Evaluation of the automated micro-solid phase extraction clean-up system for the analysis of pesticide residues in cereals by gas chromatography-Orbitrap mass spectrometry

Food analysis is a tremendously broad field that is constantly evolving. New methods have emerged to increase productivity, such as modern miniaturized and robotic analytical techniques. In this paper, a microsolid-phase extraction system (μ-SPE) for clean-up was combined with a robotic autosampler to yield ready-to-analyze extracts.

Trang 1

Contents lists available at ScienceDirect

journal homepage: www.elsevier.com/locate/chroma

Elena Hakme∗, Mette Erecius Poulsen

National Food Institute, Technical University of Denmark, Søborg, Denmark

a r t i c l e i n f o

Article history:

Received 19 April 2021

Revised 23 June 2021

Accepted 28 June 2021

Available online 3 July 2021

Keywords:

μ-SPE clean-up

Robotic system

Cereals

Pesticide residues

Evaluation study

a b s t r a c t

Food analysis is a tremendously broad field that is constantly evolving New methods have emerged to increase productivity, such as modern miniaturized and robotic analytical techniques In this paper, a micro- solid-phase extraction system (μ-SPE) for clean-up was combined with a robotic autosampler to yield ready-to-analyze extracts The system was evaluated for its applicability in routine laboratories The new, automated, high-throughput μ-SPE clean-up method was applied to acetonitrile extracts and was developed for the analysis of pesticide residues in cereals by gas chromatography-Orbitrap mass spectrometry (GC-Orbitrap-MS) The μ-SPE clean-up efficiency was demonstrated in the removal of matrix-interfering components and in the recovery of pesticides The sorbent bed mixture consisted of magnesium sulfate, primary-secondary amine, C 18, and CarbonX, and effectively retained matrix components without loss of target analytes Analysis of five types of cereals (barley, oat, rice, rye, and wheat) by GC-Orbitrap-MS showed that the method removed more than 70% of matrix components The clean-up method was validated for 170 pesticides in rye, 159 pesticides in wheat, 142 pesticides in barley, 130 pesticides in oat, and 127 pesticides in rice Spike recovery values were 70–120% for all pesticides and the repeatability, calculated as the relative standard deviation, was less than 20% The limits of quantitation achieved were 0.005 mg kg −1for almost all analytes, ensuring compliance with the maximum residue limits

1 Introduction

Pesticide residues, among the large variety of contaminants, are

continuously monitored and controlled to ensure legislative com-

pliance Pesticide residue analysis is crucial in estimating max-

imum residue limits, reviewing toxicological data, and ensuring

food safety Similar to other food analysis applications, the sample

preparation step is often the key parameter in method develop-

ment, particularly in the isolation and detection of contaminants

Besides the accuracy and validity of the method, the time required

to complete the analytical process and the cost of the consum-

ables (e.g., solvents and sorbents) used in the analysis are partic-

ularly considered It is estimated that 60–80% of the work activity

and operational costs in analytical laboratories are spent prepar-

ing samples for analysis It is also estimated that this step is re-

sponsible of 50% of the error in the ﬁnal reported data [20] There-

∗ Corresponding author

E-mail address: elehak@food.dtu.dk (E Hakme)

fore, faster, automated, cost-effective, and greener alternative sample preparation techniques with good accuracy are needed According to the literature, several sample preparation techniques, such as liquid-liquid extraction (LLE) [5], gel permeation chromatography (GPC) [14], solid phase microextraction (SPME) [24], and matrix solid phase dispersion [12], have been explored, and some have been successfully applied to the multiresidue analysis of pesticides in food Despite the effectiveness of these methods, the methods require large amounts of solvents, are time con- suming and tedious, and require intense labor The sample preparation approach known as QuEChERS (quick, easy, cheap, effective, rugged, and safe), developed by Anastassiades et al in 2003 [2], met the changing needs of multiresidue analysis and has been successfully applied to the recovery of pesticide residues in food In

2007, the QuEChERS-d-SPE was published by the Association of Of- ﬁcial Analytical Chemists (AOAC, [3]) and by the European Commit- tee for Standardization [6] In its basic scheme, the method consists of an extraction with acetonitrile, partitioning with salts to promote water separation from the organic solvent, and clean-up

of the ﬁnal acetonitrile extract with dispersive solid phase extrac-

https://doi.org/10.1016/j.chroma.2021.462384

Trang 2

tion (d-SPE) sorbents to remove organic acids, sugars, and polar

pigments

SPE sorbents, used in dispersive form or packed in a cartridge,

are demonstrated suitable to a wide variety of food and agricul-

tural products when appropriate adsorbing/sorbent materials are

selected [ 4, 10, 13] The availability of pre-packaged dispersive kits

has enabled fast sample preparation and has added advantages

of the dispersive SPE (d-SPE) in terms of time and operational

conditions However, some studies have shown that the clean-up

eﬃciency with cartridge-SPE is better than with d-SPE, because

there is better surface contact between the sorbent and the sam-

ple [1] Moreover, cartridge-SPE permits either solvent reduction

or solvent exchange prior to clean-up, as well as the possibility

of using different solvent mixtures that effectively elute the tar-

get analytes, which preserves the accuracy of the method [ 22, 23]

