摘要:

JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY APRIL 1993 VOL. 8 403 Atomic Absorption Equilibrium and Mass Spectrometry of Germanium in Graphite Atomizers Peter S. Doidge Varian Optical Spectroscopy Instruments P. 0. Box 222 Springvale Victoria 3 7 71 Australia Trevor McAllister CSlRO Materials Science and Technology Locked Bag 33 Clayton Victoria 3768 Australia The atomization of Ge in graphite atomizers has been studied by electrothermal atomic absorption spectrometry (ETAAS) and electrothermal mass spectrometry (ETMS) and by calculation of thermochemical equilibrium. The evolution of GeO(g) from alkaline or nitric acid matrices and of GeS(g) from a sulfuric acid solution of germanium at a relatively low temperature during pre-atomization or atomization have been confirmed by ETMS.No gaseous germanium carbide was detected in ETMS experiments. Equilibrium calculations at low carbon activity show how GeO(g) may be evolved at 800 K by partial reduction of GeO,. Further estimates of equilibrium for sodium germanate and sulfuric acid-GeO matrices in both tube and platform atomizers show that many of the anomalies observed in the study of Ge by ETAAS may be explained by the assumption of a temperature or matrix-dependent carbon activity of the graphite with consequent varying degrees of formation of GeO(g) or Keywords Germanium and germanium oxide; graphite furnace; atomic absorption spectrometry; mass spectrometry; equilibrium calculation GeS(g)* The atomization of Ge from graphite furnace atomizers has attracted the interest of several with varying conclusions.Appearance temperatures (T,) reported for Ge range from 1 5201-2500 K. The issue of carbon activity in the atomizer has been raised in various forms. L'v0v3 has suggested a gaseous dicarbide as an intermediate but Katskov et aL4 reported that carbides were not formed in the condensed phase and that the activation energy of atomization (E,) was close to the enthalpy of vaporization of Ge at the appearance temperature as also found by Pelieva and Martynenko.' On the other hand partial reduction to GeO(g) of the germanium species from alkaline solutions in the furnace has been observed above 1100 K by Kolb et al.5 and Dittrich et aL6 have observed GeS(g) from solutions modified by the addition of H2S04. Reports of the effect of acids on the Ge signal are varied.Pelieva and Martynenkol found that HC1 and the oxyacids depress the signal at all concentrations whereas Sohrin et al.7 found that the signal was enhanced by HN03 and HC104 at concentrations up to 0.1 mol 1-l. In this work the atomization of Ge is investigated in graphite atomizers by atomic absorption mass spectro- metry and equilibrium calculations to clarify the role of carbon to detect gaseous intermediates in the reduction of Ge and to define the optimal conditions for the deter- mination of Ge by electrothermal atomic absorption spectrometry (ETAAS). Experimental Experiments with ETAAS were performed at first by means of a Varian AA-875 with a GTA-95 atomizer. An HP-9845B was used for data collection. This apparatus was superseded by a Varian SpectrAA-30D and a SpectrAA-300Z each with a GTA-96 atomizer.Signal graphics obtained with the SpectrAA-300Z were stored and overlaid by means of signal graphics library (SGL) software (Science Traveller P/L). Pyrolytic graphite coated and uncoated atomizer tubes were used both alone and with 'I' and 'fork' pyrolytic graphite platforms (Ringsdorff-Werke Part Nos. 63 10001 3 and 63 100024). Argon came from CIG and Liquid Air. Furnace temperature was measured by means of a Leeds and Northrup 8632-C disappearing wire pyrometer. Electrothermal mass spectrometry (ETMS) and associ- ated experimental techniques have been described in detail elsewhere.s The apparatus which is shown schematically in Fig. 1 has been modified recently9 to place the graphite furnace atomizer at right angles to the ion source in order to reduce interference in the detection system from non- analysed particles and photons transmitted along the axis of the quadrupole mass spectrometer (VG SXP300).This new geometry also permitted the installation of an infrared pyrorneter (Ircon Modline 11) to record the furnace temper- ature by superimposing a digital display on the video recording of the mass spectrometer output. The pyrometer was roughly calibrated by measuring the time taken to reach 2000 "C as observed by a disappearing wire pyrometer (Pyro-Werke GMBH) at a fixed heating rate. It is not intended to obtain absolute temperature readings but rather to measure the relative T of volatile species. A standard solution of Ge [ 1000 pg ml-I in 2% KOH (Sigma)] was used to make up three different samples for ETMS diluted 9+ 1 with de-ionized water; 9+ 1 with I0/o HNO,; and 9+ 1 with 1% H2S04 respectively.ton source Viewing Mass spectrometer 4 / molecules ' and 1 u Twin Cu rod feedth roug h Fig. 1 Schematic of the ETMS apparatus showing the orientation of the furnace and the ion source404 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY APRIL 1993 VOL. 8 Equilibrium Calculations All investigations of mechanisms of electrothermal atomi- zation would benefit from an estimate of the equilibrium in the graphite-analyte system with the temperature varying from about 500 K to above 2000 K. It is the basic thermochemical property of free energy change which will determine the temperature threshold for the detection of relevant species (T,) in the gas phase.Kinetic factors which delay or expedite this appearance are seldom well known and often the subject of speculation or misunderstanding. The CSIRO (Commonwealth Scientific and Industrial Research Organization) Thermochemistry System of Turn- bull and Wadsley,lo which employs the free energy minimi- zation method of Erikssonll (see Appendix) is used. The chemical system must be carefully defined for a free energy minimization calculation. Ideally every possible product from a given set of reactants should be included but in practice there are usually only a limited number of species that need to be considered. Very restricted species sets such as are used to calculate individual reaction equilibria are useful only in the context of assumptions about the kinetics of allowed and forbidden reactions.This is essentially L'vov's approach in his hypotheses of gaseous carbides3 and gasification of involatile oxides. * Such hypotheses rely heavily on the detection of the purported intermediate species in the predicted temperature regions in amounts which would suggest that they are major atomization precursors. Equilibrium calculations with a comprehensive set of product species invariably show negligible levels of gaseous carbides. The amount of carbon included in the equilibrium calculation is as important as the choice of product species. Even though there is a favourable AG for reduction of the oxide if insufficient carbon is used reduction will be incomplete. Calculations with such low levels of carbon are of interest because they indicate what might occur when the carbon activity in the furnace is low which might happen at lower temperatures because active sites on the graphite are occupied by residual oxygen.This might also control the carbon activity of the furnace by regulating the amount of CO present as has been suggested for the reduction of PbO in the atomization of lead.13 The number of active sites may also rise with the temperature owing to a change in the surface properties of the graphite or the disposition of the matrix or both. Little is known of the variation of carbon activity of pyrolytic graphite coated graphite atomizers with temperat~re.'~ Some equilibrium calculations have at- tempted to allow for variation in carbon activity by an arbitrary variation in the partial pressure of oxygen with temperat~re,'~ but the outcome of such calculations is dependent on the amount of carbon originally assumed to be present.Results and Discussion Carbon Activity Results for Ga In and As have shown that the volatile oxides Ga,O InzO and As406 should be produced under conditions of low carbon activity which may be associated with either thermal pre-treatment or lower atomization temperatures.I6 A similar outcome was found for Ge. The result of an equilibrium calculation with equimolar amounts of C and GeO is shown in Fig. 2. The GeO(g) should be evolved at T,=800 K and carry away most of the Ge sample leaving only a minor amount of Ge to evaporate at T p 1400 K. An experiment using ETMS with the sample of Ge diluted with HN03 showed GeO+ appearing at about 900-950 K early in a 4 s atomization ramp of 500 K s-* (mass spectral 1 m 2 X v) - z ,+ 0.5 0 0 2 a A 1000 1400 1800 2200 Temperature/K Fig.2 Calculated variation of equilibrium for A GeO(g) and B Ge(g) with temperature in a system containing equimolar amounts of C and GeO 1 2 3 4 No. of moles of C x lo-* Fig. 3 Calculated variation of equilibrium amount of GeO(g) with carbon activity at 1200 K peaks mlz=86 88 and 90 in the expected Ge isotope ratios). No Ge+ was observed at higher temperatures in the atomization ramp. It might be reasoned that the HN03 sample produced GeO during drying pre-treatment with evolution of N02(g) which was observed in the mass spectrum. The GeO was then partially reduced to GeO(g) which rapidly diffuses away in vacuu before further reduc- tion can occur.Varying amounts of GeO(g) and Ge are produced in the equilibrium calculation depending on the amount of carbon present. This result is essentially similar to that of Pelieva and Martynenko,' who suggested that the reduction in their experiments followed two parallel paths through GeO(g) and directly from GeO,. Pelieva and Martynenkol have also noted the sensitivity enhancing effect of added carbon powder. The calculated amount of GeO(g) varies inversely with carbon activity at 1200 K as shown in Fig. 3. It is to be expected that considering the variety of pre-treatment techniques used by different workers and the varying age and state of the graphite surface the carbon activity would vary and therefore the amount of reduction to GeO(g) or Ge should vary. The observed T for Ge then tends to vary depending on the amount of Ge remaining after loss of GeO(g) and on the sensitivity of the detection technique. No gaseous carbide GeC,(g) was detected during ETMS experiments.One of the significant parameters found to affect theJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY APRIL 1993 VOL. 8 405 0.4 a 0.3 0 0 e 51 0.2 9 0.1 0 0.2 0.4- 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0 Ramp rate/10-3 K s-' Fig. 4 Variation of Ge signal (peak absorbance) with atomization ramp rate pyrolytic graphite coated graphite tube with Ge sample (40 ng) in HN03 (5% m/v) absorbance of Ge in ETAAS is heating rate. The results of experiments on variation of the atomization ramp rate in ETAAS are shown in Fig.4. Both peak height and integrated absorbance varies extremely sharply with the ramp rate up to the maximum 2000 K s-l. Results for the solution in HN03 are consistent with those of Johnson et ~ 1 . ~ ; there is a very large variation in signal for heating rates above 500 K s-l. It can be concluded that for optimum sensitivity the graphite atomizer must get to maximum temperature as quickly as possible and this is consistent with the hypothesis that carbon activity rises with tempera- ture so that at higher temperatures and activities less GeO(g) and more Ge are produced. Further support for this concept is given by an estimate of the variation of atomization efficiency with temperature which was calculated by means of the equationl77l8 for the characteristic mass m,.Data used in the calculation are given in Table 1. The temperature dependent parameters are given for T=2800 K and were re-calculated for other temperatures according to the formulae given by L'vovl7 and Frech and Baxter.I8 The values for the ratio of length to diameter of the absorbing vapour column were derived from measurements of m for T1 which is reported to have 100% atomization efficiency.I7J8 The value of r2/P thus derived 1 . 0 4 ~ lo-* is close to that expected from the geometrical dimensions of the tube (r=2.5 mm 1=25 mm). Table 1 Parameters used for absolute efficiency calculation with platform furnace.* Characteristic mass calculated according to the formula l 7 Diffusion coefficient 0=6.8 cm2 s-l Partition function Z(r)= 5.81 Parameter a= 0.476 Doppler width Av,=O.168 cm-' Line shape factor H(a 0)=0.589 Ratio of radius to length of absorber r2/P= 1.04 x Hyperfine factor J=0.952 Oscillator strength f=0.2 1 (gf= 1.05) Spectral line source purity factor 6=0.9 1 Excitation energy E of lower term (3P2)= 1410 cm-I (=0.175 eV) M= 72.6 g mol-I *Temperature-dependent parameters are given for a temperature of 2800 K using values reported by L'vov'~ and were re-calculated for different temperatures using formulae given by L'vov.'~ Temperatures for the calculation were taken as in L'vov'' as the wall temperature less 73 K. 80 70 60 50 0 .- + .- @ $ 0 30 $ 0 20 ." 2.5 2.7 2.9 Tern pe rat u re/l 0-3 K Fig. 5 Dependence of atomization efficiency of Ge on atomiza- tion temperature using a platform the recommended lamp current (7 mA) and other parameters discussed in the text The experimental value of rn found here for the graphite tube atomizer (GTA) should lie between those found for the heated graphite atomizer (HGA)I7 and the two-step atom- izer (TSA),I8 because the length of the GTA tube falls between those of the HGA and the TSA.This is confirmed by the results shown in Fig. 5. The efficiency rises with temperature from 10% at 2500 K to ~ 4 0 % at 2800 K and is essentially constant above 2800-3000 K. The results of Frech and BaxterI8 also show rising atomization efficiency over a similar range to about 2700-2800 K but indicate thereafter a more gradual rise from 2800-3000 K. Given the scatter in the data of the two estimates the slight difference in results between 2800 and 3000 K is unimpor- tant and it is the similarity in the finding that the increase in atomization efficiency levels off at about 2800 K which should be noted.Although essentially the same parameters have been used in the calculation as in the work of refs. 17 and 18 there is some uncertainty in the accuracy of the $values for Ge given in the literature. An $value of 0.13 has recently been found for the 265.12 nm line.I9 This value would give efficiencies about 2/3 higher than the value of 0.21 used in refs. 17 and 18. Platform Atomization It has been reported that the stabilized temperature platform gives considerably improved sensitivity for Ge in ETAAS. When platform atomizers were used in the present experiments two overlapping peaks were seen with Ge in HN03 solution (Fig.6). The second of these two peaks disappeared with decreasing atomization temperature. The double-peaked absorption profile may be explained by the evolution of GeO(g) during the pyrolysis step or a slow 0.40 I 1 3000 a Y E C 2 3 a m CI 1500 0.20 n I-" -0.05 y 1 0 1.65 3.30 Time/s Fig. 6 Germanium signals using platform atomization HN03 matrix (5% m/v) and no added Pd. Atomization temperature A 2650; and B 2850 "C406 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY APRIL 1993 VOL. 8 1 .o 7 s! L 5 0.5 E o z X Y- 1000 1400 Temperature/K Fig. 7 Calculated effect of alkaline modifier on variation of equilibrium amount of GeO(g) with temperature A Na:Ge 1:l; and B Na:Ge 1O:l atomization ramp. The first peak of the double corresponds to dissociation or reduction of the evolved GeB(g) whereas the second less volatile peak represents reduced Ge remaining on the platform.This follows from the equilib- rium calculations and ETMS experiments discussed in the previous section and from the ‘double reduction’ mecha- nism suggested by Pelieva and Martynenko.’ The effect of the platform atomizer may be illustrated by use of the CSIRO Thermochemistry System’s ‘Model’ program for calculating reactor equilibrium. The platform atomizer is defined as a three stage reactor the platform is regarded as stage 1 at 1200 K; stage 2 with 20% of the gas phase is regarded as being in contact with the tube wall at 2000 K; stage 3 is in the centre of the tube with 80% of the gas phase out of contact with the tube wall.The equilibrium for this model is shown in Table 2. Although in stage 3 GeO(g) is not substantially dissociated into Ge(g) and O(g) in the presence of the tube wall carbon as shown in the result for stage 2 GeO(g) is sufficiently reduced by a gas-solid reaction to yield a substantial conversion to Ge(g). A significant proportion of the Ge however should remain on the platform consistent with the observations given in Fig. 6. Matrix and Chemical Modifiers The equilibrium results given in Fig. 2 refer to solutions of Ge in HN03. Alkaline matrices have been investigated by several workers and Kolb et aL5 have identified Na,Ge03 by X-ray diffraction after thermal pre-treatment in a graphite atomizer of a solution of Ge in 0.1 mol 1-’ NaOH.It is possible to allow for the formation of Na2Ge03 in the equilibrium calculation. There are no published thermo- chemical data for Na2Ge03 but one can make the rough approximation of using equivalent amounts of Na20 and GeQ in the calculation. Otherwise an estimate of thermo- chemical properties can be made by analogy with Na2Si03 which has a very similar melting-point to Na2GeQ3 and therefore probably behaves in a very similar fashion in dissociating to the component oxides. The enthalpy of reaction of NazO with GeQ2 to form Na2Ge03 is taken to be equal to that of Na20 with Si02 to form Na2Si03 at - 229.7 kJ mol-I. (Data used in this calculation obtained from the Mineral Research Laboratories databasela and ref. 20.) This gives AHf (298.15 K)= 1 195 kJ mol-l for Na,Ge03.By the same analogy the entropy S” of Na2Ge03 at 298.15 K was estimated to be 127.3 J mol-1 K-l and heat capacity at constant pressure C, was taken to be equal to that of Na2Si0,. Using these data and with an input of equimolar Table 2 Reactor diagram and equilibrium amounts of Ge- containing species in Model calculation of platform atomization of Ge from GeO Equilibrium amount/mole Platform Tube wall Tube axis Species (T= 1200 K) (T=2000 K) (T= 2000 IC) Input* C 1 x 10-7 1 x 10-6 - 1.1 x 10-7 - - GeO2 - 1.8 x 7.1 x Ge(g) G a g ) output C - 9.8 x 10-7 - Ge02 - Ge(g) GeOk) - 2 x 10-13 8 x 10-13 - - 1.09 x 1 x 10-8 1.5 x 10-lQ - 4.5 x 10-12 7.1 x 10-8 *Input moles are chosen so that no element is less than 1 x low8 g atom in any of the three separate equilibria (see Appendix).amounts of Na,Ge03 and carbon the calculated T for GeO(g) is about I000 K (see Fig. 7 curve A and compare the result at 800 K for acid solutions given in Fig. 2). A T value of = 1 100- I 1 50 K was observed in the ETMS experiments using alkaline (2% KOH) solutions of Ge and Kolb et aL5 have found T = 1 100 K by molecular spectroscopy in the ETAAS of solutions of Ge in NaOH. It is noteworthy that the difference between acid and alkaline solutions found experimentally for T is similar to the calculated difference ( ~ 2 0 0 K). Adding surplus Na in the form of NaOH to the equilibrium calculation (Na:Ge= 10 1) raised the calculated T for GeQ to 1400 K (see curve B Fig. 7). The equilibrium calculation does not support the suggestion of Kolb et aL5 that surplus Na(g) reduces GeQ(g) in the gas phase at T> 1600 K.The substantially improved sensitivity ob- served by them for pre-atomization treatment above 1600 K must be due to increased carbon activity yielding relatively more Ge rather than increased reduction due to The effects of sulfuric acid on the atomization of Ge have been shown by the work of Dittrich et aL6 The reduction in sensitivity found by them was attributed to the generation of GeS(g) tentatively identified by absorption bands at 280 and 290 nm. An experiment with the sample of Ge diluted by H2SO4 using ETMS confirmed the occurrence of GeS(g) (mass spectral peaks at mlz= 102 104 and 106) with Ta= 1300 K. Dittrich et aL6 suggested that GeS(g) is formed Na(g). 0.5 0.4 8 e 5 2 0.2 0.3 0.1 0 I I I I 1 I I I 1.0 1.4 1.8 2.2 2.6 3 3.4 3.8 4.2 Log(mass Pd:Ge) Fig.8 Response (absorbance) for Ge as a function of added Pd in the ratio of masses. Response normalized to 1 ng of Ge. The absorbance for zero added Pd (not shown) was calculated to be 0.0034 from the measured peak absorbance for 18 ng of GeJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY APRIL 1993 VOL. 8 407 Table 3 Reactor diagram and equilibrium amounts of Ge and S containing species in the Model calculation of platform atomiza- tion of Ge02 sample with excess H2S04 Equilibrium amount/mole Platform (T= 1200 K) 1 x 10-6 1 x 10-7 2 x 10-7 - - - - - Tube wall (T=2000 K) 1 x 10-6 - - 2.9 x 10-17 6 . 9 ~ 10-3 4.8 x 10-9 7.2 x 10-9 2 x 10-8 Tube axis (T= 2000 K) - - I 1.1 x 10-6 2.8 x 10-l2 8 x 1 .9 ~ 2 . 9 ~ *Input moles chosen according to the same criterion as in Table 2. by a gas-phase reaction but its detection in the spectro- meter which operates in vacuum indicates that it is more likely to be the product of a solid state reaction. Dittrich et a1.6 also observed that paradoxically the platform atomizer gave a much lower improvement in sensitivity for H2S04 solutions than for HN03 solutions of Ge. This is not to be expected on the simple platform atomizer concept of dissociation in the hotter tube atmosphere of the respective diatomic molecules with GeS(g) being less stable than GeO(g). Dittrich et aL6 propose that the carbon of the tube plays an important role in sensitivity improvement in platform atomizers a suggestion which was also made by Pelieva and Martynenko,' and the equilibrium calculation using the Model program explains this result (see Table 3).The GeS(g) should form when H2S04 is added to the Ge solution but it is reduced less readily by contact with the carbon of the tube wall at 2000 K (stage 2 in Table 3) than is GeO(g) (compare Table 2). Palladium is one of the most frequently used chemical modifiers and in earlier work,21 the effect of Pd on the ETAAS signal for Ge collected as the hydride in a graphite tube has been studied. The amount of Pd required for maximum signal was about 5 pg much in excess of the characteristic mass found for Ge (about 10 pg). Plots of peak height for the atomization of Ge (tube wall) from a solution of HN03 with added Pd from 2 ng to 2 pg (Ge:Pd ratio from 1 to 1 x lo4) are shown in Fig. 8.The results are given on a logarithmic scale because the high enhancement factor required a change in the amount of Ge used in order to remain within the linear absorbance range. An enhance- ment of absorbance was found at a mass ratio (Pd:Ge) as low as 0.5 (not shown in the plot). This represents a mole ratio (Ge:Pd) of 2.9. Although Ge3Pd has not been reported intermetallic compound formation between Pd and Ge is possible and the subject has been surveyed by Samsonov and Bondarev.22 Xuanz3 has identified by X-ray photo- electron spectroscopy the products Ge9Pdz3 and GePd2 in a furnace after pyrolysis. There are no thermochemical data for germanides but the analogy of the silicides of transition metals which are well known,24 suggests that either inter- metallic compound or alloy formation with Pd would be a substantial stabilizing factor for Ge.This compound forma- tion causes a delay in the appearance time of the peak and an increase in the temperature in tube atomization such Table 4 Comparison of Ge characteristics masses with and without Pd (pyrolytic graphite platform in pyrolytic graphite coated tube temperature of atomization. 2700 K) Characteristic mass/pg* Peak height mode Peak area mode With added Pd 11.7 26 No added Pd 15.5 25 *Mass giving peak absorbance of 0.0044 or integrated absor- bance of 0.0044 s; using 200 ng of Pd (as the nitrate). that any GeO(g) which is formed is evolved at higher temperatures (T for Ge increases from 1420 K for Ge in HCl in wall atomization without Pd to 2380 K for 1 ng of Ge with 5 pg of Pd).In the platform furnace this delay is effectively accomplished by the platform itself and no benefit is conferred by using Pd as well although the peak shape is altered slightly. In Table 4 the characteristic masses for Ge using platform atomization with and without Pd are shown. Under these conditions there was no improvement in the integrated absorbance response with Pd but peaks were sharper giving a lower characteristic mass in peak height measurement; platform atomization using Pd provides a more isothermal environment for meeting the requirements of absolute analysis. Conclusion The varying results found for Ge by ETAAS may be explained in terms of varying carbon activity in the atomizer which produces differing amounts of the gaseous oxide GeO(g) depending on the heating rate and the nature of the chemical modifier and pre-atomization treatment.This conclusion is supported by the calculated equilibrium for low carbon activity in the atomizer which gives appearance temperatures for GeO(g) similar to those found in ETMS experiments from alkaline and nitric acid solu- tions of Ge. 1 2 a 4 5 6 7 8 9 10 11 12 13 14 15 16 References Pelieva L. A. and Martynenko K. P. J. Appl. Spectrosc. 1984 40 24. Johnson D. J. Sharp B. L. West T. S. and Dagnall R. M. Anal. Chem. 1975 47 1234. L'vov B. V. J. Anal. At. Spectrom. 1987 2 95. Katskov D. A. Grinshtein 1. L. and Kruglikova L. P. J. Appl. Spectrosc. 1980 33 1 175. Kolb A. Muller-Vogt G. Wendl W. and Stoessel W. Spectrochim.Acta Part B 1987 42 95 1. Dittrich K. Mandry R. Mothes W. and Judelivic J. G. Analyst 1985 110 169. Sohrin Y. Isshiki K. Kuwamoto T. and Nakayama T. Talanta 1987 34 341. Ham N. S. and McAllister T. Spectrochim. Acta Part B 1988 43 789. McAllister T. Intern. J. Mass Spectrom. Zon Proc. 1990 101 127. Turnbull A. G. and Wadsley M. W. The CSZRO Thermo- chemistry System Version V CSIRO Division of Mineral Products Port Melbourne 1986. Eriksson G. Chem. Scr. 1975 8 100. L'vov B. V. Mikrochim. Acta 1991 11 299. Salmon S. G. Davis R. H. and Holcombe J. A. Anal. Chern. 198 1 53 324. Holcombe J. A. and Koirtyohann S. R. Spectrochim. Acta Part B 1984 39 243. Cedergren A. Frech W. and Lundberg E. Anal. Chem. 1984,56 1382. McAllister T. J. Anal. At. Spectrom. 1990 5 171.408 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY APRIL 1993 VOL.8 17 L‘vov B. V. Spectrochim. Acta Part B 1990 45 633. 18 Frech W. and Baxter D. C. Spectrochim. Acta Part B 1990 45 867. 19 Komarovskii V. A. and Verolainen Ya. F. Opt. Spectros. (Engl. Transl.) 1989 66 294. 20 CRC Handbook of Chemistry and Physics ed. David R. Lide CRC Press Boca Raton 7 1st edn. 1990 p. 5-1 6. 21 Doidge P. S. Sturman B. T. and Rettberg T. M. J. Anal. At. Spectrom. 1989 4 25 1. 22 Samsonov G. V. and Bondarev V. N. The Germanides Metallurgy Moscow 1968. 23 Xuan W.-K. Spectrochim. Acta Part B 1992 47 545. 24 Schlesinger M. E. Chem. Rev. 1990 90 607. 25 Stull D. R. and Prophet H. JANAF Thermochemical Tables US Govt. Printing Office Washington 2nd edn. 1971. 26 Chase M. W. Jr. Davies C.A. Downey J. R. Jr. Frurip D. J. McDonald R. A. and Syverup A. N. J. Phys. Chem. Ref Data 1985 14 Suppl. No. 1. 27 Barin I. and Knacke O. Thermochemical Properties of Inorganic Substances Springer-Verlag Berlin 1973. 28 Barin I. Knacke O. and Kubaschewski O. Thermochemical Properties of Inorganic Substances Supplement Springer- Verlag Berlin 1977. 29 Robie R. A. Hemingway B. S. and Fisher J. R. Thermo- chemical Properties of Minerals and Related Substances U S . Geol. Survey Bull. 1452 US Govt. Printing Office Washing- ton 1978. 30 Stull D. R. Westrum E. F. and Sinke G. C. The Chemical Thermodynamics of Organic Compounds Wiley New York 1969. Paper 2/0453 7K Received August 24 1992 Accepted December I I992 Appendix The CSIRO Thermochemistry System The CSIRO Thermochemistry System is available in a package for either IBM-compatible PC or for VAX mini- computers.The package contains a series of applications programs with relevant databases of thermochemical func- tions. The two programs most applicable to electrothermal or flame atomization problems are Chemix a general equilibrium program which has been used before,8 and Model which calculates equilibrium for multi-stage reac- tors. The programs are limited to using amounts of elements greater than 1 x g atom overall in any one system. This means for instance that in the case of a graphite furnace which frequently uses 1 x 1 0-l2- 1 x l 0-9 g atom of sample the calculations must be ‘scaled’ by the addition of appropriate amounts of argon as though they occurred in a furnace of larger volume than used in practice. The databases use the published data in the compilations from the British National Physical Laboratory (1975) which has been partly superseded by newer compilations from the Scientific Group Thermodata Europe from the JANAF compilations based on the work of Stull and Prophet25 and Chase et a1.26 and from the CSIRO Mineral Products Laboratory’s own compilation based on published critical compilations from Barin and Kna~ke*~; Barin Knacke and Kubaschewski28; Robie Hemingway and Fisher29; Stull Westrum and Sinke3*. Also available is a database from the compilations of the NBSINSRDS series. Data may also be entered from literature sources other than these and it is often the case in work involving unstable intermediates in high temperature reactions that not all possible species will be available in the critical compilations. Further informa- tion on the CSIRO Thermochemistry System may be obtained from the CSIRO Division of Mineral Products P.O. Box 124 Port Melbourne Victoria Australia 3207.

