Analyst, September, 1974. Vol. 99, p p . 595-601 595 Atomic-absorption Studies on the Determination of Antimony, Arsenic, Bismuth, Germanium, Lead, Selenium, Tellurium and Tin by Utilising the Generation of Covalent Hydsides BY K. C. THOMPSON AND D. R. THOMERSON (Shandon Southern Instruments Ltd., Frimley Road, Camberley, Surrey, G U16 5ET) A method for the determination of arsenic, bismuth, germanium, lead, antimony, selenium, tin and tellurium by means of hydride generation is described. The hydrides are generated by adding the acidified sample to dilute (1 per cent. m/ V ) sodium borohydride solution. The liberated hydrides are passed directly into a 17 cm long silica tube mounted in an air - acetylene flame. The advantages of the proposed system are its simplicity, high Sensitivity, high speed of analysis and the fact that background correction facilities are not required.The generation of plumbane for analytical purposes does not appear to have been reported previously. THE hydride generation technique, with subsequent atomic-absorption spectrophometry in a suitable flame (usually an argon - hydrogen diffusion flame) , is now a well known method for the determination of arsenic. The original methods for determining arsenic made use of a zinc metal reduction1 -5 with some form of collection of the arsine prior to actual measurement. The method has been extended to ~elenium~-~ and antimony.' Pollock and West, 8 ~ ) further extended the technique to the determination of bismuth, antimony and tellurium by using a magnesium - titanium(II1) chloride reduction, and to germanium by using a reduction with sodium borohydride.Sodium borohydride has been shown to be a promising reducing agent lo -12 and has been used for the determination of arsenic, antimony, bismuth, germanium, selenium, tin and tellurium12 by the method of hydride generation. There are certain disadvantages when using the methods described : appreciable collection times are normally required for generation of the hydride; a collection vessel (or balloon) is usually required to store the hydride and liberated hydrogen; the sensitivity is rather limited because of the large dilution factor involved when the liberated hydrides are passed into an argon - hydrogen diffusion flame ; and background (non-specific) absorption is usually observed below 200 nm when the hydrides and hydrogen are passed from the collection vessel to the flame.L4n improvement in the speed of arsenic determinations by using a zinc column has been reported by Lichte and Sk0gerb0e.l~ The samples were injected on to the column and the arsine was evolved rapidly while the sample percolated through the column ; no collection vessel was used. The replacement of the flame with an electrically heated, 15 cm long, 2.5 cm i.d. silica tube14 was found to double the arsenic sensitivity compared with that obtained by use of a flame. This paper reports the use of a 17 cm long, 0.8 cm i.d. silica tube mounted in an air - acety- lene flame in order to effect atomisation of the generated hydrides. The advantages of this technique are that no collection vessel is required, that background flame absorption for arsenic, selenium and tellurium determinations is effectively eliminated, thus obviating the requirement of an automatic background corrector, that the narrow silica tube gives a relatively large increase in sensitivity compared with direct sample injection into an argon - hydrogen diffusion flame, and that very little modification to the atomic-absorption spectro- photometer, other than the mounting of two Terry clips on the grid of a wide-path air - acety- lene burner, is required. A cell for the rapid generation of hydride (without associated collection vessels) has been developed that utilises a reduction with sodium borohydride. The reduction products are Q SAC and the authors.596 THOMPSON AND THOMERSON : ATOMIC-ABSORPTION DETERMINATION OF [Analyst, Vol.99 continuously swept into the heated silica tube by using a constant flow of carrier gas. One sample can be run every 40 s while using this form of generation. The borohydride method has been found to be applicable to the determination of arsenic, bismuth, germanium, lead, antimony, selenium, tin and tellurium. The generation of plumbane (lead hydride) by using an aqueous solution of sodium borohydride does not appear to have been reported before. Some results have also been obtained for arsenic by use of a zinc column similar to that described by Lichte and Skogerboe.13 One sample could be run every 30 s and the column could be used for fifty to seventy samples before it