Analysis of Geological Materials for Bismuth, Antimony, Selenium and Tellurium by Continuous Flow Hydride Generation Inductively Coupled Plasma Mass Spectrometry Part 1. Mutual Hydride Interferences GWENDY E. M. HALL* AND JEAN-CLAUDE PELCHAT Geological Survey of Canada, 601 Booth Street, Ottawa, Ontario, Canada K1A 0E8 A study was made of mutual interferences in the determination dearth of data for the low abundance elements, Bi and Te, in soils (e.g., Table 1) indicates an absence of sufficiently sensitive of Bi, Sb, Se and Te by hydride generation inductively coupled plasma mass spectrometry (HG–ICP-MS).The design of the and robust analytical methods for their measurement. Although also an element of low abundance, the geochemistry study was based upon the valency states in which these elements would occur following two acid digestion procedures of Se is better known because it has received considerable attention in agricultural and epidemiology studies, being an commonly used for geological materials (aqua regia and HF–HClO4–HNO3–HCl), that is, as BiIII, SbV, SeIV and TeIV. essential element which exhibits a narrow range between deficiency (e.g., leading to Keshan and Keshan-Beck disease) Arsenic in its two valency states (III and V ) was also examined as a potential interferent but Ge, Sn and Pb were not, as and toxicity (selenosis, capable of causing death).4 In order to improve our understanding of the processes controlling the formation of their hydrides at the concentration of HCl used (2–4m) would be negligible.Interferents were investigated at distribution of these elements in the surficial environment and to search for anomalous concentrations in mineral exploration, concentrations up to 2000 mg l-1 while the analyte concentration was held at 0.2 mg l-1. AsV and SeIV severely analytical methods must provide the capability of accurate and precise detection at, and indeed below, the natural back- suppressed the signal for Te in both 2 and 4 m HCl, though less so in the stronger acid medium.These interferences were ground concentrations of the various media examined. At the Geological Survey of Canada (GSC), we have used negated by reduction of AsV to the non-interfering state AsIII and SeIV to SeO by addition of KI and ascorbic acid at a final hydride generation quartz tube atomic absorption spectrometry (HG–QTAAS) to determine these elements in tens strength of 0.005%.The preferred acid medium was selected as 4 m HCl for two reasons: interferences were reduced; and of thousands of geological samples (mainly rocks, soils and sediments) since 1975.5 While this technique has served well in wash-out time between samples was shorter (i.e., decreased memory effect). Mutual interferences of possible concern analysis for As and Sb, it has shortcomings of both inadequate sensitivity and numerous interferences for the other three comprise the effect of: 2500-fold excess of Bi on the measurement of Te; a 5000-fold excess of Te on Bi; a elements.Given the wide range of matrix composition expected amongst these samples, separation of the analytes from the 2500-fold excess of Bi on Sb; and a 5000-fold excess of Bi on Se. However, these relative abundances in geological materials well documented liquid phase interferents (e.g., Cu, Ni, Co) has been used routinely in preference to addition of complexing (rocks, sediments, soils, vegetation) would be extraordinary and hence implementation of the recommended scheme should reagents such as tartrate, citrate, cysteine or ethylenediaminetetraacetic acid.6 Coprecipitation with La(OH)3 is used and, if produce analysis for Bi, Sb, Se and Te by HG–ICP-MS that is free of mutual interference.high concentrations of transition metals are present (i.e., >2% in Ni, Cu), a second precipitation may be necessary.7 Keywords: Bismuth; antimony; selenium; tellurium ; hydride Suppression by precious metals is not negated by this separa- generation inductively coupled plasma mass spectrometry ; tion but their natural background concentrations are very low geological ; interferences (<5 mg kg-1).This leaves gas phase interferences of major concern, both The fact that group VA and VIA