient began with 5 minutes isocratic at 10% B followed by a linear gradient to 100% B in 30 minutes at a flow rate of 0.6 ml/min Agilent 6210 LC/MS TOF with dual spray electrospray source Positive ESI, Capillary 4000 V Nebulizer 40 psig, drying gas 9 L/min, gas temp 300 C Fragmentor 190 V, skimmer 60 V, Oct DC1 37.5 V, OCT RF V 250 V Reference Masses: mass range (m/z) 121.0509 and 922.0098, resolution: 9500 ± 500 at m/z 922.0098, m/z 50 to 1,000, Reference A Sprayer 2 is constant flow rate during the run Results and Discussion Molecular Feature Database Search Theoretical monoisotopic exact masses of compounds based on their molecular formula were calculated using an Excel spreadsheet and put into csv (comma separated values) format for use by the TOF software of the Agilent LC/TOF MS system for 600 pesticides known to ionize by positive ion electrospray. The csv file is created in the TOF software by an Excel spreadsheet tool called Formula DB Generator. The csv file is then searched automatically by the LC/TOF MS instrument at the completion of the sample run and a report generated on compounds that were found in the database. Search criteria include ppm mass tolerance (5 ppm), retention time window (0.2 minutes) if available, and minimum peak height count, which is called the compound threshold (1,000 counts or a signal-to-noise ratio of ~ 10:1 or 0.06% relative volume), and adducts and neutral fragment losses. The search routine is called a molecular feature extractor and is software recently available on the Agilent LC/TOF MS (November 2005). The molecular feature extractor finds all ions in an LC/MS TOF data file representing ions of real compounds in the sample analyzed. Noise and other extraneous ions are excluded. The resulting list of ions are then searched against the csv database using the chosen criteria and ions found are then tabulated from the full scan spectrum and checked for accuracy and retention time against the database. The molecular feature approach is more suited to large libraries because of the ease of operation and the quickness for which the search is done. Thus, 2
each ion of interest in the database is not extracted from the sample file as in a reverse search. This procedure requires much more time with LC/TOF MS data files collected in profile mode. From this point, confirmation is carried out manually by checking the positive screens for retention time match and fragment ions (if present). The sample may be reanalyzed at a higher fragmentor voltage to check for fragment ions and confirmed by authentic standard analysis. Limit of Detection The limit of detection (LOD) was determined in several matrices, including spiked food samples and solvent extracts for 100 compounds (Table 1). These compounds consist of the major classes of pesticides that are commonly used in the United States and Europe for treating crops of fruits and vegetables. The limit of detection was based on an accurate mass of less than 3 ppm and the appearance of the correct accurate mass of A+1 and A+2 isotopic signatures. The LOD was determined at various levels for pesticide work, including the EU regulation of 0.01 mg/kg for baby food, and 0.05, 0.1, and 0.5 mg/kg for various food levels, depending on pesticide and crop type. The LOD was equal to or less than 0.01 ppm for 33 compounds and less than or equal to 0.05 for 60 compounds (60%). The 0.05 mg/kg LOD is also a critical one for food monitoring of banned substances or controlled compounds. The LOD for 95% of the compounds was equal to or less than 0.2 mg/kg for food. Only six compounds were found to be insensitive with LODs of 0.5 mg/kg for food. The insensitive compounds were promecarb and aldicarb, which are two carbamate pesticides