ANALYST, MARCH 1986, VOL. 111 281 Method for Improving the Sensitivity and Reproducibility of Hydride-forming Elements by Atomic Absorption Spectrometry Nicolaos E. Parisis and Aubin Heyndrickx Department of Toxicology, State University of Ghent, Hospitaalstraat 13, 9000 Ghent, Belgium The positive effect of oxygen on the atomisation of hydride-forming elements at temperatures above 800 "C is indicated. Several different materials of construction for gas transport tubing were tested and their influence on the transportation of hydrides and on radical production was studied in order to ascertain their operating efficiency. Silanised glass and FEP tubing gave the highest sensitivity and reproducibility and silicone-rubber and nylon tubing the lowest. The combination of the above, together with the use of 2 M nitric acid as a rinsing agent for the reaction vessel, allows the convenient use of peak areas for the measurement of arsenic, selenium, bismuth, antimony and tin signals at levels of a few nanograms. Keywords: Atomic absorption spectrometry; h ydride generation; oxygen - argon carrier gas; silanised glass tubing The atomisation mechanism for volatile hydride-forming elements in a heated quartz tube was generally thought to involve thermal decomposition.However, in the last few years, some workers have proposed a new mechanism in order to explain the increased sensitivity obtained in their experi- ments. First, Dedina and Rubeskal suggested that hydride atom- isation is caused by free radicals (H-, OH*) generated in the reaction zone of a hydrogen - oxygen flame burning in the T-shaped quartz tube of the system. Welz and Melcher,2 in studies of the selenium interference with trivalent and pentavalent arsenic, found that the only theory that would explain the results was that selenium, which is volatilised earlier than arsenic, increases the deficiency of radicals so that for the arsenic that appears later there remains an insufficient number of radicals to cause atomisation to the same extent as in the absence of selenium.Moroever, arsenic has a consider- ably smaller effect on selenium than vice versa. Recently, the same workers in another study3 concluded that these radicals are formed in a reaction with oxygen at low temperatures (above 600 "C) and in a "clean" environment the concentration of radicals is well above the equilibrium concentration because their formation is a much faster process than the recombination.They found a significant enhancing effect of oxygen on the sensitivity of volatile hydride-forming elements at temperatures around 700-800 "C; they established that this cannot be due to a temperature increase in the gas phase (effective temperature) of the atomiser . In this work, several investigations have been performed with a commercially available hydride system and an electric- ally heated quartz cell atomiser. The influence of different purge gases, gas transfer tubing and reaction flasks on the sensitivity of hydride-forming elements was studied. can be heated to 1000 "C by eight insulated heating wires. This atomisation device, mounted in the sample compartment of the spectrometer, is connected with an electronic system that allows one to select the required duration of purge and reaction, between sample and reagent solution, and controls the temperature to within +20 "C by using an Ni - NiCr thermocouple and a feedback system.Piston stroke pipettes of capacity 10,25 and 50 pl (Assipette No. 100) and pipette tips from Assistent, Sondheim/Rhon, FRG, were used. Tubing Nylon tubing. From Perkin-Elmer (gas supply hose, 079873), 40 cm x 4 mm i.d. Silicone-rubber tubing. From Perkin-Elmer (transfer hose, 094140), 40 cm X 4 mm i.d. Soda-glass tubing. Made in our department from commer- cially available material, 44 cm x 4.5 mm i.d. PTFE tubing. FEP (fluorinated ethylene - propylene) tubing from Rudolf Brand, Wertheim, FRG, 40 cm X 5.7 mm i.d.Procedure Table 1 gives the important operating parameters for the instrument, lamps and hydride system used for the determina- tions. Experimental Apparatus The instrument used