JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY DECEMBER 1992 VOL. 7 1167 Anion Exchange for the Elimination of Spectral Interferences Caused by Chlorine and Sulfur in Inductively Coupled Plasma Mass Spectrometry Jan Goossens and Richard Dams Laboratory of Analytical Chemistry Institute for Nuclear Sciences University of Gent Proeftuinstraat 86 8-9000 Gent Belgium An easily applicable separation method has been developed for the accurate and simultaneous determination of V Cr Cu Zn As and Se in biological clinical and environmental samples by inductively coupled plasma mass spectrometry. Chlorine and sulfur which cause spectral interference with these elements are retained as anions on an anion-exchange resin column (Dowex-1) whereas the analytes are eluted with dilute nitric acid and collected in the eluate.In most of the examples the sample preparation is limited to a reduction with SnCI,. This new approach to the elimination of spectral interferences has been applied to soils percolate water sewage and human serum. The results are in good agreement with certified values and results obtained by d.c. plasma atomic emission spectrometry and electrothermal vaporization atomic absorption spectrometry; differences between mean values are 4%. Keywords Inductively coupled plasma mass spectrometry; spectral interference; anion exchange; determina- tion of vanadium chromium copper zinc arsenic and selenium Inductively coupled plasma mass spectrometry (ICP-MS) is a relatively new technique combining high-speed analysis multi-element capabilities and high sensitivity.However owing to the limited resolution of the quadrupole mass analyser spectral overlap occurs when ion masses differ by less than 0.5 u. In biological and environmental samples many of these interferences are caused by the matrix elements C1 and S.lJ These elements give rise to polyatomic species leading to spectral overlap with analytes of physio- logical and toxicological importance (V Cr Cu Zn As and Se). A survey of these interferences is presented in Tables 1 and 2. Several solutions to this problem have already been suggested. 'Simulated blank solution^',^ containing equal concentrations of interfering matrix elements as the sample solutions can be prepared and the apparent analyte concentrations determined. However since the formation Table 1 Interferences due to sulfur Species subject to interference Interfering polyatomic Element Isotope species Chromium 50Cr(4.35%) 34S160 Zinc 64Zn(48.90/o) 3 2 s 1 6 0 I 6 0 66Zn(27.80/o) 34S160160 67Zn(4.1 O/o) Copper 65Cu( 30.9%) 32Sl60160lH 34S160160 I H 68Zn( 18.6%) 34S160180 Table 2 Interferences due to chlorine Species subject to interference Interfering polyatomic Element I sot ope species Vanadium 51 V( 99.7%) 3 5 ~ 1 1 6 0 Chromium Wr(8 3.8%) 3 5 ~ 1 1 6 0 1 ~ 53Cr( 9.5%) 3 7 ~ 1 1 6 0 Iron 54Fe(5.80h) 3 7 ~ 1 1 6 0 1 ~ 67Zn(4. 1 Oh) 35C1160160 Zinc Arsenic 7 5 A ~ ( 1 OOOh) 40Ar35C1 Selenium 77Se( 7.5%) *OAr3'CI of polyatomic species strongly depends on matrix condi- tions this procedure requires a complete knowledge of the sample matrix composition to allow the addition of the correct amounts of all matrix elements to the blank solution. An interesting alternative is the retention of the analyte elements on cation-exchange resin columns while interfer- ing matrix ions are el~ted.~9~ A serious drawback of this method is the fact that all analyte elements have to be present as cations which implies decomposition of organo- metallic complexes and the loss of some interesting anions (e.g.As033- and Se03z-). Ion chromatographyY6 gel filtra- tion,' electrothermal vaporization* and hydride generation9 have also been successfully applied. The use of mathematical correctionslO*ll could be useful but often these corrections suffer from uncertain assump- tions systematic errors and error amplification leading to inaccurate and imprecise results.The use of alternative plasma gases (e.g. He)12 in order to shift some of the interfering species to other mass regions seems very promising while the addition of Nz to the Ar plasma gasL3J4 in order to alleviate the formation of some polyatomic species is being investigated. In the present paper a new approach