10
Full Paper Factorial designs for Cd, Cr, Hg, Pb and Se ultrasound-assisted acid leaching from human hair followed by atomic absorption spectrometric determination Pilar Bermejo-Barrera, Antonio Moreda-Pin ˜eiro and Adela Bermejo-Barrera Department of Analytical Chemistry, Nutrition and Bromatology, Faculty of Chemistry, University of Santiago de Compostela, Avenida das Ciencias s/n., E 15706, Santiago de Compostela, Spain Received 27th August 1999, Accepted 15th November 1999 An acid leaching procedure, for use as a sample pre-treatment, was developed for Cd, Cr, Hg, Pb and Se determination in human hair. A Plackett–Burman experimental design was used as a multivariate strategy for the evaluation of the effects of varying several variables at once. The effects on acid leaching of metals of nitric acid, hydrochloric acid and hydrogen peroxide concentrations, acid solution volume, exposure time to ultrasound, temperature of the ultrasonic bath, and hair particle size, have been investigated. From these studies, certain variables showed up as significant, and they were optimised by a 2 3 zstar central composite design, which involved 16 experiments. Optimum values of the variables were selected for the development of acid leaching procedures to extract trace metals from human hair. Electrothermal atomic absorption spectrometry (ETAAS) was used to determine Cd, Cr, Pb and Se, while cold vapour-atomic absorption spectrometry (CVAAS) was used for Hg measurements. An acid digestion induced by microwave energy was used to obtain the total metal concentration and also for comparative purposes. IAEA-085 and NIES No. 13 certified reference materials, which offer certified values for some elements, were used in order to verify the accuracy of the methods. Introduction Interest in human hair as a clinical sample has increased in recent years due to certain advantages offered by human hair over other clinical specimens such as blood or urine samples. These advantages have been mentioned by Burguera et al. 1 and the most important is the ten-fold higher metal concentrations in hair than are present in blood or urine. In forensic science, human hair has been demonstrated to be one of the most useful clinical samples to assess drug consumption, so drugs of abuse and/or metabolites analysis in human hair is now well established and the methods are recommended. 2 However, human hair analysis for metals is not a routine methodology; this can be due to the fact that metals reference values for human hair are not available. External contamination by dust and sweat, and cosmetic treatments, are other drawbacks that we have to take into account when metals determination in human hair is performed. These limitations can be solved using standard procedures accepted by the international community for both hair sampling and hair washing processes. These recommendations have been summarised by Bencze. 3 Never- theless, human hair has been used to assess mercury (methylmercury) impregnation in the course of environmental monitoring, 4 and according to Caroli et al. 5 the use of human hair as clinical sample must be exploited. Different sample pre-treatments for element human hair analyses, such as acid digestion procedures, 1,6–15 have been developed. The slurry sampling technique 16–22 has also been used since the firsts applications carried out by Kitagawa et al. 23 and Ebdon et al. 24 Other procedures such as enzymatic hydrolysis processes are nowadays being studied, although many limitations have been found to extract some metals. 25 The extraction of elements by the action of an acid and/or an oxidant agent, named as acid leaching, appears to be an interesting sample pre-treatment that must be developed. Leaching is not a total decomposition, and leachable recoveries of analytes are generally lower than total concentrations. Therefore, recoveries can only achieve total values if the elements are completely soluble in the leaching solvent. 26 The main advantage of this sample pre-treatment appears to be the lower acid concentrations than are required by acid digestion procedures to achieve extraction of metals which leads to small consumption of mineral acids and a reduction in nitrous vapours. In addition, the time required for sample preparation is reduced. Sediment, sludges and soils have been subjected to microwave assisted and conventional leaching using EPA method 3035B. 26 Puchyr and Shapiro 27 have developed an acid leaching process to extract several elements from seafoods and foods with high sugar content, while El Azouzi et al. 28 applied an acid leaching induced by ultrasound to extract metals from mussel samples. Acid leaching processes for speciation analysis have been also used, 29 and in this sense, the method developed by Westo ¨o ¨ 30 in order to extract methyl- mercury from biological matrices can be considered the first application of acid leaching procedures. Several means of optimisation can be used but some methods are time-consuming: for instance, the study of each variable separately. In addition, some optimisation methods, such as simplex procedures, are restrictive in the number of variables under study. As many variables are involved throughout the acid leaching process, experimental designs are powerful tools to optimise a certain procedure. Among the different groups of designs, Plackett–Burman designs (PBDs), introduced in 1946 by Plackett and Burman, 31 allow us to discover the most significant variables for a certain system with only few experiments. They are used as a screening method in order to select the variables that have influence on a system but they do not give the optimum value for each variable. PBDs constitute a variation on saturated fractional designs, allowing the evaluation of either system with few experiments; k factors can be studied in kz1 runs (only the main effects are estimated). These designs can be used only when kz1 is a multiple of 4 (i.e., k~3, 7, 11, …). This enlarges the opportunities offered by the saturated factorial design, which J. Anal. At. Spectrom., 2000, 15, 121–130 121 This journal is # The Royal Society of Chemistry 2000 Published on 10 February 2000. Downloaded by Lomonosov Moscow State University on 28/11/2013 16:45:33. View Article Online / Journal Homepage / Table of Contents for this issue

Factorial designs for Cd, Cr, Hg, Pb and Se ultrasound-assisted acid leaching from human hair followed by atomic absorption spectrometric determination

  • Upload
    adela

  • View
    213

  • Download
    1

Embed Size (px)

Citation preview

Page 1: Factorial designs for Cd, Cr, Hg, Pb and Se ultrasound-assisted acid leaching from human hair followed by atomic absorption spectrometric determination

FullPaper

Factorial designs for Cd, Cr, Hg, Pb and Se ultrasound-assisted acid

leaching from human hair followed by atomic absorption

spectrometric determination

Pilar Bermejo-Barrera, Antonio Moreda-PinÄeiro and Adela Bermejo-Barrera

Department of Analytical Chemistry, Nutrition and Bromatology, Faculty of Chemistry,University of Santiago de Compostela, Avenida das Ciencias s/n., E 15706,Santiago de Compostela, Spain

Received 27th August 1999, Accepted 15th November 1999

An acid leaching procedure, for use as a sample pre-treatment, was developed for Cd, Cr, Hg, Pb and Se

determination in human hair. A Plackett±Burman experimental design was used as a multivariate strategy for the

evaluation of the effects of varying several variables at once. The effects on acid leaching of metals of nitric acid,

hydrochloric acid and hydrogen peroxide concentrations, acid solution volume, exposure time to ultrasound,

temperature of the ultrasonic bath, and hair particle size, have been investigated. From these studies, certain

variables showed up as signi®cant, and they were optimised by a 23zstar central composite design, which involved

16 experiments. Optimum values of the variables were selected for the development of acid leaching procedures to

extract trace metals from human hair. Electrothermal atomic absorption spectrometry (ETAAS) was used to

determine Cd, Cr, Pb and Se, while cold vapour-atomic absorption spectrometry (CVAAS) was used for Hg

measurements. An acid digestion induced by microwave energy was used to obtain the total metal concentration

and also for comparative purposes. IAEA-085 and NIES No. 13 certi®ed reference materials, which offer certi®ed

values for some elements, were used in order to verify the accuracy of the methods.