ISSN:0267-9477    

DOI:10.1039/JA9930800403

出版商:RSC    

年代:1993

数据来源: RSC

摘要:

JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY APRIL 1993 VOL. 8 409 Determination of Manganese in River and Sea-water Samples by Electrothermal Atomic Absorption Spectrometry With a Tungsten Atomizer Shan Xiao-quan* Bernard Radziuk and Bernhard Welzt Department of Applied Research Bodensee werk Perkin- Elmer GmbH W-7770 Uberlingen Germany Olga VyskoCilova Laboratory of Atomic Spectrometry Institute of Chemical Technology 16628 Prague Czechoslovakia A method for the direct determination of manganese in river and sea-water reference materials using electrothermal atomization atomic absorption spectrometry with a tungsten atomizer is described. Interferences caused by the sea-water matrix are investigated and it is shown that these can be eliminated by the addition of ascorbic acid.The method is applied to the determination of manganese in sea-water reference materials and results within one standard deviation of certified values are obtained. The detection limit (30) in sea-water is 1.2 pg corresponding to 0.12 pg I-I for a 10 pl sample. Keywords Electrothermal atomic absorption spectrometry; tungsten atomizer; manganese determination; river and sea-water analysis; chemical modifier Accurate information on the concentrations and distribu- tions of trace elements in sea-water is of considerable use in the study of geological processes and of the pathways of anthropogenic input. Manganese which plays a central role in marine geological processes and is also important in biological systems,' is one of the elements occurring in unpolluted sea-water at trace concentrations between 0.02 and 10 fig 1-1.2 The determination of such elements in this matrix by electrothermal atomic absorption spectrometry (ETAAS) is made more difficult by the simultaneous volatilization of the major sea salts leading to spectral and non-spectral interferences.For this reason a number of schemes for separation from the matrix and analyte preconcentration prior to analysis have been developed. Methods based on the extraction of manganese dithiocar- bamate or quinolin-8-olate complexes have rep~rtedly~-~ yielded detection limits as low as 0.005 ,ug 1-I. A number of workers have determined a variety of trace elements including manganese in sea-water using ion-exhange col- umns filled with for example Chelex- 100 for preconcentra- tion and separation of the analyte from alkali and alkaline earth element^.^.^ Exceptionally low detection limits were achieved by these methods but the additional sample handling and the reagents used were often the cause of high and unpredic- table blank values.Moreover the methods for preconcen- tration were often time consuming and could easily lead to errors. Although it has recently been shown that semi- automatic preconcentration methods combining flow injec- tion techniques with ETAAS can lead to very high sensitiv- ity with very low blank values,* the simplest approach but with inferior limits of detection remains the direct injec- tion of the sample into the analyser. There have been a number of studies on the direct determination of manganese in sea-water.Ammonium nitrate was used by some workers as a chemical modifier9J0 but it was found to cause damage to the pyrolytic coating of the graphite tubes9 and according to Sturgeon et al.2 and Segar and Cantillo,ll to actually cause a reduction in peak height sensitivity. These workers reported that by careful choice of pyrolysis conditions most of the sea-water matrix could be removed without loss of manganese and that determination by the analyte addition technique was possible. *On leave from Research Center for Eco-Environmental Sci- ?To whom correspondence should be addressed. ences Academia Sinica Beijing China. Carnrick et a1.l2 used the determination of manganese in sea-water to demonstrate that the application of a L'vov platform in order to achieve stabilized temperature condi- tions during analyte evaporation provided increased free- dom from interferences when combined with evaluation based on integrated absorbance.It was reported by Hydes13 that interferences due to sea salts could be eliminated by the addition of 1% ascorbic acid. This worker suggested that the matrix reacted at 500 "C so that atomization took place via the oxide rather than the chloride. Using this method Tominaga and BanshoI4 obtained results with the analytical curve technique that were in agreement with those obtained by the analyte addition technique. The tungsten tube atomizer described by Sychra and co- w o r k e r ~ . ~ ~ J ~ is heated transversely and owing to a very low heat capacity very rapidly. Thus it is possible to achieve relatively isothermal conditions both spatially and tempo- rally.This might be advantageous for the analysis of samples with matrices that cause non-spectral interferences. However up to this point few studies describing the application of such atomizers to the analysis of real samples have been reported. It is the purpose of this work to investigate the applicability of the tungsten atomizer WETA-90 to the direct determination of manganese in sea- water reference materials. Experimental The tungsten atomizer workhead and power supply manu- factured at the Institute of Chemical Technology Prague Czechoslovakia and the tubes supplied by Metallwerk Plansee Reutte Austria have been described else- w h e ~ e . ' ~ J ~ The atomizer was mounted in the sample compartment of a Perkin-Elmer Model 1100 atomic ab- sorption spectrometer equipped with an Epson FX-85 printer.A manganese hollow cathode lamp (Perkin-Elmer) was operated at 20 mA. The analytical wavelength was 279.5 nm selected using a spectral bandpass of 0.2 nm. Temperature settings for the atomizer were calibrated using an Ircon Modline I1 infrared pyrometer with P3 optics (Ircon Skokie IL USA). A stock solution (Merck Darmstadt Germany) contain- ing 1000 pg 1-' of manganese as the chloride was diluted stepwise to produce working standards containing 0.2% v/v nitric acid. The nitric acid had been purified using a sub- boiling still (Kuerner Analysentechnik Rosenheim Ger- many). Ascorbic acid of the highest available purity (analytical-reagent grade Merck Darmstadt Germany)410 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY APRIL 1993 VOL.8 Table 1 Tungsten atomizer temperature programme for the determination of manganese in sea-water samples Ramp Step 1 2 3 4 5 6 7 TemperaturePC I10 120 200 1250 1251 2200 2500 Rate/"C s-I 18 1 8 210 100 30000 1000 Time/s 5.00 10.00 10.00 5.00 0.0 1 0.03 0.33 Hold time/s 40.00 1 .oo 1 .oo 30.00 0.50 1 S O 0.50 H2:Ar ratio Read 0.05 - 0.05 - 0.20 - 0.30 - 0.30 Read 0.20 Read 0.20 - was used and it was ascertained that no manganese could be detected in a 1% m/v solution. Water was purified using a Milli-Q system (Millipore Eschborn Germany). The following reference materials were used in this study Standard Reference Material (SRM) 1643b Trace Elements in Water [National Institute of Standards and Technology (NIST) Gaithersburg MD USA]; SLRS- 1 Riverine Water NASS- 1 Seawater SLEW- I Estuarine Water and CASS-2 Nearshore Seawater [National Research Council Canada (NRCC) Ottawa Canada].Standard solutions were prepared daily by dilution of the stock solution. Ascorbic acid was dissolved in the standard solutions and reference materials in order to make the final concentration in all solutions 1% m/v. Aliquots (10 pl) of the solutions were introduced into the atomizer using an Eppendorf pipette. A short length of flexible capillary tubing was added to each disposable tip used in order to prevent damage to the tungsten. The temperature pro- gramme is given in Table 1. Results and Discussion Heating Rate The low heat capacity of the tungsten atomizer makes it possible to achieve much higher heating rates than for a graphite atomizer.In the system under study it was possible to select heating rates of up to 30 K ms-l which were largely independent of the final temperature.17 Fig. 1 shows the effect of heating rate on the sensitivity for the determination of manganese. Both peak height and inte- grated absorbance increase with rate of heating up to at least 20 K ms-l although the latter somewhat more slowly. Since there was no sign of a decrease in sensitivity at the highest heating rate and since the stability of the tempera- ture conditions during atomization should increase with heating rate 30 K ms-l was used for the determinations. It should be noted that the peak height values obtained were 0.06 e 8 5 B Q 0.04 W + !!? + 8) - I ' 0.02 30 0.2 I 0 10 20 Heating rate/K ms-' Fig.1 Effect of heating rate on manganese absorbance; sample 10 pl of 10 pg 1-I of Mn in 0.2% HN03. A peak height absorbance; and B integrated absorbance more than an order of magnitude higher than those for integrated absorbance which were expressed in seconds i.e. the total absorbance divided by 54 the measurement frequency of the instrument. Since a typical peak half-width contains only 3-4 measurement cycles in the absence of interferences evaluation of peak height is preferred for this particular instrument combination. Hydrogen Concentration in the Purge Gas For tungsten atomizers it is essential that the purge gas contain some hydrogen in order to prevent oxidation of the metal and thus to prolong the lifetime of the tubes.Fig. 2 shows the dependence of the sensitivity for manganese determination on the concentration of hydrogen in the purge gas. There is no effect on the peak height or integrated absorbance for concentrations of up to 30% after which there is a significant reduction in both cases. A concentra- tion of 20°/0 well below the point at which there is a change in sensitivity but sufficient to provide a reducing atmos- phere during atomization was selected. Although it has been reported19 that hydrogen lowered the atomization temperature for manganese no significant effects on appearance time could be observed in this study. Pyrolysis and Atomization Temperature Pyrolysis and atomization temperature curves for both peak height and integrated absorbance for manganese in an aqueous reference solution are shown in Fig.3. Also shown are curves for the reference material SLEW- 1 obtained with and without the addition of ascorbic acid. The slow increase in the sensitivity with increasing pyrolysis temperature up to 1200 "C for the atomization temperature of 2200 "C might be due to the fact that the convective effects due to thermal expansion of gases decrease as the difference 8 OS4* 5 0.40 w v) 2 0.32 0.24 0.16 t B i 0.05 3 yi 0.04 2 0 10 20 30 40 50 60 Hydrogen concentration (%I Fig. 2 Effect of hydrogen concentration in argon on absorbance by manganese; sample 10 pl of 10 pug 1-1 of Mn in 0.2% HN03. A peak height absorbance; and B integrated absorbanceJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY APRIL 1993 VOL.8 41 I 0.4 8 e (D 0.2 0-0 500 1000 1500 2000 2400 Temperature/"C Fig. 3 Effect of pyrolysis and atomization temperature on absorbance by 0. ]I ng of Mn. Atomization temperature for pyrolysis study 2200 "C and pyrolysis temperature for atomization study 1200 "C. Standard solution A peak height absorbance; and B integrated absorbance. SLEW- 1 Estuarine Water G peak height absorbance; and D peak height absorbance with 1% m/v ascorbic acid between the pyrolysis and atomization stages is reduced. It should be noted that the software controlling this furnace system did not permit the introduction of a cool down step in the temperature programme which would probably have eliminated this effect. At temperatures above about 1250 "C loss of manganese due to volatilization commences.The atomization temperature curves show more or less constant sensitivity for manganese reference solutions at temperatures above 1800 "C whereas the sensitivity for both peak height and integrated absorbance continues to increase up to at least 2000 "C for the sea-water sample. A slightly higher temperature 2200 "C was selected for the analyses of matrix containing solutions. Interferences Due to Sea-water Matrix In many interference studies the effects of concommitant species have been considered singly and not necessarily at the concentrations found in the samples to be analysed. Hydes13 studied the determination of manganese in sea- water using a sodium chloride solution and a mixture of sodium chloride and sulfate in addition to sea-water.Absorbance by manganese was suppressed seriously in a solution containing 35 g 1-l of sodium chloride a concen- tration typically found in sea-water. This interference effect was much reduced when sulfate was added also at the concentration found in sea-water. The interference effects observed for the actual sea-water sample were different again. Thus only studies made with ii real sea-water matrix can be expected to be truly representative. In this work interferences were studied by comparing manganese absorbance in SLEW- 1 Estuarine Water reference material which contains a relatively high (13.1 pg 1-l) concentration of manganese and in reference solutions made up to contain the same concentration of manganese. The direct comparison made using the temperature programme given in Table 1 is shown in Fig.4(a) and (b). It is apparent that the matrix causes a change in the rate of atom formation resulting in a broader peak and change in peak height but that the overall atomization efficiency is unchanged as the integrated absorbance is constant. The addition of ascorbic 0 0.5 Time/s 1 .o Fig. 4 Cornparison of atomization profiles and absorbances for standard solutions and sea-water reference materials; broken line is background absorbance. (a) 10 pl of 13.1 pg 1-1 of Mn (peak height absorbance 0.667 and integrated absorbance 0.036 s); (b) 10 pl of SLEW-1 containing 13.1 kO.8 pg 1-* of Mn (peak height absor- bance 0.5 10 and integrated absorbance 0.037 s); and (c) 10 pl of SLEW-1 with lo/o m/v ascorbic acid (peak height absorbance 0.674; integrated absorbance 0.035 s) 0.4 ( b ) 1 0 0.5 Time/s 1 .o Fig.5 Comparison of atomization profiles and absorbances obtained after pyrolysis at 1200 "C; broken line is background absorbance. (a) 10 pI of a standard solution containing 6.6 pg I-' of Mn (peak height absorbance 0.4 16 and integrated absorbance 0.019 s); (b) 10 p1 of I + 1 diluted SLEW-1 containing 6.6 pg 1-1 of Mn with pyrolysis time 30 s (peak height absorbance 0.283 and integrated absorbance 0.016 s); and (c) 10 pl of 1 + 1 diluted SLEW-1 with pyrolysis time 60 s (peak height absorbance 0.275 and integrated absorbance 0.0 16 s)412 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY APRIL 1993 VOL. 8 15 r d L 1- 7 I 1 I I 0 1 2 3 4 Ascorbic acid concentration (% m/v) Fig.6 Effectiveness of ascorbic acid in reducing interferences due to sea-water matrix on the determination of manganese in SLEW- 1 Estuarine Water. Broken lines indicate the 95% confidence level 0 peak height absorbance; 0 integrated absorbance acid to the reference material to give a concentration of 1% m/v [Fig. 4(c)] effectively removed the interference in peak height and the integrated absorbance remained constant also. It has been reported by Sturgeon et a1.2 that manganese could be relatively easily determined in sea-water by graphite atomizer ETAAS since most of the sea-water matrix could be selectively volatilized during the pyrolysis step without loss of analyte. Hydes13 also reported that the matrix could be eliminated in 50-90 s of pyrolysis at 550 "C.As is indicated in Fig. 5 selective volatilization alone was not sufficient to remove the interference due to the matrix on manganese peak height absorbance. There was little difference in the peak height between pyrolysis for 30 and 60 s at the maximum usable temperature of 1200 "C. As shown in Fig. 6 the effectiveness of ascorbic acid in removing the interference on the peak height absorbance is dependent on the concentration of the modifier. A mini- mum of 1% m/v is required in order that the peak height absorbance measured for the undiluted SLEW- 1 reference material matches within the 9 5% confidence limits marked in the figure that for an aqueous reference solution containing the same concentration of manganese. Above this concentration the amount of ascorbic acid is not critical up to at least 4% m/v.Table 2 shows that dilution of the matrix was not helpful in improving the accuracy of the manganese determination based on peak height absor- bance. The difficulty in quantifying the integrated absor- bance at manganese concentrations below about 5 pg 1-1 also becomes apparent. Recovery of Manganese from Sea-water The reference material NASS-I contains only 0.022 k 0.007 pg 1-1 of manganese and so was convenient for the study of recoveries. Manganese was added to diluted and undiluted NASS-1 so that final concentrations ranged from 1 to 5 pg la-". Recoveries were measured with and without the addition of 1% m/v ascorbic acid by comparison with the response function for aqueous reference solutions. It is evident that good recoveries for undiluted sea-water were obtainable only upon the addition of ascorbic acid even if integrated absorbance was used for evaluation (Table 3).Determination of Manganese in Freshwater and Sea-water Samples Calibration curves for manganese in aqueous solutions containing 1% m/v ascorbic acid did not differ from those obtained without a modifier. The linear range extends to about 5 pg l-* but the curvature is slight and it is certainly Table 2 Comparison of peak height and integrated absorbance obtained for an aqueous reference solution and sea-water (SLEW-1) containing the same amount of manganese with and without the addition of L-ascorbic acid Absorbance of Mn in sea-water No ascorbic acid Mn concentration Pea,k height Integrated Sample in sea-water/pg I-' absorbance absorbance SLEW-1 13.1 0.473 0.03 1 SLEW-1 (1 + 1 ) 6.6 0.275 0.016 SLEW-1 (1+3) 3.3 0.168 0.009 Aqueous solution 13.1 - _. 6.6 - - 3.3 - - 1% m/v ascorbic acid Peak height Integrated absorbance absorbance 0.688 0.039 0.366 0.0 19 0.194 0.0 10 0.709 0.037 0.387 0.0 18 0.192 0.0 10 Table 3 Recovery of manganese from NASS-1; n= 3 Recovery (O/O) No ascorbic acid 1% m/v ascorbic acid Sea-water Mn added/mg ml-I NASS-1 ( 1 +3) 1 2 5 NASS-1 ( 1 + 1) 1 2 5 NASS- 1 1 2 5 Integrated ahsorhance 97 100 90 66 83 67 29 57 510 Peak height absorbance 76 60 49 54 41 41 36 43 34 Integrated absorbance 100 114 100 100 100 94 100 100 101 Peak height absorbance 114 102 103 106 104 99 98 100 96JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY APRIL 1993 VOL.8 413 Table 4 Determination of manganese in water samples Manganese concentration/pg 1 - I This work* Without Ascorbic Certified Water sample ascorbic acid acid (1 O/o) value NIST SRM 1643b 27.1 k 1.3 - 2 8 k 2 SLRS- 1 1.64 k 0.10 - 1.77 +.0.23 SLEW-1 - 12.47k0.76 13.1 20.8 CASS-2 - 1.88 2 0.13 1.99 +- 0.15 *n= 10 peak height evaluation. feasible to work up to 20 pg 1-l. The results for the determination of manganese in four reference materials are shown in Table 4. Addition of a chemical modifier proved to be unnecessary for the freshwater samples 1643b and SLRS- 1. The materials 1643b and SLEW- 1 were diluted 10 and 4 times respectively. Agreement with certified values is in all cases within one standard deviation. The absolute detection limit for manganese was deter- mined in NASS- 1 sea-water reference material matrix as the peak height absorbance corresponding to three times the standard deviation of ten replicate measurements and was 1.2 pg corresponding to 0.12 pg 1-' for a 10 pl sample.A 3a detection limit of 0.09 pg 1-I using peak height evaluation and a 20 pl sample volume has been reportedZo for the determination of manganese by graphite atomizer ETAAS. In sea-water 2 0 detection limits for both peak height and integrated absorbance were 0.02 pg 1-l for a 50 pl sample volume with platform atomization and chemical modifica- tion using Mg(NOj)2.21 Conclusions The tungsten atomizer under study was applicable to the determination of manganese in sea-water matrix. For this type of atomizer as well as for graphite atomizers inte- grated absorbance evaluation reduced the degree of mea- sured interference effects. However owing to the extremely shortlived absorption resulting in a very high peak height to integrated absorbance ratio quantification using peak height absorbance was easier.This became practicable for the analysis of sea-water only if ascorbic acid was added as a chemical modifier. The detection limits measured for the tungsten atomizer using a conventional AA spectrometer are in spite of the very high heating rates no better than those reported for graphite atomizers. Thus there is no analytical advantage in the application of this combination to the determination of the non-refractory element manganese. References 1 Statham P. J. Anal. Chim. Acta 1985 169 149. 2 Sturgeon R.E. Berman S. S. Desaulniers A. and Russell D. S. Anal. Chem. 1979 51 2364. 3 Weiss H. V. Kenis P. R. Korkisch J. and Steffan I. Anal. Chim. Acta 1979 104 337. 4 Klinkhammer G. P. Anal. Chem. 1980 52 117. 5 Shijo Y. Watenabe J. Akiyama S. Shimizu T. and Sakai K. Bunseki Kagaku 1987 36 59. 6 Kingston H. M. Barnes I. L. Brady T. J. Rains T. C. and Champ M. A. Anal. Chem. 1978 50 2064. 7 Corsini A. Wade G. Wan C. C.. and Prasad S. Can. J. Spectrosc. 1987 65 9 15. 8 Welz B. Microchem. J. 1992 45 163. 9 McArthur J. M. Anal. Chim. ricta 1977 93 77. 10 Montgomery J. R. and Peterson G. M. Anal. Chim. Acta 1980 117 397. 11 Segar D. .4. and Cantillo A. Y. Adv. Chern. Ser.. 1975 147 56. 12 Carnrick 6. R. Slavin W. and Manning D. C. Anal. Chem. 1981 53 1866. 13 Hydes D. J. Anal. Chem. 1980 52 959. 14 Tominaga M. and Bansho K. Anal. Chim. Acta 1985 169 171. 15 Sychra V. Kolihova D. Vyskociiova O. and Hlavac R. Anal. Chim. Acta 1979 105 263. 16 Puschel P. Formanek Z. Hlavac R. Kolihova D. and Sychra V. Anal. Chim. Acta 1981 127 109. 17 Sychra V. Doleial J. HlavaE R. PEtroS L. VyskoEilova O. Kolihova D. and Piischel P. J. Anal. At. Specctrom. 1991,6 521. 18 Shan X.-q. Radziuk B. Welz B. and Sychra V. J. Anal. At. Spectrom. 1992 7 389. 19 Suzuki M. Ohta K. and Yamakita T. Anal. Cham. 1981 53 9. 20 Guide to Techniques and Applications of Atomic Spectroscopy Perkin-Elmer Norwalk CT 199 1. 21 Carnrick G. R. Slavin W. and Manning D. C. Anal. Chem. 1991 53 1866. Paper 2/04299A Received August 10 I992 Accepted November 30 1992

