Residue analysis of 500 high priority pesticides: Better by GC–MS or LC–MS/MS?

Authors


Abstract

This overview evaluates the capabilities of mass spectrometry (MS) in combination with gas chromatography (GC) and liquid chromatography (LC) for the determination of a multitude of pesticides. The selection of pesticides for this assessment is based on the status of production, the existence of regulations on maximum residue levels in food, and the frequency of residue detection. GC–MS with electron impact (EI) ionization and the combination of LC with tandem mass spectrometers (LC–MS/MS) using electrospray ionization (ESI) are identified as techniques most often applied in multi-residue methods for pesticides at present. Therefore, applicability and sensitivity obtained with GC–EI–MS and LC–ESI–MS/MS is individually compared for each of the selected pesticides. Only for one substance class only, the organochlorine pesticides, GC-MS achieves better performance. For all other classes of pesticides, the assessment shows a wider scope and better sensitivity if detection is based on LC–MS. © 2006 Wiley Periodicals, Inc.

I. INTRODUCTION

Pesticides have been widely used throughout the world since the middle of the 20th century. Based on the compilation of the British Crop Protection Council, approximately 860 active substances are formulated in pesticide products currently (Tomlin, 2003). These substances belong to more than 100 substance classes. Benzoylureas, carbamates, organophosphorous compounds, pyrethroids, sulfonylureas, or triazines are the most important groups. The chemical and physical properties of pesticides may differ considerably. There are several acidic pesticides; others are neutral or basic. Some compounds contain halogens, others phosphorous, sulfur, or nitrogen. These heteroatoms may have relevance for the detection of pesticides. A number of compounds are very volatile, but several do not evaporate at all. This diversity causes serious problems in the development of a “universal” residue analytical method, which should have the widest scope possible.

But such multi-residue methods are urgently needed. Probably, no other use of chemicals is regulated more extensively than that of pesticides. Maximum residue levels (or tolerances) have been established for pesticides in foodstuffs and drinking water in most countries to avoid any adverse impact on public health, and to insist on good agricultural practice. Residues of systemic herbicides in soil used in the previous season may influence the growing of succeeding crops. Residues of insecticides in surface water may cause adverse effects on aquatic organisms. For these reasons a large number of laboratories are involved in the surveillance of maximum residue levels or in the identification and quantification of pesticide residues in environmental matrices. In this context the use of numerous single-residue methods is usually too expensive. It has to be noted that every company which applies for registration of a new pesticide has to provide residue analytical information. At least in the EU, this part of a registration package is not confidential.

Depending on the purpose, determination of pesticide residues may be target analysis or non-target analysis. An example of target analysis is the inspection of MRLs in food. The relevant analytes are fixed by the residue definition given in the MRL regulation. These residue definitions may include relevant metabolites or degradation products of the pesticides. In contrast, the EU regulation of residues in drinking water does not contain detailed residue definitions. Furthermore, residues in soil or surface water are not regulated at all. In such cases, metabolites or degradation products may be unknown. Their detection and identification is part of the analytical task. Both types of analysis have the need for different analytical schemes and may require different instrumentation. In this review, we want to focus on the application of mass spectrometry (MS) in target analysis.

In the past decades, the methods for trace level determination of pesticides have changed considerably. Since the early 1970s most routine pesticide residue analysis has been conducted by gas chromatography (GC) in combination with electron capture, nitrogen-phosphorous, and/or flame photometric detection. Confirmation of results required the use of a further gas chromatograph equipped with a different type of column or detector. Nowadays, using GC combined with MS, simultaneous determination and confirmation of pesticide residues can be obtained with one instrument in one analytical run. In most cases, the sensitivity obtained with GC–MS is similar to that of classical GC detectors. Selectivity of GC–MS can be adjusted by the selection of appropriate molecular and fragment ions to avoid interferences from co-extracted sample materials. Therefore, the importance of GC with ECD, NPD, or FPD detection has decreased in pesticide residue laboratories.

