Dwg

Pa

b

a

ARR2AA

KDLGHM

1

wcpitcauuh8e[el

(

0d

Analytica Chimica Acta 655 (2009) 60–65

Contents lists available at ScienceDirect

Analytica Chimica Acta

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

evelopment of a simple device for dispersive liquid–liquid microextractionith lighter than water organic solvents: Isolation and enrichment of

lycyrrhizic acid from licorice

ayman Hashemia,∗, Somayeh Beyranvanda, Reza Siah Mansurb, Ali Reza Ghiasvanda

Department of Chemistry, Faculty of Science, Lorestan University, Falakol,aflak St, Khoramabad, IranAgriculture and Natural Resource Research Center, P.O. Box 348, Khoramabad, Iran

r t i c l e i n f o

rticle history:eceived 23 June 2009eceived in revised form2 September 2009ccepted 23 September 2009vailable online 27 September 2009

eywords:

a b s t r a c t

A simple device was developed that makes the use of lighter than water organic solvents feasible in disper-sive liquid–liquid microextraction (DLLME) method. In the ordinary DLLME, the fact that a heavier thanwater organic solvent must be used, to be sedimented at the conical bottom of a centrifuge tube, limitsthe applications of this method in some extent. In the developed method, a glass tube with a narrow neckis inserted inside the centrifuge tube. After phase separation, the organic solvent is accumulated in thenarrow neck of the device and therefore, can be simply collected by a micro-syringe. The DLLME methodwith the proposed device was tested for the enrichment of glycyrrhizic acid from aqueous extracts of

ispersive liquid–liquid microextractionighter than water organic solventslycyrrhizic acidigh performance liquid chromatographyultivariate optimization

licorice before analysis by a HPLC method. n-Hexanol and acetone were used as the organic and disperserphases, respectively. Effects of pH, salt concentration and phase volumes on the extraction of the analytewere optimized using a central composite (response surface) design. Under the optimized conditions (i.e.pH 1.3, ionic strength 0.2 mol L−1, n-hexanol 140 �L and acetone 0.8 mL) an extraction recovery of 104.1(±5.1)% and an enrichment factor of 54 were obtained. The proposed method was successfully appliedfor the study of glycyrrhizic acid’s level of licorice roots grown in three regions of Lorestan province, Iran,

nditio

with different climate co

. Introduction

Licorice (Glycyrrhiza globra L.) is a traditional herbal medicineith several health benefits such as antiinflammatory, antiul-

erous, antidotal, antiallergic, immunomodulating and antiviralroperties [1–3]. Glycyrrhizic acid (GA), is the main pharmacolog-

cally active ingredient in the root of licorice which is about 170imes sweeter than sucrose [4]. GA is a weak acid containing threearboxyl and five hydroxyl groups. It is a glycosylated saponin withpentacyclic triterpenic structure bound to two glucuronic acid

nits as shown in Fig. 1 [4]. The GA content of the licorice dependspon several parameters such as the species, plant age, season ofarvest and climate conditions. Its level ranges probably from 0.2 to3.7 mg g−1 [5–7]. A number of methods such as reflux [6,7], solvent

xtraction [8,9], surfactant assisted pressurized liquid extraction10], pressurized hot water extraction [11] and microwave assistedxtraction [12] have been used for the withdrawal of GA fromicorice roots before its analysis by HPLC. Methanol, ethanol and

∗ Corresponding author. Tel.: +98 661 2202782; fax: +98 661 2200185.E-mail addresses: payman hashemi@yahoo.com, hashemi.p@lu.ac.ir

P. Hashemi).

003-2670/$ – see front matter © 2009 Elsevier B.V. All rights reserved.oi:10.1016/j.aca.2009.09.034

ns.© 2009 Elsevier B.V. All rights reserved.

water, or a mixture of them, have been among the most commonlyused solvents for this purpose [1]. However, other solvents suchas tributyl phosphate, trialkylphosphine oxide, n-hexanol and 2-ethyl-hexanol have been also practiced [8].