ISSN:0267-9477    

DOI:10.1039/JA9930800409

出版商:RSC    

年代:1993

数据来源: RSC

摘要:

JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY APRIL 1993 VOL. 8 41 5 Preconcentration of Lead(ii) in Environmental Waters Using a Chelating Resin and Subsequent Determination as a Resin-Water Suspension by Hydride Generation Atomic Absorption Spectrometry Masahiko Chikuma and Hiroyuki Aoki Osaka University of Pharmaceutical Sciences Ka wai Matsubara Osaka 580 Japan A method is described for the determination of lead(ii) in environmental water samples by hydride generation atomic absorption spectrometry combined with a resin-water suspension sampling technique after preconcen- tratisn using a chelate-forming resin. The chelate-forming resin was prepared by the modification of a commercially available anion-exchange resin and the sodium salt of a sulfonic acid derivative of dithizone disodium (4-su~fophenyl)-l-[2-(4-sulfophenyl)hydrazide]diazenecarbothioate (DzS).Lead(ii) was found to be collected completely in the pH range between 6.3 and 10 by the DzS-loaded resin. When the aqueous suspension of metal-adsorbed resin was introduced into a hydride generation system with 1 mol I-* HCI 15% v/v H202 and 5% rn/v NaBH4 (in 0.5% m/v NaQH) lead(![) in the resin was converted into its hydride effectively. Neither DzS nor DzS-loaded resin interferes with the generation of lead hydride. The recovery and relative standard deviation were found to be 104 and 4% (n=4) respectively when 2.5 pg of lead(ii) was spiked into 1 I of distilled water. The proposed method was applied to the determination of trace amounts of lead(ii) in environmental water samples. Keywords Hydride generation atomic absorption spectrometry; lead determination; dithizone; chelating resin; suspension introduction Hydride generation techniques in atomic absorption spec- trometry (AAS) are commonly used for the determination of hydride-forming elements such as antimony arsenic bismuth germanium lead selenium tellurium and tin.The principal advantages of the hydride generation technique in conjunction with AAS include the concentration and separation of the analytes from potential matrix interfer- ence caused by background absorption. However precon- centration is often required to determine ultra-trace ele- ments in environmental waters and other samples. Several preconcentration methods such as solvent extraction co- precipitation and ion exchange by ion-exchange resins and chelating resins are often used.' Enrichment of metal ions with chelating resins is the most promising among the concentration methods owing to their selectivity large concentration factor and ease of handling. Recently.pre- concentration methods by use of chelating resins and other sorbents followed by electrothermal AAS (ETAAS) with direct heating of the sorbent on a graphite furnace have been successfully applied to the determination of trace elemenk2-* If hydride-forming elements adsorbed on solid adsorbents such as chelating resins are converted directly into their hydrides without the need for a desorption procedure hydride generation AAS also would be much more useful. In this paper a specially adapted chelate-forming resin is described (i.e. a sulfonated dithizone-loaded anion-ex- change resin) and its suitability to collect lead(@ in environmental water samples with determination of the lead@) adsorbed in the resin by hydride generation AAS is examined. For the first time a sensitive and precise method for the determination of lead(n) by introducing a lead@)- bearing resin in aqueous solution to a hydride generation atomic absorption spectrometer is described. Experimental Apparatus A Nippon Jarrel-Ash AA-880 atomic absorption spectro- meter (Kyoto Japan) equipped with hydride generation apparatus and an electrically heatable quartz tube furnace was used for all experimental work. Reagents The sodium salt of a sulfonic acid derivative of dithizone disodium (4-sulfopheny1)- 1 42-( 4-sulfophenyl)hydrazide]di- azene-carbothioate (DzS) was prepared from phenylhydra- zine-p-sulfonic acid and carbon disulfide as previously rep~rted.~ A standard lead@) solution (1000 mg 1-l) was obtained from Wako Pure Chemical Industries Osaka Japan. Working solutions were prepared freshly every day by diluting appropriate aliquots of the standard solution.Hydrochloric acid and Hz02 (30% v/v) specially prepared for the analysis of heavy metals were purchased from Nakalai Tesque Kyoto Japan and Wako Pure Chemical Industries respectively. Distilled water was purified with a Milli-Q system (Millipore Bedford USA). All other re- agents used were of analytical-reagent grade. Preparation of DzS-loaded Resin Place 10 g of an anion-exchange resin Amberlite IRA 400 (200-400 mesh) and 200 ml of 5 mmol 1-1 DzS in a flask with a stopper.Shake the flask by use of a mechanical shaker until the supernatant of the mixture becomes colourless. Separate the DzS-loaded resin by filtration and wash with water and methanol. Uptake of lead(n) by DzS-loaded resin Place 50 mg of DzS-loaded resin and 50 ml of buffer (pH 1-2 HC1-KCl; pH 3-6 CH,COOH-CH,COONa; pH 7-8 KH2PB4-NaOH; pH 9- 10 NH4Cl-NH40H) containing 25 pg !-l of lead (as nitrate) in a glass tube with a stopper. Maintain the ionic strength at 0.1 by adding potassium nitrate. Shake the glass tube with a mechanical shaker for 2 h at 30 "C in the pH-dependence study of metal uptake and for 3 min-2 h in the time-dependence study. Filter off the resin and determine the lead concentration in the filtrate by ETAAS.Calculate the amount of lead sorbed by subtracting the lead remaining in the filtrate from the amount added. Procedure Filter sample solutions through a glass fibre filter (Whatman GF/C 47 mm) immediately after sampling. Place a selected416 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY APRIL 1993 VOE. 8 Ar 11 Fig. 1 Hydride generation system for AAS 1 5Oh NaBH,; 2 1 rnol 1-l HCl; 3 15Oh H,O,; 4 sample; 5 peristaltic pump; 6 flow meter; 7 mixing coil; 8 first gas-liquid separator; 9 condenser; 10 second gas-liquid separator; 11 drain; 12 quartz cell; and 13 heater. volume of water sample containing up to 10 pg of lead(@ in an Erlenmeyer flask (3 1) and adjust the pH to 7.0 with 0.05 moll-' Na2HP04-KH2P04. Add 50 mg of DzS-loaded resin to the flask and shake the flask with a mechanical shaker for 3 h.Separate the lead(@-adsorbed resin by filtration wash the resin with water and ethanol and air-dry and store the resin in a desiccator (humidity 20-30%). Add 40 mg of the lead-adsorbed resin to 40 ml of water. Mix the resin-water suspension thoroughly and introduce 12.5 ml of the suspension into the hydride generation apparatus (Fig. 1) with a mixing coil (length about 1.0 m) and a peristaltic pump. Introduce 1 moll-' HCl 15% v/v H202 and 5% m/v NaBH in 0.5% m/v NaOH solutions into the mixing coil with another peristaltic pump. Sweep the generated hydride into the quartz tube furnace (optical path length 225 mm) using argon gas. Set the instrumental parameters as follows wavelength 2 17.0 nm; spectral slit-width 0.68 nm; lamp current 15 mA; temperature of the quartz tube furnace 1000 "C; and argon flow rate 1.0 1 min-l. Results and Discussion Uptake of Lead by DzS-loaded Resin The time required for complete uptake of lead@) was 30 min at pH 7.0 and for 50% uptake was less than 5 min.The metal uptake and its rate depend on the amount of DzS immobilized and up to 1.5 mmol of DzS can be immobil- ized per gram of the anion-exchange resin. However the chelating agent is inclined to become removed from the resin when more than 1 .O mmol is loaded. The optimum pH range for the uptake of lead(@ by DzS-loaded resin was 6.3- 10. Hydride Generation Several kinds of reducing solution have been developed to generate lead hydride. A medium HC1-NaBH4-H2O2 reported by Vijan and Woodlo was chosen for the present hydride generation AAS system.The optimum concentra- tions of each component were investigated for this study. The optimum HC1 and H202 concentrations for the hydride generation of lead(@ in DzS-loaded resin were 1 .O-1.6 mol 1-l and 10-20% m/v respectively. The generation of lead hydride decreases at concentrations of HC1 above 1.8 and below 0.8 mol 1-l. The conversion of lead@) into its hydride increases with increasing concentration of NaBH under the described conditions. To prevent the quartz tube furnace from deteriorating the concentration of NaBH was set at 5% m/v. The calibration curves were linear in the range of 5-200 pg I-* of lead(@ both in the aqueous solution and in the resin-water suspension and the concentrations corre- sponding to an absorbance of 0.0044 were 4.4 and 3.5 pg 1-1 of lead(@ respectively.The slope of the calibration curve obtained from the resin-water suspension is about 1.3 times that obtained from the aqueous standard solution indicating that lead in the DzS-immobilized resin is labile. The absorbance at the blank level was 0.0004 and the detection limit [signal to noise ratio (SIN)= 31 was 0.3 pg 1-l in the aqueous solution whereas they were 0.0009 and 0.5 pg 1-l in the resin-water suspension respectively when 1 mg of lead-bearing resin was suspended in 1 ml of water. The apparent detection limit of the proposed method depends on the concentration factor and is 0.025 pug 1 - I when 1 1 of sample is treated with 50 mg of the resin. The recovery and relative standard deviation (RSD) were 104 and 4% (n=4) respectively when 2.5 pg of lead@) was spiked into 1 1 of distilled water.Effect of Chelating Agents on the Generation of Lead H ydride It is reported that some chelating agents and complexing agents interfere with the generation of lead hydride,20 whereas a chelating agent nitroso-R salt enhances lead hydride formation.' In the present hydride generation system chelating agents such as ethylenediaminetetraacetic acid (EDTA) and 8-mercaptoquinoline-5-sulfonic acid (MQS) and a chelating resin such as Chelex- 100 interfered seriously with the generation of lead hydride. On the other hand there was no interference from DzS and DzS-loaded resin. The results are summarized in Table I. These findings suggest that the hydride conversion of lead(@ is influenced by the ability of the chelating agents to form complexes or the lability of their complexes.The DzS does not react with lead@) in 1 mol 1-1 HCl and decomposes gradually in a strongly acidic medium. Interference Study It is well known that a number of transition metals mainly those of Groups 8 and 11 and other volatile hydride- Table 1 Effect of chelating agents and chelating resins on the generation of lead hydride in the presence of 300 ng mi-' of lead(r1) in aqueous solution or in suspension containing 1 mg ml-1 of resin. For one determination 12.5 ml of the solution or the suspension was introduced into the hydride generation apparatus Agent Relative sensitivity (O/O) None I00 0.1 mol 1-l DzS 102 0.1 rnol I-' MQS 62 0.1 rnol 1-l EDTA 1 DzS-loaded resin 103 Chelex- 100 0JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY APRIL 1993.VOL. 8 41 7 Table 2 Effect of foreign ions on the generation of lead hydride in the presence of 100 ng ml-I of lead(nr) in a 0.1 mol 1 - I HN03 solution and in a suspension containing 1 mg ml-l of resin Relative sensitivity (O/O) (50 mg I-') Solution Suspension Foreign ion added None As111 GeIV* SeIV CU" CO" N ill *lo0 mg I-' were added. 100.0 100.0 92.2 99.0 84.4 98.9 98.4 99.9 77.1 84.4 90.0 97.7 79.8 95-5 Table 3 Lead concentration in spring water (collected in Osaka on March 6 199 1 ) determined by the proposed method Sample Standard Found +- SDI Recovery volume/l added/pg Pg I-' (O/O) 2.5 None 0.20 2 0.0 l(n= 5) - 2.5 1.25 rfi 0.06(n=4) 105 1 .o 2.5 2.78 rfi 0.12@=4) I03 forming elements cause severe signal depression in the hydride-generation technique.Interference from hydride- forming elements and transition metals in the determina- tion of lead@) was studied. The results are shown in Table 2. Volatile hydride-forming elements such as arsenic(w) germanium(rv) and selenium(1v) and transition metals such as cobalt(rI) nickel(n) and copper(n) interfere with the hydride generation of lead(m) in 0.1 mol 1-I HNQ3. However the interference from arsenic(m) germanium- (IV) and selenium(1v) is eliminated almost completely by the proposed procedure because they are not absorbed by DzS- loaded resin under the described conditions and not introduced into the hydride generation system. On the other hand copper(rI) nickel(r1) and cobalt(r1) are almost completely adsorbed on the resin together with lead@) under these conditions.Improvement of signal depression caused by these metals might result from the stability and inertness of their DzS complexes in the resin. Welz and Melcher2I have systematically investigated the mechanism of transition metal interferences in hydride generation AAS and pointed out that it is based on the preferential reduction of the interfering metal ions to lower oxidation states followed by the capture and decomposition of the hydrides formed. The interference from transition metal ions might be reduced by the immobilization and inactivation of the free metal ions by complex formation. The interference from copper(r1) is still present probably because the DzS-copper(r1) complex is labile in the resin.Determination of Lead@) in Environmental Waters The proposed method was applied to the determination of lead@) in natural water samples collected from a spring that is not polluted. Since the reaction time required for complete uptake of metals might depend on their concen- trations and the matrix composition of the samples it was determined from sample to sample by studying the time dependence of lead uptake. The standard additions tech- nique was also used to maintain lead levels at > 1 .O pg 1-' in the samples. The recovery and relative standard deviation of lead(@ were 105 and 4.8% ( n = 4) respectively when 2.5 pg of lead@) was spiked into 2.5 1 of the spring water sample as shown in Table 3.The concentration of lead(I1) in the sample was 0.20-+0.01 pg 1-' (meankstandard deviation) for five replicate determinations. In the labora- tory it was difficult to determine such a low level of lead by ETAAS and hydride generation AAS without preconcentra- tion. For one determination in the present study 2.5 1 of the sample were used in order to obtain more precise values although 1.3 1 of sample volume was sufficient. 1 2 3 4 5 6 7 8 9 10 l i 112 13 14 15 16 17 18 19 20 21 References Minczewski J. Chwastowska J. and Dybczynski R. Separa- tion and Preconcentration Methods in Inorganic Trace Analy- sis transl. ed. Masson M. R. Ellis Horwood Chichester 1982. Chikuma M. Aoki H. Tanaka H. and Tanaka T. in New Development in Ion Exchange Materials Fundamentals and Applications eds.Abe M. Kataoka T. and Suzuki T. Kodansha Tokyo 199 1 pp. 388-394. Chikuma M. Tanaka T. andTanaka H. Biomed. Res. 7'race Elein. 1991 2 85. Okutani T. Kubota T. Sugimura N. and Tsuruta Y. Nippon Kagaku Kaishi 1991 315. Takada T. and Koide T. Anal. Chirn. Acta 1987 198 303. Isozaki A. Kumagai K. and Utsumi S. Anal. Chim. Acta 1983 153 15. Isozaki A. Soeda N. and Utsumi S. Bull. ('hem. SOC. Japan 1981 54 1364. Nakano K. Takada T. and Fujita K. Chem. Lett. 1979 869. Tanaka H. Chikuma M. Harada A. Ueda T. and Yube S. Talanta 1976 23 489. Vijan P. N. and Wood G. R. Analyst 1976 101 966. Zhang S. Han H. and Ni G. Anal. Chim. Acta 1989 221 85. Bonilla M. Rodriguez L. and Camara C. J. Anal. At. Spectrom. 1987 2 157. Castillo J R. Mir J. M. Martinez C. Val J. and Colon M. P. Mikrochim. Acta 1985 I 253. Aznarez J. Palacios F. Vidal J. C. and Galban J. Analyst 1984 109 713. Jin K. and Taga M. Bunseki Kagaku 1980 29 522. Chapman J. F. and Dale L. S. Anal. Chim. Acta 1979 111 137. Jin K. Taga M. Yoshida H. and Hikime S. Bunseki Kagaku 1978,27 759. Fleming H. D. and Ide R. G. Anal. Chim. Acta 1976,83,67. Li J. Liu Y. and Lin T. Anal. Chim. Acta 1990 231 151. Castillo J. R. Mir J. M. Val J. Colon M. P. and Martinez C. Analyst 1985 110 12 19. Welz B. and Melcher M. Analyst 1984 109 569. Paper 2/008621 Received February 18 1992 Accepted December 8 1992

ISSN:0267-9477    

DOI:10.1039/JA9930800415

出版商:RSC    

年代:1993

数据来源: RSC

摘要:

JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY APRIL 1993 VOL. 8 419 Determination of Lead and Cadmium in Human Seminal Fluid by Electrothermal Atomic Absorption Spectrometry* Jasna Jurasovic and Spomenka Telihan Clinical Toxicology Laboratoryf Institute for Medical Research and Occupational Health University of Zagreb 2 Ksaverska 4 1000 Zagreb Croatia A sensitive method is described for the determination of Pb and Cd in human seminal fluid by electrothermal atomic absorption spectrometry with Zeeman-effect background correction. The method includes deproteiniz- ation of seminal fhid with nitric acid calibration using matrix-matched standards containing a mixture of Pb and Cd and integrated absorbance measurement using a L’vov platform in a non-grooved pyrolytic graphite coated graphite tube.The detection limits (317) are 1.4 pg I-’ of Pb and 0.05 pg I- of Cd in seminal fluid. The day-to-day precisions (relative standard deviations) of duplicate determinations in 121 samples are in the range 4-1 4% at 4-48 pg I-’ of Pb and 4-1 6% at 0.2-3.6 pg I-’ of Cd in seminal fluid. The recoveries (mean k standard deviation) are 100.6 c 3.6% of Pb and 98.8 k 4.2% of Cd and the characteristic mass values are 1 1.6 pg of Pb and 0.41 pg of Cd indicating the absence of interferences. Data are presented on the levels of Pb in seminal fluid in normal subjects and those working with Pb and of Cd in seminal fluid in non-smokers and smokers. Corresponding ‘normal values’ (median and range) of Pb and Cd in seminal fluid are 8.6 (4.2-1 6.6) pg I-’ and 0.54 (0.1 7-1.67) pg I-I respectively.Keywords Lead and cadmium determination; human seminal fluid; nitric acid deproteinization; electrothermal atomic absorption spectrometry; male reproduction capacity There is increasing interest in the possible adverse effects of environmental factors on reproduction capacity in men which is mainly due to a decrease in sperm density noted in the general population over the past three decades and the possibility that the human male might be more vulnerable to toxic influences than other mammals. It has been noted however that the human sperm output is about four times less than that of other mammals (ie. the number of sperm cells produced per gram of testes) and that the human ejaculate is unique when considering the relatively high number of abnormal sperm cells that are regarded as being ‘normally’ present in a fertile male.’ Lead and Cd are inevitably present in the human environment and both are known to be toxic agents which accumulate in the human body over a lifetime (including pre-natal life particularly with Pb).Lead is mostly accumu- lated in the bones whereas Cd is mostly accumulated in the renal cortex. The biological half-lives of Pb and Cd in the human body ( i e . the time for clearance of half the body burden of the metal) are estimated to be ‘a number of years’ and ‘several years’ respectively,2 whereas data regarding various target organs are particularly scarce and ineonclu- sive. Although the storage of Pb in bones (up to 95O/o of the body burden of Pb) has long been considered to be a sequestration or removal of Pb from active sites in soft tissues recent data have shown that Pb can be released from bones under certain stresses and physiological changes3 Apart from numerous sources of occupational exposure to each of the metals the most important non- occupational sources are food water (Pb mostly from Pb pipes in contact with soft and acidic water) air (especially Pb from petrol in dense traffic areas) smoking habits (Cd and to a lesser extent Pb from tobaccoj and alcohol consumption (Pb-contaminated alcoholic beverages). There is also some evidence of the possible effect of Pb and alcohol interaction in man i.e.an ethanol-induced increase in the biologically active fraction of Pb accumulated in the organi~rn.~ Although experimental data for animals have indicated that both Pb and Cd can reduce male reproduction *Presented at the XXVII Colloquium Spectroscopicum Interna- tionale (CSI) Bergen Norway June 9- 14 I99 1..capacity veiy few data are available regarding the possible reproductive effects of Pb and/or Cd in Lead and Cd have been found both in the seminal fluid and spermatozoa of patients suspected of infertility.7 A significantly higher concentration of Cd in whole semen was found in infertile subjects than in feitile subjects whereas no significant difference in whole semen Pb concentration was observed.$ It should be mentioned that in the latter two studies7,* no information was presented regarding possible occupational exposure to Pb or Cd in the population studied.On the other hand a significantly higher concentration of Pb in seminal fluid was found in the infertile subjects than in fertile subjects with no occupational exposure to Pb.9 In subjects with no occupational exposure to Cd a trend of increasing concentration of Cd in seminal fluid was found with respect to smoking habits ie. cigarettes per day although no significant correlation with the parameters of semen quality was observed. lo However relevant human data are generally scarce and inconclusive and there is obviously a lack of published analytical methods regarding the concentrations of Pb and Cd in the human ejaculate. A recent study of men has indicated that exposure to Pb and to a lesser extent exposure to Cd through smoking habits can reduce semen quality’ and can somewhat affect male reproductive endocrine function.12 During the same study a method was developed for the determination of Pb and Cd in seminal fluid as such data are essential for establishing the location and mechanism(s) of the effects of Pb and/or Cd on reproduction capacity in men.Experimental Instrumentation A Perkin-Elmer Zeeman/S 100 atomic absorption spectro- meter equipped with an HCA-600 graphite furnace an AS- 60 autosampler and a PR-310 printer was used. The instrument was controlled and signals were processed by a Perkin-Elmer 7700 professional computer. Argon was used as a purge gas. Solid pyrolytic graphite L’vov platforms (Perkin-Elmer Part No. BO-1 09324) in grooved pyrolytic graphite coated graphite tubes (Perkin-Elmer Part No. BO-109322) were used initially and were found to produce variable atomic420 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY APRIL 1993 VOL.8 signals and analytical sensitivity. As considerably smaller variations in atomic signals were observed when using L'vov platforms in non-grooved pyrolytic graphite coated graphite tubes (Perkin-Elmer Part No. BO-09 1 504) these were subsequently used throughout the study. An Eppendorf 54 1 5-C centrifuge and Eppendorf 38 10 microtubes (1.5 ml volume) were also used. Reagents De-ionized water purified to approximately 18 MR cm-l was used for the preparation of standards and for final washing of the laboratory ware. Working standard solu- tions containing a mixture of Pb and Cd in 0.02 rnol 1-' HNO were prepared from the BDH Spectrosol solutions containing 1 g 1-1 of Pb and Cd respectively in nitrate form.Concentrated HN03 (BDH Aristar grade) was used to prepare 0.02 rnol 1-1 HN03 and 1 mol I-* FINO3 which was used for dilution and deproteinization of seminal fluid. The autosampler washing solution contained 0.2% v/v concentrated nitric acid (BDH Aristar grade) and 0.1% v/v Triton X- 100 (BDH). Precautions Against Contamination All the laboratory ware used (glass and plastic) was cleaned by soaking in 10°/o m/v HN03 for 24 h rinsed with de- ionized water soaked in 3% m/v Na2EDTA.2H20 solution for 24 h and again rinsed with de-ionized water. Such a washing procedure is essential also for the containers used for sampling of ejaculate and those for storage of seminal fluid. Because the concentrations of Pb and Cd in seminal fluid are generally very low and the sample volume is small the impact of possible contamination of a sample is relatively great.Sampling and Storage of Seminal Fluid The ejaculate was collected in a metal-free glass container by masturbation after 4 days of sexual abstinence. Follow- ing the semen liquefaction (approximately 15 min) sperma- tozoa were separated from seminal fluid by centrifugation at 3000 rev min-l for 10 min. Approximately 300 pl of the supernatant were transferred into an Eppendorf poly- propylene microtube and stored at -20 "C until required for analysis. Under the aforementioned storage conditions the concentrations of Pb and Cd in seminal fluid were found to be stable for more than 18 months. A large pool (approximately 100 ml) of the seminal fluid samples collected from non-smokers not occupationally exposed to either Pb or Cd was applied during the experiments on optimization of the method and for a matrix-matched calibration throughout the study.The seminal fluid pool was stored in a metal-free polypropylene bottle at -20 "C and the concentrations of Pb and Cd were found to be stable for a period of 2 years (all precautions were taken to prevent contamination e.g. aliquots were taken monthly by pouring instead of pipetting from the pool). The human seminal fluid samples analysed for Pb and Cd in the present study were obtained from 121 healthy volunteers (aged 20-43 years) 35 subjects with no occupa- tional exposure to either Pb or Cd and 86 subjects with long-term slight to moderate occupational exposure to inorganic Pb (printing works a factory for Pb products a ceramics factory a factory for Pb-based paints and a storage batteries factory).Out of the 121 subjects there were 79 cigarette smokers and 42 non-smokers (including nine former smokers). Analytical Procedure In an Eppendorf microtube 100 pl of a well mixed seminal fluid were added to 700 pl of 1 HN03 followed by the addition of 50 pl of 0.02 mol 1-' HN03 (the 'zero' Pb and Cd working standard solution). The microtube was capped and vortex mixed allowed to stand for 30 min centrifuged at 14000 rev min-l for 15 min and the clear supernatant was carefully poured into a polystyrene auto- sampler cup for subsequent analyses of Pb and Cd by electrothermal atomic absorption spectrometry (ETAAS).Each sample was prepared and analysed in duplicate. (All the volumes indicated above can be halved if necessary i.e. the minimum sample volume required for determina- tion of Pb and Cd in seminal fluid by the present method is The matrix-matched samples for calibration were simul- taneously prepared in the same way as the unknown sample by using 100 pl of a pooled seminal fluid and 50 pul of a working standard solution containing a mixture of Pb and Cd in 0.02 mol 1-1 HN03. The working standard solutions (prepared from an intermediate standard contain- ing 100 mg 1-' of Pb and 2 mg 11' of Cd in 0.02 mol 1-' HN03) had concentration ranges of 25-500 pg 1-1 of Pb and 0.5-10 pg 1 - I of Cd and were found to be stable at room temperature for 6 months.(Each working standard solution was always prepared and stored in the same glass bottle which was previously only rinsed with de-ionized water.) The working standard additions correspond to concentra- tions of Pb and Cd in seminal fluid of 12.5-250 and 0.25-5 pg l-l respectively whereas 0.02 moll-' HNO (the 'zero' working standard) corresponds to 0 pg 1-' of Pb and Cd. A 'blank' sample was prepared in the same way as the unknown sample except that 100 pul of de-ionized water instead of seminal fluid were used and centrifugation was omitted. Blanks were prepared and analysed in triplicate. Experience has shown that all the aforementioned samples are stable for more than 48 h when stored in an autosampler cup sealed with Parafilm at 4 "C so it is convenient that the ETAAS analyses be performed on one day for Pb and on another day for Cd. The thermal conditioning of the graphite furnace was performed daily by running the optimum temperature programme (Table 2) without any sample injection ten times after which a sample for calibration with the highest standard addition was injected and the drying temperature and efficacy of thermal conditioning were checked.The precision [relative standard deviation (RSD)] of six succes- sive replicate measurements of integrated absorbance was usually 4 1 Yo for Pb and 4 1.5% for Cd ie. an RSD of <2%0 was regarded as being acceptable. 'Three replicate atomizations were run for each sample. The recalibration controls using a sample for calibration with the highest standard addition were performed after every 12 samples.Experience has shown that the same pyrolytic graphite coated graphite tube and L'vov platform can be used for approximately 400 firings without any notable change in sensitivity and precision. 50 pl.) Results and Discussion Optimization of Procedure Initial experiments on the optimization of the method were performed by using a L'vov platform in a grooved graphite tube. Considerable variations were observed in the time of appearance of atomic absorbance signals (Fig. 1 ) analytical sensitivity and the characteristic mass when using combina- tions of different L'vov platforms and grooved graphite tubes under otherwise identical experimental conditions. This can be attributed to the variable contact between the E'vov platform and the grooves producing various extentsJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY APRIL 1993 VOL.8 42 1 of electrical and conductive heating of the L'vov platform in addition to radiative heating of the analysed sample. Smaller variations were observed when using a L'vov platform in a non-grooved graphite tube (Fig. 2) which is in agreement with the results of other worker^.*^-^^ The latter combination was therefore applied throughout the study and the observed variations in the characteristic masses of Pb and Cd during a period of 2 years were < 12% which is regarded as satisfactory. However a tendency towards a gradual decrease in sensitivity (i.e. a slight decrease in the slopes of the calibration lines for Pb and Cd) was observed in the same period.Nitric acid solution was used for the dilution and deproteinization of seminal fluid to reduce the non-atomic background signals during measurements of Pb and Cd. Nitric acid also acts as a chemical modifier by reducing halide interferences. A concentration of 1 mol 1 - I HN03 was found to be optimum i.e. dilution of 100 pl of seminal fluid in 700 pl of 1 mol 1-1 HN03 plus 50 pl of 0.02 moll-' HN03 ('zero' Pb and Cd working standard or those containing a mixture of Pb and Cd) as it produced the lowest background absorbance signals (Fig. 3) coupled with relatively the best repeatability of the specific absorbance signals for Pb and Cd. The chosen concentration of 0.4 k I Timeis Fig. 1 Variation in Pb atomic signals of deproteinized seminal fluid spiked with Pb (final concentration of 263 pg 1 - I of Pb in seminal fluid) obtained by using two different combinations of L'vov platforms and grooved tubes P n a 0.2 0.1 0 2.5 Timels 5.0 Fig.2 Variation in Cd atomic signals of deproteinized seminal fluid spiked with Cd (final concentration of 5.6 pg I - I of Cd in seminal fluid) obtained by using (a) four repeated insertions of the same L'vov platform into the same non-grooved tube and (b) four different combinations of L'vov platforms and non-grooved tubes 0.30 0.25 0.20 0.15 $ 0.10 % 0 m .f! 0.05 L) 1.6 Qp 0 + 2 = 1.2 4- 0.8 0.4 - 0.30 - 0.25 - 0.20 - 0.15 0.10 ; 0.05 $ D m 0 0.5 1 .o 1.5 2.0 Concentration of HNOJmol I-' Fig. 3 Effect of nitric acid concentration on the non-atomic background signals obtained during measurement of (a) Pb and (b) Cd in seminal fluid. The concentrations shown refer to the HN03 solution used for dilution and deproteinization of seminal fluid (e.g.1 mol I - I HN03 is equivalent to a final concentration of 0.825 rnol I-' HN03 in the analysed sample) 1 mol 1 - I HN03 is equivalent to a final concentration of 0.825 mol 1-1 HN03 in the samples analysed for Pb and Cd. Similar findings on the optimum HNO concentration for deproteinization were reported by other workers for the determination of Cd in blood plasma16 and urine.I7 Table 1 gives the optimum instrumental conditions and Table 2 the optimum temperature programmes for the determination of Pb and Cd in deproteinized seminal fluid. Temperature programmes were optimized using a sample for calibration with the highest standard addition (equiva- lent to concentrations of 250 pg 1 - I of Pb and 5 pg 1-' of Cd in seminal fluid). Fig.4 shows the effect of the pyrolysis temperature on the specific absorbance signals for Pb and Cd fat the chosen atomization temperature of 1600 "C for Pb and 1400 "C for Cd) and Fig. 5 shows the effect of the atomization temperature on the specific absorbance signals for Pb and Cd (at the chosen pyrolysis temperature of 600 "C for Pb and 400 "C for Cd). Although the chosen atomization temperatures are higher than the optimum regarding the size of the Pb and Cd signals when measured Table 1 Instrumental conditions for the determination of Pb and Cd in seminal fluid using a Perkin-Elmer Zeeman/5100 spec- trometer Lead Cadm i u rn Hollow cathode lamp current/mA Wavelengthhm Spectral bandwidthlnm Peak evaluation Baseline offset correction timels Read delay time/s Integration time/s Injection volumelpl 8 4 283.3 228.3 0.7 0.7 Integrated Integrated absorbance absorbance 1 1 0.9 0.9 3.0 3.5 20 20422 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY APRIL 1993 VOL.8 -0 Q 0.20 2 4- 0.16 0.12 0.08 CI - 0.04 Table 2 Furnace temperature programmes for the determination of Pb and Cd in seminal fluid. using a HGA-600 graphite furnace with L'vov platform in non-grooved pyrolytic graphite coated graphite tube -I 0.20 .$ - 0.16 - - 0.12 - - 0.08 - ( b ) - c - IL - - 0.04 Lead Cadmium ~ I - Temperature/ Ramp Hold Argon flow rate/ Temperature/ Ramp Hold Argon flow rate/ Step "C time/§ time/s mI min-I "C time/s time/s ml min-I Dry 160* I 45 300 160* 1 45 300 Pyrolyse 600 10 20 300 400 10 25 300 Cool-down 20 1 20 300 20 1 20 300 Atomize 1600 0 5 0 1400 0 5 0 Clean 2500 1 3 300 2500 1 3 300 Cool 20 1 4 300 20 1 4 300 'Varied between 150 and 180 'C on different days.as integrated absorbance the resulting profiles of the Pb and Cd peaks are considerably better (as indicated by the peak height absorbance data). The optimum temperature programmes for Pb and Cd include a 'cool-down' step prior to atomization because this ensures faster heating of the graphite tube and more constant conditions in the tube,I8 resulting in better repeatability of the Pb and Cd signals. The influence of non-spectral interferences on the deter- mination of Pb and Cd in seminal fluid was examined by comparing the calibration graphs for standards prepared In a different matrix obtained by integrated absorbance measurements under identical instrumental conditions.'Table 3 shows the relevant statistical data for the calibra- tion curves for Pb and Cd when using the standards prepared in A water; B 0.825 mol 1 - I HNO,; and C deproteinized seminal fluid containing 0.825 mol 1-' HNO i.e. matrix-matched standards. The standard addi- tions were equivalent to concentrations of Pb in seminal fluid of 0 25 50 75 100 125 and 250 pg 1-1 and concentrations of Cd of 0 0.5 1.0 1.5 2.0 2.5 and 5.0 pg 1-l. The original concentrations of Pb and Cd in the pooled seminal fluid used for calibration were 13 and 0.6 pg l-l respectively. All the calibration graphs were found to be linear.Although differences between the slopes A B and C 0.5 0.4 0.3 0.2 200 300 400 500 600 Tern perat u re/"C Fig. 4 Effect of pyrolysis temperature on the atomic signals of (a) Pb and (b) Cd of deproteinized seminal fluid in 0.825 an01 1-1 HN03 spiked with a mixture of Pb and Cd (final concentrations of 263 pg 1-1 of Pb and 5.6 pg 1 - l of Cd in seminal fluid) were small and were not significant between B and C for either Pb (PXl.05) or Cd (P>O.$O) the times of appear- ance of the peaks and the peak profiles were different (Fig. 6) and the matrix-matched calibration C is regarded as being optimum. By using a pool of human seminal fluid as a matrix for calibration solutions it was hoped that a time-consuming method of standard additions could be avoided (as proved to be so in our 10 years of experience in applying essentially the same method for determinations of blood Pb and Cd).Table 4 shows the slopes of the calibration graphs for Pb and Cd obtained by using ten different seminal fluid samples each measured on a different day. The observed variations in the slopes (RSD) of 3.4% for Pb and 3.9% for Cd are even smaller than the day-to-day variations in the slopes obtained by using the pooled seminal fluid viz. 5.1% for Pb and 5.8% for Cd (Table 4) indicating no need for a method of standard additions. However a matrix-matched calibration was chosen and the values for Pb and Cd in the pooled seminal fluid were obtained each day and used for internal quality control. The results obtained on 32 differ- ent days (meantSD) were 12.6kO.63 pg 1-' for Pb and 0.64+0.035 pg 1-l for Cd in the pooled seminal fluid corresponding to RSDs of4.9% for Pb and 5.4% for Cd.The main advantage of a matrix-matched calibration over the 0.6 0.4 0.2 2! n 0 m 5 0 a $ 0.24 t? Q P & 0.16 c.' 4- 0.08 0.6 0.4 0.2 8 e QI 0 1 0 0.24 3 03 Q) .K Y .- 0.16 8 0.08 O 3 0 800 1200 1600 Tern peratu re/"C Fig. 5 Effect of atomization temperature OA the atomic signals of (a) Pb and (b) Cd of deproteinized seminal fluid in 0.825 mol I - l HN03 spiked with a mixture of Pb and Cd (final concentrations of 263 pg I-* of Pb and 5.6 pg I-' of Cd in seminal fluid)JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY APRIL 1993 VOL. 8 423 Table 3 Statistical parameters of the calibration graphs for Pb and Cd obtained by using standards prepared in A water; B 0.825 mol I-' HNQ,; and C deproteinized seminal fluid in 0.825 moll-' HNO (matrixmatched standards) under identical instrumental conditions (no read delay integration time 5 s) Parameter A B c Lead- Correlation coefficient 0.9988 0.9989 0.9995 Slope/s 1 pg-l 8.2 105 x 1 0-4 8.6247 x 8.8613 x SEslo& 1 K3-l 1.2922 x 1 .0 1 6 8 ~ 7.7121 x tSl0pe 63.5 84.8 114.9 Significance of the t=2.5 19 P<0.05 t = 1.854 0.1 O>P>O.O5 difference between slopes ( t f) t=4.325 Pt0.01 Cadmium- Correlation coefficient 0.9955 0.9963 0.9963 Slopels 1 pg-l 1.9128~ 2.0295 x 2 . 0 5 4 4 ~ SEs,opds 1 Pug-' 5.7260~ 5.4965 x 4 . 0 7 8 0 ~ ts1ope 33.4 36.9 50.4 Significance of the t= 1.470. P>O.lO t=0.364 P>0.80 difference bet ween slopes (t P) t=2.014 O.lO>P>0.05 Table 4 Comparison of the slopes of the calibration graphs for Pb and Cd in seminal fluid obtained by using A ten different seminal fluids each measured on a different day; and €3 the pooled seminal fluid (other than A) measured on 32 days Slope/s 1 pg-l Sample No.1 2 3 4 5 6 7 8 9 10 Lead 9.678 x 8.879 x 8.605 x 9.052 x 9.245 x 10-4 9.176 x 8 . 7 1 4 ~ 1 0 - 4 8.828 x 1 0 - 4 8,931 x 8.963 x 1 0 - 4 A n=10 Mean = 9.007 x 1 O-* RSD = 3.4% (mo= 11.5 pg Pb)* SD=3.059 x B n=32 Mean=8.958 x SD=4.592 x RSD=5.1% (mo= 11.6 pg Pb)* A - B significance of the difference in slopes t=0.317 f 5 0 . 7 0 Cadmium 2.590 x 2.632 x 2.563 x 2.601 x 2.593 x 2.587 x 2 . 6 1 0 ~ 2.298 x 2.627 x 2.488 x Mean=2.559 x SD= 1.002 x 10-3 RSD= 3.9% (m,=0.40 pg Cd)* Mean=2.510x SD= 1.455 x 10-3 RSD=5.8% (m,-0.41 pg Cd)* t=0.981 P>0.30 *Mean value of the characteristic mass (m,) of Pb and Cd corresponding to the indicated mean value of the slope.method of standard additions is the considerably smaller volume of seminal fluid required for analyses for Pb and Cd. This is particularly important in pathological cases with reduced ejaculate volume taking into account the fact that several characteristic parameters of semen quality (e.g. the lactate dehydrogenase isoenzymes LDH-C4 fraction fruc- tose zinc acid phosphatase and citric acid) are usually also measured in the same seminal fluid specimen. Fig. 7 shows the matrix-matched calibration graphs for Pb and Cd in seminal fluid A when using a standard procedure i e .an injection volume of 20 pl and B and C when using a multiple injection (pre-concentration) proce- dure i.e. an injection volume of B 2 x 20 pl or C 3 x 20 pl. In B and C the autosampler was programmed so as to inject an additional 20 pl of the sample following the termination of the pyrolysis step of the previous aliquot(s). The optimum temperature programme (Table 2) was then continued as usual. In this way it is possible to inject a sample three times (C) without any notable change in precision (RSD) while at the same time the detection limit of the method (expressed as the concentrations of Pb and Cd in seminal fluid) is considerably lowered. However the results (Fig. 7) show that the calibration graphs are linear up to seminal fluid concentrations of approximately 263 pg 1 - I of Pb and 5.6 pg 1-1 of Cd using a standard procedure (A) 138 pg 1-1 of Pb and 3.6 pg 1-1 of Cd when using a double injection volume (B) and I 13 pg 1-1 of Pb and 2.6 pg 1-1 of Cd when using a triple injection volume (C) taking into account that the original concentrations of Pb and Cd in the pooled seminal fluid used for calibration were 13 and 0.6 pg l-l respectively.The multiple injection procedure (€3 and particularly C) is regarded as being convenient because very low concentrations of Pb and Cd in seminal fluid can be more precisely determined by applying the same samples for calibration and the same temperature programme whereas any additional step in the method could be a source of error. Analytical Parameters The detection limits of the method calculated as three times the standard deviation of ten measurements of a 'blank' sample are equivalent to concentrations in seminal fluid of 3.7 pg 1-1 of Pb and 0.16 pg I-' of Cd when using the424 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY APRIL 1993 VOL.8 0.4 1 0 Time/s Fig. 6 Variation in the atomic peak profiles of (a) Pb and (6) Cd obtained by using standards prepared in A water; B 0.825 moll-' HNO,; and C deproteinized seminal fluid in 0.825 mol I-' HNO (matrix-matched standard) under identical experimental condi- tions as shown in Tables 1 and 2 9 Addition to the concentration of Pb in seminal fluid/pg I-' % ( b ) -0 L rn $ 0.4 - 0.3 0.2 0.1 n I- /' I I 1 1 " 0 1.0 2.0 3.0 4.0 5.0 Addition to the concentration of Cd in seminal fluid/pg 1 Fig.7 Matrix-matched calibration graphs for (a) Pb and (b Cd in seminal fluid obtained by using A single injection (20 pl); B double injection (2 x 20 pl); and C triple injection (3 x 20 pl) of deproteinized seminal fluid in 0.825 mol 1 - I HN03 spiked with a mixture of Pb and Cd. (The original concentrations of Pb and Cd in seminal fluid are 13 and 0.6 pg l-' respectively) standard procedure (k an injection volume of 20 pl). The corresponding absolute detection limits are 8.8 pg of Pb and 0.38 pg of Cd. However when using an injection volume of 2 x 20 pl or 3 x 20 pl the detection limits of the method are equivalent to concentrations in seminal fluid of 2.0 pg 1-1 of Pb and 0.09 pg 1-1 of Cd or 1.4 pg 1-I of Pb and 0.05 pg 1-1 of Cd respectively. The precision (RSD) of the method calculated on the basis of seven replicate analyses of seminal fluid containing 26 pg 1-1 of Pb and 1.3 pg 1-1 of Cd was 5% for Pb and 6% for Cd when measured on the same day (within-day precision) and 7% for Pb and 10% for Cd when measured on different days (day-to-day precision). The day-to-day precision of the method was also calculated on the basis of duplicate analyses of 12 1 samples of seminal fluid covering the concentration ranges 4-48 pg 1-1 of Pb and 0.2-3.6 pg 1-1 of Cd and the RSDs were in the ranges 14-4% for Pb and 16-4% for Cd (i.e.at lower Pb and Cd concentrations in seminal fluid relatively higher RSDs were obtained). The recovery of the method was calculated on the basis of measurements carried out in 18 different seminal fluid samples.The samples originally contained 4.4-1 1.8 pg 1 - I of Pb and 0.21-1.93 pg 1 - I of Cd and were spiked with 25 50 75 or 100 pg 1-I of Pb and 0.5 1.0 1.5 or 2.0 pg 1-1 of Cd. Each measurement was performed in duplicate and the recoveries obtained (mean k SD) were 100.6 -t 3.6% (range 89-108%) for Pb and 98.8+4.2% (range 91-1 13%) for Cd in seminal fluid. The average characteristic masses of Pb and Cd obtained during a period of 2 years were 1 1.6 pg of Pb and 0.4 1 pg of Cd which are regarded as being in very good agreement with the results of other workers i.e. 12 pg of Pb and 0.35 pg of Cd19 or 10.9 pg of Pb and 0.42 pg of Cd.20 Lead in Seminal Fluid of Lead Workers and Control Subjects The concentration of Pb in seminal fluid was determined in 12 1 adult male volunteers 86 subjects occupationally exposed to Pb and 35 subjects with no occupational exposure to Pb.The results obtained expressed as median and range because of a skewed distribution were 15.3 (6.5-48.3) pg 1 - I for workers exposed to Pb and 8.6 (4.2-1 6.6) pg 1-1 in control subjects. The difference between the groups was highly significant according to the Mann-Whitney test (z= - 5.865 PC Our results for lead workers are considerably lower than the reported average of 148 pg 1 - I in ten Pb workers.Z1 However in that study no mention was made of either the method used or any data on control subjects. on the other hand our results for control subjects are similar to the reported range of 2-23 pg 1-1 in 45 adult subjects with no specified occupational exposure to Pb,' but are higher than those of Scandinavian worker^,^ viz.a reported mean 4 SD of 3.6 k 3.1 pg 1-1 in 87 adult subjects with no occupational exposure to Pb. However as indicated by the blood Pb levels it appears that the general population in Scandinavia is less exposed to Pb than the population in Croatia,22 which might explain the aforementioned difference in concentrations of Pb in seminal fluid. Cadmium in Seminal Fluid of Cigarette Smokers and Non- smokers The concentration of Cd in seminal fluid was determined in the same 121 adult male volunteers who had no occupa- tional exposure to Cd 79 cigarette smokers and 42 non- smokers. The results obtained expressed as median and range because of a skewed distribution were 0.86 (0.29-3.56) pg 1-I in smokers and 0.54 (0.17-1.67) pg 1-1 in non-smokers.The difference between the groups was highly significant according to the Mann-Whitney test (z= -4.683 Pc lo+). The results for both smokers and non-smokers reported in this paper are lower than the reported range of 0.15-10.00 pg 1-1 in 45 adult subjectsJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY APRIL 1993 VOL. 8 425 with no specified smoking habits or occupational exposure to Cd,7 but are higher than those of Scandinavian workers,'O viz. a reported meankSD of 0.28t-0.24 pug 1-1 in 27 smokers and 0.19-tO.21 pg 1 - l in 31 non-smokers. How- ever as indicated by blood Cd levels it appears that both smokers and non-smokers in Scandinavia are less exposed to Cd than those in Croatia,22v23 which might explain the aforementioned difference in concentrations of Cd in seminal fluid.Conclusions The described method for the determination of Pb and Cd in seminal fluid appears to be sufficiently sensitive and reproducible for research purposes with regard to the reproductive effects of exposure to Pb and/or Cd in men. As no reference materials with certified Pb and Cd levels in seminal fluid are available the accuracy of the method could not be adequately evaluated. However the recoveries of Pb and Cd obtained and the characteristic mass values of 11.6 pg of Pb and 0.41 pg of Cd which are in agreement with published value^,^^.^^ indicate the absence of interfer- ences in the method. This study was financially supported by the International Lead Zinc Research Organization (grant LH-378/ILZRO). References Fisher-Fischbein J.Am. J. Znd. Med. 1987 11 61 1. World Health Organization Recommended Health-based Lim- its in Occupational Exposure to Heavy Metals WHO Geneva 1980 pp. 22 and 37. Silbergeld E. K. Environ. Health Perspect. 1990 86 191. TeliSman S. PripiC-MajiC D. and Keiic S. Scand. J. Work Environ. Health 1984 10 239. Clarkson T. W. Nordberg G. F. and Sager P. R. Scand. J. Work Environ. Health 1985 11 145. Don Schrag S. and Dixon R. L. .4nnu. Rev. Pharmacol. Toxicol. 1985 25 567. 7 Pleban P. A. and Mei D. S. Clin. C'hirn. Acta 1983 133 43. 8 Umeyama T. Ishikawa H. Takeshima H. Yoshii S. and Koiso K. Fertil. Steril. 1986 46 494. 9 Saaranen M. Suistomaa U. Kantola M. Saarikoski S. and Vanha-Perttula T. Hum. Reprod.1987 2 475. 10 Saaranen M Kantola M. Saarikoski S. and Vanha-Perttula T. Andrologia 1989 21 140. 1 1 TeliSman S. CvitkoviC P. Gavella M.. and PongraEiC J. Abstracts of the International Symposium on Lead and Cad- mium Toxicology Shenyang Association for Science and Technology Shenyang 1990 p. 29. 12 TeliSman S. CvitkoviC P. RoEiC B. PrpiC-MajiC D. and Pizent A. in Heavy Metals in the Environment ed. Farmer J. G. CEP Consultants Edinburgh 1991 vol. 2 p. 13. 13 Shuttler I. L. and Delves €3. T. J. Anal. A t . Spectrum. 1987 2 171. I4 Shuttler I. L. and Delves H. T. J. Anal. At. Spectrom. 1988 3 145. I5 Shuttler I. L. Delves H. T. and Hutsch B. J. Anal. At. Spectrum. 1989 4 137. 16 Black M. M. Fell G. S. and Ottaway J. M. J. Anal. At. Spectrom. 1986 1 369. 17 Halls D. J. Black M. M. Fell G. S. and Ottaway J. M. J. Anal. At. Spectrum. 1987 2 305. 18 Falk H. Glismann A. Bergann L. Minkwitz G. Schubert M. and Skole J. Spectrochirn. Acta Part B 1985 40 533. 19 Slavin W. Sci. Total Environ. 1988 71 17. 20 L'vov B. V. Spectrochirn. Acta Part B 1990 45 633. 21 Chowdhury A.. R. Chinoy N. J. Gautam A. K. Rao R. V. Parikh D. J. Shah G. M. Highland H. N. Patel K. G. and Chatterjee B. B. Advance.7 in Contraceptive Delivery Systems 1986 2 208. 22 Assessment of Human Exposure to Lead and Cadmium Through Biological Monitoring ed. Vahter M. National Swedish Institute of Environmental Medicine and Department of Environmental Hygiene Karolinska Institute Stockholm 1982. 23 TeliSman S. AzariC J. and PrpiC-MajiC D. Bdl. Environ. Contam. 7oxicol. 1986 36 49 I . Paper 2/03 1291 Received June 15 I992 Accepted November 10 1992

