Mtm

Ia

b

Ac

a

ARRAA

KPSMEL

1

hv[aCo

0d

Journal of Chromatography A, 1216 (2009) 4539–4552

Contents lists available at ScienceDirect

Journal of Chromatography A

journa l homepage: www.e lsev ier .com/ locate /chroma

ethod validation and comparison of acetonitrile and acetone extraction forhe analysis of 169 pesticides in soya grain by liquid chromatography–tandem

ass spectrometry

onara R. Pizzutti a,∗, André de Kokb, Maurice Hiemstrab, Cristine Wickertc, Osmar D. Prestesc

Federal University of Santa Maria, Chemistry Department, Center of Research and Analysis of Residues and Contaminants (CEPARC), Santa Maria, RS, BrazilVWA – Food and Consumer Product Safety Authority, Chemistry Laboratory, R&D Group, National Reference Laboratory (NRL) for Pesticide and Mycotoxin Analysis in Food,msterdam, The NetherlandsFederal University of Santa Maria, Chemistry Department, Laboratory of Pesticide Residue Analysis (LARP), Santa Maria, RS, Brazil

r t i c l e i n f o

rticle history:eceived 20 October 2008eceived in revised form 23 March 2009ccepted 24 March 2009vailable online 28 March 2009

eywords:esticidesoya grainulti-residue method

xtractionC–MS/MS

a b s t r a c t

An acetonitrile-based extraction method for the analysis of 169 pesticides in soya grain, using liquidchromatography–tandem mass spectrometry (LC–MS/MS) in the positive and negative electrospray ion-ization (ESI) mode, has been optimized and validated. This method has been compared with our earlierpublished acetone-based extraction method, as part of a comprehensive study of both extraction meth-ods, in combination with various gas chromatography–(tandem) mass spectrometry [GC–MS(/MS)] andLC–MS/MS techniques, using different detection modes. Linearity of calibration curves, instrument lim-its of detection (LODs) and matrix effects were evaluated by preparing standards in solvent and in thetwo soya matrix extracts from acetone and acetonitrile extractions, at seven levels, with six replicateinjections per level. Limits of detection were calculated based on practically realized repeatability rela-tive standard deviations (RSDs), rather than based on (extrapolated) signal/noise ratios. Accuracies (as% recoveries), precision (as repeatability of recovery experiments) and method limits of quantification(LOQs) were compared. The acetonitrile method consists of the extraction of a 2-g sample with 20 mL ofacetonitrile (containing 1% acetic acid), followed by a partitioning step with magnesium sulphate and asubsequent buffering step with sodium acetate. After mixing an aliquot with methanol, the extract can beinjected directly into the LC–MS/MS system, without any cleanup. Cleanup hardly improved selectivityand appeared to have minor changes of the matrix effect, as was earlier noticed for the acetone method.Good linearity of the calibration curves was obtained over the range from 0.1 or 0.25 to 10 ng mL−1, withr2 ≥ 0.99. Instrument LOD values generally varied from 0.1 to 0.25 ng mL−1, for both methods. Matrix

effects were not significant or negligible for nearly all pesticides. Recoveries were in the range 70–120%,with RSD ≤ 20%. If not, they were still mostly in the 50–70% range, with good precision (RSD ≤ 20%). Themethod LOQ values were most often 10 �g kg−1 for almost all pesticides, with good repeatability RSDs.Apart from some minor pros and cons, both compared methods are fast, efficient and robust, with good

he twd 20

method performances. Tduring surveys in 2007 an

. Introduction

Nowadays, consumers tend to change their food consumptionabits towards a preference for healthy fruits and vegetables pro-iding also the right nutritional value. The benefits of soybean

Glycine max (L.) Merrill] and the importance to include this legumend its products in the daily diet have been very well known by thehinese people since more than 2000 years B.C. In the beginningf the 20th century, the main properties of soya, the high content

∗ Corresponding author. Tel.: +5555 3220 9458; fax: +5555 3220 9458.E-mail address: pizzutti@quimica.ufsm.br (I.R. Pizzutti).

021-9673/$ – see front matter © 2009 Elsevier B.V. All rights reserved.oi:10.1016/j.chroma.2009.03.064

o methods were applied successfully in a routine analysis environment,08.

© 2009 Elsevier B.V. All rights reserved.

of oils and proteins, called the attention of the occidental peoplethat started to use this grain not just for consumption, but also forindustrial and technical applications, such as amongst others, in theareas of cosmetics, textiles, inks, disinfectants, fertilizers and bio-fuel. However, the applications of soybean for food and feed are stillthe most important [1].

Soybean belongs, together with wheat, rice, corn and potatoes,to the five most important food and feed crops in the world. Brazil

is the second biggest producer of soya grain, but the first exporter.Seventy-five percent of the total Brazilian production of soybean isexported, mainly to China and the European Union [2]. The Nether-lands is the second biggest importer of soya grain in the world, afterChina.

4 atogr. A 1216 (2009) 4539–4552

awt1f(tvoaf

iaiqefl

ftacmam

[mpf

mahf

am[u

ai

2

tmdif

2

arPDGZ(

540 I.R. Pizzutti et al. / J. Chrom

Taking into account the continuous increase in consumption ofll products originating from soybean and other grains, legislationas established recently [3] in order to keep pesticide residues at

he lowest possible levels, and guarantee food and feed safety. SinceSeptember 2008, a new European Union regulation became into

orce setting around 145,000 harmonised maximum residue levelsMRLs) for the tens of thousands commodity–pesticide combina-ions [4]. Compliance with these MRLs requires an efficient controlia regulatory monitoring programs and enforcement, analysisf raw and processed food and feed [5], risk assessment, field-pplication trials, organic food verification [6] and border controlor international trade of food and feeding stuffs [7].

The development of sensitive, selective and reproducible analyt-cal methods and techniques has always been a prerequisite for thechievement of high-quality results in enforcement and monitor-ng programmes. Nowadays, other characteristics of an analyticaluantitative method are also requested, such as usage of the small-st possible amount of sample and chemicals, more environmentalriendly approaches, safer and less hazardous chemicals to the ana-ysts, cost-effective, less laborious and faster methods [8–10].

Until recently, most improvements in multi-residue methodsor pesticides have been achieved for crops with a high water con-ent, like fruits and vegetables. In the last years, research has nowlso been directed towards analysis of pesticide residues in moreomplex matrices, such as grains and its derived products, andatrices with a high fat and sugar content. These latter matrices

re now gradually becoming less difficult, thanks to the newest,ore advanced chromatographic detection techniques [11–15].Classical analytical methods based on acetone [16], acetonitrile

17] and ethyl acetate [18] extraction have been continuously opti-ized and modified in such a way to improve considerably their

erformances and applications for pesticide residues analysis, alsoor the more complex matrices mentioned above.

Recent versions of the acetone [12,13] and acetonitrile [19]ethods and its many modifications and applications [20–24] and

lso methanol-based [25] and ethyl acetate-based [26] methodsave achieved the state-of-the-art stage of multi-residue methods

or pesticide residues.As follow-up of our previous study [12] on the development of

n acetone-based extraction method, this study aimed at the opti-ization and validation of an acetonitrile-based extraction method

27] for the analysis of 169 pesticides in soya grain by LC–MS/MSsing both the positive and negative ESI ionization mode.

The results for the validation parameters, such as matrix effect,ccuracy, precision, linearity range, detection and quantitation lim-ts, sensitivity and selectivity, were compared for the two methods.

. Experimental

In this study, soya extract 1 means the extract obtained byhe acetone extraction [1,12] method (Fig. 1) and soya extract 2

eans the extract obtained by the acetonitrile extraction methodescribed in Section 2.5 of this paper and also shown schematically

n Fig. 2. For both methods cited, no clean up procedure was appliedor analysis by LC–MS/MS.

