JOURNAL O F ANALYTICAL ATOMIC SPECTROMETRY NOVEMBER 1994 VOL. 9 1203 Role of Barium Chemical Modifier in the Determination of Fluoride by Laser-excited Molecular Fluorescence of Magnesium Fluoride in a Graphite Tube Furnace* Alexander I. Yuzefovsky and Robert G. Michelt Department of Chemistry University of Connecticut Storrs CT 06269-3060 USA Some improvement in the determination of fluorine in urine and tap water by use of laser-excited molecular fluorescence spectrometry of magnesium monofluoride in a graphite tube furnace has previously been reported. The use of barium as a chemical modifier increased the size of the signal by a factor of 100. The work reported in the present paper was carried out in an attempt to elucidate the mechanism of the enhancement of the magnesium fluoride fluorescence by barium and to explain some other experimental characteristics of the method such as the vaporization temperature which was lower at 1800 "C than the 2400-2700 "C reported by other workers.The mechanism of formation of gaseous magnesium fluoride molecules from sodium fluoride and magnesium nitrate solutions in a graphite tube furnace during atomic absorption measurements was investigated with and without the presence of barium. It was shown that without chemical modification the formation of magnesium fluoride in the gaseous phase proceeded mainly via interaction between magnesium difluoride molecules and excess of free magnesium atoms [Mg(g) + MgF2(g)+2MgF(g)]. The efficiency of this process was fairly low primarily because of the difference between the vaporization temperatures of the reacting species (1 400 "C for magnesium difluoride and 1800 "C for magnesium vaporized as magnesium oxide). The presence of barium changed the mechanism of formation of magnesium fluoride. It was calculated that the formation of barium difluoride rather than magnesium difluoride was thermodynamically preferable in the first step of the mechanism.Experimental data indicated that the formation of magnesium fluoride then proceeded with higher efficiency than without barium because the reaction Mg(g) + F(g)-+MgF(g) followed the appearance of magnesium from magnesium oxide and fluorine from barium difluoride at coincidental temperatures in the range 1700-1 900 "C. Keywords Flu0 rin e ; lase r-excited molecular flu0 rescen c e spec tro rn e try; m agn esiurn fluoride ; barium ; graphite furnace In a previous publication' significant improvements in the determination of fluorine in urine and tap water by use of laser-excited molecular fluorescence spectrometry (LEMOFS) of magnesium monofluoride in a graphite tube furnace were reported. Excess of magnesium was added to the samples in order to promote the formation of the magnesium fluoride The method was extraordinarily sensitive with a detection limit of 0.3 pg.This detection limit allowed the determination of low levels of fluorine in a urine standard reference material by use of simple aqueous calibration. Physico-chemical inter- ferences were removed by dilution of the sample which was permitted by the high sensitivity of the method. The sensitivity of the analysis was aided by the use of barium as a chemical modifier which increased the temporal peak area by a factor of 100.The work reported in the present paper was carried out in an attempt to elucidate the mechanism of the enhance- ment of the magnesium fluoride fluorescence by barium and to explain some other experimental characteristics of the method such as the vaporization temperature which was lower at 1800 "C than reported by other workers (2500-2700 "C)."' In order to develop the method described above Butcher et al.' used as the standard fluorine in the form of sodium fluoride (NaF) added magnesium in the form of magnesium nitrate [Mg(N03)2 20 pg as Mg] and barium in the form of barium nitrate [Ba(NO,) 1.65 pg as Ba].These aqueous solutions were injected into the graphite furnace dried at 200 "C charred at 800 "C and cooled down to 20 "C before the fluorescence signal of magnesium fluoride was measured upon vaporization at 1800 "C. It was proposed that after decompo- sition of sodium fluoride (NaF) and magnesium oxide (MgO) at high