tsomicpectroscopy - a and b spectroscopy 29(3).pdf · •u.s.$70.00includesfirst-class...

Issues alsoavailable

electronically.

(see inside front cover)

ASPND7 29(3) 77–114 (2008)ISSN 0195-5373

AtomicSpectroscopy

May/June 2008 Volume 29, No. 3

In This Issue:Determination of Trace Elements in Brazilian Human Milk by QuadrupoleInductively Coupled Plasma Mass Spectrometry and Microwave-assisted DigestionRosilene S. Nascimento, Denise B.C. Mendes, Judith Maria Gomes Matos,Júlio C.J. Silva, Virgínia S.T. Ciminelli, Waldomiro B. Neto,and José Bento B. Silva ....................................................................................................... 77

Comparative Study of X-Ray Fluorescence and Inductively Coupled PlasmaOptical Emission Spectrometry of Heavy Metals in the Analysis of Soil SamplesA.K. Krishna, K. Rama Mohan, N.N. Murthy, and P.K. Govil .......................................... 83

Determination of Oxygen in High Purity Cadmium by Radio FrequencyGlow Discharge Optical Emission SpectrometryG. Anil, M.R.P. Reddy, S.T. Ali, N.R. Munirathnam, and T.L. Prakash ............................ 90

Determination of Trace Copper in Biological Samples by On-line ChemicalVapor Generation-Atomic Fluorescence SpectrometryLiang He, Xiaofan Zhu, Li Wu, and Xiandeng Hou .......................................................... 93

Determination of Cu, Mn, and Pb in Yogurt Samples by Flame Atomic AbsorptionSpectrometry Using Dry, Wet, and Microwave Ashing MethodsGokce Kaya, Ismail Akdeniz, and Mehmet Yaman .......................................................... 99

Copper Determination by ETAAS in Fish Tissue Cytosols With MinimalSample PretreatmentZrinka Dragun and Biserka Raspor ................................................................................. 107

PerkinElmer is a registered trademark of PerkinElmer, Inc.SCIEX and ELAN are registered trademarks of MDS SCIEX, a division of MDS Inc.Milli-Q is a trademark of Millipore Corporation, Bedford, MA, USA.Pyrex is a registered trademark of Corning Glass Works.Suprapur is a registered trademark of Merck & Co., Darmstadt, Germany.SpectrAA and GTA are trademarks of Varian, Inc.Teflon is a registered trademark39 of E.I. duPont deNemours & Co., Inc.

Registered names and trademarks used in this publication, even without specific indicationthereof, are not to be considered unprotected by law.

Printed in the United States and published six times a year by:

PerkinElmer Life and Analytical Sciences710 Bridgeport AvenueShelton, CT 06484-4794 USAhttp://www.perkinelmer.com

AtomicSpectroscopy

Vol. 29(2) March/April 2008

Guidelines for AuthorsAtomic Spectroscopy serves asa medium for the disseminationof general information togetherwith new applications andanalytical data in atomicabsorption spectrometry.The pages of Atomic

Spectroscopy are open to allworkers in the field of atomicspectroscopy. There is no chargefor publication of a manuscript.The journal has around 1500

subscribers on a worldwidebasis, and its success can beattributed to the excellentcontributions of its authors aswell as the technical guidanceof its reviewers and theTechnical Editors.The original of the manuscript

can be mailed to the editor inhard copy including electronicfile on disk or CD (or simply bye-mail) in the following manner:1. If mailed, provide text (dou-ble-spaced) and tables in hardcopy plus on disk or CD withtext and tables in .doc file; fig-ures in doc or tif files.3. Number the references in the

order they are cited in the text.5. Consult a current copy ofAtomic Spectroscopy forformat.

6. Editor’s e-mail:[email protected]: [email protected]

All manuscripts are sent to tworeviewers. If there is disagreement,a third reviewer is consulted.Minor changes in style are made

in-house and submitted to theauthor for approval.If a revision of the manuscript

is required before publication canbe considered, the paper isreturned to the author(s) withthe reviewers’ comments.In the interest of speed of

publication, a pdf file of the type-set text is e-mailed to the corre-sponding author beforepublication for final approval.Once the issue has been printed,

each author receives a final pdf fileof the article including 50 compli-mentary copies of the article andseveral copies of the completeissue.Additional reprints can be

purchased, but the requestmust be made before printing.

PerkinElmer. Inc., holds copy-right to all material published inAtomic Spectroscopy unless other-wise noted on the first page of thearticle.

Anneliese LustEditor, Atomic SpectroscopyPerkinElmerLife and Analytical Sciences710 Bridgeport AvenueShelton, CT 06484-4794 USA

EDITORAnneliese LustE-mail:[email protected]

TECHNICAL EDITORSLaura J. Thompson, AADennis Yates, ICPKenneth R. Neubauer, ICP-MS

SUBSCRIPTION INFORMATIONAtomic SpectroscopyP.O. Box 3674Barrington, IL 60011 USAE-mail: [email protected]

2008 Subscription Rates• U.S. $70.00 includes first-classmail delivery worldwide$20.00 extra for pdf file.

• Payment by check (drawn on U.S.bank in U.S. funds) made out to:“Atomic Spectroscopy”

Electronic (pdf) FileSend request via e-mail to:[email protected]

Back Issues/Claims• Single back issues are availableat $15.00 each.

• Subscriber claims for missingback issues will be honoredat no charge within 90 daysof issue mailing date.

Address Changes to:Atomic SpectroscopyP.O. Box 3674Barrington, IL 60011 USA

Copyright © 2008PerkinElmer, Inc.All rights reserved.http://www.perkinelmer.com

UMI Microfilm Editions ofAtomic Spectroscopy issuesalso available from:National Archive Publishing Co.300 N. Zeeb RoadAnn Arbor, MI 48106-0998 USA

77

*Corresponding author.E-mail: [email protected]: +55 31 3499 5708Fax: + 55 31 3499 5700

Determination of Trace Elements in BrazilianHuman Milk by Quadrupole Inductively Coupled PlasmaMass Spectrometry and Microwave-assisted Digestion

Rosilene S. Nascimentoa, Denise B.C. Mendesc, Judith Maria Gomes Matosb, Júlio C.J. Silvab,Virgínia S.T. Ciminellib, Waldomiro B. Netoa, and José Bento B. Silvaa*

aDepartamento de Química, Universidade Federal de Minas Gerais, Belo Horizonte, 31270-901, MG, BrazilbDepartamento de Engenharia Metalúrgica e Materiais, Universidade Federal de Minas Gerais, MG, Brazil

cMaternidade Odete Valadares, Belo Horizonte, MG, Brazil

Atomic SpectroscopyVol. 29(3), May/June 2008

INTRODUCTION

Human milk is the ideal food fornewborns due to its compositionand availability. In addition tomacronutrients, it also contains vit-amins and elements that are essen-tial for newborns in their firstmonths of life. What is more, tissuecell synthesis occurs at a very highrate during infancy and, after aboutfive months, the infant’s weight istwice as much as the originalweight at birth (1). It is thereforeessential that all macronutrientsand micronutrients are contained inmilk at adequate quantities in orderto ensure an appropriate develop-ment of the infant’s functions,organs, and systems. However, thechemical composition of maternalmilk changes with postpartum timefrom colostrums milk (days 1–3) tomature milk (days 14–28 after deliv-ery). When infants are not breast-fed or if breast-feeding isdiscontinued very early, formulaeare used instead of human milk.Ideally, the composition of formulamilk should be as similar as possi-ble to that of maternal milk in eachparticular lactating stage (2-4).

Using an analytical method fordetermining such trace elements inhuman milk offers certain difficul-ties: the contents of the elementsare low (mg/mL); there is a risk ofcontamination during sample treat-ment; the matrix is complex; and,in most cases, the amount of avail-

ABSTRACT

A fast procedure to determine27Al, 53Cr, 55Mn, 57Fe, 60Ni, 65Cu,64Zn, 66Zn, 78Se, 82Se, 112Cd,114Cd, 206Pb, 208Pb, 138Ba, 51V,59Co, and 75As in Brazilian humanmilk by quadrupole inductivelycoupled plasma mass spectrome-try (Q-ICP-MS) is proposed.Mature and colostrum humanmilk samples were mineralizedusing closed-pressurized andhigh-performance microwave(MW) oven digestion. Aqueouscalibration was performed withan r2 higher than 0.99 for all iso-topes investigated. A referencematerial, Infant Formula (NISTSRM 1846), was analyzed to testthe accuracy of the proposedmethodology. In addition, recov-ery assays were carried out tocheck the interference effects.Recoveries ranged between80.8% and 110.9%. The precisionof the method, expressed as arelative standard deviation (RSD),was better than 3.4% for all ana-lytes investigated. Precision andaccuracy of the method showthat it is useful for the intendedpurpose.

and sensitivity (6). Due to the com-position of human milk, sample oxi-dation is necessary, which is oftencomplicated and involves the riskof contamination and of analyte lossby adherence to the walls of thecontainer or volatilization. Besidesthe classic dry mineralization (7-10)and wet digestion (11-16), a systemof wet digestion in closed Teflon®vessels and microwave heating isalso used in this work. Thisimproves the destruction of organicmatter, shortens the time neededfor digestion, and offers advantagessuch as minimum reagent volume,reduction of possible analyte lossesby volatilization or retention, andelimination of contamination risksfrom the environment (17).

Inductively coupled plasma massspectrometry (ICP-MS) has becomethe standard analytical techniquefor rapid multi-element analysis ofbiological samples (18-20). It pro-vides multi-element capability andextreme sensitivity and selectivity.The most common methods for ICPmulti-element analysis of milk sam-ples involve previous dry-ashing(21), hot plate digestion (22), ormicrowave-assisted digestion (23)of samples. When dealing with bio-logical samples, the advantages ofmicrowave digestion comparedwith other methods are evident inthe literature (24-27). Recently,microwave digestion has also beenused for human milk (28-33). Theconditions (i.e., the type andamount of acid added, the possibleuse of coadjutants, the time-powerprogram and the need foradditional steps) must be adaptedto the characteristics of theelements and to the specific study.

able sample is limited. In addition, atechnique for determining traceelements has to be highly sensitiveand interference-free, and the sam-ple treatment has to be simple andinvolve minimum handling andreagent use. The method shouldalso be fast and allow several ele-ments to be measured (5). Somenegative aspects associated withmilk analysis by direct nebulizationof aqueous solutions in inductivelycoupled plasmas are poor accuracy

posed method was applied to fivemature milk and five colostrummilk samples of approximately10 mL each. The samples were keptfrozen at -18 ºC until analysis.

Sample Preparation

Due to the fatty nature of milkand also due to the large averagedroplet size of untreated milk, ele-mental analysis by direct nebuliza-tion is hampered by poor accuracyand sensitivity (22-30). Althoughmany digestion techniques areavailable, closed microwave-assisted digestion has been usedfrequently over the last few years tomineralize biological samples (35-37), including milk (18,29,38).

In order to determine the con-centration of the elements studied,human milk aliquots of 1.0 mLwere transferred into Teflon® ves-sels followed by the addition of2.0 mL 65% HNO3 and 0.5 mL 30%H2O2 (Suprapur®, Merck, Darm-stadt, Germany). The samples werethen digested in a high-performance microwave (MW)oven (MLS 1200 MEGA, Milestone,Bergamo, Italy) with power stepsfrom 250 to 600 W. After cooling ofthe digest vials in a water bath for30 minutes, the completely clear,colorless, and homogeneous digestsobtained were diluted 1:50 withdeionized water. The heating pro-gram for digesting the human milksamples in the microwave oven issummarized in Table II.

A 0.05-g amount of the certifiedreference material NIST SRM 1846

Pressurized closed-vessel systemshave been used in this study: theboiling point of the reagents israised by the pressure producedinside the vessel. Since higher tem-peratures are reached in a closed-vessel system, digestion iscompleted in a very short time(34).

The aim of this study is to opti-mize and validate a method fordetermining Ba, V, Se, Cr, Ni, Mn,Pb, Cu, Co, Cd, As, Fe, Zn, and Alin human milk by means ofmicrowave mineralization of thesamples and measurement by ICP-MS.

EXPERIMENTAL

Instrumentation

A PerkinElmer SCIEX® ELAN®9000 quadrupole inductively cou-pled mass spectrometer (Q-ICP-MS,PerkinElmer SCIEX, Concord,Ontario, Canada) was used for allthe measurements. The instrumentwas equipped with a cross-flownebulizer, a double pass spraychamber (Scott-type), and an alu-mina injector with a 2.00-mm inter-nal diameter (i.d.). Argon (99.98%purity, Air Liquide, Contagem,Minas Gerais, Brazil) was also used.Instrument optimization was per-formed daily prior to analysis bymonitoring the signals produced bya multi-element solution containing10 µg L–1 Mg, Rh, and Pb in theGraphics mode of analysis. Theselected conditions were thosewhich maximize the 103Rh, 24Mg,and 208Pb signals, or they were sig-nals for an element of low mass,middle mass, and one of high mass.After that, the plasma was allowedto stabilize for 30 minutes beforeinstrument optimization for furtheranalysis. With regard to the analyti-cal conditions, some preliminarymeasurements were carried out toestablish the most convenientinstrumental parameters, as summa-rized in Table I. Human milk sam-ples were mineralized by means of

a closed, pressurized, high-perfor-mance microwave (MW) digestionoven (MLS 1200 MEGA, Milestone,Bergamo, Italy) equipped with 10Teflon® vessels and able to with-stand pressures of up to 30 bar.

Reagents and Solutions

A multi-element standard solu-tion (1000 mg L–1, Std. 3PerkinElmer Pure™ Plus) contain-ing all the elements studied wasused. Nitric acid (65%) and hydro-gen peroxide (30%) were obtainedfrom E. Merck (Darmstadt,Germany). All working standardsolutions were prepared daily fromthe stock solutions by simple dilu-tion with Milli-Q™ ultrapure waterobtained with a Millipore Milli-Qwater purification system(Millipore, Milford, MA, USA).

Certified reference materialNIST SRM 1846 Infant Formula wasobtained from the National Insti-tute of Standards and Technology(Gaithersburg, MD, USA).

Samples

Human milk samples were pro-vided by the milk bank of maternityclinic Odete Valadares, Belo Hori-zonte, Minas Gerais, Brazil. The pro-

78

TABLE IICP-MS Operating

Conditions and Parametersfor Data Acquisition

RF (40 MHz), Power (W)Forward 1000 WReflected < 5 W

Gas Flow Rate (L min–1)Plasma 15Auxiliary 1.2Nebulizer 0.94

Measurements Peak HoppingSweeps 4Reading/Replicate 3Replicates 3Dwell Time 50 ms

Integration Time 600 ms

TABLE IIMicrowave Program

for Digestion of Human Milk

Step t (min) Power (W)

1 5 2502 5 4003 5 5004 1 600

Vent 10 -

138Ba, 51V, 59Co, and 75As. Thesewere chosen because they are themost abundant non-interfered iso-topes and thus provide betterresults.

The milk reference material NISTSRM 1846 was reconstituted as rec-ommended by the supplier, and theresulting solution was digested (asdescribed under microwave diges-tion procedures). In general, goodagreement between obtained andrecommended values was found.

Aqueous calibrations were car-ried out for each element. Withrespect to other methods proposedfor the determination of theelements under study in humanmilk, our method offers the advan-tages of simplicity, short time (<2 h,small volume (1 mL) in relation tothe amount required in other stud-ies (40-43). The dilution applied inthe assay (1 mL of human milk to

Infant Formula was digested in amicrowave oven with 2 mL nitricacid, 0.5 mL H2O2, and 1 mL deion-ized water. The digested samplewas diluted to a final volume of25 mL. The NIST SRM 1846 InfantFormula solutions were also pre-pared in independent triplicatesand read in random triplicates. Theheating program for digesting sam-ples of the certified materials in themicrowave oven is summarized inTable III. Optimization of two heat-ing programs was necessarybecause the certified infant formulasample required a more selectprocess of digestion after beingreconstituted according to the man-ufacturer's recommendations.

RESULTS AND DISCUSSION

Figures of Merit

The detection limits for each ele-ment was based on three times thestandard deviation of the average of10 individually prepared blank solu-tions, and the quantification limitswere based on 10 times the stan-dard deviation of the average of 10individually prepared blank solu-tions. Both the detection and quan-tification limits are listed in TableIV. Table V reports the analyticalperformance of ICP-MS in terms ofLOQs, precision, and accuracy forthe analysis of NIST SRM 1846Infant Formula. The detection limitis at or below the µg L–1 level,while the precision (expressed asrelative standard deviation, %RSD)was almost always equal to or lessthan 3.4%.