Introduction

Interest in human hair as a clinical sample has increased inrecent years due to certain advantages offered by human hairover other clinical specimens such as blood or urine samples.These advantages have been mentioned by Burguera et al.1 andthe most important is the ten-fold higher metal concentrationsin hair than are present in blood or urine. In forensic science,human hair has been demonstrated to be one of the most usefulclinical samples to assess drug consumption, so drugs of abuseand/or metabolites analysis in human hair is now wellestablished and the methods are recommended.2 However,human hair analysis for metals is not a routine methodology;this can be due to the fact that metals reference values forhuman hair are not available. External contamination by dustand sweat, and cosmetic treatments, are other drawbacks thatwe have to take into account when metals determination inhuman hair is performed. These limitations can be solved usingstandard procedures accepted by the international communityfor both hair sampling and hair washing processes. Theserecommendations have been summarised by Bencze.3 Never-theless, human hair has been used to assess mercury(methylmercury) impregnation in the course of environmentalmonitoring,4 and according to Caroli et al.5 the use of humanhair as clinical sample must be exploited.

Different sample pre-treatments for element human hairanalyses, such as acid digestion procedures,1,6±15 have beendeveloped. The slurry sampling technique16±22 has also beenused since the ®rsts applications carried out by Kitagawa etal.23 and Ebdon et al.24 Other procedures such as enzymatichydrolysis processes are nowadays being studied, althoughmany limitations have been found to extract some metals.25

The extraction of elements by the action of an acid and/or anoxidant agent, named as acid leaching, appears to be aninteresting sample pre-treatment that must be developed.Leaching is not a total decomposition, and leachable recoveriesof analytes are generally lower than total concentrations.

Therefore, recoveries can only achieve total values if theelements are completely soluble in the leaching solvent.26

The main advantage of this sample pre-treatment appears to bethe lower acid concentrations than are required by aciddigestion procedures to achieve extraction of metals whichleads to small consumption of mineral acids and a reduction innitrous vapours. In addition, the time required for samplepreparation is reduced. Sediment, sludges and soils have beensubjected to microwave assisted and conventional leachingusing EPA method 3035B.26 Puchyr and Shapiro27 havedeveloped an acid leaching process to extract several elementsfrom seafoods and foods with high sugar content, while ElAzouzi et al.28 applied an acid leaching induced by ultrasoundto extract metals from mussel samples. Acid leaching processesfor speciation analysis have been also used,29 and in this sense,the method developed by WestoÈoÈ 30 in order to extract methyl-mercury from biological matrices can be considered the ®rstapplication of acid leaching procedures.

Several means of optimisation can be used but some methodsare time-consuming: for instance, the study of each variableseparately. In addition, some optimisation methods, such assimplex procedures, are restrictive in the number of variablesunder study. As many variables are involved throughout theacid leaching process, experimental designs are powerful toolsto optimise a certain procedure. Among the different groups ofdesigns, Plackett±Burman designs (PBDs), introduced in 1946by Plackett and Burman,31 allow us to discover the mostsigni®cant variables for a certain system with only fewexperiments. They are used as a screening method in orderto select the variables that have in¯uence on a system but theydo not give the optimum value for each variable. PBDsconstitute a variation on saturated fractional designs, allowingthe evaluation of either system with few experiments; k factorscan be studied in kz1 runs (only the main effects areestimated). These designs can be used only when kz1 is amultiple of 4 (i.e., k~3, 7, 11, ¼). This enlarges theopportunities offered by the saturated factorial design, which

J. Anal. At. Spectrom., 2000, 15, 121±130 121

This journal is # The Royal Society of Chemistry 2000

Publ

ishe

d on

10

Febr

uary

200

0. D

ownl

oade

d by

Lom

onos

ov M

osco

w S

tate

Uni

vers

ity o

n 28

/11/

2013

16:

45:3

3.

View Article Online / Journal Homepage / Table of Contents for this issue

Page 2: Factorial designs for Cd, Cr, Hg, Pb and Se ultrasound-assisted acid leaching from human hair followed by atomic absorption spectrometric determination

it complements.32 Another characteristic of Plackett±Burmandesigns is the orthogonality property. In order to obtain theoptimum values for each variable involved in a certain systemcentral composite designs (CCDs) are the most widely useddesign framework for second-order RS modelling within kfactor experiments.33,34 These design structures are based onfull, or fractional, two-level factorial designs by centre pointreplication and inclusion of an axial portion.35 The optimisa-tion procedures based on these approaches are nowadays beingapplied to optimise some sample pre-treatments36±39 and someoperating conditions for some analytical techniques.40±48

The aim of the present work has been the optimisation ofaccelerated, ultrasound assisted, acid leaching procedures toextract Cd, Cr, Hg (Me-Hg), Pb and Se from human hair. Dueto there being many variables that affect the acid leachingprocess, experimental designs, such as Plackett±Burman andcentral composite designs, have been used throughout theoptimisation.

Experimental

Apparatus

A Perkin Elmer Model 1100B (Norwalk, CT, USA) atomicabsorption spectrometer, equipped with a graphite furnaceHGA-400, an autosampler AS-40 and a deuterium lamp asbackground correction system, was used for Cd, Cr, Pb andSe measurements. A Perkin Elmer Model 4100ZL atomicabsorption spectrometer equipped with a Perkin Elmer FIAS400 system with a ®ve-port ¯ow-injection valve was used for Hgdeterminations. The rotation speed of the two multi-channelperistaltic pumps was programmed and automatically con-trolled by a separate PC. Hollow cathode lamps (Perkin Elmer)operating at recommended current were used for all cases,except for Pb and Se, for which EDLs, connected to a powersupply (Perkin Elmer), were used. For Hg, an EDL (System 2),connected to a power supply (System 2) from Perkin Elmer wasemployed. A Raypa1 Model UCI-150 ultrasonic cleaner bathfrom R. Espinar S.L. (Barcelona, Spain), programmable fortemperature ranging from 0 to 90 ³C, and time up to 15 min,with a frequency of 35 kHz for the ultrasound energy and atotal volume of 4 l, was used to induce the acid leachingprocess. A vibrating ball mill, Retsch (Haan, Germany),equipped with zircon cups (15 ml in size) and zircon balls(7 mm diameter) was used to pulverise and to reduce theparticle size of the human hair samples. A laser diffractionspectrometer, Coulter Series LS100, Fraunhofer optical modelparticle sizer (Coulter Electronics, Hialeah, FL, USA) was usedto obtain the particle size of human hair samples. An Orto-Alresa centrifuge (Barcelona, Spain) was also used in order toreach the separation between the solid human hair and the acidleachates liquid phase. A Panasonic domestic microwave oven(Osaka, Japan), programmable for time and microwave powerfrom 100 to 900 W, was used for total digestion of samples. Thepoly(tetra¯uorethylene) (PTFE) bombs were laboratory-madeand hermetically sealed, and proved adequate for work at lowpressures. The PTFE bombs were cleaned by addition of 2 mlof concentrated nitric acid, and subjected to a microwaveenergy of 500 W for 3 min. The nitric acid was discharged aftereach cycle and the bombs were rinsed three times withultrapure water. This procedure was repeated three times.Finally, a new cleaning microwave cycle was carried out andthe nitric acid, after dilution to 10 ml, was analysed for eachelement. Low blank values were obtained for all cases so anadequate cleanliness of the bombs must have been reached.After each acid digestion of human hair this cleaning procedurewas repeated, for each bomb, in order to avoid contaminationbetween samples. In addition, reagent blanks were alwaysobtained for each microwave cycle.

