Modification and re-validation of the ethyl acetate-based multi-residue method for pesticides in produce

Abstract

The ethyl acetate-based multi-residue method for determination of pesticide residues in produce has been modified for gas chromatographic (GC) analysis by implementation of dispersive solid-phase extraction (using primary–secondary amine and graphitized carbon black) and large-volume (20 μL) injection. The same extract, before clean-up and after a change of solvent, was also analyzed by liquid chromatography with tandem mass spectrometry (LC–MS–MS). All aspects related to sample preparation were re-assessed with regard to ease and speed of the analysis. The principle of the extraction procedure (solvent, salt) was not changed, to avoid the possibility invalidating data acquired over past decades. The modifications were made with techniques currently commonly applied in routine laboratories, GC–MS and LC–MS–MS, in mind. The modified method enables processing (from homogenization until final extracts for both GC and LC) of 30 samples per eight hours per person. Limits of quantification (LOQs) of 0.01 mg kg⁻¹ were achieved with both GC–MS (full-scan acquisition, 10 mg matrix equivalent injected) and LC–MS–MS (2 mg injected) for most of the pesticides. Validation data for 341 pesticides and degradation products are presented. A compilation of analytical quality-control data for pesticides routinely analyzed by GC–MS (135 compounds) and LC–MS–MS (136 compounds) in over 100 different matrices, obtained over a period of 15 months, are also presented and discussed. At the 0.05 mg kg⁻¹ level acceptable recoveries were obtained for 93% (GC–MS) and 92% (LC–MS–MS) of pesticide–matrix combinations.

Introduction

For monitoring and control of pesticide residues, multi-residue methods are very cost-effective and are used in many laboratories. The pesticides are usually first extracted with an organic solvent of high or medium polarity. Typical solvents used for this purpose are acetone [1–4], ethyl acetate [5–26] (Table 1), and acetonitrile [26–31]. With all three options, pesticides are partitioned between an aqueous phase and an organic phase. With acetone and acetonitrile this is done in two successive steps, with ethyl acetate in one step. With regard to extraction efficiency, ethyl acetate has been shown to be equivalent to the water-miscible solvents for both polar and non-polar pesticides in vegetables, fruit, and dry products (after addition of water) [6, 7, 26, 32]. It is also suitable for products with a high fat content—because of the solubility of fat in ethyl acetate, pesticides are released and extracted efficiently. The extract obtained is compatible with gel-permeation chromatography (GPC), the clean-up procedure most suitable for this type of sample. Ethyl acetate is very suitable for GC analysis. It has good wettability in GC (pre)columns; this is of benefit for solvent trapping of the most volatile analytes, which is required for refocusing after injection. Its vapor pressure and expansion volume during evaporation also favor large-volume injection. Finally, it is compatible with all GC detectors. The same extract can also be used for LC analysis, after a solvent change into, e.g., methanol [11, 15–18, 26], as is done for acetone-based methods also [33].

Table 1.

Examples from literature. Conditions typically used in ethyl acetate-based multi-residue analysis

Sample (g)	Addition	EtAc (mL)	Na2SO4 (g)	Extr.	Phase separation	Re-extr.	Evap./reconst. (aliquot/to mL)	Clean-up	Evaporation (from/to mL)	Final extr. g mL−1	Inj. (μL)	Analysis
50	–	100	50	B		–	5→1	GPC	None	0.19	10	GC–NPD/ECD	1987 [5]
75	–	200	40	T	F/Na2SO4	–	100→5	GPC	Eluate→5	1.5	?	GC–NPD, FPD	1991 [6]
									(dilute)	0.3	?	GC–ECD
5	–	20	10	T	Let settle	–	10→1			2.5	1–5	GC–FPD/NPD	1992 [7]
1	5 mL H2O (wheat)									0.5
50	–	100	50	T	F/Na2SO4	–				0.5	2–8	GC–MS/FPD/ECD	1996 [4]
50	–	250	100	B	F/Na2SO4	–	All→100	GPC	Eluate→1	1	1	GC–NPD/ECD	1998 [8]
75	–	200	40	T	F/Na2SO4	–	100→5	SPE (ENV+)	3 mL	1.25	2	GC–ITD/ NPD/ECD	1999 [9]
20	–	100		T	See clean-up	–		Cartridge water abs. Polymer+ GCB/Na2SO4	50→dry→2 ace/hex	1	2	GC–MS, GC–NCI-MS	2001 [10]
												GC–FPD LC–PCR-Flu
8	2 g NaHCO3	50	70	T	F	Yes	All→20 MeOH			0.4	5–10	LC–MS–MS	2002 [11]
25	–	100	75	T	F (vac)	–	All→25 +25 cyclohexane	GPC	Eluate→1	1	1	GC×GC–TOF-MS, GC–TOF-HRMS	2003 [12]
					F/Na2SO4	Rinse							2004 [13]
30	5–6 g NaHCO3	60	30–40	T (30 °C)	F/cotton wool	–	1+0.1 IS→1	–		0.5	10	GC–TOF-MS (DMI)	2003 [14]
25	–	50	25	T	Let settle or centrifuge	–	1→1 H2O	–		0.5	20	LC–MS–MS	2003 [15]
75	NaOH if pH < 4.5	200	40	T	F/Na2SO4	–	100→5	–	5→MeOH	2.5	10	LC–MS–MS	2004 [16]
15	1 mL 6.5 mol L−1 NaOH	90	13	T	F/Na2SO4	Rinse	All→15 MeOH	–		1	10	LC–MS–MS	2004 [17]
15	1 mL 6.5 mol L−1 NaOH	90		T	F/Na2SO4	Yes 2×	All→15 MeOH	–		1	50	LC–TOF-MS	2005 [18]
10	–	50a	10	B	F	–	All→5	GPC	Eluate 35→2	2	10	GC–MS–MS	2006 [19]
6	–	50	3	B	Centr.	Yes	All→5	GPC	Eluate 84→1b	5	1	GC–MS	2006 [20]
20	–	80	50–100	T	F	Yes	All→ace/hex	SPE SAX/PSA	All→3	2.4	2	GC–ECD	2006 [21]
50	–	100	75	T	F/Na2SO4	Rinse	All→10	–		5	2	GC–NPD/MS	2006 [22]
5	–	10		T	F/Na2SO4	Rinse	All→1	–		5	10	GC–MS–MS	2006 [23]
2.5	–	5	2	T	F (syringe)	–		–		0.5	50	GC–NPD	2006 [24]
30	5–6 g NaHCO3	60	30–40	T (30 °C)	F	–	1→0.9 +0.1 IS	–		0.5	20	GC–FPD	2006 [25]
5	10 mL H2O (barley)	50	15	S	F/Na2SO4	–	25→1	GPC	Eluate→10 ACN	0.25	25	GC–TOF-MS, LC–MS–MS	2006 [26]
25	2 mL 4 mol L−1 phosphate buffer	40	25	T	Centrifuge	–	GC: -	GCB/PSA disp		0.5	20	GC– MS	This work
							LC: 0.48→1.5 (MeOH/water)	–		0.2	10	LC–MS–MS

^aEthyl acetate–cyclohexane, 1:1

^bAdditional SPE clean-up step with Florisil EtAc/Hex 1:1 5 mL evap. to 1 mL

T, Turrax; B, blender; S, shaking; F, filtration; MeOH, methanol; ACN, acetonitrile; ace, acetone; hex, hexane

Although multi-residue methods based on ethyl acetate extraction have been used for more than 20 years, and continue to be used in many laboratories (they are, for example, the official methods in Sweden and Spain and are also commonly used in the Netherlands, UK, Czech Republic, Japan, and China), the methods described in the literature frequently include steps that make them, in our opinion, unnecessary laborious. Such steps include repeated extraction, filtration, clean-up steps involving GPC for non-fatty matrices, column chromatography or solid phase extraction (SPE) manifolds and evaporative concentration. Typical examples are given in Table 1. It will be shown in this paper that most of the laborious steps can be replaced by more efficient alternatives—repeated extraction is not required, an aliquot is taken after settling or centrifugation rather than filtration, use of GCB instead of GPC for removal of chlorophyll, use of dispersive SPE instead of classical SPE for clean-up (analogous to an acetonitrile-based method [29]), and injection of larger volumes into the GC instead of manual evaporative concentration.

The objective of the work discussed in this paper was to update and improve the ethyl acetate-based multi-residue method for pesticides in vegetables and fruit in respect of straightforwardness, robustness, and ease and speed of sample and extract handling. Aspects studied include dispersive clean-up using combined GCB/PSA, the possibility of preventing unacceptable adsorption of “planar” pesticides by GCB, by addition of toluene, and large-volume (20 μL) injection in GC. The method has been validated for 341 pesticides and degradation products which are analyzed by GC–MS or LC–MS–MS. For the latter the initial raw extract was used and injected after a solvent change to methanol–water. The suitability of the method as a multi-residue, multi-matrix method is evaluated by use of analytical quality-control data generated during 15 months for 271 pesticides and degradation products for over 100 different matrices, including less common and exotic crops. Results obtained for proficiency test samples during three years are also presented.

Experimental

Chemicals and reagents

Pesticide reference standards were obtained from C.N. Schmidt (Amsterdam, The Netherlands). For GC–MS a mixed stock solution containing 135 pesticides (Table 7; concentration 50 mg L⁻¹ for each pesticide) was obtained from Alltech–Grace (Breda, The Netherlands). The full chemical names of the metabolites of phenmedipham and pyridate are methyl N-(3-hydroxyphenyl)carbamate and 3-phenyl-4-hydroxy-6-chloropyridazine, respectively. Solvents were from J.T. Baker (ethyl acetate, Resi-analysed; Deventer, The Netherlands), Labscan (toluene, Pestiscan), and Rathburn (methanol). Anhydrous sodium sulfate, ammonium formate, potassium dihydrogen phosphate, disodium hydrogen phosphate, acetic acid, and diethylene glycol (all p.A. quality) were from Merck. Water was purified by use of a MilliQ reagent-water system (Millipore).

Table 7.

Recoveries over all matrices (GC–MS analysis)

Pesticide	Quan. ion m/z	Qual. ion m/z	Fortification level (mg kg−1)	# QCs matrices (see Table 6)	Both diagn. ions 60–140%	One of diagn. ions 60–140%	Both diagn. ions >140%	Both diagn. ions <60%	Average recov. (%) Quan. ion	RSD (%)
Acrinathrin	208	289	0.10	110	107	107	3	0	97	16
Azaconazole	173	217	0.05	110	107	107	2	1	97	14
Azoxystrobin	388	344	0.05	108	97	102	0	8	96	15
Benalaxyl	206	148	0.05	110	108	109	0	1	100	13
Bifenthrin	181	166	0.05	109	109	110	0	0	102	13
Biphenyl	154	153	0.05	110	93	94	7	9	98	20
Boscalid	112	140	0.13	109	98	100	2	8	96	16
Bromopropylate	341	343	0.05	110	100	101	9	0	109	14
Bromuconazole	295	173	0.05	110	100	105	4	1	102	18
Bupirimate	273	208	0.02	110	108	109	0	1	96	15
Buprofezin	172	105	0.05	109	105	108	2	0	102	12
Cadusafos	158	159	0.05	110	105	107	1	2	104	13
Chlorfenapyr	364	328	0.04	110	103	106	2	2	102	16
Chlorfenvinphos	323	267	0.05	110	103	103	7	0	103	16
Chlorpropham	213	127	0.05	108	101	106	2	2	105	14
Chlorpyrifos	314	286	0.05	109	107	109	0	1	101	14
Chlorpyrifos-methyl	288	286	0.05	108	101	104	4	2	102	16
Chlorthal-dimethyl	332	301	0.05	110	110	110	0	0	101	14
Cinerin-1	123	150	0.11	110	104	105	4	1	101	15
Cyfluthrin	226	199	0.20	110	102	106	0	4	100	17
Cyhalothrin, lambda-	208	181	0.05	108	104	109	1	0	99	16
Cypermethrin	163	181	0.15	105	99	107	2	0	102	14
Cyproconazole	222	224	0.05	110	103	105	1	4	102	16
Cyprodinil	224	225	0.05	109	101	102	0	8	85	15
DDE, p,p′-	246	318	0.06	110	110	110	0	0	101	13
DDT, o,p′-	235	237	0.05	110	106	107	2	1	103	14
DDT, p,p′-	237	235	0.05	110	82	90	9	11	98	20
Deltamethrin	253	255	0.10	110	91	98	4	8	95	17
Diazinon	179	137	0.05	109	108	110	0	0	101	13
Dichlorvos	185	109	0.05	110	90	96	8	6	99	20
Dicloran	206	160	0.05	108	96	102	3	5	99	15
Dieldrin	263	79	0.05	110	109	109	0	1	104	14
Diethofencarb	168	267	0.05	110	107	108	1	1	100	15
Difenoconazole	323	265	0.10	107	101	106	0	4	96	16
Dimethipin	118	76	0.05	110	95	104	5	1	104	16
Dimethomorph	387	301	0.10	110	98	100	0	10	89	16
Dimoxystrobin	205	116	0.05	110	108	109	0	1	100	12
Diniconazole	270	268	0.15	64	58	62	1	1	97	17
Diphenylamine	169	167	0.05	110	107	107	0	3	101	16
Dodemorph	238	154	0.05	110	109	109	0	1	96	15
Endosulfan-alpha	195+241	239+197	0.50	110	95	100	10	0	107	12
Endosulfan-beta	195+241	237+160	0.10	110	107	107	3	0	102	14
Endosulfan-sulfate	272+229	274+237	0.05	109	102	107	2	1	104	16
EPN	157	323	0.05	110	103	106	3	1	103	17
Epoxiconazole	192	138	0.05	110	106	108	1	1	98	14
Esfenvalerate	167	125	0.15	110	102	103	4	3	106	15
Ethion	231	153	0.05	110	106	106	4	0	103	14
Ethoprophos	158	200	0.05	110	107	108	1	1	104	13
Etofenprox	376	164	0.05	110	102	104	2	4	97	15
Etridiazole	211	183	0.05	109	80	82	21	7	97	21
Fenarimol	219	139	0.05	110	106	108	1	1	103	16
Fenazaquin	160	145	0.05	110	105	105	1	4	88	16
Fenbuconazole	129	198	0.05	110	105	107	1	2	99	17
Fenitrothion	277	260	0.05	108	99	102	7	1	106	16
Fenoxycarb	186	116	0.05	110	89	101	8	1	105	17
Fenpiclonil	238	174	0.05	110	101	106	3	1	102	17
Fenpropathrin	181	141	0.05	109	101	104	6	0	103	13
Fenpropimorph	128	129	0.05	110	108	109	1	0	101	14
Fenvalerate	167	125	0.25	110	102	103	2	5	98	15
Fipronil	367	369	0.05	110	101	100	3	7	99	18
Flucythrinate	199	157	0.05	110	102	106	3	1	103	15
Fludioxonil	248	182	0.05	109	105	107	1	2	98	17
Flusilazole	233	206	0.05	110	104	107	1	2	97	15
Flutolanil	323	281	0.05	110	107	109	1	0	100	13
Flutriafol	219	123	0.04	110	102	104	5	1	103	14
Fluvalinate, tau-	250	252	0.15	110	97	99	5	6	99	15
Furalaxyl	242	95	0.05	110	106	107	3	0	101	13
Heptenophos	124	126	0.05	109	97	104	6	0	102	18
Hexaconazole	216	214	0.05	110	106	108	1	1	102	14
Iprodione	316	314	0.10	103	79	88	8	13	100	20
Jasmolin-1	164	123	0.04	110	92	104	4	2	97	15
Kresoxim-methyl	116	206	0.05	109	106	109	0	1	100	15
Lindane	183	219	0.05	110	107	110	0	0	99	15
Malathion	173	127	0.05	108	103	107	3	0	104	17
Mecarbam	329	131	0.05	110	109	110	0	0	101	15
Mepanipyrim	223	222	0.05	110	88	91	7	12	85	19
Mepronil	269	119	0.10	110	109	110	0	0	97	15
Metalaxyl	206	160	0.05	107	105	108	2	0	103	12
Methidathion	145	85	0.05	109	85	89	19	2	107	15
Metrafenone	395	393	0.05	110	104	106	2	2	94	14
Mevinphos	192	127	0.05	110	88	90	17	3	104	17
Myclobutanil	179	150	0.05	110	102	107	2	1	98	15
Nitrothal-isopropyl	236	254	0.05	110	108	108	1	1	99	13
Nuarimol	235	203	0.05	110	108	110	0	0	101	15
Oxadixyl	163	132	0.15	110	106	107	1	2	99	13
Parathion	291	109	0.05	110	105	109	1	0	105	15
Parathion-methyl	263	247	0.05	109	86	102	8	0	107	17
Penconazole	159	248	0.05	109	108	110	0	0	100	15
Pentachloroaniline	267	265	0.11	110	96	97	0	13	81	15
Pentachlorothioanisole	296	246	0.05	110	87	89	0	21	77	16
Permethrin-cis	183	163	0.05	110	108	110	0	0	101	14
Permethrin-trans	183	163	0.05	110	106	107	3	0	100	13
Phenylphenol, 2-	170	141	0.05	109	102	107	3	0	98	13
Phosalone	182	184	0.05	110	90	92	13	5	101	19
Phosmet	161	160	0.05	109	76	90	16	4	100	22
Phosphamidon	264	127	0.05	110	91	94	13	3	103	19
Picoxystrobin	335	145	0.05	110	105	109	1	0	103	12
Piperonyl-butoxide	176	177	0.05	107	106	109	1	0	100	13
Pirimiphos-methyl	276	305	0.05	110	109	109	1	0	102	13
Procymidone	283	285	0.05	108	106	108	1	1	100	14
Profenofos	337	206	0.05	108	93	102	8	0	104	17
Propargite	173	135	0.33	109	104	109	1	0	103	16
Propiconazole	259	261	0.05	109	106	107	2	1	99	14
Propyzamide	173	175	0.05	110	107	108	2	0	102	12
Prothiofos	309	267	0.05	110	108	109	1	0	99	13
Pyrazophos	221	232	0.05	110	99	99	3	8	91	18
Pyrethrins	123	160	0.36	110	87	103	7	0	105	18
Pyridaben	147	148	0.05	110	107	107	1	2	99	14
Pyridaphenthion	340	199	0.05	110	96	101	7	2	102	17
Pyrifenox	262	264	0.05	110	108	110	0	0	100	15
Pyrimethanil	199	198	0.05	110	107	106	1	3	90	14
Pyriproxyfen	226	136	0.05	110	104	107	2	1	103	16
Quinalphos	157	146	0.05	110	104	105	4	1	104	14
Quinoxyfen	307	272	0.05	110	106	106	0	4	92	14
Quintozene	237	142	0.05	110	107	107	1	2	93	16
Silafluofen	179	286	0.05	110	106	106	0	4	98	14
Spirodiclofen	312	314	0.25	110	95	96	6	8	96	19
Spiromesifen	272	254	0.05	110	105	108	1	1	96	16
Spiroxamine	100	198	0.10	110	107	109	0	1	96	13
TDE, p,p′-	235	237	0.05	110	97	100	5	5	103	14
Tebuconazole	250	252	0.15	67	66	67	0	1	97	15
Tebufenpyrad	171	318	0.05	110	107	108	1	1	100	13
Tebupirimfos	234	318	0.05	110	108	109	1	0	101	14
Tefluthrin	177	197	0.05	110	106	107	3	0	103	13
Tetraconazole	336	338	0.05	110	109	109	1	0	99	14
Tetradifon	356	229	0.15	109	109	110	0	0	99	14
Thiometon	88	125	0.05	110	108	110	0	0	104	15
Tolclofos-methyl	265	267	0.05	108	107	107	2	0	101	13
Tri-allate	268	270	0.05	110	104	105	4	1	104	13
Triazamate	242	227	0.05	110	107	107	3	0	102	14
Triazophos	285	257	0.05	109	95	100	8	2	104	18
Trifloxystrobin	131	116	0.05	110	108	109	1	0	103	14
Triflumizole	278	287	0.03	110	105	107	0	3	99	15
Trifluralin	264	306	0.05	110	107	107	2	1	101	14
Vinclozolin	212	198	0.05	107	106	109	1	0	103	11
Total				14696	13688	14057	402	300
% of # QCs					93.1	95.2	2.7	2.0

Bondesil primary secondary amine (PSA, 40 μm) was obtained from Varian (Middelburg, The Netherlands) and GCB (graphitized carbon black) was purchased as Supelclean ENVI-carb (120–400 mesh, Supelco, Zwijndrecht, The Netherlands).

