Analysis of Silicon Dioxide and Silicon Nitride Powders by Electrothermal Vaporization Inductively Coupled Plasma Atomic Emission Spectrometry Using a Tungsten Coil and Slurry Sampling PETER BARTH SUSANNE HAUPTKORN AND VILIAM KRIVAN* Sektion Analytik und Ho� chstreinigung Universita�t Ulm D-89069 Ulm Germany Slurry sampling in combination with ETV-ICP-AES was employed for the direct determination of trace amounts of impurities in silicon dioxide and silicon nitride. The ETV device consisted of a double layer tungsten coil in a quartz apparatus. Spectral interferences and background emission caused by tungsten ablation of the coil were reduced by coating the coil with tungsten carbide. The background was measured either with a high-purity sample the suspension medium or close to the analyte emission line depending on matrix and analyte or it was calculated using relative emission intensities of tungsten.The concentrations of Al B Be Ca Cd Co Cr Cu Fe Mg Mn Ni Pb and Zn were measured simultaneously whereas K and Na were determined in the sequential mode. Calibration was performed using the standard additions method. The accuracy was checked by detection between 0.035 (Mg) and 130 mg g-1 (B) and between comparison with the results of independent methods. Limits of 0.01 (Be Mg) and 34 mg g-1 (B) were achieved in silicon dioxide and silicon nitride respectively. Keywords Inductively coupled plasma atomic emission spectrometry; electrothermal vaporization; tungsten coil; slurry sampling; silicon dioxide; silicon nitride Conventional solution techniques have serious limitations particularly for the analysis of refractory inorganic materials.The decomposition of these matrices is often time consuming involves highly toxic reagents such as hydrofluoric acid and is an important source of systematic errors caused by blanks or analyte losses. For these reasons methods for the direct analysis of solid samples are of great interest. Solid sampling techniques that have been developed for inductively coupled plasma atomic emission spectrometry (ICP-AES) include the direct insertion of solids,1–4 slurry nebulization,5–8 laser ablation,9–11 arc and spark erosion12–14 and electrothermal vaporization (ETV).15–25 The ETV technique oVers a number of advantages such as a high sample introduction eYciency low sample consumption the possibility of a simple in situ analyte–matrix separation and the use of aqueous standard solutions for calibration.Moreover the simple handling the comparatively inexpensive equipment and the ease of automation make it well suited for routine analysis. However its applicability can be limited by analyte vapour condensation on the cold walls of the transport tubing interferences caused by the material of the vaporizer or the sample matrix and the insuYcient data processing capabilities of most commercial spectrometers and their respective software concerning short transient signals. Graphite tubes are the most common vaporizers for ETV.26–38 As they are widely used in ETAAS their characteristics are well known and ETAAS atomizers can easily be modified to work as vaporizers for ETV.The furnace material Journal of Analytical Atomic Spectrometry December 1997 Vol. 12 (1359–1365) Instrumentation The instrumental set-up has been described in detail in a previous paper.48 All ICP-AES measurements were performed carbon causes no spectral interferences in AES measurements and its reducing properties can support the volatilization of some analytes. Furthermore the tubes allow the introduction of large sample amounts. Nevertheless they have several disadvantages including memory eVects caused by the formation of stable carbides and the necessity of strong and expensive power supplies to attain the temperatures needed for the vaporization step.The latter is also true for furnaces made of refractory metals mostly tungsten. Metal filaments on the other hand which consume less power only allow sample volumes of typically 3–5 ml. Moreover the laboratory-made vaporizers often suVer from low reproducibility. Double layer tungsten coils as manufactured for halogen lamps form simple and inexpensive but nonetheless eYcient ETV devices.39–45 They are produced in large numbers with highly reproducible physical properties allow high heating rates and temperatures of up to 3000 °C to be achieved with low cost power supplies and only a small quartz apparatus is necessary for mounting of the