JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY DECEMBER 199 1 VOL. 6 66 1 Determination of Selenium by Electrothermal Atomic Absorption Spectrometry Part 1. Chemical Modifiers* Hana D&ekalova Veterinary Research Institute Hudco wa 70 CS-62 1 32 Brno Czechoslovakia Bohumil Ddekal Institute of Analytical Chemistry Czechoslovakia Academy of Sciences Vevefi 97 CS-611 42 Brno Czechoslo wakia Josef Komhrek and Ivan Novotny Department of Analytical Chemistry Faculty of Science Masaryk University Kotlakka 2 CS-611 37 Brno Czechoslovakia The effect of various chemical modifiers including nitrates of palladium nickel magnesium calcium lanthanum europium and aluminium on the analytical signal of selenium in a graphite furnace was studied. The signals of various selenium compounds such as selenite selenate and organic compounds representing different types of selenium forms in body fluids (selenomethionine and trimethylselenium iodide) were evaluated.The shape of the transient signal appears to be influenced not only by the chemical reactions in the graphite-selenium-modi- fier system but also very strongly by vaporization effects connected with the physical character of the charred residue. It follows that successful chemical modification involves the application of a considerable excess (higher than 1000-fold) of some metal nitrates which produce refractory oxides and no thermally stable carbides and are at the same time capable of quantitative conversion of the analyte into a single form. An integral part of the modifier action is trapping of the resulting compound by the modifier residue.Keywords Chemical modification; modifier effect; selenium determination; electrothermal atomic absorption spectrometry The determination of selenium by electrothermal atomic absorption spectrometry (ETAAS) has been the subject of many studies in recent years. The elucidation of the processes taking place in the electrothermal atomizer is still a great challenge for many atomic spectroscopists. A comparison of the results presented in many papers over the last decade indicated that for selenium these processes are complicated and variable. 1-9 The following factors play a decisive role the type of selenium cornpound1v2 and the type of graphite used for the construction of a tube or a platform (ordinary electr~graphite,~~~*~ pyrolytic graphite coated ele~trographite,~~ totally pyrolytic glassy carbon7 and graphite cloth3) the composition of the modifier the procedure for its preparation* and the compo- sition of the sheath g a ~ e s .~ - ~ Some processes taking place during the charring and atomization steps have been studied in detai11*2s4-10 but general comparison of most other results is confusing. This is obviously caused by the application of different experimental conditions. In order to study atomization and other processes in the electrothermal atomizer several workers have employed devices of different designs,'-'* in many instances of unusual c~nstruction,~*~*~ commercial and non-commerical tubes andor platforms. Experiments have often been performed under complicated non-stabilized atomization temperature condition~.~*~-~*~ During thermal pre-treatment the behaviour of various selenium compounds especially those containing (Se-") or those metabolized in biological samples and traced by in vivo incorporation of 75Se has seldom been e~amined.l-~.~*~J 1-14 Additionally modifiers of various compositions have often been utilized in different *Presented in part at the 3rd Conference on ICP and Develop- ment Trends in Atomic Spectroscopy held on the occasion of Professor E.PlSko's 60th binhday at Smolenice Czechoslovakia March 26th-30th 1990. mass ratios of modifier to analyte.14-16 Therefore the results cannot be compared directly in many instances. In general modifiers can only be applied with some limitations to the analysis of real samples by conventional ETAAS.Originally the ultimate aim of this work was to examine the stabilization and modification effects of various com- pounds on different forms of selenium in a commercially available graphite tube atomizer equipped with a L'vov platform. The work was undertaken with a view to choosing a suitable and efficient modifier for the direct analysis of body fluids.17 However owing to the complexity of the problem studied it proved to be necessary to examine the behaviour of the modifier for