speciation of arsenic by hydride generation–atomic absorption spectrometry (hg–aas) in...

19
Talanta 50 (1999) 1109–1127 Speciation of arsenic by hydride generation – atomic absorption spectrometry (HG – AAS) in hydrochloric acid reaction medium Amjad Shraim a, *, Barry Chiswell a , Henry Olszowy b a Department of Chemistry, The Uni6ersity of Queensland, St. Lucia, Qld 4072, Australia b Queensland Health Scientific Ser6ices, 39 Kessels Rd., Coopers Plains, Qld 4108, Australia Received 31 March 1999; received in revised form 7 July 1999; accepted 15 July 1999 Abstract The effects on the absorbance signals obtained using HG – AAS of variations in concentrations of the reaction medium (hydrochloric acid), the reducing agent [sodium tetrahydroborate(III); NaBH 4 ], the pre-reducing agent (L-cysteine), and the contact time (between L-cysteine and arsenic-containing solutions) for the arsines generated from solutions of arsenite, arsenate, monomethylarsonic acid (MMA), and dimethylarsenic acid (DMA), have been investigated to find a method for analysis of the four arsenic species in environmental samples. Signals were found to be greatly enhanced in low acid concentration in both the absence (0.03–0.60 M HCl) and the presence of L –cysteine (0.001–0.03 M HCl), however with L-cysteine present, higher signals were obtained. Total arsenic content and speciation of DMA, As(III), MMA, and As(V) in mixtures containing the four arsenic species, as well as some environmental samples have been obtained using the following conditions: (i) total arsenic: 0.01 M acid, 2% NaBH 4 , 5% L-cysteine, and contact time B10 min; (ii) DMA: 1.0 M acid, 0.3–0.6% NaBH 4 , 4.0% L-cysteine, and contact time B5 min; (iii) As(III): 4–6 M acid and 0.05% NaBH 4 in the absence of L-cysteine; (iv) MMA: 4.0 M acid, 0.03% NaBH 4 , 0.4% L-cysteine, and contact time of 30 min; (v) As(V): by difference. Detection limits (ppb) for analysis of total arsenic, DMA, As(III), and MMA were found to be 1.1 (n =7), 0.5 (n =5), 0.6 (n =7), and 1.8 (n =4), respectively. Good percentage recoveries (102 – 114%) of added spikes were obtained for all analyses. © 1999 Elsevier Science B.V. All rights reserved. Keywords: Arsenic; Arsenite; Arsenate; Monomethylarsonic acid; Dimethylarsenic acid; Speciation; L-cysteine; Hydride generation; Selective reduction www.elsevier.com/locate/talanta 1. Introduction Arsenic, which is potentially toxic to humans, animals, and plants [1–4] and according to recent reports [5–8] may be carcinogenic to humans, occurs naturally in many chemical forms. Arsenic * Corresponding author. Fax: +61-7-33653839. E-mail address: [email protected] (A. Shraim) 0039-9140/99/$ - see front matter © 1999 Elsevier Science B.V. All rights reserved. PII:S0039-9140(99)00221-0

Upload: independent

Post on 15-Nov-2023

0 views

Category:

Documents


0 download

TRANSCRIPT

Talanta 50 (1999) 1109–1127

Speciation of arsenic by hydride generation–atomicabsorption spectrometry (HG–AAS) in hydrochloric acid

reaction medium

Amjad Shraim a,*, Barry Chiswell a, Henry Olszowy b

a Department of Chemistry, The Uni6ersity of Queensland, St. Lucia, Qld 4072, Australiab Queensland Health Scientific Ser6ices, 39 Kessels Rd., Coopers Plains, Qld 4108, Australia

Received 31 March 1999; received in revised form 7 July 1999; accepted 15 July 1999

Abstract

The effects on the absorbance signals obtained using HG–AAS of variations in concentrations of the reactionmedium (hydrochloric acid), the reducing agent [sodium tetrahydroborate(III); NaBH4], the pre-reducing agent(L-cysteine), and the contact time (between L-cysteine and arsenic-containing solutions) for the arsines generated fromsolutions of arsenite, arsenate, monomethylarsonic acid (MMA), and dimethylarsenic acid (DMA), have beeninvestigated to find a method for analysis of the four arsenic species in environmental samples. Signals were foundto be greatly enhanced in low acid concentration in both the absence (0.03–0.60 M HCl) and the presence ofL–cysteine (0.001–0.03 M HCl), however with L-cysteine present, higher signals were obtained. Total arsenic contentand speciation of DMA, As(III), MMA, and As(V) in mixtures containing the four arsenic species, as well as someenvironmental samples have been obtained using the following conditions: (i) total arsenic: 0.01 M acid, 2% NaBH4,5% L-cysteine, and contact timeB10 min; (ii) DMA: 1.0 M acid, 0.3–0.6% NaBH4, 4.0% L-cysteine, and contact timeB5 min; (iii) As(III): 4–6 M acid and 0.05% NaBH4 in the absence of L-cysteine; (iv) MMA: 4.0 M acid, 0.03%NaBH4, 0.4% L-cysteine, and contact time of 30 min; (v) As(V): by difference. Detection limits (ppb) for analysis oftotal arsenic, DMA, As(III), and MMA were found to be 1.1 (n=7), 0.5 (n=5), 0.6 (n=7), and 1.8 (n=4),respectively. Good percentage recoveries (102–114%) of added spikes were obtained for all analyses. © 1999 ElsevierScience B.V. All rights reserved.

Keywords: Arsenic; Arsenite; Arsenate; Monomethylarsonic acid; Dimethylarsenic acid; Speciation; L-cysteine; Hydride generation;Selective reduction

www.elsevier.com/locate/talanta

1. Introduction

Arsenic, which is potentially toxic to humans,animals, and plants [1–4] and according to recentreports [5–8] may be carcinogenic to humans,occurs naturally in many chemical forms. Arsenic

* Corresponding author. Fax: +61-7-33653839.E-mail address: [email protected] (A. Shraim)

0039-9140/99/$ - see front matter © 1999 Elsevier Science B.V. All rights reserved.