Microextraction methods have attracted much attention inrecent years as alternatives for classic liquid–liquid and solventextraction procedures. Solid phase microextraction (SPME) [13]and liquid-phase microextraction (LPME) [14], both with direct andheadspace operations, have been extensively used in recent years.These techniques are simple and fast, miniaturize sample pretreat-ment processes and minimize the use of organic solvents. However,these techniques suffer from some problems such as samplecarry-over, relatively high cost, fiber fragility and relatively low pre-cisions. Recently, dispersive liquid–liquid microextraction (DLLME)was introduced as an alternative to the classic liquid–liquid extrac-tion method [15]. DLLME is simple, rapid and cost effective andprovides high enrichment factors. DLLME has been applied for theanalysis of a variety of trace organic pollutants and metal ions in the

environmental samples [15–21]. DLLME is based on a ternary com-ponent solvent system. In this method, generally, a small volumeof a heavier than water organic solvent (such as carbon tetrachlo-ride or chlorophenol) is dispersed into fine droplets in the samplesolution by the aid of a water miscible solvent as disperser. After

P. Hashemi et al. / Analytica Chimica Acta 655 (2009) 60–65 61

pstito

nomtlieo

ecdcau

2

2

Ahfd

a

2

(as(mwwB

Fig. 1. The chemical structure of glycyrrhizic acid.

hase separation followed by centrifugation, the organic solvent isedimented at the bottom of the centrifuge tube as a micro-drophat can be withdrawn by a micro-syringe. However, the fact thatn this method a heavier than water, normally chlorinated, highlyoxic, solvent has to be used limits, in some extent, the applicationsf this method.

In this paper, a simple device composed of a glass tube with aarrow neck, is presented that makes the use of lighter than waterrganic solvents feasible. The method is then applied for the enrich-ent and HPLC determination of GA in licorice roots and study of

he effect of climate conditions on its level. A recent study in ouraboratory indicated that the climate conditions can significantlynfluence the essential oil composition of cumin seeds [22]. How-ver, to the best of our knowledge, there is no report on the effectf climate conditions on the GA content of licorice roots.

A central composite (response surface) design is used forfficient optimization of the microextraction conditions. Centralomposite design is a multivariate optimization technique that pro-uces a detailed quantitative model for the response(s). This modelan predict how a response relates to the values of various factorsnd it usually requires less number of experiments compared to anivariate procedure [23].

. Experimental

.1. Reagents and materials

Glycyrrhizic acid (GA) monoammonium salt was prepared fromCROS Organics (Newjercy, USA). HPLC grade acetonitrile, n-exanol, organic solvents and other chemicals were purchased

rom Merck (Darmstadt, Germany) and used as received. Double-istilled water was used throughout.

A 0.01 mol L−1 solution of GA in ethanol was prepared and useds stock solution for the preparation of working standards.

.2. Apparatus

HPLC analysis of the samples was conducted on a Shimadzumodel L-10AD) instrument consisting of two reciprocating pumps,DGU-14A in-line degasser, a model CT10-10AC oven, a high pres-

ure manual injection valve (20 �L injection loop) and a UV/VIS

model SPD-10A) detector. The software used for the data acquire-

ent and processing was Class-vp v.R 6.1. The analytical columnas a 25 cm × 4.6 mm i.d. RP-8 column (Shim-Pack CLC-C8) packedith 5 �m particles and equipped with a 1 cm guard column (C8-197) packed with 10 �m particles of the same type.

Fig. 2. The proposed NNGT device for lighter than water organic solvents.

A 25 �L HPLC micro-syringe (F-LC, SGE, Australia) was used forthe sample withdrawal and injection. A totally glass Fisons (UK)double distiller was used for preparation of doubly distilled water.

For ultrasonic irradiation of the samples, an ultrasonic waterbath (22 kHz, model 5RS, Sonica, Italy) was used. An Eppendorf AGcentrifuge (model 5810, Hamburg) was used for centrifugation ofthe extracts.

For DLLME with lighter than water organic solvents a home-made narrow neck glass tube (NNGT) was used (Fig. 2). The glasstube was about 10 cm in length and 12 and 4 mm in diametersof the body and neck, respectively. The NNGT was placed insidea 15 mL polyethylene centrifuge tube for the centrifugation andphase separation.

