Direct chiral liquid chromatography determination of aryloxyphenoxypropionic herbicides in soil: deconvolution tools for peak processing

ORIGINAL PAPER

Direct chiral liquid chromatography determinationof aryloxyphenoxypropionic herbicides in soil:deconvolution tools for peak processing

V. Guillén-Casla & J. Magro-Moral &N. Rosales-Conrado & L. V. Pérez-Arribas &

M. E. León-González & L. M. Polo-Díez

Received: 2 February 2011 /Revised: 23 March 2011 /Accepted: 28 March 2011 /Published online: 13 April 2011# Springer-Verlag 2011

Abstract In this paper, the enantiomeric separation of twoaryloxyphenoxypropionic esters (fluazifop-butyl andquizalofop-ethyl) and a safener herbicide (mefenpyr-diethyl), which is widely used for protecting crop plants,has been studied by direct liquid chromatography (LC) withUV detection on an α1-acid glycoprotein as chiral station-ary phase. Optimization of separation conditions was doneby factorial experimental design. Experimental factors andranges selected were propanol (5–10%), phosphate bufferpH (6.5–7.0), and column temperature (15–25 °C).Responses were expressed in terms of enantioresolution(Rs) and adjusted retention time of the second elutedenantiomer (tr2′). The chemometric method used to exploredata was response surface analysis. Multiple responseanalyses were carried out to determine the combination ofexperimental factors which simultaneously optimize exper-imental responses. Under optimum conditions for enantio-separation of each herbicide, partially overlapped or fullyresolved enantiomers were obtained. Deconvolution toolswere employed as an integration method to fit chromato-graphic data and to achieve a more precise enantiomericratio (ER) and enantiomeric fraction (EF) values. Applica-bility of both direct chiral LC and peak deconvolutionmethods was evaluated in spiked soil samples at different

R/S enantiomeric ratios. Acceptable and reproduciblerecoveries between 71% and 96% with precision in therange 1–6% were achieved for herbicide-spiked levels from0.50 to 9.0 μg g–1. In addition, parameters such as Rs, ER,and EF were calculated and compared with values obtainedusing the common valley drop integration method.

Keywords HPLC . Chiral analysis . Pesticides .

Chemometrics . Peak deconvolution . Soil

Introduction

Aryloxyphenoxypropionic acids (ArPPs) are a new class ofherbicides widely used and commercialized as esters for theselective removal of most grass species from any non-grasscrop. These compounds are more active in post-emergence(foliar) applications [1], and they are often applied at thelow levels of liters per hectare together with safenerherbicides, which selectively protect crop plants fromherbicide damage without reducing activity in target weedspecies [2]. Some of these herbicides present chirality, andgenerally, the R-(+) enantiomer shows great herbicidalactivity and produces the desired effect [1]. Therefore, inorder to prevent unnecessary enantiomer use causingadverse environmental impact, only the active isomershould be employed [3, 4].

At present, direct high-performance liquid chromatogra-phy (HPLC) based on chiral stationary phases (CSPs) is apreferred method for herbicide chiral separations. Among awide variety of natural and synthetic CSPs, phenylcarba-mates of cellulose and amylose have been mainly employedin normal phase mode for enantiomeric analysis ofdiclofop-methyl and diclofop acid [5–9], quizalofop-ethyl[5], and fenoxaprop-ethyl [10]. Separation of racemic

Electronic supplementary material The online version of this article(doi:10.1007/s00216-011-4969-4) contains supplementary material,which is available to authorized users.

V. Guillén-Casla : J. Magro-Moral :N. Rosales-Conrado (*) :L. V. Pérez-Arribas :M. E. León-González : L. M. Polo-DíezAnalytical Chemistry Department, Faculty of Chemistry,Complutense University of Madrid,Avda. Complutense s/n,28040 Madrid, Spaine-mail: [email protected]

Anal Bioanal Chem (2011) 400:3547–3560DOI 10.1007/s00216-011-4969-4

http://dx.doi.org/10.1007/s00216-011-4969-4

diclofop, fluazifop, quizalofop, and fenoxaprop has beenachieved using hexane–isopropanol mixtures as mobilephase with a Pirkle-type CSP [11]. Under normal phasemode, vancomycin has also been used for the enantiosepa-ration of haloxyfop-methyl and fenoxaprop-ethyl [12] andtert-butylbenzoylated tartardiamide chiral stationary phasefor the enantioresolution of both fenoxaprop and diclofop[13]. In some cases, these normal phase LC methods haveprovided low resolutions [5], very low retention factors [5,11], or long analysis times to achieve acceptable resolution[8, 14, 15]. Reversed phase LC methods are very scarce,but some CSPs such as permethylated β-cyclodextrin andterguride have been employed for the enantioseparation ofdiclofop-methyl and fluazifop-butyl [14], and fluazifop,haloxyfop, and quizalofop [15], respectively. Si Ahmed etal. [16] have prepared a based silica-bonded phase ofhydroxypropyl-β-cyclodextrin for the enantioseparation offluazifop-butyl and haloxyfop-ethoxyethyl in reversedphase mode, achieving baseline resolution. However, someof these methods required analysis times longer than45 min to obtain acceptable resolution or the use of non-commercial chiral columns. Enantiomers of quizalofop-ethyl and diclofop-methyl have been separated with lowresolution on cellulose tris-3,5-dimethyl carbamate station-ary phase with methanol–water or acetonitrile–water as themobile phase [17]. Recently, the enantiomeric separation ofdiclofop and diclofop-methyl has been achieved withacceptable resolution using a protein-based α-acid glyco-protein (AGP) stationary phase [18].

Assuming the complex mechanisms involved in chiralinteractions, a stationary phase can exhibit differences inselectivity for compounds having similar structure, andconsequently, it is necessary to look for different experi-mental conditions to separate the enantiomers that belong tothe same chemical family. In chiral reversed phase LC,separations are accomplished by controlling the amount oforganic modifier, pH of aqueous buffer [18], and columntemperature [13], which could affect the chiral recognitionmechanism and produce changes in the enantiomericretention order or even induce conformational changes inthe protein-based CSPs [19]. Optimization of the chro-matographic conditions requires taking into account allparameters adjustable as a whole, and therefore, multivar-iate approaches and methodologies such as experimentalresponse surface analysis are commonly employed [20].

On the other hand, although it could be relatively easy tofind a chiral selector that enables some degree ofenantioseparation for the compound of interest, totalseparation is often difficult to get within reasonable analysistimes. In fact, the literature shows numerous chiral methodsin which enantiomers are partially overlapped. In thesecircumstances, the analytical responses measured throughmanual integration techniques such as the valley drop

method (VDM) may be applicable for certain purposes, butfor quantification [21] or obtaining enantiomeric fractions[22], peak deconvolution could be an effective strategy forthe mathematical treatment of the chromatographic signal,even when the peaks are fully resolved. In addition,deconvolution software is commercially available andoccasionally; it is included in the chromatograph softwarepackages. Several functions have been used to describechromatographic peaks: the Gaussian distribution function(GD), the asymmetric GD, the generalized exponentialfunction, the Lorentzian function, the polynomial modifiedGD, and the exponentially modified Gaussian (EMG)function [23], the last one being the most popular model[24, 25]. Thus, peak deconvolution has been successfullyapplied in GC × GC analysis [26] and in the determinationof pesticide residues in complex samples, where completeresolution of all compounds is rarely achieved even usingan optimum selectivity and extremely high performance ofthe column [27, 28]. In the field of chiral LC, deconvolu-tion has been rarely applied and only for the investigationof asymmetric peaks, providing good results in cases inwhich enantiomeric peaks are not symmetrical or do nothave approximately equal areas [22].