PII: S0039 -9140 (99 )00221 -0

A. Shraim et al. / Talanta 50 (1999) 1109–11271110

poisoning is a major public health problem, espe-cially in countries such as Bangladesh, where thesoil is high in arsenic compounds and the wellwater may contain as much as 300–4000 mg L−1

of arsenic [9].The toxicity of arsenic varies widely, ranging

from highly hazardous inorganic arsenicals (ar-sine, arsenite, and arsenate) to relatively harmlessorganic species (monomethylarsonate anddimethylarsenate) [10–14]. Indeed some organo-arsenicals, such as arsenobetaine and arseno-choline, are effectively non-toxic towards livingorganisms [15]. Therefore, determination of totalarsenic content in a sample does not reflect thelevel of hazard of the element actually present,and it is increasingly important that the variousforms of arsenic be determined in biological andenvironmental samples to provide a much clearerview of the risk associated with exposure to ar-senic in the environment. Reported arsenic con-centrations in some parts of Australia range up to36 000 mg kg−1 in surface soils and 300 mg kg−1

in ground water [16]; the current ANZECC/NHMRC arsenic guideline for soil is 100 mgkg−1, while the NHMRC drinking water guide-line has recently been reduced from 50 to 7 mg/l[16].

Speciation of arsenic in environmental samplesusually involves several steps including derivitiza-tion, separation, and detection; hydride genera-tion (HG), initially developed by Braman andForeback in 1973 [17], coupled to one of severalseparation and detection systems has been foundto be one of the most common techniques usedfor derivitization of arsenic. A number of detec-tion systems have been used in arsenic analysis, ofwhich the most popular and preferred one interms of simplicity, sensitivity, precision, speed,and cost is AAS.

Our aim in the work described here was todevelop a simple, rapid, and inexpensive tech-nique for the speciation of the most commonlyoccurring species in environmental samples, viz.As(III), As(V), monomethylarsonic acid (MMA),and dimethylarsenic acid (DMA) using hydridegeneration–atomic adsorption spectrometry(HG–AAS), i.e. a derivitization and detectiontechnique only, without the need for a separation

step such as HPLC and cryogenic trapping, whichwe believe has caused many complications andresulted in lengthy analytical techniques. To com-pensate for the absence of a separation step in ourmethod, a selective–reduction concept is em-ployed to generate hydrides of each arsenicspecies.

Critical evaluation of the existing literature,showed that some work has been done on the useof HG–AAS for the speciation of arsenic [18–21],and much of this work dealt with the use ofthiol-containing ligands, such as L-cysteine, L-cystine, and thioglycerol, to obtain identical re-sponse from all four arsenic species. These thiolshave been used as pre-reducing agents before theaddition of NaBH4, and have been shown toenhance the arsine signals in low acid concentra-tion and to reduce the effects of interferences[20,22–24].

In this study, the effect of HCl as a reactionmedium on the arsine generation from the fourarsenic species has been investigated. Control ofthe concentrations of the reaction medium (HCl),of the reducing and hydride generating agent(NaBH4), and of the pre-reducing agent (L-cys-teine) when used, and the employment of HG–AAS has resulted in methods for the analysis andspeciation of the four arsenic species in environ-mental samples.

2. Experimental

2.1. Equipment

A Vapour Generation Accessory (VGA-76,Varian) connected to an Atomic Absorption Spec-trometer (Spectra 300, Varian) was used in thisstudy, and operated according to manufacturer’sinstructions; instrument parameters used are sum-marised in Table 1. In this system, arsenic-con-taining solutions were pumped into a mixer andreacted with sodium tetrahydroborate(III) solu-tion; generated arsines were swept to a gas–liquidseparator using nitrogen gas and then to a heatedT-shaped absorption cell.

A. Shraim et al. / Talanta 50 (1999) 1109–1127 1111

2.2. Reagents and solutions

2.2.1. GeneralAll chemicals were of analytical-reagent grade

unless otherwise specified. All glassware wassoaked in 4 M HNO3 for a minimum of 12 h andwashed with distilled water and finally rinsed withMilli-Q reagent water before use. All water usedwas obtained from a Milli-Q reagent system (Mil-lipore), resistivity 18 MV cm. L-cysteine (mini-mum 98%, TLC) was obtained from Sigma (St.Louis, MO), sodium tetrahydroborate(III) formMerck and Alfa, As(III) atomic absorption stan-dard solution (1 mg ml−1) and As(V) (As2O5,99.999%) from Acros, MMA (disodium methy-larsenate, 99%) from Chem Service, and DMA(cacodylic acid, 98%) was obtained form Aldrich.Various concentrations of NaBH4 solution sta-bilised with NaOH were used; the concentrationof NaOH was maintained at 0.5% in experimentswhere NaBH4 concentrations exceeded 0.6%.While both reagents were kept at the same con-centration when NaBH4 was used at concentra-tion levels B0.6%.

2.2.2. Arsenic stock solutionsThe arsenic stock solutions were prepared as

follows: 1000 ppm As(III): arsenic(III) atomic ab-sorption standard solution (1 mg ml−1 As in 2%potassium hydroxide).

1000 ppm As(V): 0.03835 g of arsenic(V) oxide(As2O5) was dissolved in a minimum volume of4.0 M NaOH, neutralised by a same volume of4.0 M HCl, and the final volume was adjusted to25 ml with Milli-Q water.

1000 ppm MMA: 0.09844 g of disodium methy-larsenate (CH3AsO(ONa)2 · 6H2O) was dissolvedin 25 ml Milli-Q water.

1000 ppm DMA: 0.04699 g of cacodylic acid(CH3)2AsO(OH) was dissolved in 25 ml of Milli-Qwater. Cacodylic acid is the commercial name forDMA, the acronym used in this work.

The stock solutions of As(V), MMA, andDMA were prepared monthly and stored in glassvolumetric flasks wrapped with aluminum foil andkept refrigerated at 4°C to prevent any change inspeciation. The As(III) stock solution was alsokept refrigerated at 4°C. All arsenic solutionswere found to be stable under these conditionswhen tested after 1 month.

2.3. Analytical procedures

Solutions were prepared as required by appro-priate dilution of stock solutions and additions ofthe required volumes of hydrochloric acid andL-cysteine solutions to achieve the required con-centrations. Solutions were then rapidly mixed.When using L-cysteine, measurement of contacttime was commenced upon addition of L-cysteineand stopped at the beginning of the analysis.Throughout this study contact time refers to thetime that has been allowed for L-cysteine to reactwith the arsenic-containing solutions before thecommencement of the introduction of solutions toHG–AAS.