ISSN:0267-9477    

DOI:10.1039/JA9930800419

出版商:RSC    

年代:1993

数据来源: RSC

摘要:

JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY APRIL 1993 VOL. 8 42 7 Low Pressure Inductively Coupled Plasma Source for Mass Spectrometry* E. Hywel Evans? and Joseph A. Caruso University of Cincinnati Department of Chemistry Cincinnati OH 45221 USA The generation of a low pressure Ar inductively coupled plasma has been successfully performed without any modification to the torch box of a commercial inductively coupled plasma mass spectrometer. A water-cooled low pressure torch and interface designed to allow sampling for mass spectrometry are described. Using gas chromatography sample introduction it was possible to measure the analytical signal for a sample of 1 -bromononane. The background spectrum contained many of the polyatomic ions seen for atomospheric plasmas thought to be due to small leaks in the vacuum seals.Keywords Low-pressure plasma; inductively coupled plasma mass spectrometry; gas chromatography The pioneering work on inductively coupled plasma mass spectrometry (ICP-MS) was performed by Houk et a1.l and Date and G r a ~ . ~ - ~ Later Douglas and co-~orkers~-~ investi- gated microwave induced plasma mass spectrometry (MIP- MS). Subsequently the MIP source has received some attention in conjunction with MS8 and has been used for gas chromatography (GC) detecti~n,~-l~ pneumatic nebulisa- tion13-19 and electrothermal vaporization (ETV)20-22 sample introduction. The advantage of the MIP is the ease with which plasmas can be formed in a number of gases particularly He resulting in a mass spectrum which is less prone to Ar-containing polyatomic ions.An alternative to the generation of plasmas at atmo- spheric pressure is to generate them at low pressure. In this way air can be excluded the gas flows much reduced considerably lower power used to sustain the plasma and ICPs can be generated using He 02 N2 and other gases with much greater ease. Low pressure ICPs have been investi- gated for emission spectr~metry,~~J~ and low pressure MIPS for MS.25-28 The application of a low pressure plasma in MS is particularly attractive since the exclusion of air and consequent reduction in interferences on P and S would make it a highly sensitive and selective GC detector for these elements and indeed this has been investigated using a low pressure IVIIP.~~ This paper describes preliminary results obtained using a low pressure Ar ICP which has been designed so that very little modification to an existing ICP-MS instrument is required for operation.Experimen tall Mass Spectrometer All experiments were performed using an ICP-mass spectrs- meter (VG PlasmaQuad VG Elemental Winsford Chesh- ire UK). The pumping capacity of the expansion stage was increased from approximately 400 to 1900 1 min-' by addition of a second port on the opposite side of the interface to the existing expansion port and linked to a 1500 1 min-l rotary pump (Edwards E1M-80 Edwards High Vacuum Crawley Sussex UK). The extraction lens voltage supply was replaced with an alternate supply (Keithley 247 High Voltage Supply Keithley Instruments Taunton MA USA) variable between + 999 and - 999 V.The ion lenses were tuned by monitoring A r k + at mlz 80 and optimizing them for maximum signal. No problems *Presented at the Eighteenth Annual Meeting of the Federation of Analytical Chemistry and Spectroscopy Societies Anaheim CA USA October 6-1 I 1991. ?Present address University of Plymouth Department of Environmental Sciences Drake Circus Plymouth UK PL4 8AA. were encountered when tuning the lenses and voltages were similar to values found for atmospheric pressure plasmas with the exception of the extraction lens. The extraction lens was an exception because no distinct optimum could be found for the dry atmospheric pressure plasma with a more negative voltage resulting in greater signal but also greater continuum background. A compromise setting of - 190 V was used.This phenomenon was typical for the instrument used in this study. For the low pressure Ar plasma an optimum could be typically found at - 150 V and for the low pressure He plasma at between -80 and -95 V. No continuum background was observed with these plasmas. Data acquisition parameters for the full mass range scan were mass range 4-90 mlz; dwell time 160 ps; sweeps 100; and channels 4096. Data acquisition parameters for the chromatographic temporal scan were mass 79 mlz; and dwell time 500000 ps. Low Pressure ICP Interface A low pressure ICP interface was constructed as shown in Fig. 1. The interface was similar to that described by Creed and ~ o - w o r k e r s . ~ ~ ~ ~ ~ who studied low pressure microwave plasma mass spectrometry except that in this work the sampler was designed to accept a modified ICP torch.The sampling cone was fabricated from aluminium (University of Cincinnati Department of Chemistry Machine Shop) such that a 0.75 in ultra-Torr fitting was incorporated into it (Fig. 1). Ultra-Torr fitting Coolant out 4 Outer Heated transfer gas Low Diessure I t ! I Make-up gas I 1 Coolant in sampler Liquid cooled 7r- ICP torch E?%Q Fig. 1 and the gas chromatograph coupled to the low pressure interface Schematic diagram of the low pressure sampler and torch428 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY APRIL 1993 VOL. 8 Table 1 Dimensions of the low pressure sampler and ICP torch Torch- Water jacket 0.d. (mm) 19 Water jacket i.d. (mm) 16.5 Intermediate tube o.d./mm 16 Spacing between intermediate and outer tubeslmm x0.5 Injector tip i.d./mm 1 .o Sa mpier- Ultra-Torr i.d./in Sampling orifice i.d./mm Sampling orifice depthlmm 3 4 1 .O 2.0 Table 2 Operating conditions for GC Injector temperaturePC 250 Start temperature/"C 110 Final temperature/"C 200 Heating ratePC min-l 32 Transfer line temperature/"C 270 He column gas flow/ml min-' 3.0 Solvent venting time/s 75 A water-cooled ICP torch was fabricated (Precision Glassblowing of Colorado Englewood CO USA) such that it could be fitted into the low pressure sampler and a vacuum seal made using the ultra-Torr fitting (Fig.1). The external diameter of the torch was similar to that of a standard ICP torch so that the existing load coil and torch box on the ICP-MS instrument could be used. However it was necessary to adjust the distance between the load coil and the sampler to 30 mm to accommodate the ultra-Torr fitting. The dimensions of the low pressure sampler and torch are listed in Table 1.The cooling fluid around the torch was maintained at 10 "C. Gas Chromatograph The gas chromatograph consisted of an HP 5700 oven with a UNI K- 10 direct splitless injector. The column was a DB- 5 (40 mx0.25 mm i.d.) with a film thickness of 25 pm (J&W Scientific Austin TX USA). A six-way valve (Valco Houston TX USA) was incorporated to allow venting of the solvent. The transfer line between the gas chromato- graph and the ICP torch consisted of 1 m x & in i.d. stainless-steel tube wrapped with heating tape insulated with fibre-glass tape and maintained at a temperature of approximately 270 "C.The transfer line was connected to the rear of the torch by means of an ultra-Torr fitting (Fig. 1) and a stainless-steel T connector. Make-up gas was introduced through the stainless-steel T connector. Operat- ing conditions for GC are shown in Table 2. Reagents A stock solution of 1-bromononane 1035 pg g-l (Halogen Chemical Columbia SC USA) was prepared in hexane and subsequently diluted in methanol (HPLC grade Fisher Scientific Cincinnati OH USA). Liquid Ar (Wright Broth- ers Cincinnati OH USA) and 99.999% purity He (American Air Liquide Countryside IL USA) were used throughout. Results and Discussion Optimization For the low-pressure Ar plasma there was no necessity to use the auxiliary gas flow and a stable plasma could be maintained using a coolant gas flow of between 0.3 and 0.8 12 10 a 6 4 2 2 3 10 - 5 - 200 300 400 500 600 100 inner gas flow/mi min-' Fig. 2 Effect of (a) outer gas flow rate; and (b) inner gas flow rate at forward powers of A and C 350 and B 450 W using an Ar low pressure plasma and expansion stage pumping rates of A and B 1900 and C 400 1 min-* for 25 ng of Br injected as 1.-bromononane.Signal monitored at mlz 79 1500 rn 3 0 > C Q) + P *z 1000 c w .- - (D 0 iTj 500 0 50 100 150 Time/s Fig. 3 Signal for 25 ng of Br as 1-brornononane monitored at m/z 79 1 min-' and an inner gas flow of 0.15 1 min-l at a power of CIS0 W. Under these conditions the plasma still resembled am atmospheric pressure plasma with a clearly discernable torus-like structure ( i . ~ .the inner gas flow 'punched' the plasma). An optimization was performed by taking 2pl injections of 12.5 pug g-l of Br (25 ng of Br absolute) as a solution of 1 -bromononane in methanol monitoring mlz 79 and integrating the resultant peak.JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY APRIL 1993 VOL. 8 429 lx108 iX1o7 1x106 iX1o5 1x10~ 1x10~ 1x102 E 10 Y Y > 1 3 0 - .- 0 ; 1x108 .S iX107 .- - (II v) 1x106 1x10~ iX1o4 iX1o3 1x102 10 1 40Ar+ (skipped) / I ( b ) 14 16 + "Ar' r 30;i+ O (Skipped) 0 5 10 15 20 25 30 35 40 m/z Fig. 4 Background spectra between mlz 4 and 40 for ( a ) Ar atmospheric pressure plasma; and (6) Ar low pressure plasma The results of the optimization for the outer gas are shown in Fig. 2(a). As can be seen the outer gas flow had a considerable effect on peak area signal the best signal being achieved at the lowest gas flow of 0.3 1 min-l at both 350 and 450 W power.Similarly a lower inner gas flow rate was preferable [Fig. 2(b)] though in this case an optimum was observed at 0.25 1 min-' for both 350 and 450 W forward power. Presumably the higher gas flows caused a decrease in the plasma temperature though this was not offset by increasing the power as would be expected. A much greater Table 3 Operating conditions for atmospheric and low pressure plasmas Parameter Ar atmospheric Ar low pressure He low pressure Plasma- Forward power/W Reflected powerlW Inner gad1 min-l Intermediate gad1 min-' Outer gad1 min-I Sampling depth*/mm 1350 3 50 100 < 10 30 50 0.65 0.15 0.063 1.0 0 0 14 0.35 0 10 30 30 Mass spectrometer- Sampler material Ni A1 A1 Skimmer material Ni Ni Ni Sampler-skimmer spacinglmm 5.5 7.5 7.5 Expansion pump11 min-' 400 1900 1900 Expansion pressure/mbar 2x 100 2 x 10-1 2x 10-2 Intermediate pressurelmbar t i x 10-4 ti x 1 0 - 4 ti x 10-4 Analyser pressure/mbar 6 x 7x lo-' tl x 1 0 - 8 *Sampling depth is defined as the distance between the outside surface of the sampling orifice and the foremost turn of the load coil.430 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY APRIL 1993 VOL.8 28 29 4 0 ~ ~ 1 6 0 1 H+ Si Si+ 1x106 lx1o5 1 ~ 1 0 ~ 1x10~ 1x102 g 10 .- 5 1 3 0 Y m l X l O 6 1 40Ar40Art 1x10~ lX1o4 1x10~ 1x102 10 1 (skipped} I 40 45 50 55 60 65 70 75 80 85 90 mlr Fig. 5 Background spectra between rnlz 40 and 80 for (a) Ar atmospheric pressure plasma; and (b) Ar low pressure plasma effect on signal was caused by adding the 1500 1 min-l pump to the expansion stage to result in a potential pumping capacity of 1900 1 min-l [Fig.2(a) and (b)]. At a pump rate of 400 1 min-l the signal was much reduced compared with a pump rate of 1900 1 m1n-l. Observation of the tip of the skimmer was possible by means of a quartz window set into the side of the expansion chamber. When a small amount of air was allowed to leak into the low-pressure system the characteristic red Ha emission at 656.3 nm from moisture in the air was clearly visible. However no definite 'barrel shock' region was visible. An example of a chromatogram obtained for 1 -bromono- nane using the operating conditions listed in Table 3 is shown in Fig.3. For an injection of 25 ng of Br as l-bromo- nonane a peak area of 225000 counts was observed which was reproducible from day to day though no attempt was made to optimize the chromatographic conditions. Peak tailing could have been caused when the column effluent was switched to the low-pressure environment causing a temporary increase in the column flow rate. The advantage of the low pressure plasma is that all of the GC effluent and hence all the analyte must pass through the sampling orifice into the expansion stage. If the skimming conditions are suitably optimized it might then be possible to achieve lower detection limits. Background Spectra An example of the background spectrum obtained for the low pressure plasma with Ar as the support gas i s shown in Figs.4(b) and 5(b) and compared with the spectrum obtained for Ar at atmospheric pressure [Figs. 4(a) and 5(a)]. Spectra were obtained using the operating conditions shown in Table 3. The spectra between rnlz 4 and 40 for the plasmas studied are shown in Fig. 4(a) and (b). The spectrum for Ar at atmospheric pressure [Fig. 4(a)] was obtained with a dry plasma and contained many of the so-called gas peaks due to Q2 Nz and moisture entrainment from the air and as contamination in the liquid Ar supply. Lithium Na Mg and K peaks were due to contamination on the torch from previous experiments. The A1 peak was due to A1 in the cleaning agent used to polish the Ni cones. Many of the polyatomic peaks are difficult to assign but from a comparison of the atmospheric pressure plasma [Fig.4(a)] with the low pressure plasma [Fig. 4(b)] the gas peaks are generally of the same order of magnitude in intensity for both plasmas. The only substantial differences are those for OH+* at mlz 18 and 20 which were less intense in the low pressure plasma and WC2 and 320+2 which were more intense though by less than an order of magnitude. Also the Ar+ signal was much smaller in theJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY APRIL 1993 VOL. 8 43 1 low pressure plasma presumably because of the reduced power and gas flows. No peak due to *'Al+ was observed in the low pressure spectrum [Fig. 4(b)] despite the fact that the sampler was made of Al. Erosion of the sampler is unlikely at the low power studied but the absence of 27Al+ also indicated that there was no intense secondary discharge in the expansion stage.The most likely explanation for the persistence of the gas peaks for the low pressure plasma is that a small leak in the interface between the gas chromatograph and the torch or between the torch and the sampler allowed minute amounts of atmospheric air into the system all of which were consequently sampled. The most likely point of entry would be the interface between the hot transfer line from the gas chromatograph and the torch where the ultra-Torr seal could have suffered from thermal degradation. A comparison of the spectra at rnlz 40-80 shows that the atmospheric pressure plasma [Fig. 5(a)] suffered from interferences due to Cr+ Ni+ and Cu+ which may have been caused by a secondary discharge in the expansion chamber whereas the low pressure plasma [Fig.5(b)] had greater interferences due to 40Ar14N+ and ArAr+. A signifi- cant difference between the atmospheric and low pressure plasmas is the absence of the continuum background of approximately 10 counts in the latter case. The continuum background was a characteristic of the instrument used in this study when operating a dry atmospheric plasma. The deconvolution of many of the polyatomic interfer- ences is difficult especially since the low pressure plasma has not yet been fully optimized with respect to sampling depth sampler-skimmer spacing and sampling orifice size. It is possible for instance that the formation of polyatomic ions was due to skimming downstream of the mach disc. Helium Plasma One advantage of the low pressure plasma is the ability to form He and molecular gas plasmas easily and it was indeed possible to do this using the conditions given in Table 3.However it was not possible to increase the forward power above 100 W owing to the extremely high reflected power of 50 W but higher powers should be practical if a suitable matching network is used. The main characteristic of the low pressure He plasma was the much increased intensities of polyatomic species between mlz 4 and 40 compared with the atmospheric Ar plasma. This was presumably due to a number of factors such as the extremely low power which probably allowed greater formation of polyatomic species; the more efficient ioniza- tion of O2 [ionization potential (IP)= 13.62 V] N2 (IP= 14.53 V) and H2 (IP= 13.6 V); and the unoptimized nature of the interface which meant that the expansion gases could have been sampled from a region where recombination of the extracted gases was prevalent. Simi- larly the spectrum between rnlz 40 and 80 contained much larger peaks due to SiO+ compared with the atmospheric pressure Ar plasma.One advantage of the He low pressure plasma may be for coupling with laser ablation ETV hydride generation or other methods of gaseous sample introduction since the attenuation of 40Ar+ 40Ar40Ar+ and other Ar-containing polyatomic ions could allow the deter- mination of 39K+ 40Ca+ and isotopes of Se. Conclusions A low pressure Ar plasma was successfully generated and used with mass spectrometry. Using GC a sample of 1 -bromononane was analysed.Plasmas were easily gener- ated and sustained at low pressure in both Ar and He but the total exclusion of atmospheric air was difficult and depended on the effectiveness of several vacuum seals. Future work in this area will include the simplification of the torch design and GC interface to reduce the number of vacuum seals necessary; generation of higher power He plasmas by improving the tuning of the RF circuit; and generation of plasmas in other gases such as 02 N2 and air. The authors are grateful to the National Institute of Environmental Health Sciences for providing support through grants numbered ES03221 and ES04908. We are also thankful to the National Institute of Health Shared Instrument Grant Program for providing the VG Plasma- Quad.1 2 3 4 5 6 7 8 9 10 I I 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 References Houk R. S. Fassel V. A. Flesch G. Svec H. J. Gray A. L. and Taylor C. E. Anal. Chem. 1980 52 2283. Date A. R. and Gray A. L. Analyst 1981 105 1255. Date A. R. and Gray A. L. Analyst 1983 108 159. Gray A. L. and Date A. R. Analyst 1983 108 1290. Date A. R. and Gray A. L. Spectrochim. Acta Part B 1983 38 29. Douglas D. J. and French J. B. Anal. Chem. 1981 53 37. Douglas D. Quan E. S. K. and Smith R. G. Spectrochim. Acta Part B 1983 38 39. Satzger R. D. Fricke F. L. Brown P. G. and Caruso J. A. Spectrochim. Acta Part B 1987 42 705. Brown P. G. Davidson T. M. and Caruso J. A. J. Anal. At. Spectrom. 1988 3 763. Creed J. T. Mohamad A. H. Davidson T. M. Ataman G. and Caruso J.A. J. Anal. At. Spectrom. 1988 3 923. Suyani H. Creed J. T. Caruso J. A. and Satzger R. D. J. Anal. At. Spectrom. 1989 4 777. Mohamad A. H. Creed J. T. Davidson T. M. and Caruso J. A. Appl. Spectrosc. 1989 43 1127. Wilson D. A. Vickers G. H. and Hieftje G. M. Anal. Chem. 1987,59 1664. Satzger R. D. Fricke F. L. and Caruso J. A. J. ,4nal. At. Speclrom. 1988 3 319. Creed J. T. Davidson T. M. Shen W. Brown P. G. and Caruso J. A. Spectrochim. Acta Part B 1989 44 909. Shen W. Davidson T. M. Creed J. T. and Caruso J . A. Appl. Spectrosc. 1990 44 1003. Shen W. Davidson T. M. Creed J. T. and Caruso J. A. Appl. Spectrosc. 1990 44 10 1 1. Heitkemper D. Creed J. T. and Caruso J. A. J. Chromatogr. Sci. 1990 28 175. Shen W. L. and Satzger R. D. Anal. Chem. 1991,63 1962. Satzger R. D. and Brueggemeyer T. W. Mikrochim. Acta 1989 111 239. Satzger R. D. J. Microwave Power Electromagnetic Energy 1989 24 132. Evans E. H. Caruso J. A. and Satzger R. D. Appl. Spectrosc. 1991,45 1478. Miller D. C. Seliskar C. J. and Davidson T. M. Appl. Spectrosc. 1985 39 13. Seliskar C. J. and Warner D. K. Appl Spectrosc. 1985 39 181. Creed J. T. Davidson T. M. Shen W. and Caruso J. A. J. Anal. At. Spectrom. 1990 5 109. Olson L. K. Story W. C. Creed J. T. Shen W. and Caruso J. A. J. Anal. At. Spectrom. 1990 5 471. Story W. C. Olson L. K. Shen W. Creed J. T. and Caruso J. A. J. Anal. At. Spectrom. 1990 5 467. Eberhardt K. Buchert G. Herrman G. and Trautmann N. Spectrochim. Acta Part B 1992 47 89. Paper 2/04 I2 7H Received July 31 I992 Accepted Novem ber 2 7 I 992