temperature in the gas phase free fluorine reacts with * Presented at the XIX Annual Meeting of the Federation of Analytical Chemistry and Spectroscopy Societies (FACSS) Philadelphia PA USA September 20-25 1992 as paper No. 289. t To whom correspondence should be addressed. magnesium atoms to produce stable diatomic magnesium fluoride (MgF) molecules. The excess of magnesium moves the equilibrium of the following reaction to the Mg(g) + MgF,(g)-+2MgF(g) (1) In ref.1 there were three results that were not satisfactorily explained. Firstly why was the optimum vaporization tempera- ture for magnesium monofluoride between 1700 and 1900 "C which was approximately 1000°C lower than reported by other in analogous experiments? Secondly why did the addition of barium nitrate as the chemical modifier increase the sensitivity of the magnesium fluoride by a factor of loo? Thirdly why did an excess of barium nitrate beyond the optimum that produced the 100 times enhancement mentioned above strongly depress the fluorescence signal of magnesium fluoride? In order to provide a starting point for an explanation of these phenomena Butcher et a/.' proposed a based on the predication of the ability of barium to produce gaseous barium carbides in a graphite furnace.This mechanism was based on the paper b; Styris7 that described the formation of gaseous magnesium from involatile magnesium oxide (boil- ing-point 3600 "C8) according to the reaction BaC2(g)+2MgO(l/s)-+2Mg(g)+Ba(g)+2C0 (2) This reaction in turn would produce the excess of mag- nesium in the gas phase that is a necessary for conversion of magnesium difluoride in to magnesium fluoride according to reaction (1). The explanation of Butcher et al.' was not satisfactory for various reasons. Firstly why does gaseous fluorine which is extremely reactive at high temperature not react with barium carbide? As a result of this type of reaction it could be expected that a significant amount of fluorine could be lost owing to the formation of barium difluoride according to the reaction BaC,(g)+2F(g)~BaFz(g)+2C(s) (3)1204 JOURNAL O F ANALYTICAL ATOMIC SPECTROMETRY NOVEMBER 1994 VOL.9 The Gibbs energy per mole of gaseous products (AG*T/y) can be compared for reactions (2) and (3) at 2000 "C which is fairly close to the optimum temperature reported by Butcher et al.' According to the Second Law of Thermodynamics the most favourable reaction is that with the smallest positive or largest negative change in free energy per mole of reactant. For reaction (3) AG*2000~C/~ = - 857.72 kJ which is much more negative than for reaction (2) where AG*2000~C/~ = -92.048 kJ.' This means that reaction (3) has a higher prob- ability of occurrence than reaction (2). Thermodynamic infor- mation to describe the behaviour of barium carbide in the gaseous state could not be found so the calculations used here were based on data for barium carbide in the liquid/solid states.However this is unlikely to result in a different con- clusion because the difference between AG*=/v for reactions (2) and (3) would most likely have been larger rather than smaller for gaseous barium carbide as eqn. (3) represents a homogeneous reaction and (2) is a heterogeneous reaction. In addition for the case without the addition of any chemical modifier a magnesium monofluoride signal was observed even though it is 100 times smaller than in the presence of barium and this requires more explanation than that given by Butcher et al.' Remy" stated that the most stable compounds of the alkaline earth elements are the dihalides.For example at 1700 "C magnesium oxide in the presence of fluorine produces magnesium difluoride according to reaction (4) which has a AG*,~OOOC/V = - 246.856 kJ (4) The occurrence of the magnesium monofluoride is exper- imentally clear,' which indicates that an explanation must be available for the generation of excess of magnesium at the optimized temperature ( 1 700-1900 "C) to allow reaction (1) to proceed. Therefore the work reported in the present paper sought to understand the lower than expected vaporization temperature the increase in sensitivity caused by barium nitrate and the depression of the magnesium