The main disadvantages of the cartridge-SPE are the extended op-

eration time and procedure steps, the susceptibility to loss or

degradation of the target analytes, and other potential sources

of repeatability errors arising from the use of an SPE vacuum

manifold

In recent years, much effort has been devoted to eliminate

these drawbacks This has led to the development of robotic auto-

mated techniques Currently, an automated micro-solid phase ex-

traction (μ-SPE) clean-up method for acetonitrile extracts is avail-

able as an alternative to cartridge-SPE μ-SPE is a simple scale-

down or miniaturization of the cartridge-SPE procedure The use of

automated μ-SPE clean-up was originally reported by Morris et al

[17] for the analysis of pesticide residues in avocado and citrus

Automated mini-SPE clean-up was also evaluated, and was found

to be eﬃcient for the analysis of pesticide residues in spices, in-

cluding chili powder, turmeric, black pepper, cumin, coriander, and

cardamom [11] Lehotay et al demonstrated the clean-up eﬃciency

of mini-SPE on avocado, salmon, pork loin, and kale [15] Ederina

et al demonstrated the eﬃciency of robotic mini-SPE clean-up for

the analysis of pesticides and their metabolites in catﬁsh muscle

[18] Pandey et al also demonstrated the high-quality results of

this automated system for diverse types of analytes and food ma-

tricesf [19] The automation of the μ-SPE method for the clean-

up and pre-concentration of polyﬂoroalkyl substances from surface

water has also been demonstrated [16]

Laboratory automation is expected to increase in food testing

laboratories because of their time and space eﬃciency Thus, it is

of great importance that laboratories adopt robotic automated sys-

tems that guarantee high sample throughput without much labor,

and that, most importantly, are reliable The objective of this study

was to evaluate the performance of the automated μ-SPE tech-

nique in the analysis of 172 pesticide residues in cereals, and to

determine if the technique could be used in national and oﬃcial

routine analysis laboratories The μ-SPE clean-up method used in

this study consisted of a removal/trapping strategy, where the ma-

trix components were retained and the analytes of interest were

eluted Since the procedure is intended to be scaled-up for ap-

plication to all raw cereal products, ﬁve cereal matrices were se-

lected for the validation study Most of the pesticides included in

this study are included in the EU multi-annual control program [ 7]

Two evaluation studies were designed to demonstrate the cleanli-

ness of the extract and the clean-up eﬃciency In the ﬁrst, blank

extracts subjected to automated μ-SPE clean-up were compared

to extracts subjected to d-SPE clean-up In the second study, ace-

tonitrile extracts were spiked with pesticides prior to clean-up to

demonstrate the recovery eﬃciency of the method Finally, the

method was validated according to the guidance document on an-

alytical quality control and method validation procedures for pes-

ticide residues and analysis in food and feed [9] in terms of lin-

earity, recovery, and repeatability The matrix effect of each cereal

was evaluated for quantitation purposes

2 Material and methods

2.1 Chemicals

Pesticide standards (purity >96%) were purchased from Sigma- Aldrich and LGC Standards Pesticide standard stock solutions of

1 mg mL −1were prepared in toluene and stored at −18 °C in am- poules under an argon atmosphere A standard solution of 10 μg

mL −1was prepared from these stock solutions Working calibration standard solutions were prepared by diluting 1:1 ( v/v) with acetonitrile to obtain ﬁve concentration levels: 0.2 μg mL −1, 0.0 6 67 μg

mL −1, 0.02 μg mL −1, 0.0067 μg mL −1, and 0.002 μg mL −1 Ace- tonitrile (HPLC Grade 5) was purchased from Rathburn Chemicals μ-SPE cartridges (Cart-uSPE-GC-QUE-0.3 mL) were purchased from CTC-Analytics Supel TM QuE QuEChERS tubes containing 4 g magnesium sulfate (MgSO 4), 1 g sodium chloride (NaCl), 0.5 g sodium citrate sesquihydrate, and 1 g sodium citrate dihydrate were purchased from Thermo Scientiﬁc The clean-up sorbent Supel TM QuE (EN) tubes were purchased from Supelco

2.2 Extraction method

The samples were extracted using the citrate-buffered QuECh- ERS (EN 15662) (CEN 2008) method without clean-up In brief, 5 g

of each sample was prepared The procedural standard dichlorvos- d6 was added to all samples before extraction Then, 10 mL cold water was added, followed by 10 mL acetonitrile To aid the extraction, a ceramic homogenizer was used The tubes were shaken for

1 min by hand Next, 4.0 g of MgSO 4, 1.0 g NaCl, 1.0 g sodium citrate dihydrate, and 0.5 g sodium citrate sesquihydrate were added After 1 min of shaking by hand and centrifugation for 10 min at

4500 rpm, 8 mL of the supernatant was transferred to a clean tube and stored at –80 °C for at least 1 h The extracts were then thawed, and while they were still very cold, they were centrifuged

at 4500 rpm for 5 min at 5 °C Thereafter, 6 mL of the cold supernatant was collected

For the μ-SPE automated clean-up, the extracts were diluted (1:1 v/v) with acetonitrile and placed in 1 mL glass vials on the sample tray of the robotic autosampler A minimum volume of 500

μL is recommended to avoid the aspiration of air bubbles into the

10 0 0 μL μ-SPE syringe, or else the syringe depth in the instrumen- tal method should be modiﬁed accordingly Triphenyl phosphate (15 μl of a 0.1 μg mL −1internal standard solution) which is used

as an internal standard to check the performance of the injection system of the instrument, was added automatically on the robotic autosampler

For the d-SPE clean-up, a dispersive sorbent mixture consisting of 150 mg PSA and 900 mg MgSO 4 was added to the 6 mL extract The tubes were shaken for 30 s, and then centrifuged at

4500 rpm for 5 min at room temperature After centrifuging, 4 mL supernatant was collected and 5% formic acid was added The extracts were diluted (1:1 v/v) in 1 mL glass vials with acetonitrile, and the internal standard (triphenyl phosphate) was added

