232 Analyst, March, 1979, Vol. 104, pp. 232-240 Determination of Selenium in Soil Digests by Non -dispersive Atomic-FI uorescence Spectrometry Using an Argon = Hydrogen Flame and the Hydride Generation Technique J. Azad, G. F. Kirkbright and R. D. Snook Department of Chemistry, Imperial College, London SW7 2BP The determination of selenium at submicrogram levels by atomic-fluorescence spectrometry, based on the evolution of hydrogen selenide into an argon - hydrogen air-entrained flame, is described. Using a simple purpose-built non-dispersive atomic-fluorescence spectrometer a detection limit of 10 ng cm-3 of selenium is obtained. The technique has been applied to the determina- tion of selenium in soil digests and experiments have been carried out in order to study the interference of other elements on the determination.Procedures for the elimination of interferences from copper are recommended. Keywords Selenium determination ; atoulzic-fluoracence spectrometry ; lzydride generation ; soil digests The volume of literature published on the flame spectrometric determination of selenium is small compared with that available for many other elements. The determination of selenium by flame spectrometry presents some problems ; for example, the selenium resonance lines lie in the far ultraviolet region of the spectrum below 200nm and this frequently leads to unfavourable signal to noise ratios resulting from atmospheric and back- ground absorption of these selenium lines. Rann and Hambly,l however, obtained a sensitivity (for 1% absorption) of 1.0 pg ~ m - ~ of selenium by atomic-absorption spectro- metry (AAS) using the 196.1-nm selenium resonance line and an air - acetylene flame.With the introduction of the argon - hydrogen air-entrained flame2 the problem of flame absorption was greatly reduced but more severe interference effects were observed in AAS for selenium in this cooler flame. In order to provide higher sensitivity and to overcome some of these interferences chemical separation procedures have been developed that are based on the evolution of hydrogen selenide into the The original method for the determination of selenium by AAS in this way employed reduction with metallic zinc, and a collection and storage device was used for the evolved hydrogen selenide prior to its introduction into the flame for AAS.6 Pollock and West' extended the technique to the determination of bismuth, antimony and tellurium by AAS using a magnesium metal- titanium( 111) chloride mixture as reductant.Schmidt and RoyeI.8 reported the determina- tion of selenium, arsenic, bismuth and antimony by AAS using the hydride generation technique in a procedure in which a solution of sodium tetrahydroborate( 111) was employed as the reductant; this reagent had previously been shown by other workersV~lO to provide rapid and efficient reduction. Sodium tetral~ydroborate(II1) has been employed in this work for the generation of hydrogen selenide prior to the determination of selenium by atomic-fluorescence spectroscopy (AFS) in an argon - hydrogen air-entrained diffusion flame using a simple non-dispersive atomic-fluorescence spectrometer.This paper reports the development of a reliable, simple method for the determination of selenium in aqueous solution by this technique and a study of the chemical interference effects encountered. The atomic-fluorescence spectrometric determination of selenium was first reported by Dagnall et aZ.ll using a dispersive spectrometeir equipped with an air - propane flame giving a detection limit of 0 . 2 5 p g ~ m - ~ of selenium on aspiration of aqueous solutions using a pneumatic nebuliser. Fluorescence from the 204.0-nm selenium resonance line was observed when the flame was irradiated by radiation from a selenium electrodeless discharge lamp, the optical axis of which was aligned at 90" to the optical axis of the monochromator. In this study a similar experimental arrangement has been employed using a non-dispersiveAZAD, KIRKBRIGHT AND SNOOK 233 spectrometer with which it was possible to observe fluorescence from the 196.1-, 214.3- and 204.0-nm lines simultaneously, thus enabling a detection limit of 10 ng cme3 to be observed using discrete sample introduction via the hydride generation technique. Two procedures have