Methods based on liquid chromatography (LC) were applied more rarely in the past, because traditional UV, diode array, and fluorescence detectors are often less selective and sensitive than GC instruments. But in the last few years, the commercial availability of atmospheric pressure ionization caused a spectacular change. Compared to traditional detectors, electrospray (ESI) or atmospheric pressure chemical ionization (APCI) in combination with MS instruments have increased the sensitivity of LC detection by several orders of magnitude. Moreover, HPLC column switching techniques and extensive sample cleanup procedures become unnecessary if tandem mass spectrometers are used and operated in the selected reaction mode (SRM) (Stout et al., 1998; Hernandez, Sancho, & Pozo, 2005). Due to the suppression of most interfering signals by LC–MS/MS in the SRM, the signal-to-noise ratio increases distinctly and the full sensitivity range of LC–MS instruments can be utilized.

The applicability of GC–MS in pesticide residue analysis is summarized in pesticide analytical manuals (Thier & Zeumer, 1992; van Zoonen, 1998), applications from instrument producers (Agilent Technologies, 1999), or scientific studies (Cairns et al., 1993; Fillion, Sauve, & Selwyn, 2000; Wong et al., 2003). Several mass spectral databases contain electron impact (EI) mass spectra of many pesticides (Ehrenstorfer, 2005; NIST/EPA/NIH, 2005). Analogous presentations of the scope of LC–MS/MS in the area of pesticide residue analysis are missing. Up to now, the largest overview has been given by Lehotay et al. (2005), who applied LC–MS/MS for the determination of 144 pesticides. But a complete inventory of all available LC–MS/MS information does not exist.

Therefore, the aim of this review is to summarize all typical precursor and product ions appropriate for LC–ESI–MS/MS determination of 500 pre-selected pesticides (if these pesticides are adequately ionized by electrospray) and the sensitivity obtained. The applicability of GC–MS with EI MS is checked for the same list of pesticides. Typical fragment ions are provided, if their determination is possible. The decision between two alternatives of quantitative determination also depends on sensitivity. For this reason, the smallest analyte concentration required for GC–MS and/or LC–MS/MS is listed in addition. Finally, the achievable scope of multi-residue methods based on GC–MS or LC–MS/MS is presented.

II. SELECTION OF PESTICIDES FOR THIS COMPARISON

As noted above, approximately 860 active substances are currently used in pesticide formulations (Tomlin, 2003). In addition, several metabolites, degradation products, and “old” (persistent) pesticides have to be considered by pesticide residue analysts. Probably no technique is able to analyze all these >900 analytes completely.

For this reason, a selection of “important” pesticides was necessary. The selection was started with the exclusion of >140 pesticides, which are not important for the comparison of GC–MS versus LC–MS/MS. These pesticides are:

  • Nine dithiocarbamates, 48 biological agents (bacteria, fungi, viruses, etc.), and 29 inorganic compounds, which cannot be analyzed by multi-residue methods based on GC–MS or LC–MS/MS.

  • Thirty-five pheromones, which are less important because residues are not expected.

  • Several isomers (e.g., alpha-cypermethrin, beta-cypermethrin, theta-cypermethrin, and zeta-cypermethrin), if one of these isomers is considered.

The following criteria were taken into account to select the more important substances from the remaining pesticides:

  • 1Status of production: Selected pesticide should be listed in that part of the 13th edition of the Pesticide Manual that contains the actually produced pesticides.
  • 2Status of residue regulation in the EU or in Germany (which are completely available for us). Regulated pesticides are preferred.
  • 3Occurrence of residues: Those pesticides are preferred, which are more often found in food monitoring programs.
  • 4Inclusion of important metabolites: Metabolites and/or degradation products, which are included in the residue definition, should be considered in addition.
  • 5At least one of both detection techniques (GC–MS or LC–MS/MS) must be applicable.

Using these criteria, 422 pesticides and 42 important metabolites were chosen. In addition, 36 pesticides were selected, because their residues in food are regulated, even though these compounds are not produced any longer. The resulting total number of 500 compounds is presented in Table 1. The compilation contains 81 organophosphorus pesticides, 43 carbamates, 40 organochlorines, 26 sulfonylureas, 24 triazoles, 23 triazines, 22 other ureas, 19 pyrethroids, 12 aryloxyphenoxypropionates, and 10 aryloxyalkanoic acids. In the Pesticide Manual, the remaining 207 compounds are assigned to further 90 chemical classes.

