ANALYST. FEBRUARY 1992. VOL. 117 131 Determination of Lithium, Beryllium, Cobalt, Nickel, Copper, Rubidium, Caesium, Lead and Bismuth in Silicate Rocks by Direct Atomization Atomic Absorption Spectrometry Toshihiro Nakamura, Hideyuki Oka, Hidehiro Morikawa and Jun Sat0 Department of Industrial Chemistry, Meiji University Higashimita, Tama-ku, Kawasaki 2 14, Japan Optimum experimental conditions for the atomic absorption spectrometric determination of Li, Be, Co, Ni, Cu, Rb, Cs, Pb and Bi by direct atomization of solid silicate rock samples were investigated. A 1.0 mg portion of powdered sample (particle size 0.3-25 pm) was mixed with the same amount of graphite powder and atomized in a miniature graphite cup placed within a graphite furnace according to the heating programme that was established.Calibration was effected using aqueous standard solutions in which 500 ng of K were added for Rb and Cs as an ionization buffer. Results for nine geochemical standard reference rock samples showed good agreement with the recommended values; the correlation coefficient for the found and recommended values was 0.9993. The relative standard deviation ( n = 5) was less than 10%. Keywords: Silicate rock; minor element; direct atomization; electrothermal atomic absorption spectrometry; suppression of ionization interference The concentrations of minor and trace elements in rocks are conventionally determined by d.c. arc emission spectrometry, atomic absorption spectrometry (AAS) or inductively coupled plasma atomic emission spectrometry (ICP-AES) .Usually AAS and ICP-AES require complicated and tedious methods for the dissolution of rock samples. Also, these processes are sometimes accompanied by accidental contamination. Hence, direct atomization of these elements has an advantage over the conventional method and the electrothermal atomization of solid samples has been investigated owing to its high sensitiv- ity, low determination limits and short analytical time requirement. Since the 1940s, d.c. arc emission spectrometry has domi- nated the determination of trace components in rock sam- ples.' This technique is, however, still of low sensitivity for alkali and alkaline earth elements.' Although several attempts have been made to atomize solid rock samples in flame AAS,3-5 applications are limited owing to insufficient sensitivity.This is not only because the temperature of the flame is not sufficiently high but also because maintenance of the flame at a constant temperature is difficult. The combination of ablation with a laser beam and flame atomization was tried for the determination of copper in rocks,h.7 but the sensitivity was not sufficiently high and the reproducibility was not acceptable. Katskov and L'VOV~ first employed a furnace for the direct atomization of Pb. Leucke et al.9 accurately determined Co, Cu and Pb using a heating programme for the atomization based on the boiling-points of the analytes. Calibration was effected by using solid materials whose elemental compo- sitions were similar to those of the samples. Fuller and Thompson10 and Isozaki et al.11 successfully determined Cu and Zn in rock powders by injecting the sample powders as a slurry into a furnace. This technique involved simplified analytical processes and the reproducibility was improved. Langmyhr et aZ.12 and Siemer and Wei13 determined Pb by mixing rock powders with graphite powder to enhance the heat conductivity between the graphite furnace and the sample powder. The choice of calibrating materials is a serious concern when a solid sample is atomized directly. Few investigations have been made on the use of aqueous solutions for calibration except those by Siemer and Wei*3 and Isozaki et al.11 This investigation was concerned with further applications of direct solid atomization AAS to determine Li, Be, Co, Ni, Cu, Rb, Cs, Pb and Bi in silicates