was necessary to replace the zinc granules. EXPERIMENTAL Results were obtained by using a Shandon Southern Instruments A3400 atomic-absorp- tion spectrophotometer, Shandon Southern Instruments hollow-cathode lamps and a Kipp and Zoiien BD8 recorder used in the 10-mV range. The wavelengths used in this study are reported in Table I.TABLE I Element AS AS Bi Ge Pb Sb Se Sn Te DETECTION LIMITS AND CHARACTERISTIC CONCENTRATIONS Sample volume = 1 ml Characteristic concentration Volume of 1 per cent. Concentration Nitrogen (1 per cent. Wavelength/ m/ V NaBH,/ of HC1/ flow-rate/ absorption)/ nm ml M 1 min-1 pg ml-1 193-7 2 1.5 1.2 0.000 52 193.7 Zinc column 1.5 1.0 0.001 223.1 2 1.5 1.2 0*000 43 265.1 2 1.5 1-2 1.0 233.3 1 0.2 1.2 0.08 2 17.6 2 1.5 1.2 0.000 61 196.1 2 1.5 3.0 0.0021 224.6 2 0.5 1.2 0*000 44 214.3 2 1.5 3.0 0.0020 Detection limit pg ml-l 0.0008 0.00 15 0*0002 0.5 0.1 0.0005 0.001s 0~000.5 0.0015 (24 / Stock solutions containing 1000 pg ml -l of arsenic(III), arsenic(V), bismuth(III), germanium(IV), lead, antimony(III), selenium(IV), selenium(VI), tin(II), tellurium(1V) and tellurium(V1) were prepared. The results for the higher oxidation state of some of these elements (i.e., arsenic, selenium and tellurium) are discussed later.High-purity analytic21 grade (Aristar) hydrochloric acid and Alfa Inorganics sodium borohydride 10/32-inch pellets* (98 per cent. pure) were used. The tube was mounted above the burner grid of a wide-path air - acetylene burner. The generated hydride, contained in the stream of nitrogen, was introduced through the side-arm (A) in the middle of the silica tube.Provision was made for cooling the outside of this side-arm via annulus(B), and a cooling air flow-rate of 5 1 min --I. was used. An auxiliary nitrogen stream was injected into the two transverse tubes (CC’) that were positioned 1 cm from the ends of the silica atomising tube (Fig. 1). This transverse stream of nitrogen prevents the liberated hydrogen from igniting at one or both ends of the tube. When ignition does occur, in the absence of the transverse nitrogen stream, the resulting flame absorbs radiation weakly below 200nm and causes a small background absorption. If ignition occurs when the auxiliary gas stream is flowing, the resulting flame does not ignite on the optical axis but on the ends of tubes D and D‘ (Fig.1). Instead of these trans- verse tubes, silica windows could be positioned over the ends of the tube, leaving a Z to 2-mm gap. This device would eliminate molecular absorption as the resulting flame ignition would take place away from the optical axis but this system, when used, caused a slight loss in sensitivity. The main disadvantage was that transmission at wavelengths below 200 nni rapidly decreased with use if the windows were not frequently removed and repolished. The measurement procedure was as follows. A 1 to 2-ml amount of 1 to 2 per cent. m/V sodium borohydride solution was injected into the generator cell (see Fig. 2). The sample, containing a suitable concentration of hydrochloric acid (see below), was contained in a l-ml MLA pipette (Shandon Southern Instruments Ltd.), which formed an air-tight seal with the side-arm of the cell.When a stable base-line was obtained, the sample was injected into * Obtainable from Ralph N. Emanuel Ltd., 264 Water Road, Wembley, Middlesex. The silica atomising tube is depicted in Fig. 1.September, 19741 SOME ELEMENTS UTILISING GENERATION OF COVALENT HYDRIDES 597 B A t N2 t hydride Fig. 1. Silica atomising tube. 1 cm from ends of atomising tube. A, side-arm; B, annulus; and C , C’, D, D’, transverse tubes, Half actual size the cell and the signal monitored. The cell was then removed, emptied and the procedure repeated. Arsine could also be generated from the simple zinc column (see Fig. 3). The column contained 30-mesh, arsenic-free zinc powder (obtained from Fisher Scientific).The sample, 0.5 to 1 ml in 1.5 M hydrochloric acid, was injected through the septum cap on to the top of the column and the resulting arsine was carried in the nitrogen stream out of the bottom of the column into the silica tube. This system gave satisfactory results only for arsenic and was less sensitive than the sodium borohydride reduction method (see Table I). OPTIMISATION OF OPERATING CONDITIONS- Hydride generator design-The design of the sodium borohydride cell is shown in Fig. 2. The-gas was not bubbled through the solution but simply passed over the surface of the liquid in the cell. The volume of the cell was sufficient to ensure that actual sample carry-over due to rapid evolution of hydrogen when the acidified