various pure selenium compounds first. This topic is dealt with in this initial paper. Experimental Apparatus A Perkin-Elmer Model 3030 atomic absorption spectro- meter with deuterium background correction equipped with an HGA-500 graphite furnace an AS-40 autosampler and a PR-100 printer was used for the atomic absorption measurements.The instrumental parameters used are summarized in Table 1. Unless otherwise mentioned Perkin-Elmer pyrolytic graphite coated graphite tubes (Part No. 121 092) with L'vov platforms of a 'new' type (Part No. 121 091) with cavity dimensions of 3 x 13 mm were used for the experk ments. Also two additional platform types were used for comparative measurements. A platform of the older type (marketed recently by Perkin-Elmer under the same part number) with approximate cavity dimensions of 1.8 x 13 mm and platforms of the design resembling the 'new' type manufactured from Slovak commercial spectroscopic elec- trographite SU and SG (Elektrokarbon TopolEany Czechoslovakia) (these having different density and poro-662 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY DECEMBER 1991 VOL.6 Table 1 Instrumental parameters and settings Temperature programme TemperatureV'C Drying stage 90 140 Chamng stage Variable Atomization stage 2400 Cleaning stage 2650 Cooling stage 300 Cooling stage 20 Ramp/s Holds 10 30 10 10 20 20 1 20 ot 3 1 3 1 20 Element Internal argon flow/ ml min-I 300 300 300 300 ol 300 300 Parameter se Pd Ni Wavelengtldnm Spectral bandwidthlnm Radiation source 196.0 204.0 247.6 303.8 2.0 0.7 0.2 0.2 EDL (5 W) HCL (30 mA) HCL (1 5 mA) * Nominal values. t 'Maximum power heating' with optical feedback control (-2 K ms-l). $ Gas stop; read command selected. sity) and from RW 1 spectroscopic graphite and graphite cloth (Ringsdorffwerke Bad Godesberg Germany) were used.For measurements of the timedependent temperature of the tube or platform during the atomization step a TMR 4854 optical pyrometer (Dr. Georg Maurer GmbH Kohl- berg Germany) equipped with a VL 100 optical adaptor and an ETSZL detector-amplifier unit was used. Either a defined segment (an area with diameter of 1.5 mm) of the tube wall or the platform in the central part of the tube were imaged and focused through the sampling hole on to the pyrometer sensor. Scanning electron microscopy (SEM) was used to obtain micrographs of the platform surface. A Tesla BS 300 microscope (Tesla Czechoslovakia) employing an acceler- ating voltage of 25 kV with magnification ranging from 200x to ~ O O O O X and a Philips SEM 505 microscope employing an accelerating voltage of 20 kV equipped with an EDAX PW 9900 microprobe system for surface analysis were used.These imaging techniques enabled the detection of the modifier as a separate phase on the graphite surface if it was present in amounts of greater than 1 x lo2 ng. Reagents Stock solutions of selenium were prepared from analytical- reagent grade selenium oxide sodium selenate (Lachema Bmo Czechoslovakia) selenomethionine (Sigma) and tri- methylselenium iodide (Institute of Nuclear Biology and Radiochemistry Prague Czechoslovakia). Stock solutions of the modifier were prepared from analytical-reagent grade crystalline palladium nickel magnesium calcium lan- thanum europium and aluminium nitrates (Lachema Brno) and acidified with analytical-reagent grade HN03.The substances used for the preparation of concentrated solu- tions were checked for selenium contamination and yielded a negative result. Spectroscopic-reagent grade argon was used as a sheath gas. Procedure Typically 20 pl of a solution containing either 2 ng of selenium in the selected chemical form or the same amount of selenium modified with various amounts of the selected nitrate were dispensed into the platform cavity. The acidity of the final solution was adjusted with HN03 to 0.003 mol 1-l in the acid. An independent measurement verified that this concentration had no influence on either the magnitude of the signal or its shape. Absorbance signals for selenium were measured and recorded following the charr- ing procedure at various temperatures (see Table 1).In order to eliminate possible influences of cross-memory effects between several different modifiers on the selenium signal a new untreated tube equipped with a new platform was usually used for each nitrate tested. Measurement of Analyte Distribution If double or multiple peaks were