Anal. Chen. 1995,67,354-359
Analysis of Silicon Nitride Powders for AI, CryCu, Fe, K, Mg, Mn, Na, and Zn by Slurry=Sampling Electrothermal Atomic Absorption Spectrometry Klausehristian Friese and Viliam Krivan* Sektion Analytik und Hochstreinigung der Universitat Ulm, 0-89069Ulm, Germany
A slurry-samplingelectrothermal atomic absorption spectrometry method for the determination of Al, Cr, Cu, Fe, K, Mg, Mn, Na, and Zn in powdered silicon nitride is described. The atomization behavior of Cr, Cu, Fe and Mn in aqueous slurries could be improved by using a mixture of ( W ) H z P 0 4 and M g ( N 0 3 ) ~as chemical modifier. Radiotracer experiments using in situ labeling of the matrix with 31Si showed that silicon was retained up to -80%after the atomization stage at 2200 "Cbut it could be removed almost quantitatively in the clean-out stage at 2500 "C. The modifier facilitated this removal. For all elements, excluding Cu and Mn, quantification by a calibration curve using aqueous standards was possible. As a result of reduction of the blank by use of this technique, the limits of detection of contamination-risk elements such as K, Mg, Na, and Zn could be improved by factors between 8 and 100 over the solution technique, reaching for these elements 4, 7, 12, and 1 ng g-l, respectively. Upon comparison of the results with those obtained by solution electrothermal atomic absorption spectrometry and neutron activation analysis, good accuracy with the developed slurry sampling technique was shown. Due to its special properties including firmness combined with low density, hardness, chemical and thermal stability, low thermal expansion coefficient, and high conductivity resistance, silicon nitride has made great advances as a ceramic material.' The spectrum of applications ranges from engine parts, bearings, cutting tools, and crucibles to fusion reactor technology.lt2 In microelectronics, silicon nitride is used for patterning of substrate films and for etching, insulating, and encapsulation of integrated circuit^.^ Many of the desired properties of silicon nitride are infiuenced by trace impurities. Alkali ions are known to decrease the resistance of silicon nitride to oxidation by enhancing the mobility of oxide ions.4 Aluminum, magnesium, and iron infiuence the sintering behavior of silicon nitride powder^.^ Calcium, lithium, and sodium decrease the mechanical stability of silicon nitride ceramics at elevated temperatures6 These effects stress the necessity of appropriate analytical methods for the determination of trace impurities in powdered silicon nitride. (1) Wotting, G.; Ziegler, G. Sprechsaal 1986,119 (4), 265-271. (2) Wotting, G.; Ziegler, G. Sprechsaal 1990,123 (ll), 1102-1113. (3) Gmelin Handbook of Inorganic Chemistry, 8th ed.; Springer Verlag: Berlin, 1991: Silicon, System No. 15, Supplement, Vol. B5c, pp 3-80. (4) Riley, F. L. Sprechsaal 1985,118 (3), 225-233. (5) Evans, J. R G.; Moulson, A J. 1.Mater. Sci. 1983,18, 3721-3728. (6) Engel, W. PowderMetall. Int. 1978,10 (3), 124-127.
354 Analytical Chemistry, Vol. 67, No. 2, January 15, 7995
Silicon nitride has been analyzed after wet chemical or fusion decomposition by various methods including flame7and electrothermal atomic absorption spectrometry (ETAAS),8,9 stripping voltametry,10inductively coupled plasma and direct current plasma atomic emission spectrometry (ICP-AESand DCP-AES) ?J1-13 and inductively coupled plasma mass spectrometry (ICPMS).g Decomposition in autoclave vessels made of ITFE using a mixture of subboiled nitric and hydrofluoric acid is the method of choice. However, this digestion method has several serious disadvantages: it is time consuming and prone to contamination and requires the use of hazardous acids. Instrumental and radiochemical neutron activation analyses are powerful methods for multielement characterization of silicon nitride regarding achievable limits of detection and accuracy,14 but the dependence on nuclear reactors and radiochemical laboratories seriously limits the applicability of these methods to routine analysis. Different approaches for avoiding the digestion of silicon nitride have been tried. Nebulization of slurries for ICP-AES12J5avoids sample decomposition but can only be applied to analysis of refractory powders with particle sizes of less than 10 pm and has rather poor limits of detection. Slurry sampling ETAAS has only been applied to the determination of chromium in silicon nitride powders.16 However, this technique has been successfully used for analysis of other ceramic materials.17-19 This paper describes the application of slurry-sampling ETAAS to the determination of nine relevant impurities in silicon nitride powders. Limits of