Certified Reference Materials (CRMs) of all pesticides (purity >96%) under study were purchased from Dr. Ehrenstorfer GmbH (Augsburg, Germany). MS grade acetonitrile, acetone, ethyl acetate were obtained from J.T. Baker, Avantor, USA. Analytical grade anhydrous magnesium sulphate, sodium chloride and sodium sulphate were obtained from Rankem, India. Anhydrous magnesium sulphate was heated at 500oC for 5 hrs to remove phthalates and then cooled naturally and stored in desiccators. Primary Secondary Amine (PSA) sorbents were purchased from Agilent Technologies, Bangalore, India. Centrifuge (Superspin, Plasto-Craft), vortex mixer (Spinix, Tarson, India), rotospin (Tarson, Kolkata, India), silent crusher (Heidolph, Schwabach, Germany), Rotary vacuum evaporator with temperature controlled water bath (HS 2001 NS, Germany) were used for sample preparation. The electronic analytical balance Sartorius GD603 (Sartorius, Germany) with readability= 0.001ct/0.2 mg was used for weighing. All glassware used for the study was calibrated.

Stock solutions of individual pesticide (1000 mg/L) were prepared in 100 ml volumetric flask with hexane-toluene (1:1) mixture. Additional working standards of different concentrations were prepared by serial dilution technique from the stock solution. All the stock solutions were stored under refrigerated condition (-4°C). Multi-pesticide working standard solutions were prepared by diluting appropriate volumes of the mixture stock solution. The matrix-matched standards were prepared by evaporating appropriate volumes of spiking solutions to dryness under a stream of nitrogen, and then, diluting with the matrix extracts in acetone. The working standard solutions were also stored at -4°C.

Bitter gourd, cabbage and chilli were collected from the untreated control plots of the research field of university farm, BCKV, West Bengal, India. About 1.5 kg of each sample was collected and chopped into pieces. The pieces were homogenized in a Robot Coupe Blixer at 16990×g for 10 min. 15 g homogenized sample was taken into a 50 mL centrifuge tube with 15 mL acetonitrile following vortexing for 1 minute. The mixture was then homogenized by a Silent Crusher at 525 ×g for 1 minute. 1.5 g of Sodium chloride and 4 g of anhydrous MgSO₄ were added to it to remove the moisture. The whole mixture was centrifuged for 10 min after thoroughly mixed by a vortex mixer for 1 minute and subsequently rotospined for 5 minutes. d-SPE cleanup was done using 250 mg PSA sorbent and 750 mg anhydrous MgSO₄. Thereafter, 5 ml supernatant was transferred to 15 ml centrifuge tubes. The tube was capped, vortexed for 30s and centrifuged for 5 min at 8495 ×g speed. The supernatant (1 mL) was evaporated to dryness at 35°C under a gentle stream of nitrogen. The residue was reconstituted in 1 mL acetone and filtered through 0.2 mm nylon 6, 6 membranes to analyse in GC-MS.

The Gas Chromatography coupled with Mass Spectrometry (detector used: mass selective detector-MSD) QP 2010 Plus (Shimadzu Corp., Kyoto, Japan) was used. The oven temperature programming was: initial temperature of 40°C, hold for 1 minute, raised @ 25°C/min to 130°C, then @ 12° C/min to180°C, and finally @ 3°C/min to 280°C with a hold of 7 min. The injector and ion source temperature was 250°C. Helium was used as a carrier gas with purity-99.999%. The interface temperature was 280°C. The instrument was operated in the spit mode with split ratio of 1:10. The injection volume was 2 ìL. The MS conditions included as solvent delay (6 min); scan rate (0.50/s); and scanned mass range (50-500 m/z). All samples were analyzed in Selected Ion Monitoring (SIM) mode.

Data were acquired and processed by GCMS Lab Solution Software with version 4.45. The compound specific retention times, m/z ions & molecular mass for the identification, confirmation and quantification are represented in table 1.

The method was validated as per EURACHEM and the SANTE guidelines (SANTE/11813/2017) by evaluating linearity, limit of detection (LOD), limit of quantification (LOQ), specificity and accuracy and precision.

LOD and LOQ were experimentally determined based on signal to noise ratio (S/N) 3:1 and 10:1 respectively. LOD & LOQ was calculated using LOD=3Sa/b & LOQ =10Sa/b, where Sa is standard deviation of the response & b is the slope (Alexander, 2007). Calibration curve was prepared for checking the linearity and regression coefficients (R²).

The precision in term of intra-day repeatability was determined by calculating relative standard deviation (RSD). The Horwitz ratio pertaining to intra laboratory precision was calculated for all pesticides at LOQ level as follows: HorRat = RSD/PRSD, where PRSD = predicted RSD = 2C^-0.15 and C = concentration expressed as a mass fraction (Conc. @ LOQ = 0.10 × 10^-9). (Horwitz and Albert 2006; Linsinger and Josephs 2006).

