Use of a deactivated PTV injector liner and GCMS/MS for the quantitative determination of multiple pesticide residues in fruit and vegetables.

The quantitative determination of multiple pesticide residues in food is an iterative process given the frequent changes in monitoring specifications set by regulatory authorities, introduction of new pesticide active ingredients, variety of commodities encountered and advances in the capability of analytical instrumentation and software platforms. The method described here:•replaces our previous methodology [1] that was based on an ethyl acetate extraction [2], two different sample extract clean-up regimes depending on the commodity; either Gel Permeation Chromatography (GPC) or Solid Phase Extraction (SPE) and GC/MSMS analysis using cool on-column injection and permits higher throughput using the same QuEChERS extraction method used for LCMS/MS analysis [3]•uses PTV injection incorporating a deactivated (baffled) injection liner required to improve performance for 'difficult to analyse' pesticides e.g. captan, dichlofluanid, folpet, tolylfluanid.•has been validated for the quantitative determination of 113 different pesticides and their metabolites in a range of fruit and vegetables of high water content and high acid and high water content i.e. cabbage, lemon, pepper, plum and spinach and complies with requirements of European Commission guidance document on Analytical Quality Control and Method Validation Procedures for Pesticides Residues Analysis in food and feed - SANTE/12682/2019 [4].

Individual stock standard solutions were prepared at ~400 μg mL −1 in methanol and combined to prepare mixed standard solutions at approximately 500 μg mL −1 in ethyl acetate. These were used to prepare solvent standard mixtures in ethyl acetate of lower concentration. A range of matrixmatched calibration standards was subsequently prepared by admixing aliquots of solvent standard mixtures and matrix extract to ensure a matrix concentration of 1 g mL −1 was maintained. The matrix-matched pesticide standards were prepared to cover a concentration range that encompassed the specific analyte/commodity reporting level (RL) set by UK regulatory authorities, typically 0.5RL, 1RL, 2RL, 5RL and 10RL concentrations and included 50 μL of an injection internal standard (ISTD) and 30 μL of an analyte protectants (AP) mix for every 1 mL of matrix-matched standard. The ISTD contained pp-DDE D8 and trifluralin D14 in ethyl acetate at 2 μg mL −1 . The AP mix contained Dsorbitol and shikimic acid in acetone at 5 mg mL −1 and was included to assist with unwanted tailing and decomposition of 'difficult to analyse' pesticides in the GC inlet [5] . Matrix-matched standards of higher concentration were included if necessary. Reporting levels were typically 0.01 mg kg −1 with a few pesticides with RLs of 0.02 or 0.05 mg kg −1 .
For validation purposes, blank matrix was fortified from appropriate mixed solvent standards to generate 6 replicate 'recovery' samples at concentrations corresponding to 1 RL and 2 RL ( ≡ 10 and Table 1 Details of sorbents/salts used for dSPE clean-up for each commodity.

Spiked samples n = 6 (1 rl and 2 RL)
Portions of a suitable organic blank (10 ± 0.1 g) were weighed into centrifuge tubes (50 mL) and spiked with 100 μL of solvent standard mixture (100RL) for 1RL spikes and 200 μL of solvent standard mixture (100RL) for 2RL spikes prior to extraction using method detailed below.

Preparation of blank matrix
Portions of organic blank (10 ± 0.1 g) were weighed into centrifuge tubes (50 mL) and extracted and cleaned-up using the method detailed below. A suitable volume of dSPE blank extract (typically 10 mL) was combined and solvent exchanged into 10 mL ethyl acetate so that a matrix concentration of 1 g mL −1 was maintained.

Sample provenance and preparation
Samples were received as part of the 2019 UK/EU Pesticide Residues in Food (PRiF) annual surveillance programmes [6] and/or the corresponding UK [7] and EU [8] proficiency testing (PT) schemes. Organic fruit and vegetables were purchased from local retailers, determined to be free from pesticides targeted in this study and subsequently used as 'blank' matrix in order to prepare matrixmatched standards and fortified (spiked) samples.
Upon receipt, samples were uniquely registered, appropriately sub-sampled and stored in accordance with EC Regulation 396/2005 Annex 1 [ 9 ]. Ultimately, all samples including blank matrix samples were cryogenically milled using 'dry-ice' and frozen at −20 °C until the point of extraction.

