Vaporization and removal of silica for the direct analysis of geological materials by slurry sampling electrothermal vaporization-inductively coupled plasma-mass spectrometry MuÆda E. Ben Younes,a D. Conrad Gre�goire*b and Chuni L. Chakrabartia aDepartment of Chemistry, Ottawa-Carleton Chemistry Institute, Carleton University, 1125 Colonel By Drive, Ottawa, ON, Canada K1S 5B6 bGeological Survey of Canada, 601 Booth St., Ottawa, ON, Canada K1A 0E8 Received 26th April 1999, Accepted 4th August 1999 Reported is a method for the removal of silica for the direct analysis of solid geological samples high in silica content using slurry sampling electrothermal vaporization-inductively coupled plasma-mass spectrometry (ETV-ICP-MS).The ETV-ICP-MS vaporization curve for SiO2, sampled as a slurry, is reported for temperatures ranging from 810±2600 �C. This curve showed that silica was completely vaporized at a temperature of 2200 �C. The effect of using HF as a chemical modiÆer to remove silica as the tetraØuoride was studied.It was found that HF could completely remove any Si attributed to silica if sufÆcient modiÆer were added and an adequate reaction time allowed. At reaction hold times that were less than optimal, two Si signals were observed. The Ærst signal, which appears at an earlier time and at a temperature less than 480 �C, is attributed to the volatilization of silicon tetraØuoride. The second signal, which appears at a later time and at a temperature of about 2500 �C, is attributed to the vaporization of residual unreacted SiO2 remaining in the graphite tube, or to the vaporization of a mixture of SiO2 and SiC. 20 ml of 50% HF is effective in completely removing 0.125 mg of SiO2. No adverse effects including corrosive degradation of the graphite tube were observed over the lifetime of the tube, which exceeded 200 Ærings. HydroØuoric acid chemical modiÆer was successful in removing virtually all the silica content in natural silicate standard reference materials.Introduction Many environmental and geological materials contain high concentrations of Si as SiO2 or other more complex silicates. Before analysis of these materials for trace metals can proceed, it is necessary to break up this silicate matrix and to remove it or separate it from the analytes. Slurry sampling combined with ETV-ICP-MS is an attractive approach to the trace analysis of environmental and geological samples.Slurry sampling combines the beneÆts of both solid and liquid sampling.1 When compared to conventional sample preparation techniques, slurry sampling has several important advantages: reduced sample preparation time and lower risk of contamination; the analysis of micro amounts of sample; and full automation. When combined with ETV-ICP-MS, slurry sampling can offer additional advantages. For example, with the ETV system it is possible to use the graphite tube as a chemical reactor.Procedural steps such as the addition of digestion reagents, chemical modiÆers, etc., can all be done on-line automatically. Thus, it may be possible to complete an entire sample preparation sequence within the graphite tube starting with suitably powdered sample material. Bendicho and de Loos-Vollebregt2 and Miller-Ihli3 have reviewed the literature on solid sampling and discussed the beneÆts of ultrasonic slurry sampling, which has been used successfully in graphite furnace atomic absorption spectrometry (GFAAS). Ultrasonic slurry sampling has also been extended to ETV-ICP-MS.4±11 Chemical modiÆers are used to enhance the vaporization properties of both analyte and matrix components.One approach involves converting relatively refractory compounds into more volatile compounds. Halocarbons introduced into plasma gases have been used to improve the vaporization properties of relatively refractory analytes. Kirkbright and Snook12 used 0.1% freon-23 for the determination of B, Mo, Zr, Cr and W,13 and CCl2 and freon-12 to determine trace elements in ceramic powders and SiC.14 Ren and Salin15,16 and Allary et al.17 used freon-12 to enhance the vaporization of carbide forming elements.Matousek et al.18 used Cl2 for the determination of