Ultrahigh-Performance Liquid Chromatography ... - ACS Publications

Canadian Food Inspection Agency, Calgary Laboratory, 3650-36th Street N.W., Calgary, Alberta T2L 2L1, Canada. § ThermoFisher Scientific, 355 River Oa...
1 downloads 0 Views 8MB Size
Article pubs.acs.org/JAFC

Ultrahigh-Performance Liquid Chromatography Electrospray Ionization Q‑Orbitrap Mass Spectrometry for the Analysis of 451 Pesticide Residues in Fruits and Vegetables: Method Development and Validation Jian Wang,*,† Willis Chow,† James Chang,§ and Jon W. Wong# †

Canadian Food Inspection Agency, Calgary Laboratory, 3650-36th Street N.W., Calgary, Alberta T2L 2L1, Canada ThermoFisher Scientific, 355 River Oaks Parkway, San Jose, California 95134, United States # Center for Food Safety and Applied Nutrition, U.S. Food and Drug Administration, 5100 Paint Branch Parkway, College Park, Maryland 20740, United States §

ABSTRACT: This paper presents an application of ultrahigh-performance liquid chromatography electrospray ionization quadrupole Orbitrap high-resolution mass spectrometry (UHPLC/ESI Q-Orbitrap MS) for the determination of 451 pesticide residues in fruits and vegetables. Pesticides were extracted from samples using the QuEChERS (quick, easy, cheap, effective, rugged, and safe) procedure. UHPLC/ESI Q-Orbitrap MS in full MS scan mode acquired full MS data for quantification, and UHPLC/ESI Q-Orbitrap Full MS/dd-MS2 (i.e., data-dependent scan mode) obtained product ion spectra for identification. UHPLC/ESI Q-Orbitrap MS quantification was achieved using matrix-matched standard calibration curves along with the use of isotopically labeled standards or a chemical analogue as internal standards to achieve optimal method accuracy. The method performance characteristics include overall recovery, intermediate precision, and measurement uncertainty evaluated according to a nested experimental design. For the 10 matrices studied, 94.5% of the pesticides in fruits and 90.7% in vegetables had recoveries between 81 and 110%; 99.3% of the pesticides in fruits and 99.1% of the pesticides in vegetables had an intermediate precision of ≤20%; and 97.8% of the pesticides in fruits and 96.4% of the pesticides in vegetables showed measurement uncertainty of ≤50%. Overall, the UHPLC/ESI Q-Orbitrap MS demonstrated acceptable performance for the quantification of pesticide residues in fruits and vegetables. The UHPLC/ESI Q-Orbitrap Full MS/dd-MS2 along with library matching showed great potential for identification and is being investigated further for routine practice. KEYWORDS: UHPLC/ESI Q-Orbitrap, high-resolution mass spectrometer, pesticides, fruits and vegetables, quantification, identification, measurement uncertainty



INTRODUCTION

Chemical Residues Monitoring Program and Food Safety Action Plan by the Canadian Food Inspection Agency.5 Determining trace levels of pesticide residues and screening for a large number of pesticides in various food commodities remain as constant challenges for analytical chemists. Improved multiclass or multiresidue methodologies with high sensitivity and expanded scopes, which include as many pesticides and commodities as possible in a single method, are required for checking compliance or for studying risk assessment of consumer exposure to pesticides. Pesticides in foods are traditionally determined using gas chromatography (GC) coupled with detectors such as electron capture and mass spectrometer (MS). GC-MS continues to be a key tool in pesticide analysis because it is inexpensive and easy to operate and satisfies the required sensitivity and selectivity for both quantification and identification. However, new generations of pesticides are not amenable to GC-MS due to their thermal instabilities and polarities.6 A practical technique that solves this

Pesticides have been widely used in various combinations at different stages of cultivation and during postharvest storage. They protect crops against a range of pests and fungi and provide quality preservation. Pesticide residues, which might pose a potential risk for human health due to their subacute and chronic toxicity, could possibly remain in crops such as fruits and vegetables. It is important to control or regulate the uses of pesticides in crop production and to monitor their levels for compliance so as to ensure the safety of the food supply. Canada and the United States as well as national or international bodies such as the European Union1,2 and Codex3 have set regulations for monitoring programs and have conducted health risk assessment of pesticide residues in food and established maximum residue limits (MRLs) for domestic use and international trade of foods. In Canada, prior to the registration of a pesticide, Health Canada determines the allowable/safe amount of pesticide residues in food for human consumption and then sets science-based MRLs to ensure the Canadian food supply is safe.4 Many food commodities such as fruits and vegetables, infant foods, teas, grains, and pulses have been tested for pesticide residues under the Canadian National © XXXX American Chemical Society