been investigated for the suppression or elimination of the well known interference of copperl29l3 on the determination of selenium by the hydride genera- tion technique.In the first procedure the copper and other interfering ions are removed when selenium is coprecipitated with lanthanum from alkaline medium and the precipitate containing selenium is taken for analysis by the procedure developed.In the second procedure the interference from copper is suppressed by utilising the addition of tellurium( IV) to the analyte solution; stable copper telluride is formed and the interference of copper on the hydrogen selenide generation step is suppressed. The chemical pre-treatment and atomic-fluorescence spectrometric procedures developed have been applied to the deter- mination of selenium in soil digests obtained after digestion with a perchloric acid - nitric acid mixture. Experimental Apparatus The instrumentation employed in this work was a purpose-built non-dispersive atomic- fluorescence spectrometer and a simple hydride generation apparatus. A schematic diagram of the equipment employed is shown in Fig.1 and the details of the components employed are listed in Table I. Radiation from a microwave-excited selenium electrodeless discharge lamp (EDL) was focused on to a rotating sector and then refocused into the argon - hydrogen flame. The atomic-fluorescence radiation stimulated from selenium atoms in the flame was then observed at 90" to the incident radiation by passage through a focusing lens to the solar-blind end-window photomultiplier. The output from the photomultiplier was taken to a lock-in amplifier whose reference signal was provided by the rotating sector in the incident radiation beam from the EDL source. The analytical atomic-fluorescence signals for selenium observed at the output from the lock-in amplifier were displayed at the pot en t iometric chart recorder.E. H.T I Baffle L_? I A Loc k- in amp1 if ier L' Reference P.S.D. Hydride generation cell Chart recorder Microwave generator I I Fig. 1. Schematic diagram of equipment employed. Reagents Selenium( IV) standard solutions were prepared by dissolving pure elemental selenium (Specpure grade, Johnson Matthey Ltd.) in a minimum volume of concentrated nitric acid and diluting to volume with 5 M hydrochloric acid. The sodium tetrahydroborate(II1)234 AZAD et al. : DETERMINATION OF SELENIUM IN SOIL Analyst, Vol. 104 reagent was used as a freshly prepared 5% (m/’V) solution in 1% sodium hydroxide solution. Analytical-reagent grade lanthanum nitrate, tellurium(1V) oxide, perchloric acid, hydro- chloric acid, nitric acid and concentrated ammonia solution were used in all experiments.TABLE I INSTRUMENTATION EMPLOYED Source .. .. .. Chopper. . . . . . Microwave generator . . Photomultiplier . , Lock-in amplifier . . Phase sensitive detector optics . . . . .. Chart recorder . . .. * . .. .. .. .. .. .. .. Selenium microwave electrodeless discharge lamp operated at 2450 MHz in a &wave resonant cavity. Radiation modulated with an eight-sector mechanical chopper Programmable Rofin, Model 7500, 3-800 Hz (Rofin Ltd., Egham, Surrey) Microtron 2001 (EMS Ltd. , Wantage, Berkshire) Solar blind, Type R431, Hamamatsu Co., Japan Brookdeal Electronics, Type 450s (Brookdeal Ltd., Bracknell, Berkshire) Brookdeal Electronics, Type 41 1 (Brookdeal Ltd.) Source focused as 1: 1 image on the flame using two 7.5-cm focal length fused silica convex lenses (L, and La).Flame focused as inverted 1: 1 image on PMT using 7.5-cm focal length lens (Id3) Servoscribe, Model RE 511.20 (Smiths Industries Ltd.) Procedure With the flame ignited and argon passing through the hydride generation cell, sufficient time (approximately 20s) was allowed for the_ replacement of any air in the apparatus. A 2-cm3 volume of sodium tetrahydroborate(II1) reagent solution was then transferred into the generation cell through the side-arm. Acidified selenium standard solution (or sample solution) (1 cm3) was then pipetted into the sodium tetrahydroborate(II1) solution using a syringe pipette whose tip was fitted with a rutbber sleeve to ensure a gas-tight fit with the side-arm of the cell during sample introduction.The hydrogen selenide generated was then swept into the argon-hydrogen flame by the argon supply to