Interferences Studies

Recovery assays were carried outto check the presence of the inter-ference effects on the developedmethod. Adequate amounts of theanalytes were added to milk beforedigestion. Recovery rates were inthe range of 81-111% and are listedin Table VI. It was not possible torecover Fe at the 1.0 µg L–1 level.The normal range of iron in human

Vol. 29(3), May/June 2008

79

TABLE IIIHeating Program for

NIST SRM 1846 Infant Formulain Microwave Oven

Step t (min) Power (W)

1 1 2502 1 03 6 2504 6 4005 4.5 6006 1.5 250Vent 10

TABLE IVRelationship of Correlation Coefficient and Calibration Equation of

Analyzed Isotopes (n = 3) and Limits of Detection and Quantification

Isotopesa Correlation Calibration LOD LOQCoefficient Equation (µg L–1) (µg L–1)

138Ba 0.9999 Y = 7293.64X + 461.28 0.184 0.61551V 0.9997 Y = 8794.26X + 4140.93 1.104 3.67978Se 0.9995 Y = 279.14X - 207.89 5.408 18.0382Se 0.9998 Y = 102.45X + 9.17 0.013 0.04453Cr 0.9992 Y = 893.91X - 353.85 2.137 7.12260Ni 0.9999 Y = 1184.4X + 44.75 0.041 0.13755Mn 0.9997 Y = 11645X - 3258.4 0.909 3.029206Pb 0.9997 Y = 2483.79X + 816.01 0.941 3.138208Pb 0.9997 Y = 5163.94X + 1517.57 0.863 2.87665Cu 0.9957 Y = 1085.9X - 471.58 2.684 8.94759Co 0.9999 Y = 6266.5X + 178.24 0.065 0.215112Cd 0.9999 Y = 1474.88X + 75.04 0.110 0.367114Cd 1.0000 Y = 1905.08X + 93.63 0.126 0.41975As 0.9999 Y = 802.68X + 63.52 0.172 0.57257Fe 0.9984 Y = 239.86X - 472.24 12.525 41.74964Zn 0.9954 Y = 1075.4X - 589.07 1.787 5.95866Zn 0.9950 Y = 613.45X - 325.92 1.672 5.57327Al 0.9984 Y = 7184.7X - 7870.8 4.745 15.817

a For all the isotopes, the calibration was accomplished with eight standards rangingbetween 0.010 to 20.0 µg L–1.

milk is about 0.5 mg L–1 (39). It wasalso not possible to recover Ni at1.0 µg L–1, most likely because ofthe polyatomic interferences atm/z 60: 44Ca16O, 23Na37Cl or36Ar24Mg. The isotopes used forfinal measurements were: 27Al, 53Cr,55Mn, 57Fe, 60Ni, 65Cu, 64Zn, 66Zn,78Se, 82Se, 112Cd, 114Cd, 206Pb, 208Pb,

80

ture. The high Fe concentrationrange obtained in our study is possi-bly due to the fact that the analyzedsamples were from women wholive in an area in which there isiron ore exploration. Another possi-bility is the presence of the interfer-ing ion CaOH+. The levels of Ca inthe samples used for our studywere obtained by inductively cou-pled optical emission spectrometry(ICP-OES) and the values were veryhigh (at an average of about 2380mg L–1).

CONCLUSION

Using a quadrupole inductivelycoupled mass spectrometry (Q-ICP-MS) instrument after microwave-assisted digestion has proven to bea very useful tool for the simultane-ous multi-element determination oftrace elements in human milk sam-ples. The method proposed is rapidand simple. The short time andsmall volume required, associatedwith the values of the analyticalparameters, ensure the suitability ofthe method for determining 27Al,53Cr, 55Mn, 57Fe, 60Ni, 65Cu, 64Zn,66Zn, 78Se, 82Se, 112Cd, 114Cd, 206Pb,208Pb, 138Ba, 51V, 59Co, and 75As inhuman milk during the usual lacta-tion period (colostrum and maturemilk). Recoveries ranged between80.8 and 110.9%. The precision ofthe method, expressed as relativestandard deviation (%RSD), was bet-ter than 3.4% for all analytes investi-gated. Precision and accuracy ofthe method show that it is a usefulmethod for the intended purpose.

ACKNOWLEDGMENTS

The authors are thankful toConselho Nacional de Pesquisa eDesenvolvimento Tecnológico(CNPq), Fundação de Amparo àPesquisa do Estado de Minas Gerais(FAPEMIG), Departamento deEngenharia Metalúrgica e deMateriais, and Maternidade OdeteValadares. J.C.J. Silva and W.B. Netoobtained scholarships from CNPq

50 mL of solution) is adequate formeasuring the usual elemental con-tent.

However, microwave digestionhas certain limitations. The numberof samples that can be simultane-ously digested is restricted to thenumber of digestion vessels themicrowave can accommodate (10in our case). Moreover, precautionsmust be adopted to prevent alter-ations of the inner surface of thedigestion vessels that can lead toadsorption of the elements. There-fore, aggressive conditions, such ashigh acid concentrations, high tem-peratures, and long digestion times,should be avoided.

Quantification

Table VII presents the resultsobtained for each element in termsof means and standard deviations(SD) in mature and colostrummilks. In general, the data do notindicate the presence of potentiallytoxic metals beyond the nationalrecommended values. These con-centrations are also comparable tothe values referenced in the litera-ture, especially those of an interna-tional collaborative WHO IAEAsurvey on the contents of trace ele-ments in milk worldwide (43,44).The concentration ranges of theelements in human milk consideredin this study are in good agreementwith those reported in the litera-

TABLE VAnalytical Performance of ICP-MS for theQuantification of Elements in Human Milk

Isotope LOQ Repeat- Accuracyb Recovery(µg L–1) abilitya Certifiedc Foundc (%)

%RSD (µg L–1) (µg L–1)138Ba 0.615 1.4 Ncd - -51V 3.679 1.0 Nc - -78Se 18.03 2.1 0.1622 ± 0.0011 0.1597 ± 0.0119 98.4882Se 0.044 1.4 0.1622 ± 0.0011 0.1710 ± 0.0110 105.4353Cr 7.122 1.6 Nc - -60Ni 0.137 2.5 Nc - -55Mn 3.029 1.7 0.8112 ± 0.0055 0.9179 ± 0.0151 113.15206Pb 3.138 1.7 Nc - -208Pb 2.876 1.5 Nc - -65Cu 8.947 2.1 10.2211 ± 0.0698 9.3369 ± 0.2848 91.3559Co 0.215 0.7 Nc - -112Cd 0.367 1.6 Nc - -114Cd 0.419 1.4 Nc - -75As 0.572 0.8 Nc - -57Fe 41.749 2.9 127.9668 ± 0.8743 126.7870 ± 3.2124 99.0864Zn 5.958 3.1 Nc - -66Zn 5.573 2.6 Nc - -27Al 15.817 3.4 Nc - -

a Calculated on 10 independent samples, reading in aleatory triplicates, at the follow-ing concentrations: 5.0 µg L–1 for Ba, V, Se, Cr, Ni, Mn, Pb, Cu, Co, Cd, As and Al;15.0 µg L–1 for Fe and Zn.b Performed on the NIST SRM 1846 Infant Formula.c Mean ± S.D.d Nc: not certified.

81


and FAPEMIG. V.S.T. Ciminelli andJ.B.B. Silva are grateful to CNPq forthe research grants.

Received February 1, 2008.

REFERENCES

1. F.A.R. Martino, M.L.F. Sánchez, andA. Sanz-Medel, Anal. Chim.Acta442, 191–200 (2001).

2. E. Coni, A. Alimonti, A. Bocca, F. LaTorre, D. Pizzuti, and S. Caroli,Element Speciation in BiorganicChemistry, ed. S. Caroli, JohnWiley and Sons, pp. 255 (1996).

3. B.Lönnerdal, Physiol. Rev. 77, 643(1997).

4. P. Brätter, I. Navarro Blasco, V.E.Negretti de Brätter, and A. Raab,Analyst 123, 821 (1998).

5. M.D. Silvestre, M.J. Lagarda, R. Farré,C. Martínez-Costa, J. Brines, FoodChemistry 68, 95-99 (2000).

6. J.A. Nóbrega, Y. Gélinas, A.Krushevska, and R.M. Barnes, J.Anal. At. Spectrom. 12, 1243–1246(1997).

7. R.M. Feeley, R.R. Eitenmiller, J.B.Benton, and H. Barnhart, Am. J.Clin. Nutr. 37, 443-448 (1983).

8. H. Gunshin, M. Yoshikawa, T.Doudou, and N. Kato, Agric. Biol.Chem. 49, 21-26 (1985).

9. J.A. Lamounier, J.C. Danelluzi, and H.Vannucchi , J. Trp. Pediat. 35, 31-34 (1989).

10. M.A. Siimes, E. Vuori, and P.Kuitunen, Acta Pediatr. Scand. 68,29-31 (1979).

11. H. Benemariya, H. Robberecht, andH. Deelstra, Sci. Total Environ.164, 161-174 (1995).

12. C.E. Casey, M.C. Neville, and K.M.Hambidge, Am. J. Clin. Nutr. 49,773-785 (1989).

13. G.B. Fransson and B. Lönnerdal,Am. J. Clin. Nutr. 39, 185-189(1984).

14. M.C. Neville, R. Keller, J. Seacat,C.E. Casey, J.C. Allen, and P.Archer, Am. J. Clin. Nutr. 40, 635-646 (1984).

15. H. Robberecht, H. Benemariya, andH. Deelstra, Biol. Trace Elem. Res.49, 151-159 (1995).

16. P. R. Silva, J. G. Dorea, and G. R.Boaventura, Biol. Trace Elem. Res.1997, 59, 57-62.

17. M.A. Amaro-López, R. Moreno-Rojas, P.J. Sánchez Segarra, and G.Zurera Cosano, Alimentaria April,71-78 (1996).

18. M. Krachler K.J. Irgolic, J. TraceElements Med. Biol. 13, 157(1999).

19. T.D.B. Lyon, G.S. Fell, R.C. Hutton,and A.N. Eaton, J. Anal. At. Spec-trom. 3, 265 (1988).

20. C. Sariego Muñiz, J.M. MarchanteGayón, J.I. García Alonso, and A.Sanz-Medel, J. Anal. At. Spectrom.13, 283 (1998).

21. E. Coni, S. Caroli, D. Ianni, and A.Bocca, Food Chem. 50, 203(1994).

22. S.E. Emmett, J. Anal. At. Spectrom.3, 1145 (1988).

23. M. Krachler and E. Rossipal, Ann.Nutr. Metab. 44, 68 (2000).

24. H.M. Kingston and L.B. Jassie, Anal.Chem. 58, 2534-2541 (1986).

25. M.D. Mingorance and M. Lachica,Anal. Lett. 18, 1519-1531 (1985).

26. G.M. Schelkoph and D.B. Milne,Anal. Chem. 60, 2060-2062 (1988).

27. R.A. Stripp and D.C. Bogen, J. Anal.Toxicol. 13, 57-59 (1989).

28. A. Alegría, R. Barberá, R. Farré, M.J.Lagarda, and A. Torres, Nahrung40, 92-95 (1996).

29. T. Alkanani, J.K. Friel, S.E. Jackson,and H.P. Longerich, J. Agric. FoodChem. 42, 1965-1970 (1994).

30. E. Coni, P. Falconieri, E. Ferrante,P. Semeraro, E. Beccaloni,A. Stacchini, and S. Caroli, Ann. Ist.Super. Sanita. 26(2), 119-130(1990).

31. A. Frkovic, M. Kras, and A. Alebic-Juretic, Bull. Environ. Contam.Toxicol. 58, 16-21 (1997).

TABLE VIMetal Recoveries in Human Milk (n = 3)

Isotope Recovery (%)(from spiked samples in 3 levels of concentration)

(1.0 µg L–1) (5.0 µg L–1) (15.0 µg L–1)138Ba 95.9 ± 7.1 95.1 ± 1.3 98.5 ± 3.051V 85.2 ± 4.2 90.8 ± 0.7 95.4 ± 0.978Se 110.9 ± 5.2 102.5 ± 7.7 100.0 ± 1.282Se 104.0 ± 9.0 98.8 ± 0.7 101.1 ± 0.753Cr 102.0 ± 2.2 99.2 ± 1.2 94.1 ± 2.360Ni - 82.7 ± 0.1 82.6 ± 0.255Mn 105.8 ± 1.1 92.4 ± 0.9 91.6 ± 0.7206Pb 81.8 ± 1.1 99.9 ± 1.7 101.3 ± 0.6208Pb 83.4 ± 0.8 98.9 ± 1.2 101.6 ± 1.765Cu 101.4 ± 15.1 99.2 ± 2.3 98.2 ± 1.259Co 85.5 ± 0.3 80.8 ± 0.3 82.2 ± 1.3112Cd 109.0 ± 1.2 100.7 ± 1.5 100.9 ± 2.5114Cd 107.5 ± 1.1 100.5 ± 1.0 99.3 ± 2.175As 102.6 ± 6.8 102.3 ± 0.8 101.2 ± 0.857Fe - 93.2 ± 4.2 100.4 ± 2.664Zn 98.1 ± 14.3 100.0 ± 1.6 98.3 ± 3.266Zn 100.2 ± 5.7 101.2 ± 2.8 100.3 ± 1.927Al 103.7 ± 7.7 98.6 ± 1.6 88.8 ± 2.9

82

32. M. Krachler, F. Shi Li, E. Rossipal,and K.J. Irgolic, J. Trace Elem.Med. Biol. 12, 159-176 (1998).

33. E. Coni, B. Bocca, B. Galoppi, A.Alimonti, and S. Caroli,Microchem. Journal 67, 187-194(2000).

34. J.M. Torres, M. Llauradó, and G.Rauret, Anal. Chim. Acta 379, 135-142 (1999).

35. J.K. Friel, C.S. Skinner, S.E. Jackson,and H.P. Longerich, Analyst 115,269 (1990).

36. M.D. Mingorance, M.L. PérezVázquez, and M. Lachica, J. Anal.At. Spectrom. 8, 853 (1993).

37. M. Krachler, E. Rossipal, and K.J.Irgolic, Biol. Trace Elem. Res. 65,53 (1998).

38. T. Prohaska, G. Köllensperger, M.Krachler, K. De Winne, G.Stingeder, and L. Moens, J. Anal.At. Spectrom. 15, 335 (2000).

39. I. Kaldor and E. Ezequiel. Nature,196, 175 (1962).

40. N.F. Krebs, M. Hambidge, M.A.Jacobs, and S. Myler, J. Pediatr.Gas- troenterol. Nutr. 4, 227-229(1985).

41. M.F. Picciano and Ph.D. Guthrie,Am. J. Clin. Nutr. 29, 242-254(1976).

42. E.Vuori and P. Kuitunen, Acta Pae-diatr. Scand. 68, 33-37 (1979)

43. L.A. Vaughan, C.W. Weber, and S.R.Kemberling, Am. J. Clin. Nutr. 32,2301-2306 (1979).

43. Minor and trace elements in breastmilk, Report of a Joint WorldHealth Organization InternationalAtomic Energy Agency Collabora-tive Study, WHO, Geneva, Switzer-land (1989).

44. S. Caroli, A. Alimonti, E. Coni, F.Petrucci, O. Seno-fonte, and N.Violante, Crit. Rev. Anal. Chem.24, 363-398 (1994).

TABLE VIIConcentrations of Trace Elements in

Five Mature Human Milk Samples andFive Colostrum Human Milk Samples and

Reference Ranges for Human Milk

Isotope Concentration (µg L–1)Colostrum Mature Range

(mean ± S.D.) (mean ± S.D.) (see Ref. 45)138Ba 8.24 ± 0.33 4.90 ± 4.07 0.9 – 4751V 13.09 ± 2.65 18.02 ± 6.64 -78Se 72.99 ± 13.47 205.17 ± 62.62 5 - 2382Se 41.56 ± 10.48 68.56 ± 11.27 5 - 2353Cr 76.23 ± 8.09 98.16 ± 20.05 0.2 – 0.860Ni 11.54 ± 4.16 7.38 ± 3.00 0.5 – 1055Mn 6.75 ± 1.32 8.91 ± 4.72 3 - 40206Pb 2.55 ± 1.82 2.48 ± 2.09 2 - 30208Pb 3.34 ± 1.62 2.93 ± 1.72 2 - 3065Cu 503.17 ± 34.55 297.45 ± 129.07 180 - 75159Co 0.26 ± 0.02 0.30 ± 0.10 0.1 – 2.0112Cd 0.33 ± 0.15 0.37 ± 0.15 0.7 – 4.6114Cd 0.23 ± 0.09 0.32 ± 0.14 0.7 – 4.675As 6.29 ± 4.14 8.07 ± 5.18 0.2 - 1957Fe 1091.39 ± 378.89 1015.38 ± 305.90 200 - 80064Zn 2889.57 ± 502.03 2088.00 ± 1552.72 700 - 400066Zn 2947.50 ± 531.27 2144.07 ± 1602.38 700 - 400027Al 46.98 ± 27.90 69.41 ± 59.76 3 - 125

83Atomic SpectroscopyVol. 29(3), May/June 2008

*Corressponding author.E-mail: [email protected].: +91-040-23434700Fax: +91-040-23434651


Comparative Study of X-ray Fluorescence andInductively Coupled Plasma

Optical Emission Spectrometry of Heavy Metalsin the Analysis of Soil Samples

A. Keshav Krishna*, K. Rama Mohan, N.N. Murthy, and P.K. GovilEnvironmental Geochemistry Group

National Geophysical Research Institute,(Council of Scientific & Industrial Research)

Habsiguda, Hyderabad-500606, India

ABSTRACT

Two multielement analyticaltechniques, X-ray fluorescencespectrometry (XRF) and induc-tively coupled plasma opticalemission spectrometry (ICP-OES)were used for the determinationof the trace element compositionof soil samples from four indus-trial sites and standard referencematerials in order to evaluate thetwo methods and assess thepotential use of such advancedinstruments for soil analysis.

The distinguishing factor isthat X-ray fluorescence analysis isa non-destructive method, pos-sessing the advantage that thetotal elemental content of thesamples can be obtained withoutany complicated sample pretreat-ment. XRF measurements in thisstudy were performed onpressed pellets, which were pre-pared for each sample. For theICP-OES determination of theelements, solutions wereprepared by the application ofdigestion techniques. Also, thestandard soil reference materialsNIST-2709, NIST-2710, NIST-2711, TILL-1, and TILL-3 wereused to verify the efficiency ofthe calibration of the XRF instru-ment and were analyzed by bothmethods. Elemental concentra-tion for Co, Cr, Cu, Ni, and Zndetermined by the two analyticaltechniques (XRF and ICP-OES insoils) was statistically treated bycorrelation analysis.

INTRODUCTION

The techniques of wavelength-dispersive X-ray fluorescence analy-sis (XRF) and inductively coupledplasma optical emission spectrome-try (ICP-OES) are widely used formultielement analysis of environ-mental samples (1,2).

