ANALYST, JANUARY 1987, VOL. 112 23 Determination of Scandium in Coal Fly Ash and Geological Materials by Graphite Furnace Atomic Absorption Spectrometry and Inductively Coupled Plasma Atomic Emission Spectrometry M. Bettinelli, U. Baroni and N. Pastorelli Central Laboratory, ENEL-DCO, Via Nino Bixio 39, 29700 Piacenza, Italy Two procedures for the determination of scandium in coal fly ash and other geological materials have been developed. The first procedure consists in dissolution of the samples with a mixture of nitric, perchloric and hydrofluoric acids in a sealed PTFE vessel followed by determination by graphite furnace atomic absorption spectrometry. In the second procedure, the sample is decomposed by fusion with lithium tetraborate, the melt is dissolved in 5% nitric acid and scandium is determined by inductively coupled plasma atomic emission spectrometry using an argon plasma.The limits of determination are 0.80 and 0.63 pg g-1, respectively, for the two methods. The scandium concentration in eight coal ash samples obtained by GFAAS and ICP-AES compare well with those determined previously by NAA. Keywords: Scafidium determination; coal fly ash; geological materials; graphite furnace atomic absorption spectrometry; inductively coupled plasma atomic emission spectrometry Studies of the particle size dependence of element concentra- tions in fly ash can be classified into two categories.1 The first category consists of those studies that relate the elemental concentrations to the particle size of size-classified material (sufficient mass of size-classified material must be collected in order to allow gravimetric determination prior to elemental analysis).The second category consists of the many studies that have employed inertial cascade impactor systems for aerodynamic size classification. Because only small masses of material may be collected on the stages of the cascade impactor, the specific elemental masses of the particles deposited on each stage are often ratioed to the mass of an element that does not demonstrate a marked dependence of concentration on particle size. One of the elements that has been frequently used for this purpose is scandium. This element was selected because it is present at very low levels in the natural environment, it is essentially non-volatile at furnace temperatures , and also because most fly ash analyses are carried out by neutron activation analysis (NAA) , which gives a detection limit of about 0.002 pg of Sc.2 Gladney et aZ.,3 for example, reported the concentration of Sc in several geological environmental NBS standard materials, but referred only to NAA and flame atomic absorption spectrometric (FAAS) techniques. More recently, ICP-AES and DCP-AES techniques have been successfully employed to determine trace amounts of elements in geological and related materials, including Sc and rare earth elements.4-13 In contrast, very few papers have been published on the determination of Sc by ETA-AAS.1620 Sen Guptal6 determined Sc in silicate rocks after coprecipi- tation with calcium oxalate and hydrated iron(II1) oxide. Working in the peak-height mode (Tatom.= 2500 "C) using pyrolytically coated graphite tubes, the sensitivity for Sc was 37 pg per 0.0044 A, superior to that found (50 pg per 0.0044 A) in a tantalum-lined furnace.17 Sen Gupta later showed18 that, at 2000 "C, greater sensitivity can be achieved by using a tantalum foil-lined graphite furnace (1.2 pg per 0.0044 A) instead of a pyrolytically coated furnace (13 pg per 0.0044 A). Wu and Ma19 reported the direct determination of Sc in soils by graphite furnace AAS (GFAAS) with a pyrolytically coated graphite tube lined with both tungsten and tantalum foil. Sample aliquots (10 ~ 1 ) were atomised at 2730 "C for 10 s in a 0.5 1 min-l argon gas flow. The absolute limit of determination was 4.6 pg of Sc.Atnashev et aZ.20 reported the determination of Sc by ETA-AAS using a tungsten coil atomiser, carrying