JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY MARCH 1994 VOL. 9 177 Spectral I Materials nterferences Encountered in the Analysis of Biological by lnductively Coupled Plasma Mass Spectrometry* Hans Vanhoe Jan Goossens Luc Moens and Richard Dams Laboratory of Analytical Chemistry University of Ghent Institute for Nuclear Sciences Proeftuinstraat 86 B-9000 Ghent Belgium In order to evaluate the potential of inductively coupled plasma mass spectrometry for the analysis of biological materials a systematic study on the occurrence of spectral interferences was carried out. All polyatomic ions originating from the matrix elements Ca CI P K C Na and S that give rise to spectral overlap with the analyte elements were examined systematically. For these matrix elements which are representative of most biological materials a list of all the polyatomic ions which can be seen to interfere up to a matrix element concentration of 1 g I-' is given.In addition for each spectral interference the corresponding apparent analyte concentration was calculated for different matrix concentrations. In this way an interference factor (IF) could be calculated for each spectral overlap so that the extent and importance of each potential spectral interference can be estimated. This factor is defined as follows IF = 1 O6 x apparent concentration (analyte element)/concentration (matrix element). A study on the formation and short- and long-term stability of polyatomic ions was also carried out to evaluate the use of some simple correction procedures to overcome spectral overlaps.Finally a review of the correction procedures applied to biological materials is given together with some examples. Keywords lnductively coupled plasma mass spectrometry; spectral interference; biological material Although for trace element analysis inductively coupled plasma mass spectrometry (ICP-MS) has several advantages over other techniques such as a quasi-simultaneous multi-element capability and excellent detection limits in a large number of applications it suffers from both spectroscopic and non- spectroscopic interferences. The latter group of interferences can be overcome by internal standardization standard additions isotope dilution or sample preparation (e.g. separ- ation) techniques. This subject has recently been reviewed by Evans and Giglio.' Spectroscopic interferences on the other hand are caused by atomic or polyatomic ions that have the same nominal mass as the analyte of interest.Such interferences can be divided into two categories depending on the origin of the interference. Firstly there are isobaric interferences caused by the overlap of isotopes of different elements. They are easy to predict and can be overcome by use of alternative isotopes of the element of interest. However when an element is introduced into a plasma the monoatomic singly charged analyte ion (M') is not the only species observed in the final mass spectrum. Elements with a low second ionization energy will be partly doubly ionized (M2+) whereas for elements that have a high MO bond strength oxide (MO') and hydroxide ions (MOH') are formed.The intensities of these species can be reduced to 1-2% by adjusting instrumental settings for MO' M+ even for the most refractory elements. In addition to these species polyatomic ions originating from the plasma (Ar) from the matrix (0 H and often C N P S and C1) and from the air surrounding the plasma (C N and 0) exist. There are probably three places where these polyatomic ions are formed in the plasma itself (especially for refractory elements that form oxide ions) in the boundary layer that is formed in the vicinity of the sampling cone surface and finally in the expansion region. Vaughan and Horlick2 have given an over- view of all spectral interferences due to the formation of metal monoxide ions (MO') hydroxide ions (MOH') and doubly charged ions (M2+).When analysing biological materials the number of potential spectral overlaps is dramatically increased by the acids (HC1 HClO and H2S04) used for the digestion of the sample and/or by the large variety of matrix elements present. Information * Presented at the XXVlII Colloquium Spectroscopicurn Inter- nationale (CSI) York UK June 29-July 4 1993. on these spectral interferences is therefore necessary to obtain accurate results. Tan and Horlick3 have extensively described the background mass spectra for water 5% solutions of HNO HC1 and H,SO,. Since no quantitative data were given this paper does not allow an estimation of to what extent the polyatomic species mentioned can contribute to the analyte signal intensity. In addition when biological materials are analysed other matrix elements such as C and Na also give rise to spectral overlap. Mulligan et aL4 have reported potential interferences on trace metals in the analysis of urine.Both qualitative and quantitative data were given. Evans and Giglio' gave an overview of all spectroscopic interferences (including those for biological materials) which have been described in the literature. In the present work the polyatomic ions related to the following matrix elements which are representative of most biological materials were examined Ca C1 P K C Na and S. Since both qualitative and quantitative results were obtained in most circumstances it is possible to estimate the magnitude and hence the importance of the spectral overlap when biological samples with a known composition are ana- lysed.In addition the formation and stability of some poly- atomic ions were studied in more detail so that the use of some simple correction procedures to overcome spectral over- laps could be evaluated. Experimental Instrumentation The ICP mass spectrometer used is a VG PlasmaQuad PQ1 (VG Elemental) equipped with a Gilson Minipuls-2 peristaltic pump a Meinhard concentric glass nebulizer (Type TR-30-A3) a Scott-type double pass spray chamber with surrounding liquid jacket and a Fassel-type torch. Details of the operating conditions are summarized in Table 1. Reagents and Standards High-purity water was obtained with a Millipore Milli-Q water purification system (resistivity of 