Simultaneous multi-element determination of hydride-forming elements by ‘‘in-atomiser trapping’’ electrothermal atomic absorption spectrometry on an iridium-coated graphite tube James Murphy,a Gerhard Schlemmer,b Ian L. Shuttler,b Phil Jonesa and Steve J. Hill*a aDepartment of Environmental Science, University of Plymouth, Drake Circus, Plymouth, Devon, UK PL4 8AA bBodenseewerk Perkin-Elmer GmbH., Postfach 107161, D-88647 U� berlingen, Germany Received 4th June 1999, Accepted 4th August 1999 A simultaneous multi-element approach utilising ‘‘in-atomiser trapping’’ electrothermal atomic absorption spectrometry (ETAAS) for As, Bi, Sb and Se was developed.The approach uses flow injection methodology and hydride formation with sodium tetrahydroborate to sequestrate the hydrides of the elements of interest on an Ir precoated graphite tube. Since the eYciency of the hydride formation depends on the oxidation state of the analyte, an oV-line reduction process was included to ensure that the analyte to be determined was in the most sensitive and favourable oxidation state.Initially five elements, As, Bi, Sb, Se and Te, were considered for simultaneous ‘‘inatomiser trapping’’. The elements were split into two groups reflecting the nature of the reducing agent required by each of the elements. Group A consisted of As, Bi and Sb and used L-cysteine as the reducing agent, whilst Group B consisted of Bi, Se and Te and used concentrated HCl as the reducing agent.However, Te was later removed from Group B due to problems in identifying a set of compromise conditions which enabled all three elements to be determined simultaneously. Bismuth featured in both groups as it did not require a reduction step. Various tube coatings were considered and Ir and Zr were evaluated. Iridium was found to be well suited to this application. The characteristic masses obtained using this method were 177, 91, 107 and 90 pg for As, Bi, Sb and Se, respectively, yielding detection limits (500 ml sample loop) of 0.82, 0.04, 0.26 and 0.29 mg l-1.Precision for analytes at the 5 mg l-1 level was typically better than 3.5% RSD. The method was validated by the analysis of two Certified Reference Materials and good agreement was found with the certified values. thus oVering characteristic concentrations in the low ng l-1 Introduction range;5 (c) a decrease or elimination of the interference eVects Hydride generation atomic absorption spectrometry in both the liquid and gas phases;5 and (d) easy integration (HGAAS) is a useful and sensitive technique for the determi- with flow injection methodology, thereby facilitating increased nation of those elements which form volatile hydrides.use of automation and allowing for a higher sample throughput Traditionally, HGAAS has been performed by passing the than for batch HGAAS methods.4 volatile hydrides, once formed, into an atomiser cell, where As stated above, the ‘‘in-atomiser trapping’’ technique utildecomposition of the hydride takes place.The atomiser cell ises the graphite tube as both the preconcentrating medium can either be a flame or more commonly a heated quartz tube. and the atomisation cell. The preconcentrating surface can be When coupled with flow injection (FI ) technology, the whole either the uncoated tube6 or a tube pre-treated with a precious process, i.e. sample introduction, control of the hydride gener- metal or a carbide-forming element.7,8 As the analytes are ation stage and the final measurement step, may be easily trapped on the graphite tube, this allows the carrier gas (Ar) automated.However, more recently the potential of using a and the hydrogen gas (reaction by-product) to escape before graphite electrothermal atomiser as both the hydride trapping the atomisation step, whereas in a quartz tube atomiser, these cell and the atomisation cell (‘‘in-atomiser trapping’’) has been gases would dilute the analyte during the atomisation stage.reported by a number of workers and recently reviewed by This extra gas production can also alter the hydride pro- Matusiewicz and Sturgeon.1 We have previously reported the duction/transfer rates which in a HG-quartz tube atomiser potential of ‘‘in-atomiser trapping’’ to allow simultaneous method can influence the atomisation signal. The in situ multi-element determination of Bi and Se.2 The work described preconcentration step is expected to eliminate the possible here extends this study to consider both the possibilities and influence of the hydride generation kinetics on the signal shape limitations of the simultaneous determination of five hydride- which in quartz tube atomisers leads to extended peak widths.9 forming elements, i.e.As, Bi, Sb, Se and Te. Performing the atomisation in the graphite tube atomiser The ‘‘in-atomiser trapping’’ technique has a number of means that conventional ETAAS peaks are obtained with peak advantages for the atomisation of hydride-forming elements widths of 1–5 s instead of 10–15 s (typical for quartz tube over the use of conventional heated quartz cells.These