2.3. Field samples

Licorice (G. globra L.) root samples were collected from threeregions of Lorestan province in the west part of Iran. This provincehas a variation from warm to cold climates. The regions selected forthis study were Alashtar with cold, Khoramabad with temperateand Poldokhtar with tropical climates. Some characterizations ofthe experimental sites and collection time of the samples have beensummarized in Table 1. From each region, at least 10 root sampleswere collected. The collected samples were dried in room temper-ature, ground into about 1 mm pieces, homogenized and stored incapped bottles before the extraction.

2.4. Procedures

For direct ultrasound assisted solvent extraction of GA fromlicorice samples, 1.0 g of the ground sample was transferred intoa glass tube and 25 mL water was added. The test tube was

◦

capped, thermostated at 55 C, and sonicated for 15 min in theultrasonic water bath. The supernant solution was decantated andthe extraction procedure was repeated on the residue once more.The supernant solutions were added together, filtered through a0.45 �m membrane filter and stored at 4 ◦C before analysis.

62 P. Hashemi et al. / Analytica Chimica Acta 655 (2009) 60–65

Table 1Some characterizations of the experimental sites and collection time of the licorice samples.

Region Climate Altitude (m) Mean annual temperature Mean annual rainfall Harvest time

112

sDp

aham(wadts8wtwtH

psrlBT

Ddfvtwti

mocts3bc

TTe

For an appropriate separation of the licorice roots components,different gradient elution programs were used for the HPLC col-umn. In the literature, usually long programs, up to 140 min [9],have been reported for an appropriate separation. In this work the

Table 3The conditions used and extraction recoveries (E%) obtained for the central compos-ite design experiments.

No. Parametersa E%

DV EV IS pH

1 700 125 0.05 2.0 45.832 600 150 0.1 2.75 53.253 600 100 0.1 1.25 44.914 700 75 0.05 3.5 22.205 600 100 0.1 2.75 41.586 500 125 0.15 2.0 49.327 800 100 0.1 2.75 53.628 400 100 0.1 2.75 26.969 700 125 0.15 3.5 27.6110 600 100 0.1 4.25 16.3611 500 75 0.15 2.0 34.6412 700 125 0.15 2.0 53.90

Aleshtar (Honam) Cold 1600Khoramabad (Sarab Changaei) Temperate 1125Poldokhtar (Ghashghaei) Tropical 713

The same procedure was used for preliminary extraction of theamples before the isolation and preconcentration by the proposedLLME method. However, in this case only 0.1 g of the ground sam-le was applied.

For performing the DLLME method, the sample acidity wasdjusted on 0.05 mol L−1 (pH of about 1.3) by addition of someydrochloric acid solution. The ionic strength of the sample waslso adjusted on 0.4 mol L−1 using a sodium nitrate solution. In aicrovial, 0.8 mL acetone (as the disperser) and 140 �L n-hexanol

as the acceptor organic solvent) were mixed and the mixture wasithdrawn by a 2 mL glass syringe, equipped with a 10 cm needle,

nd rapidly injected into 4.0 mL of the sample in the proposed NNGTevice. A uniformly distributed cloud was achieved by further suc-ion and reinjection of about 2 mL of the cloudy sample using theame syringe. The mixture was then centrifuged at 3000 rpm formin for the phase separation. Since n-hexanol is a lighter thanater solvent, the extraction phase was collected at the solvent

op. By careful addition of some water drops, the organic phaseas conducted to the narrow neck of the NNGT. This phase was

hen carefully withdrawn by a micro-syringe and injected into thePLC column, after three times dilution by ethanol in a microvial.

For the HPLC separation of GA and other co-extracted com-ounds, a gradient elution with a mixture of solvents A (1%olution of acetic acid in water) and B (acetonitrile) with a flowate of 1.0 mL min−1 was used. The elution program was as fol-ows: 0–6 min, 20–30% B; 6–8 min, 30–50% B; 8–15 min, 50–60%; 15–17 min, 60–80% B; 17–23 min, 80% B; 23–25 min, 80–20% B.he chromatograms were acquired at 254 nm.