X-ray fluorescence, a versatiletechnique, is ideally suited for theanalysis of rocks, soils, dust, conta-minated land samples, mineral con-centrates and products, archaeol-ogical artifacts, synthetic materials,and metals. The non-destructivenature of the technique allowslong-term storage of the samples,which can then be re-analyzed anynumber of times for additional ele-ments as necessary. This approach,therefore, avoids problems of re-sampling and digestion of separatealiquots.

The recent development of mod-ern computer-controlled XRF spec-trometers offers unattendedoperation of the instrument for cali-bration and matrix correction.Moreover, modern X-ray fluores-cence spectrometers have veryhigh inherent precision. However,they can give analytical accuracy ofthe same magnitude only if (a) thecalibration standards have physicalcharacteristics similar to those ofthe samples and are of accuratelyknown composition and (b) correc-tions are made for matrix effects.

Inductively coupled plasma opti-cal emission spectrometry (ICP-OES), on the other hand, is anemission technique, exploiting thefact that excited electrons emitenergy at a given wavelength asthey return to ground state (3,4).The fundamental characteristic ofthis process is that each elementemits photons at a specific wave-length corresponding to its chemi-cal composition. This offers someexcellent analytical characteristics,including high precision, sensitiv-ity, and selectivity (5,6). However,in ICP-OES analysis the samplesshould be in liquid form, thus a por-tion of the original soil sample ispermanently dissolved by somekind of digestion method.

The aim of the present work wasto investigate and evaluate the abil-ity of the XRF instrument for the insitu multielement and quantitative(Co, Cu, Cr, Ni, and Zn) analysis ofsoil samples. The obtained resultswere compared with those usingICP-OES analysis after digestion ofthe samples. The mean elementalconcentrations were compared bycorrelation analysis. In addition theinternational soil standardreference materials NIST-2709,NIST-2710, and NIST-2711 (fromthe National Institute of Standardsand Technology, Gaithersburg, MD,USA), and TILL-1 and TILL-3 (fromthe Canadian Certified MaterialsProject (CCRMP, Canada) wereused for the efficiency calibrationof the XRF instrument and analyzedby both methods.

84

EXPERIMENTAL

XRF Instrumental andOperating Conditions

The XRF spectrometer used inthe present study was a PhilipsMagiX PRO Model PW 2440 wave-length dispersive X-ray fluores-cence spectrometer coupled withan automatic sample changerModel PW 2540 (Philips,Eindhoven, The Netherlands). TheMagiX PRO is a sequential instru-ment with a single goniometer-based measuring channel, coveringthe complete elemental measure-ment range from F to U, in the con-centration range from 1.0 ppm to %levels. In addition, up to two fixedchannels may be fitted with fivetube filters. It has three primary col-limators along with three detectors,which are fitted in the goniometer,and eight analyzing crystals tocover the entire wavelength rangenecessary for the application. Theinstrument is microprocessor-con-trolled for maximum flexibility andconsists of end-window X-ray tubewith an Rh anode and a maximumvoltage/current of 60 kV/125 mA ata maximum power level of 4 kW.The analytical lines and instrumen-tal parameters used for eachelement are listed in Table I.

A prerequisite in any analyticalscheme is the use of the correctoperating conditions. For the vastamount of geochemical work, mostelements are known to fall withincertain ranges and these were con-sidered when selecting the operat-ing conditions (see Table I). Ingeneral, one set of operating condi-tions is sufficient for each element,designed initially to avoid spectraloverlaps (on peaks and backgrounds)and to optimize the count rate.Also, where the choice exists, it ispreferable to count for a longerperiod on a well-separated peakthan to introduce a correction foroverlapping peaks. In quantitativeanalysis, the intensity of any givenline is proportional to the concen-

tration, but modified by a combina-tion of absorption and enhancementeffects which are, in turn, a func-tion of the composition of the sam-ple and the primary spectrum fromthe X-ray tube (7,8,9). The currentand voltage values were chosenbased on the results of a prelimi-nary investigation concerning theeffect of voltage value on the inten-sity of the characteristic peak areas.For the five elements (Co, Cr, Cu,Ni, Zn) that were determined in thepresent study we found that thecharacteristic peak area valuesapproach, an upper limit value(kcps) around a voltage of 30 kV to40 kV and any further increase inthe voltage, does not significantlyinfluence the intensity of the char-acteristic X-ray lines.

To compensate for possibleinstrument drift, a monitor samplewas analyzed at the beginning andat the end of each day when sam-ples were run. For all elements, thenet intensities were calculated bysubtracting the background fromthe raw peak intensity. The netintensity (or raw intensity wherebackground levels were not mea-sured) for a given element dividedby the average of the two monitorsums for the day gives the monitor-normalized intensity. Thedifference in count rate of the mon-itor between the start and the endof the day was typically + 0.2% rela-tive to the mean of the two runs ofthe day.

In the present study, theelements were determined usingpressed powder pellets to minimizedilution of the sample with only afew exceptions. The spectral linesused for trace element analysis inmost natural samples lie at a shorter(higher energy) wavelength thanthose of the associated major ele-ments, and the dominant problemto overcome is that of absorption.The simplest, and probably themost effective, approach to correct-ing for absorption makes use of thefact that the intensity of the back-ground, and of the coherent andincoherent scattered lines of theanode element, vary systematicallywith the mass absorption of thesample matrix. Measurement ofthis scattered radiation, usually thatof Compton Kα for most anodesprovides an absorption correctionfactor (10,11).

ICP-OES Instrumental andOperating Conditions

A PerkinElmer® Optima™ 4300DV (dual view) inductively coupledplasma optical emission spectrome-ter (ICP-OES), equipped with across-flow nebulizer, Scott spraychamber, Echelle grating, and seg-mented array charge-coupleddevice detector (PerkinElmer Lifeand Analytical Sciences, Shelton,CT, USA). The fundamental charac-teristic of this process is that eachelement emits energy at a specificwavelength corresponding to itschemical composition. Although

TABLE IInstrumental Conditions Used for X-ray Fluorescence Spectrometer

Element Line Crystal Detector Peak Bkgrda Count LLDb

(2θ) (+2θ) Time(s) (ppm)

Co Kα LiF220 Duplexc 77.837 0.877 60+40 3.2Cr Kα PX-9 Duplex 69.387 0.771 60+40 2.8Cu Kα LiF220 Duplex 65.502 0.865 60+40 2.3Ni Kα LiF220 Duplex 71.226 0.859 60+40 2.6

Zn Kα PX-9 Duplex 41.743 0.912 60+40 1.6

a Bkgrd = background.b LLD = lower limit of detection.c Duplex = flow proportional and Sealed xenon counters.

85


each element emits multiple wave-lengths, in the ICP-OES technique itis most common to select a singlewavelength for a given element.The intensity of the energy emittedat chosen wavelengths is propor-tional to the concentration of thatelement in the analyzed sample.Thus, by determining which wave-lengths are emitted by a sample andby determining their intensities, theanalyst can quantify the elementalcomposition of the given sample,relative to a reference standard. Theoperating conditions of the ICP-OESspectrometer are listed in Table II.Based on the precision and sensitiv-ity observed in each emission line,the following wavelengths werechosen: Co at 228.62 nm, Cr at267.71 nm, Cu at 327.39 nm, Ni at231.61 nm, and Zn at 206.21 nm.

Samples, Standards, and TheirPreparation

Four real soil samples (NGRI-P,NGRI-M, NGRI-D, and NGRI-V)were collected from industrial sitesin India (Patancheru, Manali, Durga-pur, and Vapi) in polythene bagsfrom the top layer only. The soilsamples were dried for 2 days at 60oC in a hot oven. The dry soil sam-ples were disaggregated with amortar and pestle. The sampleswere finely powdered to –300mesh size (U.S. standard) using aball mill having ceramic balls toavoid contamination of Cr, Ni fromsteel plates, and Co from tungstencarbide tools. The samples werethoroughly mixed using the coningand quartering method to ensurehomogeneity. Each sample wasdivided and put into PET bottlesusing a continuous-flow motorizeddivider. The bulk samples werepacked into polythene bottles andstored in a cool and dry place. Asper the recommendations of theISO Guide 35, the samples werecollected at random from the stockprepared. Two pellets of each sam-ple were prepared and analyzedthree times by both the methods.

The measured intensity rather thanconcentration of the elements wasused, since the aim was to checkthe precision of the method ratherthan accuracy. About 2 g of samplewas taken to prepare the pressedpellet. Before preparing the pressedpellets, the bottles of referencematerials and samples were manu-ally shaken. Pressed pellets (40-mmdiameter) were prepared by usingboric acid as backing material andthen pressing for one minute at25-ton pressure with a hydraulicpress (HTP 40 Herzog, Germany).The sample was considered to behomogeneous since the variationin counts per second of the X-rayintensity for the elements thatcould be measured to the highestprecision by XRF was less than 1%.The U.S. standard soil referencematerials NIST-2709, NIST-2710,NIST-2711, and the Canadian certi-fied reference materials TILL-1 andTILL-3 were used as the calibrationstandards in the XRF method and

were also analyzed by both XRFand ICP-OES in order to verify thecalibration (12,13).

For ICP-OES analysis, multiele-ment solutions prepared from sin-gle-element stock solutionsprocured from Inorganic Ventures,Lakewood, USA were taken for cali-bration. The standards used in thecalibration were entered into themethod along with the concentra-tions of the elements required andtheir units (both concentration andsample). The sample was treatedwith 10 mL (1:1) hydrochloric acidfor one hour at 60 – 80 oC. Thesupernatant was decanted (whichcontains all the alkaline earth met-als) into a 500-mL volumetric flask,and retained. For the residue, 10 mLof hydrofluoric acid (HF) and 10 mLhydrochloric acid (HCl) wereadded and evaporated to dryness bykeeping on a hot plate. The addi-tion of HF and HCl, and evapora-tion to dryness was repeated. Then,another 5 mL was added, the solu-

Table IIInstrumental Operating Conditions ofPerkinElmer Optima 4300 DV ICP-OES

RF Generator Power 1500 WRF Generator Frequency 40 MHzPlasma Flow 15 mL/minAuxilary Flow 0.2 mL/minNebulizer Flow 0.6 mL/minPump Flow 1.5 mL/minType of Nebulizer Cross-FlowType of Spray Chamber Scott-typeType of RF Generator Solid StateShear Gas AirPurging Gas NitrogenRead Delay 60 secRead Time 2 – 10 secEquilibrium Time 15 secNumber of Replicates 3Viewing Mode AxialViewing Height 15 mm

Detector Segmented Array Charge-coupledDevice Detector (SCD)

86

tion evaporated to dryness, and theresidue dissolved in a minimumamount of HCl. Then the solutionwas transferred to a 500-mL volu-metric flask (combining it with theprevious extract) and diluted to vol-ume so that the final acid volumewas approximately 5% (v/v) HCl.

Reagents

A multielement solution,prepared from single-element stocksolutions procured from InorganicVentures, was used for calibration.The standards used for the calibra-tion were entered into the methodalong with the concentration andunits of the elements required. Theprocess blank prepared was run asa reagent blank and subtractedfrom the values obtained in thesamples to reduce the effects of thematrix difference.


XRF Spectrometer Calibration

Evaluation of the XRF spectra toextract quantitative results involvestaking into account a number offactors, each possessing an impacton the final result (2,14,5). Hetero-geneity of the different particlesizes and elemental distribution ofthe samples can significantly influ-ence the final results in XRF analy-sis. The spectrometer wascalibrated after measuring theintensities of the internationalreference materials. The criteriaused to select the samples were therequired interval of concentration,the quality of the known data foreach reference material, and alsoprevious calibration tests. Calibra-tion lines were obtained with theanalytical software issued by theinstrument manufacturer and usinglinear regression of the net intensi-ties versus concentration. Matrixeffects were corrected using empir-ical coefficients, more specificallyalphas based on count rate.Although corrections were madefor matrix effects using the

software’s fundamental parametersapproach as well as the empiricalalphas based on concentrations.They did not produce acceptablecalibration lines. The empirical cali-bration, based on intensities, isachieved by trial and error and cor-rections introduced were mainlybased on considerations aboutwhich elements would morestrongly absorb the emitted intensi-ties of the element of interest.Matrix corrections based on empiri-cal coefficients are only valid forthe analysis of samples with com-positions within the interval of thestandards. For this reason,reference materials with high con-centrations were also included asstandards.

ICP-OES Calibration

A calibration blank (2% nitricacid) was run before the standards.A multielemental solution fromSpex Certiprep, WP-15-500, (Spex,Metuchen, NJ, USA) was taken forcalibration. Five dilutions were pre-pared from WP-15-500 to make afive-point calibration curve. All ofthe elements determined exist at100 ppm concentration in WP-15-500. Subsequently, the dilutionscontained 10 ppm, 20 ppm, 40 ppm,60 ppm, and 80 ppm in HNO3. Thestandards being used in the calibra-tion were entered into the methodalong with the concentrations ofthe elements required and theirunits. The process blank preparedwas run as a reagent blank and sub-tracted from the values obtained inthe samples to reduce the matrixdifference.

The analytical results obtainedfor five trace elements in the refer-ence materials are given in TableIII. The precision of triplicate analy-sis is expressed as the standarddeviation (SD). Although the SDvaries from sample to sample andfor each element, the range of theSD based on the highest and lowestconcentrations provides a goodindication of the precision

obtained. In general, the SDincreases with decreasing concen-trations. In XRF, the five elements(Co, Cr, Cu, Ni, and Zn) showedgood precision at high concentra-tions, e.g., < 1% for NIST-2710. Atlow concentrations, Co, Cr, and Nihave a SD greater than 4% for theNIST-2709, NIST-2710, and NIST-2711 standards. For ICP-OES, theprecision is generally better. Con-centration of Zn and Cu for NIST-2710 standard reference materialdiffered greatly from the certifiedvalues using the XRF methodwhich may be due to the high back-ground originating from the Znpeak. In case of ICP-OES, it may bedue to a systematic error occurringduring the analysis where the Co,Cr, and Ni concentration for thefive standards determined by eachmethod lie within the 95% confi-dence level. The accuracy was alsoin good agreement when comparedto the certified reference standardvalues (see Table III).

Analysis of Soils FromIndustrial Sites

Results from the quantitativeanalysis (mg/kg ± SD) of the foursoil samples from local industrialsites by the two analytical proce-dures (ICP-OES after applying diges-tion techniques and XRF usingpressed pellets) are presented inTable IV. Precision of the XRFanalysis for four different samples(in terms of standard deviation,SD), ranged between 4.3% and 8.0%for Co, between 0.34% and 1.39%for Cr, between 0.34% and 1.89%for Cu, between 0.2% and 1.11% forNi, between 0.09% and 0.91% forZn. For the ICP-OES analysis, preci-sion ranged between 1.1% and 2.5%for Co, between 0.1% and 1% forCu, between 0.1% and 0.54% forCu, between 0.47% and 1.12% forNi, between 0.24% and 0.71% forZn.

The mean elemental concentra-tions were subjected to linear cor-relation analysis. Linear correlation

87


coefficients for each element werecalculated in each one of the twocomparison pairs: XRF vs. ICP-OES.It was found that the correlationcoefficients (CC) were strong foralmost all elements. Co showed0.995, Cr showed 0.93, Cu showed0.99, Ni showed 0.93, and Znshowed 0.99. The accuracy of theresults was evaluated by comparingwith the certified values (CV) of theanalyzed reference materials (whencertified values are known); theresults should ideally be within theconfidence interval (CI): CV±CI.Figure 1 shows a comparison of thecertified values with respect toXRF and ICP-OES analysis for thefive elements (Co, Cr, Cu, Ni, Zn)studied. By comparing the results ofboth methods for NGRI-D analysisof the elements studied, not muchdifference was found. However, forNGRI-P, NGRI-M, and NGRI-V, theCu and Zn values differed betweenboth methods. These differencesare justified by the fact that, in ICP-OES, the bulk volume of the sampleis used in the sample preparationprocedure; while in XRF, wherepellets are used, the particle sizewas restricted to a certain fraction.

CONCLUSION

Good agreement was observedin the determination of Co, Cr, Cu,Ni, and Zn by X-ray fluorescence(XRF) spectrometry for certified orrecommended values of five refer-ence materials (NIST-2709, NIST-2710, NIST-2711, TILL-1, TILL-3).