Reagents

Chemicals were of ultrapure grade, using ultrapure water,resistance 18 MV cm21, which was obtained from a Milli-Qpuri®cation device (Millipore Co., Bedford, MA, USA).Cd(NO3)2, Cr(NO3)3, Pb(NO3)2 and Se(NO3)4 stock standardsolutions, 1.000 g l21, were supplied by Merck (Darmstad,Germany). Hg(NO3)2 stock standard solution, 1.000 g l21, wasobtained from Panreac (Barcelona, Spain). Mg(NO3)2 stockstandard solution, 2.000 g l21, used as a chemical modi®er, wasprepared from Mg(NO3)2 (Merck Ltd., Poole, Dorset, UK). Pdstock standard solution, 3.000 g l21, used as a chemicalmodi®er, was prepared from Pd 99.999% (Aldrich, Milwaukee,WI, USA). Nitric acid 70.0% was obtained from J. T. BakerB.V. (Deventer, The Netherlands). Hydrogen peroxide 33%was supplied by Panreac (Barcelona, Spain). Hydrochloricacid, 37%, was obtained from J. T. Baker B. V. and acetone,99.7%, from Carlo Erba, Milan, Italy. Sodium tetrahydrobo-rate solution, 0.01% (m/v) was prepared from sodiumtetrahydroboarte reagent (Aldrich) and it was used as reducingagent in Hg CVAAS determinations. This solution wasprepared daily by dissolution in 0.5% (m/v) sodium hydroxidesolution (prepared from sodium hydroxide supplied by CarloErba, Milan, Italy). IAEA 085 was obtained from theInternational Atomic Energy Agency (Monaco), and NIESNo. 13 from the National Institute for Environmental Studies,Japan Environment Agency (Japan).

Human hair samples

Human hair samples were taken from healthy people (lengthvaried between 2 and 3 cm), using stainless steel scissors. TheInternational Atomic Energy Agency (IAEA) hair washingprocess49 was carried out in order to provide an accurateassessment of endogenous metal content. For some experi-ments, the hair samples were pulverised in a vibrating zirconiaball mill for 20 min using a power of 75%. In order to study theparticle size achieved after this procedure, studies weredeveloped to obtain the particle size distribution after eachpulverisation time. Laser diffraction spectrometry was used,and mean particle sizes around 20 mm were reached after thistreatment. The hair samples were also cut in 1±3 cm fragmentsin order to study the effect of the particle size on the acidleaching process. In all cases, the hair samples were put intopre-cleaned polyethylene vials.

Microwave acid digestion procedure

An acid digestion induced by microwave energy was used inorder to discover the total content of the metals and also forcomparative purposes. It is well known that from a safetypoint of view, the use of domestic microwave ovens is notrecommended, and commercial systems with both temperatureand pressure control must be considered. However, the use ofdomestic microwave ovens can be a useful and safe methodol-ogy under certain conditions, such as the use of low microwaveenergy and shorted times, and carrying out the digestionprocess in an extractor hood in order to reduce the fumes in thework environment. These considerations were taken intoaccount in our case because a domestic microwave oven wasused. Details of the optimised acid digestion procedure using adomestic microwave oven, laboratory-made low pressurePTFE bombs, and nitric acid and hydrogen peroxide asreagents, are given in ref. 25. The acid digests, made up to 10 mlwith ultrapure water, were kept into polyethylene vials at 4 ³Cbefore measurements.

Acid leaching procedure

Samples, 0.2±0.3 g, of hair were directly weighted intocentrifuge tubes, 5 ml of the acid±oxidant solution were

122 J. Anal. At. Spectrom., 2000, 15, 121±130

Publ

ishe

d on

10

Febr

uary

200

0. D

ownl

oade

d by

Lom

onos

ov M

osco

w S

tate

Uni

vers

ity o

n 28

/11/

2013

16:

45:3

3.

View Article Online

Page 3: Factorial designs for Cd, Cr, Hg, Pb and Se ultrasound-assisted acid leaching from human hair followed by atomic absorption spectrometric determination

added and, after slight homogenisation by mechanical stirring,the tubes were placed inside the ultrasonic water-bath and weresubjected to ultrasound energy at 35 kHz for 10 min. Thetemperature of the water-bath was ®xed at 90 ³C. The solutionwas centrifuged at 3000 rpm for 10 min and the acid liquidphase was separated with a Pasteur pipette. A 2 ml volume ofultrapure water was added to clean the solid residue and themixture was centrifuged again at 3000 rpm for 10 min. Theliquid phase was again separated with a Pasteur pipette andcombined with the acid leachate. The solution was ®nally madeup to 10 ml with ultrapure water and placed into a polyethylenebottle at 4 ³C.

ETAAS determinations

Cd, Cr, Pb and Se determinations were carried out by ETAASunder optimum conditions (Table 1). Previous studies werereferred to in order to obtain the optimum pyrolysis andatomization temperatures. Pd(NO3)2 was used as a chemi-cal modi®er for Pb determination, while Mg(NO3)2, also achemical modi®er, was used to determine Cr. A solutioncontaining Pd(NO3)2±Mg(NO3)2 was used for Cd and Sedeterminations. Portions of the acid leachate were mixed withappropriate volumes of the chemical modi®er solution directlyinto the autosampler cups, and 20 ml were injected into thegraphite furnace. That gives a total mass of 0.4 mg of Pd(NO3)2,in the Pb determination, and also 0.4 mg of Mg(NO3)2 forthe determination of Cr, while total masses of 0.4 and 0.3 mg,for Pd(NO3)2 and Mg(NO3)2, respectively, were injected todetermine Cd, and masses of 6.0 and 5.0 mg of Pd(NO3)2 andMg(NO3)2, respectively, were sampled for Se measurement.Aqueous calibration was a real possibility for Cd, Cr, Hg andPb, while the standard addition technique was required for Sedetermination.

CVAAS determination

Hg determination was carried out by use of a cold vapourgeneration±¯ow injection system on line to atomic absorptionspectrometry, with an electrically heated quartz cell, usingsodium tetrahydroborate as reducing agent and hydrochlorideacid as carrier solution. Optimum conditions for the cold

vapour generation and optimum ¯ow injection parameters aregiven in Table 2. Aqueous calibration was used to develop themeasurements.

Calibration and sensitivity

Calibration and standard addition graphs were obtained foreach metal, and the matrix effect was observed only for Sedeterminations after the application of the F-test for acon®dence interval of 95%. Therefore, the standard additiontechnique is needed for the Se measurements, while aqueouscalibration can be applied to Cd, Cr, Hg and Pb determina-tions. However, as the acid composition of each acid leachate isdifferent, the standard addition technique was used throughoutthe development of Plackett±Burman and central compositedesigns. The mean and standard deviation, for N~9, of theslopes of the standard addition graph corresponding to eachmetal were 0.063¡0.008, 0.0080¡0.0010, 0.0071¡0.0011,0.0044¡0.007 and 0.0042¡0.0012 l mg21 for Cd, Cr, Hg, Pband Se, respectively, and good precision could be seen for thecalibration. The detection and quanti®cation limits, given by

LOD~3|s

m,

and

LOQ~10|s

m,

respectively, where s is the standard deviation of elevenmeasurements of a reagent blank and m is the slope of thecalibration or standard addition graph, were also obtained foreach case, LODs of 0.03, 0.04, 0.05, 0.73 and 0.31 mg g21 beingreached for Cd, Cr, Hg, Pb, and Se, respectively, and LOQsof 0.11, 0.12, 0.17, 2.43, and 1.04 mg g21 calculated for Cd,Cr, Hg, Pb, and Se, respectively. The characteristic mass,recommended by International Union of Pure and AppliedChemistry for ETAAS determinations,50 was 0.34¡0.02,3.5¡0.2, 10.5¡0.4, and 23.4¡0.6 pg, for Cd, Cr, Pb and Se,respectively (the results as x+s; where x is the mean and s is thestandard deviation for n~4 measurements).

