Protocol for an Electrospray Ionization Tandem Mass Spectral Product

Jun 11, 2012 - Ontario Ministry of the Environment, 125 Resources Road, Etobicoke, Ontario M9P .... Drug Administration/Center for Food Safety and App...
24 downloads 0 Views 900KB Size
Download PDF
Article pubs.acs.org/ac

Protocol for an Electrospray Ionization Tandem Mass Spectral Product Ion Library: Development and Application for Identification of 240 Pesticides in Foods Kai Zhang,*,†,‡ Jon W. Wong,†,‡ Paul Yang,§,‡ Douglas G. Hayward,†,‡ Takeo Sakuma,∥ Yunyun Zou,∥ André Schreiber,∥ Christopher Borton,∥ Tung-Vi Nguyen,§ Banerjee Kaushik,⊥ and Dasharath Oulkar⊥ †

Office of Regulatory Science, Center for Food Safety and Applied Nutrition, U.S. Food and Drug Administration, 5100 Paint Branch Parkway, HFS-706, College Park, Maryland 20740, United States § Ontario Ministry of the Environment, 125 Resources Road, Etobicoke, Ontario M9P 3 V6, Canada ∥ AB Sciex, 71 Four Valley Drive, Concord, Ontario L4K4 V8, Canada ⊥ National Research Centre for Grapes, Post Office Manjri Farm, Pune, Maharashtra 412307, India S Supporting Information *

ABSTRACT: Modern determination techniques for pesticides must yield identification quickly with high confidence for timely enforcement of tolerances. A protocol for the collection of liquid chromatography (LC) electrospray ionization (ESI)− quadruple linear ion trap (Q-LIT) mass spectrometry (MS) library spectra was developed. Following the protocol, an enhanced product ion (EPI) library of 240 pesticides was developed by use of spectra collected from two laboratories. A LC-Q-LIT-MS workflow using scheduled multiple reaction monitoring (sMRM) survey scan, information-dependent acquisition (IDA) triggered collection of EPI spectra, and library search was developed and tested to identify the 240 target pesticides in one single LC-Q-LIT MS analysis. By use of LC retention time, one sMRM survey scan transition, and a library search, 75−87% of the 240 pesticides were identified in a single LC/MS analysis at fortified concentrations of 10 ng/g in 18 different foods. A conventional approach with LC-MS/MS using two MRM transitions produced the same identifications and comparable quantitative results with the same incurred foods as the LC-Q-LIT using EPI library search, finding 1.2−49 ng/g of either carbaryl, carbendazim, fenbuconazole, propiconazole, or pyridaben in peaches; carbendazim, imazalil, terbutryn, and thiabendazole in oranges; terbutryn in salmon; and azoxystrobin in ginseng. Incurred broccoli, cabbage, and kale were screened with the same EPI library using three LC-Q-LIT and a LC-quadruple time-of-flight (Q-TOF) instruments. The library search identified azoxystrobin, cyprodinil, fludioxinil, imidacloprid, metalaxyl, spinosyn A, D, and J, amd spirotetramat with each instrument. The approach has a broad application in LC-MS/MS type targeted screening in food analysis.

T

development of many multiresidue methods for the analysis of polar and thermally labile pesticides. Using one multiple reaction monitoring (MRM)6 or two MRM data acquisitions,7,8 these methods are capable of analyzing >200 pesticides in a single LC-MS/MS analysis. The use of two MRM data acquisitions within a specific retention time (RT) window increases both the selectivity and signal-to-noise ratio (SNR) of the extracted ion chromatograms. This approach was assumed to be sufficient to fulfill the requirement of EU Directive 2002/ 657/EC9,10 for identification of unknowns. Because mass spectral fragmentation patterns are dependent on instrument,

he application of anthropogenic pesticides on crops to protect against weeds, insects, fungi, and other pests is a common agricultural practice. When applied pesticides dissipate, pesticide residues may remain on agricultural products and in the environment. These residues may pose a potential threat to human health. Regulations have been established by government agencies to ensure that concentrations of pesticides in food are lower than the maximum residue limits (MRLs),1−4 leading to the development of various analytical methods to support these MRLs. Until the late 1990s, gas chromatography−mass spectrometry (GC-MS) had been the technology of choice for pesticides analysis with the exception of polar and/or thermally labile pesticides that are not suitable for GC procedures.5 The advent of electrospray ionization (ESI)-based liquid chromatographytandem mass spectrometry (LC-MS/MS) resulted in the © 2012 American Chemical Society

