Removal of an Iron Matrix with Polyoxyethylene-Type Surfactant

Among the surfactants tested, polyoxyethylene(20)-4-isononylphenoxy ether (PONPE-20) showed the best performance: the iron leaked from the PONPE-20 co...
0 downloads 8 Views 114KB Size
Download PDF
Anal. Chem. 2005, 77, 5344-5348

Removal of an Iron Matrix with Polyoxyethylene-Type Surfactant-Coated Amberlite XAD-4 for the Determination of Trace Impurities in High-Purity Iron Hiroaki Matsumiya,* Shigeru Furuzawa, and Masataka Hiraide

Department of Molecular Design and Engineering, Graduate School of Engineering, Nagoya University, Furo-cho, Chikusa-ku, Nagoya 464-8603, Japan

Admicellar sorbents for the removal of an iron matrix were prepared for the determination of trace impurities in highpurity iron. A 1.0-g amount of Amberlite XAD-4 (macroreticular styrene-divinylbenzene copolymer) was coated with 0.14-1.3 mmol of polyoxyethylene-type surfactants, including polyoxyethylene-4-tert-octylphenoxy ethers (Triton X series) and polyoxyethylene-4-isononylphenoxy ethers (PONPEs). The surfactant-coated XAD-4 was packed into a polypropylene column (7 mm i.d. × 50 mm high). A 5.0-cm3 volume of sample solution was passed through the column at a flow rate of 0.5 cm3 min-1. Milligram amounts of iron(III) were effectively sorbed on the column from 8 mol dm-3 hydrochloric acid solutions. Among the surfactants tested, polyoxyethylene(20)-4-isononylphenoxy ether (PONPE-20) showed the best performance: the iron leaked from the PONPE-20 column was 4 µg when 25 mg of iron(III) was introduced onto the column. Trace elements, such as Ti(IV), Cr(III), Mn(II), Co(II), Ni(II), Cu(II), Zn(II), Ag(I), Cd(II), Pb(II), and Bi(III), were not retained on the column and thus quantitatively recovered in the column effluent. The effective separation of trace elements from an iron matrix allowed their accurate determinations by inductively coupled plasmamass spectrometry or graphite furnace atomic absorption spectrometry. The detection limits (3σ blank) were in the nanogram per gram range. The proposed method was successfully applied to the determination of trace impurities in high-purity iron samples. Iron and its alloys are essential materials in various fields of industry. The physical and chemical properties of iron metals strongly depend on their impurities, even at microgram per gram or lower levels.1-4 Therefore, the determination of trace impurities in high-purity iron metals is of great importance for the control of their quality as well as for a better understanding of the synergistic action and correlation of impurities. * To whom correspondence should be addressed. Tel.: +81-52-789-3591. Fax: +81-52-789-3241. E-mail: [email protected]. (1) Takaki, S.; Kimura, H. Scr. Met. 1976, 10, 1095-1100. (2) Abiko, K.; Kimura, H.; Nakane, Y. Trans. 1SIJ 1984, 24, B19. (3) Kimura, H. ISIJ Int. 1994, 34, 225-233. (4) Abiko, K. Phys. Status Solidi, A 1997, 160, 285-296.

