JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY JUNE 1994 VOL. 9 707 Halocarbon-assisted Slurry Vaporization in Inductively Coupled Plasma Atomic Emission Spectrometry for the Analysis of Silicon Nitride Powder" Gyula Zaray lmre Varga and Tibor Kantort Institute for Inorganic and Analytical Chemistry L. Eotvos University P. 0. Box 32 H- 15 18 Budapest 7 12 Hungary A conventional Babington-type nebulizer was applied for introduction of samples in the form of solutions and slurries (1% m/m) produced from a finely dispersed silicon nitride (Si3N4) powder (mean particle size 0.54 pm). Calibration with solution standards for the slurry method resulted in 20-40% negative deviations in the results for impurities of Al Fe Ca Mg and Ti compared with the dissolution-based analysis. The addition of Freon-12 (CCI,F,) as a possible halogenation agent to the plasma resulted in decreased negative deviations by factors of between 1.2 and 2.8 which suggests that the degree of evaporation of the slurry particles was increased by halogenation.The fact that the negative deviations could not be completely eliminated could perhaps be explained by a difference in the efficiency of nebulization (as the sample introduction process) for the solutions and the slurries which applies rigorously to the present sample type and nebulizer system. With the introduction of Freon no degradation in the linearity of the analytical curves was found in contrast to earlier observation by other workers. The line-to-background intensity ratios (wavelength >220 nm) were not decreased at the rate of halocarbon introduction eventually used.Keywords lnductively coupled plasma atomic emission spectrometry; silicon nitride slurry; halogenation; Freon- 72 Pneumatic nebulization of slurries has been found to be a fairly easy and readily accessible method of sample introduc- tion for the inductively coupled plasma atomic emission spec- trometric (ICP-AES) analysis of finely dispersed In principle accurate results can be expected if calibration is based on certified reference samples similar in physico-chemical properties (including particle size distribution) to the samples. The approach of replacing powdered standards by standards in solution has led to moderate success in general although small or large negative errors in the results are often observed for several sample types and different constituents.The high solid V-groove nebulizer was designed' specifically for slurry nebulization and resulted in a lowering of the negative errors but these could not be eliminated ~ompletely.~?~ One of the possible reasons for this is related to the lower degree of evaporation of slurry particles relative to the smaller particu- lates formed from solution droplets in the plasma.24 To decrease this potential source of error Ebdon and Goodal16 suggested halogenation and introduced a mixture of Freon-116 (C,F,) and argon as the aerosol carrier (nebulizer) gas. Although they reported an apparent improvement in 'analytical recovery' the overall result was discouraging in that increased curvature of the analytical graphs towards the concentration axis (for Al Fe Mg and Ti) was found for solution standards with the introduction of Freon.In the present work the applicability of a commercial Babington-type nebulizer to slurry nebulization was specifically investigated. The sample was a finely dispersed silicon nitride powder (mean particle size of 0.54 pm) analysed first by a conventional dissolution-based method the results of which were then used as reference data for the slurry method. Even applying standard additions in the form of solutions to the slurry as an attempt at matrix matching negative errors were found similar to those observed in the works cited above. It therefore became a vital question as to whether the halogen- ation suggested in ref. 6 can improve the results of the present slurry nebulization technique.The method of Freon introduction selected for use was * Presented at the XXVIII Colloquium Spectroscopicurn Internationale (CSI) York UK June 29-July 4 1993. t To whom correspondence should be addressed. similar to that used earlier in flame atomic absorption spec- trometry (AAS),7 in that it was supplied after the spray chamber into the injector flow rather than into the high- pressure gas that feeds the nebulizer.6 In AAS experiments (using an acetylene-air flame) the depression effect of A1 on the Mg signal (as a typical solute vaporization interference) could be eliminated at a Freon-12 concentration of 1.8% v/v in the total flame gases.7 Experimental Instrumentation and Operating Conditions Spectrometer.Labtam 8440 Plasmalab 1440 vacuum polych- romator with Paschen-Runge mounting of the