Analyst, March, 1968, Vol. 93, @. 148-152 Investigations on Long-path Absorption Tubes in Atomic-absorption Spectroscopy BY IVAN RUBESKA AND BEDRICH MOLDAN (Geological Survey of Czechoslovakia, Prague 7 , KostelnZ 26, Czechoslovakia) Although it is known that the use of long tubes in atomic absorption brings considerable increase in sensitivity, little is known about various factors affecting the sensitivity. In this work the r61e of fuel flow-rate for several elements (copper, silver, gold, cadmium, tin, lead, antimony, bismuth, manganese and rhodium) is studied and the results are discussed. A heated, 45-cm long, alumina tube is used, the burner being a Beckman type fed by air and hydrogen. THE use of absorption tubes in atomic-absorption spectroscopy has two major advantages, the first being the considerable increase in the sensitivity of determination for several elements, and the second the improvement in the geometric stability of the flame gases in relation to the optical axis.This is particularly important when measuring lines near 2000 A where the flame gases absorb.1 It may, therefore, be expected that the flame-in-tube technique will be more widely used. To make full use of the possibilities offered by absorption tubes, investigations were carried through to find what factors affect the sensitivity of the elements in absorption tubes. EXPERIMENTAL APPARATUS- The experimental arrangement has been described previously.2 It consists of a 45-cm long, ceramic tube heated in an electric furnace and placed in the optical axis.Tubes of 1 7 , l l and 9-mm i.d. are used. At one end of the tube an air - hydrogen flame from a Beckman burner is introduced. Light from the source is focused at the entrance of the tube and is collected at the exit by a spherocylindrical lens and focused on the monochromator slit. The molecular spectra are scanned with a hydrogen lamp and a Zeiss GlBl recorder. Because a simple d.c. measuring device was used our investigations were confined to the wavelength region below 3300A. RESULTS SENSITIVITY AND FUEL FLOW- It has already been observed that the sensitivity of elements in absorption tubes depends markedly on the fuel flow-rate.3 For all the elements investigated, i.e., copper, silver, gold, cadmium, tin, lead, antimony, bismuth, manganese and rhodium, a relatively steep absorbance increase was observed at a definite value of the hydrogen-to-air flow ratio.For most of the elements, namely copper, silver, lead, antimony and bismuth, the dependences had the form reproduced in Fig. 1. It was found that for a fuel-lean flame the small absorbance value suddenly increased, after which it remained constant. The critical value at which this change takes place depends, primarily, on the tube diameter. The greater the diameter of the tube the more ambient air is entrained, and the higher the necessary hydrogen flow. To a lesser extent it also depends on the burner position. The more the burner is tilted the more air is entrained, and the higher the required hydrogen flow. The sensitivity obtained is generally higher the smaller the inclination of the burner from the vertical.Evidently less air is entrained and the flame gases are less diluted. How- ever, to secure a laminar flow through the tube the angle between the burner and the optical axis cannot exceed a definite value. This was about 60" for the 17-mm tube, 45" for the 11-mm tube and 25" for the 9-mm tube. With this inclination the critical hydrogen-to-air flow ratios were 2.8, 1.9 and 1.5, respectively. 