5344 Analytical Chemistry, Vol. 77, No. 16, August 15, 2005

Modern instrumental analytical methods, such as graphite furnace atomic absorption spectrometry (GFAAS) and inductively coupled plasma-mass spectrometry (ICPMS), are so sensitive as to enable one to readily access to the fractional nanogram per cubic centimeter level of trace elements in a sample solution.5,6 However, the determination of trace elements accompanied by a large amount of matrix elements often requires preliminary matrix removal procedures to guarantee the accuracy and precision of the analytical results.7 In addition, if the matrixes are not sufficiently eliminated, the analytical instrument would be seriously contaminated. Liquid-liquid extraction of iron(III) from hydrochloric acid into some oxygenic organic solvents, such as diethyl ether and methyl isobutyl ketone, is a well-known technique for separating iron from other elements.8,9 In this extraction scheme, iron(III) in the aqueous phase is converted into the chloride complexes and then extracted into the organic phase by the strong, protonmediated interaction between the anionic iron(III) chloride complexes and the electron-donating oxygens of the organic solvent molecules; in other words, solvated ion pairs, e.g., [H+, FeCl4-], are extracted into the organic phase. Although the ion association extraction is facile and thus frequently employed for the removal of an iron matrix,10-13 there has been a growing demand for an alternative, organic solvent-free method from the viewpoint of preventing harmful impacts to human health and ecological systems. Ion-exchange separation is also frequently performed for the separation of trace analytes from an iron matrix and does not require organic solvents.13-18 However, highly hazardous hydro(5) Ingle, J. D., Jr.; Crouch, S. R. Spectrochemical Analysis; Prentice Hall: Englewood Cliffs, 1988. (6) Montaser, A., Golightly, D. W., Eds. Inductively Coupled Plasmas in Analytical Atomic Spectrometry; VCH Publishers: New York, 1987. (7) Zolotov, Yu. A.; Kuz’min, N. M. Preconcentration of Trace Elements; Elsevier: Amsterdam, 1990. (8) Dean, J. A. Chemical Separation Methods; Van Nostrand Reinhold Co.: New York, 1969; p 53. (9) Mizuike, A. Enrichment Techniques for Inorganic Trace Analysis; SpringerVerlag: Heidelberg, 1983; pp 46-50. (10) Chen, J. S.; Berndt, H.; Klockenka¨mper, R.; To ¨lg, G. Fresenius J. Anal. Chem. 1990, 338, 891-894. (11) Okano, T.; Matsumura, Y. Tetsu to Hagane´ 1991, 77, 1951-1958. (12) Dongling, L.; Xiaoyan, H.; Haizhou, W. Anal. Chim. Acta 2001, 449, 237241. (13) Takada, K. Phys. Status Solidi, A 1997, 160, 561-565. 10.1021/ac0504114 CCC: $30.25