Cr, V, Ti, W, and Zr. PolytetraØuoroethylene was also used as a Øuorinating reagent to promote the vaporization of refractory elements.19±21 Goltz et al.22 used freon-23 to enhance analyte signals for the rare-earth elements in ETV-ICP-MS and Ng and Caruso23 added ammonium chloride (7% m/v) as a means of vaporizing analytes as chlorides.It has been reported in GFAAS studies that samples high in silica content cause serious effects on analyte signals. Bendicho and de Loos-Vollebregt24 reported that when analyzing glass slurries a rapid deterioration in the condition of the graphite tube ensued and a decrease in absorbance signals occurred after 50±100 Ærings.Eams and Matousek25 and Mu» ller-Vogt and Wendl26 attributed the rapid deterioration of the graphite tube when silica was present to the chemical attack of silica on graphite. Bendicho and de Loos-Vollebregt27 analyzed glass samples and reported that this effect can be minimized by using HF, and thus the deterioration effects of the graphite tube could be reduced and the lifetime of the graphite tube extended. For ETV-ICP-MS, microgram quantities of silicate materials vaporized directly into the plasma will result in signiÆcant analyte signal suppression.For this reason, silicate matrices must be eliminated prior to the vaporization step. McIntyre et al.11 used NH4F, NH4F±HF, and HF as Øuorinating reagents for the removal of SiO2. The use of NH4F and NH4F± HF was ineffective in removing silica; however, HF was able to remove much larger quantities of silica and made possible the direct determination of Ra in silicate reference materials.11 Although published separately, the work of McIntyre et al.11 J.Anal. At. Spectrom., 1999, 14, 1703±1708 1703 This Journal is # The Royal Society of Chemistry 1999was completed in this laboratory following the research reported here. The objective of this work was to determine the optimum experimental conditions for the complete removal of silica as silicon dioxide and silica as silicate in geological reference material powders using ultrasonic slurry sampling ETV-ICP-MS.Experimental Instrumentation A Perkin-Elmer SCIEX (Concord, Ontario, Canada) Elan 5000a ICP mass spectrometer equipped with an HGA-600 MS electrothermal vaporizer, a Model AS-60 autosampler, and a USS-100 ultrasonic mixing probe were used. The ultrasonic mixing probe (constructed of high purity Ti) used a mixing time of 30 s at a power level of 12 Wto provide adequate mixing and suspension of slurries. To facilitate larger sample particle sizes, the PFA tip provided with the autosampler was replaced with larger thin-walled PFA capillary tube (id 0.81 mm).Pyrolytic graphite-coated tubes were used throughout. The experimental conditions for both the Elan 5000 and the HGA-600 MS are given in Table 1. A PTFE tube of 80 cm and 6 mm id was used to interface the HGA-600 MS to the plasma torch. Optimization of the plasma and mass spectrometer was accomplished using solution nebulization, prior to switching to ETV mode. No further optimization of the ICP mass spectrometer was required with the exception of small (°50 ml min21) variations in the carrier Ar Øow rate.The operation of the HGA-600 MS was completely computer controlled. During the drying and pyrolysis steps of the temperature program, opposing Øows of argon (300 ml min21) originating from both ends of the graphite tube removed water and other vapours through the dosing hole of the graphite tube. During the high temperature or vaporization step, the dosing hole was sealed by a pneumatically activated graphite probe.Once the graphite tube was sealed, a valve located at one end of the HGA workhead directed the carrier argon gas Øow, originating from the far end of the graphite tube, directly to the argon plasma at a Øow rate of 850 ml min21. Feriments, 30Siz (3.12% abundance) was used to monitor the vaporization of silica. Data in all Ægures and tables are corrected for any background signal present. Standards and reagents High purity argon gas (99.995%, Matheson Gas Products, Ottawa, Ontario, Canada) was used.Solutions were prepared with ultra-pure water obtained from a Milli-Q water