observed selenium distribution between the platform and tube wall was checked. After the charring step the platform was removed from the tube and the fraction of selenium retained on the tube wall was determined using peak area evaluation. The selenium remaining on the platform surface was measured following the re-insertion of the platform into the grooves of the tube. Measurements performed under the same conditions but without the removal and re-insertion proce- dure showed that the above technique is justified i.e.that the peak areas from the wall and platform are additive. Results and Discussion Selenium in the Absence of a Modifier Selenium signals of various separately sampled selenium compounds were measured under a variety of charring conditions with different graphite platform materials (see Figs. 1-3). Commercially available ('new' type) pyrolytic graphite platforms did not show a significant amount of reactivity with the different selenium compounds as can be seen in Fig. 1. It appears that the extent of selenium stabilization in the atomizer corresponds to the thermal behaviour of the particular selenium compounds. While selenate and selenite are volatile above 300 0C,497 seleno- methionine decomposes below 200 "C and trimethyleselen- ium iodide dissociates into its fairly volatile constituents dimethyl selenide and methyl iodide without distinct phase changes at even lower temperatures.'* As has been ob- served the signals corresponding to the remaining selenium show a similar shape (appearance time time of peak maximum etc.) for all the selenium compounds selected.Charring temperature had no influence on the signal shape. Finally variation of the volume of solution dispensed between 5 and 80 pl (while keeping the analyte massJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY DECEMBER 199 1 VOL. 6 v) s f Y Q L t A 0 Tube temperaturePC Fig. 1 Chamng curves for various selenium compounds. Sele- nium (2 ng) was dispensed as 20 pl solutions of A selenate; B selenite; C selenomethionine; and D trimethylselenium iodide onto the 'new' ordinary pyrolytic graphite platform 0 500 lo00 1500 Tube temperaturePC Fig.2 Chamng curves for platforms of different graphite ma- terials. Selenium (2 ng) as selenite (20 pl) was placed onto the platform. P commercially available 'new' pyrolytic graphite; Po 'older' SU SG and RW home made from the corresponding electrographite materials; and GT graphite cloth. Charring tem- perature 500 "C constant) showed no influence on the magnitude and shape of the signal. Measurements with selenium using amounts differing by two orders of magnitude (2 and 200 ng) showed no changes in signal shape with concentration in contrast to the observations made by Di5dina et al.5*6 who used a platform made from polycrystalline graphite. In contrast to totally pyrolytic graphite a pronounced reactivity of the platform graphite with inorganic selenium compounds was observed when the platform was made of polycrystalline electrographite material.In analogy with the observations of Chung et al.3 and Di5dina et al.,596 the graphite substrate stabilized selenate and selenite (see Fig. 2). Time-resolved signals displayed double peaks (see Fig. 3) the first of which decreased with increasing charring temperature and the second of which remained constant up to 1300 "C. When the commercially available 'new' type of platform was used the latter peak showed a slight signal tailing ( ( 5 % of the integrated absorbance signal) (see Fig. 3). Additional measurements at a charring temperature of 500 "C confirmed that the double peaks are not due to selenium deposition on the tube wall.In all instances less than 2Oh of the selenium applied was transferred to the tube wall. Unlike the inorganically bound selenium organic sele- SG p 0.5 2 0 P" RW 2 Time/s 4 663 Fig. 3 Time-resolved signals for selenium using platforms made from different graphite materials (for specification see caption to Fig. 2) nium compounds particularly trimethylselenium iodide did not show any interaction with graphite. The first peak on the signal trace was considerably larger than the second and disappeared at lower temperatures analogous to the situation shown in Fig. 1. Stabilization of organic selenium compounds therefore seems to be considerably less pro- nounced in contrast to selenite as shown in Fig.2. Selenium in the Presence of a Modifier Selenium