monofluoride signal by excess of barium nitrate. Why is fluorine not lost as barium fluoride? Finally without any modifier what processes govern the formation of magnesium monofluoride? Note that for the chemical reactions discussed here only major products are shown and the thermodynamic calcu- lations were performed based only on these products.With eqn. (4) as an example the oxygen from the right-hand side of the reaction after transfer from the reaction zone into the gaseous phase of the furnace could react further with gaseous carbon to produce some CO or CO but these 'secondary reactions' do not significantly affect the mechanisms that are proposed here. Accordingly such reactions are not considered in detail in the remainder of the present paper. Chemical modifiers such as barium nitrate have been widely employed in atomic absorption spectrometry (AAS) in the context of stabilized temperature platform furnace (STPF) technology6?" and are also used in laser-excited atomic fluor- escence spectrometry (LEAFS) and LEMOFS.Many workers have attempted to systemize the practice of the use of chemical modifiers12-14 but such reviews have demonstrated that the physico-chemical reactions involved during the vaporization cycle have often been very complicated contradictory and poorly understood. A better understanding of the chemistry of the reactions in the graphite furnaces should facilitate the appropriate selection of chemical modifiers for particu- lar analyses. The work described in the present paper demon- strates one instance where simple spectrometric measurements with an atomic absorption instrument together with thermo- dynamic calculations can make a contribution to this endeavour.Experimental A11 measurements were made by AAS with a Perkin-Elmer Model 5000 spectrometer equipped with Zeeman-effect back- ground correction an HGA-500 graphite furnace and an AS-40 autosampler. Standard hollow cathode lamps ( HCLs) were used as light sources. Argon which contained less than 1 x oxygen was the sheath gas for the atomizer. Spectrometric measurements which were made during the atomization step were taken under gas-stop or gas-flow (300 ml min- ') conditions. Standard experimental conditions such as wavelength lamp currents and slit-widths for spectro- metric measurements were chosen according to S l a ~ i n . ~ In the experiments with magnesium the samples were vapor- ized from standard platforms made from anisotropic graphite or from laboratory-made tantalum platforms which were used to exclude the possibility of significant interactions between oxide and carbon during the atomization step.15 Tantalum platforms were made from 0.05 mm thick tantalum foil and were 8 x 3 mm in size with rims 1 mm high.16 The furnace heating programme was chosen according to Slaviq6 except for the atomization temperature which was varied between 1700 and 2500 "C.In experiments on the thermal decomposition of magnesium difluoride MgF and sodium fluoride NaF solid samples were used in some cases (salts were of 99.99% purity from Aldrich Chemicals Milwaukee WT USA). Particles of the fluorides with total mass between 100 and 300 pg were placed directly on a standard graphite platform by use of tweezers.Then the platform on which the particles were set was placed into the graphite furnace and the sample was heated slowly (7 "C s-') from 1000 "C while the absorption signals were recorded. In experiments on the decomposition of barium oxide a solution of Ba(NO,) was used. The heating programme consisted of a drying step at 150"C a pyrolysis step at < 1200 "C and atomization with a slow heating rate of 7 "C s-' while the absorption signals were recorded. All experiments were repeated at least 3-5 times to achieve good reproducibility. A slow heating rate was used where appropriate to compensate for the thermal lag in the tempera- ture of the platform relative to the graphite tube. This meant that the platform would be at the same temperature as the wall during the measurements of absorbance.At the normal analytical fast heating rates the platform temperature tends to lag behind the wall temperature by 200-300°C. Results and Discussion The experiments