is presented whereby C1 and S anions are removed from the sample solution by retention on a Dowex-1 resin column in the NO3- form.15J6 Dowex-1 is a strongly basic anion-exchange resin with a quaternary ammonium functionality; C1- C104- S042- S032- and some other anions are retained on the resin while analyte elements including trace metals subject to interference (V Cr Cu Zn As and Se) are eluted. The main advantage of this procedure is that both anionic and cationic analytes can be simultaneously sepa- rated from C1 and S. Moreover there is no need for the analyte elements to be present in free cationic form.The only anions that cannot be held at all by the resin are complex anions or organic anions which because of their size and configuration cannot enter the interior of the resin particles. Experimental Instrumentation A VG PlasmaQuad (VG Elemental Winsford UK) was used for the ICP-MS measurements; details of the instru- ment and operating conditions are summarized in Table 3.1168 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY DECEMBER 1992 VOL. 7 Table 3 PlasmaQuad operating conditions Plasma- R.f. power Forward/W Reflected/W Plasma/I min-l Nebulizer/l min-' Auxiliary/l rnin-l Gas flows Nebulizer Spray chamber Ion sampling- Sampling cone Skimmer cone Vacuum- 1350 ( 5 13.5 0.725 0.9 1 Meinhard concentric type Scott-type double bypass pumped at 0.9 ml min-'* water cooled Nickel 1.0 mm orifice Nickel 0.75 mm orifice Expansion stage/mbart 2.3 Intermediate stage/mbar 2.0 x 1 0 - 4 Analyser stage/mbar 4.6 x 1 0 - 4 * Increased up to 7 ml mind' for qualitative on line experiments.t 1 bar=105 Pa. The ion lens settings were optimized for the internal standard (74Ge or '%I). The measurements were performed in the mass scanning data-acquisition mode limited mass regions (Am) were scanned entirely. The number of chan- nels n was always chosen so that nlAm was a minimum of 20 and thus accurate integration of the signal peaks was possible.The dwell time per channel was 160 ,us. A standard number of 200 sweeps per measurement was applied. For some qualitative experiments data were collected by setting the quadrupole at a fixed mlz value with use of the multichannel analyser for time-resolved analysis at a single mass. Comparative analyses were carried out by electrothermal vaporization atomic absorption spectrometry (ETV-AAS) and d.c. plasma atomic emission spectrometry (DCP-AES). The ETV-AAS instrument used was a Perkin-Elmer (Nor- walk CT USA) Model 3030 equipped with a deuterium-arc background corrector and a Perkin-Elmer Model HGA 400 graphite furnace atomizer. The DCP instrument was purchased from Applied Research Laboratories (ARL) Ecublens Switzerland and was a DCP Spectrajet 111. Reagents and Solutions The anion-exchange resin Dowex-1 X8 (mesh size 100-200; technical grade) commercially available in C1- form (Serva Heidelberg Germany) was used.No pre-treatment or purification other than described under 'Column Prepa- ration' was carried out. Commercial standard solutions of 1 g 1-l [Fluka (Buchs Switzerland) Janssen Chimica (Beerse Belgium) Merck (Darmstadt Germany)] were used to prepare the synthetic samples. For calibration purposes use was made of com- mercial standard solutions (1 g 1-I) for V Cr and As (Fluka) whereas pure metals (>99.95%) were used to prepare standard solutions of Cu Zn and Se by dissolving the metal in concentrated nitric acid followed by appropri- ate dilution. All acids used were purified by sub-boiling distillation except for perchloric acid (Merck Suprapur) and hydrofluoric acid (J.T. Baker Instra-Analyzed; Philips- burg NJ USA; Suprapur). Millipore (Milford MA USA) Milli-Q water was used throughout. A Sn*I solution was prepared by dissolving approximately 15 g of SnCl2-2H20 (pro analysi) in 5 ml of boiling 10 moll-' HC1 followed by dilution to 100 ml with water. Sample Preparation A synthetic sample was prepared containing V Cr Mn Fe Co Ni Cu Zn As and Se at the 500 pg 1-l level in 0.14 rnol I-' HN03. Amounts of HCl and H2S04 equivalent to 3 g 1-l of C1 and 3 g 1-2 of S were added. Four authentic samples all containing high concentrations of C1 and/or S were analysed sewage percolate water light sandy soil and human