that fragment easily in the electrospray source and give low abundance for the MH + ion. Likewise, malathion oxon and dimethoate are two organophosphate pesticides that fragment easily. Thus, these compounds could be more sensitively detected by the use of the more abundant fragment ion rather than the MH +. For example, Figure 1 shows the mass spectrum of dimethoate. The MH + ion is not the major fragment ion of the spectrum, and in fact has an intensity about three to four times less than the m/z 124.9819 ion. Furthermore, it must be taken into consideration that the LOD is affected by the matrix for two reasons. One is suppression of ionization and the second is interfering ions of nearly identical mass. Suppression of ionization has been tested in our previous work for some of these pesticides in food [7], including pepper, broccoli and tomato, melon, orange, and lemon, so we have experience on which matrices are most difficult. For example, Figure 2 shows the matrix chromatogram for pesticide-free pepper, which gives a complicated chromatogram. The MF database search identified ~3,000 compounds of which none were pesticides. These peaks contained signal-to-noise ratio of 10:1 or greater and present an extremely difficult matrix for which to find ions, especially at trace levels. Spiked food extracts of these difficult matrices were used to establish LOD for the compounds shown in Table 1. An example printout is shown in Figure 3. The report contains formula, compound, SH H 3 CO P S H N CH 3 OCH 3 C 5 H 13 NO 3 PS 2 + Exact mass: 230.0069 O Figure 1. Dimethoate mass spectrum showing low intensity of MH + ion and importance of using characteristic fragment ions to lower LODs on some compounds of low intensity, specifically, m/z 124.9819 ion. 3
Figure 2. Blank pepper sample showing complexity of the sample ~3000 accurate mass peaks were detected in this sample at signal to noise of 10:1 or greater. accurate mass of the neutral molecule, error in mda and error in ppm, retention time error in minutes, and a description (specifically, fungicide). The mass spectrum of the MH + and isotope signature of the compound are also shown, which is useful for a quick check and partial confirmation of the formula, especially since most pesticides show an interesting A+2 ion from a halogen or sulfur atom. A+2 Ions and Empirical-Formula Confirmation Because the MF database search is a screening program, a formula is not unequivocally confirmed at this point with a mass accuracy window of 5-ppm. The molecular formula (not molecular identification) may be confirmed by the A+2 isotopic signature of the compound. For example, 70% of the 600-compound database contains either S, Cl, or Br, which give the isotopic cluster of the A+2 ions. The accurate mass of the isotope along with the intensity profile can then be used as a first level confirmation of the empirical formula. Thus, this adds a great deal of confidence to the screening data of the accurate mass but of course does not yet meet the standards for identification of the molecule, a topic discussed later. For example, let us examine the isotopic signature of imazalil in a pear extract (Figure 4). The measured mass of the MH + was 297.0564 and the chlorine 37 isotope was 299.0533. Thus the difference in mass is 1.997 mass units, which is the mass defect of a chlorine 37 atom relative to the chlorine 35 atom that has been replaced. Furthermore, the intensity of the A+2 peak is about 2/3 of the A peak, which is consistent with two chlorine atoms in the molecule. This is further established by the presence of an A+4 peak at 301.0503. Thus, these 4
