ANALYST, MARCH 1986, VOL. 111 269 Investigations into the Improvement of the Analytical Application of the Hydride Technique in Atomic Absorption Spectrometry by Matrix Modification apd Graphite Furnace Atomisation Part 1. Analytical Results Klaus Dittrich and Rita Mandry Karl-Marx-University, Chemistry Section, Analytical Centre, Talstrasse 35, 7010 Leipzig, GDR Hydride generation AAS is characterised as a very sensitive analytical technique using the commercial AAS-3 hydride system with quartz tube atomisation. Studies of matrix interferences are reported for the trace elements As, Sb, Se and Te. The hydride-forming elements As, Sb, Bi, Se, Te, Ge, Sn and Pb were used as matrices. Two types of matrix interference can be distinguished: matrix interference in the liquid phase of hydride generation, and matrix interference in the gaseous phase of hydride atomisation.The strong matrix interference in the liquid phase was diminished by matrix modifications and new reagents are characterised. To diminish or avoid matrix interference in the gaseous phase, a new type of graphite tube atomiser was developed. The dimensions of the graphite paper atomiser are comparable to those of commercial quartz tubes, but the heated graphite atomiser volume is very small (comparable to HGA 500 tubes). The analytical applicabilities of both systems were characterised and compared. The absolute detection limits are between 0.1 and 0.5 ng. The relative detection limits in graphite paper atomisers are 2-1000 times better than in quartz tube atomisers.Keywords: Atomic absorption spectrometry; h ydride technique; matrix modification; graphite furnace atomisation In the last 30 years, atomic absorption spectrometry (AAS) has become one of the most commonly used methods in trace analysis. This development was connected with the introduction of some new techniques, especially atomisation, for example AAS with different flames, AAS with electrothermal atomisation in graphite tubes or graphite rods, AAS with evaporation of solid material by lasers, AAS with sputtering and AAS with the generation of volatile hydrides, called hydride AAS for short. The reasons for all these developments have been to improve the analytical criteria of AAS methods, such as detection ability, accuracy, selectivity, reproducibility and applicability. The initial development of hydride AAS was carried out by Holakl in 1969, but in 1955 Erdey et aZ.2 generated volatile hydrides for analytical measurements in d.c.arc atomic emission spectrography . Holak used the well known Marsh reaction in AAS. Arsine (AsH3) was generated by nascent hydrogen (from Zn/H+), transported by an argon stream and collected in a cold trap (liquid nitrogen). After collection, the trap was rapidly heated to room temperature. The volatile arsine was transported by argon into an air - acetylene flame for atomisation. The absorption was measured at 193.7 nm. Up to now three distinct operations have been performed in hydride AAS: hydride generation, hydride transportation and hydride atomisation. The procedure of all steps has been changed and improved in the last decade.Hydride generation. Pollock and West (1972)3 introduced as the reducing agent Mg - Ti3+ - H+ mixtures and Goulden and Brooksbank (1974)4 used A1 - H+ for this purpose. As late as 1972 Braman et aZ.5 introduced BH4- - H+ solutions in atomic emission and Schmidt and Royer6 transferred this hydride generation reaction system to AAS. The latter system is mostly used today, because the homogeneous reduction reaction is faster than the heterogeneous reduction by M - H+ systems. In general, d.5-10% solutions of NaBH4, stabilised by 0.5-270 NaOH, are used. Mostly dilute HCl is used, but HN03, citric acid, oxalic acid, tartaric acid and maleic acid can also be used. Hydride transportation. There are two methods: the direct transfer mode and a collection mode.The latter was necessary for the slowly working M - H+ reduction systems. Since the introduction of the BH4- - H+ reduction, the direct transfer mode has become possible and useful. Nevertheless, in some instances the collection mode was used for enrichment and concentration, but according to Chapman and Dale7 this is useful only for the most stable