Received: August 5, 2014 Revised: September 24, 2014 Accepted: September 29, 2014

A

dx.doi.org/10.1021/jf503778c | J. Agric. Food Chem. XXXX, XXX, XXX−XXX

Journal of Agricultural and Food Chemistry

Article

Table 1. Pesticides, Exact Mass, and UHPLC Retention Time for Data Processing and Quantification and UHPLC/ESI QOrbitrap MS Method Performance Results

B

dx.doi.org/10.1021/jf503778c | J. Agric. Food Chem. XXXX, XXX, XXX−XXX

Journal of Agricultural and Food Chemistry

Article

Table 1. continued

C

dx.doi.org/10.1021/jf503778c | J. Agric. Food Chem. XXXX, XXX, XXX−XXX

Journal of Agricultural and Food Chemistry

Article

Table 1. continued

D

dx.doi.org/10.1021/jf503778c | J. Agric. Food Chem. XXXX, XXX, XXX−XXX

Journal of Agricultural and Food Chemistry

Article

Table 1. continued

E

dx.doi.org/10.1021/jf503778c | J. Agric. Food Chem. XXXX, XXX, XXX−XXX

Journal of Agricultural and Food Chemistry

Article

Table 1. continued

F

dx.doi.org/10.1021/jf503778c | J. Agric. Food Chem. XXXX, XXX, XXX−XXX

Journal of Agricultural and Food Chemistry

Article

Table 1. continued

G

dx.doi.org/10.1021/jf503778c | J. Agric. Food Chem. XXXX, XXX, XXX−XXX

Journal of Agricultural and Food Chemistry

Article

Table 1. continued

a

Numbers or text in bold font and underlined indicates ionization form or charge state for data processing or quantification. bExact mass or measured accurate mass was used for quantification. The electron mass (0.000549 amu) is subtracted when exact mass is calculated. cBold and underlined indicate pesticides with recoveries not in the range of 81−110%. dBold and underlined indicate pesticides with intermediate precision >20%. eBold and underlined indicate pesticides MU > 50%. fFor data in red font, the method performance was based on three spike levels, i.e., 90.0, 240.0, and 400.0 μg/kg, due to its poor sensitivity. For data in blue font, the method performance was based on two spike levels, i.e., 240.0 and 400.0 μg/kg, due to its poor sensitivity. gColumn number.

analytes are found, they are further identified using Q-Orbitrap MS/MS and QqTOF MS/MS. This paper discusses the development and validation of a method that can rapidly quantify and accurately identify 451 pesticides in fruits and vegetables at low (∼10) ppb concentration levels using an UHPLC/ESI Q-Orbitrap along with the QuEChERs extraction procedure. The UHPLC/ESI Q-Orbitrap MS operated in full MS scan mode for screening and quantification was evaluated, and the UHPLC/ESI QOrbitrap Full MS/dd-MS2 (i.e., data-dependent scan) mode for identification was demonstrated. The UHPLC/ESI Q-Orbitrap mass spectrometer proved to be a very promising and powerful tool for the determination of pesticide residues in fruits and vegetables. This is the first validated UHPLC/ESI Q-Orbitrap method to quantify 451 pesticide residues in fruits and vegetables. The method proved to be robust and sensitive enough to determine pesticide residues at low ppb concentration levels. The sample extraction was also quick and simple enough to allow for high-throughput testing of routine samples, which will greatly benefit monitoring programs while satisfying regulatory purposes.

problem is liquid chromatography−mass spectrometry (LCMS), which has been widely used to quantify LC-amenable (or thermally labile) pesticides and to confirm their identities in fruits and vegetables at low parts-per-billion (ppb) concentration levels.7,8 Liquid chromatography coupled with quadrupole tandem mass spectrometry with electrospray ionization (LC/ESI-MS/ MS) as an interface is the most commonly used LC-MS system for pesticide residue analysis. This system configuration allows for increased selectivity and sensitivity when operated in multiple-reaction monitoring (MRM) mode. Current advances in mass spectrometers have made it possible to determine hundreds of targeted pesticides in a single analysis. However, LC-MS/MS requires extensive compound-dependent parameter optimization. Alternatively, an LC can be coupled to full scan mass spectrometers such as Orbitrap and time-of-flight (TOF) mass spectrometers. These types of MS have been increasingly used for quantification, identification, characterization, and structural elucidation of pharmaceuticals, pesticides, veterinary drugs, unknown contaminants, and their transformation or degradation products in foods and environmental samples.9−12 The Orbitrap and TOF mass spectrometers (recently advanced models) offer high resolution (>20000 fwhm), accurate mass measurement (70000 fwhm for routine practice and still be sensitive enough to quantify at trace ppb concentration levels. The resolving power and accurate mass measurement capability of the Q-Orbitrap (99.0%), and LC-MS acetonitrile (Chromasolv, 2.5 L) were purchased from Sigma-Aldrich Corp (Canada). ENVIRO CLEAN MYLAR pouches (6.0 g of anhydrous magnesium sulfate (MgSO4) and 1.5 g of anhydrous sodium acetate) and ENVIRO CLEAN extraction columns (900 mg of MgSO4, 150 mg of C18, and 300 mg of primary−secondary amine (PSA), 15 mL centrifuge tubes) were purchased from United Chemical Technologies, Inc. (Bristol, PA, USA). Acetic acid (glacial acetic acid, reagent grade, 99.7%), acetonitrile (distilled in glass), and methanol (distilled in glass) were obtained from Caledon Laboratories H