Received: March 27, 2012 Accepted: June 8, 2012 Published: June 11, 2012 5677

dx.doi.org/10.1021/ac300844d | Anal. Chem. 2012, 84, 5677−5684

Analytical Chemistry

Article

world samples or effects of analyte concentrations.23 Libraries had been developed and applied to screen pesticides in blood16 and urine,17 pharmaceutical residues in environmental waters,18 and pesticides in fruits.19 These studies were limited by either lacking a defined protocol for library collection and application or testing only a few target analytes. We therefore selected 240 pesticides that are regulated by the U.S. Environmental Protection Agency (U.S. EPA) and European Union (EU) and are amenable to electrospray ionization (ESI) based LC-MS analysis in positive polarity mode. We optimized a LC-Q-LIT MS system, developed a protocol, and collected EPI library spectra at the U.S. Food and Drug Administration/Center for Food Safety and Applied Nutrition laboratory (CFSAN, College Park, MD) concurrently with the Laboratory Services Branch of the Ontario Ministry of the Environment (LSB, Toronto, Ontario, Canada), using the same protocol for the same 240 pesticides. We combined the collected spectra to form a single EPI library and carried out this study to investigate how to best utilize the EPI library. The workflow involving QuEChERS (quick, easy, cheap, effective, rugged, and safe) sample preparation,7,8 LC-Q-LIT-MS, and the EPI library was developed in order to identify targeted pesticides and was evaluated by use of spiked and incurred samples. Operational parameters for LC separation, CE used in the CID experiment, and ion trap fill time were optimized. These parameters that could affect the quality and consistency of the resulting EPI spectra are discussed for future reference. Finally, the results obtained from three laboratories for the same incurred foods were evaluated to assess the applicability of the protocol, workflow, and EPI library.

LC mobile phase, analyte concentration, and matrix effects in the ESI source, there are challenges using RT, two MRM transitions, and their relative ion intensity ratios for identification of target analytes.7,11 These include the selection of nonspecific transitions that may be the same as that of an interfering matrix compound or the weak response of the second MRM transition at low intensity, which could lead to false positive/negative identification. These challenges prompted the need for additional approaches to complement existing methods to achieve unambiguous identification of pesticides. The soft ionization nature of the ESI process of LC-MS allows spectra formation providing useful molecular weight information through formation of intense pseudomolecular ions. Identifications attempted by using a single pseudomolecular ion cannot meet current identification criteria.9,10 One must use collision-induced disassociation (CID) in the source region or utilize MS/MS to collect product ions useful for identification. However, CID in the source region would generate spectra with significant background interference. The Q-LIT can perform MRM experiments as survey scan and IDA triggered CID experiments in a collision cell to acquire full scan product ion spectra of preselected precursor ions from the survey scan. This approach has been exploited for both quantitative and qualitative analysis of small molecules.12−14 When single collision energy (CE) is used for CID, the generated spectra could lack enough characteristic fragments if the CE was too low or lack pseudomolecular ions if the CE was too high. Both ion-trap MS and LIT-MS have the unique ability of conducting collision energy spread (CES) experiment to collect an “average” MS spectrum of different CE values in one single EPI scan, resulting in a full scan spectrum with both molecular and fragment ion information that can be used in library search-based identification15−19 for increased confidence. Several attempts have been made to create LC-MS libraries20−24 and search algorithms26 that can be used for the identification of chemicals in LC-MS/MS analysis. In a CID experiment, analytical standards were introduced into different MS/MS systems through a syringe pump20,22 or a LC through an ESI interface.21,23−26 Fragment-rich EPI spectra were collected via ion trap,20,26 triple quadrupole,18 Q-LIT,21−23 and quadrupole time-of-flight (Q-TOF) MS with CID experiment utilizing single or multilevel CE20−23 or a tuning point protocol25 to build EPI spectral libraries. Due to the ionization process of different ESI sources, including but not limited to clustering of neutral species and in-source fragmentation, and the lack of standardized protocols, there was limited success in creating an EPI spectral library with features similar to the NIST GC-MS library (http://www.nist. gov/srd/nist1a.cfm) that can be used independent of LC-MS/ MS instrument platforms. The general consensus derived from these studies was that, by use of prescribed LC and MS/MS parameters, fragment-rich EPI spectra could be obtained from the CID experiment along with characteristic molecular and fragment ions. This suggests that the creation of EPI mass spectral libraries and its potential application in targeted screening can be a viable approach. There are limited studies on the viability of using LC-Q-LIT or LC-LIT analysis and an EPI-based library for identification. Only one EPI library has been developed for the identification of compounds of forensic and clinical toxicology. The library was tested with analytical standards under optimized laboratory conditions without considering possible matrix effects of real-