advan- atomiser) for a similar mass of analyte,2 and the careful tages can be summarised as: (a) the high absolute sensitivity control of the atomisation parameters, i.e. temperature, time intrinsic to electrothermal atomisation,3 where the character- and heating rates, is possible, leading to improved reproducistic masses are generally lower than for direct transfer ibility of the atomisation process and can reduce interferences.HGAAS with quartz tube atomisers;4 (b) the high relative A key requirement in achieving the simultaneous multielement determination of hydride-forming elements with sensitivity due to in situ preconcentration from larger samples, J. Anal. At. Spectrom., 1999, 14, 1593–1600 1593ETAAS is the selection of the instrumental parameters and and tetravalent Se.Based on a consideration of the chemistries involved the selected elements were divided into groups the hydride generation chemistries. The selection of the instrumental parameters is relatively straightforward, i.e. easily reflecting the optimum oxidation state required for hydride generation. Group A contained As, Sb and Bi, and used L- achieved by taking the lowest trapping temperature and the highest atomisation temperature from the optimum instrumen- cysteine as the reducing agent and Group B contained Bi, Se and Te and employed elevated temperatures with concentrated tal conditions for the group of elements to be determined.However, with respect to the hydride generation chemistries, HCl as the reducing agent. Bismuth was grouped with As and Sb due to it also being a Group Vb element; however, it can these have to be a compromise of the optimum single element conditions and these may show considerable variation. Clearly, also be determined with Se as shown in our earlier study.2 therefore, the sensitivity for a single element being determined will depend on the oxidation state of the analyte in the sample.Experimental To achieve the pre-reduction step required, potassium iodide10 or a mixture of potassium iodide and ascorbic acid11 Instrumentation has been used to reduce Asv to AsIII, but recently, L-cysteine A simultaneous electrothermal atomic absorption spectrometer has found favour12,13 since it is compatible with FI systems (Model SIMAA 6000, Bodenseewerk Perkin-Elmer, and uses reagents at lower concentrations.Other advantages U� berlingen, Germany) with a transversely heated graphite of L-cysteine include more eYcient reduction of As/SbV to the atomiser (THGA) with longitudinal Zeeman-eVect back- trivalent oxidation state, and the stabilising eVect on the ground correction equipped with a Model AS-72 autosampler analyte solutions.12,14 was employed for this study.Standard THGA tubes with For Se and Te, the highest oxidation state is hardly reduced integrated platforms and Perkin-Elmer EDL ‘‘System 2’’ elec- at all by tetrahydroborate,15 owing to the extremely slow trodeless discharge lamps wused. The instrumental param- kinetics of the reduction. Therefore, if total Se and Te conceneters used are shown in Table 1 and the THGA programme trations are required, a pre-reduction step must be used is shown in Table 2.implemented to reduce the hexavalent state to the tetravalent Flow injection (FI ) hydride generation was performed using state. For the reduction of SeVI to SeIV, the use of concentrated a commercial FI system (Perkin-Elmer FIAS 400) equipped hydrochloric acid is considered an eYcient and successful with a Model AS-91 autosampler. The control programme for method, with the chloride ion being the reductant.15,16 the FI system is shown in Table 3.Tygon tubing was used for For an ‘‘ideal’’ multi-element analysis, a simple preall reagent lines. The sample and acid carrier lines (id 1.52 mm) treatment step should be used, which converts all the elements had flow rates of 12 and 6 ml min-1, respectively, while the to their optimum oxidation state for hydride generation. Lreductant line (id 1.14 mm) had a flow rate of 4 ml min-1. Cysteine will reduce the As/SbV to the lower trivalent oxidation The PTFE capillary of the furnace autosampler was discon- state, but it will also reduce SeVI or SeIV to the elemental state, nected from the autosampler arm and replaced by a quartz while concentrated hydrochloric acid will reduce Se/TeVI to capillary (id 1.0 mm) and PTFE transfer line (id 2.0 mm) to the tetravalent state, but it has no eVect on reducing the convey the hydrides from the gas–liquid separator.The quartz As/SbV to the trivalent state.16 The ideal reductant that will capillary was carefully adjusted so as to inject the hydrides meet all the criteria discussed above (i.e.reduce the hydrideover the heated platform of the THGA without touching the forming elements to their most sensitive oxidation state) has platform or the tube walls. The optimum distance between the yet to be identified. However, various diVerent approaches quartz tip and the heated platform surface was 1.0 mm. have been tried to find the ‘‘ideal’’ reducing agent for simul- The graphite tube and platform