JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY OCTOBER 1991 VOL. 6 553 Study of Internal Standardization for Analysis of Powdered Samples Using a Theta Pinch Discharge Zuwei Wang* and Alexander Scheeline School of Chemical Sciences University of Illinois 1209 W. California Street Urbana lL 61801 USA Theta pinch discharges are capable of sampling ceramics and other refractory substances for atomic emission spectrometry. Previously only qualitative information has been obtained. Research on the quantitative aspects of this technique has become feasible owing to improvements in the detection system (a charge coupled array detector and echelle spectrometer) and completion of preliminary parametric studies. This initial quantitative study of the direct sampling of powdered samples using single theta pinch discharges reports the shot-to-shot reproducibility [relative standard deviation (RSD) 20-30%] internal standard correlation (RSD 10-1 5%) and sampling curves for several elements.Keywords Theta pinch discharge; atomic emission spectrometry; internal standard; direct solid sampling Considerable problems remain for the elemental analysis of ceramics and other refractory solids. Most methods require lengthy dissolution fusion or dilution pre-treatment. Con- tamination and loss of volatile species can occur during these procedures in spite of precautions. Developments in the field of ceramics require analysts to research into reliable direct methods of elemental analysis further and also to study suitable certified reference materials for calibration.' Many publications have appeared on elemental analysis of ceramics.Some direct methods have been reviewed.lJ Most research on the direct analysis of ceramics and other refractory materials by plasma spectrometry has focused on techniques such as spark ablati0n,~9~ laser ablati~n,~ nebu- lization of ~lurries,~?~ direct i n ~ e r t i o n ~ ~ ~ - ~ and electrothermal atomi~ation.~ These techniques have advantages in terms of speed ease of sample preparation and low detection limits (in some instances) compared with wet methods but still need further improvement to meet the requirement of fast routine analysis for quality control with adequate accuracy independent of the form particle size melting-point thermal history and concomitant composition of the ana- lyte material. The theta pinch discharge has been developed as an additional technique for direct sampling of ceramics and other refractory materials because of its high sampling energy.Previously only qualitative information could be confidently obtained in this laboratory owing to deficien- cies in the available detection technology; qualitative performance was however promising. Research on the quantitative aspects of this technique has become feasible because a charge coupled array detector and Cchelle spectrometer have been used for detection17J8 and pre- liminary parametric studies have been completed. l4 Initial quantitative studies of spectrochemical analysis of powdered samples by single theta pinch discharges with investigations of the shot-to-shot reproducibility internal standard correlation and sampling curves for several elements are presented. Experimental Low-pressure Argon Plasma According to earlier work reported by White,16 low-pressure argon is a satisfactory discharge gas for the theta pinch.The optimum argon pressure is element dependent but a pressure of 20 Torr (2.666 x lo3 Pa) gives a satisfactory sampling efficiency and signal-to-noise ratio for the elements determined in this work. * On leave from the Dalian Institute of Chemical Physics Dalian China. Mirrored Coil for Energy Transfer A mirrored coil described previously,lOJ1 was used for the analysis of powdered samples. It has a satisfactory energy transfer efficiency for the direct sampling of solids in the theta pinch discharge. A comparison of this design with other types of coil is currently being made.Polymer Cup for Sample Introduction Samples were held 10 mm from the axis of the discharge tube in a holder made of poly(methy1 methacrylate). This material was chosen because polycarbonate holders decom- pose giving rise to a carbon coating on the analyte powders. Typically a pellet specimen with dimensions 2.25 mm deep and 5.25 mm in diameter was