2.3 Chromatographic separation and high-resolution mass spectrometry

The analyses were performed on an GC-Exactive MS (Thermo Fisher Scientiﬁc), consisting of a Trace 1300 Series GC, a TriPlus RSH Autosampler GC-liquids, and an Exactive GC-Orbitrap-MS The samples were injected in a programmable temperature va- porizer (PTV) through a PTV baﬄe liner (2 × 2.75 × 120 mm) designed for Thermo GCs (Siltek) The injection volume was 1 μL and the injection temperature was set to 70 °C Helium was used

as the carrier gas at a ﬂow rate of 1.2 mL/min for analyte separation on a Thermo Scientiﬁc Trace GOLD TG-5SILMS column (30 m

Trang 3

Fig 1 Schematic of the TriPlus RSH robotic PAL autosampler

length × 0.25 mm i.d × 0.25μm ﬁlm thickness) The GC oven pro-

gram started with an initial temperature of 60 °C, which was held

for 1.5 min, followed by a ramp of 25 °C/min to 90 °C This tem-

perature was held for 1.5 min, followed by a ramp of 25 °C/min up

to 180 °C, then up to 280 °C at 5 °C/min Finally, to clean the col-

umn, the temperature was raised to 300 °C at a rate of 10 °C/min

and held for 12 min

The analyses were performed in electron ionization (EI) positive

mode Eluting peaks were transferred through an auxiliary transfer

line into the EI source The EI source and the transfer line tem-

peratures were set to 280 °C The instrument operated at a reso-

lution of 60k and the automatic gain control (AGC) target was set

to 1 × 106 The MS data were acquired in a scan mode covering a

mass range from 50 to 500 m/z

The instrument was tuned using the Thermo Scientiﬁc Exactive

GC Tune software (v 2.9 SP3 Build 290204) The vacuum inside

the Orbitrap Analyzer was maintained below 1 × 10−9 mbar The

instrument method was developed on Thermo Scientiﬁc XCalibur

software The full scan MS data were processed using a quantita-

tion master method on the Thermo Scientiﬁc TraceFinder 4.1 soft-

ware The studied compounds were transferred into the quantita-

tion method from an in-house compound database The database

included retention time, target ion, and at least 2 conﬁrming ions

for each compound The Genesis algorithm was used for peak in-

tegration The method is shown in the Supplementary Material

2.4 TripPlus RSH autosampler

The μ-SPE system is coupled to the GC-Orbitrap-MS The TriPlus

RSH robotic PAL autosampler comprises three tools: the μ-SPE tool

(LS3) that holds a 10 0 0 μL syringe, the analyte protectant or in-

ternal standard tool that holds a 25 μL syringe (LS2), and the in-

jection tool (LS1) that holds a 10 μL syringe A schematic of the

robotic PAL autosampler is presented in Fig.1

The system also contains a standard wash module, a solvent

station module, and a fast wash module The solvent station mod-

ule was not used because the experiment was done without con-

ditioning of the cartridges and without additional solvent elution,

which also saved solvents and time The standard wash module

tray holds 2 mL glass vials, reserved for internal standards or ana-

lyte protectants, and three 25 mL glass vials, reserved for aliquots

of blank extracts or acetonitrile for automated matrix-matched cal-

ibration curves and automated sample dilution, respectively The

fast wash station for fast syringe washing is connected to the au-

tosampler It uses two solvents: acetonitrile (fast wash station position 1) and a mixture of acetonitrile, methanol, and water (1:1:1

v/v/v) (fast wash station position 2)

The μ-SPE tray holder, also attached to the PAL bus, has three slots The ﬁrst slot is the sample tray where the crude extracts obtained from acetonitrile extraction were placed The second slot

is the eluate tray where empty vials were placed to collect the cleaned-up μ-SPE extracts μ-SPE cartridges were placed in the third slot

2.5 μ-SPE clean-up workﬂow

Table 1 shows the automatic μ-SPE program steps and their duration In the automatic tool change station, the 10 0 0 μL μ- SPE syringe was automatically selected The syringe was robotically moved to the fast wash station module and rinsed with pure acetonitrile (2 rinsing cycles) A 300 μL aliquot of crude extract was loaded into the syringe after 3 ﬁlling strokes The tool with the ﬁlled μ-SPE syringe was moved to the third slot to pick one μ-SPE cartridge and then back to the eluate tray to load the extract into the cartridge Approximately 240 μL cleaned extract was eluted at

a ﬂow rate of 30 μL −1 and collected in the empty vial placed

in the eluate tray Once the clean-up was completed, the syringe was moved back to the fast wash station to be rinsed again with

Table 1

Automatic μ-SPE program steps with a total duration of 13 min Time (mm:ss) Steps

0:30 Required tool selected

Syringe wash: 2 cycles at wash position 1 01:30 Load sample onto μ-SPE

Perform 3 ﬁlling strokes 04:30 Load sample onto μ-SPE cartridge: 300 μL 05:30 Syringe wash: 2 cycles at wash position 1 06:30 Required tool selected

07:30 Syringe wash: 2 cycles at wash position 1

Rinse 08:30 Add 15 μL internal standard

Perform 3 ﬁlling strokes 09:30 Add internal standard: 15 μL 10:30 Required tool selected