were pre-treated with Ir taneous multi-element hydride generation determinations.after the new tube had been thermally conditioned. Using a Uggerud and Lund17 used a reducing agent consisting of two standard micropipette, a 50 ml volume of IrCl3 (1000 mg l-1) separate reagents (concentrated HCl and thiourea) to deterstock standard (Inorganic Ventures, Lakewood, NJ, USA) mine As, Bi, Sb, Se and Te by HG-ICP-AES. First, the was injected manually onto the platform and then the THGA concentrated HCl was added to the sample oV-line, to reduce was run using the programme previously described in the the Se and Te to the lower oxidation state and then the literature.8 This complete sequence was performed twice, and thiourea was added again oV-line to reduce any As/Sbv to the then the THGA tube was ready for use and no further coating lower+3 oxidation state. This combination of reducing agents or conditioning was necessary during the lifetime of the tube.works well, but the thiourea also slowly reduces the Se/TeIV The pre-treatment was suYcient to last the lifetime of the to the elemental state; therefore, once the thiourea has been tube, provided that the clean-out temperature was not excessive added, the analysis must be performed within a short period (>2300 °C). It was also possible to prepare batches of tubes of time for precise and reproducible results. Bowman et al.18 in this way and store them under dust-free conditions for took this idea of the two-component reducing agent one step future use.The system was controlled using AA Winlab further when they determined As, Sb and Se in water samples software (Perkin-Elmer Version Beta 2.38). by HG-ICP-MS. The concentrated HCl at an elevated tempera- The combined FI procedure and synchronisation of the ture of 80 °C was still added in an oV-line manner, but the atomiser time/temperature programme was controlled by the thiourea was added on-line to eliminate the time restraint. sequences shown in Table 4.This operation took 2 min and Stroh and Vollkopf19 used a diVerent approach to determine consisted of six individual steps, which started with the sample As, Bi, Hg, Sb, Se and Te in water and sea-water samples at loop being filled with fresh sample, sequestration of the hydride ultra-trace levels by FI-vapour generation-ICP-MS. The on the tube and finally the atomisation and measurement step.sample was split into two sub-samples, with one sample being The atomiser ran for 90 s (which included a 60 s trapping used to determine As and Sb by using a mixed reducing agent time) and the FI stage lasted for 75 s. Before the first replicate (5% m/v potassium iodide–5% m/v ascorbic acid), while the of any new sample, fresh sample was pumped into the sample second sub-sample was used to determine Se and Te using loop expelling any previous sample left in the tubing to waste.concentrated HCl at an elevated temperature as the reducing agent. Bi and Hg were found not to require any reducing agent; Reagents therefore, they could be determined in either sub-sample. The object of the study presented here was to assess the All reagents used were of pro analysi grade (Merck, Darmstadt, Germany) unless otherwise stated. Dilutions were made using performance of a multi-element approach where elements are present in their optimum oxidation states, i.e.trivalent As/Sb de-ionised water (Nanopure, Barnstead, Boston, MA, USA). 1594 J. Anal. At. Spectrom., 1999, 14, 1593–1600Table 1 Instrumental parameters Parameter Arsenic Bismuth Antimony Selenium Wavelength/nm 193.7 223.1 217.6 196.0 Lamp type As EDL Bi EDL Bi EDL Se EDL Lamp current/mA 400 380 380 280 Read time/s 5 5 5 5 Read delay/s 0 0 0 0 Carrier Group A 1% (v/v) hydrochloric acid Carrier Group B 10% (v/v) hydrochloric acid Reductant 0.5% (m/v) sodium tetrahydroborate in 0.5% (m/v) sodium hydroxide Reducing agent Group A 1% (m/v) L-cysteine in 0.1 mol l-1 hydrochloric acid Reducing agent Group B 50% (v/v) hydrochloric acid Signal measurement Integrated absorbance Signal type Background-corrected AA The reductant solution, 0.5% m/v NaBH4 in 0.5% m/v Table 2 Furnace programme NaOH, was prepared by dissolving 5.0 g of sodium hydroxide pellets followed by 5.0 g of sodium tetrahydroborate (95% Temperature/ Ramp time/ Hold time/ Gas flow/ assay, Riedel-de Ha�en, Seelze, Germany) and diluting to Step °C s s Read ml min-1 1000 ml.This solution was prepared daily and stored in a plastic Nalgene bottle. The carrier solution was either 1 or 1 a 1 60 50 2 a 1 20 250 10% v/v HCl depending on the group of analytes being deter- 3 2200 0 5 Yes 0 mined. The solutions were prepared by diluting either 10 or 4 2300 1 3 250 100 ml of concentrated hydrochloric acid (Suprapur, Merck) to 1000 ml. This solution was prepared when needed.aGroup A 300 °C and Group B 250 °C. The working metal standards were made by serial dilutions with either (a) 1% v/v HCl and 1% m/v L-cysteine for As, Bi and Sb or (b) 10% v/v HCl for Bi and Se from the original Table 3 Optimised programme for flow injection systema stock solutions, 1000 mg l-1 as SeVI, 1000 mg l-1 as SbIII, 1000 mg l-1 as BiIII (Merck) and 1000 mg l-1 as AsIII (Fixanal, Time/ Pump No. 1/ Pump No. 2/ s rpm rpm Position Riedel-de Ha�en). Research-grade argon was used for both the FI system and the internal and external gas streams for the Prefill 20 100 0 Fill THGA system. 