Rapid determination of phenoxy acid residues in rice by modified QuEChERS extraction and liquid chromatography–tandem mass spectrometry

Rapid determination of phenoxy acid residues in rice by modified QuEChERS extraction and liquid chromatography–tandem mass spectrometry

a n a l y t i c a c h i m i c a a c t a 6 2 6 ( 2 0 0 8 ) 10–20 available at www.sciencedirect.com journal homepage: www.elsevier.com/locate/aca Ra...

1MB Sizes 0 Downloads 4 Views

a n a l y t i c a c h i m i c a a c t a 6 2 6 ( 2 0 0 8 ) 10–20

available at www.sciencedirect.com

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

Rapid determination of phenoxy acid residues in rice by modified QuEChERS extraction and liquid chromatography–tandem mass spectrometry Urairat Koesukwiwat a , Kunaporn Sanguankaew a , Natchanun Leepipatpiboon a,b,∗ a

Chromatography and Separation Research Unit, Department of Chemistry, Faculty of Science, Chulalongkorn University, Bangkok 10330, Thailand b National Research Center for Environmental and Hazardous Waste Management (NRC-EHWM), Chulalongkorn University, Bangkok 10330, Thailand

a r t i c l e

i n f o

a b s t r a c t

Article history:

A new method for the analysis of phenoxy acid herbicide residues in rice, based on the use of

Received 5 April 2008

liquid extraction/partition and dispersive solid phase extraction (dispersive-SPE) followed by

Received in revised form

ultra-performance liquid chromatography–electrospray ionization tandem mass spectrom-

19 July 2008

etry (UPLC–ESI–MS/MS), is reported. 5% (v/v) formic acid in acetonitrile as the extraction

Accepted 21 July 2008

solvent and inclusion of citrate buffer helped partitioning of all the analytes into the ace-

Published on line 29 July 2008

tonitrile phase. The extract was then cleaned up by dispersive-SPE using C18 and alumina neutral as selective sorbents. Further optimization of sample preparation and determina-

Keywords:

tion allowed recoveries of between 45 and 104% for all 13 phenoxy acid herbicides with

Herbicide residues

RSD values lower than 13.3% at 5.0 ␮g kg−1 concentration level. Limit of detections (LODs) of

Phenoxy acid

0.5 ␮g kg−1 or below were attained for all 13 phenoxy acids. Quantitative analysis was done

Quick, Easy, Cheap, Effective, Rugged,

in the multiple-reaction monitoring (MRM) mode using two combinations of selected pre-

and Safe

cursor ion and product ion transition for each compound. This developed method produced

Ultra-performance liquid

relatively higher recoveries of the acid herbicides with a smaller range of variation and less

chromatography–mass

susceptibility to matrix effects, than the original QuEChERS (Quick, Easy, Cheap, Effective,

spectrometry

Rugged, and Safe) method. © 2008 Elsevier B.V. All rights reserved.

Rice

1.

Introduction

Herbicides are considered to be indispensable for the production of an adequate food supply for the increasing world population via the control of weed and insect borne diseases. In rice production, weed growth can be prolific and is the major constraint on crop yields causing seriously reduced or even no rice production. Consequently, herbicides are being used with both increasing frequency and amounts to control weeds and provide protection against damages caused by



Corresponding author. Tel.: +66 2 2187608; fax: +66 2 2541309. E-mail address: [email protected] (N. Leepipatpiboon). 0003-2670/$ – see front matter © 2008 Elsevier B.V. All rights reserved. doi:10.1016/j.aca.2008.07.034

some insects, fungi, etc. In Thailand [1,2], chlorophenoxy or phenoxy acid herbicides (Fig. 1) are widely used for weed and broadleaf plants control, by mixing into commercial fertilizers. It prevents rice from competition and damage by reducing or prohibiting the germination of grass, broadleaf weeds, and seedlings. The intensive use of these herbicides may, however, make them an emerging issue in directed-seed planting systems and pose a serious environmental threat to non-target organisms. Consumers can be subject to indirect exposure due to the presence of residual phenoxy acids in rice. Indeed,

a n a l y t i c a c h i m i c a a c t a 6 2 6 ( 2 0 0 8 ) 10–20

11

Fig. 1 – Chemical structure of phenoxy acid compounds.

exposure to even very low doses of some phenoxy acids may be responsible for human organ mechanism [3]. Consequently, EU have established legal directives to restrict the use and control the maximum residue levels of some phenoxy acids in rice for human consumption at 0.050 mg kg−1 according to EU Directive 86/362/EEC (for cereals) [4]. The polar nature and high water solubility of phenoxy acids make their specific extraction very difficult. The monitoring of phenoxy acids at trace levels can be an extremely challeng-

ing task in the case of complicated matrices because of the large of co-extracted components that may adversely affect the extraction efficiency and instrument performance for quantitative detection. For this reason, the maximum residue limits (MRLs) imposed by food safety legislation can usually be achieved by sample preparation techniques that provide a high extraction efficiency of the phenoxy acid analytes. Traditionally, a diverse array of sample preparation methods have been used, with currently (i) liquid–liquid extraction

