Determination of Main and Minor Components of Silicon Based

Jul 21, 2000 - For the determination of main and minor components in silicon-based ceramic powders, a decomposition by a combustion with elemental ...
0 downloads 0 Views 106KB Size
Download PDF
Anal. Chem. 2000, 72, 3875-3880

Determination of Main and Minor Components of Silicon Based Materials by a Combustion with Elemental Fluorine. Separation of Gaseous Fluorination Products by Carrier Gas Distillation and Gas Mass Spectrometry Katrin Russe†

Fachbereich Chemie, Universita¨t Dortmund, D-44221 Dortmund, Germany Heinrich Kipphardt‡

Institute for Reference Materials and Measurements, B-2440 Geel, Belgium Jose´ A. C. Broekaert*

Institut fu¨r Analytische Chemie, Universita¨t Leipzig, Linne´ str. 3, D-04103 Leipzig, Germany

For the determination of main and minor components in silicon-based ceramic powders, a decomposition by a combustion with elemental fluorine and separation of the volatile fluorination products by a carrier-gas distillation with a subsequent detection by quadrupole mass spectrometry is described. The necessity and success of the separation step is demonstrated for the determination of boron as a minor constituent in SiC, where the spectral interferences of silicon on the boron signals are decreased considerably. The method developed is shown to be directly applicable to determination of silicon in Si3N4, SiC, and SiO2. The determination of nitrogen in Si3N4 requires additional effort, to separate nitrogen from the excess of fluorine. For the determination of boron, a complete mobilization of BF3 is assured by the presence of an adequate amount of GeF4. Analysis results obtained with different types of calibration show a precision of 30 µg for boron at the milligram-per-gram level and a precision between 0.5 and 2% (m/m) for the main components, silicon and nitrogen. Within these standard deviations, the results agree well with the values expected from the stoichiometry, with the results for silicon and boron obtained by wet chemical decomposition and slurry techniques in combination with ICP-OES and with the results for nitrogen obtained by carrier gas heat extraction. The combustion with oxygen and the determination of the resulting volatile oxides (i.e., H2O, CO2, SO2) and nitrogen by gas* Corresponding author. Tel.: +49 341 9736100. Fax: +49 341 9736115. E-mail: [email protected]. † Presently working at: Institute for Reference Materials and Measurements, B-2440 Geel, Belgium. ‡ Presently working at: Federal Institute for Materials Research and Testing (BAM), D-12200 Berlin. 10.1021/ac991238c CCC: $19.00 Published on Web 07/21/2000

© 2000 American Chemical Society

Figure 1. Typical reaction products of a combustion with elemental fluorine.

analytical methods, such as IR spectrometry or gas mass spectrometry, is used as a standard method for quantitative elemental analyses of organic materials. Whereas the application of this method is limited to the few elements which form volatile oxides, an analogous method based on decomposition with elemental fluorine (fluorine volatilization (FV) analysis) has been shown to be of potential interest for quantitative elemental analyses of inorganic materials.1,2 At temperatures of 400-600 °C, elemental fluorine reacts with nearly all materials. Even advanced ceramics such as compact pieces of SiC, for which the high resistance to chemical and thermal attack makes a wet-chemical decomposition difficult,3 are decomposed quantitatively by reaction with elemental fluorine within a short period of time. The reaction products are mainly fluorides in high oxidation states (Figure 1).2 Many nonmetals form volatile fluorides which can be directly detected by IR spectrometry or gas mass spectrometry. Additionally, gas mass spectrometry is applicable for the detection of the reaction products of oxygen and nitrogen, which are released mainly in the form of the IR-inactive elements. Moreover, the slightly volatile (1) Jacob, E. Fresenius’ Z. Anal. Chem. 1989, 333, 761-62. (2) Jacob, E. Fresenius’ Z. Anal. Chem. 1989, 334, 641-42. (3) Broekaert, J. A. C.; Garten, R. P. Analytiker-Taschenbuch; Springer: Berlin, 1996; Vol. 14, Chapter 6.