hydrides, such as AsH3, SbH3, BiH3 and SeH2. Hydride atomisation. The most intense development has taken place in this field. The air - acetylene flame used at first has disadvantages owing to its high background absorption and sensitivity due to plasma dilution. In 1972, Chu et al.8 introduced the electrically heated quartz tube. Knudson and Christian9 also used the graphite tube very early.Special flame techniques, the flame-heated quartz tube6 and the flame-in- tube technique,lo were introduced in 1974. Today the electrically heated quartz tube is most commonly used, because this technique is simple and for many analytical samples the other techniques including the graphite tube do not give advantages. The following advantages and disadvantages of hydride AAS can be summarised: 1. In many instances the traces of the analyte can be separated from the sample, which improves the accuracy. 2. The efficiency of sample introduction into the plasma is very high (nearly loo%), which gives good sensitivity. 3. Large solution volumes can be used, in special instances with the collection mode, which gives good sensitivity. 4. The method can be automated and a flow injection mode of operation is possible.5. Contributions to chemical speciation are possible. l1 For a long time the first point was the major advantage. However, recently it has become more evident that this is also the reason for the main disadvantage, owing to the existence or occurrence of matrix interferences (see also reference 12). In this paper we present proposals for the systematic270 ANALYST, MARCH 1986, VOL. 111 characterisation of matrix interferences, and for avoiding or diminishing such effects. For this purpose we have used the most complicated conditions: the determination of volatile hydride-forming elements (As, Sb, Se and Te) in volatile hydride-forming matrices (As, Sb, Bi, Se, Te, Ge, Sn and Pb). We have used a new AAS-3 hydride system with a quartz tube atomiser (Carl Zeiss, Jena, GDR) and have also developed a new atomisa- tion system, a long-path graphite tube atomiser, based on a new material, on graphite paper. In the following paper, some explanations about the causes of matrix interference are given.Experimental Apparatus An AAS-3 atomic absorption spectrometer (Carl Zeiss), a hydride system for AAS-3 (hydride generation system) and quartz tube atomiser (QT) (controlled by computer) (see Fig. 1) and a graphite paper atomiser (GPA) (laboratory construc- ted) (see Fig. 2) were used. The graphite paper used had a thickness of about 0.1 mm. A rectangular piece was cut out (92 x 33 mm) and a tube of length 92 mm and diameter 9 mm was formed. The overlap was 3-4 mm. The ends of the tube were stabilised by inner graphite rings (0.d.8.8 mm; i.d. 7 mm; length 5 mm) and jammed between two water-cooled brass half-rings. To avoid thermal destruction we used four small outer graphite rings (0.d. 9.5 mm, i.d. 9.1 mm, length 1 mm). The tubes were heated by direct resistance heating (0-20 V; 0-200 A). The pyrometrically measured temperatures are shown in Table 1. The medium lifetime of such a tube at atomisation tempera- tures of about 2000 "C and a heating period of 30 s is about Ar Ar iser NaBH4 Sample A B Fig. 1. AAS-3 hydride device (Carl Zeiss, Jena, GDR) Brass (water-cooled) n Graphite stabilisation Graphite Quartz window f Graphite I I I Quarktube I paper tube, I electrically 1 I heated Ar Hydride shielding Ar, H2 gas Fig. 2. Graphite paper atomiser (GPA) (laboratory-constructed) 70-100 heating cycles.The paper was made by EKL (Berlin, GDR). The shielding gas was argon. The outer flow-rate was 20 1 h-1 and the inner flow-rate was controlled by the hydride system (6-40 1 h-1). For comparison we also used graphite tube cuvettes of the Beckman type (Type 1268) and Carl Zeiss type (AE3). The light sources were electrodeless discharge lamps for As (9 W), Sb (7 W), Se (8 W) and Te (11 W) (Westinghouse, USA). Reagents Reagents were of analytical-reagent grade from Laborchemie Apolda, GDR. Stock solutions of As3+, Sb3+, Se032- and Te032- with concentrations of 1 mg ml-1 were prepared from As203, Sb2O3, Se02 and Te, respectively. Stock solutions for matrix elements were prepared at concentrations of 50-100 