by using a graphite furnace and graphite cups.The following aspects are discussed: (1) particle size of sample powder, (2) mixing ratio of graphite powder, (3) temperatures for drying, pyrolysis and atomiza- tion, (4) amount of sample powder and ( 5 ) suppression of ionization interference. The effective use of aqueous standard solutions for obtaining calibration graphs is also discussed. Experimental Apparatus A Hitachi 2-8000 Zeeman atomic absorption spectrometer was used in conjunction with monatomic hollow cathode lamps and a data-processing module which compensates for background absorbance. The operating conditions are given in Table 1. Pure argon was used as the inert gas. The atomization was conducted with a graphite crucible furnace of a cup type with a custom-made miniature cup (3.0 mm external diameter x 3.8 mm external height, 2.5 mm internal diameter x 3.0 mm internal height) after Atsuya and Itoh14 from a spectroscopic graphite rod (Nippon Carbon).The Table 1 Operating conditions for the determination of Li. Be, Co. Ni, Cu, Rb, Cs. Pb and Bi in silicate rock samples Parameter Li Be co Ni c u Rb cs Pb Bi Analytical line/nm 670.8 234.9 240.7 232.0 324.8 794.8 852.1 283.3 306.8 SI i t-wid t h/nm 0.4 1.3 0.2 0.2 1.3 0.4 1.3 1.3 0.2 Lamp current/mA 10.0 10.0 10.0 10.0 7.5 10.0 6.0 7.5 6.0 Absorbance Ar sheath gas flow rate/ml min-I Ar carrier gas flow rate/ml min-1 Peak area 30 200132 0.15 0.10 ANALYST, FEBRUARY 1992, VOL. 117 (a) - - miniature cups were heated to 2600 "C before use to remove any contamination.The powder mill was an Ishikawa Model AGA grinder with an agate mortar and pestle. The powder mixer was an Iwaki Model MA-1 mill with a polyethylene cylinder (24 mm X 12 mm i.d.) and a single polyethylene ball (9 mm diameter). Powdered samples (0.5-1 .O mg) weighed on a Mettler Model M3 microbalance were introduced into the miniature cup using paper. The miniature cup was handled with titanium or brass tweezers. Liquid samples were injected into the cup within the graphite furnace with a Model 4700 Eppendorf micropipette (10 yl). Particle size distributions were measured with a Shimadzu Model SA-CP3L centrifugal particle size analyser. U I I Reagents and Samples Graphite powder of about 50 pm particle size was prepared from spectroscopic graphite (Nippon Carbon) by grinding in an agate mortar.Standard solutions for calibration were prepared from commercially available 1000 ppm standard solutions (Junsei Kagaku) by dilution before use. All reagents used were of analytical-reagent grade and water was de- ionized. The samples were nine geochemical standard reference rock samples issued by the Geological Survey of Japan: JG-1, JGb-1, JP-1, JR-1, JR-2, JA-1, JB-la, JB-2 and JB-3. Procedure Rock samples were ground in order to make the average particle radius less than 2 pm (with no particles exceeding 10 pm). A portion of the powdered sample was mixed with the same amount of graphite powder. A 1.0 mg amount of the 0'11 ~ 0.35 0 0.09 0.08 $ TI 0 $ 0.07 L 0.06 0.05 + - I , I I 10.15 0 50 100 150 Fig.1 Variation in the absorbance of A, Cu; B, Rb; and C, Pb in silicate rock samples with grinding time. A 0.50 mg amount of rock powder (JG-1) mixed with 0.50 mg of graphite powder was atomized. The rock powder (3 g) was ground with an Ishikawa grinder (Type AGA) 0.04 Time/min -0 -0 C & 7 4- m 6 H R\\ B 0 50 100 150 Ti me/m i n Fig. 2 Variation in relative standard deviation (n = 5) of the absorbance of A, Cu; B, Rb; and C, Pb in silicate rock powder (JG-1) with grinding time mixture was weighed in the tared miniature graphite cup and inserted into the graphite furnace. The mixture was dried, pyrolysed and atomized according to the heating programme described later. Absorbance was determined by integration of the spectral lines in the absorbance-time spectrum.Analyte concentrations were determined by comparison