TABLE IIIComparison Between XRF and ICP-OES of Certified and

Recommended Values for the Reference Materials NIST-2709, NIST-2710, NIST-2711, TILL-1, TILL-3 (concentration in mg/kg ± SD)

Code Certified XRF ICP-OES± SD ± SD

Co (mg/kg)NIST-2709 13.4 17.1 ± 0.77 14.8 ± 0.03NIST-2710 10.0 13.8 ± 0.75 10.9 ± 0.07NIST-2711 10.0 12.7 ± 1.24 11.4 ± 0.14TILL-1 18.0 18.6 ± 1.40 21.0 ± 0.14TILL-3 15.0 17.3 ± 0.95 16.1 ± 0.11Cr (mg/kg)NIST-2709 130 106 ± 1.97 115 ± 0.19NIST-2710 39.0 40.3 ± 0.98 34.6 ± 0.42NIST-2711 47.0 44.2 ± 1.59 43.8 ± 1.01TILL-1 65.0 66.5 ± 1.44 62.3 ± 0.30TILL-3 123 120 ± 0.87 114.3 ± 0.16Cu (mg/kg)NIST-2709 34.6 35.6 ± 0.56 31.9 ± 0.44NIST-2710 2950 2513 ± 6.02 2491± 1.52NIST-2711 114 104 ± 0.55 101.9 ± 0.03TILL-1 47.0 45.9 ± 0.51 43.2 ± 0.14TILL-3 22.0 20.1 ± 1.41 20.4 ± 0.05Ni (mg/kg)NIST-2709 88.0 82.4 ± 2.1 84.2 ± 0.11NIST-2710 14.3 11.8 ± 0.55 13.1 ± 0.20NIST-2711 20.6 13.1 ± 3.20 18.5 ± 0.16TILL-1 24.0 20.7 ± 1.06 23.4 ± 0.33TILL-3 39.0 35.8 ± 1.55 40.4 ± 0.31Zn (mg/kg)NIST-2709 106 108 ± 0.92 100 ± 0.2NIST-2710 6952 5346 ± 11.8 6917 ± 1.2NIST-2711 350 316 ± 1.43 357 ± 0.41TILL-1 98.0 98.0 ± 1.02 99.5 ± 0.43TILL-3 56.0 58.3 ± 0.51 57.3 ± 0.47

TABLE IVComparative Results of the Chemical Composition of Soils From Industrial Area (mg/kg ± SD)

Sample Method Co Cr Cu Ni Zn

NGRI-P ICP-OES 17.8 ± 0.4 210 ± 0.98 47.1 ± 0.18 27.1 ± 0.9 228 ± 1.1XRF 18.8 ± 0.8 233 ± 3.2 50.7 ± 0.9 39.4 ± 0.4 230 ± 1.7

NGRI-M ICP-OES 11.1 + 0.7 479 ± 1.1 357 ± 0.8 110 ± 0.2 1225 ± 2.1XRF 10.8 ± 0.8 488 ± 1.7 372 ± 1.4 108 ± 1.1 1102 ± 3.2

NGRI-V ICP-OES 34.3 ± 0.3 388 ± 0.1 500 ± 0.4 88.1 ± 0.1 564 ± 4.1XRF 36.7 ± 1.0 493 ± 1.0 542 ± 1.0 79.1 ± 0.2 572 ± 0.2

NGRI-D ICP-OES 21.3 ± 0.3 136 ± 1.1 30.9 ± 0.1 55.6 ±0.4 83.9 ± 0.1XRF 20.7 ± 1.0 139 ± 0.6 33.2 ± 0.3 55.1 ± 0.1 81.9 ± 0.7

88

Fig.1. Comparison of certified value with XRF and ICP-OES for elements Co, Cr, Cu, Ni, Zn (mg/kg).

89


Multi-element and non-destructiveanalysis is an object of increasinginterest in the environmental pollu-tion monitoring research field. Fourdifferent soil samples from indus-trial sites (Patancheru, Manali, Vapi,and Durgapur in India) were ana-lyzed by XRF spectrometry and alsoby inductively coupled plasmaoptical emission spectrometry (ICP-OES). Elemental mean concentra-tions show statistically goodcorrelation with a 95% confidenceinterval when comparisons wereperformed between the two analyti-cal techniques.

XRF has long been theinstrumental method of choice formost igneous geochemists. XRF isa fully multi-element technique,which is theoretically capable ofdetermining an ever greater rangeof elements than ICP-OES, includ-ing the halogens. In practice, ele-ments lighter than F are highlyinsensitive and ICP-OES is the bestmethod for B, Be, and Li, and givesbetter detection limits for Na, Mg,and Al. XRF is more sensitive forsome high mass elements such asNb, Rb, Sn, Th, and U. Unlike ICP-OES, XRF is principally a solid sam-ple-based technique so a directcomparison of detection limits isdifficult, although in principle theperformance of the two methods isbroadly similar. The choicebetween a destructive (ICP-OES) ora non-destructive (XRF) method ofanalysis depends on the specificdemands of each investigation.

Most importantly, sample prepa-ration plays a crucial role not onlybetween different analytical meth-ods (ICP-OES or XRF), where thesample may be analyzed in the formof a liquid or a solid, but also in thesame analytical method (XRF)where the sample may be analyzedin pellet form. Nevertheless, astrong to moderate correlation wasnoticed between the mean elemen-tal concentrations obtained by XRFanalysis applied on pellets for all

five elements. However,heterogeneity problems that affectthe quantification procedure aremore prominent in this case. Con-sequently, the XRF spectrometeremployed is a promising tool thatmay be used for quick screeninganalysis of environmental sampleswithout the need for time-consum-ing sample preparation procedures.Whereas in the case of ICP-OES,sample pretreatment methodsdepend on the type of soil and thechemical method itself. Incompletedissolution may cause systematicunder-estimation of the elementsby ICP-OES measurements. Thistype of error can only beascertained if the matrix composi-tion of the soil reference material(SRM) and the sample does not dif-fer significantly, or the concentra-tions are compared with thosedetermined by non-destructive ana-lytical methods.

ACKNOWLEDGMENTS

The authors are thankful to Dr.V.P. Dimri, Director, National Geo-physical Research Institute, Hyder-abad, for his support andpermission to publish this paper.

Received January 29, 2008.

REFERENCES

1. B. Holynska, X-Ray Spectrometry 22,192 (1993).

2. K. Janssens, F. Adams, and A.Rindby,Microscopic X-Ray FluorescenceAnalysis, Wiley, Chichester, UK(2000).

3. Andrea Somagyi, Mihely Braun, andJozsef Posta, Spectrochim. Acta,Part B, 52, 2011 (1997).

4. D.N. Papadopoulon, G.A.Zachariadis, A.N. Anthemidis, N.C.Tsirliganis, and J.A.Stratis, Spec-trochim. Acta, Part B, 1877(2004).

5. Philip Atannaker, Menua Haukka,and S.K. Sen, Chem. Geol. 42, 319(1984).

6. Mark. E. McComb and Hyman D.Gesser, Talanta 49, 869 (1999).

7. R. Van Grieken, A. Markowicz, andP. Very, X-Ray Spectrom. 20, 271(1991).

8. P.K. Harvey, Edited by T.S Ahmedali,Short Course, Vol, 7, GeologicalAssociation of Canada, Montreal,Canada, pp. 221-257 (1989).

9. X.Y. Han, S.J. Zhuo, R.X. Shen,P.L.Wang, andA. Ji, Journal ofQuantitative Spectroscopy andRadioactive Transfer, 97 (1), pp.68-74, (2006).

10. M. Hoeing and A.M. de Kersabiec,Spectrochim. Acta, Part B, 51,1297 (1996).

11. Tim S. Brewer and Peter K. Harvey,Spectroscopy Europe 17(3), 10-16(2005).

12. K. Govindaraju, Geostand. Newsl.18, 1 (1994).

13. Dean-Claude Germanique,Geostand. Newsl. 18(1), 91(1994).

14. Jenkins R.W. Gould and D. Gedcke,Quantitative X-Ray spectrometry,Marcel Dekker, Inc., New York(1995).

15. M. Milazzo and C. Cicarali, X-RaySpectrom. 26,



Determination of Oxygen in High Purity Cadmium byRadio Frequency Glow DischargeOptical Emission Spectrometry

G. Anil*, M.R.P. Reddy, S.T. Ali, N.R. Munirathnam, and T.L. PrakashUltra High Purity Materials Division

Center for Materials for Electronics Technology (C-MET), IDA, Phase-III, HCL Post, Cherlapally,Hyderabad – 500 051, India

ABSTRACT

A method was established forthe determination of trace levelsof oxygen in high purity cadmiumby radio frequency glowdischarge optical emission spec-trometry (RF-GD-OES). The para-meters optimized for thedetermination of oxygen in cad-mium are nitrogen purge time,RF power, argon pressure, pre-integration time, and analysisduration. The method developedwas evaluated for the determina-tion of oxygen in Cd using anoxygen-nitrogen analyzer. Theanalysis results were satisfactory,which indicates that this methodcould fully satisfy the determina-tion of trace gases (oxygen) inhigh purity Cd metal.

INTRODUCTION

Semiconductor compounds areusually synthesized from highpurity elements, typically 99.9999(6N) purity and above. Ultrapurecadmium compounds (CdTe,CdZnTe, HgCdTe, CdS,CdSe, etc.)are used in various opto-electronicapplications, such as solar cells, IRdetectors, imaging devices, due totheir sensitivities to x-rays, gammarays, and photonic (UV-Visible-IR)radiation. These materials may stillcontain non-metallic contamina-tion, predominantly oxygen at the100-ppm levels. Cd is a reactivemetal and even its pure form oftencontains gaseous elements (oxy-gen) picked up mainly from theatmosphere during the purificationprocesses. High purity Cd is ofgreat importance in the semicon-ductor industry (1). High purity Cdused for semiconductor applica-tions should be void of oxygenbecause this impurity at variousconcentration levels (ppm to ppb)has deleterious effects for all theabove applications. For example,trace oxygen present in a crystalgrowth system affects theelectronic properties of the materi-als, such as crystals sticking towalls, which leads to detrimentalstrains and crystallographic imper-fections, etc. Therefore, the mea-surement of oxygen is a criticalfactor in semiconductor materialprocessing.

In this current work, a methodwas established to determine tracelevels of oxygen in high purity Cdusing radio frequency glowdischarge optical emissionspectrometry (RF-GD-OES).

EXPERIMENTAL

Instrumentation

All measurements wereperformed using a JY 10,000 RF1.0-m monochromator ( Jobin-Yvon, Longjumeau Cedex, France).A standard Marcus-Grimm source(stainless steel) was used, with a2400/2 grooves/mm holographicgrating. The power supplied to theplasma is by a radio frequency gen-erator at a frequency of 13.56 MHz.The sputtering rate (2-4) of the Cdwas set at a moderate power of25 W. Jobin-Yvon emission spec-trometry software (Quantum IQ-

2.22) was employed for controllingthe data acquisition parameters.High-purity argon (99.999%) wasused as discharge gas at a constantpressure of 1200 pascals.High–purity nitrogen (99.999%)was used as the purge gas at a con-stant pressure of 2 bars for 48hours. A limiting anode of orificediameter of 4 mm was used for theanalysis. The RF-GD-OES analysisparameters are given in Table I. Allsamples were polished on siliconcarbide abrasive paper (120, 400,600, 800 grit) and then rinsed withmethanol prior to analysis.

RESULTS AND DISCUSSIONS

Raw cadmium (99.9%) contain-ing 150 ppm of oxygen was puri-fied by vacuum distillation to99.999% (5N) with respect tometallic impurities. This material’s

* Corresponding author.E-mail: [email protected] address:Department of Chemistry,102, BiosystemResearch Complex, Clemson University,51 New Cherry Street,Clemson, S.C, 2964, USA

TABLE IRF-GD-OES Operating

Conditions for OxygenDetermination in Cd

Parameters

Spectrometer Jobin-YvonJy-10, 000 RF

GratingHolographic 2400/2 grooves/mmSpectral Range 120–800nmPower 25 WPressure(Ar) 0.0 12 barNitrogen Purge 2 barsPurge Time 48 hoursPhase 700Module 500Flush Time 10 sPre-integration 5 sAnalysis 10 s

Wavelength 130.217 nm

91


oxygen was around 75 ppm. The5N material was further refined to6N (99.9999%) by three consecu-tive vacuum distillations. The oxy-gen level was reduced to around40 ppm. To further decrease theoxygen levels in 6N Cd, the metalwas melted under hydrogen atmos-phere (hydrogen reduction),which reduced the oxygen levels towell below 1 ppm. To analyzethese various ranges of oxygen lev-els in the material using RF-GD-OESrequires nitrogen purging since theprominent (sensitive) lines of oxy-gen are in the far ultraviolet (UV)region. To standardize RF-GD-OESanalysis, temporal stability profilesneed to be monitored during sput-tering for plasma stabilizationtimes. It was found that at 25 Wpower, plasma sputtering stabilizedafter 5 seconds. A nitrogen purge at2 bar pressure for 48 hours wasrequired for identification anddetermination of oxygen. The lineselected (5) for the analysis of oxy-gen in cadmium was 130.217 nm.This line had no interference fromthe cadmium matrix. Since RF-GD-OES is a comparative technique, astandard material (Cd) with aknown amount of oxygen isrequired. An in-house calibrationstandard of cadmium for oxygenwas used to evaluate the oxygencontent in raw, 1st distilled and2nd distilled cadmium using an oxy-gen–nitrogen analyzer and RF-GD-OES (6-8). The results obtainedusing the two instruments werecomparable and were found to bein agreement (see Table II). The RF-GD-OES oxygen peaks in the cad-mium samples are shown in Figures1a, 1b, and 1c, respectively. Whileanalyzing low levels of oxygenunder inert atmosphere, the refer-ence line used (C-193.091 nm) wasnot very stable; hence, a slight shiftin oxygen peaks was observed. Thereproducibility of the analyteresults indicate that they are indeedreal and above their respectivedetection limit. The limit of detec-

tion (LOD) was generated using therelative standard deviation of thebackground (RSDB). The method ofcalculation was proposed byBoumans and coworkers (9). Sincethere was no solid blank sampleavailable, the background signalfrom the oxygen peak obtainedusing RF-GD-OES was used for cal-culating the LOD. The LOD calcu-lated for oxygen in cadmium was0.8 ppm.

CONCLUSION

The application of radiofrequency glow discharge opticalemission spectrometry (RF-GD-OES) has demonstrated that it iswell suited for the determination ofoxygen in semiconductor materials.Trace levels of oxygen can be eval-uated using RF-GD-OES and thedetection limits calculated werefound to be 0.8 ppm. The method-ology described can also be appliedfor depth profiling of oxygen con-tent in semiconductor materials.

ACKNOWLEDGMENTS

The authors are thankful to theExecutive Director, C-MET, for hisencouragement during this work.

Received November 23, 2007.

REFERENCES

1. S.T. Ali, R.C. Reddy, N.R. Munirath-nam, C. Sudheer, G. Anil, and T.L.Prakash , Sep. Purif. Technol. 52,288 (2006).

2. C.Perez, R.Pereiro, N.Bordel, andA.Sanz-Medel, JAAS 15 , 67 (2000).

3. G. Anil, M.R.P. Reddy, D.S. Prasad,S.T. Ali, N.R. Munirathnam, andT.L. Prakash, Mater. Charact. 58,92 ( 2007).

4. G. Anil, M.R.P. Reddy, S. RajeshKumar, N.R. Munirathnam, andT.L. Prakash , Accepted in At.Spectrosc, (2007).

5. K.Wagatsuma, Bunseki Kagaku, 48,457 (1999).

6. V.Lavoine, H.Chollet, J.C. Hubinois,S.Bourgeois, and B.Domenichini,JAAS 18, 572 (2003).

7. M.L.Hartenstein and R.K.Marcus,JAAS 12, 1027 (1997).

8. G. Anil, M.R.P. Reddy, D. S. Prasad,N.R. Munirathnam, andT.L.Prakash , Annali diChimica,,97, 10,1039, (2007).

9. P.W.J.M.Boumans andJ.J.A.M.Vrakking, SpectrochimActa. Part B, 42, 819 (1987).

TABLE IIComparison of Oxygen Results (Results in ppm)

Samples Oxygen Nitrogen Analyzer RF-GDOES

Raw Cadmium 152±2.8 148±0.9

1st Distilled Cadmium 75±1.2 69±0.5

2nd Distilled Cadmium 40±1.0 43±0.2

92

Fig. 1 (a, b, c). Oxygen peaks in the cadmium samples using RF-GDOES.



* Corresponding author.E-mail: [email protected]

Determination of Trace Copper in Biological Samplesby On-line Chemical Vapor Generation-Atomic Fluorescence Spectrometry

Liang Hea, Xiaofan Zhub, Li Wub, and Xiandeng Houa,b*a College of Chemistry, Sichuan University, Chengdu, Sichuan 610064, P.R. China

b Analytical & Testing Center, Sichuan University, Chengdu, Sichuan 610064, P.R. China

ABSTRACT

On-line chemical vapor gener-ation atomic fluorescence spec-trometry (CVG-AFS) was, for thefirst time, used to determinetrace copper in biological sam-ples by merging acidified samplesolution with potassium tetrahy-droborate aqueous solution inthe presence of micro-amountsof 1,10-phenanthroline. Nitricacid, for both sample digestionand chemical vapor generation,was used as the acid medium.CVG conditions and instrumen-tal parameters were optimizedfor the best CVG efficiency,good gas/liquid separation, andefficient atomization/excitation.Under the optimized conditions,a limit of detection of 4 ng mL–1

was obtained for copper, with alinear dynamic range of overthree orders of magnitude. Theproposed method was success-fully applied to the determina-tion of copper in biologicalcertified reference materials.

INTRODUCTION

Chemical vapor generation(CVG) is a prevalent sample intro-duction technique in analyticalatomic spectrometry owing to itsefficient matrix separation,enhanced analyte transportefficiency (approaching 100%),more efficient analyte atomization,and potential speciation analysis(1,2). Hydride generation-atomicfluorescence spectrometry (HG-AFS) is a representative applicationof CVG in analytical atomic spec-trometry. It has additional advan-tages of simple instrumentation,flexible sampling amount, wide lin-ear dynamic range, good stability,and low instrumental and runningcosts. Although CVG, especially thehydride generation techniques, hasbeen used with atomic spectrome-try for routine analysis of varioussamples, the elemental coverage ofthis technique is still limited, usu-ally the eight traditional hydridegeneration elements, plus mercury,cadmium, and zinc. Recently, therange of elements amenable to CVGhas been expanded to more transi-tion elements and some noble ele-ments (3,4).

Copper is an essential elementfor the functions of the humanbody such as hematopoiesis, cyto-genesis, and endocrine, whereastoxicity would attack in case ofexcess intake. Therefore, it isimportant to develop sensitive,rapid, simple, and reliable analyticalmethods to monitor copper infood, biological, geological, or envi-

that the sensitivity can be increasedby an order of magnitude whenlow concentrations of acetic acid,instead of hydrochloric or nitricacid, is used for CVG for copper (8).