grating ( 1440 grooves mm-' 1 m focal length) simultaneous detection of the lines A1 I 396.152 Ca I1 393.367 Fe I1 259.940 Mg I1 279.553 Si I 251.611 and Ti I1 334.940nm. Integration time 5 s and background correction by shifting the entrance slit. ICP source. A 27.12 MHz crystal controlled generator 1.2 kW output power demountable torch of medium size (outer quartz tube of 17.5 mm i.d.) sample injection tube of 1.8 mm i.d. argon flow rates (outer + intermediate + inner) of (14 + 1 + 0.8) dm3 min-' respectively and observation height in the plasma of 16 mm unless stated otherwise. Sample introduction system. Babington-type GMK nebulizer aerosol carrier argon flow of 0.7 dm3 min-l impact bead distance of 3 mm Gilson Minipuls-2 peristaltic pump and sample delivery of 3 ml min-'.Freon introduction system. A T-shaped glass junction 4 cm long was incorporated between the outlet of the nebulizer chamber and the inlet of the torch sample capillary in order to feed a mixture of 100 cm3 min-l of argon and 10 cm3 min-l of Freon-12 (Union Carbide Westerlo Belgium) unless stated otherwise. The flow rates of argon used for dilution of the Freon and for sample nebulization were kept constant (total 0.8 dm3 rnin-l see above). The flow rate of Freon was moni- tored by a 'pressure difference' flow meter which was filled with paraffin oil and calibrated by a soap-bubble gas volu- metric device. The argon+Freon mixture was streamed in a polyethylene tube and it was concluded that the adsorption- desorption equilibrium of Freon on the internal tube wall708 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY JUNE 1994 VOL.9 could be attained in 3-5 min depending on the flow rate. A favourable practice was to first set a higher flow rate of Freon than that to be selected for a particular experiment for 3 min and then set the required value and allow 5 min before sample introduction. The reflected power could also be used for checking the constancy of Freon introduction. Characteristics of Sample The silicon nitride (Si,N,) powder sample was manufactured by Starck Berlin Germany. The particle size distribution of the Type LC-10 powder was determined in the form of a suspension (0.05 YO Calgon liquid) using Micrometrics equip- ment.After ultrasonic treatment for 8 h the mean particle size was dS0=0.54 pm. Slurry Preparation A 1 f0.01% m/m concentration of silicon nitride powder was achieved in 1% m/m HCl (see below) the suspension was agitated in an ultrasonic bath for at least for 5 min before nebulization and stirred continuously during nebulization by a magnetic stirrer. Dissolution Procedure for Solution Analysis A CEM Model MDS-2100 microwave digestion apparatus was used for sample dissolution in an HF+HN03 mixture. Silicon nitride (150mg) was weighed into a 120ml volume poly(tetrafluoroethy1ene) (PTFE) vessel and 10 ml of HF (40% m/m) and 1 ml of HNO (65% m/m) were added to digest the samples at 170°C (for 30 min). After cooling the solution was transferred into a PTFE beaker and heated gently to dryness.Then 10 ml of HCl (18% m/m) were added and the evapor- ation was repeated. Again 10 ml of 1% m/m HCl were added for dissolution of the dry residue. Three solutions were pre- pared simultaneously as described above all were combined and transferred into a 50ml calibrated flask and made up to the mark with doubly distilled water. Calibration Standard additions Standard additions in the form of solution to the powdered sample was performed as follows. Four sample aliquots of 0.4 g were weighed in 50 ml glass beakers 10 ml of 1% m/m HCl were added to each and three of the samples were spiked with linearly increasing volumes (50 100 and 200 pl) of a multi-element (Al Fe Ca Mg and Ti) stock solution. The first addition approximated the impurity contents according to available information.For the 'zero standard addition' the stock solution was replaced with water. The liquid phases of the suspensions were adjusted to volumes of 40 ml again with the 1% m/m HCl. Stundard and blank solutions The standard and blank solutions were also made in the same manner as above using the same stock solution and diluent but without addition of the silicon nitride powder. Evaluation of intensity data The intensity data (measured with background correction) of triplicate runs was evaluated by linear regression analysis and the unknown concentrations were calculated on the basis of standard additions to slurries and also external calibration using matrix-free standard solutions. Blank concentrations were determined on the basis of solution standards and sub- tracted from the former results.Results and Discussion Linearity of the Analytical Curves With Introduction of Freon Ebdon and Goodal16 