.1. Chemicals and reagents

Methanol, toluene, acetone, acetonitrile (pesticide grade), andnhydrous magnesium sulphate and sodium acetate (analyticaleagent grade) were purchased from Merck (Darmstadt, Germany).

esticides reference standards (purity > 97%) were obtained fromr. Ehrenstorfer (Augsburg, Germany); Riedel de Haën (Seelze,ermany); Hayashi (Osaka, Japan), Bayer (Leverkusen, Germany);eneca (Wilmington, DE, USA); Janssen (Beerse, Belgium), NovartisBasel, Switzerland), Agrevo (Frankfurt, Germany), Rohm and Haas

Fig. 1. Scheme of the acetone extraction method for the determination of pesticidesin soya grain.

(Philadelphia, PA, USA), Cyanamid (Princeton, NJ, USA) and DuPont(Wilmington, DE, USA).

2.2. Standard solutions

Standard stock solutions were prepared at 1.00 mg mL−1 mainlyin toluene. However, for those compounds with solubility problemsin toluene, methanol or acetone were used. A stock standard mix-ture solution containing all 169 pesticides studied, was preparedat 1 �g mL−1 of each pesticide, in methanol (0.1% acetic acid). Thissolution was used as spiking solution and also to prepare the stan-dard solutions to obtain the calibration curves from the pesticidesin solvent.

Blank soya extracts (1 and 2) were used for preparation ofstandard solutions in matrix. Hence, all standard solutions wereprepared in solvent, in soya extract 1 (acetone extraction) and soyaextract 2 (acetonitrile extraction) in order to evaluate the matrixeffect for each extraction method. The standards in matrix extract(matrix-matched standards) were used for the calculation of recov-eries.

2.3. Analytical instrumentation

The chromatographic system consisted of a Waters Alliance

2695 separation module (Milford, MA, USA), equipped with a qua-ternary solvent delivery system, degasser, autosampler and columnheater. An Alltima C18 analytical column (150 mm × 3.2 mm I.D.,5 �m particle size) from Alltech (Deerfield, IL, USA), fitted with a C18guard column (4 mm × 3.0 mm I.D.) purchased from Phenomenex

I.R. Pizzutti et al. / J. Chromatogr.

Ft

(ts

e(d

2

fraltt

stmsctmrtflT

ig. 2. Scheme of the acetonitrile extraction method for the determination of pes-icides in soya grain.

Torrance, CA, USA) and kept in a column heater (at a constantemperature of 35 ◦C), were applied to perform chromatographiceparations.

A Waters Quattro Ultima tandem mass spectrometer (Manch-ster, UK), equipped with an electrospray ionization (ESI) interfaceZ-spray), operating in the positive or negative ion mode, was theetection system used.

.4. Chromatographic and mass spectrometric conditions

For the ESI negative ionization mode, an isocratic run was per-ormed during a total time of 10 min, using a mobile phase (flowate at 0.3 mL min−1) constituted of 90% methanol in 5 mmol L−1

mmonium formate. A divert valve was placed between the ana-ytical column outlet and the mass spectrometer inlet, and duringhe first 1.0 min of the chromatographic run the flow was divertedo waste.

For the ESI positive ionization run mode, the gradient programtarted at 25% methanol in 5 mmol L−1 ammonium formate solu-ion and was directly ramped linearly over the first 15 min to 95%

ethanol in 5 mmol L−1 ammonium formate solution. This compo-ition was held for a further 10 min before returning to the initialondition in 0.1 min. While the elution and acquisition was con-inued until 31 min, the column was re-equilibrated at the initial

obile phase composition until 32.5 min, which was also the totalun time. The flow (0.3 mL min−1) was diverted to waste duringhe first 2.5 min of the chromatographic run. The autosampler wasushed with methanol between analytical runs to avoid carry over.he injection volume of each sample was 5 �L.

A 1216 (2009) 4539–4552 4541

The mass spectrometer ion source parameters were: capillaryvoltage 2.0 kV; sample cone voltage 35 V (constant for all pesti-cides); source temperature 100 ◦C and desolvation gas temperature350 ◦C. The flow rates for desolvation gas and cone gas (N2) wereset at 100 and 600 L h−1, respectively. Multiple reaction monitoring(MRM) experiments were conducted with a dwell time and interchannel delay time of 20 and 10 ms, respectively, for all pesticides.Ions were collisionally fragmented with argon at 3.5 × 10−3 mbar.Optimization of the collision energy (CE) for each individual pes-ticide was done by infusion of the pesticide directly into the LCeffluent using a syringe pump (Harvard, Kent, UK). For instrumentcontrol, data acquisition and processing, MassLynx and QuanLynxsoftware 4.0 was used. Precursor and product ions monitored, timewindows/functions and optimized collision energies, for both ESIpositive and negative ionization modes were described in our pre-vious paper about this comprehensive study [1,12].

2.5. Acetone and acetonitrile extractions

The acetone extraction procedure was already described in part1 of this study [12] and is just shown schematically in Fig. 1. To per-form the acetonitrile extraction (Fig. 2), an amount (2.0 ± 0.02 g)of soya grain (milled and homogenized) was weighed into a 50 mLPTFE centrifuge tube and 4.0 g of water was added, allowing soak-ing for 1 h. A volume (20 mL) of acetonitrile containing 1% of aceticacid was added to the tube, and manual shaking was performedduring ±4.5 s. Then, 2.0 g anhydrous magnesium sulphate and 2.5 gsodium acetate, were added to each tube and immediate vigor-ous, manual shaking was performed, in order to avoid coagulation,which could have a negative influence on the partitioning process.Besides, buffering was applied at this step, which provides a bet-ter stability for some pH-dependent pesticides. The acetonitrileextracts (upper layer) were decanted into other tubes, containing2.0 g of anhydrous magnesium sulphate. The tubes were shakenmanually for 20 s. After centrifugation at 3000 rpm for 1 min, analiquot of 0.5 mL extract was diluted with 0.5 mL of methanol intoan autosampler vial. The final extract before injection into theLC–MS/MS system corresponds with a matrix-equivalent concen-tration of 50 mg mL−1.

2.6. Method validation study

One single chromatographic run was performed to analyse 155pesticides by LC–MS/MS in the ESI positive mode and another runfor the 14 pesticides analysed in the ESI negative mode.

2.6.1. Calibration curves, linearity, LOD and LOQThe evaluation of the calibration curves’ linearity was done

based on injections of the standard solutions prepared in organicsolvent (acetonitrile/methanol, 1:1) and also in blank soya extract,at the concentrations 0.1, 0.25, 0.5, 1.0, 2.5, 5.0 and 10.0 ng mL−1.These solutions were each injected six times (n = 6). The range ofpesticide concentrations in the matrix extract corresponds witha range of residue levels in the soya bean samples from 2 to200 �g kg−1.

Calculations were performed of the average peak areas, rela-tive standard deviations (RSDs) and calibration curve equations,and also the determination coefficients (r2) and linear ranges weredetermined for each pesticide analysed.

Both from the calibration curves and the repeatability (RSD)

data, at the lowest concentration levels of each individual pesti-cide of the in total 169 pesticides studied, the instrument limits ofdetection (LOD) and limits of quantification (LOQ) (LODi and LOQi,respectively) and also the method LOD and LOQ (LODm and LOQm,respectively) were estimated.