coil causing only low analyte vapour dilution. Vaporization from the tungsten coil has been applied in our laboratory to the analysis of silicon carbide by ETV-ICP-AES using the slurry technique.46 A very similar device was used for the trace characterization of silicon dioxide by slurry ETV-ICP-MS.47 In a previous paper we described an improved set-up for ETV-ICP-AES48 using a tungsten coil.In the present work this set-up was applied to the trace elemental analysis of silicon dioxide and silicon nitride. EXPERIMENTAL 2 2 Samples and Reagents The analysed samples were SiO -1 -325 mesh 99.9% pure lot No. X8653 (Cerac Milwaukee WI USA) SiO -2 Aerosil 200 LOS 7638605 (Novartis Basle Switzerland) and silicon nitride LC12 (H. C. Starck Goslar Germany). Particle sizes of the silicon dioxide samples were estimated by electron microscopy to be lower than 10 and 1 mm for samples SiO2-1 and SiO2-2 respectively.The silicon nitride sample had a median particle size of 0.48 mm with a Gaussian grain size distribution; 90% of the sample had a particle diameter of <0.80 mm. element stock standard solutions (1 g l-1 Merck Darmstadt Multi-element standard solutions were prepared using single Germany). The hexane used for the coating of the coil was of ‘reinst’ quality (Merck). Doubly distilled water additionally purified with a Milli-Q system (Millipore Neu-Isenburg Germany) was used throughout. 1359 on a JY-24 sequential spectrometer extended with a JY-74 polychromator (Jobin-Yvon Longjumeau France). The ETV device consisted of a double layer tungsten coil Type 64655 HLX supplied by Osram (Munich Germany) connected to a laboratory-made 0–24 V 250 W power supply.The quartz apparatus for mounting the tungsten coil and the quartz interface to the ICP were also laboratory-made. A Type HD 70 ultrasonic probe supplied by Bandelin Electronic (Berlin Germany) was used for homogenization of the slurries. The power supply the coating gas valve and the ultrasonic probe were all controlled by a portable 286-computer PP-1601 (Charisma Taiwan). Data acquisition and spectrometer control was performed on a 386-IBM clone. Procedure A description of the tungsten carbide coating (TCC) and the high voltage discharge cell (HVDC) is given in a previous paper.48 The signals of the 15 polychromator channels and the monochromator photomultiplier were recorded via a 16 channel analogue-to-digital card (resolution 12 bit).Data acquisition and processing was performed by a program written in PowerBasic (Kirschbaum Software Emmering Germany). The baseline was measured for 5 s before the vaporization step and corrected automatically. The data acquisition is also described in more detail in a previous paper.48 The entrance slit of the polychromator was calibrated using a Pb hollow cathode lamp. The monochromator was set to the peak maximum of the emission lines using continuous signals obtained by either heating the coil to 2000 °C (W) or by introducing the aerosol of a pneumatic nebulizer into the carrier gas stream (Na). Slurries were prepared by mixing up to 300 mg of silicon dioxide and 80 mg of silicon nitride with 10 ml of water in 15 ml polystyrene beakers.After conditioning and pre-coating a new tungsten coil slurry aliquots (20 ml ) were pipetted during ultrasonic agitation on the coil and the voltage-time programme was started. Depending on sample and analyte diVerent methods of background evaluation were applied which are discussed in detail under Results and Discussion.libration was performed using the standard additions method. The ETV and ICP-AES parameters are summarized in Tables 1 and 2. RESULTS AND DISCUSSION Spectral Interferences and Background Possible spectral interferences of the analyte elements caused by the matrix (silicon) or the vaporizer material (tungsten) were investigated by recording the emission spectra over a range of ±0.3 nm around the emission lines used in the determinations (see Table 2).For this purpose aqueous solutions of Si (1000 mg ml-1) W (100 mg ml-1) and a mixture of the elements to be determined (10 mg ml-1 for each element) were processed. Sample introduction was performed by pneumatic nebulization. The superimposed spectra showed no spectral interferences resulting from Si. However overlapping with tungsten emission lines was observed for the analyte lines of Mn Cr Co and Fe (see also Table 2) whereby the extent of the interference decreased from Mn to Fe. The significance