were separated into two parts. The first part was designed to investigate the mechanism of the formation of gaseous molecules of magnesium monofluoride from sodium fluoride and magnesium nitrate without any barium modifier. The second part was designed to investigate the mechanism of the formation of gaseous magnesium monofluoride from the same components but with the presence of a barium salt modifier. Formation of Magnesium Monofluoride From Sodium Fluoride and Magnesium Nitrate Without Barium Modifier The slow heating rate vaporization curve of sodium as it evolved from the decomposition of solid sodium fluoride salt in the graphite tube is shown in Fig.1. Sodium atoms appeared in the gaseous phase at a temperature at least 500°C lower at about 1250 "C than the literature' boiling-point of sodium fluoride (b.p. 1695 "C m.p. 993 "C) which could be a result of the involvement of carbon from the graphite surface in the reduction of sodium fluoride. It was assumed that decompo- sition of sodium fluoride on a graphite platform proceeds by the following reactions NaF(s/l)-+Na(g) + F(g) or (5)JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY NOVEMBER 1994 VOL. 9 1205 I I I 1000 1400 1800 2200 TemperaturePC Fig. 1 Vaporization curve of an individual NaF solid particle (x 100 pg) from a graphite platform The reactive fluorine atoms in the form of F(g) and/or F,(g) almost certainly interact with solid and/or liquid mag- nesium oxide formed on the platform by the decomposition of magnesium nitrate to produce stable magnesium difluoride according to reaction (4).The following experiments supported this thesis. The vaporization of magnesium from magnesium difluoride salt by measurement of magnesium atomic absorption signal is shown in Fig. 2(a) and the vaporization of the same metal from a mixture of magnesium oxide and sodium fluoride salt in Fig. 2(b). In the latter case magnesium and sodium fluoride were introduced onto the platform in the form of a magnesium 0 0.8 + Q 0.4 f! v) 2 0 0.8 0.4 0 1000 1400 1800 2200 TemperaturePC Fig. 2 Vaporization curves of Mg from a graphite platform (a) indivi- dual MgF particle about 300pg in mass; (b) mixture of individual particles of NaF ( ~ 1 0 0 p g ) and 5 pl of lOOOpgml-' Mg(NO,) solution; and (c) mixture of individual particle of NaF (x 100 pg) and 20 pl of 1000 pg ml-' Mg(N03)2 solution nitrate solution and a solid salt respectively.A comparison of Fig. 2(u) and (b) indicates that the appearance temperature of the thermal dissociation is about the same at about 1400"C in both cases. Thus it appears that both curves are the result of decomposition of the same molecule magnesium difluoride which is only possible for the mixture of magnesium oxide and sodium fluoride if reaction ( 6 ) occurs first M€Fz(s/l)-+MgF(g) + F(g) MgF,(s/U + MgF,(g) -+MgF(g) + F(g) or ( 6 ) When the concentration of magnesium oxide was signifi- cantly increased and the concentration of sodium fluoride was kept the same two peaks were observed in the magnesium thermal dissociation curve [Fig.2(c)]. The first one was prob- ably due to the decomposition of magnesium difluoride. The second was possibly due to the decomposition of magnesium oxide. Note that although it does not have a significant effect on the present arguments the small delay in the vaporization temperature of magnesium difluoride in Fig. 2(b) and (c) compared with (a) could be explained in terms of the formation of a carbon film on the sample particles owing to the presence of an excess of metal o ~ i d e . ' ~ ' ~ Butcher et ul.' observed only a fairly small magnesium fluoride fluorescence signal without the presence of the barium modifier.According to the thesis of eqn. ( l ) it is necessary to have an excess of free magnesium atoms in the gas phase in order to produce a significant amount of gaseous magnesium fluoride from magnesium difluoride. In Fig. 2(c) it is indicated that in the presence of excess of magnesium only a small fraction of the magnesium vaporizes as magnesium difluoride while most of it is vaporized as magnesium oxide beginning at a temperature that is more than 250°C higher. Optimization of the atomization temperature