4 atogr. A 1216 (2009) 4539–4552

acTtl(trcrw

tcL

olw

2

ftsetcot

M

wtas

aai[

2

ebeTthwatwa

ecaepwu

esti

cid

esw

ith

ace

rtai

nr2

,lin

ear

ran

geof

the

cali

brat

ion

curv

es,e

stim

ated

inst

rum

ent

and

met

hod

LOD

(LO

Di

and

LOD

m),

der

ived

from

stan

dar

dso

luti

ons

pre

par

edin

solv

ent,

inbl

ank

soya

extr

act

1an

dbl

ank

btai

ned

byac

eton

ean

dac

eton

itri

leex

trac

tion

s,re

spec

tive

ly).

An

alys

isw

asp

erfo

rmed

byLC

-ESI

–MS/

MS

inth

ep

osit

ive

and

neg

ativ

em

odes

.

Perc

enta

geof

pes

tici

des

LC-E

SI–M

S/M

Sp

osit

ive

(nu

mbe

rof

pes

tici

des

:15

5)LC

-ESI

–MS/

MS

neg

ativ

e(n

um

ber

ofp

esti

cid

es:

14)

r2Li

nea

rR

ange

LOD

i(n

gm

L−1),

LOD

m

(�g

kg−1

)r2

Lin

ear

Ran

geLO

Di(n

gm

L−1),

LOD

m

(�g

kg−1

)

90%

(140

/155

)≥0

.999

90%

(140

/155

)0.

1or

0.25

–10

ng

mL−1

2.0

or5.

0–20

0�

gkg

−1

84%

(130

/155

)0.

1or

0.25

(LO

Di)

,2.0

or5.

0(L

OD

m)

93%

(13/

14)≥0

.999

70%

(10/

14)

0.1

or0.

25–1

0n

gm

L−12.

0or

5.0–

200

�g

kg−1

79%

(11/

14)

0.1

or0.

25(L

OD

i),2

.0or

5.0

(LO

Dm

)90

%(1

40/1

55)

0.25

–10

ng

mL−1

5.0–

200

�g

kg−1

50%

(7/1

4)≥0

.999

64%

(9/1

4)0.

1or

0.25

(LO

Di)

,2.0

or5.

0(L

OD

m)

100%

(14/

14)≥0

.99

95%

(147

/155

)≥0

,999

93%

(14

4/15

5)0.

1or

0.25

–10

ng

mL−1

2.0

or5.

0–20

0�

gkg

−1

91%

(141

/155

)or

0.25

(LO

Di)

,2.0

or5.

0(L

OD

m)

80%

(11/

14)≥0

.999

80%

(11/

14)

0.1

or0.

25–1

0n

gm

L−12.

0or

5.0–

200

�g

kg−1

50%

(7/1

4)0.

1or

0.25

(LO

Di)

,2.0

or5.

0(L

OD

m)

542 I.R. Pizzutti et al. / J. Chrom

The LODi was defined as the minimum concentration of annalyte that can be identified, measured and reported, with 99%onfidence that the analyte concentration is greater then zero.he LODi was calculated from the RSD% of the average detec-or responses (peak areas) of six replicate injections at the fourowest, detectable concentration levels, via the formula: LODing mL−1) = 3 × RSD × concentration (provided that the response ofhe blank was zero). From these calculated values, a best estimated,ounded LODi value was established. As a consequence, this con-entration should always have been really injected and detectableepeatedly (with a signal-to-noise ratio >2) all six times at that level,hile the RSD should not have exceeded 33%.

The estimated LOQi was defined as the minimum concentra-ion of an analyte that can be identified and quantified with 99%onfidence. The (estimated) LOQi was calculated via the formula:OQi = 10 × RSD × concentration, that is LOQi = 3.3 × LODi.

The real LOQm was based on the accuracy and precision data,btained via the recovery determinations and was defined as theowest validated spike level meeting the requirements of a recovery

ithin the range 70–120% and a RSD ≤ 20% [28].

.6.2. Matrix effect evaluation and comparisonThe matrix effect was evaluated by comparing the peak areas

rom standard solutions (n = 6) of the 169 pesticides in solvent (ace-onitrile/methanol, 1:1) with the peak areas obtained from standardolutions, of the same pesticides, prepared either in blank soyaxtract 1 or blank soya extract 2, at all seven concentrations men-ioned in Section 2.6.1. This way, it was possible to calculate andompare the positive or negative matrix effect, that is an increaser decrease of the detector response, respectively, for both extrac-ion methods. The matrix effect was calculated via the formula:

atrix effect (%) =[

(X2 − X1)X1

]× 100 (%)

here X1 = average area of the pesticide standard in solvent (ace-onitrile/methanol, 1:1), at a specific concentration; X2 = averagerea of the pesticide standard in blank soya extract 1 or 2, at theame concentration.

When the average matrix effect exceeds circa 20%, it can gener-lly be considered to having a significant effect on the quantitativenalytical results [29], but it is also necessary to take the repeatabil-ty (expressed as RSD values) of the average peak areas into account30].

.6.3. Accuracy and precisionThe accuracy and precision of both the acetone and acetonitrile

xtraction method were evaluated through recovery experimentsy spiking pesticides to a blank soya sample at three different lev-ls (10, 50 and 100 �g kg−1), six replicates at each level (n = 6).he spiking procedure was performed by adding the standard mix-ure solution containing 169 pesticides (Section 2.2) to milled andomogenized soya grain that had been previously soaked withater during 1 h, before applying either the acetone (Fig. 1) or

cetonitrile extraction method (Fig. 2). The contact time of the pes-icides with the wetted sample before addition of extraction solventas kept at a minimum. Each blank soya extract (1 and 2) were also

nalysed six times.In order to check whether the evaporation procedure (a gen-

ral step applied in many methods for pesticide residues analysis)ould cause losses of some pesticides, the spiking procedure (only

t 100 �g kg−1) was also done after the extraction and before thevaporation steps. The blank soya extract 1 (10 mL) was theretolaced into a 25 mL tube and the evaporation was conducted in aater bath, with a bath temperature starting at 45 ◦C and contin-ing to 62 ◦C, until near dryness. The remaining residue of solvent Ta

ble

1Pe

rcen

tage

ofp

soya

extr

act

2(o

Solv

ent

Soya

extr

act

1

Soya

extr

act

2

I.R. Pizzutti et al. / J. Chromatogr. A 1216 (2009) 4539–4552 4543

Table 2Average matrix effect (in %, n = 6) for a selected number of pesticides, analysed by LC–MS/MS (ESI positive mode). Soya extract 1 and 2 are obtained from acetoneand acetonitrile extraction, respectively.

Pesticide Matrix effect (%)*

Soya extract 1 Soya extract 2

Concentration (ng mL−1) Concentration (ng mL−1)

5 2.5 5 2.5

Acephate −2 4 −74 −70Methamidophos −22 −18 −77 −75Aldicarb sulfoxide 28 14 −70 −76Azamethiphos −23 −12 0 13Boscalid −25 −10 1 −6Butocarboxim sulfoxide 7 2 −52 −52Omethoate −5 −1 −75 −75Fenoxycarb −36 −29 −36 −41Fenpyroximate −25 −13 −21 −10Flusilazole −26 −22 −12 −8Mepanipyrim −42 −35 −17 −24Paclobutrazole −39 −41 −16 −14Spinosad A/D −26 −22 −25 −5

k area

wpser5t(

c

3

3

ticcm(

Fp

Tebufenozide −36Vamidothion sulfoxide 0

* Matrix effect (%) = [(X2 − X1)/X1] × 100 (%), where X1 and X2 = average pea

as allowed to evaporate in the air (in a fume hood) at room tem-erature. The extract reconstitution was done by adding 1.0 mL ofolvent or standard mixture solution (50 ng mL−1). The blank soyaxtract 2 (5 mL) was transferred to a 25 mL tube and the evapo-ation was conducted in a RapidVap System (speed: 50%; temp.:0 ◦C; time: 20 min; vacuum: 160 mbar). The residue was reconsti-uted in 1.0 mL of methanol or 1.0 mL of a standard mixture solution50 ng mL−1).