3. Results and discussion

To understand the role of NaBH4 as a reducingand HG agent in the analysis of arsenic, thefollowing mechanism has been proposed [25,26]:

RnAs(O)(OH)3−n+H++BH4−

�RnAs(OH)3−n+H2O+BH3 (1)

RnAs(OH)3−n+ (3−n)BH4− + (3−n)H+

�RnAsH3−n+ (3−n)BH3+ (3−n)H2O (2)

Table 1Operating conditions of the HG–AAS system

Instrument mode Absorbance

ConcentrationCalibration modeIntegrationMeasurement mode

Slit width (nm) 0.5NormalSlit height

Wavelength (nm) 193.7Flame Air–acetylene

NormalSample introductionDelay time (s) 40Time constant 0.05

2.0Measurement time (s)Replicates 3

OnBackground correction7Sample flow rate (ml min(1)

NaBH4 flow rate (ml min(1) 1

A. Shraim et al. / Talanta 50 (1999) 1109–11271112

Table 2Steps followed for the analysis of arsenic speciesa

NaBH4 concentration L-cysteine concentrationSteps Contact timeHCl concentration

CStep 1 0V NAStep 2 C V 0 NA

VStep 3 C C CV CC CStep 4

Step 5 CC C V

a V=varied; C=constant; 0=zero; NA=not applicable.

BH3+3H2O�H3BO3+3H2. (3)

While in the presence of L-cysteine the proposedmechanism is shown below [23,25,27]:

2RSH+R*n AsO(OH)3−n

�RS–SR+R*n As(OH)3−n+H2O (4)

R*n As(OH)3−n+ (3−n) RSH

�R*n As(SR)3−n+2H2O (5)

R*n As(SR)3−n+ (3−n)BH4−

�R*n AsH3−n+ (3−n)BH3+ (3−n)RS− (6)

BH3+3H2O�H3BO3+3H2. (7)

We have used these proposed mechanisms to de-velop a method for the speciation of arsenic usingthe steps shown in the Table 2.

3.1. Effect of HCl concentration when using 0.6%NaBH4 solution in the absence of L–cysteine

The effect of using 0.001–9.6 M HCl on theabsorbance signals in the absence of L-cysteine isshown in Fig. 1. The absorption signals of As(III),MMA, and DMA sharply increase with increase inacid concentration over a range of 0.01–0.10 M,while the increase in the As(V) absorption signal ismuch slower. Further increase in the acid concen-tration results in:

(a) a sharp decrease in the DMA signal up to anacid concentration of 1.0 M, then a slow decreaseto negligible values beyond 4.0 M,

(b) a broad maximum for MMA over an acidconcentration range of 0.2–1.0 M, and a slowdecrease thereafter to a negligible signal at 9.60 M,

(c) a slow increase in the As(III) signal up to an

acid concentration of 2.0 M and a very broadplateau afterwards, and

(d) a much slower increase in the As(V) signal,when compared to the As(III) signal, up to an acidconcentration of 4 M, and a broad plateau there-after.Our results are similar to those obtained byworkers who used similar instrumentation andanalytical procedure. For example Anderson et al.[18] and Hakala and Pyy [28] used a HG–AAS anda HCl concentration of up to 5.0 M and reportedsimilar results. However, changes in experimentalconditions may cause major differences in results,as shown in the work of Rude and Puchelt [19], whoused flow injection analysis (FIA)–HG–AAS andHCl concentrations of up to 5.0 M, and found thefollowing:1. at HCl concentration of \1.0 M, they obtained

an increasing response with a linear function(r=0.995) for As(III) which disagrees with ourand the findings of others [18,20,29], in whichAs(III) showed a constant response. This in-creasing linear response is most probablycaused by incomplete generation of arsine,which is due to the use of a low KBH4 concen-tration (0.2%). As reported below, results ob-tained by us when using low NaBH4

concentrations are similar to those of Rude andPuchelt [19];

2. the low KBH4 concentration used by Rude andPuchelt [34] yielded a negligible As(V) ab-sorbance across the whole acid concentrationrange;

3. MMA has an almost identical response toDMA over the whole acid range (0.0–5.0 M)covered by the study of Rude and Puchelt [19],and both MMA and DMA showed a negligibleresponse in 2.0 M and higher HClconcentrations.

A. Shraim et al. / Talanta 50 (1999) 1109–1127 1113

Further examination of Fig. 1, shows thefollowing:1. the sharp increase in the absorption signals of

the arsenic species, especially As(III), MMA,and DMA, with increase in the acid concentra-tion over the low range (0.1–0.6 M), suggeststhat an investigation into the use of higherNaBH4 concentrations may be warranted todetermine whether this condition might in-crease the signal of As(V) and produce identi-cal responses from all four arsenic species andthus enable the determination of total arsenic;

2. at low acid concentration (0.01–0.1 M), DMAshows higher signals when compared to theother three species. An investigation into theuse of NaBH4 concentrations seemed appro-priate and suggested that a low acid and lowNaBH4 concentrations could result in a condi-

tion where all species signals, except that ofDMA, are reduced to negligible values, thusleading to the speciation of DMA in the pres-ence of the other three species,

3. the observation that the DMA and MMAresponse signals decrease to negligible levels athigh HCl concentrations whilst As(III) andAs(V) essentially provide a constant positiveresponse prompted another investigationwhere an As(III) speciation might be possible,with the As(III) signal retained, by use of lowNaBH4 concentration at high acidconcentrations.

To address the above-mentioned points the effectof NaBH4 concentration over the range 0.02–2.0% when using low (0.005–0.1 M) and moder-ate to high acid concentrations (2.0–8.0 M) hasbeen investigated.

Fig. 1. Effect of HCl concentration on the absorption signals of As(III), As(V), MMA, and DMA (40 ppb As each) when using 0.6%NaBH4, in the absence of L-cysteine.

A. Shraim et al. / Talanta 50 (1999) 1109–11271114

Fig. 2. Effect of NaBH4 concentration on the absorption signals of As(III), As(V), MMA, and DMA (40 ppb As each) when using0.06 M HCl (A) and 0.10 M HCl (B) in the absence of L-cysteine.