given by Butcher et al.' indicated that in the presence of barium the sensitivity at 1700-1900 "C is better than at all higher tempera- tures. This is logical given the observations from Fig. 2(c) that there is not a sufficient excess of magnesium for reaction (1) to proceed until temperatures above about 1700 "C which is after the magnesium difluoride has already vaporized and probably mostly diffused out of the furnace.At higher tempera- tures between 2000 and 2700 "C the sensitivity drops steadily.' An increase in the rate of diffusion of both magnesium mono- fluoride and magnesium difluoride from the furnace would result in less signal from magnesium monofluoride and more importantly less time for the excess of magnesium to inter- act with magnesium difluoride to produce magnesium mono fluoride. Formation of Magnesium Monofluoride From Sodium Fluoride and Magnesium Nitrate with Barium Modifier The addition of barium nitrate as a chemical modifier enhanced the magnesium monofluoride fluorescence signal by a factor of 100.Now two concurrent reactions (4) and (7) can occur during the decomposition of sodium fluoride reaction (9 as a result of the presence of barium oxide 2BaO(s/l) + 2F2(g)+2BaF2(s/l)+ O,(g) (7) In order to decide which reaction is preferable the AG*T/v for both reactions was calculated. It was assumed that these reactions must occur in a range of temperatures up to a maximum of 1200 "C which was the highest possible tempera- ture of the complete decomposition of sodium fluoride from the graphite platform (Fig. 1). An inspection of the results of calculations of the Gibbs energies for reactions (4) and (7) through the temperature interval between 500 and 1200 "C always showed a preference for reaction (7) over (4).For example at 1200 "C the difference between the AG*T/v values of the two reactions was 83.68 kJ mol-' which favours reac-1206 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY NOVEMBER 1994 VOL. 9 0.8 m m c (u 0 (D + g 0.4 z 0 tion (7) and this difference increased with decrease in tempera- ture BaF2(s/l)+Ba(g)+2F(g)+ [BaF(g)?] or BaF,(s/l)+BaF,(g)-+Ba(g)+ W g ) + CBaF(g)?l (8) For both reactions (4) and (7) together with reaction (8) and without barium modifier reaction (6) free fluorine appears in the gas phase after the thermal dissociation of the corresponding fluorides but the temperature of vaporization of barium fluoride is higher than the other fluorides and approaches the vaporization temperature of magnesium oxide [Fig. 2(c)]. This statement can be verified by the following results in Fig.3(u) is shown the slow heating rate vaporization curve of barium as it evolved from decomposition of barium fluoride. Barium and sodium fluoride were introduced onto the platform in the form of a solution of a metal nitrate and a particle of a solid salt respectively. The fluoride appeared on the platform according to reaction (7) between barium oxide produced during pyrolysis of barium nitrate salt and fluor- ine produced during decomposition of sodium fluoride. In addition the results in Fig. 4(u) were obtained by atomization of the same mass of magnesium from a graphite platform under normal analytical conditions and the variation in absorption signal with atomization temperature is shown in . 1200 1600 2000 TemperaturePC Fig.3 Vaporization curves of Ba from a graphite platform A mixture of 10 pl of 1000 pg ml-' Ba(NO& solution and individual particles of NaF (z 100 pg); and B 20 p1 of 1000 pg ml-' Ba(NO,) solution 1600 1800 2000 2200 2400 TemperaturePC Fig. 4 Relationship between absorption signal and atomization tem- perature for 15 pg of Mg from Mg(NO,) solution A from a graphite platform; and B from a tantalum platform the figure. It can be seen that the two processes of vaporization of magnesium from magnesium oxide [Fig. 4(u)] and fluorine from barium difluoride [Fig. 3(a)] occur at the same tempera- ture. In this instance the formation of magnesium monofluor- ide molecules occurs in the gas phase [reaction (9)] as a result of a collision of the two free atoms Mg(g) +F(g)+MgF(g) (9) For simplicity it can be assumed that each collision of the two different types of free atom produces