0 SAC and the authors.RUBESKA AND MOLDAN A 0.5 - p 0.4 - 8 0, f 0.3 - -E 0 D f C I I l l l l 149 0.4 $ 0.3 -2 0 2 0.2 0. I For gold and rhodium the absorbance increase was followed by an absorbance fall again, as seen in Fig. 2. For manganese the sudden sensitivity increase was followed by a further slight increase up to a maximum value, shown in Fig.3, and for tin the sensitivity increase was slow and shifted considerably towards higher hydrogen flow values, as shown in Fig. 4. Whereas for all other elements the curves for different tube diameters were simply shifted on the fuel-flow scale, with only minor effects on the maximum sensitivities, for tin a much higher sensitivity was found for small diameter tubes. Flow-rate, litres per minute Fig. 3. Absorbance of 0.1 p.p.m. of manganese: curve A for corundum tube 9-mm diameter; curve B, for ll-mm diameter; curve C, for pyrolan Fig. 4. Absorbance of tin. Same tube 17-mm diameter Flow-rate, litres per minute conditions as in Fig. 3 DISCUSSION OF THE RESULTS The relatively high increase in absorbance at a critical hydrogen-to-air ratio suggests that a basic change in the experimental conditions takes place.Most probably the mixture of hydrogen and total air, i.e., including the entrained air, becomes fuel-rich. By shielding the flame gases from the surrounding atmosphere a split flame, with the primary reaction zone at the entrance end and the secondary reaction zone at the exit end of the tube, results. The interconal gases then fill the whole tube length. Because the flame is split along a horizontal axis a considerable sensitivity increase is possible.150 RUBESKA AND MOLDAN : INVESTIGATIONS ON LONG-PATH [Analyst, Vol. 93 To confirm the existence of hydroxyl radicals along the whole tube length the absorption of the hydroxyl band head at 3089 A was measured for different hydrogen flows (see Fig.5). The dependence is similar to that of the previously mentioned elements, i.e., with an increase 456789 Flow-rate. litres per minute Fig. 5. Absorbance of OH radi- cals with increasing hydrogen flow. Pyrolan tube 17-mm diameter. Band head 3098 A measured at the critical values. The decrease of the hydroxyl concentration when increasing the hydrogen flow beyond the critical value is evidently caused by the formation of hydrogen radicals according to the reaction- H, + OH- + H,O + H+, which proceeds immediatel~.~ The maximum absorbance value was about twenty times higher than that found for a hydrogen flame burning freely in air, i.e., without the tube and combusted solely with entrained air. Because the concentration of hydroxyl radicals cannot vary considerably the high absorbance is evidently caused by the longer absorption path length with the absorption tube. The energy available in the tube comes mainly from the re-combination reactions- H+ + OH- + M + H,O + M+ AH = -5.05 eV H+ + M + H, + M+ AH = -4-5 eV.and H+ + In both reactions M represents a third body capable of absorbing the energy released by the re-combination.4 It may be a water molecule but any other molecule, for instance, the oxide of the metal being determined, has the same effect. Oxides of all the elements investi- gated, with the exception of tin, have a smaller dissociation energy than the energy released by the re-combination of the hydrogen and hydroxyl radicals (see Table I). Any energy transfer may, therefore, result in their dissociation.TABLE I COMPARISON OF ENHANCEMENT FACTOR, DISSOCIATION ENERGY OF THE OXIDES AND VAPOUR PRESSURE Silver . . Cadmium Gold . . Bismuth Rhodium Antimony Manganese Lead . . Copper Tin - . Absorbance (maximum) Absorbance with fuel-lean flame .. .. 1-4 .. .. 1.7 .. .. 2.2 . . .. 15 . . .. 34 .. .. 45 .. .. 60 . . .. 65 .. .. -200 . . .. -200 * t mb = 9.3 x = 6.03; PBle = 1'74. P8bs = 1.02; Dissociation energy, eV 1-4 3.8 4.0 3.8 4.0 4.1 4.9 5.7 - - Pressure, ton- T = 1200" K 1-23 x > 760 4.69 x 10-7 6.03* < 10-10 6.54t 3.61 x 4.47 x 10-1 8-15 x 1-95 x lowsMarch, 19681 ABSORPTION TUBES IN ATOMIC-ABSORPTION SPECTROSCOPY 151 The absorbance increase at a critical value of the hydrogen flow is in correlation with the dissociation energy of the particular oxide.In Table I the elements are arranged in order of the observed enhancement factor (ie., the ratio of the maximum absorbance to absorbance with a fuel-lean flame), with their dissociation energies and vapour tensions. This correlation holds, with the exception of antimony, which is, however, predominantly present in the vapour phase as a dimer and tetra me^-,^ and the concentration of free antimony atoms is thus lowered. The existence of antimony molecules (Sb,) was confirmed by the absorption spectrum,6 as shown in Fig. 6. I I 2200 23000 