3.2. Effect of NaBH4 concentration in theabsence of L-cysteine

3.2.1. Use of low concentrations of HClThe effect of using 0.1–2.0% NaBH4 and low

acid concentrations (0.06 and 0.1 M) on the ab-sorption signals of the four arsenic species isshown in Fig. 2. The use of NaBH4 concentra-tions up to 2.0% failed to produce identical sig-nals from the four arsenic species, and thus underthese conditions, determination of a total arsenicsignal is not possible. Further increase in NaBH4

concentrations up to 4.0% was found to causelarge instability in the signals, and could not beused to check possible increases in signal re-sponses. This instability may be caused by eitheror both of the following two reasons:1. it was observed that the preparation of NaBH4

solutions with concentrations \1.0% resultsin cloudy solutions (an indication of undis-solved particles) which settle down with timeand become clear. The settled particles tend todissolve when mixed with acid solution insidereaction tubes, producing a non-homogeneous

reaction mixture with localised cells of NaBH4

concentrations thus causing the signal instabil-ity. Filtering the solution, or leaving it to settleand withdrawing clear portions partially re-duced this instability;

2. The use of high concentrations of NaBH4 mayresults in a vigorous production of arsines, aswell as the production of large quantities of H2

gas, where both can cause large signal instabil-ity. Several droplets of solution were observedat the top of the gas-liquid separator whenusing high NaBH4 concentrations, which sup-port this assumption.

It is also clear from Fig. 2 that the use of NaBH4

concentrations of B0.6% has resulted in a de-crease in the signals of the four arsenic species.Fig. 2A and B show that DMA still yields arelatively large signal, whereas the signals of theother three species are reduced to lower values.Nevertheless the signals of As(III), As(V), andMMA effectively result in a total interference of60–70% in the DMA signal. In order to reducethis interference, the use of lower acid concentra-tions has been investigated; results are shown inFig. 3.

A. Shraim et al. / Talanta 50 (1999) 1109–1127 1115

Fig. 3A and B show that the use of low NaBH4

concentrations, when using 0.03 and 0.01 Macid, resulted in a high DMA signal with lowinterference from As(III) and negligible signalsfor both As(V) and MMA. On the other hand,at a NaBH4 concentration of 2.0% in 0.01 MHCl (Fig. 3B), DMA showed a reduced res-ponse, but with no interference from theother three species. Further decrease in acidconcentration to 0.005 M eliminated all the inter-

ferences and produced significant signals fromDMA at NaBH4 concentrations of B0.6% (Fig.3C).

To eliminate the interference of As(III) in theDMA signal when using 0.01 M acid (see Fig.3B), the use of NaBH4 concentrations of lowerthan 0.1% has been investigated; as shown in Fig.3D, this approach has eliminated the interferenceof As(III) in the DMA signal when NaBH4 con-centrations of 0.02–0.075% are used.

Fig. 3. Effect of NaBH4 concentration on the absorption signals of As(III), As(V), MMA, and DMA (40 ppb As each) when usingA: 0.03; B: 0.01; C: 0.005 and D: 0.01 M HCl in the absence of L-cysteine.

A. Shraim et al. / Talanta 50 (1999) 1109–11271116

Fig. 4. Effect of NaBH4 concentration on the absorption signals of As(III), As(V), MMA, and DMA (40 ppb As each) when usingA: 2; B: 4; C: 6; and D: 8 M HCl in the absence of L-cysteine.

The use of 0.005 M HCl and 0.1% NaBH4

in the absence of L-cysteine has produced excel-lent results for the speciation of DMA in amixture containing all four arsenic species (Table3A).

On the other hand the use of 0.07 and 2.0%NaBH4 with 0.01 M HCl yields very large errors.Although the other three arsenic species show nosignals when analysed separately (see Fig. 3B andD), they yielded a combined interference of 27.3

and 56.3% when NaBH4 concentrations of 0.07and 2.0%, respectively were used.

3.2.2. Use of moderate to high concentrations ofHCl

The effect of using 0.02–2.0% NaBH4 andmoderate to high acid concentrations (2, 4, 6, and8 M) on the absorption signals of the four arsenicspecies is shown in Fig. 4; the results indicate thatthe use of NaBH4 concentrations of 50.1% atany acid concentration produces high signals from

A.

Shraim

etal./

Talanta

50(1999)

1109–

11271117

Table 3Experimental conditions for speciation of arsenic in mixtures containing the four arsenic species

%NaBH4Speciesa As usedc Nd Avg e %Errf dg Equationh R2i LWRj DLk[HCl], M % L-cysteine TbNo.

80 7 20.2 0.8 0.46 y=0.005408xA 0.9987DMA 0–40 0.90.005 0.0 N/A 0.180 7 20.0 0.1 0.59 y=0.005587x 0.99280.05 0–80As(III) 0.6B N/A0.04.0

0.05 80 8 19.7 –1.6 0.36 y=0.009332x 0.9963 0–40 0.86.0 0.0 N/A10 1.0 40 3 39.6 –1.0 0.82 y=0.030760x 0.9986 0–20 1.0C 0.01TAs 4.0

40 7 40.8 1.9 0.72 y=0.032440x 0.99132.0 0–105.0 1.150.010.6DMA 80 5 21.7 8.7 0.56 y=0.012853x 0.9961 0–20 0.51.0 4.0 4D

30 0.03 80 4 20.4 2.2 0.56 y=0.005693x 0.9952 0–80 1.84.0E 0.4MMA

a Species for which analysis was undertaken.b Contact time after which the first reading was taken.c Total arsenic in the mixture solution, equal concentrations of each of the four arsenic species were added (ppb).d Number of readings taken for each speciation analysis.e Average concentration of arsenic species found (ppb).f Percentage error from the average.g S.D. (ppb).h Equation of calibration curve.i R2 value of calibration curve.j Linear working range of calibration curve (ppb).k Detection limit (ppb) calculated from three times the S.D. of the replicated blanks.

A. Shraim et al. / Talanta 50 (1999) 1109–11271118

As(III) only and negligible signals from the otherthree species. The increase in acid concentrationfrom 2 to 8, when using NaBH4 concentrations of0.02–2.0%, slightly increased the As(V) signal,but has little effect on the As(III) signal, while thesignals of the organic arsenicals decrease withincrease in the acid concentration.

From the results shown in Fig. 4, it appearslikely that a signal for As(III) only can be ob-tained when acid concentrations of 2–8 M andlow concentrations of NaBH4 are used. To checkthis, the analysis of arsenic solution mixtures us-ing 4 and 6 M HCl and 0.05% NaBH4 has beencarried out and found to yield excellent results(see Table 3B).

Controlling the concentrations of acidand NaBH4 in the absence of L-cysteine hasresulted in methods for the speciation ofDMA and As(III). To find methods for the speci-ation of the other species and the analysis oftotal arsenic, the use of L-cysteine has been intro-duced.