Table 1. Typical ions selected for GC-EI-MS or transitions used in LC-ESI-MS/MS and the sensitivity obtained with both techniquesThumbnail image of Thumbnail image of Thumbnail image of Thumbnail image of Thumbnail image of Thumbnail image of Thumbnail image of Thumbnail image of Thumbnail image of Thumbnail image of Thumbnail image of Thumbnail image of Thumbnail image of Thumbnail image of Thumbnail image of
  • aEU regulation can be found on the websites: http://europe.eu.int/eur-lex/en/search/search_lif.html or http://europe.eu.int/eur-lex/lex/en/repert/035020.htm

  • b[M + NH4]+ used as quasimolecular ion.

  • cRegulated metabolite in the EU.

  • dAnalyte requires special HPLC conditions for detection with ESI–MS/MS.

  • eQuasimolecular ion was [M–OH]+.

  • fReference 2 does not report the product ion.

  • gQuasimolecular ion was [(M–O)/2]+.

  • hQuantitative degradation of the pesticide occurs in the GC injector.

  • iReference 1 does not report the product ion.

  • The placement of some compounds into categories is somewhat arbitrary because some pesticides contain several or more characteristic structural features. If the mode of action is considered, 172 herbicides, 171 insecticides, 105 fungicides, and 52 pesticides from other pesticide types (acaricides, bactericides, herbicide safeners, molluscicides, nematicides, plant growth regulators, and synergists) are selected. Approximately 90% of those pesticides that are regulated by the EU Commission are included in the table. It should be noted that the majority of the excluded pesticides belongs to the group of herbicides, which typically causes lower amounts of residues in food.

    By selection of such a large number of pesticides, we tried to include most analytes being important in pesticide residue analysis. But without a doubt, every selection must be incomplete. Several metabolites may be missing, as well as some pesticides, which are important for other reasons not considered here. Nevertheless, this selection was not put together to promote a particular analytical technique.

    III. SELECTION OF INSTRUMENTS AND IONIZATION TECHNIQUES

    The choice of the most appropriate instruments to handle the majority of samples and analytes is one of the most important decisions on investments in residue analytical laboratories. The same decision was necessary for the comparison presented here.

    A. GC–MS

    Ionization of pesticides in GC–MS can be done by EI, and positive or negative chemical ionization (PCI, NCI). For ion separation, single quad instruments are used most frequently. Additionally, GC–MS systems with quadrupole ion traps, time-of-flight (TOF) mass spectrometers or tandem mass spectrometers are available.

    Most of the published studies on residue analysis by GC–MS report on results obtained by single quadrupole instruments and EI ionization. Advantages of EI ionization are a low influence of molecular structure on response, and a large number of characteristic fragments. Extensive studies describe the simultaneous determination of 245–400 pesticides by GC–EI–MS with single quadrupole mass filters (Cairns et al., 1993; Fillion, Sauve, & Selwyn, 2000; Stan, 2000; Chu, Hu, & Yao, 2005). The use of ion traps in scan mode is more simple because no selection of characteristic ions is necessary during data acquisition. In full scan mode these instruments are quite sensitive, and confirmation by library search is possible at lower concentrations. But, compared to single quad instruments running in selected ion monitoring mode (SIM), identical pesticides are covered and the sensitivity do not differ significantly (Cairns et al., 1993).

    Chemical ionization is used more rarely. Positive or negative CI–MS give better selectivity for several pesticides compared to EI. This results in chromatograms with reduced matrix interference (Hernando et al., 2001). But the signal intensity of different pesticides (if identical amounts are injected) varies much more compared to EI ionization. Preferentially, GC–MS with chemical ionization is focused on special substance classes only, for example, organohalogen pesticides (Artigas, Martinez, & Gelpi, 1988; Chaler et al., 1998), pyrethroids (Ramesh & Ravi, 2004), and organophosphates (Russo, Campanella, & Avino, 2002). It is rarely used in multi-residue methods, because it is not a universal ionization technique. Finally, mass spectra produced by chemical ionization usually contain a smaller number of fragments, thus offering less information.