serum. For the analysis of sewage and percolate water no further sample pre-treatment was carried out The sewage containing 1.1 g 1-l of C1- and 6.0 g 1-' of S042- was filtered and provided by LISEC (Limburgs Studiecen- trum voor Toegepaste Ecologie).The percolate water was analysed by several laboratories in the framework of a ring analysis campaign organized by VITO (Vlaams Instituut voor Technologisch Onderzoek). Further the aqua regia soluble content of a candidate BCR (Community Bureau of Reference) certified reference material (CRM 142R Light Sandy Soil) was determined. The aqua regia soluble phase of heavy metals in soils and sludges is often of more ecological importance than the total content since it can be considered as the bio-available fraction for plants and crops. Approximately 1 g of the Light Sandy Soil was heated under reflux in 10 ml of aqua regia (7.5 ml of HC1+2.5 ml of HN03) and filtered closely following the procedure described by the BCR (DIN 38 414-57).The filtrate obtained was diluted to 1000 ml with 0.14 mol 1-' HN03. Blank solutions were prepared in the same way. For the analysis of freeze-dried human serum 1 g of sample was digested with concentrated HN03 and HC104 and the digest was evaporated to near-dryness. The decom- position was necessary to convert the organically bound S into S042-. The sample was again diluted to 10 ml with 0.14 mol 1-l HN03. The freeze-dried human serum was the 'second generation' biological reference material prepared by Versieck et af.17 This material was collected and stored under rigorously controlled conditions in order to avoid contamination so that the concentrations of most trace elements were fairly similar to those expected in normal human serum.The addition of HN03 and of strong acids in general to the samples prior to elution was limited as much as possible to avoid column overload. To achieve complete separation lihe HNO concentration in the sample solutions should not exceed 0.3 rnol 1-l. Column Preparation ,4 40 ml (wet volume) portion of resin was transferred into a polyethylene tube (20 cm in length 2.7 cm i.d.) and converted into the NO3- form by rinsing the column with 1.4 mol 1-l HN03. The column was next washed with 0.0014 mol 1-l HN03 until the same pH value was measured at the inlet and outlet. A column of this size has a theoretical ion capacity of approximately 50 mequiv (1.2 mequiv ml-l wet resin) but the amount of adsorbable material in the sample should not exceed 10°/o of this value.For regeneration purposes the column was rinsed with 1.4 mol 1-1 HNO until the effluent no longer tested positive to C1- (precipitation of AgCl by addition of 0.01 mol 1-1 AgN03). The presence of HC104 could cause regeneration problems because of the high affinity of the resin for C104-. In this instance it was necessary to renew the resin partially before regeneration. For the experiments described below regeneration was carried out after each elution. Elution Procedure Depending on the analyte concentration and the amount of sample available 2.5-50 ml of the sample solution wereJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY DECEMBER 1992 VOL.7 1169 applied to the column. After the sample drained into the resin bed a maximum volume of 120 ml of the eluent (0.0014 mol 1-I HN03) were added in small portions and elution was carried out. A constant elution rate of 10 ml min-l was obtained by using a peristaltic pump [Ismatec MS-4/8 (Zurich Switzerland)]. A single separation required about 15 min. Preliminary Studies Choice of a Suitable Reductant Owing to the relatively high acid strength of H3V04 H3As0 and H,SeO partial retention can occur for V As and Se if these elements are present in high oxidation states. Therefore when at least one of these elements has to be determined a suitable reductant has to be added to the sample for the reduction of Vv AsV and Sevl to VIV As111 and SeIV respectively.A theoretical approach to this problem is presented in Fig. 1. For a given pH the redox potential of the reductant should be situated in the region between lines 3 and 4. The upper limit is the maximum redox potential for quantitative (> 99%) reduction of the three elements mentioned above while