employed. The cup was held on a mandrel of length 155 mm which was held by one of the pinch chamber end caps. In order to prevent contamina- tion a clean sample cup was used for each specimen. Sample Powders and Sample Preparation Analytical-reagent grade powders of the various oxides were used as received. Mixtures were prepared by weighing appropriate amounts on an analytical balance (total mass approximately 500 mg) then hand grinding in an agate mortar for 30 min.Approximately 30 mg of the mixture were then compressed into a pellet in the sample cup. Although a pellet press was also used it was found that placing the powder in the cup and packing it manually was more convenient and generated larger signals because of the ease with which the loose surface powder can be sampled. Particle size was generally 10 pm or less as measured by optical microscopy. Detailed studies of particle size effects are planned. For all experiments the diluent powder was A1203. Impurity concentrations as measured in an assay of a portion of the powder using inductively coupled plasma atomic emission spectrometry are listed in Table 1.In order to determine these concentrations the A1203 was fused with LiB02 dissolved in HN03 and 0.6 mol dm-3 0.12 mol dmP3 HC1 and then aspirated. A 100 mg portion of the sample was ultimately diluted to 50 ml of solution. Reagent blanks were subtracted from all measured concen- trations. Background equivalent concentrations (BEC) in the plasma were no higher than 2.5 ppm (equivalent in the undigested oxide powder) for the elements listed in Table 1 except for potassium which had a BEC of 1 ppm in the plasma or equivalently 500 ppm in the oxide powder. Binary or ternary mixtures of oxide powders were employed to obtain working curves. For experiments on554 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY OCTOBER 1991 VOL. 6 Table 1 Impurities in the A1203 matrix Element Concentration (ppm) Na Ti Ca Fe K Mg 2300 1030 420 345 <loo 23 measurement precision the mixture compositions given in Tables 3 and 4 were used.Charge Coupled Detector and Echelle Spectrometer Previous work16 has demonstrated that it is not possible to use the theta pinch for quantitative analysis unless internal standardization is routinely employed and real-time diag- nosis of self-reversal is possible which necessitates the use of multichannel detection. An kchelle spectrometer de- signed and constructed in the laborato~y,~~J* with a charge coupled array detector (CCD) (Thomson-CSF 7882 CCD Photometrics Tucson AZ USA) cooled by liquid nitrogen was employed. A wide range of wavelengths was accessible allowing simultaneous monitoring of many elements.All the data presented here were obtained from individual pinch firings. While much of this work could have been accomplished using a gated photomultiplier tube (PMT) and a direct To vacuum I Discharge vessel .Sample holder r - - I/ Controller w Power Grounded wall ichelle spectrometer IE EE488 I I Fig. 1 Experimental set-up. Discharge vessel shown both in the top and end-on view. Mirrors Mpl and Mp2 are off-axis paraboloids Table 2 Experimental parameters Parameter Value Voltage 26 kV Current (peak) 85 kA Capacitance 6.05 pF Argon pressure 19.5 Torr Sample position (distance off axis of discharge vessel) (2.600 x lo3 Pa) 10 mm 1 Exposure duration (No. of discharges) reading spectrometer two characteristics of the pinch make the use of the CCD-Cchelle spectrometer combination particularly effective.Firstly the linewidths and shapes appear to be irreproducible from shot to shot but are consistent between elements on individual firings. Thus the ability to select an effective linewidth on each firing improves the precision. Furthermore any unanticipated interferences may be more readily seen with full spectral coverage (CCD or photographic plate) than with isolated line measurements (PMT). Secondly the pinch produces a significant amount of radiofrequency interference (RFI). While use of fibre optics for digital signal transmission and a Faraday cage around the discharge limits the difficulties analogue measurement during a discharge is nevertheless difficult. By recording optical signals on the CCD during the discharge and reading the data when the RFI has subsided the RFI ceases to be a problem.This is not