of this interference depends on the amount of tungsten ablated from the coil during vaporization.As has been reported in our previous work,48 the tungsten emission signal obtained during the vaporization step when aqueous solutions are processed consists of two sections. Section 1 a sharp peak at the beginning of the vaporization stage is obviously caused by species volatilizing at comparatively low 1360 Journal of Analytical Atomic Spectrometry December 1997 Vol. 12 Table 1 ETV parameters used for slurry ETV-ICP-AES Voltage–time programmes Voltage/mV Temperature/°C Ramp/s Hold/s Pre-coating of the coil— 0 20 2400 Decreasing 2400 14 500 0 14 500 Decreasing 10 10 10 1 0 23 0 0 0 0 0 Coating gas 1st to 11th 21st to 31st and 41st to 51st second Measurement— (Drying) — — — — 2600 2700 (Vaporization) (Clean out) 90 710 610 500 17 100 19 500 (Re-coating) 0 14 500 11 200 Decreasing 2400 2200 Decreasing (Cool down) 20 10 40 30 5 3 9 9 4 45 0 0 0 0 0 1 0 0 0 0 0 Coating gas 117th to 125th second Coating gas flow 9–16 ml min-1 depending on matrix and slurry concentration Sample volume 20 ml Table 2 ICP-AES parameters used for slurry ETV-ICP-AES Plasma gas (Ar) 14 l min-1 0.3 (K Na 0.9) l min-1 Rf power Intermediate plasma gas (Ar) Aerosol carrier gas (Ar–H2 6.5% v/v H2) 0.7 l min-1 900 W Emission lines Wavelength/nm Interfering W emission line Analyte emission line Element Al B 228.629 267.728 259.964 Be Ca Cd Co Cr Cu Fe Mg 257.617 Mn Ni Pb Zn K Na W 396.152 208.959 313.042 393.366 226.502 228.616 267.716 324.754 259.940 279.553 257.610 231.604 220.353 213.858 766.490* 588.995* 208.819* *=Monochromator.temperatures while section 2 appearing only after about 1.1 s reflects the formation of volatile tungsten compounds at higher coil temperatures and/or the vaporization of tungsten metal. Both sections of the signal can be eVectively reduced by applying a HVDC reducing the trace amounts of oxygen in the carrier gas and TCC of the coil.48 In the following tungsten ablation and the eVect of a HVDC and TCC in the presence of silicon nitride and silicon dioxide are discussed in detail.Silicon nitride The vaporization of silicon nitride might lead to the intermediate formation of tungsten nitrides (WN W2N WN2) and/or silicides. Tungsten nitrides are unstable compounds easily decomposing into the elements.49 Without using a HVDC and TCC section 1 and section 2 of the tungsten signal are significantly reduced compared with water when silicon nitride slurries are processed (see Table 3). With increasing slurry concentrations section 1 of the signal increases whereas section 2 decreases [see Fig. 1(i )]. As can be seen from Fig. 1(i ) and (ii ) the HVDC and TCC have virtually no eVect on section 1 of the tungsten signal. For high sample amounts e.g. 160 mg as used for analysis this part of the signal was even slightly increased.Section 2 of the tungsten signal was totally suppressed by the HVDC and TCC under all conditions [see Fig. 1(ii )]. As a consequence of the diVerent behaviour of section 1 and section 2 of the tungsten signal a significant reduction of tungsten ablation by the HVDC and TCC was only achieved for low sample amounts and long integration times (see Table 3). Under the conditions used for analysis (sample amount per vaporization 160 mg) however the application of the HVDC and TCC was generally found to be dispensable. The optimum coating gas flows for sample amounts of 40 and 160 mg were 11 and 9 ml min-1 respectively. Silicon dioxide Compared with silicon nitride silicon dioxide causes a much higher tungsten ablation. Applying a sample amount of 40 mg per vaporization tungsten ablation was approximately a factor of 4 higher for silicon dioxide than for silicon nitride.The pronounced release of tungsten during the vaporization step is obviously caused by the formation of volatile tungsten oxide WO3 (mp 1472 °C bp 1750 °C),50 and tungsten Fig. 1 Tungsten ablation during the vaporization of diVerent amounts of silicon nitride. Signal measured via the monochromator at the tungsten emission line at 208.819 nm. Vaporization temperature 2600 °C. (i) without high-voltage discharge cell (HVDC) and tungsten carbide coating (TCC); (ii) with HVDC and TCC. (A) 40; (B) 100; and (C) 160 mg silicon nitride per vaporization. Table 3 Tungsten ablation during the vaporization of diVerent amounts of silicon nitride measured via the monochromator at an interferencefree tungsten emission