Table 1. Limits of Detection for Pesticides in Food Samples with Retention Time and Accurate Mass Accurate mass LOD Compound Ret time Formula molecule food mg/kg Atrazine 21.1 C 8H 14N 5Cl 215.0938 0.005 Azoxystrobin 24.0 C 22H 17N 3O 5 403.1168 0.005 Benalaxyl 26.8 C 20H 23NO 3 325.1678 0.005 Buprofezin 27.2 C 16H 23N 3OS 305.1562 0.005 Cyanazine 22.0 C 9H 13N 6Cl 240.0890 0.005 Diazinon 27.6 C 12H 21N 2O 3PS 304.1010 0.005 Difenconazole 26.4 C 19H 17Cl 2N 3O 3 405.0647 0.005 Isomer 26.6 C 19H 17Cl 2N 3O 3 405.0647 0.005 Difenoxuron 21.3 C 16H 18N 2O 3 286.1317 0.005 Dimethomorph 22.2 C 21H 22NO 4Cl 387.1237 0.005 Fenamiphos 23.9 C 13H 22NO 3PS 303.1058 0.005 Imazalil 18.0 C 14H 14N 2OCl 2 296.0483 0.005 Imazapyr 20.0 C 13H 15N 3O 3 261.1113 0.005 Imazaquin 20.0 C 17H 17N 3O 3 311.1270 0.005 Irgarol 21.2 C 11H 19N 5S 253.1361 0.005 Irgarol metabolite 17.0 C 8H 15N 5S 213.1048 0.005 Isoproturon 21.3 C 12H 18N 2O 206.1419 0.005 Mebendazole 18.2 C 16H 13N 3O 3 295.0957 0.005 Metolachlor 25.6 C 15H 22NO 2Cl 283.1339 0.005 Metribuzin 15.0 C 8H 14N 4OS 214.0888 0.005 Nicosulfuron 17.0 C 15H 18N 6O 6S 410.1009 0.005 Prochloraz 23.0 C 15H 16Cl 3N 3O 2 375.0308 0.005 Prometon 16.6 C 10H 19N 5O 225.1590 0.005 Prometryn 19.0 C 10H 19N 5S 241.1361 0.005 Propazine 23.0 C 9H 16N 5Cl 229.1094 0.005 Propiconazole 25.9 C 15H 17CI 2N 3O 2 341.0698 0.005 Isomer 26.1 C 15H 17CI 2N 3O 2 341.0698 0.005 Simazine 18.8 C 7H 12N 5Cl 201.0781 0.005 Spinosyn A 20.9 C 41H 65NO 10 731.4608 0.005 Spinosyn D 21.9 C 42H 67NO 10 745.4765 0.005 Spiroxamine 19.6 C 18H 35NO 2 297.2668 0.005 Isomer 19.7 C 18H 35NO 2 297.2668 0.005 Terbuthylazine 23.4 C 9H 16N 5Cl 229.1094 0.005 Terbutyrn 20.4 C 10H 19N 5S 241.1361 0.005 Triflumazole 25.9 C 15H 15ClF 3N 3O 345.0856 0.005 Acetamiprid 16.3 C 10H 11N 4Cl 222.0672 0.01 Acetochlor 23.0 C 14H 20NO 2Cl 269.1183 0.01 Alachlor 23.0 C 14H 2ONO 2Cl 269.1183 0.01 Bensultap 21.4 C 17H 21NO 4S 4 431.0353 0.01 Bromuconazole 23.8 C 13H 12N 3OCl 2Br 374.9541 0.01 Carbaryl 21.3 C 12H 11NO 2 201.0790 0.01 Carbendazim 6.2 C 9H 9N 3O 2 191.0695 0.01 Carbofuran 20.4 C 12H 15NO 3 221.1052 0.01 Cartap 3.1 C 7H 15N 3O 2S 2 237.0606 0.01 Chlorfenvinphos 26.5 C 12H 14Cl 3O 4P 357.9695 0.01 Cyproconazole 23.4 C 15H 18N 3OCl 291.1138 0.01 Cyromazine 2.9 C 6H 10N 6 166.0967 0.01 Deethylatrazine 15.3 C 6H 10N 5Cl 187.0625 0.01 Deisopropylatrazine 12.1 C 5H 8N 5Cl 173.0468 0.01 Dichlorvos 20.0 C 4H 7Cl 2O 4P 219.9459 0.01 Dimethenamide 24.0 C 12H 18NO 2SCl 275.0747 0.01 Dimethoate 16.3 C 5H 12NO 3PS 2 228.9996 0.01 Diuron 21.0 C 9H 10N 2OCl 2 232.0170 0.01 5
Table 1. Limits of Detection for Pesticides in Food with Retention Time and Accurate Mass (Continued) Accurate mass LOD Compound Ret time Formula molecule food mg/kg Ethiofencarb 21.8 C 11H 15NO 2S 225.0823 0.01 Fenuron 15.7 C 9H 12N 2O 164.0950 0.01 Imidacloprid 15.7 C 9H 10N 5O 2Cl 255.0523 0.01 Lenacil 19.2 C 13H 18N 2O 2 234.1368 0.01 Malathion 1 22.5 C 10H 19O 6PS 2 330.0361 0.01 Malathion 2 16.9 C 10H 19O 6PS 2 330.0361 0.01 Metalaxyl 21.2 C 15H 21NO 4 279.1471 0.01 Methiocarb 23.5 C 11H 15NO 2S 225.0823 0.01 Methomyl 12.1 C 5H 10N 2O 2S 162.0463 0.01 Monuron 18.7 C 9H 11ClN 2O 198.0560 0.01 Nitenpyram 11.9 C 11H 15CIN 4O 2 270.0884 0.01 Oxadixyl 19.1 C 14H 18N 2O 4 278.1267 0.01 Profenfos 28.6 C 11H 15BrCIO 3PS 371.9351 0.01 Promecarb 24.0 C 12H 17NO 2 205.1341 0.01 Propachlor 25.6 C 11H 14NOCl 211.0764 0.01 Prosulfocarb 29.0 C 14H 21NOS 251.1344 0.01 Thiabendazole 3.7 C 10H 7N 3S 201.0361 0.01 Thiacloprid 17.7 C 10H 9N 4SCl 252.0236 0.01 Thiocyclam 4.5 C 5H 11NS 3 181.0054 0.01 Aldicarb 18.5 C 7H 14N 2O 2S 190.0776 0.05 Aldicarb sulfoxide 6.0 C 7H 14N 2O 3S 206.0725 0.05 Bendiocarb 20.6 C 11H 13NO 4 223.0845 0.05 Chlorotoluron 20.4 C 10H 13N 2OCl 212.0716 0.05 Flufenacet 25.0 C 14H 13N 3O 2SF 4 363.0665 0.05 Hydroxyatrazine 11.5 C 8H 15N 5O 197.1277 0.05 Lufenuron 