JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY NOVEMBER 1994 VOL. 9 1285 Determination of Trace Impurities in High-purity Graphite by Electrothermal Atomic Absorption Spectrometry and Inductively Coupled Plasma Atomic Emission Spectrometry* 2. Hladky and M. Figera Department of Analytical Chemistry Faculty of Chemical Technology Slovak Technical University Radlinskeho 9 812 37 Bratislava Slovakia The process for the determination of B and Si in high-purity graphite used for spectral analysis elaborated here is based on matrix combustion without any loss of analyte in an oxygen atmosphere with the addition of alkali. Sub-boiling distilled nitric acid and water were used to dissolve ashes. Silicon B Cd and Cu were determined by inductively coupled plasma atomic emission spectrometry and the other elements (Cr Co Mo V Ni and Ti) were determined by electrothermal atomic absorption spectrometry.Carbon is the most commonly used material in the production of electrodes for emission spectral analysis as well as being the material of electrothermal atomizers used in atomic absorption analysis. The level of impurity should not be greater than 10-5-10-6%. The purifying processes used in the production of carbon electrodes for spec- troscopy are adequate for obtaining these values for most elements with the exception of B Si and those elements whose carbides are extremely non-reactive. Precision given as the relative standard deviation for both Si and B at the ppm level was within 1O0/o. The accuracy of the whole procedure was controlled using spiked samples.Keywords Silicon and boron determination ; high-purity graphite; inductively coupled plasma; electrothermal atomic absorption spectrometry The one component common to electrothermal atomizers in atomic absorption spectrometry (ETAAS) and electrodes used in atomic emission spectrometry (AES) are graphite tubes bars or differently formed electrodes. The analytical performance of a measuring system is very dependent on the shape and properties of the material from which it is constructed. Some of the most important properties required mainly by the tube material of an electrothermal atomizer have been described Graphite most closely satisfies these requirements and there- fore is the most widely used material. However even electro- graphite carbon does have some limitations the most important of which are its porosity and that some elements can form highly refractive carbides or nitrides on its ~urface.~ Therefore the use of a suitable method with low limits of determination of impurities is necessary and is described in this paper.Experimental Chemicals and Solvents Standard solutions of B Si Cd Cu V Ni Cr Co Mo and Ti with concentrations of 1 mg ml-' were used for the preparation of the calibration solutions and the standard additions for spiked samples. All solutions were prepared using beakers made of poly- (tetrafluoroethylene) (PTFE) and micropipettes with tips made of polyethylene. Solutions were kept in these beakers that had been conditioned for a few days with purified de-ionized water and with the blank solutions. High-purity water was produced from de-ionized water with the Barnstead Nano-Pur System ( Wilhelm Werner Germany) and secondarily purified by distillation with apparatus made completely of PTFE (Berghof Germany).All measurements were carried out in a clean laboratory. Processing of the Samples The determination of B Si and the other elements was performed on samples prepared by the following methods.6 *Presented at the XXVIII Colloquium Spectroscopicum Internationale (CSI) York UK June 29-July 7 1993. Method A Graphite (5-10 g) was combusted in an oxygen atmosphere in the apparatus shown in Fig. 1 for 2 h. Sub-boiling distilled nitric acid and water were used to dissolve the residual ashes and to bring the volume up to 10m1 respectively. Method B Matrix combustion was carried out as in Method A in an oxygen atmosphere without any loss of analytes with the addition of alkali occurring in the solid phase (graphite sintered by 0.1 g of solid NaHCO,).Method