18 Mi2 cm). Concentrated HN03 ( 14 moll- ') was purified by sub-boiling distillation in a quartz still using analytical-reagent grade HNO (Union Chemique Belge) as feedstock.Pre-cleaned polyethylene Cali- brated flasks and glass pipettes were used throughout.178 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY MARCH 1994 VOL. 9 Table 1 Operating conditions for ICP-MS Stage Parameter Plasma R.f. power Forward Reflected Gas flow Plasma Intermediate Aerosol Peristaltic pump Nebulizer Spray chamber Ion sampling Sampling cone Skimmer cone Sampling depth Vacuum Expansion stage Intermediate stage Analyser stage 1.35 kW <10 w Condition 13 1 min-' 11 min-' 0.725 1 min-' Minipuls 2 (Gilson) pumped at 0.9 ml min- Meinhard Tr-30-A3 concentric glass nebuli Double-pass Scott type water-cooled (10 "C) Nickel 1.0 mm orifice Nickel 0.75 mm orifice 10 mm (from load coil) 2.4 mbar* 1.0 x lop4 mbar 4.0 x mbar - 1 zer * 1 bar= 1 x lo5 Pa.For qualitative and quantitative study of the spectral inter- ferences for each matrix element investigated (Ca C1 P K C Na and S) ten solutions were prepared with concentrations of 0.05 0.1 0.5 1 5 10 50 100 500 and lOOOmgl-'. Each solution also contained In as an internal standard (concen- tration of 100 pg 1-'). Stock solutions with a concentration of 10 g 1-' of the matrix element were prepared by dissolving an appropriate amount of the solid salt or by diluting a concen- trated solution Ca(N03),.4H20 (UCB pro analysi) sub- boiled HC1 (10 mol 1-') H,P04 (Merck pro analysi) KNO (UCB pro analysi) acetic acid (Merck Suprapur) NaNO (Carlo Erba pro analysi) and sub-boiled H2S04 (18 moll- ').All solutions were prepared with 0.14moll-' HNO as the solvent and the diluent. In order to calculate the apparent analyte concentrations four multi-element standard solutions with a concentration of 100 pg 1-' for each element were used (external calibration) the first solution contained V Cr Mn Fe Co Ni Cu Zn and In (internal standard); the second contained Ti Mo and In (internal standard); the third contained Ge As Se and In (internal standard); and the fourth Ca Sc Ga Br and In (internal standard). All these solutions were prepared from commercially available 1 g 1-l atomic absorption spectrometry standard solutions and stored in polyethylene flasks. The blank solution used consisted of 100 pg 1-l In (internal standard) in 0.14 moll-' HNO,.Use of an Internal Standard In order to correct for the matrix-induced ion signal fluctu- ations and instrumental drift an internal standard was used. Although Vanhaecke et aL5 suggested that an internal standard with a mass close to that of the analyte is to be preferred for all elements In of mass 115 u was used [1'51n(95.7%)] because other internal standards (with a mass below 80u) could give rise to additional spectral interferences. Measuring Procedure Optimization of the ion signal Before each measuring procedure the aerosol gas flow rate and the ion lens settings which are generally accepted to have a strong effect on the signal intensity were optimized in order to give maximum signal intensity for '%+ . The optimum aerosol gas flow rate was found to fluctuate around the value of 0.725 1 min-' mentioned in Table 1.Other parameters including r.f. power intermediate and plasma gas flow rates and temperature of the spray chamber were not changed during the experiments described in the present work. Measuring conditions The qualitative and quantitative data for the spectral inter- ferences were obtained using the scanning conditions summar- ized in Table 2. The range between m/z 42 and 126 was scanned 100 times with a dwell time of 320 ps per channel. In this way one measurement required approximately 1 min. Five replicate measurements were made for each solution. In order to reduce memory effects the following measuring sequence was used firstly a blank was measured then the ten matrix element solutions with increasing concentration and finally the four standard solutions were measured.After meas- uring each solution the sample introduction system was rinsed for 2 min with 0.14 mol 1-' HNO,. Calculations For each solution the signal (peak area integrated over an m/z value of 0.8 around the peak maximum) of each nuclide of interest was normalized to the signal of 'IsIn (internal standard). The mean and standard deviation of the five resulting normalized. signals of each solution were calculated. External calibration was used to calculate the apparent concen- tration of the analyte elements. In addition for each spectral interference an interference factor (IF) was calculated as follows apparent concentration (analyte element) concentration (matrix element) IF= lo6 - with the matrix element and the analyte element concentrations expressed in the same mass per volume unit.In this way it is possible to estimate the extent and importance of each spec- tral overlap. Results and Discussion Spectral Interferences Qualitative and Quantitative Data The interference factors were calculated based on the results that were obtained for the range of matrix element solutions containing various concentrations. For each matrix element concentration for a given spectral interference an interference factor was calculated based on the formula given under Calculations. A mean of all interference factors obtained was then calculated together with a standard deviation. Normally all results were taken into account except for those which have a relatively high uncertainty [percentage relative standard deviation (RSD) above 20%].In cases where only one or two results are available an indicative value is given. The relatively small uncertainty in the interference factors obtained for most of the spectral interferences indicates a good and in most cases linear correlation between the matrix element concentration and the apparent analyte concentration. It is an exception for the uncertainty obtained to be high whereby the spectra1 overlap at low matrix element concentrations is underestimated. Table 2 Scanning conditions used to obtain qualitative and quantitat- ive data for spectral interferences Mass range No. of channels Dwell