puriÆcation system (Millipore Corp., Mississauga, Ontario, Canada). The nitric and hydroØuoric acids used were analytical-reagent grade (Anachemia, Rouses Pint, NY, USA). Finely divided high-purity precipitated silica and high purity silicon carbide were obtained from Spex Industries, Metuchen, NJ, USA. Standard reference materials used were SY-2 syenite rock containing 60.11% SiO2 and BHVO-1, a basalt containing 49.94% SiO2.Preparation of slurries The slurries were prepared by accurately weighing the solid sample (e.g., 0.10 g) and placing it into a 10 ml plastic test-tube. Water was added as the liquid medium. The slurry was mixed using a vortex mixer from which was removed (during agitation) a 1 ml aliquot. If required the 1 ml aliquot was diluted further until a 1 ml aliquot was produced which contained the desired mass to volume ratio of solid to diluent.Results and discussion Freon, used as halogenation agent, has become a popular gasphase matrix modiÆer used to enhance the volatility of refractory elements in ETV-ICP-AES12,15±17 and ETV-ICPMS. 22,28 In most published work, freon has been mixed with the argon gas during the high temperature vaporization step. Freon decomposes at high temperatures to give highly reactive chlorine or Øuorine radicals.These radicals react with sample and matrix components producing relatively volatile analyte chlorides and Øuorides. In this work freon-23 was used as a possible Øuorinating reagent to remove the SiO2 by conversion to volatile SiF4. The results obtained for these experiments indicated that either large quantities of freon or very long (minutes) reaction times are required to complete the removal of silica. Even when using a pyrolysis temperature of 1000 �C, signiÆcant quantities of carbon soot were produced as a result of freon decomposition.If either the freon concentration or the reaction time were increased to levels required to completely remove silica, excess soot was produced causing unwanted plasma effects. Because of this, the use of gas-phase matrix modiÆers was abandoned in favour of using HF. Table 1 Instrument operating conditions and data acquisition parameters for ETV-ICP-MS ICP mass spectrometer– RF power 1100W Coolant Ar Øow rate 15.0 l min21 Auxiliary Ar Øow rate 900 l min21 Carrier Ar Øow rate 850 l min21 Sampler and skimmer Ni HGA-600 MS electrothermal vaporizer– Sample volume 20 ml ModiÆer volume 20 ml Clean up step 1 s ramp, 2650 �C, for 10 s Digestion step 10 s ramp, 60 �C, variable hold time Dry step 5 s ramp, 100 �C for 30 s Pyrolysis step 1 s ramp, 100 �C for 7 s Vaporization step 1 s ramp, 2500 �C for 6 s Data acquisition2 Dwell time 20 ms Scan mode Peak hop transient Points/spectral peak 1 Signal measurement Integrated counts Resolution 0.7 u at 10% peak height Fig. 1 Vaporization curve for 20 ml slurry sample containing 125 mg of SiO2. 1704 J. Anal. At. Spectrom., 1999, 14, 1703±1708Vaporization curve for silicon dioxide As an initial study on the removal of SiO2, a vaporization curve was constructed to determine at which temperature SiO2 vaporization begins and at which temperature the vaporization is complete. The vaporization curve for SiO2 was obtained without using a chemical modiÆer and is shown in Fig. 1. These data were obtained using 20 ml of the SiO2 slurry containing 125 mg of SiO2. The ETV-ICP-MS signal intensity was measured at vaporization temperatures ranging between 810 and 2600 �C. No pyrolysis step was used. The ETV-ICP-MS vaporization curve (Fig. 1) gives an appearance temperature for Si (derived from the vaporization of SiO2) of approximately 1600 �C and reaches a maximum at about 2200 �C. These temperatures coincide with the melting point (1610 �C)29 and the boiling point (2230 �C)29 of SiO2, respectively.A single Si signal was observed over the entire range of vaporization temperatures studied. Clearly, a pyrolysis temperature of 2200 �C cannot be used to remove silica since, at this temperature, many analyte elements are vaporized as well. HydroØuoric acid matrix modiÆer The ETV heating program step during which HF is added and the reaction between sample and HF can best be referred