possible using gated integration with PMTs. For all measurements the background was subtracted by interpolating a linear baseline between points adjacent to the line of interest. Lines were sufficiently broad that wings of nearby lines might have led to overestimation of the continuum with a consequent underestimation of line intensity. Integrated intensities rather than peak heights were employed for all lines. The equipment used is shown in Fig. 1 and some important experimental parameters are shown in Table 2. According to the spatially resolved emission experiments the best sampling position is about 10 mm off the axis of the discharge ve~se1.l~ The CCD shutter was opened for 2 s encompassing the moment of discharge firing resulting in collection of time-integrated data.Results and Discussion Discharge Reproducibility and Internal Standardization Although a precedent from other discharge systems sug- gested that internal standardization would probably im- prove reproducibility adequate characterization of internal standard signals in the pinch has not previously been reported. The shot-to-shot reproducibility of A1 I at 396.2 nm and Sr I1 at 42 1.6 nm in the theta pinch is shown in Fig. 2. A total of 20 single theta pinch discharges were used. For ease of visual comparison the raw A1 I signal was multiplied by a scale factor of 2.5. The intensity fluctua- tions for the individual lines are pronounced. The ratio of Sr I1 to A1 I emission is also shown and has a lower relative standard deviation (RSD) (10%) and range (20% of the mean) than the raw emission data (RSD 20%; range 7000 I 1 200 A v) I-’ .- 5 6000 2 5 5000 f 5 4000 .- >.v) I-’ .- 3000 4- c .- p 2000 g 1000 4d I I-’ - 0 2 4 6 8 10 12 14 16 18 20 Pinch number Fig. 2 Shot-to-shot reproducibility and internal standard correla- tion. A Sr I1 42 1.6 nm emission; B scaled A1 I 396.2 nm emission; and C ratio of Sr 1I:Al I emissionJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY OCTOBER 199 1 VOL. 6 555 Table 3 Improvement of reproducibility by using several elements as internal standards (20 replicates per sample) RSD (O/O) Element Internal standard Without With Sample and line/nm and line/nm internal standard internal standard A* A1 I 396.2 Sr I1 421.6 B* Mg I1 279.5 Sr I1 421.6 C t Mg I1 280.3 Ca I1 317.9 D$ Ti I1 376.5 Ti I1 379.5 * Composition (Yo m/m) MgO 5 ; MnO 5; SrO 5; and A1203 85.t Composition (Oh m/m) MgO 5 ; MnO 5; CaO 5; and A1203 85. $ Composition (Yo m/m) TiO 20; and A1203 80. 22 19 23 18 10 10 13 5 Table 4 Improvement of reproducibility by using the matrix element Al I as an internal standard (20 replicates per sample) RSD (O/O) Element Internal standard Without With Sample and line/nm and line/nm internal standard internal standard E* Mg I1 280.3 A1 I 396.2 Ft Ti I1 376.1 A1 I 396.2 G$ Ti I1 368.5 A1 I 396.2 H* Ca I1 317.9 A1 I 396.2 23 21 30 29 15 8 9 13 * Composition (Yo m/m) MgO 5; MnO 5; CaO 5; and A1203 85. t Composition (Yo m/m) MgO 5; Ti02 5; and A1203 90.$ Composition (Oh m/m) MnO 5; Crz03 5; TiOz 5 ; and A1,03 85. 80% of the mean). Sample transport rather than photon shot noise is the cause of the signal fluctuations. The transport varies between discharges for a number of reasons including differences in the packing of sample particulates plasma fluid dynamics and energy transfer into the plasma and from the plasma to the sample. Very little of the sample is consumed when the pinch is fired; the top layer of powder is vaporized. While the change in the surface is evident the thickness of the pellet does not appear to change to the naked eye. Erosion depth is certainly less than 0.5 mm and might be substantially less. Clearly intensity reproducibility of 20-30% RSD does not meet the analytical requirements for fast routine quality control analysis for the production of ceramics.Fig. 2 suggests however that internal standardization is a viable alternative for general analysis. Several elements were used as internal standards. Im- provement in the RSD was different for each element (shown in Table 3 for some elements). The last line pair (Table 3) is an example of self-correlation between two Ti I1 lines giving an RSD of about ~O/O which is the best of any of the examples. Apparently much of the uncertainty in emission intensity is due to sample