Analytical Chemistry, Vol. 72, No. 16, August 15, 2000 3875

Figure 2. Experimental setup.

fluorides formed by many transition metals can be detected by mass spectrometry, when using a heated inlet system. In principle, the high dynamic range of mass spectrometers enables them to perform both, determinations of main components as well as trace analyses. Whereas IR spectrometry combined to a combustion with elemental fluorine has already been successfully applied to determine main and minor components in metallic and ceramic samples with high precision,4,5 the special advantages of mass spectrometric detection have not been fully exploited up to now. Because of the corrosive nature of inorganic fluorides, a special quadrupole mass spectrometer with electron impact ionization and with a fluorine-resistant gas inlet system needs to be used. Quadrupole mass spectrometers have the advantage of being robust, rather inexpensive, and fast in scanning, but they are restricted by a low resolution. The latter is a major disadvantage, since ionization by electron impact in itself results in mass spectra containing many signals. For complex mixtures, such as the fluorination products, these signals often cannot be unambiguously assigned to fragments. The determination of nitrogen in the presence of silicon, for example, is hampered by the fact that their fluorination products, N2 and SiF4, build fragment ions (e.g., 14N2+ and 28Si+, m/z ) 28), which spectrally interfere. Also, traces of boron in a silicon-containing matrix cannot be determined, as the fragment ions of BF3 (10BFn+, 11BFn+, n ) 1-3) and SiF4 (29SiFn-1+, 30SiF + n-1 ) have identical nominal masses. These limitations can be overcome either by the use of highresolution mass spectrometry or by a separation of the fluorination products. In this study, carrier gas distillation is shown to be useful for the separation of many of the volatile fluorides. The separation is performed in a simple and inexpensive setup, which is directly coupled to the fluorination apparatus and the mass spectrometer. This approach is shown to be sufficient to reduce isobaric interferences and to make the combination of fluorine combustion and gas mass spectrometry suitable for a determination of silicon, boron, and nitrogen as main and minor components in siliconcontaining ceramics. (4) Jacob, E.; Harris, M. Nichtmetalle in Metallen ′90; DGM Verlag: Oberursel, Germany, 1990; 79-85. (5) Kaiser, G.; Mayer, A.; Friess, M.; Riedel, R.; Harris, M.; Jacob, E.; To¨lg, G. Fresenius’ J. Anal. Chem. 1995, 352, 318-26.

3876

Analytical Chemistry, Vol. 72, No. 16, August 15, 2000

EXPERIMENTAL SECTION Reagents. The fluorine gas was supplied and has been purified6 by MAN, Munich, Germany. CoF3 with a purity of 99% (Alfa, Deisenhofen, Germany) was used as an alternative fluorinating agent. For the fluorination experiments, graphite (spectroscopically pure, Ringsdorff, Bonn, Germany), BN and tungsten (Riedel de Hae¨n, Seelze, Germany), silicon wafer material (Elmos, Dortmund, Germany), P3N5 (Hoechst, Hu¨rth-Knappsack, Germany), KBr, NaBF4, sulfur and selenium (pro analysi, Merck, Darmstadt, Germany), germanium (99,999%, Alfa, Deisenhofen, Germany), SiC (A10, Starck, Goslar, Germany), SiC (S 933, Elektroschmelzwerk Kempten, Germany), Si3N4 (Sigma, St. Louis, MO), and SiO2 (IRMM-018, IRMM, Geel, Belgium) have been used. Apparatus. In Figure 2 a diagram of the experimental setup is shown. The apparatus consists of three parts, namely, a unit for the sample decomposition by fluorination, a unit for the separation of volatile fluorination products by carrier-gas distillation, and a quadrupole mass spectrometer for the detection of volatile reaction products. Features of each of these parts of the setup are given below. Fluorination. A detailed description of the fluorination apparatus with a unit for fluorine dosing, a vacuum, and an absorber system and a combustion cell is given in refs 7-9. For a decomposition with elemental fluorine, a cylindrical reactor made of pure nickel (DIN 17740, Deutsche Nickel AG Schwerte, Germany) with a volume of 10 mL is used. Up to 20 mg of a sample are weighed in a nickel crucible and placed in the reactor. The reactor is sealed by a copper gasket and connected to the fluorination apparatus via a shutter valve made of monel (Nupro). Subsequently, the system is evacuated. The fluorine is transferred from a storage vessel into the reactor by cooling the reactor with liquid nitrogen; thereby, dosing is performed under pressure control by using a shutter valve (Nupro) controlled by a pressure transducer (MKS, Munich, Germany). The reactor can rapidly be heated to 750 °C (6) Jacob, E.; Christe, K. O. J. Fluorine Chem. 1977, 10, 169-172. (7) Richts, U.; Garten, R. P. H.; Jacob, E.; To ¨lg, G. Fresenius’ J. Anal. Chem. 1994, 349, 251-53. (8) Kipphardt, H.; Garten, R. P. H.; Jacob, E.; Broekaert, J. A. C.; To¨lg, G. Mikrochim. Acta 1997, 125, 101-05. (9) Kipphardt, H.; Hendriksen, F.; Valkiers, S.; Taylor, P. D. P.; De Bie`vre, P. Int. J. Mass Spectrom. 1999, 189, 27-37.