to as the digestion step.The effect of added HF on the removal of SiO2 in ETV-ICP-MS is shown in Fig. 2. For these experiments, a pyrolysis temperature of 100 �C and a vaporization temperature of 2500 �C were used following a digestion step of 60 �C (10 s ramp). The duration of the digestion step (hold time) was varied from 30 s to 300 s. A 20 ml aliquot of 50% hydroØuoric acid was added to 20 ml of the SiO2 slurry sample in the graphite tube. The ETV-ICP-MS signal intensity for Si decreased with increased hold time for the digestion step indicating that most of the silica was removed with a hold time of about 150 s.For digestion step hold times greater than 150 s, little additional SiO2 is lost. This observation is in agreement with the work of McIntyre et al.11 who used HF for the removal of silica from geological materials analysed for Ra using slurry sampling ETV-ICP-MS. The actual Si signals obtained, corresponding to the data given in Fig. 2, are shown in Fig. 3. The Si ETV-ICP-MS signals shown in Figs. 3a and 3b have two distinct signals widely separated in time. The Ærst signal appears at about 1 s Fig. 2 Effect of digestion step hold time on the integrated signal for Si(125 mg) using 20 ml 50% HF. Fig. 3 ETV-ICP-MS signals for 125 mg SiO2 using 20 ml 50% HF at hold time of (a) 30 s, (b) 60 s, (c) 90 s, (d) 120 s, (e) 150 s, (f) 180 s, (g) 300 s. J. Anal. At. Spectrom., 1999, 14, 1703±1708 1705reaching a maximum at about 2 s, and the second smaller signal appears at about 5 s with a maximum at approximately 6 s into the high temperature vaporization step.This suggests that two different Si species are released during the vaporization step and that these two species have very different volatilities. Increasing the hold time above 60 s (Figs. 3c±3g) shows no change in the Si signals with the exception that the second silica signal persists and is somewhat variable indicating that some residual unreacted silica remained in the graphite tube.Origin of Si ETV-ICP-MS signals To investigate the nature of the two Si species observed, the effect of changing the ramp time (time to maximum temperature) or graphite tube heating rate was studied. For these experiments an HF digestion step of 60 �C with a hold time of 30 s, a pyrolysis temperature of 100 �C (7 s hold time) and a vaporization temperature of 2500 �C were used. The effect of changing the graphite tube heating rate during the vaporization step is shown in Fig. 4. The ramp time was varied between 1 s and 10 s giving a heating rate of between 2400 �C s21 and 240 �C s21, respectively. At a ramp time of 1 s, the Ærst signal (Fig. 4a) appeared at 1 s with a signal maximum at about 2 s. The second signal occurred at about 5 s reaching a maximum at 7 s. As the ramp time was increased, there was a large shift in the appearance and signal maximum times of the second signal, and a correspondingly small shift in the signal times for the Ærst signal.The higher the ramp time (slower heating rate) the greater the shift in the second signal. Fig. 4c shows the ETV-ICP-MS signal for Si at a ramp time of 10 s. The appearance time for the Ærst signal was 2 s and for the second signal between 12 and 19 s. The fact that a radical change in the heating rate of the graphite tube results in almost no change in the signal position, height and width of the Ærst Si signal while resulting in a very large change in the shape and size of the second signal suggests that thbut have widely different vaporization characteristics.Since the Ærst, larger, signal appeared at 2 s using a 10 s ramp time, it can be calculated that the appearance temperature of this Si species is about 480 �C. The melting point and boiling point of SiO2 are 1610 and 2230 �C respectively, indicating that the Ærst signal cannot be SiO2 but could be due to SiF4 (bp 286 �C),29 which was formed by the addition of HF as a chemical modiÆer in the digestion step.Both Øuorine and HF are known to intercalate in graphite30 and it is possible that SiF4 is also thermally stabilized through intercalation. From this, it can be surmised that although most of the SiF4 was lost during the reaction and