Determination of Impurities in Boron Nitride Powder by Slurry Sampling Electrothermal Atomic Absorption Spectrometry VILIAM KRIVAN* AND THOMAS RO » MMELT Sektion Analytik und Ho » chstreinigung, Universita » t Ulm, D-89069 Ulm, Germany A slurry sampling ETAAS method for the determination of Al, for a modiÆer beaker (autosampler location 0). Furthermore, the original PTFE sampling capillary of the autosampler was Ca, Cd, Co, Cu, Fe, K, Mg, Mn, Na, Ni, Pb, Si and Zn in powdered boron nitride is described.The main aspects in the replaced by one with a larger internal diameter (0.8 mm) to avoid Æltration effects. Pyrolytic graphite coated graphite development of the method included preparation of slurry samples, pyrolysis and atomization parameters, minimization ringed tubes with fork-shaped platforms (Part No. BO 505057) and pyrolytic graphite coated graphite tubes (Part No. of the matrix effect by chemical modiÆcation and calibration. The accuracy was checked by comparison of the results with BO 091504) were used for platform and wall atomization, respectively.Suspensions were pre-treated in a Sonorex RK those obtained by analysis of sample digests by AAS and ICPAES. The LODs achieved were of the order of 2 (Cd) and 400 255 H (Bandelin Electronic, Berlin, Germany) ultrasonic bath. For the determination of Ca, Mg and Na by the Øame (Si) ng g-1. technique, a Perkin-Elmer Type 400 atomic absorption spec- Keywords: Electrothermal atomic absorption spectrometry; trometer was used.boron nitride matrix; slurry sampling; trace analysis A Jobin-Yvon JY-24 spectrometer (Longjumeau, France) extended with a JY-50P polychromator (15 elements) was used Boron nitride, mainly because of its outstanding thermal and for sequential (elements Na and K) and simultaneous (all other electrical properties, has become an important material in elements) measurements. The JY-50P allows wavelength scan- various technological Æelds.1±3 In many instances, the relevant ning by moving the entrance slit of the polychromator. Control properties are inØuenced by trace impurities.Therefore, there of the spectrometer and data collection were performed using is a need for adequate analytical methods for the trace charac- a computer system and the ISA/Jobin-Yvon software package. terization of this material. Scanning electron micrographs of the boron nitride powders ETAAS is one of the most powerful routine methods for the were obtained with a Zeiss DSM 962 microscope (Zeiss, determination of trace impurities in a large variety of materials.Oberkochen, Germany) with magniÆcation ranging from 100 For analysis by solution ETAAS, boron nitride can be decom- to 10000 times at an accelerating voltage of 20 kV. posed by fusion with alkali salts.4,5 However, the application of these procedures to boron nitride samples of higher purity Sample and Reagents grades is seriously limited by the blank, and often the introduction of high salt concentrations can become a problem for the The boron nitride powder examined (Elektroschmelzwerke, analysis.Digestion procedures using acid mixtures containing Kempten, Germany) was a commercially available material. It hydroØuoric acid in an autoclave have proved to be more was a Æne powder with typical particle size in the range useful for this purpose.6±8 Nevertheless, these methods are time 0.2±0.7 mm, with the largest particle size not exceeding 1.5 mm.consuming and also limited by the blank values. The particle size was estimated by SEM. For the estimation Slurry sampling ETAAS has proved to be a very advanta- of the LODs of some elements, a sample was prepared by geous method, especially for the analysis of high-purity mate- grinding a high-purity boron nitride sheet obtained from rials that are available in powdered form, as it requires no Ringsdorffwerke (Bonn, Germany). The resulting powder con- sample decomposition and only little sample pre-treatment sisted of particles with a maximum thickness of 5 mm and and therefore the contamination risk is