12

a n a l y t i c a c h i m i c a a c t a 6 2 6 ( 2 0 0 8 ) 10–20

[5–8], (ii) subcritical water extraction [9], (iii) dynamic liquid–liquid–liquid microextraction (D-LLLME) [10], (iv) solid phase extraction (SPE) [6,11–19], (v) stir-bar sorptive extraction (SBSE) [20], (vi) supported liquid membrane extraction (SLME) [21], (vii) supercritical fluid extraction (SFE) [22] and (viii) solid phase microextraction (SPME) [23] being widely adopted for detection of phenoxy acid residues. However, these methods are mainly reported for the analysis of residual phenoxy acids in soil [7–9,12] and water [5,10,11,13–16,18,20,21,23], but few applications reported and characterized in fruits and vegetables [17,24]; barleys and cereals [25]; and other food crops [7,22]. Moreover, these methods use large amount of organic solvents, tedious, time-consuming, and also lack sufficient selectivity, which are an important consideration for trace analysis. Recently, the QuEChERS method (Quick, Easy, Cheap, Effective, Rugged, and Safe) has been developed as an attractive alternative method for sample preparation [26], and has received the distinction of recognition as an official method for use in the detection of multiple pesticide residues in fruits and vegetables. The QuEChERS method is particularly popular for determination of polar, middle polar, and non-polar pesticide residues in various food matrices [27–35] because of its simplicity, inexpensive, amenable to high throughput, and relatively high efficiency results with a minimal number of steps. Chromatographic methods dedicated to the determination of phenoxy acid residues such as GC [8,11], GC/MS [5,6,12,20,23,25], and CE [16,18,22,36] have been reported. However, despite the great efficiency of GC and GC/MS in pesticide analysis, phenoxy acid residues cannot be analyzed directly due to their lack of volatility and instead critically require derivatisation (e.g. with diazomethane, BF3 , BSTEA, MTBSTFA) to confer volatility to the analytes prior to analysis. The derivatisation requires a fair amount of sample manipulations and time-consuming. More importantly, the toxicity of many GC-derivatising agents to human health has limited their usefulness in the field of non-volatile herbicides. Consequently, HPLC [6,9,10,12,14,17–19,21], LC–MS [13,24], and LC–MS/MS [13,15,24,25] have become valuable techniques in the multiresidue pesticide analysis. They provide the most efficient confirmatory tools that can discriminate the residues at ultra trace levels for the reliable and sensitive detection to help ensure food safety. Despite the extensive worldwide use of phenoxy acids in agricultural crops and the restrictive regulations on their MRLs in food, it is surprising that no specific methods have been reported for the determination of these acid herbicides in rice. However, this is likely due to their chemical properties, especially at low residue levels. Moreover, the matrices in rice both polar (carbohydrate and sugar) and non-polar (starch, macromolecule, and pigment) compounds make the difficulties of extraction. As results, an effective and reliable sample preparation and sensitive determination methods are invariable and essential requirements. The aim of this work was to develop a rapid and effective method for the analysis of low levels of phenoxy acid herbicide residues in rice samples that fulfill the requirements of excellent sensitivity and unequivocal confirmation of the residues detection according to the EU guidelines. This paper describes the extension of the systematic QuEChERS approach to the

determination of phenoxy acids in rice. The method proved to be an efficient, reliable, and sensitive for 13 phenoxy acid herbicides most commonly used in rice production. The influence of several parameters was investigated and reported.

2.

Experimental

2.1.

Materials and reagents

Methanol, ethyl acetate, and acetonitrile as high purity solvents were obtained from Merck (Darmstadt, Germany). Formic acid was supplied by Fluka (Buchs, Switzerland). Magnesium sulfate (MgSO4 ), sodium chloride (NaCl), tri-sodium citrate dihydrate (tri-Na), and di-sodium hydrogencitrate sesquihydrate (di-Na) all of analytical grade were purchased from Merck (Darmstadt, Germany). Before use, MgSO4 was activated by heating at 600 ◦ C for 7 h and cooled and kept in desiccators. Primary secondary amine (PSA, 40 ␮m), octadecyl (C18, 40 ␮m), alumina neutral (25 ␮m) sorbents were from Varian (Harbor City, CA, USA). Graphitized carbon black (GCB, 400 meshes) was from Supelco (Bellefonte, PA, USA). A Milli-Q ultrapure water system from Waters (Milford, MA, USA) was used throughout the study.

2.2.

Phenoxy acid standards

All herbicide standards, namely acifluorfen, clopyralid, cyclanilide, 2,4-dichlorophenoxyacetic acid (2,4-D), 4-(2,4-dichlorophenoxy) butyric acid (2,4-DB), 2methoxy-3,6-dichlorobenzoic acid (dicamba), fluroxypyr, 2-methyl-4-chlorophenoxyacetic acid (MCPA), 4-(2-methyl4-chlorophenoxy) butyric acid (MCPB), mecroprop (MCPP), picloram, quinclorac, (2,4,5-trichlorophenoxy) acetic acid (2,4,5-T), and triclopyr were purchased from Dr. Ehrenstorfer GmbH (Augsburg, Germany) with the highest available purity grade. Each individual stock solution was prepared as 1.00 g L−1 in methanol and stored at −18 ◦ C for up to 2 months. The working solutions were prepared daily by appropriately diluting the stock solutions with mobile phase and stored at 4 ◦ C until use.

2.3.