mg ml-1 by dissolving the appropriate amounts of As203, Sb203, Se02, Te, Ge02, SnCI2 and Pb(NO& in 1 M hydrochloric or nitric acid.Procedure Vessel A (Fig. 1) contains the alkali stabilised solution of sodium tetrahydroborate(II1) (NaBH4). Vessel B is for the sample and can be changed rapidly. The system is controlled by different Ar flows using the computer (Fig. 3). The acidifed sample is placed in vessel B. After initiation, two gas streams [the first purge stream (36 1 h-1) and the carrier stream (18 1 h-l)] purify the system (about 30 s are necessary). In the reaction phase (reduction by BH4- - H+), the purge stream is closed and the pressure gas valve is opened Gas flow Start Purge phase Reaction phase Purge phase Purge gas 36 I h-' Carrier gas 18 I h-1 Pressure gas 6 I h-1 Purge gas 36 I h-1 Carrier gas 18 I h-1 Measurement 1 I I I Auto-zero Signal integration Computerisation Temperature of quartz tube atorniser constant I I Temperature of graphite paper atomiser I I I I Heating phase Constant-temperature phase Cooling phase Fig.3. Scheme of procedure using hydride device Table 1. Relationship between electrical conditions and the resulting temperatures of the GPA PotentiaUV Primary 110 120 130 140 150 160 170 180 190 200 210 220 230 240 Secondary 9.5 10.2 11 12 13 14 14.8 15.2 16.1 16.9 17.6 18.4 19.1 20.5 Temperature/"C 1450 1600 1700 1800 1900 1980 2100 2120 2200 2280 2310 2400 2460 2560 Heating rate/ "C s-' 100 150 200 300 333 670 750 1000 1000 1300 1500 1700 2000 2300ANALYST, MARCH 1986, VOL. 111 271 (6 1 h-1).Depending on the duration of the reaction phase a variable amount of the reagent is transferred from vessel A into vessel B. After the reaction phase, the second purge (or trans- portation) phase is initiated. The pressure gas supply is closed and the second purge gas supply is opened. During the reaction phase and the second purge phase, the hydride - hydrogen - argon gas mixture is transported to the atomiser. Every time a small gas stream (6 1 h-1) avoids the introduction of air. The first seconds of the reaction phase are used to auto-zero the instrument. After this, the measuring phase is started (peak-height and -area integration modes are possible). If the system is in operation with the quartz tube atomiser, the tube is heated before to the desired temperature (up to 1000 "C) for all procedures.If the system is operated with the graphite paper atomiser, the graphite tube is heated at the end of the first purge period by resistance heating to the desired temperature (up to 2600 "C). The heating period is completed in 2-3 s. The hold time for the selected temperature is given by the total duration of the reaction and the second purge phases. Subsequently the tube is cooled rapidly by water cooling. Results and Discussion Investigations of the Commercial Hydride System (AAS-3, QT) Analyses of pure dilute solutions of As, Sb, Se and Te In order to guarantee high sensitivity, it is necessary to optimise the concentration of the acid used, the amount of NaBH4 and the sample volume. As the sample volume is an independent parameter, two volumes (0.2 and 10 ml) were used for the investigations. It was observed that a concentration of 2 M hydrochloric acid was the best (1 and 3 M HC1 led to signal depressions of about 5-10%).The optimum NaBH4 concentration was 3% and the optimum amounts were 0.8 ml for the smaller and 1.3 ml for the larger sample volume. The possibility of the direct introduction of the NaBH4 (introduction time <0.5 s) was also examined. It appeared that the AA signal height and shape were unchanged, but the reagent volumes required were smaller (0.5 and 1.1 ml, respectively). The analytical results obtained under the optimum condi- tions are shown in Table 2. Table 2 shows that the absolute reciprocal sensitivities (for 1% absorption, to a first approxi- mation this value is similar to a 30 detection limit criterion) are one order of magnitude better for the smaller sample volumes.There are two reasons for this: (1) with small sample volumes the reduction is faster owing to the faster mixing and higher concentration of the reagent; the homogeneity of the solution is also more rapidly achieved; and (2) with smaller volumes the amount of NaBH4 