Syringe wash: 1 cycle at wash position 2

12:00 Move to sample at position 1 13:00 Perform 3 ﬁlling strokes

Aspirate 1 μL Inject sample

Trang 4

acetonitrile (2 rinsing cycles) The tool holding the 25-μL syringe

was then selected The syringe was moved to the fast wash mod-

ule to be washed with acetonitrile (2 cycles) The syringe was then

moved to the standard wash module After 3 ﬁlling strokes, 15 μL

internal standard was added to the cleaned-up extract Then, the

required injection tool, holding a 10-μL syringe, was selected The

syringe was washed with the acetonitrile, methanol, and water sol-

vent mixture (one rinsing cycle) The syringe was moved to the

eluate tray Three ﬁlling strokes were performed before 1 μL was

aspirated and injected

2.6 Experiment 1: extract cleanliness assessment

Whether cartridge-SPE, d-SPE, or μ-SPE, the critical piece in all

SPE methods is the selection of the sorbent It is necessary to

consider chemical and physical characteristics that allow maximal

interaction between the sorbent and the analytes, which ensures

selectivity of extraction, removal, or preconcentration of analytes

present in analytical matrices In order to check the cleanliness

of the extracts obtained with μ-SPE, blanks of barley, wheat, oat,

rice, and rye were extracted using both the conventional QuEChERS

extraction method with manual dispersive SPE clean-up and the

robotic μ-SPE clean-up system Both extracts were injected onto

the GC-Orbitrap-MS The same amount of matrix was used in both

SPE methods, both in the clean-up step (0.5 g mL −1) and in the

injection step (0.25 g mL −1, following a 1:1 dilution with acetoni-

trile) In these comparison experiments, two factors were consid-

ered: differences in automation and sorbents In the manual d-

SPE clean-up, 25 mg of PSA/mL of extract and 150 mg of MgSO 4

/mL of extract were used In the μ-SPE, 40 mg of PSA/mL of ex-

tract and 66 mg of MgSO 4/mL of extract were used, in addition to

40 mg/mL of C18 and 66 mg/mL of CarbonX

The total ion chromatograms (TICs) of blanks obtained with

the two different clean-up procedures were overlaid using XCal-

ibur software For a closer examination of the clean-up effective-

ness, the deconvolution plugin software (v 1.3), in conjunction with

the TraceFinder software (v 4.1), was used The software automati-

cally deconvoluted coeluted chromatographic peaks into multiple

components by aligning mass spectral peaks, according to their

slightly different retention times The software also automatically

performed a peak search and a library search Combined with the

unknown screening functionality of TraceFinder software, the de-

convolution software was used to do a cross-sample overlay of an-

alytes

2.7 Experiment 2: calibration assessment

In order to evaluate the possible loss of pesticides during clean-

up by μ-SPE, and to assess if a procedural standard calibration was

required, a preliminary evaluation study was carried out before

proceeding with the validation study Matrix-matched calibrations

are routinely used for quantitation of pesticide residues in food

matrices, and they use standards prepared from blank extracts of

the same matrix The blanks used to prepare matrix-matched cali-

brations were extracted in the same manner as the samples After

extraction and clean-up, the blank extracts were diluted with a se-

ries of calibration standards The procedural standard calibration

approach is typically used when dealing with diﬃcult matrices to

compensate for matrix effects and low extraction recovery asso-

ciated with certain pesticide/commodity combinations It consists

of spiking a series of blank test portions with different amounts

of analytes prior to extraction In order to assess the clean-up ef-

ﬁciency, the matrix effect, and the recovery of pesticides on the

robotic μ-SPE system, rather than in the whole acetonitrile extrac-

tion method, blanks of rye were spiked before and after the clean-

up step

For this purpose, a semi-procedural standard calibration was prepared by spiking a series of blank test portions of rye with different amounts of analyte just before the clean-up step It is re- ferred to it as a “semi-procedural calibration” because the spiking was done just prior to clean-up, and not prior to the whole extraction method The extracts (0.5 g mL −1 matrix) obtained from the QuEChERS extraction were diluted 1:1 with a standard mixture of

172 pesticides prepared in acetonitrile at 0.2 μg mL −1, 0.0 6 67 μg

mL −1, 0.02 μg mL −1, 0.0067 μg mL −1, and 0.002 μg mL −1 The ﬁnal concentrations prepared were 100 μg kg −1, 33 μg kg −1, 10 μg kg −1,

3 μg kg −1, and 1 μg kg −1, respectively The vials were placed on the robotic autosampler for automated μ-SPE clean-up The amount

of cleaned-up matrix injected in this experiment was 0.25 g mL −1 The semi-procedural standard calibration was compared to a matrix-matched calibration The set of matrix-matched calibration curves was prepared using the blank extracts (0.5 g mL −1 blank matrix extract obtained from the QuEChERS extraction) The extracts were placed on the robotic autosampler for μ-SPE clean-

up After clean-up, the eluates were diluted 1:1 with a series of standards, giving a series of matrix-matched calibration samples at

100 μg kg −1, 33 μg kg −1, 10 μg kg −1, 3 μg kg −1, and 1 μg kg −1 The amount of matrix injected onto the GC system was 0.25 g mL −1, which enabled the comparison of the two calibrations In this latter calibration, the pesticides were not loaded into the μ-SPE cartridge; therefore, the matrix-matched calibration was considered a reference calibration to evaluate pesticide recovery

The slopes of the two calibration curves were compared The data obtained from this experiment were also processed to calcu- late the pesticide recovery after the robotic μ-SPE clean-up

2.8 Experiment 3: method validation

Method validation was performed according to SANTE/ 12682/2019 guidelines Five sets of semi-procedural calibration curves were prepared using extracts of blank samples of each of the ﬁve matrices (barley, oat, rice, rye, and wheat) as described in the previous section The extracts were cleaned-up using the automatic μ-SPE robotic system The linearity range was determined for the 172 compounds in the 0.0033 0.1 μg

mL −1 range In gas or liquid chromatography systems, the matrix effect is caused by the unwanted interference of compounds during ionization in the MS source or during injection, and it can dramatically influence the analysis for both identification and quantification of an analyte Usually, the matrix effect is calculated

as the percentage difference between the slopes of the matrix- matched calibration curves and the solvent calibration curve In this study, the matrix effect was investigated by comparing the slopes of the semi-procedural calibration curves, obtained with barley, rice, oat, and wheat, to the semi-procedural calibration curves prepared with rye, which was chosen as a representative matrix among the cereals included in the current study The rye matrix provides good protection of analytes and has a moderate matrix effect compared to those of other cereal matrices