a minimum value is imposed by the possible reduction of Asill to Aso. Taking this into consideration SnC12 and CuCl are suitable reductants; this was confirmed experimentally. Tin(I1) chloride was pre- ferred because of its greater stability and solubility and because of the fact that Cu may be an analyte of interest itself A standard volume of 100 pl of Sn1I solution was added to the samples unless mentioned otherwise. It is important to note that unreduced organometallic com- 1.2 1 0.8 .m 0.6 0.4 5.U C P 5 0.2 d o -0 -0.2 -0.4 0 1 2 3 4 5 6 7 PH Fig. 1 Maximum redox potential for quantitative reduction of SeV1 (line I) Vv (line 2) and AsV (line 3) to SeIV VvIv and As111 respectively. Minimum redox potential for quantitative oxidation of Aso ( 1 mg I-') to As111 (line 4). The redox potential of the reductant chosen should be situated in the region between lines 3 and 4 plexes (e.g. arsenobetaine) are not retained by the resin. A trace amount of HF (1 00 p1 of a 5% solution) was always added to the eluate to prevent the precipitation of Sn(OH)2. Effect of the Eluent Concentration on Retention and Recover- ies As already mentioned dilute HNO was used as the eluent for the separations. The concentration of the HN03 (0.00 14 moll-]) is a compromise suppression of the dissociation of acid analytes (e.g.H,As03) requires a higher concentration whereas a lower concentration is favourable in view of the competition between NO3- and all other adsorbable species for the active sites on the resin. The elution diagrams of Cr3+ and H3As03 are shown in Fig 2. It was found that all positively charged analytes (V3+ VOz+ Cr3+ MnZ+ Fe3+ Co2+ Ni2+ Cu2+ and Zn2+) show very similar elution characteristics. Arsenic however is slightly retarded and peak broadening can be observed. For a high elution yield a sufficient volume of eluate must be collected. Selenious acid (H2Se03) was eluted quantitatively with an eluent concentration of a0.014 mol I-* HNO only. Eluent concentrations in excess of 0.05 moll-' HN03 were however found to yield incomplete retention of C1-.The latter ion was chosen as an indicator for possible 'bleed off since it has the lowest selectivity coefficient of all the retained anions of interest (Table 4).18 This coefficient is the degree of preference of an anion-exchange resin for a certain anionic species relative to another (conventionally Cl-). It shows no rigid correlation with a simple chemical or physical property of the anions. However a certain propor- tionality to the acid strength is observed. The elution parameters applied for the analysis of all authentic samples are listed in Table 5. Amount of the Eluent and Elution Rate For complete elution of cationic species (V3+ Cr3+ etc.) H3As03 and H2Se03 a minimum eluent volume of 60 80 Table 4 Selectivity coefficients for anions on Dowex-1 resinI8 Hydroxide Fluoride Acetate Dihydrogen phosphate Hydrogen carbonate Chloride Hydrogen sulfite Nitrate Hydrogen sulfate Perchlorate 0.09 0.09 0.17 0.25 0.32 I .oo 1.3 3.8 4.1 (32)* * Selectivity coefficient measured for Dowex-2 and not for Dowex- 1 however both resins are similar and the selectivity coefficients are comparable for other anions.I ' 2 Table 5 Elution parameters 0 200 400 600 800 1000 1200 1400 Time/s Resin Type Dowex- I X8 (mesh size 100-200) Counter ion NO3- Wet volume 40 ml Eluent Concentration 0.0014 rnol I-' HN03* Volume 100- 120 ml Elution rate 10 ml min-' Samples Volume less than 50 ml; maximum equivalents of adsorbable materials 10% of the theoretical column capacity; maximum HN03 concentration for com- plete separations 0.3 rnol I-' Fig.2 Elution diagrams of I Cr3+ and 2 H3As03 * 0.014 mol-I HN03 if quantitative elution of Se is required.1170 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY DECEMBER 1992 VOL. 7 Table 6 Analysis of a synthetic sample Recovery (O/o) after elution with 0.0014 rnol I-' HN03 on a NO3- Element (mlz value monitored) loaded column Vanadium (5 1) Chromium (53) Manganese (55) Iron (57) Cobalt (5 9) Nickel (60) Copper (65) Zinc (66) Arsenic (75) Selenium (77) Chlorine (35) Sulfur (48) 102.0 f 0.3* 103.0 ? 2.4 100.0 * 2.4 102.4+- 1.6 98.5 * 2.2 100.1 k 0.3 100.0 * 1.6 99.0 k 0.8 99.0 k 2.6 98.7k 1.1$ (0.1 t o . I * Uncertainties are expressed as standard errors of the mean. t Corrected for Arc interference.