essentially lower.The length of between 10 and 100 mm. method has already been successfully applied to the analysis All reagents and standard solutions used were obtained from of silicon based ceramics,9,10 zirconium dioxide,11 graphite12 Merck (Darmstadt, Germany). Laboratory-reagent grade etha- and molybdenum oxide13 and to the determination of silicon nol and nitric acid of pro analysi quality were puriÆed by sub- in boron nitride,14 titanium dioxide and zirconium dioxide.15 boiling distillation. HydroØuoric acid as well as magnesium In the present paper, slurry sampling ETAAS was applied nitrate and calcium nitrate, used as chemical modiÆers, were to the determination of 14 trace impurities in powdered boron of Selectipur quality.Doubly distilled water was used for the nitride.Solution ETAAS and ICP-AES were used as indepen- preparation of slurries, standards and chemical modiÆer dent reference methods. solutions. EXPERIMENTAL Procedure for Slurry Sampling ETAAS Apparatus A volume of 15 ml of a mixture of sub-boiled ethanol and doubly distilled water (1+1) containing the appropriate All absorption measurements using the ETAAS technique were carried out with a Perkin-Elmer Zeeman 5000 atomic absorp- amount of the chemical modiÆer (see Table 1) was prepared in a cleaned polystyrene beaker and stirred using a PTFE-covered tion spectrometer equipped with an HGA-500 graphite furnace, an AS-40 autosampler and a 3600 Data Station.This system magnetic bar. Aliquots of 20 ml were taken by the autosampler to check the blank value. For the preparation of sample was adapted to slurry sampling introduction by Æxing a remote controlled rotating magnet over the position commonly used slurries, 50±350 mg of the boron nitride powder were added Journal of Analytical Atomic Spectrometry, February 1997, Vol. 12 (137±141) 137Table 1 Instrumental parameters and experimental conditions for the analysis of boron nitride by slurry sampling ETAAS Gas Øow*/ Element Wavelength*/nm Slit/nm Pyrolysis/°C Atomization/°C ml min-1 ModiÆer� Tube type� Al 256.8 (309.3) 0.7 1700 2500 200 (300) A I Ca 239.9 (422.7) 0.7 900 2700 0 (300) A II Cd 228.8 0.7 900 2200 0 A I Co 242.5 0.2 1400 2500 0 A I Cu 324.7 0.7 1300 2500 0 A I Fe 344.1 (248.3) 0.2 1400 2500 300 (0) A I K 769.9 (766.5) 1.4 900 2300 300 A II Mg 202.5 (285.2) 0.7 900 2200 100 (300) – – Mn 279.5 0.2 1400 2200 300 (0) A I Na 330.3 (589.0) 2.0 (1.4) 900 2200 0 (300) A I Ni 232.0 0.7 1400 2700 0 A I Pb 283.3 0.7 850 2200 0 A I Si 251.6 0.2 800 2700 300 (0) A, B I Zn 213.9 0.7 700 2200 300 C I * Values in parentheses were used only for estimation of LODs.� Chemical modiÆers (per atomization): A, 50 mg of Mg(NO3)2; B, 50 mg of Ca(NO3)2 ; and C, 25 mg of Mg(NO3)2.� Tube type: I, pyrolytic graphite coated ringed tube with fork-shaped platform; and II, pyrolytic graphite coated tube. to this solution. The resulting slurry was pre-treated in the the slurries. Therefore, an alternative medium was required which would effectively reduce the surface tension, could be ultrasonic bath for 15 min in order to break-up particle agglomerates. Under continuous stirring with a PTFE-coated easily puriÆed and would be miscible with water (which is important for calibration by standard additions using aqueous magnetic bar, 20 ml aliquots of the homogenized slurries were dispensed automatically into the tube by the autosampler from standard solutions).A mixture of equal volumes of sub-boiled ethanol and doubly distilled water proved to fulÆl these position 0, which is usually assigned for the vessel containing the modiÆer solution. The optimized instrumental parameters requirements well. In spite of the relatively low boiling-point of ethanol, losses in mass during 2 h of continuous stirring in and other experimental conditions used are summarized in Table 1.Integrated absorbance evaluation was used for all a closed beaker with a hole in the lid did not exceed 1.5%. However, because of ted surface tension, the maximum measurements. QuantiÆcation