with the aid of a high-frequency furnace (Linn, Hirschbach, Germany). CF4, BF3, SiF4, PF5, SF6, SeF6, GeF4, and WF6 were prepared by combusting 5-10 mg of graphite, pieces of a silicon-wafer, BN, P3N5, and Ge, Se, and W powder with a 2-fold excess of fluorine at 350 °C within 10 min, respectively. For the determination of the main constituents, silicon and nitrogen, 5-10 mg of Si3N4, SiC, and SiO2, respectively, together with 5 mg of elemental sulfur (as internal standard) were decomposed with a 2-fold molar excess of fluorine at 350 °C within 10 min. For the determination of nitrogen in Si3N4, CoF3 was also used as a fluorinating agent. At high temperatures, CoF3 namely releases F2 in a reversible reaction (CoF3 a CoF2 + 1/2 F2). For the fluorination with CoF3, the sample was heated together with a molar excess of CoF3 in the nickel reactor. Since relatively high temperatures are needed to release fluorine, even at moderate pressure (70 kPa at 514 °C10), and long reaction times are required to reach equilibrium, the decomposition was performed at 650 °C and a longer decomposition time (30 min) was used. For the determination of boron in a SiC powder, the silicon signal was used for internal standardization. To avoid losses of BF3 due to the formation of nonvolatile complexes11 (e.g., NaF(s) + BF3(g) f NaBF4(s)), 6 mg of germanium powder were added to 8 mg of SiC before fluorination. GeF4 is even a stronger Lewis acid than BF3 (2 NaBF4(s) + GeF4(g) f Na2GeF6(s) + 2 BF3(g)) and, accordingly, prevents the formation of nonvolatile complexes with BF3. Safety Considerations. Fluorine is a highly toxic and corrosive gas. To avoid any contact with fluorine and its reaction products, which are also often toxic and corrosive, the fluorination was performed in an all-metal apparatus. Nickel, copper, and stainless steel soon are passivated by a fluoride layer, which prevents a complete fluorination of these materials. Aluminum builds up a fluoride passive layer as well, but its use is limited by its relatively low melting point of 660 °C. The tightness of the apparatus, and especially of the nickel reactor, after assembling was carefully checked with the aid of an ion-getter pump. Except during the combustion process itself, fluorine was stored and handled only at reduced pressure. Additionally, a ventilation hood was placed above the setup so as to suck off gases, which might have escaped from the reactor in the case of accident. Further, for each individual fluorination, only small amounts of fluorine (max, 4 mmol) were used. Finally, a cryosorption pump filled with activated TiO2 was used to pump off excesses of fluorine and corrosive fluorides and to make them harmless by the formation of titanium fluoride oxide compounds and O2. From our experience (more than 1000 fluorinations) with these precautions, the fluorination technique is considered safe. Removal of Excess Fluorine. Elemental fluorine was found to considerably lower the sensitivity of the mass spectrometer and to have an influence on the intensity of the nitrogen signal. Therefore, the excess of fluorine was removed before the separation and the detection of the fluorination products. The reaction products were frozen by cooling the reactor with liquid nitrogen, and the remaining gases were pumped into the TiO2 absorber. (10) Christe, K. O.; Wilson, R. D. Inorg. Chem. 1987, 26, 2554-56. (11) Jacob, E. U.S. Patent 5,081,043, 1992.