the flame. The selenium atomic-fluorescence signal was recorded at tlhe potentiometric chart recorder ; the signal duration observed was approximately 8 s for ii 5 pg ~ m - ~ selenium standard solution. The optimum operating conditions established for the procedure, with the particular instru- mental arrangement employed, are summariseti in Table 11. TABLE I1 OPTIMUM OPERATING CONDITIONS FOR DETERMINATION OF SELENIUM Microwave power to source . . .. .. .. .. .. Reflected power from cavity .. .. .. .. .. Applied voltage to PMT . . .. .. .. .. .. Hydrogen flow-rate . . .. .. .. .. .. * . Argon flow-rate . . .. .. .. .. .. .. Hydride generation cell volume .. .. .. .. .. Sodium tetrahydroborate(II1) reagent volume (5% m/ V ) Selenium sample solution volume . . .. .. .. .. .. 50 W 12 w 600 V 3.3 dma min-1 6.0 dm8 min-1 46 cma 2 cm3 1 cm3 Determination of Selenium in Soil Digests One-gram amounts of soil samples were weighed into a series of test-tubes, 3.5 cm3 of oncentrated nitric acid were added to each sample and the test-tubes were covered and llowed to stand overnight. A few glass boiling beads were added to each tube and thenMarch, 1979 DIGESTS BY NON-DISPERSIVE ATOMIC-FLUORESCENCE SPECTROMETRY 235 1.5 cm3 of concentrated perchloric acid (72% m/V) were added to each. The tubes were then transferred into a cold aluminium digestion block, the temperature of which was increased steadily to 100 "C over a period of 30 min.The block was maintained at this temperature for 30 min and then the temperature was increased to between 190 and 200 "C and main- tained at this temperature until digestion of the soil was complete. The final temperature of 200 "C should not be exceeded if charring and the loss of selenium by volatilisation are to be avoided. The test-tubes were then removed from the digestion block and allowed to cool. A 2-cm3 volume of potassium bromide solution (2% m/V) was added to each and the test-tubes were allowed to stand in boiling water for 15 min to ensure complete reduction of selenium(V1) to selenium(1V). The solutions were then centrifuged and the residues rejected. The supernatant solution was taken for analysis; either the lanthanum nitrate - ammonia or the tellurium(1V) addition procedure was applied in order to eliminate inter- ference from copper.The solutions were then made 5 M with respect to hydrochloric acid and analysed by the hydride generation technique using the atomic-fluorescence spectro- meter. Procedures for Suppression of Interferences Lanthanum nitrate coprecipitation procedure Lanthanum nitrate (0.5 cm3 of a 5% m/V solution) was added to each solution prepared for analysis using the digestion procedure described above, 2 cm3 of ammonia solution were then added and the solutions were mixed. After standing for 1 min the solutions were centrifuged and the liquid discarded. The precipitate was then dissolved in the appropriate amount of 5 M hydrochloric acid. Telluyiam(I V ) procedure Tellurium(1V) oxide (0.3 cm3 of a 0.1 M solution) was added to each solution prepared using the digestion procedure described above and then diluted to 5 cm3 with 5 M hydro- chloric acid.Results and Discussion Optimisation of Experimental Parameters Pure aqueous standard selenium(1V) solutions were used in order to optimise the experi- mental variables in the instrumental system, and to provide the best attainable sensitivity and precision in the determination of selenium by the atomic-fluorescence spectrometric technique, utilising the generation of hydrogen selenide for introduction of the analyte into the argon - hydrogen air-entrained flame. The operating power for the microwave-excited EDL source, argon and hydrogen gas flow-rates to the flame, photomultiplier operating voltage, hydride generation cell volume and sodium tetrahydroborate(II1) and selenium sample solution volumes used were each varied independently in order to establish optimum conditions for the determination of selenium.The optimum conditions established in this way are summarised in Table 11. E$ect of hydrochloric acid alzd sodium tetrahydroborate(III) concentrations The effect on the intensity of the atomic-fluorescence signal, observed for 0.5 pg of selenium introduced into the hydride generation cell in 1 cm3 of solution, of variation in