signals were measured in the presence of various amounts of selected modifying compounds at various charring temperatures and for different chemical forms of selenium. The signal traces for selenate are summarized in Fig. 4 however the same trends were observed for selenite selenomethionine and trimethylselenium iodide. Charring curves in the presence of palladium and magnesium nitrates as chemical modifiers are shown in Figs. 5 and 6. Modifica- tion of all the selenium compounds studied with the remaining (nickel aluminium calcium europium and lanthanum) nitrates led to charring curves with the same characteristic form as with magnesium nitrate as shown in Fig. 6. It was noted that in several instances (e.g.for calcium and lanthanum nitrates) peak shape and peak area values changed dramatically with the amount of modifier present (see Fig. 4). In comparison with commonly used palladium and nickel compounds the application of the remaining nitrates produces surprisingly similar results but for a greater molar excess of the modifier. It can generally be stated that the shape and the position of the signals varies little with the form of the selenium compound and with the charring temperature. It follows that the type of atomization reaction does not appear to be the limiting factor influencing the shape of the signal.664 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY DECEMBER 199 1 VOL. 6 1.4 1 (a) 0 1 100 10 A i o o o 1 1 I a 1.3 - ( C ) 300 3000 300 000 s 0 1.2 -(el O r 0 1 2 1 .o 0 30 000 1.0 - ( f ) 0 - 0 1 2 3 0 P E! 0 w x $ I- 0 1 2 3 Time/s Fig.4 Absorbance versus time profiles for 2 ng of selenium as selenate in the presence of the selected modifier (a) palladium nitrate (b) nickel nitrate (c) magnesium nitrate (d) aluminium nitrate (e) europium nitrate u> calcium nitrate and (g) lanthanum nitrate. The modifier to selenium molar ratio is indicated on the corresponding traces. A charring temperature of 600 "C was chosen except for 300 "C for aluminium nitrate. The atomization temperature was 2400 "C except for nickel nitrate where 2100 "C was used. The broken line represents the signal of the modifier element (3 ng of palladium and 100 ng of nickel) and the dotted line the background signal of the aluminium matrix.A is the temperature profile of the tube; and B the platform An increasing amount of the modifier shifts the time of appearance of the signal to a greater value (see Fig. 4) so that the atomization of selenium takes place under the stabilized temperature conditions in the atomizer. For small amounts of some of the modifiers used splitting of selenium signals was observed (see Fig. 4). The first part of the split peak disappeared above a chamng temperature of 600 "C while the second peak remained up to 900-1 200 "C. For organic selenium compounds the first peak was larger than that for the inorganic form. The signal splitting disappeared as the amount of modifier increased 5 0 Q confirming the critical effect of the amount of modifier on b 0.5 the stabilization efficiency.0 The extent of selenium stabilization in its various n chemical forms in the atomizer as a function of the amount of the modifier and of the charring temperature applied is summarized in Figs. 5 and 6. It was found that the selenium signals for the four selenium compounds examined pos- sessed similar shape and magnitude if the molar excess of the modifier was approximately 300-fold for palladium (1 pg) 30 000-fold for magnesium nitrate (200 pg) 2000-fold for aluminium nitrate (20 pg) and 2000-fold for europium nitrate (20 pg). For nickel calcium and lanthanum nitrates at their maximum applicable modifier concentration ( 100 0 500 1000 1500 10 and 0.5 mg ml-l) the signals of trimethyselenium iodide were only about 50,90 and 80% respectively of the selenite 0.5 v) Y Tube temperature/"C selenium molar ratio A 0; B 1; C,- 10; D 100; E 1000; F 10 000; and G 40000 on the physico-chemical parameters of compounds likely to be produced in the atomizer e.g.selenate and selenite. ForJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY DECEMBER 199 1 VOL. 6 665 0 I -.