In this work, copper CVG is forthe first time coupled to atomic flu-orescence spectrometry (AFS) byusing a commercial intermittent-flow CVG-AFS instrument. Theintermittent-flow method is a com-promise sample introductionmethod, which has characters ofboth flow injection and continuous-flow methods, demonstrating lesscross contamination, less sampleconsumption, low running cost,and the possibility of adjusting thepump rate at different stages tooptimize the vapor generation con-ditions (9). The influence of experi-mental conditions on the analyticalperformance was studied in detail,and it was successfully applied tothe determination of trace copperin biological samples.

EXPERIMENTAL

Instrumentation

A Model AFS-2202 double-chan-nel non-dispersive atomic fluores-cence spectrometer with aprogrammable intermittent-flowvapor generation apparatus (BeijingKechuang Haiguang InstrumentCo., Beijing, P.R. China) was usedthroughout this work. Details aboutthe construction of the instrumentcan be found elsewhere (10).A copper hollow cathode lamp,specially designed for AFS, wasemployed as the excitation source.The argon carrier and shield gaswere controlled individually by twoflow meters which could operate ata low flow rate.

ronmental samples. In 1996, Stur-geon et al. (5) were first to detectvolatile copper species by mergingan acidified sample solution into anaqueous sodium tetrahydroboratesolution at room temperature, andsuccessfully applied this method todetermine copper in environmentalmatrices by CVG inductively cou-pled plasma optical emission spec-trometry (CVG-ICP-OES).Subsequently, the significantenhancement (10-fold) of the cop-per-containing vapor generationefficiency was observed by addingtrace amounts of 1,10-phenanthro-line (6,7). It has further been found

94

Figure 1 shows the schematicdiagram of the intermittent-flowvapor generation apparatus. Twogas-liquid separators were used inorder to eliminate the interferenceof vapor and the fluctuation ofAr/H2 flame atomizer. A PTFE tub-ing (i.d. 1.0 mm) was used as thestorage coil (90 cm in length) andtransport tubing (40 cm in length),whereas the silicone rubber tubing(i.d. 4.5 mm, 15 cm in length) wasused as the reaction coil. The oper-ation of the pump can beprogrammed in several steps inwhich the duration and rotationrate of the pump can be manipu-lated, and the working program isdescribed in Table I.

Reagents

A 100-mg L–1 working solution ofcopper was prepared by serial dilu-tion of a copper stock standardsolution of 5000 mg L-1. A 1% 1,10-phenanthroline solution was pre-pared by dissolving 0.5 g of1,10-phenanthroline (AR, TianjinKermel Chemical Reagent Co.,Tianjin, P.R. China) in 50 mL of20% ethanol solution. Potassiumtetrahydroborate solution was pre-pared daily by dissolving appropri-ate amounts of KBH4 (AR, ChengduKelong Chemical Reagents Co.,Chengdu, P.R. China) in 0.05%potassium hydroxide solution.Nitric, hydrochloric, sulfuric,formic, propionic, and malonicacids were also purchased fromChengdu Kelong Chemical

Reagents Co., and acetic acid wasordered from Chengdu ChanglianChemical Reagents Ltd. (Chengdu,P.R. China). The certified referencematerials used in this work wereGBW 07604 Poplar Leaf (Instituteof Geophysical and GeochemicalExploration, CAGS, Langfang, P.R.China) and GBW(E) 080193 BovineLiver (Institute of Rock and MineralAnalysis, CAGS, Beijing, P.R.China).

Procedure

For one measurement cycle, theacidified sample solution of 1.4 mL(determined by the pump rate andsampling time) was sucked in Step1 to fill the storage coil but not mixwith the reductant, and Step 2 wasof 6-second duration to change thesampling tube over to the carrier.At Step 3, the sample solution ofthe fixed volume was pushed bythe carrier into the reaction coil tomix with the reductant for the CVGprocess. The volatile species andhydrogen gas separated from thegas-liquid separator was swept bythe carrier gas into the atomizerand ignited at the outlet of thequartz tube (the atomizer). It is notnecessary to add any auxiliaryhydrogen to maintain the argon-hydrogen (Ar/H2) flame, since thereis enough of hydrogen produced inthe CVG process. When Step 3 fin-ished, the pump was stopped againand made ready for the next mea-surement cycle.

Sample Pretreatment

Certified reference materialsGBW 07604 Poplar Leaf (0.2500 g)and GBW(E) 080193 Bovine Liver(0.1000 g) were accuratelyweighed and dissolved in 10 mL ofa 4:1 nitric and perchloric acid mix-ture, and subsequently the samplesolutions were heated at 170 oC tonear dryness. After cooling, theresidue was diluted to 25 mL with1% nitric acid and 0.0005% 1,10-phenanthroline for the measure-ment by CVG-AFS.

TABLE IWorking Program for the Intermittent-Flow Reactor

Step Flow Rate (mL·min–1) Time (s) FunctionSample/Carrier KBH4

1 10.3 8.6 8 Put the sampling tube intosample solution

2 0 0 6 Change-over of samplingtube into carrier solution

3 10.3 8.6 16 Readout and cleanup

4 0 0 6 Get ready to return to Step 1

Fig. 1. Schematic diagram of the intermittent-flow vapor generation for atomicfluorescence spectrometry. GLS: gas/liquid separator.

95



Optimization of Chemical Reac-tion Conditions for CVG

Previous reports (5,6) showedthat either hydrochloric acid ornitric acid could be used in vaporgeneration of copper with atetrahydroborate system, thoughnitric acid is generally avoided as itsdecomposition products interferewith hydride generation. Severalorganic acids were also tested byXu et al. (8), and it was found thatthe acetic acid of relatively lowconcentration could enhance thesignal of copper significantly over abroader concentration range com-pared to nitric acid. In this work,the influence of several inorganicand organic acids, includinghydrochloric, nitric, sulfuric,formic, acetic, propionic and mal-onic acids, were investigated on theatomic fluorescence intensity of500 µg L-1 copper, and the resultsare shown in Figure 2. It can beseen from Figure 2 that each inor-ganic acid or malonic acid providesgood performance while the effectof acetic acid or propionic acid isunsatisfactory. The exact identityof the volatile copper species isunknown, but the copper alkylmight be generated when organicacids (such as acetic or propionicacid) are used for the acidicmedium. Presumably, the good per-formance of acetic acid in Xu’swork was probably due to the effi-cient atomization of the resultingvolatile species in ICP, comparedwith the relative lower temperatureof the Ar/H2 flame in this work.The significant impact of the atom-izer temperature on the signalresponse of copper has also beenvalidated by Luna et al. (11). Asshown in Figure 3, the generationefficiency of copper volatile speciesdepended strongly on the organicacidity. Taking into considerationthat it is essential to use nitric acidin many digestion procedures andthus no need to remove it from ana-

lyte solution prior to analysis, nitricacid was chosen for further study inthis work. High acidity leads to thegeneration of a large amount ofhydrogen gas which dilutes the ana-lyte and causes the instability of theAr/H2 flame, while low acidityresults in insufficient generation ofhydrogen to maintain the flame. Itwas found that 1% nitric acid wasthe optimal reaction medium, forboth high vapor generationefficiency for copper and the stabil-ity of the Ar/H2 flame.

In this work, potassium tetrahy-droborate is used not only as areductant but also as the hydrogensupply to maintain the Ar/H2 flame.Figure 4 shows the effect of its con-centration on the atomic fluores-cence intensity of 100 µg L-1copper. Abundant hydrogen is pro-duced when its concentration isover 1.5%, resulting in a dramaticdilution and shorter residence timeof the volatile copper species in theatomizer. Therefore, 1.25% potas-sium tetrahydroborate was chosenfor further experiments for thehighest signal intensity of copper.

Several complexing reagents,such as ethylene diaminetetraacetic acid (EDTA), thiourea,L-ascorbic acid, 2-picolinic acid,1,10-phenanthroline and potassiumiodide, were investigated in thiswork as sensitivity-enhancingreagents, but only 1,10-phenanthro-line showed a significant enhance-ment effect as reported by Zhouand colleagues (6). Figure 5 showsthat the optimal concentration of1,10-phenanthroline is 0.0005%,with an enhancement factor ofabout 12.

Optimization of InstrumentalParameters

As expected, the signal intensityof copper increases with the lampcurrent. The signal-to-noise ratioreached a maximal plateau whenthe lamp current was around 80 mA.Observation height is the distance

from the quartz furnace outlet tothe point where the atomic fluores-cence signal is measured. Thecloser observation point (< 9 mm)results in a higher blank because ofthe scatter of the radiation sourceon the atomizer outlet, but the sig-nal intensity of copper is remark-ably reduced as the observationheight increases. As a consequence,an observation height of 10 mmwas chosen for this study.

The carrier gas flow rate greatlyaffects the efficiency of CVG, aswell as transport and atomization ofvolatile species. At a low flow rate,the resultant volatile species mightdecompose quickly in the reactionsolution due to a lack of effectiveseparation from the reaction media.However, at a high flow rate, thedilution of the volatile species andthe short residence time of the ana-lyte in the Ar/H2 flame would bedominant, resulting in the decreaseof the signal intensity. Consideringboth CVG and atomizationefficiency, 200 mL min-1 was thebest carrier gas flow rate. In addi-tion, 400 mL min-1 argon was usedas the shield gas in order to preventextraneous air from entering theflame, alleviating the interferencefrom the alien environment.

Rapid mixing of sample andreductant solution is of paramountimportance to the overall efficiencyof the generation process as well asfast gas/liquid separation of thevolatile species from spent solutionbecause the nascent volatile speciesis relatively unstable and easilydecomposed in the aqueous phase(4,11–13). Increasing sample /reductant flow rate by adjusting thepump rotater speed of the intermit-tent-flow system at Step 3 is the eas-iest way to enhance the efficiencyof this process, while an auxiliarycarrier gas also assists the transportof the resultant volatile species tothe detection unit. Therefore, themaximal rotater speed of the pro-grammable pump of 10.3 mL min-1

96

Fig. 2. Effect of acid medium on the fluorescence intensity of500 µg L–1 copper. Concentration of acids: 1% hydrochloricacid; 1% nitric acid; 0.5% sulfuric acid; 0.5% formic acid;0.75% acetic acid; 0.5% propionic acid; and 1% malonic acid.

Fig. 5. Effect of 1,10-phenanthroline concentration on thefluorescence intensity of 100 µg L-1 copper.

Fig. 4. Effect of potassium tetrahydroborate concentration (in0.05% potassium hydroxide) on the atomic fluorescence inten-sity of 100 µg L-1 copper. Error bars denote one standard devi-ation (same for Figure 5).

Fig. 3. Effect of concentration of nitric, formic or malonicacid on the fluorescence intensity of 500 µg L–1 copper.

97


(corresponding to the sample /reductant ratio of 1.2) was adaptedin this work.

Little difference was foundbetween the silicone rubber andTeflon® tubing for the efficiency ofcopper volatile generation whenoptimizing the reaction coil. Thelength of the reaction coil from 5 to30 cm on the atomic fluorescenceintensity of copper was studied,and 15 cm was found optimal. Twogas-liquid separators were used inorder to eliminate the interferenceof vapor and the fluctuation ofAr/H2 flame atomizer. The effect ofthe length of the transport tubebetween the first and secondgas/liquid separators was also inves-tigated. Good signal size can beobtained even if the transport tubeis as long as 40 cm. A short trans-port tube led to instability of theflame, thus the measurements. Con-sidering both the sensitivity andprecision of the measurements, a40-cm transport tube was selectedfor use. The optimized instrumentalparameters are summarized inTable II.

Evaluation of Interference

In the present work, the poten-tial interferences caused by somemetals in environmental, biological,and geological samples were stud-ied carefully, and the results arelisted in Table III. There is no signif-icant interference of most commonco-existing metals except Ni(II) athigh concentrations. Althoughmasking reagents, such as thiourea,ammonium fluoride, and organicacids (formic, acetic, propionic,and malonic acid), were used in anattempt to minimize the interfer-ences from Ag(I) and Ni(II), nonewas found to be effective.

Analytical Applicability

Under the optimized conditions,the analytical figures of merit of theproposed method for the determi-nation of copper were obtained

and compared with similar atomicspectrometric techniques, as sum-marized in Table IV. At least 0.99 ofR2 (linear correlation coefficient)showed good linearity of the cali-bration curves. The limit of detec-tion (LOD) of 4 µg L-1 is better thanor equivalent to those by atomicabsorption spectrometricdetection, and the linear dynamicrange from LOD to 1000 µg L-1 aremuch better.

In order to validate the potentialapplication of the proposedmethod, two certified referencematerials, GBW 07604 Poplar Leafand GBW (E) 080193 Bovine Liver,were analyzed for trace copper. Asseen from Table V, good agreement

was obtained between the deter-mined and certified values.

CONCLUSION

The commercial HG-AFS instru-ment was successfully employedfor the determination of trace cop-per by using the CVG technique,thus expanding its measurable ele-ments to non-traditional hydridegeneration element. The proposedmethod has the advantage of highsensitivity, fairly broad lineardynamic range, low sampleconsumption, cost-effectiveness,and ease of use. It has potentialapplication in the biological analy-sis for trace copper. Further workshould be directed to the identifica-

TABLE IIOptimized Instrumental Parameters of the Proposed Method

(100 µg L-1 Cu)

Lamp Current 80 mAObservation Height 10 mmCarrier Gas Flow Rate 200 mL min-1

Shield Gas Flow Rate 400 mL min-1

Sample / Reductant Flow Rate 10.3 mL min-1

Length of Reaction Coil 15 cm

Length of Transport Tube 40 cm

TABLE IIIInterference Study of Various Co-existing Metal Ions

on the Proposed Method (500 µg L-1 Cu)

Co-existing Concentration Cmetal ion / RecoveryMetal Ions (mg L-1) CCu(II) (%)

Ca(II) 500 1000 92K(I) 500 1000 93Na(I) 500 1000 87Mg(II) 100 200 94Zn(II) 100 200 96Mn(II) 100 200 90Al(III) 50 100 91Cr(III) 25 50 95Fe(III) 25 50 91Co(II) 25 50 94Ni(II) 25 50 82Ag(I) 1 2 86

Au(III) 1 2 95

98

tion of the generated volatilespecies, thus elucidating the CVGmechanism.

ACKNOWLEDGMENTS

We acknowledge the financialsupport for this project from theNational Natural Science Founda-tion of China (No.20775051) andthe Ministry of Education of China(NCET-04-0869).

Received November 12, 2007.

REFERENCES

1. R.E. Sturgeon, X. Guo, and Z. Mester,Anal. Bioanal. Chem. 382(4), 881(2005).

2. Y.H. He, X.D. Hou, C.B. Zheng, andR.E. Sturgeon, Anal. Bioanal.Chem. 388(4), 769 (2007).

3. P. Pohl, Trends Anal. Chem. 23 (1),21 (2004).

4. Z.X. Li, J. Anal. At. Spectrom. 21(4),435 (2006).

5. R.E. Sturgeon, J. Liu, V.J. Boyko, andV.T. Luong, Anal. Chem. 68(11),1883 (1996).

6. H.Y. Zhou, S.K. Xu, and Z.L. Fang,Spectrosc. Spect. Anal. 20(4), 525(2000).

7. S.K. Xu, H.Y. Zhou, X.G. Du, D.A.Zhu, Chem. J. Internet 3(7), 29(2001).

8. S.K. Xu, R.E. Sturgeon, Y. Guo, W.M.Zhang, and H.F. Zhao, Ann. Chim.-Rome 95 (7-8), 491 (2005).

9. X.W. Guo and X.M. Guo, Spectrosc.Spect. Anal. 15(3), 97 (1995).

10. T.Z. Guo, M.Z. Liu, and W.Schrader, J. Anal. At. Spectrom.7(4), 667 (1992).

11. A.S. Luna, R.E. Sturgeon, and R.C.de Campos, Anal. Chem. 72(15),3523 (2000).

12. C. Moor, J.W.H. Lam, and R.E. Stur-geon, J.Anal. At. Spectrom. 15(2),143 (2000).

13. X.C. Duan, R.L. McLaughlin, I.D.Brindle, and A. Conn, J. Anal. At.Spectrom. 17(3), 227 (2002).

14. A.S. Luna, H.B. Pereira, I. Takase,R.A. Goncalves, R.E. Sturgeon, andR.C. de Campos, Spectrochim.Acta B 57(12), 2047 (2002).

TABLE IVAnalytical Figures of Merit of the Proposed Method

in Comparison With Similar Techniques

Method Upper LOD Linear RSD ReferenceLinear (3σ) Correlation (%)Range Coefficient(µg L-1) (µg L-1) (R)

a FI-CVG-AAS 150 1.8 0.9997 2.6 (100 µg L-1) (6)b CVG-QT-AAS 15,000 800 - 9 (10,000 µg L-1) (11)c CVG-GF-AAS 200,000 2800 0.9958 11 (5000 µg L-1) (14)d CVG-AFS 1000 4 0.9978 2.5 (500 µg L-1) This work

a Flow-injection Chemical Vapor Generation Atomic Absorption Spectrometry;b Chemical Vapor Generation Quartz Tube Atomic Absorption Spectrometry;c Chemical Vapor Generation Graphite Furnace Atomic Absorption Spectrometry;d Chemical Vapor Generation Atomic Fluorescence Spectrometry.