For GC–MS, in addition to the mixed stock solution, individual stock solutions of other pesticides were prepared in ethyl acetate. From these, additional mixed solutions were prepared in ethyl acetate. For LC–MS–MS analysis, individual stock solutions were prepared in methanol. Mixed solutions were prepared from the individual stock solutions and diluted with methanol. The mixed solutions were used for fortification of samples and for preparation of matrix-matched standards.

The extraction solvent was a solution of internal standard (0.05 mg L⁻¹ antor (diethatyl-ethyl)) in ethyl acetate. Matrix-matched standards were prepared by addition of mixed solutions to control sample extracts. Dilution of the sample extract with mixed solution was never more than 10%.

Instrumentation

GC–MS analysis

GC–MS analysis was performed with a model 8000 Top GC equipped with a Best PTV (programmed temperature vaporizer) injector, an AS800 autosampler, and a Voyager mass spectrometer (Interscience, Breda, The Netherlands). The instrument was controlled by Masslab software. The injector was equipped with a 1 mm i.d. liner with porous sintered glass on the inner surface. The GC was equipped with a 30 m × 0.25 mm i.d., 0.25 μm film, HP-5-MS column and a 2.5 m precolumn (same as the analytical column, connected by means of a press-fit connector).

For PTV injection in solvent-vent mode 20 μL was injected at 5 μL s⁻¹. The solvent was vented at 50°C in 0.67 min using a split flow of 100 mL min⁻¹. The split valve was then closed and the analytes retained in the liner were transferred to the GC column by ramping the temperature at 10° s⁻¹ to 300°C. Total transfer time was 2.5 min after which the split was re-opened.

Helium was used as carrier gas at constant flow (1.5 mL min⁻¹). The oven temperature was maintained at 90°C for 2 min after injection then programmed at 10° min⁻¹ to 300°C which was maintained for 10 min. The transfer line to the MS was maintained at 305°C.

Mass spectrometry was performed with electron-impact (EI) ionization (electron energy 70 eV) at a source temperature of 200°C. Data were acquired in full-scan mode (m/z 60–400), after a solvent delay of 5.5 min, until 30 min. Scan time and inter-scan delay were 0.3 and 0.1 s, respectively, resulting in 2.5 scans s⁻¹. The detector potential was 450 V.

Masslab software (Interscience, The Netherlands) and an Excel macro developed in-house were used for data handling and quantitative data evaluation.

LC–MS–MS analysis

LC was performed with an Agilent, model 1100 instrument comprising degas-unit, pump, autosampler, and column oven. A 4 mm × 2 mm i.d. C₁₈ guard column (Phenomenex) and a 150 mm × 3 mm i.d. LC column (Aqua, 5 μm C₁₈, Phenomenex) were coupled to a triple-quadrupole mass spectrometer (model API2000 or API3000, Applied Biosystems, Nieuwerkerk a/d Yssel, The Netherlands). Analyst 1.2 and, later, 1.4 were used for instrument control and data handling. Additional data processing was performed using an Excel macro developed in-house.

Compounds were separated by elution with a gradient prepared from methanol–water–1 mol L⁻¹ ammonium formate solution, 20:79.5:0.5 (component A) and methanol–water–1 mol L⁻¹ ammonium formate solution, 90:9.5:0.5 (component B). The composition was changed from 100% A to 100% B in 8 min and was then isocratic until 24 min. The composition was then changed back to 100% A in 1 min and the column was re-equilibrated for 10 min before the next injection. The flow rate was 0.3 mL min⁻¹ which was introduced into the MS without splitting. The injection volume was 20 μL and 10 μL for the API2000 and API3000, respectively.

Data were acquired in multiple-reaction-monitoring (MRM) mode. Electrospray ionization (ESI) (called turbo ion spray for the instruments used) mass spectrometry was performed in positive-ion mode. For the API2000 the nebulizer gas, turbo gas, and curtain gas were 20, 50, and 40 arbitrary units (a.u.), respectively. The ion-spray potential was 5000 V. Nitrogen was used as collision gas (4 psi). For the API3000 the nebulizer gas and curtain gas were 12 and 10 a.u. and the turbo gas was 7.5 L min⁻¹. The ion spray potential was 2000 V. Nitrogen was used as collision gas (4 psi). For both instruments, the pause time was 5 ms. The dwell times for the pesticide transitions varied between 10 and 25 ms. The precursor and product ions and the collision energy (data for API3000) for each pesticide or degradation product are listed in Table 8. In the acquisition method one transition for each pesticide was measured. All transitions were acquired in one time window. The total cycle time was 2.24 s resulting in 8–10 data points across the peak. To measure the second transition a second method was created and run if confirmation was needed.

Table 8.

LC–MS–MS settings and performance-validation characteristics

Pesticide	tr (min)	Precurs.	Prod. ion 1	DP	FP	CE	CXP	Prod. ion 2	CE	CXP	Vegetables	n	0.01 mg kg−1		0.1 mg kg−1		Fruits	n	0.01 mg kg−1		0.1 mg kg−1		MS–MS
											Matrix		Rec. (%)	RSD (%)	Rec. (%)	RSD (%)	Matrix		Rec. (%)	RSD (%)	Rec. (%)	RSD (%)
Abamectinea,c	21.7	891	305	46	340	33	22	145	49	10	Cuc/lett	4	66	18	68	15	Apple/grape	4	159	39	155	46	API2000
Acephate	5.5	184	143	31	150	11	12	95	33	6	Cuc/lett	4	80	21	75	9	Apple/grape	4	80	8	76	10	API2000
Acetamiprid	10.5	223	126	91	270	29	10	177	11	14	Cuc/lett	4	99	3	96	7	Apple/grape	4	117	4	98	6	API2000
Aldicarba	11.7	208	116	16	110	11	8	89	21	6	Cuc/lett	4	103	20	91	12	Apple/grape	4	99	13	109	13	API2000
Aldicarbsulfon	7.9	223	86	32	200	21	12	148	13	6	Cuc/lett	4	104	9	83	4	Apple/grape	4	120	5	91	3	API2000
Aldicarbsulfoxide	7.2	207	132	46	300	9	10	89	19	6	Cuc/lett	4	109	12	89	4	Apple/grape	4	109	4	86	3	API2000
Asulam	3.6	231	156	41	260	15	12	92	33	6	Cuc/lett	4	35	28	29	38	Apple/grape	4	13	23	10	30	API2000
Azamethiphos	12.1	325	183	36	220	23	14	112	53	8	Cuc/lett	4	101	6	94	4	Apple/grape	4	106	11	91	10	API2000
Azinfos-methyl	13.5	318	132	41	60	23	6	160	15	6	Lettuce	5	93	16	88	4	Orange	5	69	11	79	9	API3000
Bendiocarb	12.2	224	167	16	100	13	10	109	25	18	Lettuce	5	102	10	96	6	Orange	5	86	8	108	6	API3000
Bifenazate	13.9	301	198	16	110	13	16	170	27	14	Lettuce	5	35	9	33	7	Orange	5	93	7	83	5	API3000
Bitertanol	15.2	338	269	21	120	13	20	99	21	8	Lettuce	5	96	10	81	7	Orange	5	93	10	81	8	API3000
Butocarboximb	11.6	213	75	41	300	21	4	156	17	12	Lettuce	5	101	23	93	12	Orange	5	72	16	89	19	API3000
Butoxycarboxim	7.7	223	106	36	250	13	8	166	11	10	Cuc/lett	4	116	9	104	15	Apple/grape	4	118	4	95	7	API2000
Carbaryl	12.5	202	145	101	370	13	12	127	37	10	Cuc/lett	4	100	4	95	4	Apple/grape	4	111	12	100	10	API2000
Carbendazim	11.4	192	160	46	230	23	12	132	43	10	Cuc/lett	4	104	2	102	10	Apple/grape	2	122	1	105	1	API2000
Carbofuran	13.3	222	165	46	290	17	12	123	29	10	Cuc/lett	4	124	12	111	13	Apple/grape	4	104	11	93	4	API2000
Carbofuran, 3-OH	10.4	238	220	31	210	9	16	163	19	12	Lettuce	5	91	8	94	4	Orange	5	100	7	91	6	API3000
Carboxin	12.6	236	143	11	350	21	2	93	51	2	Lettuce	5	87	9	80	2	Orange	5	89	9	84	6	API3000
Chlorbromuron	14.0	295	206	41	350	27	12	182	25	4	Lettuce	5	100	19	86	5	Orange	5	83	30	84	6	API3000
Chlorfluazuron	18.1	542	385	40	270	29	30	158	29	12	Lettuce	5	79	7	89	5	Orange	5	74	21	86	8	API3000
Clofentezin	15.5	303	138	51	280	21	10	102	61	8	Cuc/lett	4	93	17	76	10	Apple/grape	4	127	24	101	16	API2000
Clomazone	13.6	240	125	31	190	25	8	89	67	6	Lettuce	5	97	4	104	6	Orange	5	90	5	89	9	API3000
Clothianidin	10.2	250	132	36	70	23	10	169	17	10	Lettuce	5	99	11	100	2	Orange	5	110	4	100	3	API3000
Cycloxydim	14.9	326	280	46	260	19	22	180	29	14	Cuc/lett	4	18	118	82	9	Apple/grape	4	38	45	70	28	API2000
Cymoxanil	11.1	199	128	18	120	13	10	111	25	8	Cuc/lett	4	83	13	95	7	Apple/grape	4	90	8	99	2	API2000
Cyromazine	7.1	167	85	40	240	26	6	125	25	10	Cuc/lett	4	96	10	78	11	Apple/grape	4	96	7	81	3	API2000
Demeton	13.6	259	89	26	180	13	6	198	11	16	Lettuce	5	97	14	85	4	Orange	5	76	15	76	9	API3000
Demeton-S-methyl	12.5	231	89	31	50	21	4	61	37	4	Lettuce	5	93	5	86	4	Orange	5	81	6	81	8	API3000
Dem-S-meth-sulfone	8.8	263	169	41	350	23	6	109	41	4	Lettuce	5	104	12	92	6	Orange	5	100	2	97	4	API3000
Desmedipham	13.1	301	182	51	340	13	14	154	25	12	Cuc/lett	4	86	10	88	3	Apple/grape	4	95	22	83	15	API2000
Diafenthiuron	18.1	385	329	41	260	27	22	278	45	18	Lettuce	5	0	–	0	–	Orange	5	104	9	92	7	API3000
Dichlofluanidec	14.1	333	224	46	270	17	18	123	37	8	Cuc/lett	4	21	116	36	116	Apple/grape	4	33	82	54	68	API2000
Dicrotophos	9.5	238	112	41	270	17	8	193	13	16	Cuc/lett	4	110	5	99	3	Apple/grape	4	100	12	93	10	API2000
Diflubenzuron	14.5	311	158	46	270	19	12	141	47	10	Cuc/lett	4	79	15	84	1	Apple/grape	4	101	6	102	12	API2000
Dimethirimol	13.1	210	71	51	290	45	4	98	37	8	Lettuce	5	99	7	97	5	Orange	5	91	10	105	5	API3000
Dimethoate	10.6	230	199	11	350	13	4	125	29	2	Lettuce	5	98	7	96	4	Orange	5	109	17	95	6	API3000
Diniconazole	15.6	326	70	56	310	63	14	159	45	16	Lettuce	5	78	10	93	6	Orange	5	94	16	96	5	API3000
Disulfotonc	15.7	275	89	11	90	27	6	61	41	10	Lettuce	5	53	6	64	7	Orange	5	85	16	86	4	API3000
Disulfoton-sulfone	12.8	307	97	31	150	39	8	153	17	14	Lettuce	5	113	10	105	7	Orange	5	81	6	106	8	API3000
Disulfoton-sulfoxide	12.8	291	185	26	140	17	16	213	15	14	Lettuce	5	111	10	115	6	Orange	5	92	5	101	5	API3000
Diuron	13.3	233	72	36	210	37	4	46	35	6	Lettuce	5	111	6	101	7	Orange	5	94	7	94	6	API3000
DMSA	11.6	201	92	26	150	25	6	137	13	10	Lettuce	5	102	13	97	4	Orange	5	85	13	87	7	API3000
DMST	12.3	215	106	26	160	21	8	151	13	10	Lettuce	5	97	5	95	6	Orange	5	84	13	85	5	API3000
Ethiofencarb	12.8	226	107	36	220	21	8	169	9	14	Cuc/lett	4	81	30	94	5	Apple/grape	4	99	17	94	20	API2000
Ethiofencarbsulfon	9.7	258	107	36	240	21	6	201	11	16	Cuc/lett	4	120	10	105	5	Apple/grape	4	101	8	97	11	API2000
Ethiofencarbsulfoxide	9.9	242	107	31	180	23	8	185	13	14	Cuc/lett	4	114	13	97	2	Apple/grape	4	127	10	107	7	API2000
Ethirimol	13.3	210	140	51	370	31	12	98	37	6	Cuc/lett	4	96	3	88	6	Apple/grape	4	86	26	81	26	API2000
Famoxadonea	14.6	392	331	11	130	15	22	238	25	18	Lettuce	5	90	15	80	1	Orange	5	88	9	80	1	API3000
Fenamiphos	14.5	304	217	41	350	29	4	234	21	4	Lettuce	5	87	8	87	4	Orange	5	93	7	93	5	API3000
Fenamiphos-sulfone	12.2	336	308	81	360	23	22	266	29	20	Lettuce	5	102	8	94	5	Orange	5	81	16	86	8	API3000
Fenamiphos-sulfoxide	12.1	320	171	56	230	27	14	233	35	14	Lettuce	5	114	10	94	4	Orange	5	97	8	108	5	API3000
Fenhexamid	14.2	302	97	51	290	35	8	55	59	8	Lettuce	5	84	15	82	4	Orange	5	85	6	84	5	API3000
Fenpyroximate	19.3	422	366	61	360	21	26	135	43	10	Cuc/lett	4	98	8	95	9	Apple/grape	4	111	9	104	10	API2000
Fensulfothione	13.0	309	281	46	260	21	22	253	25	18	Lettuce	5	96	7	89	3	Orange	5	101	23	83	8	API3000
Fensulfothion-sulfone	13.0	325	269	36	120	21	18	191	33	12	Lettuce	5	103	10	98	8	Orange	5	85	6	100	6	API3000
Fenthion	13.9	279	231	26	130	21	16				Lettuce	5	111	31	81	8	Orange	5	38	22	74	8	API3000
Fenthion-sulfone	12.5	311	125	51	320	29	8	279	25	22	Lettuce	5	95	6	90	4	Orange	5	101	1	89	6	API3000
Fenthion-sulfoxide	12.4	295	280	46	230	25	20	109	45	8	Lettuce	5	93	2	94	6	Orange	5	94	8	87	6	API3000
Fipronil	14.1	437	368	66	370	23	26	290	37	16	Lettuce	5	70	24	88	11	Orange	5	92	28	90	12	API3000
Flucycloxuron	17.3	484	289	66	360	15	20	132	49	10	Cuc/lett	4	113	4	104	3	Apple/grape	4	163	38	121	26	API2000
Flufenoxuron	17.1	489	158	101	360	27	12	141	65	10	Cuc/lett	4	107	17	90	8	Apple/grape	4	172	50	102	8	API2000
Formetanate	12.2	222	165	36	190	19	14	120	37	8	Lettuce	5	100	14	103	6	Orange	5	95	6	95	7	API3000
Fosthiazate	12.7	284	104	31	200	23	6	228	15	22	Lettuce	5	99	8	102	6	Orange	5	84	2	98	6	API3000
Furathiocarb	16.5	383	195	76	370	25	16	252	19	18	Cuc/lett	4	55	32	55	38	Apple/grape	4	87	17	84	7	API2000
Hexaflumuronc	15.2	461	158	51	300	27	10	141	61	10	Cuc/lett	4	91	24	82	7	Apple/grape	4	171	15	114	16	API2000
Hexythiazox	17.4	353	168	41	270	35	12	228	21	18	Cuc/lett	4	99	19	84	15	Apple/grape	4	120	26	84	11	API2000
Hymexazolc	5.8	100	54	66	360	21	4	44	29	2	Cuc/lett	4	76	34	50	49	Apple/grape	4	45	15	22	20	API2000
Imazalil	15.0	297	159	46	290	33	12	201	29	16	Cuc/lett	4	90	4	76	12	Apple/grape	4	111	7	90	13	API2000
Imidacloprid	10.0	256	175	41	240	25	14	209	21	18	Cuc/lett	4	99	9	81	11	Apple/grape	4	121	12	89	7	API2000
Indoxacarb	15.1	528	249	41	240	23	18	150	35	10	Lettuce	5	60	32	73	5	Orange	5	84	6	78	6	API3000
Iprovalicarb	14.1	321	119	31	160	29	10	203	13	18	Lettuce	5	108	5	104	7	Orange	5	97	4	90	10	API3000
Isoxaflutole	12.9	360	251	46	270	19	22	220	55	22	Lettuce	5	76	18	90	15	Orange	5	86	18	98	5	API3000
Linuron	13.8	249	160	46	290	25	12	182	21	14	Cuc/lett	4	103	16	86	11	Apple/grape	4	90	26	101	6	API2000
Metamitron	10.7	203	175	51	290	23	14	104	31	6	Cuc/lett	4	80	11	87	17	Apple/grape	4	97	17	95	9	API2000
Methabenzthiazuron	13.3	222	165	31	200	21	12	150	45	12	Lettuce	5	106	4	98	6	Orange	5	84	8	107	9	API3000
Methamidofos	4.6	142	94	41	240	21	6	125	19	8	Cuc/lett	4	83	16	79	19	Apple/grape	4	86	11	81	5	API2000
Methiocarb	13.8	226	169	46	300	13	14	121	25	10	Cuc/lett	4	94	9	95	4	Apple/grape	4	101	5	94	1	API2000
Methiocarbsulfon	10.7	258	122	56	370	25	8	201	13	16	Cuc/lett	4	109	12	99	11	Apple/grape	4	94	9	87	6	API2000
Methiocarbsulfoxide	10.1	242	185	46	290	19	14	170	31	14	Cuc/lett	4	116	5	101	3	Apple/grape	4	126	8	104	2	API2000
Methomyl	8.8	163	88	21	130	13	6	106	13	8	Cuc/lett	4	153	22	136	19	Apple/grape	4	125	14	103	7	API2000
Methoxyfenozide	13.8	369	313	24	200	13	24	133	34	10	Lettuce	5	93	7	91	4	Orange	5	91	13	91	3	API3000
Metobromuron	13.1	259	170	46	280	25	12	148	21	12	Cuc/lett	4	112	19	99	6	Apple/grape	4	96	9	99	12	API2000
Metoxuron	11.6	229	72	31	190	37	4	46	35	2	Lettuce	5	104	8	100	4	Orange	5	95	8	102	4	API3000
Monocrotofos	9.2	224	127	41	240	21	10	193	11	16	Cuc/lett	4	108	5	90	4	Apple/grape	4	111	8	98	10	API2000
Monolinuron	12.8	215	126	41	260	23	8	148	19	12	Cuc/lett	4	104	7	98	6	Apple/grape	4	111	7	107	8	API2000
Omethoate	6.5	214	125	36	230	29	10	183	15	14	Cuc/lett	4	98	13	85	13	Apple/grape	4	102	5	86	2	API2000
Oxamyla	8.0	237	72	21	160	23	4	90	11	6	Cuc/lett	4	107	31	90	7	Apple/grape	4	128	14	97	9	API2000
Oxamyl-oxim	6.6	163	72	36	230	17	4	90	25	6	Cuc/lett	4	100	6	85	3	Apple/grape	4	118	3	101	9	API2000
Oxycarboxin	10.9	268	175	26	170	19	14	147	35	10	Lettuce	5	98	6	96	4	Orange	5	85	22	78	5	API3000
Oxydemeton-methyl	8.5	247	169	41	230	19	14	109	35	8	Cuc/lett	4	98	11	89	7	Apple/grape	4	104	5	96	4	API2000
Paclobutrazole	13.8	294	70	36	320	45	4	125	51	10	Lettuce	5	96	9	87	8	Orange	5	77	67	69	6	API3000
Pencycuron	15.4	329	125	56	340	35	10	218	23	18	Cuc/lett	4	100	5	77	3	Apple/grape	4	118	4	92	9	API2000
Phenmedipham	13.2	301	168	51	290	13	14	136	29	10	Cuc/lett	4	99	7	96	5	Apple/grape	4	108	11	84	11	API2000
Phenm.-metabolite	10.0	168	136	31	200	14	10	108	26	8	Cuc/lett	4	107	9	103	5	Apple/grape	4	101	14	96	17	API2000
Phorate	15.5	261	75	26	150	21	4	47	45	8	Lettuce	5	96	27	91	11	Orange	5	104	2	88	6	API3000
Phorate-sulfone	12.9	293	171	26	150	17	10	115	37	10	Lettuce	5	114	10	95	9	Orange	5	83	6	104	4	API3000
Phorate-sulfoxide	12.8	277	199	41	270	17	6	97	45	4	Lettuce	5	99	8	96	3	Orange	5	98	6	91	4	API3000
Phosphamidon	11.7	300	174	41	250	19	14	127	33	10	Lettuce	5	101	5	107	5	Orange	5	98	7	100	7	API3000
Picolinafen	16.4	377	238	56	220	41	14	256	29	20	Lettuce	5	81	8	96	6	Orange	5	103	8	99	5	API3000
Pirimicarb	13.0	239	72	26	360	31	4	182	23	12	Lettuce	5	99	6	96	4	Orange	5	89	9	92	3	API3000
Pirimicarb, desmethyl	11.6	225	72	21	360	33	4	168	21	6	Lettuce	5	103	4	98	3	Orange	5	31	14	42	15	API3000
Prochloraz	15.4	376	308	46	310	13	22	70	41	16	Cuc/lett	4	90	15	78	13	Apple/grape	4	84	38	94	64	API2000
Profoxydim	16.2	466	280	66	140	27	20	180	35	12	Lettuce	5	33	25	30	6	Orange	5	49	34	55	5	API3000
Propamocarb	8.5	189	102	31	190	25	6	144	19	12	Lettuce	5	75	4	72	4	Orange	5	22	14	18	8	API3000
Propoxur	12.2	210	111	31	210	19	8	168	11	14	Cuc/lett	4	114	3	100	5	Apple/grape	4	118	3	98	5	API2000
Prothiocarb	7.4	191	146	46	240	21	12				Cuc/lett	4	85	26	63	37	Apple/grape	4	106	5	83	10	API2000
Pymetrozine	9.0	218	105	56	370	27	8	201	9	16	Cuc/lett	4	65	26	85	8	Apple/grape	4	47	7	71	7	API2000
Pyraclostrobin	15.1	388	194	1	350	19	6	163	33	6	Lettuce	5	72	13	77	6	Orange	5	87	4	83	7	API3000
Pyridate metabolite	10.4	207	77	56	340	45	6	104	31	8	Cuc/lett	4	100	12	87	4	Apple/grape	4	89	9	75	5	API2000
Rotenone	14.7	395	213	101	370	31	16	192	33	14	Cuc/lett	4	93	13	93	8	Apple/grape	4	94	16	94	30	API2000
Sethoxydim	15.2	328	178	46	260	25	14	220	19	18	Cuc/lett	4	67	39	88	3	Apple/grape	4	59	34	96	28	API2000
Spinosyn A	22.0	733	142	96	280	43	12	98	83	6	Lettuce	5	95	9	93	6	Orange	5	97	4	92	2	API3000
Spinosyn D	24.1	747	142	96	110	47	12	98	89	4	Lettuce	5	86	3	93	6	Orange	5	99	7	92	5	API3000
Tebuconazole	14.8	308	70	61	140	51	6	125	53	8	Lettuce	5	80	6	93	3	Orange	5	95	8	96	4	API3000
Tebufenozide	14.5	353	133	26	180	23	10	297	13	22	Cuc/lett	4	103	16	86	11	Apple/grape	4	106	42	78	33	API2000
Temephos	16.3	467	125	71	320	39	10	419	35	32	Lettuce	5	62	27	81	6	Orange	5	92	7	95	9	API3000
Tepraloxydim	12.7	342	250	31	180	19	28	166	29	12	Lettuce	5	44	19	60	7	Orange	5	73	15	62	4	API3000
Terbufos	16.7	289	103	11	120	13	10	57	37	8	Lettuce	5	73	27	75	8	Orange	5	80	24	81	12	API3000
Terbufos-sulfone	13.5	321	171	21	130	19	12	115	39	6	Lettuce	5	108	4	101	11	Orange	5	99	6	93	10	API3000
Terbufos-sulfoxide	13.5	305	187	6	110	17	10	97	59	8	Lettuce	5	106	3	103	5	Orange	5	98	5	97	9	API3000
Thiabendazole	12.2	202	175	56	370	35	12	131	45	10	Cuc/lett	4	87	12	101	3	Apple/grape	4	98	2	92	7	API2000
Thiacloprid	11.0	253	126	41	210	27	8	90	53	16	Lettuce	5	97	9	102	3	Orange	5	102	6	116	7	API3000
Thiametoxam	9.0	292	211	46	270	19	24	132	33	10	Lettuce	5	94	4	97	4	Orange	5	101	9	99	6	API3000
Thiocyclamd	12.6	182	137	21	160	21	12	73	29	14	Lettuce	5	96	11	89	6	Orange	5	100	15	82	11	API3000
Thiodicarb	12.7	355	88	20	130	31	6	108	21	8	Cuc/lett	4	37	115	42	98	Apple/grape	4	83	4	79	4	API2000
Thiofanox	12.9	219	57	11	90	19	6	61	15	4	Lettuce	5	nd	81	93	21	Orange	5	nd	–	84	30	API3000
Thiofanox-sulfone	10.2	251	57	16	350	26	2	76	21	4	Lettuce	5	110	16	101	5	Orange	5	85	25	85	8	API3000
Thiofanox-sulfoxide	9.8	235	104	31	320	17	4	57	27	2	Lettuce	5	110	2	105	3	Orange	5	109	11	88	6	API3000
Thiometonc	13.0	247	89	16	110	23	6	61	45	8	Lettuce	5	96	17	100	9	Orange	5	87	11	100	2	API3000
Thiophanate-methyl	12.1	343	151	30	210	25	12	311	17	23	Cuc/lett	4	66	8	75	16	Apple/grape	4	41	59	37	98	API2000
Tolylfluanidea	14.7	364	238	31	210	19	18	137	41	10	Cuc/lett	4	31	116	42	115	Apple/grape	4	75	93	24	81	API2000
Triadimefon	14.0	294	197	31	180	23	12	225	19	18	Lettuce	5	92	10	86	6	Orange	5	89	7	78	7	API3000
Triadimenol	14.1	296	70	16	130	31	4	99	21	8	Lettuce	5	101	7	87	6	Orange	5	89	7	82	9	API3000
Triazoxide	13.5	248	68	56	320	47	4	95	37	6	Lettuce	5	99	102	76	19	Orange	5	43	107	69	10	API3000
Trichlorfon	10.6	257	109	46	260	27	8	221	15	18	Cuc/lett	4	116	16	104	22	Apple/grape	4	114	8	99	4	API2000
Tricyclazole	11.5	191	136	56	360	39	10	163	31	12	Cuc/lett	4	105	5	92	6	Apple/grape	4	96	11	83	3	API2000
Triflumuron	14.9	359	156	30	200	23	12	139	47	10	Cuc/lett	4	94	9	92	7	Apple/grape	4	118	12	109	8	API2000
Triforine	13.2	435	390	12	100	13	30	215	40	15	Cuc/lett	4	98	13	101	6	Apple/grape	4	97	10	93	9	API2000
Vamidothion	10.4	288	146	46	300	19	12	118	31	8	Cuc/lett	4	111	16	96	3	Apple/grape	4	119	11	104	7	API2000