ISSN:0267-9477    

DOI:10.1039/JA9930800427

出版商:RSC    

年代:1993

数据来源: RSC

摘要:

JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY APRIL 1993 VOL. 8 433 ‘Zone Model’ as an Explanation for Signal Behaviour and Non-spectral Interferences in Inductively Coupled Plasma Mass Spectrometry Frank Vanhaecke and Richard Dams* Laboratory of Analytical Chemistry Ghent University Institute for Nuclear Sciences Proe ftuinstraat 86 8-9000 Gent Belgium Carlo Vandecasteele Department of Chemical Engineering Katholieke Universiteit Leuven de Croylaan 46 B-300 1 Heverlee Belgium The zone model is a simplified representation of the plasma resulting from the findings of an optimization study for a VG PlasmaQuad PQ1 inductively coupled plasma (ICP) mass spectrometer (VG Elemental Winsford Cheshire UK). According to this model for every nuclide there is a zone in the central channel of the ICP where a maximum density of singly charged ions occurs.The position of such a zone of maximum M+ density is a function of the mass number of the nuclide and the zone can undergo a spatial displacement under the influence of an alteration of an instrumental parameter or the introduction of a different matrix. This representation not only enables an explanation of a large number of observations from the optimization study but also allows an understanding of why both matrix induced signal suppression and enhancement were observed why for a given matrix the extent to which the signal intensities were altered differed from day to day and finally why the extent to which a signal is influenced by the matrix was seen to be a function of the mass number of the corresponding nuclide.Although the zone model might not completely reflect the genuine physical reality in all its facets it provides a phenomenological model for the variation of ion signals with mass number operating parameters and matrix composition. Keywords Inductively coupled plasma mass spectrometry zone model signal behaviour non-spectral interference matrix effect Inductively coupled plasma mass spectrometry (ICP-MS) is a versatile and powerful technique for trace and ultra-trace element determination mainly in aqueous solution. Since its introduction the technique has aroused great interest and has become increasingly popular and widespread in the analytical community. Although its usefulness for a variety of analytical tasks has been proven many fundamental properties of ICP-MS are not yet fully understood.Some aspects of signal behaviour and the origin of secondary-ion species eg. M2+ MO+ and MOH+ ions are still topics of discussion while conflicting reports exist on the influence of the matrix on signal intensities (non-spectral interfer- ences). In the present paper a simple model based on experiences with a VG PlasmaQuad PQ 1 ICP mass spectro- meter is presented allowing the user to understand and often predict many aspects of signal behaviour and matrix effects. Experimental The instrument used is a VG PlasmaQuad PQl ICP mass spectrometer (VG Elemental Winsford Cheshire UK) equipped with a Fassel torch a Gilson Minipuls-2 peristal- tic pump a Meinhard type Tr-30-A3 concentric glass nebulizer and a double pass Scott-type spray chamber with surrounding liquid jacket the temperature of which is controlled with a recirculating water refrigeration-heating system.Sampling cones (1.0 mm orifice) and skimmer cones (0.75 mm orifice) are made of nickel. The development of the zone model is based on the results of an optimization study and the results of a study of non-spectral interferences. Both studies are described in detail in previous publications. l,* *To whom correspondence should be addressed. Introduction of the Zone Model In an earlier optimization study it was established that the M+ signal intensity is strongly dependent on the nebulizer gas flow rate.’ At the optimum nebulizer gas flow rate the M+ signal exhibits maximum intensity and both decreasing and increasing the flow rate relative to this optimum value leads to a decrease in signal magnitude.This behaviour has also been established by several other worker^.^-'^ The observed behaviour can be explained as follows in the central channel of the ICP there is a zone where a maximum density of singly charged ions O C C U ~ S . ~ ~ - ~ ~ The position of this ‘M+ zone’ is a function of several instru- mental parameters and is strongly dependent on the nebulizer gas flow rate. Increasing the nebulizer gas flow rate results in a faster analyte transport through the plasma. If the residence time in the plasma required for ionization remains unaltered an increase in the flow rate causes a spatial displacement of the M+ zone in the direction of the sampling cone. An increase in the nebulizer gas flow rate could however also broaden the central channel of the ICP and hence decrease the temperature.Because of the temperature drop a longer residence time in the plasma would then be required for ionization so that this pheno- menon also causes the M+ zone to shift in the direction of the sampling cone. Whatever the relative contribution of both processes an increase in nebulizer gas flow rate leads to a spatial shift of the M+ zone in the direction of the sampling cone whereas a decrease in the flow rate leads to a spatial shift of the M+ zone in the opposite direction (Fig. 1). If the M+ zone is positioned far away from the sampling cone the signal intensity is low. By increasing the nebulizer gas flow rate the M+ zone is pushed fowards in the direction of the sampling cone and the signal intensity increases.A maximum signal intensity is obtained if the centre of the M+ zone is sampled. Finally a further increase in the flow rate leads to a decrease in the signal intensity since under these conditions the M+ zone cannot be434 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY APRIL 1993 VOL. 8 Load coil M' zone Sampling cone I I Fig. 1 Schematic representation of the dependence of the spatial position of the M+ zone on the nebulizer gas flow rate 1 decreasing nebulizer gas flow rate; 2 increasing nebulizer gas flow rate. For clarity the ICP and the central channel are not shown C C .- 50 100 150 200 .- Mass number of the nuclide CI 8 Fig. 2 The optimum nebulizer gas flow rate (at which a maximum signal intensity for M+ is obtained) as a function of the mass number of the nuclide efficiently sampled because it is pushed sideways away from the sampling cone.On plotting the optimum nebulizer gas flow rate (at which a maximum signal intensity for M+ is obtained) as a function of the mass number a gradual decrease is observed (Fig. 2). The heavier the element the lower the optimum nebulizer gas flow rate. A plot of the optimum nebulizer gas flow rate as a function of the first ionization energy shows no correlation. These results suggest that the position of the M+ zone depends on the mass number of the nuclide. Since for instance a higher flow rate is needed to obtain maximum signal intensity for beryllium than for uranium one can imagine the zone of maximum M+ density for uranium to be closer to the sampling cone than the zone of maximum density for beryllium.Generally the heavier the element the closer the position of the corresponding M+ zone to the sampling cone leading to the graphic representation of the 'zone model' in Fig. 3. Physical Background A probable explanation for the observed dependence of the optimum nebulizer gas flow rate on the mass number of the nuclide can be given by analogy to one used by Ronan et al.14 These workers showed that the efficiency of ion extraction from a d.c. discharge cell is proportional to the mass number. This relationship is a result of the fact that for a d.c. ion source all ions extracted are transported to Nebulizer gas -_l___t \ Tokh Plasma sampling cone Fig.3 Schematic representation of the plasma according to the zone model. The M+ zones for Be In and U and the centre of the respective zones are indicated. For clarity the M' zones are depicted over the full width of the plasma and not restricted to the central channel and the perturbation of the plasma as a result of the plasma sampling process is not shown the exit orifice of the d.c. cell by gas flow and that electrical extraction only becomes important at distances very close to the exit orifice. Under these conditions Ronan et a1.14 showed mathematically that the volume from which ions are extracted is a function of the mass number heavy ions coming from a larger extraction region than light ones. Of course equations for a glow discharge might not always be directly applicable to the ICP but there is evidence that an analogous reasoning holds for ICP-MS.As for a d.c. discharge cell it can be shown that the motion of ions in an ICP is dominated by gas flow and random diffusion. It is generally accepted that in the ICP-MS sampling process the gas flow extracts ions from the plasma into the vacuum system and the only opposition to this is random diffusion. Although the purpose of the extraction lens is of course not to extract positive ions from the plasma but to separate positive ions from negatively charged ions (and electrons) and neutral particles once they have passed the skimmer and to lead these positive ions into the rest of the electrostatic lens stack it is interesting to evaluate the possible (unintentional) effect of the applied voltage on the sampling process.Both the gas flow and the electric field would then act to extract ions from the plasma into the vacuum system and this would only be counteracted by random diffusion. In analogy to Ronan et al.14 it is possible to compare the critical radius r (the distance from the sampling cone at which an ion is equally likely to diffuse back into the plasma as it is to be extracted) for electrical extraction and transport by gas flow allowing the allocation of the dominant mechanism. Firstly an estimation is made of the critical radius r for electric field extraction. Taking into account that the Debye length (AD) defines the radius over which the electrical field of an ion shields an applied potential r has to be <AD or =AD to first order.with k= 1.38 x J K-* T,=electron temperature x 10,000 K,lS &,=permittivity of vacuum= 1/36n109 F m- 1 n,= electron density = 1 021 111-~ (ref. 15) and e- electron charge= 1 . 6 ~ C. Secondly by equalling the diffusion velocity and the gas velocity an estimation is made for rc for transport by gasJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY APRIL 1993 VOE. 8 435 flow. According to Ronan et a1.,I4 the diffusion velocity is given by 3 r rC i I r n where v= average particle velocity = 2 \i (3) 2nm T,=gas temperature ~ 5 0 0 0 K,IS and m=particle mass so that for Ar v=800 m s-*. 1 A,, = mean free path = J k n d (4) with n=number of particles m-3= m-3 (estimation assuming a plasma pressure of 1 atm (1 atm= 101.325 kPa) and a temperature of 5000 K),16 d=atomic dia- meter=3 x 10-IQ m so that Amfp=2 x m and D=diffu- sion coefficient. If in a first approximation the sampler surface is considered to be flat the critical volume around the sampler orifice defined by the critical radius is given by The time required to empty this volume is V/F with F=flow rate through the orifice =3 x 1W5 m3 s - ' .' ~ As a result in analogy to Ronan et al.,14 the gas velocity is given by the distance travelled (rc) divided by the time taken 3 F - rCF vss= - - V 2nr At the critical radius the diffusion velocity eqn. (2) and the gas velocity eqn. (6) are equal so that after rearrangement and use of the average particle velocity eqn. (3) ( 7 ) is obtained. Introduction of the appropriate values gives for argon rc=3 x m. Since the r obtained in this way is much larger than the r obtained for electrical extraction it is shown that transport to the sampling orifice is dominated by the gas flow regime while the ion optics (electrical extraction) do not come into play until after the skimmer.Under these conditions the dependence of the optimum nebulizer gas flow rate on the mass number of the nuclide can be explained as follows. For every nuclide the zone of maximum M+ density is defined by both the ionization and the diffusion processes. The idea is that there is an ionization region which attempts to increase the ion density and a diffusion region,which attempts to decrease the ion density. The overlap of these regions defines the 'true' ion density (M+ zone). If it is assumed that the ionization region is approximately the same for all elements (for a given nebulizer flow rate i.e. temperature profile) the lighter elements will show a maximum density lower in the plasma (i.e.at higher nebulizer flow rates) than do heavier elements since low mass ions diffuse radially more rapidly than do heavier ones (Fig. 4). Maximum density for low mass ions Low mass ion Plasma flow (increasing densi I Maximum density for high mass ions Fig. 4 Physical explanation for the dependence of the position of the Mf zone on the mass number of the corresponding nuclide. The ionization region is assumed to be approximately the same for all elements (for a given nebulizer gas flow rate i.e. temperature profile). Since the diffusion coefficient is inversely proportional to mass the light elements diffuse as a function of mass to a larger volume element of the plasma column than the heavy elements.Since the ion density is defined by both the ionization and the diffusion process the lighter elements show a maximum density lower in the plasma Application of the 'Zone Model' Observations from the optimization study During the optimization study,' it was found that the effect of a number of instrumental parameters on the signal intensity was mass dependent. This was the case for the nebulizer gas flow rate the r.f. power and the spray chamber temperature for example. Since the zone model postulates that the position of the M+ zone depends on the mass number of the corresponding nuclide while the zones can undergo a spatial displacement under the influence of an alteration of an instrumental parameter the observed behaviour can easily be rationalized. One particular case the dependence of the signal intensity on the spray chamber temperature is dealt with in detail.Analogous reasoning can be used for the other instrumental parameters. During the optimization study it was seen that the influence of the spray chamber temperature on the M+ signal intensity depends on the mass number of the element investigated. In the temperature range studied (4-22 "C) and using reference settings,' the lighter elements (e.g. beryllium or aluminium) show a maximum at an intermedi- ate temperature while for the heavier elements (e.g. thorium) the M+ signal intensity decreases gradually as a function of spray chamber temperature.This behaviour is illustrated for aluminium and thorium in Fig. 5. The water loading in the plasma is related to the spray chamber temperature.I7J8 The addition of water vapour into the plasma might have complicated consequences involving thermal effects conductivity changes and changes in the local. ionization conditions. Since undissociated water in the plasma represents a thermal buffer (energy required to vaporize/dissociate water17) raising the spray chamber temperature might cause the plasma to become cooler and less energetic as a result of higher water loading. Hence it can be assumed that a longer residence time in the plasma will be required for ionization so that the spray chamber temperature affects the position of the zones of maximum M+ density; increasing the spray chamber temperature shifts the M+ zone downstream in the plasma.If the initial position of the zones is such that the extraction conditions434 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY APRIL 1993 VOL. 8 - v) Y I _ _ Temperature/"C Fig. 5 M+ signal intensity (arbitrary units) for A Al+ and B Th+ as a function of the temperature of the spray chamber ( a 1 ( b ) Fig. 6 Explanation for the dependence of M+ signal intensities on the spray chamber temperature [temperature increasing from (a) to (c)] according to the zone model. The M+ zones for A1 and Th and the centre of these respective zones (0 Al; + Th) are indicated on the graph U 0.70 0.75 0.80 0.85 0.90 a)" Nebulizer gas flow rate/l min-' Fig. 7 CeO+ signal intensity (arbitrary units) as a function of the nebulizer gas flow rate for three different r.f.powers A 1200; B 1350; and C 1500 W are at an optimum for indium the centre of the M+ zone is positioned in front of the sampler for aluminium and behind the sampler for thorium [Fig. 6(a)]. As a result of the described spatial shift an increase in spray chamber temperature leads to gradually worsening extraction condi- tions for the 'heavy' elements [Fig. 6(a)-(c)]. For the light elements on the other hand at first the extraction condi- tions improve and the signal intensity increases and at an intermediate temperature the extraction conditions become optimum [Fig. 6(b)] and a maximum signal intensity is observed. Beyond this temperature also for aluminium an increase in spray chamber temperature leads to a gradual decrease in signal intensity.The MO+ signal shows essentially the same behaviour as a function of the nebulizer gas flow rate as the M+ signal but for the same power the MO+ signal peaks at a higher nebulizer gas flow rate. Hence the presence of an analogous maximum density zone for the oxide ion (MO+) located closer to the load coil can be assumed enabling a better understanding of the oxide-ion intensity and of the behav- iour of the ratio (MO+:M+) as a function of the nebulizer gas flow rate and the ref. power. This concept allows for instance the explanation of the effect of the r.f. power on the MO+ intensity. Fig. 7 shows that the attenuation of the MO+ intensity observed with increasing r.f. power at the reference nebulizer gas flow rate (approximately 0.725 1 min-l) is to be attributed to a shift of the MO+ signal behaviour plot rather than to an increased breakdown of MO+ ions.The observed behaviour of the M2+ signal intensity and the ratio (M2+:M+) however is more compli- cated as a result of the occurrence of a secondary or pinch discharge.19 Observations frorrt the study of non-spectral interferences The ICP-MS system suffers from non-spectral interferences i.e. signal suppression or enhancement caused by the matrix. Using a number of synthetic matrices of different types (acid organic and high-salt content) and one natural matrix (sea-water) the phenomenon of matrix effects was studied. An investigation into whether the use of an internal standard added to the blank standard and sample solutions would allow accurate correction was carried out.The results of this study have been described in detail in an earlier publication.2 The matrix effect was seen to be rather unpredictable; both signal suppression and enhancement occur while the extent to which a signal is altered by a given matrix can differ significantly from day to day. Further- more not all M+ signal intensities are influenced to the same extent; the signal suppression or enhancement ob- served was in all cases studied seen to be a function of the mass number of the nuclide. Although these non-spectral interferences have been discussed extensively in the litera- ture no consensus about their origin has been reached. Different mechanisms including ionization suppression,z0 ambipolar diffusion,*I increased collisional rate in the plasma22 and space charge effectsz3 have been postulated to be the origin of the observed effects. None of the mecha- nisms however was able to explain all observations. Particularly those cases in which 'heavy' nuclides are suppressed to a larger extent than the 'lighter' ones remain unclear when currently cited mechanisms are considered.Probably a complex combination of several mechanisms is at the origin of the observed behaviour. The zone model of the plasma however allows one to understand and explain qualitatively all observations if the assumption is made that the M+ zones can undergo a spatial displacement as a result of the introduction of another matrix. The validity of this assumption is illustrated by the spatial shift of atomic and ionic lines under the influence of the introduction of another matrix in ICP-atomic emission spe~trometry~~ and was confirmed by the following experiments.A series of synthetic and natural matrices was spiked with a number of elements at the 100 pg 1-1 level. For each matrix investigated the M+ signal intensity was recorded as a function of the nebulizer gas flow rate immediately after recording an analogous signal behaviour plot for the (spiked) reference matrix (0.14 mol I-' HN03). Figs. 8 and 9 illustrate that the optimum nebulizer gas flow rate and hence the position of the M+ zone are functions of the matrix composition. A similar behaviour has also been reported for a SCIEX Elan instr~ment.~~ Possibly both the ionization process (local ionization conditions) and t h e diffusion process (ambipolar diffusion) and hence the spatial distribution of analyte species depend on the matrix composition.Recently it has been shown thatJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY APRIL 1993 VOL. 8 437 0.5 0.6 0.7 0.8 0.9 Nebulizer gas flow rate/l min-’ Fig. 8 Th+ signal behaviour plots for two different matrices A 0.14 mol 1-1 HN03 (reference matrix) and B sea-water diluted 7- fold 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.1 Nebulizer gas flow rate/l min-’ Fig. 9 T1+ signal behaviour plots for two different matrices A 0.14 mol 1 - I HN03 (reference matrix) and B 0.5 mol 1-l CH3COOH the magnitude of the shift is a function of the matrix concent rat ion. 26 An important consequence of the observed shift of the signal behaviour plot is that the difference between signal suppression and enhancement becomes somewhat artificial; for the same nuclide and for the same matrix both signal suppression and signal enhancement can be seen depending on the nebulizer gas flow rate.On the other hand both suppression and enhancement can be observed for a given nuclide and at a fixed nebulizer gas flow rate depending on the direction and the magnitude of the shift of the signal- behaviour plot. The direction and magnitude of this shift are respectively a function of the nature of the matrix and the concentration of the matrix. In both Figs. 8 and 9 however an ‘absolute’ effect can be observed; in both cases the maximum signal intensity is seen to be a function of the matrix composition.Hence the extent to which the signal intensity is altered is of course also affected by this ‘absolute’ effect but the importance of the latter is often eclipsed by the effect on the signal intensity resulting from the spatial displacement of the M+ zones. In the frame of the zone model an ‘absolute’ signal suppression is to be attributed to a dilation of the corresponding M+ zone and/or a decreased ion density in the zone. An ‘absolute’ signal enhancement can then be attributed to a contraction of the corresponding M+ zone and/or an increased ion density in the zone. In the physical reality other mecha- nisms e.g. space charge effects23 and ambipolar diffusion,21 might play a major role in the ‘absolute’ effects. Hence the zone model is not meant to provide a complete and universal explanation for non-spectral interferences but it allows a better understanding of why (i) both signal suppression and enhancement can occur (ii) for a given matrix the extent to which signal intensities are altered can Fig. 10 Explanation for non-spectral interferences according to the zone model.The M+ zones for Li and U and the centre of these respective zones (0 Li; + U) are indicated. See text for details of ( 4 5 (b) and (d I I I I I 0 50 100 150 200 250 Mass number of the nuclide Fig. 11 Residual signal intensity (relative to the signal intensity in a 0.14 moll-’ HN03 matrix) for several elements present in a sea- water matrix diluted l 0-fold differ from day to day and finally (iii) the extent to which a signal is influenced is a function of the mass number of the corresponding nuclide.The first point has already been illustrated the last two will be dealt with in the following paragraphs. The signal intensity is a function of the relative position of the M+ zone with regard to the sampling cone. The position of such a zone is affected not only by different controllable instrumental parameters e.g. the nebulizer gas flow rate and the r.f. power but also by a number of parameters which can hardly be adjusted or only with insufficient reproducibility. These parameters include the pressure in the expansion chamber the condition of the sampling cone and skimmer (aperture sizes) and the torch position. As a result even if a completely reproducible matrix-induced shift could be assumed the matrix effect would not be reproducible owing to day by day differing initial conditions.The zone model postulates that the position of the M+ zone is dependent (in a continuus way) on the mass number of the corresponding nuclide. Even if the introduction of a given matrix caused an equal spatial shift of the M+ zone for all elements for every nuclide the effect on the signal intensity would be a function of the initial and the ‘new’ position of the corresponding M+ zone relative to the sampling cone and hence be a function of the mass number. In this way the continuus variation of the matrix effect as a function of the mass number can be understood. Signal suppression with the heavy nuclides being more suppressed than the light ones can be explained as follows.If the position of the M+ zones relative to the sampling cone is initially such that the centre of the IisIn+ zone coincides with the apex of the sampling cone the ’Li+ zone is situated in front of the 238U+ zone behind the sampling cone [Fig.438 JOURNAL OF A.NALYTICAL ATOMIC SPECTROMETRY APRIL 1993 VOL. 8 0.60 0.65 0.70 0.75 0.80 0.85 0.90 Nebulizer gas flow rate/l min-’ Fig. 12 Be+ and TI+ signal behaviour plots for two different matrices A Be+ and C TI+ in 0.14 mol I-’ HN03 (reference matrix) and B Be+ and D TI+ in sea-water diluted 7-fold. At nebulizer gas flow rates between the values indicated by arrows T1+ signal intensities are suppressed to a greater extent than Be+ signal intensities as a result of introduction of the sea-water matrix 10(a)].If the introduction of a given matrix causes the M+ zones to shift in the direction of the sampling cone [Fig. lO(a)-+(b)] the extent of signal suppression will be much larger for uranium than for lithium since the distance between the centre of the M+ zone and the apex of the sampling cone increase much more for uranium than for lithium. In an analogous way signal suppression with the light nuclides being more suppressed than the heavy ones can be the result of a spatial shift of the M+ zones in the opposite direction [away from the sampling cone Fig. I O(a)-+(c)]. Finally besides being caused by the ‘absolute’ effects mentioned earlier a signal enhancement might also be the result of a spatial shift of the M+ zone if this shift leads to a better relative position of the M+ zone with regard to the sampling cone.However care must be taken in interpreting experimental observations using representations such as Fig. 10. This is illustrated by the observations made for a sea-water matrix. For this matrix it was established that (i) using the reference setting for the nebulizer gas flow rate the signals of ‘heavy’ nuclides are more suppressed than those of the ‘lighter’ ones (Fig. 11) and (ii) the signal behaviour plot (Fig. 8) shows a shift to a higher nebulizer gas flow rate so that a spatial displacement of the M+ zones away from the sampling cone or in the direction of the load coil can be assumed. In the light of the explanation illustrated in Fig. 10 these observations seem contradictory. For every gas flow rate the extent to which the signal intensity is suppressed or enhanced depends on the initial and the ‘new’ shape and position of the signal behaviour plot as illustrated for sea- water in Fig.12. Fig. 12 shows that although a spatial shift of the M+ zone in the direction of the load coil can be assumed at nebulizer gas flow rates between the indicated values signal suppression with the ‘heavier’ nuclides being more suppressed than the ‘lighter’ ones can be observed. Although Fig. 10 is only a schematic and simplified representation of the physical reality in which dilation/ contraction of the zones absolute suppression or enhance- ment effects etc. are not taken into account it holds for the simpler cases. Representations as used for sea-water (Fig.12) allow an understanding of all cases studied. References 1 2 3 4 5 6 7 8 9 10 1 1 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 Vanhaecke F. Vandecasteele C. Vanhoe H. and Dams R. Mikrochim. Acta 1992 108 4 1 . Vanhaecke F. Vanhoe H. Dams R. and Vandecasteele C. Tdanta 1992 39 737. Long S. E. and Brown R. M. Analyst 1985 111 901. Gray A. L. and Williams J. G. 9. Anal. At. Spectrom. 1987 2 599. Zhu G. and Browner R. F. Appl. Spectrosc. 1987 41 349. Horlick G. Tan S. H. Vaughan M. A. and Rose C. A. Spectrochim. Acta Part B 1985 40 1555. Vaughan M. A. and Horlick G. Appl. Spectrosc. 1986 40 434. Lam J. W. H. and Hsrlick G. Spectrochim. Acta Part B 1990,45 1327. Jakubowski N. Feldmann I. and Stuewer D. in Applications of Plasma Source Mass Spectrometry eds. Holland G. and Eaton A. N. Royal Society of Chemistry Cambridge 199 1 p. 79. Ross B. S. Yang P. Chambers D. M. and Hieftje G. M. Spectrochim. Acta Part B 1991 46 1661. Caughlin B. L. and Blades M. W. Spectrochim. ‘4cta Part B 1985,40 1539. Furuta N. Spectrochim. Acta Part B 1986 41 I 1 15. Vickers G. H. Wilson D. A. and Hieftje G. M. Spectrochim. Acta Part B 1990 45 499. Ronan G. Clark J. and Ketchell N. Mikrochirn. .4cta 1989 111 231. Douglas D. J. and French J. B. J. Anal. At. Spectrom. 1988 3 743. Houk R. S. and Thompson J. J. Mass Spectrom. Rev. 1988 7 425. Hutton R. C. and Eaton A. N. J. Anal. At. Spectrom. 1987 2 595. Zhu G. and Browner R. F. .I. Anal. At. Spectrom. 1988 3 781. Douglas D. J. and French J. B. Spectrochim. Acta Part B 1986 41 197. Olivares J. A. and Houk R. S. Anal. Chem. 1986 58 20. Gregoire D. C. Spectrochim. Acta Part B 1987 42 895. Beauchemin D. McLaren J. W. and Berman S. S. Spectro- chim. Acta Part B 1987 42 467. Gillson G. R. Douglas D. J. Fulford J. E. Halligan K. W. and Tanner S. D. Anal. Chem. 1988,60 1472. Blades M. W. and Horlick G. Spectrochim. Acta Part B 1981 36 881. Longerich H. P. J. Anal. At. Spectrom. 1989 4 665. Goossens J. De Smaele T. Moens L. and Dams R. Fresenius’ Z. Anal. Chem. Puper 2/048 740 Received September 10 I992 Accepted December 4 1992