Then, the obtained recoveries and RSD data were evaluated andompared, considering the different stages of spiking.

. Results and discussion

.1. Calibration curves, linearity, LOD and LOQ

In Table 1, the summarized results for the linearity of the calibra-ion curves are given for the pesticides standard solutions prepared

n solvent and those prepared either in soya extract 1 or 2. Theoefficient of determination (r2) and linear range of the calibrationurves, both for pesticides analysed by ESI positive and negativeodes, are shown. Approximately 90% (140 pesticides) and 95%

147 pesticides), respectively, of the pesticides in solvent or in soya

ig. 3. Comparison of number and percentage of pesticides with a LOQm of 10, 50 or 100 �erforming the acetonitrile and acetone extraction method and LC–MS/MS analysis in the

−47 −22 −231 −76 −66

of the pesticide standard solution in solvent and soya extract, respectively.

extract 2, analysed by ESI positive mode, showed r2 ≥ 0.999 and alinear range between 0.1 or 0.25 and 10.0 ng mL−1. The same resultswere obtained in the ESI negative mode for 93% (13) and 80% (11) ofthe pesticides in solvent and soya extract 2, respectively. This corre-sponds with an application range with equivalent pesticide residuelevels in the sample of 2 or 5 to 200 �g kg−1.

The estimated, rounded instrument LODi values, based on theRSD of repeatedly injected solutions (n = 6) at that concentration,were 0.1 or 0.25 ng mL−1 for 91% and 50% of the pesticides in matrixextract 2, detected in the positive and negative mode, respectively,compared with corresponding percentages of 84% and 79% for thepesticides in pure solvent. Based on the LODi, corresponding LOQm

values could be estimated of 2 or 5 �g kg−1, being near the targetmethod LOQ of 10 �g kg−1, which was therefore also selected as thelowest spike level for the recovery studies.

When comparing the results of soya extract 2 with those of soyaextract 1, it can be noticed from Table 1 that for the 155 pesticides

determined by the positive mode, the results obtained from soyaextract 2 are slightly better than those from soya extract 1. However,the reverse situation can be observed for the pesticides analysedby the negative mode, which means that the soya extract 1 seemsslightly better.

g kg−1, not detected (n.d.) or not qualifying for the quantitation criteria (n.q.), whenESI positive and negative modes.

4544 I.R. Pizzutti et al. / J. Chromatogr. A 1216 (2009) 4539–4552

Table 3Average recovery and RSD (in %, n = 6) and method LOQm, obtained by acetonitrile extraction of soya samples, spiked at 10, 50 and 100 �g kg−1, and analysed by LC–MS/MS(ESI positive mode).

Pesticide Spike levels (�g kg−1) LOQm (�g kg−1)

100 50 10

Recovery (%) RSD (%) Recovery (%) RSD (%) Recovery (%) RSD (%)

Acephate 78 6.0 84 11.8 66 9.3 50Methamidophos 73 8.4 67 8.4 64 14.8 100Acetamiprid 92 3.0 88 5.3 81 6.0 10Aldicarb sulfoxide 79 8.3 75 2.4 69 11.6 50Aldicarb sulfone 92 4.6 88 5.2 81 6.0 10Azaconazole 101 1.1 96 2.9 81 8.9 10Azamethiphos 93 2.4 93 7.0 83 2.6 10Azinphos-methyl 89 7.0 81 12.0 91 12.7 10Azoxystrobin 95 3.1 89 5.1 77 5.6 10Bitertanol 99 6.9 93 7.9 88 27.1 50Boscalid 94 4.1 89 1.7 84 10.1 10Bromuconazole 95 4.9 94 1.0 87 17.1 10Bupirimate 86 5.8 88 9.8 81 5.4 10Ethirimol 61 2.7 49 19.1 50 12.4 n.q.Buprofezin 88 4.4 84 4.8 68 11.0 50Butocarboxim sulfoxide 76 5.8 75 9.9 79 22.0 50Butocarboxim sulfone 88 2.5 80 3.0 80 3.0 10Carbaryl 92 5.4 89 7.9 85 6.7 10Carbendazim 84 3.1 80 7.9 77 5.4 10Carbofuran 95 1.9 93 4.6 82 2.3 10Carbofuran, 3-hydroxy 92 3.8 83 4.0 86 3.9 10Carpropamid 94 3.8 93 0.9 88 6.8 10Chlorbromuron 85 5.9 89 6.1 69 20.2 50Clofentezine 90 10.3 88 3.6 69 18.2 50Clothianidin 88 2.9 85 5.3 80 3.4 10Cyazofamid 76 6.2 81 5.7 76 5.8 10Cymoxanil 80 0.9 82 7.0 70 17.5 10Cyproconazole 102 4.6 102 2.5 77 21.3 50Cyprodinil 95 1.6 93 2.1 85 5.1 10Demeton-O-sulfoxide 96 2.0 88 5.7 81 4.6 10Oxydemeton-methyl 90 2.0 86 3.3 77 5.2 10Demeton-S-methyl. sulfone 88 2.7 86 3.7 80 3.8 10Desmedipham 45 9.2 57 13.0 48 12.4 n.q.Dichlofluanid 24 14.2 34 20.1 25 29.4 n.q.Dichlorvos 83 4.2 98 8.1 79 3.2 10Trichlorphon 11 33.0 26 38.5 14 48.0 n.q.Dicrotophos 90 4.0 91 9.9 78 18.0 10Diethofencarb 94 2.1 90 5.0 84 3.7 10Difenoconazole 153 7.4 153 1.2 148 9.5 n.q.Dimethoate 95 2.3 91 5.8 79 9.0 10Omethoate 69 6.7 75 6.5 72 12.3 10Dimethomorph 97 2.2 94 4.0 87 8.0 10Dimoxystrobin 100 3.5 91 4.5 70 5.8 10Diniconazole 98 9.7 98 4.7 90 23.4 50Diuron 88 4.5 83 6.8 80 7.7 10Dodemorph 69 4.7 88 5.9 70 6.9 10Epoxiconazole 103 3.6 94 6.2 89 7.0 10Ethiofencarb sulfoxide 78 4.3 78 5.5 75 4.7 10Ethiofencarb sulfone 68 6.6 69 4.9 65 7.5 n.q.Ethiprole 83 4.2 85 3.6 92 35.0 50Etofenprox 60 35.0 109 8.2 66 12.5 n.q.Famoxadone 95 9.0 94 12.1 91 33.7 50Fenamidone 98 4.7 95 6.1 78 9.0 10Fenamiphos 76 3.2 83 17.1 75 8.9 10Fenarimol 95 3.9 92 8.2 67 9.2 50Fenazaquin 66 28.2 99 11.0 86 5.6 10Fenbuconazole 97 4.0 90 10.1 83 8.1 10Fenhexamid 94 7.2 83 4.7 67 9.8 50Fenoxycarb 99 4.3 95 4.5 79 4.8 10Fenpropidin 80 9.2 76 4.2 63 10.6 50Fepropimorph 77 15.9 109 7.7 85 8.0 10Fenpyroximate 79 16.5 74 27.0 75 15.2 10Fenthion 108 9.6 112 11.6 127 12.6 50Fenthion sulfoxide 91 2.8 91 4.5 82 6.4 10Flufenacet 101 8.7 91 9.5 89 17.0 10Fluquinconazole 97 6.8 97 6.7 102 23.5 50Flusilazole 95 8.5 92 5.4 80 13.2 10Flutolanil 104 2.8 98 4.4 85 3.2 10Flutriafol 97 2.2 94 2.9 79 3.7 10Fosthiazate 93 2.6 89 6.0 83 5.9 10Furathiocarb 89 7.6 88 2.5 73 11.6 10Halofenozide 100 5.8 92 9.9 76 11.5 10