Sample extraction
10 g ± 0.1 g of sample was weighed into a plastic centrifuge tube and shaken for at least 1 min. This minimum shaking time was important to ensure that adequate extraction efficiency was obtained. The contents of one QuEChERS EN15662 extraction salt packet were added and shaken immediately for at least 1 min to avoid conglomeration. The salt packet contained magnesium sulfate (4.0 g), sodium chloride (1.0 g), disodium hydrogen citrate sesquihydrate (1.0 g) and trisodium citrate dehydrate (0.5 g). Samples were centrifuged at room temperature for 5 min at 30 0 0 rpm (1452 R.C.F).
Dispersive SPE (dSPE) clean-up 6 mL of the crude QuEChERS extract was transferred into an appropriate dSPE clean-up tube depending on the matrix type and shaken for approximately 30 s i.e. if the dSPE tube did not contain graphitised carbon black (GCB) or for approximately 2 min if the dSPE tube did contain GCB. Tubes were centrifuged at 30 0 0 rpm (1452 R.C.F.) for 5 min. Addition of GCB to the dSPE tube is recommended for pigmented fruit and vegetable matrices. Table 1 has details of sorbents/salts used for dSPE clean-up for the commodities tested.

Solvent exchange
2 mL of dSPE cleaned-up extract was transferred to an evaporation tube containing 20 μL of 5% formic acid in acetonitrile. This acidification step was carried out straight away in order to protect base labile pesticides. At least 5 drops of dodecane were added as a 'keeper' to preserve volatile pesticides. Extracts were evaporated to dryness at 35 °C under nitrogen and made up to 2 mL with ethyl acetate. 100 μL of ISTD and 60 μL of AP mix were added and the samples were filtered through 1.2 μm glass fibre filters into autosampler vials for GCMS/MS analysis. ISTD and AP were added to all matrix-matched standards, fortified 'recovery' samples and real samples at the same stage of the protocol i.e. at the end prior to filtration into autosampler vials.

GCMS/MS instrumentation
The gas chromatography triple quadrupole mass spectrometry (GCMS/MS) system was a TSQ 80 0 0 Evo GCMS system. The GCMS/MS incorporated a TRACE 1310 gas chromatograph fitted with a TriPlus RSH autosampler and a programmable temperature vapourising (PTV) injector. All GCMS/MS system components were supplied by Thermo Fisher Scientific, Waltham MA, USA. The PTV injector incorporated a deactivated Topaz 2.0 mm ID baffled inlet liner along with a Thermo Scientific TG-5SILMS 30 m x 0.25 mm x 0.25 μm GC column (both Thames Restek UK Ltd., High Wycombe, UK). This injector/detector configuration was used throughout. Prior to introduction of this liner, issues were seen with degradation of susceptible pesticides. Pesticides such as captan, dichlofluanid, folpet and tolylfluanid degraded rapidly, often after only 10-20 injections and this was not deemed stable enough for reliable quantification. Introduction of the deactivated liner stabilised the response for these analytes.

Analytical method
All matrix-matched standards, sample extracts, spiked samples, matrix blank and reagent blanks were presented batch-wise to the GCMS/MS system. GC and MS conditions detailed in Table 2 were combined with the data acquisition method that incorporated relevant time-scheduled data acquisition parameters detailed in Table 3 . Acquisition of a minimum of two transitions per pesticide allowed flexibility in the use of different transitions for quantifier and qualifier depending on the matrix. TraceFinder software was used to process the data.
Linearity was assessed over five calibration levels at the method validation stage. A linear calibration curve fit was obtained for all pesticides over the range of standards used (0.5 -10 RL). This range was selected because it covered the residue concentrations routinely encountered during annual surveys. If residues exceeded this calibration range, then higher calibration standards were included, or the extract was diluted accordingly. A weighted linear regression (1/x) was used. The deviation of the back-calculated concentration of the calibration standards was calculated using the equation: Deviation of back-calculated concentration (%) = (measured concentration (from calibration function)) -true concentration) x 100/true concentration. Deviation of back-calculated concentrations for each standard were ≤ ± 20% if positive residues were detected.
Ion ratios of spikes and positive samples had to be within ± 30% (relative) of the average of the calibration standards run in the same analytical sequence. Individual measurements were accepted when both quantifier and qualifier transitions were detected within the retention time window ( ± 0.1 min compared to the corresponding standard in the same analytical sequence).