In this paper, the chiral recognition ability of the α1-acidglycoprotein CSP for the enantiomeric separation of threeracemic herbicides (fluazifop-butyl, quizalofop-ethyl, andmefenpyr-diethyl) by direct LC with UV detection underreversed phase conditions has been investigated. As far aswe know, reversed phase LC methods are not commonlyused for herbicide chiral separation, and in the case ofsafener herbicides, no methods for their enantiomericseparation are available. In this work, the optimumchromatographic conditions have been obtained by exper-imental designs, which provided equations able to evaluatethe performance of chiral separations. Mathematical decon-volution has also been studied to determine peak areas andto calculate Rs, enantiomeric ratio (ER), and enantiomericfraction (EF). It can be a useful integration tool to obtainmore precise results when enantiomers are fully or partiallyresolved. The applicability of the developed chiral methodhas been evaluated analyzing spiked soil samples.

Experimental

Chemicals and standards

All reagents and solvents were of analytical grade, andpurified water from a Milli-Q system was used (Millipore,Bedford, MA, USA). Methanol, 2-propanol, acetonitrile,acetone, dichloromethane, ethyl acetate, and hexane ofgradient HPLC grade (Scharlau, Barcelona, Spain), sodiumdihydrogen phosphate, sodium hydroxide, and hydrochloric

3548 V. Guillén-Casla et al.

acid (Panreac, Barcelona, Spain) were employed. Glacialacetic acid (99.5%) was from Carlo Erba (Milan, Italy) andammonium hydroxide (33%) from Riedel de Häen (Seelze,Germany).

Herbicide standards were used as received from commer-cial sources. (R,S)-ethyl 2-[4-(6-chloroquinoxalin-2-yl)oxyphenoxy]propanoate (quizalofop-ethyl, CAS no. 76578-14-8, 96.4% pure) and (R,S)-diethyl 1-(2,4-dichlorophenyl)-5-methyl-4H-pyrazole-3,5-dicarboxylate (mefenpyr-diethyl,CAS no. 135590-91-9, 99.7% pure) were supplied by Riedel

de Haën (Sigma-Aldrich, Seelze, Germany). (R,S)-butyl 2-[4-[5-(trifluoromethyl)pyridin-2-yl]oxyphenoxy]propanoate(fluazifop-butyl, CAS no. 69806-50-4, 98.8% pure); ethyl(R)-2-[4-(6-chloro-2-quinoxalyloxy)phenoxy]propionate(quizalofop-p-ethyl, CAS no. 100646-51-3, 98.2% pure);and butyl (R)-2-[4-(5-trifluoromethyl-2-pyridyloxy)phenoxy]propionate (fuazifop-p-butyl, CAS no. 79241-46-6, 96.1%pure) were provided by Fluka (Sigma-Aldrich, St. Louis,MO, USA). The chemical structures of herbicides are shownin Fig. 1.

Quizalofop-ethyl

Fluazifop-butyl

Mefenpyr-diethyl

O(R)

O CH3

O

H3C H

N

NCl

O

O(S)

O CH3

O

H3C H

N

NCl

O

N O

F

F

F

O(R)

O CH3

O

H3C H

N O

F

F

F

O(S)

O CH3

O

H3C H

Cl

Cl

N

N

(R)

O

O

CH3

H3C

O CH3

O

Cl

Cl

N

N

(S)

O

O

CH3

H3C

O CH3

O

Fig. 1 Chemical structures ofmefenpyr and studied ArPPsherbicides

Direct chiral LC determination of ArPP herbicides in soil 3549

Analyte stock solutions (200 mg L1) were prepared inmethanol and stored in the dark at 4 °C for 1 monthmaximum. Fresh working standard solutions were dailyprepared in mobile phase by suitable dilution of stocksolutions as required.

Soil samples

A sandy loam soil (4% organic matter, pH 7.9) was freshlycollected from the region of Madrid. This soil, which had notreceived any pesticide application for at least 5 years, wastaken from a surface layer of 3-cm depth in a 20×30-cm areawithout vegetation cover (40 kg m2).

Equipment

Experiments were carried out in a HPLC chromatographpurchased from Jasco (Tokyo, Japan) that included a four-channel gradient pump PV-2089 Plus Solvent Module and aprogrammable variable wavelength UV-2075 Plus detector.For data acquisition and processing, all components wereinterfaced to a PC data station equipped with BorwinWorkstation software (version 1.50) for Windows.

An external stainless steel loop with a volume of 20 μLwas placed into a manual Rheodyne® injection valve (Cotati,CA, USA). Chromatographic separation of the herbicideenantiomers was carried out in a α1-acid glycoprotein Chiral-AGP™ column (100×3.0 mm, 5 μm) supplied by Chrom-Tech (Cheshire, U.K), thermostated in a Jasco CO-2067column oven. Mobile phases were also thermostated in a P-Selecta Presciterm bath (Barcelona, Spain).

Extraction of the herbicides from soil was carried out inan ultrasound bath provided by P-Selecta, a vortex mixerfrom VELP Scientifica (Usmate, Italy), and a mechanicalshaker model Vibromatic from P-Selecta. A Unicencentrifuge model 21 supplied by Ortoalresa (Madrid, Spain)was used for centrifugation of the soil extracts. Solventevaporation under nitrogen stream was carried out bymeans of a VacElut 20-place vacuum manifold suppliedby Varian (Harbor City, CA, USA).

Procedure

Enantiomeric determination of herbicides by LC-UV

Direct chiral separation of herbicides was performed byisocratic elution using the following mobile phase compo-sitions and temperatures: 2-propanol/10 mM phosphatebuffer, pH 7.0 (7:93, v/v) at 17 °C for fluazifop-butyl;2-propanol/10 mM phosphate buffer, pH 7.0 (5:95, v/v) at20 °C for quizalofop-ethyl; and 2-propanol/10 mM phos-phate buffer, pH 6.5 (9:91, v/v) at 15 °C for mefenpyr-diethyl. Flow rate was set to 0.80 mL min−1 and UV

detection was done at 230 nm for fluazifop-butyl andquizalofop-ethyl, and at 307 nm for mefenpyr-diethyl.

The individual optimization study of chiral resolutionfor each herbicide was done using factorial experimentaldesign. To obtain the response function, three factorsincluding 2-propanol percentage on the mobile phase(5–10%), pH of 10 mM phosphate buffer (6.5–7.0), andcolumn temperature (15–25 °C) were considered, andtwo (pH) or three levels (2-propanol and temperature)were studied for each selected factor. All the experi-ments were carried out using racemic standard solutionsof 5 μg mL−1, prepared in the mobile phase. Mobilephases were filtered through a nylon membrane filter(0.45 μm), sonicated, and degassed under vacuum for5 min before use.

Responses studied were expressed as enantioresolution(Rs) and adjusted retention time of the second elutedenantiomer (tr2′). Chromatographic analyses of racemicherbicides were carried out according to each designedexperiment (Table 1). Data analysis was performed byemploying the software package Statgraphics Plus version5.1 (Statistical Graphics Corp., Rockville, MD, USA).

Integration of partially resolved enantiomers was doneby the VDM using standard chromatographic software. Inthis technique, which is commonly used for integratingpartially resolved chromatographic peaks, a vertical line ismanually dropped from the valley between the two peaks tobaseline.

For peak deconvolution, chromatographic data weredirectly fitted into an EMG function using PeakFit version4.12 (SeaSolve Inc., San Jose, CA, USA) commercialsoftware. Previously, experimental chromatograms wereconverted to ASCII (.txt) format and were sectioned toexclude unnecessary parts. Fitting was iterated until the r2

yielded a value higher or equal to 0.970.