3.3. Effect of HCl concentration when using 0.4%L-cysteine, 0.6% NaBH4 solutions, and constantcontact time

The effect of HCl concentration (0.001–9.2 M)and 0.6% NaBH4, in the presence of 0.4% L-cys-teine, on the absorption signals of the four arsenicspecies is shown in Fig. 5. The presence of L-cys-teine produces a very rapid increase in the signalsfor the four arsenic species with increase in acidconcentration over a lower and narrower range of0.001–0.03 M HCl compared to the signals ob-tained in the absence of L-cysteine (see Fig. 1).The four species showed similar maxima at anacid concentration 0.03–0.06 M with very similarabsorption signals for As(III), As(V), and DMAand a lower signal for MMA (Fig. 5B). As dis-cussed later, improvements in the MMA signaland production of similar signals from all fourspecies have been achieved by using more concen-trated solutions of L-cysteine and NaBH4 allowingthe determination of total arsenic content.

Fig. 5. Effect of HCl concentration on the absorption signals of As(III), As(V), MMA, and DMA (40 ppb As each) when using 0.6%NaBH4 and 0.4% L-cysteine after a contact time of 2 h.

A. Shraim et al. / Talanta 50 (1999) 1109–1127 1119

Chen et al. [23] used a HG–DCP–AES tostudy the effect of low concentrations of HCl andHNO3 (0.002–0.1 M) on the arsine generationfrom As(III) and As(V) only, in the absence andthe presence of L-cysteine. In agreement with ourfindings, they reported that the As(III) signal, inthe presence of L-cysteine, was increased by morethan 75% when using an acid concentration of:0.02 M. They also reported that the reductionof As(V) to As(III) in the presence of L-cysteine isslow and time-dependent. Our results showedidentical responses from both As(III) and As(V)over the entire HCl concentration range (0.001–9.2 M) in the presence of 0.4% L-cysteine.

Le et al. [20] have also studied the effect of HClconcentration (\0.0–4.0 M) on the arsine gener-ation from As(III), As(V), MMA, and DMA inthe presence and the absence of many pre-reduc-ing agents, such as L-cysteine and methionine,using a FIA–HG–AAS with 2.5% NaBH4. Theyobtained maximum and identical responses fromthe four arsenic species when using 0.3–0.7 MHCl after 10–20 min of contact with 2.5% L-cys-teine. These results are different to ours and to thework of Anderson et al. [18] in which very similarresponses from the four arsenic species were ob-tained when using much lower and narrower acidconcentration ranges of 0.01–0.03 M HCl and0.06–0.1 M mercaptoacetic acid, respectively.Reasons behind these differences are unclear, butthe use of high concentrations of L-cysteine andNaBH4 may be responsible. Also, the reportedresults for the determination of MMA and DMAin acid concentration above 1 M are inconsistent;thus Le et al. [20] found that DMA gave a highersignal than MMA, whereas our results show theopposite. The increase in the As(III) and As(V)signals at \2 M acid was also much slower thanthat found by us. The use of high concentrationsof L-cysteine and NaBH4 by Le et al. [20] may beresponsible for these differences.

Further examination of Fig. 5 shows thefollowing:1. the use of high acid concentrations of ]6.0 M

produces similar signals from all species, ex-cept DMA, which showed a lower signal; wehave been unsuccessful in our attempts toincrease the DMA signal,

2. obtaining a signal for only DMA at an acidconcentration of :0.6 M is possible if theinterferences of MMA, As(III), and As(V) sig-nals are eliminated,

3. obtaining an MMA signal at an acid concen-tration of 2–3 M is also possible, if the inter-ference of the other three species is alsoeliminated,

4. the presence of L-cysteine produces similarsignals from both As(III) and As(V) over thewhole acid concentration range, supportingthe proposed mechanism shown in Eqs. (4)–(7).

Based on these observations, we proceeded tostudy the effect of using NaBH4 and L-cysteineconcentrations other than 0.6 and 0.4%, respec-tively in 0.01, 0.6, and \2.0 M HCl.

3.4. Effect of NaBH4 concentration

3.4.1. Effect of NaBH4 concentration onobtaining a total arsenic signal when using 0.01M HCl, 2.5% L-cysteine, and constant contacttime

The effect of NaBH4 concentration when using0.01 M HCl, 2.5% L-cysteine over a contact timeof 50–70 min is shown in Fig. 6. The MMAsignal has been enhanced and similar signals fromthe four arsenic species have been produced, whena minimum NaBH4 concentration of 0.6% is em-ployed; this allows for determination of a totalarsenic signal. A long contact time has been usedto make sure that the reduction of arsenic(V)species to arsenic(III) analogues, and subsequentgeneration of arsine is complete. Fig. 6 also indi-cates that increase in NaBH4 concentration up to0.6% results in sharp increase in absorption sig-nals of all four arsenic species; no more increase isobserved with increase in NaBH4 concentrationbeyond 0.6%.

3.4.2. Effect of NaBH4 concentration onobtaining a sole signal for DMA when using 0.6M HCl, 0.4% L-cysteine, and constant contacttime

Fig. 5 indicates that at an acid concentration of:0.6 M, obtaining a signal for only DMA maybe possible if the signals for the other three spe-

A. Shraim et al. / Talanta 50 (1999) 1109–11271120

Fig. 6. Effect of NaBH4 concentration on the absorption signals of As(III), As(V), MMA, and DMA (40 ppb As each) when using0.01 M HCl and 2.5% L-cysteine after a contact time of 50–70 min.

cies are reduced to negligible values. Based onthis, the effect of NaBH4 concentrations of 0.02–2.0% and 0.4% L-cysteine at a contact time ofabout 2 h has been investigated as shown in Fig.7. It is evident that a DMA signal, although low,can be obtained if NaBH4 concentrations between0.05 and 0.1% are employed. The use of NaBH4

concentrations of \0.6% will significantly in-crease the signals of the other species and there-fore increase their interference with the DMAsignal.

3.4.3. Effect of NaBH4 concentration onobtaining a sole MMA signal when using low tomedium HCl concentrations, 0.4% L-cysteine, andconstant contact time

The effect of NaBH4 concentrations of 0.02–2.0% and 0.4% L-cysteine at a contact time ofabout two h when using 1.5–6.0 M acid, has beeninvestigated as shown in Fig. 8. The use of lowconcentrations of NaBH4 (B0.03%) when using3–4 M acid (Fig. 8B and C) produces a high

signal only from MMA with negligible interfer-ence from the other three species. The use of anacid concentration of B3 M (Fig. 8A) has in-creased the interference caused by DMA, whilethe use of acid concentrations of \4 M (Fig. 8D)has increased the As(III) and As(V) interferenceswith the MMA signal.