was a Perkin-Elmer Model 372 atomic absorption spectrometer, equipped with an external elec- trodeless discharge lamp (EDL) power supply. Perkin-Elmer electrodeless discharge lamps were used for all elements. Results are shown on a four-digit electronic display. A simple diagram of the apparatus is shown in Fig. 1. The hydride-generation device is a Perkin-Elmer MHS-20 mercury - hydride -system. The volatilised hydrides are atomised in an electrically heated quartz tube (165 X 14 mm i.d.) closed at both.ends by quartz windows.The quartz tube Arsine generating chamber W 3% NaBH, in 1% NaOH Fig. 1. Hydride-generation apparatus282 ANALYST, MARCH 1986, VOL. 111 _ _ _ _ ~ ~ ~~~~ - Table 1. Operating parameters MHS-20 EDL Wavelength/ Element power/W nm As(II1) . . . . 8 193.7 Se(IV) . . . . 6 196.0 Bi(V) . . . . 8 223.1 Sb(II1) . . . . 8 217.6 Sn(IV) . . . . 8 224.6 Slit nrn S 0.7 40 0.2 48 0.2 25 0.2 30 0.2 20 width/ Purge I/ Reaction/ Purge II/ 4 20 4 20 4 30 4 50 4 40 S S Other conditions were as follows: measuring mode, inte- grated peak area; reading time, 20 s; background correction, none; quartz cell temperature, 880 k 20 "C; reaction volume, 10 ml; standard dilution liquid, 1.5% mlV H2S04 [except for Sn(IV), 0.75% mlV H2S04]; inlet pressure, 250 kN m-2; and hydride-generating reagent, 3.92 _+ 0.03 ml of 3% NaBH4 in 1% NaOH, 18-22 "C.For all the measurements, 10 ml of 1.5% mlVH2S04 and an appropriate aliquot of the standard solution (10-50 pl) are added to the polypropylene reaction flasks before attachment to the system. When the start button of the control unit is activated, the purge gas flows through the reaction flask for a pre-selected time (purge I) at a flow-rate of 1000 ml min-1, and purges the air from the system. Immediately after, the gas flow-rate is reduced to 400 ml min-1 and a pre-selected amount of tetrahydroborate solution flows through a capillary and the immersion tube into the bottom of the reaction vessel for a pre-selected time (reaction time). At the end of the reaction time, a second selected purge time (purge 11) follows at a flow-rate of 1000 ml min-1, in order to remove all gases from the system.During all the experiments, after each determination of a blank or standard, the reaction vessel was rinsed with 2 M nitric acid (3 x 5 ml) and doubly distilled water (3 X 5 ml). This treatment, as emphasised by Moody and Lindstrom4 and Reamer et a1.,5 is necessary in order to remove all traces of analyte from the walls of the hydride-generation chamber. In this way, the reproducibility of the results and the stability of the base line were greatly improved. Background correction is usually essential for graphite furnace atomic absorption. However, in hydride systems this is generally not so because the analyte is separated from the sample matrix and only very few elements form hydrides and enter the quartz cell atorniser.6 Volatile organic compounds, which could eventually interfere, are decomposed during preliminary ashing of the sample.Background correction could improve the precision of the results for arsenic and selenium owing to the small absorption (0.010 and 0.040 A s for arsenic and selenium, respectively) from the oxygen present in the purge gas. However, this absorption phenome- non is also the same for the blank. This, together with the instability of the deuterium hollow-cathode lamp at wavelengths less than 200 nm, which increases the base-line noise and the detection limit and influences the precision of low-level measurements dramatically, compared with measurements made using the electrodeless discharge lamp alone, does not permit its use.Reagents Sodium tetrahydroborate(III) solution, 3 Yo m1V. Dissolve 30 g of analytical-reagent grade sodium tetrahydroborate(II1) powder (Merck) and 10 g of analytical-reagent grade sodium hydroxide pellets (UCB) in 400 ml of doubly distilled water, dilute to 1 1 and filter. Store in a refrigerator at 4 "C, where it remains stable for at least 1 week. Sulphuric acid working solution, 1.5% mlV. The solution is obtained by diluting the appropriate volume of 96% mlV