-I '1 0 500 1000 1500 \ X Tube temperaturePC Fig. 6 Chamng curves for 2 ng of selenium as (a) selenate and selenite (b) selenomethionine and (c) trimethylselenium iodide in 20 pl of solution for various amounts of magnesium nitrate. The broken lines were obtained for the modifier mixture of magnesium and palladium nitrates in the molar ratio of 300000 (magne- sium) 4000 (palladium) 1 (selenium) i.e. 2 mg of magnesium nitrate with 10 pg of palladium.Modifier to selenium molar ratio A 0 B 300; C 3 000; D 30 000; and E 300 000 example the affinity of some metal ions such as NiZ+ CuZ+ and especially PdZ+ for the functional groups containing selenium is well known.19 Chemisorption of Se02(g) and Se(g) by MgO at 500 "C by CdO at 700 "C and by La203 at 1100 "C has also been d e s ~ r i b e d . ~ ~ - ~ ~ Therefore the transformation of all the selected selenium chemical forms into SeO by the excess of nitrate present can be expected in addition to sorption of SeO by the metal oxides formed. It is appropriate to mention in this context that there are no references in the literature suggesting that selenium can be stabilized by the formation of metal selenites which are usually thermally stable up to 600-700 "C and in some instances up to 1000 "C (for calcium and lanthan~m).~O-,~ The formation of such thermally stable selenium com- pounds during the analysis of biological samples e.g.body fluids could be advantageous as it could provide a common oxidative ashing procedure unlike the method of selenide formation which usually requires different chemical condi- tions. The modifiers tested except for calcium and lanthanum nitrates and large amounts of palladium nitrate did not give rise to any selenium or modifier memory effects. After 0 1 2 3 4 5 Tirnels Fig. 7 Influence of platform contamination by lanthanum on the selenium signal for 2 ng of selenium as selenite in 20 pl of solution containing A and B lanthanum nitrate (10 pg 900-fold molar excess); and C magnesium nitrate (2 mg 300000-fold molar excess).Traces A and B were recorded before and after contamina- tion caused by 100 pg of lanthanum nitrate (9000-fold molar excess) respectively. Trace C is not affected by the contamination the cleaning step the platform behaved in a similar manner to the untreated one based on measurements with pure selenium solutions. Such measurements showed peak char- acteristics closely resembling those obtained with a new platform. The lifetime of the platforms was 300-400 firing cycles without any perceptible change in the analytical properties. Calcium and lanthanum nitrates and large amounts of palladium nitrate ( 100 ,ug of palladium) caused considerable corrosion of the platform graphite owing to carbide formation and intercalation.The peak height and the peak area values first increased with an increasing amount of the modifier (see Figs. 5 and 6). This corresponds proportionally to the stabilization effects. However signal depression and broadening were caused by a much larger amount of a modifier (see Fig. 4). Remarkable differences were observed between the effects caused by the following nitrates calcium versus magnesium lanthanum versus europium and palladium versus magne- sium. A larger amount of a refractory matrix probably influences the atomization processes and thereby reduces the selenium signal. During the atomization step at temper- atures above 1600 "C calcium and lanthanum oxides produce thermally stable ionic carbide~,~O-~~ in contrast to magnesium and europium ~ x i d e s .~ l J ~ These carbides obvi- ously prevent the release of trapped selenium from the modifier residue (see Fig. 4). Similar effects of slow vaporization from a compact molten residue enhanced by intercalationZS might be expected with the use of a larger amount of palladium modifier. The influence of the platform memory effect on the selenium signal caused by lanthanum nitrate is shown in Fig. 7. This effect cannot be eliminated by repeating the cleaning step. The use of magnesium nitrate modifier prevented this problem. It can therefore be concluded that the magnesium nitrate matrix traps selenium and minimizes the contact between sele- nium compounds and the lanthanum oxide residue. These results also suggest that there is no evidence for a gas-phase interference caused by the decomposition products of lanthanum oxide (oxygen releasing effect) during the atomization step.A similar effect caused by a calcium nitrate residue which is finally converted into the less thermally stable carbide,2z was found to be relatively smaller. Exact interpretation of most of the results mentioned above characterized by general and common trends of changes in the