time per channel No. of sweeps Measuring time 42-126 u 2048 320 ps 100 = 1 minJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY MARCH 1994 VOL.9 179 Calcium An overview of the spectral interferences arising from Ca is given in Table 3. In addition to spectral interferences due to the formation of polyatomic ions some Ca isotopes also give rise to an isobaric overlap 46Ca (0.0040/,) coincides with 46Ti (8.0%) and 48Ca (0.19'%0) with 48Ti (73.8%). The interference factors for 46Ti and 48Ti are respectively 543 and 4010. These overlaps reduce the result for determination of Ti in a Ca-rich matrix. Some conclusions can be made from Table 3. Firstly the formation of CaO' and CaOH' is dominant. Since Ca has six stable isotopes with mass numbers between 40 and 48 the corresponding spectral interferences are spread out between m/z 56 and 65.The Ca polyatomic ions mainly interfere with the determination of Fe. The main Fe isotope 56Fe (91.7%) as well as 57Fe (2.20%) are subject to interference to a large extent interference factors of 650 and 3710 were observed respectively. The problems experienced with calcium oxide and hydroxide interferences in the analysis of human serum and urine have been given by Vanhoe et aL6 and Vaughan and Templeton7 respectively. Other elements subject to interference are Ni (58Ni and 60Ni) and to a lesser extent Co ("Co) and Zn (64Zn). Secondly calculations show that the ratios CaOH+ CaO+ and CaO+:Ca+ are 0.1 and 7.4 x respectively. The contribution of Ca160H+ to the total signal (Ca160H+ and Cai70+) was found to be more than 99%. Thirdly the formation of minimum amounts of 40Cai60160 + was only observed at high Ca concentrations (500 mg 1-' and above).Finally it should be mentioned that neither ArCa' nor Ca2+ could be observed. The detection of such species is however hampered by the presence of a background peak (Ar2+). Chlorine The spectral interferences arising from C1 are summarized in Table 4. The important formation of C10' and ClOH' should first be noted. Since C1 has two isotopes (mass numbers of 35 and 37 respectively) these polyatomic species can be found in the mass spectrum between m/z 51 and 54. The occurrence of such polyatomic ions substantially hampers the determination of two elements in a Cl matrix V and Cr. Both "V (99.7%) which is almost monoisotopic and ',Cr (9.5%) suffer from interferences from 35C1160 + and 37Cli60+ respectively. The high interference factors found 29 and 88 respectively indicate the importance of these interferences.In addition they are observed at a C1 concentration of only 5 mg 1-'. The combi- nation of Cl with less abundant isotopes of 0 170 and l80 contributes to a lesser extent to an increase in the blank signal. These combinations only play a significant role at Cl concen- trations of 500mg1-' or higher. The ratio CIOHf:CIOf is about 3.2 x whereas the contribution of C1160H+ to the total signal (Cli60H+ and Cl170+) was established to be Secondly the formation of 35Cli60160+ was observed at a C1 concentration of 1 g 1- '. The size of this spectral interference is however very small analogous to that of CaO,'. Thirdly at C1 concentrations of 500 mg 1-' and higher a signal significantly different from the background signal was observed at m/z 49. This is attributed to the formation of 35C114N+ [spectral overlap with 49Ti (5.5%)] as suggested by Tan and H ~ r l i c k .~ The importance of this interference is limited because several other isotopes of Ti are available and because the size of the signal is very small. The ratio for ClN+ C10+ is about 2.5 x which is initially in contradiction with the ratio of the bond strengths (333.9 kJ mol-' 272 kJ mol-' = 1.23).8 However the concentration of N f and 0' in the plasma has to be taken into account when such polyatomic ions are formed. Gray' reported a concentration of 3 x 1014 cm-3 for 0' and only 2.5 x 10" cm-3 for N + (for 0.14 mol I-' HNO,).Fourthly an important CI interference problem is caused by the formation of ArCl' (ArCl' CIOf = 8.0 x This mainly hampers the determination of As in that 75As (100%) (monoisotopic) suffers from an interference from 40Ar35C1 +. In addition such polyatomic ions are observed at C1 concen- trations of 5-10mg1-' and higher. For the determination of Se the interference of 77Se (7.6%) with 40Ar37C1f can be %99%. Table 3 Calcium spectral interferences. Stable isotopes of Ca 40Ca (96.6%) 42Ca (0.65%) 43Ca (0.14%) 44Ca (2.08%) 46Ca (0.004%) and 48Ca (0.19%). The most important polyatomic ions are printed in italics. Values given are apparent analyte concentration (Fg 1-') &standard deviation of the mean n = 5 Ca concentration/mg I-' Analyte 56Fe (91.7%) 57Fe (2.20%) 58Ni (68.3%) 60Ni (26.1 Yo) 64Zn (48.6%) 72Ge (27.4%) 59c0 (100%) 5 10 50 100 500 <2.2 40f14 177f13 229f 15 372 f 22 - < 15 137f4 314f 11 1861 f 92 - <0.30 0.39f0.12 0.78k0.17 3.7 & 0.5 - - - < 0.04 0.24 & 0.09 - - <0.10 0.74+0.44 11.35 1.0 - < 0.47 - - - <0.10 0.35 2 0.15 - - - 1000 560 f 30 4270 f 140 7.3 & 0.6 1.1 &0.1 28.5 k 1.1 0.84 & 0.18 1.5f0.1 I F 650f 130 3710f 570 7.5 f 0.3 1 26$4 0.8 1.5 Table 4 Chlorine spectral interferences.Stable isotopes of C1 35Cl (75.8%) and 37Cl (24.2%). The most important polyatomic ions are printed in italics. Values given are apparent analyte concentration (pg 1-') fstandard deviation of the mean n=5 Analyte 49Ti (5.5%) 52Cr (83.8%) 53Cr (9.5%) 54Fe (5.8%) 55Mn (100%) 67Zn (4.1 %) 70Ge (20.5%) 75As (100%) 77Se (7.6%) 51v (99.7%) Interferen t 35C1'4N 35C1160 37C1'4N 35C1160H 35C1170 37C1160H 37C1170 3 7 ~ 1 1 6 0 3 7 ~ ~ 0 35~1160160 35C135C1 40Ar35 CI 40Ar37C1 C1 concentration/mg 1-' 1 5 - - < 0.07 0.17 & 0.02 < 0.26 0.40 f 0.04 - - - - - - < 0.08 0.14 f 0.03 - - 10 50 - <0.71 0.31 k0.02 1.2f0.1 < 0.03 0.96 f 0.04 4.2 f 0.2 - - - 0.19 0.02 0.76f 0.14 < 1.3 3.8 f0.5 100 1.4f.0.2 2.8k0.1 0.11 f.0.04 7.8 f 0.5 < 5.5 < 0.02 <0.10 1.6 f 0.2 7.6 f 2.5 - 500 7.5 5 0.4 15.5f.0.5 0.5f0.1 44.5 f 0.6 6.6 f 2.5 0.06 f 0.02 < 0.43 0.17 f 0.02 7.6 k 0.2 39.9 f 1.7 ~ 1000 15.0 + 0.7 33.8 5 2.6 1.2 * 0.1 99f 11 16.1 +4.7 0.07 & 0.03 3.0 + 0.4 0.30 + 0.09 16.0k 1.1 86+ 14 ~ I F 14.7 f 0.6 29k4 1.1 f O .1 8 8 2 9 16 0.07 