out the pulsed atomisation in a laminar flow of Ar - N2 (10 + 1). However, no characteristic amounts data for Sc were reported by Atnashev et aZ.,20 but it was reported that the results compare well with corresponding values obtained with a L'vov graphite cell. In this work we compared two different procedures for sample dissolution: acid attack with nitric - perchloric - hydrofluoric acids and lithium tetraborate fusion. Scandium was determined in the solution derived from the mixed acid digestion procedure by GFAAS, whereas in the fusion solution we used the ICP-AES technique. Experimental Reagents Standard solutions of Sc were prepared from a 1000 mg 1-1 stock standard solution for atomic absorption spectrometry (Aldrich-Chemie , Steinheim, FRG) by dilution with de-ion- ised water or lithium tetraborate.Perchloric acid (70% mlv), nitric acid (65% m/V) and hydrofluoric acid (40% mlv) were all Suprapur reagents (E. Merck, Darmstadt, FRG). Lithium tetraborate solution (5 g 1-1) was prepared from Baker Analyzed reagent flux grade material (J. T. Baker Chemicals, Denveter, The Netherlands). Water was purified in a Milli-Q system (Millipore, S.P.Q. Milan, Italy). Apparatus A Perkin-Elmer 5000 atomic absorption spectrometer equipped with a D2 arc background corrector, an AS-40 autosampler and an HGA-500 graphite furnace was used for the Sc determinations. A 7500 data station was used for the display and storage of the fast atomisation signals.Peak absorbance and integrated absorbance signals were calculated using the HGA Graphics I1 software and PR 210 printer - plotters were used for printing out the analytical information and the high-resolution peak profiles. The optimum instrumental parameters for Sc determinations (Table 1) were established after extensive investigations. Pyrolytically coated graphite tubes were used in all determinations. A Perkin-Elmer ICP/6000 inductively coupled plasma atomic emission spectrometer was used for all Sc determina-24 ANALYST, JANUARY 1987, VOL. 112 Table 1. AAS instrumental operating conditions Model 5000 spectrophotometer: Wavelength . . . . . . . . . . . . 391.2 nm Calibration mode . . . . . . . . . . Peak area Integration time .. . . . . . . . . 6 s Background corrector . . . . . . . . Deuterium arc Spectral slit width . . . . . . . . . . 0.2 nm HGA-500 graphite furnace: Step, n Temperature/"C Ramp time/s Hold time/s 1 80 1 4 2 120 10 10 3 500 10 10 4 1700 20 10 5 2700 O* 6 6 2800 1 3 Purge gas . . . . . . . . . . . . Argon, interrupted Sample volume . . . . . . . . . . 20 pl Alternative volume . . . . . . . . . . 20 pl * Maximum power heating mode; read activated at -2 s. Table 2. ICP-AES instrumental operating conditions . . . . . . . . Incident RF power 1250 W Reflected RF power . . . . . . . . <5 w Plasma gas flow-rate . . . . . . . . 15 1 min-1 Auxiliary gas flow-rate . . . . . . . . 0.3 1 min-1 Nebuliser gas pressure . . . . . . . . 26 lb in-2 Viewing height . . . . . . . . .. 14 mm above load coil tions in lithium tetraborate solutions; the operating conditions used are given in Table 2. For sample decomposition, a Perkin-Elmer Autoclave 3 was used. Neutron activation analyses were performed with the TRIGA Mark I1 reactor at the University of Pavia. Standards and samples were irradiated for 50 h at a flux of 1.2 X 1012 neutrons cm-2 s-1; the radionuclide used was 46Sc with a half-life of 83.80 d and a characteristic peak at 889 keV. Acid Dissolution Procedure About 0.25 g of sample was taken with 4.0 ml of 65% nitric acid, 2.0 ml of 70% perchloric acid and 4.0 ml of 40% hydrofluoric acid in the PTFE beaker of the Autoclave-3 and heated in a drying oven for 3-5 h at 150 "C. After cooling the contents of the PTFE beaker were slowly evaporated to dryness on a low-temperature (200 "C) alumi- nium block.After the addition of 1 ml of nitric acid and 10 ml of water, the solution was transferred into a polypropylene bottle and diluted to 100 ml with