Suprapur sulphuric acid (Merck) to 1 1. Nitric acid rinsing solution, 2 M. Prepared by diluting the appropriate volume of 65% mlV analytical-reagent grade nitric acid (Merck) to 1 1. Stock standard solutions of As3+, Sb3+ and Bis+, 1000 mg 1-1. Baker atomic spectral standards.Working standard solutions were prepared fresh daily by dilution with doubly distilled water. Stock standard solution of Se4+. Prepared by diluting Titrisol solution (Merck) containing 1000 g of selenium (as SeOz) to 1 1 with doubly distilled water. Aliquots were diluted with 1.5% mlV hydrochloric acid to obtain appropriate working standard solutions. Stock standard solution of Sn4+. Prepared by diluting a Titrisol solution (Merck) containing 1000 g of tin (SnC14) to 1 1 with 10% mlV hydrochloric acid. A 50-pl volume of this solution was further diluted to 50 ml, and 1 ml further to 5 ml, both with 10% mlV hydrochloric acid, to obtain a 0.2 pg ml-l working standard solution. Carrier Gases High-purity argon. High-purity 99% argon + 1% oxygen (Cargal 1). High-purity nitrogen.These gases were used to purge the system and were obtained from L'Air Liquide, Afdeling Precigaz, Liege , Belgium. Results and Discussion Effect of Gas Transport Tubing Different kinds of plastic and glass transfer tubing were tested in order to study their influence on the determination of 10 ng of As(II1) by using the peak area as a measuring mode. In the first part of the experiments, nitrogen was the carrier gas for the hydrides. Silicone-rubber and nylon tubing gave identical results. The reproducibility was very low and the blank values were unstable. This is the main reason why other workers (see Table 4) have found it difficult to use peak areas, especially at analyte levels of 10 ng. The use of glass tubing did improve the sensitivity and reproducibility but the base line remained unstable.A repeated number of blank determinations returned the absorbance values to the zero level. This indicates that a portion of the hydride was not atomised the first time but was absorbed on the surface of the reaction vessel or the gas transfer tubing, or both. This portion of the hydride was atomised when the repeated number of blank determinations was carried out. This is possibly the main reason for the low reproducibility of the standards. Subsequently, the glass tubing was silanised in the following manner. It was thoroughly cleaned, dried and filled with a 5% mlV dimethyldichlorosilane solution in toluene. The organic solvent was allowed to evaporate at room temperature in a fume-cupboard for 2 h. Finally, the tube was emptied, heated for 1 h in an oven at 110 "C and a nitrogen stream was blown through it for 5 min.As is known, dimethyldichlorosilane reacts with the surface hydroxy groups and deactivates the283 ANALYST, MARCH 1986, VOL. 111 Table 2. Precision of replicate analyses of 10 ng of As(II1) Peak area/A s Gas transfer Carrier tubing gas Silicone-rubber* N2 Glass N2 FEP N2 Silanised glass N2 . . . . . . . . . . . . . . . . . . . . Silicone-rubber . . . . Ar FEP . . . . . . . . Ar Silicone-rubber . . . . 99% V/V Ar - 1% v / v o * Glass . . . . . . . . 99% V/V 1% v / v o 2 Silanisedglass . . . . 99% V/V 1% v/v 0 2 FEP . . . . . . . . 99% V/V 1% v / v o 2 Ar - Ar - Ar - * Silicon-rubber and nylon tubing gave similar results. t After 200-220 determinations.3: After 25-30 determinations. Peak height, A Mean F standard (mean k standard deviation/ deviation) A s 0.062 f 0.004 0.265 k 0.029 0.323 4 0.023 0.382 4 0.013 0.385 F 0.013 0.414 4 0.019 0.479 k 0.011 0.063 k 0.003 0.688 k 0.021-t 0.710 k 0.017 0.082 !c 0.002 0.726 k 0.011$ 0.728 !c 0.009$ No. of analyses 10 10 10 10 10 10 10 10 10 10 Relative standard deviation, YO 10.9 7.1 3.4 3.4 4.6 2.3 3.0 2.4 1.5 1.2 glass surface, and this treatment further improved the sensitivity and reproducibility. When not used, the silanised glass tubing was cleaned with 3 x 3 ml of toluene and 3 x 3 ml of methanol, dried with a stream of nitrogen for 5 min and kept in a desiccator; it is well known that moisture destroys the silanised glass surface. After the tubing