Five samples each of barley, oat, rice, rye and wheat matrices were spiked before extraction at each of three concentration levels (5 μg kg −1, 10 μg kg −1, and 50 μg kg −1) to study the extraction effectiveness Therefore, in total, the validation study was performed on 75 spiked samples The trueness and the precision

of the method were evaluated by calculating the recovery and the repeatability, respectively Acceptable mean recovery is usually within the range of 70 – 120% Repeatability refers to the variation

in repeated measurements made on the same subject under identi- cal conditions Therefore, repeatability was evaluated by calculating the relative standard deviation (RSD) based on the recovery results

of the ﬁve spiked samples of each matrix at each spiking concentration level The precision of the method was also investigated by

Trang 5

calculating the reproducibility, which was derived from the range

of recovery results obtained with different matrices of the cereals

at each of the concentration levels

3 Results and discussion

3.1 Performance expectations of the automated sample preparation

system

The time needed for performing manual d-SPE clean-up on a

batch of four samples was approximately 14 min The addition of

clean-up salts to the collected QuEChERS extracts took 2 min Mix-

ing using an automatic agitator took 1 min Centrifugation was per-

formed in 5 min, according to the citrate-buffered QuEChERS (EN

15662) (CEN 2008) method Collecting the ﬁnal extracts took up

to 4 min, and the addition of the internal standard was achieved

in 2 min, resulting in a total clean-up time of 14 min Using the

robotic μ-SPE, the automated clean-up and addition of the inter-

nal standard for a batch of 4 samples took a total of 52 min

(13 min/sample) Surface-level thinking would lead to the conclu-

sion that the automated μ-SPE system is not advantageous in terms

of saving time However, a robotic system that could be operating

24/7 undoubtedly enables higher productivity because more sam-

ples can be processed outside of the normal work schedule or even

overnight Hence, overall, the robotic system results in a signiﬁcant

reduction of labor Moreover, the automated μ-SPE system allows

more consistent preparation by avoiding human laboratory errors

in the ﬁnal clean-up step

Although the system was coupled to the GC-Orbitrap-MS, the

autosampler was not equipped with a cooling system With a non-

thermostatic autosampler, samples had to stand in the sample

tray at room temperature for a long time, since the GC analy-

sis time of each injection was 45 min Therefore, in the case of

non-availability of a thermostatic tray, a stand-alone robotic μ-SPE

clean-up system is recommended Yet, a μ-SPE coupled to a chro-

matographic and spectrometric system would be advantageous, be-

cause the clean-up of an extract can be performed while another

extract is being analysed

3.2 Experiment 1: extract cleanliness assessment

The integrated area of the TIC obtained with a blank of wheat

extract (0.5 g mL −1) cleaned-up with d-SPE was 1.2 × 10 9 The in-

tegrated area of the TIC of the same blank (0.5 g mL −1) after μ-SPE

clean-up was 3.34 × 108 The μ-SPE method resulted in the re-

moval of approximately 70% more matrix interferences than the d-

SPE method Fig.2shows the overlay of the TICs of the two blanks

The TIC corresponding to d-SPE shows the most intense peaks at

9.34 min (corresponding to linoleic acid (C 18H 32O 2)), at 16.7 min

(corresponding to linolenic acid (C 18H 30O 2)), and at 31.47 min (cor-

responding to campesterol (C 28H 48O)) The comparison of these

proﬁles showed that the extract obtained with d-SPE seems to

have had a higher concentration of interfering compounds or ma-

trix components In the μ-SPE sample, these unwanted matrix

components remained bound to the μ-SPE sorbent and had a much

lower concentration in the ﬁnal extracts The two adsorbents (C 18

and CarbonX) embedded in the μ-SPE cartridge also allowed the

adsorption and removal of fatty acids and other matrix interference

compounds Fig.3 shows the cross-sample peak overlay of ligno-

ceric acid methyl ester, a saturated fatty acid with the chemical

formula C 23H 47COOH, in rye blanks after d-SPE and μ-SPE clean-

up methods The most effective removal of this matrix interfer-

ence compound was by μ-SPE Moreover, the number of peaks de-

tected in a blank of rye after deconvolution analysis was 172 and

123 peaks for d-SPE and μ-SPE, respectively The same results were

observed with the four other cereal matrices

3.3 Experiment 2: calibration assessment

The deviation between the slopes of the matrix-matched calibration and the semi-procedural-standard calibration was less than 25% for almost all of the compounds Additionally, pesticide recovery was calculated at each of the ﬁve spiked levels This recovery study is not a full validation study, but rather an estimation of the pesticide recovery (or loss) and an assessment of the robotic clean-up method Recovery results between 70 120% were considered successful Fig.4shows the percentage of compounds recovered after clean-up at the ﬁve different spiked concentration levels Almost all of the compounds were successfully recovered

at a concentration level of 100 μg kg −1 For instance, at each of the concentration levels (100, 33, 10, 3, and 1 μg kg −1), 98%, 94%, 112%, 99%, and 109% of toclophos-methyl were recovered, respectively Poorer results were obtained with ditalimphos, where only 58%, 59%, 57%, 55%, and 50% of the analyte were recovered from the samples with concentration levels of 100, 33, 10, 3, and 1 μg

kg −1, respectively Some compounds, such as spiroxamine, fenhexamid, fenpropidin, deltamethrin, and iprodione were not recovered

at high levels For the ﬁrst two acidic compounds and the cationic potential compound (fenpropidin), recovery values obtained were less than 15%, probably due to interaction with PSA The two latter compounds exhibited a signal enhancement after passing through the μ-SPE, with recovery around 140% On average, 85% of the compounds were successfully recovered at the levels of 33, 10, and