Instrumental and chromatographic conditions

A Micromass Quattro PremierTM XE benchtop tandem quadrupole mass spectrometer was interface via electrospray ionization (ESI) to an Acquity Ultra Performance LC (UPLC) System (Waters). The phenoxy acid analytes were separated on an Acquity BEH C18 (2.1 mm × 100 mm, 1.7 ␮m) column (Waters). The mobile phases, which were composed of 0.1% (v/v) formic acid in water (A) and 0.1% (v/v) formic acid in methanol (B) were pumped at a flow rate of 0.200 mL min−1 . Simultaneous separations were completed using a gradient profile of: 0.0 min/10% B, 5.0 min/90% B, 8.0 min/90% B, 8.1 min/10% B, and 10.0 min/10% B, respectively. The injection volume was 10.0 ␮L. All 13 phenoxy acid herbicides were eluted within 9 min. For operation in ESI–MS/MS mode, nitrogen was used for the cone and desolvation gas flows set at 30 and 997 L h−1 , respectively. A capillary voltage of 0.80 and 1.20 kV were used

198.78 [M−H]− > 140.64 [M–H–CH2 CO2 ]− 218.84 [M−H]− > 175.01 [M–H–CO2 ]− 218.93 [M−H]− > 161.08 [M–H–CH2 CO2 ]− 212.79 [M−H]− > 140.70 [M–H–CH3 CHCO2 ]− 227.01 [M−H]− > 141.28 [M–H–(CH2 )3 CO2 ]− 239.93 [M−H]− > 195.96 [M–H–CO2 ]− 246.65 [M−H]− > 160.59 [M–H–(CH2 )3 CO2 ]− 252.55 [M−H]− > 194.65 [M–H–CH2 CO2 ]− 254.60 [M−H]− > 196.53 [M–H–CH2 CO2 ]− 255.61 [M−H]− > 197.53 [M–H–CH2 CO2 ]− 359.97 [M−H]− > 315.87 [M–H–CO2 ]− 191.91 [M+H]+ > 173.96 [M+H − H2 O]+ 240.95 [M+H]+ > 194.90 [M+H − H2 O–CO]+

in negative and positive ionization modes, respectively. The extractor voltage was set at 3.00 V and a source temperature of 120 ◦ C was chosen. MS/MS experiments were carried out in multiple reactions monitoring mode (MRM). All other ESI and MS parameters were optimized individually for each phenoxy acid compounds. However, the increased sensitivities obtained by scanning a limited mass range or by SIM are essential. Two mass ions were acquired for each residue confirmation. These were used so that quantification and confirmation could be performed with a single injection assuming that the ion ratios between masses or transitions were consistent for standards and samples.

2.4.

Sample preparation

18.0 7.0 13.0 20.0 12.0 7.0 10.0 13.0 15.0 15.0 10.0 11.0 13.0 22.0 12.0 21.0 15.0 18.0 14.0 15.0 13.0 10.0 20.0 15.0 19.0 24.0 2.0 1.97

2.8

200.62 221.04 221.04 214.65 228.67 242.06 249.09 255.03 255.48 256.47 361.66 192.00 241.46 MCPA Dicamba 2,4-D MCPP MCPB Quinclorac 2,4-DB Fluroxypyr 2,4,5-T Triclopyr Acifluorfen Clopyralid Picloram

C9 H9 ClO3 C8 H6 Cl2 O3 C8 H6 Cl2 O3 C10 H11 ClO3 C11 H13 ClO3 C10 H5 Cl2 NO2 C10 H10 Cl2 O3 C7 H5 Cl2 FN2 O3 C8 H5 Cl3 O3 C7 H4 Cl3 NO3 C14 H7 ClF3 NO5 C6 H3 Cl2 NO2 C6 H3 Cl3 N2 O2

3.1 1.9 2.7 3.6

ES− ES− ES− ES− ES− ES− ES− ES− ES− ES− ES− ES+ ES+

0.5–8.5

1.5–8.5

0.10 0.05 0.10 0.10 0.10 0.08 0.10 0.10 0.10 0.10 0.25 0.10 0.10 0.0–8.5

Precursor ion (MS1) > Product ion (MS2)

MRM transition (m/z) Collision energy (V) Cone voltage (V) Dwell time (s) Retention time window (min) Polarity M.W.

pKa [14,17]

13

Rice samples were prepared similar to the original QuEChERS method.

M.F. Compound

Table 1 – Herbicide molecular formulas, molecular weights, polarities, retention times, and optimized MS/MS parameters for the 13 phenoxy acid herbicides selected for residue analysis in rice sample by UPLC–ESI–MS/MS

a n a l y t i c a c h i m i c a a c t a 6 2 6 ( 2 0 0 8 ) 10–20

Step I : A thoroughly homogenized sample 10 g sample of rice was weighed in a plastic centrifuge tube. Step II : 5.0 mL of water and 10.0 mL of acetonitrile were added and the tube was shaken vigorously for 1 min by hand ensuring that the solvent interacted well with the entire sample. Step III : 4 g of anhydrous MgSO4 and 1 g of NaCl were added into the mixture and the shaking step was repeated for 1 min. Step IV : After centrifugation (3800 rpm, 1 min), 1.00 mL of the clarified supernatant was introduced into a new centrifuge tube containing 0.25 g of PSA sorbents and 1.50 g of anhydrous MgSO4 . The mixture was then shaken for 1 min and centrifuged for 1 min. Step V : 5.0 mL of the upper phase extract was evaporated to dryness at 40 ◦ C and reconstituted in 1.00 mL of mobile phase and filtered through a 0.20 ␮m filter prior to UPLC–ESI–MS/MS analysis. Fortified phenoxy acids in rice samples were prepared by adding known amounts of phenoxy acid standards mixture into 10 g of representative rice sample (step I) and extracted following the above procedures (steps II–V).

2.5.

Method performance

The limit of detection (LOD), estimated for product ion of each analyte, is defined as the lowest concentration that the analytical method can reliably differentiate from background levels. LOD values of each phenoxy acid were estimated for spiked rice samples at 0.01 ␮g kg−1 and calculation based on the detection signal being three times over the average of background noise assayed.