sample was added to the cell did not occur to any appreciable extent.This property was demonstrated with cadmium, which should not It was possible to obtain a result every 40 s. +- N2in N2 in Zinc 30 Fig. 2. Hydride generator cell (containing Fig. 3. Zinc column (for generation of NaBH,). One third actual size arsine). One third actual size698 THOMPSON AND THOMERSON ATOMIC-ABSORPTION DETERMINATION OF [AIZalyst, VOl. 99 form a volatile hydride although any carry-over will be readily atomised; the high sensitivity for cadmium should make detection of any carry-over very easy. A 1 pg ml-l solution of cadmium in 1.5 M hydrochloric acid gave a negligible signal with 1 ml of a 2 per cent. m/V solution of sodium borohydride. If the sodium borohydride concentration was increased to 4 per cent.m/V a small signal was observed. The cadmium signal increased rapidly with further increases in the bordhydride and hydrochloric acid concentrations and it gave a sharper peak than the hydride peaks. Severe memory effects were also observed with cadmium, but not with the hydrides. When the generator cell was replaced by a much smaller cell, and the PVC and silicone rubber tubing connecting the cell outlet to the silica tube was shortened from 60 to 40 cm, the cad- mium response, unlike the hydride response, increased greatly (100 times). It is extremely unlikely that the lead (plumbane) peaks observed were due to carry-over, for a number of reasons : firstly, the sensitivity for cadmium at 228.5 nm is much greater than that for lead at 283.3 nm; secondly, the lead response greatly decreased with increasing acid concentration ; and thirdly, the response to lead did not change appreciably (unlike the cadmium response) when the cell shown in Fig.2 was replaced by the much smaller cell. The nitrogen inlet was at the top of the column and the exit at the bottom. Evolution of hydrogen did not occur to the same extent as with the borohydride generation system. Acid concentration-The response to arsenic, bismuth, germanium, antimony, selenium and tellurium when using sodium borohydride was not very dependent on the hydrochloric acid concentration (1 to 4 M). For tellurium, a slight increase was observed with increasing acid concentration from 1 to 4 M. With tin, the response decreased markedly if the acid concentration exceeded 0-7 M.These results are similar to those obtained by Fernandez.12 With lead, both the acid and the borohydride concentrations were critical. The optimum concentration of hydrochloric acid was 0.2 M when a suitable excess of sodium borohydride was used. With the zinc column the arsenic solutions were prepared in 1.5 M hydrochloric acid.13 Sodium borohydride solutiofi concentration-For all of the elements studied except lead, the sodium borohydride concentration was not very critical; 2 ml of 1 per cent. wz/V sodium boro- hydride solution for a 1-ml volume of sample was found to be satisfactory. When using certain samples reg., determining arsenic in a 10 000 pg ml-l solution of iron (11) or in the diluted solution resulting from the fusion with sodium peroxide of a tungsten alloy], it was found that the borohydride concentration should be increased to 2 per cent.m/V in order to ensure adequate reduction of arsenic, bismuth, germanium, antimony, selenium, tin and tellurium. Other solution components can consume sodium borohydride and consequently minimise hydride formation. This increased borohydride concentration gave a slight reduction in sensitivity, probably due to the greater volume of hydrogen that is evolved resulting in dilution of the hydrides. The decrease in sensitivity was more marked for selenium than for the other elements. In the determination of lead the sodium borohydride concentration was critical. For a 1-ml solution of sample in 0.2 M hydrochloric acid, the optimum volume of 1 per cent.mjV sodium borohydricle solution was 1 ml. If the borohydride concentration was increased the response slowly declined, and if the acid concentration was increased the response rapidly decreased to zero. It would appear that for generation of the unstable lead hydride the final solution in the cell must contain an excess of borohydride. The relatively poor sensitivity obtained for lead indicated a poor efficiency of conversion to the hydride (less than 5 per cent.). The use of an ice-bath in the preparation of all samples and for the immersion of the generator cell did not markedly increase the lead response. Similarly, if the sodium borohydride solution was mixed with a neutral lead solution and allowed to stand in the generator cell for 1 to 10 