3 0.3 15.6 f 0.5 7 9 f 5180 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY MARCH 1994 VOL.9 overcome by the use of other isotopes. Some of these (76Se 78Se and are however subject to interference by Ar,'. Problems encountered with the determination of As and Se have been reported for serum,1s12 ~ r i n e ~ ? ~ '-19 protein solu- t i o n ~ ' ~ and other biological materials such as bovine l i ~ e r . ' ~ ~ ~ Finally it can be noted that at very high Cl concentrations (> 500 mg 1-') the formation of Cl,' was observed. Phosphorus The spectral interferences arising from P are presented in Table 5. Firstly the formation of PO' and POH' is important. Since P is monoisotopic ("P) these polyatomic ions are situated at between m/z 47 and 49 interfering with three isotopes of Ti.These spectral interferences however do not hamper the determination of Ti in a P matrix because 46Ti (8.O%) which is free from interferences by P can be used. The ratio of POH' :PO' is about 2.8 x In addition the contribution of 31P180' to the total signal at m/z 49 is about 80%. The remaining part is due to the formation of 31P160H2+. A ratio for POH2' PO' of 5.1 x Secondly there is the important formation of PO2' at P concentrations of 5 mg 1-' and higher. This polyatomic ion influences the determination of Cu and to a lesser extent that of Zn. A ratio for PO,' :PO' of 8.3 x lo- was observed. Calculations show that the signal at m/z 64 can be attributed fully to the formation of 31P160160H' and not to that of 31P160170f (about 0.2%).The formation of a tetraatomic ion was also observed at m/z 79 i.e. PO,'. The intensity of the signal however is very small (PO,' PO' = 3.6 x Thirdly at high P concentrations (> 50 mg 1-I) the forma- tion of ArP+ was observed. This polyatomic ion influences the determination of Ga but its importance is rather limited ( I F = 3.5 and ArP' PO' = 5.5 x lop3). Mulligan et ~ 1 . ~ have reported on the spectral interferences of P (PO PO3 and H2P03) in the analysis of urine. was found. Potassium The spectral interferences arising from K are given in Table 6. Although K has three stable isotopes 39K (93.3%) 40K (0.01%) and 41K (6.7%) a significant spectral interference due to KOf was only observed at m/z 55. The formation of KO' is rather small in comparison with other XO' polyatomic ions owing to a low XO bond strength:8 276 kJ mol-' in comparison with 596.6 kJ mol-' for PO.In addition the detection limits for Fe at m/z 56 and 57 are relatively high 2.2 and 15 pg1-' respectively so that the formation of KOH' was not observed at all. At K concentrations of lOOmgl-' and higher the formation of ArK' was noticed at m/z 79 and 81 (spectral interferences with the two isotopes of Br) analogous to that of ArC1' and of ArP'. In addition to peaks due to the expected formation of KO+ and ArK' the mass spectrum of K also contained some other peaks. These are presented in Fig. 1 which gives the spectrum between m/z 90 and 100 for a 1 g I-' solution of K. As can be seen peaks are found at m/z 94 95 96 and 97 Taking into account the relative intensities of the peaks they can only be attributed to the formation of K20' and of ArKO'.Although these polyatomic ions have not been described in the literature other combinations seem most unlikely. A similar formation of Na20' and of ArNaO' was observed with solutions of Na (see later). Finally it should be pointed out that an experimental ratio for 39K39K'60+ 40Ar39K160' of 0.22 is in contrast with a ratio for 39K160f 40Ar39K' of about 1. As has been described by Lam and Horlick,20 such polyatomic ions are probably formed at the interface by collision of monatomic and/or polyatomic ions. In this way K,O' can be formed by collision of K with KO (the formation of K2+ was not observed the detection of such a species is however hampered by the presence of Ar2') whereas collisions of Ar with KO and of ArK with 0 can lead to the formation of ArKOf.Mulligan et aL4 have reported spectral interferences of K (KO and ArK) in the analysis of urine. Carbon The spectral interferences arising from C are summarized in Table 7. Two groups of spectral interferences were observed. The first group is due to the formation of polyatomic ions consisting of C and 0. Although not given in Table 7 CO' and COH' are predominantly formed at m/z values between 28 and 32. These interfere with all isotopes of Si. Besides CO' CO,' is also formed (at m/z 44 and 45) and to a lesser extent CO,' (m/z 60). The following ratios were observed 0.14 and 8.1 x for CO,' CO' and CO,' CO' respectively. In addition it can be calculated that the contribution of 12C160160H' to the total signal at m/z 45 is about 75% and that the ratio for C02H+ :CO,+ is 6.1 x The second group is due to the formation of polyatomic ions consisting of Ar and C. These polyatomic ions mainly affect the determination of Cr because two isotopes of Cr ',Cr (83.8%) and 53Cr (9.5%) are subject to interference by 40Ar12C+ and 40Ar1'3C' respectively at C concentrations of 100 mg 1-' and higher.The spectral overlap is however small ( I F values of 2.4 and 2.1 respectively). Nevertheless problems experienced with the determination of Cr in plasma protein solutions have been reported by Lyon et a[.' Carbon spectral interferences are much more severe when organic solutions are measured. An overview of the polyatomic ions obtained has been given by Hutton.21 Sodium The spectral interferences arising from Na are summarized in Table 8.The vicinity of the intense peak of 40Ar' hampered the detection of 23Na160f and this ion was therefore not investigated further. Firstly in contrast with other matrix elements such as P and C the formation of NaO,' was not observed probably because of the relatively low bond strength? 