elements form gaseous in transport to the measurement cell (in transport kinetics and hydrides at ambient temperature (e.g., AsH3, SbH3 , BiH3, SeH2 efficiency) and in the atomiser itself, an area of considerable and TeH2) has long been used as a mode of improving their debate in terms of mechanisms.8–11 Two mechanisms were method of measurement in many different sample types, includ- proposed by Dedina8: the interferents may reduce the popuing geological materials.1 These elements are chalcophile in lation of radicals responsible for the atomisation of the analyte nature, that is, they have an affinity for S and hence are found by accelerating their decay (radical population interference); concentrated in sulfides. They are used in mineral exploration as pathfinders for epithermal precious metal and base metal Table 1 Average natural ranges in mg kg-1 reported for As, Bi, Sb, Se and Te in various geological media3 deposits.2 The following average concentrations, in mg kg-1, of these elements in igneous rocks provide a general guide to Element Magmatic rocks Shale Limestone Soil their expected background values: As, 2; Bi, 0.1; Sb, 0.1; Se, As 0.5–2.5 5–13 1.0–2.4 <1–95 0.1; and Te, 0.002.However, the compositions of many types Bi 0.001–0.15 0.05–0.50 0.10–0.20 of rocks differ, and perhaps a better appreciation of element Sb 0.1–1.0 0.8–1.5 0.3 0.05–2.3 ranges can be gleaned from Table 1, taken from summary Se 0.01–0.05 0.6 0.03–0.1 0.02–2.3 tables by Kabata-Pendias and Pendias.3 Note the enriched Te 0.001–0.005 0.01 background of these elements in sedimentary shale media.The Journal of Analytical Atomic Spectrometry, January 1997, Vol. 12 (97–102) 97and the interferents accelerate the decay of the analyte atoms themselves (analyte decay interference). Using thermodynamic calculations coupled with spectroscopic and thermal investigations, Dittrich and Mandry11 found the main cause of gas phase interference to be the formation of diatomic molecules (e.g., AsSb) and therefore suggested the use of a higher atomisation temperature (>2000 °C).They classified mutual hydride interferences into the following three categories: (1) those which involve only molecule formation in the gaseous phase (As in Sb matrix, Sb in As); (2) those which include both molecule formation in the gaseous phase, and chemical reaction and adsorption in the liquid hydride generation phase (As, Sb and Se in Bi, Se, Ge, Sn and Pb matrices); and (3) those which originate by chemical reaction and adsorption in the liquid phase (Te in As, Sb, Bi, Se, Ge, Sn and Pb matrices, and As, Fig. 1 Schematic diagram of the GSC hydride generation system. Sb and Se in Te matrices). With consideration to the relative Introduction of KI is optional (see text). natural abundances of these elements, the determination of Te in samples containing significant amounts of As would appear to be the most challenging. Indeed, using the GSC deliver solutions to the HG system and for this work the HG–QTAAS method5 the presence of 2000 mg l-1 of As was sample tubing was placed in test-tubes manually (rather than found to suppress the signal for 5 mg l-1 of Te by 60%, a ratio using the Gilson 222 autosampler). The hydrides were trans- of interferent to analyte which is to be expected in geoanalysis.ported via Tygon tubing (about 3.2 mm id) directly to the ICP Clearly, the combination of hydride generation with induc- torch without the spray chamber in place. Stability of the tively coupled plasma mass or atomic emission spectrometry hydride signal was improved by the additional flow of Ar at (ICP-MS, ICP-AES) should be much freer of these gas phase 0.1 l min-1 to the gas–liquid separator (Fig. 1) and by consist- interferences than HG–QTAAS. Wickstrøm et al.,12 for ent pumping of the liquid waste to the drain. Hardware and example, reported significant reduction of gas phase inter- operating conditions are given in Table 2. Two isotopes each ferences from Sn and As in the analysis of Ni alloys and of Se, Sb and Te were measured in these tests to ensure low-alloy steels for Se by HG–ICP-AES compared with consistent response. A period of about 50 s was allowed to HG–QTAAS.Uggerud