transport but a residual uncertainty is due to other factors which might include a small contribution from shot noise in the photometric signal readout and digitization noise in the CCD excita- tion fluctuations between discharges or in different spatial regions of a single discharge and background fluctuations.In all these examples A1 was vaporized from A1203 the major matrix substance. Other elements were present in the amount of approximately 5% m/m in the form of the oxide. The major matrix element Al was also used as an internal standard in order to avoid the need for mixing a separate internal standard with each specimen. The shot-to- shot reproducibility was improved (Table 4). By using A1 I at 396.2 nm as an internal standard the reproducibility of Ti I1 is improved by a factor of 3 compared with the use of no internal standard and is nearly as good as the self- correlation (Table 3). Plausible reasons for the good correlations include their similar boiling-points and wave- lengths.Apparently whether the emitting species is an atom or ion is not a main consideration for internal standard selection with this discharge. Ti Sampling From TiOz Titanium is an important trace element in several types of ceramics. The variation of concentration of this element can affect some qualities of the ceramics such as electrical conductivity. When the 396.2 nm neutral emission line of A1 (the major matrix element other than oxygen) is used as an internal standard for Ti 11 both the reproducibility and linearity are satisfactory as shown in Fig. 3. A typical concentration of Ti in the ceramics studied is about 0.5Oh m/m which is in the linear region of the working curve. Thus direct quantitative determination of minor elements such as Ti in ceramics is possible using a theta pinch discharge. I A [Ti] ("A m/m) Fig.3 Working curve for Ti as Ti02 in A1203 with Al as an internal standard. Abscissa concentraton (O/o d m ) includes the amount of Ti known to be in the blank556 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY OCTOBER 199 1 VOL. 6 03 1 I I I Mg Sampling From MgO Another important element present in ceramics is Mg. While Mg is a minor or trace element in some ceramics it was investigated here as a major species. Data are shown for Mg sampling from MgO in Fig. 4. The intensities were measured at three different wavelengths 279.5 nm (reso- nance line) 279.1 nm and 279.8 nm (non-resonance lines). All three lines give rise to non-linear working curves in the absence of internal standardization. Generally in emission sources the resonance lines of Mg I1 are reversed when the concentration of Mg is high.This has been confirmed both in the literature19 and by spark experiments using this detection system.20 Thus one would expect to be able to detect reversal in the pinch. Initially dips in the centre of the 279.5 nm line were interpreted as evidence for reversal. Later investigation showed that the line shape is irreprodu- cible from one firing to the next. While self-reversal is the likely cause of the central dip the extent of a central dip varies substantially from shot to shot rendering definitive experiments difficult. Furthermore line broadening is sufficient that a significant portion of the light falls outside most conceivable Doppler absorption linewidths.The use of Mn I1 at 294.9 nm proved to be acceptable as an internal standard for Mg. Linear working curves as seen in Fig. 4(b) were obtained. There appears to be a reduction in the total sample uptake as the concentration of MgO is increased (Fig. 5). The Mn emission decreases as the Mg concentration increases although the Yo m/m of Mn is constant and the at.-% of Mn changes by less than 3 parts per thousand. Magnesium oxide has a higher melting- and boiling-point than either of the other oxides. For MgO to alter the signal from Mn there must either be a change in x 1200 I- 1 I ( b ) 4t 0 3 6 9 12 [MgO] (% m/m) Fig. 4 Sampling of Mg in theta pinch. (a) No internal standardiza- tion and (b) with Mn I1 294.9 nm used for internal standardization. Intensities measured at A 279.1 nm; B 279.5 nm (resonance line); and C 279.8 nm.Concentration in O/o m/m; three replicate measurements per point 600 5 500 0 400 0 > 300 cn w 8 9 4- In .