3.

Results and disscussion

3.1.

UPLC–ESI–MS/MS conditions

For the purpose of finding the retention times and the best resolution between the analyte peaks, preliminary experiments were carried out to systematically vary the strength of

14

a n a l y t i c a c h i m i c a a c t a 6 2 6 ( 2 0 0 8 ) 10–20

the mobile phase and fragmentor voltage in full scan mode using phenoxy acid standard solutions. However, increased sensitivities were obtained by scanning a limited mass range. Under ESI–MS/MS conditions, precursor ion of each phenoxy acid was selected for collision-induce dissociation (CID) experiment, which generated product ions. Consequently, the one product ion with high intensity, representing the characteristic of each compound, was set to monitor for quantification and for identification. Table 1 shows the values of the MRM transitions used for quantification and identification of selected herbicides and other optimal conditions for phenoxy acids analysis. Acquisition MRMs was performed under timescheduled conditions, the various selected transitions were programmed into different time acquisition windows in order to avoid a decrease in sensitivities. The molecular weights of 13 phenoxy acids were concluded on the basis of their negative and positive ion ESI mass spectra, which showed precursor ions [M−H]− and [M+H]+ in MS1. Collision induced fragmentation leads to the loss of alkyl and carboxylic acid group [M–H–RCO2 ]− , which was in accordance with the fragmentation pathway of phenoxy acids in negative ESI. The loss of fragment corresponding to a mass of [M+H–H2 O]+ from clopyralid and [M+H–H2 O–CO]+ from picloram were presented in the positive ESI mass spectra. These characteristic fragment ions were used in MS2 for quantification and confirmation of the phenoxy acid residues. The possible structures of all analytes were deduced by careful studies on their MS/MS spectra. The developed method is also highly selective with the monitoring of specific MRM of each analyte, which is essential to reduce the risk of false positive results. The general minimum criteria required by the European Commission (Document no. SANCO/2007/3131) for reliable identification and quantification of pesticide residues is for data from two ions for m/z > 200 [37], and so this guideline, that is increasingly being adopted in the food and feed, was achieved.

3.2.

Optimization of extraction conditions

As previously stated, the analysis of phenoxy acids in complicated matrices is very difficult due to their chemical and physical natures. Several analysis methods that have been employed before are summarized in Table 2, which also clearly illustrates the sparsity of assays developed for crop analysis compared to soil or water. In this study, the proposed extraction method was based on the extraction, liquid–liquid partition and dispersive-solid phase clean up procedure. Several factors have been shown to affect the performance in original QuEChERS such as the nature of the solvent addition, the nature of solid phase sorbent, pretreatment or modification of the sample (pH adjustment), most specifically for the nature of the sample matrices and analytes separation were also studied.

3.2.1.

Extraction and liquid–liquid partition

In general, liquid extraction is the fundamental method utilized for the isolation of acid herbicides from various food matrices such as by the Luck method [25]. Many aspects such as the ability to extract equally efficiently different pesticides of wide polarity ranges, the selectivity involved in extraction and clean up steps, and the compatibility with the separation techniques, all have to be considered and then optimized. Therefore, the choice of solvent is one of the most important decisions in compiling a multiresidue extraction method. For any liquid–liquid extraction, the partition coefficient of the analytes between two liquid phases is an important aspect to be optimized. Phenoxy acids are relatively strong acids (pKa < 4) [14,17] and are more stable at low pH. Therefore, the pH of the extraction is also important and should be suitably controlled. In this study, two types of extraction/partition organic solvent, (A) acetonitrile and (B) acidified acetonitrile with formic

Table 2 – Comparison of various analytical methods developed for the analysis of phenoxy acid herbicides in several matrices Sample preparation LLE Subcritical water extraction SPE SPE-solid phase derivatisation SPE SPE SPE SPE SPE SPE SPE SBSE SPME SFE SLM D-LLLME MIP QuEChERS QuEChERS Modified QuEChERS NR refers to not reported.

Matrix

Analysis/detection

Water Soil Water Water Water Water Water Water Water Soil Apple juice Water Water Food crops Ground water Water Water Barley Fruit and vegetable Rice

GC–MS HPLC/DAD HPLC/UV–vis GC–ECD GC–MS LC–APCI–MS/MS HPLC/UV–vis and GC–MS LC–MS/MS MECC–CE/UV–vis HPLC/UV–vis and GC-MS Capillary-LC/UV–vis GC–MS GC–MS MEKC-CE/UV–vis HPLC/PDA HPLC/UV–vis HPLC/UV–vis and CZE-CE LC–ESI–MS/MS LC–TOF/MS and LC–ESI–MS/MS UPLC–ESI–MS/MS

LOD (␮g kg−1 ) 0.01–0.06 NR 2.0–3.0 NR NR 0.01–0.07 0.05 0.001–0.075 2.0–5.0 NR 5.0–18.0 0.001–0.8 NR 0.6 pg 0.3–0.6 0.1–0.4 NR 0.2–2.0 15.0 and 5.0 0.5–5.0

Recovery (%) NR 94–113 >95 75–137 80–120 NR >80 91–98 NR 82–90 84–99 70–130 87–112 95–99 80–126 94–112 87.0–103.5 60–70% NR 45–104