and Lund13 studied the effect of 10 lapse between initial uptake of sample solution and actual and 100 mg l-1 of As and Sb on the HG–ICP-AES signal measurement. intensity of 100 mg l-1 of As, Sb, Bi, Se and Te. The signals for Bi and Se were mostly enhanced by As and Sb (up to 128%), Reagents and Solutions whereas that for Te was strongly depressed (e.g., to 2 and 22% in 100 and 10 mg l-1 of As, respectively). All reagents were of analytical-reagent grade, purchased mostly Since the early preliminary report by Powell et al.14 demon- from Baker (Phillipsburg, NJ, USA); distilled, de-ionised water strating the detection capability of continuous flow was used throughout for dilution of standard solutions.The HG–ICP-MS in the ng l-1 range for As, Se, Sb, Bi and Te, NaBH4 reducing agent, stabilised in 0.1 M NaOH and filtered, there have been relatively few applications of this technique in was prepared daily.Standard solutions containing 1000 mg l-1 the literature, and very few indeed focusing on the analysis of geological materials.15 Mutual interference (amongst others) in Table 2 HG–ICP-MS hardware and operating conditions the measurement of Se by flow injection HG–ICP-MS was studied recently:16 at an interferent5Se ratio of 1000, As and Inductively coupled plasma Sb at the III or V valency state had no effect, while at an Rf generator 2.5 kW, frequency 27.12 MHz interfereent5Se ratio of 100, BiIII and TeIV had no effect.Nebuliser Meinhard C concentric glas The intention of this paper is to report the mutual inter- Torch SCIEX, ‘long’ Distance from load coil to ferences encountered in continuous flow HG–ICP-MS and to sampler orifice 16 mm design appropriate methodology to counteract them for accu- Rf power 1.1 kW rate and efficient analysis of geological materials.Mutual Plasma Ar flow rate 12 l min-1 interferences from Group IVA elements were not investigated Intermediate Ar flow rate 2.1 l min-1 as their hydride formation would be minimal at the strength of HCl (2–4 M) used in the generation medium here. Part 2 Mass spectrometer Sample Nickel, 1.14-mm orifice will describe the decomposition and separation procedures Skimmer Nickel, 0.89-mm orifice developed, together with results for geological standard Ion lens settings B=17, El=83, P=13, S2=28 reference materials.Resolution High (0.6 u 10% peak height) Data acquisition parameters Measurement time/isotope= 1.8 s; dwell time=200 ms in EXPERIMENTAL multichannel mode (peak hopping), 3 sweeps/3 Instrumentation readings/3 replicates each solution The ICP mass spectrometer employed was a SCIEX Perkin- Isotopes measure 78Se, 82Se, 121Sb, 123Sb, 128Te, Elmer (Thornhill, Ontario, Canada) ELAN Model 250, 130Te, 209Bi upgraded to the 500.Matheson mass flow controllers were used for the plasma, intermediate and carrier gases. [This Hydride generation Reducing agent 1% NaBH4 in 0.1 mol l-1 instrument was replaced recently with the VG PlasmaQuad NaOH PQII+ (VG Elemental, Winsford, Cheshire, UK) and all Flow rate of reducing agent 1.1 ml min-1 previous results were verified with this equipment.] The hydride Flow rate of acidified sample 3.6 ml min-1 generation system was built at the GSC and is shown sche- Flow rate of carrier gas (Ar) 1.9 l min-1 matically in Fig. 1. A Minipuls 3 peristaltic pump was used to 98 Journal of Analytical Atomic Spectrometry, January 1997, Vol. 12each of BiIII, SeIV and TeIV were purchased from Plasmachem (Bradley Beach, NJ, USA). AsIII and AsV stock solutions at 100 mg l-1 were both made by using the primary standard As2O3 dissolved in a small amount of NaOH and then acidified with HNO3. To prepare the AsV standard, a portion of this solution was oxidised by addition of H2O2 and boiled in an HNO3 medium.AsIII was kept in 2% HCl whereas AsV was kept in 2% HNO3. SbIII and SbV stock solutions at 100 mg l-1 were prepared from Sb2O3 and Sb2O5, respectively; both were kept in 1 M HCl. Fe and La were added as their trivalent chlorides in the interference studies. Procedure The element of lowest natural abundance, Te, was used for optimisation studies of: carrier gas flow rate; plasma power; Fig. 2 Optimisation of Ar carrier gas flow rate