Table 1 ETAAS operating conditions and graphite furnace programa for the determination of Cd, Cr, Pb and Se in acid leachates from human hair

Cd Cr Pb Se

Operating conditionsWavelength/nm 228.8 357.9 283.3 196.0Slit/nm 0.7 0.7 0.7 2.0Background correction D2 Ð D2 D2

Injection volume/ml 20 20 20 20

aPyrolytic coated graphite tubes with L'vov platforms; integrated absorbance measurement.

Graphite furnace temperature programs Step Temperature/³C Ramp/s Hold/s Ar ¯ow/ml min21

Cdd Drying 150 15 20 300Pyrolysis 600 10 15 300Atomization 1600 0 2 0 (READ)Cleaning 2200 2 3 300

Crb Drying 150 15 20 300Pyrolysis 1200 10 15 300Atomization 2500 0 5 0 (READ)Cleaning 2650 1 2 300

Pbc Drying 150 15 20 300Pyrolysis 800 15 10 300Atomization 1600 0 3 0 (READ)Cleaning 2200 2 3 300

Sed Drying 150 15 20 300Pyrolysis 1200 15 10 300Atomization 2100 0 5 0 (READ)Cleaning 2650 2 2 300

bMg(NO3)2 as chemical modi®er for Cr (0.4 mg). cPd(NO3)2 as chemical modi®er for Pb (0.4 mg). dMg(NO3)2±Pd(NO3)2 as chemical modi®erfor Cd (0.3±0.4 mg) and Se (5.0±6.0 mg).

J. Anal. At. Spectrom., 2000, 15, 121±130 123

Publ

ishe

d on

10

Febr

uary

200

0. D

ownl

oade

d by

Lom

onos

ov M

osco

w S

tate

Uni

vers

ity o

n 28

/11/

2013

16:

45:3

3.

View Article Online

Page 4: Factorial designs for Cd, Cr, Hg, Pb and Se ultrasound-assisted acid leaching from human hair followed by atomic absorption spectrometric determination

Results and discussion

Seven factors were selected to be examined. The factors andtheir levels, low (2) and high (z), are described in Table 3.Three factors are related to the acid solution composition,which are nitric acid [HNO3], hydrochloric acid [HCl], andhydrogen peroxide [H2O2]. Other factors correspond to theultrasonic stirring, such as the ultrasonic water-bath tempera-ture and the exposure time to ultrasound, which are symbolizedby T and t, respectively. The acid solution volume, V, and thehair particle size, W, are also factors to be taken into account.The ultrasound energy was ®xed at a frequency of 35 kHz forall experiments. Firstly, an additional variable, frequencyenergy, with low level at 0 (without ultrasonic stirring) and highlevel at 35 kHz, was studied. The results obtained showed thatthis variable was the most signi®cant, and its importance led toincorrect values for the other variables (the other variablesinvolved in the process being insigni®cant). So, we decided thatultrasonic stirring was necessary and we ®xed the ultrasoundenergy at the maximum allowed by the ultrasonic water-bath,35 kHz.

The effect of changing a factor from a low level to a highlevel value was examined on a selected response such aspercentage recovery, according to the following equation:

Recovery~� �acid leaching

� �acid digestion

|100

where [ ]acid leaching is the metal concentration obtained after theacid leaching procedure (each experiment, 1±12 in Table 4, or1±16 in Tables 5 and 6) and [ ]acid digestion is the metalconcentration found after an acid digestion induced bymicrowave energy. A recovery close to 100% would showquantitative extraction of metals. As ®ve metals are studied(Cd, Cr, Hg, Pb and Se), ®ve responses will be obtained. Bothmetal concentrations after acid leaching and acid digestionprocedures were measured using the standard additiontechnique.

Plackett±Burman designs

The original Plackett±Burman approach is based on balancedincomplete blocks51 and suggest designs for eight, twelve,sixteen, etc., variables or factors. For the evaluation of sevenfactors at two levels a Plackett±Burman design with only twelveexperiments is described instead of the 27~128 required for afull factorial design. In this work, a folded Plackett±Burman2763/32, Type III resolution design, with four degrees offreedom and twelve runs, was carried out using the Stat-graphics Version 5.0 routine.52 The Plackett±Burman matrix isshown in Table 4, where the low (2) and high (z) levels arethose speci®ed in Table 3. The results of the design (%recovery) are also shown in Table 4 and visualised by using astandardised (P~95.0%) main effect Pareto chart, Fig. 1,which is a combination of the individual Pareto charts for eachmetal. From inference tests, the results produced a minimum tvalue at the 95.0% con®dence interval of 2.8, and we considerthat a factor is signi®cant when the t value for a certain factor ishigher than 2.8.

Sometimes, the effect of some factors (less signi®cant) canlead to wrong results on the importance of the most signi®cantvariables. In addition, the evaluations of interactions betweenfactors are also important. Such interactions are not allowed bya design of this type, and to evaluate them, the effects of somevariables were omitted, and the effect of the signi®cant factorsand interactions between signi®cant factors were again studied.After the study of results, shown as combined standardised(P~95%) two-factor interactions Pareto charts, in Fig. 2(a, b),and combined standardised (P~95%) main factor Pareto chartin Fig. 1, we can make the following comments.

In¯uence of the acid solvent composition. From the results ofthe Plackett±Burman design (Fig. 1), as well as centralcomposite design (which will be developed later), it is clearlyseen that high concentrations of nitric and hydrochloric acidsprovide a signi®cantly higher recovery, which is close to 100%for some responses. An increase of nitric acid concentrationleads to high Cr, Pb and Se acid leaching ef®ciency, whilehydrochloric acid concentration controls the Cd, Cr, Hg andPb acid leaching. Hydrogen peroxide concentration appears tobe insigni®cant for the acid leaching of all of the metals understudy. In addition, two order interactions between nitric acidconcentration and acid solvent volume ([HCl]/V) are found tobe signi®cant (negative sign) for the Hg acid leaching (Fig. 2b),while two order interactions between hydrochloric acid andnitric acid concentrations ([HCl]/[HNO3]) are signi®cant for Crand Pb and the two order interaction between hydrochloricacid concentration and temperature ([HCl]/T) is signi®cantonly for Pb.

In¯uence of the acid solvent volume. Acid solvent volume wasa factor that did not produce any signi®cant effect for themetals, except for Hg [Fig. 1 and 2(b)]. As has been commentedbefore, the main mercury species in human hair is methylmer-cury, and this compound is easily leached with hydrochloricacid. This result is in accordance with those previously

Table 2 Hg cold vapor generation program for CVAASa

Step

Flow rate/ml min21

Time/s Valve Read FunctionPump 1

Pump 2

Carrier solution Reducing solution

1 10 9 5 Fill Sampling step2 0 9 5 Inject * Hg cold vapor generationaWavelength, 253.7 nm; slit, 0.7 nm; sample loop, 500 ml; argon ¯ow rate, 100 ml min21; quartz cell temperature, 20 ³C; measurement mode,peak height; carrier solution HCl, 0.1% (m/m); reducing solution NaBH4, 0.01% (m/v).

Table 3 Experimental ®eld de®nition for the Plackett±Burman andcentral composite designs

Variable Symbol Low level (2) High level (z)

HNO3 concentration/M [HNO3] 0.0 4.8HCl concentration/M [HCl] 0.0 4.8H2O2 concentration/M [H2O2] 0.0 2.4Acid solvent volume/ml V 3 7Ultrasonic water-bath

temperature/³C T 20 90Exposure time to

ultrasounds/min t 10 180Human haira

particle size/mm W 20 2500aThe hair sample mass was 0.3 g for all experiments.