required is smaller, which leads to a smaller dilution of hydride by excess of hydrogen. As the larger volume is more than one order of magnitude greater than the small volume used, Table 2 shows that the large volume gives, as expected, better relative reciprocal sensitiv- ity and therefore also better detection limits. Optimisations of sample volume are possible in both directions, depending on the requirements of specific analy- tical problems.Investigation of matrix interferences of volatile hydride- forming matrices Table 3 shows the systems evaluated and Fig. 4 summarises the results of the investigations. Similarly to our earlier results for the determination of Se and Te,13 we found in all instances more or less strongly decreased signals. From the results in Fig. 4, it can be concluded that trace analysis is impossible in most instances. Classification of matrix interferences Interferences by different matrices have been investigated in hydride generation AAS for many systems. The influence of heavy metals has been frequently described.12 These metal ions only influence the procedure in the liquid phase of hydride generation.In our case it is possible that the matrix interferences could occur in all three steps of the hydride technique. Table 4 indicates the possible causes of interfer- ence. Possibilities of avoidance of matrix interferences in the liquid hydride generation phase Many investigators have described methods for decreasing matrix interferences in the liquid phase of hydride generation. A survey of these papers is given in reference 12. Recent publications by Welz and Melcherl4-16 also deal with this problem and have attempted to offer explanations in some instances. Welz and Melcherl7 concluded that the influence of heavy metals does not consist in the simple formation of soluble selenides, such as CuSe and CoSe, but that the true influence is connected with adsorption and destruction of hydrides on surface-active precipitates of the heavy metals formed by reduction. Procedures recommended for avoiding or decreasing matrix interferences have included change of acidity of the solution, formation of complexes (masking) with the matrix interferents, and formation of precipitates of the matrices.Complexation reagents used include EDTA, KI, citric acid, thiosemicarbazide 1 ,lo-phenanthroline, oxalic acid, thiourea and pyridylaldoxime (for details, see reference Two interesting examples of precipitation reactions should be mentioned. The determination of As in the presence of Se 12). Table 2. Results of hydride AAS using the AAS-3 with hydride system and quartz tube atomiser (QT) Reciprocal sensitivity per 1% A Sample volume 0.2 ml Sample volume 10 ml Relative/ Trace ng ml- element Absolutehg (p.p.b.) As ., . . 0.17 0.9 Sb . . . . 0.11 0.6 Se . . . . 0.2 1.0 Te . . . . 0.19 0.95 Relative/ ng ml-l Absolutehg (p. p. b. ) 1.6 0.16 0.8 0.08 1.8 0.18 1.1 0.11 Table 3. Matrix systems investigated Trace element Matrix substances As . . . . . . Sb, Bi, Se, Te, Ge, Sn, Pb, InSb Sb . . . . . . As, Bi, Se, Te, Ge, Sn, Pb, GaAs, InAs Se . . . . . . As, Sb, Bi, Te, Ge, Sn, Pb, GaAs, InAs, InSb Te . . . . . . As, Sb, Se, Ge, Sn, Pb, GaAs, InAs, InSb Table 4. Possible causes of matrix interferences in hydride AAS Step Interference Hydride generation . . Reduction of matrix (loss of NaBH,); reaction of matrix or reduced matrix with trace amounts of hydride (e.g. precipitation); adsorption of hydride on precipitates Decomposition of unstable hydrides on surface (only for hydride-forming matrices) atomisation mechanism; molecule formation (only for hydride- forming matrices) Hydride transportation .. Hydride atomisation . . Relatively unknown; change in272 ANALYST, MARCH 1986, VOL. 111 1 oc 80 6C 40 20 $ F 0 d - 100 80 60 0 Pb Ge Te I I I I 1 'b) Pb Bi Ge Te Sb Pt: GE As Sn Sb I 0 2 4 6 8 10 0 2 4 6 8 10 Mat rix/pg Fig. 4. used, 10 ng) Interference of hydride-forming matrices on the determination of (a) As, (b) Se, ( c ) Sb and (d) Te, using QT (amount of analyte is actually improved in the presence of Cu2+ ions, because these bind the generated H2Se as soluble CuSe.17J8 Selenium determinations in the presence of Cu2+ ions are improved by the addition of Te02, because the