Cuc, cucumber

Lett, lettuce

^aNH₄ adduct

^bNa adduct

^cLOQ level 0.05 mg kg⁻¹

^dLOQ level 0.02 mg kg⁻¹

Sample preparation

Vegetable and fruit samples were taken from batches of samples as received from the food industry and trade for routine multi-residue analysis. After removal of stalks, caps, stems, etc., as prescribed by 90/642/EEC Annex I [34], an amount corresponding, at least, to the minimum size of laboratory samples (usually 1–2 kg [35]) was homogenized in a large-scale Stephan food cutter. A subsample (25 g) was weighed into a centrifuge tube. Fortification was performed at this stage. Phosphate buffer (pH 7, 4 mol L⁻¹, 2 mL) and extraction solution (ethyl acetate with internal standard, 40 mL) were then added. Just before Turrax extraction anhydrous sodium sulfate (25 g) was added. After Turrax extraction (1 min) the tubes were centrifuged (sets of four).

For GC–MS analysis, Eppendorf cups were prefilled with 25 mg PSA and 25 mg GCB. To avoid a weighing step, scoops were made in-house for this purpose. Their accuracy was established to be 25 ± 2 mg (n = 10). For clean-up, 0.8 mL extract and 0.2 mL toluene were added to the cup with the SPE materials. The cups were then closed and the samples were vortex mixed for 30 s and centrifuged (up to 24 at one time). One aliquot was transferred to an autosampler vial with insert, and a second aliquot was transferred to an autosampler vial and stored under refrigeration as back-up extract. The calculated amount of initial sample in the final extract was 0.5 g mL⁻¹.

For LC–MS–MS analysis the initial extract (3.2 mL for the API2000 and 0.48 mL for the API3000) was transferred to a disposable glass tube. After addition of a solution of diethylene glycol in methanol (10%, 200 μL) the extract was evaporated to “dryness” under a gentle flow of nitrogen gas at 35°C (up to 36 tubes in a heater block). The residue was reconstituted in methanol (1 mL and 0.75 mL for the API2000 and API3000, respectively), by use of vortex mixing and ultrasonication (5 min). The extract was then diluted 1:1 with component A. After centrifugation one aliquot was transferred to an autosampler vial with insert, and a second aliquot was transferred into an autosampler vial and stored under refrigeration as back-up extract. The final extract concentration was 1 g mL⁻¹ and 0.2 g mL⁻¹ for the API2000 and API3000, respectively.

For dry products (e.g. cereals) 5 g was weighed and 20 mL water was added. After soaking for 2 h samples were processed as described above. A larger amount of extract was taken for evaporation to compensate for the reduced amount of sample processed and to bring the final extract concentration to 0.2 g mL⁻¹.

With the final method, one person can process 30 samples in eight hours. Here processing includes specific preparation before homogenization (i.e. removal of caps from strawberries, etc.), homogenization of the samples, extraction, cleaning the Turrax between samples, clean-up for GC–MS, and solvent switch for LC–MS–MS, i.e. from laboratory sample to ready-to-inject solutions in autosampler vials.

Quantification

GC–MS

For each pesticide the concentrations were calculated for two diagnostic ions. In previous validation work (not published) using the same software it was found that for most pesticides automatic integration and repeatability of response were better when peak height, rather than area, was used. Peak height was therefore used, with few exceptions (e.g. pesticides prone to tailing, for example 2-phenylphenol). All responses were normalized to the response of the internal standard (antor). One-point calibration was performed using a fixed matrix-matched standard (tomato, see Results and discussion section) at a level corresponding to five times the LOQ. The linearity of the plot of MS response against concentration was verified periodically over the range 0.01 to 1–5 mg kg⁻¹. For most pesticides linearity was adequate (relative response within 20% of the calibration standard) up to at least 1 mg kg⁻¹.

LC–MS–MS

The internal standard (antor) was evaluated qualitatively only to confirm injection of the sample extract. Because of unpredictable and varying matrix effects for several of the matrices included in this work, normalization against the internal standard was not considered feasible. For each sample matrix that was fortified, a matrix-matched standard was also prepared by spiking the final extract of the corresponding control sample. Peak area was used for quantification. One-point calibration was performed using the matrix-matched standard at a level corresponding to five times the LOQ. Linearity of the MS response against concentration was verified periodically over the range 0.01 to 1 mg kg⁻¹. For most pesticides, the relationship was linear (relative response within 20% of the calibration standard) up to at least 0.5 mg kg⁻¹.

Validation

Initial method validation was performed in accordance with EU guidelines [36, 37]. Two times five portions of the homogenized sample were spiked with a mixture of pesticides at a low level (0.01 mg kg⁻¹ or lower) and at a level ten times higher. Together with two unfortified control portions of the sample, they were processed and analyzed as outlined above.

Additional method-performance data were acquired by analyzing fortified samples concurrently with each batch of samples. The spike level (0.05 mg kg⁻¹ for most pesticides) was five times the LOQ. With each batch different products were selected as much as possible. In the compilation the emphasis was on products which are less frequently reported in the literature to challenge the applicability of the method as a “multi-matrix method”. For this purpose samples were not pre-screened for absence of pesticides and, consequently, occasionally recoveries could not be determined, because of the relatively high levels incurred. Such results were eliminated from the data set.

Spectrophotometric measurement of removal of chlorophyll

For evaluation of the removal of chlorophyll by GCB and comparison with GPC, a lettuce extract was prepared by extracting 25 g lettuce with 40 mL ethyl acetate after addition of 25 g anhydrous sodium sulfate. As a reference, 0.8 mL ethyl acetate was added to 3.2 mL of this extract to bring the extract concentration to 0.5 g mL⁻¹. For dispersive SPE, 100 mg GCB was added to sets of duplicate tubes and 3.2 mL extract was added to all tubes. Solvent was then added to four sets of tubes: set one 0.8 mL ethyl acetate, set two 0.4 mL ethyl acetate and 0.4 mL toluene (i.e. 10% toluene), set three 0.8 mL toluene (20% toluene), and set four 0.8 mL xylene (20% xylene). The extracts were vortex mixed and centrifuged.

For GPC clean-up, 2.5 mL lettuce extract was injected on to a 40 cm × 28 mm i.d. Biobeads SX3 column with 1:1 ethyl acetate–cyclohexane as eluent. The fraction collected was such that at least 50% of the pyrethroids were recovered (fraction from 105–200 mL). The eluate was first concentrated, by rotary evaporation at 40°C, to approximately 5 mL, then transferred to a tube for further concentration, under nitrogen gas, to 2.5 mL.

Final extract concentration before and after clean-up was always 0.5 g mL⁻¹. Aliquots of the extracts were transferred to a cuvet for spectrophotometric analysis at 450 nm. If required, the extracts were diluted with ethyl acetate to bring absorption within the linear range. The amount of chlorophyll in the uncleaned extract was defined as 100%. For calibration purposes the uncleaned extract was diluted 10, 20, 40, 50 and 100 times with ethyl acetate and a calibration plot was constructed. Chlorophyll remaining after clean-up was determined from the decrease in absorption at 450 nm compared with the absorption of the uncleaned lettuce extract.

Results and discussion

Monitoring of residues in fresh produce for the food industry, especially trade and retail, calls for rapid turnaround, preferably within one or two days. This means sample preparation must be rapid and straightforward. With regard to cost and waste, consumption of solvents and reagents should be low. At the same time, EU directives with regard to sample definition (90/642/EEC, [34]) and laboratory sample size (2002/63/EC [35]) for residue analysis should be respected. This means, for example, that that a total of 2 kg grapes (after removal of stalks), five whole melons, or 1 kg strawberries (after removal of caps) must be processed. The actual analysis is performed on a subsample of the laboratory sample, after appropriate comminution. The more thorough the comminution, the smaller the subsample can be and the lower the amount of solvent needed for extraction. It has, furthermore, been reported that for well homogenized samples extraction by vortex mixing or shaking, instead of high-speed blending (Turrax) suffices for effective extraction [29], although there is still some debate on this matter [38].