Refs. [5] [9] [14] [11] [3] [13] [6] [15] [16] [6] [17] [20] [23] [22] [21] [10] [18] [25] [24] This work

a n a l y t i c a c h i m i c a a c t a 6 2 6 ( 2 0 0 8 ) 10–20

15

Fig. 2 – The effect of organic solvents and dispersive-SPE sorbents on recovery of 13 phenoxy acid herbicides (n = 3) spiked at 0.05 mg kg−1 . (A) Acetonitrile–PSA (original QuEChERS method), (B) 5% (v/v) formic acid in acetonitrile – 0.25 g PSA, and (C) 5% (v/v) formic acid in acetonitrile – 0.25 g C18. UPLC–ESI–MS/MS conditions are described in Sections 2.3 and 3.1.

acid were evaluated for their extraction efficiency (step II, Section 2.4). The results using acetonitrile as in the original QuEChERS method were used as standard method to compare with the acidified acetonitrile extraction (5% (v/v) formic acid in acetonitrile). For 12/13 of the tested phenoxy acid herbicides significantly better recoveries were achieved when using 5% (v/v) formic acid in acetonitrile as the extraction solvent (Fig. 2). Only clopyralid recovery was not improved and remains undetected with and without inclusion of the formic acid in the extraction. Considering the ionizable and high water soluble characteristics, the acid dissociation equilibrium for phenoxy acid analytes would be pushed towards their protonated neutral forms by the addition of formic acid into solvent extraction. However, rice is composed of high sugar and carbohydrate contents and under these conditions acetonitrile can be separated more easily from water with the addition of the amount of NaCl to induce phase separation (the acetonitrile become water immiscible) and partition the neutral form of phenoxy acids into the acetonitrile phase resulting in the improved extraction recoveries. Additionally, the removal of residual water by MgSO4 adsorption helps to promote the partitioning of analytes into the acetonitrile phase. Moreover, the merit of acetonitrile is that it is compatible with the RP–LC mobile phase and minimizes the co-extraction of interfering matrix components [27,35]. From these results, 5% (v/v) formic acid in acetonitrile was used as an extraction solvent (step II, Section 2.4) throughout the further study.

3.2.2.

Dispersive-SPE clean up

After phenoxy acid analytes were extracted into a water soluble solvent (5% (v/v) formic acid in acetonitrile) followed by

partitioning of the analyte molecules in organic solvent in the presence of a salt mixture (salting out effect), the acetonitrile phase was further cleaned up and dried by mixing with the SPE sorbents and anhydrous MgSO4 . The sorbents were chosen to retain matrix components and allow the analytes of interest into the acetonitrile phase. During the process of sample preparation, it was found that different dispersive sorbents had a significant influence on the purification and recovery of phenoxy acid herbicide extracts. In QuEChERS method, PSA sorbent, a weak anion exchanger which strongly interacts with acid compounds, was used to remove various co-extractive interferences. The polar organic acids, sugars and fatty acids co-extracted from the QuEChERS extracts including the phenoxy acids of interest were strong retained on the PSA which resulted in low extraction recoveries. Thus, we also attempted purification using C18 in place of PSA (Fig. 2). Rice samples spiked with 0.05 mg kg−1 phenoxy acids subjected to both 0.25 g PSA or 0.25 g C18 dispersive sorbents (step IV, Section 2.4) revealed significantly different recovery for 12/13 of the tested phenoxy acid herbicides. For 9/13 analytes, a significant increase in the recovery was noted, ranging from 5 to 20% recoveries (acifluorfen & MCPB, respectively) when C18 was used in place of PSA. However, two of the other four samples showed a weak (2%) or significant (25%) decrease in yield with C18 compared to PSA whilst picloram showed a dramatic decrease in yield (84–0%) and clopyralid still remain undetected. C18 is a reversed phase sorbent which is effective at trapping (binding) and removing starch and sugar from rice samples and provides the cleanest extract. As shown in the chemical structure of clopyralid and picloram, it can be explained that nitrogen atoms in the molecules were induced by the high electronegativity of oxygen and chlorine atom,

16

a n a l y t i c a c h i m i c a a c t a 6 2 6 ( 2 0 0 8 ) 10–20

made the electrons permanently transferred from nitrogen and provided the positive charge. When C18 sorbents were dispersed into sample solution, the residual silanol group on the C18 sorbent surface bound strongly with the positive charge of clopyralid and picloram and shown the undetectable recoveries.

3.2.3.

Selectivity of dispersive-SPE clean up

At this stage, the recoveries of some phenoxy acids were increased up to 93% (triclopyr) with most (10/13) exceeding 70% extraction recoveries. However, the recovery of MCPA, and particularly, picloram, was reduced when using C18 as a dispersive sorbent. Thus, it was necessary to further improve the extraction efficiency by adjustment of other parameters. As phenoxy acid herbicides are pH dependent compounds in terms of net charge and hydrophilic-hydrophobic tendencies. To minimize the ionization of these compounds, buffering with citrate salts (composed of 1 g tri-Na and 0.5 g di-Na) should be introduced in the extraction/partitioning step (step III, Section 2.4) to manipulate the compromise pH value of 5–5.5. At this pH range, most acidic analytes were existed in their neutral forms and shown the low ionization rate [35,38]. When citrate buffer (pH ∼ 5–5.5) was added in the modified QuEChERS method, improvement in the recovery was noted for 4/13 of the phenoxy acids (Fig. 3), especially MCPA (∼45%). However, the percent of recoveries for seven other phenoxy acids were reduced, ranging from 3–18% recoveries. The other two herbicides, clopyralid and picloram, remains unchanged and undetected. Moreover, adding citrate buffer elevated the pH of rice sample and as a consequence dramatic increase in the amount of co-extracts (weak acid compounds) in the raw extracts. These co-extracts might strongly bind with the analyte molecules resulting in decreased recoveries. From here