shape of the signal could not be based solely on chemical interactions chemical modification or stabili- zation and on the formation of a new chemical b ~ n d * ~ * ~ ' (see also refs. 28-31 for comparison of common trends). The physical and physico-chemical properties of dried and charred residues such as the grain size of particles heat of vaporization melting- and boiling-points of the compounds generated also seem to play an important role by critically determining the shape and position of the signal.Vaporiza-666 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY DECEMBER 199 1 VOL. 6 tion effects might therefore act as a limiting step that influences the form of the final transient signal. Vaporization of a relatively refractory oxide matrix which is produced from the nitrate modifier during the charring step and traps selenium limits the vaporization and consequently also the atomization of selenium. It could be expected that the modifier allows selenium to vaporize and atomize only under such conditions where the matrix of the refractory modifier itself is volatilized. This is demonstrated for example by signal traces in Fig.8. A comparison of the signals for selenium and some modifier elements (see Fig. 4) also confirmed this conclusion. The matrix of the modifier traps the selenium in its bulk not only by chemical reaction but also by physical interactions such as sorption occlusion dissolution formation of solid solutions and melts depending on the pre-treatment condi- tions. This agrees very well with the situation where the tested mixtures of the modifier and trimethylselenium iodide were only dried at 140 "C (see Fig. 9). Chemically stable trimethylselenium iodide could not react with nitrate at this temperature and under these chemical conditions and thus it could not be converted into the selenite form. However the same signal shift and the signal increase were observed for this highly volatile selenium compound1* as for the selenite form.The position of the signal was not varied by higher temperature pre-treatment in comparative measurements using a charring step. For all selenium compounds tested the signal splitting disappeared if the modifier was used in a larger excess (see Fig. 4). It follows that this effect can only be caused by physical trapping in the matrix of the modifier. The SEM micrographs of the charred modifier residue on the platform surface showed that there were differences in the grain size and the spatial distribution depending on the amount of modifier. For example with a relatively small amount of modifier i.e. 300 ng of palladium well-separated particles with an approximate diameter of 1 x 1 0-1 pm were Fig.10 Scanning electron micrograph of 300 ng of palladium deposited as nitrate solution on the platform surface. Charring temperature of 1100 "C was applied. Magnification 5000x was used on an SEM Philips 505 microscope I I I J G 1 2 3 Time/s Fig. 8 Influence of modifier composition on the signal shape for 2 ng of selenium in 20 p1 of solution containing A palladium nitrate ( 10 pg of palladium 4000-fold molar excess); B magnesium nitrate (2 mg 300 000-fold molar excess); and C mixture of the same amounts of palladium and magnesium nitrates as for A and B I I I 0 1 2 3 Time/s Fig. 9 Time-resolved signals for dried non-charred residue containing 2 ng of selenium as trimethylselenium iodide and also with 200-fold molar excess of aluminium calcium europium lanthanum and magnesium nitrates.A drying and charring temper- ature of 140 "C was applied before atomization. The dotted line represents the signal of the non-modified solution 10 pm Fig. 11 Scanning electron micrograph of 100 pg of palladium deposited as nitrate solution on platform surface. Charring temper- ature 1 100 "C was applied. Magnification 5000x was used on a BS 300 Tesla microscope observed on the graphite surface (see Fig. 10). On the contrary with a larger amount of modifier e.g. 100 pg of palladium a significant agglomeration of particles was noticed (see Fig. 11). Similar micrograph patterns were observed for the remaining nitrates tested. The differences in the shape of the signal (see Fig. 4) such as changes in width and broadening could be influenced by the physical conditions of the residue as it vaporizes more easily from small separated particles or drops than from either robust agglomerates or from a compact film of