Homogenization

For homogenization there are several possibilities. Food choppers or kitchen blenders are often used. Very thorough homogenization can be achieved with the latter, but it is not possible to process the entire laboratory sample at once. For this reason, large-scale food choppers are more suited. With such devices, homogeneity is not always optimum, as can be observed with, e.g., tomatoes, for which small pieces of skin drift in the “soup” obtained after homogenization. Subsampling of very small amounts is, therefore, not acceptable after this procedure, because the subsample would be insufficiently representative of the original sample. More thorough homogenization can be achieved after addition of dry-ice or liquid nitrogen (cryogenic homogenization). This procedure is recommended when reducing the subsample for analysis to 10 g. This procedure is more laborious, however, because it involves cutting the sample into pieces, freezing the sample (usually overnight), cryogenic comminution, then dissipation of the dry-ice or liquid nitrogen before further processing or storage. It also puts higher demands on the cutter (blades) and requires additional precautions for the operators (protection against low temperatures and noise). Cryogenic comminution has been recommended for some pesticides because it reduces their degradation during this step [39].

In recent years the food trade and retail have been intensifying their residue-monitoring programs and require analytical data before harvest, before accepting an assignment, or before releasing their products from distribution centers to supermarkets. For fresh produce this means there is a much pressure on laboratories for rapid turnaround (24–48 h). This is difficult to achieve when the analysis involves overnight freezing for cryogenic comminution. Thus, for reasons of ease and speed, it was decided to retain the current procedure—ambient homogenization of the entire laboratory sample by use of a large scale food cutter (thus accepting the consequence that for a limited number of pesticides the concentration found might be an underestimate). Because of non-optimum homogenization with the food cutter, subsamples should not be too small, and further comminution is required for efficient extraction of systemic pesticides. This can be achieved during extraction by use of an Ultra Turrax. We have previously established the minimum size of subsample that did not negatively affect the repeatability of the analysis. This was done with samples which contained residues. For subsamples (n = 7) of 50 and 25 g, the relative standard deviation (RSD%) was below 8% for several pesticide–matrix combinations. For pear leaves (regarded as a difficult matrix to homogenize) containing bromopropylate, phosalone, and tolylfluanide it was observed that the RSD increased from <8% to 14–18% when the amount of subsample was reduced from 25 g to 12.5 g. From this it was concluded that, with our procedure, 25 g was the minimum required amount of subsample.

pH adjustment

In the ethyl acetate-extraction procedure analytes are extracted and partitioned between water (from the matrix itself, or added water for dry crops) and ethyl acetate in one step. For basic and acidic compounds the partitioning can be affected by pH, which can vary substantially with the matrix. Because the same extract is to be used not only for GC–MS but also for LC–MS–MS (after changing the solvent to methanol) which, preferably, should also include analysis of basic and acidic pesticides, control of pH was regarded as necessary. A pH of approximately 6 was chosen as compromise for efficient extraction of basic and acidic compounds. Although acidic pesticides were not included in this work, data in the literature (for barley without pH adjustment, i.e. non-acidic conditions [26]) indicate they are extracted into ethyl acetate.

For pH adjustment others have used sodium hydroxide [16–18] or sodium hydrogen carbonate [11, 14, 25] (Table 1). A disadvantage of this is that the amount of salt needed depends on the acidity of the sample. Addition of too much will result in a high pH and possible degradation of base-sensitive pesticides. To keep the method as straightforward as possible the pH was adjusted using a solution of concentrated phosphate buffer (4 mol L⁻¹, 2 mL). A solution was preferred over addition of solid salts because this enabled use of a dispenser and eliminated additional weighing of the salts. The buffer resulted in appropriate pH adjustment for most matrices, although there were exceptions, for example lemon and lime.

Extraction

The two conditions most relevant to extraction efficiency are the sample-to-solvent ratio and addition of salt, which in ethyl acetate-based multi-residue methods has always been sodium sulfate.

The amount of ethyl acetate (in mL) relative to the amount of sample (in g) is, typically, at least 2:1. This ratio has been used for many years (Table 1). It results in good extraction efficiency and is practical with regard to achieving phase separation and avoidance of emulsions. To avoid sacrificing decades of method history no attempts were made to reduce the ratio; to do so might also adversely affect recovery and/or complicate phase separation. Larger amounts (as used by several other laboratories; Table 1) result in greater solvent consumption and more dilute extracts. In previous work [15] it has been shown that the efficiency of extraction of polar pesticides improves with the amount of salt added. When 50 mL ethyl acetate and 25 g sample were used, 25 g sodium sulfate was sufficient to obtain recoveries of 80% or better, even for very polar and highly water-soluble compounds, for example acephate and methamidophos. Because these recoveries were obtained with a single extraction it was found unnecessary to perform repeated extraction, as some laboratories are doing [11, 18, 20, 21]. For addition of the sodium sulfate an automatic salt-dispenser coupled to a balance, as is used in our laboratory, or a scoop, was found to be very convenient.

The extraction procedure involves successive addition of buffer, extraction solution (ethyl acetate with internal standard), and sodium sulfate to the centrifuge tube containing the sample, after which the pesticides are extracted and partitioned in one step using a Turrax. During this step the subsample is further comminuted for efficient extraction of the pesticides from the matrix. Vortex mixing, shaking or sonication were regarded as less efficient for subsamples that were homogenized in a large-scale food cutter under ambient conditions, but this was not investigated, partly because a variety of samples containing residues would be required to do so in an appropriate manner.

It was noted from the literature that filtration is often performed to separate the solid pellet from the liquid. Again, there is no real need for this step, which involves additional glassware and, occasionally, rinsing (diluting) of the extract. For many samples a clear ethyl acetate extract is obtained after settling; if not the tubes can be centrifuged. This is no more laborious than filtration and does not involve additional glassware.

Because the same Turrax is used for several samples, carry-over is an aspect to be considered. Between samples the Turrax is cleaned first by rinsing with water, by means of a flow-through beaker, then by brief immersing in two beakers containing ethyl acetate. Using this procedure, carry-over was tested by analyzing a blank after a sample that had been fortified at 5 mg kg⁻¹. Carry-over was less then 0.1%, indicating that the straightforward cleaning procedure was sufficient to avoid cross-contamination up to 5 mg kg⁻¹ when setting reporting limits not lower than 0.01 mg kg⁻¹.

GC–MS analysis

Clean-up

In ethyl acetate-based multiresidue methods either no clean-up or GPC clean-up is performed. This has hardly changed over the years (Table 1). In contrast with acetone and acetonitrile-based methods, in which SPE is commonly employed, this has been reported only occasionally for ethyl acetate-based methods. Obana et al. [10] used a cartridge packed with layers of water-absorbing polymer and GCB. Sharif et al. [21] described a clean-up using SAX/PSA but the scope of the method was restricted to organochlorine and organophosphorus pesticides. Zhang et al. [20] used a clean-up based on Florisil and achieved adequate recovery of many pesticides but not the more polar organophosphorus pesticides. It has been stated that in GC analysis with use of highly selective detectors, for example MS–MS no clean-up is required, even when injecting 15 mg equivalent of matrix (green bean, tomato, pepper, cucumber, marrow, egg plant, and water melon [40]). Other laboratories experienced problems with contamination of the GC inlet and tried to solve this by automatic exchange of liner inserts [14, 41]. This is in agreement with our experience that injection of 10 mg matrix equivalent, especially for leafy vegetables, does result in rapid deterioration of system performance because of accumulation of non-volatile material in the inlet. This makes the system less robust, and frequent exchange of the liner (daily) and GC–pre column (weekly) is required. Another problem encountered with injection of the uncleaned extracts was a shift in the retention times of pesticides relative to that of the calibration standard for some sample extracts. This shift was insufficiently corrected by automatic adjustment of retention times relative to that of the internal standard. Typically, shifts were in the range 0.05–0.20 min and were most abundant for the “azole” pesticides. Such shifts can complicate automatic peak assignment during data-handling. When data acquisition is performed in a non-continuous mode (e.g. selected-ion monitoring or MS–MS) such shifts also increase the risk of pesticides shifting from their acquisition window. For injection of relatively large amounts of matrix (e.g. 10 mg) in GC analysis clean-up for removal of bulk co-extractants is therefore regarded as a prerequisite for robust analysis of a wide variety of vegetable and fruit matrices.

For vegetables and fruit matrices, chlorophyll (MW ∼900) and other pigments, for example carotenoids (e.g. β-carotene, MW 537) are typical bulk co-extractants. Most of these compounds are of low volatility and are not apparent as interferences in the chromatograms; they do, however, accumulate in the liner of the GC and eventually have an adverse effect on transfer of analytes to the column and/or on peak shape. Because of its high molecular weight, chlorophyll can be removed by GPC. A disadvantage is that the extract is strongly diluted and reconcentration by rotary evaporation is almost inevitable when LODs of 0.01 mg kg⁻¹ are required. Such a step would contribute substantially to overall sample-preparation time. Although a very efficient on-line combination of GPC and GC–MS was described recently [42], avoiding GPC whenever possible would be even more straightforward. Solid-phase extraction is an alternative clean-up procedure which involves less dilution and is less laborious. Even more efficient is SPE in the so-called dispersive mode, as described by Anastassiades et al. [29]. Here the solid phase is simply added to the extract, thereby avoiding typical SPE procedures such as conditioning, sample transfer, elution, and evaporative reconcentration. The pesticides partition between the solid phase and the solvent and after vortex mixing and centrifugation the supernatant is ready for analysis.

Two stationary phases, graphitized carbon black (GCB) and phases with amino functionality, have been shown to be particularly effective for removing co-extracted material from the raw extract while not removing most of the pesticides; this makes them very suitable for wide-scope methods [28, 29, 31, 38, 43–45].

Initially, a method was envisaged using SPE column clean-up with GCB, because for leafy vegetables this was found to be the only sufficiently effective alternative to GPC. After the publication on dispersive SPE [29] it was decided to investigate this approach, thus sacrificing some clean-up potential (as has been reported in the literature [31]) for ease and speed.

GCB is well known to adsorb planar molecules, including chlorophyll and other pigments but also pesticides with planar functionality. In acetonitrile-based methods, toluene (typically 25%) is often added to the eluent to desorb these pesticides also from the SPE column [28, 38, 43, 45]. One of the objectives of this work was to investigate the possibility of using GCB in a dispersive clean-up step without unacceptable losses of planar pesticides. First we investigated which pesticides, dissolved in ethyl acetate, are adsorbed by GCB. A somewhat arbitrary, 25 mg mL⁻¹ GCB phase was added to standard solutions. After vortex mixing and centrifugation the solution was analyzed by GC–MS (165 pesticides) and, after changing the solvent to methanol, by LC–MS–MS (another 70 pesticides), and the responses were compared with those from untreated standard solutions. For 35 pesticides (15%) adsorption was observed (Table 2). In addition to the pesticides included in this test, it is known from the literature [44] that chinomethionate, furametpyr, and pyraclofos are also adsorbed by GCB (from acetone–cyclohexane, 1:4).

Table 2.

Pesticides adsorbed by GCB^a

Strong adsorption (rec. 0–50%)	Medium adsorption (rec. 50–70%)	Not consistent
Measured by GC–MS
Chlorothalonil	Azinphos-ethyl	Phosmet
Cyprodinil	Azinphos-methyl	Prochloraz
Fenazaquin	Chlorpyrifos-methyl	Pyrazophos
Hexachlorobenzene	Dicloran	Trifluralin
Mepanipyrim	EPN
Pentachloroaniline	Fenamiphos
Phosalone	Phorate
Pyrimethanil	Quintozene
Quinoxyfen
Measured by LC–MS–MS
Carbendazim	Fenpyroximate
Clofentezine	Flufenoxuron
Desmedipham	Tricyclazole
Diflubenzuron	Triflumuron
Flucycloxuron	Thiophanate-methyl
Hexaflumuron
Phenmedipham
Pymetrozine
Thiabendazole

^aPesticides in ethyl acetate, 25 mg GCB mL⁻¹ solvent

rec., recovered

To investigate how much toluene is required to prevent adsorption of planar pesticides by GCB in dispersive SPE, the partitioning experiment was repeated with standard solutions of 10, 20, or 30% toluene in ethyl acetate. This was done for the GC–MS pesticide mixture only.

As is apparent from Fig. 1, even 10% toluene dramatically improved recovery. With 20% toluene recovery of all pesticides was higher than 65%. It should be noted that this experiment with standard solutions is the worst case. For real samples chlorophyll and carotenoids will also affect the distribution in favor of the pesticides in solution. Use of 30% of toluene further improved recovery only slightly. Twenty percent was regarded as optimum with regard to distribution and ease of solvent elimination in large-volume injection (see below). In addition to toluene, two alternative analogues, benzene and xylene, were also considered. Benzene, was not tested because it could not be used in routine practice because of its carcinogenic properties (although it would have been favorable with regard to solvent elimination). Xylene was tested in a similar way as toluene. Results obtained for hexachlorobenzene and chlorothalonil by use of the two solvents are compared in Fig. 2. Slight but consistently better recovery was obtained with xylene—>70% recovery could now be obtained for all pesticides. Because of its greater volatility, however, toluene was finally selected.

Fig. 1.

Effect of the amount (%) of toluene in ethyl acetate on recovery of pesticides adsorbed by GCB (25 mg mL⁻¹). hcb, hexachlorobenzene; pca, pentachloroaniline; ctn, chlorothalonil; mep, mepanipyrim; cypr, cyprodinil; pyri, pyrimethanil; fena, fenazaquin; quin, quinoxyfen; pyra, pyrazophos; epn, EPN

Fig. 2.

Comparison of toluene and xylene as additives for preventing adsorption of planar pesticides by GCB in dispersive SPE

Obviously, toluene is also likely to affect adsorption of chlorophyll and/or carotenoids and might reduce the effectiveness of clean-up. To investigate this, a lettuce extract was prepared, the dispersive clean-up experiments were performed with different amounts of toluene, and removal of chlorophyll was verified. Visually it was clearly apparent that, despite addition of toluene, the intense green color turned light yellow, indicating that chlorophyll was removed to a large extent. To enable more quantitative evaluation, the extracts were also measured with a spectrophotometer at 450 nm. For comparison, the same extracts were also cleaned by GPC. The results are presented in Table 3. Without toluene, chlorophyll was very effectively removed. Absorption at 450 nm was reduced by 94%. Toluene, as expected, reduced adsorption of chlorophyll, but removal was still 87% or 78%, after addition of 10% or 20% toluene in ethyl acetate, respectively. Similar to observations with the planar pesticides, adsorption was reduced slightly more by use of xylene than by use of toluene. With GPC, chlorophyll removal was 60%. It should be noted here that the elution window was relatively wide, to include pyrethroids within the scope of the method. The elution windows for chlorophyll (and carotenoids) partially overlap those for pyrethroids, as has also been reported by others [44]. From these experiments it can be concluded that chlorophyll has more affinity than the planar pesticides for GCB. In dispersive SPE toluene effectively prevents unacceptable adsorption of planar pesticides while to a large extent maintaining its cleaning properties in respect of chlorophyll. Dispersive GCB not only enables much faster chlorophyll removal, it is also more effective when including pyrethroids in the scope of the method. For non-fatty vegetable and/or fruit matrices, therefore, GPC is not required and dispersive GCB clean-up is a much faster alternative without sacrificing scope.

Table 3.

Removal of chlorophyll by dispersive SPE (GCB) and GPC

Clean-up procedure	Chlorophyll removal (%)
Dispersive SPE, 100% ethyl acetate	94
Dispersive SPE, 10% toluene in ethyl acetate	87
Dispersive SPE, 20% toluene in ethyl acetate	78
Dispersive SPE, 20% xylene in ethyl acetate	71
GPC (fraction incl. pyrethroids)	60

Sample extract: lettuce 0.5 g mL⁻¹. Dispersive SPE: 25 mg GCB mL⁻¹. GPC: wide scope elution window, i.e. including pyrethroids.

The GCB clean-up enabled continuous injection of extracts of leafy vegetables without rapid system deterioration. With some matrices, however (e.g. plums, grapefruit), retention time shifts were still observed. In addition, depending on the matrix, quite intensive interferences could be observed in the GC–MS TIC chromatograms. Further clean-up by PSA, complementing the GCB clean-up by removing compounds such as organic acids and sugars by hydrogen bonding, was therefore investigated. To keep sample clean-up as straightforward and rapid as possible focus was on a combined dispersive GCB/PSA clean-up.

After the outcome of the GCB experiments, partitioning of the pesticides and co-extractants will be between PSA and ethyl acetate–toluene, 8:2. Because no information was available about the distribution of pesticides between these two phases, this was obtained by analyzing pesticide standards in ethyl acetate–toluene, 8:2, with and without PSA. Preliminary experience with dispersive PSA clean-up revealed that with some matrices (e.g. cereals) 25 mg mL⁻¹ did not result in complete elimination of interfering compounds (e.g. fatty acids) typically removed by PSA. Partitioning with a much larger amount of adsorbent (200 mg mL⁻¹) was, therefore, also studied.

With 25 mg mL⁻¹ losses of 30–40% were observed for sixteen pesticides, most probably as a result of adsorption, although the possibility of degradation induced by the basic nature of the PSA material could not be fully excluded. The findings were confirmed by the experiment with 200 mg PSA mL⁻¹ (Table 4). The pesticides for which interaction with PSA was observed all had a C=O or P=O group in common (except for chlorothalonil). Our findings are not in full agreement with those of Anastassiades et al. [29] who did not observe losses as a result of using PSA. For this there can be two explanations. In our experiment adsorption was tested with standard solution rather than matrix. Co-extractants in matrix are likely to compete with the pesticides during adsorption. Second, with our method the organic phase (ethyl acetate–toluene, 8:2) is less polar than the acetonitrile phase; this could result in a stronger interaction between the polar functionality of the pesticides and amino functionality of PSA. From our results it became clear that with regard to the amount of PSA “the more, the better” does not apply. Another observation was that a hump appeared in the TIC chromatogram after a 20-μL injection of solvent mixed with 200 mg PSA mL⁻¹. This hump, which eluted between 6 and 12 min, consisted of many peaks and a variety of masses. Cleaning of the PSA by washing with ethyl acetate (3 × 20 mL for 1 g), then drying by rotary evaporation, eliminated this contamination without affecting the clean-up properties. To keep the method straightforward, 25 mg PSA mL⁻¹ was used as default, and the material was not cleaned before use.