one diatomic mol- ecule and that the collision frequency per unit volume depends directly only on the number density of both atoms.I9 The residence time of gas-phase atoms in a tube furnace is depen- dent upon temperature which affects the rate of diffusion from the tube.The significant decrease in the fluorescence signal of magnesium fluoride at temperatures higher than 1800 "C reported by Butcher et al.' was probably the result of the decreased residence time of atoms in the gas phase which activated a lesser number of collisions in a gas phase. As a significant excess of magnesium was used,' it seems reasonable to assume that the collision frequency depended on the concen- tration of free fluorine atoms in the gas phase.The following conclusions can be drawn. Without a barium modifier the relatively weak signal of magnesium fluoride was due to free atoms of magnesium and fluorine that appeared in the gas phase at different optimum temperatures. The concen- tration of magnesium atoms reached a maximum just as the concentration of fluorine atoms was significantly decreased [Fig. 2(c)]. The presence of a barium modifier significantly delayed the appearance of free fluorine atoms in the gas phase by pro- duction of relatively more thermostable fluorides. Magnesium and fluorine atoms vaporized at the same temperature which meant that each atom reached its highest concentration in the gas phase simultaneously.These conditions produced the highest concentration of gaseous magnesium fluoride molecules in the graphite tube. The existence of optimum concentrations of barium chemical modifier can be explained as follows according to the mechan- ism that was proposed above an increased amount of barium caused an excess of free barium atoms in the gas phase after the decomposition of barium oxide. The slow heating rate vaporization curve of barium as it evolved from the decompo- sition of barium nitrate is shown in Fig. 3(b). This appearance temperature for barium atoms coincides approximately with the optimum temperature for the formation and detection of magnesium fluoride. It follows that when the number of free barium atoms in the gas phase increased beyond the optimum the number of collisions between barium and fluorine atoms became significant.These interactions can happen directly via collisions [eqn. (10u)l or through interaction of barium and magnesium fluoride [eqn. (ll)] Ba(g) + F(g)+BaF(g) (10) (11) Both reactions dramatically decrease the number of mag- nesium monofluoride molecules and with further increased barium concentration would depress the fluorescence signal of magnesium monofluoride completely as reported in ref. 1. As a last comment for discussion an attempt was made to explain the decomposition of magnesium oxide in a graphite tube at the rather low temperature of 1800 "C [Fig. 2(c)]. There has been controversy concerning the theories of the formation of free metal species in the gaseous phase of the graphite furnace.The most comprehensive theory is the reduction of oxides by carbon (ROC) the mechanism of which was first proposed by L'vov and Savin2' and has been exten- sively investigated over the last ten years in different labora- torie~."-~~ Some contradictions in the details of the ROC theory have r e ~ e n t l y ~ ' ~ ~ been pointed out. The ROC theoretical mechanism is based on two concurrent Ba(g) + MgF(g)-+BaF(g) + Mg(g)JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY NOVEMBER 1994 VOL. 9 1207 autocatalytic reactions the first reaction occurs on the surface of the graphite tube and the second reaction occurs on the surface of the oxide The appearance temperature of metal atoms in the gas phase produced via the ROC mechanism has been shown to be significantly lower than the appearance temperature of atoms produced by thermal decomposition.'1~'4 In other words if all the assumptions hold true the ROC mechanism is always preferable relative to thermal decomposition. The difference in the analytical signal and atomization temperature of mag- nesium due to the two possible mechanisms is demonstrated in Fig.4. The same mass of sample was vaporized from two different surfaces graphite (a) and