■

EXPERIMENTAL SECTION Chemicals and Sample Preparation. The majority of the 240 pesticide standards were obtained from the U.S. Environmental Protection Agency (U.S. EPA) Pesticide Repository (Ft. Meade, MD). The others were obtained through Fluka/ Sigma−Aldrich (St. Louis, MO), EQ Laboratories (Dr. Ehrenstofer, Atlanta, GA) and Wako Chemicals USA (Richmond, VA) and are listed in Supporting Information (Table S1). Methanol, acetonitrile, HPLC-grade water, formic acid, ammonium formate, anhydrous magnesium sulfate, and sodium chloride were purchased from Fisher Scientific (Pittsburgh, PA). Four deuterium (2H) isotope-labeled internal standards (diazinon-d10, dimethoate-d6, diuron-d6, and dichlorvos-d6) were purchased from CDN Isotopes (Montreal, Quebec, Canada). All samples were prepared by QuEChERS (quick, easy, cheap, effective, rugged, and safe) protocols by use of kits purchased from United Chemical Technologies (UCT, Bristol, PA). Separate stock solutions of analytical standards, including those for the isotope-labeled internal standards, were prepared by weighing 10−75 mg of each individual compound and dissolving in 10 or 25 mL of acetonitrile, methanol, or methanol/water (50:50 v/v) in volumetric flasks or calibrated plastic tubes (Simport, Quebec, Canada). Intermediate solutions and spike solutions were prepared in 200 mL volumetric flasks by mixing the 240 stock solutions and used in the preparation of solvent-only calibration standards (SOCS) and matrix-matched calibration standards (MMCS) and used in the study for quantitation and identification. All incurred samples were prepared at CFSAN and analyzed at CFSAN, LSB, and AB Sciex product application laboratories (PAL). Food and environmental vegetation and biota samples 5678