found increased curvature of the analyt- ical curves towards the concentration axis for the analytes of interest (Al Fe Mg and Ti) when an argon+Freon-116 mixture was used for the nebulization of solution standards. They added 4% v/v Freon to the aerosol carrier gas which at the maximum corresponded to a Freon flow rate of 52 cm3 min-'. As the linearity of the analytical curves is of prime importance for simplicity of calibration investigation of this parameter was given prominence in the present research. However a different method of Freon introduction (described under Experimental) was applied from that in earlier work.6 As demonstrated for two elements (A1 and Fe) in Fig.1 rectilinear analytical curves (logarithmic plots) were found in the 1-100 mg 1-' concentration range independent of the introduction of Freon which applies also to the other elements (Ca Mg and Ti) investigated. In this instance multi-element standard solutions were nebulized and the concentration range was similar to that investigated in ref. 6 except for Al where the concentration in the earlier work extended to 1000 mg 1-'. It should be noted that linear intensity versus concentration functions were also found by addition of solution standards to the slurries (see Experimental) and setting the observation heights to 20 and 25 mm (16 mm in Fig. 1). Other Spectral Characteristics Versus Freon Flow Rate By increasing the flow rate of Freon up to 40 cm3 min-' the net line intensities decreased for all constituents in the slurry and the solution by factors of between 2.5 and 3.2.Aluminium was selected for demonstration as shown in Fig. 2 because of the considerable separation of the 'slurry' and 'solution' curves (see also below). As is also shown in Fig. 2 the line-to- background intensity ratios (which are important from the point of view of detection limits) vary according to the maxi- mum of the curves peaking at a Freon flow rate of around 25 cm3 min-'. More general information on the variation of the spectral background on introduction of Freon is presented in Fig. 3 where the intensity ratio (with :without Freon) is plotted as a function of wavelength for 23 analytical lines that can be detected with the polychromator used.Data are shown for nebulization of high-purity water and for a blocked sample uptake tube i.e. for an almost dry plasma. The flow rate of Freon was 10cm3min-' which was also selected for the 5x10' I 5 ' I I I 1 x 10-1 1 10 1 ~ 1 0 2 1 ~ 1 0 3 Concentration/mg I-' Fig. 1 Analytical curves for A1 and Fe determined with solution standards A without Freon; B with 10 cm3 rnin-.' Freon; and C with 40 cm3 min- ' FreonJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY JUNE 1994 VOL. 9 709 > L .- 1.6 Q) C c .- 1.4 3 2 E 1.2 n 2 1.0 m .- - U 0.8 i 0.6 I I I I I 0 10 20 30 40 Flow rate (Freon-12)/cm3 min-' Fig. 2 (a) Net line intensity and (b) normalized line-to-background intensity ratio for Al A in slurry and B in solution as a function of Freon flow rate ~ 0 150 200 250 300 350 400 450 500 Wavelengthhm Fig.3 Background intensity ratio with and without the introduction of Freon at a flow rate of 10 cm3 min-' for 23 analytical lines in the 180-460 nm wavelength range x without water nebulization and 0 with water nebulization analytical work (see below). As can be seen the back- ground intensity ratio (with without Freon) is higher for a dry plasma than for a wet plasma above 250nm and the opposite is valid below 250 nm. On the other hand for a wet plasma this ratio is lower than unity above 250nm which means that a lower background is found with halocarbon introduction than without it. The background is increased by introduction of Freon below 250 nm and this increase is higher for a wet plasma than for a dry plasma.These findings are probably due to the band emission of CO (and chemically related) species,' while the lower background above 250 nm (mostly the electron continuum) could be the result of a drop in excitation-ionization temperature on introduction of Freong (see below). By visual inspection of the plasma a green colouration of the lower zone of the inner core can be observed at and above a Freon flow rate of 20cm3 min-' which is much more intensive without nebulization of water. This phenomenon has also been observed when carbon tetrachloride was nebulized into the ICP and is related to the band emission of C2 species.' The decrease in intensity of C2 bands in the wet plasma is the consequence of oxidation of carbon along with the formation of CO (see above).In the work described in ref. 9 the 'excitation temperatures' of the ICP were compared when nebulizing aqueous and carbon tetrachloride solutions using a desolvation system. The rate of