was carried out using the standard additions method by spiking the slurries with aqueous dispensible volume into the cavity of the platform was limited to 20 ml.standard solutions. In the slurry sampling technique, sampling errors can arise from variations in the number of particles, and the mass of Procedure for Solution AAS and ICP-AES the individual particles, uncertainties in the volume pipetted For the analysis of the boron nitride powder by solution AAS and an inhomogeneous distribution of the analyte element and ICP-AES, 1.5 g of sample were digested in 150 ml PTFE throughout the solid sample.The characteristics required for liners (Berghof System III, Berghof, Eningen, Germany) with minimization of the sampling errors include small particle size, a mixture of 10 ml of sub-boiled nitric acid and 10 ml of 40% narrow particle size distribution and high slurry concentration. hydroØuoric acid for 4 h at 200°C. The resulting digest con- However, for the boron nitride powder under investigation, tained a crystalline fraction of ammonium boroØuoride.After the applicable slurry concentration was limited by the stability decanting the liquid phase into a 50 ml calibrated Øask, the of the slurry, analyte concentration in the sample and reduced crystalline solid was dissolved in water by standing in the lifetime of the graphite tube. Using the ultrasonic pre- ultrasonic bath for 1 h, the solution was transferred into the treatment, the large particle agglomerates disintegrated, how- calibrated Øask and then diluted with doubly distilled water ever, permanent stirring was necessary to avoid coagulation.to 50 ml. This solution was used for the determinations by The linear working range of the slurry concentrations ranges FAAS, ETAAS and ICP-AES. from about 1 to 15 g l-1. With higher concentrations, no The wavelength, slit and Øame conditions, respectively, for sufficiently stable slurries could be obtained by magnetic the determination of Ca, Mg and Na by FAAS were as follows: stirring whereas lower slurry concentrations led to low Ca, 422.7 nm, 0.7 nm, C2H2±N2O; Mg, 285.2 nm, 0.7 nm, reproducibility.C2H2±air; Na, 589.0 nm, 1.4 nm, C2H2±air. For quantiÆcation, The sampling efficiency was determined by weighing the the standard additions method was used. graphite tube before and after 20-fold pipetting of 20 ml aliquots For the determination of Al, Cd, Co, Cu, Fe, K, Mn, Ni, Pb of a 15 g l-1 slurry followed by drying. From three replicates and Zn in boron nitride digests by ETAAS, 20 ml aliquots of of this experiment, a mean slurry sampling efficiency of 98±4% the sample solution were dispensed automatically into the was estimated.atomizer. The standard additions method was used for calibration. The most important experimental conditions are Matrix Behaviour and Chemical ModiÆcation summarized in Table 2. The operating parameters used in the analysis of the boron Boron nitride reacts with carbon to form boron carbide at nitride digest by ICP-AES are summarized in Table 3.Standard temperatures around 2000°C.16 The formation of boron car- solutions used for calibration contained 10 ml of sub-boiled bide is expected to take place via a solid-phase reaction on the nitric acid and 10 ml of 40% hydroØuoric acid in a volume surface of the platform. This has two advantageous effects on of 50 ml. the atomization performance. Firstly, it favours the release of the analyte elements from the matrix. Furthermore, the conversion of boron nitride into the refractory boron carbide hinders RESULTS AND DISCUSSION the occurrence of non-speciÆc spectral interferences by the Slurry Preparation and Sampling matrix during the atomization stage.However, the boron carbide remains as a residue that cannot be removed from the Owing to the low wettability of boron nitride, pure water could not be used as a dispensing medium for preparation of platform during the clean-out step. Thus, the composition and 138 Journal of Analytical Atomic Spectrometry, February 1997, Vol. 12Table 3 