Slurry sampling of silicon nitride powder combined with fluorination assisted electrothermal vaporization for direct determination of titanium, yttrium and aluminum by ICP-AES Peng Tianyou, Jiang Zucheng* and Qin Yongchao Department of Chemistry, Wuhan University, Wuhan 430072, China Received 25th November 1998, Accepted 18th May 1999 The vaporization behavior of silicon and three refractory trace elements (Al, Ti and Y) were studied in the presence and absence of a PTFE emulsion as fluorinating reagent and applying an electrothermal ICP-AES coupled system.It was found that during a 60 s ashing step at 700 °C about 90% of 100 mg of Si3N4 can be decomposed and evaporated without considerable losses of the trace elements investigated. Calibration could be carried out by the standard addition method and the calibration curve method applying spiked slurries and aqueous standard solutions with peak height intensity measurements, respectively. The detection limits varied from 0.11 mg g-1 (Al ) to 0.09 mg g-1 (Ti) with RSD 1.9–4.2%.Advanced ceramics, owing to their unique properties, are refractory carbides and eliminates memory eVects, but also markedly decreases matrix eVects and the influence of the increasingly applied in various fields of industry. For example, particle size of the samples. In the present work, we directly silicon nitride (Si3N4) ceramics have high chemical resistance determined trace amounts of refractory elements (Al, Ti and and excellent mechanical properties at high temperatures and Y) in Si3N4 ceramic powders with slurry sampling ETV-ICP- they are used as high-density, corrosion- and heat-resistant AES by using PTFE as halogenating reagent.Selective volatil- materials for turbine blades and ceramic engines. The presence ization between the analytes and the matrix was used to of metallic impurities such as Al and Ti in Si3N4 ceramics improve the sensitivity and reduce the matrix interference.adversely aVects their mechanical properties. Moreover, several substances (e.g., Y2O3) play an important role in the production of the Si3N4 ceramics. Therefore, in addition to the Experimental characterization of compact ceramics, the analysis of the basic Instrumentation and main operating conditions products is also important so as to optimize production procedures, including the exclusion of contamination at the A ca. 27 MHz ICP spectrometer with a power of 2 kW(Beijing various production stages and tailoring the properties of the Broadcast Equipment Factory, Beijing, China) and a convenmaterials by controlling the levels of impurities.Trace impurit- tional plasma silica torch were used. A laboratory-made graphite furnace vaporizer15–18 was employed as the vaporiz- ies in ceramic powder have routinely been determined by the ation device. The radiation from the plasma was focused as a use of ICP-AES1–5 and AAS.6,7 151 straight image on the entrance slit of a WDG500–1A Solid sampling ICP-AES methods can be recommended monochromator (Beijing Second Optics, Beijing, China) owing to the elimination of the time-consuming dissolution having a reciprocal linear dispersion of 1.6 nm mm-1.The procedure. Direct solid sampling techniques for ICP include evolved components were swept into the plasma excitation the direct insertion technique,8,9 slurry nebulization1–3,10 and source through a 35 cm×4 mm id polythene tube by a stream electrothermal vaporization (ETV).11,12 ETV seems to be a of argon.The transient signals were detected with an R456 very promising technique for routine applications, because of type photomultiplier tube (Hamamatsu, Tokyo, Japan) and a the inexpensive equipment, simple handling and the possibility laboratory-built direct current amplifier, and recorded with a of calibration with aqueous standards. Furthermore, selective U-135C recorder (Shimadzu, Kyoto, Japan).volatilization between the analytes and the matrix through sequential volatilization of the sample components is feasible. Standard solutions and reagents Recently, several approaches to solid sampling ETV have been described, including the application of pelletized solids,13 of Standard solutions (1 g L-1) of Al, Ti and Y were prepared from their Specpure oxides