dx.doi.org/10.1021/jf503778c | J. Agric. Food Chem. XXXX, XXX, XXX−XXX

Journal of Agricultural and Food Chemistry

Article

Ltd. (Canada). Water (18.2 MΩ·cm) used for reagent and sample preparation was obtained from a Barnstead Nanopure system (Thermo Scientific, USA). Pesticide standards (Table 1, column 1) were obtained from EQ Laboratories Inc. (USA), Riedel-de Haen AG (Germany), or Chem Service (USA). Internal standards carbendazimd4 and carbofuran-d3 were purchased from EQ Laboratories Inc. (USA), and thiabendazole-d4 was from Chemical Synthesis Services (Northern Ireland). LC vials were Mini-UniPrep syringeless filter devices with polypropylene housing and PVDF 0.45 μm membrane (Whatman Inc., USA). Preparation of Standard Solutions. Individual pesticide standard stock solutions were generally prepared at a concentration of 4000.0 μg/mL in methanol/acetonitrile (50 + 50, v/v). Due to their poor solubility in methanol, carbendazim was prepared at 200.0 μg/ mL, and some pesticides were prepared at 1000.0 or 2000.0 μg/mL. Intermediate pesticide standard mix working solutions were prepared at 10.0 or 15.0 μg/mL in methanol/acetonitrile (50 + 50, v/v), from stock solutions. Because of the large number of pesticides (>450), intermediate solutions were prepared in separate 200 mL (15 μg/mL, including 210 stocks) and 250 mL (10 μg/mL, including 241 stocks) volumetric flasks. Stock and intermediate solutions were stored at −20 °C. A six-level pesticide standard mix working solution was prepared by transferring 0.15, 0.75, 3.0, 6.0, 9.0, and 15.0 mL of 10.0 μg/mL and 0.1, 0.5, 2.0, 4.0, 6.0, and 10.0 mL of 15.0 μg/mL intermediate working solution into six separate 50 mL volumetric flasks for their respective concentration levels and then making up to volume with methanol. The resulting concentrations were 0.03, 0.15, 0.60, 1.20, 1.80, and 3.00 μg/mL, which were used to construct the matrix-matched standard calibration curves. Four-level sample spike pesticide standard working solutions were prepared by transferring 1.5, 13.5, 36.0, and 60.0 mL of 10.0 μg/mL and 1.0, 9.0, 24.0, and 40.0 mL of 15.0 μg/mL intermediate working solution into four separate 100 mL volumetric flasks and making up to volume with methanol for their respective concentration levels. The resulting concentrations were 0.15, 1.35, 3.60, and 6.00 μg/mL, which were used for sample fortification. Internal standard working solutions (5.0 and 0.5 μg/mL) including carbofuran-d3, carbendazim-d4, and thiabendazole-d4 were prepared in a mixture of methanol/acetonitrile (50 + 50, v/v). All working solutions were stored at 4 °C. Preparation of Reagent Solutions. Acetonitrile/acetic acid (99 + 1, v/v) was prepared by mixing 990 mL of acetonitrile with 10 mL of acetic acid. Ammonium acetate (0.1 M) was prepared by weighing 7.7 g of ammonium acetate and dissolving it in 800 mL of water. After transfer into a 1000 mL volumetric flask, the solution was made up to volume with water. Solvent buffer was a mixture of 0.1 M ammonium acetate and methanol (50 + 50, v/v). UHPLC/ESI Q-Orbitrap Parameters. The UHPLC/ESI QOrbitrap system consisted of an Accela 1250 LC pump and an Accela open autosampler coupled with a Q Exactive mass spectrometer (ThermoFisher Scientific, Germany). The system was controlled by Xcalibur 2.2 software. (a). UHPLC. UHPLC mobile phase A was 4 mM ammonium formate and 0.10% formic acid in water, and mobile phase B was 4 mM ammonium formate and 0.10% formic acid in methanol. The UHPLC column utilized was a Hypersil Gold, 100 mm × 2.1 mm, 1.9 μm column (Thermo Scientific, USA). The UHPLC guard column was an Accucore aQ 10 × 2.1 mm, 2.6 μm Defender cartridge (Thermo Scientific, USA). The UHPLC gradient profile and flow rate are shown in Figure 1. The column oven temperature was set at 45 °C, and the autosampler temperature was set at 5 °C. The injection volume was 5 μL, and the total run time was 14 min. (b). Q-Orbitrap Parameters. The Q-Exactive ion source was equipped with a heated electrospray ionization (HESI) probe, and the Q-Orbitrap was tuned and calibrated using positive LTQ calibration solution once a week. The Q-Exactive was operated in either full MSSIM or full MS/dd-MS2 positive mode. In full MS-SIM, the Q-Oritrap performed full MS scan without high-energy collision dissociation (HCD) fragmentation. The full MS scan range was set from m/z 80 to 1100 (0−12.0 min). The mass resolution was set to 70000 fwhm at m/ z 200, and the instrument was tuned for maximum ion throughput.