© 2005 American Chemical Society Published on Web 06/30/2005

fluoric acid is often required for sufficient separation.14-16 Alternatively, masking an iron matrix with a large amount of complexing agents, such as oxalate17 and ethylenediamine-N,N,N′,N′tetraacetate,18 is sometimes performed, which potentially causes appreciable contamination. Coprecipitation is another organic solvent-free technique and is occasionally employed for preconcentrating trace elements in high-purity iron.13,19,20 However, a large amount of coprecipitant is, in turn, introduced into analytical instruments. In the present study, we took an alternative approach, in which surfactant aggregates formed on solid surfaces, namely, admicelles or hemimicelles, were employed for collecting an iron matrix. Since our first report on the analytical use of admicellar sorbents for the preconcentration of trace metals in water,21 the potential and utility of admicellar sorbents in trace analysis have been investigated by us21-31 and other researchers.32-36 The reported applications are generally based on the partitioning of hydrophobic metal complexes or organic compounds between the aqueous sample solution and the surfactant aggregates by hydrophobic interaction. On the other hand, other interactions have been scarcely utilized in the admicelle-mediated separation.27,36 Several reports describe the strong interaction between some anionic metal complexes (e.g., Co(SCN)42-, AuCl4-, and GaCl4-) and polyoxyethylene-type nonionic surfactants, in which ethylene oxide chains behave in a manner similar to that of oxygenic organic solvents (e.g., ethers) in the ion association extraction.37-39 It has also been reported that Amberlite XAD-4 resin (macroreticular styrene-divinylbenzene copolymer) acts as an effective (14) Yamada, K.; Kujirai, O.; Hasegawa, R. Anal. Sci. 1993, 9, 385-390. (15) Coedo, A. G.; Lo´pez, T. D.; Alguacil, F. Anal. Chim. Acta 1995, 315, 331338. (16) Fujimoto, K.; Shimura, M. Bunseki Kagaku 2001, 50, 175-182. (17) Oguma, K.; Kato, K.; Kurashima, Y.; Seki, T, Ono, A.; Ishibashi, Y. Tetsu to Hagane´ 1999, 85, 119-123. (18) Yamaguchi, H.; Itoh, S.; Hasegawa, S.; Ide, K.; Kobayashi, T. Tetsu to Hagane´ 2004, 90, 48-50. (19) Danzaki, Y. Fresenius J. Anal. Chem. 1996, 356, 143-145. (20) Itagaki, T.; Ashino, T.; Takada, K. Fresenius J. Anal. Chem. 2000, 368, 344349. (21) Hiraide, M.; Sorouradin, M. H.; Kawaguchi, H. Anal. Sci. 1994, 10, 125127. (22) Hiraide, M.; Ohta, Y.; Kawaguchi, H. Fresenius J. Anal. Chem. 1994, 350, 648-650. (23) Hiraide, M.; Iwasawa, J.; Hiramatsu, S.; Kawaguchi, H. Anal. Sci. 1995, 11, 611-615. (24) Hiraide, M.; Iwasawa, J.; Kawaguchi, H. Talanta 1997, 44, 231-237. (25) Hiraide, M.; Shibata, W. Anal. Sci. 1998, 14, 1085-1088. (26) Hiraide, M.; Hori, J. Anal. Sci. 1999, 15, 1055-1058. (27) Saitoh, T.; Akita, S.; Torii, T.; Hiraide, M. J. Chromatogr., A 2001, 932, 159-163. (28) Hiraide, M.; Ishikawa, A. Anal. Sci. 2002, 18, 199-201. (29) Saitoh, T.; Nakayama, Y.; Hiraide, M. J. Chromatogr., A 2002, 972, 205209. (30) Hiraide, M.; Itoh, T. Anal. Sci. 2004, 20, 231-233. (31) Saitoh, T.; Matsushima, S.; Hiraide, M. J. Chromatogr., A 2004, 1040, 185191. (32) Manzoori, J.; Sorouraddin, M. H.; Haji-Shabani, A. M. J. Anal. At. Spectrom. 1998, 13, 305-308. (33) Shemirani, F.; Akhavi, B. T. S. Anal. Lett. 2001, 34, 2179-2188. (34) Absalan, G.; Mehrdjardi, M. A. Sep. Purif. Technol. 2003, 33, 95-101. (35) Dadfarnia, S.; Haji-Shabani, A. M.; Gohari, M. Talanta 2004, 64, 682-687 (36) Merino, F.; Rubino, S.; Pe´rez-Bendito, D. Anal. Chem. 2004, 76, 38783886. (37) Sotobayashi, T.; Suzuki, T.; Yamada, K. Chem. Lett. 1976, 77-80. (38) Akita, S.; Yang, L.; Takeuchi, H. Hydrometallugy 1996, 43, 37-46. (39) Kinoshita, T.; Akita, S.; Nii, S.; Kawaizumi, F.; Takahashi, K. Sep. Purif. Technol. 2004, 37, 127-133.

Figure 1. Chemical structures of surfactants tested.