line (208.819 nm); n=10.Vaporization temperature 2600 °C Integration time=1.2 s Amount of Si3N4 per vaporization/mg II† I* 0 40 100 175±2 148±19 245±29 1510±45 159±25 263±32 160 860±145 730±90 * I Without high-voltage discharge cell (HVDC) and tungsten carbide coating (TCC). † II With HVDC and TCC. Journal of Analytical Atomic Spectrometry December 1997 Vol. 12 2O5 W4O11 (subl. p. 800–900 °C).50 Tungsten suboxides W ablation increased with increasing sample amount. Aside from the tungsten interference the vaporization of silicon dioxide in amounts of 40 mg or more also caused an increased unspecific background emission influencing analyte lines not directly interfered with by tungsten.This observation was not made when processing silicon nitride. Possible reasons for the observed unspecific background emission are a high plasma loading leading to an increased continuum emission a shift of the background emission intensity maximum to a diVerent height in the plasma and especially at higher wavelengths an enhancement of stray light. These eVects are more pronounced for silicon dioxide than for silicon nitride because tungsten ablation from the coil and therefore tungsten introduction into the plasma is much higher with the former matrix. Another explanation for the observed diVerences between the two matrices could be their diVerent thermal properties leading to diVerent behaviour in the plasma.Molecular bands of SiO (between 210.0 and 292.5 nm)51 and WO (between 350 and 590 nm)52 may also contribute to the background. Molecules containing Si alone however can be excluded as a source of unspecific emission because otherwise silicon nitride would show a similar unspecific background increase to silicon dioxide. Both the unspecific emission and the interference by tungsten can be significantly reduced by TCC. The eVect of TCC on the background of 12 elements measured using the high-purity sample SiO2-2 is shown in Fig. 2. The TCC improves the reproducibility of the measured background values and reduces the background by a factor of 2.3 (Fe) to 3.1 (Pb) with a mean value for all elements of 2.6±0.3 (as the SiO2-2 sample still 2 Fig.2 EVect of tungsten carbide coating (TCC) on the background of 12 elements measured using the high-purity sample SiO -2. Concentrations of all elements except for Ca and Mg below the limits of detection; sample amount 200 mg per vaporization. High-voltage discharge cell used throughout. Vaporization temperature 2600 °C; five replicate measurements; integration time=2.5 s. Tungsten signal; peak area/mV s Integration time=2.5 s Integration time=1.7 s II† I* II† I* 175±3 162±20 252±30 3180±40 795±40 336±24 175±2 157±20 249±29 2180±75 252±26 266±32 870±150 750±95 865±150 735±90 1361 contains detectable amounts of Ca and Mg these elements were not considered for factor calculation).Obviously the refractory tungsten carbide can withstand the attack of the matrix much better than tungsten. Nevertheless when analysing a slurry the corrosion of the carbide layer and thus the introduction of tungsten into the plasma cannot be completely avoided. The corrosion increases with increasing slurry concentrations leading not only to a higher background emission but also requiring higher coating gas flows to restore the tungsten carbide layer. For the maximum applicable sample 16 ml min-1. amount of 600 mg the optimum coating gas flow was Background Correction In ETV-ICP-AES background correction is aggravated by the transient nature of the emission signals because a simultaneous determination of the background beside the emission line and the emission value at the line maximum is not possible with a Paschen–Runge mount still utilized in many spectrometers including that used in the present work.Moreover for the analyte lines directly interfered with by tungsten the background has to be determined on the peak maximum. Therefore diVerent background correction strategies were developed depending on the matrix type the concentrations of the respective analyte elements and the nature of the background. As mentioned above the presence of the silicon dioxide matrix causes a considerable increase in background emission. Therefore it is in principle not possible to determine the background by simply using the suspension medium water. There are two possible ways of estimating the background when processing the sample slurry.The measurement could be performed close to the