introduction of the solvent vapour could be varied by operating the condenser of the desolvation unit in the tempera- ture range from -10 to 20°C. At the lowest condenser temperature (the lowest rate of vapour introduction) the plasma temperature was lower by about 700 K for carbon tetrachloride compared with that for introduction of water at an observation height of 10mm. A considerable decrease in the ratio of the ionic to atomic line intensities (using the most sensitive lines of Mg) was also found with the introduction of carbon tetrachloride indicating a depression in the ionization.It could be expected that similar trends in changes of the excitation parameters prevail with introduction of Freon as those deter- mined for carbon tetrachloride introduction,' as outlined above. It was anticipated that the decomposition products of the fluorinated hydrocarbons generated in the plasma would react with the hot parts of the torch thus resulting in corrosion after long-term application. Indeed this corrosion was noted in ref. 6 on introducing 52cm3 min-' of Freon-116 at an intermediate argon flow rate of 0.4 dm3min-'. As is known the latter component of the argon supply to the torch plays a dominant role with respect to heating of the tip of the quartz sample introduction capillary by the plasma. Therefore the effect of the Freon flow rate was studied with increasing intermediate argon flow rates by monitoring the Si 251.61 1 nm line as an indicator of Si release from the torch.The results are depicted in Fig. 4 (curve B) when using a 1 dm3 min-' intermediate argon flow rate which was found to be sufficient to diminish corrosion of the tip of the sample introduction capillary. Also shown is the variation in the Si line intensity when a silicon nitride slurry is nebulized (curve A) which is similar to that found for the other slurry components (see Fig. 2 and more details are given below). Axial Intensity Profiles In Fig. 5 net line and background intensities and relative line- to-background intensity ratios are plotted as a function of plasma observation height (POH) for slurry nebulization 201 7 I " 0 10 20 30 40 Flow rate (Freon-12)/cm3 min-' Fig.4 Line intensity of Si (251.611 nm) as a function of Freon flow rate A slurry nebulization; and B water nebulization710 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY JUNE 1994 VOL.9 1 lo5 -5 2 1x103 & 2 1x102 Y m C .- - -a C m .- ,z 1x10’ w al C 4- - 1 1 x lo-’ A 4 8 12 16 20 24 28 Plasma observation height/mm Fig.5 Axial plasma profiles of some spectral parameters for slurry nebulization A Si line intensity; B Mg line intensity; C background intensity of Mg line; and D Mg line-to-background intensity ratio with the introduction of Freon (10 cm3 min-l) relative to that without Freon (0 argon; 0 Freon; and 0 Freon argon) without and with the use of Freon (10 cm3 min-l). The shape of the corresponding curves of the five analytes (note the logarithmic scale of the ordinate) were similar and Mg was however selected for demonstration purposes.The main characteristics of these curves correspond to those determined for a matrix-free solution; deviations in the finer details will be shown later. In summary the line intensity maxima are found at a POH of 16 mm for the analytes (14 mm for Si) the log of the background intensity decreases linearly with POH to a first approximation and all of these intensities are between 0.59 and 0.83 of those obtained when no Freon was present in the plasma. This last observation for line intensities again suggests a lower excitation temperature with introduction of Freon as is presumed above on the basis of the work cited in ref.9. Although this explanation is plausible it has not been proved so far. The curve representing the line-to-background intensity ratio with the use of Freon related to that in the absence of Freon shows a slight maximum in the 14-18 mm POH range which also applies to the other analytes. The maximum values were found to be close to unity (lower for Mg and Fe and higher for Ti Ca and Al). This means that no gross change in the detection limits are expected with the introduction of Freon for the elements investigated under optimum plasma conditions. The comparison of axial intensity profiles with and without introduction of Freon was also aimed at finding experimental evidence for halocarbon assisted vaporization of slurry par- ticles in addition to the analytical results discussed in the next section.As is known the main excitation parameters (tempera- ture and electron pressure) change dramatically along the vertical axis of the plasma and also these parameters vary because of the introduction of Freon. The latter variation can be related to a higher energy