Operating parameters for ICP-AES Spectrometer– Outer plasma gas (Ar) 14 l min-1 Intermediate plasma gas (Ar) 0.2 l min-1 Aerosol carrier gas (Ar) 0.3 l min-1 Rf power 900 W Nebulizer V-groove, HF resistant spray chamber Solution uptake 1.5 ml min-1 Emission lines used– Element Wavelength/nm Al 396.152 Ca 393.366 Cd 226.502 Co 228.616 Cu 324.754 Fe 259.940 K 766.490 and 769.896 Mg 279.553 Mn 257.610 Na 588.995 and 589.592 Ni 231.604 Pb 220.353 Zn 213.856 the structure of the atomizer surface is continuously changing.As a result of the retention of boron carbide, the availability of free carbon for the reaction with boron nitride will decrease with an increasing number of analysis cycles. This inØuences the conditions for the atomization, especially of elements trapped in the bulk, and, consequently, the resulting absorption signals. Therefore, in order to improve the performance characteristics of the method, use was made of the possibility of chemical modiÆcation.As seen from scanning electron micrographs or even observed visually, addition of magnesium nitrate to the sample slurry led to a considerable reduction in the amount of boron carbide retained on the graphite tube surface. A possible reason for this phenomenon could be seen in reactions of magnesium oxide with boron nitride by which the more volatile boron oxide is formed. However, further investigations are necessary to explain the mechanism leading to the effect of the magnesium nitrate modiÆer.From a comparison of the integrated absorbances for manganese obtained in a series of analysis cycles with a boron nitride slurry without and with the addition of 50 mg of magnesium nitrate, shown as an example in Fig. 1, it is evident that magnesium nitrate as a chemical modiÆer improves both the sensitivity and the long-term reproducibility. Fig. 1 Dependence of the integrated absorbance of manganese on the number of analysis cycles with a boron nitride slurry (200 mg of boron nitride per atomization): without modiÆer (broken line); with Table 2 Instrumental parameters and experimental conditions used in analysis of boron nitride digests by ETAAS Pyrolysis Atomization Gas Øow/ Chemical Tube Element Wavelength/nm Slit/nm Temperature/°C Ramp/s Hold/s Temperature/°C Ramp/s Hold/s ml min-1 modiÆer* type� Al 308.2 0.7 1700 15 15 2700 0 3 300 A I Cd 228.8 0.7 400 20 15 2500 1 4 0 None II Co 242.5 0.2 400 25 25 2500 0 4 0 None II Cu 324.8 0.7 400 10 15 2300 0 4 50 None II Fe 346.6 0.7 400 10 20 2500 0 4 50 None II K 769.9 1.4 900 20 10 1800 0 5 300 None II Mn 279.5 0.2 1400 25 5 2200 0 3 100 A II Ni 232.0 0.2 400 10 20 2500 0 4 50 None II Pb 283.3 0.7 850 15 10 2200 0 3 0 B I Zn 213.9 0.7 700 15 10 1900 0 3 300 C I * Chemical modiÆers (per atomization): A, 50 mg of Mg(NO3)2 ; B, 200 mg of PO43-+10 mg Mg(NO3)2 ; and C, 25 mg of Mg(NO3)2.� Tube type: I, pyrolytic graphite coated ringed tube with fork-shaped platform; and II, pyrolytic graphite coated tube with L'vov platform. 50 mg of Mg(NO3)2 per atomization (solid line). For ETAAS conditions see Table 1. Journal of Analytical Atomic Spectrometry, February 1997, Vol. 12 139In addition, it also increases the tube lifetime. With one Calibration graphite tube, up to about 200 atomizations with 150 mg of In an accurate calibration, the analyte element present in the sample could be performed.However, as is evident from the sample and in the standard has to behave similarly during the Ægure, the signal became stable only after the execution of the charring and atomization steps. This assumption is best met 10±15 atomization cycles, which were necessary to condition by using powdered standards with a matrix composition close the inside surface of the atomizer. The lifetime of the tubes to that of the sample.However, such a calibration would ended either by a mechanical breakdown or by a more or less require a relatively high Ænancial expenditure, and nandard complete coverage of the surface with boron carbide causing of this type is actually available for boron