- c 200 - 100 I 0 I 1 0 3 6 9 12 [MgO] (“A m/m) Fig. 5 Intensity of Mn I1 as a function of [Mg]; [Mn] = 6% m/m in an A1203 matrix. Concentration in O/o m/m; three replicate measurements per point excitation conditions as the concentration of Mg changes or there must be a change in the uptake of all species despite the fact that the specimens are mixtures of pure powders. Although there is no evidence of changes in the excitation of the support gas with the introduction of the analyte there is weak evidence that sampling is changed proportionally for all analyte species even when the melting-point of a component of the sample changes.An alternative explanation is that particle size and hardness are critical experimental variables and MgO causes clumping of the mixed powder samples. Larger particles give smaller signals for all species. While the linearity seen in the data presented here is promising examination of the effects of particle size and analyte speciation must be pursued. The effects of changes in melting-point and hardness must be isolated from the effects of particle size alone. Conclusion Sample transport variability is the major noise source in theta pinch emission spectrometry. Thus internal standardization is a useful means of improving the reprodu- cibility of analytical measurements. Not only is the signal- to-noise ratio improved but the working curve linearity improves also.While much work must still be carried out to develop the theta pinch as a routine analytical source the feasibility of obtaining quantitative measurements has now been demonstrated. While we are aware of the Generalized Internal Standard method21*22 and n o r m a l i z a t i ~ n ~ ~ ~ ~ ~ techniques which could benefit from the multiplicity of available analytical lines an attempt to characterize the use of such methods together with the pinch has not yet been made. They are a logical means of using a large amount of data that has been obtained but not yet processed. This work was supported by the Office of Basic Energy Sciences US Department of Energy (Grant DE-FG02 84ER 132 18). The technical assistance of Duane L.Miller is greatly appreciated. References 1 Broekaert J. A. C. Graule T. Jenett H. Tolg G. and Tschopel P. Fresenius 2. Anal. Chem. 1989 332 825. 2 Ishizuka N. and Morikawa H. Bunseki 1986 471; Anal. Abstr. 1987 49 4B187. 3 Broekaert J. A. C. Leis F. Raeymaekers B. and Zaray Gy. Spectrochim. Acta Part B 1988 43 339.JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY OCTOBER 199 1 VOL. 6 557 4 Aziz A. Broekaert J. A. C. Laqua K. and Leis F. Spectrochim. Acta Part B 1984 39 109 1. 5 Technical Information PQ 709 A VG Elemental Winsford Cheshire 1987. 6 Raeymaekers B. Graule T. Breokaert J. A. C. Adams. F. and Tschopel P. Spectrochim. Acta Part B 1988 43 923. 7 Reisch M. Nickel H. and Mazurkiewicz M. Spectrochim. Acta Part B 1989 44 307. 8 Karanassios V. and Horlick G.Spectrochim. Acta Part B 1990,45 85 119 and 129. 9 Kirkbright G. F. and Snook R. D. Anal. Chem. 1979 51 1938. 10 Kamla G. J. and Scheeline A. Anal. Chem. 1986 58 923. 11 Kamla G. J. and Scheeline A. Anal. Chem. 1986 58 932. 12 White J. S. and Scheeline A. Anal. Chem. 1987 59 305. 13 Scheeline A. and White J. S. Spectrochim. Acta Part B 1988 43 55 1. 14 White J. S. Lee G. H. and Scheeline A. Appl. Spectrosc.. 1989 43 99 1. 15 Scheeline A Lee G. H. and White J. S. Appl. Spectrosc. 1990,44 108 1. 16 White J. S. Ph.D. Thesis University of Illinois 1988. 17 Scheeline A. Bye C. A. Rynders S. W. and Miller D. L. in Optical and Spectroscopic Techniques for the 19903 SPIE Proceedings V. eds. Learner J.M. and McNamara B. J. SPIE Bellingham WA 1990 vol. 1318 p. 44. 18 Scheeline A. Bye C. A. Rynders S. W. and Miller D. L. Appl. Spectrosc. 1991 45 334. 19 Coleman D. M. and Walters J. P. J. Appl. Phys. 1977 48 3297. 20 Bye C. A. personal communication 1990. 21 Lorber A. and Goldbart Z. Anal. Chem. 1984 56 37. 22 Lorber A. Goldbart Z. and Eldan M. Anal. Chem. 1984 56 43. 23 Beaty J. S. and Belmore R. J. J. Test. Eval. 1984 12 212. 24 Practice for Fundamental Calculations to Convert Intensities into Concentration in Optical Emission Spectrochemical Analy- sis Annual Book of ASTM Standards 3.06 E l 58-86 American Society for Testing and Materials Philadelphia PA 1990. Paper I /00023C Received February Ist 1991 Accepted May 30th 1991