3 μg kg −1 The highest percentage of compounds (27%) was lost in the lowest spiking level (1 μg kg −1) during the clean-up, likely by adsorbing onto the μ-SPE bed sorbents Compound loss could also have occurred in the injector In the injector, matrix components protect the analytes from thermal decomposition and block them from adsorption onto the active sites of the GC system Thus, in

a cleaner extract, compounds are no longer protected from degradation According to SANTE guidelines, recovery values outside the

70 120% range can be accepted if the results are consistent, and a correction factor can be applied However, due to the very low recovery of the compounds as mentioned above, a correction factor for recovery was not used Instead, a more accurate approach was adopted, which consisted of the use of a semi-procedural standard calibration for routine analysis

Therefore, in the light of these results, a semi-procedural standard calibration was used for accurate quantitative method validation, and it is recommended for use in routine analysis to compensate for possible clean-up automation and extraction eﬃciency errors or the retention/loss of compounds, especially at the lowest concentration levels

3.4 Experiment 3: method validation

The method showed a linear response over the studied concentration range of 1–100 μg kg −1 with the four matrices, and it had a coeﬃcient of correlation greater than 0.99 The matrix effect percentage was calculated for each of the 172 pesticides Fig.5 shows the percentage of compounds that exhibited a weak, moderate, and strong matrix effect The matrix effect is caused by co- eluting compounds from the matrix, which generate a signal sup- pression or enhancement A strong matrix effect corresponds to a value above ±50%. A weak matrix effect is less than ±25%. A moderate matrix effect is between ±25% and ±50% An eﬃcient extraction and clean-up method will generate clean extracts and retention of all matrix-interfering components, and thus will have a smaller matrix effect A weak matrix effect was observed for 75%

of the compounds in wheat, in comparison to rye For oat and barley matrices, in comparison to rye, 65% of the compounds showed

a weak matrix effect In rice, 45% of the compounds showed weak matrix effect A moderate matrix effect was observed for 14% of

Trang 6

Fig 2 Total ion chromatograms of a blank of wheat extracted with d-SPE (blue) and μ-SPE clean-up (red)

Fig 3 Cross-sample peak overlay of lignoceric acid methyl ester in 0.5 g mL −1 rye blank extracted with d-SPE (black; peak area: 5667.091), μ-SPE clean-up of 0.5 g mL −1 matrix (red; peak area: 75.399), and μ-SPE clean-up of 0.25 g mL −1 matrix (green; peak area: 420.236)

Fig 4 Percentage of compounds with recoveries within the range 70–120% after

clean-up of acetonitrile extracts spiked at ﬁve concentration levels of spiking (1, 3,

10, 33, 100 μg kg −1 )

the compounds in barley, 8% of the compounds in oat, 25% of the

compounds in rice, and 17% of the compounds in wheat The obvi-

ous explanation for a weak-to-moderate matrix effect is that the

matrix-interfering components were successfully retained by the

μ-SPE sorbents Although matrix effect was not signiﬁcant com-

pared to rye, signal enhancement between 0 and + 20% was ob-

served for almost all pesticides in wheat, but mainly in rice, bar-

ley, and oat For all compounds showing a weak-to-moderate ma-

trix effect, the matrix-matched calibration prepared with the rye

blank was used for the qualitative and quantitative analyses of

Fig 5 Percentage of compounds showing a weak ( ±25%), moderate ( ǀ25–50 ǀ%), and

strong ( > ± 50%) matrix effect in the barley, oat, rice and wheat matrices compared

to the rye matrix

those compounds in different kinds of cereal samples (barley, oat, rice, and wheat) Preparing semi-procedural calibration curves with each type of cereals is a tremendous effort in routine analysis laboratories, and a signiﬁcant reduction of labor is achieved by preparing one semi-procedural calibration with rye

A strong matrix effect was observed for 9% of the compounds in wheat, 14% of the compounds in barley, 27% of the compounds in oat, and 30% of the compounds in rice The more complex a ma-

Trang 7

Fig 6 Extracted chromatograms of pirimiphos-methyl and chlormephos at a spiking level of 5 μg kg −1 in rye, wheat, barley, oat, and rice

trix, and the higher the amount of fatty components it contains,

the stronger the matrix effect Cereal grain is a complex, hetero-

geneous mixture of a relatively wide range of chemical substances

The gross composition differs among cereals The total amount of

fatty acids in wheat, rye, barley, rice, and oat are 2, 2.3, 2.4, 2.9,

and 6.5 g per 100 g matrix, respectively [21], which explains why

the strongest matrix effect was observed in rice (30% of the com-

pounds) and in oat (27% of the compounds) In cases of a strong

matrix effect, the analyte should be quantiﬁed using standard ad-

dition or an external matrix-matched calibration prepared with the same matrix as the sample Therefore, using a semi-procedural calibration of rye, the current study validated 170, 159, 142, 130, and

127 compounds in rye, wheat, barley, oat, and rice, respectively Table2shows the limit of quantitation (LOQ), recovery, repeatability obtained for each compound at the spiking levels of 5, 10, and 50 μg kg-1, and the corresponding MRLs Successful results had a recovery between 70 and 120% and a relative standard deviation (RSD) of less than 20% These analytical ﬁgures validate the

Trang 8

Table 2

Compound recoveries (%), LOQs (mg kg −1 ), repeatability (%) for each spiking level in cereal matrices, and corresponding MRLs (EU pesticides

database, 2021) [ 8 ]

Compound MRL (mg.kg −1 ) Matrix LOQ (mg kg −1 ) Spiking levels (mg kg −1 )

Recovery% (RSD)