by applying a conventional method. the miniature cup technique,14 of halocarbon vapor to promote A 60% m/v PTFE emulsion (viscosity 7×10-3–15×10-3 Pa fast vaporization4 and of slurry sample injection.15–18 Of these, s) was purchased from the Shanghai Institute of Organic the last mentioned method has demonstrated its potential as Chemistry (Shanghai, China).All other reagents were of a promising method for direct solid analysis. analytical reagent grade or better. Doubly distilled, de-ionized Generally, it is diYcult to vaporize completely ceramic water and HF and HNO3 were further purified by applying powder from the ETV device because of its refractory nature.sub-boiling distillation. Si3N4 powder (d<80 mm) was provided Favorable conditions for the selective volatilization of analyte by the Northwest Iron and Steel Research Institute of China from refractory materials could be achieved with the use of a (Xi’an, China). halogenating reagent (e.g., CF2Cl2 12 and PTFE15,16). Our previous studies15–18 indicated that the direct determination of Preparation of sample trace elements in high-purity Y2O3, SiO2 and biological samples could be carried out by slurry sampling fluorination An Si3N4 powder sample (50 mg, d<80 mm) was weighed into assisted ETV-ICP-AES.The experimental results also showed calibrated flasks, 0.5 mL of 60% m/v PTFE, 0.5% agar and 0.1% Triton X-100 solutions were added and the mixture was that the addition of PTFE not only prevents the formation of J. Anal. At. Spectrom., 1999, 14, 1049–1053 1049diluted to 5 mL with water.The resulting mixture (1% m/v slurry) was dispersed with an ultrasonic wave vibrator for 20 min and the calibrated flasks were shaken prior to any sampling. Solutions containing the same amounts of Si and PTFE were prepared by dissolving the same mass of sample with HF–HNO3 (1+1) in a high pressure system at 180 °C for 12 h.2 These samples were used to study the vaporization behavior. For the standard addition method, the slurries prepared as above were spiked with appropriate amounts of aqueous multielement standard solutions; for the calibration curve method, a multi-element standard series containing PTFE (6% m/v) were also prepared from standard solutions. Analytical procedure After the plasma had stabilized, 10 mL of sample were pipetted into the furnace.After being dried and ashed, the analyte was vaporized and carried into the plasma by argon carrier gas, Fig. 1 EVect of ashing temperature on the signal intensities of Al, Ti, the emission signal from the plasma was recorded and the Y and Si (vaporization temperature: 2700 °C).Al, 0.2 mg L-1 with peak height was measured for quantification. Calibrations PTFE; Al¾, 0.2 mg L-1 without PTFE; Ti, 0.2 mg L-1 with PTFE; were carried out by the standard addition and calibration Ti¾, 0.2 mg L-1 without PTFE; Y, 0.5 mg L-1 with PTFE; Y¾, 10mg curve methods. Six replicates were analyzed by using slurry L-1 without PTFE; Si, 0.4 mg L-1 with PTFE; Si¾, 10mgL-1 and solution sampling.A similar number of experiments were without PTFE. performed using a blank solution without sample to evaluate the blank values. Results and discussion Optimization of ETV-ICP-AES system The ETV-ICP-AES operating parameters were optimized on the basis of signal-to-background ratios by using standard solutions of analytes containing 6% m/v PTFE. From a comparison with pneumatic nebulization (PN)-ICP-AES, there were no obvious diVerences in the operating parameters between PN-ICP-AES and ETV-ICP-AES with regard to rf power and observation height but the carrier gas flow rates diVered.The experimental parameters selected are given in Table 1. Typical ashing and vaporization curves for the elements of interest in standard solutions with/without PTFE are shown in Fig. 1 and 2. Our previous studies demonstrated that both the analyte and matrix could react with the pyrolysis products of PTFE in the graphite tube and be vaporized in the form of the corresponding fluorides, when the temperature in the graphite tube reaches the chemical decomposition temperature Fig. 2 EVect of vaporization temperature on the signal intensities of of PTFE (400 °C). Therefore, the addition of PTFE can Al, Ti, Y