simultaneously generated H2Te forms insoluble CuTe.1'21 Except for the example of the determination of As in the presence of Se, no other literature information was available for reducing the matrix effects of our analytical combinations.We therefore attemp- ted to obtain appropriate improvements by matrix modifica- tion procedures. Investigation of the influence of Cu2+ ions on the determina- tion of As and Sb in the presence of Se and Te. The results of these investigations are illustrated for As in Fig. 5 . The effects observed for Sb were the same and are not shown here. Fig. 5 shows that the influence of the Se matrix is stronger than that of the Te matrix. The general reason for the interference lies in the formation of insoluble compounds such as As2Se3, Sb2Se3, Sb2Se3, As2Se3 and Sb2Tes, and in the adsorption of these compounds on the surface of elemental Se and Te precipitates.The addition of increasing concentrations of Cu2+ ions reduced the depression of both Se and Te on the AA signal. The influence of Cu2+ ions is much stronger for the Te matrix than the Se matrix. We assume that this is due to the smaller solubility constant of CuTe compared with CuSe. Particularly with the Se matrix but also for the Te matrix, it is impossible to avoid completely the matrix interference. From this fact we deduce that there are other matrix interferences than those in solution, e.g., in the gaseous phase. Investigations of the influence of ED TA on the determination of As, Sb, Se and Te in the presence of Bi, Ge, Sn and Pb matrices.The results of these investigations for a trace element to matrix ratio of 1 : 1000 are shown in Table 5. The amount of EDTA was sufficient to complex the metal ions fully. With the Ge and Pb matrices only small improve- ments were observed because Pb2+ ions do not have a strong 0 0.01 0.10 1.00 0 0.01 0.10 1.00 Mat r i xipg Fig. 5. Influence of the matrix modifier Cu2+) on the trace determination of As and Sb in the presence of (A Se and (b) Te matrix (amount of analyte, 10 ng) Table 5. Improvement in the determination of As, Sb, Se and Te in the presence of hydride-forming matrices (Bi, Sn, Pb, Ge; 10 pg) achieved by addition of EDTA Improvement factor Trace element (10 ns) Bi Sn Pb Ge As . . . . . . 7 5* 1.05 1.15 Sb . .. , . . 3.3 loo* 1.05 1.06 Se . . . . . . 1.3 30 * 1.15 1.7 Te . . . . . . - 1.3* 1.7 1.7 * 5 M HCI.ANALYST, MARCH 1986, VOL. 111 273 influence and complex formation between Ge*+ ions and EDTA is not high. A successful outcome was achieved for the complexation of Bi3+ ions, especially for the determination of Sb and As. Te determinations are impossible in the presence of Bi3+ ions, irrespective of the presence or absence of EDTA. Interference by Sn2+ ions, which in most instances is the strongest influence, is diminished by EDTA. The use of a higher concentration of hydrochloric acid ( 5 M) is advan- tageous. Investigation of the influence of citric and tartaric acids on the determination of Se in the presence of Sb and Te matrices. The presence of a 100-fold excess of the complexing reagent (10 pg) improves the analytical results only by a factor of 1.5 (Sb) or 1.4 (Te).Investigations Using the AAS-3 Hydride Generation Systems Combined with the Graphite Paper Atomiser As indicated above, only in some instances is it possible to avoid matrix interference using chemical matrix modification. We therefore assumed that very strong matrix interferences occur in the gaseous phase in addition to the solution phase. In order to evaluate such a hypothesis, it is necessary to provide some variation of the temperature of the atomiser. Because further enhancement of the temperature (> 1000 "C) in quartz tube atomisers is impossible, we investigated the use of a graphite tube atomiser. Characterisation of graphite tubes for hydride atomisation After the first application of graphite tubes in hydride atomisation by Knudson and Christian,g a number of other workers used this technique.12 In most instances small graphite tubes of the HGA 2100 type (Perkin-Elmer) have been used.This graphite tube is not very useful for hydride atomisation, because the inner volume is small (see Table 6) in relation to the large gas volume produced in the generation reaction (argon - hydrogen - hydride mixture up to 80 ml). The analytical sensitivities achieved are therefore relatively poor. We therefore attempted to develop a new graphite tube atomiser, which avoids this disadvantage, and which has dimensions similar to those of the quartz tube atomiser widely Table 6. Dimensions of commercial quartz tube and graphite tube atomisers and of the new developed graphite paper atomiser Graphite tubes HGA 500 Beckman Dimensions QT AE3 1268 GPA Length/mm .. . . . . 