with calibra- tion graphs for standard solutions. For Rb and Cs, 20 ppm of K were added to each calibration standard solution to suppress the ionization interference. Results and Discussion Particle Size and Rock Sample Minor elements are present in rocks forming solid solutions with rock-forming minerals or at the boundary of the minerals as oxides, sulfides, etc. They are heterogeneously distributed in rocks. Therefore, sufficient grinding of samples to fine powders is required to make the sample powders homo- geneous for better reproducibility. Wilson15 reported that it is desirable for the number of particles to exceed 1 x 105 to obtain favourable analytical repeatability (i.e., relative standard deviations of less than 10%) in the analysis of solid samples.His calculation showed that 1 mg of powder contains 7 x 103 particles for a particle size of 50 pm and 8 x 105 particles for a particle size of 10 ym. At a size of 5 ym, the number of particles in 1 mg of powder is 7 x 106, and sampling errors become less important. Fig. 1 shows the variation in absorbance of Cu, Rb and Pb with grinding time for a 3.0 g rock sample (JG-1). The variation in the associated relative standard deviations for each element is shown in Fig. 2. The absorbance of Be, Co, Ni, Cu, Pb and Bi increased abruptly in the first 20 min of grinding, which was followed by a gradual increase up to 160 min. For Li, Rb and Cs, the absorbance was constant and essentially independent of grinding time.The relative stan- dard deviation decreased slightly for the first 20 min, then levelled off. Both the median and the mode of the particle size distribution also decreased abruptly in the first 5 min and levelled off up to 160 min. The variation in the relative standard deviations of the absorbance showed a similar trend. The optimum grinding time was longer than 20 min for Be, 1.0 1 ( b ) I I 1 I I 1 I 0.4 I I I 1 I I 1 I I 0 10 20 0 10 20 0 10 20 Time/s Fig. 3 Absorbance-time profiles for (a) Cu; (b) Rb; and (c) Pb in A , aqueous solution; B, silicate rock powder (JG-1); and C, a mixture of rock powder and graphite powderANALYST, FEBRUARY 1992. VOL. 117 133 Co, Ni, Cu, Pb and Bi, although 20 min was sufficient for Li, Rb and Cs for good reproducibility. Based o n data obtained so far, the following grinding conditions are considered to be applicable: a 3.0 mg rock sample is ground for 20 min and the particle size range of the powder is 0.3-25 pm (median 2.4 pm; mode 2.0 pm), 12% of the particles being larger than 10 pm.According to Wilson,l-5 this size distribution is suitable for obtaining sufficien ti y accurate results. Addition of Graphite Powder Mixing o f graphite powder with solid powders in the graphite furnace gives a sharp spectral line12.13 owing to the improved atomization. This is because the total effective surface area of bulk samples is made larger and the heat conductivity is improved. Fig. 3 shows the change in the profiles of spectral lines with the addition of graphite powder. Fig. 4 shows the variation in the absorbance of Cu, Rb and Pb with amount of graphite powder added. Stable and maximum absorbance was obtained with 0.5 mg of rock powder mixed with 0.5 mg of graphite powder.Fig. 5 shows the secondary electron images of rock powder (JR-1) after heating at 2400 "C with and without graphite powder. The presence of graphite powder gives a sharp spectral line and enhances the intensity of the spectral lines for Li, Be, Ni, Cs, Pb and Bi. No change in the half-width of the spectral lines of Co, Cu and Rb was observed, although the absorbances of these elements were intensified. The mixing ratio of graphite powder to sample powder for maximum absorbance for each element is as follows: Li, >0.8; Be, 1.5-3.0; Co, 1.0-1.5: Ni, 1.0-1.5; Cu, 0.0-1.0; Rb, 0.8-1.0; Cs, 0.8-2.0; Pb, 1.G-1.5; and Bi, 0.8-1.5.A mixing ratio of graphite powder of 1.0 appeared to be suitable, although for Be the value was 2.0. The proposed method is effective even for a sample with a concentration that is too high for the dynamic range; a sample containing elements at levels as high as a few hundred ppm can be analysed simply by increasing the amount of graphie powder. 