2 3-Hydroxycarbofuran 0.01 3

11 Bromophos-ethyl 0.01 Barley, oat, rice, rye, and wheat 0.005 95 (9) 84 (5) 82 (7)

12 Bromopropylate 0.01 Barley, oat, rice, rye, and wheat 0.005 114 (7) 90 (6) 90 (9)

13 Bromuconazole 0.03 2 Barley, oat, rice, rye, and wheat 0.005 107 (8) 92 (5) 94 (8)

18 Carbosulfan 0.01 1 Barley, oat, rice, rye, and wheat 0.005 103 (8) 88 (7) 91 (5)

22 Chlorfenvinphos 0.01 Barley, oat, rice, rye, and wheat 0.005 111 (12) 88 (9) 88 (9)

23 Chlormephos 0.01 3 Barley, oat, rice, rye, and wheat 0.005 86 (8) 90 (36) 91 (11)

24 Chlorobenzilate 0.02 Barley, oat, rice, rye, and wheat 0.005 106 (9) 89 (7) 95 (7)

25 Chlorpropham 0.01 Barley, oat, rice, rye, and wheat 0.005 89 (10) 90 (13) 91 (7)

27 Chlorpyrifos-methyl 0.01 Barley, oat, rice, rye, and wheat 0.005 104 (9) 88 (8) 83 (8)

30 Cyhalothrin-lambda 0.05 1 , 2 Barley, oat, rye, and wheat 0.005 96 (20) 91 (12) 96 (14)

32 Cyproconazole 0.1 2 Barley, oat, rice, rye, and wheat 0.005 110 (9) 93 (7) 96 (8)

34 Demeton-S-methyl 0.02 1 , 2 Barley, oat, rice, rye, and wheat 0.005 101 (13) 77 (12) 81 (10)

40 Difenoconazole 0.1 Barley, oat, rice, rye, and wheat 0.005 111 (12) 86 (6) 90 (17)

41 Dimethoate 0.02 2 Barley, oat, rice, rye, and wheat 0.005 112 (12) 85 (11) 85 (12)

45 Ditalimphos 0.01 3 Barley, oat, rice, rye, and wheat 0.005 91 (11) 70 (11) 82 (17)

47 Endosulfan-alpha 0.05 1 Barley, oat, rice, rye, and wheat 0.005 86 (9) 85 (5) 85 (6)

48 Endosulfan-beta 0.05 1 Barley, oat, rice, rye, and wheat 0.005 90 (11) 83 (6) 82 (6)

49 Endosulfan-sulfate 0.05 1 Barley, oat, rice, rye, and wheat 0.005 112 (11) 90 (5) 86 (8)

52 Epoxiconazole 0.6 2 Barley, oat, rice, rye, and wheat 0.005 87 (11) 98 (7) 93 (9)

58 Fenamiphos 0.02 1 Barley, oat, rice, rye, and wheat 0.005 109 (19) 97 (11) 98 (15)

64 Fenitrothion 0.05 Barley, oat, rice, rye, and wheat 0.005 112 (10) 85 (8) 86 (8)

66 Fenpropathrin 0.01 Barley, oat, rice, rye, and wheat 0.005 115 (9) 94 (6) 95 (12)

68 Fenpropimorph 0.15 2 Barley, oat, rice, rye, and wheat 0.005 92 (13) 78 (14) 74 (17)

( continued on next page )

Trang 9

Table 2 ( continued )

Recovery% (RSD)

71 Fenthion-sulfone 0.01 1 Barley, oat, rice, rye, and wheat 0.005 112 (9) 91 (6) 95 (12)

73 Fenvalerate 0.2 2 Barley, oat, rice, rye, and wheat 0.005 117 (13) 82 (58) 78 (16)

75 Fluazifop-butyl 0.01 Barley, oat, rice, rye, and wheat 0.005 107 (9) 91 (7) 97 (8)

80 Flutriafol 0.15 2 Barley, oat, rice, rye, and wheat 0.005 117 (12) 93 (8) 101 (12)

86 Heptenophos 0.01 3 Barley, oat, rice, rye, and wheat 0.005 103 (13) 85 (10) 77 (15)

92 Isofenphos-methyl 0.01 3 Barley, oat, rice, rye, and wheat 0.005 100 (8) 88 (7) 90 (6)

93 Isoprothiolane 0.01 2 Barley, oat, rice, rye, and wheat 0.005 110 (7) 98 (6) 98 (8)

96 Kresoxim-methyl 0.08 2 Barley, oat, rice, rye, and wheat 0.005 106 (9) 94 (6) 91 (6)

103 Metalaxyl 0.01 1 , 2 Barley, oat, rice, rye, and wheat 0.005 103 (8) 89 (7) 91 (5)

104 Metconazole 0.06 2 Barley, oat, rice, rye, and wheat 0.005 117 (8) 92 (8) 102 (9)

109 Methoxychlor 0.01 Barley, oat, rice, rye, and wheat 0.005 117 (18) 89 (6) 81 (14)

114 Myclobutanil 0.01 Barley, oat, rice, rye, and wheat 0.005 106 (10) 90 (7) 94 (6)

115 Nuarimol 0.01 3 Barley, oat, rice, rye, and wheat 0.005 109 (16) 111 (11) 97 (8)

119 Paclobutrazol 0.01 Barley, oat, rice, rye, and wheat 0.005 105 (10) 90 (8) 93 (8)

121 Parathion 0.05 2 Barley, oat, rice, rye, and wheat 0.005 111 (10) 96 (11) 90 (9)

122 Parathion-methyl 0.02 1 Barley, oat, rice, rye, and wheat 0.005 117 (12) 86 (9) 85 (10)

127 Phenthoate 0.01 3 Barley, oat, rice, rye, and wheat 0.005 108 (11) 84 (7) 83 (8)

132 Pirimicarb-desmethyl 0.01 3 Barley, oat, rice, rye, and wheat 0.005 108 (8) 97 (14) 86 (6)

133 Pirimiphos-methyl 0.5 2 Barley, oat, rice, rye, and wheat 0.005 100 (10) 105 (5) 92 (8)

137 Propargite 0.01 Barley, oat, rice, rye, and wheat 0.005 114 (19) 105 (17) 90 (11)

138 Propiconazole 0.09 2 Barley, oat, rice, rye, and wheat 0.005 110 (9) 96 (6) 94 (8)

( continued on next page )