Figure 1. UHPLC gradient profile and 451 pesticides elution distribution. AGC (automatic gain control) target or the number of ions to fill CTrap was set at 1.0 × 106 with a maximum injection time (IT) of 250 ms. All quantitative data in this study were acquired using full MS-SIM mode. Targeted identification was achieved by full scan MS, and if a targeted pesticide was present, its precursor ion scan, provided by an inclusion list, triggered a data-dependent MS2 (dd-MS2) scan. During full MS scan, the mass resolution was set at 70000 fwhm, AGC target at 1.0 × 106, maximum IT 250 ms, and scan range from m/z 80 to 1100. If the targeted compound was detected within the 10 ppm mass error window and achieved by a designated intensity threshold (i.e., setting of 1.7 × 105), the precursor ion in the inclusion list was then isolated by the quadrupole and sent to the HCD collision cell for fragmentation via the C-trap. The inclusion list consists of precursor ions that are of interest for targeted identification and is provided in Table 1, columns 3−5. The precursor ion was fragmented with stepped normalized collision energy (NCE) to generate the resulting dd-MS2 product ion spectrum. At this stage, the mass resolution of the Orbitrap analyzer was set at 35000 fwhm, AGC target at 2 × 105, maximum IT at 120 ms, and isolation window at 1.0 m/z, NCE at 40 ± 50%, underfill ratio at 10%, intensity threshold at 1.7 × 105, apex trigger at 2−4 s, and dynamic exclusion at 10.0 s. Other Q-Exactive generic parameters were sheath gas flow rate set at 60, Aux gas flow rate at 30, sweep gas flow rate at 2, spray voltage (kV) at 3.50, capillary temperature (°C) at 350, S-lens level at 55.0, and heater temperature (°C) at 350. Sample Preparation and Extraction Procedure. Sample extraction and cleanup procedures followed the buffered QuEChERS14 or AOAC Official Method 2007.0115 with a slight modification. For the fortification experiment, fruit and vegetable samples (15.0 g/ sample) were weighed into individual 50 mL polypropylene centrifuge tubes (VWR International, Canada). One milliliter per four-level sample spike pesticide standard working solution(s) was added into four centrifuge tubes to provide 10.0, 90.0, 240.0, and 400.0 μg/kg of pesticide equivalent in samples, followed by the addition of 300 μL of 5.0 μg/mL internal calibration standard working solution (100.0 μg/kg equivalent in samples). The centrifuge tubes were capped, mixed, and left to stand for 15 min at room temperature. Then, 14 mL of acetonitrile/acetic acid (99 + 1, v/v) mixture was added to individual samples and mixed for 45 s, followed by adding 1.5 g of anhydrous sodium acetate and 6.0 g of anhydrous magnesium sulfate from an ENVIRO CLEAN MYLAR pouch. One milliliter of organic solvent (methanol/acetonitrile, 50 + 50, v/v), resulting from the addition of the pesticide standard working solutions, was accounted for and added I