3.5. Effect of contact time

3.5.1. Effect of contact time on obtaining a totalarsenic signal when using 0.01 M HCl, 4.0%L-cysteine, and 1.0% NaBH4

In an attempt to find a short analysis time fordetermination of total arsenic signal when usingthe experimental conditions described in Section3.4.1, but with an L-cysteine concentration in-creased to 4%, the effect of contact time of 1–20min on the absorption signals of solutions con-taining single arsenic species was investigated. Itwas found that a minimum of 5–10 min wasneeded before obtaining similar signals from all

A. Shraim et al. / Talanta 50 (1999) 1109–1127 1121

four species. It was also noticed that signals of allspecies were largely increased with increase incontact time from 1 to 5 min, except for As(III),where the signal reached a maximum from thebeginning and remained so until the end of theanalysis time.

To confirm these results, the analysis oftotal arsenic in a solution mixture containing40 ppb as total arsenic (10 ppb each of thefour species) using the above-mentioned ex-perimental conditions was investigated. The re-duction of arsenic species was found to becomplete after 8–10 min under these experi-mental conditions, and very good results fortotal arsenic were achieved. However the useof higher concentrations of L-cysteine (5%)and NaBH4 (2.0%) reduces the reduction timeto 5 min; very good results for total arsenic werealso obtained (Table 3C). Therefore the useof 0.01 M HCl under these experimental condi-tions can be used for the determination of totalarsenic.

Low concentrations of HCl and HNO3 in thepresence L-cysteine, have been used in literaturefor the determination of total arsenic content[20,22–24], and two techniques similar to ourshave been employed in these studies e.g. FIA–HG–AAS [20,22] and HG–Direct CurrentPlasma–AES [23,24]. Le et al. [20] allowed 10–20min as a contact time between the sample andL-cysteine when using FIA–HG–AAS for thedetermination of total arsenic in urine samples.For each 10 ml sample in 2% L-cysteine, theconcentration and the flow rate of the HCl andthe NaBH4 were 0.5 M and 3.4 ml min−1, and0.65 M and 3.4 ml min−1, respectively. In com-parison, our results have shorter contact time(5–10 min) when using lower acid concentration(0.01 M), and NaBH4 concentrations of 1.0–2.0%at a higher L-cysteine concentration (4.0%).

In the study by Yin et al. [22], the contact timeallowed for the determination of total inorganicarsenic when using FIA–HG–AAS was muchlonger than ours; they found that the reduction of

Fig. 7. Effect of NaBH4 concentration on the absorption signals of As(III), As(V), MMA, and DMA (40 ppb As each) when using0.60 M HCl and 0.40% L-cysteine after a contact time of 2 h.

A. Shraim et al. / Talanta 50 (1999) 1109–11271122

Fig. 8. Effect of NaBH4 concentration on the absorption signals of As(III), As(V), MMA, and DMA (40 ppb As each) when usingA: 1.5; B: 3; C: 4 and D: 6 M HCl and 0.4% L-cysteine after a contact time of 2 h.

As(V) (500 ml of 2 ppb As) to As(III) was com-pleted within 60, 40, and 20 min when using 0.04,0.08, and 1.6 M, respectively of L-cysteine in 0.024M HCl (or 0.029 M HNO3) and 0.5% NaBH4.Flow rates of acid and NaBH4 were 7.8 and 5.4ml min−1, respectively. In the present study, theAs(V) reduction was completed in B10 min com-pared to 40 min in their work.

In the other two studies [23,24], the reductionof As(V) required a longer time to complete, but

when the sample and L-cysteine were boiled for ashort time before the introduction of NaBH4, thereduction was immediate.

In the work of Chen et al. [23], 0.01 M HCl (orHNO3) has been employed to determine the totalinorganic arsenic using HG–DCP–AES. TheAs(V) content was determined as follows: 1 ml of2% L-cysteine was injected into a 5 ml sample (thefinal L-cysteine concentration was 0.33%), fol-lowed by 0.5% NaBH4. No As(V) signal was

A. Shraim et al. / Talanta 50 (1999) 1109–1127 1123

detected when the NaBH4 was added just after theaddition of L-cysteine, but when the reaction wasgiven 5 min before the addition of NaBH4, asmall signal was detected, which increased withincrease in contact time. The As(V) reduction wascomplete within 35, 60, and 135 min when 1.0,0.5, and 0.25 g, respectively of L-cysteine wereadded to 100 ml of 50 ppb As(V) when using 0.02M acid. But when the solution was heated inboiling water for 5 min, As(V) was completelyreduced to As(III). We have been able to obtaincomplete reduction of As(V) at room temperaturein a much shorter time (B10 min) when using4.0% L-cysteine.

In the other study, Brindle et al. [24] haveachieved an immediate and complete reduction ofAs(V) to As(III) after mixing the sample solutionwith L-cysteine and NaBH4 under the followingconditions: in a continuous flow HG–DCP–AESthe sample (10 ml min−1), L-cysteine (0.7%, 1.6ml min−1), and HNO3 (0.02 M, 10 ml min−1)were mixed and heated, online to 95–98°C,

cooled to room temperature, and reacted withNaBH4 (0.5%, 1.6 ml min−1).

It is evident from the last two studies [23,24],that the introduction of a heating step has elimi-nated any delay and spontaneously and com-pletely produced arsine from As(V). We willconsider this in future work.

3.5.2. Effect of the contact time on obtaining aDMA signal when using 0.6 M HCl, 4.0%L-cysteine, and 0.1 and 0.6% NaBH4

Use of 0.6% NaBH4 and applying the sameexperimental conditions of Section 3.4.2 (0.6 MHCl, 0.4% L-cysteine), results in negligible signalsfrom all species including DMA, even after theapplication of contact time of 40 min. As a resultthe L-cysteine concentration was increased to4.0% to obtain a high DMA signal within reason-able contact times. The effect of contact time,under the new experimental conditions is shownin Fig. 9, which shows high DMA signals withnegligible signals from the other three species up

Fig. 9. Effect of contact time on the absorption signals of As(III), As(V), MMA, and DMA (40 ppb As each) when using 0.60 MHCl, 4.0% L-cysteine, and 0.60% NaBH4.

A. Shraim et al. / Talanta 50 (1999) 1109–11271124

Fig. 10. Effect of contact time on the absorption signals of As(III), As(V), MMA, and DMA (40 ppb As each) when using 4.0 MHCl, 0.03% NaBH4, and A: 0.4% and B: 1% L-cysteine.