256.5 kJ rriol-' for NaO 596.6 kJ mol-I for PO and 1076.5 kJ mol-' for CO. Secondly there is the important formation of Na,' at Na concentrations of 50 mg 1-' and higher. The signal at m/z 47 can be attributed to the formation of Na,H+. A ratio for Na,H' Na,' of 9.8 x lo- was observed. Thirdly there is the remarkable formation of 23Na23Na160 ' with a ratio for NazO+ Na2' of 4.5. Fourthly the formation of 40Ar23Na+ was observed which affects the determination of Cu.A ratio for A.rNa' Na20' of about 1 was noted. Finally at m/z '79 the formation of 40Ar23Na160+ was observed. The intensity is however relatively small (ArNaO' ArNa' =; 5.8 x lo- and ArNaO+ NaNaO' = 4.9 x Lyon and Fell2 studied the spectral interferences of Na20 and of ArNa on Cu in serum in detail. It should also be noted that for the calculation of the interference factors for Na,' and Na,Of the square of the Na concentration must be used [IF = apparent analyte concentration x 106/(Na con- ~entration)~] and that for Na,H' and ArNa+ no relationship could be found. Sulfur The spectral interferences arising from S are given in Table 9. The formation of SN' mainly of mass 46 was not observed although this species has been suggested by Tan and H ~ r l i c k .~ It is clear from Table 9 that the combination of S with 0 gives rise to most of the spectral interferences. A first group of interferences is due to the formation of SOf and SOH'. Since S has four stable isotopes with mass number between 32 and 36 these polyatomic ions can be found between m/z 48 andTable 5 f standard deviation of the mean n = 5 Phosphorus spectral interferences. Stable isotope of phosphorus 31P (100%). The most important polyatomic ions are printed in italics. Values given are apparent analyte concentration (pg 1-I) P concentration/mg I-' Anal yte 47Ti (7.3%) 48Ti (73.8%) 49Ti (5.5%) 63Cu (69.2%) 64Zn (48.6%) Interferent 31~160 3 1 ~ 1 6 0 ~ 3 1 ~ 1 7 0 31~180 3 i p l 6 0 ~ ~ 31p160160 3 1 p 1 6 0 1 6 0 ~ 31p160170 31p160180 2 40Ar31 P 31p160160160 0.1 0.5 1 5 - - < 0.04 0.13f0.02 - - < 0.02 0.59 f 0.02 <0.61 4.3 f 0.5 8.2 0.4 41.9 f 0.3 - - - - 10 78.2 f 1.8 0.24 f 0.01 <0.35 0.90 & 0.04 50 366 & 8 0.99 k 0.05 0.86f0.16 4.1 f0.2 100 659 & 24 1.7 & 0.1 1.9f0.2 7.5 f 0.5 500 3060 f 160 7.7 f 0.5 8.0k 1.4 30.8 f 0.4 1000 6410 f 380 14.9 f 1.2 17.0f 1.5 59.6k2.5 I F 6850f 700 17f2 17_+ 1 65f8 0.46 f 0.04 < 0.04 - - 1.4f0.2 < 0.01 0.17f0.02 < 0.3 1 2.8 k 0.3 0.13 f 0.03 0.38 & 0.06 0.93 f 0.12 11.0f0.4 0.94 f 0.17 1.7k0.2 4.9 _+ 0.4 23.4f1.0 1.73 f 0.04 3.4 f 0.3 9.7 f0.3 2523 1.6 k0.3 3.5 * 0.2 9.6 f 0.3 65Cu (30.8%) 71Ga (39.9%) 79Br (50.7%) Table 9 concentration (pg 1-I) fstandard deviation of the mean n=5 Sulfur spectral interferences.Stable isotopes of S 32S (95.0%) 33S (0.75%) 34S (4.2%) and 36S (0.02%). The most important polyatomic ions are printed in italics. Values given are apparent analyte 0.05 0.1 0.5 1 0.27k0.07 0.44k0.08 0.98f0.05 1.5f0.1 - < 1.7 - - < 0.43 4.1 f0.3 - - 5 5.7f0.1 3.2 f 0.3 5.0 f 0.3 10 10.8 * 0.1 4.5 f 1.0 8.7 & 1.9 7.8 f 0.2 1.8fO.1 - 50 50.2 _+ 0.7 10.3 _+ 0.6 33.5 & 2.2 100 90.2 f 1.6 15.5 k0.6 57.6 f 1.7 < 0.02 61.2 f 0.4 8.6 0.4 500 445 f 9 1000 842 _+ 8 144f3 511f8 0.30 f 0.08 500f 5 64.8 f 2.5 I F 980f 120 150+6 580 k 70 0.3 6302 110 80f 15 Analyte 48Ti (73.8%) 49Ti (5.5%) 50Ti (5.4%) 52Cr (83.8%) 64Zn (48.6%) "Zn (27.9%) 6 5 C ~ (30.8%) F W W "P 75.9 k 1.4 272 f 2 0.14f0.03 275 f 8 36.3 f 0.5 0.33f0.08 0.7010.15 1.4f0.3 2.1f0.2 - - - < 0.09 4.5 f 0.4 1.2f0.1 34.2 f 0.8 4.9 T 0.3 1.3 f 0.2 - 1.7 f 0.3 < 3.5 - 3.9 & 0.2 5.5 f 1.5 < 0.69 < 0.5 1 <0.36 < 1.6 6.8 f 0.2 6.7 f 0.9 2.0 _+ 0.6 0.69 f 0.05 1.6 f 0.3 5.3 f 1.0 24.7 * 0.7 17.6 f 3.2 2.5f0.1 1.7 & 0.4 8.1 f0.2 9.8f4.0 44.0 _+ 0.4 23.4 _+ 2.0 3.1 f0.4 2.1 k0.2 15.4 f 0.3 12.6 +_ 2.7 60f 16 29f8 4.1 f 1.3 3fl 15.9 f 0.4 16+5 67Zn (4.1 YO) 68Zn (18.8%) 72Ge (27.4%) 'lBr (49.3%) 82Se (9.2%)182 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY MARCH 1994 VOL.9 Table 6 are printed in italics. Values given are apparent analyte concentration (pg 1 - I ) f standard deviation of the mean n = 5 Potassium spectral interferences. Stable isotopes of K 39K (93.30/b) 40K (0.01%) and 41K (6.7%).The most important polyatomic ions K concentration/mg 1-' Anal yte "Mn (100%) 79Br (50.7%) 81Br (49.3%) 95Mo (15.9%) 96Mo (16.7%) 9 4 ~ 0 (9.3%) 9 7 ~ 0 (9.6%) In terferen t 40Ar39K 40Ar41K 3 9 ~ 1 6 0 3 9 ~ 3 9 ~ 1 6 0 4 0 ~ ~ 3 9 ~ 1 6 0 3 9 ~ 4 1 ~ 1 6 0 4 0 ~ ~ 4 1 ~ 1 6 0 10 50 100 < 0.08 0.1 1 + 0.02 0.29 f 0.04 - < 0.45 0.57 f 0.13 - < 1.3 - - <0.11 - < 0.04 0.17 f 0.05 - - < 0.05 - < 0.02 0.08 f 0.03 - 500 2.9 & 0.2 17.2 f 4.1 3.2 f 0.5 1.30f0.13 9.0 f 1.6 0.32 f 0.08 2.4 f 0.4 1000 5.7 * 1.1 59518 9.0f 1.0 4.0 rt 1.2 38 k 14 0.99 f 0.34 9.5 & 3.5 I F 5.8f0.1 47k17 9 4 38 1 10 500 I 400 - - a C I u a a 300 - L v) 200 - +- =I 0 u 100 - 90 91 92 93 94 95 96 97 98 99 100 m/z Fig. 1 Mass spectrum between 90 and 100 m/z of a 1 g 1-1 K solution 53.This causes a dramatic spectral overlap with three isotopes of Ti including the most abundant isotope 48Ti (73.8%). The interference factors vary from 150 to 980. The important formation of SO' can be explained by the relatively high bond strength' (521.7 kJ mol-I) which is comparable to that of P (596.6 kJ mol-l). A ratio for SOH' :SO' of 5.6 x was found. A second group contains the polyatomic ions SO,' and S02H' which can be found at between m/z 64 and 69. They mainly hamper the determination of Zn because almost all Zn isotopes are subject to interference. In addition the intensity of SO,' is remarkable a ratio for SO2' SO' of 0.31 was noted. This is higher than for similar XO,':XO+ ratios. Examples are ratios of 8.3 x for PO2+ PO' and of 0.14 for CO,' CO'.Although Tan and Horlick suggested that in addition to SO,' S2' can be formed it can be seen from the present data that a significant contribution due to the formation of such species is dubious. Up to a S concentration of 1 g I-' a linear response between the S concentration and the signal at m/z 64 (64Zn s9Co) was obtained indicating that SO2' is formed almost exclusively. A third group of polyatomic ions contains SO,' and SO,Hf. The most important polyatomic ion of this group namely 32S160160160' could not be detected because of an intense spectral overlap with 40Ar40Ar'. Less intense peaks were seen at m/z 81 and 82. For SO,' :SO+ a value of 1.9 x lop2 was found. Finally at S concentrations of 100 mg 1-1 and higher the formation of ArS' was also observed. The intensity of this signal is however small (ArS' SO' = 8.8 x Spectral