Analytical characteristics from herbicide standardsolutions

Detection (LOD) and quantitation (LOQ) limits, linearityrange, and precision were established under optimumexperimental conditions (Enantiomeric determination ofherbicides by LC-UV).

LODs and LOQs were defined as each enantiomer’s signalto baseline noise peak ratio at a height of 3 and 10,respectively. They were estimated experimentally by analysisof a series of decreasing concentrations of herbicide solutions.

Linearity was determined at nine concentration levelsin the range 0.250–40 μg mL−1 for quizalofop-ethyl,0.125–40 μg mL−1 for mefenpyr-diethyl, and 0.625–40 μg mL−1 fluazifop-butyl enantiomers. Results wereanalyzed by linear regression using areas obtained by peakdeconvolution.


The intra-day precision was evaluated by injecting12.0 μg mL−1 of each racemic herbicide standard solutionfive consecutive times on the same day (n=5). The inter-day variation was similarly conducted during three succes-sive days (N=15). Relative standard deviation (RSD, %)was taken as a measure of precision and calculated for Rs

and tr2′ of the second eluted enantiomer.

Analysis of soil samples

Prior to laboratory experiments, soil was ground in amarble mortar, air-dried for about 24 h at room temperatureto the point of soil moisture suitable for sieving to amaximum particle size of <2 mm, and mixed manually toachieve homogeneity. Then, soil was stored at roomtemperature until analysis.

Soil samples (1 g) were spiked with racemic mefenpyr-diethyl in the concentration range 0.50–2.5 μg g−1.Fluazifop-butyl and quizalofop-ethyl were added at differ-ent R/S ratios to soil samples in the range 2.5–9.0 and 1.0–9.0 μg g−1, respectively. These spiked levels were selectedtaking into account their usual application doses and theexpected residue levels in soil. Samples were left tostabilize for 30 min in dark conditions at room temperatureto allow solvent evaporation and sample equilibration. Theywere prepared by triplicate at each spiked level andextracted three times at room temperature by sonication

during 5 min each. Extractions were carried out with 2 mLof acetone for quizalofop-ethyl and 2 mL of acetonitrileacidified with 1% acetic acid for both mefenpyr-diethyl andfluazifop-butyl. After each extraction step, liquid extractswere separated from solid soils by centrifugation at4,200 rpm for 10 min and collected in a centrifuge beaker.Final extracts were evaporated to dryness in a vacuummanifold under a nitrogen stream and reconstituted in0.5 mL of the mobile phase. A total of 20 μL of theresulting solution was injected in the LC system andanalyzed under conditions described in Section 2.3.2 toestablish extraction recoveries at the three different spikedlevels.

Analytical characteristics from herbicide-spiked sampleswere similarly determined as explained above for standardsolutions. Calibration (n=5) was performed with a blanksoil spiked with increasing amounts of the racemicherbicides in order to reach concentration levels for eachenantiomer ranging from 0.20 to 1.5 μg g−1 for mefenpyr-diethyl and from 2.0 to 7.5 and 0.40 to 2.5 μg g−1 forfluazifop-butyl and quizalofop-ethyl, respectively. LODsand LOQs were estimated analyzing spiked blank soilsamples at low concentrations. The precision of the soilextraction method was tested by recovery experiments withfortified samples. Precision was expressed as the repeat-ability (RSD, %) of the extraction recovery determinationsat the studied spiked levels.

Experiment Factors

2-Propanol (%) Buffer pH Column temperature (°C)

1 10.0 7.0 15

2 10.0 7.0 20

3 5.0 6.5 15

4 7.5 6.5 15

5 5.0 6.5 25

6 5.0 7.0 25

7 10.0 6.5 20

8 10.0 7.0 25

9 10.0 6.5 15

10 7.5 7.0 15

11 7.5 6.5 25

12 10.0 6.5 25

13 7.5 7.0 20

14 5.0 7.0 15

15 5.0 7.0 20

16 5.0 6.5 20

17 7.5 6.5 20

18 7.5 7.0 25

19 7.5 7.0 20

20 7.5 6.5 20

Table 1 Plan of experiments forfactorial design


Results and discussion

Optimization of the chiral LC method

Experimental design

Concentration of organic modifier, buffer pH, and temper-ature are chromatographic parameters having importanteffects on the enantiomeric separation in AGP stationaryphases [18]. Consequently, 2-propanol percentage (5–10%),phosphate buffer pH (6.5 and 7.0), and column temperature(15–25 °C) were also assumed the most important factorsfor the optimization of the chiral separation of fluazifop-butyl, quizalofop-ethyl, and mefenpyr-diethyl on the α1-acid glycoprotein CSP. Due to the target analytes being inthe ester form, buffer pH mainly affects the stationaryphase. Based on specifications of the AGP chiral column,and considering that pH values lower than 6.5 decreaseretention times and resolution, only two pH levels aroundthe optimum one for the AGP stationary phase wereconsidered.

Factorial experimental designs for each racemic herbi-cide were carried out as described in “Enantiomericdetermination of herbicides by LC-UV.” To evaluate theseparation quality, Rs and the adjusted retention time of thesecond eluted enantiomer (tr2′) were selected as experimen-tal responses. They were calculated using the VDM

integration method, and the obtained results are includedin Table 2. Responses (y) were correlated with theexperimental factors (x1, 2-propanol ratio; x2, buffer pH;x3, column temperature) by means of a second-orderpolynomial equation:

y ¼ a0 þ a1x1 þ a2x2 þ a3x3 þ a4x1x2 þ a5x1x3

þ a6x2x3 þ a7x21 þ a8x

23 ð1Þ

Table 3 shows the parameter values that fit experimentaldata into Eq. 1 for each herbicide and the statistic p values.As can also be seen in this table, mathematical models withacceptable correlation coefficients (R2) higher than 0.902and good standard errors of estimation (SEE) in the range0.005–0.484 were achieved, showing the reliability of theequations.

Equation parameters with p values lower than 0.05indicate factors that affect significantly the experimentalresponses. In the studied domain and for all racemicherbicides, 2-propanol percentage affected significantly Rs

and tr2′, which decreased with the increase in the propanolratio. In addition, and for all herbicides, temperature mainlyaffected the tr2′ response, which decreased at the highertemperatures studied. Regarding buffer pH, it affected bothexperimental responses of quizalofop-ethyl. The quadraticterms were less important for all the studied analytes, and

Experiment Fluazifop-butyl Quizalofop-ethyl Mefenpyr-diethyl

Rs tr2′ (min) Rs tr2′ (min) Rs tr2′ (min)

1 0.67 13.15 0.41 9.43 0.89 7.05

2 0.60 11.40 0.36 8.44 0.89 6.35

3 1.49 40.03 0.60 24.49 1.87 21.76

4 1.18 22.93 0.44 15.44 1.95 12.04

5 1.17 30.88 0.67 19.93 1.89 18.31

6 1.23 28.63 0.74 18.17 1.91 17.90

7 0.64 10.82 0.30 8.03 0.86 6.03

8 0.54 10.21 0.36 7.62 0.80 5.77

9 0.70 12.07 0.30 8.83 0.87 6.50

10 1.04 20.23 0.54 13.40 1.32 11.13

11 0.78 16.36 0.46 11.61 1.24 9.39

12 0.54 9.55 0.36 7.20 0.87 5.41

13 1.03 17.93 0.57 12.15 1.30 10.41

14 1.57 37.73 0.70 22.93 1.85 21.31

15 1.19 32.38 0.75 20.59 2.10 19.44

16 1.36 34.22 0.73 21.38 1.97 19.73

17 0.93 18.82 0.56 13.35 1.19 10.50

18 0.81 16.02 0.56 11.17 1.24 9.48

19 0.93 18.10 0.55 12.47 1.36 10.31

20 0.85 18.48 0.43 12.86 1.22 10.35

Table 2 Experimental responsesobtained from the modeling setfor racemic herbicides


only in some cases did the interaction between factors havea significant influence.