adsorbent for polyoxyethylene-type nonionic surfactants.40 Therefore, the combination of polyoxyethylene-type nonionic surfactants and XAD-4 resin should afford novel admicellar sorbents to retain iron(III) in hydrochloric acid media. The use of admicellar sorbents would be advantageous over other micelle-mediated separation schemes, such as micellar-enhanced ultrafiltration41,42 and cloud point extraction,42-44 because of freedom from difficulties encountered in separating small micelles from the bulk aqueous phase and handling viscous surfactant solutions. It is also emphasized that the use of admicellar sorbents for removing matrix elements (not collecting trace analytes) has not been attempted so far. Herein, we report a facile and organic solvent-free method for the removal of an iron matrix. The adsorption behavior of five types of surfactants (Figure 1) on XAD-4 resin and the iron removal performance of the resulting surfactant-coated XAD-4 resins were examined. The optimized column sorption system offered the selective and almost quantitative (more than 99.9%) removal of an iron matrix, allowing the determination of trace impurities in high-purity iron by ICPMS and GFAAS. EXPERIMENTAL SECTION Apparatus. A Seiko SPQ-6500 ICP-mass spectrometer (Chiba, Japan) was used for the determination of 48Ti+, 55Mn+, 59Co+, 60Ni+, 63Cu+, 64Zn+, 107Ag+, 114Cd+, 208Pb+, and 209Bi+ under the following plasma conditions: rf power, 1.2 kW; sampling depth, 12 mm; argon flow rates (in dm3 min-1), 18 for outer, 0.8 for intermediate, and 1.0 for carrier. A Perkin-Elmer Model AAnalyst 600 graphite furnace atomic absorption spectrometer equipped with a Zeeman effect background corrector (Norwalk, CT) was used for the determination of chromium and iron under the following furnace operating conditions: The graphite tube was warmed during 1 s to 110 °C and held for 30 s. The tube was further heated during 15 s to 130 °C and held for 30 s; it was then heated during 10 s to a pyrolysis temperature of 900 °C and held for 20 s. The tube was immediately heated to an atomization temperature of 2300 (Cr) or 2100 °C (Fe) and held for 5 s. Cleanup was done at 2450 °C for 3 s. The wavelengths used were 357.9 (Cr) and 248.3 nm (Fe). A Jasco model V-550 UV/visible double-beam spectrophotometer (Tokyo, Japan) was used with a 1-cm quartz cell for the determination of surfactants at 277 nm. Separation procedures were carried out in a Hitachi model ECV-843 BY clean bench. Materials and Reagents. An Amberlite XAD-4 resin (Rohm and Haas, Philadelphia, PA) was pulverized to 100-200 µm and (40) Jones, P.; Nickless, G. J. Chromatogr. 1978, 156, 87-97. (41) Dunn, R. O.; Scamehorn, J. F.; Christian, S. D. Sep. Sci. Technol. 1985, 20, 257-284. (42) Stalikas, C. D. Trends Anal. Chem. 2002, 21, 343-355. (43) Hinze, W. L.; Pramauro, E. CRC Crit. Rev. Anal. Chem. 1993, 24, 133177. (44) Tani, H.; Kamidate, T.; Watanabe, H. J. Chromatogr., A 1997, 780, 229241.

Analytical Chemistry, Vol. 77, No. 16, August 15, 2005

5345

washed with ethanol and water before use. Polyoxyethylene-4tert-octylphenoxy ethers with average numbers of ethylene oxide units of 9.5, 30, and 40 (Triton X-100, Triton X-305, and Triton X-405, respectively) were purchased from Nacalai Tesque (Kyoto, Japan). Polyoxyethylene-4-isononylphenoxy ethers with average numbers of ethylene oxide units of 10 and 20 (PONPE-10 and PONPE-20, respectively) were purchased from Tokyo Kasei (Tokyo, Japan). Working solutions of the surfactants were prepared by dissolving the respective surfactants in water to give concentrations of 0.30 (Triton X-100), 0.10 (Triton X-305, Triton X-405), 0.050 (PONPE-10), and 0.030 M (PONPE-20) (1 M ≡ 1 mol dm-3). A high-purity iron (99.998%, Puratonic grade, Johnson Matthey GmbH, Karlsruhe, Germany) was used for the optimization of the experimental conditions. Certified reference materials for high-purity iron, NIST 2168 (chips) and CMSI 1010b (powder), were obtained from National Institute of Standards & Technology (Gaithersburg, MD) and China Metallurgical Standardization Research Institute (Beijing, China), respectively. Hydrochloric acid and nitric acid were of ultrapure grade (Kanto Kagaku, Tokyo, Japan). Commercial standard metal solutions were purchased from Katayama Kagaku (Osaka, Japan). All reagents used were of reagent grade, unless otherwise stated. Water was purified with a Milli-Q Gradient water system (Millipore, Billerica, MA). Preparation of Surfactant-Coated XAD-4 Column. A 1.0-g amount of XAD-4 was added to 10 cm3 of surfactant solution. After the suspension was gently stirred for 4 h, the supernatant solution was discarded and the remaining XAD-4 was packed into a polypropylene column (7 mm i.d. × 50 mm high). The column was washed with 50 cm3 of water and then conditioned with 5.0 cm3 of a 2 + 1 mixture of 12 M HCl and 3 M HNO3 before use. Analytical Procedure. Sample Dissolution. A 0.25-g sample of iron, placed in a 20-cm3 PTFE beaker, was dissolved in 10 cm3 of a 1 + 1 mixture of 2 M HCl and 2 M HNO3 by heating at ∼60 °C. The resulting solution was cooled to room temperature and transferred to a 50-cm3 PTFE volumetric flask. After adding 33 cm3 of 12 M HCl and 3.0 cm3 of 16 M HNO3, the solution was diluted to the mark with water to give the sample solution (Fe 5.0 mg cm-3, HCl 8 M, HNO3 1 M). The blank solution was prepared by the same procedure as described above without iron. Matrix Removal. A 5.0-cm3 aliquot of the sample solution (or the blank solution) was passed through the surfactant-coated XAD-4 column at a flow rate of 0.5 cm3 min-1. The column was subsequently washed twice with 2.0 cm3 of a 2 + 1 mixture of 12 M HCl and 3 M HNO3. The column effluent was collected in a 25-cm3 PTFE beaker and evaporated to near-dryness. After cooling to room temperature, the residue was dissolved in 2.0 cm3 of 0.1 M HNO3 and subjected to analysis by ICPMS or GFAAS (after dilution, if necessary). Calibration graphs were constructed using 0.1 M HNO3 solution containing metals of interest at picogram to nanogram per milliliter levels. RESULTS AND DISCUSSION Coating XAD-4 with Polyoxyethylene-Type Surfactants. First, the amounts of the surfactants sorbed on XAD-4 were investigated as follows. A 1.0-g amount of XAD-4 was added to 10 cm3 of water containing different amounts of the surfactant examined. The concentrations of the surfactants (5.0 mM-0.40 M) were greater than their respective critical micelle concentra5346 Analytical Chemistry, Vol. 77, No. 16, August 15, 2005