3.5.3. Effect of contact time on obtaining a soleMMA signal when using 4.0 M HCl, 0.2, 0.4,and 1.0% L-cysteine, and 0.03% NaBH4

Attempts to study the effect of contact time onspeciation of MMA was undertaken using 4.0 MHCl, 0.4 and 1.0% L-cysteine, and 0.03% NaBH4,and results are shown in Fig. 10. Use of 0.2%L-cysteine after a contact time of 20 min was alsoexamined and found to result in negligible signalsfrom all species. Increasing the concentration ofL-cysteine to 0.4% results in a linearly increasingsignal for MMA with increase in contact timefrom 1.8 to 35 min, while the other three speciesexhibited negligible signals (Fig. 10A). Therefore,under these conditions, an MMA signal can beobtained with good intensity and little interfer-ence from the other three species after a contacttime of around 20 min. However, the sole MMAsignal may also be obtained in a shorter contacttime, but with less intensity. On the other hand,the use of L-cysteine concentrations of \0.4%(1.0% as shown in Fig. 10B), under the above-mentioned conditions, yields a higher MMA sig-

to a contact times B18 min. It is also clear fromFig. 9 that a maximum of 12 min (the lowest timetried) can be safely used to obtain a signal forDMA only, in the presence of the other threespecies; shorter times can also be used.

Even though no interferences from the otherthree species on the DMA signals were foundwhen solutions of single arsenic species wereanalysed (see Fig. 9), the use of 0.6 M HCl, 0.6%NaBH4, and 4.0% L-cysteine for the analysis of amixture of the four arsenic species has producedhuge errors (70%) in the DMA analysis. Theincrease in HCl concentration to 1 M, under theabove experimental conditions, has significantlyreduced the interferences and produced very goodresults after a contact time of only 4 min (Table3D). By comparison, the use of 0.005 M HCl,0.1% NaBH4 in the absence of L-cysteine hasproduced better results with an error of only0.81%. However, the analysis of actual environ-mental samples will decide which method is moreappropriate for DMA speciation.

A. Shraim et al. / Talanta 50 (1999) 1109–1127 1125

nals over shorter contact times; however theDMA signal starts to appear after 6 min, andconsequently interferes with the MMA signal.Therefore, the best conditions, under these cir-cumstances, for obtaining a sole signal for MMAwith minimal interference from the other threespecies, are 4.0 M HCl, 0.03% NaBH4, and 0.4%L-cysteine after any contact time between 10 and35 min.

The speciation of MMA in a solution mixturecontaining all four arsenic species has beenachieved by using 4 M HCl, 0.4% L-cysteine, and0.03% NaBH4, but a minimum contact time of 30min should be provided to obtain good repro-ducible results (see Table 3E).

3.6. Calibration cur6es

Using appropriate conditions developed in thisstudy, five different calibration curves were con-structed for the speciation analysis. Arsenic spe-cies used to construct a calibration curve were thesame ones for which the speciation analysis was

undertaken. Although any species could be usedfor TAs analysis, As(III) was used in this study.

3.7. Analysis of en6ironmental samples

Two environmental water samples were ob-tained from Coen dam, Queensland, Australia; adam close to gold mining activities. The firstsample was taken before water treatment purifica-tion (BWT), while the other one was taken aftertreatment (AWT). Using experimental conditionsshown in Table 3, five sub-samples of each watersample were analysed; the first was used for theanalysis of TAs in the presence of L-cysteine, thesecond two sub-samples were used for the specia-tion of DMA and As(III) in the absence of L-cys-teine, and the last two sub-samples were used forthe speciation of DMA and MMA in the presenceof L-cysteine. Details of the results and experi-mental conditions applied to obtain the results aresummarised in Table 4.

The AWT water sample was found, as ex-pected, to contain much less TAs when compared

Table 4Results and details of experimental conditions used for the analysis of water samples

%L-cyst[HCl]Sample ID Arsenic (ppb)Analysis %NaBH4Ta(, min

Foundb Addedc Totald %Recoverye

0.01 5.0 5 2.0 38.9 8.0 47.5BWTf 107.5TAs2.8 16.0 19.1AWTg 101.9

87.518.380.00.80.1BWTf N/A0.00.005DMAAWTg 4.9 80.0 25.4 102.5BWTf 105.521.380.00.20.05N/A0.04.0As(III)

21.980.00.0 109.5AWTg

DMA 1.0 4.0 5 0.6 0.3 80.0 21.0 103.5BWTf

AWTg 0.0 80.0 20.6 103.0BWTf MMA 4.0 0.4 45 0.03 1.6 80.0 23.8 111.0

80.0AWTg 0.1 113.522.836.8As(V)hBWTf

AWTg 2.7

a Contact time (min).b Total arsenic concentration found (ppb) in unspiked sample.c Total arsenic concentration added (ppb) to spike the sample, equal quantities of each of the four arsenic species were added.d Total arsenic concentration found (ppb) in spiked sample.e %Recovery of added spike.f Water sample before water treatment.g Water sample after water treatment.h Calculated by difference, i.e. As(V)= [TAs−{As(III)+MMA+DMA}].

A. Shraim et al. / Talanta 50 (1999) 1109–11271126

to the BWT sample; TAs concentration of 2.8 ppbwas found which represents the arsenic residueleft after water treatment.

Two methods for the speciation of DMAwere applied; the first one in the absence ofL-cysteine, while the second method was in thepresence of L-cysteine. The first method yielded aDMA value of 0.8 ppb for the BWT sample, whilea much larger value of 4.9 ppb was obtained forthe AWT sample. The second method gave DMAvalues of 0.3 and 0.0 ppb for BWT and AWTsamples respectively. The DMA value obtainedfor the AWT sample when using the first methodappears to be unrealistically high and was re-jected.

As expected, As(V) was the main speciesfound in both samples; its concentration ]95%of TAs. The pH value of each sample wasfound to be :6.5, and as the dominancy of theinorganic arsenic species in natural and groundwaters is controlled by the pH values andthe oxidising or reducing conditions of suchwaters, it would be expected that As(V) wouldbe the most dominant species in oxygenatednatural waters, as it is the most thermody-namically stable species under these conditions[30–33]. Results for other natural water samples,which have been analysed for their arsenicspecies concentration by various workers, confirmthese predictions; As(V) was found to bethe predominant species in these studies [30,34–36] with values of greater than 90% of totalarsenic.