124 J. Anal. At. Spectrom., 2000, 15, 121±130

Publ

ishe

d on

10

Febr

uary

200

0. D

ownl

oade

d by

Lom

onos

ov M

osco

w S

tate

Uni

vers

ity o

n 28

/11/

2013

16:

45:3

3.

View Article Online

Page 5: Factorial designs for Cd, Cr, Hg, Pb and Se ultrasound-assisted acid leaching from human hair followed by atomic absorption spectrometric determination

obtained for methylmercury acid leaching from human hair byconventional optimisation methods,21 which showed hydro-chloric acid concentration and hydrochloric acid volume as themost important variables.

In¯uence of temperature. The increase in the water-bathtemperature showed a highly signi®cant effect on the recoveryof all metals, except Hg. This result is in agreement with the factthat methylmercury is easily leached, and only stirring(provided by 35 kHz ultrasound) is required: no additional(heat) energy is necessary. However, more drastic conditionsthan for Hg are necessary in order to leach Cd, Cr, Pb and Sefrom human samples. Therefore, the temperature of theultrasonic water-bath was signi®cant for the acid leaching ofthese metals. The use of sonication at high temperaturescontrasts with the results obtained by El Azouzi et al.,27 whohave carried out an acid leaching process induced byultrasound on mussel samples at room temperature. Probably,the metals' linkages to human hair proteins are stronger thanthe corresponding linkages to soft tissues such as mussel.

In¯uence of exposure time to ultrasounds. In the range 10±120 min, the Cd, Cr, Hg, Pb and Se recoveries are not affected.This fact offers an important practical advantage due to thetime for the acid leaching can be shorted. In previous studiesdeveloped by El Azouzi et al.,27 the sonication time was about180 min.

In¯uence of the particle size. Particle size (varying between20 mm and 2±3 cm) was the only factor that did not produceany signi®cant effects on the recovery of all metals tested. It

would can be expected that the particle size played animportant rule on the metals acid leaching from a ®brousmatrix as human hair. The insigni®cance of the particle size canbe explained by the high energy supplied by the ultrasound(frequency of 35 kHz) and the temperature, which increased thecontact between the sample and the acid solvent. This,therefore, allows rapid human hair sample pre-treatmentbecause a previous pulverisation stage (required when theslurry sampling technique is used) can be omitted. The samplescan be subjected to the acid leaching process just after thewashing procedure.

In conclusion, three factors, nitric acid concentration[HNO3], hydrochloric acid concentration [HCl], and tempera-ture of the ultrasonic water-bath, T, were the most signi®cantvariables in the Cd, Cr, Pb and Se acid leaching. A second setof two factors, hydrochloric acid concentration [HCl] andacid solvent volume, V, was shown to be signi®cant for Hg(Me-Hg).

Final optimisation by central composite designs

Having screened out the variables that did not have asigni®cant effect on the response, the remaining three or twofactors were optimised to provide the maximum metalrecovery. A central 23zstar orthogonal composite designwith six degrees of freedom and involving 16 experiments wasperformed, optimising the variables [HNO3], [HCl] andtemperature of the ultrasonic water-bath, T, for the Cd, Cr,Pb and Se acid leaching, while a central 22zstar orthogonalcomposite design with four degrees of freedom and 10experiments47 was performed for the set [HCl]/V in the Hg

Table 4 Plackett±Burman design (n~12) for the signi®cant variables determination

Run [HNO3] [HCl] [H2O2] T t V w

Recovery (%)

Cd Cr Hg Pb Se

1 z 2 z 2 2 2 z 91.5 54.1 5.6 81.6 43.82 z z 2 z 2 2 2 11.1 97.6 86.0 110.5 88.73 2 z z 2 z 2 2 56.3 70.1 84.0 60.0 18.54 z 2 z z 2 z 2 108.5 106.1 42.6 102.7 93.85 z z 2 z z 2 z 9.8 97.5 85.6 111.5 83.16 z z z 2 z z 2 40.8 64.2 65.6 72.6 53.87 2 z z z 2 z z 19.7 75.6 87.2 89.2 70.08 2 2 z z z 2 z 22.5 14.5 2.4 52.2 23.19 2 2 2 z z z 2 22.2 12.7 49.3 68.5 53.7

10 z 2 2 2 z z z 77.5 49.6 56.8 83.8 23.811 2 z 2 2 2 z z 64.5 45.1 87.2 54.3 25.312 2 2 2 2 2 2 2 110.1 5.4 5.2 47.7 7.0

Table 5 Central 23zstar orthogonal composite design (n~16) for the set [HNO3]/[HCl]/T in the Cd, Cr, Pb and Se acid leaching

Run [HNO3]/M [HCl]/M Temperature/³C

Recovery (%)

Cd Cr Pb Hg

1 K0a K0

a K0b 77.3 37.9 85.4 35.1

2 2 2 2 18.2 0.0 11.7 28.33 z 2 2 36.4 10.5 64.8 24.94 2 z 2 77.3 25.3 70.0 17.65 z z 2 68.2 49.5 88.2 45.76 2 2 z 31.8 0.0 29.1 10.67 z 2 z 100.0 36.8 84.6 74.68 2 z z 120.0 36.8 78.5 21.39 z z z 86.4 89.5 100.0 92.6

10 2ac K0 K0 27.3 4.2 66.8 21.211 zad K0 K0 77.3 28.4 85.8 38.812 K0 2ac K0 63.6 13.7 64.3 3.513 K0 zad K0 13.6 78.9 75.3 24.714 K0 K0 2ae 68.2 13.7 65.2 7.115 K0 K0 zaf 81.8 49.5 93.9 95.416 K0 K0 K0 77.3 41.1 74.1 45.9aK0~2.4 M. bK0~55 ³C. c2a~20.689252 M. dza~5.489252 M. e2a~4.23043 ³C. fza~100.7696 ³C.

J. Anal. At. Spectrom., 2000, 15, 121±130 125

Publ

ishe

d on

10

Febr

uary

200

0. D

ownl

oade

d by

Lom

onos

ov M

osco

w S

tate

Uni

vers

ity o

n 28

/11/

2013

16:

45:3

3.

View Article Online

Page 6: Factorial designs for Cd, Cr, Hg, Pb and Se ultrasound-assisted acid leaching from human hair followed by atomic absorption spectrometric determination

(Me-Hg) acid leaching. Although any factor was signi®cant forCd, the set [HNO3]/[HCl]T was also considered. The factorsthat were shown to be insigni®cant by the Plackett±Burmandesign were ®xed at convenient values. The human hair masswas also 0.3 g for all runs.

The experimental ®eld de®nition for these designs is given inTable 3, while Tables 5 and 6 show the central compositedesign matrices together with the responses obtained for eachmetal. It must be observed that experiments 1 and 16 (Table 5for Cd, Cr, Pb and Se), and 1 and 10 (Table 6, for Hg) are thesame, being the centres of the design (central compositedesign). A study of the estimated response surfaces for eachpair of variables, [HNO3]/[HCl], [HNO3]/T and [HCl]/T, forCd, Cr, Pb and Se, and [HCl]/V for Hg, showed the optimumvalues for each variable and for each metal. In Fig. 3(a±f) someof these estimated response surfaces are shown. The commentsfor each case are the following.