I.R. Pizzutti et al. / J. Chromatogr. A 1216 (2009) 4539–4552 4545

Table 3 (Continued )

Pesticide Spike levels (�g kg−1) LOQm (�g kg−1)

100 50 10

Recovery (%) RSD (%) Recovery (%) RSD (%) Recovery (%) RSD (%)

Hexaconazole 105 4.3 101 4.9 48 33.3 50Hexythiazox 72 8.3 72 17.6 48 23.0 50Imazalil 77 3.3 65 10.9 67 20.8 100Imidaclopride 83 6.1 79 2.9 71 10.0 10Indoxacarb 73 10.8 70 15.1 53 28.2 50Iprovalicarb 71 4.4 61 9.8 48 16.4 100Isoprothiolane 76 2.1 72 6.9 54 22.9 50Isoxaflutole 74 3.9 66 11.7 54 27.2 100Isoxathion 70 9.0 72 11.0 54 20.1 50Kresoxim-methyl 68 8.4 69 19.7 47 24.6 n.q.Linuron 72 7.0 70 4.7 61 21.3 50Malathion 75 9.9 77 8.6 49 27.4 50Mefenacet 75 3.7 70 6.5 55 16.2 50Mepanipyrim 71 7.5 70 8.6 59 23.7 50Mephosfolan 87 4.2 80 3.8 78 3.2 10Mepronil 70 4.6 62 9.4 53 25.4 100Metalaxyl 82 4.1 75 3.5 68 6.2 50Metconazole 68 6.3 63 9.3 57 31.6 n.q.Methidathion 78 6.2 74 9.4 57 18.5 50Methiocarb 79 3.7 72 4.3 63 16.4 50Methiocarb sulfoxide 83 5.5 77 5.2 81 7.3 10Methiocarb sulfone 83 4.2 76 6.4 75 7.1 10Methomyl 102 10.8 89 9.3 107 4.7 10Thiodicarb 55 14.0 57 17.0 33 11.7 n.q.Metobromuron 79 4.2 74 2.1 62 11.5 50Metoxuron 63 7.3 60 4.5 60 11.7 n.q.Methoxyfenozide 68 5.9 57 5.2 46 27.4 n.q.Monocrotophos 86 4.3 78 9.9 86 10.8 10Monolinuron 82 2.9 79 5.9 70 7.3 10Myclobutanil 66 10.5 61 15.2 55 33.5 n.q.Nitenpyran 82 6.1 81 7.5 86 14.7 10Nuarimol 74 5.6 70 4.8 63 15.3 50Oxadixyl 78 18.6 70 16.4 70 8.5 10Oxamyl 79 7.4 70 5.6 76 12.9 10Oxamyl-oxime 76 8.1 70 6.1 82 8.8 10Oxycarboxin 82 4.3 82 5.7 81 5.7 10Paclobutrazole 72 10.1 63 21.6 n.d. 100Penconazole 75 2.2 69 12.3 56 29.2 100Pencycuron 8 9.6 8 13.6 n.d. n.q.Phenmedipham 67 3.7 58 9.8 50 21.3 n.q.Phosphamidon 88 4.8 83 6.0 84 5.2 10Picoxystrobin 71 4.2 66 9.9 45 17.8 100Piperonyl butoxide 75 5.3 71 13.0 51 22.7 50Pirimicarb 86 5.0 75 4.1 69 5.7 50Pirimicarb, desmethyl 84 5.1 78 4.0 78 7.3 10Prochloraz 72 4.8 64 8.6 48 21.3 100Profenofos 73 4.5 72 11.1 51 24.9 50Propiconazole 75 3.8 70 8.4 n.d. 50Propoxur 88 3.7 84 4.3 83 3.2 10Propyzamide 72 5.4 68 7.3 51 17.5 100Pymetrozine 65 6.3 63 4.3 63 12.9 n.q.Pyraclostrobin 69 4.9 64 12.8 49 17.0 n.q.Pyridaben 73 7.6 66 19.9 46 20.1 100Pyridaphenthion 75 5.9 71 6.0 52 24.8 50Pyrifenox 75 4.3 77 6.2 163 30.9 50Pyrimethanil 77 3.8 73 5.4 60 10.7 50Pyriproxyfen 68 6.7 67 14.6 50 15.4 n.q.Spinosad A/D 64 11.9 75 10.9 n.d. n.q.Spirodiclofen 73 6.6 72 15.4 55 14.9 50Spiroxamine 56 12.3 62 7.7 n.d. n.q.Tebuconazole 71 5.4 65 6.4 53 28.9 100Tebufenozide 58 13.2 49 13.8 39 16.9 n.q.Tebufenpyrad 71 5.1 66 11.5 48 27.8 100Tetraconazole 65 10.0 82 5.9 169 8.4 n.q.Thiabendazole 86 5.3 82 2.7 75 8.0 10Thiacloprid 81 6.5 79 6.5 76 5.6 10Thiamethoxam 81 6.3 83 6.2 81 5.5 10Thiofanox sulfoxide 82 5.0 75 6.6 74 24.8 50Thiofanox sulfone 94 5.8 63 14.5 n.d. 100Thiometon sulfoxide 85 5.0 80 2.6 81 5.2 10Tiometom-sulfona 82 4.9 79 4.0 69 11.8 10Tolclophos methyl 64 6.5 72 15.4 58 18.0 n.q.Tolylfluanid 85 4.1 86 5.0 129 20.1 50Triadimefon 75 4.5 67 7.9 56 16.5 100

4546 I.R. Pizzutti et al. / J. Chromatogr. A 1216 (2009) 4539–4552

Table 3 (Continued )

Pesticide Spike levels (�g kg−1) LOQm (�g kg−1)

100 50 10

Recovery (%) RSD (%) Recovery (%) RSD (%) Recovery (%) RSD (%)

Triadimenol 66 15.7 55 22.2 n.d. n.q.Tricyclazole 85 6.5 81 5.6 78 3.0 10Trifloxystrobin 68 6.0 66 11.5 52 16.6 n.q.Triflumizole 74 4.5 64 7.9 42 26.5 50Triforine 60 6.8 56 8.6 n.d. n.q.Triticonazole 71 10,8 63 11.1 58 22.7 100Vamidothion sulfoxide 73 7.4 73 2.6 68 10.8 50VZ

n ike le

3

oepeeiri

rstotoatsEefetttat

TA(

P

2CDFFFFFFHLNTT

n

amidothion sulfone 80 6.6 78oxamide 74 12.3 64

.d.: not detected; n.q.: not qualifying for the quantitation criteria for at least one sp

.2. Matrix effect of soya extracts

Matrix effect is the observed effect of an increase (enhancement)r decrease in detector response (a positive or negative matrixffect, respectively) of a pesticide present in a matrix extract com-ared with the same pesticide present in just solvent. The matrixffect in LC–MS is usually caused by extracted matrix components,luting at the same retention time as the pesticide and therebynfluencing the ionization process occurring in the ion source. Thisesults in a decreased (“ion suppression”) or increased number ofons formed.