$ 0.014 rnol I-' HN03 used as eluent. Recovery (O/O) after elution with 0.014 rnol I-' HN03 on an acetate loaded column t 60 98.2 f 0. It 100.2 f 1 .o 102.3 f 1.8 99.1 f 2.4 100.3 f 0 . 6 99.9 -+ 1.7 102.1 k 1.6 98.2 * 1.9 < 30 t o . 1 t o . 1 Table 7 Determination of Cr and V in sewage Determined valuelpg 1-1 Element ICP-MS DCP-AES ETV-AAS Chromium 1 14.0 f 3.4* 115.7k2.7 112.2k3.4 Vanadium 62.5 -+ 2.6 64.3f 1.8 60.7 k 4.1 * Uncertainties are expressed as standard errors of the mean. Table 8 Determination of Cr in aqua regia soluble phase of Light Sandy Soil Determined valuelpg g-I Sample ICP-MS DCP-AES ETV-AAS I 84.6 83.7 87.9 2 87.2 85.2 87.7 3 81.0 85.0 83.3 4 85.1 88.0 87.0 5 80.0 84.7 79.1 Mean 83.6k 1.3* 85.32 k 0.72 85.0 & I .7 * Uncertainties are expressed as standard errors of the mean.and 100 ml respectively is required. Further research on the possibilities of on-line connection of the resin columns to the ICP-MS system to limit sample dilution is being undertaken. Elution rates of 10 ml min-I were used for off-line separations whereas for the qualitative on-line experiments the rate was decreased to 7 ml min-I. Analysis of Synthetic Samples After reduction 10 ml of a synthetic sample were applied and elution was effected with 120 ml of 0.0014 mol 1-1 HN03. The eluate was collected and Ge was added as an internal standard. Further dilution to 150 ml was accomplished with 0.14 mol 1-I HN03. Recoveries of all metals and anions added were measured by ICP-MS. This procedure was repeated twice and average recoveries and standard deviations were calculated.All recoveries except for Se were near 100%. Elution with 0,014 mol 1-' HN03 yielded quantitative recoveries for all the elements includ- ing Se. In both instances the concentrations of C1- and Sod2- in the eluate were below ICP-MS detection limits (0.1 and 0.3 mg I-' for C1 and S respectively). Chlorine was monitored at mlz=35 for S the SO signal at mlz=48 was used. Analytical blanks for the procedure were below the ICP-MS detection limit for all elements. In an additional experiment the column was loaded with acetate anions by means of a 2 moll-' NH,Ac solution and 0.014 mol 1 - I HN03 was used as the eluent. The main advantages of this modification are that the HN03 concen- tration of the samples and the eluent is less critical and that complete retention of phosphates is possible.Both of these advantages are owing to the low selectivity coefficient of acetate for Dowex- 1. The most serious drawback is the low recovery of Se and V (the latter owing to reduction problems) and the possible Arc interference on Cr. The latter procedure can however be used to separate Cu Zn other cationic species and As from C1 S and P in samples with HN03 or other strong acids in too high concentrations for the columns loaded with NO3-. The capabilities of this modification were tested by elution and analysis of syn- thetic samples only. Instead of 0.0014 mol 1-' HN03 (see above) 0.014 mol 1-' HN03 was used as the eluent. The results of the analysis of the synthetic samples for both procedures are listed in Table 6.Results and Discussion Analysis of Sewage After reduction 20 ml of sewage were treated as described above. Germanium was added as an internal standard to the eluate in which Cr and V were determined. As can be seen from Table 7 the results for Cr and V are in good agreement with independent analyses by other techniques (DCP-AES and ETV-AAS). Originally the interference by Cl on 51V T r and 53Cr was 250 30 and 350% of the net signals respectively. Analysis of Percolate Water Percolate water (25 nil) was treated after reduction. Three replicates were prepared and Ge was added as an internal standard to the eluates. Arsenic was determined and a mean value of 49.9 -t- 2.8 ,ug 1-L was obtained which is in excellent agreement with the accepted value (50.0 pg 1-I).Before separation an 'apparent' As concentration of 85 ,ug 1-l was found. Analysis of Light Sandy Soil For the determination of the aqua rcgia soluble content of (3- in Light Sandy Soil five different solutions and severalJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY DECEMBER 1992 VOL. 7 1171 Table 9 Determination of Cu and Zn in a human serum reference materialI7 Element Determined value/pg g-I Certified value/pg g-’ Copper 11.15 k0.5 I* 1 1 . 