TABLE VAnalytical Results for Reference Materials

Sample Matrix Found by Certifiedthis Method Value(mg kg-1) (mg kg-1)

GBW 07604 Poplar Leaves 9.1 ± 1.0 9.3 ± 3.0

GBW (E) 080193 Bovine Liver 92.2 ± 5.7 91.6 ± 5.7



*Corresponding author.E-mail: [email protected]: 90 424-2330062

Determination of Cu, Mn, and Pb in Yogurt Samplesby Flame Atomic Absorption Spectrometry

Using Dry, Wet, and Microwave Ashing MethodsGokce Kayaa, Ismail Akdenizb, and Mehmet Yamanc*

a Adiyaman University, Sciences and Arts Faculty, Department of Chemistry, Adiyaman, Turkeyb Bozok University, Sciences and Arts Faculty, Department of Chemistry, Yozgat, Turkey

c Firat University, Sciences and Arts Faculty, Department of Chemistry, 23119, Elazig, Turkey


ABSTRACT

The digestion proceduresbased on dry and wet ashing andclosed vessel-assisted microwaveoven were compared to deter-mine a rapid, reliable, and simpledigestion method for the flameatomic absorption spectrophoto-metric analysis of yogurt samples.Various temperatures and peri-ods for dry ashing, variousamounts of the digesting mix-ture/sample for wet ashing, anddifferent powers of themicrowave oven were examined.The obtained data show that themicrowave oven is an excellentmethod because it is fast andresults in fewer losses and conta-minants. In dry ashing, consider-able lead losses were observedwhen 500 oC was applied for aperiod of eight hours. The resultswere compared with thosereported in the literature. It wasfound that the Pb concentrationof all studied samples (except fortwo samples–taken from top ofAl container) were higher thanthe maximum limit (20 ng g-1) setby FAO/WHO for milk.

quency and attention disorders (3).The acceptable daily intake (ADI)recommended by authorized agen-cies (the FAO/WHO) and the U.S.Recommended Daily Allowances(US RDA) for adults was establishedas 2.0 to 4.0 mg/day for Cu and 2.0to 5.0 mg/day for Mn which isdependent on sex, age, and otherhealth factors such as pregnancy(4). Lead is known to be a highlytoxic metal that accumulates in thehuman body (especially in the brainand kidneys). Except for occupa-tional exposure, the main sourcesof the trace metals Cu, Mn, and Pb

INTRODUCTION

The essential trace elementsdetermined in this study (Cu, Mn,and Pb) have four major functionsin the body and are used as stabiliz-ers, elements of structure, for hor-monal functions, and as co-factorsin enzymes (1). Copper and man-ganese are the prosthetic group ofsome metalloenzymes containingsuper oxide dismutase (SOD),which is an important antioxidantenzyme for cellular protection fromreactive oxygen species (ROS) (2).Copper helps to form hemoglobinin the blood, facilitates the absorp-tion and use of iron so that the redblood cells can transport oxygen tothe tissues, and assists in regulatingthe blood pressure and heart rate.Copper also takes a role in activat-ing more than 30 enzymes includ-ing caeruloplasmine, cytochromeoxidase, lysine oxidase, dopamine–hydroxylase, ascorbate oxidase,and tyrosinase, some of which areinvolved in collagen synthesis.Moreover, copper is necessary forthe healthy development of theconnective tissue, nerve coverings,and the bones. Manganese is essen-tial for the proper formation andmaintenance of bone, cartilage andconnective tissue. It contributes tothe synthesis of proteins and thegenetic material, and helps to pro-duce energy for the body fromfoods (2). On the other hand,excessive exposure to these essen-tial metals can cause some diseasessuch as anemia, liver and kidneydamage, stomach and intestinal irri-tation, aggressive behavior, delin-

are from foods and beverages. As aresult, it has been recommendedthat all trace metal contents bedeclared for all prepared foods.There are some published studieson the determination of Cu, Pb, andMn concentrations as well aspapers on Zn, Fe, Ca, and Mg levelsfound in yogurt (5-13). Because oftheir generally low levels in thismatrix, the concentrations of Cu,Pb, and Mn in yogurt samples mustbe determined with great care.

In the wet and dry ashing proce-dures, the sample digestion stepsare often the most time-consumingsteps of the analysis. These meth-ods are laborious and tedious, oftenaccompanied with a high tendencyfor contamination. Some authorsreported that 15 to 24 hours arerequired for dry ashing (6,7) or onenight (about 12 hours) for both wetand dry ashing (10,11) for thedigestion of yogurt or similar sam-ples. In addition, some potentialproblems such as incomplete disso-lution, precipitation of insolubleanalyte, contamination, and loss ofsome volatile elements can causecontradictory results. With closed-vessel microwave digestion, theanalysis time, amount of reagents,and risk of contamination orvolatile analyte loss can be greatlyreduced. Thus, use of themicrowave oven for reliable andrapid sample digestion seems anattractive procedure. A digestiontime of several hours for conven-tional wet digestion can be reducedto a few minutes by usingmicrowave energy (13). Analyticalapplications of microwave energyinclude both a commercial domes-tic microwave oven (14,15) for bio-logical samples and a commercial

100

microwave oven equipped withtemperature and pressure regula-tors for biological and environmen-tal matrices (16).

Yogurt is gaining in popularitydue to its acceptability byconsumers as well as its nutritionalproperties and potentially benefi-cial effects for human health.Yogurt can also be a good source ofessential nutrients as well as miner-als in the human diet. On the otherhand, undesirable metals may enteryogurt, or the contents of the exist-ing nutritional metals may increasein the fermentation procedure andduring transport to the consumerdue to leaching from the process-ing containers. There are somestudies (7-13) that evaluated theextent and nature of mineral redis-tribution arising from yogurt in theprocessing centers.

In this study, the digestion meth-ods of dry, wet, and microwaveoven were examined to determinea fast and efficient sample ashingmethod. Furthermore, the concen-trations of Cu, Mn, and Pb in Turk-ish yogurt samples fermented indifferent containers made of alu-minum, Cr-steel, and plastic materi-als were compared to examine thecontamination sources from con-tainers. The concentrations of thesemetals were determined by usingflame atomic absorption spectrome-try (FAAS); the slotted tube atomtrap (STAT) system was also studiedfor Cu and Pb determinationsbecause their concentrations arebelow the sensitivity of FAAS.

EXPERIMENTAL

Instrumentation

A Model ATI UNICAM 929atomic absorption spectrophotome-ter (AAS) equipped with an ATIUNICAM hollow cathode lamp wasused for the determinations (UNI-CAM, Cambridge, England). Theoptimum conditions for AAS aregiven in Table I. The Slotted Tube

Atom Trap (STAT), made in a glassworkshop in Ankara, Turkey, wasused for improving the sensitivityof copper and lead at the optimumconditions described elsewhere(17,18). A domestic microwaveoven (Kenwood MW 460,Kenwood Ltd, New Lane-Havant,UK) and specially made Teflon®bombs (produced in a glass work-shop in Ankara, Turkey) were usedfor the digestion of yogurt.

Reagents

Unless stated otherwise, allchemicals used were of analyticalreagent grade. Throughout all ana-lytical work, double distilled waterwas used (MilliQ system). All glass-ware was of Pyrex® material andwas kept permanently filled with1 mol L-1 nitric acid when not inuse. In the digestion procedure,Pyrex glass, concentrated nitricacid (65%, Merck, Darmstadt, Ger-many), and hydrogen peroxide(35%, Merck) were used.

Certified Reference Samples

BCR 151 Spiked Skim Milk Pow-der (Institute for Reference Materi-als and Measurements, Geel,Belgium) was used as certified ref-erence material (CRM). Stock solu-tions of metals (1000 mg L-1) wereprepared by dissolving their nitratesalts in a suitable volume of1.0 mol L-1 HNO3. The yogurt sam-ples were fermented in containersmade of aluminum or plastic mate-rials or chromium steel, and werepurchased at local stores.

Preparation of Samples

The yogurt samples describedabove were purchased at differentlocal stores in Elazig, Turkey.Yogurt samples are generally madewith cow’s milk (scarcely withsheep’s milk) in this region in thesummer. In this study, two samplesof yogurt were fermented in con-tainers made of plastic material bytwo big production centers, whilethe other samples were producedin private houses. In addition,Pb(NO3)2 was added to indepen-dent portions of one milk sample.Then, both the Pb-added and with-out Pb-added milk samples wereyeasted for the fermentation ofyogurt. These samples were ana-lyzed to examine the volatile lossesof Pb. Different portions of thesame yogurt samples were exam-ined using the ashing methods (dry,wet, and microwave oven) for com-parison. Furthermore, yogurt sam-ples were taken from the top,central, and bottom of the contain-ers to examine any leaching of themetals from the container.

Dry Ashing

Approximately 8.0 g of yogurtsamples were transferred to Pyrexglass containers and dried in anelectric oven at 100 oC. The driedsamples were ashed at the selectedtemperatures of 475 and 500 oC forthe various ashing periods lastingfrom 2 to 8 hours. Minimum vol-umes of the nitric acid-hydrogenperoxide (1/1) mixture were addedto the ashed samples and driedwith occasional stirring on a hot

TABLE IOperating Parameters for FAAS

Parameter Cu Mn Pb

Wavelength (nm) 324.8 279.5 217.0HCL Current (mA) 3.0 9.0 7.5Acetylene Flow Rate (L/min) 0.5 0.5 0.5Air Flow Rate (L/min) 4.0 4.1 4.0

Slit (nm) 0.5 0.2 0.5

101


plate at temperature below boilingpoint (158 oC) of hydrogen perox-ide. After drying, 1.5 mL of 1.0 molL-1 HNO3 was added and, if neces-sary, centrifuged. The clear solu-tions were analyzed by FAAS. Theblank digests were carried out inthe same way.

Wet Ashing

Approximately 8.0 g of yogurtsamples were transferred to Pyrexglass containers, 8 mL of the con-centrated HNO3/H2O2 (1/1) mix-ture was added, and the mixtureswere dried on a hot plate withoccasional stirring. Then, the sameprocedure was reapplied by addingthe next 8 mL of the same digestionmixture. Various solvent/sampleratios from 1:1 to 3:1 were exam-ined. After drying, 1.5 mL of1.0 mol L-1 HNO3 was added and, ifnecessary, centrifuged. Therequired total digestion time wasabout 3 hours. The clear solutionswere analyzed by FAAS. The blankdigests were carried out in thesame way.

Digestion by Microwave Oven

For microwave oven digestion,the steps as outlined below werefollowed by using the HNO3/H2O2

(1/1) mixture as digestion reagent.

Weight: 8.0 g of yogurt⇓Add 8 mL of HNO3/H2O2

(1/1) mixture⇓Place into a water bath at85 oC for 90 minutes withoccasional stirring

⇓Add 4.0 mL of the abovedigestion mixture

⇓Put into Teflon bomb of400 cm3 and place insidemicrowave oven

⇓

Radiate at the selectedpower (360 or 450 W),for 2+2+2 min sequentially(if necessary, minimumvolume of the samedigestion mixture wasadded)⇓Add, 1.0 mL of 0.1 mol L-1

HNO3

⇓Transfer into the Pyrextube, measure the finalvolume (final volume wasfound to be about 3 mL)

The clear solutions were analyzedby FAAS. The blank digests werecarried out in the same way.

Method Validation

To determine the best digestionmethod, the dry and wet ashing andmicrowave oven methods wereexamined using various conditions.In the wet ashing procedure, differ-ent proportions of solvent/sample(1-, 2-, 2.5-, and 3-fold) were stud-ied. To determine volatilization lossand optimum conditions with thedry ashing procedure, two different

temperatures (475 and 500 oC)were examined by using variousoxidation periods ranging from2.0–8.0 hours.

The standard additions methodwas studied by adding differentamounts of Cu, Mn, or Pb to thesame yogurt sample (see Figures1–3). Then, microwave ashing wasapplied as described above.

To determine the accuracy ofthe method, certified referencematerial BCR 151 Skim Milk Pow-der was digested by using the vari-ous ashing methods.

RESULTS

Analytical Performance

Smith et al. and Yaman et al.reported (16,17) that the sensitivi-ties of Pb and Cu increased 6-foldand 3-fold, respectively, by usingSTAT. The obtained calibrationgraphs were found to be linear inthe concentration ranges asdescribed below and the equationsof the curves were as follows:

Fig. 1. Calibration graphs obtained with the standard additions method and withstandards (the concentrations are in ng mL–1).

102



Y= 314 XR2=0.9998 for Cu rangingfrom 0.025–0.4 mg L-1 bySTAT

Y= 268 X + 0.5R2=0.9999 for Pb rangingfrom 0.05–0.8 mg L-1 bySTAT

Y= 138 XR2=0.9971 for Mn rangingfrom 0.05–1.0 mg L-1

For microwave ashing, the metallevels in the reagent blanks were16 µg L-1 for Pb, 10 µg L-1 for Cu,30 µg L-1 for Mn, with standarddeviations of 4.0, 2.0, and 6.0,respectively. The limit of detection,defined as three times of the stan-dard deviation of the blanks, were12, 6, and 18 µg L-1, respectively.The effect of contamination waseliminated by subtracting the val-ues obtained for the blanks. It wasobserved that the blank values forwet ashing were higher than formicrowave ashing since a higheramount of digestion mixture wasrequired for the wet ashingmethod.

Comparison of DigestionMethods

In the wet ashing procedure,different proportions of solvent /sample were studied. It was foundthat two-fold proportions were suf-ficient for total digestion (Table II).

In the dry ashing procedure,maximum Cu and Mn concentra-tions were obtained within the3–8 hour range at 500 oC and the4.0–8.0 hour range at 475 oC (Fig-ures 4 and 5). In addition, Cu didnot reach maximum concentrationwhen 450 oC was applied. Similarly,maximum Pb concentrations werefound for 3.0–8.0 hours of diges-tion at 475 oC. On the other hand, adecrease of about 25% in Pb con-centration was observed after ash-ing for 8.0 hours at 500 oC (Figure6). This decline was not observedwhen Pb(NO3)2 was added to the

103


yogurt sample before the digestionprocedure. This result implies thatPb in milk and yogurt may bebound to lactate (which is themajor component in yogurt), andPb-lactate may be lost due to itsprobably higher volatilization prop-erties. To verify this assumption,Pb(NO3)2 was added to the milksamples. Then, both this Pb-addedmilk and the same milk without Pbadded were yeasted for fermenta-tion of the yogurt. It was observedthat 15% of the added Pb was lostafter ashing for eight hours at500 oC.

In the examination of the stan-dard certified reference materialBCR 151 Skim Milk Powder it wasfound that slightly higher Cu levelswere obtained by using microwaveoven at 360 W in comparison to theother ashing procedures.

Accuracy

The accuracy of the method wasstudied by examining the certifiedreference material BCR 151 SkimMilk Powder. The results are givenin Table II. It can be seen that therecovery values were 97% for Cu,100% for Pb by using microwaveoven at 360 W, dry ashing at 475oC, and the wet ashing methods.In addition, the accuracy of themethod was studied by examiningthe recovery of each metal from theyogurt samples fortified with thesemetals. The spiked metal to each ofthe samples was 50 µg kg-1. Afterdigestion by microwave oven, therecoveries were found to be at least95% for all metals.

The standard additions methodwas used to investigate possibleinterferences caused by the matrix.The slopes of the calibration curvesfor all studied elements were com-pared with the slopes obtained bythe standard additions method.From Figures 1–3 it can be seenthat the slopes of the calibrationcurves were found to be very close(0.314 for Cu, 0.268 for Pb, and

TABLE IIComparison of Dry, Wet, and Microwave Oven Ashing Methods

for Metals Determination in the Same Yogurtand in Standard References Material Samples

Digestion Method Cu Mn Pb Confidence(proportion of (ng/g) (ng/g) (ng/g) Limit (90%)solvent/sample)

Dry Ashing, 8 h at 475 oC 100±7 33±3 39±4Wet Ashing (one-fold) 75±6 19±2 26±3Wet Ashing (1.5-fold) 85±7 27±2 28±3Wet Ashing (two-fold) 109±10 35±3 40±5Wet Ashing (2.5-fold) 112±10 27±2 38±5Wet Ashing (three-fold) 110±11 32±2 37±4Microwave (MW) Oven 360 W 111±7 34±2 40±3MW Oven 450 W 107±6 35±2 39±3CRM BCR 151 Milk PowderCertified Value 5.23 - 2.002

CRM MW oven 360 oC 5.21 - 1.996 0.007CRM MW Oven 450 oC 5.11 - 1.998 0.011CRM Dry Ashing, 475 oC 5.18 - 2.061 0.059

CRM Wet Ashing,HNO3/H2O2 (1/1) 5.12 - 2.022 0.029

Fig. 4. Effect of ashing periods (h) on Cu concentrations in yogurt by using dryashing.

104

lowed in exactly the same way asdescribed above, using the sameglassware and the same reagentsthroughout. The results showedthat there were no contaminationor adsorption losses. The analyticalparameters obtained in the recov-ery assays using microwave ovendigestion showed that the methodis simple and fast for the determina-tion of Cu, Mn, and Pb in yogurtand in the CRM samples. Thesedata show that the method is reli-able at the 90% confidence level.

Applications

Due to the above results,microwave oven was chosen forthe digestion of all other samples.It can be seen from Table III thatthe Mn concentration in the souredyogurt sample was significantlyhigher than in all other yogurt sam-ples. In the literature (6,8), thetrace metal levels in yogurt matri-ces are reported in the ranges of0.04–0.17 mg kg-1 for Cu and0.03–0.06 mg kg-1 for Mn. In thisstudy, the metal concentrationswere found in the ranges of0.011–0.498 mg kg-1 for Cu,0.01–0.179 mg kg-1 for Mn, and0.019–0.126 mg kg-1 for Pb.

DISCUSSION

In 2000, the Joint FAO/WHOExpert Committee on Food Addi-tives established the provisionaltolerable weekly intake (PTWI) of25 µg Pb per kg of body weight forall groups including adults, infants,and children (19). However, thereis no apparent risk of exceeding thePTWI for Pb from yogurt even ifone kilo of yogurt were to be con-sumed per person per week. In thiscase, the mean intake of Pb fromyogurt is lower than 50% of PTWIset by the WHO. On the otherhand, the maximum Pb concentra-tion in milk was limited to 20 ng g-1

by different authorized agenciesand are suggested for milk productsdepending on their dilution factors(20,21). In view of these

ences. Thus, calibration graphswith diluted nitric acid can be used.