de-ionised water. If any undissolved residue was visible at this stage, the solution was filtered through a 0.5 pm Fluoropore membrane filter and the residue solubilised with 2.0 ml of 70% perchloric acid and 5.0 ml of 65% nitric acid. The mixture was heated on the plate nearly to dryness and, after washing with 5.0 ml of water, was then heated until the evolution of dense white fumes. The PTFE beaker was then cooled, 1.0 ml of nitric acid plus 10 ml of water were added to dissolve the salts and the solution was combined with that in the polypropylene bottle.This final solution was then diluted to 100 ml. Fusion Procedure Approximately 0.25 g of sample and 0.5 g of lithium tetra- borate were placed in a 50-ml platinum crucible and thor- oughly mixed. The sample - flux mixture was fused in a muffle furnace at 1000 k 50 "C for 45 min. When the fusion was complete, the cooled crucible was placed in a 100-ml beaker, a small PTFE-coated stirrer was inserted and 25 ml of 5% nitric acid were added. The solution was heated for 15-20 min at 50-60 "C on a magnetic stirrer and then transferred into a 100-ml poly- propylene flask. The same procedure was repeated with a second aliquot of nitric acid and, when dissolution was complete, the contents of the flask were adjusted to volume with de-ionised water. Blanks containing only lithium tetraborate were also fused in the manner described above.These solutions are stable over a period of several months and may be used for the determination of ten major elements, plus some trace elements. Results and Discussion Optimisation of HGA Operating Parameters In order to select the optimum ashing temperature, a study of the effect of this parameter on the absorption signal of Sc was carried out. Aliquots of 20 pl of a 50 pg 1-1 Sc solution were injected into the pyrolytically coated graphite tubes and drying and atomisation cycles of 120 "C for 10 s and 2700 "C for 6 s, respectively, were employed. The integrated absorption signal produced during the atomisation step was detected at 391.2 nm and recorded by the HGA Graphics I1 software. On varying the ashing temperature between 1000 and 2000 "C, we observed that the signal remained constant at temperatures up to 1800 "C (Fig.1). Hence, an ashing temperature of 1700 "C was selected for all further determina- tions. Interferences Sen Guptalg reported that there are no inter-element interfer- ences in the determination of lanthanides in GFAAS and that interferences from associated common elements can be eliminated by heating the sample at about 1800-2000 "C before atomisation. The final solution in work of the Sen Guptals is, however, relatively free from other elements because it is first submitted to a double calcium oxalate and a single hydrous iron(II1) oxide co-precipitation step. Our final solution, in contrast (250 mg of sample diluted to 100 ml), contains several milligrams of matrix elements such as Si, Al, Fe, Mg, Ca, Na and K. For this reason we verified the interference effects of a synthetic solution containing 100 mg 1-1 of Fe, Mg, Na, K and P, 500 mg 1-1 of A1 and 1000 mg 1-l of Ca and Si on the absorbance signal of 20 pg 1-1 of Sc.Fig. 2 shows that under the experimental conditions reported in Table 1 the interfer- ences due to the principal elements in our matrices are negligible. GFAAS Determinations According to the concept of "characteristic mass" (rn,) developed by Slavin and Carnrick,21 we calculated this quantity by injection of the analyte in aqueous solution (1% HN03) and by the method of standard additions. The instrumental parameters and the experimental con- ditions are given in Table 1 and the results for rn, are reported in Table 3.The characteristic mass (m,) averaged on ten scandium determinations in aqueous solution YO HN03) was 31.2 rt 4.0 pg per 0.0044 A s (CV = 12.7%) but with two different values, 34.6 f 1.9 pg per 0.0044 A s (n = 5, CV = 5.6%) and 27.9 k 1.2 pg per 0.0044 A s (n = 