1 k0.4 Zinc 9.8 1 f 0.25 9.6 f 0.2 * Uncertainties are expressed as 95% confidence limits. blanks were treated 30 ml of each being applied to the column. After elution Ge was added to the eluates as an internal standard. Since V As and Se were of no interest no preliminary reduction was needed.Interference by Cl on 52Cr and 53Cr was about 30 and 3OO0h of the net signals respectively while after elution both isotopes yielded identical results. The determined contents for Cr are in good agreement with the results obtained by ETV-AAS and DCP-AES (Table 8). Analysis of Human Serum Copper and Zn both subject to interference from poly- atomic S species were determined in the human serum reference material. A digestion prior to elution was necessary to convert the organically bound S into S042-. After separation In was added to the eluate. Copper was determined with use of the T u isotope because 63Cu was still subject to interference from 40Ar23Na. Without separation the interference by S on the 65Cu isotope amounted to 25% of the net signal. The determina- tion of Zn in human serum is limited also since all isotopes are subject to interference from S (68Zn to a small extent only).After separation all the Zn isotopes yielded identical results. The contents determined are in good agreement with certified values for both elements and are listed in Table 9. Conclusions Spectral interference on V Cr Cu Zn As and Se caused by C1 and S in ICP-MS can be overcome by separation on a Dowex-1 column in the NO3- form. This allows the accurate determination of these elements together with other analytes in many environmental and biological samples. In most instances sample preparation can be limited to a reduction with SnC12 and analytical blanks of the procedure are below ICP-MS detection limits. Organi- cally bound C1 or S however necessitates acid digestion prior to the separation whereby the addition of mineral acids is limited as much as possible to avoid column overload.Sample volumes of up to 50 ml were used. A dilution of the sample by a factor of 3-10 is inherent to the separation method. Direct connection of the columns to the ICP-MS instrument could alleviate this drawback and will be the subject of further research. Grateful acknowledgement is made to Dr. C. Vandecasteele for his valuable scientific contribution. The authors are also grateful to Ing. R. Steegmans from LISEC (Limburgs Studiecentrum voor Toegepaste Ecologie) for providing sewage samples. References 1 Tan S. H. and Horlick G. Appl. Spectrosc. 1986 40 445. 2 Lyon T. D. B. Fell G. S. Hutton R. C. and Eaton A.N. J. Anal. At. Spectrom. 1988 3 265. 3 Vanhoe H. Vandecasteele C. Versieck J. and Dams R. Anal. Chem. 1989 61 149. 4 Plantz M. R. Fritz J. S. Smith F. G. and Houk R. S. Anal. Chem. 1989 61 149. 5 McLaren J. W. Mykytiuk A. P. Willie S. N. and Berman S. S. Anal. Chem. 1985 57 2907. 6 Sheppard B. S. Shen W.-I. Caruso J. A. Heitkemper D. T. and Fricke F. L. J. Anal. At. Spectrom. 1990 5 431. 7 Lyon T. D. B. Fell G. S. Hutton R. C. and Eaton A. N. J. Anal. At. Spectrom. 1988 3 60 1. 8 Whittaker P. G. Lind T. Williams J. G. and Gray A. L. Analyst 1989 114 675. 9 Branch S. Corns W. T. Ebdon L. Hill S. and O’Neill P. J. Anal. At. Spectrom. 1991 6 155. 10 Munro S. Ebdon L. and McWeeny D. J. J. Anal. .4t. Spectrom. 1986 1 2 1 1. 1 I Ridout P. S. Jones H. R. and Williams J. G. Analyst 1988 113 1383. 12 Montaser A. Chan S.-K. and Koppenaal D. W. .4nal. Chem. 1987 59 15 1. 13 Branch S. Ebdon L. Ford M. Foulkes M. and O’Neill P. J. Anal. At. Spectrom. 1991 6 151. 14 Evans E. H. and Ebdon L. J. Anal. At. Spectrom. 1989 4 299. I5 Faris J. P. and Buchanan R. F. Anal. Chem. I964,36 I 157. 16 Caletka R. Hausbeck R. and Krivan V. J. Radioanal. Nucl Chem. 1990. 142 383. 17 Versieck J. Vanballenberghe L. De Kesel A. Hoste J. Wallaeys B. Vandenhaute J. Baeck N. and Sunderman F. W. Anal. Chim. Acta 1988 204 63. 18 Peterson S. Ann. N. Y. Acad. Sci. 1954 57 144. Paper 2/02181A Received April 28 1992 Accepted July 7 1992