analyte emission line with any sample or on the line maximum using a sample with analyte concentrations below the limit of detection. However only the second method allows a correction of the direct line interferences by tungsten. Fortunately the high-purity sample SiO2-2 was appropriate for this kind of background evaluation. This was further verified by measuring the emission intensities at the respective line maxima and near to the lines when processing slurries of both silicon dioxide samples with equal matrix concentrations. Except for those elements interfered with by tungsten the background values obtained for the two diVerent samples were within the measurement uncertainties the same.It can be assumed that this is also true for the analyte lines that overlap the tungsten lines. In spite of its high purity the sample SiO2-2 still contains detectable amounts of Ca Mg and Na and can therefore not be used for the evaluation of the background at the emission lines of these elements. For high analyte concentrations i.e. Ca and Mg in both samples and Na in the SiO2-1 sample water could be employed in the correction of the background because very low matrix concentrations (0.1–1 g l-1) were suYcient for the analysis and therefore the unspecific background caused by the matrix was low compared with the respective analyte and blank signals. For the determination of Na in the SiO2-2 sample and in silicon nitride the background had to be measured beside the analyte emission line.On account of an argon emission line close to the Na line used the background measured at wavelengths below the line is significantly higher than the background measured above it. Thus background correction was performed by measuring the background at both sides of the analyte line using a sample slurry and interpolating the background at the line maximum. As the silicon nitride matrix caused no significantly increased background emission it was possible to evaluate the background for those elements not interfered with by tungsten at the maximum of the emission lines using water. Nevertheless 1362 Journal of Analytical Atomic Spectrometry December 1997 Vol.12 Analysis of Samples Fig. 3 shows the ETV signals obtained for an aqueous standard solution a slurry and a slurry spiked with a standard solution for Fe in silicon dioxide and Mn in silicon nitride as typical examples. On account of the high heating rate achievable by the tungsten coil very short signals with durations usually below 1 s were obtained for both slurries and standard solutions. Nevertheless the signals obtained for slurries were Fig. 3 ETV-ICP-AES signals obtained for (i) Fe in silicon dioxide (A) background signal (50 mg SiO2-2) (B) silicon dioxide slurry (50 mg SiO2-1) (C) silicon dioxide slurry spiked with 20 ng Fe and (D) aqueous standard solution containing 10 ng Fe; and for (ii) Mn in silicon nitride slurry (A) background signal (B) silicon nitride slurry (160 mg Si3 N4) (C) silicon nitride slurry spiked with 1 ng Mn and (D) aqueous standard solution containing 1 ng Mn.the true background can only be determined in the presence of the matrix. Therefore for the analyte elements (B Cd Ni Pb and Zn) with concentrations below their respective limit of detection in the analysed sample a sample slurry was employed for background determination. For the elements Co Cr Fe and Mn which were directly interfered with by tungsten neither of the above-mentioned correction methods is applicable to the silicon nitride matrix as on the one hand the tungsten emission was not identical for water and silicon nitride and on the other hand no adequately pure sample was available for a direct background evaluation.For this reason the tungsten emission constituting the greatest part of the background on the interfered analyte lines was calculated via the emission on the interference-free tungsten line at 208.819 nm. The relative emission intensities of tungsten on the interfered lines (=background at the interfered analyte lines measured with the polychromator divided by the tungsten emission signal at 208.819 nm measured with the monochromator) could be determined in extra runs prior to the analysis using water provided that they do not diVer for water and slurry. In fact a diVerence of about 15% for the relative emission intensities of water and silicon nitride was observed. This could be attributed to a shift of the emission intensity maximum in the