consumption and also to the increase in the total gas flow rate in the plasma channel. Therefore ratioing of line intensities is necessary to compensate for the changes in excitation parameters and thus to obtain information on changes in the vaporization of sample particles. The better this compensation is approximated the more reasonable is the expectation that the difference in line intensity ratios (with without Freon) is due to the difference in the vapour concentration of the emitting analyte.Two approaches of this ratioing are represented by Figs. 6 and 7 and will be discussed. To avoid possible problems in understanding the terms ‘vaporization’ and ‘atomization’ as they are used here will be clarified. By (high-temperature) vaporization a heterogeneous phase transition (from condensed phase to gas phase) process is implied which often also involves a chemical decomposition 0‘ I I I I I 1 0 ‘ I I I I I I I 16 20 24 28 0 4 8 12 Plasm a observation he ig ht/m m Fig. 6 Axial plasma profiles of intensity ratios measured with Freon (10 cm3 min-l) and without Freon (i.e. with argon) when nebulizing 0 a solution and 0 a slurry for (a) Si; (b) Ca; (c) Ti; and ( d ) Fe I I I I I I ~ 0 4 8 12 16 20 24 28 Plasma observation height/mm Fig.7 Analyte-to-Si intensity ratios for the slurry 0 wit1 out and 0 with the introduction of Freon (10 cm3 min-’) as a fui ction of plasma observation height for (a) Mg; (b) Ti; (c) Fe; and ( d ) t 1JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY JUNE 1994 VOL. 9 71 1 and/or chemical transformation. For an unambiguous differ- entiation the term gas-phase atomization can be used which means a homogeneous phase chemical process (gas-phase dissociation of molecules). In the literature on analytical atomic spectrometry the term 'atomization' most often means both vaporization and gas-phase atomization together which is accepted terminology if no mechanistic interpretations are concerned. In the present work the possible effects of Freon on the vaporization and on the gas-phase atomization must be distinguished and thus the more specific meaning of the terms should be utilized.The best compensation for the changes in excitation param- eters is expected when ratioing is made for the same spectral line without and with a supply of Freon and this ratioing is made separately for slurry and solution nebulization. In addition to the curves seen in Fig. 6 for Fe Ti Ca and Si it should be noted that the corresponding curves for A1 and Mg are similar to those shown for Fe and Ti respectively. The Si curve (determined only for a slurry) shows a relatively high value at the lowest observation height (6 mm) and the same applies for Fe and A1 (not shown) for both slurries and solutions.On the other hand the curves for Ca Mg (not shown) and Ti start with the lowest value at an observation height of 6 mm. In this low-lying zone of the plasma desolv- ation is probably far from complete and desolvation can be different for solution droplets and slurry particles. It has been shown that the excitation parameters are strongly influenced by the degree of desolvation," so the interpretation of the change in line intensities at an observation height of 6 mm is complicated by too many factors. On considering the range of observation heights of between 10 and 22 mm slightly higher intensity ratios (with without Freon) were found for all analytes present in the slurry than those present in the solutions (demonstrated for three elements in Fig. 6).It was expected that evaporation of the residue particles formed from droplets of the matrix-free solution also readily takes place without Freon introduction. Therefore the slightly higher intensity ratios found for the slurry relative to the solution could be explained by the increase in the degree of evaporation of slurry particles under halogenation. The gas-phase atomization of the elements of interest could be influenced by the introduction of Freon predominantly through the formation of the most stable monofluorides as also discussed in ref. 6. The dissociation energies of these species" are as follows A1F 6.89 TiF 5.90 CaF 5.42 SiF 5.03 and MgF 4.77 eV. The data for FeF are not given in ref. 11 but they can be assumed to be similar to those of MnF (4.4eV). It is expected that the formation of fluorides takes place to a greater extent in the lower temperature regions of the higher plasma zones and also the value of the dissociation energy is reflected in the magnitude of the decrease in line intensities.Measurement points of up to an observation height of 26 mm (Fig. 6) however do not indicate a decrease in the intensity ratios (with without Freon) i.e. a decrease in gas- phase atomization owing to