ISSN:0267-9477    

DOI:10.1039/JA9930800433

出版商:RSC    

年代:1993

数据来源: RSC

摘要:

JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY APRIL 1993 VOL. 8 439 Qxychlorine Ions in Inductively Coupled Plasma Mass Spectrometry Effect of Chlorine Speciation as CI- and C1Q4- Henry P. Longerich Department of Earth Sciences and Centre for Earth Resources Research Memorial University of Newfoundland St. John's Newfoundland Canada A 16 3x5 Differences in the inductively coupled plasma mass spectra of 0.2 rnol I-1 HCI and 0.2 rnol I-' HC104 are reported. An enhancement was observed of the ion signal in the mass spectra of several oxychlorine species (CIO,+ HC102+ CIO3+ HC103+ HC104+ and H2Cl84+) in a solution of HC104 compared with the spectra of an equimolar solution of HCI. Spectra of Ca in HCIO4 showed enhanced formation of CaCIO+ CaCI02+ CaCIO,+ and CaCI04+ compared with equimolar solutions of Ca in HCI.Similarly solutions of Mg in HC104 showed enhanced formation of MgC102+ MgC103+ and MgC104+ compared with equimolar solutions of Mg in HCI. Other major rock-forming elements (Na Al K Sc Ti Mn and Fe) did not produce enhanced signals from oxychlorine pslyatomic ions. Keywords Inductively coupled plasma mass spectrometry; polyatornic ion; chlorine; alkaline earth metal; rocks In the course of analysis using inductively coupled plasma mass spectrometry (ICP-MS) of digests of rock materials prepared using HC104 previously unobserved interferences were found specifically that of 40Ca35C11604+ at rnlz 139 and 40Ca37C11604+ at rnlz 141 which caused significant interferences in the determination of 139La+ and 141Pr+. The use of HC1104 is advantageous in the digestion of geological samples as the ClO,- salts of most elements are soluble and the high temperatures obtained in the digestion aid in the complete oxidation of the sample material.Hence HC104 continues to be used by many workers in geological sample dissolutions. The observation of these polyatomic ions of Ca C1 and 0 along with axychlorine polyatornic ions in solutions of HClO containing no added Ca has interesting implications for the study of the physics of an ICP and processes that occur when ions are extracted into the vacuum of the mass spectrometer. The formation of these oxychlorine ions clearly demonstrates that the desire of analysts for the ICP to be a perfect ion source that is not affected by the speciation of the elements in a sample is not satisfied.Clear differences were observed in the ICP mass spectra of a solution of HCl compared with a solution of HC104 that was prepared at equimolar concentrations and thus contained nearly identical concentrations of the elements H 0 and C1. The alkaline earth elements Ca and Mg form a variety of oxychlorine polyatomic ions when aspirated in HCIO which are absent or decreased in spectra of solutions of Ca and Mg in HCI. This paper reports unambiguous evidence that some polyatomic ions observed in IGP mass spectra are related to the speciation in the original sample solution and are not entirely formed by reaction in the expansion stage of an ICP-MS system. Experimental Instrumentation The ICP-MS instrument was a SCIEX Perkin-Elmer Model 250 Elan.The early modifications of the instrument and description of the laboratory facilities have been reported. A computer system upgrade for Model 250 and 500 instruments which made available Model 5000 software to these instruments has been installed.* The upgraded computer system uses a Gilson Model 212B autosampler and version I .04-ICPS software.z A standard Scott-type spray chamber was used with a Meinhard C-2 nebulizer both of which were identical with those supplied with the original installation. The spray chamber was wrapped with Tygon tubing secured with black electrical tape through which tap water flowed at approximately 10°C thus forming a water 'jacket'. Reagents and Solutions Solutions of mixed acids were prepared using distilled HNO HCl and HC104.The MC1 was prepared from Fisher Reagent ACS grade concentrated acid. Equal volumes of concentrated acid and Nanopure type 1 grade water (17 Mi2 cm) were mixed and distilled in a sub-boiling quartz still (Quartz and Silice). The distilled 6 rnol 1-I acid was collected in a Nalgene 20 1 polyethylene carboy. The HC104 was prepared from Fisher Reagent ACS grade 70% stock solution which was distilled in a non-boiling two-bottle still.3 The bottles used were Nalgene Teflon fluorinated ethylene propylene (FEP) 1 1 bsttlesjoined with an in-house machined virgin poly(tetrafluoroethy1ene) (PTFE) connec- tor and heated with an infrared sun lamp. The HNO was prepared from Fisher Reagent Grade ACS concentrated acid. Equal volumes of concentrated acid and de-ionized water were mixed and distilled as for HC1.The concentra- lion of the resulting acid was 8 mol 1 - I . For ICP-MS analysis three mixed acid solutions were prepared in triplicate all of which were nominally 1 moll-' HNO (0.87-0.98 mol 1-l). The three solutions were ( I ) 1 mol 1-' HNO,; (2) 0.2 mol I-' HC1 (0.23-0.24 mol 1-l) and 1 mol I-* HNO,; and (3) 0.2 moll-' HC104 (0.19-0.2 1 moll-l) and 1 rnol l-'HNO,. Thus all mixed acid solutions were 1 rnol 1-' HNO with (1) no C1 (2) C1 as C1- or (3) C1 as C104-. Stock solutions containing nominally 100 or 200 yg g-' of each of nine geologically abundant elements (Na Mg Al K Ca Sc Ti Mn and Fe) were prepared from Plasma Grade powders (Spex Industries Metuchen NJ USA) in 1 moll-' HNO,. Aliquots of these stock solutions were mixed with aliquots of distilled 6 mol 1-l HCl distilled concentrated HC104 1 mol 1-' HWO or water as required producing solutions of the nine elements at nominal concentrations of 90 p g g-l (83-100 pg g-l) in (1) 1 mol 1-I HN03 (2) 0.2 moll-' HCl and 1 moll-' HNO and (3) 0.2 moll-l HC104 and 1 mol 1-I HNO,.To monitor the instrumental sensitivity and the formation of polyatomic ions Th was added to each solution at a nominal concentration of 100 ng g-' (88-126 ng g-'). The final concentrations of HN03 (nominally 1 moll-') varied over a limited range from 0.83 to 1.00 mol 1 - I . The HCI concentration in solutions (nominal concentration 0.2 mol 1-l) varied from 0.23 to 0.24 mol l-! and the HC104 concentration in solutions440 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY APRIL 1993 VOL.8 (nominal concentration 0.2 mol 1-l) were slightly more dilute varying from 0.19 to 0.2 1 mol 1-'. Mean crustal abundances of the major rock-forming elements are given in Table 1 This suite of oxides compose >99% of the earth's crust. The next most abundant element is Sr which has a mean crustal abundance of only 260 pg g-l. To cover a range of element groups Sc was included so that elements with oxidation numbers of 1 2 3 and 4 were included in the set of elements. In most procedures used to digest rock samples for the determination of trace elements using ICP-MS Si is removed. When HF digestion is used Si is volatilized as SiF4. In an Na20 sinter procedure Si is extracted in water as a soluble Na salt.4 Therefore potential interferences of Si were not studied.In our laboratory rock digests are now typically prepared at a final concentration of 0.5 g of rock per kilogram of solution (0.1 g of rock in 200 g of solution) a reduction from the higher concentration of 1 g kg-l previously reported and less than the higher concentrations of 2 g kg-l being used by other laboratories. For a typical crustal rock prepared at a final concentration of 0.5 g kg-l the resulting solution concentrations are given in Table 1. The total dissolved solids in a typical rock digest is thus approximately 270 pg g-l being higher for ultramafic rocks that contain less Si and lower in high-Si granitic rocks. The concentrations of these major rock-forming elements in solution that were determined in this study (90 pg g-l) are higher than would typically be found in solutions submitted for ICP-MS analysis (Table 1).study of the variation of the formation of various rare earth oxides as a function of operating conditions.s A typical instrument set-up for this facility is to set ThO+:Th+< When ThO+:Th+x lo% the interfer- ence of the oxide ions of Ba and the rare earth elements (especially Ba with Eu Pr with Gd and Nd with Tb) are small enough that accurate corrections can be made when analysing typical silicate rock samples. In a typical analysis this results in the instrument being operated at a Th sensitivity that is less than the maximum but always more than 50% of maximum sensitivity.8 In this study to enhance the formation of polyatomic ions a slightly higher ThO+:Th+ level was used.The mean ThO+:Th+ value for all samples was 13.4% with a range of 10.0-14.4%. For the spectra of the acids in the absence of other major elements the range was slightly smaller (12.6-14.4%). There was a gradual increase in ThO+:Th+ over the course of the data acquisition. Instrumental operating conditions are given in Table 2. To attenuate the count rate in the rnlz range 34-37 which contains the signals from the isotopes of C1 [35Cl (76%) and 37Cl (24O/o)] a rod offset (Omni Range) of 6.4 was used which attenuated the count rates by a factor of 78. The attenuation factor was determined from an analysis of a 0.02 moll-' solution of HCl. All count rates reported in the mlz range 34-37 are multiplied by the attenuation factor of 78 hence some reported ion signals exceed the maximum instrumental count rate of 3 x lo6 ions s-l.Count rate data was collected using the Elan software. The Elan software was used to create a 'report' file that Data Acquisition Mass spectra were obtained on each of 36 different solutions. Spectra were obtained on three acid solutions which contained (1) HN03 (2) HCl and HN03 and (3) HC104 and HN03. Spectra were also obtained on solutions containing added elements (Na Mg Al K Ca Sc Ti Mn and Fe) in each of the three acid mixtures. The autosampler was used without operator intervention. The instrument was optimized using a 100 ng g-l solution of Cs for which > 1 x lo5 ions s-l (> 1 x lo6 ions s-l for a 1 ,ug g-' solution) were obtained. It is imperative that some measure of the plasma temperature as measured by the formation of a polyatsmic ion be reported. For a variety of elements Horlick et aL6 initially showed that the monopositive elemental ion to metal oxide ion ratio varies over a large range as the operating conditions are changed.The applica- tion of Th and Tho to monitor polyatomic ion formation was first suggested by Lichte et aL7 for the determination of the rare earth elements. Our laboratory reported a detailed Table 1 Mean crustal abundances of the major elementsS and the concentrations of these elements in solutions digested at 0.5 g of rock per kilogram of solution Oxide Na,O MgO Si02 K2O CaO SC203 Ti02 MnO FeO Total A 1 2 0 3 Crustal abundance (%) 3.1 5.3 15.9 57.3 1 . 1 7.4 t0.01 0.9 0.2 9.1 100.3 Element Na Mg A1 Si K Ca s c Ti Mn Fe Solution elemental concentrat ion1 Pg g-' 1 1 16 42 0 5 26 0 3 1 35 273 Table 2 ICP-MS operating conditions Inductively coupled plasma- Plasma gas Forward powerlW Reflected powerlW Gas flow rated min-l- Plasma (outer) Auxiliary (intermediate) Nebulizer (inner) Sampling distance (load coil to sample aperture)/mm Sampling aperture/mm Skimmer aperture/mm B lens P lens E-1 lens S-2 lens Measurement mode Time per mass/s Dwell time/ms Sweeps per replicate Number of replicates Resolution Points across peak Delay times/+- Sample delay Wash delay Rabbit delay Interface- Ion lens settings1 V- Data acquisition parameters- Mass range scanned- Rod offset 0 0 6.4 (attenuation 78) 0 0 0 0 Argon 1200 <5 13.0 1 .o 0.87 21 mm I .1 (diameter nickel 0.9 (diameter) nickel 7.6 - 5.6 -13.3 0.0 Multi-channel 50 30 1 Normal 1 1.5 70 180 60 mlz 5-14 23-3 1 34-37 39 42-2 10 23 1-239 247-256JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY APRIL 1993 VOL.8 44 1 lX1o7 I 1x106 '; 1x10~ 2 iX1o4 .- Z iX103 5 1x102 C 0 7 v) C .- - m v) 10 1 13 HCI + HNO HCIO + HNO 0 HN03 ' 35 36 49 51 52 67 68 70 75 80 83 84 100 101 232248 mlz Fig. 1 Comparison of ion signals of acid solutions. For clarity only the polyatomic ions of the 3sCl isotope are shown. Note the greatly enhanced ion signals in the HClO. solution due to ClQ+ (m/z 67) HC102+ (mlz 68) C103+ (mlz 83) HCIO,+ ( m / ~ 84) HC104+ (mlz loo) H2ClO.4+ (mlz 101) contained the count rate data in an ASCII format. These files were translated using a compiled in-house written BASIC program which translated the data into a Lotus formatted spreadsheet file and saved the file to floppy disc. This translate program is executed under the SCIEX recommended DOS shell vpix. Results and Discussion Mass Spectra of HCl and HC104 Solutions The formation of various polyatomic ions of C1 was noted by most early ICP-MS workers owing to their interference with several important analytes especially 35C1160+ with 51V+ and 40Ar35C1+ with 75A~+.Tan and HorlickIo reported a complete and detailed mass spectrum of a 5% solution of HCl along with the interpretation of the polyatomic species which were observed in the spectrum. There are two isotopes of C1 [35Cl (75.8%) and 37Cl (24.2%)] two of H [lH (99.9%) and 2H (0.01%)] two of N [I4N (99.6%) and 15N (0.36%)] three of 0 [ l 6 0 (99.8%) 1 7 0 (0.08%) and l 8 0 (0.2%)] and three of Ar [36Ar (0.34%) 38Ar (0.07%) and 40Ar (99.6%)]. The species reported by Tan and Horlick were combinations of these isotopes. The C1-containing species reported were C1+ ClH+ ClN+ C10+ ClOH+ C102+ and ArCl+.The mass spectrum of HC1 (containing HNO,) deter- mined in this study is comparable to that reported by Tan and HorlickIo and therefore is not shown in detail. The interpretation of the molecular ions suggested by Tan and Horlick in HCl solution is maintained. Fig. 1 shows the spectral ion signals of species which contain the major Cl isotope 35Cl+ along with those of 36Ar+ 40Ar2+ 232Th+ and 232Th160+. Appropriate ion signals were recorded for the other combinations of the isotopes of the major elements in the solutions (H N 0 C1 and Ar).New species which are detected in the spectra of solutions of HC104 but are undetected in solutions of HC1 are shown in Fig. 1. A detailed discussion of the spectrum in Fig. 1 follows. The data in Fig. 1 are means of three analyses of each acid or acid mixture. mlz 35. The ion signals at mlz 35 in the C1-containing solutions are primarily due to T l + . As the concentrations of C1 in these two solutions are similar the ion signals are comparable for both the HCI and HC104 solutions. Some of the slightly higher ion signals in the HCl solution are due to the higher C1 concentration (0.24 mol 1-l) in the HCl compared with the HCD4 solutions (0.20 mol 1 - I ) . mlz 36. The ion signals at mlz 36 are comparable in all solutions and are from 36Ar+.Any enhancement due to ,TlH+ is small. These data are shown as they serve as a mediocre internal standard demonstrating that there was no large change in instrumental sensitivities between analyses. mlz 49. The ion signals at mlz 49 in C1-containing solutions are primarily from ,T114N+ with some contribu- tion from 37C112C+. The ion signals relative to that of 35Cl+ are comparable for the HC1 and HC104 solutions. mlz 51. The ion signals at rnlz 51 in the C1-containing solutions are primarily from 35C1160+. The ion signals relative to that of 35Cl+ are slightly enhanced in the HC104 solutions compared with the solutions of HCl. mlz 52. The ion signals at mlz 52 in the C1-containing solutions are primarily from 1H35C1160+. The ion signals relative to that of 35Cl+ are comparable for the HCl and HC104 solutions.mlz 67. The ion signals at mlz 67 in the C1-containing solutions are primarily from 3sC11602+. The ion signals in the HC104 solutions are dramatically enhanced by a factor of more than 300 compared with the solutions of HCl. This observation is very strong evidence that the C1-0 bonds in C104- are not entirely broken in the plasma and that these spectral artifacts are not due to reactions as the molar concentration of the elements C1 and 0 are nearly identical in both solutions. This observation is the first to suggest that the speciation of C1 in the sample (Cl- compared with Clod-) affects spectral ion signals. mlz 68. The ion signals at mlz 68 in the C1-containing solutions are primarily from 1H35C11602+.The ion signals in the HC104 solutions are clearly enhanced compared with the solutions of HCl. Although this effect is not as great as that seen at mlz 67 the enhancement is none the less very clear. mlz 70. The ion signals at rnlz 70 in the CI-containing solutions are primarily from T 1 2 + . These polyatomic ions were identified by Tan and Horlicklo the triplet of 35C12+ at mlz 70 35C137C12+ at rnlz 72 and 37C12+ at rnlz 74 being observed. The ion signals relative to that of 35Cl+ are comparable for the HCl and HC104 solutions. mlz 75. The ion signals at mlz 75 in the C1-containing solutions are primarily from 40Ar35C1+ the important interferent in C1-containing solutions which interferes with the very important monoisotopic 75A~+. The ion signals relative to that of T l + are comparable for the HC1 and HC104 solutions.mlz 80. The ion signals at mlz 80 are comparable in all solutions and are from 40Ar2+. These data are shown because as for the 36Ar+ ion signals they serve as a mediocre internal standard demonstrating that there was no large change in instrumental sensitivities between the samples.lI They also serve as an indication of the degree of formation of polyatomic ions. mlz 83. The ion signals at mlz 83 in the C1-containing solutions are primarily from 35C11603+. The ion signals in a solution of HC104 are dramatically enhanced (3.6 x lo4 ions s-l) compared to those in the spectrum of HC1 which are only slightly above the background. This species was not detected by Tan and Horlicklo in a solution of HCl and similarly was also not observed in this study.The ion signals at mlz 83 seen in the HNO blank and the HCl solution are the background in addition to ion signals from 83Kr+ from Kr contamination in the Ar plasma gas atmospheric gas entrainment [partial pressure of Kr= 1 x atm (1 atm= 101.325 kPa)] and atmospheric Kr dissolved in the sample solution. The ratio of the ion signals for the HC104 solution relative to that of the HC1 solution at mlz 83 is the highest of any observed in Fig. 1. A442 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY APRIL 1993 VOL. 8 comparable strong peak (1.2 x 10 ions s-I) was observed from the less abundant 37C11603+. rnlz 84. The ion signals at mlz 84 in the C1-containing solutions are primarily from 1H35C11603+. The ion signals in the HC104 solutions are clearly enhanced compared with the spectrum of HCl especially when compared with the CI ion signals at rnlz 35.There is a large background contribution at rnlz 84 from the major isotope 84Kr+ which diminishes the ratio of the ion signals in the HClO solution relative to the HC1 and HNO solutions. rnlz 100. The ion signals at mlz I00 in the C1-containing solutions are primarily from 1H35C11604+. As at rnlz 83 the ion signals in the HClO solutions are dramatically en- hanced. As with the other heavy polyatomic ions this species is not detected in the HCl solution where the total count rate of less than 10 ions s-' is from background. Interestingly there was no signal which was above the background at rnlz 99 where the signal of a C104+ ion would be observed.The bonding of H with ClO is necessary to form a stable ion. mlz 101. The ion signals at rnlz 101 in the C1-containing solutions are primarily from 1H235C11604+. As at mlz 100 the ion signals in the HClO solutions are dramatically enhanced. Enhanced spectral ion signals were also observed at mlz 103 from 1H237c11604+. mlz 232 The ion signals at mlz 232 are comparable in all solutions and are from 232Th+ which was added to all solutions at a nominal concentration of 100 ng g-l. These data serve as an internal standard demonstrating that there was no large change in instrumental sensitivities between the samples. rnlz 248. The ion signals at rnlz 248 are comparable in all solutions and are from Tho+. These data are shown as an indication of the operating conditions specifically the apparent plasma temperature.I2 It is important that the ThO+:Th+ ratio or a similar ratio from a strong oxide- forming element be included in the analytical data as changes in operating conditions especially the nebulizer gas flow rate allows the degree of formation of polyatomic ions to be varied over many orders of magnitude.Spectra of HC104 Solutions Containing Major Rock-forming Elements Mass spectra were obtained for solutions which contained approximately 90 pg g-l of one of nine major rock-forming elements (Na Mg Al K Ca Sc Ti Mn and Fe) in the three acid mixtures previously described. Very interesting sxy- chlorine metal polyatomic ions were observed only in the spectra of Ca and Mg in HC104. As with the acid spectra only the spectral features that were notable and not previously reported are shown in Figs.2 and 3. There were no observed unexpected spectral artifacts in the solutions of the other elements (Na Al K Sc Ti Mn and Fe) except for a range of contaminants from the single-element starting materials. Ca oxychlorine spectra The spectral ion signals for selected masses for solutions containing nominally 90 pg g-' of Ca in the three different acid mixtures are shown in Fig. 2 along with the data from the solution containing 1 moll-' WN0 alone (blank). This 1 mol 1-l HNO spectrum is the same as the data shown in Fig. 1 and is included for comparison. A detailed discussion of the spectrum in Fig. 2 follows. rnlz 35 36. The ion signals at mlz 35 ( T l + ) and 36 (36Ar+) were previously discussed. mlz 48.The ion signals at rnlz 48 are due to one of the minor isotopes of Ca (0.19%) and are shown to demonstrate 1x1071 ' 35 36 48 mfz Fig. 2 Comparison of spectral ion signals for solutions containing 90 pug g-' of Ca. Note the greatly enhanced ion signals in the HC104 solution due to CaClO+ (mlz 91 and 931 CaCIO,+ (rnlz 107 and 109) CaC103+ (mlr 123 and 125) and CaCI04+ (rnlz 139 and 141) the comparable Ca concentration in all solutions except the 'blank' HNO solution which contained no Ca. mlz 75 77. These ion signals are primarily due to 40Ca35C1+ and 4oCa37C1+ and are absent from the solutions that contained Ca with no C1 and the solution that contained neither Ca nor C1. There is not a large difference between the HC1 solution and the HClO solutions of Ca.There is a significant background in the C1-containing solutions from 4oAr35C1+ at mlz 75 (1200-1900 ions s-l) and from 4oAr37C1+ (400-600 ions s-l) at mlz 77. However the ion signals are significantly enhanced when C1 and Ca are present. rnlz 80. The ion signals at rnlz 80 are comparable in all solutions and are from ,0Ar2 (see above). rnlz 91 93. The ion signals are from 40Ca35Cl'60+ and 40Ca37C1160+. There is a significant enhancement of the ion signals of these polyatomic ions in the HC1Q4 solution of Ca compared with the HCl solution of Ca. This is a very interesting and unexpected ion demonstrating a speciation effect in the two acid solutions. mlz 107 109. These ion signals are from 4UCa35C1'602+ and 40Ca37C11602+. As for the monooxide there is a significant enhancement of the ion signals of these polya- tomic ions in the HC104 solution of Ca compared with the HCl solution of Ca.rnlz 123 125. These ion signals are 40Ca35C11603+ and 40Ca37C11603+. As for the mono- and dioxides there is a significant enhancement of the ion signals of these polya- tomic ions in the HCIO solution of Ca compared with the HCl solution of Ca. rnlz 139 141. These ion signals are due to 40Ca3sC116Q4+ and 40Ca37C11604+. As for the mono- di- and trioxides there is a significant enhancement of the ion signals of these polyatomic ions in the HClO solution of Ca compared with the HC1 solution of Ca. The ion signals in the Ca solutions that do not contain HC104 are due to trace amounts of 13gLa+ and 141Pr+ contained in the single-element Ca material.This contamination is confirmed by the presence of the characteristic La and Pr spectra in the solution of Ca in HN03. The ion signals which are at background levels in the HNO blank demonstrate that the signals are not related to the water or HN03 used. These interferences can be significant in the determination of low levels of La and Pr in Ca-containing rock materials that are digested with the aid of HClO,.JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY APRIL 1993 VOL. 8 443 HCl + HNO H HCIO + HNO 0 HNO D Blank lX1o7 I r I I X l O 6 0 T cn 5 iX1o5 1x10~ lX1o3 1x102 .- .- - CT z 10 4 ' 25 35 77 80 91 108 123 232 248 rnlz Fig. 3 Comparison of spectral ion signals for solutions containing 90 pg 1-1 of Mg. Note the greatly enhanced ion signals in the HC104 solution due to 24Mg35C11602+ (mlz 9 11 2sMg35C11603* (mlz 108) and 24Mg3sC11604+ (mlz 123) rnlz 232 248.The ion signals are from 232Th+ and 232Th160+ and are an indication of the operating condi- tions (see above). Mg oxychlorine spectra In addition to the interesting polyatomic ions found in the HClO solutions of Ca analogous ions were found in the spectra of the other alkaline earth element studied Mg. The spectral ion signals for selected masses for solutions containing nominally 90 pg 1-l of Mg in the three different acid mixtures are shown in Fig. 3 along with the data for the solution containing I moll-' HN03 alone (blank). The HN03 blank is the same as shown in Figs. 1 and 2 and is included for comparison. A detailed discussion of the spectrum in Fig.3 follows. rnlz 25. The ion signals at rnlz 25 are from a minor isotope (10%) of Mg and are shown to confirm the similar Mg concentration in the solutions. mlz 35. The ion signals at mlz 35 in the C1-containing solutions are from 35Cl+. rnlz 77. Some of the ion signals at rnlz 77 are probably from 24Mg35C1160+ the ion signals being comparable for the solutions that contained C1 (1600-2200 ions s-l). This is higher than observed in the equivalent acid solutions that did not contain Mg where the ion signals were 400-600 ions s-l. Only data from the minor isotope of Mg (1 1%) are shown as there are larger background interferences at mlz 75 (40Ar35C1+) and 76 (3sAr2+) which interfere with the other polyatomic ion of the other istopes of Mg 24Mg35C1160+ and 25Mg35C1'60+.The pres- ence of these ions is expected by analogy with the CaC10+ ions that were observed in solutions containing Ca and Cl. rnlz 80. The ion signals of ,0Ar2 at mlz 80 are comparable in a11 solutions. mlz 91. There is a significant enhancement of this ion signal from the polyatomic ion 24Mg35C11602+9 in the HC104 solution compared with the HC1 solution. This ion is comparable to the equivalent 40Ca35C116Q2+ ions observed in the Ca in HCD4 spectrum previously noted. mlz 108. This signal is from 25Mg35C11603+ This minor isotope of Mg (10%) is shown due to interference from trace amounts of Ag at mlz 107 and 109 with the other two isotopes of Mg. These polyatomic ions are analogous to the polyatomic ions CaC103+ found in the HClO solution of Ca and absent in the HC1 solution of Ca.mlz 123. The ion signal is from 24Mg35C11604+. There is a very significant enhancement of the ion signal of this polyatomic ion in the HClO solution of Mg compared with the HCl solution of Ca. rnlz 232 248. These ion signals are shown as an indication of the operating conditions (see above). The occurrence of species of MgClO in the mass spectra is probable but not clearly observed in the spectra of these solutions owing to the presence of ArCl+ Ar2H+ and Ar2+ polyatomic ions in the same mass range as the formation of MgCIO+. The occurrence of other oxychlorides of Mg (MgC102+ MgC103+ and MgC104+) is however clearly observed. Other elements (Nu Al K Sc Ti Mn and Fe) As mentioned spectral data were obtained for seven other major rock-forming elements in order to identify possible interferences when determining trace element concentra- tions in rock digests for which HClO was used in the preparation.In this study there were no observed ion signals that were not from contaminants in the reagents or which were not due to ions previously discussed in this paper. No polyatomic metal oxychlorine ions were ob- served where the metal was Na Al K Sc Ti Mn or Fe. Conclusions The ICP mass spectra of solutions containing HC104 demonstrate significantly enhanced signals from H,ClQ,+ (x= 1 2 3 or 4; y=O 1 or 2) polyatornic oxychlorine ions when compared with an equimolar solution of HCl both solutions containing the nearly same concentrations of the elements H 0 and C1 but with different speciation.The greatest enhancement is from the higher oxides (x= 2,3 and 4). The alkaline earth elements Ca and Mg form polyatomic ions with ClO+ C102+ C103+ and Clod+ that are greatly enhanced in solutions of HC104 when compared with equimolar solutions of HC1. Perchlorate ions (ClO,+) were not observed alone but only when combined with H (HC104+ or H2C104+) Ca (CaCIO,+),or Mg (MgC104+). Other major rock-forming elements (Na Al K Sc Ti Mn and Fe) did not show mass spectral differences when in HCIO solutions compared with equimolar HCl solutions or solutions that contained no C1. The ICP contains polyatomic ions the structure of which is related to the speciation of the elements in the sample solution and this bonding is preserved in the plasma and through the two stages of expansion into the mass spectro- meter.These ions are not entirely formed from recombina- tion of ions and neutral species but exist in solutions and these species are not completely converted into atoms either neutral or ionized in the ICP. The ions ClW C102+ C103+ and C104+ are clearly related to the presence of C104- ions in the sample solution. The ions formed by the combination of these oxychlorides with H Ca or Mg could be formed either in solution or as an ion combination in the plasma or expansion stage. The data reported here do not support either mechanism of formation. This research was initially undertaken in order to determine the source of previously unobserved inter- ferences with the light rare earth elements. For the determination of low concentrations of La and Pr in light rare earth element-depleted rock samples [e.g. BIR- 1 United States Geological Survey (USGS) reference materi- al] the use of HClO in the sample pxeparation can cause significant interferences.The final concentration of HC104 in a preparation depends on the time and temperature of the sample digestion and on the composition of the444 JOURNAL OF sample. The magnitude of the effect of the interferences of CaC104+ is therefore not easily amenable to accurate prediction and correction. Also the interference is not amenable to correction by use of a reagent blank as a reagent blank that contains no rock-forming elements will not when taken to dryness contain C104- salts The study of complete mass spectra especially when new sample types and new digestion procedures are undertaken is therefore very important.This work was carried out as part of the development of the ICP-MS facility at Memorial University of Newfoundland. The instrument was purchased with a Natural Science and Engineering Research Council of Canada (NSERC) major installation grant and is operated with assistance of an NSERC infrastructure grant and numerous NSERC operat- ing grants including one to the author. The generous support of the Memorial University to the operation of the facility is acknowledged. Careful reading by B. J. Fryer S. E. Jackson and P. King is appreciated. Preparation of the solutions by P. King and instrument operation by B. Chapman are greatly appreciated. ANALYTICAL ATOMIC SPECTROMETRY APRIL 1993 VQL. 8 1 2 3 4 5 6 7 8 9 10 1 1 12 References Longerich H. P. Strong D. F. and Kantipuly C. J. Can. J. Spectrosc. 1986 5 1 1 1. Longerich H. P. ICP In$ Newsl. 1992 18 32. Mattinson J. M. Anal. Chem. 1972 44 1715. Longerich H. P. Jenner G . A. Fryer B. J. and Jackson S. E. Chem. Geol. 1990 83 105. Taylor S. R.. and McLennan S . M. The Continental Crust Its Composition and Evolution Blackwell Oxford. 1985 p. 67. Horlick G. Tan S. H. Vaughan M. A. and Rose C. A. Spectrochim. Acta Part 63 1985 40 1555. Lichte F. E. Meier A. L. and Crock J. G. Anal. Chem. 1987,59 1150. Longerich H. P. Fryer B. J. Strong D. F. and Kantipuly C. J. Spectrochirn. Acta Part 18 1987 42 75. Jenner G. A. Longerich H. P. Jackson S. E. and Fryer €3. J. Chem. Geol. 1990 83 133. Tan S. H. and Horlick G. Appl. Spectrosc. 1986 4Q 445. Beauchemin D. McLaren J. W. and Berman S. S. Spectro- chim. Acta Part 3 1987 42 467. Longerich H. P.. J. Anal. At. Spectroin. 1989 4 491. Paper 2/034 73E Received June 30 1992 Accepted November 18. I992