However, due to the similarity in vapor pressure, a separation of F2 and N2 or O2 cannot be realized in this way. For a selective removal of F2, the reaction products were expanded into a second reactor filled with 500 mg of KBr.12 Here, the remaining F2 could be quantitatively removed at 300 °C within 15 min through the reaction: KBr + 1/2 F2 f KF + 1/2 Br2. After fluorinations with CoF3 as a reversible fluorine store, no excess of fluorine was found when the reactor was kept at room temperature. Calibration. Gaseous standards for calibration were prepared either by a fluorination of mixtures of solids or by mixing the pure gases. For the determination of silicon with sulfur as internal standard, a calibration was performed by a fluorination of mixtures of elemental silicon and elemental sulfur. To determine boron in SiC, a calibration was done by fluorinating mixtures of NaBF4, elemental silicon, and germanium powder. Since the small amounts of NaBF4 required could not be weighed with a satisfactory precision, they were added as an aqueous solution which subsequently was dried. For the determination of nitrogen with sulfur as the internal standard, synthetic gas mixtures of N2 and SF6 were prepared by a successive pressure-controlled release of the pure gases into a gas store. At lower concentrations, exponential dilution13 was used for the preparation of the gas mixtures. Carrier Gas Distillation. Since most of the inorganic fluorides react with the materials typically used for the packing of GC columns, a low-temperature distillation in an all-metal apparatus was used for the separation of the fluorination products. The carrier gas distillation system used in this work is similar to the one proposed by Cady et al.14 and by Rudzitis.15 However, here a column without packing was used. The separations were performed in a spirally wound tube of stainless steel applying a temperature gradient and slowly raising the temperature of the whole column. The temperature characteristic was realized by subsequently cooling the spiral with liquid nitrogen, emptying the Dewar flask, and replacing it immediately around the column. The fluorination products, which were released according to their vapor pressure, were directly transported into the mass spectrometer with the aid of an argon carrier gas flow. The distillation was performed with a low carrier gas flow and a low pressure, as required for the on-line mass spectrometric measurements. Further details on the separation procedure are given in ref 16.16 Detection. The quadrupole mass spectrometer (QMS 420, Balzers/Li) with its fluorine-resistant gas inlet system and electron impact ionization has been previously described in detail by Kipphardt et al.8 Ion currents with up to eight defined m/z ratios can be recorded, time-dependent, in the peak-jumping mode. To correct for short time fluctuations, ion currents for each m/z ratio were measured twice in a symmetrical order. For quantitative determinations, peak areas were measured. (12) Brenninkmeijer, C. A. M., Ro ¨ckmann, T. Rapid Commun. Mass Spectrom., 1998, 12, 470-83. (13) Ritter, J. J.; Adams, N. K. Anal. Chem. 1976, 48, 612-619. (14) Cady, G. H.; Siegwarth, D. P. Anal. Chem. 1959, 31, 618-20. (15) Rudzitis, E. Anal. Chem. 1967, 39, 1187-88. (16) Russe, K.; Broekaert, J. A. C. Fresenius’ J. Anal. Chem. 1998, 361, 582-84.

Analytical Chemistry, Vol. 72, No. 16, August 15, 2000

3877

Figure 3. Influence of the temperature (monitored at the bottom of the spiral) on the ion currents for different fragments of fluorides (a) and their vapor pressure (b).