For optimization of the extraction conditions by the proposedLLME method a half fraction central composite (response surface)esign was performed using Minitab statistical software [24]. Fouractors of n-hexanol (organic solvent) volume, acetone (disperser)olume, pH of the aqueous phase and salt (sodium nitrite) concen-ration were included in the design. The extraction recovery (E%)as considered as the dependent variable to be optimized using

he model. Table 2 shows the abbreviations and levels of each factorncluded in the design.

For calculation of the extraction recovery (E%) of the proposedethod, 100 times of the amount extracted of GA (mmol) into the

rganic phase was divided by its initial amount in the sample. Foralculation of the enrichment factor (EF), the GA concentration inhe receiving phase was divided by its initial concentration in theample. The detection limit of the method was calculated from

�, using the calibration curve data. The standard deviation of thelank (sb) was estimated from that of the residuals (sy/x) for thisalculation.

able 2he factors included in the central composite design and the low and high levels forach factor.

Parameter Abbreviation Factor’s levels

− +

Extractant (n-hexanol) volume (�L) EV 125 75Disperser (acetone) volume (�L) DV 700 500Sample pH pH 3.5 2.0Ionic strength (mol L−1) IS 0.15 0.05

2.9 ◦C 576 mm June 20077.9 ◦C 438 mm June 20072.9 ◦C 433 mm June 2007

3. Results and discussion

3.1. Ultrasound assisted solvent extraction of GA

It is known that ultrasound power can substantially improvethe extraction efficiency of a solvent extraction method [25]. There-fore, the extraction of GA from licorice root samples was conductedby the assistance of this power and in an ultrasonic bath. In somepreliminary studies, different solvents such as ethanol, methanol,water and different mixtures of them were tested. It was foundthat water and mixtures of water and an alcohol are more effi-cient solvents than the pure alcohols. Water was finally selectedas the solvent, because it was more appropriate for performing thesubsequent DLLME procedure. Maximum extraction efficiency withwater was obtained at a temperature of 55 ◦C, after 15 min sonica-tion in the ultrasonic bath. For ensuring the complete extractionof GA from the samples, four more successive extractions wereperformed on the residue of the first extract. It was found that,respectively, 84.4% and 15.1% of GA is extracted in the first andsecond steps and extraction in the following steps is negligible.Therefore, extraction in two successive steps was considered asquantitative and used in subsequent experiments.

13 500 125 0.05 3.5 32.0114 500 125 0.15 3.5 34.4515 600 100 0 2.75 41.3016 600 50 0.1 2.75 0.0017 600 100 0.1 2.75 44.4718 700 75 0.05 2.0 18.6519 700 125 0.05 3.5 44.9120 500 75 0.05 2.0 25.4421 700 75 0.15 3.5 18.2722 500 75 0.15 3.5 25.3823 700 75 0.15 2.0 37.4924 600 100 0.1 2.75 37.9125 500 75 0.05 3.5 25.3226 500 125 0.05 2.0 40.3127 600 100 0.2 2.75 48.02Optb 800 140 0.2 1.3 74

a For the full names of the parameters and their units, see Table 2.b The predicted optimized conditions by the model and the predicted E%.

Chimica Acta 655 (2009) 60–65 63

sao

3

sTlnovct(a

etesfwfamt

3

op

estvviI

mptbl

TE

P. Hashemi et al. / Analytica

eparation time was reduced to about 20 min by selecting appropri-te solvents and a suitable gradient program with the appearancef GA peak at 14.7 min in the chromatogram.

.2. DLLME of GA

In some preliminary experiments it was found that chlorinatedolvents cannot efficiently extract GA from its aqueous solutions.herefore, some lighter than water organic solvents, such as ethy-acetate, cyclohexane, petroleum ether, tributyl phosphate and-hexanol were examined for this purpose. Since withdrawal of therganic solvents collected as a thin layer on the sample surface wasery difficult or even impossible, a simple NNGT device was fabri-ated and inserted inside the centrifuge tube (see Fig. 2). By usinghis device, the organic phase was conducted to a narrow spacethe neck of the device) and therefore could be easily suctioned bymicro-syringe.