plasma in the presence of the matrix and the diVerent observation heights of the monoand polychromator.Consequently all determined relative emission intensities were corrected accordingly. The calculation of the relative emission intensities and the correction of the analyte emission intensities were performed automatically by the data processing program. usually broader than those obtained for aqueous standard solutions whereby the width of the signals increased with increasing slurry concentrations. However this did not impede the analysis because integrated signals were used for quantification. Apparently the excitation conditions in the plasma for the slurry and standard solution were not identical as for both matrices the sensitivity obtained for the slurry was lower than that obtained for the aqueous standard.Therefore the calibration had to be performed using the standard additions method. Maximum applicable matrix amounts per vaporization for silicon dioxide were found to be 50 mg for Mn and Cr 200 mg for Na and 600 mg for all other elements. For Mn Cr and Na the matrix mass applied was limited by the background emission and for the remaining elements by the signal suppressing eVect of the matrix. The matrix amount of silicon nitride slurries was limited to a maximum value of 160 mg by the lifetime of the tungsten coil. Under these conditions approximately 25 vaporizations could be performed with a single tungsten coil before it broke or melted which is suYcient for one entire analysis including blank sample slurry and standard additions with two diVerent concentrations.The lifetime of the coil was considerably longer when analysing silicon dioxide reaching up to about 100 vaporizations. It has to be pointed out that the tungsten coils employed are extremely inexpensive and could easily be exchanged in less than 1 min without switching oV the plasma. The contents of the analytes determined in silicon dioxide and silicon nitride by the developed slurry ETV-ICP-AES method are listed in Tables 4 and 5 together with results obtained by several independent methods. With the exception of Na in SiO2-2 and in Si3N4 and Cr in Si3N4 the agreement of the results can be considered as excellent.This confirms the applicability of the proposed background correction procedures and the calibration. The concentration of Na in the SiO2-2 sample determined by slurry ETV-ICP-AES is higher by a factor of about 5 than Al B Ca Co Cr Cu Fe Mg Na Ni Pb Zn Table 4 Results of the determination of 13 elements in silicon dioxide by slurry ETV-ICP-AES and comparison with results by independent methods (all values in mg g-1) 8±1 0.4±0.1 the values obtained by the other methods excluding ICP-MS. The results for Cr and Na in silicon nitride are higher by factors of about 2 and 1.5 respectively than the ETAAS results. One possible reason for the observed deviations is a faulty background correction especially for the determination of Na in SiO2-2 for which the background amounts to about 50% of the overall signal.Moreover the result obtained for the SiO2-1 sample which has a much higher Na content is in good agreement with the results of the independent methods indicating that the calibration is accurate. The calibration can become a problem despite the use of the standard additions method if either the vaporization or the excitation conditions diVer for analyte contained in the solid material and analyte originating from the spike. Whereas the former is less likely to cause errors on account of the quantification via integrated signals the latter cannot be excluded and is probably responsible for the high Cr result in Si3N4. The eVect was still more pronounced for the determination of K for which slurry ETV-ICP-AES gave a result that was approximately a factor of 150 higher than the results obtained by ETAAS similarly also for K in silicon dioxide.ICP-MS55 2500±3 — 1.6±0.1 3.30±0.01 <0.15 109±1 — — 0.4 — 3 0.02 0.8 <0.03 400 0.3 289 19 0.029 198 2.6 — — 5.5 <0.03 1 <0.25 This work Sample Slurry ETAAS53 Slurry ETV-ICP-MS47 Slurry ETV-ICP-AES 3100±250 3300±400 1.1±0.2 — — 3200±600 <70 — — SiO2-1 SiO2-2 SiO2-1 SiO2-2 SiO2-1 SiO2-2 — — — <2 <130 <130 131±19 0.4±0.1 <70 <70 — — 0.8±0.15 <0.014 2.1±0.5 — 3.7±0.6 <0.02 1.62±0.06 SiO2-1 SiO2-2 SiO2-1 SiO2-2 SiO2-1 SiO2-2 <9 <9 <3 <3 294±30 <0.02 1.7±0.2 <0.05 369±18 <3 <0.07 360±30 0.4±0.1 SiO2-1 SiO2-2 SiO2-1 SiO2-2 — <2 62±9.5 1.2±0.1 0.50±0.01 <6 — — — — <0.1 