pyrolysis steps, some SiF4 remained and perhaps became intercalated into the graphite tube only to be released later at higher temperatures during the vaporization step.In a separate study, Gre�goire et al.31 reported that small quantities of HCl were retained within the graphite tube by intercalation even after pyrolysis at 400 �C. Fig. 5 shows ETV-ICP-MS signals for Si obtained from the vaporization of (a) SiO2 and (b) SiC, both without modiÆer. A ramp time of 1 s was used for the vaporization step. Fig. 5a shows that the signal for SiO2 occurred between 3.5 and 10 s with the maximum occurring at about 5.9 s, which was coincident with the second Si signal shown in Fig. 4a. Fig. 5b shows that the signal for SiC occurred between 3.5 and 12 s with the maximum at about 7.3 s. This may suggest that the small second signal in Fig. 4a, which appeared at a later time, could be due to unreacted SiO2 remaining following the digestion step and then released at a later time during the vaporization step. It may also be attributed to the release of a mixture of SiO2 and SiC or SiC alone, which may be formed at high temperatures.Optimization of HF matrix modiÆcation for silica removal In Fig. 3g it was shown that, for a digestion step hold time of 300 s, the Ærst larger Si signal essentially disappeared and that only a small decrease in the second Si signal was observed. This indicated that using only 20 ml HF possibly did not convert all of the SiO2 to SiF4. To investigate the effect of increasing the amount of HF on the ETV-ICP-MS signal intensity for Si, the volume of 50% HF used was increased from 20 ml to 70 ml.To ensure an adequate reaction time, a digestion step hold time of 300 s was used. Fig. 6 shows that the second signal (from Fig. 3) appeared at slightly earlier times with increasing amounts of HF and that the second signal was almost completely removed when 70 ml of HF was used. This suggests that there was enough HF and a sufÆcient holding time for the Fig. 4 ETV-ICP-MS signals for 125 mg SiO2 using 20 ml 50% HF at different vaporization ramp times, (a) 1 s, (b) 5 s, (c) 10 s.Hold time at the digestion step was 30 s. Fig. 5 ETV-ICP-MS signals for (a) SiO2 and (b) SiC without using HF as a chemical modiÆer; vaporization step ramp time of 1 s. 1706 J. Anal. At. Spectrom., 1999, 14, 1703±1708SiO2 to react completely with HF to produce SiF4 and that there was also enough time to volatilize SiF4 during the digestion and pyrolysis steps. To obtain the most practical conditions for removing silica, we also determined the reaction step holding time required to eliminate SiO2 when using 70 ml of 50% HF.The results show that the Si signal intensity decreases as the digestion step hold time is increased and essentially disappears at a hold time of 150 s. This result is similar to that obtained when using 20 ml of 50% HF (Fig. 3) with the exception that virtually no second Si signal is apparent. This may mean that the second signal observed earlier was due to the presence of unreacted silica or silica that had escaped contact with the HF.For the sake of brevity, this data is not given here. An important Ænding resulting from this study was that at a hold time of 180 s (using 70 ml 50% HF) the graphite tube failed catastrophically with the end of the graphite tube downstream from the dosing hole (end nearest to plasma) completely breaking off. To determine whether this effect was due to the presence of silica or hydroØuoric acid, the experiment was repeated using only 70 ml of 50% HF and 20 ml H2O.When using only HF, the tube also failed indicating that tube fracture was the result of the corrosive effect of HF on graphite. From this we conclude that using 20 ml of 50% HF and a reaction hold time of 150 s at the digestion step is optimal for the removal of silica while maintaining the integrity of the graphite tube. The small quantities of silica remaining when using 20 ml of HF are not signiÆcant and did not detract from the determination of Ra in silicate materials using slurry sampling ETV-ICP-MS.11 Silica removal from geological reference materials Silica is, of course, the simplest of