Initial attempts to assess the accuracy ofthe methods developed in this work were under-taken using the standard addition method. Table4 indicates that very good recoveries were ob-tained (101.9–113.5%); equal concentrationsfrom all four arsenic species were added toeach of the two water samples, ie. for the specia-tion of As(III), 20 ppb of each of the fourspecies was added to give a total arsenic con-centration of 80 ppb. From these samples 21.1and 21.9 ppb As(III) for the BWT and AWTsamples, respectively were recovered; these resultsrepresent percentage recoveries of 105.5 and109.5% for the BWT and AWT samples, respec-tively.

4. Conclusions

The results reported in this paper, for theanalyses and speciation of arsenic using methodsdeveloped in this work employing the selective-re-duction–HG–AAS technique, show that theseanalyses can be quickly and accurately under-taken using this simple and inexpensive technique.The only exception is the analysis of MMA,which needs a minimum of 30–45 min to providereliable results. The reduction of contact time incase of MMA may be achieved if different suit-able experimental conditions such as a heatingstep are introduced to this system.

There is no doubt that a much more compre-hensive analysis program of environmental sam-ples from a wide variety of sources has to beundertaken before the usefulness of these methodscan be accurately assessed.

Acknowledgements

The authors would like to thank Claire Mooreand Ron Sumner of the Queensland Health Scien-tific Services for their help in operating theHGAAS.

References

[1] J.P. Gustafsson, G. Jacks, Appl. Geochem. 10 (1995) 307.[2] A.R. Marin, P.H. Masscheleyn, W.H. Patrick, Plant Soil

152 (1993) 245.[3] J.R. Abernathy, Role of Arsenical chemicals in agricul-

ture, in: W.H. Lederer, R.J. Fentsterheim (Eds.), Arsenic:Industrial, Biomedical, Environmental Perspective’s,VNR, New York, 1983, pp. 57–62.

[4] K. Ringwood, Arsenic in the Gold and Base-Metal Min-ing Industry, Australian Minerals and Energy Environ-ment Foundation (AMEEF), Melbourne, Australia, 1995,pp. 1–33.

[5] M. Piscator, Life Sci. Res. Rep 33 (1986) 59.[6] M. Buat-Menard, P.J. Peterson, M. Havas, E. Steinnes,

D. Turner, Group Report: Arsenic, in: T.C. Hutchinson,K.M. Meema (Ed.), Lead, Mercury, Cadmium Arsenic inthe Environment, SCOPE 31, Pub. John Wiley and Sons,Chichester, England, 1987, pp. 43–48.

[7] G. Stohrer, Arch. Toxicol. 65 (1991) 525.[8] C. Hopenhayn-Rich, M.L. Biggs, A.H. Smith, D.A.

Kalman, L.E. Moore, Environ. Health Perspect. 104(1996) 620.

A. Shraim et al. / Talanta 50 (1999) 1109–1127 1127

[9] J. Beard, L. Cruces, New Sci. 28 (1998) 10.[10] R.W. Whitcare, C.S. Pearse, Miner. Ind. Bull. 17 (1974)

1.[11] G.M.P. Morrison, G.E. Batley, T.M. Florence, Chem.

Brit. 25 (1989) 791.[12] B. Amran, F. Lagrade, M.J.F. Leroy, A. Lamotte, M.

Olle, M. Albert, G. Rauret, J.F. Lopez-Sanchez, Tech.Instr. Anal. Chem. 17 (1995) 285.

[13] W.R. Cullen, K.J. Reimer, Chem. Rev. 89 (1989) 713.[14] J.M. Wood, Science 183 (1974) 1049.[15] S. Caroli, F.L. Torre, F. Petrucci N. Violante, Arsenic

speciation and Health Aspects, in: S. Caroli (Ed.), Ele-ment Speciation in Bioinorganic Chemistry, John Wileyand Sons, New York, 1996, pp. 445–463.

[16] A. Hinwood, R. Bannister, A. Shugg, M. Sim, Water 25(4) (1998) 34.

[17] R.S. Braman, C.C. Foreback, Science 182 (1973) 1247.[18] R.K. Anderson, M. Thompson, E. Culbard, Analyst 111

(1986) 1143.[19] T.R. Rude, H. Puchelt, Fresenius Z. Anal. Chem. 350

(1994) 44.[20] X.C. Le, W.R. Cullen, K.J. Reimer, Anal. Chim. Acta

285 (1994) 277.[21] I.D. Brindle, X.C. Le, Anal. Chem. 61 (1989) 1175.

[22] X. Yin, E. Hoffmann, C. Ludke, Fresenius Z. Anal.Chem. 355 (1996) 324.

[23] H. Chen, I.D. Brindle, X.C. Le, Anal. Chem. 64 (1992)667.

[24] I.D. Brindle, H. Alarabi, S. Karshman, X.C. Le, S.Zheng, H. Chen, Analyst 117 (1992) 407.

[25] A.G. Howard, J. Anal. At. Spectrom. 12 (1997) 267.[26] J. Aggett, A.C. Aspell, Analyst 101 (1976) 341.[27] A.G. Howard, C. Salou, Anal. Chim. Acta 333 (1996) 89.[28] E. Hakala, P. Lauri, J. Anal. At. Spectrom. 7 (1992) 191.[29] P.H. Masscheleyn, R.D. Delune, W.H. Patrick, J. Envir.

Qual. 20 (1991) 96.[30] Battelle, A Report: Speciation of Selenium and Arsenic in

Natural Waters and Sediments. Vol. 2: Arsenic Specia-tion, No. EPRI EA-4641, Pacific Northwest Laboratories,Sequim, Washington, 1986.

[31] L.E. Hunt, A.G. Howard, Mar. Pollut. Bull. 28 (1994) 33.[32] H. Hasegawa, Y. Sohrin, M. Matsul, M. Hojo, M.

Kawashima, Anal. Chem. 66 (1994) 3247.[33] J. Stummeyer, B. Harazim, T. Wippermann, Fresenius Z.

Anal. Chem. 354 (1996) 344.[34] P. Thomas, K. Sniatecki, J. Anal. At. Spectrom. 10 (1995)

616.[35] C.J. Hwang, S.J. Jiang, Anal. Chim. Acta 289 (1994) 205.[36] R.J.A. Van Cleuvenbergen, W.E. Van Mol, F.C. Adams,

J. Anal. At. Spectrom. 3 (1988) 169.

.