3.1±0.6 6.1±0.5 SiO2-2 SiO2-1 SiO2-2 SiO2-1 SiO2-2 SiO2-1 <2 125±13 128±15 130±30 150±10 0.4±0.1 <0.7 0.21±0.03 0.19±0.01 18±3 14±3 17±4 17.9±0.6 <0.04 0.012±0.002 <0.03 73±4 6±9 8±10 9 0 1.0±0.2 1.5±0.2 1.1±0.2 — <0.006 1.0±0.1 <0.4 5±1 <6 <20 <20 <2 <2 — — — — SiO2-2 SiO2-1 SiO2-2 Journal of Analytical Atomic Spectrometry December 1997 Vol.12 <5 0.21±0.09 Table 5 Results of the determination of ten elements in silicon nitride by slurry ETV-ICP-AES and comparison with results obtained by ETAAS (all values in mg g-1) Independent methods This work Slurry ETV-ICP-AES Slurry ETAAS56 Solution ETAAS56 470±40 — 380±20 — 371±15 0.055±0.006 43±2 — — 3.8±0.1 3.5±0.3 7.0±0.8 1.1±0.1 1.56±0.07 1.2±0.1 62±2 59±1 50±5 2.3±0.1 42±3 55±4 46±8 1.8±0.1 12.0±0.6 <1.5 Al Be Ca Cr Cu Fe Mg Mn Na Zn 2.0±0.1 6.5±0.4 0.34±0.02 Independent methods ICP-AES54 INAA54 Solution ETAAS53 — — — — — — — 2990±150 0.8±0.1 — — — — — <2.5 — — — 0.55±0.1 0.0017±0.0001 3.0±0.4 — 2.1±0.1 <0.02 1.60±0.01 — 3.5±0.4 0.007±0.001 1.8±0.1 0.015±0.004 — — 348.0±0.2 0.5±0.2 <0.007 390±24 0.7±0.1 <0.02 233±1 0.80±0.01 147±1 — — 14.0±0.2 <0.02 79±1 10.9±0.1 <0.05 61.6±0.5 0.8±0.1 — — — — — — — — <0.01 1.1±0.1 0.10±0.01 1363 On closer investigation it was found that the emission intensity maximum for K originating from the standard solution appeared very early in the plasma inside the load coil.The emission maximum for K originating from the matrix however appeared much later near the observation height of the monochromator. This shift is obviously caused by the additional energy needed for the release of the analyte from the matrix aerosol.As a result the sensitivity obtained at the chosen observation height for K originating from the spike is lower than the sensitivity for K originating from the matrix. For a wet aerosol i.e. when using pneumatic nebulization the emission intensity maximum for K is also shifted to a higher position in the plasma near the observation height owing to the energy needed for the evaporation of the water. As this is the case for both the analyte from the sample and from the standard the problem described above is specific for sample introduction via ETV. However none of the above factors should influence the determination of Na in Si result obtained for a second sample was 9.4±0.5 mg g-1 which 3N4 by slurry ETV-ICP-AES as the is in good agreement with the results obtained by slurry ETAAS (8.2±0.5 mg g-1) and solution ETAAS (12±2 mg g-1).The limits of detection (LODs) achievable for the analysis of silicon dioxide and silicon nitride by slurry ETV-ICP-AES are listed in Table 6 and were calculated as three times the standard deviations of background measurements. For silicon nitride the LODs were mainly limited by the small applicable sample amount of only up to 160 mg per vaporization. However the LODs achievable in silicon dioxide are between a factor of 1.3 and 67 higher than those in silicon nitride although 3.75 times higher sample amounts can be applied. The reason for this is the considerably higher matrix-produced background showing also higher fluctuations.LOD/mg g-1 SiO2 Si3N4 2 130 — 0.4 — 70 9 3 3 0.035 2 1 6 20 2 1 34 0.01 0.03 0.8 3 0.4 0.3 0.4 0.01 0.03 0.04 3 14 1.5 CONCLUSIONS For most analytes the combination of a tungsten coil as a simple and inexpensive ETV device with ICP-AES and the slurry sampling technique provides a reliable method for the analysis of silicon-based materials. Compared with the conventional nebulization of solutions the developed method considerably reduces the risk of contamination is less time consuming easier to apply and avoids the use of the highly toxic hydrofluoric acid as digestion medium. However owing to rather small applicable sample portions and tungsten interferences for the analytes Mn Cr Co and Fe only moderate LODs can be achieved.The method is well suited for the fast routine determination of impurities at the mg g-1 level. Table 6 Limits of detection (LODs) achieved by slurry ETV-ICPAES for the analysis of silicon dioxide and silicon nitride Element Al B Be Ca Cd Co Cr Cu Fe Mg Mn Na Ni Pb Zn 1364 Journal of Analytical Atomic Spectrometry December 1997 Vol. 12 However coupling of the tungsten coil vaporizer to an ICP-MS instrument allows considerably lower LODs to be achieved.47 REFERENCES 1 Umemoto M. Hayashi K. and Haraguchi H. Anal. Chem. 1992 64 257. 2 Blain L. and Salin E. D. Spectrochim. Acta Part B 1992 47 205. 