dx.doi.org/10.1021/ac300844d | Anal. Chem. 2012, 84, 5677−5684

Analytical Chemistry

Article

were prepared by QuEChERS procedures7,8 with four 2Hlabeled surrogates as internal standards. Eighteen sample matrices including almond, avocado, broccoli, cabbage, corn, ginkgo, ginseng, honey, kale, olive oil, orange, raisin, salmon, saw palmetto, shrimp, acetonitrile, spinach, and tomato were spiked with the 240 pesticides and used in the EPI library study. Incurred broccoli, cabbage, ginseng, kale, orange, peach, and salmon samples were purchased locally or collected from the field. Following the procedures described in two previous studies,7,8 organic brand samples were used as “blank” to prepare MMCS standards at concentrations ranging from 1 to 100 ng/mL for the quantitative analysis. Prepared samples were stored at −40 °C in the freezer until ready for analysis. Prior to LC-MS/MS analysis, quality control blanks and fortified sample extracts were prepared by diluting 200 μL of sample extracts with 300 μL of acetonitrile and 500 μL of sample buffer and filtered through 0.2 μm micrometer nylon membrane filters (Sun SRi, Rockwood, TN) directly into the autosampler vials. For the interlaboratory comparison, two sets of prepared sample extracts, sample blank extracts and standards in acetonitrile, were packed in coolers with ice packs and shipped to LSB and PAL overnight, then prepared by the prescribed procedure and analyzed as is. Instrumental Conditions for EPI Reference Spectra Collection. All reference EPI spectra in the library were obtained independently at CFSAN and LSB under the same conditions as described below. Pesticide standards were introduced individually into the two AB Sciex 4000 quadruple linear ion trap (Q-LIT) (Concord, Ontario, Canada) via Shimadzu Prominence/20 series LC (Columbia, MD) systems so that the retention times of each pesticide could be obtained. Separation was achieved by an Restek Ultra Aqueous C18 column (100 mm × 2.1 mm, 3.0 μm) and a Trident Direct inline guard cartridge holder with filter (10 mm × 2.1 mm, 3.0 μm) (Bellefonte, PA). The LC mobile phases consist of A = 10 mM ammonium formate/0.1% formic acid/water and B = 10 mM ammonium formate/0.1% formic acid/methanol. Gradient elution at 0.5 mL/min flow rate started at 5% B, ramped to 60% B in 5 min via linear gradient mode and then to 95% B by 12.5 min via exponential gradient mode (pump B curves 3−6), held for 2 min to ensure the removal of organic contaminants at the end of the LC run. Typical full width at half-height (fwhh) of the extracted ion chromatogram (XIC) was about 15 s. Total run time was 18 min including 3.5 min of column conditioning time. Injection volume was 20 μL and the column temperature was set at 40 °C. Ionization source dependent parameters in positive ionization mode were set as follows: curtain gas (CUR), 30 psi; ion spray voltage, 5000 V; nitrogen collision gas (CAD), high; source temperature (TEM), 500 °C; ion source gases 1 and 2 (GS1 and GS2), each 50 psi. Resolution at Q1 and Q3 was set to unit. Optimized compound-dependent parameters such as declustering potential (DP), entrance potential (EP), collision energy (CE), and cell exit potential (CXP) were used in the course of EPI spectra collection on the 4000 Q-LIT systems are summarized in Table S1 (Supporting Information). For each pesticide, the reference EPI spectra were generated at column loadings of 200 and 2000 pg (CFSAN) and 300 and 3000 pg (LSB), under ESI positive mode, and at four collision energies of 20, 35, 50, and 35 eV with a CES (collision energy spread) of 15 eV (35 ± 15 eV). The system would collect spectra at CE 20, 35, or 50 eV with a full dwell time. By use of CE 35 with CES of 15 eV, ions are collected in the LIT