1 10 100 0 Fill The reducing agent for Group A elements was L-cysteine. 2 5 100 80 Fill A stock solution of 10% L-cysteine in 0.5 mol l-1 HCl was 3 40 0 80 Inject made by adding 10 g of L-cysteine (Biochemistry, Merck) to aSample loop 500 ml; argon gas flow 150 ml min-1. 4.5 ml of concentrated HCl (Suprapur, Merck) and diluting to 100 ml and kept for 1 week.From this solution, enough Lcysteine was added to make the final concentration in the standard or sample 1% m/v. The reducing agent for Group B Table 4 Sequence control programme for ETAAS and FIAS elements was concentrated hydrochloric acid (apur, Merck) and suYcient acid was added to each sample to give Step Operation a final concentration of 50% v/v HCl. Sample solution pumped from the autosampler vessel to fill A Zirconium trapping solution: 0.02 mol l-1 ZrOCl2 8H2O the 500 ml sample loop, excess sample passes to waste.B Furnace Step 1, the tube is pre-heated to 250 or 300 °C, i.e. A 0.02 mol l-1 solution of ZrOCl2 8H2O was prepared by the trapping temperature. In parallel, the flow injection dissolving 0.06445 g of the salt in 10 ml of water. A total of system pumps fresh carrier and reductant solutions into 100 ml was injected onto the platform in two aliquots of 50 ml the manifold, and the argon gas sweeps the gas–liquid followed by the temperature programme shown in Table 5.separator clean. Sample loading is completed. The coating lasted for the lifetime of the tube.20 C The autosampler arm moves the quartz pipette tip from the standby position into the graphite tube. The sample is mixed with the carrier stream followed by the Results and discussion tetrahydroborate solution and argon gas and then passed through a reaction coil. The mixture enters the gas–liquid Blank contamination problem separator where the hydride vapour is swept into the graphite tube by the argon gas flow, and the waste is During the initial experiments, elevated blank values for As pumped out.A 0.45 mm membrane filter is positioned and to a lesser extent for Sb were observed. It was confirmed between the gas–liquid separator and the delivery tube to prevent excessive moisture from entering the graphite tube. Table 5 Furnace programme for graphite tube pre-treatment The tube is maintained at the trapping temperature.D The autosampler arm moves back to its standby position. Temperature/ Ramp time/ Hold time/ Gas flow/ E Pumps on the flow injection system are stopped to reduce Step °C s s Read ml min-1 reagent and sample consumption. F The atomiser time/temperature programme runs to 1 110 10 50 250 completion. A drying step to removes excess hydrogen and 2 130 30 50 250 any water vapour before the atomisation step and then a 3 1200 30 20 250 clean-out step cleans the tube before the next sample is 4 a 1 3 Yes 250 introduced. A maximum temperature of 2300 °C is used to prevent the iridium coating from being removed.aFor Ir 2000 °C and for Zr 2600 °C. J. Anal. At. Spectrom., 1999, 14, 1593–1600 1595experimentally that the contamination was due to the sodium experiments, Bi was also monitored, and it was found that, with prolonged heating times, both the accuracy and reproduc- tetrahydroborate as noted in the literature.20 DiVerent batches of sodium tetrahydroborate were therefore assessed, giving ibility of the results decreased.Based on these findings, it became clear that a single set of each batch tested gave slightly diVerent blank values, and thus aVected the detection limits of the method. The ‘‘cleaning-up’’ conditions that would reduce Te and Se completely, but without losses of either Se or Bi, would be diYcult to identify of the sodium tetrahydroborate was therefore considered. Activated alumina has previously been used to preconcentrate and that these three elements were incompatible in terms of the proposed simultaneous procedure.Consequently, it was oxyanions21 and it was assumed that, since the sodium tetrahydroborate was in alkaline solution, this approach may be decided to remove Te from Group B. Overall, the optimised conditions for the Se reduction step were found to be 70 °C used. Initially, an on-line column approach appeared to be the best option since no special treatment would then be for 120 min in 50% v/v HCl; under these conditions reproducible results were obtained for Bi.needed and the reductant would be cleaned-up immediately prior to use, keeping the method as simple as possible. Comparison between Ir and Zr trapping materials In preliminary experiments, it became clear that although the on-line activated alumina column was eVective in reducing Whilst many diVerent trapping materials have been proposed, the contamination levels, the limited column capacity (approxi- Ir and Zr are the most