Based on practical experience in routine analysis of pesticideesidues in food, a matrix effect ≥20% can generally be considered asignificant and has to be taken into account when reporting quanti-ative results accompanied with a typical measurement uncertaintyf 50% [28]. Matrix effects are always variable with time and condi-ion of instrument used and it should thus be realized that the databtained during validation, are just indicative and should thereforelways be continuously monitored for the actual figures during rou-ine application of a method. The matrix effect comparison betweenoya extract 1 and soya extract 2, for the pesticides analysed in theSI positive ionization mode, is shown in the Table 2. The matrixffects are shown for two concentrations (5 and 2.5 ng mL−1), onlyor those selected pesticides which showed a significant matrixffect for at least one or both of the 2 extracts. It can be concluded

hat the matrix effect for most of the 155 pesticides studied inhe positive ESI mode is negligible or not significant. Even thoughhe extracts have not been cleaned up, the LC–MS interface canpparently handle these relatively dirty extracts. Less than 10% ofhe total number of pesticides shows a significant negative matrix

able 4verage recovery and RSD (in %, n = 6) and method LOQm, obtained by acetonitrile extracESI-negative mode).

esticide Spike levels (�g kg−1)

100 50

Recovery (%) RSD (%) Recovery (%)

,4-D n.d. n.d.hlorfluazuron 92 11.6 87iflubenzuron 124 25.6 82ipronil 80 11.4 86luazinan 79 8.1 91lucycloxuron 164 36.9 n.d.ludioxanil 93 12.6 88lufenoxuron 103 11.7 104lusulfamide n.d. n.d.exaflumuron 90 2.9 91ufenuron 82 3.9 88ovaluron 92 4.9 88eflubenzuron 83 6.5 94riflumuron 122 27.4 81

.d.: not detected; n.q.: not qualifying for the quantitation criteria for at least one spike le

6.2 79 6.3 109.9 50 24.1 100

vel.

effect for both extraction methods. However, it is striking to see thatfor some specific pesticides, there is a difference for which extrac-tion method the significant matrix effect is observed. For the mostpolar pesticides, that is those eluting first in the LC–MS (typicallyretention time range 4–7 min), the negative matrix effect is higherfor acetonitrile than for acetone extracts: e.g. methamidophos,acephate, butocarboxim sulfoxide, omethoate, aldicarb sulfoxideand vamidothion sulfoxide. This could be explained by the moreefficient extraction of polar matrix interferences from the soyamatrix by acetonitrile compared to acetone, causing more ion sup-pression. For fenoxycarb, fenpyroximate and spinosad, the matrixeffects are rather similar for both extraction methods. For somemore apolar (late eluting) pesticides, the negative matrix effectappears to be higher for acetone than for acetonitrile extracts: e.g.azamethiphos, boscalid, paclobutrazole, flusilazole, tebufenozideand mepanipyrim. This indicates that acetone is probablyextracting some non-polar matrix components more efficientlythan acetonitrile, causing more ion suppression for non-polarpesticides.

All pesticides tested and detected in the negative ESI negativemode showed no significant matrix effect for both the acetone andacetonitrile extraction methods.

3.3. Accuracy, precision, method LOQ and selectivity

The method performance evaluation was done according to thevalidation criteria for quantitative methods for pesticide analysis infood and feed, laid down in the SANCO document 2007/3131 [28].Typically a recovery within the range of 70–120% and a repeata-bility RSD ≤ 20% is required. Some recoveries below 70% are still

tion of soya samples, spiked at 10, 50 and 100 �g kg−1, and analysed by LC–MS/MS

LOQm (�g kg−1)

10

RSD (%) Recovery (%) RSD (%)

n.d. n.d.12.9 107 20.1 509.5 122 20.6 n.q.10.2 95 16.7 107.5 101 7.0 10

n.d. n.q.12.8 102 5.6 1012.1 n.d. 50

n.d. n.d.7.6 109 20.6 509.5 91 10.4 107.1 106 5.0 105.6 97 6.8 104.4 98 9.0 n.q.

vel.

I.R. Pizzutti et al. / J. Chromatogr. A 1216 (2009) 4539–4552 4547

F modea

apCr

t1stmtfnptj

hhti1

ig. 4. Total ion chromatograms obtained by LC–MS/MS ESI positive and negativettenuated), by acetonitrile extraction method.

cceptable for a multi-residue method including a large number ofesticides, provided that a good precision (RSD ≤ 20%) is obtained.orrection for recovery during routine application would then beecommended.

The summarized results of the recoveries (%) and RSD (%) ofhe pesticides spiked to the blank soya samples at 10, 50 and00 �g kg−1 and extracted by the acetonitrile-based method, arehown in Tables 3 and 4 for the LC–MS/MS analysis in the posi-ive and negative ionization mode, respectively. The real (validated)

ethod LOQm, defined as the lowest validated spike level meetinghe requirements mentioned above, are included in Tables 3 and 4or each individual pesticide. When both of the two criteria wereot met for all the spike levels tested, it was concluded that theesticide could not be quantified reliably. When at least one of thewo criteria was not met for all the spike levels, the pesticide canust be determined semi-quantitatively or qualitatively.

In total, 70 of the 155 pesticides analysed in the positive mode

ad a LOQm of 10 �g kg−1. Those pesticides with higher LOQ values,ad either a recovery <70% or a RSD > 20%, or both, at one or two ofhe spike levels, but were always detectable. All pesticides analysedn the positive mode could at least be detected at levels of 50 and00 �g kg−1.

s of (A) blank soya extract; soya spiked at (B) 10 �g kg−1 and (C) 100 �g kg−1 (10×

The number and percentage of pesticides within each recov-ery range or not detected (n.d.) at the 10, 50 and 100 �g kg−1

spike levels, are shown in Fig. 3. Most of the pesticides anal-ysed by LC-ESI–MS/MS positive mode (Table 3) showed goodrecoveries. Significantly low recoveries were observed for ethiri-mol, desmedipham, dichlofluanid, trichlorphon, pencycuron andtebufenozide. This may have been likely caused rather by degra-dation of the pesticide in the acetonitrile solvent.

From the 14 pesticides analysed by LC-ESI–MS/MS negativemode (Table 4), two pesticides, 2,4-D and flusulfamide, could not beextracted and were not detected at all. For comparison, it is interest-ing to note, that with the acetone extraction, flusulfamide could notbe extracted either, but on the other hand, 2,4-D was at least partlyextractable with acetone. The different behavior is probably causedby an expected different pH (that is, a higher pH with acetonitrile)during the extraction step. Flucycloxuron is only detectable at thehighest spike level, but very irreproducible. For the other pesticides,

recoveries and RSD values were mostly acceptable.

Selectivity of the method is best illustrated with figures of chro-matograms. The total ion chromatograms for 155 pesticides in theESI positive mode and 14 pesticides in the ESI negative mode areshown in Fig. 4 for acetonitrile extracts of blank soybean and spiked

4548 I.R. Pizzutti et al. / J. Chromatogr. A 1216 (2009) 4539–4552

F zole (s f soya

abep1aw

ts

ig. 5. Multiple reaction monitoring chromatograms for azoxystrobin and tebuconaoya extract; (D) standard solution 5 ng mL−1 in blank soya extract and (E) extract o

t 10 and 100 �g kg−1. The good sensitivity of this method cane typically seen from the high responses of the pesticides in theight overlapping scan functions at the 10 �g kg−1 spike level, com-ared with the 10× attenuated total ion chromatograms at the00 �g kg−1 spike level. The selectivity of the detection method is

lso clearly visible from the chromatograms of the blank extracts,hich show hardly any interfering matrix peaks.