RESULTS AND DISCUSSION Separation of Fluorination Products. First, the separation system was tested by distilling different fluorides independently. The ion currents for the respective most abundant fragments were recorded and are displayed in Figure 3a as a function of the temperature monitored at the bottom of the separation tube, which is the coldest part. The shape of the signals obtained corresponds to the shape of the respective vapor-pressure curves17 (part b, Figure 3). Only the shape of the signal corresponding to CF4 differs, as CF4 cannot be condensed completely at the temperature of -196 °C (boiling point of liquid nitrogen). Provided the mixtures of fluorination products behave as ideal gases, the results obtained for the single components allow one to predict whether the respective fluorides can be separated completely by the carrier gas distillation procedure. However, even components with similar boiling points can be at least partially separated, and their existence can be identified on the basis of the peak shapes. Since F2, N2, and O2 cannot be condensed by means of liquid nitrogen, these elements pass the spiral at once without being retained at all. After a decomposition of Si3N4 with sulfur as the internal standard and a subsequent removal of the excess of fluorine by a reaction with KBr, it was found from the mass spectrum of the reaction products (part a, Figure 4) that N2, SiF4, SF6, and Br2 have been formed. A chromatogram illustrating the separation (17) Landolt, H.; Bo ¨rnstein, R. Zahlenwerte und Funktionen aus Naturwissenschaft und Technik; Springer: Berlin, 1960; Vol. II, 2a.

3878 Analytical Chemistry, Vol. 72, No. 16, August 15, 2000

Figure 4. (a) Mass spectrum obtained during the removal of the excess of fluorine from fluorination products with KBr after a decomposition of Si3N4 with S as the reference substance. (b) Ion currents for species with the nominal masses 28 (s), 85 (- • - •), 127 (- - -), and 160 (• • • •) during the separation of N2, SiF4, SF6, and Br2.

of these products by carrier gas distillation is given in part b of Figure 4. During the separation, the temperature measured at the bottom of the column continuously rose from -196 °C up to about -30 °C. It was found that the chronological order of the signals observed corresponds to the order expected from vapor-pressure curves of the respective reaction products and that N2 and Br2 are separated quantitatively from the other reaction products. The signals for SiF4 and SF6 partially overlap, which is understandable from the small difference in their respective vapor pressures. However, the compounds seem not to influence each other. As expected, a separation of the reaction products of SiO2 leads to a similar chromatogram with a signal for O2 instead of N2. The mass spectrum of the fluorination products of SiC (Figure 5, part a) shows a great number of signals stemming from SiF4

Table 1. Determination of Silicon as a Main Constituent in Si3N4, SiC, and SiO2 mass abundance of silicon method fluorine volatilization, carrier gas distillation + mass spectrometry fluorine volatilization + FTIR spectrometry pressure digestion + ICP-OES stoichiometric value

Si3N4 60 ( 2%

SiO2 46 ( 2%

69 ( 1%

46.7%

67.8 ( 0.2 5,a 68.8 ( 0.2 5,a 55-62 5,a 56-63 5,a 70.0%

60 ( 1%18 60.1%

SiC (S 933)

a These results were obtained for two powders of SiC different from the sample used in this work and therefore can only serve to indicate the precision attainable by the different methods.

Figure 5. (a) Mass spectrum of the decomposition products obtained by the fluorination of a powder of SiC. Nominal masses, for which signals for the fragments BFn+ (n ) 0-3) are expected, are marked by an (*). (b) Ion currents for species with the nominal masses 49 (s), 85 (• • • •), and 129 (- • - •) during a separation of BF3, SiF4, and GeF4, obtained after a decomposition of SiC with an addition of Ge for a quantitative mobilization of BF3

and various carbon fluorides, which makes it impossible to unambiguously identify whether signals for the fragments of small amounts of BF3 are also present. When using germanium in order to obtain a quantitative mobilization of BF3, the spectrum becomes even more complex, due to additional signals for fragments of GeF4. A chromatogram obtained during the carrier gas distillation of a mixture of BF3, SiF4, and GeF4 is given in Figure 5b. In this case, the chromatogram shows discrepancies from ideal behavior, since, in addition to the release expected according to the vaporpressure curve, a fraction of BF3 occurs simultaneously with GeF4. A third fraction is released by desorption, when the column is brought back to room temperature after 40 min. Furthermore, BF3 and SiF4 are not completely separated by distillation, which leads to overlapping signals of 11BF2+ and 30SiF+ at m/z ) 49. A comparison of the peak shape with the one for a noninterfered signal of a SiF4 fragment (i.e., 28SiF3+ at m/z ) 85) nevertheless allows one to identify and to determine even small amounts of BF3 in the presence of SiF4.