For protonation of the carboxylic acid group of GA and its betterxtraction to the organic phase, the test solutions were acidifiedo pH 2. Using acetone as the disperser indicated a more efficientxtraction of the analyte by using n-hexanol as the solvent. Thisolvent is known as an efficient solvent for solvent extraction of GArom aqueous licorice leaching solutions [8]. Therefore, n-hexanolas used as the extraction solvent in subsequent experiments. Dif-

erent water miscible organic solvents such as methanol, ethanolnd acetone were tested as disperser for the DLLME system. Sinceore stable and uniform cloudy solutions were shaped with ace-

one, this solvent was selected as the disperser.

.3. Optimization by the central composite design

After performing some preliminary experiments, optimizationf the extraction conditions by the proposed DLLME method waserformed by a half fraction central composite design.

The data given in Table 3 show the extraction recoveriesxpressed as E% for 27 experiments designed by using the MINITABoftware [23]. Table 4 shows the regression coefficients for eacherm in the model and the t-Student’s statistics and probabilityalues for the significance of the terms. It is obvious that n-hexanololume (EV) and pH, with the smallest p-values, are the most signif-cant factors. The quadratic term of EV2 and the interaction betweenS and pH were also significant in 95% confidence level.

Fig. 3 depicts the response surface plots of extraction recovery

odeling for some of the factors. The curvature in these kinds of

lots can indicate possible interactions between factors. Accordingo Fig. 3a, in a high ionic strength the extraction efficiency increasesy decreasing the pH. On the other hand, in a low ionic strength

ower extraction efficiencies are obtained in low pH levels. This

able 4stimated regression coefficients for E% obtained for the central composite design.

Term Coef T P

Constant 41.3185 10.193 0.000DV 2.3055 1.609 0.134EV 9.4766 6.612 0.000IS 1.6589 1.157 0.270pH −5.5212 −3.852 0.002DV2 −0.6020 −0.396 0.699EV2 −4.0184 −2.643 0.021IS2 0.4899 0.322 0.753pH2 −3.0160 −1.984 0.071DV × EV 1.8951 1.080 0.302DV × IS −0.9382 −0.534 0.603DV × pH −0.6453 −0.368 0.720EV × IS −1.3723 −0.782 0.449EV × pH −1.5833 −0.902 0.385IS × pH −3.9907 −2.273 0.042

Fig. 3. The response surface plots for the extraction recovery (E%) as a function ofIS and pH (a) and EV and pH (b).

is a demonstration of the interaction between IS and pH that is inconcurrence with the results of Table 4. No considerable interactionof such is observed between EV and pH in Fig. 3b, as expected.

The response surface model with the estimated regression coef-ficients (see Table 4) was used for the response optimization,including all the factors. The conditions for the predicted optimumpoint have been shown in the last row of Table 3. To evaluate thevalidity of the prediction by the model, three experiments wereperformed under the suggested optimized conditions and a meanextraction recovery of 70.6 ± 2.6 was obtained which shows a fairagreement between the predicted and experimental results.

3.4. Effects of some individual factors

In the optimized conditions suggested by the central compos-ite design, the extraction of GA was still not quantitative. Furtherstudies showed that gentle shaking of the formed cloudy solutionor suction and reinjection of a part of it by a syringe increases theextraction recovery. The results indicated that only one time of thesuction and reinjection was sufficient for achievement of a uni-form cloudy solution and quantitative recovery of the analyte. Thismethod was more efficient than gentle shaking of the NNGT andtherefore, was used in all subsequent experiments. By using thismethod, about 100% recoveries could be obtained.

After performing a multivariate optimization procedure, it isoften desirable to study, in some details, the individual effect of

some important factors. This will give a better view of the effectsof the experimental conditions on the response which is useful inmethod development practices. Here, the individual effect of pHwas studied because of its significance. The suction and reinjectionmethod was used during the study for a better homogenization.