dx.doi.org/10.1021/jf503778c | J. Agric. Food Chem. XXXX, XXX, XXX−XXX

Journal of Agricultural and Food Chemistry

Article

Figure 2. UHPLC/ESI Q-Orbitrap Full MS/dd-MS2 chromatograms and product ion spectra. A mixture of 509 pesticides (50 μg/L in solvent buffer) and single O-Phenylphenol (50 μg/L in solvent buffer) were injected and analyzed: (left) Mixture of 509 pesticides; (right) O-Phenylphenol; (A1, A2) extracted ion chromatograms of m/z 171.08044 with a mass tolerance of 5 ppm, if detected; (B1, B2) dd-MS2 chromatograms of m/z 171.08044, if detected; (C1, C2) dd-MS2 product ion spectra of m/z 171.08044, if triggered at 8.39 min; (D1, D2) dd-MS2 product ion spectra of m/z 171.08044, if triggered at 10.16 min. general, a quadratic function was applied to the calibration curves based on the line of best fit. Occasionally, linear regression may be used for quantification. The 1/x weighting was used to improve the accuracy for quantification of pesticides at low concentrations. Responses for the unknown concentration or fortified samples were compared to the curves to calculate the amount of pesticide residues (μg/kg (ppb)) in samples. Matrix-matched calibration standards were prepared fresh for each batch of samples. Experimental Design and Method Validation. The method was validated according to the nested experimental design, which was described elsewhere.16 The main factors of variances associated with the method performance or measurement uncertainties of an in-house validated method were concentrations or spiked levels of analytes, matrix effects, day-to-day variation, and within-day variation of the method. The last two factors are designated the intermediate precision. In this study, there were a total of five fruits (i.e., apple, banana, grape, orange, and strawberry) and five vegetables (i.e., carrot, potato, tomato, broccoli, and lettuce). For each matrix, samples were spiked at 10.0, 90.0, 240.0, and 400.0 μg/kg, in triplicate. Spike experiments were repeated on two different days or by two analysts. Overall recovery, intermediate precision, and measurement uncertainty were calculated using a combined computer program that consisted of SAS codes (SAS Software release 9.3, SAS Institute Inc., USA) along with a Microsoft Excel (Microsoft Office 2010) workbook.

up to a total of 15 mL organic solvent for extraction. The centrifuge tubes were capped, shaken at 1500 rpm using a Geno/Grinder 2010 (SPEX SamplePrep, USA) for 1 min, and then centrifuged at 3000 rpm (∼2100g) for 3 min using an Allegra 6 centrifuge (Beckman Coulter Inc., USA). Supernatants were transferred (7 mL/sample) into individual 15 mL polypropylene centrifuge tubes or ENVIRO CLEAN extraction columns that contain 900 mg of MgSO4, 150 mg of C18, and 300 mg of PSA for cleanup. The centrifuge tubes were capped, shaken by hand for 45 s, and centrifuged at 3000 rpm (∼2100g) for 3 min. Ninety microliters of each sample extract and 410 μL of solvent buffer were transferred into separate Mini-UniPrep vials (Whatman Inc., USA). The vials were capped, vortexed for 30 s, and pressed to filter the solution. Sample extracts were ready for UHPLC/ESI Q-Orbitrap injection. Preparation of Matrix-Matched Calibration Standards and Calculation. Matrix-matched calibration standards were prepared by adding standards and internal standards to blank sample extracts after sample extraction and cleanup. A blank fruit or vegetable sample (15.0 g/sample) was weighed into a 50 mL centrifuge tube, and the sample was processed through the extraction procedure as described above (but using 15 mL of acetonitrile/acetic acid (99 + 1, v/v) mixture for extraction). After 90 μL of blank sample extracts and 377 μL of solvent buffer had been transferred into each of six Mini-UniPrep vials (Whatman Inc., USA), 15 μL of each six-level pesticide standard mix working solution and 18 μL of 0.50 μg/mL internal calibration working solution were added to the individual vials, providing 5.0, 25.0, 100.0, 200.0, 300.0, and 500.0 μg/kg per standard and 100.0 μg/ kg per internal standard equivalent in samples. The vials were capped, vortexed for 30 s, and pressed to filter the solution. Sample extracts were ready for UHPLC/ESI Q-Orbitrap injection. Matrix-matched standard calibration curves for each individual pesticide were constructed using TraceFinder 3.1 (optimized for Environmental and Food Safety) software. Concentration (μg/kg (ppb)) versus the ratio (analyte area/IS area) of each individual pesticide was plotted. Deuterium-labeled standards carbendazim-d4, carbofuran-d3, and thiabendazole-d4 were used as internal standards for their respective native compounds for quantification. Other pesticides used carbofuran-d3 as an internal standard for quantification. In



RESULTS AND DISCUSSION QuEChERs. Pesticides were extracted from fruits and vegetables (15 g/sample) following the buffered QuEChERS method14 or AOAC Official Method 2007.01.15 The whole procedure consisted of two steps including (1) extraction with acetonitrile containing 1% acetic acid, MgSO4, and sodium acetate or (2) cleanup by dispersive solid-phase extraction (dSPE) using MgSO4, PSA, and C18. Previous methods used ChloroFiltr, a polymeric-based sorbent used in the cleanup of chlorophyll-rich matrices such as lettuce and broccoli.14 To simplify the method and avoid confusion, the d-SPE that J