interferences due to S (interfering with Cu Zn and Br) have been reported for human ~ e r u m ~ ~ and urine.4 Formation and Stability of Polyatomic Ions A study of the formation and short- and long-term stability of polyatomic ions was carried out in order to evaluate the use of some simple correction procedures to overcome spectral overlaps.The signals for s9Co+ (as a reference) CIO' SOz' and ArCl' m/z 51 64 and 75 respectively as a function of time Table7 Carbon spectral interferences. Stable isotopes of C "C (98.8%') and I3C (1.1Y0). The most important polyatomic ions are printed in italics. Values given are apparent analyte concentration (pg 1-l) f standard deviation of the mean n = 5 C concentration/mg I-' Analyte 44Ca (2.08%) 48Ti (73.8%) 49Ti (5.5%) "Cr (83.8%) 53Cr (9.5%) 6oNi (26.1 %) 45sc (100%) Interferent 12~160160 1 2 ~ 1 6 0 1 6 0 ~ 1 3 ~ 1 6 0 1 6 0 36Ar1 ' C 36 Ar 13C 40Ar12C 40Ar'3C 1 2 ~ 1 6 ~ 1 6 ~ 1 6 ~ 10 50 100 500 < 14 43.8 +:!.7 96.6-+_ 5.8 501 f 19 < 0.06 0.18 + 0.01 0.29 2 0.07 1.0 _+ 0.1 - - < 0.04 0.10f0.04 - - <0.35 1.7f0.5 < 0.02 0.13 +0.01 0.28 f 0.09 1.1 f0.2 - <0.1.5 0.66 0.07 1.2f0.3 - <0.10 - - 1000 985 k 39 2.0k0.1 0.19 & 0.02 2.4 0.3 2.2 * 0.2 1.7k0.3 0.90 k 0.15 I F 957 f 56 2.6 k 0.8 0.2 3 2.4 f 0.3 2.1 f 0.5 0.9 Table 8 Sodium spectral interferences. Stable isotope of Na 23Na (100%).The most important polyatomic ions are printed in italics. Values given are analyte concentration (pg 1-I) &standard deviation of the mean n = 5 Na concentration/mg 1-' Analyte Interferen t 10 50 100 500 1000 I F 46Ti (8.0%) 23 Nuz3 Nu <0.13 1.1 k0.3 6.7 f 2.3 189 k 31 589f81 610f 140" 47Ti (7.3%) 23 Nuz3 Nu H - <0.10 0.89k0.17 3.6k0.1 4.9 f 0.5 62Ni (3.59%) 23Na23Nu160 < 1.1 1.8 f 1.1 18.2 f 7.0 609 k 80 1890f270 2050 k 340* 2.7 f 1.1 92+ 12 281 + 30 63Cu (69.2%) 40Ar23Nu <0.16 ?3r- (50.7%) 40Ar23Na160 - - < 0.45 1.7 k0.4 5.9 +_ 1.6 6 - * (Concentration matrix element)' x x interference factor ( I F ) = apparent analyte concentration.JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY MARCH 1994 VOL.9 183 (the first 2 min after the start of aspiration of the corresponding solution) are given in Fig. 2. A first significant signal is obtained for both 5 9 C ~ f and the polyatomic ions after 40-50s.The signals due to the polyatomic ions however are less stable and fluctuate to a larger extent the RSD calculated between the first and the second minute amounted to 0.86% for 59C0+ 1.1% for ClO' 3.1% for SO,' and 1.9% for ArCl'. The behaviour described can be extrapolated to almost all poly- atomic ions (the exceptions will be described further). The larger fluctuation on the signals from polyatomic ions indicates that temperature fluctuations in the plasma changing sampling conditions etc. exert a different influence on the formation of monatomic and polyatomic ions. The behaviour of some polyatomic ions is different from that described in Fig. 2. This can be seen in Fig.3 where the signals for 59C0+ (as a reference) Na,' Na,O' and ArNa+ (m/z 46 62 and 63 respectively) as a function of time (the first 2 min after the start of aspiration of the corresponding solution) are presented. In contrast to most of the polyatomic ions no constant and stable signal is obtained for polyatomic ions of Na after the observation of the first significant signal. In addition at that moment the signal increases almost linearly for at least 1 min. It is remarkable that this behaviour was only observed for polyatomic ions containing Na or K (similar observations were made for ArK+ at m/z 79). Both matrix elements have a low ionization energy 5.14 and 4.34eV respectively. The behaviour of ArNa+ (m/z 63) has been confirmed by Lyon and They proposed the condensation of Na on the surface of the interface with an accumulation of deposit on the surface as the run proceeds as the origin of the established behaviour.The ArNa' is subsequently produced and transported from the surface to the supersonic jet at the interface where it is detected by the mass spectrometer. The A 0 60 120 Timels Signal intensity for A 59C0; B ArC1' (m/z 75) (x 5); C SO2+ (mjz 64) (x 0.1); and D C10' (m/z 51) as a function of time (the first 2 min after the start of aspiration of the corresponding solution) 0 30 60 90 120 Time/s Fig. 3 Signal intensity for A s9C0; B Na2' (m/z 46); C ArNa' (m/z 63); and D Na20+ (m/z 62) as a function of time (the first 2 min after the start of aspiration of the corresponding solution) interface as the origin of species such as ArX+ has also been suggested by Lam and Horlick.,' A similar explanation can be given for the formation of Na,' Na,O+ and ArK'.A more or less constant signal for Na,' Na,O' and ArNa+ (m/z 46 62 and 63 respectively) is obtained after an aspiration time of about 10 min as shown in Fig. 4. For the long-term experiment (several hours) a stable and reproducible signal was obtained for ClO' SO,' and ArC1+ (m/z 51 64 and 75 respectively) whereas the signals of Na,' Na,O+ and ArNa' (m/z 46 62 and 63 respectively) were subjected to short- and long-term fluctuations. Reduction of and/or Correction for Spectral Interferences In order to reduce or to correct for some of the spectral interferences encountered in the analysis of biological materials several procedures have been proposed.Some of these are given here and illustrated. In several cases an isotope subject to less or no interference can be used. An example is the determination of Zn in human serum via the 68Zn (18.8%) isotope ( I F 4.1) which has less interference and not via the 64Zn (48.6%) isotope ( I F 630). Since the concentration of S in human blood plasma24 is about 1 81-' it can be calculated from Table 9 that the following apparent Zn concentrations are expected 630 60 29 and 4 pg I-' when 64Zn 66Zn 67Zn and 68Zn respectively are used for the Zn determinations. Comparison of these values with the Zn concentration in human serum2' (about 1 mg l-') shows that the spectral overlap for 68Zn is only about 0.4% and thus can be neglected. This was confirmed by the analysis of a reference serum reported by Vandecasteele et a1.26 For spectral overlap with polyatomic ions containing H N 0 