Figure 2. Sorption of surfactants on 1.0 g of XAD-4.

tions (sub-mM levels)45,46 and thus high enough for the formation of surfactant aggregates. After the suspension was gently stirred for a given time, the supernatant solution was analyzed by spectrophotometry to determine the sorption yield. The sorption proceeded nearly completely within 2 h. A 4-h stirring was sufficient for the equilibrium sorption. Although an attempt was made to accelerate the surfactant sorption by sonication (28 kHz, 210 W), it was not effective in the acceleration of the sorption. Figure 2 shows the relations between the amounts of the surfactants added and sorbed. All the surfactants examined here were sufficiently retained on XAD-4. The maximum amounts of the surfactants sorbed on 1.0 g of XAD-4 (in mmol) were 1.3 for Triton X-100, 0.40 for Triton X-305, 0.22 for Triton X-405, 0.20 for PONPE-10, and 0.14 for PONPE-20. In the fields of surface and colloid chemistry, the adsorption behavior of polyoxyethylenetype surfactants, including Triton X-100 and Triton X-405, on polystyrene surfaces has been discussed in detail recently.47-49 According to these studies, the adsorption is driven mainly by the attractive interaction between the hydrophobic tail (i.e., alkylated phenolic moiety) of the surfactant molecule and the resin surface. Some organized structures, such as monolayer,47 hemisphere,48 and hemicylinder,48 have been proposed for the adsorbed surfactant aggregates formed at high concentrations. In these adsorption models, the hydrophobic tails are estimated to be oriented toward the resin surface. As can be seen in Figure 2, the retention of the surfactants having the same hydrophobic tail decreased as the number of ethylene oxide unit increased, which can be explained by the repulsive interaction between the ethylene oxide chains.49 In the following studies, the resulting surfactantcoated XAD-4 resins were packed into respective columns and used for the sorption of an iron matrix. Sorption Behavior of Iron. As described in the introduction, the ethylene oxide chains of the surfactants were expected to interact with iron(III) in HCl media; hence, the iron samples were (45) Schick, M. J.; Atlas, S. M.; Eirich, F. R. J. Phys. Chem. 1962, 66, 13261333. (46) Crook, E. H.; Fordyce, D. B.; Trebbi, G. F. J. Phys. Chem. 1963, 67, 19871994. (47) Romero-Cano, M. S.; Martı´n-Rodrı´guez, A.; de las Nieves, F. J. J. Colloid Interface Sci. 2000, 227, 322-328. (48) Jo´dar-Reyes, A. B.; Ortega-Vinuesa, J. L.; Martı´n-Rodrı´guez, A.; Leermakers, F. A. M. Langmuir 2003, 19, 878-887. (49) Jo´dar-Reyes, A. B.; Ortega-Vinuesa, J. L.; Martı´n-Rodrı´guez, A. J. Colloid Interface Sci. 2005, 282, 439-447.