for 0.2 ppb (mg l-1) HCl concentration; and NaBH4 concentration. Response under of Te by HG–ICP-MS. these conditions, optimal for Te, was then tested for Bi, Sb and Se. The acid decomposition procedures to be used (aqua regia and HF–HClO4–HNO3–HCl) result in the following valency states: AsV; Bi2I; SbV; SeIV and TeIV (discussed in Part 2 of this series). Thus, Bi, Se and Te are in their reactive forms (cf., SeVI, TeVI), whereas SbV must be converted to SbIII for greatest sensitivity, a reduction usually carried out with KI and ascorbic acid.5 The effects of SeVI and TeVI as possible interferents therefore were not required to be evaluated but both valency states of As and Sb were included in the tests.The following interferents, at levels of 50, 100, 250, 500, 1000 and 2000 mg l-1, were examined for their effect on the signal for 0.2 mg l-1 of TeIV: AsIII, AsV, BiIII, SbIII, SbV and SeIV . This study was carried out at two concentrations of HCl, 2 and 4 M, and was repeated in 4 M HCl in the presence of KI at various concentrations.Care was taken to measure the interferent solutions alone for their possible Te content. Fig. 3 Optimisation of plasma power for 0.2 ppb (mg l-1) of Te by The signal intensity for 0.2 mg l-1 of SeIV in 4 M HCl was HG–ICP-MS. then examined in the presence of 50–2000 mg l-1 of AsV, AsIII, SbV, SbIII , Bi and Te. KI was not included in the test for Se as it is known to reduce Se to the unreactive SeO state.Bi as an analyte was examined similarly, with and without KI present. Finally, SbIII as an analyte was examined, using both SbIII directly from the standard solution (made as SbIII) and SbIII prepared by reducing SbV with KI, which would be required after either acid decomposition. The effect of Fe, at concentrations of 100–2000 mg l-1, was also examined on the signal of 0.2 mg l-1 of Bi, Sb, Se and Te.Fe can be a major constituent of geological samples and comes through the separation procedure with the analytes. Any influence from La, at concentrations of 0.1–1% in the analyte solution, was also checked for the four analytes. RESULTS AND DISCUSSION Fig. 4 Effect of HCl concentration in analyte solution on the 0.2 ppb Optimisation of Hydride Generation Conditions (mg l-1) 130Te signal intensity. Dependence of the Te signal on carrier gas flow rate, plasma power and HCl concentration is shown in Figs. 2–4, respectively. The concentration of NaBH4 in the range 0.5–3% was 5 M. An HCl concentration of 4 M was chosen but interference studies for Te were also carried out at 2 M HCl to test for not critical but signals tended to be noisier at 3%; 1% NaBH4 was selected. Fortunately, the other three analytes behaved in changes in severity with acid strength. The washout time required in 4 M HCl appears to be in the order of 100 s (Fig. 4), a similar manner to Te and therefore optimum conditions for all four elements were chosen to be 1.9 l min-1 carrier gas but this can be decreasedsubstantially to about 60 s if solutions containing analyte at similar concentrations are being run.flow rate and 1.1 kW power. Increase in signal intensity with HCl concentration levels off at 2 M (Fig. 4), but a higher Under these operating conditions, the net intensity (blank subtracted) obtained by hydride generation in 4 M HCl was concentration of 4 M was found to be beneficial in decreasing the amount of wash-out time required between samples (Fig. 5). compared with that found by nebulisation of the analytes in 4% HNO3 solution. Solutions containing 0, 0.2 and 0.5 mg l-1 Again, the behaviour of Te was typical of all analytes. The percentage relative standard deviation (RSD) for 10 readings analyte were used for HG, whereas standard solutions of 0, 5, 10 and 20 mg l-1 analyte were used for nebulisation ICP-MS. of a 0.2 mg l-1 Te solution at various concentrations of HCl was: 2.9 at 1 M; 2.7 at 2 M; 1.7 at 3 M; 1.8 at 4 M; and 2.1 at Summary data are given in Table 3, showing a maximum Journal of Analytical Atomic Spectrometry, January 1997, Vol. 12 99Fig. 5 Effect of HCl concentration on wash-out time for 2 ppb (mg l-1) of Te. Table 3 Comparison of sensitivity by hydride generation (HG) versus nebulisation (NEB) ICP-MS for Bi, Sb, Se and Te (10 readings for each solution) (Net c s-1/mg l-1)HG/ Analyte (Net c s-1/mg l-1)NEB c s-1 for 0.2 mg l-1 by HG Bi 264 57 040±548 Sb 359 69 660±727 Se 1110 13 270±240 Fig. 6 Effect of AsV on 130Te at 0.2 ppb (mg l-1): (a), in 2 M HCl; (b), Te 897 13 480±160 in 4 M HCl. improvement for Se of over a 1000-fold increase in sensitivity by HG. Mutual Interferences T ellurium An interference in this work is defined as an effect causing a positive or negative change in the analyte signal of 10% or more. No interference is evident in the measurement of 0.2 mg l-1 of Te, in 2 or 4 M HCl, in the presence of up to 2000 mg l-1 of AsIII, SbIII or SbV.However, AsV causes a dramatic decrease of the Te signal in 2 M HCl [Fig. 6(a)]. This decrease, though certainly still a problem, is considerably less severe in the 4 M HCl medium [Fig. 6(b)]. Thus, at a ratio of 500051 As to Te (not uncommon in geological materials), a suppression of about 35% can be expected. This interference was not reduced by changing the strength of NaBH4.Increasing concentrations of Bi beyond 500 mg l-1 led to a severe suppression of the signal for 0.2 mg l-1 Te [Figs. 7(a) and (b)]. As with AsV, the degree of suppression is greater in 2 M HCl. Furthermore, once these high concentrations of Bi had been run, the 0.2 mg l-1 Te alone did not regain its initial intensity of about 26×103 c s-1 (counts s-1), which is unlike the situation in the interference by AsV. The generation system and torch required cleaning to restore performance. Bismuth hydride is known to be both thermally and kinetically unstable, Fig. 7 Effect of Bi on 130Te at 0.2 ppb (mg l-1): (a), in 2 M HCl; (b), in even at 25°C.A rate constant of 0.24 min-1 at room tempera- 4 M HCl. ture has been determined17 for the first order reaction BiH3�Bi+3/2H2 intensity of the 0.2 mg l-1 Te signal and again the effect was more severe at 2 M rather than 4 M HCl [Figs. 8(a) and (b)]. Thus, the plating out of Bi on the walls of the tubing, separator and torch is causing sequestering and decomposition of TeH2.Unlike the situation with Bi, the original sensitivity was obtained for 0.2 mg l-1 of Te alone when run at the end of the Fortunately, this suppression is not evident at levels of Bi below about 500 mg l-1 (i.e., a 2500-fold excess), allowing interference study. The tolerance limits for both Se and Bi appear to be at about 250051 excess over Te but the suppres- accurate determination of Te in the majority of geological samples.sion by Se is less drastic than that of Bi [cf., Figs. 7(b) and 8(b)]. Of the three mutual interferences discussed above, that from Selenium at levels above 500 mg l-1 also decreased the 100 Journal of Analytical Atomic Spectrometry, January 1997, Vol. 12(by way of their precipitation as TeO and BiOI) as well as from Se but the strength of KI was much higher in that case (0.25%). Given the reduced degree of interference generally in 4 M rather than 2 M HCl, studies of Bi, Sb and Se were carried out in the higher strength acid.Other workers have recommended high acid strengths to reduce interferences. For example, using HG–QTAAS, Hershey and Keliher19 studied the effect of progressively increasing HCl strength from 1.2 to 7.2 M and found this to reduce the interferences of: Se on Bi; Au and Ir on Se and Bi; W on Se; and Ir and Pt on Sb. Bismuth There was no interference on 0.2 mg l-1 Bi from AsIII, AsV, SbIII, SbV or Se at levels up to 2000 mg l-1.In one of the trials there was an indication of a slight enhancement of the Bi signal in the presence of 500 mg l-1 or more of SbIII, but the change in intensity was only about 7–9%. A marked suppression is evident for 0.2 mg l-1 of Bi in the presence of greater than 1000 mg l-1 of Te (Fig. 9); the intensity is reduced by about 50% at a Te concentration of 2000 mg l-1 (10000-fold excess). In fact, analysis of the 0.2 mg l-1 Bi alone following the solution containing 2000 mg l-1 of Te produces a signal of only 70×103 c s-1 (cf., 80×103 c s-1 at the beginning of the run), suggesting memory effects of Te in the generation system and possibly the central channel of the torch.However, this residual effect