Wavelength, A Fig. 6. Absorption spectrum of antimony(I1). The solution sprayed contained 5000 p.p.m. of antimony, 1 g of tartaric acid, 6 ml of nitric acid per 100 ml (see NO bands in the spectrum). Lead was present as an impurity in antimony The slow absorbance increase of manganese following the sudden increase is probably caused by the formation of MnOH7 (dissociation energy = 3.15 eV), which dissociates readily as the OH concentration in post-flame gases decreases with increasing hydrogen flow.The absorption decrease observed for gold is probably caused by the formation of AuH (dissociation energy = 3.1 eV). For rhodium no values for the hydride are available. However, rhodium is the only element for which the saturation point of its vapour pressure is e~ceeded,~ so that the observed curve may be influenced by condensation of rhodium with decreasing temperature. Tin presents the most intriguing problem and at this point our conclusions are rather speculative. It has already been observed that the dissociation of tin(I1) oxide in a hydrogen - air flame is anomalously high and that a hydrogen - air flame provides a much higher con- centration of free tin atoms than the oxy-hydrogen flame.* The presence of nitrogen does not cause this difference because Gibson, Grossman and Cookeg found, by using a Beckman burner, that no great change in absorbance of tin occurred when nitrogen was replaced by argon.They suggest, therefore, that the higher concentration of tin atoms in a hydrogen - air flame with a slower burning velocity is caused by a longer stay of the tin(I1) oxide molecules in the reaction zone where reduction of tin oxide takes place. In absorption tubes this reducing zone may be extended by increasing the hydrogen flow. The hydrogen flow required for maximum sensitivity is, therefore, considerably higher than for the elements with smaller dissociation energies.The reduction of tin may probably involve a reaction with hydrogen molecules, because tin(1V) oxide is easily reduced by hydrogen, even at temperatures below 800" C. For elements other than tin no great difference in sensitivity is found when nitric acid or hydrochloric acid solutions are used, whereas for tin nitric acid has a slight depressive effect and hydrochloric acid a strong enhancing one. This is possibly because of the preferen- tial formation of tin(I1) chloride in the presence of hydrochloric acid and the reduction facilitated by the exothermal reaction- SnCl + H = Sn + HCl.152 RUBESKA AND MOLDAN The decrease of sensitivity with increasing tube diameter could then be attributed to a decrease of the partial pressure of hydrochloric acid in the flame gases because of a somewhat greater dilution by ambient air.CONCLUSIONS From these experiments it may be concluded that the best sensitivities are achieved when working under fuel-rich conditions, i e . , when the primary and secondary reaction zones are at the entrance and exit ends of the tube, respectively. The sensitivity enhancement for different elements, when changing over from a normal to the split flame, increases with increasing dissociation energy of the metal oxide. 1. 2. 3. 4. 5. 6. 8. 9. REFERENCES Rann, C. S., and Hambly, A. N., Analytica Chim. Acta, 1965, 32, 346. RubeSka, I., Ibid. 1968, 48, 187. RubeSka, I., and Stupar, J., Atomic Absorption Newsletter, 1966, 5, 69. Fenimore, C. P., “Chemistry in Premixed Flames,” Pergamon Press, Oxford and New York, 1964. Nesmeyanoff, A. N., Davlenye para khimicheskikh elementov, Izd. A. N. USSR, Moscow, 1961. Pearse, R. W. B., and Gaydon, A. G., “The Identification of Molecular Spectra,” Second Edition, Chapman & Hall Ltd., London, 1950. Cottrell, T. L., “The Strengths of Chemical Bonds,” Butterworths Scientific Publications, London; Academic Press Inc., New York, 1954. Gilbert, P. T., in Lippincott, E. K., and Margoshes, RI., Editors, “Proceedings of the Xth Collo- quium Spectroscopicum Internationale, Maryland, 1962,” Spartan Books, Washington, D.C., 1963, p. 171. Gibson, J. H., and Grossman, W. E. L., and Cooke, W. D., Analyt. Chem., 1963, 35, 266. Received September 28th, 1967