sequentially from a CE set to 20, 35, and 50 eV with 1/3 of the dwell time at each CE setting and then are scanned out together into an “average” spectrum. Nitrogen was used as the collision gas and the corresponding pressure was set to “high”. The mass spectral range of EPI spectra was from m/z = 50 to the precursor m/z plus 50 amu. The dynamic fill time (DFT) function was used to determine the fill time of the LIT automatically. The product ions were scanned out of the LIT at a rate of 4000 amu/s. Analyst 1.5.2 software program was used to control the instruments and to export collected spectra into a Microsoft Access database as the library used throughout the study. All participating laboratories used this library generated above throughout the study. Reporter 3.0.1 software program and visual comparison were used to perform library searches for samples analyzed. IDA-EPI Spectrum Parameters for Pesticide Identification in Foods. The EPI library was evaluated by use of incurred samples and fortified samples injected and separated on Shimadzu Prominence/20 series LC (Columbia, MD) systems that were interfaced to three different AB Sciex (Forest City, CA) quadrupole linear ion trap (Q-LIT, 5500, 4000 and 3200) and a high resolution quadrupole-time-of-flight (q-TOF) systems by use of an ESI interface in CFSAN, LSB, and PAL. CFSAN and LSB used a 4000 LC-Q-LIT, while PAL used a 5500 and a 3200 LC-Q-LIT and a 5600 Q-TOF MS for testing incurred foods. Mass spectrometers used in the analysis were tuned according to the manufacturer’s specifications for mass accuracy and sensitivity by use of poly(propylene glycol) prior to data collection. Scheduled MRM data and EPI spectra were acquired and processed for all compounds in positive ionization. LC conditions were same as those used for reference spectra collection. The IDA-EPI parameters used to collect data included the retention times and MRM transitions (one transition for each pesticide and internal standard) of the 240 target analytes (Table S1, Supporting Information). We set the IDA criteria to select the four (4000 Q-LIT and 5600 Q-TOF) or two (5500 Q-LIT) most intense precursor ions after dynamic background subtraction of survey scan and exclude the former target ions for 3 s. Mass tolerance for precursor ions was 250 mDa. Intensity threshold was 1000 cps. In the analysis of incurred samples, only one EPI spectrum was collected at CE of 35 ± 15 eV and it was searched against library spectra collected with the same energy spread. For each EPI experiment, DP was set to 60 eV while product ions were scanned out of the LIT at 4000 amu/s from m/z = 50−800. Source-dependent parameters were the same as those used in the library collection. The search algorithm in the AB Sciex software used the 16 most intense ions in the sample spectrum, normalized to the product of mass and intensity with removal of ions less than 5% of the base peak. The resulting spectrum was searched against the eight most intense ions in the library spectra. Hits must have the correct retention time window for the reference standard of ±20 s and match at least one ion. Reporter 3.0.1 software program and a visional comparison were used to report and verify search results for samples analyzed. Quantitation was performed with the peak area ratios of the MRM transitions in the survey scan of pesticides to that of the transition of the internal standard, diazion-d10, and compared to concentrations of matrix-matched calibration standards by use of Analyst 1.5 or MultiQuan 2.0. 5679

dx.doi.org/10.1021/ac300844d | Anal. Chem. 2012, 84, 5677−5684

Analytical Chemistry

Article

Figure 1. Protocol and workflow used for EPI library collection and target analyte screening.

■

RESULTS AND DISCUSSION The library collection protocol is described in Figure 1. This protocol was tested by using standards and 18 different spiked and incurred matrices. The workflow shown in Figure 1 was used, involving QuEChERS sample preparation, sMRM survey scan, CID at four CE values to collect EPI spectra, and finally library search against the spectral library for identification. Therefore, under ideal conditions, a library search of acquired spectra from a SOCS should identify all 240 pesticides. This was never the case at any of the concentrations tested. Various factors such as lower analyte concentrations, instrument parameters used, the specific pesticide, and sample matrix affected the number of identified pesticides. In the following discussion, the number of pesticides identified at various operational parameters was used to evaluate and optimize these LC-Q-LIT parameters to the extent possible. LC Optimization Parameters. Liquid chromatography can separate analytes from each other and sample matrix components to minimize the matrix effects and interfering coeluents in LC-MS analysis. It is an integral part of acquiring good analytical data.27,28 LC separation modulates analyte and matrix introduction to the MS system, improving SNR and allowing practical duty cycles. The LC separation should be optimized to control the rate of analyte introduced into the ion trap, to avoid overwhelming EPI spectra generation by the MS. We first optimized the LC separation to achieve the fastest throughput (minimum LC run time) while maintaining the highest number of identified analytes from library search by minimizing coelutions to the extent required. The LC was operated at 0.5 mL/min flow rate to maintain a constant (∼15 s) full width at half-height (fwhh) for the XIC of each pesticide. This ensures that at least five data points are collected with a Q-LIT duty cycle of 3 s. This 3-s duty cycle allows the Q-LIT to collect up to four EPI spectra triggered by the predefined IDA threshold. Utilizing the same mobile phase, six LC gradient programs running from 10 to 45 min were used to analyze six SOCS with concentrations from 1 to 100 ppb. EPI spectra obtained from these 36 analyses were subjected to the library search, and the number of identified compounds at each concentration and LC separation time is plotted in Figure 2. With the exception of the

Figure 2. Effect of LC separation on the number of identified compounds on 4000 Q-LIT.