and Si (ashing temperature: 700 °C). Al, 0.2 mg L-1 with prevent the formation of refractory carbides at high tempera- PTFE; Al¾, 0.2 mg L-1 without PTFE; Ti, 0.2 mg L-1 with PTFE; tures, which has also been proved by using XRD analyses of Ti¾, 0.2 mg L-1 without PTFE; Y, 0.5 mg L-1 with PTFE; Y¾, 10mg the residues of ETV after recycling heating.16–18 As can be L-1 without PTFE; Si, 0.4 mg L-1 with PTFE; Si¾, 10mgL-1 without PTFE.seen from Fig. 1 and 2, the ashing temperatures of the same Table 1 ETV-ICP-AES operating conditions Incident power/kW 1.2 Carrier gas (Ar) flow rate/L min-1 0.5 (Al, Si); 0.6 (Ti); 0.7 (Y) Auxiliary gas (Ar) flow rate/L min-1 0.8 Coolant gas (Ar) flow rate/L min-1 16 (Al, Y, Ti); 18 (Si) Observation height/mm 12 Entrance slit width/mm 25 Exit slit width/mm 25 Wavelength/nm Al 308.215, Ti 334.941, Y 371.03, Si 251.611 Drying temperature/°C 100, ramp 10 s, holds 20 s Ashing temperature/°C 400, ramp 10 s, holds 20 s ( Ti, Si) 1200, ramp 10 s, holds 20 s (Al ) 1300, ramp 10 s, holds 20 s (Y) Vaporization temperature/°C 1500, 3 s (Si), 2340, 3 s (Ti), 2240, 3s (Al ), 2460, 4 s (Y) Clear-out temperature/°C 2700, 4 s Sample volume/mL 10 1050 J.Anal. At. Spectrom., 1999, 14, 1049–1053elements decreased significantly in the presence of PTFE. It is obvious from Fig. 2 that PTFE has a great influence on the vaporization behavior of both the analyte and matrix, and the vaporization curves reach their plateaux at lower temperature. This means that the fluorination reactions between the analytes and the pyrolysis products of PTFE are complete. In contrast, no signal plateaux were found in the tested temperature range in the absence of PTFE.Ashing time Fig. 3 shows the eVects of the ashing time on the signal intensities of the tested elements at an ashing temperature of 700 °C. It is obvious that the ashing time has no great influence on the signal intensities of Al, Ti and Y. However, the signal intensity of Si declines on prolonging the ashing time owing to the formation of gaseous SiF4 . Based on these results, there is a possibility of removing the matrix (Si) at the ashing stage by prolonging the ashing time.Fig. 4 EVect of matrix (Si) on the signal intensity of Al, Y and Ti in the presence of PTFE. Conditions: 10 mL of standard solutions with EVects of matrix concentrations of Al 0.2, Ti 0.2 and Y 0.5 mg L-1 and containing 0.6 mg of PTFE. The experimental results show that the signal intensities of the analyte and the plasma discharge were not stable with the large amounts of matrix (Si) entering the ICP with the analyte. this paper, unless stated otherwise, the PTFE concentration in Fortunately, as shown in Fig. 3, selective volatilization between the sample was 6% m/v.the matrix and the analyte could be carried out by prolonging the ashing time at an ashing temperature of 700 °C. Therefore, Comparative investigation of fluorinating vaporization behavior a 60 s ashing time was chosen to reduce the matrix influence Fig. 5 shows the typical emission profiles of Y in solution or as much as possible, and the analytical signal of Ti decreased slurry at the same concentration. In the absence of PTFE, the only slightly.Fig. 4 shows the largest tolerable matrix concensignal intensity of Y in the solution or slurry of Si3N4 was trations are 12 g L-1 for Al and Y and 20 g L-1 for Ti at an very weak, and a broad signal profile with tailing was recorded. ashing temperature of 700 °C. Obviously, the matrix inter- Furthermore, the residual signals for Y were almost the same ference observed in this work is low.A possible reason is that as the original signal. However, in the presence of the fluorin- the relative vaporization eYciency of silicon is about 90% (see ating reagent PTFE, a sharper, more intense and symmetrically Table 2) at 700 °C. The relative vaporization eYciency is shaped peak without tailing was obtained for Y, and there defined as the ratio of the signal intensity of an element at a was no memory eVect on the performance after vaporization certain ashing temperature and a 60 s ashing time to that of of Y in solution and slurry.Similar results were also observed the element at a 400 °C ashing temperature, which is discussed