had been used for more than 500 determinations within a period of 3 weeks, the silanisation process had to be repeated.When argon was used as the carrier gas, all results were higher than those obtained with nitrogen. This difference could be caused by the lower heat capacity of argon compared with nitrogen, resulting in a higher atomisation temperature when using argon. This is supported by the fact that when nitrogen was used, the cell temperature indicator on the front panel of the controller was frequently switched off during the determination, indicating that the feedback circuit was not able to maintain the temperature at the pre-selected value. This was not so when argon or 99% V/V argon - 1% V/V oxygen was used. The results are shown in Table 2. Effect of 99% V/V Argon - 1% V/V Oxygen as a Purge Gas The use of nylon or silicone-rubber tubing together with 99% V/V argon - 1% V/V oxygen as a purge gas slowly and stably increased the sensitivity, which reached its maximum level after about 200 determinations.The reproducibility was low and memory effects made the method time consuming and impractical. Cleaning the nylon and silicone-rubber tubing with methanol and drying with nitrogen destroyed the previously increased sensitivity. As indicated by Reamer et al. 5 with radiotracer techniques for hydride generators, plastic materials initially retain or decompose a considerable portion of the generated hydride, but as subsequent reactions are performed, the amount retained or decomposed decreases and larger portions can enter the atomiser, resulting in an increase in sensitivity.It is possible that the available absorption sites on the walls of the 40 crn long plastic tubing are being filled with the metal or the hydride and therefore deactivating the surface towards further hydride absorption. On washing the tubing with methanol it is possible that the surface reverted to its original high absorption characteristics, resulting in a decrease in the signal for the same amount of generated hydride. With glass tubing, better results were obtained, but the memory effects still remained a serious problem. Silanised glass and FEP tubing gave the most satisfactory results, with a stable base line even if a standard as high as 50 ng of As(II1) was determined. A series of five standards of 10 ng of As(II1) also did not leave any memory effect.The probability of free atom formation from the hydride is proportional to the number of collisions with free radicals. This indicates that the atomisation efficiency increases as the number of radicals increases. This is the reason why, when a quartz cell is used that has been cleaned with hydrofluoric acid for 15 min, the peak-area values of a repeatedly determined standard slowly increase and after about 30 determinations the sensitivity reaches its maximum value. The surface of the quartz cell has an important effect on sensitivity. It probably catalyses the formation of radicals and subsequently the atomisation of hydrides. The quartz cell can be stored in a desiccator and re-used for another day without any decrease in sensitivity.The replacement of 99% V/V argon - 1% V/V oxygen with nitrogen does not destroy the sensitivity immediately. The peak area of the first standard has the same value as when 99% V/V argon - 1% V/V oxygen is used; that for the second standard will be smaller, the third even smaller, and so on. The population of the radicals in the cell was high enough to atomise the first standard completely. However, with the use of nitrogen, the consumption of radicals is greater than its production, with the result that the second standard will not be completely atomised and will give smaller values. As Welz and Melcher3 suggested, when nitrogen or argon is used, the gas bubbles through the sample solution during the purge time and drives the dissolved air out of the solution.If the purge time is less than 60 s, dissolved oxygen still remains in the sample, and this, together with the hydrogen generated from sodium tetrahydroborate(II1) during the reaction time, pro- duces a limited number of radicals in the heated quartz cell. Results obtained with the use of 99% V/V argon - 1% V/V oxygen are given in Table 2. The sensitivities attained for the different hydride-forming elements using 99% V/V argon - 1% V/V oxygen as a purge284 ANALYST, MARCH 1986, VOL. 111 Table 3. Precision of replicate analyses using silanised glass tubing and 99% v/v argon - 1% v/V oxygen Peak area Mean k standard Element deviation/ (10 ns) A s As(1II) . . 