minutes and then 1 ml of 0-2 M hydrochloric acid was added to the cell, no increase in response was observed.Carvier gas-Nitrogen and argon gave similar responses ; nitrogen was used because it was cheaper. The optimum nitrogen flow-rate was 1.2 1 min -l for all of the elements except selenium and tellurium when a flow-rate of 3.0 1 min -l was used. With lower gas flow-rates the precision of determination of selenium and tellurium decreased and the selenium signal took appreciably longer t o return to the base-line. The total auxiliary transverse nitrogen flow-rate was 3 lmin-l. With the zinc column the optimum gas flow-rate was 1 lmin-l, The zinc column is shown in Fig. 3. The acid concentrations used in this study are listed in Table I.September, 19741 SOME ELEMENTS UTILISING GENERATION OF COVALENT HYDRIDES 599 although the rate was temporarily increased to 3 1 min-1 just before the addition of a new sample.This increase ensured a stable base-line with no memory effects. FZame-An air - propane flame was tried initially, but it gave much lower results for arsenic than the air - acetylene flame, probably because of incomplete breakdown of the hydride when the silica tube was surrounded by the relatively cool air - propane flame. All of the results given were obtained by using a stoicheiometric air - acetylene flame, supported on a wide-path air - acetylenc burner. Damj?kg-In order to attain the optimum signal to noise ratio the A3400 spectrophoto- meter was operated at damping position 1 (time constant = 0.5 s).The pen recorder response was 0-8 s (full-scale deflection). RESULTS DETECTION LIMITS AND CHARACTERISTIC CONCENTRATIONS- listed in Table I. shown in Fig. 4. Detection limits (20) and characteristic concentrations (1 per cent. absorption) are A typical arsenic trace is These correspond to a l-ml addition of sample. 1 ml 0.01 pg mi-’ arsenic (in 1.5 M HCI) I Time - Fig. 4. Typical arsenic trace (2 ml of 1 per cent. NaBH, solution in the generator cell) With germanium small blank signals were observed, which were thought to be caused by the deposition of involatile germanium metal from the decomposition of the hydride on the inside of the silica tube. The hydrogen and the small amount of hydrochloric acid liberated when a subsequent blank was added to the generator cell probably caused volatilisation of some of this metal film.The only other significant blank was observed with arsenic, which was traced to arsenic in the sodium borohydride. By using 2 ml of a 1 per cent. m/V solution of the sodium boro- hydride reagent from Alfa Inorganic the blank corresponded to 0-0025 pg of arsenic, and when sodium borohydride from BDH Chemicals Ltd. was used the corresponding figure was 0.005 pg. The background signal at 193.7 nm when using a deuterium hollow-cathode lamp corresponded to less than 0.0008 pg of arsenic for both reagents. The relative blank could be reduced by using less borohydride. Various methods for removing this blank were tried; passing helium through a 1 per cent. m/V borohydride solution has been reported to remove arsenic as the hydride.1° On passing nitrogen through 1 per cent. m/V sodium borohyride solution for 1 hour no significant reduction in the arsenic blank was observed.However, if 2 pg of arsenic(II1) were added to 20 ml of 1 per cent. m/V sodium boro- hydride solution, this additional arsenic “blank” could be removed by bubbling nitrogan through the solution for 10 minutes ; nevertheless, the original reagent blank signal remained.600 THOMPSON AND THOPV.:ERSON : ATOMIC-AESQRPTIQN DETERMINATION OF [Analyst, Vol. 99 With arsenic(V) a longer bubblicg time (30 minutes) was required to remove the added arsenic. Increasing the hydrochloric acid concentration from 1.5 M to 3 M merely increased the blank by 25 per ccnt. and did not affect the sensitivity to arsenic.Hence, it would appear that the arsenic blank originates mainly from the sodium borohydride rather than the hydrochloric acid and that the arsenic present in the borohydride is not in the I11 or V oxidation states. Recrystallisation of the sodium borohydride would probably decrease the reagent blank and would give an improved arsenic detection limit (see Fig. 4). CALIBRATION GRAPHS- The graph obtained for tin (not shown) was very similar to that for bismuth. The linearity of the arsenic, selenium and tellurium graphs could be improved by using an electrodeless microwave15~ l6 or radioirequency-poweredl2 lamp. Some calibration graphs are shown in Fig. 5. w 0 Fig. 5 (a) and ( b ) . Calibration graphs (conditions as described in Table I) PRECISION STUDIES- It was important