5, CV = 4.4%) for two different lots of pyrolytically coated graphite tubes. The characteristic amounts found by the standard additions method for NBS 1633, 1633a, 1645, 1646, 278 and 688 standards compare very well with those in water. The precisions for the reported characteristic mass data appear to be better than 10-20%.ANALYST, JANUARY 1987, VOL. 112 25 0.050 a 0.040 Table 3. Characteristic mass data (pg per 0.0044 A s) for Sc in aqueous solution and in different NBS materials A ( a ) - - Sample Water (1% HN03) NBS 1633 NBS 1633a NBS 1645 NBS 1646 NBS 278 NBS 688 Number of samples 5 3 3 3 3 3 3 (n) Characteristic mass (mo) Lot A* 34.6 2 1.9 32.4 k 1.7 33.3 f 1.2 33.9 2 1.5 32.5 f 2.5 34.3 f 2.0 32.8 k 1.9 * Lots A and B are two different lots of pyrolytically coated graphite tubes.Lot B* 27.9 k 1.2 25.3 k 0.8 25.3 f 1.8 24.7 k 1.7 24.8 f 2.6 29.2 f 1.1 25.1 f 2.6 Average 31.2 k 4.0 30.0 f 5.9 29.2 f 4.5 29.3 f 5.2 28.7 k 4.8 31.8 f 3.1 28.9 k 4.6 c v , Yo 12.7 19.7 15.4 17.9 16.7 9.8 16.0 0.21 v) a 0.09 1 1 I I I I I I 1 I I I 1000 1100 1200 1300 1400 1500 1600 1700 1800 1900 2000 2100 Ashing temperaturePC Fig. 1. Typical ashing curve for 1 ng o f Sc in 1% HN03 0 1 2 3 4 Time/s Fig. 2. of Sc and (b) 20 pg 1-1 of Sc plus Fe, Mg, Na, K and P (100 mg l-l), A1 (500 mg 1-l) and Ca and Si (loo0 mg 1-1) High-resolution peak profiles for ( a ) 20 pg 1 Fig.3. Graphic spectral scans of Sc I1 at 361.384 nm. Conditions: 100 pg 1-1 of Sc in A, lithium tetraborate; B, NBS 1633a solution; C, NBS 688 solution; and D, NBS 1646 solution. Background positions: B1, -0.126 nm and B2, +0.140 nm After about 50-70 firings of a new pyrolytically coated graphite tube at the recommended temperatures, the inte- grated absorbances decrease because of the destruction or loss of some of the pyrolytic coating and the tube should be discharged. The detection limit, DL (calculated as the concentration corresponding to three times the standard deviation of the blank), was 2.0 pg 1-1, whereas the lowest quantitatively determinable concentration, LQD (defined as the concentration corresponding to ten times the standard Table 4.Results for NBS reference materials (pg g-l) This work NBS Certified standard GFAAS ICP-AES values Coal fly ash: 1633 25.1 23.5 28.7k2.0 - 1633a 40.62 5.1 41.3 k 4.8 (40)* River sediment: Estuarine sediment: Obsidian rock: Basalt rock: 1645 (1.8) (1.6) (2) 1646 9.4 f 3.1 10.9 k 1.9 (10.8) 278 6.1 f 1.3 3.5 k 0.1 (5.1) 688 39.1 k 2.4 40.4 2 0.7 (38.1)* * Values reported but not certified by NBS. Other values3 26.6 2 1.7 38 2 3 2.6 10.4 4.9 f 0.5 36.2 Table 5. Comparative concentrations of Sc (pg g-l) in different coal ash samples This work Sample GFAAS ICP-AES NAA BMAl . . . . . . 2 6 f 1.7 29f0.6 27f1.4 BMA2 . . . . . . 20k1.1 20f0.4 GMAl .. . . . . 3352.0 36f0.6 3 4 f 1 . 7 GMA2 . . . . . . 3.550.08 2.8k0.03 3 k 0 . 2 BCLl . . . . . . 8k0.09 9k0.07 10f0.5 11 f 0.5 BCL2 . . . . . . 951.0 1OkO.4 BMIl . . . . . . 25k1.3 28k0.6 25k1.3 BM12 . . . . . . 2420.7 25k0.3 2921.5 19 5 1.0 deviation of the blank and a dilution factor of 400), was 2.7 pg g-1 of Sc. ICP Determinations The choice of the Sc I1 line at 361.384 nm was based on a systematic consideration of the detection limit, the expected elemental concentration in the sample solution and spectral interferences. Fig. 3 shows a scan from 0.5 nm below to 0.5 nm above the scandium 361.384 nm line, overlaid with scans of three NBS reference materials. These, and other scans not reported in the figure, indicate that, for these and similar types of samples, the Sc emission line is sufficiently free from background interferences to be utilised in a reasonably straightforward manner.Background correction must be carried out accurately because the background under the analyte line is extremely structured and could cause spectral interference problems, particularly if there were any broadening as the concentrations increased.26 ANALYST, JANUARY 1987, VOL. 112 The