favoured, because these coatings only mately 30 replicates) was a serious disadvantage. The next need to be applied once during the lifetime of the tube.Both step was to clean the tetrahydroborate oV-line. The contami- of these materials have been applied successfully to single nated tetrahydroborate was passed through an activated alumelement analysis.7,8,20 In this study, a comparison was made ina column and collected in a fluorinated ethylene propylene to assess the suitability of both materials for simultaneous (FEP) bottle prior to the start of analysis.A large column multi-element analysis. The results are shown in Fig. 1. (150×10 mm id) was made from plastic tubing and a plastic Using the Zr-coated tube, it was not possible to find a restrictor was placed in one end of the tube, with a glass wool temperature where As, Bi and Sb could be successfully trapped plug. Approximately 2 g of activated alumina were slurried together [Fig. 1(a)–(c)]. Similar results were found for Bi and with a small volume of water and carefully poured into the Se. The findings are in line with the work of Haug and Liao,20 column. The alumina was allowed to settle and a glass wool who concluded that in terms of the optimum trapping tempera- plug was inserted above the alumina. The column was activated tures, As, Bi and Sb are all mutually exclusive. Thus, it was by passing 1% v/v HCl through it for 5 min followed by concluded that Zr was not an eVective trapping material for distilled water for 5 min and finally the alkaline reductant was this study.Using an Ir coating, As, Bi and Sb could be then passed through the column for a further 5 min or until successfully trapped at 300 °C [Fig. 1(d)] while Se and Bi the waste solution was alkaline. The column was left in the required a trapping temperature of 250 °C. To the best of our alkaline solution when not in use.This procedure removed knowledge, this is the first reported comparison of the applicaover 80% of the As contamination and 30% of the Sb bility of Ir- and Zr-coated graphite tubes to simultaneous contamination. multi-element hydride trapping. This oV-line approach gave similar blank values to the on-line method (As 0.005 s-1, and Sb 0.003 s-1) but the Long-term stability tests useful lifetime of the procedure was greatly increased, as 500 ml of ‘‘contaminated’’ reductant could be cleaned-up in In order to determine the lifetime of the Ir-coated tubes, tests were run overnight to check the stability of the analyte signal 2 h by slowly passing the solution through the column using a pump rate of 5 ml min-1.This volume of reductant was in terms of integrated absorbance and the precision of the signal (RSD %) for ten replicates. A solution of 5 mg l-1 As, suYcient for 160 analytical cycles and a fresh batch of reductant was prepared daily following re-activation of the column Bi and Sb, Group A, and Bi and Se, Group B, was used and all other instrumental parameters were as stated in Tables 1–3.as previously described. The results obtained are shown in Fig. 2. For Group A, initially for all three elements [Fig. 2(a)–(c)], Choice of reducing agents the integrated absorbance was high with good precision, but the mean peak absorbance then dropped to a constant level Although the instrumentation used in this study would facilitate the simultaneous determination of As, Bi, Sb, Se and Te for the next 120+ samples.This is as expected as the instrument will be most sensitive just after the graphite tube has as described above, the analytes were split into two groups, the criteria being those analytes which could be determined been freshly treated with the trapping material. Overall, however, for the first 5 h, i.e. 150 cycles, the results are steady and with the same reducing agent and conditions.Grouping the elements into two distinct sets in this way ensured that the reproducible, but then for As and Sb, but not for Bi, the precision is degraded and the average RSD exceeds 5%. This elements would be in their most favourable states prior to the hydride generation reaction. is thought to be because both As and Sb require L-cysteine as the reducing agent but L-cysteine also acts as a complexing During the initial experiments, Te was included in Group B along with Bi and Se.Tellurium requires the same reducing agent for these two analytes.13 When using a 1% v/v HCl carrier solution, the L-cysteine is not stable for long periods agent as Se, i.e. 50% v/v HCl, but the optimum temperature for reduction is 100 °C for 20 min. This temperature is 20 °C of time, as lower acid carrier concentrations (<0.1 mol l-1 HCl) are recommended with 1% m/v L-cysteine.12 higher than the optimum Se temperature. It is known that Se can be lost from solution as the chloride,16 and hence the It was also found in these studies that Sb was more stable (i.e.gave more reproducible results) in 10% v/v HCl carrier temperature of the reduction vessel should be kept as low as possible in order to eliminate any loss of Se. The results solution than in 1% v/v HCl