Table 4.

Adsorption of pesticides by PSA

Pesticide	Recovery (%)
Acephate	43a
Acrinathrin	41b
Asulam	0a
Carbaryl	56b
Chlorothalonil	17b
Cycloxidim	39a
Dichlorvos	33b
Dimethoate	62b
Hymexazol	0a
Mevinphos	62b
Phosmet	25b
Phosphamidon	63b
Profenofos	56b
Pyridate	40a
Pyridate-metabolite	7a
Sethoxydim	48a

^aAfter partitioning with ethyl acetate, 25 mg mL⁻¹ and LC–MS–MS analysis

^bAfter partitioning with ethyl acetate–toluene, 8:2, 200 mg PSA mL⁻¹ and GC–MS analysis

The clean-up proved effective at reducing retention time shifts. As an example, for a plum extract without clean-up, the retention times of 24 pesticides (out of 140) were shifted by more than 0.05 min compared with the calibration standard. After clean-up this occurred for three pesticides only. With other matrices also shifts were reduced, but for some matrices (herbs, e.g. parsley) deviations were still quite common.

As an illustration of the removal of co-extractants from the ethyl acetate extract (or, in fact, from the ethyl acetate–toluene, 8:2, extract) by dispersive GCB/PSA clean-up, GC–MS total ion current chromatograms of extracts obtained with and without clean-up are shown in Fig. 3. The most apparent differences are indicated. Several abundant matrix peaks are removed or strongly reduced. For lettuce, the overall background level between 15 and 25 min was also reduced. This clearly visible clean-up was mainly caused by the PSA material. With GCB alone differences between cleaned and uncleaned were much less apparent. The main benefit of GCB was prevention of rapid build up of non-volatile material (chlorophyll) in the liner, which enables prolonged use of the system without maintenance. Experience with method for more than three years and analysis of over 15,000 vegetable and fruit samples shows that, on average, the liner must typically be replaced weekly (after 150–200 injections; iprodion, dimethipin, and chlorfenapyr are the first for which response is lost). Further GC–MS maintenance consists in replacement of pre-column once of twice a month. The GC column is replaced approximately twice a year. The source of the MS is cleaned once a month.

Fig. 3.

GC–MS chromatograms. Overlay total ion chromatograms (TICs) obtained after 20 μL injection of an extract of mandarin (top) and lettuce (bottom) without (higher peaks) and with clean-up

In a continuing search for even further simplification of sample preparation, the possibility of combined extraction and dispersive SPE clean-up in one step was investigated. For two matrices (lettuce and mandarin, fortified with 140 pesticides, triplicate experiments) the solid phase materials (GCB/PSA, relative amounts similar to previous experiments) were added directly to the centrifuge tube containing the sample, sodium sulfate, and the extraction solvent (to which 20% toluene had been added). After Turrax extraction and centrifugation, the extract was ready for injection into the GC. Recovery was compared with that obtained by use of dispersive clean-up after separation of the ethyl acetate extract from the sample mixture. As could be seen from the color of the extract (the lettuce extract was almost colorless) the GCB remained effective. Adsorption of chlorophyll is based on planarity (shape) rather than polarity and, therefore, this will occur from both the aqueous and the organic phases. As was to be expected, the same was not true for PSA. The presence of water prevented adsorption of co-extractants with a hydroxyl group, i.e. almost identical GC–MS total-ion chromatograms were obtained from extracts which were not cleaned and from those cleaned in the centrifuge tube. Pesticide recovery obtained after use of successive or simultaneous dispersive SPE clean-up was very similar, although recovery of some pesticides in the combined approach was too high, because of co-elution of interferences. The final method therefore used successive extraction and dispersive SPE clean-up.

Large-volume injection

GC–MS analysis of sample extracts was performed in full-scan mode. This enables detection of any GC–amenable pesticide. Because system LOQ for a quadrupole mass spectrometer in full-scan mode is limited, conservatively estimated at 100 pg, 10 mg matrix equivalent must be introduced into the GC to reach a target LOQ of 0.01 mg kg⁻¹. With an extract concentration of 0.5 g mL⁻¹, this means 20 μL must be introduced into the GC. Off-line tenfold evaporative concentration and then 2 μL injection could also be performed, but this would involve clean-up of larger volumes of extract, the risk of loss of the volatile pesticides (e.g. dichlorvos), and an additional step in sample preparation. Although large-volume injection in GC is a well established technique [47, 48], many routine laboratories are still reluctant to apply it; if they do, the volume is often restricted to 5–10 μL. Such volumes can be accommodated in liners with a frit or even in empty (baffled) liners when injection speed is carefully adjusted. For larger volumes there is a risk of flooding [46], i.e. that extract is lost as liquid through the split exit. To prevent this, liners can be packed with a variety of materials. Packing materials often have the disadvantage of a large surface area with active sites, however, resulting in degradation and/or adsorption of thermo labile and/or polar pesticides; problems can also be encountered with splitless transfer of higher boiling pesticides (e.g. deltamethrin) from the liner to the GC column. Other disadvantages can be a pressure drop over the liner (slows down solvent elimination) and liner-to-liner variability requiring re-optimization of the solvent-elimination process after liner replacement. A means of by-passing the disadvantages of packed liners while still achieving accommodation of 20–50 μL of liquid was described in 1993 by Staniewski and Rijks [49]. They developed a liner with a sintered porous glass bed on the inner surface wall of the liner. The liquid is retained in the porous glass bed. The potentially active glass surface area is relatively small compared with the materials in packed liners. The gas flow is not obstructed, because the centre of the liner is empty. This enables efficient solvent vapor removal during solvent elimination and efficient transfer of analytes to the analytical column during splitless injection after solvent elimination. Since the early 2000s such liners have been commercially available for PTV injectors from several suppliers, and since then our laboratory has implemented 20 μL as default injection volume for ethyl acetate.

After the development of the dispersive GCB clean-up, the solvent to be introduced into the GC contained 20% toluene, which might effect the processes involved in large-volume injection differently from 100% ethyl acetate. Because toluene does not evaporate azeotropically with ethyl acetate and is less volatile, it will be the main solvent left at the end of the evaporation process. Injection of 20 μL 20% toluene in ethyl acetate means that 4 μL toluene is introduced. The PTV used in this work was equipped with a 1 mm i.d. porous glass bed liner that could hold approximately 30 μL within the zone that is appropriately heated during splitless transfer. Up to this volume there is no need for optimization of injection speed. To obtain information about splitless transfer of the last few microliters of toluene after solvent elimination, cold splitless injections of 1, 2, and 3 μL of standards in 100% toluene were performed. Even with 2-μL volumes peak distortion (fronting peak shape) was observed for pesticides of medium volatility. With 1 μL injections peak shape was good and for several pesticides even better than for ethyl acetate. On injection of 20 μL standard in ethyl acetate–toluene, 8:2, in the solvent-vent mode, no peak distortion was observed, indicating that less then 2 μL toluene remained in the injector after the solvent-vent step. As observed earlier with large-volume injection of ethyl acetate, the vent time (here set at 40 s using an initial PTV temperature of 50°C) was not at all critical, even for the most volatile pesticide (dichlorvos). Venting for 35 or 50 s did not dramatically affect responses or peak shape of the pesticides. In our experience, this phenomenon is typical for porous glass bed liners and contributes to the robustness of the method.

Validation of GC–MS method

In the past a method based on simple ethyl acetate extraction followed by direct GC–MS analysis of the raw extract [4] had been validated for concentrations in the range 0.05–0.5 mg kg⁻¹. The modified method described here involved a dispersive clean-up step, large-volume injection, and injection of ten times more matrix into the GC. Re-validation was therefore required, and focused on method performance at low concentrations. This was done using lettuce as matrix. The validation set consisted of two control samples, five fortifications at 0.001–0.05 mg kg⁻¹ and five fortifications at a level ten times higher. Over 200 pesticides were included in the validation procedure. The results are presented in Table 5. For the 0.01–0.5 mg kg⁻¹ concentration range the EU criteria (recovery 70–110%, RSD 30%, 20%, or 15% for ≤0.01, >0.01–0.1, and >0.1–1 mg kg⁻¹, respectively [37]) were met for 184 of the 201 pesticides included in the validation. At a level a factor of ten lower (fortification in the 0.001–0.01 mg kg⁻¹ range for most pesticides) 147 pesticides could still be detected and for most (78%) of these recovery and RSDs were acceptable. For many pesticides S/N ratios were surprisingly good and background-corrected mass spectra often contained sufficient diagnostic ions (or were even recognizable mass spectra) to enable identity confirmation, as is illustrated in Fig. 4. The limits of detection, defined as S/N = 3 for one favorable diagnostic ion for each pesticide, were determined on the basis of the signals from the low fortification levels and the average noise observed in duplicate control samples. The LOD was at or below 0.001 mg kg⁻¹ for 78 pesticides, between 0.001 and 0.005 mg kg⁻¹ for 73 pesticides, between 0.005 and 0.01 mg kg⁻¹ for 29 pesticides, between 0.01 and 0.05 mg kg⁻¹ for 16 pesticides, and higher for four pesticides.

Table 5.

GC–MS re-validation data for pesticides in lettuce

	Pesticide	tR (min)	m/z (quant)	Level (mg kg−1)	Rec. (%)	RSD (%)	Level (mg kg−1)	Rec. (%)	RSD (%)	LOD (mg kg−1)
1	Acephate	10.45	136	0.026	35	4	0.257	58	9	0.006
2	Acrinathrin	22.06	289	0.018	118	15	0.178	94	9	0.003
3	Aldrin	16.58	265	0.003	139	25	0.031	94	2	0.002
4	Atrazine	14.17	215	0.002	91	21	0.018	98	7	0.002
5	Azinphos-methyl	21.64	160	0.01	119	9	0.098	110	7	0.009
6	Azoxystrobin	25.80	344	0.01	82	8	0.099	92	5	0.003
7	Benalaxyl	19.82	148	0.005	85	9	0.047	90	8	0.002
8	Benzoylurea (deg)a	8.90	141		113	5	0.025	110	6
9	Bifenthrin	20.91	181	0.007	84	9	0.068	89	13	≤0.001
10	Biphenyl	9.81	154	0.006	97	10	0.063	101	5	≤0.001
11	Bitertanol	22.89	170	0.003	83	9	0.031	90	4	0.002
12	Bromophos	17.02	331	0.003	99	7	0.032	105	2	≤0.001
13	Bromopropylate	20.94	343	0.003	103	13	0.032	89	5	0.001
14	Bromuconazole	20.86	173	0.002	109	12	0.024	91	6	≤0.001
15	Bupirimate	18.72	273	0.003	61	8	0.032	91	5	0.001
16	Buprofezin	18.68	172	0.002	85	14	0.019	92	8	0.001
17	Cadusafos	13.46	158	0.002	117	18	0.021	92	11	0.001
18	Carbaryl	15.84	115	0.004	93	9	0.04	93	8	0.002
19	Carbofuran	14.10	164	0.003	88	7	0.033	93	3	0.002
20	Chlordane, alpha-	17.81	373	0.001	*	*	0.015	92	4	0.002
21	Chlordane, gamma-	18.12	373	0.002	84	7	0.015	96	4	0.001
22	Chlorfenvinphos	17.47	323	0.003	84	6	0.03	97	5	0.001
23	Chloroaniline, 3-	7.49	127	0.002	*	*	0.025	25	46	0.003
24	Chlorobenzilate	19.10	251	0.005	*	*	0.05	95	4	0.010
25	Chlorothalonil	15.05	264	0.004	146	15	0.042	136	9	≤0.001
26	Chlorpropham	13.08	171	0.006	*	*	0.059	95	6	0.015
27	Chlorpyrifos	16.67	314	0.003	102	16	0.034	102	5	0.002
28	Chlorpyrifos-methyl	15.70	286	0.001	105	5	0.015	102	6	≤0.001
29	Chlorthal-dimethyl	16.77	301	0.005	90	7	0.051	91	4	0.001
30	Cinerin-1	18.67	150	0.053	84	3	0.528	93	6	0.041
31	Clofentezine	22.45	304	0.014	*	*	0.14	101	14	0.050
32	Cyfluthrin I	23.33	226	0.041	91	7	0.407	93	6	0.023
33	Cyfluthrin II	23.60	226	0.041	100	8	0.407	88	8	0.016
34	Cyhalothrin-lambda	21.91	181	0.003	110	10	0.029	93	6	0.002
35	Cypermethrin-I	23.65	163	0.018	107	29	0.184	96	5	0.008
36	Cypermethrin-II	23.83	181	0.018	94	16	0.184	97	5	0.006
37	Cypermethrin-III	24.07	181	0.018	96	10	0.184	96	6	0.013
38	Cyproconazole	18.97	222	0.006	72	20	0.059	88	7	0.001
39	Cyprodinyl	17.19	224	0.005	105	25	0.051	85	10	≤0.001
40	Cyromazine	14.47	166	0.013	*	*	0.13	82	56	0.040
41	DDE, o,p′-	17.90	248	0.002	*	*	0.015	92	3	0.009
42	DDE, p,p′-	18.50	248	0.001	110	11	0.015	100	5	≤0.001
43	DDT, o,p′-	19.32	235	0.001	102	9	0.015	94	7	0.001
44	DDT, p,p′-	20.28	235	0.002	86	11	0.016	95	8	0.001
45	Deltamethrin	25.44	253	0.022	114	9	0.223	106	5	0.014
46	Demeton-S-methyl-sulfone	16.11	169	0.03	71	15	0.302	91	9	0.004
47	Desmethylpirimicarb	15.42	152	0.003	*	*	0.026	76	7	0.005
48	Diazinon	14.70	137	0.002	98	14	0.019	94	3	0.001
49	Dichlofluanid	16.41	224	0.004	79	9	0.044	98	8	≤0.001
50	Dichlorvos	8.00	185	0.002	107	6	0.018	92	7	≤0.001
51	Dicloran	13.96	206	0.003	96	16	0.029	106	2	0.003
52	Dicofol (as DCBP)	16.75	250	0.005	*	*	0.049	126	33	0.010
53	Dieldrin	18.56	263	0.004	*	*	0.041	95	6	0.005
54	Diethofencarb	16.53	267	0.005	98	5	0.046	96	6	0.001
55	Difenoconazole-I	25.12	323	0.029	94	10	0.288	95	3	0.006
56	Difenoconazole-II	25.36	323	0.029	91	9	0.288	99	3	0.003
57	Diflubenzuron (deg)	6.63	153	0.005	124	9	0.05	107	2	0.002
58	Dimethoate	13.97	125	0.009	*	*	0.091	91	4	0.017
59	Dimethomorph	25.88	301	0.021	95	7	0.207	87	5	0.002
60	Diniconazole	19.54	268	0.002	*	*	0.018	89	12	0.003
61	Diphenylamine	12.76	169	0.003	86	10	0.028	72	15	≤0.001
62	Disulfoton	14.81	88	0.005	101	5	0.05	96	3	0.002
63	DMSA	13.19	200	0.005	87	9	0.052	92	7	0.002
64	DMST	14.37	214	0.005	*	*	0.053	73	32	0.019
65	Dodemorph	16.95	154	0.005	67	26	0.046	91	7	0.002
66	Edifenfos	18.07	310	0.005	96	10	0.05	94	8	0.001
67	Endosulfan-alpha	18.08	239+197	0.005	*	*	0.047	93	5	0.010
68	Endosulfan-beta	19.19	195+241	0.005	*	*	0.046	87	1	0.020
69	Endosulfan-sulfate	19.98	274+237	0.005	82	10	0.047	97	4	0.004
70	Endrin	20.94	245	0.005	*	*	0.051	90	8	0.006
71	EPN	20.57	169	0.01	103	23	0.099	94	7	0.001
72	Epoxiconazole	20.55	194	0.007	*	*	0.066	92	1	0.010
73	Esfenvalerate	24.77	125	0.004	*	*	0.036	98	5	0.008
74	Ethion	19.36	231	0.003	*	*	0.03	97	3	0.007
75	Ethoprofos	12.86	158	0.003	88	17	0.026	93	5	0.001
76	Etofenprox	23.85	164	0.005	100	11	0.049	93	5	0.004
77	Etridiazole	10.74	211	0.014	95	8	0.138	98	4	0.001
78	Etrimfos	15.01	292	0.003	96	4	0.025	93	5	≤0.001
79	Famoxadone	25.90	330	0.01	97	9	0.1	96	5	0.003
80	Fenamiphos	18.23	303	0.015	97	6	0.154	91	11	≤0.001
81	Fenarimol	22.13	139	0.004	*	*	0.038	101	4	0.008
82	Fenazaquin	21.22	160	0.003	152	12	0.027	114	8	0.001
83	Fenbuconazole	23.30	129	0.003	*	*	0.03	92	3	0.006
84	Fenhexamid	20.10	177	0.003	*	*	0.026	90	7	0.004
85	Fenitrothion	16.25	260	0.001	*	*	0.015	95	8	0.003
86	Fenoxycarb	20.89	116	0.015	117	8	0.154	94	4	0.002
87	Fenpiclonil	20.78	238	0.007	88	5	0.071	92	8	0.003
88	Fenpropathrin	21.05	181	0.005	77	13	0.05	92	13	0.001
89	Fenpropimorph	16.63	128	0.001	*	*	0.01	93	2	0.002
90	Fenthion	16.63	278	0.002	99	7	0.023	99	5	≤0.001
91	Fenvalerate	24.54	167	0.004	*	*	0.036	103	8	0.006
92	Fipronil	17.57	367	0.002	81	6	0.024	94	9	≤0.001
93	Flucythrinate-I	23.77	199	0.017	93	11	0.174	92	1	0.004
94	Flucythrinate-II	18.51	199	0.017	94	6	0.174	93	4	0.004
95	Fludioxonil	19.05	248	0.003	113	13	0.027	97	3	0.001
96	Flufenoxuron (deg)	14.79	331	0.012	104	13	0.118	118	19	0.005
97	Flusilazole	18.70	233	0.006	68	8	0.055	87	6	≤0.001
98	Flutolanil	18.30	323	0.003	81	9	0.025	86	8	≤0.001
99	Fluvalinate, tau-	24.80	250	0.025	95	11	0.245	95	5	0.004
100	Folpet	17.65	147	0.016	96	16	0.159	91	15	0.009
101	Fonofos	14.55	246	0.005	94	6	0.049	92	7	0.001
102	Formetanate	15.27	122	0.05	*	*	0.498	102	62	0.188
103	Formothion	15.27	170	0.005	102	13	0.049	89	4	0.004
104	Fuberidazole	15.79	184	0.005	83	29	0.051	55	17	0.001
105	Furalaxyl	17.59	242	0.005	95	10	0.051	101	9	0.002
106	Heptachlor	12.19	272	0.001	*	*	0.014	92	5	0.003
107	Heptachlorepoxide-I	17.45	353	0.003	*	*	0.033	97	12	0.004
108	Heptachlorepoxide-II	17.36	353	0.001	96	13	0.015	94	8	≤0.001
109	Heptenophos	12.24	124	0.003	95	5	0.03	93	3	≤0.001
110	Hexachlorobenzene	18.33	284	0.005	75	28	0.049	96	15	0.001
111	Hexaconazole	18.32	216	0.002	*	*	0.02	87	7	0.003
112	Imazalil	18.37	215	0.005	79	50	0.05	77	14	0.002
113	Iprodione	20.75	316	0.012	108	7	0.12	95	4	0.004
114	Isofenphos	17.46	213	0.005	*	*	0.051	93	3	0.010
115	Jasmolin-I	19.36	123	0.053	*	*	0.528	77	5	0.100
116	Kresoxim-methyl	18.73	206	0.014	95	6	0.139	91	9	0.005
117	Lindane	14.41	183	0.002	86	18	0.02	99	6	0.001
118	Linuron	16.35	248	0.005	*	*	0.048	79	9	0.010
119	Lufenuron (deg)	11.48	176	0.011	123	20	0.114	76	34	0.004
120	Malathion	16.43	173	0.003	*	*	0.034	98	5	0.005
121	Mecarbam	17.49	329	0.003	*	*	0.029	93	5	0.004
122	Mepanipyrim	18.07	222	0.001	*	*	0.013	92	8	0.002
123	Mepronil	19.54	269	0.002	*	*	0.023	87	10	0.005
124	Metalaxyl	15.95	206	0.003	92	10	0.028	97	5	0.002
125	Metaldehyde	8.87	89	0.005	*	*	0.05	111	62	0.021
126	Methacrifos	11.28	180	0.003	97	17	0.029	85	4	≤0.001
127	Methamidophos	7.75	141	0.026	36	24	0.258	47	15	0.005
128	Methidathion	17.82	145	0.003	81	20	0.03	101	5	0.001
129	Methiocarb	16.26	168	0.002	109	59	0.02	77	46	0.001
130	Methoxychlor	21.03	228	0.002	*	*	0.025	90	10	0.003
131	Metoprene	17.56	73	0.01	104	5	0.103	93	3	0.003
132	Mevinphos	10.36	192	0.003	104	16	0.03	99	1	≤0.001
133	Monocrotophos	13.43	192	0.046	84	8	0.456	88	7	0.021
134	Myclobutanil	18.66	150	0.006	*	*	0.055	97	5	0.012
135	Nuarimol	20.28	314	0.005	*	*	0.049	89	7	0.008
136	Omethoate	12.39	156	0.005	57	19	0.054	53	14	0.002
137	Oxadixyl	19.38	163	0.012	*	*	0.124	92	4	0.038
138	Oxydemeton-methyl (deg)	6.63	110	0.005	*	*	0.052	79	7	0.010
139	Paclobutrazole	18.11	238	0.007	197	28	0.07	90	6	≤0.001
140	Parathion	16.69	291	0.011	106	26	0.106	91	6	0.004
141	Parathion-methyl	15.71	263	0.002	88	7	0.021	94	2	≤0.001
142	Penconazole	17.35	248	0.003	90	10	0.03	94	4	≤0.001
143	Permethrin-cis	22.65	183	0.005	101	7	0.049	98	7	0.003
144	Permethrin-trans	22.77	183	0.001	*	*	0.011	98	7	0.001
145	Phenothrin-I	21.40	183	0.005	97	8	0.05	92	9	0.001
146	Phenothrin-II	21.51	123	0.005	93	6	0.05	93	10	0.004
147	Phenthoate	17.53	274	0.005	103	8	0.048	91	5	0.001
148	Phenylphenol, 2-	11.56	170	0.005	96	6	0.052	95	4	0.001
149	Phorate	13.56	260	0.005	98	6	0.05	92	5	0.001
150	Phosalone	21.61	182	0.001	117	5	0.009	101	5	≤0.001
151	Phosmet	20.90	160	0.005	123	16	0.052	100	4	≤0.001
152	Phosphamidon-I	14.75	127	0.011	93	16	0.105	90	3	0.002
153	Phosphamidon-II	15.49	127	0.011	89	9	0.105	91	2	0.005
154	Piperonyl butoxide	20.36	176	0.004	*	*	0.037	89	10	0.010
155	Pirimicarb	15.25	166	0.002	101	9	0.02	95	5	≤0.001
156	Pirimiphos-methyl	16.26	233	0.002	*	*	0.016	87	2	0.004
157	Prochloraz	22.97	180	0.004	*	*	0.038	101	6	0.007
158	Procymidone	17.68	285	0.003	104	15	0.029	91	7	0.001
159	Profenofos	18.42	337	0.005	97	8	0.052	95	10	0.001
160	Propargite	20.31	350	0.01	*	*	0.102	96	7	0.020
161	Propham	10.73	179	0.005	97	5	0.049	94	5	0.001
162	Propiconazole-I	19.89	259	0.014	92	5	0.141	89	9	0.003
163	Propiconazole-II	20.02	259	0.014	90	5	0.141	87	9	0.002
164	Propoxur	12.62	110	0.002	96	6	0.02	92	7	≤0.001
165	Propyzamide	14.58	175	0.005	76	39	0.046	99	2	0.001
166	Prothiofos	18.37	267	0.003	85	19	0.032	101	9	0.001
167	Pyrazophos	22.17	221	0.003	137	11	0.03	145	4	≤0.001
168	Pyrethrins	19.62	123	0.053	*	*	0.528	99	13	0.087
169	Pyridaben	22.82	147	0.005	96	9	0.051	94	3	0.001
170	Pyridaphenthion	20.80	199	0.005	99	10	0.048	93	5	0.003
171	Pyrifenox-I	17.39	262	0.011	84	7	0.106	95	6	0.003
172	Pyrifenox-II	14.68	264	0.011	*	*	0.106	90	6	0.170
173	Pyrimethanil	14.65	198	0.002	135	14	0.02	123	4	≤0.001
174	Pyriproxyfen	21.65	136	0.002	119	18	0.024	91	6	≤0.001
175	Quinalphos	17.55	146	0.004	70	9	0.041	87	8	0.002
176	Quinoxyfen	19.90	272	0.001	113	13	0.014	105	13	≤0.001
177	Quintozene	14.50	237	0.005	106	10	0.046	108	2	0.003
178	Simazine	16.17	201	0.004	91	9	0.039	95	7	0.002
179	Spiroxamine	15.67	198	0.018	99	17	0.176	81	2	0.009
180	TDE, o,p′-	18.67	235	0.003	99	5	0.028	95	4	≤0.001
181	TDE, p,p′-	19.36	235	0.001	86	10	0.014	90	7	≤0.001
182	Tebuconazole	20.28	250	0.009	*	*	0.089	91	9	0.031
183	Tebufenpyrad	21.12	171	0.005	92	17	0.052	87	7	0.001
184	Tecnazene	12.56	203	0.005	108	6	0.048	99	6	0.002
185	Teflubenzuron (deg)	8.12	197	0.003	174	25	0.025	124	25	0.002
186	Tefluthrin	14.91	197	0.001	*	*	0.014	89	14	0.002
187	Terbufos	14.46	231	0.005	100	8	0.052	95	3	≤0.001
188	Tetraconazole	16.85	336	0.003	95	3	0.026	88	6	≤0.001
189	Tetradifon	21.44	356	0.003	*	*	0.03	94	8	0.010
190	Thiometon	13.78	88	0.005	93	5	0.055	100	3	≤0.001
191	Tolclofos-methyl	15.80	265	0.001	91	6	0.01	102	5	≤0.001
192	Tolylfluanid	17.42	238	0.003	85	17	0.031	96	2	0.002
193	Triadimefon	16.75	208	0.007	90	14	0.065	97	6	0.005
194	Triadimenol	17.85	168	0.005	*	*	0.053	85	2	0.029
195	Triazamate	17.95	242	0.003	*	*	0.028	90	10	0.010
196	Triazophos	19.62	257	0.005	109	37	0.054	89	20	0.001
197	Trifloxystrobin	19.92	116	0.006	91	13	0.055	88	11	0.002
198	Triflumizole	17.70	278	0.007	102	15	0.066	80	15	0.001
199	Trifluralin	13.33	306	0.002	92	19	0.019	94	8	≤0.001
200	Vamidothion	17.95	87	0.019	*	*	0.187	100	5	0.045
201	Vinclozolin	15.71	198	0.005	97	16	0.047	93	7	0.003