Cadmium. It can be seen, from Fig. 3(a), that the acidleaching ef®ciency is increased when the temperature is higher,

achieving the optimum for the highest temperature value, while[HCl] values between 3.0 and 3.5 M were optimum. The highest90 ³C value for T was selected due to its being the maximumtemperature supplied by the ultrasonic water-bath. From bothresponse surfaces, [HNO3]/[HCl] and [HNO3]/T (not given),and in the same way as for temperature, an optimum for[HNO3] was not reached; the highest value tested (4.8 M) wasagain chosen. Higher nitric acid concentrations than 4.8 Mwere not studied because they can lead to acid digestionconditions. However, it can be seen in Table 4 that Cdrecoveries of about 100% were achieved, and these results wereconsidered satisfactory.

Chromium. The [HNO3]/[HCl] and [HCl]/T estimatedresponse surfaces for Cr [Fig. 3(b) and (c), respectively]showed that a maximum was not obtained for [HCl] and Tfactors, while a [HNO3] in the 4.0±4.5 M range can beconsidered as optimum. In this case, the acid leaching ef®ciencywas directly proportional to both factors, [HNO3] and T, and itpeaked at the upper levels tested.

Mercury (methylmercury). From Fig. 3(d), it appeared thatthe region corresponding to the experimental de®nition ®elddid not include the optimum for either [HCl] or V factors. Inthe same way, for Cd and Cr, new factorial designs for highervalues of both factors were not considered because higher [HCl]and V levels than the maximum values shown in Table 3 ledto over pressure into the test tubes and vapour losses wereobserved. In addition, Hg recoveries close to 100% wereattained (Table 6), so the results were considered satisfactory.

Lead and selenium. Similar results to those achieved for Cdhave been attained for Pb and Se. Maximum values were notincluded in the experimental regions for the factors [HNO3]and T, while an optimum can be seen for [HCl], as is shown in

Table 6 Central 22zstar orthogonal composite design (n~10) for theset [HCl]/V in the Me-Hg acid leaching

Run [HCl]/M V/ml Me-Hg recovery (%)

1 K0a K0

b 63.22 2 2 0.03 z 2 77.64 2 z 0.05 z z 109.16 2ac K0 0.07 zad K0 97.98 K0 2ae 37.79 K0 zaf 79.9

10 K0 K0 71.6aK0~2.4 M. bK0~5.0 ml. c2a~20.1874 M. dza~4.9874 M. e2a~2.843 M. fza~7.156 M.

Fig. 1 Standardised (P~95%) main effects Pareto chart for the Plackett±Burman 2763/32, type III resolution design.

126 J. Anal. At. Spectrom., 2000, 15, 121±130

Publ

ishe

d on

10

Febr

uary

200

0. D

ownl

oade

d by

Lom

onos

ov M

osco

w S

tate

Uni

vers

ity o

n 28

/11/

2013

16:

45:3

3.

View Article Online

Page 7: Factorial designs for Cd, Cr, Hg, Pb and Se ultrasound-assisted acid leaching from human hair followed by atomic absorption spectrometric determination

Fig. 3(e) and (f) for Pb and Se, respectively. The optimumvalues for [HCl] in the Pb and Se acid leaching were about3.5 M. In the same way, higher values for [HNO3] and T werechosen.

Therefore, the optimum values for the factors under studyare those shown in Table 7. In addition, the ®xed values for theinsigni®cant factors are also given.

Effect of sample mass on the acid leaching ef®ciency

The effect of the human hair mass on the acid leachingef®ciency was studied, using the optimum conditions shown inTable 7. Different sample masses in the 0.1±0.6 g range weretested, and results as recovery (%) are shown in Fig. 4. It can beseen that Cd, Cr, Pb and Se recoveries close to 100% were

obtained for all masses tried. Therefore, the human hair massdoes not affect the acid leaching ef®ciency, and the same acidleaching conditions can be used for all masses. This offers apractical advantage in that sensitivity can be increased byincreasing the sample mass (up 0.6 g). However, Hg recoverieslower than 100% were reached for human hair masses higherthan 0.3 g. This different behaviour observed for the acidleaching of Hg from high human hair masses can be attributedto the effect of the volume of the acid solvent. Because of thesigni®cance of this factor, the acid solvent volume must beincreased when the sample mass is higher. Therefore, the Hgacid leaching ef®ciency decreases for high sample masses whena ®xed (optimum) acid solvent volume is used. This suggeststhat only certain mass : acid solvent volume ratios are suitable.Therefore, the Hg acid leaching is quantitative for masses

Fig. 2 Standardised (P~95%) two-factor interactions, [HNO3]/[HCl]/T (a) and [HCl]/V (b) Pareto charts for the Plackett±Burman 2763/32, type IIIresolution design.

Table 7 Optimum acid leaching conditionsa,b

Cd Cr Pb Se

[HNO3]/M 4.8 3.323.8 4.8 4.8[HCl]/M 3.3±3.8 4.8 3.3±3.8 3.3±3.8Temperature/³C 90 90 90 90

a[H2O2], 0.5 M; volume, 5 ml; sonication time, 10 min; particle size, 2±3 cm; human hair mass, 0.2 g.

Me-Hg

[HCl]/M 4.8V/ml 7.0b[H2O2], 0.5 M; [HNO3], 0.5 M; ultrasonic water-bath temperature, 15 ³C (room temperature); sonication time, 10 min; particle size, 2±3 cm,human hair mass, 0.2 g.

J. Anal. At. Spectrom., 2000, 15, 121±130 127

Publ

ishe

d on

10

Febr

uary

200

0. D

ownl

oade

d by

Lom

onos

ov M

osco

w S

tate

Uni

vers

ity o

n 28

/11/

2013

16:

45:3

3.

View Article Online

Page 8: Factorial designs for Cd, Cr, Hg, Pb and Se ultrasound-assisted acid leaching from human hair followed by atomic absorption spectrometric determination

between 0.1 and 0.3 g, so we have chosen 0.2 g as arepresentative human hair mass due to the good sensitivitythat is reached for metals, and this mass will be used in futureexperiments.

Precision and accuracy

The precision of the acid leaching process and ETAAS/CVAAS determination was evaluated through the analysis ofa human hair sample, which was subjected to the acidleaching procedure (under optimum conditions for each caseaccording to Table 7) eleven times. Cd, Cr, Hg, Pb and Sewere determined in each acid leachate and the results,expressed as RSDs (%) were 8.6, 8.0, 6.0, 7.1 and 8.3%,respectively.

Two human hair reference materials, IAEA 085 and NIESNo. 13, were analysed in order to assess the accuracy of themethods. IAEA 085 offers certi®ed contents only for Hg andMe-Hg, and an informative value for Se. NIES No. 13certi®es all elements, except Cr. They were subjected threetimes to the optimum acid leaching conditions according to

Table 7 and each leachate was analysed also three times foreach metal. The results are listed in Table 8 and they were inaccordance with the certi®ed contents for all cases, except forSe in NIES No. 13, for which the found value,1.55¡0.01 mg g21, is close to the certi®ed concentrationrange, 1.79¡0.17 mg g21, so approximately 86.6% of theselenium content in NIES No. 13 was extracted. In addition,it must be said that the Hg concentration found after theapplication of the proposed method was in agreement withthe Me-Hg certi®ed content for both IAEA 085 and NIESNo. 13. This con®rms that only organic mercury is leachableby the action of hydrochloric acid.