carrier solution. AsIII could not be determined satisfactorily with a 10% v/v HCl carrier solu- obtained in this study showed that for an oV-line reduction, if the temperature was increased to 100 °C for 3 h to facilitate tion; therefore, to allow the simultaneous determination of AsIII and SbIII, a compromise was reached which used a carrier the reduction of Te, significant losses of Se were observed. The 3 h reduction time was required to produce the best results solution of 1% v/v HCl.These issues, with respect to the HCl carrier concentration and the stability of L-cysteine in the for Te. If the temperature was reduced to 70 °C, optimum for the Se reduction, then the results for Te were irreproducible, solutions, explain the degradation in precision noted after 5 h of operation. It has been observed by Welz and Sucmanova12 indicating incomplete reduction. During the course of these 1596 J.Anal. At. Spectrom., 1999, 14, 1593–1600Fig. 1 Simultaneous signals for As, Bi and Sb (Group A) at the 5 mg l-1 level; atomisation temperature 2200 °C, sample volume 500 ml. (a)–(c) Zr-coated tube with trapping temperature of: (a) 300 °C; (b) 600 °C; (c) 800 °C.(d) Ir-coated tube with a trapping temperature of 300 °C. that the integrated absorbance of an AsIII standard (10 mg l-1) interferent (0.01, 0.1, 1.0, 10, 100 mg l-1) were then added in 1 mol l-1 HCl with no reductant decreased almost linearly and the new response (n=3) was recorded. An interference with time to 50% of its original value within 3–6 h. This was defined as any signal which gave a deviation of ±10% highlights the instability of the AsIII standard (which is oxidised from the standard response.The levels at which such deviations to AsV) which must be correctly stabilised to prevent poor were observed are shown in Table 6. precision. Bi does not appear to be aVected by the acid Of the four analytes, SeIV was the most aVected with two concentration of the L-cysteine and both the precision and the out of six metals (CoII and PbII) giving an interference at the integrated absorbance values remain steady during the full 9 h 0.01 mg l-1 level, i.e.only a 2-fold increase above the original of the experiment. Se concentration. Two other metals, FeIII and NiII, gave an For the Group B elements [Fig. 2(d) and (e)], Bi showed a interference at the 0.1 mg l-1 level, i.e. a 20-fold increase. The slight increase in integrated absorbance over the 9 h period worst interference on AsIII was from four elements at the (i.e. 270 runs) with a precision of approximately 2% for the 0.1 mg l-1 level, FeIII, NiII, PbII and AgI.However, As can be whole run. determined free from interference in the presence of up to Overall, these results indicate that Ir as the trapping reagent 10 mg l-1 CoII and CuII. SbIII was the least aVected analyte, for simultaneous multi-element analysis is reliable, shows good and could tolerate up to 10 mg l-1 CuII, NiII, AgI and PbII precision and is stable for reasonable working periods, i.e. 5 h and 100 mg l-1 FeIII and CoII without interference eVects.The for As and Sb and up to 9 h for Bi and Se. The reduced interference eVects on Bi varied between Groups A and B. In analysis time for As and Sb is due to instability in the chemistry Group A, both FeIII and AgI gave an interference at the over prolonged periods of time rather than degradation of the 0.1 mg l-1 level, while only Ag gave the same level of intertrapping material used. ference in Group B. In Group A, there was no interference from CoII up to 100 mg l-1 although under Group B con- Interference studies ditions, CoII interfered at the 1 mg l-1 level.The results of this brief interference study show that Se is Gas phase interferences are caused by the presence of other the element that is potentially aVected the most by interferents, volatile elements and result from the interferent removing the with four elements giving an interference at the 0.1 mg l-1 hydrogen radicals required by the analyte hydride for atomislevel, i.e.a 2-fold increase in concentration of the interferent ation.22 This process happens in conventional hydride generover the original Se concentration. At the 0.1 mg l-1 inter- ation within the quartz cell,23 but since ‘‘in-atomiser trapping’’ ference level, i.e. a 20-fold increase in interferent concentration utilises higher atomisation temperatures and the atomisation over the analyte concentration, it is possible that only FeIII mechanism is diVerent to that in a quartz tube atomiser, this may interfere with the Group A elements (As and Bi) and Se type of interference may be eliminated.The mutual interference in Group B, but all the other interferent concentrations in the among the hydride-forming elements is much lower than in samples should be below this interferent level. Therefore, when externally heated quartz tube atomisers.24 This interference analysing natural water samples, care must be taken either to should therefore be expected to depend on the total mass of reduce the potential interferences whenever possible, by