for Al and Ti. The possible reasons for this are that the below. Once the concentration of the matrix (Si) exceeds the temperature supplied by the conventional ETV device is not tolerance level, the signal intensity of the analyte starts to high enough to destroy and vaporize repeatedly the refractory decline.The reason may be a lack of suYcient PTFE, which ceramics, but using PTFE as a chemical modifier, the fluoride leads to incomplete vaporization or a change in the vaporizproduced from decomposition of PTFE can react very strongly ation rate of the samples. However, our previous studies17,18 with the ceramic. Hence, the addition of PTFE not only showed that the optimum concentration was 6% m/v, otherwise eliminates the diVerence in the form in which the samples the stability of the plasma decreased markedly.Therefore, in exist, but also prevents the formation of refractory carbides, promotes vaporization eYciency and improves the analytical sensitivity. Moreover, the profiles and the height of the emission signal of the analytes in the slurry are very similar to those in the solution in the presence of PTFE. Hence we can conclude that the vaporization behavior of the analytes in slurry and in solution are very similar; the standard solutions can be used for the calibration of slurry samples and the systematic errors in the analytical results can be ignored.As described above, the low level matrix eVects can be attributed to the removal of most of the matrix at the ashing stage, but once the ashing temperature exceeds 400 °C, the analyte (Ti) is partially lost owing to the formation of TiF4 with a very low sublimation temperature (284 °C). To obtain quantitative information about the vaporization processes, a solution containing PTFE (6% m/v) and Al, Ti, Y and Si (0.4 mg L-1) was prepared from the standard solutions.The signal intensities of the above elements were then determined subsequent to various ashing steps carried out at ashing temperatures of 400, 500, 700, 900, 1100 or 1300 °C for 60 s. Fig. 3 EVect of the ashing time on the signal intensities of the four The relative vaporization eYciencies of the analytes are given elements investigated in presence of PTFE at an ashing temperature in Table 2.As can be seen, when the ashing temperature is of 700 °C. Conditions: 10 mL of standard solutions with concentrations increased from 400 to 1100 °C, the losses of Al and Y are of Al 0.2, Ti 0.2, Y 0.5 and Si 0.4 mg L-1 and containing 0.6 mg of PTFE. negligible, but the relative vaporization eYciencies of Ti and J. Anal. At. Spectrom., 1999, 14, 1049–1053 1051Table 2 EVect of ashing temperature on the relative vaporization eYciency of analytes (mean±s, n=3) with an ashing time of 60 s Element 400 °C 500 °C 700 °C 900 °C 1100 °C 1300 °C Al 100.1±3.0 98.9±2.6 98.6±2.2 95.2±2.4 93.7±2.2 87.2±2.0 Ti 99.3±2.5 91.7±2.0 85.3±1.9 73.2±2.2 50.5±2.0 42.3±1.7 Y 99.8±2.7 98.9±2.3 98.3±2.1 94.2±2.3 93.1±1.9 90.9±1.3 Si 98.0±3.2 43.2±3.0 10.3±2.9 — — — Fig. 5 Typical signal profiles of Y using a standard solution or Si3N4 slurry with the same analyte concentration and with or without PTFE: Fig. 6 Calibration curves (peak height) of Al, Ti and Y determined a, Y in solution with PTFE; b, Y in solution without PTFE; c, Y in using multi-element standard solutions containing PTFE (6% m/v).slurry with PTFE; d, Y in slurry without PTFE; a¾, b¾, c¾ and d¾, residual signals detected during the second heating period. Sample analysis Si decrease. Table 2 shows that when the ashing temperature and time are chosen as 700 °C and 60 s, respectively, about The proposed method was applied to determine refractory 90% of the silicon nitride is released without significant losses elements (Al, Y and Ti) in a real sample (Si3N4).The sample of the analyte of interest. In other words, under the above was also analyzed by dissolution-based PN-ICP-AES. The conditions, the determination of Al, Ti and Y in Si3N4 powders analytical results (Table 4) are in good agreement. is reliable and accessible. Analytical characteristics Conclusions The detection limit, the lowest concentration level that can be The use of PTFE not only eVectively destroys the skeleton of determined to be