    Available GC–TOF instruments can be operated in two different modes. One type offers very high scan rates, allowing the separation of overlapping peaks by automated mass spectral deconvolution of overlapping signals (de Koning et al., 2003; Patel et al., 2004). This can result in up to 30,000 peaks from cigarette smoke (Dalluge et al., 2002). Another type of GC–TOF instruments offers high mass resolution, allowing data evaluation with a narrow mass window of 0.02 Da (Cajka & Hajslova, 2004). However, most TOF instruments suffer from a reduced dynamic range (Dalluge, Roose, & Brinkman, 2002). For this review, no sufficient information on multi-analyte GC–TOF was available.

    In analogy to CI–MS and GC–TOF, a good suppression of matrix background is obtained by GC–MS/MS systems (Goncalves & Alpendurada, 2004). Even with extracts of tobacco, excellent selectivity and sensitivity were observed (Haib, Hofer, & Renaud, 2003). MS/MS experiments can be performed using ion trap (Gamon et al., 2001; Aguera et al., 2002; Martinez Vidal, Arrebola, & Mateu-Sanchez, 2002) and triple quadrupole mass analyzers (Leandro, Fussell, & Keely, 2005). Some limitations in GC–MS/MS arise from the absence of a universal soft ionization mode, which could be used for the efficient production of molecular ions of most pesticide classes. Chemical ionization generates high-intensity ions of only some pesticides classes. EI ionization is more universal, but often the total ion current is spread on many fragments, resulting in a low intensity of parent ions of MS/MS experiments. Up to now, the prospects of GC–MS/MS are not totally clear. GC–MS/MS acquisition parameters are published for a small percentage of selected pesticides. Therefore, it is too early to choose GC–MS/MS instead of GC—MS for a comparison with the most appropriate LC–MS(/MS) approach.

    B. LC–MS

    If pesticides are not amenable to GC, the application of LC is the best alternative. Likewise, LC may be combined with single quadrupole instruments, quadrupole ion traps, triple quadrupole (tandem) mass spectrometers, TOF spectrometers, or hybrid quadrupole TOF instruments.

    In contrast to GC–MS, single quadrupole mass spectrometers are not used in the majority of recent studies dealing with LC–MS. A disadvantage of single quadrupole instruments (and ion traps operated in the SIM mode) is the high intensity of background signals produced from sample matrix and HPLC solvent clusters. Due to this chemical noise in real samples very low limits of quantification cannot be achieved, even if the sensitivity of these instruments is high (Hernandez, Sancho, & Pozo, 2005).

    The chemical background can be reduced significantly if tandem MS in combination with selected reaction monitoring (SRM) is applied. Even if a co-extracted matrix component has the molecular mass of a pesticide, usually both isobaric ions can be separated in SRM experiments, because their fragmentation in the collision cell most often results in different product ions. Therefore, tandem mass spectrometers offer excellent sensitivity and unsurpassed selectivity. For this reason, triple quadrupole mass analyzers have been the most often applied MS detectors until now (Pico, Blasco, & Font, 2004). Quadrupole ion traps may also be operated in the MS/MS mode, which reduces the background to a level known from tandem mass spectrometers. However, ion collection, fragmentation, and mass analysis of fragments is a step by step process in traps and requires much more time than in triple quadrupole instruments, which do this in parallel. Furthermore, ion traps suffer from a limited dynamic range, a smaller potential to fragment very stable ions and the inefficiency to trap low mass fragments (Pico, Blasco, & Font, 2004).

    Time-of-flight mass spectrometers in combination with LC are more often used in high-resolution mode (typical mass error <2 mDa), which provides better discrimination of background (Hogenboom et al., 1999; Ferrer et al., 2005). The main advantage of this type of instrument is the identification of unknown peaks in a sample even if analytical standards are not available (Garcia-Reyes et al., 2005; Thurman, Ferrer, & Fernandez-Alba, 2005). But, this advantage is usually not needed in the enforcement of maximum residue levels. Furthermore, identification of pesticides in samples is less certain by LC–TOF–MS than identification of pesticides by GC–EI–MS (Maizels & Budde, 2001).