tantalum (b) and the results were compared. When the graphite platform was used the ROC mechanism was assumed to be the predominant mechan- ism for the formation of gaseous metal atoms. A tantalum platform was used to prevent interactions between carbon and oxide so that the decomposition of the metal oxide developed via thermal decomposition.Consequently the results in Fig. 4 show that the maximum absorbance signal for magnesium produced by thermal decomposition occurred at a temperature about 500 "C higher than for magnesium produced according to the ROC mechanism. Despite the above discussion there remains some doubt about why the optimum vaporization temperature in the present work was lower at 1800°C than the 2500-2700°C reported by other workers with' and without3 barium nitrate. Unfortunately previous workers did not provide the data that were obtained to optimize the vaporization temperature. Hence it was difficult to discern whether or not the experimen- tal conditions had been properly optimized in their work. At the same time the reported sensitivities for their methods were lower by more than one order of magnitude compared with the work by Butcher et al.' The lack of data and the contradictory results between researchers make it difficult to rationalize why these other researchers used such high vaporiz- ation temperatures.Throughout the present paper it was tacitly implied that all the information derived from atomic absorption measurements are directly transferable to molecular fluorescence measure- ments. This is not true per se since molecular fluorescence could also be affected by a change in collisional environment but the present workers could find no reason to disbelieve this implication in the context of the data presented above. This work was supported by an American Chemical Society Division of Analytical Chemistry Fellowship sponsored by Perkin-Elmer (awarded to A.I.Y.).1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 References Butcher D. J. Irwin R. L. Takahashi J. and Michel R. G. J. Anal. At. Spectrom. 1991 6 9. Dittrich K. Hanisch B. and Stark H. J. Fresenius'Z. Anal. Chem. 1986,324,497. Anwar J. Anzano J. M. Petrucci G. and Winefordner J. D. Microchem. J. 1991 43 77. Searcy A. W. in Progress in Inorganic Chemistry ed. Cotton F. A. Wiley Interscience New York 1963 vol. 3. Frech W. Lundberg E. and Cedergren A. Prog. Anal. At. Spectrosc. 1985 8 257. Slavin W. Graphite Furnace AAS A Source Book Perkin-Elmer Norwalk CT 1984. Styris D. L. Anal. Chem. 1984 56 1070. Handbook of Chemistry and Physics ed. Weast R. C. CRC Press Cleveland OH 62nd edn. 1982.Barin I. Knacke O. and Kubaschewski O. Thermochemical Properties of Inorganic Substances Springer-Verlag New York 1977. Remy H. Anorganische Chemie Akademische Verlagsgesellschaft Geest & Portig K.G. Leipzig 1972. Slavin W. Manning D. C. and Carnrick G. R. At. Spectrosc. 1981 2 137. Tsalev D. L. and Slaveykova V. I. J . Anal. At. Spectrom. 1992 7 147. Tsalev D. L. Slaveykova V. I. and Mandjukov P. V. Spectrochim. Acta Rev. 1990 13 225. Tsalev D. L. Bibliography Chemical Modification in Electrothermal Atomization Atomic Absorption Spectrometry 1973-1989 RP 143 Bodenseewerk Perkin-Elmer Uberlingen 1991. L'vov B. V. Spectrochim. Acta Part B 1989 44 1257. L'vov B. V. Nikolaev V. G. Novichikhin A. V. and Polzik L. K. Spectrochim. Acta Part B 1988 43 1141. L'vov B. V. Dokl. Akad. Nuuk SSSR 1985 283 1415. Welz B. Curtius A. J. Schlemmer G. Ortner H. M. and Birzer W. Spectrochim. Acta Part B 1986 41 1175. Atkins P. W. Physical Chemistry W. €3. Freeman San Francisco 1982. L'vov B. V. and Savin A. S. Zh. Anal. Khim. 1982 37 2116. L'vov B. V. Polzik L. K. Romanova N. P. and Yuzefovsky A. I. J. Anal. At. Spectrom. 1990 5 163. Gilmutdinov A. K. Zacharov Y. A. and Ivanov V. P. Zavod. Lab. 1989 55 31. Bendicho C. and de Loos-Vollebregt M. T. C. Spectrochim. Acta Part B 1990 45 547. L'vov B. V. Spectrochim. Acta Part B 1989 44 1257. Holcombe J. A. Styris D. L. and Harris J. D. Spectrochim. Acta Part B 1991 46 629. Round Table Discussion XXVII-CSI Pre-Symposium J . Anal. At. Spectrom. 1992 7 471. Paper 31075691 Received December 24 1993 Accepted June 30 1994