C This is similar to Method B except that the addition of alkali occurred in the liquid phase (graphite saturated by the solution Quaflz tube Oven maximum 600°C Sintered filter Pt boat cylinder (flow 2- Fig. 1 Apparatus used for combustion in oxygen atmosphere Table 1 Measurement conditions for ICP-AES Optics Grating Resolution Wavelength range Power/W Nebulizer Observation height/mm Argon consumption/l min- ' Uptake rate/ml min-' Computer Czerny-Turner arrangement (f= 1 m) Holographic 2400 grooves mm-' 13 pm (first order) 165-800 nm 1200 glass Meinhard 15 14 (on the whole) 1.6 (without peristaltic pump) PDP-l1/03 (built into the spectrometer) language ARLEB1286 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY NOVEMBER 1994 VOL.9 Table 2 Measurement conditions for ETAAS ~~~ ~ ~ Element Wavelength/nm Bandwidth/nm Lamp current/mA TD/t,*/”C S - l TA/tAt/°C S - ’ Cr 357.9 0.5 8.0 1200/25$ 260015 c o 240.7 0.2 12.0 1250/25$ 250015 Mo 313.3 0.2 12.0 1400/20 2800/3 v 318.5 0.2 12.0 1450120 280013 Ni 232.0 0.2 9.0 11 50/25$ 260015 Ti 365.4 0.2 13.0 1500/20 280013 * TD and t temperature and time of decomposition respectively. t TA and t A temperature and time of atomization respectively. 10.1% solution of Mg(N0,)2 used as chemical modifier. of 0.1 g of NaHCO in 0.5 ml H20 and dried under IR radiation). Samples prepared using each of the methods (A B and C) were spiked by standard additions of B Si and Cd before com bus tion.The efficiency of this procedure was tested on synthetic samples composed from boron nitride and carbide and silicon carbide as the most stable compounds of these elements. Instrumentation The analytical measurements were carried out by inductively coupled plasma atomic emission spectrometry (ICP-AES) using a sequential spectrometer ICP 3510 (ARL USA) under the conditions summarized in the Table 1 or by ETAAS using a Pye Unicam (UK) SP9 atomic absorption spectrometer fitted with a PU 9090 Data Graphic System and PU 9095 Video Furnace Programmer using the standard operating conditions summarized in Table 2. Results and Discussion Because the level of impurities in graphite should not be higher than to (level required for production of high- purity graphite suitable for use in the production of graphite tubes for ETAAS) we could make the choice of the spectral line with satisfactory low detection limits.7 The program SCAN provided by the ICP-manufacturer can be used to scan around a chosen spectral line.The lowest quantity determinable (LQD equal to the limit of determination in solution) and background equivalent con- centration (BEC) values for suitable lines were determined by the standard programme limit of detection (provided by the manufacturer) using a solution with a content of 10 pg m1-l. Results are summarized in Table 3. As can be seen from Table 3 the best lines for B Si and Cd determination seem to be 249.678 251.611 and 228.802 nm Table 3 Analytical characteristics for some suitable lines Element B B B Si Si Si Cd Cd Cd c u c u linelnm 182.640 208.959 249.678 212.452 251.61 1 288.158 214.438 226.502 228.802 324.757 224.700 LQD */pg ml - ’ 0.1000 0.0200 0.0125 0.1000 0.0500 0.1000 0.01 60 0.01 50 0.0200 0.0056 0.0200 BEC/pg ml - ’ 2.000 0.400 0.250 2.000 1 .ooo 2.000 0.348 0.300 0.121 0.150 0.400 ~~ * The lowest quantity determinable (k the concentration which yields a signal differing from that of the blank by 100 where cr is the standard deviation of the blank intensity) values were obtained with synthctic solutions of elements and represents only ‘theoretical’ values and will differ from that obtained using real sample solution with a known content of elements.4 ~ 1 0 ~ I 1 o4 104 1 o4 0 2 49.631 249.669 249.706 Wavelengthtnm Fig. 2 Spectra obtained around B line 249.678 nm for 0 sample; 0 ultrapure water; 0 blank; and A standard 10 ppm of B. Integration time 1 s and scan range k0.039 nm respectively which were also free of possible spectral coinci- dence in the sample analysis. Figs. 2 3 and 4 show the results of the spectral scans around these lines for the samples ultrapure water (sub-boiling