    The use of a hybrid quadrupole time-of-flight instrument (Q–TOF) allows the most certain confirmation. This confidence is based on the combination of retention time, mass of the quasi molecular ion selected by the quadrupole mass filter, and the complete collision induced mass spectrum obtained by the TOF analyzer (Hernandez et al., 2004). Unfortunately, the sensitivity of Q–TOF instruments in relation to triple quadrupole analyzers is one order of magnitude lower (Hernandez et al., 2004; Nunez, Moyano, & Galceran, 2004). Additionally to this drawback, a smaller linear range restricts the use of Q–TOF for the quantification of residues.

    All LC–MS instruments can be equipped with at least three types of soft ionization techniques, that is, ESI, APCI, and photoionization. Up to now, articles on photoionization of pesticides have been rarely published (Takino, Yamaguchi, & Nakahara, 2004). ESI and APCI are applied more often. Comparing the suitability of ESI versus APCI for the ionization of many pesticides, electrospray was identified as more universal technique (Thurman, Ferrer, & Barcelo, 2001; Klein & Alder, 2003; Jansson et al., 2004; Hernandez, Sancho, & Pozo, 2005).

    C. Final Decision

    Any of the instruments discussed above have special merits, but none of them can detect the full range of all pesticides. However, if the selection of the most appropriate techniques is focused on the enforcement of maximum residue levels, simultaneous identification, and quantification of a very large number of target analytes will be more important than the detection, identification, and quantification of non-regulated (non-target) pesticides and/or metabolites. Under these conditions, EI ionization and single quadrupole MS was identified as the preferred GC detection system. If LC is used, most benefits should be obtained from tandem mass spectrometers operating in the electrospray mode. Therefore, in the next section scope and sensitivity of GC-EI–MS will be compared to pesticide detection by LC–ESI–tandem MS.

    IV. COMPILATION OF EXISTING DATA

    Characteristic ions of EI mass spectra, which are applied to the determination of pesticides by GC–MS, as well as typical transitions from precursor to product ions used for LC/tandem MS, are presented in Table 1.

    Most of the cited articles contain information on the sensitivity of the instrument or method used. However, a comparison of such data is difficult. In some cases, sensitivity is based on the signal-to-noise ratio of peaks in chromatograms of standards. In other studies, sensitivity is derived from the limit of quantification (LOQ) of the complete analytical method. In the latter case, the type of matrix and the concentration of the final extracts have to be considered. Furthermore, sensitivity of instruments has improved significantly in the last years. Finally, several parameters of GC–MS or LC–MS/MS measurement influence the sensitivity. In GC–MS such parameters are the dwell time, but also the type and length of column or temperature program. If LC–MS/MS is used, chromatography (e.g., type of solvent, buffer), ionization (e.g., ionization voltage, temperature, gas pressure), or parameters of ion measurement (e.g., dwell time, collision energy) can influence the sensitivity obtained. Therefore, data on sensitivity from different studies are often not comparable.

    To compare the sensitivity of mass spectrometric determination of pesticides avoiding these problems, the limits of quantification presented in Table 1 were estimated by GC–MS and LC–MS/MS under identical conditions each.

    A. GC–MS Data

    Gas chromatography–mass spectrometry (GC–MS) has been practiced in analyzing pesticides for several decades and most characteristic ions are available from pesticide analytical manuals, applications of instrument producers, or spectral databases supplied by producers of analytical standards. In addition, some excellent articles covering GC–MS analysis of a broad range of pesticides are published (Fillion, Sauve, & Selwyn, 2000; Wong et al., 2003; Chu, Hu, & Yao, 2005). Only for some pesticides, the characteristic ions in Table 1 are obtained from recent articles or studies conducted by the pesticide industry.