28.6 C 17H 8N 2O 3Cl 2F 8 509.9784 0.05 Metamytron 14.0 C 10H 10N 4O 202.0855 0.05 Methidathion 24.1 C 6H 11N 2O 4PS 3 301.9619 0.05 Methiocarb sulfone 17.4 C 11H 15NO 4S 257.0722 0.05 Molinate 24.8 C 9H 17NOS 187.1031 0.05 Parathion ethyl 27.3 C 10H 14NO 5PS 291.0330 0.05 Propanil 17.0 C 9H 9NOCl 2 217.0061 0.05 Triclocarban 27.5 C 13H 9CI 3N 2O 313.9780 0.05 Aldicarb sulfone 11.4 C 7H 14N 2O 4S 222.0674 0.1 Bromacil 18.5 C 9H 13N 2O 2Br 260.0160 0.1 Bromuconazole 23.8 C 13H 12N 3OCI 2Br 374.9541 0.1 Butylate 15.0 C 11H 23NOS 217.1500 0.1 Diflubenzuron 25.0 C 14H 9N 2O 2ClF 2 310.0321 0.1 Flufenoxuron 29.2 C 21H 11N 2O 3ClF 6 488.0362 0.1 Fluroxypyr 18.8 C 7H 5N 2O 3Cl 2F 253.9661 0.1 Hexaflumuron 27.2 C 16H 8N 2O 3Cl 2F 6 459.9816 0.1 Imazalil degradate 14.6 C 11H 10N 2OCl 2 256.0170 0.1 Iprodione 25.4 C 13H 13N 3O 3Cl 2 329.0334 0.1 Pendimethalin 25.0 C 13H 19N 3O 4 281.1376 0.1 Teflubenzuron 27.6 C 14H 6N 2O 2Cl 2F 4 379.9742 0.1 Captan 24.4 C 9H 8CI 3NO 2S 298.9341 0.5 Chloropyrifos methyl 28.2 C 7H 7CI 3NO 3PS 320.8950 0.5 Spiromesifen 23.0 C 23H 30O 4 370.2144 0.5 Thiosultap 3.2 C 5H 13NO 6S 4 310.9626 0.5 6
Figure 3. Example of a report from the molecular feature database search. data are excellent for the confirmation of the molecular formula for imazalil (however not for the identification of imazalil). The mass accuracy was 0.8 mda off or 2.7 ppm error. Thus, this is an example of how one confirms the database screening of a pesticide molecular formula. The database report prints the isotopic signature and accurate masses to help for the manual screening of the database hits. When the formula does not contain an A+2 atom (that is, consists of only C, H, N, and O), then formula confirmation is not readily possible by this method. The reason being that only an A+1 peak is present and this peak is dominated by the carbon 13 signal of the molecule, which in itself is not enough for formula confirmation. Thus, for these pesticides (about 30% of the database) more data are required (specifically, retention time or fragment ion). Screening of Fruits and Vegetables Table 2 shows the results of screening six fruit and vegetable samples from a nearby grocery store (apple, pear, tomato, potato, pepper, and cucumber) and one commercial brand of olive oil for the 600 pesticides in the MF database search. The MF database search found from 617 to 2,681 accurate mass peaks in the sample chromatograms. The least complicated sample matrix was the tomato with 617 peaks, and the apple was the most complex sample with 2,681 peaks. The sensitivity of the MF database search was set at a signal-to-noise ratio of 10:1. The quantity of peaks found approximately doubles with decreasing the signal-to-noise ratio from 20:1 to 10:1. The value of 10:1 is chosen in order to obtain good values for the isotopic signature of the compound, which is the A+1 and A+2 isotope signatures with the maximum instrument sensitivity. The accuracy window of the MF database search is set at 5 ppm in order to be well within the mass accuracy of the LC/TOF MS system, which typically operates at less than 3 ppm and often at 1 to 2 ppm or approximately 0.3 mda [3]. The number of pesticides found in the 5 ppm mass window of these samples varied from 8 to 41 compounds. The only criterion to be included in this match was that the MH + ion was within 5 ppm of the database value. Thus, as an example, the pepper sample, which had 2,402 peaks, had only 41 of these peaks that met the 5-ppm accuracy window (Table 2). Of these 41 peaks only three formulae were confirmed based on the correct isotope signature and retention time match (for compounds not containing an A+2 isotope), which was checked not only in the printout of the automated database match, but also by manual confirmation of the data file. The confirmation of pesticides varied from no detections in the potato sample, one pesticide in olive oil, three pesticides in pepper and tomato, and five pesticides in the cucumber and apple. The most common compound found in the fruit and vegetable samples was imazalil, which is a postharvest fungicide used for transport and storage of fruits and vegetables before their sale. Other compounds included organophosphate insecticides, such as diazinon, phosmet, and malathion and the 7