Even more illustrative for the selectivity are the multiple reac-ion monitoring chromatograms for the individual pesticides, ashown in Fig. 5, for example, for the two transitions of azoxystrobin,

LC–MS/MS ESI positive mode), (A) 1st transition and (B) 2nd transition for (C) blankspiked at 10 �g kg−1, by acetonitrile extraction method.

404 → 372 and 404 → 344, measured in the same run. Even at the10 �g kg−1 spike level, for both the first (quantitation) and the sec-ond (confirmation) transition, the very good signal-to-noise (S/N)ratio is clearly visible.

Another example is also given in Fig. 5 for the two MRM

transitions of tebuconazole (308 → 70 and 435 → 125), also show-ing a good S/N ratio for the first MRM transition. However, inthis case, due to a very unfavorable ion ratio, the peak fromthe second (confirmation) MRM chromatogram is below thedetection limit and therefore the 10 �g kg−1 level can be con-

I.R. Pizzutti et al. / J. Chromatogr. A 1216 (2009) 4539–4552 4549

F LC–Me spike

sot[

tlia1a

ig. 6. Multiple reaction monitoring chromatograms for lufenuron and fludioxonil (xtract; (D) standard solution 5 ng mL−1 in blank soya extract and (E) extract of soya

idered as the validated LOQ, but not as a (confirmed) limitf identification (LOI), because the ion ratio of the two MRMransitions cannot be confirmed within the acceptable range28,31].

The LC-ESI–MS/MS negative mode also shows similar selec-ivity, as can be seen in Fig. 6 for the MRM chromatograms of

ufenuron, an insecticide belonging to the class of benzoylureansecticides, which can be detected more sensitively in the neg-tive mode than in the positive one. Lufenuron has a LOQm of0 �g kg−1 based on the 1st MRM transition MRM (509 → 339)nd also the LOI is at the same level, as can be concluded from

S/MS ESI-negative mode), (A) 1st transition and (B) 2nd transition for (C) blank soyad at 10 �g kg−1, using the acetonitrile extraction method.

the S/N ratio of the 2nd MRM transition (509 → 326) at thisconcentration.

In the example of fludioxonil (Fig. 6), it is illustrated that, despitethe good selectivity of the MS/MS detection method, the LC separa-tion is also very important. The 1st MRM transition (247 → 180) isdistinctly selective, but the 2nd MRM transition (247 → 126) gives

not only a significant response at the retention time of fludioxonil,but also at a retention time around 2.9 min. Due to the good resolu-tion, the identification can be confirmed via the ion ratio still beingwithin the acceptable range relative to that of the matrix-matchedstandard solution.

4550 I.R. Pizzutti et al. / J. Chromatogr. A 1216 (2009) 4539–4552

F angea (A) an

3m

cscowp(otFeL

1thdbl

ig. 7. Comparison of number and percentage of pesticides within each recovery rcetone and acetonitrile extraction methods and LC–MS/MS analysis in the positive

.4. Comparison of the acetonitrile and acetone extractionethods performances

In this study, an acetonitrile-based extraction method, whichomprises in fact the recently AOAC International collaborativelytudied and validated, buffered QuEChERS (acronym for quick, easy,heap, efficient, rugged and safe) method [27], however, with-ut applying dispersive solid-phase extraction (SPE) as cleanup,as compared with an earlier developed, very fast and sim-le one-step acetone extraction/dichloromethane-light petroleumb.p. 40–60 ◦C) partitioning method [1,12], without any bufferingr cleanup. The performance characteristics for the majority ofhe 169 pesticides studied were acceptable for both methods. Inigs. 3 and 7, the comparison between the acetonitrile and acetonextraction methods is illustrated via bar graphs, for the obtainedOQm and recovery data, respectively

The number of pesticides with a LOQm at the target level of0 �g kg−1 and acceptable RSDs is slightly higher for the acetoni-

rile method than the acetone method. It should be kept in mind,owever, that this is not caused by the sensitivity of the LC–MS/MSetection method, which is essentially the same for both methods,ut by the differences in recoveries and RSDs at the lowest spike

evel.

or not detected (n.d.) at 10, 50 and 100 �g kg−1 spike levels, when performing thed negative (B) ionization mode. Number of pesticides is depicted next to the bars.

On an average, the recoveries were 10–15% lower and/or theRSDs were somewhat higher at the lowest levels with the acetonemethod, causing that in the latter method less pesticides were ful-filling the strict requirements of both a recovery range 70–120% andRSD ≤ 20%. For both methods, detection at 50 and 100 �g kg−1 wasalways achieved.

To prove whether the somewhat lower recoveries of the acetonemethod might have been caused by possible pesticide losses duringthe evaporation step of the acetone method, a separate experimentwas performed for the determination of the recoveries from spikedsoya extracts (corresponding with 100 �g kg−1). It can be concludedfrom the data, that the evaporation step does not lead to any sig-nificant losses, and therefore, the extraction step is responsible forthe deviation from the 100% recovery.

The speed and user friendlyness of the acetone extraction in realpractice are slightly better than for the acetonitrile method, despitethe extra evaporation step, which can be easily performed in batchand offers the additional advantage of reconstitution in the optimal

solvent methanol for LC analysis of the very early eluting pesticides,without any further dilution of the original extract. On the otherhand, the acetonitrile extract, after dilution with methanol, can bedirectly injected, thus allowing faster sample extract introductionin the LC–MS/MS instrument.

I.R. Pizzutti et al. / J. Chromatogr. A 1216 (2009) 4539–4552 4551

Table 5Surveys of pesticide residues in soya products using acetone (2007) and acetonitrile (2008) extraction and LC–MS/MS analysis.

2007 2008

Soya Product Pesticide Residue level (�g kg−1) Soya Product Pesticide Residue level (�g kg−1)

Soya beans Piperonyl butoxide 3 Soya beans (organic) Carbofuran <2Soya beans (11×) n.d.* Piperonyl butoxide <2Soya pellets n.d. Soya beans (organic) Piperonyl butoxide <2Soya pellets Piperonyl butoxide 6 Soya beans (organic) Tetraconazole <2Soya proteins Chloorpyrifos <2 Piperonyl butoxide <2Soya proteins Tebuconazole <2 Soya beans (organic) Clofentezine <2Soya proteins Chloorpyrifos 2 Piperonyl butoxide <2

Cyproconazole 4 Soya beans (organic) Piperonyl butoxide <2Epoxiconazole 7 Carbofuran <2Flutriafol 2 Soya beans (5×) n.d.Myclobutanil 2 Soybean meat (frozen) n.d.Tebuconazole <2 Soya (minced) n.d.Tetraconazole 2 Soya beans for tofu n.d.

Soya proteins Epoxiconazole 7 Soya pellets n.d.Soya proteins Chloorpyrifos <2 Soya chunks Pyrimethanil 2

Flutriafol <2 Methoxyfenozide <2Piperonyl butoxide 2 Carbendazin 8Propoxur 8

Soya proteins Piperonyl butoxide 4Soya proteins Cyproconazole <2

Tebuconazole 2SS

naemstcmutm2nw

oaiaTtocpmpdcaa

4

nLbS

oya proteins (4×) n.d.oya sauce Thiabendazole 4

* n.d.: no residues detected.

Another typical difference is the use of the Polytron homoge-izer for acetone extraction in the open 250-mL PTFE tubes, whichllows the extraction of 5 g soya sample amounts with 100 mLxtraction solvent. Using the typical 50 mL tubes of the QuEChERSethod, with manual shaking for the acetonitrile extraction, the

oya sample amount has to be limited to not more than 2 g and dueo the voluminous matrix, not more than 20 mL extraction solventan be used. The extraction ratios for the acetone and acetonitrileethods were thus 20:1 and 10:1, respectively. The lower solvent

se and exclusion of chlorinated solvents is a clear environmen-al advantage of the acetonitrile method. Upscaling the acetonitrile

ethod to a 5-g sample and a 50-mL extraction volume in the open50-mL PTFE tubes is an alternative, but the advantage the origi-al low-solvent usage and safety aspects of closed extraction tubesould be lost then.