the con- centration of hydrochloric acid in the sample solution was investigated. The sodium tetrahydroborate( 111) concentration was maintained constant at 5% (m/V) for this experi- ment. The results obtained are shown in Fig.2. Variation in the acid concentration present in the selenium sample solution has a pronounced effect on the efficiency of generation of hydrogen selenide only when the solution is less than 0.8 M with respect to hydrochloric acid; in all further work the hydrochloric acid concentration of solutions to be analysed was maintained at 5 M. Using 1 cm3 volumes of solution containing 0.5 pg of selenium, which were 1 M with respect to hydrochloric acid, the effect on the selenium atomic-fluorescence signal of vari- ation in the concentration of sodium tetrahydroborate(II1) solution in the generation cell236 AZAD et al. : DETERMINATION OF SELENIUM IN SOIL Analyst, Vol. 104 was investigated. The results obtained are shown in Fig. 3; little variation in hydride generation efficiency was observed over the concentration range 2 4 % (m/V) of sodium tetrahydroborate(II1). A concentration of 5y4, sodium tetrahydroborate( 111) was chosen for use in all further work.100 100 III- 2.0 4.0 6.0 ---i 0.5 1 .o Concentration of HWM Na BH4 concentration, % m/V O 0.1 Fig. 3, Effect of sodium tetra- hydroborate(II1) concentration on the centration on the determination of 0.5 determination of 0.5 pg cm-S of sele- pg cm-S of selenium. nium. Fig. 2. Effect of hydrochloric acid con- Calibration graph, limit of detection and $recisio;vt With the optimum instrumental operating conditions and reagent concentrations, analytical calibration graphs for selenium were found to be rectilinear for selenium solutions containing between 10 and 500 ng cmW3 of selenium in 1 cm3 sample volumes, i.e., 10500 ng of selenium (Fig.4). The relative standard deviation obtained in the repetitive determina- tion of selenium in a solution containing 100ng~m-~ was 2.5%. The detection limit for selenium, defined as that mass of selenium required to produce a signal to noise ratio of 2 for the atomic-fluorescence signal, was 10 ng of selenium under the conditions employed. A significant background blank signal was observed for selenium, equivalent to approxi- mately 16 ng cmV3 of selenium, caused by the presence of selenium as impurity in the sodium tetrahydroborate(II1) reagent ; this blank was corrected for by subtraction in all quantitative analytical work undertaken. Selenium concentration/pg cmv3 Fig.4. Aqueous calibration graph for the determinaMon of selenium by non-dispersive atomic-fluorescence spectrometry.March, 197'9 DIGESTS BY NON-DISPERSIVE ATOMIC-FLUORESCENCE SPECTROMETRY 237 Interference efects and their sufi$ression and elimination The determination of selenium by atomic-absorption spectrometry utilising the hydride generation technique is well known to be subject to interference from a number of heavy metal ions, and in particular copper(II), which depress the efficiency of the hydrogen selenide generation by sodium tetrahydroborate( 111). Similar interferences were expected in the atomic-fluorescence spectrometric procedure developed here and were confirmed in experi- ments in which the effects of metal ions on the atomic-fluorescence signal produced for 500ng of selenium were recorded.Table I11 illustrates typical results of the depressive effects of the presence of some metal ions on the analytical signals observed for 1 pg CM-~ selenium solutions. It is clear that the presence of copper(I1) as a concomitant element causes serious interference; in the presence of 1000 pg cm-3 no atomic-fluorescence signal was obtainable for selenium. As copper(I1) concentrations in soil digests were expected to be sufficiently high to interfere with the selenium determination, two procedures were investigated to minimise or eliminate interference from copper. TABLE I11 DEPRESSIVE EFFECT OF METAL IONS ON THE ANALYTICAL SIGNALS OBSERVED FOR 0.5 pg OF SELENIUM Concentration of interfering element 1 000 pg cm-8.Element Na(1) . . .. K(1) . . .. .. .. .. ::\::\ Ca(I1) . . Ba(I1) . . .. .. Hg(W Al(I1) . . .. Depression of signal, 0 2 0 0 2 0 0 0 % Element Fe(I1) . . .. Fe (I1 I) .. Pb(I1) . . .. Zn(I1) . . .. Co(I1) . . .. Cu(I1) . . .. A m * - .. Ni(I1) . . .. Depression