introduction of Freon. (This means that the formation of monofluoride could probably be studied at higher plasma zones not considered in the present work.) The measurement points of the analytes in Fig. 6 show a tendency to reach constant and equal values for both the solution and the slurry in the POH region of 22-26 mm.This tendency can be explained by supposing that either the analytes are vaporized from the slurry particles or the whole particles vaporized to completion also without the introduction of Freon up to this observation zone. It should be remembered that finely dispersed silicon nitride was used in the present experiments which was therefore less appropriate for a clear demonstration of halocarbon assisted vaporization. Also because of the use of a relatively high concentration of HC1 in the slurry (see Experimental) partial dissolution of certain slurry components could take place. This could be another reason for the apparent decrease in the difference in the degree of vaporization of slurry particles and the residue particles from droplets of solution.In Fig. 7 reference is made to Si when the slurry is nebulized without introduction of Freon and separately when Freon was supplied to the plasma. This ratioing corresponds to the use of the internal reference method. The Ca Si curve (not shown) was similar to that depicted for Ti Si. Here compensation for the change in excitation parameters is limited because of the difference in the ionization energies of the analytes and the Si and also in the excitation energies of the corresponding spectral lines. In addition the possible change in the degree of vaporiz- ation of the major Si component under the effect of halogen- ation influences the information that can be drawn for the analytes in this respect. In spite of these limitations the higher intensity ratios (analyte silicon) seen for the case of introduc- tion of Freon in the POH region of 14-22 mm can be explained by an increase in the degree of vaporization of the analytes relative to the major Si component.The declining tendency of the curves (the A1 Si curve is an exception) at higher obser- vations heights can be explained by supposing that the degree of vaporization increases only for the major element Si (POH=22-26 mm). On the other hand the almost constant values for the Al Si curves with and without introduction of Freon above a POH of 14mm can be considered as an indication that the selectivity of vaporization for this element from the silicon nitride matrix is lowest irrespective of the presence of Freon. The changes in line intensity ratios due to introduction of Freon shown in Figs. 6 and 7 are rather small and if the possible limitations in compensation for changes in the exci- tation parameters are considered one cannot make an unam- biguous statement about the assistance provided by introduction of Freon towards vaporization.Further evidence should be anticipated from the analytical results discussed below. Analytical Results As described under Experimental two calibration methods were selected for slurry analysis (i) using matrix-free solution standards and (ii) using solution-spiked slurries as an approxi- mation to matrix matching. The preparation of acidic solution standards of matched Si matrix is problematic and was not pursued. The results found by the conventional dissolution- based method were considered as reference values and these are shown in Table 1 together with results for the different versions of the slurry method.It was interesting to observe the effect of the slurry matrix Table 1 Results for analysis of a silicon nitride sample using solution and slurry nebulization methods without and with the use of Freon-12. Average relative standard deviation (RSD) for the slurry results was 3.2% Method Element concentration/mg kg - Sample dissolution* Slurry At no Freon Deviation (YO) Slurry BS no Freon Deviation (YO) Slurry At with Freon Deviation (YO) Slurry BS with Freon Deviation (YO) A1 392 239 233 268 - 31.6 253 - 35.5 - 39.0 - 40.6 Fe 56.6 47.7 - 15.7 45.3 - 20.0 52.9 - 6.5 52.6 -7.1 Ca 51.5 45.1 - 12.4 40.6 -21.2 48.7 - 5.4 46.5 - 9.7 Mg 51.3 40.8 -20.5 37.4 - 27.1 44.2 43.0 - 13.8 - 16.2 Ti 12.6 10.3 - 18.3 9.3 - 26.2 11.8 - 6.3 11.4 - 9.5 ~ ~~~ ~ ~~~ * Reference method. t Calibration based on matrix-free standard solutions.1 Calibration based on addition of solution standards to the slurry (approximate matrix matching).712 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY JUNE 1994 VOL. 9 on the slopes of the linear analytical curves (expressed as the ratio of slopes found with solution-standard spiked slurries and with matrix-free solution standards). The averages of this ratio for the five