silicate materials and so it is possible that the removal of Si from more complex silicates may be more difÆcult relative to pure silica.The method developed above was applied to two geological reference materials: syenite SY-2 (60.11% SiO2) and basalt BHVO-1 (49.94% SiO2). Fig. 7 shows the ETV-ICP-MS signals for Si obtained when 0.125 mg BHVO-1 basalt reference material was vaporized (a) without using a chemical modiÆer and (b) using 20 ml of 50% HF as a chemical modiÆer with a reaction time of 150 s.Fig. 7a shows a single large Si signal appearing at about 3.9 s reaching a maximum at about 5.8 s. This signal can be attributed to SiO2. Fig. 7b shows a smaller signal appearing at about 1.6 s with a maximum at about 2.9 s with a very small shoulder appearing at about 4 s.The Ærst signal can be attributed to the release of SiF4 as discussed earlier in this paper. The shoulder may be due to a very small quantity of unreacted silica. This result demonstrates that almost all the silica was removed by using 20 ml of 50% HF (150 s reaction time) and that complex silicate materials can be successfully digested using HF. The results obtained for SY-2 were identical to those reported above for BHVO-1. References 1 S. C. Stephen, D. Littlejohn and J.M. Ottaway, Analyst, 1985, 110, 573. 2 C. Bendicho and M. T. C. de Loos-Vollebregt, J. Anal. At. Spectrom., 1991, 6, 353. 3 N. J. Miller-Ihli, Anal. Chem., 1992, 64, 964A. 4 D. C. Gre�goire, N. J. Miller-Ihli and R. E. Sturgeon, J. Anal. At. Spectrom., 1994, 9, 605. 5 R. W. Fonesca and N. J. Miller-Ihli, Appl. Spectrosc., 1995, 49, 1403. 6 M. Liaw and S. Jiang, J. Anal. At. Spectrom., 1996, 11, 555. 7 N. J. Miller-Ihli, Spectrochim. Acta, Part B, 1996, 51, 1591. 8 S. Hauptkorn, V.Krivan, B. Gercken and J. Pavel, J. Anal. At. Spectrom., 1997, 12, 421. 9 M. Liaw, S. Jiang and Y. Li, Spectrochim. Acta, Part B, 1997, 52, 125. 10 R. W. Fonesca, N. J. Miller-Ihli, C. Sparks, J. A. Holcombe and B. Shaver, Appl. Spectrosc., 1997, 51, 1800. 11 R. St. C. McIntyre, D. C. Gre�goire and C. L. Chakrabarti, Spectroscopy, 1998, 13, 18. 12 G. F. Kirkbright and R. D. Snook, Anal. Chem., 1979, 51, 1938. 13 G. Zaray, T. Kantor, G. Wolff, Z. Zadgorska and H. Nickel, Mikrochim.Acta, 1992, 107, 345. 14 G. Zaray, F. Leis, T. Kantor, J. Hassler and G. Tolg, Fresenius' J. Anal. Chem., 1993, 346, 1042. 15 J. M. Ren and E. D. Salin, Spectrochim. Acta, Part B, 1994, 49, 555. 16 J. M. Ren and E. D. Salin, Spectrochim. Acta, Part B, 1994, 49, 567. 17 J. Allary, G. Hernadez and E. D. Salin, Appl. Spectrosc., 1995, 49, 1796. 18 J. P. Matousek, R. T. Satumba and R. A. Bootes, Spectrochim. Acta, Part B, 1989, 44, 1009. 19 M. Huang, Z. Jiang and Y. Zeng, J. Anal. At. Spectrom., 1991, 6, 221. 20 M. Huang, Z. Jiang and Y. Zeng, Anal. Sci., 1991, 7, 773. 21 Z. Jiang, B. Hu, Y. Qin and Y. Zeng, Microchim. J., 1996, 53, 326. 22 D. M. Goltz, D. C. Gre�goire and C. L. Chakrabarti, Spectrochim. Acta, Part B, 1995, 50, 1365. 23 K. C. Ng and J. A. Caruso, Analyst, 1983, 108, 476. 24 C. Bendicho and M. T. C. de Loos-Vollebregt, Spectrochim. Acta, Part B, 1990, 45, 679. 25 J. C. Eams and J. P. Matousek, Anal. Chem., 1980, 52, 1248. 26 G. Mu» ller-Vogt and W. We, 1981, 53, 651. 27 C. Bendicho and M. T. C. de Loos-Vollebregt, Spectrochim. Acta, Part B, 1990, 45, 695. 28 B. Wanner, P. Richner and B. Magyar, Spectrochim. Acta, Part B, 1996, 51, 817. Fig. 7 ETV-ICP-MS signals for SiO2 from 125 mg of basalt (BHVO-1), (a) without using HF as a chemical modiÆer and (b) using 20 ml of 50% HF chemical modiÆer. Digestion step hold time: 150 s. Fig. 6 Effect of HF volume on the ETV-ICP-MS signals for SiO2, (a) 20 ml, (b) 40 ml, (c) 60 ml, (d) 70 ml of HF. Hold time at the digestion step was 300 s. J. Anal. At. Spectrom., 1999, 14, 1703±1708 170729 Handbook of Chemistry and Physics, ed. D. R. Lide, 74th edn, CRC Press, Cleveland, OH, 1994. 30 W. Slavin, Graphite Furnace AAS: A Source Book, The Perkin Elmer Corporation, Norwalk, CT, 1984, p. 45. 31 D. C. Gre�goire, D. M. Goltz, M. M. Lamoureux and C. L. Chakrabarti, J. Anal. At. Spectrom., 1994, 9, 919. Paper 9/03269J 1708 J. Anal. At. Spectrom., 1999, 14, 1703±17