^aBenzoylurea(deg) = 2,4-difluorobenzamide

LOD: Amount for which S/N = 3, or in the event of an interfering peak, the average peak height for fortified sample (n = 5) should be 3.3 times the average peak height for control sample (n = 2)

*Fortification level below LOD as defined above

Underlined values are outside EU criteria for method validation

Fig. 4.

GC–MS extracted-ion chromatograms obtained from lettuce with (upper traces) and without fortification with pesticides, and the corresponding mass spectra (upper, reference spectra; lower, background-corrected spectra from the sample). a, b, 0.005 mg kg⁻¹ disulfoton (m/z 88); c, d, 0.002 mg kg⁻¹ fipronil (m/z 367); e, f, 0.006 mg kg⁻¹ biphenyl (m/z 154)

This initial validation clearly showed it is possible to introduce 10 mg of matrix equivalent of generic extracts obtained after ethyl acetate extraction of leafy vegetables. Adequate quantitative data are obtained for most of the pesticides at levels of 0.01 mg kg⁻¹ or even below. Detection limits were usually well below 0.01 mg kg⁻¹ after full-scan acquisition with a single-quadrupole MS. This means that for most pesticides at the target LOQ of 0.01 mg kg⁻¹ (i.e. the lowest maximum residue limit set in the EU for vegetables and fruit), the signal-to-noise ratio is adequate for reliable automatic integration of peaks and that confirmation of identity of the pesticide is possible from its mass spectrum or at least one or two other diagnostic ions.

Pesticides that did not meet the EU criteria for quantitative analysis, and/or for which relatively high LODs were obtained, included many compounds known to be troublesome in GC analysis because of to their high polarity or thermal lability. Typical examples are acephate, cyromazine, dicofol (screened for as its degradation product dichlorobenzophenone), dimethoate, imazalil, metaldehyde, methamidophos, methiocarb, omethoate, and the benzoylureas (measured as one common and one compound-specific degradation product). The relatively low recovery of the polar organophosphorus pesticides (acephate, methamidophos, and omethoate) can be attributed to the GC measurement and not to poor extraction efficiency, as was apparent from LC–MS–MS analysis of samples using the same extraction technique (see section LC–MS–MS analysis). For several other polar or labile pesticides adequate quantitative data were obtained during this initial validation, but from previous experience and the results obtained after implementation of the method it was clear that for such compounds LC–based analysis is more robust than GC–MS analysis. Typical examples include carbaryl, carbofuran, clofentezin, monocrotophos, and oxydemeton-methyl.

Analytical quality-control data from routine GC–MS analysis

The initial validation data are continuously being supplemented by performance data generated as part of the analytical quality-control during routine analysis of the samples, to gain insight into reproducibility, robustness, recovery, and selectivity with other matrices. For this, with each analytical batch, one of the samples submitted for routine analysis was spiked with 135 pesticides at five times the target LOQ level (i.e. samples were spiked with 0.05 mg kg⁻¹ of most of the pesticides). A compilation was made of recovery data from a period of 15 months which included analysis of approximately 100 different vegetable and fruit commodities. Given the wide variety of commodities, matrix-matched calibration is quite tedious and would substantially increase the number of standard solutions to be analyzed in the GC sequence. It was therefore decided to select one relatively simple matrix (tomato) as default for matrix-matched calibration, i.e. recoveries for all commodities were calculated against the tomato-matrix standard. For each pesticide, calculations were performed for two diagnostic ions. All together this resulted in approximately 30,000 values.

According to the current EU guideline on quality control in pesticide residue analysis [37], the recovery obtained during routine analysis should be within 60–140%. An overview of the percentage of recovery values within or outside the 60–140% criterion for a wide variety of matrices is presented in Table 6. With such large number of pesticides (or, actually, diagnostic ions) and matrices, one failing combination or more occurred for most matrices. There are several causes for this. Main reasons for recovery below 60% could be poor extraction efficiency or incomplete transfer of the pesticides to the GC column (e.g. adsorption and/or degradation in a contaminated inlet). Higher recovery may occur when a compound from the matrix generates the same diagnostic ion as a pesticide and co-elutes with that pesticide (i.e. detection was not selective). Another reason could be that the matrix effect induced in the GC inlet [50] for a pesticide in a particular matrix is more pronounced than that in the tomato-based calibration standard.

Table 6.

Overview of percentage of recovery values^a within or outside the EU 60–140% criterion [37] after GC–MS analysis

	Matrix	Percentage of all recovery valuesa
		60–140%	<60%	>140%
1	Beetroot	100	0	0
2	Cucumber (1/2)	100	0	0
3	Mint (1/2)	100	0	0
4	Sharonfruit (1/2)	100	0	0
5	Witloof	100	0	0
6	Asparagus	99	1	0
7	Bean sprouts	99	0	1
8	Corn syrup	99	0	0
9	Fennel leaves	99	0	1
10	Grape	99	0	1
11	Kohlrabi (1/3)	99	1	0
12	Lima bean	99	0	1
13	Pak choi (1/2)	99	0	1
14	Pear concentrate	99	0	1
15	Pumpkin	99	0	1
16	Salsify	99	0	0
17	Sharonfruit (2/2)	99	0	1
18	Strawberry	99	0	1
19	Sugar pea	99	1	0
20	Taro	99	0	1
21	Bitter cucumber	98	0	2
22	Cucumber (2/2)	98	1	1
23	Egg plant	98	0	2
24	Kidney bean	98	1	1
25	Kohlrabi (2/3)	98	1	1
26	Mushroom	98	0	2
27	Pineapple	98	1	1
28	Sweet pepper	98	0	2
29	Tomato puree (processed)	98	0	2
30	Turnip	98	1	0
31	Turnip tops (1/2)	98	0	2
32	Alfalfa	97	1	2
33	Cauliflower	97	1	2
34	Cherry	97	0	3
35	Chestnut	97	2	1
36	Endive	97	0	3
37	Fig	97	0	3
38	Kangkung (1/2)	97	1	2
39	Kangkung (2/2)	97	2	1
40	Ladies’ fingers	97	0	3
41	Mango	97	0	3
42	Pear puree (processed)	97	0	3
43	Sorrel	97	3	0
44	Soybean sprouts	97	0	3
45	Asparagus bean	96	1	3
46	Orange	96	2	2
47	Potato leaves	96	2	2
48	Rhubarb	96	2	2
49	Artichoke	95	0	5
50	Tangelo	95	2	3
51	Tarrragon	95	3	2
52	Wine (red)	95	1	4
53	Apricot	94	0	6
54	Chives (1/3)	94	3	3
55	Chives (2/3)	94	4	2
56	Dill leaves	94	4	2
57	Melon puree (processed)	94	1	5
58	Mineola	94	1	6
59	Pak choi (2/2)	94	2	4
60	Sugar water	94	6	0
61	Broad bean	93	1	6
62	Celery leaves (1/4)	93	3	4
63	Chervil	93	5	2
64	Dates	93	7	0
65	Sweetcorn (1/3)	93	4	3
66	Carrot	92	1	7
67	Haricot bean	92	0	8
68	Oregano	92	5	3
69	Parsnip	92	2	6
70	Fennel	91	0	9
71	Green pea (1/2)	91	4	5
72	Passion fruit (1/2)	91	2	7
73	Celery leaves (2/4)	90	6	4
74	Green pea (2/2)	90	1	9
75	Lemon puree	90	8	2
76	Mint (2/2)	90	5	5
77	Pomegranate	90	1	9
78	Purslane	90	1	9
79	Water cress	90	2	8
80	Lettuce	89	7	4
81	Chili pepper (1/2)	88	6	6
82	Chinese cabbage	87	0	13
83	Passion fruit (2/2)	87	3	10
84	Bamboo shoots	86	0	14
85	Celery leaves (3/4)	86	7	7
86	Honey	86	14	0
87	Potato puree (processed)	86	14	0
88	Sugar pea	85	0	15
89	Turnip tops (2/2)	85	0	15
90	Lime	84	4	12
91	Blueberry	83	2	16
92	Potato	83	15	2
93	Celery leaves (4/4)	82	3	15
94	Green pea	82	1	17
95	Apple pulp (processed)	81	6	13
96	Cassava	81	9	10
97	Chives (3/3)	81	7	12
98	Kohlrabi (3/3)	78	0	22
99	Parsley (1/2)	78	6	16
100	Thyme (1/3)	78	2	20
101	Kale	77	6	17
102	Chili pepper (2/2)	76	15	9
103	Coriander leaves	76	18	6
104	Sweetcorn (2/3)	75	18	7
105	Sweetcorn (3/3)	74	9	17
106	Parsley (2/2)	73	20	7
107	Thyme (2/3)	73	3	24
108	Rocket	72	3	25
109	Thyme (3/3)	66	29	5
110	Golden berry (physalis)	65	1	34

^aRecoveries at 0.05 mg kg⁻¹ (0.10–0.30 mg kg⁻¹ for 22 pesticides). Calculated for 135 pesticides, two diagnostic ions each, against a standard prepared in blank tomato extract. The pesticides included are listed in Table 7

Failing pesticide–matrix combinations were most abundant for herbs, kale, sweetcorn, and golden berry, for which up to 35% of recovery values (calculated using the two diagnostic ions for each pesticide) were outside the 60–140% range. These products contain larger amounts of co-extractants than most other vegetables and fruits, which may result in insufficient detection selectivity, enhanced response as a result of a matrix effect (more shielding of active sites in the inlet), and contamination of the inlet. For this type of product more selectivity, e.g. by use of MS–MS would be beneficial. Such detection is also more sensitive than single quadrupole full-scan detection and would enable reduction in the amount of matrix introduced, thus reducing build up of contamination. Overall, when data for all 110 QC samples were included, recovery was acceptable for 91% of the diagnostic ions measured.