64 P. Hashemi et al. / Analytica Chimica Acta 655 (2009) 60–65

Fo

Fclip

osTtEio

3

au4oaaT2

3

ffcaosrdoccp

TTt

ig. 4. Influence of pH on the extraction recovery (E%) and enrichment factor (EF)f GA in the DLLME procedure, under the optimized conditions.

ig. 4 indicates the large variation of E% and EF with pH. This indi-ates that protonation of the carboxylic group of GA occurs in quiteow pH values. In such acidic conditions, GA gets uncharged ands extracted, more efficiently, into the fine droplets of the organichase.

The individual effect of ionic strength was also studied becausef its importance in natural samples. The results showed a relativelymall influence of ionic strength on the extraction efficiency of GA.he E% increased moderately with increasing the salt concentra-ion showing a flat maximum at about 0.4 mol L−1. However, the% variation in the region from 0.2 to 0.5 mol L−1 was statisticallynsignificant. A more or less similar, but less significant, trend wasbserved for EF.

.5. Analytical performance

Replicated experiments in the optimal conditions (i.e. 0.8 mLcetone, 140 �L n-hexanol, 0.2 mol L−1 sodium nitrate, pH 1.3 andsing the suction and reinjection method) for the extraction of.0 mL of 2 × 10−5 mol L−1 GA resulted in an extraction recoveryf 104.1% with a relative standard deviation of 5.1% (for n = 5) andn enrichment factor of 54. The calibration curve was linear overrange of 2 × 10−5 to 1 × 10−3 mol L−1 GA, with a R2 of 0.999.

he detection limit of the method for 3� was calculated to be× 10−7 mol L−1 of GA.

.6. The DLLME of licorice samples

The developed DLLME method was applied for extraction of GArom water extracts of licorice samples collected from three dif-erent regions of Lorestan province, Iran, with different climateonditions (Table 1). The data, summarized in Table 5, indicate goodgreement between the results of the DLLME and direct injectionsf the samples. It should be noted that for the DLLME of the licoriceamples, some precipitate was observed between the two sepa-ated phases after centrifugation. This precipitate was somewhatispersed in the supernant organic phase when a salt concentration

f 0.2 mol L−1 was used. This problem, however, could be over-ome by increasing the salt concentration to 0.4 mol L−1. Fig. 5ompares the chromatograms of a licorice sample, before and afterreconcentration by the proposed method, under the optimized

able 5he results of the DLLME-HPLC analysis of the extracts of three licorice roots underhe optimized experimental conditions.

Licorice sample GA (mg g−1), DLLME GA (mg g−1), direct E% EF

Alashtar 11.18 (±0.23) 11.81 (±0.02) 94.7 49Khoramabad 11.28 (±0.01) 11.80 (±0.04) 95.6 51Poldokhtar 20.51 (±0.72) 20.70 (±0.15) 99.1 52

Fig. 5. The chromatograms of the extracts of Khoramabad’s licorice sample, before(a) and after (b) preconcentration by the DLLME method, under the optimized con-ditions. S, stands for the solvent (n-hexanol) peak in chromatogram b.

conditions. Evidently, the DLLME method has not only preconcen-trated but also purified the analyte in a considerable extent (notethe scales of the vertical axes). It is due to the fact that hydrophilic(ionic) compounds are not extracted by the organic phase. Sincein the preliminary extraction by water, lipophilic species has beenalready removed, only GA and a few other components are pre-concentrated by the DLLME method. Due to the high enrichmentfactors, the method can be used for the determination of tracelevels of GA in licorice samples. Performing a one-way ANOVAtest on the results indicated that climate conditions have signifi-cant effect on the GA level. Obtaining an F-ratio of 299 indicatesthe high significance of this effect. The test showed that the GAlevel in the sample from tropical region was significantly higherthan those of the temperate and cold regions. However, therewas not a significant difference between the temperate and coldregions.