and/or Ar (present in the plasma) a correction can usually be made by the use of a blank solution containing the solvent.However one should be aware of the fact that the extent to which species such as Ar,' and ArO' are formed is dependent on the total matrix composition of the sample solution.'' An example was given by Vandecasteele et al.27 who accurately determined Br in human serum. Both bromine isotopes 79Br and "Br were corrected for the Ar,H + overlap by subtraction of the blank. Vanhoe et aL6 evaluated the use of a so-called simulated blank solution to correct for spectral interferences owing to the serum matrix elements Na S and Ca. This solution contained the same amount of the interfering matrix elements as the serum solution that was analysed.Accurate results were obtained for Fe Co Cu and Zn. They concluded that accurate and precise results can only be obtained when the correction factor is limited to less than 10% of the total signal. A similar method was used by Friel et a[.,* to correct for Ca and C1 spectral interferences during the analysis of several biological reference materials. The use of correction formulae based on the isotopic abun- 25 n a I I I I 0 4 8 12 16 20 Time/m in Fig. 4 Signal intensity for A ArNa' (m/z 63); B Na,' (m/z 46); and C Na20+ (m/z 62) as a function of time (19 min)184 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY MARCH 1994 VOL. 9 dances of the interfering matrix element is another method used to correct for spectral overlap.Spectral interferences arising from C1 can be corrected for in this way; "V and 53Cr are subject to interference from 35C1160+ and 37C1160+ respectively (Table 4). The ratio of these two interferences is comparable to the ratio of the relative abundances of 3sCl and 37Cl (75.8% 24.2% = 3.132). So the contribution of 35C1160f at m/z 51 (in order to determine V) can in the absence of Cr be calculated from the 37C1160+ value at m/z 53. This correc- tion method has been applied to lobster hepatopancreas by Ridout et al.,29 whereas Park et determined Cr in human serum [National Institute for Standards and Technology (NJST) Standard Reference Material (SRM) 909 Freeze-Dried Human Serum] neglecting the presence of V in the sample.The interference of 40Ar35C1+ on 75As can be corrected for in an analogous way. The 40Ar35C1+ and 40Ar37Clf give an interference with 75As and 77Se respectively (Table 4). This method was also applied successfully to lobster hepatopancreas by Ridout et a1.29 Another illustration of this correction method is given in Table 10. Selenium was determined in a reference serum. Since the Cl concentration in human blood plasma24 is about 3.55 g l-' it can be deduced from Table 4 that the spectral interference of 40Ar37C1+ on 77Se gives an apparent Se concentration of about 280 pg I-' (compared with a serum Se c~ncentration~~ of about 100 pg 1-l). A correction for this overlap was made in two ways. Before the analysis of 8-fold diluted serum samples the ratios of 40Ar35C1+ 40Ar37Cl+ and 35C1160+ 40Ar37C1+ were determined experimentally with 1 % HCl.The ratios found were 2.960 and 18.90 respectively. In this way the mass fractionation introduced by the ICP mass spectrometer was taken into account. These ratios did not differ significantly from those obtained by spiking the diluted serum samples with an appropriate amount of HCl and were used to correct for the 40Ar37C1+ interference on 77Se. Both results agree well with the certified value. The errors introduced by the presence of As and V in the serum solution could be calculated from the certified concentrations and were 1 and less than 0.1% respectively and thus could be neglected. Another method of correcting for spectral interferences is based on the use of multivariate correction methods.A four- component principal component analysis technique has been applied by Vaughan and Templeton7 and by Xu et ~ 1 . ~ ~ to separate Ca Na and K interferents from the Ni data in order to determine Ni accurately in urine and serum. In addition to the use of blank solutions or simple equations a more extensive sample preparation procedure can also be applied. Separation of the analyte from the interfering matrix element is the most popular technique. Lyon and co-workers10*22 used size-exclusion chromatography (gel fil- tration) to de-salt the protein solution to be analysed in order to determine Se in serum. This method offers the potential of separating the proteins from the inorganic salt fraction of a serum sample.Sheppard and c o - w o r k e r ~ ~ " ~ ~ used ion chroma- tography to eliminate the interference due to ArC1+ on the determination of As in urine whereas Heitkemper et ~ 1 . ' ~ and Beauchemin et a1.33 coupled high-performance liquid chroma- tography with ICP-MS for the speciation of As compounds in urine and a dogfish muscle reference material respectively. At the same time they avoided the ArCl+ interference on As. Table 10 Results for Se for the 'second-generation' biological reference material Freeze-Dried Human Serum Se as "Se Correction with 40Ar35C1 Correction with 35Cl'60 Certified value3' Se concentration/pg I-' 95.1 (1.7)* 100.3 (4.5)t 95.5 f 4.61 * Starrdard deviation n=4. 7 Standard deviation n = 5. 95% confidence limits. Goossens and C O - W O I T ~ ~ ~ S .