Application

Five human hair samples from healthy people were analysed bythe optimised ultrasound assisted acid leaching procedure.Each human hair sample was prepared as acid leachates twice,and also two acid digests by the microwave assisted aciddigestion (reference method) were prepared. Each acid leachateor acid digest was subjected to ETAAS/CVAAS twice. Resultsare listed in Table 9 as x+s, where x is the mean and s is thestandard deviation for n~4 measurements. The mean metalrecoveries, also expressed as x+s, where x is the mean and s isthe standard deviation for n~4 measurements, are given. Itmust be said that as two acid leachates and two acid digestswere obtained from the same hair sample, two metalconcentrations were obtained after the acid leaching and twoothers after the acid digestion process. By this means, and togive the recovery range for each metal in each sample, the metalrecovery, based on the equation

Recovery~� �acid leaching

� �acid digestion

|100

was calculated for the metal concentration in each acid leachate(two) using the metal concentration in each acid digest (alsotwo). This then gives us four recoveries for each metal in eachsample, and the recovery ranges, expressed as x+s, are listed inTable 9.

Finally, it can be seen that metal recoveries close to 100%

Fig. 4 Effect of the human hair sample mass on the element recoveryfrom acid leachates: Cd (Y), Cr (#), Hg (&), Pb ($) and Se (%).

Fig. 3 Some estimated response surfaces from the central compositedesign: Cd, [HCl]/T (a), Cr [HCl]/[HNO3] (b), Cr [HCl]/T (c), Hg [HCl]/V (d), Pb [HCl]/T (e) and Se [HCl]/T (f).

Table 8 Analysis of certi®ed reference materials

IAEA 085 NIES No. 13

Certi®ed valuea/mg g21 Found valueb/mg g21 Certi®ed value/mg g21 Found valueb/mg g21

Cd Ð Ð 0.23¡0.03 0.25¡0.01Total Hg 22.0±24.4 4.42¡0.20

22.8¡0.2 3.9¡0.2(Me-Hg) (21.5±24.3) (3.8¡0.4)Pb Ð Ð 4.6¡0.4 4.4¡0.1Se 1.1c 1.20¡0.01 1.79¡0.17 1.55¡0.01aFrom ref. 53. bExpressed as x+s for N~9 measurements. cInformative value.

128 J. Anal. At. Spectrom., 2000, 15, 121±130

Publ

ishe

d on

10

Febr

uary

200

0. D

ownl

oade

d by

Lom

onos

ov M

osco

w S

tate

Uni

vers

ity o

n 28

/11/

2013

16:

45:3

3.

View Article Online

Page 9: Factorial designs for Cd, Cr, Hg, Pb and Se ultrasound-assisted acid leaching from human hair followed by atomic absorption spectrometric determination

were obtained for all metals, except for Se, yielding goodagreement between both procedures (ultrasound±acid leachingand microwave assisted±acid digestion).

Conclusions

Rapid optimisation of sample pre-treatments that are affectedby many factors, such as the ultrasound assisted acid leachingprocess, can be achieved by experimental design. Theapplication of Plackett±Burman designs, as factor screening,shows that hydrochloric and nitric acid concentrations, and thetemperature of the ultrasonic water-bath, are the mostsigni®cant variables in Cd, Cr, Pb and Se acid leaching,while only hydrochloric acid and acid solvent volume controlsthe Hg acid leaching and only organic mercury, mainlymethylmercury, is extracted. Ultrasound stirring was foundnecessary in order to obtain quantitative extractions; however,the exposure time to ultrasound was insigni®cant. This was dueto the fact that the low level tested, 10 min, was suf®cient toobtain quantitative results. This value for the exposure time toultrasound is lower than previously reported and was ®xed asoptimum. Under the optimum conditions found, quantitativerecoveries for all metals, except Hg, are reached, and the resultsobtained are comparable to those obtained by means ofclassical sample pre-treatment based on acid digestion inducedby microwave energy. Thus, the proposed procedure is moreattractive due to a minimum reagents cost and reduced time(times of less than 30 min compared with the acid digestionprocess, about 1 h) and also due to the reduction of nitrousvapours produced.

References

1 J. L. Burguera, M. Burguera, C. E. RondoÂn, C. Rivas,J. A. Burguera and O. M. AlarcoÂn, J. Trace Elem. ElectrolytesHealth Dis., 1987, 1, 21.

2 M. Chiarotti, Forensic Sci. Int., 1993, 63, 161.3 K. Bencze, Fresenius' J. Anal. Chem., 1990, 338, 58.4 WHO, Environmental Health Criteria 101-Methyl-mercury, World

Health Organisation, Geneva, 1990.5 S. Caroli, A. Alimonti, E. Coni, F. Petrucci, O. Senofonte and

N. Violante, Crit. Rev. Anal. Chem., 1994, 24, 363.6 H. Matusiewicz, J. Anal. At. Spectrom., 1989, 4, 265.7 H. Matusiewicz, A. Suszka and A. Ciszewski, Acta. Chim. Hung.,

1991, 128, 849.8 P. Quevauviller, J.-L. Imbert and M. Olle, Mikrochim. Acta, 1993,

112, 147.9 I. Harrison, D. Littlejohn and G. S. Fell, J. Anal. At. Spectrom.,

1995, 10, 215.10 J. Kubova, V. Hanakova, J. Medved and V. Stresko, Anal. Chim.

Acta, 1997, 337, 329.

11 A. Ciszewski, W. Wasiak and W. Ciszewska, Anal. Chim. Acta,1997, 343, 225.

12 T. H. Nguyen, J. Boman, M. Leermarkers and W. Baeyens,Fresenius' J. Anal. Chem., 1998, 360, 199.

13 P. Canada-Rudner, A. GarcõÂa-de-Torres, J. M. Cano-PavoÂn andE. RodrõÂguez-CastelloÂn, J. Anal. At. Spectrom., 1998, 13, 243.

14 E. Buseth, G. Wibetoe and I. Martinsen, J. Anal. At. Spectrom.,1998, 13, 1039.

15 M. Yin and B. Li, Spectrochim. Acta, Part B, 1998, 53B, 1447.16 P. Bermejo-Barrera, A. Moreda-PinÄeiro, J. Moreda-PinÄeiro and

A. Bermejo-Barrera, J. Anal. At. Spectrom., 1997, 12, 301.17 P. Bermejo-Barrera, A. Moreda-PinÄeiro, J. Moreda-PinÄeiro and

A. Bermejo-Barrera, Anal. Chim. Acta, 1997, 349, 319.18 P. Bermejo-Barrera, A. Moreda-PinÄeiro, J. Moreda-PinÄeiro and

A. Bermejo-Barrera, Talanta, 1998, 45, 1147.19 P. Bermejo-Barrera, A. Moreda-PinÄeiro, J. Moreda-PinÄeiro and

A. Bermejo-Barrera, Mikrochim. Acta, 1998, 128, 122.20 P. Bermejo-Barrera, A. Moreda-PinÄeiro, J. Moreda-PinÄeiro and

A. Bermejo-Barrera, Fresenius' J. Anal. Chem., 1998, 360, 707.21 P. Bermejo-Barrera, A. Moreda-PinÄeiro, J. Moreda-PinÄeiro and

A. Bermejo-Barrera, Fresenius' J. Anal. Chem., 1998, 360, 712.22 P. Bermejo-Barrera, E. M. Verdura-Constenla, A. Moreda-PinÄeiro

and A. Bermejo-Barrera, Anal. Chim. Acta, 1999, in the press.23 K. Kitagawa, Y. Hotta and Y. Yasui, Anal. Sci., 1988, 4, 163.24 L. Ebdon, A. S. Fisher, H. G. M. Perry and A. A. Brown, J. Anal.