On the basis of the same data, an overview by pesticide is presented in Table 7. For each pesticide two diagnostic ions from the full-scan data were integrated and concentrations were calculated. In routine practice, however, the most convenient way of reviewing the data is by using one and the same diagnostic ion for each pesticide, irrespective the matrix. On the basis of the data set obtained (nearly 14,700 pesticide–matrix combinations) the most favorable of the two diagnostic ions, i.e. the ion for which the highest number of recoveries within 60–140% was obtained, was assigned as the Quan ion (default quantification ion). By using this ion, acceptable recoveries were obtained for 93% of pesticides–matrix combinations. This also means that 7% or, in absolute figures, 1008 of the pesticide–matrix combinations did not meet the criterion. 40% of these failing combinations could be accepted after use of the alternative ion, for which calculations were also performed automatically during data processing. Low recoveries (<60%) for both diagnostic ions were obtained for 2.7% of pesticide–matrix combinations. High recoveries (>140%) were obtained for 2% of the combinations. For this latter group manual evaluation of other ions, if available and sufficiently abundant, could further increase the number of acceptable recoveries. Because this is a time-consuming process, it was not done routinely. In the event of deviating recovery, assessment of the results to be reported was based on visual evaluation of the extracted ion chromatograms of the two diagnostic ions at least. On the basis of on the findings it was then concluded the pesticide could not be determined in that specific matrix, or only at higher levels.

It should be noted that the above evaluation applies to a level five times the reporting level, which was set at 0.01 mg kg⁻¹, or the LOQ if higher than 0.01 mg kg⁻¹. At lower levels interferences may have a larger effect and, consequently, more frequent deviations from the 60–140% criterion (most probably >140%) may be observed. For higher levels, the opposite would be true.

Pesticides for which low recoveries (<60%) were frequently obtained (10–21 of 110 QC samples) included iprodione and p,p′-DDT (degradation in inlet), dimethomorph (polar, relatively non-volatile, could be troublesome in splitless transfer), pentachloroanisole, pentachloroaniline, and mepanipyrim (no clear explanation, but probably related to the dispersive SPE clean-up). There were no indications for poor extraction efficiency.

High recovery (>140%) frequently occurred for etridiazole, methidathion, mevinphos, phosmet, phosalone, phosphamidone, and endosulfan-alpha (10–21 times out of 110 QC samples, often in herbs and peas). This was attributed to matrix effects and interferences.

Overall, the pesticides that failed most frequently (11–28 times out of 110) during routine analytical quality control were (in descending order) etridiazole, iprodione, methidathion, pentachlorothioanisole, mevinphos, phosmet, p,p′-DDT, mepanipyrim, phosalone, phosphamidon, biphenyl, dichlorvos, spirodiclofen, pentachloroaniline, deltamethrin, tau-fluvalinate, and pyrazophos. These would be the most relevant for inclusion in alternative methods, for example GC–MS–MS or LC–MS–MS.

Average recovery and RSD were calculated for pesticide–matrix combinations that passed the acceptable recovery criterion. The results are included in Table 7. Average recovery was usually close to 100% and RSDs approximately 15%. For the pesticides known to be adsorbed by GCB systematically lower average recovery (77–90%) was obtained, which is in agreement with the results obtained during method development.

These comprehensive data show that with a relatively inexpensive single-quadrupole MS detector in full-scan mode it is possible to obtain reliable quantitative data down to the 0.01 mg kg⁻¹ level, or even lower, for a wide range of pesticides in a wide variety of matrices after generic rapid sample preparation based on extraction with ethyl acetate. Unified calibration based on a tomato-matrix standard is, furthermore, a feasible approach. One should, however, be aware there are also limitations and that some pesticide–matrix combinations cannot be determined in the 0.01–0.1 mg kg⁻¹ range, and that for other pesticides calibration against the corresponding matrix instead of tomato is required to bring quantitative results within the AQC criteria, especially for MRL violations, when more stringent criteria apply. The data also reveal that the only way to gain full insight into analyte recovery and method selectivity with a wide variety of matrices is by performing analytical quality control on all pesticides which are reported, rather than on a subset, as is suggested in the EU guideline [37]. A subset will suffice for demonstration of adequate sample preparation and injection but will not reveal limitations in the selectivity of GC–MS.

GC single-quadrupole MS remains an effective tool for routine GC analysis of pesticide residues. For many vegetable and fruit matrices there is no real need to change to more advanced (and expensive) MS techniques, for example MS–MS (which has limited scope) or accurate mass TOF-MS (which has a limited dynamic range). Use of such equipment would be justified for more complex matrices and when low μg kg⁻¹ LOQs are required—for example analysis of some pesticides in baby food.

LC–MS–MS analysis

Clean-up

The ethyl acetate extraction procedure is also appropriate for many pesticides not amenable to GC analysis [11, 15, 16, 18, 26]. Typically no clean-up is performed (Table 1). One reason for this is that with regard to chromatographic performance LC columns tend to be more tolerant of injection of bulk matrix than GC columns. In our experience, continual injection of 20 mg equivalent of vegetable and fruit extracts does not result in deterioration of chromatographic performance or unacceptable contamination of the ion source (the system used here was an API2000). In LC–MS co-extracted matrix does have an effect on the response, however, by interfering with the ionization process. This results in suppression (sometimes enhancement) of the response to a pesticide in a matrix compared with that in a solvent standard [51] and complicates quantification of pesticides in the samples. The possibility of reducing matrix effects by use of dispersive SPE clean-up was investigated in a similar way as for GC. First, the effectiveness of the clean-up step was investigated by addition of 25 mg GCB and 25 mg PSA to 1 mL raw extract of a mixed spinach–grape–onion sample (1:1:1, 1 g ml^-1). Seventy pesticides (the ones in Table 8 with API2000 in the MS-MS column) were added after clean-up and analyzed by LC–MS–MS. The response was compared with that of solutions of equal concentration in the raw extract and a solvent standard. Clean-up increased the number of pesticides for which no pronounced matrix effect (less than 20% suppression or enhancement) was observed from 38 to 84%. Several of the pesticides (Tables 2 and 4) were adsorbed by the SPE material, however. Although adsorption by the GCB could have been avoided or reduced by addition of toluene (although less practical when changing from extraction solvent to methanol/water), it was concluded that PSA was not compatible with a generic method for pesticides amenable to LC–MS–MS. It was therefore decided not to include a clean-up step for LC–MS–MS analysis and to use the initial raw ethyl acetate extract. Another reason for not further pursuing clean-up in LC–MS–MS analysis was that the sensitivity of current triple-quadrupole instruments enables injection of only small amounts of matrix into the LC–MS–MS system (e.g. 2 mg) while still achieving the desired limits of quantification. Experiments showed that tenfold dilution of 1 g mL⁻¹ extracts increased the number of pesticides for which no pronounced matrix effect occurred from 65 to 82% and from 10 to 65% for cucumber and cabbage, respectively.

Routine experience with LC–MS–MS analysis for over four years, both with the API2000 (20 mg matrix) and the API3000 (2 mg matrix) has shown that injection of uncleaned extracts does not result in special maintenance requirements. The source is cleaned with a tissue daily. The LC column typically lasts for 6 months.

Changing the solvent

Because ethyl acetate is less suitable for direct injection in reversed phase LC, the solvent was changed. Because only small amounts of the raw extract need to be evaporated (less than 0.5 mL in the final method) and evaporation blocks enable simultaneous evaporation of many (typically 24–36) extracts, this step adds little to the overall sample-preparation time. Changing the solvent was even regarded as advantageous. It resulted in more freedom in selection of the final solvent to be injected into the LC, which can be critical for very polar compounds (e.g. in acetonitrile-based extraction methods, injection of 100% acetonitrile easily leads to band-broadening for methamidophos). It is also easier to compensate for the smaller amount of sample processed for dry crops (because of the need for addition of water) by evaporating a larger amount of the ethyl acetate extract.

In previous work [15] a small amount of a diethylene glycol (added as solution in methanol) was added, because this was found to facilitate reconstitution, thereby improving recovery for some pesticide–matrix combinations. It was also shown that the evaporation step did not require special attention and that continuing the process for another half hour after completion of evaporation of the solvent did not affect recovery. The same procedure was therefore used here without re-evaluating the real need for it. Reconstitution was performed by first dissolving in methanol (ultrasonication) and then dilution with LC mobile phase component A.

Validation of LC–MS–MS method

The LC–MS–MS method was validated in three separate studies, one using the API2000 with injection of 20 mg matrix equivalent and the other two using the API3000 with injection of 2 mg matrix equivalent. A total of 140 pesticides and degradation products were included. In contrast with the full-scan acquisition in GC–MS, in LC–MS–MS data were acquired for a fixed, limited, set of pesticides. Although many pesticides from the GC–MS method can also be analyzed by LC–MS–MS, emphasis was on pesticides that were not, or less, amenable to GC analysis.

Recovery, based on matrix-matched calibration, and repeatability were evaluated at the 0.01 and 0.1 mg kg⁻¹ level for vegetable and fruit matrices; the results are listed in Table 8. Although acceptable performance data were obtained for most of the pesticides, low recovery and/or high variability were observed for some. Among these were compounds that were also reported as troublesome by other workers using alternative multi-residue methods, e.g. asulam [30]. Low recovery could be partly attributed to poor extraction efficiency (asulam, hymexazol, and, in orange, propamocarb) or degradation during sample preparation (cycloxydim, sethoxydim, profoxydim, tepraloxydim, dichlofluanide, tolylfluanide, thiodicarb, thiophanate-methyl, and, in lettuce, disulfoton and furathiocarb). The degradation seems to be related to the change of solvent, as is apparent from comparison of GC–MS and LC–MS–MS validation data for dichlofluanide, tolylfluanide, and disulfoton. Fortunately, for many of these the degradation products formed are also part of the residue definition and are included in the method. Indeed, elevated recovery was observed for the degradation products when determined in the same validation set as the parent compound. In the analysis, therefore, degradation is not necessarily a problem, because the results (expressed as defined in the residue definition) have to be summed. In routine analytical quality control (see below) the data were evaluated this way.

Analytical quality-control data from routine LC–MS–MS analysis

In the same way as for GC–MS analysis, the initial validation data are continually being supplemented by performance data generated as part of analytical quality control during routine analysis of samples. With each set of analytical samples at least one was fortified with the full quantitative suite (i.e. 136 pesticides and degradation products) at the 0.05 mg kg⁻¹ level. A compilation was made from all the data generated over a period of 12 months, which included data for more than one hundred vegetable and fruit matrices. A limited number of dry matrices (flour, milk powder) were also included in the set. The data were evaluated for one transition for each pesticide, using the API3000 and injection of 2 mg equivalent of matrix (10 μL of a 0.2 g mL⁻¹ extract). Examples of typical extracted ion chromatograms are shown in Fig. 5.

Fig. 5.

Typical extracted ion chromatograms obtained by LC–MS–MS analysis of vegetable and fruit extracts (calibration standard in mango matrix, 10 pg μL, corresponding to 0.05 mg kg⁻¹)

For all fortified samples the matrix effect was also established by analyzing the corresponding matrix-matched standard, at the same level as in the extract of the fortified sample, against a solvent standard. Suppression (or enhancement) of up to 20% was regarded as acceptable for quantification. The number of compounds for which the response in matrix relative to that in solvent was between 80 and 120% is given in Table 9 for each matrix. Whereas for beetroot, asparagus, and kangkung little or no matrix effects exceeding 20% were observed, such effects were much more common for herbs and citrus fruits.

Table 9.

Overview of matrix effects and recovery^a within or outside the EU 60–140% criterion [37] after LC–MS–MS analysis

	N	Matrix effects			n*	Recovery
		# Pesticides				# Pesticides
		Rel. resp. 80–120%	>20% suppr.	>20% enhanc.		Calc. using solvent std			Calc. using matrix-matched std
						60–140%	<60%	>140%	60–140%	<60%	>140%
Corn syrup (2/2)	135	134	0	1	104	97	4	3	99	4	1
Beetroot	135	133	1	1	104	101	3	0	101	3	0
Corn syrup (1/2)	135	132	2	1	104	98	4	2	100	2	2
Kangkung	135	132	2	1	104	91	11	2	94	8	2
Green pea	135	131	3	1	104	97	5	2	99	3	2
Asparagus	135	130	4	1	104	97	7	0	98	6	0
Coco nut	135	130	4	1	104	63	41	0	59	45	0
Papaya	135	130	3	2	104	96	4	4	98	4	2
Cauliflower	135	129	1	5	104	101	2	1	102	1	1
Fennel	135	129	4	2	104	100	3	1	101	3	0
Cherry (2/3)	135	128	7	0	104	100	4	0	100	4	0
Cherry (1/3)	135	127	7	1	104	92	8	4	98	2	4
Ladies’ fingers	135	127	8	0	104	97	7	0	97	7	0
Mango (1/2)	135	127	6	2	104	97	3	4	98	2	4
Cherry (3/3)	135	126	8	1	104	100	4	0	102	2	0
Mango juice	135	126	3	6	104	101	1	2	104	0	0
Mushroom	135	126	7	2	104	102	2	0	103	0	1
Taro	135	126	7	2	104	96	4	4	99	1	4
Plum (3/3)	135	125	8	2	104	95	7	2	100	4	0
Fennel leaves (2/2)	135	124	5	6	104	99	2	3	99	2	3
Milk powder	135	124	6	5	104	58	45	1	59	45	0
Grape	135	123	9	3	104	98	3	3	98	3	3
Spinach	135	123	12	0	104	94	8	2	96	5	3
Tamarind	135	123	8	4	104	67	37	0	79	25	0
Cassava	135	122	7	6	104	87	16	1	78	26	0
Raspberry (1/3)	135	122	12	1	104	84	20	0	92	12	0
Sweet pepper	134	122	10	2	103	100	1	2	100	1	2
Apple puree	135	121	5	9	104	99	5	0	97	7	0
Corn flour	135	121	1	13	104	95	6	3	95	7	2
Courgette	135	121	7	7	104	100	2	2	100	3	1
Tomato puree	135	121	10	4	104	101	3	0	103	1	0
Raspberry (2/3)	135	120	15	0	104	98	5	1	100	3	1
Broccoli	135	119	14	2	104	90	10	4	93	8	3
Flour (2/2)	135	119	2	14	104	95	2	7	97	3	4
Peach (1/2)	135	119	16	0	104	99	5	0	100	4	0
Mango (2/2)	134	117	12	5	103	96	6	1	100	3	0
Milk/flour mix	135	117	12	6	104	43	60	1	55	49	0
Bitter cucumber	135	116	17	2	104	99	2	3	99	1	4
Melon puree	135	116	18	1	104	99	5	0	103	1	0
Tomato	135	116	13	6	104	93	8	3	96	5	3
Lettuce, crinkley	134	114	19	1	103	97	3	3	97	2	4
Pear	134	114	14	6	103	97	6	0	99	4	0
Flour (1/2)	135	113	14	8	104	73	28	3	85	19	0
Plum (1/3)	135	113	13	9	104	93	6	5	98	2	4
Celery leaves (1/3)	135	112	22	1	104	90	12	2	97	3	4
Purselane	135	112	23	0	104	96	6	2	98	4	2
Apricots	135	111	23	1	104	90	13	1	97	6	1
Artichoke	135	111	17	7	104	91	12	1	95	8	1
Cucumber	135	110	15	10	104	99	5	0	101	3	0
Horseradish powder	135	110	15	10	104	88	11	5	97	5	2
Tarrragon (2/2)	135	110	8	17	104	96	4	4	94	6	4
Avocado (1/2)	135	109	22	4	104	81	21	2	90	13	1
Haricot bean	135	109	25	1	104	83	20	1	90	13	1
Kiwi	135	109	10	16	104	97	6	1	100	2	2
Peach (12/2)	135	108	24	3	104	88	14	2	93	9	2
Raspberry (3/3)	135	107	26	2	104	80	22	2	90	11	3
Blackberry	133	106	17	10	102	91	10	1	91	9	2
Diced pumpkins	135	106	27	2	104	95	8	1	100	3	1
Plum (2/3)	135	106	23	6	104	86	18	0	85	18	1
Yam	135	106	1	28	104	97	6	1	96	8	0
Avocado (2/2)	134	103	29	2	103	68	34	1	80	22	1
Dill leaves	135	103	15	17	104	94	7	3	93	9	2
Honey	106	103	3	0	82	82	0	0	82	0	0
Chervil	135	102	29	4	104	95	9	0	98	5	1
Parsley	135	102	29	4	104	95	4	5	99	1	4
Nectarine	134	101	29	4	103	92	8	3	98	4	1
Bean sprouts	106	100	5	1	82	76	6	0	78	4	0
Sweetcorn (1/2)	106	99	6	1	82	76	5	1	77	3	2
Beetroot leaves	135	98	32	5	104	85	19	0	99	5	0
Chestnuts	106	98	1	7	82	76	4	2	79	3	0
Pomegranate (1/2)	135	97	37	1	104	84	20	0	100	4	0
Pomegranate (2/2)	135	97	37	1	104	84	20	0	100	4	0
Pear syrup	106	95	3	8	82	79	3	0	80	2	0
Alfalfa	106	94	11	1	82	75	7	0	78	4	0
Fennel leaves (1/2)	106	92	8	6	82	74	5	3	76	2	4
Chili pepper	135	91	40	4	104	95	8	1	101	1	2
Turnip tops	106	90	15	1	82	76	2	4	78	0	4
Blueberry	135	89	43	3	102	66	36	0	91	11	0
Litchi	135	88	45	2	104	78	26	0	99	4	1
Salak	135	88	42	5	104	82	20	2	99	4	1
Pepper powder	106	87	16	3	82	54	27	1	70	11	1
Celery leaves (2/3)	135	85	41	9	104	93	10	1	100	2	2
Lemon	134	84	47	3	104	78	20	6	97	3	4
Physalis	135	83	48	4	104	71	33	0	99	5	0
Maize (feed)	135	81	53	1	104	95	6	3	93	3	8
Sweetcorn (2/2)	135	80	50	5	104	79	22	3	98	6	0
Coriander (1/2)	135	79	56	0	104	68	34	2	95	6	3
Mangostan	135	76	40	19	104	46	54	4	69	35	0
Celery leaves (3/3)	134	75	58	1	103	86	16	1	99	2	2
Laos	135	73	57	5	104	70	33	1	99	4	1
Chives	135	71	57	7	104	98	5	1	102	1	1
Coriander (2/2)	135	65	60	10	104	83	21	0	98	6	0
Tea (black)	136	65	69	2	104	60	43	1	87	14	3
Lemon puree	135	53	80	2	104	68	36	0	103	1	0
Ginger	135	46	86	3	104	68	34	2	98	3	3
Grapefruit (1/2)	133	46	87	0	102	43	59	0	98	1	3
Grapefruit (2/2)	135	46	88	1	103	61	41	1	97	3	3
Oregano	135	46	75	14	104	52	50	2	87	16	1
Kumquat	135	38	95	2	104	47	56	1	94	6	4
Lime	134	38	94	2	103	48	52	3	96	4	3
Tarrragon (1/2)	135	38	95	2	104	41	63	0	90	13	1
Italian herb mix	135	33	101	1	104	54	49	1	95	8	1
Total QC results	13497	10488	2566	443	10395	8618	1613	164	9533	708	154
Percentage of total results		78	19	3		83	16	2	92	7	1

^aRecovery at 0.05 mg kg⁻¹ (higher for seven pesticides). The pesticides included are listed in Table 10

N is the total number of individual compounds (pesticides and metabolites) added to the matrix

n* is the total number of pesticides added to the matrix. Compounds belonging to the same residue definition counted as one

In contrast with GC, for which matrix effects are mainly caused by shielding of active sites in the inlet and were, to some extent predictable (in relation to the matrix load injected and the lability and/or polarity of analyte), in LC–MS–MS matrix effects are much less predictable. Although they do depend on the amount of matrix introduced into the system, and also tend to be more abundant in complex (“aromatic”) matrices, it cannot be readily predicted for which pesticides the effects occur. For this reason use of one matrix-matched standard as representative calibrant for a whole range of commodities, which worked reasonably well in GC–MS analysis, was not feasible in LC–MS–MS analysis. Consequently, critical evaluation of the matrix effect was required; if unacceptable suppression occurred there was no alternative to quantification by use of the appropriate matrix-matched calibration standard or, when not available, by standard addition.