0.05 1 I I 1 Heating Programme Optimum temperatures for the drying, pyrolysis and atom- ization steps for each element were determined with JG-1 and for Pb with JB-la. Fig. 6 shows the variation in absorbance for Cu, Rb and Pb with temperature. The heating programme established is given in Table 2. Sample Amount Fig. 7 shows the variation in absorbance with amount of sample.The absorbances for Li, Be, Ni, Cu, Cs, Pb and Bi are proportional to sample amount from 0.2 to 1.0 mg and for Co and Rb up to 0.7 and 1.5 mg, respectively. When the sample amount exceeds the upper limit or the content of the element is more than 2-3 ng, the proportionality no longer holds. However, when aqueous standard solutions are atomized, proportionality holds up to 10-30 ng. This result implies that an increase in sample amount inhibits the diffusion of atomic vapour. Hence, the optimum sample amounts for Li, Ni, Cu, Rb, Cs, Pb and Bi are 1.0 mg and those for Be and Co, which are more sensitive than other elements, are 0.2 and 0.5 mg, respectively. Suppression of Ionization Interference For an element with an ionization potential higher than 4.6 eV, the ionization interference can be ignored.16 The ioniza- tion potentials of Rb and Cs are 4.2 and 3.9 eV, respectively, hence suppression of the interference is required.Potassium, with a low ionization potential, is usually added to suppress the interference. Rock samples usually contain significant amounts of Na and K which suppress the interference, but those elements are not present in standard solutions.134 ANALYST, FEBRUARY 1992, VOL. 117 0 500 1000 1500 TemperaturePC 2000 2500 Fig. 6 Variation in the absorbance of (a) Cu; (b) Rb; and (c) Pb in silicate rock samples (JG-1) with A, drying; B, pyrolysis; and C, atomization temperature Table 2 Analytical conditions for the determination of Li, Be, Co, Ni, Cu, Rb, Cs and Pb in silicate rock samples Parameter Li Be c o Ni c u Rb c s Pb Bi Sampleamount/mg 1.0 0.2 0.5 1 .o 1 .o 1 .o 1 .o 1.0 1.0 Grinding time*/min 20 Particle sizet/pm 2.0 (0.3-25) Mixing ratio of sample Atomization conditions: to graphite powder 1:l 1:2 1: 1 1 : l 1 : l 1:l 1:l 1 : l 1: 1 Drying 120,30$ 120,30 150,30 150,30 200,30 120,30 120,30 150,30 120,30 Pyrolysis 1400,30 1900,30 900,30 900,30 700,30 1300,30 1600,30 1000,30 1500,30 Atomizing 2600,20 2600,20 2600,20 2600,20 2600,20 2400,30 2600,lO 2600,15 2600,lO * 3.0 g of silicate rock samples ground with an Ishikawa type AGA grinder.t Mode diameter. $ The first value in each pair is temperature in "C and the second is time in seconds. (u C n 0.6 p n (0 U 0.4 CJ, C 4- - 0.2 0 0 0.5 1.0 1.5 2.0 Arnountlmg Fig. 7 Variation in the absorbance of A, Cu; B, Rb; and C, Pb in silicate rock powder with amount of sample Fig.8 shows the variation in absorbance for 5 ng of Rb and Cs with the amount of K. Maximum absorbance was obtained for Rb and Cs by adding 40 and 100 ng of K (eight times the amount of Rb and 20 times the amount of Cs) or more, respectively. For more than 500 ng, the data processor cannot compensate for the background absorbance. Application to Geochemical Standard Rock Samples The results for nine geochemical standard rock samples 0 1 2 3 4 Amountlmg Fig. 8 Variation in the absorbance of 5 ng of A, Rb; and B, Cs in 10 pl of aqueous solution with amount of K added as an ionization suppressor obtained under the proposed conditions are given in Table 3. The calibration standard solutions for Rb and Cs contain K in a 30-fold excess.Each value in Table 3 is the average of five measurements; the relative standard