Trang 10

Table 2 ( continued )

Recovery% (RSD)

140 Propyzamide 0.01 Barley, oat, rice, rye, and wheat 0.005 96 (12) 85 (7) 87 (5)

141 Prosulfocarb 0.01 Barley, oat, rice, rye, and wheat 0.005 106 (9) 89 (10) 87 (5)

142 Prothiofos 0.01 3 Barley, oat, rice, rye, and wheat 0.005 100 (7) 86 (6) 83 (9)

147 Pyrimethanil 0.05 2 Barley, oat, rice, rye, and wheat 0.005 88 (9) 74 (33) 85 (8)

149 Quinoxyfen 0.02 2 Barley, oat, rice, rye, and wheat 0.005 99 (9) 83 (5) 80 (9)

153 Tebuconazole 0.3 2 Barley, oat, rice, rye, and wheat 0.005 113 (7) 108 (10) 100 (11)

154 Tebufenpyrad 0.01 Barley, oat, rice, rye, and wheat 0.005 112 (9) 94 (6) 97 (8)

157 Tetraconazole 0.05 2 Barley, oat, rice, rye, and wheat 0.005 101 (9) 90 (6) 94 (6)

159 Thiamethoxam 0.02 2 Barley, oat, rice, rye, and wheat 0.005 93 (18) 81 (8) 78 (16)

160 Thiometon 0.01 3 Barley, oat, rice, rye, and wheat 0.005 80 (9) 74 (18) 75 (9)

161 Tolclofos-methyl 0.01 Barley, oat, rice, rye, and wheat 0.005 94 (8) 90 (8) 88 (5)

162 Triadimefon 0.01 Barley, oat, rice, rye, and wheat 0.005 103 (11) 89 (7) 90 (7)

163 Triadimenol 0.01 2 Barley, oat, rice, rye, and wheat 0.005 112 (15) 93 (11) 100 (10)

167 Triﬂoxystrobin 0.3 2 Barley, oat, rice, rye, and wheat 0.005 109 (9) 93 (7) 94 (8)

171 Vinclozolin 0.01 Barley, oat, rice, rye, and wheat 0.005 99 (8) 88 (6) 88 (5)

1 Metabolites included in the residue deﬁnition

2 Assignment of MRLs for rye where MRLs for cereals’ category is not mentioned

3 Application of a general default MRL of 0.01 mg.kg −1 where a pesticide is not speciﬁcally mentioned

effectiveness of the developed method Reproducibility, estimated

as the relative response deviation among cereal samples, was less

than 20% for almost all of the studied compounds Fig 6 shows

the overlay of the ion chromatograms of pirimiphos-methyl and

chlormephos in the ﬁve cereal matrices at a spiking level of 5 μg

kg-1 The μ-SPE method was not feasible for the determination

of two compounds, 3-hydroxycarbofuran and methacrifos, which

were not successfully validated, with 3-Hydroxycarbofuran being

diﬃcult to analyze by GC LOQs reached were below the MRLs ex-

cept for 11 compounds Among these, some were also not validated

in house with QuEChERS and d-SPE (hexaconazole and formothion)

and others were validated using LC-MS/MS (fenhexamid, ethiofen-

carb, and vamidothion) Dichlorvos is very volatile which can ex-

plain its loss during the analysis and the relatively high LOQ ob-

tained Additionally, and based on laboratory experience, the GC-

Orbitrap is not as sensitive as the triple quadrupole MS with the

Advanced Electron Ion (AEI) source used in the laboratory routine

analysis

4 Conclusion

The main beneﬁt of μ-SPE is the increase in laboratory pro-

ductivity and sample throughput, with an associated reduction of

labor The best strategy for accurate pesticide determination and

quantitation is the use of semi-procedural matrix calibration The

automated μ-SPE system could be used as a standalone system, or

it could be coupled to a high-sensitivity analytical instrument In

the latter case, the addition of some features, such as a thermo-

static autosampler, a larger size tray, and an automatic de-capping and capping system, would be optimal

Authors’ contribution

Elena Hakme (conception and design of the study, acquisi- tion of data, analysis and/or interpretation of data, drafting the manuscript)

Mette Erecius Poulsen (conception and design of the study, re- vising the manuscript critically for important intellectual content)

Declaration of Competing Interest

The authors declare that they have no known competing ﬁnan- cial interests or personal relationships that could have appeared to inﬂuence the work reported in this paper

Acknowledgments

The authors received funding from the European Union Refer- ence Laboratory of pesticide residues in cereals and feeding stuff (EURL-CF) The authors thank Thomi Preiswerk from CTC Analytics for his technical support

Supplementary materials

Supplementary material associated with this article can be found, in the online version, at doi: 10.1016/j.chroma.2021.462384

Tiêu đề	Evaluation of the automated micro-solid phase extraction clean-up system for the analysis of pesticide residues in cereals by gas chromatography-Orbitrap mass spectrometry
Tác giả	Elena Hakme, Mette Erecius Poulsen
Trường học	National Food Institute, Technical University of Denmark
Chuyên ngành	Food Analysis
Thể loại	Research article
Năm xuất bản	2021
Thành phố	Søborg

Định dạng
Số trang	11
Dung lượng	1,68 MB