At. Spectrom., 1990, 5, 321.25 P. Bermejo-Barrera, S. FernaÂndez-Nocelo, A. Moreda-PinÄeiro and

A. Bermejo-Barrera, J. Anal. At. Spectrom., 1999, 14, 1893.26 E. M. L. Lorentzen and H. M. Kingston, Anal. Chem., 1996, 68,

4316.27 R. F. Puchyr and R. Shapiro, J. Assoc. Off. Anal. Chem., 1986, 69,

868.28 H. El-Azouzi, M. L. Cervera and M. de-la-Guardia, J. Anal. At.

Spectrom., 1998, 13, 533.29 J. Szpunar, V. O. Schmitt, O. F. X. Donard and R. èobinÂski,

Trends Anal. Chem, 1996, 15, 181.30 G. WestoÈoÈ , Acta Chem. Scand., 1968, 22, 2277.31 R. L. Plackett and J. P. Burman, Biometrika, 1946, 33, 305.32 R. A. Olivero, J. M. Nocerino and S. N. Deming, in Chemometrics

in Environmental ChemistryÐStatistical Methods, ed. J. Einax,Springer-Verlag, Berlin, Germany, 1995, pp. 95±96.

33 R. A. Olivero, J. M. Nocerino and S. N. Deming, in Chemometricsin Environmental ChemistryÐStatistical Methods, ed. J. Einax,Springer-Verlag, Berlin, Germany, 1995, pp. 98±99.

34 R. G. Brereton, in ChemometricsÐApplications of Mathematicsand Statistics to Laboratory Systems, Ellis Horwood, Chichester,Sussex, UK, 1990, ch. 2, pp. 65±83.

35 W. P. Gardiner and G. Gettinby, Experimental Design Techniquesin Statistical PracticeÐA Practical Software-based Approach, EllisHorwood, Chichester, Sussex, UK, 1998, ch. 11, pp. 335±338.

36 M. J. Blanco, R. Ribo , X. TomaÂs and J. Obiols, Anal. Proc., 1994,31, 353.

37 M. P. Llompart, R. A. Lorenzo and R. Cela, J. Microcolumn Sep.,1996, 8, 163.

38 M. P. Llompart, R. A. Lorenzo and R. Cela, J. Chromatogr. Sci.,1996, 34, 43.

39 I. Lavilla, B. PeÂrez-Cid and C. Bendicho, Fresenius' J. Anal.Chem., 1998, 361, 164.

Table 9 Element concentration in human hair samples (N~4)

Element Procedure 1 2 3 4 5

Cd/mg g21 Microwave digestion 0.026¡0.001 0.069¡0.001 0.067¡0.002 0.022¡0.001 0.024¡0.001Ultrasonic acid leaching 0.023¡0.002 0.085¡0.002 0.070¡0.003 0.019¡0.001 0.022¡0.001Recovery (%)a 92.4¡5.2 102.1¡2.1 96.3¡3.5 107.1¡1.1 110.7¡2.1

Cr/mg g21 Microwave digestion 0.36¡0.01 0.38¡0.01 0.34¡0.01 0.44¡0.01 0.35¡0.01Ultrasonic acid leaching 0.38¡0.01 0.42¡0.01 0.34¡0.02 0.47¡0.02 0.36¡0.01Recovery (%) 105.8¡1.1 109.0¡1.7 99.8¡0.4 106.0¡3.4 103.6¡2.0

Hg/mg g21 Microwave digestion 1.03¡0.21 0.74¡0.11 0.79¡0.15 0.82¡0.10 0.56¡0.06Ultrasonic acid leaching 0.94¡0.16 0.64¡0.13 0.73¡0.15 0.71¡0.17 0.43¡0.07Recovery (%) 89.2¡2.3 92.0¡3.1 93.5¡2.0 84.6¡1.7 91.1¡2.2

Pb/mg g21 Microwave digestion 2.43¡0.13 5.47¡0.13 2.43¡0.23 1.86¡0.24 1.82¡0.27Ultrasonic acid leaching 2.74¡0.25 5.51¡0.21 2.16¡0.10 2.26¡0.21 2.30¡0.20Recovery (%) 108.7¡4.2 100.7¡3.8 93.7¡2.9 108.1¡3.6 107.1¡2.3

Se/mg g21 Microwave digestion 0.39¡0.03 0.43¡0.02 0.27¡0.02 0.15¡0.01 0.45¡0.03Ultrasonic acid leaching 0.36¡0.02 0.34¡0.01 0.31¡0.01 0.18¡0.01 0.33¡0.02Recovery (%) 80.4¡3.5 70.0¡0.5 113.9¡4.5 118.0¡3.3 74.6¡4.9

aThe mean recovery corresponds to four recovery values obtained by application of the equation Recovery~([ ]acid leaching/[ ]acid digestion)6100,for each metal concentration in acid leachates with respect to each metal concentration in each acid digest (also N~4).

J. Anal. At. Spectrom., 2000, 15, 121±130 129

Publ

ishe

d on

10

Febr

uary

200

0. D

ownl

oade

d by

Lom

onos

ov M

osco

w S

tate

Uni

vers

ity o

n 28

/11/

2013

16:

45:3

3.

View Article Online

Page 10: Factorial designs for Cd, Cr, Hg, Pb and Se ultrasound-assisted acid leaching from human hair followed by atomic absorption spectrometric determination

40 V. Simeonov, A. Voulgaropoulos and M. Sofoniou, Fresenius'J. Anal. Chem., 1987, 329, 444.

41 M. M. Rogan, K. D. Altria and D. M. Goodall, Chromatographia,1994, 38, 723.

42 G. Lespes, F. Seby, P.-M. Sarrandin and M. Potin-Gautier,J. Anal. At. Spectrom., 1994, 9, 1433.

43 Y. Van der Heyden, K. Luypaert, C. Hartmann, D. L. Massart,J. Hoogmartens and J. de Beer, Anal. Chim. Acta, 1995, 312, 245.

44 Y. Van der Heyden, C. Hartmann, D. L. Massart, L. Michel,P. Kiechle and F. Erni, Anal. Chim. Acta, 1995, 316, 15.

45 S. Boonkerd, M. R. Detaevernier, Y. Van der Heyden,J. Vindevogel and Y. Michotte, J. Chromatogr. A, 1996, 736, 281.

46 I. Koch, C. F. Harrington, K. J. Reimer and W. R. Cullen,Talanta, 1997, 44, 771.

47 I. Koch, C. F. Harrington, K. J. Reimer and W. R. Cullen,Talanta, 1997, 44, 1241.

48 M. Legret and L. Divet, Analusis, 1998, 16, 97.49 Report on the Second Research Co-ordination Meeting of IAEA,

Neuherberg, Germany, 1985.50 A. M. Ure, L. R. P. Butler, B. V. L'vov, I. Rubeska and

R. Sturgeon, Pure Appl. Chem., 1992, 64, 253.51 W. P. Gardiner and G. Gettinby, Experimental Design Techniques

in Statistical PracticeÐA Practical Software-based Approach, EllisHorwood, Chichester, Sussex, UK, 1998, ch. 11, pp. 227±228.

52 Statgraphics Plus V. 5.0, Reference Manual, Manugistics Inc.,Rockville, MD, USA, 1992.

53 S. F. Heller-Zeisler, R. M. Pau and R. Zeisler, Fresenius' J. Anal.Chem., 1998, 360, 419.

Paper a906960g

130 J. Anal. At. Spectrom., 2000, 15, 121±130

Publ

ishe

d on

10

Febr

uary

200

0. D

ownl

oade

d by

Lom

onos

ov M

osco

w S

tate

Uni

vers

ity o

n 28

/11/

2013

16:

45:3

3.

View Article Online