Figure 3. Effect of the concentration of HCl on the sorption of 10 mg of iron on the surfactant-coated XAD-4 column. XAD-4, 1.0 g. Surfactants loaded (in mmol): 1.3 for Triton X-100, 0.40 for Triton X-305, 0.22 for Triton X-405, 0.20 for PONPE-10, and 0.14 for PONPE-20. [HNO3] ) 1 M.

Figure 4. Sorption capacity of the surfactant-coated XAD-4 column for iron. XAD-4, 1.0 g. Surfactants loaded (in mmol): 1.3 for Triton X-100, 0.40 for Triton X-305, 0.22 for Triton X-405, 0.20 for PONPE10, and 0.14 for PONPE-20. [HNO3] ) 1 M. [HCl] ) 8 M.

dissolved in HCl. The complete oxidation of the iron matrix to iron(III) required the addition of a small amount of HNO3, otherwise the sorption recovery decreased considerably (vide infra). The effect of the HCl concentration on the sorption efficiency was investigated as follows. A 5.0-cm3 volume of sample solution containing 10 mg of iron(III) was introduced onto the surfactantcoated XAD-4 column at different HCl concentrations, and the column effluent was analyzed for iron. As shown in Figure 3, the relations between the HCl concentration and the amount of iron found in the effluent showed a reversed bell-shaped profile, irrespective of the kind of surfactant. The sorption efficiency increased with increasing HCl concentration up to 8 M and slightly decreased at 10 M, which was similar to the extraction behavior of iron(III) in the ion association extraction with oxygenic organic solvents.9 In contrast, an untreated XAD-4 (without the surfactants) retained only ∼5% of the iron introduced onto the column. These results suggest that the sorption of iron onto the surfactant-coated XAD-4 is ascribed to the surfactant molecules loaded on the XAD-4 and that the surfactant aggregates provide an environment similar to that of the oxygenic organic solvents employed for the ion association extraction. The highest sorption was observed at the HCl concentration of 8 M, which was thus chosen as the optimum acidity. The sorption capacities of the surfactant-coated XAD-4 columns were investigated using sample solutions containing different amounts of iron(III). As shown in Figure 4, the PONPE-20-coated XAD-4 column showed the highest capacity: the maximum sorption of iron was 30 mg, and an almost quantitative (more than 99.9%) removal of iron was achieved for up to 25 mg of iron. In the absence of HNO3, however, 4.6% of the iron was leaked from the PONPE-20 column even when introducing 10 mg of iron. The column effluent showed light green, suggesting that most of the iron being not retained on the column was iron(II). Therefore, the insufficient oxidation of iron seems to be responsible for the decrease in the sorption efficiency. Multielement Separation of Trace Elements from Iron Matrix. Because the PONPE-20 column was most promising, it was employed for the multielement separation of trace elements from an iron matrix. To assess the reliability of the proposed method for the determination of trace impurities at the fractional