nitride anyway. poorer reproducibility of the results. These observations are in With chemical modiÆcation (see the above section), satisfac- good agreement with those made in the determination of tory similarity of the signal proÆles originating for slurries, silicon in boron nitride using a Perkin-Elmer Model 4100 ZL spiked slurries and aqueous standard solutions was observed atomic absorption spectrometer.14 Magnesium nitrate was used indicating that the use of matrix-matched solid standards for as the chemical modiÆer in the determination of all analytes quantiÆcation was not necessary.Therefore, the applicability excluding, obviously, magnesium. of the calibration curve method, which, because of its simplicity When determining Si by ETAAS, for most of the matrices and savings in time, was considered as the Ærst choice, was reported in the literature, calcium nitrate has proved to be the compared with that of the method of standard additions.This chemical modiÆer of choice.17±22 Magnesium nitrate, known was carried out by comparison of the characteristic masses as a modiÆer of universal applicability and great efficiency,23 obtained for spiked boron nitride slurries and for matrix-free has, in general, not been found to be as effective as calcium standard solutions for Æve selected elements.While for Mg, nitrate.20,22 However, when the Perkin-Elmer Model 4100 ZL Mn and Zn, the characteristic masses obtained by both atomic absorption spectrometer with a THGA graphite furnace methods were in good agreement, for Al and Ca, the character- was used, magnesium nitrate was a more effective modiÆer for istic masses obtained from slurries were lower by 33 and 25%, the determination of Si in boron nitride by slurry sampling respectively, compared with those from standard solutions.ETAAS, leading to a signiÆcantly better reproducibility com- Thus, the use of the standard additions method was uniformly pared with calcium nitrate.14 On the other hand, using the preferred for all analyte elements. same spectrometer system, calcium nitrate proved to be the most suitable chemical modiÆer for the determination of Si in Analysis of Boron Nitride and Comparison of Results titanium dioxide and zirconium dioxide.15 For titanium dioxide, magnesium nitrate even led to a reduction in sensitivity.The developed slurry sampling ETAAS method was applied Therefore, the efficiency of both calcium nitrate and magnesium to the analysis of a boron nitride material which was also nitrate as chemical modiÆers was investigated. Surprisingly, analysed by solution ETAAS and ICP-AES. The results are using the HGA-500 graphite furnace for the determination of summarized in Table 4.The contents are given as mean values Si in boron nitride, no signiÆcant difference between these two with standard deviations for each method based on Æve modiÆers was found with respect to the sensitivity, the repro- separate analyses. ducibility and the signal shape: for 50 replicate measurements For the elements Cu, Fe, K, Mg, Mn, Na and Zn, the of a boron nitride slurry with a matrix concentration of standard deviations of slurry sampling and solution AAS are 3.3 g l-1, corresponding to about 6 ng of Si per atomization, at about the same level.The considerably higher standard deviations obtained for Al and Ca with slurry sampling ETAAS mean absorbance signals of 0.028±0.004 s (RSD 15.3%) and in comparison with solution AAS and ICP-AES indicate 0.034±0.006 s (RSD 16.1%) were obtained on applying 50 mg possibly inhomogeneous distribution of these two elements in of Mg(NO3)2 and 50 mg of Ca(NO3)2, respectively, per atomizthe material.ation as chemical modiÆers. However, with magnesium nitrate, NAA is usually used as a quasi reference method for checking the tube lifetime was slightly longer than that with calcium the accuracy of a new method in this laboratory. However, nitrate. Thus, the choice of the most effective modiÆer for Si because of the extremely strong absorption of thermal neutrons is determined