In order to easily visualize the most important effects onRs and tr2′, estimated response surfaces were obtained foreach herbicide through their respective equations. Thesethree-dimensional graphs were in agreement with dataincluded in Table 3, and they showed more or less distortedplanes with some factor interactions of less importance. Asan example, response surfaces obtained for quizalofop-ethylcan be observed in Fig. 2.

To get the optimum chromatographic conditions for thechiral separation of each herbicide and taking into account theobtained experimental responses, multiple response analyseswere carried out. The optimization criterion was Rs≥1.0 andminimum tr2′. This chemometric procedure allowed deter-mining the combination of experimental factors whichsimultaneously optimize the two studied responses and,therefore, that maximize the desirability function over theselected region. As a compromise, the following mobilephase compositions and temperatures were selected asoptimum for direct chiral LC: 2-propanol/10 mM phosphatebuffer, pH 7.0 (7:93, v/v) at 17 °C for fluazifop-butyl;2-propanol/10 mM phosphate buffer, pH 7.0 (5:95, v/v) at20 °C for quizalofop-ethyl; and 2-propanol/10 mMphosphate buffer, pH 6.5 (9:91, v/v) at 15 °C for mefenpyr-diethyl. Under these conditions, the estimated responsesurfaces predicted acceptable Rs values of 0.75, 1.08, and1.21 for quizalofop, fluazifop, and mefenpyr esters, respec-tively, and tr2′ around 8.2 min (mefenpyr), 20 min

2-propanol (%)

0,3-0,40,4-0,50,5-0,60,6-0,70,7-0,8

-1 0 1-1

01

0,36

0,46

0,56

0,66

0,76

2-propanol (%)

7,0-10,010,0-13,013,0-16,016,0-19,019,0-22,022,0-25,0

-1 0 1-1

01

7

11

15

19

23

RS

T (oC)

T (oC)

tr'2

(a)

(b)

Fig. 2 Estimated response surfaces (normalized) of quizalofop-ethylobtained for enantiomeric resolution (a) and adjusted retention time ofthe second eluted enantiomer (b) using the two variables with thehighest influence at pH 7.0T

able

3Fitted

parameter

values

obtained

from

thequ

adratic

mathematical

mod

eldescribedin

Eq.

1

Herbicide

(racem

icmixture)

a 0a 1

a 2a 3

a 4a 5

a 6a 7

a 8R2

SEE

Fluazifop-butyl

Rs

0.9282

−0.3600

−0.0015

−0.1367

−0.0033

0.0463

0.0150

0.0261

0.0311

0.970

0.071

(0.0001)

(0.9264)

(0.0001)

(0.8737)

(0.0923)

(0.4792)

(0.4440)

(0.3644)

t r2′

18.32

−11.389

−0.4179

−2.874

0.7257

1.600

0.1661

3.888

0.5668

0.999

0.484

(0.0001)

(0.0026)

(0.0001)

(0.0003)

(0.0001)

(0.2596)

(0.0001)

(0.0280)

Quizalofop-ethyl

Rs

0.5246

−0.1750

0.0345

0.0133

−0.0017

−0.0125

−0.0117

0.0132

−0.0218

0.956

0.041

(0.0001)

(0.0031)

(0.2838)

(0.8905)

(0.4068)

(0.3454)

(0.5006)

(0.2753)

t r2′

12.70

−6.494

−0.3380

−1.570

0.4626

0.7343

0.1022

1.913

0.2069

0.997

0.375

(0.0001)

(0.0020)

(0.0001)

(0.0013)

(0.0002)

(0.3660)

(0.0001)

(0.2590)

Mefenpyr-diethyl

Rs

1.341

−0.5342

−0.0135

−0.0667

−0.0125

−0.0021

0.0483

0.0411

0.0236

0.902

0.187

(0.0001)

(0.7530)

(0.2428)

(0.8212)

(0.7541)

(0.3900)

(0.6446)

(0.7905)

t r2′

10.39

−6.779

−0.0436

−1.127

0.1987

0.5606

0.0715

2.491

0.1149

0.999

0.196

(0.0001)

(0.3413)

(0.0001)

(0.0049)

(0.0001)

(0.2325)

(0.0001)

(0.2315)

Valuesof

paregivenin

parentheses(p≤0

.05indicatin

gtheeffectssignificantly

differentfrom

zero

atthe95%

confidence

levelarein

italics)

SEEstandard

errorof

estim

ation


(quizalofop), and 21 min (fluazifop). These expected valueswere experimentally assessed and results were in agreementwith those predicted, achieving RSD from the predictedvalues in the range 0.80–4.2% and 0.30–3.8% for Rs and tr2′,respectively. In addition, the variance analysis of one factorshowed that there were no significant differences betweenthe experimental results and those predicted by Eq. 1 with aconfidence level of 95%.

Peak deconvolution

Herbicide enantioresolution on the AGP chiral column wasestimated using the VDM as integration tool, and values inthe range 0.73 for quizalofop-ethyl and 1.2 for mefenpyr-diethyl racemates were achieved under optimum conditionsestablished thorough experimental design methodologies.To obtain more precise peak areas especially for partiallyoverlapped enantiomers, an integration method based onmathematical deconvolution was applied. Thus, chromato-graphic data were fitted into an EMG function available inthe commercial software, which contains four independentparameters (“Enantiomeric determination of herbicides byLC-UV”):

y ¼ b02b3

expb222b23

þ b1 � x

b3

� �

� efrx� b1ffiffiffiffiffiffiffi2b2

p � b2ffiffiffiffiffiffiffiffiffiffiffiffiffi2� b3

p� �

þ b3b3j j

� �ð2Þ

where x and y represent the time and UV response,respectively, b0 is the peak area, b1 the peak center, b2 thepeak width, and b3 the distortion parameter. This function isthe convolution of a primary Gaussian with a first-orderdecay, and it was chosen as the mathematical model forfitting asymmetrical and width chromatographic peaks,which are typical in chiral LC.

This function mathematically described the chromato-graphic peaks well, producing acceptable fits of thechromatographic data with r2>0.970 for all chromatogramsand randomly distributed residuals. To evaluate the reli-ability of the peak modeling, deconvolved peaks wereintegrated and employed to obtain Rs, ER (Eq. 3) and EF(Eq. 4).

ER ¼ A1

A2ð3Þ

EF ¼ A1

A1 þ A2ð Þ ð4Þ

where A1 and A2 correspond to the peak areas of the firstand second eluted enantiomers on the chiral column,respectively.

Both ER and EF are standard descriptors of enantiomeric(chiral) signatures, although some authors prefer the secondone when performing environmental calculations. In thecase of racemic standards, the values of ER and EF are 1.0and 0.5, respectively [29]. Table 4 shows a comparisonbetween experimental values of Rs, ER, and EF, calculatedfrom chromatograms obtained under the conditions de-scribed in Table 1 and the corresponding values obtainedfor the same chromatograms after deconvolution. Similar Rs

values were obtained when VDM (experimental chromato-grams) and deconvolution (mathematical model) wereapplied. In fact, the data obtained after deconvolution werefitted into Eq. 1 to model Rs, and the predicted values underthe optimum chromatographic conditions were similar tothose previously predicted after modeling of the experi-mental data (“Experimental design”). In addition, theequations obtained for Rs showed good R2 and SEE values,which are in agreement with those included in Table 3. Incontrast, experimental ER and EF values obtained prior andafter peak deconvolution were substantially different fromthat corresponding to racemates, especially for fluazifop-butyl and quizalofop-ethyl. In the case of fluazifop-butyl,experimental ER and EF values around 1.0 and 0.5,respectively, were only obtained when experimental Rs

was higher than 1.3 (experiments 13, 14, and 16). Amongthe three studied chiral herbicides, the highest enantioreso-lution was obtained for mefenpyr-diethyl, and probably forthis reason, the mean values of ER and EF estimated fromthe experimental and deconvolved chromatograms weremore similar, particularly the EF ratios. In fact, when a two-sample t test for mean comparisons of ER and EF wasapplied, significant differences between VDM and thedeconvolution method with a confidence level of 95%were observed. Only in the case of mefenpyr-diethyl, andparticularly for EF values, were no significant differencesobserved. Therefore, deconvolution seems to be moreappropriate that the VDM as integration tool specially forthe calculation of ER and EF from racemic mixtures.