This indicates that carrier gas distillation, in principle, is of use for the separation of various fluorination products. However, for each particular case, special attention has to be paid to possible interactions between the different fluorination products and between the fluorination products and the walls of the apparatus. Calibration Curves and Detection Limits. Calibration curves obtained after a decomposition of mixtures of solids, as well as after mixing pure gases, were found to be strictly linear in the ranges of concentrations of interest. Calibration by the decomposition of mixtures of solids has the advantage that calibration standards and samples are treated exactly in the same way, and thereby, influences due to the decomposition step and the transport of the gases (stemming, e.g., from interactions with apparatus and reactor surfaces or memory effects) are considered. On the other hand, a direct mixing of pure gases is much less time-consuming. To evaluate the power of detection of the method, a calibration for the determination of N2 in SF6 by exponential dilution was used. For these rather inert gases, a detection limit of 47 µg of SF6/gram of N2 (based on a 3σ concept) was found. Taking into account that a total amount of only 1 µmol of gas was used for one separation, this value corresponds to an absolute amount of 9 × 10-12 mol of SF6. Relative standard deviations for signal ratios are typically 1-2%, and limitations in the precision were found to be mainly due to the separation step. The calibration for the determination of boron in silicon-containing samples leads to much higher detection limits of 350 µg of B/gram of Si or 250 µg of B/gram of SiC, respectively. Because of the rather difficult evaluation of the three signals of BF3 and the correction for the overlapping SiF4 signal, relative standard deviations amount to 6%. Moreover, high blanks for BF3, resulting from memory effects, limit the power of detection. Determination of Silicon as the Main Constituent in Si3N4, SiC, and SiO2. The applicability of the fluorination method for the quantitative analysis of refractory samples is obvious from the determination of the mass abundance of silicon in three different silicon-containing powders, namely Si3N4, SiC (S 933), and SiO2. The mean values of the results obtained (Table 1) are in good agreement. The standard deviations given result from four sample decompositions, each with a 4-fold separation and detection, (18) Mann, S.; Geilenberg, D.; Broekaert, J. A. C.; Jansen, M. J. Anal. At. Spectrom. 1997, 12, 975-79.

Analytical Chemistry, Vol. 72, No. 16, August 15, 2000

3879

Table 2. Determination of Nitrogen as a Main Constituent in Si3N4 method

mass abundance of nitrogen

fluorine volatilization, carrier gas distillation + mass spectrometry: (a) decomposition with CoF3 (b) removal of F2 with KBr carrier gas heat extraction stoichiometric value