is much stronger in the reverse situation, i.e., Fig. 8 Effect of Se on 130Te at 0.2 ppb (mg l-1): (a), in 2 M HCl; (b); the influence of Bi on the determination of Te. This interference 4 M HCl. pattern was unchanged in the presence of KI (at a level of 0.005%).The apparent enhancement of the signal for 0.2 mg l-1 AsV, the most abundant of the elements, is the most serious. of Bi below about 1000 mg l-1 of Te is due to the Bi content However, the simple solution is to reduce AsV to AsIII by of the Te interferent solution added (see 1000 mg l-1 Te blank, addition of KI and ascorbic acid. This was carried out on-line Fig. 9). Given the expected relative abundances of Bi and Te using 5% KI but was rejected for two reasons: firstly, blank in geological media, interference from Te does not pose a values for Bi rose to unacceptably high levels; and secondly, a problem.long coil was needed to provide sufficient time for the reduction to occur. It was decided instead to add a much smaller quantity Antimony of KI manually several hours prior to analysis. Test solutions containing 0.2 mg l-1 of Te, 50 mg l-1 of As and 10 mg l-1 of There was no interference on 0.2 mg l-1 SbIII from AsIII, AsV, Bi in 4 M HCl were made up to the following concentrations Se or Te at levels up to 2000 mg l-1.However, a slight in KI and ascorbic acid: 0.001; 0.002; 0.005; 0.0075; 0.01; 0.0125; enhancement was evident in the presence of 250 mg l-1 of Bi, 0.02; and 0.025%. By monitoring the As intensity at 75 u, it which became essentially constant from 500 to 2000 mg l-1 was found that the AsV was fully reduced to AsIII at a minimum (Fig. 10). This enhancement, of the order of 10%, is seen KI concentration of 0.005%, whereas the Te signal began to whether the test is carried out with SbIII (from SbIII stock decrease steadily at KI concentrations above 0.0125%.solution) or SbV is used and the solution made 0.005% in KI. Presumably this decrease was due to the formation of unreac- This relatively small effect on the Sb signal contrasts sharply tive TeO. The signal for Bi was unaffected by any of these with the severe suppression created by 1000 mg l-1 or more of concentrations of KI.Thus, an optimum KI and ascorbic acid Bi on the Te signal. Excess amounts of Bi over Sb, by several concentration in the 4 M HCl of 0.005% was chosen and added orders of magnitude, would be extremely rare in geoanalysis 3 h prior to the determination of Te. Repeated interference studies under these conditions showed no interference from AsV (or Sb) and, furthermore, no interference from Se, presumably having been reduced to the unreactive SeO state (a more favourable reduction than for Te).Subsequent analyses of numerous samples have shown that these solutions should not be left for more than 12 h before analysis as reductions in Te signals have been observed. Using this scheme, the remaining interference from Bi (unaffected by the presence of KI), shown in Fig. 7(b), is tolerable for the vast majority of geological samples. The associated mineralogy of these elements suggests that where the concentration of Bi is high, it is likely that Te is also anomalous and hence dilution could be used to negate this suppression.Clearly, the presence of KI would negate the possibility of measuring Se in t same solution as that for Te but Bi and Sb could both be determined with Te. It should be noted that Barth et al.18 have suggested the addition of KI in the Fig. 9 Effect of Te on 209Bi at 0.2 ppb (mg l-1) in 4M HCl. determination of Sb to eliminate interferences from Te and Bi Journal of Analytical Atomic Spectrometry, January 1997, Vol. 12 101addition of a small amount of KI and ascorbic acid, at a strength of 0.005% in 4 M HCl. Other mutual interferences of possible concern comprise the effect of: a 5000-fold excess of Te on Bi; a 2500-fold excess of Bi on Sb; and a 5000-fold excess of Bi on Se. However, the likelihood of these concentration ratios arising in geological materials is minimal. If the levels of low abundance elements, such as Te and Bi, were high (i.e., in mineralised samples), then it is probable that Se and Sb would also be elevated and dilution could