3 Fujimoto K. Okano T. and Matsumura Y.Anal. Sci. 1991 7 549. 4 Umemoto M. and Kubota M. Spectrochim. Acta Part B 1991 46 1275. 5 Farinas J. C. Moreno R. and Mermet J. M. J. Anal. At. Spectrom. 1994 9 841. 6 Halicz L. Brenner J. B. and YoVe O. J. Anal. At. Spectrom. 1993 8 475. 7 Ebdon L. and Goodall P. J. Anal. At. Spectrom. 1992 7 1111. 8 Lobinski R. Van Borm W. Broekaert J. A. C. Tscho� pel P. and To� lg G. Fresenius’ J. Anal. Chem. 1992 342 563. 9 Baldwin D. P. Zamzov D. S. and D’Silva A. P. Anal. Chem. 1994 66 1911. 10 Dittrich K. Mohamad I. Nguyen H. T. Niebergall K. Pfeifer M. and Wennrich R. Fresenius’ J. Anal. Chem. 1990 337 546. 11 Arrowsmith P. Anal. Chem. 1987 59 1437. 12 Hu W. D. Anal. Chim. Acta 1991 245 207. 13 Ono A. Suchi M. and Chibe K. Appl.Spectrosc. 1987 41 970. 14 Aziz A. Broekaert J. A. C. Laqua K. and Leis F. Spectrochim. Acta Part B 1984 39 1091. 15 Zucheng J. Bin H. Yongchao Q. and Yun’e Z. Microchem. J. 1996 53 326. 16 Plantikow-Vossga�tter F. and Denkhaus E. Spectrochim. Acta Part B 1996 51 261. 17 Hu B. Jiang Z. Qin Y. and Zeng Y. Anal. Chim. Acta 1996 319 255. 18 Alary J. F. Hernandez G. and Salin E. D. Appl. Spectrosc. 1995 49 1796. 19 Moens L. Verrept P. Boonen S. Vanhaecke F. and Dams R. Spectrochim. Acta Part B 1995 50 463. 20 Nickel H. and Zadgorska Z. Spectrochim. Acta Part B 1995 50 527. 21 Golloch A. Haveresch-Kock M. and Plantikow-Vossga�tter F. Spectrochim. Acta Part B 1995 50 501. 22 Zaray G. and Kantor T. Spectrochim. Acta Part B 1995 50 489. 23 Ren J.M. and Salin E. D. Spectrochim. Acta Part B 1994 49 555. 24 Zaray G. Leis F. Kantor T. Hassler J. and To� lg G. Fresenius’ J. Anal. Chem. 1993 346 1042. 25 Boonen S. Verrept P. Moens L. and Dams R. F. J. J. Anal. At. Spectrom. 1993 8 711. 26 Berryman N. G. and Probst T. U. Fresenius’ J. Anal. Chem. 1996 355 783. 27 Escobar M. P. Smith B. W. and Winefordner S. D. Anal. Chim. Acta 1996 320 11. 28 Kantor T. Fresenius’ J. Anal. Chem. 1996 355 606. 29 Sparks C. M. Holcombe J. A. and Pinkston T. L. Appl. Spectrosc. 1996 50 86. 30 Goltz D. M. Gre�goire D. C. and Chakrabarti C. L. Spectrochim. Acta Part B 1995 50 1365. 31 Ren J. M. Rattray R. Salin E. D. and Gregoire D. C. J. Anal. At. Spectrom. 1995 10 1027. 32 Fonseca R. W. and Miller-Ihli N.J. Appl. Spectrosc. 1995 49 1403. 33 Argentine M. D. Krushevska A. and Barnes R. M. J. Anal. At. Spectrom. 1994 9 1121. 34 Marawi J. Wang J. and Caruso J. A. Anal. Chim. Acta 1994 291 127. 35 Matousek J. P. and Mermet J. M. Spectrochim. Acta Part B 1993 48 835. 36 Ediger R. D. and Beres S. A. Spectrochim. Acta Part B 1992 47 907. 37 Carey J. M. Evans E. H. Caruso J. A. and Shen W. L. Spectrochim. Acta Part B 1991 46 1711. 38 Etoh T. Yamada M. and Matsubara M. Anal. Sci. 1991 7 1263. 39 Parsons P. J. Qiao H. Aldoas K. M. Mills E. and Slavin W. Spectrochim. Acta Part B 1995 50 1475. 40 Bruhn C. G. Ambiado F. E. Cid H. J. Wo� rner R. Tapia J. and Garcia R. Anal. Chim. Acta 1995 306 183. 50 Handbook of Chemistry and Physics ed. Weast R. C. CRC Press 41 Krug F. J. Silva M. M. Oliveira P. V. and No�brega J. A. Boca Raton FL 62nd edn. 1982. Spectrochim. Acta Part B 1995 50 1469. 51 Gmelins Handbuch der Anorganischen Chemie Silicium system-no. 15 part B Verlag Chemie Weinheim/Bergstrasse 8th edn. 1959. 42 Gine� M. F. Krug F. J. Sass V. A. Reis B. F. No�brega J. A. and Berndt H. J. Anal. At. Spectrom. 1993 8 243. 43 Ivanova E. Havesov I. Berndt H. and Schaldach G. Fresenius’ 52 Gmelins Handbuch der Anorganischen Chemie T ungsten systemno. 54 Supplement Volume B2 Oxide Springer-Verlag Berlin- J. Anal. Chem. 1990 336 320. Heidelberg-New York 8th edn. 1979. 53 Hauptkorn S. and Krivan V. Spectrochim. Acta Part B 1996 51 1197. 54 Fritz M. and Krivan V. unpublished work. 55 Baumann H. and Pavel J. Mikrochim. Acta 1988 III 423. 56 Friese K.-Ch. and Krivan V. Anal. Chem. 1995 67 354. 44 Berndt H. and Schaldach G. J. Anal. At. Spectrom. 1988 3 709. 45 Williams M. and Piepmeier E. H. Anal. Chem. 1972 44 1342. 46 Barth P. and Krivan V. J. Anal. At. Spectrom. 1994 9 773. 47 Hauptkorn S. Krivan V. Gercken B. and Pavel J. J. Anal. At. Spectrom. 1997 12 421. 48 Barth P. Hauptkorn S. and Krivan V. J. Anal. At. Spectrom. in the press. 49 Gmelin Handbook of Inorganic Chemistry T ungsten system-no. Paper 7/05322C Received July 23 1997 54 Supplement Volume A5b Springer-Verlag Berlin-Heidelberg- New York 8th edn. 1993. Accepted September 12 1997 1365 Journal of Analytical Atomic Spectrometry D