detection, calibration by using aqueous standard solutions, quality of peak shape, reproducibility of the absorbance signal, and accuracy were the most important factors considered in the development of this method. (7) Gorgenyi, T.; Ziegler, G. Ber. Dtsch. Keram. Ges. 1990,67 (6), 252-256. (8) Grossmann, 0.; Muller, E. In CAS 6. Colloquium Atomspektrometrische Spurenanalyfik; Web, B., Ed.; Bodenseewerk Perkin-Elmer: Oberlingen, Germany, 1991; pp 691-697. (9) Franek, M.; Krivan, V.; Gercken, B.; Pavel, J. Mikrochim. Acta 1994,113, 251-259. (10) Kaplin, A A; Pichgina, V. M. Zh.Anal. Mzim. 1984,39, 664-670. (11) Ishizuka, T.; Uwamino, Y.; Tsuge, A.Bunseki Kakagu 1984,33,486-490. (12) Grade, T.; von Bohlen, A; Broekaert, J. A; Grallath, E.; Glockenkhper, R; Tschopel, P.; Tolg, G. Fresenius J. Anal. Chem. 1989,335,637-642. (13) Nathansohn, S.; Czupryna, G. Spectrochim. Acta 1983,388, 317-322. 345-358. (14) Franek, M.; Krivan, V. Anal. Chim. Acta 1992,264, (15) Ziray, G.; Varga, I.; Khtor, T. J. Anal. Atom. Spectrom. 1994,9,707-712. (16) Muller, E. Analytiktrefen 1990Atomspektroskopie, Neubrandenburg, Germany, November 5-9, 1990; Poster 3. (17) Hauptkom, S.: Schneider, G.; Krivan, V. J. Anal. Atom. Spectrom. 1994,9, 463-468. (18) Dofekal, B.;Krivan. V. J. Anal. Atom. Spectrom. 1992,7, 521-528. (19) Hauptkom, S.; Krivan, V. Spectrochim. Acta 1994,498, 221-228. 0 1995 American Chemical Society 0003-2700/95/0367-0354$9.00/0
Table 1. Experimental Parameters for Analysis of Powdered Slllcon Nltrlde by Slurry Sampling ETAAS
element experimental conditions
Al
wavelength (nm) spectral bandwidth (nm) pyrolysis tempu ("C) atomization tempb ("C) gas flow during atomization (mL min-l) modifer b g of POd3-/pg of Mg(N03)~) additional drying step linear working range (s) charact mass @g) method of addition aq std calibration
309.3d 0.70 1300 2300 gas stopd -/lo no 0.05
357.9 0.70 1400 2300 200/10
yes
33 33
7.3 7.3
cu
Fe
K
Mg
Mn
Na
Zn
324.8 0.70 1000 2300/2400"
766.5 0.70 1100 1800 gas stop
Yes
0.06
no 0.30
279.5 0.20 1300 2200/23W gas stop 100/5 Yes 0.12
213.9 0.70 700 1800 gas stop -/5
yes
285.2 0.70 1100 2000 250' 70/Yes 0.18
589.0 0.20
100/5
373.7f 0.20 1200 2100/2200e gas stop 100/5
23 18
126 120
2.8 3.0
3.0 2.8
8.3 6.6
5.3 5.2
Cr
gas stop gas stop 0.35
0.09
-/-
lo00 1800 25of
-/-
no 0.25
no
0.12
1.1 1.1
Ramp time 10 s and hold time 20 s for charring. * Atomization with maximum ower and read time of 5 s. Additional drying performed at 180 "C with 20 s ramp and 10 s hold time. Maximum gas flow (250 mL mir-l) antless sensitive line at 257.5 nm used for determination of Al in ShN4-B. e Atomization temperature for sample portions 50.4 mg/>0.4 mg. f Less sensitive lines or maximum gas flow during atomization used for high concentrations. EXPERIMENTAL SECTION
Instrumentation. All absorption measurements were made with a Perkin-Elmer 4100 ZL spectrometer equipped with a transversely heated graphite atomizer m G A ) , an AS70 autosampler, and a U S 1 0 0 controller for the Vibracell VC 50-1 ultrasonic processor (Sonics & Materials, Danbury, CT) with a Ti-Al-V alloy probe. Tubes with integrated platforms (?art No. 504 033, Perkin-Elmer, Oberlingen, Germany) were used. Background correction was performed by using the longitudinal inverse Zeeman effect. Sample digestion was performed in the autoclave system DAB I1 in 25mL PTFE inserts (Berghof, Eningen, Germany). The distribution of sample constituents remaining on the platform after thermal pretreatment was examined with a DSM 962 scanning electron microscope (Zeiss, Oberkochen, Germany). The radioactivity of used to label the matrix, was measured with a Geiger-Muller counter. y-Rays of the radiotracers used for the adsorption study were counted with a high-resolutiony-ray spectrometer system (EG & G Ortec, Munich, Germany) equipped with an intrinsic germanium detector. Reagents. Silicon nitride powders were decomposed with 40% hydrofluoric acid of Selectipur quality (Merck, Darmstadt, Germany) and concentrated nitric acid subboiled from pro analysi quality (Merck). Magnesium nitrate and calcium nitrate of Suprapur quality and ammonium dihydrogen phosphate of Selectipur quality, used as chemical modifiers, were supplied by Merck. Doubly distilled water was used for the preparation of slurries. For the determination of sodium, potassium, and magnesium, doubly distilled water was additionally purified with the MilliQ system (Millipore GmbH, Neu-Isenburg, Germany). Samples. The silicon nitride powders Si3N4-A (type LC 12 SX) and SisN4-B (type LC 12) were supplied by H. C. Starck (Goslar, Germany). Si3N4-A had a median particle size of 0.48 pm with a Gaussian grain size distribution; 90%of the sample had a particle diameter of