distilled water from PTFE appar- atus) blank (composed of all reagents used in the determi- nation) and 10 ppm standards. The scan for Cu at 324.754 nm is not shown because it was similar to that for Cd i.e. free of spectral coincidence and the signal was lower than the limit of detection. Calibration graphs were constructed for the concentration range 0-10 mg 1-l (B Si Cd and Cu) for the ICP-AES method.1 o4 104 1 o4 1 o4 0 251.564 251.602 251.689 Wavelengthh m Fig.3 Spectra obtained around S line 251.611 nm for 0 sample; 0 ultrapure water; 0 blank; and A standard 10 ppm of Si. Integration time 1s and scan range k0.039 nmJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY NOVEMBER 1994 VOL. 9 1287 7.5 x lo4 6 x lo4 - v) C 3 Y .- 2 4.5 x lo4 F 4? .e 3 1 0 4 c ._ - > C al c - 1.5 x lo4 0 228.775 228.793 228.830 Wave le ngt hlnm Fig. 4 Spectra obtained around Cd line 228.802 nm for 0 sample; 0 ultrapure water; 0 blank; and A standard 10 ppm of Cd Table 4 Results for the determination of some elements in high-purity graphite calculated on the mass of the solid material; '<' indicates that the measured value was lower than limit of determination calculated on the mass of sample Element B Si Cd c u Cr c o Ni Mo V Ti ICP-AES/ng g-' 332.3 -t 7.1 25444.3 & 127.2 < 20.0 0.7 < 6.0 & 0.4 ETAAS/ng g- - 4.35 & 0.54 12.69k 1.08 45.51 k4.41 63.59k4.51 374.05 k 55.73 823.00 k 27.98 Table5 Recovery of standard additions of B Si and Cd from the spiked samples by ICP-AES (samples were spiked with standard solutions of B Si and Cd whose volumes and concentrations were recalculated on the mass of samples so that concentration added was 10.00 pg ml-' in the final solution) Similarly the calibration graphs for determinations by ETAAS were prepared for the concentration ranges of O-lOpgl-' (Cr Co and Ni) O-lOOpgl-l (Mo and V) and 0-1000 pg I-' (Ti). The signals were measured as the peak heights.Table 4 gives the results of analyses of samples for a Slovakian producer of graphite used for spectral analysis. The samples were prepared by the different methods A B and C and results obtained by ICP-AES from these samples and from the spiked samples were compared with one another (Table 5). In Table 4 are summarized the results obtained using method C. This method of preparation of samples was used because it gave the best recoveries 97% on average. Conclusion The proposed method for the determination of trace impurities in high-purity graphite for spectral use has been applied for the determination of B Si Cd Cu Cr Co Mo V Ni and Ti in spectral electrographite. The accuracy and efficiency of this method for the treatment of the samples were tested on spiked samples with recoveries from 74 to 97% depending on the preparation methods used and with the use of boron nitride and carbide and silicon carbide as the most inert compounds of B and S .Precision as the relative standard deviations (RSD) for Si and B were within & 10% at the ppm level and & 15% at the ppb level for the other elements calculated from three indepen- dent determinations from different portions of the samples. Methods B and C have been shown to be suitable for the determination of impurities in high-purity graphite. References Products for Analysis Technology Ringsdorff Werke GmbH Bonn FRG E 555/89e. Lersmacher B. and Knippenberg W. F. Philips Tech. Rev. 1977 37 189. L'vov B. V. Spectrochim. Acta 1961 17 761. Dymott T. C. Wassall M. P. and Whiteside P. J. Analyst 1985 110 467. Luguera M. Madrid Y. and Camara C. J. Anal. At. Spectrom. 1991 6 669. De Keyser W. L. and Cypres R. Bull. Centre Phys. Nucleare Univ. Bruxelles 1952 35 28. Winge R. K. Fassel V. A. Peterson V. J. and Floyd M. A. Inductively Coupled Plasma Atomic Emission Spectroscopy An Atlas of Spectral Information Elsevier Amsterdam 1985. Method of Recovery of Recovery of Recovery of Average recovery preparation B/pg ml Si/pg ml - ' Cd/pg ml - ' w.) A 7.65 k0.28 6.78 k 0.61 7.81 0.36 74.1 B 8.96k0.33 7.72k0.56 8.15k0.25 82.8 C 9.48 0.25 9.13 k 0.48 10.51 0.56 97.1 Paper 31039358 Received July 7 1993 Accepted June 6 1994