of signal, 36 20 40 21 20 99 80 65 % The procedure reported by Bedard and Kerbyson,12 in which the interference of copper on the determination of selenium, by AAS via generation of hydrogen selenide, was elimi- nated by removal of the selenium from sample solutions by coprecipitation from an alkaline medium with lanthanum, was investigated. This procedure was observed to give good recovery of selenium using a double precipitation; the mean recovery for 10 replicate analyses was 99% with a relative standard deviation of 2.5%.The procedure was found to be most efficient when the lanthanum hydroxide precipitate was filtered off as soon as possible after precipitation. The effect on the recovery of selenium, monitored as the selenium atomic- fluorescence signal intensity, of elapsed time between precipitation and filtration is shown in Fig. 5. The effect of variation in pH of the solution on the selenium recovery by the coprecipitation procedure was found not to be critical provided that the pH was maintained above 9.0. The second procedure investigated for suppression of the interference from copper utilises the addition of tellurium( IV) to sample solutions immediately before the hydride generation procedure ; this procedure has been described elsewhere by Kirkbright and Taddia.13 In this procedure the interference from copper is suppressed by formation of copper telluride, which is more stable than copper selenide.The addition of excess of tellurium(1V) results in some suppression of the atomic-fluorescence signal observed for selenium ; Fig. 6 shows the effect of variation in the tellurium(1V) concentration added to 0.5 p g ~ m - ~ selenium sample solutions on the signal recorded. A constant suppression of approximately 30% is attained at tellurium(1V) concentrations between 0.06 and 0.08 M. The presence of 0.06 M tellurium(1V) enables relatively high concentrations of copper to be tolerated in the deter- mination of selenium; Fig. 7 shows the effect of increasing copper concentration on the atomic-fluorescence signal observed from 0.5 pg ~ m - ~ selenium sample solutions containing 0.06 M tellurium(1V) solution.Copper does not cause interference at levels up to 50 p g cmq3 although at higher concentrations the selenium recovery decreases. Addition of a con- centration of tellurium(1V). greater than 0.06 M to sample solutions would permit extension238 AZAD et al.: DETERMINATION OF SELENIUM IN SOIL Analyst, VoZ. 104 .- f c W W v) - J a - ' 10 20 30 I ' 0 a.04 0.06 0.08 Time lapsed before filtratiodmin Fig. 5. Effect of time elapsed Tellurium (IV) oxide concentration/M before filtration on the recovery of selenium observed when using the Fig. 6. Effect of tellurium(1V) con- lanthanum nitrate coprecipitation pro- centration, added to 0.5pgcm-S of cedure to remove interferences.selenium solutions, on the generation of selenium hydride. of the tolerance of the procedure to higher concentrations of copper(I1). Fig. 8 shows a comparison of the analytical calibration graphs obtained for aqueous selenium solutions in the presence and absence of copper utilising tellurium( IV) to suppress copper interference. These results confirm the restoration of the selenium signal to approximately 70% of its value in the absence of copper when tellurium(1V) is employed to suppress copper inter- f erence . I , I , 0 40 80 120 Copper concentration/pg cm-3 Fig. 7. Effect of copper on the deter- mination of 0.5 p g cm-s of selenium in the presence of 0.06 M tellurium (IV) . B C _- - C 6 Selenium concentration/pg cm-3 Fig. 8. Comparison of analytical calibrations in the presence and absence of 0.06 M tellurium(1V) oxide.A, Aqueous selenium; B, aqueous selenium + 0.06 M telluiium(1V) + 20 pg cm-S of copper; and C, aqueous selenium + 20 pg cm-8 of copper. Determination of Selenium in Soil Digests Soil samples were digested using a mixture of perchloric and nitric acid; care was taken not to char the sample during digestion at 200 "C and to avoid loss of selenium by volatilisa- tion. As the hydride generation procedure is only applicable to selenium in its oxidation state of four it was necessary to reduce any selenium(V1) produced in the strongly oxidising digestion mixture by the addition of potassium bromide solution after digestion. Sample