    For estimation of sensitivity, several mixed standards in solvent containing all pesticides were analyzed. A state-of-the-art GC–MS system (Agilent 6890N GC and 5975 inert MSD) using a pulsed pressure injection of 1 µL onto a HP-5 MS column (30 m × 0.25 mm × 0.25 µm), EI ionization, and a dwell time of 40 msec were applied for each characteristic ion. The same GC conditions were used for all injections. Obviously, a longer dwell time would result in better sensitivity, but—at the same time—it would reduce the number of pesticides analyzed in one run.

    Standard solutions containing 10,000, 1,000, 100, 10, and 1 ng/mL were injected. If a pesticide was not detected at the highest concentration in the SIM mode, no data were added to Table 1. In all other cases, characteristic ions are presented. The LOQ was set to the lowest concentration, which gave a signal-to-noise ratio of ≥10 for the most intense peak of the analyte.

    B. LC–MS/MS Data

    Typical transitions from precursor to product ions are taken from recent publications, since analogous data collections do not exist for LC–MS/MS. SRM transitions from studies conducted with triple quadrupole or quadrupole ion trap instruments are preferred. If such studies were not found, data on those product ions are cited, which are produced in single quadrupole instruments by increasing the potential between the entrance capillary and the first skimmer (fragmentor or cone voltage). Often, transitions for a selected pesticide are published by more than two authors. In such cases, the citation in Table 1 prefers the first or at least the earlier publications. However, no transitions were found in published studies for approximately one-third of the pesticides listed in Table 1. In such cases, typical transitions are taken from unpublished studies of pesticide producers or from the world wide web (BfR, 2005).

    The sensitivity of LC–MS/MS instruments was assessed using a triple quadrupole mass spectrometer (API 4000, Applied Biosystems) by injection of 20 µL analytical standard on a short reversed phase column (Phenomenex Aqua, 50 mm × 2 mm × 5 µm), using a gradient of methanol/water containing 5 mmol/L ammonium formate. Approximately 100 pesticide transitions were acquired simultaneously after ESI using an identical dwell time of 20 msec for each SRM transition.

    The batch used at the LC–MS/MS instrument included standard solutions with concentrations of 100, 10, 1, and 0.1 ng/mL. If a pesticide was not detected at the highest concentration in the SRM mode, no data were added to Table 1. In all other cases, typical transitions are presented. The LOQ was set to the lowest concentration, which gave a signal-to-noise ratio of ≥10 for the most intense peak of the analyte.

    V. CONCLUSIONS FROM COMPILED DATA

    A. Comparison of Scope of Both Techniques

    The data in Table 1 demonstrate that more pesticides and their metabolites can be analyzed by LC and ESI than by GC–MS. It is well known that sulfonyl or benzoyl ureas and many carbamates or triazines can be better or exclusively detected by LC–MS/MS techniques. Furthermore, a wider scope of LC–MS/MS was found for most of the other chemical classes too, for example, the organophosphorus pesticides. Only 49 compounds out of 500 exhibited no response, if LC–MS/MS in combination with positive and negative ESI was used. On the other hand, 135 pesticides/metabolites could not be analyzed by GC/MS using EI ionization, most often because of incompatibility with evaporation of the intact molecule in the GC injector.

    A more detailed overview presenting separate data for several chemical classes is given in Table 2. The data presented in this table demonstrate clearly that several pesticides, which are identified typically by an electron capture detector in GC measurements, do not show a sufficient LC–MS/MS response. This is well known for organochlorine compounds, but it is also valid for other pesticides like benfluralin, chlozolinate, dinobuton, etridiazole, flumethralin, nitrofen, or vinclozolin. The only exceptions are fenchlorphos, which is better detected by GC with flame photometric or nitrogen-phosphorus detection and biphenyl, which can be analyzed by GC–MS, only.

    Table 2. Pesticides, which are not covered by GC–MS or LC–MS/MSThumbnail image of
  • *This calculation/list does not contain four organotin compounds, four quaternary ammonium salts, glyphosate, and picloram, which require special LC conditions for ESI–MS/MS detection.