on, citrate salts should be added in step III (Section 2.4) to manipulate the extract solution to pH ∼ 5–5.5. However, the complexity of the rice matrix requires optimization of the whole dispersive-SPE procedure (step IV, Section 2.4) in order to obtain quantitative recoveries of phenoxy acid analytes with a high degree of selectivity. Concerning the sorbent materials, the previous experiment was made using C18 which improved most recoveries. Since both the specific and non-specific interactions between co-extractive matrices and sorbents are based on their physicochemical structures and polarities, mixed-mode materials containing two sorbents were evaluated for the isolation of co-extractive compounds from acidified acetonitrile extracts. The mixed of C18-GCB and C18-alumina neutral were selected for this purpose. The samples were carried out by dispersing these mixed sorbents in the acidified acetonitrile extracts and then the mixture was shaken and centrifuged to evenly distribute the SPE materials and facilitate the clean up process. Comparing the extraction recoveries suggested that lower extraction efficiency was actually attained when using C18 mixed with GCB (Fig. 4). This could be explained by the fact that GCB provides three adsorption mechanisms, (i) anionexchange due to the positively charged oxonium group, (ii) hydrophobic interactions between the graphite surface and the aromatic structure of the analytes, and (iii) hydrogen bonding between protonated functional groups of the analytes and the carbonyl groups of the GCB [15]. For this reason, phenoxy acids trapped on the GCB sorbent will result in reduced extraction recoveries. Thus although GCB is very useful for removal of coloring substances (pigments) and sterol compounds in rice samples [31,32,38] but it appears to be less appropriate for the isolation and preconcentration of this tested herbicides.

Fig. 3 – The effect of citrate buffer addition (in step III, Section 2.4) on the recovery of 13 phenoxy acid herbicides (n = 3) spiked at 0.05 mg kg−1 . (C) No buffer addition: 4 g MgSO4 + 1 g NaCl and (D) citrate buffer addition: 4 g MgSO4 + 1 g NaCl + 1 g tri-Na + 0.5 g di-Na. UPLC–ESI–MS/MS conditions are described in Sections 2.3 and 3.1.

a n a l y t i c a c h i m i c a a c t a 6 2 6 ( 2 0 0 8 ) 10–20

17

Fig. 4 – The effect of dispersive-SPE type and sorbent mass (step IV, Section 2.4) on the recovery of 13 phenoxy acid herbicides (n = 3) spiked at 0.05 mg kg−1 . UPLC–ESI–MS/MS conditions are described in Sections 2.3 and 3.1.

When dispersive SPE was performed using a combination of C18 and 0.25 g of alumina neutral, a greater improvement in the overall recoveries of analytes was seen compared to C18 or C18-GCB, respectively (Fig. 4). However, recovery of clopyralid and picloram was still not attained. Rice contains carbohydrates, proteins, fats, vitamins, minerals, and water as the major contents, and using only C18 sorbents might not be

Fig. 5 – Schematic presentation of the optimal sample preparation method for the simultaneous extraction of 13 phenoxy acid herbicides from rice samples.

sufficient to trap all the (non-polar) co-extracted compounds. Using alumina neutral as the selective sorbent (with a pH of 6–8) will separate vitamins, glycosides and plant sterols [39], and also shows a strong retaining activity for fats and lipids. Thus it is possible that some fat, vitamins, and minerals in the rice extracts were cleaned and released bound analytes from the active sites. Therefore, alumina neutral sorbents helped bringing the overall recovery values for 11/13 phenoxy acid herbicides tested to be within the acceptable range (higher than 75%) for trace analysis (Fig. 4). However, the apparent exceeding recovery of MCPA (120%) could be explained as the interference of the co-extracted matrix components or a matrix induced chromatographic response. However, the 120% recovery is not extremely large according to the CODEX analyte recovery of pesticide and veterinary drug at the spiking level of >0.01 mg kg−1 ≤ 0.1 mg kg−1 , the range of mean recovery are between 70 and 120%. Therefore, the further clean up step led to improvement in clopyralid and picloram recoveries was also studied. If it is assumed that the co-extracted compounds are highly hydrophobic and have strong Van der Waals interactions with the non polar sorbents, then a small amount of sorbent material would be sufficient for their removal. Larger amounts, 0.25 g alumina neutral used here, would serve to also allow binding through the weaker interactions of phenoxy acid analytes and sorbents. Indeed, because phenoxy acids are neutral at this pH and strongly adsorbed on the surface of alumina neutral this could explain the reduction of clopyralid and picloram recoveries. Therefore, a lower amount (0.10 g) of alumina neutral was evaluated and, indeed, the

18

a n a l y t i c a c h i m i c a a c t a 6 2 6 ( 2 0 0 8 ) 10–20

Fig. 6 – The UPLC–ESI–MS/MS multiple-reaction ion MRM chromatograms of rice sample spiked with 13 phenoxy acid herbicides at 0.05 mg kg−1 using analytical conditions described in Sections 2.3 and 3.1.