detection limit was 1.6 pg 1-1, whereas the lowest quantitatively determinable concentration was 2.1 pg g-1 of sc. Comparison of GFAAS and ICP Results Results for scandium determinations obtained in this work compare well with NBS reference values (reported but not certified) and literature values. The accuracy data are given only for those samples for which the Sc concentration exceeded the LQD value, as quantitative determinations can usually be made with satisfactory accuracy and precision only above this concentration level.The scandium concentration determined by ICP-AES for the NBS 1645 sample was found to be less than the estimated DL, whereas the value for NBS 278 appears to be biased low. The poor recoveries also obtained for this element using the acid dissolution - HGA technique indicate that incomplete solubilisation (dissolution or fusion) for the NBS 1645 sample was responsible for the low scandium results. It is evident from these data that accurate determinations can be carried out by both the methods; the fusion - ICP-AES method gives a better precision then the acid digestion - GFAAS procedure.Similar conclusions can be drawn from the data in Table 5 relative to other coal ash samples analysed for their Sc content either by mixed acid digestion - GFAAS and fusion - ICP-AES procedures or by neutron activation analysis. Conclusions Both the fusion - ICP-AES and acid digestion - GFAAS procedures are capable of the successful routine determina- tion of Sc in coal fly ashes, provided that the samples contain relatively high concentrations of this element (i.e., 1C20 times the LQD). The major source of error in the experimen- tal reproducibility lies (mostly for acid solubilisation) in the preparation of the sample powder solution. In general, the data obtained for fusion - ICP-AES or acid digestion - GFAAS determinations are as good as those obtained for NAA determinations.4. 5. 6. 7 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. References Fischer, G. L., and Natusch, D. F. S., in Karr, C., Editor, “Analytical Methods for Coal and Coal Products,” Volume 111, Academic Press, New York, 1979, Chapter 54, p. 489. Weaver, J. N., in Karr, C., Editor, “Analytical Methods for Coal and Coal Products,” Volume I, Academic Press, New York, 1979, Chapter 12, p. 377. Gladney, E. S., Burns, C. E., Perrin, D. R., Roelandts, I., and Gills, T. E., “1982 Compilation of Elemental Concentration Data for NBS Biological, Geological and Environmental Standard Reference Materials,” NBS Special Publication 260-88, National Bureau of Standards, Washington, DC, 1984. Brenner, I. B., Watson, A. E., Steele, T. E., Jones, E. A., and Goncalves, M., Spectrochim. Acta, Part B, 1981,36,785. Crock, J. G., and Lichte, F. E., Anal. Chem., 1982,54, 1329. Uchida, U., Uchida, T., and Iida, C., Anal. Chim. Acta, 1980, 116, 433. Floyd, M. A., Fassel, V. A., and D’Silva, A. P., Anal. Chem., 1980, 52, 2168. McLaren, J. W., Berman, S. S., Boyko, V. J., and Russel, D. S., Anal. Chem., 1981,53, 1802. Fries, T., Lamothe, P. J., and Pesek, J. J., Anal. Chim. Acta, 1984, 159,329. Nadkarni, R. A., Anal. Chem., 1980,52,929. Cantillo, A. Y., Sinex, S. A., and Helz, G. R., Anal. Chem., 1984,56, 33. Bolton, A., Hwang, J., and Vander Voet, A., Spectrochim. Acta, Part B, 1983,38, 165. Barnes, R. M., and Mahanti, H. S., Spectrochim. Acta, Part B, 1983,38, 193. Sen Gupta, J. G., Talanta, 1981, 28, 31. Sen Gupta, J. G., Geostand. Newsl., 1982, 6, 241. Sen Gupta, J. G., Anal. Chim. Acta, 1982, 138, 295. L’vov, B. V., and Pelieva, L. A., Can. J. Spectrosc., 1978,23, 1. Sen Gupta, J. G., Talanta, 1985, 32, 1. Wu, Z,, and Ma, Y., Fenxi Huaxue, 1983, 11, 423; Anal. Abstr., 1984, 46, 6G10. Atnashev, V. B., Muzgin, V. N., and Atnashev, Yu. B., Zh. Anal. Khim., 1982, 37, 1590; Anal. Abstr., 1983,44,6B16. Slavin, W,, and Carnrick, G. R., Spectrochim. Acta, Part B, 1984,39,271. Paper A615 Received January 6th, 1986 Accepted August 26th, 1986