components were 1.078+0.026 and 1.035 kO.019 without and with introduction of Freon respect- ively. It follows that the analytical sensitivities were enhanced slightly in the presence of the slurry matrix and this enhance- ment effect was smaller in the presence of Freon. Considering the negative errors found with the four versions of the slurry method (Table 1) it is interesting that these are slightly larger when calibration is based on the spiked slurries as a consequence of the noted matrix effect. It is clear that both of the calibration methods used are imperfect in several aspects and because of a reason as yet unknown better error compensation results when matrix-free standard solutions are used for calibration in the case of the present sample and under conditions used.The other important consideration from Table 1 is that on introduction of Freon the negative deviations are decreased significantly for Fe Ca Mg and Ti (by factors between 2.2 and 2.8); the improvement is smallest for A1 (by a factor of 1.14).Conclusions According to the data in Table 1 the negative deviations in the results found with the slurry nebulization method decreased with the introduction of Freon but were not completely eliminated. A comparison of the axial intensity profiles of the plasma (Figs. 6 and 7) and that of the results without and with introduction of Freon- 12 together indicate an enhance- ment in the degree of vaporization for the constituents of the slurry under the effect of this halocarbon as a halogenating agent. The question emerged as to whether the negative deviations could be further decreased by applying more favour- able conditions to the evaporation of slurry particles. Therefore measurements were conducted at higher POHs (20 and 25 mm) and also the flow rate of Freon was doubled (20 cm3 min-l).Eventually practically no further improvement could be obtained which suggests that the remainder of the negative deviations found with the introduction of Freon (Table 1) are probably predominantly because of the difference in nebuliz- ation efficiency of the solution and the slurry. In this respect both earlier opinions are in agreement according to whether the insufficient degree of evaporation or the less efficient nebulization was attributed to the negative errors obtained with the slurry methods (as disputed in ref. 5). In the present work the highest negative deviation was consist- ently found for Al with and without halogenation which could be explained by a higher loss for this element during nebuliz- ation if its concentration is higher in the coarser fraction of the sample.However from the A1 Si curve seen in Fig. 7 it can also be concluded that the degree of evaporation is the smallest for this element from a silicon nitride matrix and the halogenation has only a slight influence on this behaviour. The final recommendation as a result of the present work for the expedient use of the slurry-ICP method can be summarized as follows. If high accuracy is mandatory in a certain analytical task sample introduction by slurry nebuliz- ation must be calibrated with the use of certified reference samples matched to the samples in all possible respects. This can best be achieved in laboratories by quality control of well defined types of materials. The halogenation method investi- gated here can be beneficial in broadening the tolerance with respect of the ‘physical’ non-uniformity of samples and stan- dards. However unambiguous proof for this expectation is anticipated in the future. The authors are grateful to the National Scientific Research Foundation (Hungary) for the support under project numbers OTKA 2278 and OTKA 2786. 1 2 3 4 5 6 7 8 9 10 11 References Ebdon L. and Cave M. R. Analyst 1982 107 172. Raeymaekers B. Graule T. Broekaert J. A. C. Adams F. and Tschopel P. Spectrochim. Acta Part B 1988 43 923. Huang M. and Shen X.-E. Spectrochim. Acta Part B 1989 44 957. Gervais L. S. and Salin E. D. J. Anal. At. Spectrom. 1991 6,41. Halicz L. Brenner I. B. and Yoffe O. J. Anal. At. Spectrom. 1993 8 475. Ebdon L. and Goodall P. Spectrochim. Acta Part B 1992 47 1247. Kantor T. Atomic Spectroscopy Methods for Solid Sample Analysis and for Studying High Temperature Vaporization Processes Thesis (in Hungarian) Library of Hungarian Academy of Sciences 1985. Pearse R. W. B. and Gaydon A. G. The Identijcation of Molecular Spectra Chapman and Hall 4th edn. London 1976. Pan Ch. Zhu G. and Browner R. F. J. Anal. At. Spectrom. 1990 5 537. Hobbs S. E. and Olesik J. W. Spectrochim. Acta Part B 1993 48 817. CRC Handbook of Chemistry and Physics ed. Wheast R. C. Chemical Rubber Co. Cleveland OH 54th edn. 1974. Paper 31049260 Received August 13 1993 Accepted February 2 1994