10 min LC program, the other five LC programs generated comparable results. The 18-min LC gradient was selected for the rest of the study because it gave the maximum number of identified pesticides at the concentration of the solvent standard (2.5 ng/mL) approximately equivalent to default MRL (10 ng/mL) in a food (Figure 2). Effect of Declustering Potential. Declustering potential (DP) is the voltage between the sprayer of the ESI source and the entrance of the MS analyzer. Each molecule has an optimal DP. The higher the DP that is imposed on a molecule, the better clustering of neutral molecules such as water and solvent can be minimized, resulting in more protonated or ammoniated molecular ions entering the MS analyzer; and thus improved SNR (signal-to-noise ratio) of the experiment. Increasing the DP beyond the optimal value will induce fragmentation of the pseudomolecular ions (in source CID) before they enter the mass analyzer and decrease the SNR of the library spectra. DP is typically optimized for each transition by infusing individual pesticides into the ion source. These optimized DPs in the survey scan cannot be used in IDA-EPI spectra generation. A single DP setting must be used during the EPI spectra generation. To determine the optimal DP for EPI spectra generation, SOCS at 2.5, 10, 25, and 100 ng/mL, with DP settings from 40 to 100 at 20 eV increments, were tested. The numbers of identified compounds found from these analyses 5680

dx.doi.org/10.1021/ac300844d | Anal. Chem. 2012, 84, 5677−5684

Analytical Chemistry

Article

EPI Library Evaluations with Spiked Foods. The EPI library was tested with 18 spiked food sample matrices at three concentrations (1, 10, and 100 ng/g), prepared by QuEChERS extraction and analyzed via LC-Q-LIT. Quality control samples include three SOCS solutions at the same concentrations and three blanks. These samples were analyzed by use of the workflow shown in Figure 1. The resulting numbers of identified pesticides in each matrix are listed in Table 2 along with the false negative rates. The EPI library search could identify an average across 18 matrices of 82%relat and 84% of the pesticides at 10 or 100 ng/g, respectively (Table 2). The corresponding relative standard deviations (RSD) are less than 5%. Figure 4 shows the trend of false negative identification rate between the 18 matrices at 1, 10, and 100 ng/g. It is clear that false negative identification rates decreased with increasing concentrations. At the same concentration levels, different sample matrices did affect the number of identified pesticides. For example, at 10 ng/g, 87% of the pesticides were identified in broccoli, 85% in cabbage, and 87% in solvent, while 73% of the pesticides were identified in gingko, 76% in orange, and 77% in saw palmetto. This observation is in line with our earlier study8 that broccoli and cabbage had the least matrix effects on a similar group of pesticides at the same fortifying concentrations. In more complex matrices such as ginkgo and orange, the number of identified pesticides decreased (Table 2), as was the case for gingko and orange, while the number of nondetected pesticides increased at all spiking concentrations compared to other matrices. This suppression may have reduced the signal response of sMRM in the survey scan, failing to trigger the corresponding EPI experiments, or it is also possible that coeluted matrix components with the same m/z triggered EPI experiments in place of the true target pesticide. EPI Library Evaluation with Incurred Foods. The IDAbased EPI experiments will increase the target acquisition duty cycle, which will cause a reduction in the data sampling across the eluting chromatographic pesticide peaks and could affect quantitative results. This effect is best tested with incurred samples. Quantified pesticides were measured from four incurred matrices analyzed by both sMRM LC-MS/MS (∼15 data points across the fwhh) and IDA-EPI MS with library search for identification (∼5 data points across the fwhh). As shown in Table 3, both LC-MS/MS and IDA-EPI analyses identified a total of 11 pesticides in these four samples. Pesticides found within the calibrated range were from 1.2 ng/g for terbutryn in orange to 49 ng/g for fenbuconazole in peach. The same results were found by both procedures with a slight difference in the carbaryl result in peach at a concentration of 6 and 8 ng/g from the IDA-EPI based data acquisition and MS/ MS data acquisition with two MRM transitions, respectively. These experiments demonstrate that, like MRM data acquisition, IDA-triggered EPI data acquisition can provide both qualitative and quantitative results with similar accuracy and precision. The value of an instrument-specific library will be insignificant and not practical. We evaluated the universality of the IDA-EPI library on different instrument platforms to demonstrate the broader applicability of this IDA-EPI library. Three laboratories tested the EPI library using the same incurred samples, three different LC-Q-LIT systems, and a LCQ-TOF in a blind test format, that is, laboratories involved had no prior knowledge of the sample types, incurred pesticides, or their concentrations. Library search results showed that three