37.4 ( 0.5% 38.9 ( 0.6% 38.2 ( 0.3%18 39.9%

respectively. These values are comparable to those obtained by a lengthy wet chemical digestion and a determination by ICPOES.5,18 For SiC, in particular, the fluorination method has the advantages of a much shorter decomposition time and an absence of losses compared with pressure digestion. Because of the need for a separation procedure and the use of small amounts of samples in gas mass spectrometry, the precision obtained is nearly 1 order of magnitude lower than the precision reported for IRspectrometric detection of fluorination products.5 Determination of Nitrogen and Oxygen as Main Constituents in Si3N4 and SiO2. Whereas N2 and O2 cannot be detected by IR spectrometry, such detection is possible by gas mass spectrometry even down to low concentrations. Both a treatment with KBr after decomposition with elemental fluorine as well as the use of CoF3 as the agent for a reversible release and uptake of fluorine can be applied for the determination of nitrogen in Si3N4 (Table 2). However, when using CoF3 for fluorination, small amounts of N2O were found to be formed. As this byproduct was not considered for evaluation, the value obtained by this method might be slightly too low. The procedure used for the determination of nitrogen is, in principle, suitable for a determination of oxygen as well. However, one must keep in mind that, when fluorinated in the presence of oxygen, some elements form oxyfluorides. In those cases, a determination of oxygen becomes more difficult, since each oxygen-containing fluorination product has to be considered. Under the experimental conditions reported, from the mass spectra no evidence for the presence of oxyfluorides of silicon, boron, sulfur, or germanium was found. A determination of oxygen was restricted either by blank contributions for oxygen in the procedure with KBr or by losses resulting from reactions with copper gaskets at the high temperatures required in decompositions using CoF3. Determination of Boron as a Minor Constituent in SiC. The results obtained for the determination of boron in a SiC powder (A10), by the method developed, as well as those obtained with ICP-OES after wet-chemical decomposition,19 with slurry nebulization ICP-OES,19,20 and with instrumental proton activation analysis,21 are listed in Table 3. The good agreement between these results demonstrates that a determination of a minor component is possible even when the measured species are not separated quantitatively from the interfering matrix species. Losses of BF3 due to the formation of tetrafluoroborate salts were (19) Docekal, B.; Broekaert, J. A. C.; Graule, T.; Tscho ¨pel, P.; To ¨lg, G. Fresenius J. Anal. Chem. 1992, 342, 113-17. (20) Broekaert, J. A. C.; Lathen, C.; Brandt, R.; Pilger, C.; Pollmann, D.; Tscho ¨pel, P.; To ¨lg, G. Fresenius J. Anal. Chem. 1994, 349, 20-25. (21) Pensis, L. Diploma Thesis, University of Gent, 1991.

3880

Analytical Chemistry, Vol. 72, No. 16, August 15, 2000

Table 3. Determination of Boron as a Minor Constituent in SiC method

boron concn [µg/g]

fluorine combustion, carrier gas distillation + mass spectrometry wet chemical decomposition + ICP-OES slurry nebulization ICP-OES

930 ( 30

instrumental proton activation analysis

920 ( 419 902 ( 920 1070 ( 1519 110021

prevented successfully in the presence of GeF4. This is a typical case showing that the high reactivity of some of the volatile fluorides must always be taken into account. Like BF3 and GeF4, the fluorides of arsenic, antimony, bismuth, niobium, and tantalum are also known to have Lewis acid properties and to form nonvolatile complexes with Lewis bases such as KF or NiF2. Taking interactions of a mixture of element fluorides a priory into account is very difficult. Therefore, the determination of these elements using the method presented here might be limited to simple systems of basically known composition. CONCLUSIONS A combination of a combustion with elemental fluorine and quadrupole gas mass spectrometry has been shown to be a feasible method for the analysis of refractory samples such as ceramic powders. Where the applicability up to now was limited by the low resolution of quadrupole mass spectrometers, a separation of the reaction products by a carrier gas distillation was found to be appropriate to decrease isobaric interferences for a number of reaction products. The on-line coupling of a decomposition with elemental fluorine, a separation of the reaction products, and mass spectrometric detection was further shown to be a practicable solution for quantitative analyses. Because of its short decomposition times, the method developed promises to be an alternative for time-consuming wet-chemical decomposition. In addition, it is possible to determine the nitrogen content after a decomposition at 350 °C. This makes the method also attractive to be used as an alternative for carrier-gas heat extraction, for which, in the case of ceramic powders, problems resulting from the high temperatures required are reported.18 For stoichiometry determinations, a still higher precision is desirable, which might be achieved by an automation of the up-to-now mainly hand-operated carrier-gas distillation procedure as well as by the use of isotope dilution mass spectrometry and simultaneously registrating mass spectrometers. The speed of analysis, which is now mainly limited by the speed of the carrier-gas distillation (each separation takes at least 20 min), might be increased by using ion-mobility spectrometry or high-speed multicapillary column chromatography for a separation of the fluorination products. ACKNOWLEDGMENT The authors thank the “Deutsche Forschungsgemeinschaft” (DFG), Bonn, for financial support through research Grant Br 932/15-1. Support on instrumentation and working materials by Balzers (Liechtenstein) and Deutsche Nickel AG (Schwerte, Germany) is gratefully acknowledged. Received for review October 22, 1999. Accepted June 6, 2000. AC991238C