adding interferent trapped in the furnace rather than the interferent a complexing agent (e.g., EDTA or a greater concentration of concentration in the sample.The literature identifies six metals L-cysteine) or confirm the absence of any potential interferent. (AgI, CoII, CuII, FeIII, NiII and PbII) which may be considered Overall, Group A elements appear to be more resistant to as potential interferences.A 5 mg l-1 multi-element standard the interfering elements than Group B. This is probably due of these elements was analysed (n=3) in the absence of interferent to give a base response. Increasing amounts of the to the presence of the reducing agent (L-cysteine) used in J. Anal. At. Spectrom., 1999, 14, 1593–1600 1597Fig. 2 Long-term stability as indicated by integrated absorbance (s-1) and precision (%RSD, n=10). Concentration for all elements 5 mg l-1, sample volume 500 ml, trapping time 40 s, atomisation temperature 2200 °C.(a)–(c) Group A, trapping temperature 300 °C, (a) As; (b) Bi; (c) Sb; (d) and (e) Group B, trapping temperature 250 °C, (d) Bi; (e) Se. Table 6 Results from interference study. Level (mg l-1) at which the Analysis of Certified Reference Materials interferent produces a deviation of 10% or more from the 5 mg l-1 Two reference materials (RMs) were analysed: High Purity metal standard Standard (HPS) Certified Reference Material No. 490915 AsIII BiIII SbIII BiIII SeIV Trace Metals in Drinking Water and National Institute of Standards and Technology (NIST) Standard Reference CuII 10 1 10 10 1 Material 1643c Trace Elements inWater. As the concentrations FeIII 0.1 0.1 >100 1 0.1 of the RMs were outside the calibration range for all the CoII 10 >100 >100 1 0.01 analytes of interest, appropriate dilutions were made during NiII 0.1 10 10 1 0.1 the oV-line digestion step.AgI 0.1 0.1 10 0.1 0.1 PbII 0.1 1 10 100 0.01 For Group A, two separate dilutions had to be made. The As required a 20-fold dilution to bring it within range, while Sb and Bi only required a 2.5-fold dilution. During the dilution step, the appropriate amount of 10% m/v L-cysteine was added to each calibrated flask. After the 30 min reaction period, the analysis Group A. This reducing agent also has a releasing eVect on was performed. For Group B, a 2-fold dilution was used.The the potential liquid phase interferences.9 Many workers have solution was heated for 120 min at 70 °C, and, once cool, used used the dual nature of L-cysteine as a reducing agent and to for analysis. All solutions were analysed under the conditions minimise or eliminate the liquid phase interferences in HGAAS.13,24 stated in Tables 1–3, and the results are shown in Table 7. 1598 J. Anal. At. Spectrom., 1999, 14, 1593–1600Table 7 Results obtained for Certified Reference Materials HPS CRM SRM 1646c Certified value/mg l-1 Experimental/mg l-1 Certified value/mg l-1 Experimental/mg l-1 As 80.00±0.40 80.8±2.1 82.10±1.2 80.9±0.5 Bi 10.00±0.05 9.7±0.4 12 12.6±0.2 Sb 10.00±0.05 11.1±1.2 NDa ND Bi 10.00±0.05 9.2±0.0 12 11.7±0.0 Se 10.00±0.05 9.7±0.0 12.7±0.7 12.1±0.0 aND=Not determined.The HPS RM was certified for all four elements of interest ETAAS is possible using conditions derived from single element analysis. in this study, and good agreement was achieved between the experimental and the certified values as shown in Table 7. The Advantages of simultaneous multi-element analysis over single element analysis include speed with respect to the overall NIST Trace Elements in Water SRM was certified for As and Se, with a recommended but not certified value for Bi and no analysis time (the time reduction being proportional to the number of elements run together), and reduced consumption value was provided for Sb. Again, good agreement between the found and certified values was achieved, with all results of reagents per sample.The disadvantage is reflected by the higher detection limits obtained for multi-element analysis falling within the certified range of two standard deviations (Table 7). when compared with single element analysis; this is because in the latter the conditions can be fully optimised with respect to the instrumental and chemical parameters to attain the best Analytical figures of merit possible detection limits for each element. However, while a The characteristic mass (as determined for integrated flow injection approach with 500 ml sample volumes was absorbance) and instrument and method detection limits for utilised in this study, improvements in the detection limit can the method are shown in Table 8.The instrumental detection be obtained by using larger sample volumes by either increasing limit (pg) is given as equal to the mean of the blank signal the size of the sample loop, by using multiple trapping steps (Yblank) for ten replicates plus three times the standard devi- with a small sample loop (as the trapping step is independent ation of the blank (sblank) for ten replicates (IDL= of the atomisation step), or by applying continuous-flow Yblank+3×sblank).The method detection limit (MDL) is