Figure 3 shows the experimental and deconvolvedchromatograms obtained for each racemic herbicide underselected chromatographic conditions for chiral separation inwhich partial and resolved enantiomers can be observed.

Performance of the chiral LC method

Analytical characteristics from racemic herbicide standards

Evaluation of the analytical curve linearity (n=8) was donebased on analyses of the racemic herbicide solutionsprepared in the mobile phase under optimum conditions.The concentration ranges investigated were 0.125–40 μg mL−1 for mefenpyr-diethyl, 0.250–40 μg mL−1 forquizalofop-ethyl, and 0.625–40 μg mL−1 for fluazifop-


Table 4 Comparison between experimental and deconvolution values obtained from the modeling set

Experiment Fluazifop-butyl Quizalofop-ethyl Mefenpyr-diethyl

Rs ER EF Rs ER EF Rs ER EF

1 0.67 (0.68) 0.48 (1.02) 0.32 (0.51) 0.41 (0.46) 0.27 (0.91) 0.21 (0.48) 0.89 (1.12) 0.70 (0.88) 0.41 (0.47)

2 0.60 (0.67) 0.46 (1.12) 0.31 (0.53) 0.36 (0.43) 0.26 (0.87) 0.21 (0.46) 0.89 (1.10) 0.73 (0.87) 0.42 (0.47)

3 1.49 (1.93) 1.03 (1.08) 0.51 (0.52) 0.60 (0.66) 0.47 (1.06) 0.32 (0.51) 1.87 (2.20) 1.01 (0.98) 0.50 (0.50)

4 1.18 (1.41) 0.91 (0.90) 0.48 (0.47) 0.44 (0.41) 0.28 (1.16) 0.22 (0.54) 1.95 (1.51) 0.09 (1.00) 0.08 (0.50)

5 1.17 (1.35) 0.81 (1.07) 0.45 (0.52) 0.67 (0.65) 0.50 (1.18) 0.33 (0.54) 1.89 (2.24) 1.04 (1.01) 0.51 (0.50)

6 1.23 (1.51) 0.84 (0.98) 0.46 (0.50) 0.74 (0.74) 0.56 (0.87) 0.36 (0.47) 1.91 (2.23) 1.01 (1.01) 0.50 (0.50)

7 0.64 (0.65) 0.45 (1.07) 0.31 (0.52) 0.30 (0.33) 0.28 (1.16) 0.40 (0.54) 0.86 (1.03) 0.67 (0.84) 0.40 (0.46)

8 0.54 (0.53) 0.35 (1.10) 0.26 (0.52) 0.36 (0.39) 0.21 (1.00) 0.17 (0.50) 0.80 (1.05) 0.60 (0.81) 0.38 (0.45)

9 0.70 (0.77) 0.52 (1.05) 0.34 (0.51) 0.30 (0.33) 0.37 (1.00) 0.38 (0.50) 0.87 (1.07) 0.69 (0.84) 0.41 (0.46)

10 1.04 (1.15) 0.80 (0.99) 0.44 (0.50) 0.54 (0.56) 0.42 (1.17) 0.30 (0.54) 1.32 (1.58) 0.92 (0.98) 0.48 (0.50)

11 0.78 (0.88) 0.62 (1.01) 0.38 (0.50) 0.46 (0.44) 0.30 (1.05) 0.23 (0.51) 1.24 (1.51) 0.85 (0.95) 0.46 (0.49)

12 0.54 (0.55) 0.37 (1.13) 0.27 (0.53) 0.36 (0.35) 0.21 (1.18) 0.17 (0.54) 0.87 (1.03) 0.65 (0.80) 0.39 (0.44)

13 1.03 (1.15) 0.82 (1.01) 0.45 (0.50) 0.57 (0.58) 0.44 (1.14) 0.31 (0.53) 1.30 (1.51) 0.93 (0.99) 0.48 (0.50)

14 1.57 (1.78) 1.06 (1.19) 0.51 (0.54) 0.70 (0.72) 0.59 (1.07) 0.37 (0.52) 1.85 (2.22) 0.99 (0.97) 0.50 (0.49)

15 1.19 (1.73) 0.93 (1.07) 0.48 (0.52) 0.75 (0.71) 0.64 (1.06) 0.39 (0.51) 2.10 (2.48) 1.05 (1.00) 0.51 (0.50)

16 1.36 (1.43) 1.02 (1.18) 0.51 (0.54) 0.73 (0.70) 0.58 (1.18) 0.37 (0.54) 1.97 (2.34) 1.07 (1.01) 0.52 (0.50)

17 0.93 (1.07) 0.79 (1.04) 0.44 (0.51) 0.56 (0.56) 0.40 (1.10) 0.29 (0.52) 1.19 (1.48) 0.83 (0.98) 0.45 (0.50)

18 0.81 (0.94) 0.62 (0.97) 0.38 (0.49) 0.56 (0.55) 0.37 (1.02) 0.27 (0.51) 1.24 (1.46) 0.89 (0.96) 0.47 (0.49)

19 0.93 (1.05) 0.71 (1.00) 0.41 (0.50) 0.55 (0.56) 0.40 (1.03) 0.29 (0.51) 1.36 (1.55) 0.96 (0.97) 0.49 (0.49)

20 0.85 (1.01) 0.66 (0.98) 0.40 (0.50) 0.43 (0.41) 0.25 (1.14) 0.20 (0.53) 1.22 (1.46) 0.84 (0.94) 0.46 (0.48)

Mean value 0.71 (1.05) 0.41 (0.51) 0.45 (1.06) 0.30 (0.51) 0.82 (0.94) 0.45 (0.48)

CLa 0.10 (0.03) 0.08 (0.03) 0.11 (0.05) 0.04 (0.04) 0.11 (0.03) 0.09 (0.03)

a Confidence limit (95%, n=20)

Values obtained from peak deconvolution are given in parentheses. All data showed correspond to herbicide racemates

Experimental response

Fit to Eq. (2)

Deconvolved peak

0

500

1000

1500

2000

2500

0 5 10 15 20 25 30 35

0 5 10 15 20 25 30 35

AU

Time (min)

0

1000

2000

3000

4000

5000

6000

0 2 4 6 8 10 12

AU

Time (min)

0

500

1000

1500

2000

2500

3000

3500

4000

AU

Time (min)

a b

c

Fig. 3 Chiral experimental anddeconvolved chromatogramsobtained at the 12.0-μg mL-1

concentration level under opti-mum conditions for enantiose-paration and illustration of thefit of Eq. 2 to experimentalchromatograms. The examplepresented shows Rs experimen-tal=1.20, Rs predicted=1.21, ERexperimental=0.84, and ERdeconvolved=1.03 for racemicmefenpyr-diethyl (a); Rs experi-mental=1.12, Rs predicted=1.08, ER experimental=0.84,and ER deconvolved=1.03 forracemic fluazifop-butyl (b); Rs

experimental=0.73, Rs predicted=0.75, ER experimental=0.65,and ER deconvolved=1.06 forracemic quizalofop-ethyl (c)


butyl. All enantiomers showed good linearity within thetested intervals. Correlation coefficients (R2) for all peakareas were between 0.9872 and 0.9988.