dx.doi.org/10.1021/jf503778c | J. Agric. Food Chem. XXXX, XXX, XXX−XXX

Journal of Agricultural and Food Chemistry

Article

Table 2. Isobaric Pesticides and Their Exact Mass, UHPLC Retention Time, and UHPLC/ESI Q-Orbitrap MS Method Performance

a Number or text in bold font and unerlined indicates ionization form or charge state for data processing or quantification. bExact mass or measured accurate mass was used for quantification. The electron mass (0.000549 amu) is subtracted when exact mass is calculated. cBold and underlined indicate pesticides with recoveries not in the range of 81−110%. dBold and underlined indicate pesticides intermediate precision >20%. eBold and underlined indicate pesticides MU > 50%. fFor data in red font, the method performance was based on three spike levels, i.e., 90.0, 240.0, and 400.0 μg/kg, due to its poor sensitivity. For data in blue font, the method performance was based on two spike levels, i.e., 240.0 and 400.0 μg/kg, due to its poor sensitivity. gColumn number.

UHPLC. Four hundred and fifty-one pesticides and three isotopically labeled standards (Table 1, column 1) were chromatographically separated within 12 min under a gradient profile (Figure 1A) using a Hypersil Gold (100 mm × 2.1 mm, 1.9 μm) column with methanol and water, each containing 4 mM ammonium formate and 0.10% formic acid, as mobile phases. Initially, a Waters Acquity UPLC BEH C18 (100 mm × 2.1 mm, 1.7 μm) column with acetonitrile and water (10 mM ammonium acetate)14 was chosen for the study. However, higher detection limits were observed for some of the compounds. To improve sensitivity, different LC methods were compared, and then an UHPLC method, which used Hypersil Gold (100 mm × 2.1 mm, 1.9 μm) column, was adopted.17 The Hypersil Gold (100 mm × 2.1 mm, 1.9 μm) column generally provided better chromatographic resolution

contained 900 mg of MgSO4, 150 mg of C18, and 300 mg of PSA was used for all matrices in the cleanup step for this method. An experiment (data not shown) proved that no obvious differences regarding matrix effects were observed between d-SPEs with or without ChloroFiltr. In addition, sample extracts were diluted 5.56 times (90 μL of extracts + 410 μL of solvent buffer). The Mini-UniPrep vials, consisting of a polyvinylidene difluoride (PVDF) membrane, filtered some of the light green coextractives or precipitates in the final extracts after sample extract dilution prior to UHPLC/ESI Q-Orbitrap injection. Therefore, the QuEChERS method without ChloroFiltr proved to be a simple, practical, and fast extraction procedure to achieve UHPLC/ESI Q-Orbitrap quantification of pesticide residues at trace level in fruits and vegetables. K

dx.doi.org/10.1021/jf503778c | J. Agric. Food Chem. XXXX, XXX, XXX−XXX

Journal of Agricultural and Food Chemistry

Article

Figure 3. UHPLC/ESI Q-Orbitrap Full MS/dd-MS2 chromatograms and product ion spectra. A mixture of 451 pesticides (50 μg/L in solvent buffer) was injected and analyzed: (A) extracted ion chromatogram of m/z 226.16624 with a mass tolerance of 5 ppm; (B) dd-MS2 chromatogram of m/z 226.16624; (C) dd-MS2 product ion spectrum of m/z 226.16624 at 6.53 min; (D) dd-MS2 product ion spectrum of m/z 226.16624 at 6.65 min.

Figure 4. UHPLC/ESI Q-Orbitrap Full MS/dd-MS2 chromatograms and product ion spectra. Individual secbumeton or terbumeton (50 μg/L in solvent buffer) was injected and analyzed: (left) secbumeton; (right) terbumeton; (A1, A2) extracted ion chromatogram of m/z 226.16624 with a mass tolerance of 5 ppm; (B1, B2) dd-MS2 chromatogram of m/z 226.16624; (C1, C2) dd-MS2 product ion spectra of m/z 226.16624 at 6.43 and 6.54 min, respectively; (D1, D2) dd-MS2 product ion spectra of m/z 226.16624 at 6.61 and 6.71 min, respectively.

and higher responses compared to an Acquity UPLC BEH C18 column for improved sensitivity. The gradient profile also was modified so that the elution of pesticides was optimized over the 12 min run time (Figure 1B). Under the modified gradient profile, 79.6% of pesticides were eluted between 6 and 10 min or 91.4% between 4 and 10 min. The span of the retention time

distribution reduced the number of concurrent pesticides and provided sufficient time for the dd-MS2 experiment between full MS scans. Under most circumstances, an extracted ion presented as a single LC peak or baseline separation from other peaks, which was a result of superior resolving power of both UHPLC and L

dx.doi.org/10.1021/jf503778c | J. Agric. Food Chem. XXXX, XXX, XXX−XXX

Journal of Agricultural and Food Chemistry

Article

Figure 5. UHPLC/ESI Q-Orbitrap Full MS/dd-MS2 chromatograms and product ion spectra. Individual secbumeton or prometon (50 μg/L in solvent buffer) was injected and analyzed: (left) secbumeton; (right) prometon; (A1, A2) extracted ion chromatograms of m/z 226.16624 with a mass tolerance of 5 ppm; (B1, B2) dd-MS2 chromatograms of m/z 226.16624; (C1, C2) dd-MS2 product ion spectra of m/z 226.16624 at 6.43 and 6.46 min, respectively; (D1, D2) dd-MS2 product ion spectra of m/z 226.16624 at 6.61 and 6.63 min, respectively.