4. Conclusion

The simple DLLME device proposed in this work could besuccessfully applied for the extractions with lighter than watersolvents. The proposed device overcomes the necessity for usingtoxic chlorinated organic solvents in the ordinary DLLME methodand extends, largely, the area of organic solvents applicable in thistechnique. Application of the proposed method for the extraction

of GA from licorice samples was successful and could both pre-concentrate and partially purify the analyte. The proposed methodrequires much less amounts of the sample and is quite appropriatefor trace analysis applications. The DLLME device in this work wasused only as a case study, but the proposed device can be easily

Chim

al

R

[

[[[[[

[[[

[[

[[

P. Hashemi et al. / Analytica

pplied for the DLLME of a number of other solutes with severalighter than water organic solvents.

eferences

[1] V.K. Gupta, A. Fatima, U. Faridi, A.S. Negi, K. Shanker, J.K. Kumar, N. Rahuja, S.Luqman, B.S. Sisodia, D. Saikia, M.P. Darokar, S.P.S. Khanuja, J. Ethnopharmacol.116 (2008) 377.

[2] G.A. Tolstikov, L.A. Baltina, E.E. Shults, A.G. Pokrovskii, Russ. J. Bioorg. Chem. 23(1997) 691.

[3] R.M. Kondratenko, L.A. Baltina, S.R. Mustafina, N.V. Makarova, K.M. Nasyrov,G.A. Tolstikov, Zh. Org. Farm. Khim. 35 (2001) 39.

[4] J.R. Hennell, S. Lee, C.S. Khoo, M.J. Gray, A. Bensoussan, J. Pharm. Biomed. Anal.47 (2008) 494.

[5] E.A. Spinks, G.R. Fenwick, Food Addit. Contam. A 7 (1990) 769.

[6] K. Sagara, Y. Ito, T. Oshima, M. Kawaura, T. Misaki, Chem. Pharm. Bull. 33 (1985)

5364.[7] T.-H. Tsai, C.-F. Chen, J. Chromatogr. A 542 (1991) 521.[8] G.-G. Niu, Y.-C. Xie, J.-F. Lou, H.-Z. Liu, Sep. Purif. Technol. 44 (2005) 189.[9] Y.-C. Wang, Y.-S. Yang, J. Chromatogr. B 850 (2007) 392.10] A.T.W. Eng, M.Y. Heng, E.S. Ong, Anal. Chim. Acta 583 (2007) 289.

[

[

[

ica Acta 655 (2009) 60–65 65

11] M. Mukhopadhyay, P. Panja, Sep. Purif. Technol. 63 (2008) 539.12] X. Pan, H. Liu, G. Jia, Y.Y. Shu, Biochem. Eng. J. 5 (2000) 173.13] H. Lord, J. Pawliszyn, J. Chromatogr. A 902 (2000) 17.14] L. Zhu, C.B. Tay, H.K. Lee, J. Chromatogr. A 963 (2002) 231.15] M. Rezaee, Y. Assadi, M.R. Milani, E. Aghaee, F. Ahmadi, S. Berijani, J. Chromatogr.

A 1116 (2006) 1.16] E. Zhao, W. Zhao, L. Han, S. Jiang, Z. Zhou, J. Chromatogr. A 1175 (2007) 137.17] P. Liang, E. Zhao, F. Li, Talanta 77 (2009) 1854.18] M.A. Farajzadeh, M. Bahram, B. Ghorbani Mehr, J.A. Jonsson, Talanta 75 (2008)

832.19] A. Daneshfar, T. Khezeli, H.J. Lotfi, J. Chromatogr. B 877 (2009) 456.20] L. Fu, X. Liu, J. Hu, X. Zhao, H. Wang, X. Wang, Anal. Chim. Acta 632 (2009)

289.21] M.B. Melwanki, W.S. Chen, H.Y. Bai, T.Y. Lin, M.R. Fuh, Talanta 78 (2009) 618.22] P. Hashemi, M. Shamizadeh, A. Badiei, A.R. Ghiasvand, K. Azizi, Chro-

matographia, in press.

23] R.G. Breretton, Chemometrics, Data Analysis for the Laboratory and Chemical

Plant, John Wiley & Sons Ltd., England, 2003.24] Inc. Minitab, Minitab Statistical Software, Release 15 for Windows, State Col-

lege, Pennsylvania, 2006.25] P. Hashemi, B. Delfan, A.R. Ghiasvand, M. Alborzi, F. Raeisi, Acta Chromatogr.,

in press.