are listed in Table 1. Both 40 and 60 eV DPs had similar results. DP 60 eV was chosen for all remaining EPI experiments. Table 1. Effect of Declustering Potential on the Number of Identified Compounds on 4000 Q-LIT declustering potential concn (ppb)

40 eV

60 eV

80 eV

100 eV

2.5 10 25 100

194 200 202 202

190 197 203 204

164 179 193 184

128 149 157 163

Effect of Collision Energies and Dynamic Fill Time. Collision energy (CE) used in CID experiments determines fragmentation pattern and pseudomolecular ion intensity of a specific compound. It plays a vital role in the identification of a compound by use of EPI library spectra. Depending on the instruments used and the application of the libraries, different types of CE including rotational CE,19 fixed CE24−26 or a combination of fixed CE and CES21−23 were used in the CID experiments to generate library spectra for library search applications. In this study, the effect of CE on the library search of SOCS at concentrations from 1 to 100 ng/mL was evaluated by use of EPI spectra collected at CE 20, 35, and 50 and CES 35 ± 15 eV. Mass spectra obtained in this manner were then searched against these libraries to determine which CE would give the highest number of identified compounds. The highest number of identified compounds was found for CE 35 with a CES of 15 eV rather than a fixed CE. We also evaluated the effect of ion trap fill time and DFT parameter by carrying out IDA-EPI experiments with four SOCS at concentrations of 2.5, 10, 25, and 100 ng/mL and five ion trap fill times of 20, 50, 100, and 250 ms and that determined by the DFT. Library searches were conducted on these 20 analyses to determine the number of identified compounds in each analysis, and the results are shown in Figure 3. Figure 3 clearly shows that DFT provided the highest

Figure 3. Effect of ion trap fill time used in an IDA-EPI experiment on the number of identified compounds on 4000 Q-LIT.

number of identified pesticides at 2.5 ng/mL, which is approximately equivalent to the default MRL in a food. This value of 2.5 ppb is also in line with our previous works,7,8 where method detection limits determined by the U.S. EPA protocol were in the range 2−5 ng/g. 5681

dx.doi.org/10.1021/ac300844d | Anal. Chem. 2012, 84, 5677−5684

Analytical Chemistry

Article

Table 2. Effect of Concentrations and Matrix on the Number of Identified Compounds on 4000 Q-LITa 1.0 ppb matrix

no. ID

NM

ND

almond avocado broccoli cabbage corn ginkgo ginseng honey kale olive oil orange raisin salmon saw palmetto shrimp solvent spinach tomato