Cl +1.997 H 2 C O 1 37 Cl A+2 N NH Cl +1.997 1 35 Cl A+4 Figure 4. Isotope cluster of the m/z 297 ion and imazalil structure in a pear extract. oxon of malathion, which is a pesticide degradate. The insect growth regulator buprofezin was found in a tomato sample, as was thiophanate methyl and carbendazim, both fungicides. The software originally used for this work did not address saturated peaks and compounds at high concentrations such as imazalil in the pear sample and buprofezin in the tomato sample were incorrectly identified. The latest software handles saturated peaks and correctly identified these compounds. The accuracy of all confirmed samples had an absolute-value average of 0.3 mda or 1.2 ppm and a standard deviation of 0.25 mda and 1.0 ppm, respectively (Table 2 ). The absolute-value average for retention time match was 0.07 minutes and standard deviation of 0.09 minutes. Thus, the windows chosen for the database are chosen with enough margin of error to find 99% of the compounds based on two standard deviations of the mean for mass accuracy and retention time. Screening of Baby-Food Samples Of the 100 tested compounds, 33 met the baby-food screening level of 0.01 mg/kg. The only compound detected by the database search was a trace level of imazalil in the puree of pear, banana, and orange. Furthermore, examination of the label shows that lemon juice was also used in the baby food, which was a possible source for the imazalil. The, concentration though was approximately 0.0005 mg/kg, which is considerably less than the levels for baby-food safety of 0.01 mg/kg. Imazalil is easily screened because it contains the characteristic A+2 chlorine signature with two chlorine atoms. The error in identification was 0.3 md or 1 ppm. None of the other compounds of the database was detected. Confirmation was not possible on this sample because of the low signal of the fragment ion and the low concentration of the suspected imazalil (0.0005 mg/kg). The sample was screened as safe, though, based on the health limit of 0.01 mg/kg for baby food. Approximately 10 different baby-food samples have been 8
Table 2. Screened Pesticides in Food Samples Using the MF Database Pesticide Retention Peaks matches Pesticides confirmed Error Error time error FALSE FALSE Sample screened < 5 ppm LC/TOF MS mda ppm (min.) negative positive Apple 2681 12 Imazalil 0 0 Imazalil degradate 1 3.9 0.08 Iprodione 0.22 0.7 0.12 Fluqinconazole 0.05 0.1 Difenoconazole 0.74 1.8 Olive oil 1678 10 Terbuthylazine 0.06 0.2 0.09 0 0 (Deisopropylatrazine) 1 Pepper 2402 41 Imazalil 0.3 1 0.07 0 Diazinon 0.11 0.3 0.12 Buprofezin 1 3.3 0.1 Tomato 617 8 Buprofezin 0 (Benzthiazuron) 1 Carbendazim Thiophanate methyl Cucumber 1619 17 Thiabendazole 0.2 1 0.01 0 0 Malathion isomer 1 0.04 0.1 0.05 Malathion isomer 2 0.22 0.7 0.03 Malathion oxon 0.25 0.8 0.08 Imazalil 0.1 0.3 0.05 Pear 1209 14 Imazalil 0 0 Carbendazim 0.21 1.1 0.05 Imazalil degradate 0.51 2 0.03 Phosmet 0.31 1 0.04 Potato 1150 11 None 0 0 Mean 0.30 1.20 Standard deviation 0.25 1.00 screened, including a variety of brand-name vegetables and fruits and, fortunately, no positive detections have been found for the pesticides in the MF