UHPLC chromatographic resolution. However, this assumption did not account for impurities present in the mixture, which possibly led to false identification, and therefore the injection of each individual pesticide became necessary. Figure 2 showed an example of potential misidentifications when a mixture of pesticides was injected into the system during the method development stage. O-Phenylphenol has a m/z 171.08044, if protonated. In full MS scan, two peaks were found at 8.37 and 10.19 min (Figure 2A1), respectively, for m/z 171.08044 with a mass tolerance of 5 ppm, and dd-MS2 was triggered twice at 8.39 and 10.16 min (Figure 2B1), resulting in product ion spectra as shown in Figure 2C1,D1. To identify the O-Phenylphenol peak, its neat standard (50 ppb in solvent) was injected into the system. No peaks were detected to trigger ddMS2 for O-Phenylphenol (Figure 2A2,B2,C2,D2), indicating the peaks were results of chemical impurities. Therefore, it was required that individual pesticide standards were injected in full MS/dd-MS2 mode to obtain both retention times and product ion spectra during the method development stage. This step proved to be critical because it provided data for the identification of individual pesticides in a mixture or sample matrices and for generating exact mass spectral library in future. Another challenge was to identify or differentiate isomeric compounds in the mixture of pesticides. Table 2 lists 22 pairs or groups of isomers among 451 pesticides that required both LC retention times (LC resolution) and dd-MS2 product ion spectra (MS fragmentation) for identification and differentiation. For example, secbumeton, terbumeton, and prometon are isomers and share the same precursor ion ([M + H]+) with m/z 226.16624. When a mixture of standards was injected, the three compounds eluted at 6.54 and 6.62 min (Figure 3A), and dd-MS2 was triggered twice at 6.53 and 6.65 min,

Q-Orbitrap MS, with the exception of coeluting isomeric compounds. Pesticides were eluted between 1.0 and 11.0 min, and their peak shapes were of Gaussian distribution with a baseline peak width of 5−10 s. The retention times were reproducible under ±0.2 min within and between batches for most of the pesticides. The scan rate of 3 Hz with the resolution, set at 70000 fwhm at m/z 200, is more than adequate to generate sufficient data points for quantification. For example, at least 18 data points across the chromatographic peak were generated with a 6 s (0.1 min) baseline peak width. Q-Orbitrap MS. When operated in its full MS-SIM mode in a range from m/z 80.0 to 1100.0 with 70000 fwhm at m/z 200, the Q-Orbitrap acquired full MS scan data for the screening and quantitation of the 451 pesticides listed in Table 1. When operated in full MS/dd-MS2 mode, the Q-Orbitrap acquired product ion spectra with accurate mass measurement for identification according to a list of targeted exact masses (Table 1, columns 3−5). In the current study, “stepped” normalized collision energy (NCE) was used for dd-MS2 fragmentation. The Q-Orbitrap performed a three-step 40 ± 50% NCE (i.e., the center energy 40 NCE; plus one above, 60 NCE; and one below, 20 NCE) fragmentation on the precursor ion. All fragments created in the three steps were collected sequentially in the HCD and sent together to the Orbitrap analyzer for single-scan detection. Stepped NCE worked for most but not all of the pesticides because some pesticides are not optimally fragmented in the 40 ± 50% NCE range. At the initial stage of method development, we prepared a mixture of 509 pesticides and injected it into the system. It was thought that individual pesticides in the mixture would have been identified simply on the basis of high mass-resolving power and accurate mass measurement in conjunction with M

dx.doi.org/10.1021/jf503778c | J. Agric. Food Chem. XXXX, XXX, XXX−XXX

Journal of Agricultural and Food Chemistry

Article

Figure 6. UHPLC/ESI Q-Orbitrap MS matrix effects. The 451 pesticides were prepared in sample extracts (a total of 10 matrices) and solvent buffer at a concentration of 100 μg/kg equivalent in sample: (left) fruits; (right) vegetables.