Detection and quantitation limits were calculated follow-ing the procedure described in “Analytical characteristicsfrom herbicide standard solutions.” The obtained values foreach herbicide enantiomer are shown in Table 5, in which itcan be observed that LODs and LOQs were lower than orequal to 0.110 and 0.367 μg mL−1, respectively

Precision was verified for each herbicide under selectedchromatographic conditions. The intra-day and inter-dayvariability, expressed as RSD (%), was evaluated for Rs andtr2′. The results are summarized in Table 5. As can be seen, theintra-day variation (n=5) was lower than 4.8% for Rs and<4.7% for tr2′. Similar RSD values were estimated for inter-dayprecision (N=15). Good precision between different days wasalso obtained (≤5.0% or ≤5.7% for Rs and tr2′, respectively),showing the effectiveness of the proposed chiral separations.

Accuracy at different R/S ratios

Under selected chromatographic conditions for each herbi-cide (“Experimental design”), the developed LC chiralmethod was also evaluated with standard solutions contain-ing different concentrations of both (R)- and (S)-enantiomers.However, only standards of pure (R)-enantiomers offluazifop-butyl and quizalofop-ethyl were commerciallyavailable. Therefore, herbicide racemates were spiked withthe (R)-isomer in order to obtain enantiomer mixtures ofdifferent R/S compositions. By this way, we could alsodetermined that the first eluted enantiomer of fluazifop butylwas the (R)-isomer, while for quizalofop-ethyl, the firsteluted enantiomer was the (S)-isomer.

Standard solutions with different R/S ratios ranging from80:20 to 99:1 were analyzed, maintaining constant the totalinjected concentration. Similar to precision studies carried

out with racemates (“Analytical characteristics from herbi-cide standard solutions”), this concentration was fixed at12.0 μg mL−1 for both herbicides. Taking into account theLOQ determined for the (S)-enantiomer (Table 5), theminimum R/S enantiomeric ratio that allowed its quantifi-cation was set at 90:10 for fluazifop-butyl and 95:5 forquizalofop-ethyl.

Several injections were done at different R/S ratios. Theintra-day variability, expressed as RSD (%), was evaluatedfor Rs and tr2′. In addition, EF values were calculated usingboth the VDM and deconvolution integration methods. Ascan be seen in Table 6, acceptable precision with RSD valueslower than 5.3% was obtained for both herbicides. RegardingEF, the values calculated by deconvolution were moresimilar to theoretical ones, showing again its effectiveness.

Herbicide enantioselective determination in soil samples

The usefulness of the developed LC procedure combinedwith the peak deconvolution integration method wasevaluated from soil samples spiked at several concentrationlevels with herbicide racemates and mixtures with differentR/S enantiomeric ratios. Previously, an extraction procedurewas individually optimized and validated for each analyte.

Optimization of the sample preparation

Extraction efficiency studies were carried out with aliquotsof 1 g of soil for each herbicide spiked with 0.50 μg g−1 ofmefenpyr-diethyl, 1.0 μg g−1 of quizalofop-ethyl, or2.5 μg g−1 of fluazifop-butyl. In order to achieve themaximum analyte recovery, several parameters such asnature and volume of the extraction organic solvent as wellas extraction time and number of consecutive extractionswere evaluated. All experiments were conducted in triplicateand the herbicide quantification was performed following the

Table 5 Analytical characteristics obtained from racemic herbicide standard solutions

Herbicide Linearity LOD(μg mL−1)

LOQ(μg mL−1)

Intra-dayprecision, n=5(RSD, %)

Inter-dayprecision, N=15(RSD, %)

Concentration range(μg mL−1)a

R2 1st 2nd 1st 2nd Rs tr2′(min)b

Rs tr2′ (min)b

1st 2nd

Fluazifop-butyl 0.625–40 0.9988 0.9924 0.110 0.367 3.7 2.8 4.5 2.31

Quizalofop-ethyl

0.250–40 0.9950 0.9944 0.054 0.063 0.180 0.210 1.2 4.7 1.0 5.70

Mefenpyr-diethyl

0.125–40 0.9938 0.9872 0.024 0.027 0.080 0.090 4.8 0.80 5.0 0.66

1st and 2nd referred to the first and second eluted enantiomers, respectivelya For each enantiomerb Adjusted retention time of the second eluted enantiomer


procedure described in “Analysis of soil samples”. Addi-tionally, and for each tested condition, a blank of the soilaliquot was always analyzed to control co-extraction ofendogenous matter and possible interferences.

Previous experiments were carried out to select sonica-tion, vortex, or mechanical shaking as the extractiontechnique. Under the same conditions, ultrasonicationresulted in less amount of co-extracted soil organic matter,avoiding the presence of interfering compounds at the sameretention time as the target herbicides. Thus, ultrasoundmixing was selected as the most appropriate technique forthe extraction treatment.

Three consecutive extractions of 5 min each were carriedout with 2 mL of different solvents (methanol, acetonitrile,acetone, ethyl acetate, dichloromethane, and hexane). Asreported in the literature for the extraction of differentherbicide families from soils [30], mixtures of acetonitrile oracetone with 1% acetic acid or 1% ammonia were alsotested. The best recoveries were obtained using acetone forthe extraction of quizalofop-ethyl and acetonitrile acidifiedwith 1% acetic acid for both mefenpyr-diethyl and fluazifop-butyl. Therefore, it seems clear that the addition of an acidicmodifier contributes to desorption of two of the studiedherbicides, and it promotes their extraction from the soilmatrix. Although this extraction solvent is quite acidic(apparent pH 4.38), hydrolysis of the ester forms of theherbicides was not observed during the extraction procedure.

Several volumes of the extracting solvent between 2 and10 mL were studied. Volumes higher than 2 mL did notlead to significant higher recoveries, and thus, a volume of2 mL was selected for further studies. The ultrasonicationtime was evaluated at 2, 5, and 10 min. When the extractionwas made for 2 min, a considerable decrease in therecoveries was observed for all the herbicides. A shakingtime of 10 min led to similar recoveries, and no furtherimprovement was achieved. In an attempt to reduce theanalysis time while maintaining the extraction yield, an

extraction time of 5 min was selected. Finally, the numberof consecutive extractions was studied. In order to achievethe highest recoveries, three extractions were required.

Analytical characteristics and recovery study

Performance characteristics of the optimized extractionmethods were established from spiked blank soil, studyinglinearity, LODs, LOQs, and accuracy (expressed as recov-ery). Linearity and correlation coefficients of the calibrationgraphs (n=5) for each enantiomer were obtained in therange 0.20–1.5 μg g−1 for mefenpyr-diethyl, 2.0–7.5 μg g−1

for fluazifop-butyl, and 0.40–2.5 μg g−1 for quizalofop-ethyl. Peak area was used as the response and goodlinearity of the calibration curves was observed for theherbicide enantiomers at concentrations within the testedintervals, with R2 higher than 0.9771 in all cases. LODs andLOQs were calculated following the procedure described in“Analysis of soil samples.” The obtained values for eachenantiomer are shown in Table 7. As can be observed, goodresults were achieved for all herbicides with LOD and LOQvalues lower or equal than 0.115 and 1.875 μg g−1 soil,respectively.