digestion recoveries were evaluated by adding to l-g soil samples a known amount of selenium prior to their digestion.The recovery of the added selenium was then deter- mined. The results of these experiments are shown in Table IV.March, 1979 DIGESTS BY NON-DISPERSIVE ATOMIC-FLUORESCENCE SPECTROMETRY TABLE IV RECOVERY OF SELENIUM ADDED TO SOIL SAMPLE NO. 4 239 Selenium concentration Selenium added/ Selenium determined] Recovery, in sample/pg 8-1 Pg P8 % 0.7 f 0.014 0.1 0.83 104 0.7 f 0.014 0.2 0.87 97 0.7 f 0.014 0.3 0.94 94 0.7 f 0.014 0.4 1.18 107 Nine soil samples were digested using the procedure described. Each sample was then analysed by both the lanthanum coprecipitation and the tellurium(1V) methods of interference suppression. The results obtained for the selenium content of the soils analysed are shown in Table V for both methods of interference suppression. As can be seen from the table there is no appreciable quantitative difference between the results obtained by both methods.These results also show extremely good agreement with those obtained by the hydride generation technique and optical-emission spectrometry using an inductively coupled argon plasma source. TABLE V COMPARISON OF RESULTS FOR THE SELENIUM CONTENT OF SOIL DIGESTS Selenium found r L % soil sample 1 2 3 4 5 6 7 8 9 Lanthanum nitrate method Mean, SD, RSD, 0.37 0.017 4.5 0.36 0.016 4.4 0.24 0.010 4.1 0.70 0.014 2.0 18.7 0.64 3.4 111 2.93 2.6 0.29 0.014 4.8 0.31 0.020 6.4 0.30 0.023 7.6 A r > p.p.m. p.p.m. % Tellurium( IV) method Mean, p.p.m. 0.35 0.35 0.23 0.68 18.6 110 0.28 0.30 0.29 SDI p.p.m. 0.009 0.012 0.015 0.015 0.49 2.45 0.012 0.015 0.018 RSD,’ % 2.6 3.4 6.5 2.2 2.6 2.2 4.2 5.0 6.2 ICP* method, p.p.m.0.38 0.33 0.23 0.69 19.2 - 0.28 - * ICP = optical-emission spectrometry using an inductively coupled argon plasma source. Conclusion It has been demonstrated that selenium can be determined in soil digests using the technique of non-dispersive atomic-fluorescence spectrometry. The technique is both sensitive and precise. Although the hydride generation procedure is normally subject to severe interference from copper, this effect has been eliminated by employing chemical pre- treatment of the samples, using lanthanum hydroxide as a coprecipitant or the addition of tellurium(1V) to forrn stable copper telluride during reduction. Both methods have been applied successfully to the determination of selenium in soil digests.It is difficult to recommend which procedure is the most suitable for selenium determinations as each has its own advantages; the lanthanum hydroxide coprecipitation is well established and with care leads to excellent recoveries of selenium even in the presence of high concentrations of copper (approximately 2000 pg cm-3). In order to obtain good recoveries of selenium, however, re-precipitation must be employed. The tellurium( IV) procedure is a simple method but lowers the detection limit by 30% and at the concentrations employed in this study is only effective in removing interferences from copper up to a concentration of 5 0 ~ g c m - ~ of copper. This limitation may not be a serious disadvantage, however, as copper levels in soil digests should seldom exceed this figure.240 AZAD, KIRKBRIGHT AND SNOOK References Rann, C. S., and Hambly, A. W., Analytica Chim. Ada, 1965, 32, 346. Kahn, H. L., and Schallis, J. E., Atom. Absorption Newsl., 1968, 7, 5. Fernandez, F. J., and Manning, D. C., Atom. Absorption Newsl., 1971, 10, S6. Clinton, 0. E., Analyst, 1977, 102, 187. Siemer, D., and Hagemann, L., Analyt. Lett., 1!)75, 8, 323. Yamamoto, Y. Y., Kumamaru, Y., Hayashi, Y . , and Kanke, M., Analyt. Lett., 1972, 5, 717. Pollock, E. N., and West, S. J., Atom. Absorption Newsl., 1972, 11, 104. Schmidt, F. J., and Royer, J. L., Analyt. Lett., 1973, 6, 17. Pierce, F. D., Lamoreaux, T. C., Brown, H. R., and Fraser, R. S., Appl. Spectrosc., 1976, 38, 38. Fernandez, F. J., Atom. Absorption Newsl., 1973, 12, 93. Dagnall, R. M., Thompson, K. C., and West, T. S., Talanta, 1967, 14, 557. Bedard, M., and Kerbyson, J. D., Can. J. Speabosc., 1975, 21, 64. Kirkbright, G. F., and Taddia, M., Atom. Absovpion Newsl., in the press. Received September 7th, 1978 Accepted October 17th, 1978 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13.