the melt or alloy.Effect of the Modifier An attempt has been made to formulate a general hypothe- sis about the effects of modifying agents on the selenium signals based on a comparison of the experimental results discussed above with literature data.I-l0J5 It has been noted that the amount of modifying compound critically affectsJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY DECEMBER 199 1 VOL. 6 667 the data observed and the signal characteristics i.e. appearance time peak area and peak shape. The following conclusions were therefore drawn as a function of the amount of modifier present. In the absence of a modifier the processes are determined by mutual affinity of ‘active sites’ on the graphite plat- for the chemical forms of selenium being used and/or by their decomposition and vaporization character- istics as mentioned above.The extent of graphite activa- tion and interaction with selenium is very difficult to define as the amount of selenium is very low (1 x 10-lo-1 x lo-” mol). Therefore different stabilization efficiences distorted and split signals have often been observed. Stabilization of selenium by graphite is therefore of no practical impor- tance as concluded by D&dina et U Z . ~ ~ ~ particularly for the analysis of biological samples. In the presence of a ‘small’ amount of modifier i.e. with an equimolar ratio or with a molar excess of one or two orders of magnitude over selenium and probably over active graphite sites the processes in a modifier-analyte- graphite three-component system are more complicated.Depending on the affinity the modifier either reacts only with the selenium compound to form a new thermally stable precursor such as selenide’ or ~elenite,~,~ or graphite also affects the reaction producing ternary compo~nds.~J~ A slight excess of the chemical modifier might trap the new species in the matrix and thereby also affect the atomiza- tion. Therefore it appears that some confusing data have been presented in the literature as unfortunately the majority of work has been oriented towards the study of a ‘small’ amount of modifier. In the presence of ‘large’ amounts of modifier i.e. with the molar excess over selenium of greater than 1000-fold the incidental interactions of analyte with graphite are minimized and the reaction with the modifier prevails.Analyte compounds can be transformed into thermally stable compounds or can be trap~ed~O9~j by adsorption chemisorption and other physico-chemical processes in the crystal cage (lattice34) of the matrix of the modifier. In line with these physico-chemical phenomena the extent of analyte trapping controls the processes of atom formation to a considerable extent. Thus the vaporization of the carrier matrix and its interaction with graphite influences the release of the analyte. For example carbide formation or intercalation of the modifier residue during the atomiza- tion step has a negative affect on the release of analyte.Therefore a larger amount of a suitable modifying agent except for compounds forming thermally stable carbides should be applied for practical use. Conclusion According to observations the resulting transient signal for selenium appears to be determined not only by various solid-phase chemical reactions which have usually been accepted and discussed as critical phenomena but also strongly by the effects of physical and physico-chemical processes e.g. by occlusion and sorption in the matrix of the modifier and by vaporization of the residue of the modifier. Some refractory oxides and metals are efficient modifying carriers. Corresponding nitrates are the most suitable precursors of these carriers as the other salts such as chlorides and sulphates are potential interferents* during the charring and atomization procedures and they do not easily or completely form oxides.Nitrates are also the most suitable oxidizing agents to facilitate the mineralization of an organic matrix. The limitation that the chemical modifier should not form a thermally stable carbide during the atomization restricts the choice of a suitable agent from a group of metal nitrates. With regard to all the experimental limitations and other considerations a mixture of palladium and magnesium nitrates with a considerably larger amount of the latter than usual is proposed as the most efficient modifier for the determination of selenium particularly in biological fluids as both compounds react with selenium compounds at low and high temperatures and cause negligible corrosion of the graphite platform.At the same time the large amount of magnesium nitrate is expected to eliminate negative com- petitive effects of the sample matrix during the ashing procedure as will be reported in future work dealing with this topic. The results mentioned above clarify why the magnesium nitrate based modifiers have been popular for a wide range of application^.