We have successfully applied both methods in the routine lab-ratory in surveys during 2007 and 2008, using the acetone andcetonitrile extraction method, respectively. The residues detectedn the soya samples are shown in Table 5. It is striking that almostll residue levels detected are within the range of 1–10 �g kg−1.his could be explained by the fact that samples were mostlyaken from big shipments, representing usually large, mixed lotsr consignments. Besides, a relatively high percentage of organi-ally grown samples were taken for the 2008 survey. For a futureroject, we are planning to critically compare both extractionethods side-by-side for a number of representative, positive sam-

les with a sufficiently high residue level, in order to be able toraw conclusions about significant differences in extraction effi-iencies of incurred residues. Unfortunately, from the samplesnalysed so far, only a limited number would qualify for suchn aim.

. Conclusions

Two fast and efficient extraction methods for the determi-ation of 169 pesticides in the difficult soya grain matrix byC–MS/MS analysis were optimized and validated. An acetonitrile-ased extraction method (buffered QuEChERS without dispersivePE), was compared with an earlier developed one-step acetone

extraction/dichloromethane-light petroleum partitioning method,without any buffering or cleanup. The method performancewas evaluated against the criteria for method validation ofthe most recent EU guidelines, SANCO document 3131/2007[28].

The performance characteristics for the majority of the 169pesticides studied were acceptable for both methods. The num-ber of pesticides with a LOQm at the target level of 10 �g kg−1

and acceptable RSDs is slightly higher for the acetonitrile methodthan for the acetone method. However, the speed and userfriendlyness of the acetone extraction in real practice is slightlybetter than for the acetonitrile method. Both methods havebeen proved to be successful as a real quantitative, multi-residue method for pesticide residues analysis in soya grainsamples and some derived products during application in routinesurveys.

Acknowledgements

We gratefully acknowledge VWA – Food and Consumer ProductSafety Authority, The Netherlands and the European Commission(Alfa II Programme B – Project EUROLANTRAP No. AML/B7-311/97/0666/II0461-FA-FCD-FI) for funding this project.

References

[1] I.R. Pizzutti, Ph.D. Thesis, Validation of extraction methods and developmentof a GPC clean up method for multi-residue analysis of pesticides in soybeansby GC–MS, GC–MS/MS and LC–MS/MS, Federal University of Santa Maria, SantaMaria, Brazil, 2006.

[2] E.M.A. Smaling, R. Roscoe, J.P. Lesschen, A.F. Bouwman, E. Comunello, Agr.Ecosyst. Environ. 128 (2008) 185.

[3] Regulation (EC) No. 396/2005 of the European Parliament and of the Council of23 February 2005 on maximum residue levels of pesticides in or on food andfeed of plant and animal origin and amending Council Directive 91/414/EEC,Off. J. Eur. Union L70 (March) ( 2005) 1.

[4] Commission Regulation (EC) No. 149/2008 of 29 January 2008 amending Reg-ulation (EC) No. 396/2005 of the European Parliament and of the Council by

establishing Annexes II, III and IV setting maximum residue levels for productscovered by Annex I thereto, Off. J. Eur. Union L58 (March) (2008) 1.

[5] T.D. Nguyen, J.E. Yu, D.M. Lee, G. Lee, Food Chem. 110 (2008) 207.[6] E.G. Amvrazi, N.G. Tsiropoulos, J. Chromatogr. A 1216 (2008) 2789.[7] M. Mezcua, C. Ferrer, J.F. García-Reyes, M.J. Martínez-Bueno, M. Sigrist, A.R.

Fernández-Alba, Food Chem. 112 (2009) 221.

4 atogr.

[

[

[

[[[

[[[

[

[

[

[

[[

[[[[

SANCO/3131/2007, Brussels, 31 October 2007.[29] J. Zrostlíková, J. Hajslová, M. Godula, K. Mastovská, J. Chromatogr. A 937 (2001)

552 I.R. Pizzutti et al. / J. Chrom

[8] S. Walorczyk, J. Chromatogr. A 1208 (2008) 202.[9] A. Hercegová, M. Dömötörová, E. Matisová, J. Chromatogr. A 1153 (2007) 54.10] C. Lesueur, P. Knittl, M. Gartner, A. Mentler, M. Fuerhacker, Food Control 19

(2008) 906.11] A.R. Fernández-Alba (Ed.), Comprehensive Analytical Chemistry, vol. 43, Else-

vier, Amsterdam, 2005.12] I.R. Pizzutti, A. de Kok, R. Zanella, M.B. Adaime, M. Hiemstra, C. Wickert, O.D.

Prestes, J. Chromatogr. A 1142 (2007) 123.13] M. Hiemstra, A. de Kok, J. Chromatogr. A 1154 (2007) 3.14] L. Alder, K. Greulich, G. Kempe, B. Vieth, Mass Spectrom. Rev. 25 (2006) 838.15] A.G. Frenich, P. Plaza-Bolanos, J.L. Martínez-Vidal, J. Chromatogr. A 1203 (2008)

229.16] M.A. Luke, J.E. Froberg, H.T. Masumoto, J. Assoc. Off. Anal. Chem. 56 (1975) 1020.17] P.A. Mills, J.H. Onley, R.A. Guither, J. Assoc. Off. Anal. Chem. 46 (1963) 186.

18] A.H. Roos, A.J. van Munsteren, F.M. Nab, Th L.G.M. Tuinstra, Anal. Chim. Acta 196

(1987) 95.19] S.J. Lehotay, A. de Kok, M. Hiemstra, P. van Bodegraven, J. AOAC Int. 88 (2005)

595.20] I.R. Pizzutti, M. De Kroon, O.D. Prestes, P. Rensen, A. De Kok, 7th European

Pesticide Residue Workshop Book of Abstracts, Berlin, 2008, p. 114.

[[

A 1216 (2009) 4539–4552

21] I.R. Pizzutti, P. Rensen, O.D. Prestes, M. De Kroon, A. De Kok, 7th EuropeanPesticide Residue Workshop Book of Abstracts, Berlin, 2008, p. 115.

22] I.R. Pizzutti, A. De Kok, J. Scholten, P. Rensen, 122nd Annual Meeting of AOACBook of Abstracts, Dallas, TX, 2008, p. 157.

23] J.M. Lee, J.W. Park, G.C. Jang, K.J. Hwang, J. Chromatogr. A 1187 (2008) 25.24] C. Díez, W.A. Traag, P. Zomer, P. Marinero, J. Atienza, J. Chromatogr. A 1131 (2006)

11.25] J. Klein, L. Alder, J. AOAC Int. 86 (2003) 1015.26] H.G.J. Mol, A. Rooseboom, R. van Dam, Anal. Bioanal. Chem. 389 (2007) 1715.27] S.J. Lehotay, J. AOAC Int. 90 (2007) 485.28] European Commission, DG-SANCO, Method validation and quality control

procedures for pesticide residues analysis in food and feed, Document no.

73.30] A. Kruve, A. Künnapas, K. Herodes, I. Leito, J. Chromatogr. A 1187 (2008) 58.31] S.J. Lehotay, K. Mastovska, A. Amirav, A.B. Fialkov, T. Alon, P.A. Martos, A. de Kok,

A.R. Fernandez-Alba, Trends Anal. Chem. 27 (2009) 1070.