that any air in the generator cell should be flushed through the system prior to injection of the sample.The results of some precision studies are summarised in Table 11. TABLE 11 PRECISION STUDIES WITH SODIUM BOROHYDRIDE Sample volume = 1 ml Concentration/ Number of Relative standard Element pg ml-l measurements deviation, per cent. As 0.01 10 Bi 0.1 10 Sb 0.02 10 Se 0.1 10 Sn 0.02 10 Te 0.1 10 OXIDATION STATE OF THE ELEMENT WITH ARSENIC, SELENIUM AND TELLURIUM- All previous results were obtained by using arsenic(III), selenium(1V) and tellurium(1V) solutions. With 1 nil of 1 per cent. nz/V sodium borohydride solution the signal from 0.1 pg of arsenic(V) was 37 per cent. of that from 0.1 pug of arsenic(II1). On increasing the sodiumSeptember, 19741 SOME ELEMENTS UTILISING GENERATION OF COVALENT HYDRIDES 601 borohydride concentration to 4 per cent.mlV, the arsenic(V) signal was found to be 90 per cent. of that obtained for arsenic(II1). Solutions of selenium(V1) and tellurium(V1) gave a negligible response compared with equivalent amounts of selenium(1V) and tellurium(IV), even when 1 ml of 4 per cent. m/V sodium borohydride solution was used. The selenium(V1) and tellurium(V1) solutions and the corresponding selenium(1V) and tellurium(1V) solutions were simmered with an equal volume of aqua regia (3 + 1 V/V hydrochloric acid - nitric acid) for 15 minutes, allowed to cool and then diluted to twice the volume of sample originally taken. It was then possible to add 1 ml of the resulting solution directly to the generator cell containing 2 ml of 2 per cent.m/V sodium borohydride solution. Under these conditions tellurium(V1) was quantitatively reduced by the hydrochloric acid to the tellurium(1V) state, but less than 50 per cent. of the selenium(V1) was reduced. For solid samples, dissolution in aqua regia followed by a suitable dilution (1 + 1 or greater) with distilled water is recommended. This technique would ensure that the resulting solution contained arsenic(V), selenium(1V) and tellurium(1V). The only other element in this study that commonly occurs in two oxidation states is tin. It was assumed that all dilute tin solutions (concentration of less than 0.1 pug ml -l) would be in the tin(1V) oxidation state. With the zinc column the signal from 0.1 pg of arsenic(V) was 55 per cent.of that from arsenic(II1). The prior addition of tin(I1) chloride and potassium iodide to the sample is normally recommended for the reduction of arsenic(V) when using zinc in order to generate arsine . CONCLUSIONS The hydride generation technique using sodium borohydride, coupled with a flame-heated, narrow-bore, silica atomising tube, constitutes a very sensitive method for the determination of arsenic, bismuth, antimony, selenium, tin and tellurium. The method can also be used to determine germanium and lead. (This is thought to be the first report of the generation of plumbane for analytical purposes.) The proposed method has several advantages over the use of zinc granules for the generation of hydride, followed by atomisation of the liberated hydride in an argon - hydrogen flame, viz., negligible background absorption for arsenic and selenium, increased sensitivity, applicability to a larger range of elements and increased speed of analysis. The authors thank Mr. R. G. Godden for many helpful suggestions and the Directors of Shandon Southern Instruments Ltd. for permission to publish this paper. 1 . 2. 3. 4. 5. 6. 7 . 8. 9. 10. 11. 12. 13. 14. 14. 16. REFERENCES Holak, W., A7zalyt. Chem., 1969, 41, 1712. Madsen, R. E., Atom. Absorption Newsl., 1971, 10, 57. Dalton, E. F., and Malanoski, H. J., Ibid., 1971, 10, 92. Manning, D. C., Ibid., 1971, 10, 123. Fernandez, F. J., and Manning, D. C., Ibid., 1971, 10, 86. Yamamoto, Y . , Kumamaru, T., Hayashi, Y . , and Kanke, M., Analyt. Lett., 1972, 5, 717. Yaniamoto, Y., Kumamaru, T., Hayashi, Y., and Tsujino, R., Ibid., 1972, 5, 410. Pollock, E. N., and West, S. J., -4tom. Absorption ArewsE., 1972, 11, 104. Sullivan, E. A., “Sodium Borohydride : Handling, Uses, Properties, Analytical Procedures,” Braman, R. S., Justen, L. L., and Forebank, C. C., Analyt. Chem., 1972, 44, 2196. Fernandez, F. J., Atom. Absovption Newsl., 1973, 12, 93. Lichte, F. E., and Skogerboe, R. K., Analyt. Chem., 1972, 44, 1480. Chu, R. C., Barron, G. P., and Baumgarner, P. A. W., Ibid., 1972, 44, 1476. Dagnall, R. M., Thompson, K. C., and West, T. S., Talanta, 1968, 15, 677. , , Ibid., 1973, 12, 6. -- Booklet No. 01915, Ventron Corporation, Beverley, Mass., U.S.A. , , Ibid., 1967, 14, 557. --- Received March 12th, 1974 Accepted March 26th, 1974