  • B. Comparison of Sensitivity

    Both, GC–MS- and LC–MS-based methods, reveal a significant variation of sensitivity, covering at least a range of 3–4 orders of magnitude, depending on the pesticide. However, a comparison of the median of the limits of quantification clearly shows much higher sensitivity if determinations are based on LC and tandem MS. Most analytes may be quantified reliably by LC–MS/MS (at least in standard solutions) at concentrations between 0.1 and 1 ng/mL. In contrast, the median of the limits of quantification observed by GC–MS is distinctly higher, that is, at 100 ng/mL. The distribution of LOQ data from Table 1 is summarized separately for both techniques in Figure 1. Nearly the same distribution is found for organophosphorus pesticides, which are most often analyzed by GC methods up to now (Fig. 2). An analogous pattern is found for many other chemical classes of pesticides.

    Figure 1.

    Distribution of limit of quantification (LOQ) data of all pesticides/metabolites.

    Figure 2.

    Distribution of LOQ data of all organophosphorus pesticides.

    Another approach that brings to the same conclusion is presented in Figure 3. In this figure the percentage of those pesticides that show a better response with GC–MS is compared with the percentage of compounds that are quantified with higher sensitivity by LC–MS/MS. In addition to those 47 pesticides, which are not detected by LC–MS/MS at a level of 100 ng/mL, only two analytes (acrinathrin and procymidone) are analyzed with better sensitivity by GC-MS. Finally, 19 pesticides (bromophos-ethyl, chlormephos, chlorobenzilate, chlorpyrifos-methyl, cyanofenphos, cyanophos, cycloate, cyhalofop-butyl, dichlofenthion, diphenylamine, esfenvalerate, fenitrothion, fenvalerate, lambda-cyhalothrin, methacrifos, parathion-methyl, phorate, prothiofos, tolclofos-methyl) are detected with an equal LOQ by GC–MS.

    Figure 3.

    Comparison of GC–MS sensitivity versus LC–MS/MS sensitivity of individual pesticides summarized for different pesticide classes.

    The better performance of LC–MS/MS is probably determined by several reasons. Among them the higher injection volume used in LC–MS/MS (20 µL vs. 1 µL) and the lower amount of fragmentation during ionization (ESI vs. EI) may explain some of these differences.

    C. Conclusions and Perspectives for Multi-Residue Methods

    Gas chromatography (GC) coupled to EI-MS and LC combined with tandem MS are the most important detection techniques in pesticide residue analysis today. The comparison of scope and sensitivity of both techniques presented above has illustrated the better performance of LC–MS/MS.

    While establishing the measurements of hundreds of LOQs, another important advantage of LC–MS/MS became clear. Due to the small peak width in GC, the cycle time in GC–MS methods must be 1 sec or shorter. Since all ions are recorded using a dwell time of 40 msec, not more than 25 characteristic ions can be recorded in one time window. Assuming 10 time windows in one GC run, 250 ions or 83 pesticides with 3 characteristic ions each can be analyzed in parallel theoretically. The peak width in LC measurements is usually higher, often allowing a typical cycle time of 2.5 sec. Based on the dwell time of 20 msec used for the data in Table 1, approximately 125 SRM transitions can be acquired simultaneously in one time window. Assuming 5 time windows per LC run in that case, 625 SRM transitions are obtained with one injection. Since two SRM transitions are often sufficient to quantify and confirm a result, up to 312 pesticides can be analyzed theoretically in one run. In practice, the theoretical numbers calculated above cannot be reached because usually more pesticides elute in the middle than in the beginning or end of the chromatograms. However, irrespective of this limitation, the number of analytes covered in one LC–MS/MS run is at least two or three times higher than the number of pesticides measured in parallel by GC–MS in the SIM mode.