174 150 185 163 128 112 138 180 182 140 102 174 134 131 139 183 138 142

23 35 23 30 48 48 44 27 22 41 54 33 33 40 38 23 42 40

43 55 32 47 64 80 58 33 36 59 84 33 73 69 63 34 60 58

avg SD RSD

150 25 17%

10 ppb false neg, %

no. ID

NM

ND

28 38 23 32 47 53 43 25 24 42 58 28 44 45 42 24 43 41

203 195 208 205 191 175 191 203 209 195 182 205 192 185 195 209 196 189

16 12 16 12 24 31 22 23 14 21 20 17 20 27 16 19 18 24

21 33 16 23 25 34 27 14 17 24 38 18 28 28 29 12 26 27

38 11 28%

196 10 5%

100 ppb false neg, %

no. ID

NM

ND

15 19 13 15 20 27 20 15 13 19 24 15 20 23 19 13 18 21

208 214 216 210 205 204 210 209 216 212 205 209 215 204 208 214 211 204

15 7 14 14 19 14 16 20 13 15 16 11 12 16 13 12 15 19

17 19 10 16 16 22 14 11 11 13 19 20 13 20 19 14 14 17

18 4 22%

210 4 2%

false neg, % 13 11 10 13 15 15 13 13 10 12 15 13 10 15 13 11 12 15 13 2 14%

a

No. ID, number identified. NM, no match: collected spectrum did not match any reference spectrum in the library. ND, not detected: EPI experiment was not triggered Because the intensity of precursor ion was lower than the predefined threshold. No spectra were collected.

Figure 4. Trend of false negative identification rate by concentration (1, 10, and 100 ng/g) and matrix on 4000 Q-LIT (x-axis labels: almond 1 = 240 pesticides spiked into almond matrix at 1 ng/g; almond 10 = 240 pesticides spiked into almond matrix at 10 ng/g; almond 100 = 240 pesticides spiked into almond matrix at 100 ng/g; etc.).

present in the forms of spinosyns A and D and spinosyns J and L, respectively; with the A and J forms being the respective dominant components (95%).29 These four compounds eluted within 30 s under the LC separation conditions, which makes them ideal for evaluation of the sensitivity of a LC-Q-LIT system and the ability of the workflow to deliver identification without false negatives. By use of the 4000 Q-LIT system, spinosyn A was found to be at ∼10 ppm in cabbage while spinosyn D was identified at a concentration of ∼600 ppb.

vegetation samples, broccoli, cabbage, and kale, contained a total of 9 target pesticides, the same as identified by LC-MS/ MS with two MRM transition per pesticide (data not shown). Results obtained from these three samples are listed in Table 4 according to the target compounds found and instruments used. As the library collection parameters were optimized and collected from two 4000 Q-LIT systems at CFSAN and LSB, the results obtained from these two laboratories matched well. The two pesticides spinosad and spinetoram are known to 5682

dx.doi.org/10.1021/ac300844d | Anal. Chem. 2012, 84, 5677−5684

Analytical Chemistry

Article

Table 3. Pesticides Identified and Quantified in Four Incurred Foods by Use of LC-MS/MSa or LC-MRM-IDA-EPIb on the Same 4000 Q-LIT peach MS/MS (ng/g) azoxystrobin carbaryl carbendazim diazinon fenbuconazole imazalil propiconazole pyridaben teflubenzuron terbutryn thiabendazole a

orange

EPI (ng/g)

fit

8.0 ± 0.6 3.3 ± 0.9

6.0 ± 0.9 3.1 ± 0.2

93 98

49 ± 0.7

49 ± 1.1

83

14 ± 0.3 26 ± 3.1

13 ± 0.3 23 ± 3.1

MS/MS (ng/g)

salmon

EPI (ng/g)

2.3 ± 0.03

2.1 ± 0.1

fit

MS/MS (ng/g)

ginseng

EPI (ng/g)

fit

MS/MS (ng/g)

EPI (ng/g)

fit

3.4 ± 0.1

3.2 ± 0.2

98

100c

95

1.1 ± 0.1 >100c

1.3 ± 0.1 >100c

90 91

96 91 3.4 ± 0.4

3.1 ± 0.3

98

Two MRMs; abbreviated MS/MS in table. bEPI library search; abbreviated EPI in table. cOut of calibration range (1−100 ppb).

Table 4. Library Identifications and Fit Values Obtained from Three Laboratories in Three Incurred Foodsa cabbage compd found azoxystrobin cyprodinil fludioxinil imidacloprid metalaxyl spinosyn J spinosyn A spinosyn D spirotetramat

broccoli

FDA

MOE

AB 3200

AB 5500

AB 5600

90 99 97

100 74 99

93