database search with the exception of imazalil shown above. The baby food samples represent the most difficult samples to screen because of the low LODs required. Weakness and Strengths of MF Database Search The only weakness of the database is the loss of mass accuracy because of interferences in the matrix. This problem is easily solved by the addition of a second ion (fragment ion) or a sodium or ammonium adduct ion for added confidence from matrix interference. The use of accurate mass LC/TOF MS combined with database searching is a powerful example of a new wave of monitoring technology for identification of pesticides in food and water. The use of classical fragmentation libraries with comparison of fragmentation patterns is not needed in LC/TOF MS. Rather the use of molecular formula and the calculated accurate mass, especially when combined with one or two fragment ions of accurate mass, gives the identity of compounds without the worry of the intensity of the fragment ions and how this may vary from instrument to instrument and matrix to matrix. For example, the Molecular Hunter Software works in conjunction with the Molecular Feature Database using the.mhd files to link ions in groups according to their exact retention time (matching within 0.005 minutes). Therefore, it is possible to find and differentiate the fragment ions of a pesticide, such as imazalil, from the background ions of the matrix. Thus, it is the view of the authors that a large problem in LC/MS libraries is on the verge of being solved with the use of accurate mass 9
databases for pesticide screening in food with a molecular feature algorithmic approach using the MH + ion and a major fragment ion. References 1. PAN Europe position on the European Commission Proposal for a Regulation of the European Parliament and of the Council on maximum residue levels of pesticides in products of plant and animal origin COM(2003) 117 final, 2003/0052 (COD). 2. Phillip L. Wylie, Michael J. Szelewski, Chin-Kai Meng, and Christopher P. Sandy, Comprehensive Pesticide Screening by GC/MSD using Deconvolution Reporting Software, Agilent Technologies, publication 5989-1157EN. 3. Imma Ferrer, E. M. Thurman, 2005, Measuring the mass of an electron by LC/TOF MS: A study of twin ions, Analytical Chemistry, 77, 3394 3400. 4. E. Michael Thurman, Imma Ferrer, 2005, Identification of unknown pesticides in food using both LC/MSD TOF and ion trap MS n, Agilent Technologies, publication 5989-1924EN. 5. M. Anastassiades, S. J. Lehotay, D. Stajnbaher, and F. J. Schenck, Fast and Easy Multiresidue Method Employing Acetonitrile Extraction/Partitioning and Dispersive Solid-Phase Extraction for the Determination of Pesticide Residues in Produce, (2003) Journal of AOAC International, 86:412 431. 6. S.J. Lehotay, K. Maštovská, A.R. Lightfield, Use of Buffering and Other Means to Improve Results of Problematic Pesticides in a Fast and Easy Method for Residue Analysis of Fruits and Vegetables, (2005) Journal of AOAC International, 88:615 629. 7. Imma Ferrer, Juan F. Garcia-Reyes, M. Mezcua, E. M. Thurman, A. R. Fernandez-Alba, 2005, Multiresidue pesticide analysis in fruits and vegetables by liquid chromatography time-offlight mass spectrometry, J. Chromatography A, 1082, 81 90. www.agilent.com/chem For More Information For more information on our products and services, visit our Web site at www.agilent.com/chem. Agilent shall not be liable for errors contained herein or for incidental or consequential damages in connection with the furnishing, performance, or use of this material. Information, descriptions, and specifications in this publication are subject to change without notice. Agilent Technologies, Inc. 2006 Printed in the USA October 26, 2006 5989-5496EN