The possibility of sample conta-mination was studied by subtract-ing the values obtained for theblanks. Adsorption loss can beexcluded as the procedure was fol-

0.138 for Mn) to the valuesobtained with the standard addi-tions method (0.312 for Cu, 0.262for Pb, and 0.132 for Mn). In otherwords, all standard additions curveswere parallel in value to the calibra-tion curves. These results indicatethe absence of chemical interfer-

Fig. 6. Effect of ashing periods (h) on Pb concentrations in yogurt by usingdry ashing.

Fig. 5. Effect of ashing periods (h) on Mn concentrations in yogurt by usingdry ashing.

105


TABLE IIIComparison of Metal Concentrations in

Yogurt Samples Fermented in the ContainersMade From Different Materials Sampled

From Bottom, Central, and Top(The results are mean values of three independent

portions of the same sample.)

Con- From Bottom, Cu Mn Pbtainer Central, or Top (ng/g) (ng/g) (ng/g)

Al Bottom 141±14 26±3 49±5Al Bottom 111±12 34±3 40±4Al Bottom 15±2 21±2 89±9Al Bottom (soured) 34±3.1 *179±15 78±8Al Bottom 87±9.1 22±2 100±9Al Average for Bottom 78±52 26±6 71±26Al Central 184±19 20±2 55±5Al Central 77±8 10±2 25±3Al Central 52±5 10±2 126±11Al Central 50±5 25±3 40±5Al Average for Central 91±63 16±8 62±45Al Top 108±9 10±2 47±5Al Top 72±6 26±2 19±2Al Top 82±7 30±3 35±4Al Top 60±7 20±3 45±5Al Top 46±4 24±2 20±2Al Average for Top 74±23 22±7 33±13Al Average for Top 74±23 22±7 33±13Average for all Al Containers 81±9 21±5 55±20Plastic Central 30±3 17±2 45±5Plastic Central 63±5 16±2 39±4Plastic Central 71±7 13±2 26±3Plastic Central 11±2 17±2 81±8Plastic Average for Central 44±28 16±2 48±24Plastic Top 71±8 10±1 56±5Plastic Top 18±2 13±2 83±8Plastic Top *317±33 23±2 30±4Plastic Top *498±39 28±2 39±5Plastic Average for Top 45±37 19±8 52±23Average for all Plastic

Containers 45±1 18±2 50±3Cr-steel Top 64±55 10±1 44±5Cr-steel Top 45±5 20±3 36±4Cr-steel Top 42±5 27±3 30±4

Cr-steel Average for Top 50±12 19±9 37±7

*These data are not included in the calculation of mean values.

regulations, all Pb concentrations (except in two sam-ples of our study) were found to be higher than theallowable limit of concentration.

With regard to leached Mn at the sampling points(including at top, central, and bottom) of thecontainer, no significant differences were observed byusing the statistical tests. However, there is a signifi-cant difference between Pb concentration in yogurtsamples taken from the top, bottom, and centralpoints of aluminum containers. It was observed thatthe average Cu concentrations in yogurt fromaluminum containers were significantly higher thanfrom plastic containers. The average Cu levels inyogurt samples from Cr-steel containers were found tobe close to those from plastic containers. The highestCu concentrations were found in two yogurt samplestaken from the top point of the plastic containers.After examination, it was found that the milk of thesetwo yogurts was heated in big copper containers, andthen, the milk was transferred to plastic containers foryogurt producing. For all other yogurt samples, themilk used for yogurt to be transferred into plastic con-tainers was heated in aluminum containers. So, it canbe concluded that the Cu leached from the coppercontainers into the milk due to the Cu container usedin the heating step.

CONCLUSION

This study shows that the determination of thetrace metals Cu, Mn, and Pb in yogurt by flame atomicabsorption spectrometry (FAAS) using closed-vesselmicrowave (MW) digestion is rapid and simple. Theobtained analytical parameters make this method suit-able for the determination of Cu, Mn, and Pb (thelimit of detection; 6, 18, and 12 µg L-1, respectively)in this matrix. The total time of digestion was 4 hoursfor dry ashing, 3 hours for wet ashing, and 2 hours forMW ashing. It is perfectly safe to use a domesticmicrowave oven at the studied conditions because ofpredigestion of the sample in a water bath at 85 oC for1.5 hours and using the durable Teflon bomb (durableto 360 oC) with sufficient volume (400 cm3) to pre-vent high pressure in Teflon bomb. From this study itcan also be concluded that the contribution fromyogurt of total Pb intake (PTWI; 25 µg Pb per kg ofbody weight) is toxicologically insignificant and,therefore, yogurt is only a minor source of essentialelements including Cu and Mn. On the other hand,no differences in Mn levels were observed from eitherthe plastic and aluminum containers (mean 18 and19 ng g-1, respectively), as well as from different sam-pling points including top and central for plastic con-tainers (mean 19 and 16 ng g-1, respectively).

Received December 13, 2007.

106

REFERENCES

1. L.E. Feinendegen and K. Kasperek,Trace Elem. Anal. Chem Med. Biol.1-17 (1980).

2. W. Mertz, Trace elements in humanand animal nutrition. Vol. 1-2, SanDiego, CA, USA, Academic Press(1987).

3. M. Yaman, Cur. Med. Chem. 13(21),2513 (2006).

4. L. Noel, J.C. Leblanc, and T. Guerin,Food Add. Contam. 20(1), 44(2003).

5. L.M. Klevay, J. Am. Col. Nutr. 17(4),322 (1998).

6. M.A. De la Fuente, F. Montes, G.Guerrero, and M. Juarez, FoodChem. 80, 573 (2003).

7. P.J. Sanchez-Segarra, M. Garcia-Mar-tinez, M.J. Gordillo-Otero, A. Diaz-Valverde, M.A. Amaro-Lopez, andR. Moreno-Rojas, Food Chem. 70,85 (2000).

8. M. Yaman, M. Gunes, and S. Bakird-ere, Bullet. Environ. Contam. Tox-icol. 70(3), 437 (2003).

9. Y.W. Park, Small Rum Res. 37, 115(2000).

10. I. Karadjova, S. Girousi, E. Iliadou,and I. Stratis, Mikrochim. Acta134(3-4,: 185 (2000).

11. R.M. Rojas, P.J.S. Segarra, M.G.Martinez, M.J.G. Otero, and M.A.A.Lopez, Milchwissenschaft-Milk Sci-ence International 55(9), 510(2000).

12. .A.O. Musaiger, M.A. Ahmed , andM.V Rao, Food Chem. 61(1/2), 49(1998).

13. M. Yaman and M. Durak,Spectrosc Lett. 38, 405 (2005).

14. M. Yaman and N. Cokol, At. Spec-trosc. 25(4), 185 (2004).

15. R.A. Nadkarni, Anal. Chem. 56,2233 (1984).

16. F.E. Smith and E.A. Arsenault,Talanta 43(8), 1207 (1996).

17. M. Yaman and I. Akdeniz, Anal. Sci.20, 1363 (2004).

18. M. Yaman, Anal. Biochem. 339, 1(2005).

19. WHO, Fifty-third report of the JointFAO/WHO Expert Committee onFood Additives, WHO TechnicalReport Series 896, Geneva,Switzerland (2000).

20. Turkish food codex-Turk GidaKodeksi Teblig. Resmi Gazete, 23Eylul 2002, Sayı: 24885. Ankara:Basbakanlık Basimevi (2002).

21. T. Berg and D. Licht, Food Add.Contam. 19(10), 916 (2002).

107

*Corresponding author.E-mail: [email protected]: 385-1-4680216Fax: 385-1-4680242

Copper Determination by ETAAS in Fish Tissue CytosolsWith Minimal Sample Pretreatment

Zrinka Dragun* and Biserka RasporRuđer Bošković Institute, Division for Marine and Environmental Research,

Bijenička cesta 54, P.O. Box 180, 10002 Zagreb, Croatia


INTRODUCTION

During the fall and spring sea-sons of 2005 and 2006, a metal pol-lution study was carried out in theSava River (Croatia) as part of theEuropean Union Sixth FrameworkProgramme project (SARIB project,INCO-CT-2004-509160). Amongothers, metal bioavailability in theriver water was assessed using thebioindicator organisms (1,2). Fishare often used as bioindicators ofwater pollution with metals (3),and since the European chub(Squalius cephalus L.) is the fishspecies widespread in Europeanfreshwater, it was selected for thisstudy. Two representative tissues,gills and intestine, were chosen forthe metal determinations. As organsin direct contact with the ambientwater and ingested food, they areexpected to respond quickly tochanges in metal exposure (4).

For environmental analyses, thedetermination of total metal con-centrations in digested tissues is theusual procedure. From an analyticalpoint of view, the problem ofmatrix complexity is thereby elimi-nated, but valuable information onsubcellular metal distribution andtrophically available metal levels islost. In our study, metals weretherefore determined in tissuecytosol fractions, which containheat-sensitive proteins (such asenzymes) and heat-stable proteins(such as metallothioneins) (5). Thepresence of increased metal con-centrations in the fraction contain-ing heat-sensitive proteins(potentially metal-sensitive fraction)is expected to be associated with

To avoid the analytical interfer-ences of a complex organic matrix,tissue cytosols should be digestedprior to Cu determination. How-ever, microwave digestion requiresthe use of additional chemicals(mineral acid, hydrogen peroxide)and expensive equipment. It is,therefore, a costly and time-consuming procedure, as well as apotential source of sample contami-nation. It should also be pointedout that the concentrations of met-als in the cytosols of fish tissues arerather low (1,2), and any additionalsample dilution could result in con-centrations below the instrumentdetection limit. Considering thesefacts, we measured the concentra-tions of several metals (Fe, Zn, Mn,Cu, and Cd) directly in undigestedgill and intestine cytosols to avoidpossible sample contamination andover-dilution, and to make the pro-cedure as simple and low cost aspossible. Iron and Zn weremeasured by flame atomic absorp-tion spectrometry (AAS), while Mn,Cu, and Cd were measured by elec-trothermal atomic absorption spec-trometry (ETAAS) due to their lowconcentrations in diluted cytosols(Cu and Mn <30 µg L–1; Cd <2 µg L–1)(1,2). Metal determination byETAAS in fish tissue cytosols with-out pretreatment has, to our knowl-edge, never been reported in theliterature. However, the concentra-tions of Al, Cd, Co, Cr, Cu, Fe, Mn,Mo, Ni, Pb, and Se were measuredin different types of undigestedsamples (milk, urine, serum, blood,wheat flour slurry, coconut water)but usually with the use of variouschemicals as dilution or modifiersolutions. The oxidant mixture(HNO3+H2O2) was most frequentlyapplied, but the use of ethanol, ter-tiary amines, Triton® X-100 and Pdwas also reported (7-15).

ABSTRACT

The metal concentrations inthe fish tissue cytosols offer animportant parameter in environ-mental monitoring studies as pos-sible indicators of metalbioavailability in ambient water.The aim of this study was toexamine the possibility of directCu determination in the cytosolsof chub gills and intestines byelectrothermal atomic absorptionspectrometry (ETAAS) with mini-mal sample pretreatment andwithout the use of modifiers.Tube wall and platform atomiza-tions were tested. When tubewall atomization was applied, aprogressive decline of Cuabsorbance was observed withincreasing firing count. On theother hand, stability of Cuabsorbance throughout thegraphite tube lifetime (up to250 firings) was obtained withplatform atomization. Signal sta-bility, together with excellentmeasurement repeatability (≤5%)and acceptable Cu recovery fromspiked samples (77%), makeETAAS with platform atomizationapplicable to the determinationof low Cu concentrations(<27 µg L–1) in environmentalsamples with complex organicmatrices, such as undigested fishtissue cytosols, even without theuse of chemical modifiers.

some biological impairment (6).Metals sequestered by metalloth-ioneins, on the other hand, are con-sidered detoxified, and thus notavailable to more sensitive cellularfractions (4). The cytosolic metalconcentrations, as potential indica-tors of metal bioavailability in ambi-ent water, represent the importantand very interesting parameter inenvironmental monitoring studies.

108

The main objective of our studywas to solve the problems encoun-tered as a consequence of the com-plex organic matrix in the direct Cudetermination in undigested fishtissue cytosols. The comparativemeasurements on tube wall andplatform atomization wereperformed by ETAAS without useof chemical modifiers. Due todecreased background, improvedlong-term stability and minimalchemical interferences, the use ofplatforms is recommended for themeasurement of metals in sampleswith complex matrices (16). How-ever, we first attempted to measureCu by applying tube wall atomiza-tion due to lower cost of simplegraphite tubes compared to tubeswith integrated platforms. Theresults obtained with tube wall andplatform atomization werecompared, and a more favorableapproach for Cu measurement inthe undigested fish tissue cytosolswas additionally analytically charac-terized. The detection and quantifi-cation limits, measuring ranges,repeatability, and Cu recovery fromspiked samples are listed, providingnew information about the rarelyreported direct metal determina-tion in cytosolic types of samplesby ETAAS (17).

EXPERIMENTAL

Isolation of Gill and IntestineCytosols

The gills (ranging from 0.5 to 1.0 g)and the intestines (ranging from1.0 to 2.0 g) were obtained fromspecimens of European chub(Squalius cephalus L.) caught inthe Sava River (Croatia), then cutup and diluted in five volumes ofcooled homogenizing buffer (1,2).The applied gill homogenizationbuffer (GHB) was 100 mM Tris-HCl/Base buffer (Sigma Chemical Co.,USA) with a pH equal to 8.1 at 4°C,and supplemented with the reduc-ing agent, 1 mM dithiotreitol(Sigma Chemical Co., USA). The

intestine homogenization buffer(IHB) additionally containedinhibitors of proteolitic activity,0.5 mM phenylmethylsulphonylfluoride (PMSF) and 0.006 mM leu-peptine (LEU) (18). High puritywater (Milli-Q™ Water System, Mil-lipore, Madrid, Spain) was used forpreparation of the homogenizationbuffers. Homogenization was per-formed in ice-cooled tubes using aGlas-Col homogenizer (USA), set at6,000 rpm. The homogenates werethen centrifuged in the SorvalRC28S centrifuge (Kendro, USA) at50,000×g for 2 h at 4°C. After cen-trifugation, supernatant (S50),which is a water-soluble cytosolictissue fraction, was separated fromthe pellet, then five times dilutedwith Milli-Q water (1:5), and deep-frozen until subsequent metaldetermination. Diluted cytosolscontained proteins [determinedaccording to Lowry et al. (19)] in aconcentration of approximately2–4 mg mL–1 (1,2).

Standard and ReferenceSolutions

Regarding the protein content,it was impossible to match thecomposition of the standard solu-tions and the cytosol samples.

Thus, the Cu calibration standardswere prepared by appropriate dilu-tions of a certified Cu standard solu-tion (1,000 mg L–1, ICP standard in2–3% HNO3, Merck, Darmstadt,Germany) with five times dilutedGHB and IHB (as described above).The reference solution SW-HM-47,distributed by Vituki (Budapest,Hungary) as a part of the QualcoDanube intercomparison study,was used to examine the possibleinfluence of the calibrationstandard composition on Cu deter-mination. This reference solution isan acidified sample of river waterwith an assigned Cu concentrationof 14.1 µg L–1.Instrumentation

A Varian SpectrAA™ 220 (Aus-tralia) deuterium-corrected atomicabsorption spectrometer, equippedwith a Varian GTA™ 100 graphitetube atomizer, an autosampler, anda multielement hollow cathodelamp (Cu/Zn) was employed in thisstudy. Argon (99.99% purity) wasused as a purge gas. The instrumen-tal parameters are presented inTable I. The Cu absorbance signalwas calculated by using the peakarea mode to minimize potentialdifferences in the atomization rateand signal shape between the cali-

TABLE ICommon Instrumental Parameters for Cu Measurement

in Undigested Gill and Intestine Cytosols by ETAASWith Tube Wall and Platform Atomization

Blank Gills 20 mM Tris HCl/Base, Background Correction D2 Lamp0.2 mM DTT Copper Standards 9.0 µg L–1

Blank Intestine 20 mM Tris HCl/Base, 18.0 µg L–10.2 mM DTT, 27.0 µg L–10.1 mM PMSF,

1.2 µM leupeptineSampling Mode Automix Calibration Algorithm QuadraticMeasurement Mode Peak Area Sample Volume 20 µLWavelength 324.8 nm Total Volume 35 µLSlit Width 1.0 nm aCopper Bulk Conc. 18.0 µg L–1Lamp Current 7.0 mA Gas Type Argon

a Copper bulk concentration refers to the concentration of Cu standard solutionwhich is used for preparation of the other two calibration standards in automixsampling mode.

109


1a). However, with an increasingnumber of tube firings, thedecrease of Cu absorbance signalbecame obvious (Figure 1b). It was,therefore, impossible to comparethe results obtained with theincreasing number of firings withthose obtained with the newgraphite tube.

Tube ageing was previouslyreported as one of the main factorsthat can affect metal determinationby ETAAS (23). This problem hadto be solved to obtain more reliableCu measurements, and thus the sta-bility of the Cu absorbance signalwas compared for tube wall andplatform atomization. Cu concen-trations were recorded periodicallyin the selected samples of gill andintestine cytosols during tube uti-lization, approximately up to 200tube firings. The study wasperformed separately for gill andintestine cytosols to establish iftheir different matrices damage thetube at different rates. For intestinecytosols, the same sample (IN-1,Table III) was used for both tubewall and platform atomization mea-surements, so the startingabsorbance was the same in bothcases (Figure 2). For gill cytosols,due to the small sample volume,two samples of different concentra-tions were used: G-1 for tube wallatomization and G-2 for platformatomization (Table III). When tubewall atomization was used, a pro-gressive decline of Cu absorbancesignal was observed (Figure 2).After ∼200 firings, the signaldecreased approximately 30%, andthe rates of tube impairmentcaused by gill and intestine cytosolmatrices were comparable. Whentube wall was replaced with plat-form atomization, the Cuabsorbance signal recorded after∼200 firings was comparable to thesignal obtained with the new tube.It varied only slightly (signal RSD3–4%) throughout tube usage (Fig-ure 2). A possible explanation forsignal decline is that the determina-

gram to ensure that the tempera-ture of the platform is the same foreach sample injection (22).