145 28 63 92 I.d./mm . . . . . . 18 6 9 9 Internal volume/cm3 . . 37 0.8 4.8 5.9 Heated graphite volume/cm3 . . . . - 0.6 3.6 0.5 used for this technique. Table 6 shows the dimensions of these different types of atomisers. The Beckman Type 1268 atomiser has the longest path length of the commercial graphite tube atomisers. Its disadvantage is the high energy that is needed (10 V, 1000 A = 10 kW), because the mass of graphite to be heated is very large. The graphite paper atomiser developed in this work has a longer path length (150%) compared with the Type 1268 graphite tube, but the heated graphite volume is only 13%.In comparison with the HGA 500 graphite tube type and AE 3 (Carl Zeiss), the enlargement of the inner dimensions is very high, but the heated graphite volume is only 83%. Depending on the temperature required, a power of 3-4 kW is needed. Investigation of pure dilute solutions of volatile hydride- forming elements, As, Sb and Te in graphite tube atomisers The results of these investigations are shown in Table 7. Only with the GPA can the same analytical sensitivities as with the QT be achieved. The results support our assumption that the reason for the relatively infrequent application of graphite tube atomisers in hydride AAS is the poorer analytical sensitivities obtained in normal graphite tube atomisers compared with the QT atomiser.Investigation of the influence of volatile hydride-forming matrices on the determination of As, Sb, Se and Te using the GPA The results of this investigation are shown in Fig. 6. The optimum atomisation temperatures were 1850 "C (Se, Te) and 2000 "C (As, Sb). Fig. 6 shows the improvement of the AA signal depression ( i ) using an enhanced atomisation temperat- ure, (ii) by chemical matrix modification, using the QT atomiser, and (iii) by an enhanced atomisation temperature and chemical matrix modification. It can be seen that the increase in atomisation temperatures available in the GPA gives the best analytical results for the determination of As, Sb and Se in the presence of As, Sb, Bi and Se matrices. Only the effects of a Te matrix on As- and Sb determinations, and of all matrices on Te determinations, are not improved by the use of higher temperatures, In all instances the combination of the Table 7.Results of hydride AAS using different atomisers (reciprocal sensitivity/ng per 1% A ) . Sample volume, 0.2 ml Graphite tube, Trace Quartz Graphite Beckman element tube, AE3 tube, AE3 1268 As . . . . 8.17 0.35 0.20 Sb . . . . 0.11 0.18 0.14 Se . . . . 0.20 0.32 0.24 Te . . . . 0.19 0.29 0.24 Graphite tube, graphite paper 0.16 0.11 0.21 0.20 Table 8. Results of the determination of trace amounts of hydride-forming elements in hydride-forming matrices using GPA at 2000 "C and optimum matrix modification in comparison to QT without matrix modification. Values: reciprocal sensitivities per 0.01 A * Relative detection limits,* p.p.m.with respect to matrix element Trace element Atomiser As Sb Bi Se Te Ge Sn Pb As . . . . . . QRA GPA Sb . . . . . . QRA GPA Se . . . . . . QRA GPA Te . . . . . . QRA GRA - 100 5 180 3 600 600 120 8 1100 2900 230 190 - 400 1 120 3 240 30 1900 1100 2600 260 2 0.7 1800 450 3 2 - 120 8 2900 - - - 140 740 8 45 4 150 40 190 90 1040 30 530 7 600 120 390 290 4 2 0.4 0.4 30 12 290 190 * Because the 0.01 absorbance is in all instances to a first approximation equal to three times the relative standard deviation it is possible to write detection limits.274 ANALYST, MARCH 1986, VOL. 111 1 oc 80 60 40 20 $ E 0 d - 100 80 60 40 20 0 Sb Bi Se Te matrix As 4s Sb Bi Te matrix Se AS Bi Se Te matri) 4s Sb Se matrix Fig. 6. Improvement of AA signal depressions (-) caused by hydride-forming matrices in QT using chemical matrix modification (- - -) (QT) and enhancement of temperature (1900-2000 "C) (GPA) (-.