    The comparison of both techniques would remain incomplete, if the influence of matrix on the determination is not considered. Matrix effects on the analyte transmission from the GC injector to the column (Hajslova & Zrostlikova, 2003) or inhibition of ESI (Bester et al., 2001; Stuber & Reemtsma, 2004) are well known phenomena. In both cases the use of matrix matched standards can reduce the problem, but preparation of such standards is laborious and appropriate sample materials without any residues are not generally available. Therefore, the use of surface protectants is an interesting alternative (Anastassiades, Mastovska, & Lehotay, 2003), which is applicable for GC methods but not for LC–MS/MS. In several cases, the influence of coeluting matrix peaks on the atmospheric pressure ionization can be reduced by the ECHO technique (Zrostlikova et al., 2002), but a general compensation of matrix effects is not obtained (Alder et al., 2004). Using LC–MS/MS, the simplest alternative is the dilution of extracts. However, such dilution requires residue concentrations distinctly above the LOQ. If no other choice exists, the method of standard addition will solve this problem of accurate quantification.

    In addition to the effect on response, matrix components produce several additional signals in chromatograms. Such interferences are not seldom if extracts of complex matrices (i.e., herbs or tea) are analyzed by GC–MS. False positive identifications of pesticides may be a consequence. Matrix interference is significantly reduced if tandem MS is used. From that reason, LC–MS/MS methods do not require such an extensive cleanup and sophisticated chromatographic separation (Stout et al., 1998). Different molecules that share the same transition are more rarely found than molecules producing fragments of identical mass. As a consequence, peak identification, integration, and data processing are much easier and faster in LC–MS/MS, and require less manual corrections compared to GC–MS (Lehotay et al., 2005).

    The discussion of many aspects of determination of pesticide residues by GC–MS and LC–MS/MS clarified that neither MS in combination with GC nor the LC-based technique may solve all problems of residue analysts. Both techniques and additional ones are needed today and will be needed in future. However, the benefits of LC–MS/MS in terms of wider scope, increased sensitivity, and better selectivity are obvious. These characteristics, together with the ability to perform most determinations without derivatization, make LC–MS/MS the preferred technique currently available for the determination of pesticide residues.

    Acknowledgements

    We thank Volker Happel and Marilyn Menden for their support needed for the determination of limits of quantification. Nataša Marković, Birgit Mueller, and Annamaria Melcher provided important technical assistance throughout this work.

    Biographical Information

    original image

    Lutz Alder studied chemistry in Berlin/Germany with a focus on organic and analytical chemistry. After his Ph.D. in 1978, he managed the mass spectrometry laboratory at Humboldt University for several years. He has been employed by the Federal Health Office, which is now the Federal Institute for Risk Assessment (BfR) since 1991. Most of his research has been concerned with residue analytical methods and its standardization for official use. He has published 50 scientific journal articles and book chapters.

    original image

    Kerstin Greulich studied chemistry at Dresden Technical University focusing on environmental chemistry. She received her Ph.D. at Humboldt University in 2004. At present she is employed at the Federal Institute for Risk Assessment (BfR). Her research interests involve residue analytical methods, environmental analysis, ecotoxicology, and herpetology. She has published 12 scientific journal articles and book chapters.

    original image

    Baerbel Vieth studied chemistry at the Humboldt University in Berlin. After finishing the Ph.D. in 1981, her research has been focused on the development and application of HPLC methods in trace analysis of drugs, enzyme activities, and of pesticides. In 1991, she joined the former Federal Health Office of Germany, which is now the Federal Institute for Risk Assessment (BfR). Besides residue analytical methods for pesticides, she has been responsible for evaluation of residues in human milk with special focus on POPs. She has published about 30 articles in scientific journals.

    At present, Lutz Alder, Kerstin Greulich, and Baerbel Vieth are responsible for the evaluation of residue analytical methods provided for registration of new pesticides in Germany and contribute to authorization of plant protection products in Europe.

    original image

    Guenther Kempe studied food chemistry and analytical chemistry in Dresden/Germany. He started his career in the Hygiene Institute Chemnitz, which became the State Institute for Food and Health Protection of Saxony/Germany in 1991. Most of his research has been concerned with pesticide residue analysis in food. Today he is the head of the residue laboratory of the State Institute and chairmen of the Working Group 'Pesticides' of the Gesellschaft Deutscher Chemiker (Chemical Society of Germany). His publication list contains 20 scientific articles and book chapters.

    Ancillary