Matrix effects triggered by vegetable (bitter gourd, chilli, cabbage) extracts were evaluated by GC-MS analysis through comparing the slopes of the analytical curves obtained using solvent and the matrix extracts. The evaluation of the inflfluence of co-extracts on chromatographic responses of pesticides was performed using the equation (Salvia, Cren-Olivé & Vuliet 2013):

Since the SIM mode of Mass Spectroscopy monitors the selected fragments for each pesticide, the use of mass spectrometry detector, through the SIM mode has greater selectivity and sensitivity in the analysis of pesticides. Interfering compounds may increase or decrease the analytical signal. Thus, matrix-matched calibration was used in order to minimize as well as evaluate the matrix effect.

Linearity range equation, correlation coefficient and statistical data are shown in table 1. The values of correlation coefficients ranged 0.988 – 0.998 which were considered satisfactory. The method was validated in terms of precision RSD (Relative Standard Deviation) presented in table 2. According to the SANTE/11813/2017 guidelines, RSD (%) . 20% are suitable for multipesticide residue methods. The matrix matched recovery ranged from 79% to 109%. The results of recovery percentage can also be considered satisfactory according to criteria established by SANTE/11813/2017.

In present study three vegetables (cabbage, chilli, & bitter gourd) were selected for evaluation of matrix effect. These three vegetables have different level of pigments such as chlorophyll, carotenoids, polyphenols and lycopene. Due to the presence of these compounds (co-extract, pigment, etc.) the evaluation of the matrix effect becomes significant.

Fig. 2 showed that these vegetables exhibited different matrix effect (positive or negative) to the different pesticides. Two different matrix effects are there indicating chromatographic signal enhancement due to interactions of the matrix compounds with the remaining active sites of the liner, column and detector of the instrument and suppression due to the pesticides’ interaction and co-elution with the matrix compounds (Hajšlová 1998). The influence of the three matrices on the recovery for the lowest fortification level is illustrated Fig. 2. The value of %ME in the range between -20% and +20% and it can be considered as insignificant (European Commission; Walorczyk 2014). For this study about 73.3% ME of pesticides were significant and Lamda-cyhalothrin and Bifenthrin showed most significant influence of the matrix effect irrespective of substrates. Most of the investigated pesticides have positive ME while Malathion & Heptachlor showed negative ME. This indicated a suppression of the analytical signal. The intensity of matrix effect increased with fortification levels.

In chilli, Lamda cyhalothrin showed maximum positive ME (+109.3%) and Malathion showed maximum negative ME (-68.61%). Results showed that chilli was the most matrix sensitive substrate among the three and cabbage is the least matrix sensitive as maximum ME(%) of pesticide in cabbage were in the insignificant range. The chromatographic recovery response of the pesticides tested in this study came within the significant range as prescribed by SANTE, 2017 when the analytes were dissolved in the matrix extracts compared to neat solvent. Though results varied for each pesticide and vegetable substrate, generally recoveries were higher for decreasing concentrations in the sample. Calibration curves prepared with different matrix extracts were significantly different to the corresponding curves of standards prepared in the pure solvent (Fig. 1). Recovery values ranged from 79% to 108% in all cases when matrix matched standards were used for calibration. Use of matrix match standard eliminated the effect of co-extracts, pigments etc. and the evidence of these are the value of matrix-match recovery percentage shows in Fig. 3. The comparison between matrix-match recovery (%) & pure solvent based recovery (%) showed that the recovery value was more acceptable in case of matrix-match standard. Analyte signal is influenced by the complex nature of the samples and the physical & chemical properties of co-extractives (polarity, molecule size, thermal stability, volatility, etc.). Thus, the components of the matrix have a direct influence on the quantification of pesticides, so that the matrix effect is more significant in complex samples such as fruit & vegetables (Maštovská & Lehotay 2004; Moreno-González 2014; Restrepo et al. 2014; Silva et al. 2014; Zhang et al. 2014; Domínguez et al. 2014).

The present method of modified QuEChERS coupled with GC-MS showed a simple, quick, cheap and environmental friendly procedure to estimate residues of 23 multi class pesticides accurately in three different matrices. The validation parameters (selectivity, linearity, detection and quantitation limits, accuracy and recovery) were satisfactory as per SANTE, 2017. However, most of the pesticides showed a significant matrix effects (>20%) when calculated with standards prepared in pure solvents. This was because of matrix effect. Recovery values using matrix-matched calibration curve showed the elimination of the matrix effects. The use of matrix-matched calibration standards in the case of all matrices gave reliable results and also validated as per SANTE guidelines. It has been essential to use matrix-matched calibration standards in routine analysis of pesticide residues in food by chromatographic methods to avoid an error caused by the presence of matrices.