by both the nature of the matrix and the type by boron, NAA could not be applied to analysis of this of graphite furnace.material. Therefore, solution ETAAS and ICP-AES were used Acceptable determinations of Cr in boron nitride were hindered by serious memory effects. Formation of chromium diboride, which melts without decomposition at 2280°C,24 is Table 4 Trace elements content of boron nitride determined by slurry the most probable reason for this effect. In the determination sampling ETAAS, solution ETAAS and ICP-AES of Ti, the formation of titanium diboride causes an even more Content/mg g-1 serious difficulty: at an atomization temperature of as high as 2700°C, no proper absorption signal could be obtained.These Slurry Solution non-spectral interferences could not be reduced sufficiently by Element ETAAS ETAAS ICP-AES any of the chemical modiÆers tested. Consequently, Cr and Ti Al 23±3 21.5±0.9 23±1 were not considered as appropriate analyte elements for the Ca 169±25 170±4* 155±6 proposed method.Cd <0.002 <0.001 <0.7 For the estimation of the levels of the background attenu- Co <0.1 <0.06 <0.4 Cu 0.78±0.05 0.76±0.1 0.73±0.08 ation occurring in processing boron nitride slurries with mag- Fe 58.5±4 59±4 60±2 nesium nitrate as the chemical modiÆer, measurements were K 1.5±0.3 1.3±0.3 <3.6 performed over the wavelength range from 206 to 588 nm Mg 8.5±0.7 9.1±0.5* 6.3±0.2 using the two-line method at the resonance wavelengths of the Mn 2.4±0.2 2.5±0.1 1.95±0.05 elements Bi, Cu, Co, Cd, Ni and Mo.Atomization temperatures Na 34±3 32±1* 24±1 Ni <0.6 0.40±0.03 <0.6 were between 2000 and 2500°C and the gas stop mode in the Pb <0.09 <0.01 <5 atomization step was used to obtain the maximum amount of Si 91±19 ND� ND� reaction products in the gaseous phase. Even by applying the Zn 0.42±0.05 0.38±0.03 0.40±0.08 maximum possible amount of boron nitride of 300 mg, no signiÆcant background absorption was observed at any of the * Determined by FAAS.� ND=not determined. wavelengths investigated. 140 Journal of Analytical Atomic Spectrometry, February 1997, Vol. 12as independent methods. The deviations between the mean The LODs of Co, Cu and Pb obtained by slurry sampling are slightly higher compared with those of the solution tech- values for slurry sampling and solution ETAAS are between 0 and 10%, with the exception of K where the deviation is 16%. nique, whereas the LODs for Cd and Mn are at about the same level.Only for Ni did the slurry sampling technique, However, even in the latter case, the deviation of the means is within the standard deviations. Also, the results for ICP-AES owing to the occurrence of memory effects, provide an essentially higher LOD than the solution technique. The high LOD are in good accordance with those of the two ETAAS methods, excluding Mg, Mn and Na, for which the contents determined for Al for solution ETAAS can be explained by a signal depression caused by hydroØuoric acid which was also by ICP-AES are lower by 26, 17 and 29%, respectively, than those of slurry sampling ETAAS, and the agreement is outside observed in the analysis of zirconium dioxide11 and quartz.25 The reason for the fairly low LODs achieved by ICP-AES is the overlap of the standard deviations of the respective means.Altogether, the comparison of the results of slurry sampling because boron represents an exceptionally favourable matrix element for AES, as it gives rise to a relatively low background ETAAS with those of the two independent methods shows good agreement, proving that results of satisfactory accuracy and line-poor spectrum.Indeed, addition of appropriate amounts of high-purity boric acidank and standard can be obtained by the proposed technique. solutions did not signiÆcantly affect the signal. Limits of Detection CONCLUSIONS The LODs achievable in the analysis of boron nitride samples by slurry sampling ETAAS are shown in Table 5, along with Owing to its simplicity, rapidity, reliability and high detection power, slurry sampling ETAAS