Table 3 – Retention times, average recoveries, standard deviations and method of detection limits for the 13 phenoxy acid herbicides in rice with the modified QuEChERS method Compound

Clopyralid Picloram Quinclorac Dicamba 2,4-D MCPA Triclopyr MCPP 2,4,5-T Fluroxypyr 2,4-DB MCPB Acifluorfen a

Retention time (min) 3.45 4.27 6.93 7.06 7.97 8.01 8.29 8.41 8.45 8.49 8.58 8.62 8.62

LOD (mg kg−1 )

Mean recoverya (%) 0.005 mg kg−1 94 53 104 85 94 67 85 61 78 85 76 95 90

± ± ± ± ± ± ± ± ± ± ± ± ±

9.5 3.8 11.8 5.6 5.5 12.7 11.1 4.2 7.8 5.3 13.3 8.2 1.0

0.01 mg kg−1 77 57 87 92 94 45 80 45 66 84 59 93 91

± ± ± ± ± ± ± ± ± ± ± ± ±

2.0 1.5 7.8 9.0 3.6 3.2 1.5 2.0 4.9 13.0 2.1 2.6 3.5

0.02 mg kg−1 68 54 70 85 97 70 81 53 81 83 86 95 92

± ± ± ± ± ± ± ± ± ± ± ± ±

0.6 0.6 0.6 1.5 4.0 1.0 5.0 4.7 7.5 3.1 5.1 4.0 4.2

Values obtained from three repetition extractions. NC refers to not controlled.

0.05 mg kg−1 66 57 66 89 84 68 78 68 76 74 72 74 79

± ± ± ± ± ± ± ± ± ± ± ± ±

1.2 0.6 1.2 2.5 0.6 5.3 4.4 9.2 2.5 0.6 2.0 0.6 0.6

This method 0.005 0.005 0.005 0.001 0.0005 0.005 0.003 0.005 0.001 0.001 0.001 0.003 0.0005

MRL (mg kg−1 ) [4]

Ref. [25] – – – 0.0012 0.0002 0.0009 – 0.0020 – – – – –

NC NC NC NC 0.050 0.050 NC NC 0.050 0.050 0.050 0.050 NC

a n a l y t i c a c h i m i c a a c t a 6 2 6 ( 2 0 0 8 ) 10–20

use of 0.10 g as opposed to 0.25 g alumina neutral, resulted in the highest observed recoveries of all 13 phenoxy acids ranging from 71.5 to 97.9% (Fig. 4). The optimal QuEChERS condition for simultaneous extraction of 13 phenoxy acid residues in rice obtained from this study was concluded as shown in Fig. 5. Fig. 6 shows multiple reaction monitoring (MRM) chromatograms of 13 phenoxy acid analytes in a rice sample.

3.3.

Method performance

Table 3 summarizes the recovery data for rice samples fortified at four different concentration levels. Three replicate rice samples of each concentration level were extracted and partitioned using 5% (v/v) formic acid in acetonitrile and citrate buffer mixture salts followed by clean up using 0.25 g C18 combined with 0.10 g alumina neutral sorbents. The average recoveries ranged between 45 and 104% with 0.6–13.3% RSD. The phenoxy acid concentration limits that this analytical process can reliably differentiate from background noises or LODs were estimated for fortified rice samples based on signals at three folds higher than the background level. The LOD values in this study ranged from 0.0005 to 0.01 mg kg−1 for all 13 phenoxy acids, which is below the maximum residue limits (MRLs) required by the EU legislation [4].

4.

Conclusions

In this work, the potential of a modified QuEChERS methodology for the extraction of phenoxy acid herbicide residues in rice that were suitable for quantitative analysis has been clearly demonstrated. The optimal sample preparation procedure involved the following: extraction/partition of rice samples with acidified acetonitrile (5% (v/v) formic acid in acetonitrile) and citrate salts mixture (1 g tri-Na and 0.5 g di-Na), and cleaned up using 0.25 g C18 mixed with 0.10 g alumina neutral as dispersive sorbents prior to analysis by UPLC–ESI–MS/MS (Fig. 5). Although not consistently effective across all tested phenoxy acid residues, however, the use of citrate buffer and mixed SPE sorbents during the extraction helped improve the recovery of phenoxy acid analytes within the acceptable range. Whilst the selectivity of this method including the use of C18 and alumina neutral at the appropriate ratio allows for clean extracts to be obtained which permits better detection limits and dramatically reduces sample matrices. Although the difficult phenoxy acid residues are prone to strong matrix interactions in dry samples [25], satisfactory recoveries (up to 104%) were obtained at the level of 5–50 ␮g kg−1 with a sensitivity at 5 ␮g kg−1 or less for all 13 evaluated phenoxy acid herbicides. This novel developed method is more sensitive than conventional solid phase extraction or liquid–liquid extraction. UPLC–ESI–MS/MS could analyze all 13 phenoxy acid herbicides within a short chromatographic run time of 9 min. Therefore, the modified QuEChERS method can be regarded as a strong alternative method to current extraction techniques for fast, simple, inexpensive and real-time screening within the MRL limits of these types of compounds.