Method validation and proficiency test results
Validation data for five fruit and vegetable matrices of high water and high water and high acid content are presented in Table 4 . To facilitate a more robust assessment of reproducibility at the initial method validation stage, spikes were extracted and taken through the procedure independently by two different analysts i.e. each analyst prepared three spikes at two spiking levels. Mean recoveries outwith 70-120% and%RSD > 20% required in SANTE guidelines are highlighted in bold italics . Lower recoveries are expected for planar compounds where the dSPE clean-up contained GCB since these compounds e.g. chlorothalonil, hexachlorobenzene, pentachloroaniline and pyrazaphos are retained by the GCB. These lower mean recoveries and mean recoveries just outwith 70-120% obtained for 10% of pesticides at RL and/or 2RL spiking levels and in one or more matrices were accepted for use for a multi-residue method on the basis that any positive residue must be accompanied by an acceptable recovery and these validation data are augmented with ongoing batch recovery data as generated by routine use of the method.%RSD > 20% were obtained for 13% of pesticides in the method at RL and/or 2RL spiking level. Again, these were accepted as part of a multi-residue method since they were just above the required 20%. Use of isotopically labelled internal standards added at the extraction stage could be used to correct recoveries or repeat of the clean-up. A clean-up tube without GCB could be used to achieve better recoveries for planar compounds.
Good response for 'difficult to analyse' pesticides e.g. captan, dichlofluanid, folpet and tolylfluanid can be seen in Fig. 1 which shows extracted ion chromatograms for a plum spiked sample fortified at RL (equivalent to 10 ppb).
Our laboratory used the described method to participate in 4 proficiency test rounds (one European Union proficiency test and 3 Fapas proficiency tests and achieved acceptable z-scores for all pesticide/commodity combinations. Z-scores need to be within the range −2 to 2 to be acceptable. Results are presented in Table 5 . Fig. 2 shows extracted ion chromatograms for 4 pesticides detected and quantified in Fapas peach test proficiency material (round 19,279). The sample contained deltamethrin, diphenylamine, propargite and quintozene at 0.1, 0.06, 0.04 and 0.1 mg kg −1 (100, 60, 40 and 100 ppb), respectively.

Application of method to statutory samples
The method was successfully applied throughout 2019 to statutory UK/EU samples of cabbage (96), lemon (96), pepper (120), plum (96) and spinach (96). Results have not been provided here since at the time of writing the results were still to be reviewed by the UK's Expert Committee on Pesticide Residues and published on the UK Government website [10] . Captan at 0.03 mg kg −1 was easily detected and quantified in a plum survey sample with extracted ion chromatogram presented in Fig. 3 .

Additional Information
Introduction of the PTV injector with a deactivated liner has significantly reduced the amount of preventative maintenance required on the GCMS system. Previously cool on-column injection that incorporated a 1 m coil of fused silica retention gap was used. The retention gap had to be replaced before every batch of samples (~30-40 injections), typically daily meaning that the instrument was down for over an hour per day and sensitivity and retention times had to be checked on a daily basis. Now, the PTV injection liner is only replaced on a monthly basis along with a column trim if peak shapes start to deteriorate. This has been a significant saving on analyst time previously spent on GC maintenance. The PTV also facilitates large volume injection which would be of benefit to allow injection of more sample if levels for pesticides are lowered in future.
Captan and folpet are susceptible to degradation in the GC-MS inlet [11] . Our GC-MS method preserves these pesticides as captan and folpet and we routinely observe minimal degradation to their metabolites (tetrahydrophthalimide (THPI) and phthalimide respectively). The GC-MS method has transitions for THPI and phthalimide included since these metabolites are part of the legal residue definition for captan and folpet. A phthalimide/THPI RL spiked sample is extracted with every batch of samples and included with each analytical run to check for any residues of these metabolites. If detected, then samples are quantified with separate phthalimide or THPI matrix standards to ensure correct quantification of the full residue definition for captan and folpet. The plum survey sample containing captan at 0.03 mg kg −1 displayed in Fig. 3 did not contain THPI above the reporting level of 0.01 mg kg −1 .