Recovery of the pesticides from the fortified samples was calculated relative to that from a solvent standard and a matrix-matched standard and tested against the 60–140% criterion for evaluation of routine analytical quality-control samples [37]. A total of more than 10,000 recovery values were evaluated. Without matrix-matched calibration, acceptable recovery was obtained for 83% of the pesticides. Deviating recoveries were usually too low, mainly because of ion suppression, as is apparent from the results obtained from determination of recovery using matrix-matched calibration, for which 92% met the criterion.

Concentrating on performance at the pesticide level (Table 10) enables easy identification of troublesome pesticides. All compounds belonging to the same residue definition were summed (according to the residue definition) and counted as one, thereby compensating for possible conversion during sample pretreatment. This way the low recovery of dichlofluanide and the corresponding high recovery of DMSA were acceptable for most matrices because recovery for the sum met the criterion. Pesticides for which multi-matrix analysis under fixed conditions was less favorable included asulam, bifenazate, cyromazine, furathiocarb, propamocarb, pymetrozine, and thiocyclam (low recovery because of varying extraction efficiency and/or degradation). As already observed during validation, the method was also less suitable for cycloxydim, profoxydim, sethoxydim, and tepraloxydim. For these compounds recovery was too high, possibly because of degradation in the calibration standard used for preparation of the matrix-matched standards.

Table 10.

Recovery over all matrices (LC–MS–MS)

		# ACQ samples	# Recov. 60–140%	# Recov. <60%	# Recov. >140%	Average recov. (%)a	RSD (%)a
1	Abamectin	102	100	2	0	86	17
2	Acephate	102	93	9	0	78	13
3	Acetamiprid	102	97	5	0	90	11
	Aldicarb	102	101	0	1	91	13
	Aldicarb-sulfone	102	102	0	0	92	12
	Aldicarb-sulfoxide	102	96	6	0	84	13
4	Asulam	102	69	32	1	85	17
5	Azamethiphos	102	102	0	0	89	12
6	Azinfos-methyl	102	96	5	1	87	15
7	Bendiocarb	93	93	0	0	88	12
8	Bifenazate	98	60	37	1	85	18
9	Bitertanol	102	98	4	0	84	15
	Butocarboxim	102	101	1	0	88	14
	Butoxycarboxim	102	101	1	0	91	12
10	Carbaryl	102	100	1	1	87	13
	Carbendazim	100	97	2	1	93	14
	Carbofuran	102	100	1	1	92	12
	Carbofuran,3-hydroxy-	102	102	0	0	93	11
11	Carboxin	102	97	5	0	84	13
12	Chlorbromuron	102	98	4	0	86	14
13	Chlorfluazuron	102	93	8	1	87	15
14	Clofentezine	102	89	13	0	80	15
15	Clomazone	93	89	3	1	85	12
16	Clothianidin	93	91	2	0	91	12
17	Cycloxydim	102	68	11	23	104	19
18	Cymoxanil	102	102	0	0	91	15
19	Cyromazine	102	49	53	0	74	12
20	Demeton	102	102	0	0	89	14
	Demeton-S-methyl	102	100	2	0	87	14
	Demeton-S-methylsulfone	102	101	1	0	91	12
21	Desmedipham	102	96	6	0	83	14
	Dichlofluanid	102	36	66	0	80	19
22	Dicrotophos	102	100	2	0	89	14
23	Diflubenzuron	102	98	4	0	82	15
24	Dimethirimol	93	90	3	0	89	11
	Dimethoate	102	101	1	0	90	12
25	Diniconazole	93	84	8	1	86	16
	Disulfoton	93	67	25	1	75	13
	Disulfoton-sulfone	93	93	0	0	88	12
	Disulfoton-sulfoxide	93	89	0	4	96	16
26	Diuron	93	92	1	0	87	14
	DMSA	102	41	0	61	109	17
	DMST	102	96	1	5	104	16
	Ethiofencarb	102	99	3	0	86	14
	Ethiofencarb-sulfone	102	102	0	0	90	13
	Ethiofencarb-sulfoxide	102	101	1	0	92	15
27	Ethirimol	102	98	4	0	88	12
28	Famoxadone	102	95	7	0	83	14
	Fenamiphos	102	100	2	0	89	14
	Fenamiphos-sulfone	102	102	0	0	91	12
	Fenamiphos-sulfoxide	93	92	1	0	90	11
29	Fenhexamid	102	96	6	0	85	12
30	Fenpyroximate	102	92	10	0	87	13
	Fensulfothion	102	102	0	0	88	11
	Fensulfothion-sulfone	93	91	2	0	85	12
	Fenthion	102	99	3	0	87	14
	Fenthion-sulfone	102	99	2	1	88	15
	Fenthion-sulfoxide	102	102	0	0	93	14
31	Flucycloxuron	102	94	8	0	88	15
32	Flufenoxuron	102	93	9	0	87	14
33	Fosthiazate	93	93	0	0	90	12
34	Furathiocarb	102	79	20	3	84	16
35	Hexaflumuron	102	90	10	2	85	18
36	Hexythiazox	102	91	11	0	85	15
37	Imazalil	101	92	9	0	83	14
38	Imidacloprid	102	99	3	0	90	14
39	Indoxacarb	101	96	5	0	86	16
40	Iprovalicarb	93	92	1	0	87	13
41	Isoxaflutole	93	83	10	0	82	14
42	Linuron	102	97	4	1	85	12
43	Metamitron	102	97	5	0	88	15
44	Methabenzthiazuron	93	93	0	0	88	13
45	Methamidophos	102	90	12	0	75	12
	Methiocarb	102	100	2	0	85	13
	Methiocarb-sulfone	102	84	18	0	78	15
	Methiocarb-sulfoxide	102	99	2	1	88	12
46	Methomyl	102	89	0	13	101	14
47	Methoxyfenozide	102	101	1	0	85	14
48	Metobromuron	102	97	4	1	87	12
49	Metoxuron	93	93	0	0	89	12
50	Monocrotophos	102	101	1	0	90	12
51	Monolinuron	102	101	1	0	86	14
	Omethoate	102	99	3	0	83	12
	Oxamyl	102	100	2	0	89	12
	Oxamyl-oxime	102	101	1	0	88	12
52	Oxycarboxin	102	102	0	0	91	12
	Oxydemeton-methyl	102	97	5	0	86	13
53	Paclobutrazole	102	101	1	0	87	12
54	Pencycuron	102	96	6	0	81	14
	Phenmedipham	102	94	7	1	83	14
	Phenmedipham-metabolite	102	100	2	0	93	15
	Phorate	102	68	34	0	74	19
	Phorate-sulfone	93	93	0	0	88	12
	Phorate-sulfoxide	102	101	1	0	90	12
55	Phosphamidon	93	93	0	0	89	10
56	Picolinafen	93	86	6	1	84	15
	Pirimicarb	102	101	0	1	89	12
	Pirimicarb, desmethyl-	102	100	1	1	90	12
57	Prochloraz	101	94	7	0	83	14
58	Profoxydim	99	54	32	13	99	21
59	Propamocarb	101	9	92	0	70	15
60	Propoxur	102	100	2	0	88	16
61	Pymetrozine	102	73	29	0	89	20
62	Pyraclostrobin	102	95	7	0	85	14
63	Pyridate-metabolite	102	92	9	1	86	15
64	Rotenone	102	93	9	0	81	15
65	Sethoxydim	102	72	3	27	106	19
66	Spinosyn-A	93	88	5	0	82	17
	Spinosyn-D	93	82	11	0	83	15
67	Tebuconazole	93	90	3	0	86	16
68	Tebufenozide	102	99	3	0	86	14
69	Temephos	102	94	8	0	87	16
70	Tepraloxydim	102	62	0	40	114	14
	Terbufos	93	62	30	1	77	15
	Terbufos-sulfone	93	90	3	0	86	13
	Terbufos-sulfoxide	93	92	1	0	88	12
71	Thiabendazole	98	92	5	1	86	13
72	Thiacloprid	93	90	3	0	88	12
73	Thiametoxam	93	91	2	0	89	13
74	Thiocyclam	93	64	29	0	78	16
	Thiodicarb	102	62	40	0	82	16
	Thiofanox	102	98	3	1	85	14
	Thiofanox-sulfone	102	102	0	0	90	13
	Thiofanox-sulfoxide	102	101	1	0	92	14
75	Thiometon	93	88	4	1	87	16
	Thiophanate-methyl	102	83	19	0	77	12
	Tolylfluanid	101	36	65	0	76	22
	Triadimefon	102	99	3	0	85	13
	Triadimenol	102	98	3	1	87	12
76	Triazoxide	102	90	9	3	84	16
77	Trichlorfon	102	101	0	1	87	12
78	Tricyclazole	102	96	6	0	87	12
79	Triflumuron	101	89	10	2	84	18
80	Triforine	102	97	3	2	87	15
81	Vamidothion	102	101	1	0	89	11
82	Sum aldicarb	102	101	1	0	88	11
83	Sum butocarboxim	102	101	1	0	90	11
84	Sum carbendazim	101	97	4	0	83	12
85	Sum carbofuran	102	102	0	0	92	10
86	Sum dimethoate	102	100	2	0	86	10
87	Sum dichlofluanid	102	89	1	12	107	17
88	Sum disulfoton	93	89	4	0	86	13
89	Sum ethiofencarb	102	102	0	0	89	11
90	Sum fenamiphos	102	101	1	0	90	11
91	Sum fensulfothion	102	102	0	0	86	11
92	Sum fenthion	102	102	0	0	89	12
93	Sum methiocarb	102	100	2	0	83	12
94	Sum methomyl	102	100	2	0	87	12
95	Sum oxamyl	102	101	1	0	88	10
96	Sum oxydemeton-methyl	102	101	1	0	88	11
97	Sum phenmedipham	102	101	1	0	88	13
98	Sum phorate	102	97	5	0	81	12
99	Sum pirimicarb	102	101	1	0	90	12
100	Sum terbufos	93	88	5	0	81	13
101	Sum thiofanox	102	102	0	0	89	11
102	Sum tolylfluanid	101	95	6	0	80	15
103	Sum triadimefon	102	99	3	0	86	13

^aAverage and RSD for recoveries within 60–140% range

Matrix-matched calibration, API3000

Level = 0.05 mg kg⁻¹ for most pesticides/metabolites

Bold indicates pesticides, including metabolites that are part of residue definition, if appropriate

Averaging acceptable recoveries reveals there is some bias, because the values are mostly approximately 87% (in contrast with the GC–MS data, for which the average was approximately 100%). It was noted that for dry crops relatively low recovery (typically between 60–70%) was obtained for all pesticides. The cause is not clear. This bias can also be seen in tables in other papers (barley [26], soya grain [33]).

Independent evaluation of method performance by proficiency testing

From results obtained over the years from participation in proficiency tests, an additional and independent verification of method performance could be made. The data are summarized in Table 11 and clearly show that good quantitative data were consistently obtained from both GC–MS and LC–MS–MS, with method performance good (Z-score<2) 54 times, doubtful (2 < Z < 3) three times, and never poor. It also shows that the calibration approach (one-point calibration, tomato-matrix standard for GC and matrix-matched standard for LC) is fit-for-purpose.

Table 11.

Results from the analysis of Fapas (series 19) proficiency test samples (2003–2005)

Sample	Pesticide	MRM	Spike level added (μg kg−1)	Inter-lab. result (μg kg−1)	TNO result (μg kg−1)	Z-score TNO
#53 Apple	Fenpropathrin	GC–MS	500	405	528	1.7
	Parathion-methyl	GC–MS	70	59	47	−0.9
	Tetradifon	GC–MS	140	115	91	−0.9
	Triazofos	GC–MS	140	119	74	−1.7
	Vinchlozolin	GC–MS	60	53	53	0.0
#52 Cucumber	Iprodione	GC–MS	100	94	89	−0.3
	Methomyl	LC–MS–MS	28	25	28	0.5
	Thiabendazole	LC–MS–MS	50	128	113	−0.5
#51 Pear	Carbendazim	LC–MS–MS	150	116	60	−2.2
	Dodine	not in MRM	60	59	*	*
	Imazalil	LC–MS–MS	400	237	273	0.8
#49 Melon	Chlorpropham	GC–MS	10	9	11	1.0
	Chlorpyrifos	GC–MS	8	8	7	−0.7
	Dimethoate	LC–MS–MS	15	19	15	−0.9
	Pirimicarb	LC–MS–MS	20	19	16	−0.7
#48 Tomato	Azoxystrobin	GC–MS	Not given	201	166	−0.9
	Bifenthrin	GC–MS	Not given	83	99	0.9
	Buprofezin	GC–MS	Not given	108	131	1
	Chlorpyrifos-methyl	GC–MS	Not given	319	281	−0.6
	Procymidone	GC–MS	Not given	712	668	−0.4
#47 Grapefruit	Diazinon	GC–MS	Not given	262	294	0.6
	Heptenophos	GC–MS	Not given	168	234	1.9
	Malathion	GC–MS	Not given	715	690	−0.2
	Methidathion	GC–MS	Not given	567	540	−0.3
#46 Lettuce	Bromopropylate	GC–MS	80	67	51	−1.1
	Dimethoate	LC–MS–MS	300	285	316	0.6
	Oxadixyl	GC–MS	120	127	134	0.3
	Penconazole	GC–MS	100	82	51	−1.7
	Tolclofos-methyl	GC–MS	160	137	75	−2.1
#42 Apple	Chlorfenvinphos	GC–MS	90	71	50	−1.3
	Chlorpyrifos	GC–MS	400	259	241	−0.3
	Methamidophos	LC–MS–MS	60	44	31	−1.3
	Monocrotophos	LC–MS–MS	80	58	56	−0.1
	Omethoate	LC–MS–MS	150	108	103	−0.2
	Trifluralin	GC–MS	100	59	62	0.2
#41 Basil	Kresoxim-methyl	GC–MS	150	94	86	−0.4
	Procymidone	GC–MS	120	87	78	−0.5
	Propyzamide	GC–MS	100	81	59	−1.2
	Vinclozolin	GC–MS	60	47	44	−0.3
#38 Tomato	Azoxystrobin	GC–MS	150	137	132	−0.2
	Bupirimate	GC–MS	100	83	62	−1.1
	Chlorpyrifos-methyl	GC–MS	80	72	53	−1.2
	Quinalphos	GC–MS	140	124	105	−0.7
#37 Lemon	Diazinon	GC–MS	80	42	42	0.0
	Fenitrothion	GC–MS	100	78	80	0.1
	Metalaxyl	GC–MS	120	94	93	0
	Methidathion	GC–MS	150	109	154	1.9
#35 Lettuce	Carbendazim	LC–MS–MS	80	53	31	−1.9
	lambda Cyhalothrin	GC–MS	80	66	54	−0.8
	Metalaxyl	GC–MS	120	94	86	−0.4
#34 Apple	Diphenylamine	GC–MS	50	39	29	−1.2
	Pirimiphos-methyl	GC–MS	50	41	42	0.1
	Propargite	GC–MS	200	162	172	0.3
	Tetradifon	GC–MS	100	83	38	−2.5
#29 Sweet pepper	Dichloran	GC–MS	200	179	200	0.6
	Mecarbam	GC–MS	100	90	120	1.5
	Methamidophos	LC–MS–MS	60	51	54	0.3

Conclusions

The ethyl acetate-based multi-residue method has been modified to meet today’s demands in respect of ease and speed of sample preparation. For GC–MS analysis, combined GCB/PSA dispersive clean-up enables prolonged injection of vegetable and fruit extracts (10 mg matrix equivalent) without maintenance. Retention time shifts induced by some matrices compared with the calibration standard are reduced by the clean-up procedure. Interferences are partially removed, resulting in cleaner (extracted ion) chromatograms. The last two benefits aid correct automatic peak assignment and confirmation. Addition of toluene during dispersive clean-up prevented unacceptable adsorption of planar pesticides by GCB yet removal of chlorophyll and other pigments was still sufficient. Use of liners with a sintered porous glass bed on the inner wall makes 20 μL injection non-critical and robust. In GC, use of a universal matrix-matched standard (tomato) is a feasible means of compensating for the matrix effects of many other vegetable and fruit samples. For most pesticides, LOQs of 0.01 mg kg⁻¹ can be obtained by GC–MS with full-scan acquisition.

The same initial extract (i.e. without any clean-up) can be used for LC–MS–MS analysis, after changing the solvent to methanol–water. LC–MS–MS is relatively tolerant of injection of matrix—despite the absence of any clean-up no special maintenance was required. Matrix-induced suppression was observed for several matrices, however, especially herbs and citrus, and must be evaluated for all pesticide-matrix combinations. In contrast with the GC–based method, use of a universal matrix-matched standard to compensate for matrix effects was not feasible.

Evaluation of analytical quality control data for 271 pesticides and degradation products in over one hundred matrices showed that, at the 0.05 mg kg⁻¹ level, recovery was acceptable for 92% (LC–MS–MS) and 93% (GC–MS) of all pesticide–matrix combinations. It also revealed that the method fails in the other 7–8% because of lack of specificity (mostly in GC–MS) or because of poor extraction efficiency and/or degradation (LC–MS–MS). The only way to identify these limitations is by thorough and continual evaluation of the quantitative performance of the method for all the pesticides (rather then a “representative subset”) in all the matrices.