be used to negate interference.Attempts have not been made here to elucidate the mechanisms of interference, other than to distinguish between the mode of suppression by Bi on Te and Se (by action of deposited Bi sequestering the hydride) and that of AsV on Te (by action during hydride generation). Rather, our goal has been to obviate such interferences by sample treatment (addition of Fig. 10 Effect of Bi on 121Sb at 0.2 ppb (mg l-1) in 4M HCl. KI) and hence, using this scheme, mutual interferences in HG–ICP-MS are not of concern for the analysis of geological materials. Clearly, analysis for these four elements must be carried out in two runs, one for Se without KI and the other for Te, Bi and Sb with KI. Arsenic may be determined with either run, as AsV with Se or alternatively as AsIII with the other elements. The authors are grateful to Nimal de Silva for converting the original continuous flow HG scheme to a miniaturised version with improved conditions of mixing and flow.Thanks are also given to Gilles Gauthier and Alice MacLaurin for assistance in this development. REFERENCES 1 Nakahara, T., Spectrochim. Acta, Rev., 1991, 14, 95. 2 Rose, A. W., Hawkes, H. E., and Webb, J. S., Geochemistry in Fig. 11 Effect of Bi on 78Se at 0.2 ppb (mg l-1) in 4M HCl. Mineral Exploration, Academic Press, New York, 1979. 3 Kabata-Pendias, A., and Pendias, H., T race Elements in Soils and and, if so, calibration by standard addition would appear Plants, CRC Press, Boca Raton, Florida, 1985. appropriate. 4 Merian, E., Clarkson, T. W., and Fishbein, L., Metals and their Compounds in the Environment, UHC Verlagsgesellchaft, Germany, 1991, pp. 1153–1158. Selenium 5 Aslin, G. E. M., J. Geochem. Explor., 1976, 6, 321. 6 Wickstrøm, T., Lund, W., and Bye, R., J. Anal. At. Spectrom., Neither As nor Sb, in the III or V valency state at concentrations 1995, 10, 803.up to 2000 mg l-1, interfered in the determination of 0.2 mg l-1 7 Bedard, M., and Kerbyson, J. D., Anal. Chem., 1975, 47, 1441. of Se. Tellurium had no effect on the Se signal. Bi appeared to 8 Dedina, J., Anal. Chem., 1982, 54, 2097. 9 Welz, B., and Stauss, P., Spectrochim Acta, Part B, 1993, 48, 951. produce an enhancement initially, but there was as much as 10 Walcerz, M., Bulska, E., and Hulanicki, A., Fresenius’ J.Anal. 0.06 mg l-1 of Se in the 1000 mg l-1 Bi test solution (Fig. 11). Chem., 1993, 346, 622. Clearly, there was a suppression at levels of Bi greater than 11 Dittrich, K., and Mandry, R., Analyst, 1986, 111, 277. 1000 mg l-1, probably with decomposition of SeH2 with 12 Wickstrøm, T., Lund, W., and Bye, R., J. Anal. At. Spectrom., elemental Bi, as with Te. It is interesting that this suppression 1995, 10, 809. is much less severe for Se compared with Te, indicating greater 13 Uggerud, H., and Lund, W. J. Anal. At. Spectrom., 1995, 10, 405. 14 Powell, M. J., Boomer, D. W., and McVicars, R. J., Anal. Chem., stability of its hydride. As is the case for Te and Sb, this 1986, 58, 2867. interference by Bi does not create a problem in geological 15 Hall, G. E. M., J. Geochem. Explor., 1992, 44, 201. samples. 16 Quijano, M. A., Gutie�rrez, A. M., Conde, M. C. P., and Camara, Concentrations of Fe (added as FeCl3 ) and La (added as C., J. Anal. At. Spectrom., 1995, 10, 871. LaCl3) of up to 2000 mg l-1 and 1%, respectively, in the 4 M 17 Fujita, K., and Takada, T., T alanta, 1986, 33, 203. HCl test solution did not have any effect on the signal 18 Barth, P., Krivan, V., and Hausbeck, R., Anal. Chim. Acta, 1992, 263, 111. intensities of Bi, Sb, Se or Te (each at 0.2 mg l-1). Thus, with 19 Hershey, J. W., and Keliher, P. N., Spectrochim. Acta, Part B, a dilution factor of 40 in the analytical preparation scheme, 1986, 41, 713. up to8% Fe in the sample can be toleratedwithout interference. Paper 6/05398J Received August 1, 1996 CONCLUSIONS Accepted October 25, 1996 Of the four analytes studied, Te is the most susceptible to mutual interference, significant suppression beginning at interferent5analyte ratios of 1000 for AsV and 2500 for Bi and Se. However, interference from AsV and Se can be negated by 102 Journal of Analytical Atomic Spectrometry, January 1997, Vol.