deviations are less than 10%. Values recommended by Ando et al. 17 are also given in Table 3 for comparison. Fig. 9 shows the correlation between the found and the recommended values. The correlation coefficient of 0.9993 indicates that the present results agree well with the recommended values, and that an aqueous standard solution is effective for the calibration in the direct atomization AAS of solid rock samples.ANALYST, FEBRUARY 1992, VOL. 117 135 Table 3 values17 Results for the determination of Li, Be, Co, Ni, Cu, Rb, Cs, Pb and Bi in GSJ standard reference rock samples (in ppm), and literature Li Be co Ni c u This RSD* Ref.This RSD* Ref. This RSD* Ref. This RSD* Ref. This RSD* Ref. Sample work (Yo) 17 work (Yo) 17 work (YO) 17 work (YO) 17 work (%) 17 JG-1 84.1 2 85.9 3.3 4 3.1 4.2 5 4.0 6.1 6 6.0 5.2 3 5.6 JGb-1 4.6 4 4.3 0.37 8 0.36 60.3 11 61.6 25.9 6 25.4 85.6 2 86.8 JR-1 63.6 6 62.3 3.2 4 3.1 0.70 5 0.65 0.60 5 0.66 1.4 8 1.4 JR-2 87.5 3 83 3.8 3 3.4 0.36 8 0.4 0.93 6 0.84 1.4 12 1.4 JA-1 10.8 2 10.5 0.57 7 0.50 10.2 8 11.8 5.9 2 5 42.4 10 42.4 JB-la 11.9 7 11.5 1.5 5 1.4 39.5 9 39.5 128.9 0.3 135 56.5 5 55.5 JB-2 8.3 3 8.0 0.28 5 0.27 36.1 8 39.8 19.9 6 19 222 5 227 JB-3 6.8 0.4 7.2 0.74 5 0.74 33.4 5 36.3 38.6 9 38.8 195 1 198 JP-1 1.5 4 1.8 N.D.? <1 127 9 116 - - 2460 4.4 6 5.7 Rb c s This RSD* Ref. This RSD* Sample work (YO) 17 work (Yo) JG-1 173.2 0.3 181 11.8 2.0 JGb-1 5.1 6 4 N.D.JP-1 N.D. <1 N.D. JR-1 260 0.7 257 23.1 2.4 JR-2 286 3 297 26.6 4.1 JA-1 11.6 3 11.8 N.D. JB-la 41.2 1 41 1.8 12 JB-2 6.2 9 6.2 N.D. JB-3 13.7 5 13 N.D. * Relative standard deviation (YO), n = 5. t N.D. = not detected. Pb Bi Ref. 17 0.52 0.014 0.51 0.65 0.009 0.033 0.020 - - Ref. 17 10.2 0.27 >o. 1 20.2 26 0.64 1.2 0.90 1.1 This work 25.1 2.2 N.D. 17.9 20.4 6.1 7.2 5.1 5.0 RSD* Ref. 3 26.2 7 1.9 6 19.1 1 21.9 4 5.8 4 7.2 6 5.4 4 5.5 (Yo) 17 0.11 This RSD* work (YO) 0.48 19 N.D. N.D. 0.48 11 0.65 8.9 0.95 5.9 N.D. N.D. N.D. 0 4 8 12 16 20 24 Reported value (ppm) Fig. 9 Correlation between reported values and those obtained with the proposed method for X , Li; +, Be; A, Co; 0, Ni; 0, Cu; 0, Rb; 0, Cs; A, Pb; and H, Bi in GSJ standard rock samples.The correlation coefficient is 0.9993 The authors thank S. Morita, H. Moriya and S. Miyanomoto for technical assistance, and the Machine Shop, Meiji University, for making the miniature cup. 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 Reeves, R. D., and Brooks, R. R., Trace Element Analysis of Geological Materials, Wiley-Interscience, New York, 1978, Govindaraju, K., Mevelle, G., and Chouard, C., Anal. Chem., 1974,46, 1972. Katskov, D. A., Kruglikova, L. P., L’vov, B. V., and Polzik, L. K., Zh. Prikl. Spektrosk., 1974, 20, 739. Katskov, D. A., Kruglikova, L. P., and L’vov, B. V., Zh. Anal. Khim., 1975, 30, 238. Karyakin, A. V., Pchelinsev, A. M., Shidlovskii, A. I., Vul’fson, E. K., and Tsimgarelli, M. N., Zh. Prikl. Spectrosk., 1973, 18, 610. Vul’fson, E. K., Karyakin, A. V., and Shidlovskii, A. I., Zh. Anal. Khim., 1973, 28, 1253. Katskov, D. A., and L’vov, B. V., Zh. Prikl. Spektrosk., 1969, 10, 382. Leucke, W., Eschermann, F., Lennartz, U., and Papastamat- aki, A. J., Neues Jahrb. Mineral. Abh., 1974, 120, 178. Fuller, C. W., and Thompson, I., Analyst, 1977, 102, 141. Isozaki, A., Morita, Y ., and Utsumi, S., Bunseki Kagaku, 1990, 39, 605. Langmyhr, F. J., Stubergh, J. R., Thomassen, Y., Hassen, J. E., and Dolezal, J., Anal. Chim. Acta, 1974, 71, 35. Siemer, D. D., and Wei, H. Y., Anal. Chem., 1978, 50, 147. Atsuya, L., and Itoh, K., Bunseki Kagaku, 1982,31, 708. Wilson, A. D., Analyst, 1964, 89, 416. Suzuki, M., and Ohta, K., Bunseki, 1984, 125,416. Ando, A., Mita, N., and Terashima, S., Geostand. Newsl., 1987, 11, 159. p. 151. References 1 Brooks, R. R., and Boswell. C. R., Anal. Chim. Acta, 1965,32, 339. Paper 1 I03891 E Received July 29, 1991 Accepted September 25, 1991