19

Acknowledgement The authors gratefully acknowledge The Royal Golden Jubilee Ph.D Program (PHD0036/2550) and the Thailand Research Fund and Commission on Higher Education, Research Grant for MidCareer University Faculty (TRF-CHE-RES-MR) (RMU5180009) for providing financial support. Special acknowledgement to the National Center of Excellence for Petroleum, Petrochemicals, and Advanced Materials (NCE-PPAM).

references

[1] Data of Imported Agricultural Hazardous Substances: January 2003, Department of Agriculture, Bangkok, Thailand. [2] Accessed January 2008. http://www.siamsafety.com. [3] I.R. Pereiro, R.G. Irimia, E.R. Cano, R.C. Torrijos, Anal. Chim. Acta 524 (2004) 249. [4] The Health and Consumer Protection Directorate General of the European Commission Home Page. http://ec.europa.eu/ food/plant/protection/pesticides/index en.htm (accessed January 2008). [5] M.I. Catalina, J. Dalluge, R.J.J. Vreuls, U.A.T. Brinkman, J. Chromatogr. A 877 (2000) 153. [6] C.W. Thorstensen, O. Lode, A.L. Christiansen, J. Agric. Food Chem. 48 (2000) 5829. [7] S. Roy, S.B. Singh, J. Chromatogr. A 1065 (2005) 199. [8] M.J. Santos-Delgado, E. Crespo-Corral, L.M. Polo-Diez, Talanta 53 (2000) 367. [9] J.L. Luque-Garcia, M.D. Luque de Castro, J. Chromatogr. A 959 (2002) 25. [10] J. Wu, K.H. Ee, H.K. Lee, J. Chromatogr. A 1082 (2005) 121. [11] L.Z. Yu, M.J.M. Wells, J. Chromatogr. A 1143 (2007) 16. [12] C.W. Thorstensen, A. Christiansen, J. Agric. Food Chem. 49 (2001) 4199. [13] R.B. Geerdink, A. Kooistra-Sijpersma, J. Tiesnitsch, P.G.M. Kienhuis, U.A.T. Brinkman, J. Chromatogr. A 863 (1999) 147. [14] M. Biesaga, A. Jankowska, K. Pyrzynˇıska, Microchim. Acta 150 (2005) 317. [15] A. Lagana, A. Bacaloni, I. De Leva, A. Faberi, G. Fago, A. Marino, Anal. Chim. Acta 462 (2002) 187. [16] A. Farran, S. Ruiz, J. Chromatogr. A 1024 (2004) 267. [17] N. Rosales-Conrado, M.E. Leon-Gonzalez, L.V. Perez-Arribas, L.M. Polo-Diez, J. Chromatogr. A 1076 (2005) 202. [18] C. Baggiani, C. Giovannoli, L. Anfossi, C. Tozzi, J. Chromatogr. A 938 (2001) 35. [19] H. Zhang, T. Song, W. Zhang, W. Hua, C. Pan, Bioorg. Med. Chem. 15 (2007) 6089. [20] J.B. Quintana, R. Rodil, S. Muniategui-Lorenzo, P. Lopez-Mahia, D. Prada-Rodriguez, J. Chromatogr. A 1174 (2007) 27. [21] J.-F. Liu, L. Torang, P. Mayer, J.A. Jonsson, J. Chromatogr. A 1160 (2007) 56. [22] F.M. Lanc¸as, S.R. Rissato, M.S. Galhiane, Chromatographia 50 (1999) 35. [23] I. Rodriguez, E. Rubi, R. Gonzalez, J.B. Quintana, R. Cela, Anal. Chim. Acta 537 (2005) 259. [24] J.F. Garcia-Reyes, M.D. Hernando, C. Ferrer, A. Molina-Diaz, A.R. Fernandez-Alba, Anal. Chem. 79 (2007) 7308. [25] C. Diez, W.A. Traag, P. Zommer, P. Marinero, J. Atienza, J. Chromatogr. A 1131 (2006) 11. ˇ [26] S.J. Lehotay, M. Anastassiades, D. Stajnbaher, F.J. Schenck, J. AOAC Int. 86 (2003) 412. [27] K. Banerjee, D.P. Oulkar, S. Dasgupta, S.B. Patil, S.H. Patil, R. Savant, P.G. Adsule, J. Chromatogr. A 1173 (2007) 98. [28] C. Lesueur, P. Knittl, M. Gartner, A. Mentler, M. Fuerhacker, Food Control 19 (2008) 906.

20

a n a l y t i c a c h i m i c a a c t a 6 2 6 ( 2 0 0 8 ) 10–20

[29] C. Lesueur, M. Gartner, A. Mentler, M. Fuerhacker, Talanta 75 (2008) 284. [30] S. Wang, Y. Xu, C. Pan, S. Jiang, F. Liu, Anal. Bioanal. Chem. 387 (2007) 673. [31] P. Payá, M. Anastassiades, D. Mack, I. Sigalova, B. Tasdelen, J. Oliva, A. Barba, Anal. Bioanal. Chem. 389 (2007) 1697. [32] D.A. Lambropoulou, T.A. Albanis, Anal. Bioanal. Chem. 389 (2007) 1663. [33] A. Hercegova, M. Domotorova, E. Matisova, J. Chromatogr. A 1153 (2007) 54. [34] T. Cajka, J. Hajslova, O. Lacina, K. Mastovska, S.J. Lehotay, J. Chromatogr. A 1186 (2008) 281.

[35] R. E. Majors, LCGCeurope website 2007. http://www.lcgceurope.com (accessed January 2008). [36] M. Miura, Y. Terashita, K. Funazo, M. Tanaka, J. Chromatogr. A 846 (1999) 359. [37] Document no. SANCO/2007/3131, Method Validation and Quality Control Procedures for Pesticide Residues Analysis in Food and Feed, 31/October/2007, http://ec.europa.eu/food/plant/resources/qualcontrol en.pdf. (accessed December 2007). [38] M. Anastassiades, CVUA Stuttgart, Quechers website 2007, http://www.quechers.com (accessed November 2007). [39] M.J. Lagarda, G. Garcia-Llatas, R. Farre, J. Pharm. Biomed. Anal. 41 (2006) 1486.