To evaluate the efficiency and applicability of theproposed extraction methodology, a recovery study wascarried out. Several amounts of the racemic herbicides(Table 7) were added to aliquots of 1 g soil to obtain finalconcentrations higher than or equal to the highest herbicideapplication rate in agricultural crops (4 L of the herbicideformulation per hectare) [31]. Considering the soil sam-pling area (40 kg m−2) and the amount of active ingredientpresents in the commercial formulations (5% or 12.5%, w/v),the highest application rate corresponded to a spiked level of0.50 μg g−1 for the racemic mefenpyr-diethyl, 1.0 μg g−1

for quizalofop-ethyl, and 2.5 μg g−1 for fluazifop-butyl.Recoveries were determined by an analysis of the soil matrixspiked with the racemic herbicides at three concentration

Table 6 Precision and chromatographic parameters obtained from herbicide standard solutions of different R/S enantiomeric ratios

Herbicide R/S ratio Intra-day precision (RSD, %)a EF VDM (mean ± CLc) EF EMG (mean ± CLc) EFe theoretical value

Rs tr2′ (min)b

Fluazifop-butyl 80:20 2.1 0.66 0.69±0.02 0.77±0.02 0.80

90:10 4.5 2.0 0.78±0.01 0.86±0.01 0.90

Quizalofop-ethyl 80:20 3.5 1.7 0.14±0.01 0.22±0.01 0.20

90:10 5.3 1.3 0.05±0.01 0.12±0.01 0.10

95:5 4.0 0.83 0.02±0.01 0.05±0.01 0.05

a n=3, peak integration was done by deconvolution (EMG)b Adjusted retention time of the second eluted enantiomer ((S)-isomer for fluazifop-butyl and (R)-isomer for quizalofop-ethyl)c Confidence limit (95%, n=3)e In accordance with Eq. 4, the value of enantiomeric fraction ranges from 0 to 1 for all R/S compositions


Tab

le7

Analytical

characteristicsandchromatog

raph

icparametersfrom

herbicides

spiked

tosoilsamples

Herbicide

LOD

(μgg−

1)

LOQ

(μgg−

1)

R/S

ratio

Add

eda

(μgg−

1)

Recov

ery

(%)b

(mean±

RSD)

Rs(m

ean±CLc )

ER(m

ean±CLc )d

EF(m

ean±CLc )e

1st

2nd

1st

2nd

1st

2nd

VDM

EMG

VDM

EMG

VDM

EMG

Theoretical

value

Fluazifop

-bu

tyl

50:50

2.5

74±4

71±3

1.5±0.1

1.70

±0.05

1.7±

0.2

1.00

±0.02

0.63

±0.05

0.50

±0.01

0.50

0.46

90.56

31.56

31.87

550

:50

4.0

77±2

75±4

1.3±0.1

1.35

±0.02

1.2±

0.1

1.07

±0.03

0.61

±0.06

0.52

±0.02

0.50

50:50

6.0

85±2

83±3

1.2±0.1

1.25

±0.03

1.2±

0.1

1.03

±0.03

0.54

±0.03

0.50

±0.01

0.50

80:20

9.0

91±2

88±5

0.9±0.4

1.17

±0.03

––

0.73

±0.02

0.79

±0.01

0.80

Quizalofop-

ethy

l50

:50

1.0

76±2

74±3

0.8±0.1

0.92

±0.04

0.7±

0.1

0.93

±0.04

0.42

±0.05

0.48

±0.03

0.50

50:50

2.5

79±4

82±2

0.8±0.1

0.87

±0.03

0.6±

0.1

1.07

±0.03

0.39

±0.05

0.52

±0.02

0.50

0.115

0.09

80.38

40.32

650

:50

4.0

86±1

90±1

0.7±0.1

0.80

±0.01

0.6±

0.1

1.00

±0.01

0.39

±0.07

0.50

±0.01

0.50

80:20

4.0

83±2

94±1

0.9±0.1

0.90

±0.05

––

0.30

±0.01

0.19

±0.01

0.20

90:10

9.0

87±3

96±2

1.3±0.1

1.23

±0.08

––

0.08

±0.01

0.11

±0.01

0.10

Mefenpy

r-diethy

l50

:50

0.5

84±6

84±6

1.5±0.1

1.58

±0.05

1.4±

0.2

1.03

±0.02

0.58

±0.08

0.53

±0.02

0.50

0.04

70.05

60.15

50.18

750

:50

1.5

90±4

88±2

1.3±0.1

1.36

±0.03

1.2±

0.1

1.02

±0.01

0.53

±0.07

0.50

±0.01

0.50

50:50

2.5

94±1

91±1

0.8±0.2

0.94

±0.02

1.4±

0.1

1.07

±0.03

0.59

±0.05

0.51

±0.01

0.50

1stand2n

dreferred

tothefirstandsecond

eluted

enantio

mers,respectiv

ely

aTo

talconcentrationof

herbicide

bn=3

cCon

fidencelim

it(95%

,n=3)

dTheoretically

ER=1.0forracemates.In

accordance

with

Eq.

3andfordifferentR/S

ratio

s,ERcanrangefrom

0to

infinity,which

couldbe

misleadingin

data

interpretatio

neIn

accordance

with

Eq.

4,thetheoretical

valueof

theenantio

meric

fractio

ncanrangefrom

0to

1forallR/S

compositio

ns


levels in the range 0.50–2.5 μg g−1 for mefenpyr-diethyl,1.0–4.0 μg g−1 for quizalofop-ethyl, and 2.5–6.0 μg g−1 forfluazifop-butyl, which are around or slightly above than thatcorresponding to the recommended herbicide applicationdoses in agricultural soils. They were calculated using matrix-matched calibration standards. Repeatability, expressed asRSD (%), was evaluated performing three replicates for eachspiked level. As can be observed in Table 7, the bestrecoveries were obtained for mefenpyr-diethyl. In general,the highest recovery was achieved at the highest spikedlevels for all herbicides.

On the other hand, soil was spiked with fluazifop-butyland quizalofop-ethyl at different R/S enantiomeric ratios.Considering the LOQs determined for herbicide racematesin the soil matrix (Table 7), recoveries were calculated atconcentration levels of 4.0 and 9.0 μg g−1 and R/S ratios of80:20 and 90:10. Table 7 shows the recoveries obtained forboth herbicides.

In general, high and relatively constant recoveries,between 71% and 96%, with good precision, between 1%and 6%, have been obtained all over the spiked ranges andR/S enantiomeric ratios evaluated, demonstrating thesuitable extraction efficiency of the developed analyticalmethod for a difficult matrix as soil, without cleanup.

Application of deconvolution tools to spiked soil samples

As explained above, deconvolution seems to be a suitableintegration method for the calculation of chiral descriptorssuch as ER and EF when herbicide standard solutions areanalyzed. However, in complex samples, peak integration

could be affected by the co-extracted endogenous matter ofthe matrix, which can partially overlap with the analytes ofinterest and can lead to errors in the estimation of peakareas. In these cases, and also in chiral LC determination,deconvolution tools could be of a great interest to obtainmore precise areas and, consequently, ER and EF values.

Following the procedure described in “Analysis of soilsamples,” the chromatographic data obtained from the soilsamples spiked with the racemic herbicides at differentconcentration levels were fitted into the EMG function forcomplete resolution of the enantiomers to baseline. As canbe observed in Fig. 4a, c, a peak of organic matter partiallyoverlapped the first and the second enantiomers ofquizalofop-ethyl and mefenpyr-diethyl, respectively. There-fore, this matrix peak was also selected and deconvolvedtogether with the enantiomer peaks of the two herbicides.