--); sum of improvements (-).Amount of analyte element, 10 ng per 0.2 ml; amount of matrix element, 10 yg per 0.2 ml; EDTA, 1 M; Cu2+, 1 mg per 0.2 ml; citric acid, 5%; and tartaric acid, 5% Table 9. Results of the determination of As, Sb, Se, Te in thin layers of AIIIBV semiconductor materials using hydride AAS with GPA. Conditions: layer separation by chemical etching; dimensions of layer, 10 x 10 x 0.0005 mm; sample, 0.25 mg. Values in parentheses are improvement factors relative to QT Relative detection limit/atoms ~ m - ~ Trace element GaAs InAs InSb - 9 x 1016 As - Sb . . . . . . 2.6 x 1016 5.6 x 1016 Se . . . . . .4.8 (lOo) x 10'8 5.1 x 10'8 Te . . . . . . 5.7 x 1018 . . . . . . (100) 4 x 1017 (100) (100) ( 5 ) ( 5 ) (5) ( 5 ) (10) - 5.4 x 10'8 6.1 x 10'8 Conclusions Application of the graphite paper atomiser , which allows atomisation temperatures up to 2600 "C is very useful for the determination of trace amounts of hydride-forming elements in the presence of hydride-forming matrices. The improvement factors for application of the GPA are 1-3 orders of magnitude. In spite of the strong improvement by high-temperature atomisation, it is necessary to apply matrix modifiers. The improvements that can be obtained by GPA atomisa- tion and matrix modifications show that matrix interferences occur in different ways in the liquid and gaseous phases. To explain the reasons, especially for matrix interferences in the gaseous phase, the results of further experiments are given in the following paper.GPA and chemical matrix modification gave the best results (see Fig. 6 and Table 8). Table 8 shows the results obtained for the determination of AS, Sb, Se and Te in the presence of hydride-forming matrices using the GPA and chemical matrix modification. In all instances significant improvements were achieved, the improvement factors being between 1.5 and 1000. The combination of GPA and chemical matrix modification has been used for the trace determinations of As, Sb, Se and Te in AIIIBV semiconductor materials. As shown in Table 9, considerable analytical improvements were obtained in many instances, Several conclusions can be drawn from these results. In particular, the application of the higher atomisation temperat- ures in the GPA for hydride AAS is very useful, particularly if volatile hydride-forming elements are present as matrices.Strong matrix interference appears to occur in the gaseous phase in addition to the liquid phase of the QT atomiser for this type of matrix. 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. References Holak, W., Anal. Chem., 1969, 41, 1712. Erdey, L., Gegus, E., and Koscis, E., Acta Chim. Hung., 1955, 7, 343. Pollock, E. W., and West, S. J., At. Absorpt. Newsl., 1972, 11, 104; 1973, 12, 6. Goulden, P. D., and Brooksbank, P., Anal. Chem., 1974, 46, 1431. Braman, R. S . , Justin, L. L., and Foreback, C. C., Anal. Chem., 1972, 44, 2195. Schmidt, F. J., and Royer, J. L., Anal. Lett., 1973, 6 , 17. Chapman, J . F., and Dale, L. S . , Anal. Chim. Acta, 1979,111, 137. Chu, R. C., Barron, G. P., and Baumgarner, P. A . W., Anal. Chem., 1972, 44, 1476. Knudson, E. J., and Christian, G. D., Anal. Lett., 1973, 6 , 1039. Siemer, D. D., and Hagemann, L., Anal. Lett., 1975, 8, 323. Aggett, J., and Aspell, A. C., Analyst, 1976, 101, 341. Nakahara, T., Prog. Anal. At. Spectrosc., 1983, 6 , 163.ANALYST, MARCH 1986, VOL. 111 275 13. 14. 15. 16. 17. Dittrich, K., Vorberg, B., and Wolthers, H., Tuluntu, 1979,26, 747. 104, 232. Welz, B., and Melcher, M . , Analyst, 1984, 109, 569. Welz, B., and Melcher, M., Analyst, 1984, 109, 573. Welz, B., and Melcher, M. , Analyst, 1984, 109, 577. Welz, B., and Melcher, M., “Wissenchaftliche Beitrage der Karl-Marx Universitat, Analytiktreffen 1982, Atomspektros- kopie,” Karl-Marx-Universitat, Leipzig, 1983, p. 165. Welz, B., and Melcher, M., Anal. Chim. Actu, 1981,131, 131. Kirkbright, G. F., and Taddia, M., At. Absorpt. Newsl., 1979, 18. 68. 20. 21. Azad, J., Kirkbright, G. F., and Snook, R. D., Analyst, 1979, Bye, R., Engrik, L., and Lund, W., 2. Anal. Chem., 1984,318, 25. 18. 19. Paper A5182 Received March 5th, 1985 Accepted September 20th, 1985