proved to be a method well those for solution ETAAS and ICP-AES.As the concentrations of Al, Ca, Fe, K, Mg and Na in the analysed sample were suited to the determination of a large number of impurities in powdered boron nitride at concentrations down to the ultra- relatively high, for the estimation of the LODs of these elements, a high-purity boron nitride sample prepared by trace level.grinding (see Sample and Reagents section) was used. For all three methods, the LODs are expressed as three standard REFERENCES deviations of the blank resulting from ten replicates, for slurry and solution ETAAS, and from seven replicates for ICP-AES. 1 Fister, D., Ceram. Eng. Sci. Proc., 1985, 6, 1305. 2 Knoch, H., and Hunold, K., T ech. Mitt. Essen, 1987, 80, 31. Also given in the table are LODs for FAAS, which was used 3 Koval'chuk, I.A., Markov, A. V., and Mil'vidskii, M. G., Izv. instead of solution ETAAS for the determination of Ca, Mg Akad. Nauk SSSR, Neorg. Mater., 1988, 24, 324. and Na, because of the high contents of these elements in 4 Belaya, K. P., and Kustova, L. V., Zavod. L ab., 1986, 52, 40. the sample. 5 Sucha, A., and Macak, J., Pokroky Praskove Metal., 1986, 2, 14. From a comparison of the LODs of slurry sampling ETAAS 6 Morikawa, H., Uwamino, Y., Iida, Y., Tsuge, A., and Ishizuka, T., and solution ETAAS, the superiority of the slurry sampling Bunseki Kagaku, 1988, 37, 218. 7 Okano, T., Fujimoto, K., Matsumura, Y., and Harimaya, S., technique for the ubiquitous analytes Al, Fe, K, Mn and Zn is Kawasaki Seitetsu Giho, 1989, 21, 113. evident. This is mainly due to the availability of high-purity 8 Martynova, L. M., Zh. Anal. Khim., 1994, 49, 444. water and ethanol as the suspension medium, and to the 9 Docekal, B., and Krivan, V., J.Anal. At. Spectrom., 1992, 7, 521. possibility of controlling the actual blank of the medium for 10 Friese, K.-C., and Krivan, V., Anal. Chem., 1995, 67, 354. each beaker prior to the preparation of the slurry, leading to 11 Schneider, G., and Krivan, V., Spectrochim. Acta, Part B, 1995, a signiÆcant reduction in the blank values. For Ca, Mg and 50, 1557. 12 Scha»ffer, U., and Krivan, V., Spectrochim. Acta, Part B, 1996, Na, a similar improvement in the LODs compared with 51, 1211. solution ETAAS can be assumed. 13 Docekal, B., and Krivan, V., J. Anal. At. Spectrom., 1993, 8, 637. 14 Hauptkorn, S., and Krivan, V., Spectrochim. Acta, Part B, 1994, 49, 221. Table 5 LODs for trace elements in boron nitride achievable by 15 Hauptkorn, S., Schneider, G., and Krivan, V., J. Anal. At. slurry sampling ETAAS, solution ETAAS, Øame AAS and ICP-AES Spectrom., 1994, 9, 463. 16 Gmelin, Handbook of Inorganic and Organometallic Chemistry, LOD/mg g-1 Boron, Syst. No. 13, Boron Compounds, Springer Verlag, Berlin, 8th edn., 1991, 4th supplement vol. 3a, p. 53. Slurry Solution Flame 17 Thomson, K. C., Godden, R. G., and Thomerson, D. R., Anal. Element ETAAS ETAAS AAS ICP-AES Chim. Acta, 1975, 74, 289. Al 0.08 2.3 – 0.4 18 Rawa, J. A., and Henn, E. L., Anal. Chem., 1979, 51, 452. Ca 0.08 – 2.8* 1 19 Lythgoe, D. J., Analyst, 1981, 106, 743. 20� 20 Taddia, M., J. Anal. At. Spectrom., 1986, 1, 437. Cd 0.002 0.001 – 0.7 21 Gitelman, H. J., and Alderman, F. R., J. Anal. At. Spectrom., Co 0.15 0.05 12� 0.8 1990, 5, 687. Cu 0.04 0.01 8� 0.3 22 Holden, A. J., Littlejohn, D., and Fell, G. S., Anal. Proc., 1992, Fe 0.06 0.6 15� 0.4 29, 260. K 0.01 0.35 – – 23 Tsalev, D. L., Slaveykova, V. I., and Mandjukov, P. B., Mg 0.005 – 0.15* 0.7 Spectrochim. Acta Rev., 1990, 13, 225. 0.8� 24 Gmelin, Handbook of Inorganic and Organometallic Chemistry, Mn 0.02 0.05 – 0.06 Chromium, Syst. No. 52, Verlag Chemie, Weinheim, 1962, part B Na 0.01 – 0.7* 0.3 vol. 8, p. 343. Ni 0.6 0.03 – 1.7 25 Hauptkorn, S., and Krivan, V., Spectrochim. Acta, Part B, 1996, Pb 0.04 0.01 – 5 51, 1197. Si 0.4 – – – Zn 0.04 0.3 2 0.4 Paper 6/04884F Received July 11, 1996 * The present work. � From ref. 8. Accepted September 3, 1996 Journal of Analytical Atomic Spectrometry, February 1997,