Acceptable fits with r2>0.977 were obtained for all soilchromatograms. Deconvolved peaks of the herbicideenantiomers were integrated, and parameters such as Rs,ER, and EF were calculated. Table 7 includes the obtainedvalues for racemates from the experimental chromatogramsand from the same chromatograms after deconvolution. Ascan be seen, Rs was again similar before and afterdeconvolution of chromatographic data. However, thetwo-sample t test for mean comparisons revealed significantdifferences in the ER and EF values for all the herbicides atthe soil with the highest spiked levels. Thus, when thevalley drop method was used for peak integration, values inthe range 0.65–1.7 and 0.39–0.63, with confidence limits(CLs) between 0.03 and 0.2, were estimated for ER and EF,respectively. In contrast, the obtained values by deconvo-

0

200

400

600

800

1000

1200

0 2 4 6 8 10 12

Time (min)

AU

0

100

200

300

400

500

600

700

800

0 5 10 15 20 25 30

Time (min)

AU

100

600

1100

1600

2100

2600

3100

3600

4100

0 4 8 12 16 20 24 28

Time (min)

AU

Experimental response

Fit to Eq. (2)

Deconvolved peak

Non-spiked soil

a b

c

Fig. 4 Chromatograms of anon-spiked and a spiked soilsample with 1.5 μg g−1 ofracemic mefenpyr-diethyl (a),4.0 μg g−1 of racemic fluazifop-butyl (b), and 4.0 μg g−1 ofracemic quizalofop-ethyl (c).The figure shows deconvolvedchromatograms and the fit ofEq. 2 to experimentalchromatograms


lution were more similar to those corresponding to racemicmixtures (ER=1.0, EF=0.5), and better CLs ranging from0.01 to 0.04 were achieved.

In the same way, deconvolution was applied for thecalculation of EF in soil samples spiked with fluazifop-butyl and quizalofop-ethyl at different R/S enantiomericratios. As can be seen in Table 7, EFs calculated bydeconvolution were again more similar to the theoreticalones than those obtained by VDM. Consequently, theintegration methods based on deconvolution tools seem tobe also the choice when complex samples containingherbicides at different R/S ratios are analyzed.

Figure 4 shows a typical chromatogram obtained from asoil sample spiked with 1.5 or 4.0 μg g−1 of each herbicideand for another non-spiked soil sample under optimumextractions conditions, as well as the deconvolved enan-tiomers in the soil matrix and the fit of Eq. 2 to theexperimental chromatograms.

Conclusions

A new LC method under reversed phase conditions has beendeveloped for the chiral separation of fluazifop-(R,S)-butyl,quizalofop-(R,S)-ethyl, and mefenpyr-(R,S)-diethyl herbi-cides on a commercial α1-acid glycoprotein chiral column.This method avoids the common problems of the muchemployed normal phase LC methods for the chiral separationof ArPPs such as lack of reproducibility and long equilibra-tion times. This work has also demonstrated the effectivenessof mathematical deconvolution as an integration methodusing algorithms available in commercial software. Moreprecise peak areas when enantiomers are partially or fullyresolved were obtained. Consequently, better results in thecalculation of both ER and EE of racemates and other R/Senantiomeric ratios could be achieved.

On the other hand, a method for the analysis of chiralherbicides in soil has been proposed. The optimized sampleextraction procedure combined with direct chiral LC-UVdetection and peak deconvolution provides a simple,effective, and reproducible method for herbicide determi-nation in soil matrices at the low levels of micrograms pergram. Thus, the developed procedure has allowed obtainingacceptable and constant spiked recoveries of the enan-tiomers with negligible matrix effect.

Acknowledgments The present work has received financial supportfrom the project CTQ2008-05263/BQU, funded by the SpanishMinisterio de Ciencia e Innovación, Subdirección General deProyectos de Investigación. Vanesa Guillén-Casla also wishes tothank the Comunidad de Madrid for the support through a “PersonalInvestigador de Apoyo” grant.

References

1. Weed Science Society of America (1994) Herbicide handbook,7th edn. Weed Science Society of America

2. Davies J (2001) Pestic Outlook 12:10–153. Renner R (2000) Environ Sci Technol 34(9):222A–226A4. Spangler LA, Mikalojczyk M, Burdge EL, Kielbasinski P, Smith

HC, Lyzwa P, Fisher JD, Omelanczuk J (1999) J Agric FoodChem 47:318–321

5. Peng W, Shuren J, Donghui L, Weili S, Hongjun Z, Zhiqiang Z(2006) J Chromatogr Sci 44(10):602–606

6. Gu X, Wang P, Liu D, Lu Y, Zhou Z (2008) Pestic Biochem Phys92:1–7

7. Wang P, Jiang S, Liu D, Zhang H, Zhou Z (2006) J Agric FoodChem 54:1577–1583

8. Gu X, Yuele Y, Wang P, Dang Z, Zhou Z (2010) Food Chem121:264–267

9. Lin KD, Cai XY, Chen SW, Liu WP (2006) Chin J Anal Chem5:613–616

10. Wang P, Liu DH, Jiang SR, Xu YG, Gu X, Zhou ZQ (2008)Chirality 20:40–46

11. Chunxiu P, Baochun S, Xuejun Z, Datong Z, Xiuzhu X, Chin(2006) J Anal Chem 34(2):159–164

12. Mojtahedi MM, Chalavi S, Ghassempour A, Tabar-Heydar K,Sharif SJ, Malekzadeh M, Aboul-Enein HY (2007) BiomedChromatogr 21:234–240

13. Fanali S, D’Orazio G, Rocco A (2006) J Sep Sci 29:1423–143114. Weber K, Kreuzig R, Babadir M (1997) Chemosphere 35:13–

2015. Padiglioni P, Polcaro CM, Marchese S, Sinibaldi M, Flieger M

(1996) J Chromatogr A 756:119–12716. Si Ahmed K, Tazerouti F, Badjah-Hadj-Ahmed AY, Meklati BY

(2005) Chromatographia 62:571–57917. Tian Q, Lv C, Wang P, Ren L, Qiu J, Li L, Zhou Z (2007) J Sep

Sci 30:310–32118. Guillén-Casla V, Pérez-Arribas LV, León-González ME, Polo-

Díez LM (2008) J Agric Food Chem 56:2303–230919. Persson BA, Andersson S (2001) J Chromatogr A 906:195–20320. Mongay-Fernández C (2005) Quimiometría. Servicio de Publica-

ciones, Universidad de Valencia, Valencia21. Kiesswetter R, Brandl F, Kastner-Pustet N, Mannschreck A

(2003) Chirality 15:S40–S4922. Asher BJ, D’Agostino LA, Way JD, Wong CS, Harynuk JJ (2009)

Chemosphere 75:1042–104823. Nikitas P, Pappa-Louisi A, Papageorgiou A (2001) J Chromatogr

A 912:13–2924. Li J (2002) J Chromatogr A 952:63–7025. Dasgupta PK (2008) J Chromatogr A 1213:50–5526. Kong H, Ye F, Lu X, Guo L, Tian J, Xu G (2005) J Chromatogr A

1086:160–16427. Krupcík J, Mydlová J, Spánik I, Tienpont B, Sandra P (2005) J

Chromatogr A 1084:80–8928. Norli HR, Christiansen A, Holen D (2010) J Chromatogr A

1217:2056–206429. Harner T, Wiberg K, Nordstrom R (2000) Environ Sci Technol

34:218–22030. Díez C, Barrado E, Marinero P, Sanz M (2008) J Chromatogr A

1180:10–2331. Registro de productos fitosanitarios (2010) Ministerio de Medio

Ambiente y Medio Rural y Marino, Spain. http://www.mapa.es/es/agricultura/pags/fitos/registro/menu.asp. Accessed 21 October2010


http://www.mapa.es/es/agricultura/pags/fitos/registro/menu.asp

http://www.mapa.es/es/agricultura/pags/fitos/registro/menu.asp

Direct chiral liquid chromatography determination of aryloxyphenoxypropionic herbicides in soil: deconvolution tools for peak processing

Documents