based sampling.27 on the instrument detection limit but takes into account the The optimised reducing conditions used for Group B line of regression from the calibration graph such that MDL elements in this study were determined by the need for Se to (mg l-1)=(IDL-intercept)/gradient.be in the tetravalent state for the hydrides to be formed, but The detection limit for As is dependent on the background for the Group A elements, e.g. As and Sb, the hydrides can levels from the sodium tetrahydroborate. Consequently, the be formed from either the trivalent or the pentavalent state.detection limit for As could be improved further if this source This paper describes the determination of hydride-forming of contamination could be eliminated. elements in the lowest oxidation state by using L-cysteine as When the detection limits for this simultaneous multi- the reducing agent. The use of L-cysteine also appears to element method are compared against those of single element reduce the interference eVects on the analytes of interest. methods,7,20,24,25 the single element method detection limits The ‘‘in-atomiser trapping’’ technique reported here can are lower by a factor of between 2 and 6 depending on the easily be expanded from single element analysis to multielement.This is to be expected as the single element methods element analysis provided that an Ir-coated tube is used. The use fully optimised parameters for each element, while the analysis may be set up and left to operate unaided with only simultaneous multi-element method uses a set of compromise occasional re-calibration for up to 5 h with good precision conditions based on the suite of elements being determined.and accuracy. Iridium was the only coating material found to However, when the absolute detection limits of the simul- be applicable to this multi-element procedure, as a universal taneous multi-element method (As 107 pg, Se 73 pg) are trapping temperature could not be found for a Zr-coated tube. compared with those of a previously published simultaneous The conditions developed were shown to produce acceptable method26 (As 92 pg, Se 120 pg), the absolute detection limits results for the analysis of two reference materials.are very similar. Acknowledgements Conclusion The authors thank Professor Dimiter Tsalev (University of The simultaneous multi-element determination of combi- Sofia, Bulgaria), Dr. Michael Sperling and Miss Michaela nations of hydride-forming elements by ‘‘in-atomiser trapping’’ Feuerstein (Bodenseewerk Perkin-Elmer GmbH.) for their helpful discussions and advice.Table 8 Analytical figures of merit References Group A Group B 1 H. Matusiewicz and R. E. Sturgeon, Spectrochim. Acta, Part B, 1996, 51, 377. AsIII BiIII SbIII BiIII SeIV 2 J. Murphy, G. Schlemmer, I. L. Shuttler, P. Jones and S. J. Hill, Anal. Commun., 1997, 34, 359. Characteristic mass/pg 175 136 107 91 90 3 H. O. Haug and Y. Liuo, J. Anal. At. Spectrom., 1995, 10, 1069. Instrumental detection 107 28 27 43 73 4 D. L. Tsalev, A. D’Ulivo, L. Lampugnani, M. Di Marco and limit/pg R. Zamboni, J. Anal. At. Spectrom., 1995, 10, 1003. Method detection 0.82 0.04 0.26 0.12 0.29 5 J. Dedina and D. L. Tsalev, Hydride Generation Atomic Absorption limit/mg l-1 Spectrometry,Wiley, Chichester, 1995. J. Anal. At. Spectrom., 1999, 14, 1593–1600 15996 R. E. Sturgeon, S. N. Willie, G. I. Sproule and S. S. Berman, 19 A. Stroh and U. Vollkopf, J. Anal. At. Spectrom., 1993, 8, 35. J. Anal. At. Spectrom., 1987, 2, 719. 20 H. O. Haug and Y-P. Liao, Fresenius’ J. Anal. Chem., 1996, 356, 7 X.-p. Yan and Z.-m. Ni, J. Anal. At. Spectrom., 1991, 6, 483. 435. 8 I. L. Shuttler, M. Feuerstein and G. Schlemmer, J. Anal. At. 21 M. Sperling, S. Xu and B.Welz, Anal. Chem., 1992, 64, 3101. Spectrom., 1992, 7, 1299. 22 P. Barth, V. Krivan and R. Hausbeck, Anal. Chim. Acta, 1992, 9 X-p. Yan and Z-m. Ni, Anal. Chim. Acta, 1994, 291, 89. 263, 111. 10 J. F. Tyson, S. G. OZey, N. J. Seare, H. A. B. Kibble and 23 B. Welz and P. Strauss, Spectrochim. Acta, Part B, 1993, 48, 951. C. Fellows, J. Anal. At. Spectrom., 1992, 7, 315. 24 Y. An, S. N. Willie and R. E. Sturgeon, Spectrochim. Acta, Part 11 J. Dedina and B. Welz, J. Anal. At. Spectrom., 1992, 7, 307. B, 1992, 47, 1403. 12 B. Welz and M. Sucmanova, Analyst, 1993, 118, 1417. 25 Z.-m. Ni, B. He and H.-b. Han, J. Anal. At. Spectrom., 1993, 13 D. L. Tsalev, A. D’Ulivo, L. Lampugnani, G. Pellegrini and 8, 995. R. Zamboni, J. Anal. At. Spectrom., 1996, 11, 989. 26 S. Garbos, M. Walcerz, E. Bulska and A. Hulanicki, Spectrochim. 14 B. Welz and M. Sucmanova, Analyst, 1993, 118, 1425. Acta, Part B, 1995, 50, 1669. 15 R. Bye and W. Lund, Fresenius’ Z. Anal. Chem., 1988, 332, 242. 27 H.-W. Sinemus, J. Kleiner, H.-H. Stabel and B. Radziuk, J. Anal. 16 R. Bye, Talanta, 1990, 37, 1029. At. Spectrom., 1992, 7, 433. 17 H. Uggerud and W. Lund, J. Anal. At. Spectrom., 1995, 10, 405. 18 J. Bowman, B. Fairman and Catterick, J. Anal. At. Spectrom., 1997, 12, 313. Paper 9/04468J 1600 J. Anal. At. Spectrom., 1999, 14, 1593–1600