microgram per gram level, the losses of trace elements during the separation procedure were evaluated as follows. Titanium(IV), Cr(III), Mn(II), Co(II), Ni(II), Cu(II), Zn(II), Ag(I), Cd(II), Pb(II), and Bi(III) were taken as examples of trace elements because they were expected to pass through the column, considering their extraction behavior in the ion association extraction with oxygenic organic solvents.9 The high sensitivity of ICPMS and GFAAS for these elements is another reason for the selection. Synthetic sample solutions were prepared by adding 0.2-40 ng each of these elements to 5.0 cm3 of 8 M HCl containing 1 M HNO3 and 25 mg of high-purity iron. The sample solution was passed through the column, which was subsequently washed twice with 2.0 cm3 of 8 M HCl containing 1 M HNO3. For concentrating the desired elements as well as removing HCl, the column effluent (i.e., 5.0 cm3 of sample solution plus 4.0 cm3 of washing solution) was evaporated and the residue was dissolved in 2.0 cm3 of 0.1 M HNO3. The resulting solution was subjected to the determination of heavy metals by ICPMS. However, the determination of Cr at the nanogram per cubic centimeter level was difficult because of serious spectral overlaps (52Cr+ and 40Ar12C+; 53Cr+ and 37Cl16O+), though most of the HCl was removed by evaporation. Therefore, GFAAS was employed for the determination of Cr. The iron accompanying the desired elements was 3.9 ( 0.6 µg (n ) 5), which did not interfere with the determination by ICPMS and GFAAS. As described above, the PONPE-20 column, packed with 1.0 g of the sorbent, was capable of sorbing up to 30 mg of iron. Therefore, the breakthrough volume is calculated to be 6.0 cm3 when a sample solution prepared as described in Sample Dissolution (Fe, 5.0 mg cm-3) is applied to the column. Figure 5 shows the results of the recovery tests. The blank values, obtained by the same procedure as described in Analytical Procedure using a blank solution (without iron), were as follows (in ng, n ) 6): 4.4 ( 0.3 for Ti, 4.2 ( 0.3 for Cr, 0.85 ( 0.04 for Mn, 0.15 ( 0.01 for Co, 3.7 ( 0.2 for Ni, 7.2 ( 0.2 for Cu, 17 ( 2 for Zn, 0.28 ( 0.02 for Ag, 0.29 ( 0.02 for Cd, 6.3 ( 0.5 for Pb, and 0.10 ( 0.01 for Bi. The relations between the amounts of metals added and found showed good linearity with slopes of 0.97-1.02, indicating that nanogram quantities of the desired heavy metals were quantitatively recovered in the final solution. For Mn, Co, Ag, Cd, and Bi, the linearity still existed even at the picogram level. The reproducibility was also acceptable: the Analytical Chemistry, Vol. 77, No. 16, August 15, 2005

5347

Figure 5. Multielement separation of trace heavy metals from 25 mg of iron with the PONPE-20 column. Table 1. Detection Limitsa metal

detection limit/ng g-1

Ti Cr Mn Co Ni Cu

30 40 4 2 30 30

Table 2. Analytical Results for High-Purity Iron metal

detection limit/ng g-1

Zn Ag Cd Pb Bi

200 3 2 60 1

a Calculated as the analyte contents corresponding to three times the standard deviations of multiple blanks (n ) 6).

relative standard deviations were in the range of 1 (Ni and Ag) to 6% (Zn) for five replicate runs at 20 ng each of heavy metals. The small and reproducible blanks, as well as the high sensitivity of ICPMS and GFAAS, seem to contribute to the accurate and precise results obtained here. The graphs also indicate that the iron metal used (99.998% purity) contained some heavy metals at fractional microgram per gram levels. The impurity contents estimated from the intercepts were as follows (in µg g-1): 0.02 for Ti, 0.16 for Cr, 0.02 for Mn, 0 for Co, 0.21 for Ni, 0.08 for Cu, 0.15 for Zn, 0.010 for Ag, 0.001 for Cd, 0.28 for Pb, and 0 for Bi. The detection limits were calculated as the analyte contents corresponding to three times the standard deviations of multiple blanks (n ) 6). As shown in Table 1, the detection limits approached the nanogram per gram levels. The PONPE-20 column was substantially stable during the separation procedure and regenerated by simply washing the column with 5.0 cm3 of water. No marked decrease in the separation efficiency was observed after its repeated use five times. Determination of Trace Impurities in High-Purity Iron. To demonstrate the practical applicability, the proposed method was applied to the analysis of certified reference materials for highpurity iron (NIST 2168 and CMSI 1010b). A 0.25-g amount of the iron sample was taken to prepare the sample solution and a 1/10 aliquot was subjected to the matrix removal protocol, followed by the ICPMS or GFAAS determination, as described in Analytical Procedure. 5348 Analytical Chemistry, Vol. 77, No. 16, August 15, 2005

concn in sample/µg g-1 metal Ti Cr Mn Co Ni Cu Zn Ag Cd Pb Bi

determineda

certified

Sample: NIST 2168