respectively (Figure 3B). The corresponding product ion spectra are shown in Figure 3C,D, and the observed differences revealed a product ion at m/z 184.11899 and relative ion intensities. To identify and differentiate the three compounds, individual standards were injected, and chromatograms and mass spectra are shown in Figures 4 and 5. Secbumeton (6.52

min) and terbumeton (6.62 min) were chromatography resolved (Figure 4A1,A2) but shared the same major fragment ions with different relative intensities (Figure 4C1,D1,C2,D2). Secbumeton (6.52 min) and prometon (6.53 min) coeluted (Figure 5A1,A2) but possessed different fragment ions (m/z 170.10349 and 184.11919, Figure 5, panels C1 and D1 and C2 N

dx.doi.org/10.1021/jf503778c | J. Agric. Food Chem. XXXX, XXX, XXX−XXX

Journal of Agricultural and Food Chemistry

Article

Figure 7. UHPLC/ESI Q-Orbitrap MS method performance for analysis of 451 pesticides in fruits and vegetables: (left) fruits; (right) vegetables; (A) overall recovery; (B) intermediate precision; (C) measurement uncertainty.

Matrix-matched standard calibration curves along with isotopically labeled standards were required to compensate for matrix effects and were used to assess the UPLC/ESI QOrbitrap MS quantitative accuracy. Three commercially available deuterium-labeled standards, that is, carbendazim-d4, carbofuran-d3, and thiabendazole-d4, were used as internal standards for quantifying their respective native compounds, and carbofuran-d3 was utilized for all other pesticides. The calibration curves were observed to be linear or quadratic with coefficient of determinations (R2) ≥ 0.97. Because of matrix effects, ion source contamination, or other unidentified factors, the responses of some pesticides either decreased or increased slightly over time. To average out the response changes during the course, the matrix-matched standard calibration curves were constructed on the basis of the two injections, that is, before and after spike samples, so as to improve the method performance. Quantification and Method Performance. The UHPLC/ESI Q-Orbitrap MS method was validated according to a nested design, reported elsewhere, to evaluate the method performance characteristics including accuracy expressed as overall recovery, intermediate precision, and measurement uncertainty (MU).16 Four factors, concentrations or spiked levels of pesticides, matrix effects, day-to-day variation, and within-day variation, were considered for the evaluation, and the experimental details were described under Materials and Methods. Data were grouped into two sets, that is, fruits and

and D2, respectively). Prometon (6.53 min) and terbumeton (6.62 min) were chromatographically resolved (Figures 4A2 and 5A2) and possessed different fragment ions (m/z 184.11919 and 170.10338, Figures 5C2,D2 and 4C2,D2, respectively). Therefore, dd-MS2 was required to differentiate between secbumeton and prometon. Among the 22 pairs or groups of isomers, only 5 pairs were not chromatographically resolved, and dd-MS2 was necessary for identification when detected in full MS mode. Matrix Effects. Sample matrix could either enhance or suppress the ionization of pesticides; its effects might vary from sample to sample and ultimately affect the UHPLC/ESI QOrbitrap MS quantitative results. To evaluate matrix effects, the responses of pesticides in sample extracts were compared to those pesticide standards prepared in solvent buffer at 100 μg/ kg equivalent in sample. In fruit matrices (Figure 6), pesticides experienced fewer matrix effects in apple, banana, and grape than in orange and strawberry. About 90% pesticides had ion suppression 20% of pesticides in potato, tomato, and lettuce were subjected to ion enhancement (≥10%), whereas >10% of pesticides in carrot and broccoli experienced ion suppression (>20%). O

dx.doi.org/10.1021/jf503778c | J. Agric. Food Chem. XXXX, XXX, XXX−XXX

Journal of Agricultural and Food Chemistry



vegetables, so that statistics were valid. The method performance results are summarized in Tables 1 and 2 and are illustrated in Figure 7. Depending on the type of sample matrices, about 94.5% of the pesticides in fruits or 90.7% in vegetables had recoveries between 81 and 110%; 99.3% of the pesticides in fruits and 99.1% of the pesticides in vegetables had an intermediate precision ≤20%; and 97.8% of the pesticides in fruits or 96.4% of the pesticides in vegetables showed measurement uncertainty ≤50%, which was a recommended default value in EU Document No. SANCO/12495/2011 for pesticide analysis and enforcement decisions (MRL exceedances).18 Although five pairs of isomeric compounds that were not chromatographically resolved (Table 2) had very good method performance, they needed to be quantified separately when found in samples. Overall, the UHPLC/ESI Q-Orbitrap MS can serve as a liable and practical tool for quantification. Pesticide Identification. The identification of any pesticides using UHPLC/ESI Q-Orbitrap mass spectrometry was based on mass accuracy and chromatographic retention time. SANCO/12495/201118 required ≥2 diagnostic ions (preferably the precursor ion and its product ion) with mass accuracy