~ ~ ' ~ ~ used an anion-exchange resin column (Dowex-1) to separate Cl and S from the trace metals in serum and urine. In this way they were able to determine Cu Zn As and Se in serum and As and Se in urine. Plantz et al.34 used an on-line sample treatment method to separate the analyte elements (V Cr Ni Co and Cu) from alkali and alkaline earth elements and anions. The separation consisted of complexation of the metals mentioned with the reagent bis(carboxymethy1) dithiocarbamate. The proposed method was applied to the analysis of a urine standard. Janghorbani et al.35 developed a chemical separation of Br from S based on distillation from acidified solutions to eliminate the spectral overlap with S03Ht. Correct Br isotope ratios could be obtained.Serfass et Amarasiriwardena et and Patterson et used several extraction procedures to separate Zn from the matrix. In this way they were able to determine the Zn content and the Zn isotope ratios accurately in a number of biological materials such as blood plasma faeces and urine. An on-line anodic stripping voltammetry system was coupled with ICP-MS by Pretty et ~ 1 . ~ ' to avoid the Na spectral interferences allowing the determination of Cu in urine. The use of a different sample introduction method is often a good alternative. The most promising technique is electro- thermal vaporization (ETV). Since the sample is introduced into the plasma in the absence of any accompanying solvent the levels of certain polyatomic ions can be significantly reduced.Whittaker et aL4' showed the possibility of the accurate determination of isotope ratios for Fe in serum by ETV-ICP-MS. In addition the use of an ashing step and/or the addition of a chemical modifier can eliminate some matrix elements. Carey et showed the selective elimination of C1 by the addition of NH40H (with the formation of volatile NH4C1). Another possibility is the use of hydride generation (HG) to separate C1 from As and Se to determine both elements in biological materials. Ting et al.42 and Buckley et ~ 1 . ~ ~ have reported the successful determination of Se in several biological materials by HG-ICP-MS. More recently mixed-gas plasmas have been investigated to reduce or eliminate some spectral interferences. With the addition of N to the aerosol or the plasma gas Branch et ~ 1 .l ~ and Wang et a1.18 were able to determine As in urine accurately. Hill et ~ 1 . ' ~ have determined V As and Se in several biological materials by the addition of methane to the nebulizer gas An alternative method is the addition of an organic solvent to the sample. Goossens et d." reported on the accurate determi- nation of As and Se in serum and urine by the addition of 4% ethanol to the sample in combination with careful adjustment of the aerosol gas flow rate. Finally the most effective way to overcome spectral inter- ferences is of course the use of a mass spectrometer with sufficient resolution to resolve peaks that have similar m/z values. Bradshaw et ~ 1 . ~ ~ reported a double focusing magnetic sector mass spectrometer that was able to achieve a resolution up to 10000.So far no applications with biological materials have been reported. References Evans E. H. and Giglio J. J. J. Anal. At. Spectrom. 1993 8 1. Vaughan M. A. and Horlick G. Appl. Spectrosc. 1986 40 434. Tan S. H. and Horlick G. Appl. Spectrosc. 1986 40 445. Mulligan K. J. Davidson T. M. and Caruso J. A. J . Anal. At. Spectrom. 1990 5 301. Vanhaecke F. Vanhoe H. Dams R. and Vandecasteele C. Talanta 1992 39 737. Vanhoe H. Vandecasteele C. Versieck J. and Dams R. Anal. Chem. 1989 61 1851. Vaughan M. A. and Templeton D. M. Appl. Spectrosc. 1990 44 1685. Handbook of Chemistry and Physics ed. Weast R. C. Astle M. J. and Beyer W. H. CRC Press Boca Raton 64th edn. 1984.JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY MARCH 1994 VOL.9 185 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 Gray A. L. Spectrochim. Acta Part B 1986 41 151. Lyon T. D. B. Fell G. S. Hutton R. C. and Eaton A. N. J. Anal. At. Spectrom. 1988 3 601. Goossens J. Vanhaecke F. Moens L. and Dams R. Anal. Chim. Acta 1993 280 137. Goossens J. Moens L. and Dams R. J. Anal. At. Spectrom. 1993 8 921. Lyon T. D. B. Fell G. S. Hutton R. C. and Eaton A. N. J. Anal. At. Spectrom. 1988 3 265. Heitkemper D. Creed J. Caruso J. and Fricke F. L. J. Anal. At. Spectrom. 1989 4 279. Sheppard B. S. Shen W.4 Caruso J. A. Heitkemper D. T. and Fricke F. L. J. Anal. At. Spectrom. 1990 5 431. Branch S. Ebdon L. Ford M. Foulkes M. and O’Neill P. J. Anal. At. Spectrom. 1991 6 151. Sheppard B. S. Caruso J. A. Heitkernper D.T. and Wolnik K. A. Analyst 1992 117 971. Wang J. Evans E. H. and Caruso J. A. J. Anal. At. Spectrom. 1992 7 929. Hill S. J. Ford M. J. and Ebdon L. J . Anal. At. Spectrom. 1992 7 1157. Lam J. W. H. and Horlick G. Spectrochim. Acta Part B 1990 45 1327. Hutton R. C. J. Anal. At. Spectrom. 1986 1 259. Lyon T. D. B. and Fell G. S. J. Anal. At. Spectrom. 1990 5 135. Goossens J. and Dams R. J. Anal. At. Spectrom. 1992 7 1167. International Commission on Radiological Protection Report of the Task Group on Reference Man ICRP Publication 23 Pergamon Press Oxford 1975. Versieck J. and Cornelis R. Trace Elements in Human Plasma or Serum CRC Press Boca Raton 1989. Vandecasteele C. Vanhoe H. and Dams R. J Anal. At. Spectrom. 1993 8 781. Vandecasteele C. Vanhoe H. Dams R. and Versieck J. Anal. Lett. 1990 23 1827. Friel J. K. Skinner C. S. Jackson S. E. and Longerich H. P. Analyst 1990 115 269. 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 Ridout P. S. Jones H. R. and Williams J. G. Analyst 1988 113 1383. Park C. J. Smith D. C. and Vanloon J. C. Trace Elements Med. 1990 7 103. Versieck J. Vanballenberghe L. De Kesel A. Hoste J. Wallaeys B. Vandenhaute J. Baeck N. Steyaert N. Byrne A. R. and Sunderman F. W. And. Chim. Acta 1988 204 63. Xu S. X. Stuhne-Sekalec L. and Templeton D. M. J. Anal. At. Spectrorn. 1993 8 445. Beauchemin D. Siu K. W. M. McLaren J. W. and Berman S. S. J. Anal. At. Spectrom. 1989 4 285. Plantz M. R. Fritz J. S. Smith F. G. and Houk R. S. Anal. Chem. 1989 61 149. Janghorbani M. Davis T. A. and Ting B. T. G. Analyst 1988 113 405. Serfass R. E. Thompson J. J. and Houk R. S. Anal. Chim. Acta 1986 188 73. Amarasiriwardena C. J. Krushevska A. Foner H. Argentine M. D. and Barnes R. M. J. Anal. At. Spectrom. 1992 7 915. Patterson K. Y. Veillon C. Moser-Veillon P. B. and Wallace G. F. Anal. Chim. Acta 1992 258 317. Pretty J. R. Blubaugh E. A. Evans E. H. Caruso J. A. and Davidson T. M. J. Anal. At. Spectrom. 1992 7 1131. Whittaker P. G. Lind T. Williams J. G. and Gray A. L. Analyst 1989 114 675. Carey J. M. Evans E. H. Caruso J. A. and Shen W.-l. Spectrochim. Acta Part B 1991 46 1711. Ting B. T. G. Mooers C. S. and Janghorbani M. Analyst 1989 114 667. Buckley W. T. Budac J. J. Godfrey D. V. and Koenig K. M. Anal. Chem. 1992 64 724. Bradshaw N. Hall E. F. and Sanderson N. E. J. Anal. At. Spectrom. 1989 4 801. Paper 3/043 15 K Received July 21 1993 Accepted October 15 1993