^^ The authors thank Dr. J. Kozel (Institute of Nuclear Biology and Radiochemistry Prague) for a donation of a pure trimethylselenium iodide standard Dr. M. Svoboda (Institute of Physical Metallurgy Czechoslovak Academy of Sciences Brno) and J. Kudrna (Veterinary Research Insti- tute Brno) for the SEM micrographs Dr. Z. Slovak Institute of Fine Chemicals Lachema Brno) and Dr.M. 5 ucmanova (Perkin-Elmer Austria) for instrumental sup- port of this work and Professor J. P. MatouSek (University of New South Wales Australia) and the referees of this paper for helpful comments and criticism during the preparation of the manuscript. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 References Welz B. Schlemmer G. and Vollkopf U. Spectrochim. Acta Part B 1984,39 501. Cedergren A. Lindberg I. Lundberg E. Baxter D. C. and Frech W. Anal. Chim. Acta 1986 180 373. Chung C.-H. Iwamoto E. Yamamoto M. Yamamoto Y. and Ikeda M. Anal. Chem. 1984 56 829. Droessler M. S. and Holcombe J. A. Spectrochim. Acta Part B 1987 42 981. DEdina J. Frech W. Lindberg I. Lundberg E. and Cedergren A. J Anal. At. Spectrom.1987 2 287. D5dina J. Frech W. Cedergren A. Lindberg I. and Lundberg E. J. Anal. At. Spectrom. 1987 2 435. Styris D. L. Fresenius 2. Anal. Chem. 1986 323 710. Voth-Beach L. M. and Shrader D. E. J. Anal. At. Spectrom. 1987 2 45. Droessler M. S. and Holcombe J. A. J. Anal. At. Spectrom. 1987 2 785. Styris D. L. Prell L. J. Redfield D. A. Holcombe J. A. Bass D. A. and Majidi V. Anal. Chem. 1991 63 508. Saeed K. Thomassen Y. and Langmyhr F. J. Anal. Chim. Acta 1979 110 285. Aadland E. Aaseth J. Radziuk B. Saeed K. and Thomas- sen Y. Fresenius 2. Anal. Chem. 1987 328 362. Lindberg I. Lundberg E. Arkhammar P. and Berggren P.-O. J. Anal. At. Spectrom. 1988 3 497. Alexander J. Saeed K. and Thomassen Y. Anal. Chim. Acta 1980 120 377. Schlemmer G. and Welz B. Spectrochim.Acta Part B 1986 41 1157. Kirkbright G. F. Hsiao-Chuan S. and Snook R. D. At. Spectrosc. 1980 1 85. DoEekalovB H. C.Sc. Thesis Masaryk University Bmo 1990. Rheinboldt H. in Houben- Weyl Methods of Organic Chemis- try S- Se- Te- Verbindungen ed. Muller E. Georg Thieme Stuttgart 1955 band IX pp. 974 979 980 1046 and 1187. Livingstone S. E. Comprehensive Inorganic Chemistry Per- gammon Press Oxford 1975 ch. 13. Gmelin s Handbook of Inorganic Chemistry Selenium Syst. No. 10 Supplment BI Compounds Springer Berlin 198 1 pp. 96 and 138. Gmelin ’s Handbook of Inorganic Chemistry Magnesium Syst. No. 27 Part B Verlag Chemie Weinheim 1939 p. 299. Gmelin s Handbook of Inorganic Chemistry Calcium Syst. No. 28 Part B Verlag Chemie Weinheim 196 1 pp. 822 and 835. Gmelin 3 Handbook of Inorganic Chemistry Lanthanides Syst. No. 39 Part CI Springer Berlin 1974 pp. 166 173 212 and 374.668 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY DECEMBER 1991 VOL. 6 24 25 26 27 28 29 30 Duval C. Inorganic Thermogravimetric Analysis Elsevier Amsterdam 1963. Selig H. and Ebert L. B. Advances in Inorganic Chemistry and Radiochemistry Academic Press New York 1980 ch. 23 Teague-Nishimura J. E. Tominaga T. Katsuva T. and Sakurada O. Takahashi H. and Tap M. Bunseki Kagaku 1989,38,411. Hinds M. W. Katyal M. and Jackson K. W. J. Anal. At. Spectrom. 1988 3 83. Rettberg T. M. and Beach L. M. J. Anal. At. Spectrom. 1989 4 427. Slavin W. Carnrick G. R. and Manning D. C. Anal. Chem. 1982,54 621. 3 1 32 p. 281. 33 Matsumoto K. Anal. Chem. 1987 59 1647. 34 35 Hinds M. W. and Jackson K. W. J. Anal. At. Spectrom. 1990 5 199. Heuttner W. and Busche C. Fresenius 2. Anal. Chem. 1986 323 674. Tsalev D. L. Slaveykova V. I. and Mandjukov P. B. Spectrochim. Acta Rev. 1990 13 225. Tsalev D. L. Slaveykova V. I. and Mandjukov P. B. lecture presented at the VII Polish Spectroanalytical Conference and X CANAS Torun September 5th-9th 1988. Schlemmer G. and Welz B. Spectrochim. Acta Part By 1986 41 1157. Paper 0/03541F Received August lst 1990 Accepted July 23rd 1991