statistically diVerent from a blank, is defined Si3N4, prohibits the formation of refractory carbides, removes as three times the within-batch standard deviation of a signal matrix and memory eVects and promotes the vaporization blank determination, corresponding to the 99% confidence eYciency, but eVects selective volatilization between the analevel.The blank signal was the signal obtained from the lytes and the matrix.Moreover, the experimental results also vaporization of 10 mL of 6%m/v PTFE slurry containing indicated that the use of PTFE can eliminate eVectively the 0.05% agar and 0.01% Triton X-100. Table 3 presents the diVerence in the form in which the sample exists, hence detection limits calculated for a ca. 10gL-1 Si3N4 slurry, calibration could be carried out with aqueous standards. This compared with those obtained using conventional ETV and method can be expected to become a routine method for the PN-ICP-AES. As can be seen, compared with conventional direct analysis of high purity and refractory ceramic materials.ETV and PN-ICP-AES, the detection limits for Al, Y and Ti are improved to diVerent extents. The precision (RSD) for six replicate measurements were in the range 1.9–4.2%. Acknowledgements Fig. 6 shows the calibration curves for Al, Ti and Y determined by ETV-ICP-AES using multi-element standard solu- This work was supported by the National Natural Science Foundation and the Education Commission Foundation of tions and PTFE as chemical modifier.The graphs are linear over a concentration range of three orders of magnitude. China. Table 3 Comparison of detection limits and RSDs PN-ICP-AES/mg l-1 ETV-ICP-AES/mg g-1 Wavelength/ Element nm Solution19 Slurry20 Literature21 This worka RSDa (%) Al 308.215 50.0 120.0 17 0.11 4.1 Ti 334.941 12.0 12.0 0.95 0.09 1.9 Y 371.03 10.0 12.0 — 0.10 4.2 aSampling volume 10 mL, six replicate measurements, concentrations of Al, Y and Ti 0.2, 0.2 and 0.5 mg L-1, respectively. 1052 J. Anal. At. Spectrom., 1999, 14, 1049–1053Table 4 Concentrations of Al, Y and Ti in Si3N4 powder determined by slurry sampling and fluorination assisted ETV-ICP-AES Standard addition Calibration curve Calibration curve PN-ICP-AESb/ Element methoda/mg g-1 methoda/mg g-1 methodb/mg g-1 mg g-1 Al 10.6±1.2 10.2±1.0 10.5±1.0 11.0±0.9 Y 7.20±1.1 7.42±0.79 7.50±0.55 7.80±0.49 Ti 3.02±0.40 3.27±0.27 3.25±0.21 3.32±0.3 aSlurry sampling for ETV.bDissolved in HF–HNO3 and dispensed into the ETV. 12 P. Barth, S. Haupton and V. Krivan, J. Anal. At. Spectrom., 1997, References 12, 1359. 13 V. Karanassios, J. M. Ren and E. D. Salin, J. Anal. At. Spectrom., 1 I. Varga, Gy. Zaray and J. Szepvolyi, Mikrochim. Acta, 1989,III, 1991, 6, 527. 381. 14 I. Atsaya, T. Itoh and T. Kyrotaki, Spectrochim. Acta, Part B, 2 Gy. Zaray, A. Farakas and I. Varga, Acta Chim. Hung., 1991,128, 1991, 46, 103. 489. 15 T. Y. Peng and Z. C. Jiang, Anal. Sci., 1997, 13, 595. 3 Gy. Zaray, I. Varga and T. Kantor, J. Anal. At. Spectrom., 1994, 16 T. Y. Peng, Y. C. Qin and Z. C. Jiang, J. Rare Earths, 1997, 9, 707. 15, 149. 4 T. Kantor and Gy. Zaray, Fresenius’ J. Anal Chem., 1992, 342, 927. 17 Z. C. Jiang, B. Hu, Y. C. Qin and Y. Zeng, Microchem. J., 1996, 5 S. Mann, D. Geilenberg, J. A. C. Broekaert and M. Jansen, 53, 326. J. Anal. At. Spectrom., 1997, 12, 975. 18 B. Hu, Z. C. Jiang, Y. C. Qin and Y. Zeng, Anal. Chim. Acta, 6 K.-Ch. Friese and V. Krivan, Anal. Chem., 1989, 335, 637. 1996, 319, 255. 7 T. Nakamura, Y. Noike, Y. Koizami and J. Sato, Analyst, 1995, 19 M. Franek and V. Krivan, Fresenius’ J. Anal Chem., 1992, 342, 120, 89. 118. 8 Gy. Zaray, J. A. C. Broekaert and F. Leis, Spectrochim. Acta, Part 20 R. Lobinski, W. Van Brom, J. A. C. Broekaert, P. Tschopel and B, 1988, 43, 241. G. Tolg, Fresenius’ J. Anal Chem., 1992, 342, 563. 9 M. Reisch, H. Nickel and M. Mazurkiewicz, Spectrochim. Acta, 21 Gy. Zaray, T. Kantor, G. WolV, Z. Zadgorska and H. Nickel, Part B, 1989,44, 307. Mikrochim. Acta, 1992, 107, 345. 10 B. Raeymaekers, T. Graule, J. A. C. Broekaert, F. Adams and P. Tschopel, Spectrochim. Acta, Part B, 1988, 43, 1923. 11 L. Ebdon and P. Goodall, Spectrochim. Acta, Part B, 1992, 47, 1247. Paper 8/09221D J. Anal. At. Spectrom., 1999, 14, 1049–1053 1053