Effect of Tube Ageing

At first, Cu determination wasperformed directly in undigestedcytosols by a well-establishedETAAS method with tube wallatomization, which is commonlyused for Cu measurement in acidi-fied solutions (21). The methodwas applied to more complex sam-ple matrices without anyadjustments and without use ofmodifiers. The obtained results ini-tially seemed to give reliable infor-mation on Cu concentrations in fishtissue cytosols. Excellent repeatabil-ity was achieved after Cu determi-nation in duplicates, with relativestandard deviations (RSD) mostlybelow 3%, while low backgroundabsorbance pointed to completeremoval of the complex matrix dur-ing sample ashing at 800°C (Figure

bration standards and the samples.It was shown experimentally thatdifferences in matrix compositioncan alter peak width and height;therefore, absorbance integrationused in the peak area modeproduces smaller variations in thesignals than peak height measure-ment (20). The cytosolic Cu con-centrations were measured by tubewall atomization, using partitionedpyrolitically coated graphite tubes,and by platform atomization, usingthe universal tubes with integratedplatforms. The heating programsfor tube wall (21) and platformatomization are presented in TableII. There is a time delay betweenthe temperature of the tube and thetemperature of the platform, andtherefore the heating program usedfor graphite tubes with platformsinvolved higher drying, ashing, andatomization temperaturescompared to the heating programused for tube wall atomization. Acool-down step is also included atthe end of the platform heating pro-

TABLE IIHeating Programs for Cu Measurement by ETAAS

With Tube Wall and Platform AtomizationaHeating Program for bHeating Program forTube Wall Atomization Platform Atomization

Stage Temp. Time Gas Stage Temp. Time GasFlow Flow

(°C) (s) (L min–1) (°C) (s) (L min–1)

Hot Inject 60 Hot Inject 100Drying 85 5.0 3.0 Drying 100 5.0 3.0

95 40.0 3.0 210 25.0 3.0120 10.0 3.0 300 10.0 3.0

Ashing 800 5.0 3.0 Ashing 1000 5.0 3.0800 2.5 3.0 1000 30.0 3.0800 0.5 0.0 1000 2.0 0.0

Atomization 2300 1.1 0.0 Atomization 2600 0.8 0.02300 2.0 0.0 2600 2.0 0.0

Clean 2800 2.0 3.0 Clean 2800 2.0 3.02800 5.0 3.0 2800 3.0 3.0

Cooling 40 21.9 3.040 35.0 3.0

a Heating program for tube wall atomization was taken from Varian GTA manual(21).b Heating program for platform atomization was recommended by Varian experts(personal communication).

110

tion of trace metals in biologicalsamples with complex organicmatrices gives rise to the formationof carbonaceous residues inside thetube, which could seriously alterthe effectiveness of the drying andashing temperatures (24).

According to Welz et al. (25),the analytical useful lifetime ofgraphite tubes is defined by thenumber of firings which can bemade until the analytical signaldrops to about 80% of its initialvalue, and/or the RSD begins todeteriorate significantly. The severeeffect of tube ageing observedwhen Cu is measured in chub tis-sue cytosols using tube wall atom-ization significantly reduced thetube lifetime. Based on our analy-ses, it is limited to approximately130–150 firings (Figure 2). How-ever, even within this limited num-ber of firings, the signal was notstable, but continually decreasing(Figure 2). Therefore, unlike tubewall atomization, the use of plat-form atomization can be recom-mended for Cu measurements inundigested fish tissue cytosols evenwithout the use of modifiers, sinceit provides longer stability of theabsorbance signal. Based on ourpractical experience, graphitetubes with integrated platforms canbe used up to 250 firings withoutsignal deterioration.

Method Characterization

Since the main purpose of eachanalytical measurement is to obtainreliable information about the ana-lyte content in the analyzed sample(26), it is important to characterizethe method performance wheneverapplied to new types of samples.For the characterization of the Cudetermination in undigested chubtissue cytosols by ETAAS with plat-form atomization, the limit ofdetection, limit of quantification,linearity of measuring range,repeatability, and recovery fromspiked samples were assessedmainly following the International

Fig. 1. Illustration of Cu peak profile for gill cytosol sample (G1) obtained by tubewall atomization: (a) in the new tube; (b) after approximately 150 firings.

Fig. 2. Dependence of individual Cu absorbance signals periodically measured ingill and intestine cytosols on the number of tube firings for tube wall and platformatomizations, as indicated in the legend.

111


Recommendation R100 (27). Fivetimes diluted cytosols of six gill andfive intestine samples were used,with the Cu concentrations rangingfrom 7.25–16.66 µg L–1 and18.51–26.45 µg L–1, respectively(Table III). Due to the differencesin matrix composition of gill andintestine cytosols, method charac-terization was performed separatelyfor each type of cytosol.

Calibration

The preliminary results indicatedthat Cu concentrations in the fivetimes diluted cytosols of chub tis-sues mainly fall into the concentra-tion range of 10-20 µg L–1. Since itis optimal that the calibration rangeencompasses values up to 150% ofthe expected measurements (28),the linearity was tested up to theCu concentration of 27 µg L–1,

which was also roughly indicatedas the upper calibration limit for Cudetermination by ETAAS (21).According to some recommenda-tions (29), it is sufficient to calcu-late the regression coefficient as ameasure of linearity; and when itreaches at least 0.999, it can beassumed that the analytical signal isproportional to the measured ana-lyte concentration in the definedcalibration range. The regressioncoefficients for Cu calibrationstraight lines were equal to 0.997and 0.998 for gills and intestinecalibrations, respectively (Figure 3a–b), and thus nearly compliedwith the requirements for linearity.However, better fit was achievedwhen the quadratic calibrationcurve was applied, with higherregression coefficients (r=0.9999in both cases) and lower intercepts(a=0.003-0.004). The slopes of thequadratic calibration curves(b=0.013-0.014) were comparablefor both calibrations (Figure 3 a-b).

TABLE IIIConcentrations of Cu in Diluted Gill and Intestine Cytosols

Measured by ETAAS With Platform Atomization (an=10)

Gill Cu Intestine CuCytosols (µg L–1) Cytosols (µg L–1)G-1 14.03±0.56 IN-1 18.51±0.69G-2 7.25±0.08 IN-2 21.55±1.18G-3 8.55±0.11 IN-3 20.65±0.52G-4 9.60±0.15 IN-4 19.24±0.41G-5 16.66±0.45 IN-5 26.45±0.20

G-6 14.59±0.40

a n = number of replicates.

Fig. 3 (a-b). Copper calibration curves obtained by ETAAS with platform atomization; LOD and LOQ values are denoted on thegraph, as well as the regression equations: (a) Cu calibration standards prepared with GHB (Tris/DTT; see text); (b) Cu calibra-tion standards prepared with IHB (Tris/DTT/PMSF/LEU; see text).

112

The differences in the compositionof calibration standards obviouslydid not affect the size of Cuabsorbance signal.

Method Detection Limits andMeasuring Ranges

For the determination of thedetection and quantification limits(LOD and LOQ, respectively), Cuconcentrations were measured con-secutively 10 times in the blanksolutions, and their compositionsare provided in Table I. The LODwas calculated on the basis of threestandard deviations of the blank(27), and the LOQ on the basis of10 standard deviations of the blank(30).

For gill calibration (standardsprepared with GHB; see sectionabove), the Cu LOD amounted to

11.3–16.9 µg L–1. Copper concen-tration in SW-HM-47 (obtainedwhen the calibration curve wasprepared with the acidified stan-dards) was 14.7 µg L–1. Similar Cuconcentrations were obtained fromtwo calibration curves, preparedwith gill and intestine homogeniza-tion buffers (described above), andamounted to 14.8 and 15.2 µg L–1,respectively. Based on comparableresults, whether the calibrationcurve was created with acidifiedsolutions (104%) or homogeniza-tion buffers (105–108%), it can beconcluded that the complex com-position of the calibration standardsused in this study did not affect theaccuracy of the Cu measurement.

Since the certified referencematerial with a matrix correspond-ing to gill or intestine cytosol wasnot commercially available, the reli-ability of Cu determination in thecytosols using platform atomizationwas tested by establishing Curecovery from the spiked samples.The volumes of 300 µL of five-timediluted gill or intestine cytosolswere mixed with the appropriatevolumes of Cu calibrationstandards. Although the cytosolswere additionally diluted with cali-bration standards, the Cu recoverystill should indicate if the complexcytosol matrix affects the measure-ment reliability. Copper recoveryfrom spiked gill cytosols (Table V)was 77±9%, while it was higher forspiked intestine cytosols (89±2%).Unlike the components of thehomogenizing buffers, the organicmatrix clearly caused the decreasein Cu recovery. Still, the recoveriesof added Cu from both the gill andthe intestine cytosols can beregarded as acceptable results forenvironmental monitoring studies,in which the comparability of theresults is more important than theabsolute accuracy.

0.27 µg L–1 (Figure 3a), while forintestine calibration (standards pre-pared with IHB; see section above)the LOD amounted to 0.82 µg L–1(Figure 3b). The presence of twoadditional components in the intes-tine calibration standards, PMSFand leupeptine, evidently increasedthe noise (standard deviation of theblank). Therefore, the measuringrange, which is defined as therange of Cu concentrations fromthe LOQ value to the upper limitof the calibration range (26), isbroader for the gill calibrationswith the lower limit set at 1 µg L–1.The lower limit of the measuringrange for the intestine calibrationswas set at 3 µg L–1.Repeatability

The repeatability was determinedfor the gill and intestine calibrationstandards (see above) with fourconcentrations selected within thecalibration range. It was also estab-lished for six gill and five intestinecytosol samples. The indicator ofthe measurement repeatability wasRSD (%) of 10 consecutively mea-sured Cu concentrations in eachcalibration standard solution and ineach cytosol sample (27).

The obtained RSDs of the Cumeasurements in standard solutionswere <5%, whether the calibrationstandards were prepared with GHBor IHB (Table IV). The repeatabilityobserved for the Cu determinationin undigested cytosols was alsoexcellent, and the RSDs weremainly ≤5% for both gill and intes-tine cytosol samples (Table IV).

Influence of Calibration Stan-dard and Sample Compositionon Cu Recovery

The influence of calibration stan-dard composition on Cu recoverywas tested using the reference acid-ified river water sample (SW-HM-47), with the assigned Cuconcentration of 14.1 µg L–1 andthe range of acceptance from

TABLE IVRepeatability of Cu

Measurements by ETAASWith Platform Atomization in

Gill and Intestine Cytosols,as well as in two Types of

Calibration Standards ( a n=10)cG cIN(% RSD)

b Standards 4 µg L–1 0.8 1.7

9 µg L–1 4.6 3.3

18 µg L–1 4.7 1.1

24 µg L–1 0.5 0.9

c Samples G-1/IN-1 4.0 3.7G-2/IN-2 1.1 5.5G-3/IN-3 1.3 2.5G-4/IN-4 1.6 2.1G-5/IN-5 2.7 0.7

G-6 2.7

a n = number of replicates.b Standards for gill calibrationprepared with Tris/DTT buffer; stan-dards for intestine calibrationprepared with Tris/DTT/PMSF/LEUbuffer.c Gill cytosol samples (G), intestinecytosol samples (IN).

113


CONCLUSION

The severe deterioration of Cuabsorbance signal as a consequenceof tube ageing was observed duringthe direct Cu measurements inundigested fish gill and intestinecytosols by ETAAS with tube wallatomization. Due to the longer sig-nal stability (up to 250 tube firings),excellent repeatability, and accept-able recovery from spiked samples,ETAAS with platform atomizationcan be recommended for the deter-mination of low Cu concentrationsin fish tissue cytosols, requiring nei-ther previous sample digestion noruse of additional modifiers.

ACKNOWLEDGMENTS

The financial support by theMinistry of Science, Education andSport of the Republic of Croatia(project No. 098-0982934-2721) isacknowledged. This study was car-ried out as a part of the EuropeanFP6 project SAva RIver Basin: Sus-tainable Use, Management and Pro-tection of Resources(INCO-CT-2004-509160). Theauthors are especially grateful toM.Sc. Vlatka Filipović Marijić andM.Sc. Marijana Podrug for isolationof the intestine and gill cytosols,and determination of the proteinconcentrations.

REFERENCES

1. Z. Dragun, B. Raspor, and M. Podrug,Chemosphere 69, 911 (2007).

2. V. Filipović Marijić, and B. Raspor,Toxicol. Lett. 172S; S159 (2007).

3. A. Chovanec, R. Hofer, and F.Schiemer, Fish as bioindicators. In:B.A. Markert, A.M. Breure, andH.G. Zechmeister (eds) Bioindica-tors and Biomonitors: Principles,Concepts and Applications. Else-vier Science Ltd., Amsterdam, pp.639-676 (2003).

4. L.D. Kraemer, P.G.C. Campbell, andL. Hare, Environ. Pollut. 138, 324(2005).

5. W.G. Wallace, and S.N. Luoma, Mar.Ecol.-Prog. Ser. 257, 125 (2003).

6. W.G. Wallace, T.M.H. Brouwer, andG.R. Lopez, Environ. Toxicol.Chem. 19, 962 (2000).

7. P. Viñas, N. Campillo, I. López-Gar-cía, and M. Hernández-Córdoba,Anal. Chim. Acta 356, 267 (1997).

8. N. Campillo, P. Viñas, I. López-Gar-cía, and M. Hernández-Córdoba,Anal. Chim. Acta 390, 207 (1999).

9. N. Campillo, P. Viñas, I. López-Gar-cía, and M. Hernández-Córdoba,Talanta 48, 905 (1999).

10. M. González, M. Gallego, and M.Velcárcel, Talanta 48, 1051 (1999).

11. N. Campillo, P. Viñas, I. López-Gar-cía, and M. Hernández-Córdoba,Anal. Biochem. 280, 195 (2000).

12. A.A. Almeida, and J.L.F.C. Lima,Atom. Spectrosc. 22, 324 (2001).

13. P.R.M. Correia, E. de Oliveira, andP.V. Oliveira, Talanta 57, 527(2002).

14. P.C. Aleixo, and J.A. Nóbrega, FoodChem. 83, 457 (2003).

15. J. Naozuka, and P.V. Oliveira, J.Braz. Chem. Soc. 17, 521 (2006).

16. L.M. Voth-Beach, VarianInstruments At Work, No. AA-54(1985).

17. Z. Dragun, and B. Raspor, J. Anal.Atom. Spectrom. 20, 1121 (2005).

18. V. Filipović Marijić, and B. Raspor,Mar. Pollut. Bull. 54, 935 (2007).

19. O.H. Lowry, N.J. Rosebrough, A.L.Farr, and R.J. Randall, J. Biol.Chem. 193, 265 (1951).

20. W. Slavin, D.C. Manning, and G.R.Carnrick, Atom. Spectrosc. 2, 137(1981).

21. E. Rothery, Analytical methods forgraphite tube atomizers. VarianAustralia Pty Ltd., Mulgrave, publi-cation No. 85-100848-00 (1988).

22. L.M. Voth, Varian Instruments AtWork, No. AA-45 (1985).

23. M.J. Cal-Prieto, A. Carlosena, J.M.Andrade, P. López-Mahía, S. Muni-ategui, and D. Prada, J. Anal. Atom.Spectrom. 18, 29 (2003).

24. R. Sabé, R. Rubio, and L. García-Beltrán, Anal. Chim. Acta 419, 121(2000).

25. B. Welz, G. Schlemmer, H.M. Ort-ner, and W. Wegscheider, Prog.Anal. Spectrosc. 12, 111 (1989).

26. P. Konieczka, Crit. Rev. Anal.Chem. 37, 173 (2007).

27. International organization of legalmetrology, International Recom-mendation R100: Atomic Absorp-tion Spectrometer systems formeasuring metal pollutants inwater. OIML Secretariat, USA,TC16/SC2 (2002).

28. M. Thompson, S.L.R. Ellison, and R.Wood, Pure Appl. Chem. 74, 835(2002).

29. M. Dobecki, Zapewnienie jakościanaliz chemicznych (in Polish).Instytut Medycyny Pracy im. Prof.J. Nofera, Łódź (2004).

30. J.K. Taylor, Quality Assurance ofChemical Measurements. LewisPublishers Inc., Chelsea (1987).

TABLE VCopper Recovery From Chub

Gill and Intestine CytosolsSpiked With Standard Solutions(Volumes of 300 µL of diluted gillor intestine cytosols were mixed

with the appropriate volumes of Cucalibration standards; the results

are referring to the ETAAS measure-ments with platform atomization.)

Sample Added c RecoveryCu of Added Cu

(µg L–1) (µg L–1) (%)a G-1 10.91 8.07 74.0

G-2 5.14 4.38 85.2

G-3 8.18 6.80 83.1

G-4 9.60 7.44 77.5

G-5 16.62 13.20 76.5

G-6 9.60 5.81 60.5

Average Recovery 76.6±8.9b IN-1 12.00 10.59 88.2IN-2 15.00 13.87 92.4IN-3 15.00 13.29 88.6IN-4 12.00 10.38 86.5IN-5 15.00 13.49 89.9

Average Recovery 89.1±2.2

a G = gill cytosol.b IN=- intestine cytosol.c Average of three consecutivemeasurements.

008108_03

PerkinElmer Life andAnalytical Sciences710 Bridgeport AvenueShelton, CT 06484-4794 USAPhone: (800) 762-4000 or(+1) 203-925-4602www.perkinelmer.com

Vol. 29(2), March/April 2008

tsomicpectroscopy - a and b spectroscopy 29(3).pdf · •u.s.$70.00includesfirst-class...

Documents