0.726+0.011 Se(1V) . . . . 0.469 k 0.007 Sb(II1) . . . . 0.420+0.010 Bi(V) .. . . 0.442+0.008 Sn(1V) . . . . 0.599+0.015 Relative No. of standard analyses deviation, % 10 1.5 10 1.5 10 2.4 10 1.8 10 2.5 Table 4. Calculated sensitivities obtained by different recent methods Sensitivity Element (long) Absorbance Mode Se(1V) . . . . 0.100 Peak area As(II1) . . . . 0.070 Peak height As(II1) . . . . 0.080 Peakheight Sb(II1) . . . . 0.036 Peakheight Bi(II1) . . . , 0.028 Peak height Se(1V) . . . . 0.025 Peak height Sn(1V) . . . . 0.030 Peak height As(II1) . . . . 0.100 Peak height Sb(II1) . . . . 0.001 Peak height Sb(II1) . . . . 0.017 Peakheight Sb(II1) . . . . 0.045 Peak height Bi* . . . . . . 0.015 Peak height Se(1V) . . . . 0.025 Peak height * Not specified. Reference 7 3 8 9 9 9 9 10 11 12 13 14 15 gas, the operating parameters in Table 1 and silanised glass tubing are listed in Table 3.A comparison with recently reported values in the literature is shown in Table 4. Effect of Reaction Flask Material Reamer et aZ.5 found that glass and polypropylene reaction vessels exhibit the greatest absorption of selenium and silanised glass the least. In our experiments with the same materials, statistically no difference was observed in the peak-area values for 10 ng of selenium. The possible reason is that in this system a 100-fold smaller standard of selenium and a much smaller reaction vessel are used. In addition, the sodium tetrahydroborate(II1) solution flows through a capil- lary into the lowest part of the V-shaped reaction flask, which achieves more complete hydride generation. Conclusions Oxygen has an effect on the determination of volatile hydride-forming elements, not only at low quartz cell tem- peratures, as indicated by Welz and Melcher,3 but also at temperatures above 800 "C, possibly by accelerating the production of radicals that may take part in the atomisation mechanism of the hydrides. The use of silanised glass or FEP tubing aids the transportation of gases and the maximum sensitivity can be attained after a few determinations.This method permits the use of peak areas as a measuring mode for the routine determination of hydride-forming elements. One of the advantages of peak-area over peak-height values is the smaller dependence on or independence of fluctuations of parameters such as the valence state of the analyte, reaction speed and time, sodium tetrahydroborate(II1) concentration, acid concentration, temperature changes of the quartz cell and gas flow-rates. 1. 2. 3. 4. 5 . 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. References Dedina, J., and Rubeska, I., Spectrochim. Acta, Part B, 1980, 35, 119. Welz, B., and Melcher, M., Anal. Chim. Acta, 1981, 131, 17. Welz, B., and Melcher, M., Analyst, 1983, 108, 213. Moody, J. R., and Lindstrom, R. M., Anal. Chem., 1977, 49, 14, 2264. Reamer, D. C., Veillon, C., and Tokousbalides, P. T., Anal. Chem., 1981, 53, 245. Dong So0 Lee, Anal. Chem., 1982, 54, 1682. Lloyd, B., Holt, P., and Delves, H. T., Analyst, 1982,107,927. Welz, B., and Melcher, M., Analyst, 1984, 109, 573. Welz, B., and Melcher, M., Spectrochim. Acta, Part B, 1981, 36, 5 , 439. Siemer, D., Koteel, P., and Jariwala, V., Anal. Chem., 1976, 48, 836. AznArez, J., Palacios, F., Ortega, M. S., and Vidal, J. C., Analyst, 1984, 109, 123. Chapman, J . F., and Dale, L. S . , Anal. Chim. Acta, 1979,111, 137. De Doncker, K., Dumarey, R., Dams, R., and Hoste, J., Anal. Chim. Acta, 1983, 153, 33. Terashima, S . , Anal. Chim. Acta, 1984, 156, 301. Verlinden, M., Baart, J., and Deeistra, H., Talanta, 1980, 27, 633. Paper A51232 Received June 27th, 1985 Accepted September 20th, 1985