ISSN:0267-9477    

DOI:10.1039/JA9930800439

出版商:RSC    

年代:1993

数据来源: RSC

摘要:

JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY APRIL 1993 VOL. 8 445 Determination of Nickel in Serum and Urine by Inductively Coupled Plasma Mass Spectrometry Sonny X. Xu Lidija Stuhne-Sekalec and Douglas M. Templeton* Department of Clinical Biochemistry University of Toronto 100 College Street Toronto Canada M5G 1 L5 Methods are described for the measurement of Ni in human serum and urine by inductively coupled plasma mass spectrometry. In serum interferences arising from Ca- Na- and K-containing polyatomic species are corrected for by using principal components analysis with an internal Rh standard. The method is accurate to a level of 5 pg I-j as determined with the available reference materials and a value an order of magnitude lower was achieved in pooled human serum. Within-run imprecision was 6% [relative standard deviation (RSD)] at the 1.5 pg I-' level and the pattern of interferences was stable across an analytical run of at least 15 samples.Nickel was measured directly on diluted urine following quantitative precipitation of Ca2+ with oxalic acid. The method is accurate to 40pg I-' and has a within-run imprecision of 7% (RSD) at 4.5pg I-' of Ni measured in human urine. Both 61Ni and 62Ni are available for stable-isotope tracer studies with "NNi being preferred because only Ca- dependent polyatomic interferences are present at this mlz value and these can be overcome in urine by oxalate precipitation. Principal components analysis is also shown to distinguish between Ni sources enriched in both isotopes simultaneously present with naturally abundant Ni and therefore dual isotope tracer studies are feasible. The methods described are applicable at the concentration levels of Ni in body fluids that have been observed in human biokinetic studies.Keywords Inductively coupled plasma mass spectrometry; nickel; human serum and urine; principal component analysis; stable isotope The determination of Ni in serum1t2 and urine314 is largely the province of electrothermal atomic absorption spectro- metry. After absorption by several routes most of a dose of soluble Ni is cleared by urinary excretion with a half- time of 12-48 h (ref. 5 ) and urinary Ni levels are useful in monitoring workplace exposure.6 Levels of Ni in serum and plasma of healthy individuals are more difficult to determine probably being in the range 0.1-0.5 pg l-l and the quest to establish reference intervals for serum Ni continues.' Until the natural level in the general popula- tion is established with more certainty the usefulness of many reports of increased levels remains questionable.In part uncertainty arises from expected values very near the detection limit of atomic absorption techniques. The detection limits for the determination of Ni in aqueous solutions by inductively coupled plasma mass spec- trometry (ICP-MS) (<0.05 cg 1 - 1 ) 8 7 9 are very attractive in this context but before they can be realized in biological matrices means of overcoming polyatomic interferences are needed.lo Fasting volunteers have been given aqueous solutions of NiS04 at natural isotopic abundancell or Ni(N03)2 enriched in 61Ni (ref.12) and an oral absorption of about 30% of the administered dose has been demon- strated. This is much higher than the absorption of Ni taken with food (typically< 1 O/o) and raises important questions about the bioavailability and clearance of Ni in various populations. Such questions are amenable to investigation with the use of stable isotopes which circum- vents ethical considerations prohibiting the use of radio- isotopes and somewhat relaxes the stringent clean room conditions required to avoid contamination with naturally abundant Ni. For these reasons an evaluation of the usefulness of ICP- MS in determining Ni at natural levels in body fluids has been undertaken and methods for stable-isotope tracer studies applicable to human subjects developed.A pre- liminary report on the administration of 62Ni to rats and 61Ni to a human volunteer has been published.I2 Here the analytical methods that are applicable to such studies are described. Experimental Reagents and Standards Elemental standards were from SPEX Industries (Edison NJ USA) and included Ni Ca Na K and Rh at concentrations of 1.00 10.0 10.0 1.00 and 1.00 mg 1-' respectively. Isotopically enriched 61Ni (found 88.03%) and 62Ni (found 94.85%) were obtained from Oak Ridge National Laboratories (Oak Ridge TN USA) as metal powders and dissolved in HN03 (Suprapur; Merck Darmstadt Germany). Oxalic acid was from Gen- eral Chemical (New York USA). De-ionized water (18 MS2 cm purity; MilliQ system Millipore Bedford MA USA) and Suprapur HN03 were used for all dilutions and digestions.The reference materials used are listed in Table 1. Table 1 Certified reference materials used in this study Source Reference material [Ni]/pg I-' Nycomed Oslo Seronorm Trace Elements in Serum 5 .O _+ C. 5* Nycomed Oslo Seronorm Trace Elements in Urine 30* SRM 2670 Toxic Metals in Freeze-Dried Urine 70t SRM 2670 Elevated Levels 300t NIST (Gaithersburg MD USA) *Recommended Value. thformation value. *To whom correspondence shouId be addressed.446 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY APRIL 1993 VOL. 8 Table 2 Typical ICP-MS operating parameters optimized for the analysis of aqueous Ni calibration standards R.f. power Forwardlk W 1.15 Reflected powerlW <5 Inner 1 Intermediate 2.1 Outer 12 Scanning mode Elemental Measurement mode Sequential Measurements per peak 3 Measurement timels 1 .o Repeats per integration 6 Counting precision (Yo) 0.1 Gas flow ratesll mind' Ion lens voltages Bessel Plate Photon Stop Einzel + 4.3 - 8.8 - 4.9 - 14.9 Instrumentation Analyses were performed on an Elan 250 ICP mass spectrometer (Perkin Elmer SCIEX Thornhill Ontario Canada) equipped with a Merinhard Type G nebulizer standard long ICP torch and a mass flow controller (Tylan FZ-280) on the inner gas flow line.With daily optimization of operating parameters and ion lens voltages a typical sensitivity for Ni measured at mlz=58 was 600000 ions s-l for a 100 pg 1-l solution. The detection limit (blank+3a) of Ni in aqueous standards was 0.03 pg l-* Typical optimized operating parameters for the determina- tion of Ni in calibration standards are given in Table 2.Principal Components Analysis To correct for spectral overlaps primarily from CaO+ and CaOI1+ species principal components analysis (PCA) was used as described previo~sly.~J~ Data were acquired at mlz 58 60 61 and 62 at low resolution. No corrections for isobaric overlap were applied at the time of measurement. Rhodium (100 pg 1-l) was used as an internal standard. Aqueous solutions of Ni (100 pg l-l) Ca (100 mg 1-l) and Na plus K (1.5 g 1-l of Na plus 0.5 g 1-l of K) were prepared from the elemental standard solutions and used to obtain the required target vectors. These are referred to as target solutions below. Reproduced data were obtained by sum- ming the product of the target scores and target vectors.The PCA was carried out on a Macintosh SE-30 computer with a program written in Matlab using algorithms from refs. 14 and 15. Analysis of Serum and Urine To 200 pl of semm in a screw capped poly(tetrafluorsethy- lene) digestion vessel (30 ml capacity) were added 300 pl of HN03. The mixture was warmed on a hot-plate to about 9s "C for at least 4 h and then made up to 2.00 ml in a calibrated flask with water and Rh (100 pg 1-l) as internal standard for analysis by PCA To 4 ml of urine in a 15 ml plastic Falcon tube were added 12.6 mg of oxalic acid (General Chemical) to a concentration of 25 mmol l-l. The tube was shaken and centrifuged in a bench top centrifuge to precipitate calcium oxalate. An aliquot of the superna- tant (0.8 ml) was mixed with 0.2 ml of l mg l-' Rh and 1.2 ml of 0.l0/o HN03 for analysis by ICP-MS against an external calibration graph at mlz= 60.All glass and plastic ware including plastic pipette tips contacting the sample Table 3 Signal intensities at mlz values of the Ni isotopes during analysis of pooled human serum as described under Experimental. Intensities are expressed relative to a Rh internal standard at mlz= 103 rnlz Relative intensity ( x lo2) Ni isotopic abundance (Yo) 58 2.167 60 1.375 61 0.382 62 0.150 67.8 26.2 1.25 3.66 during processing were acid-washed by soaking in 10% analytical-reagent grade HN03 overnight or longer fol- lowed by repeated rinsing in de-ionized water. Results and Discussion Nickel in Serum Nickel has five stable isotopes with relative atomic masses (and abundances in natural Ni) of 58Ni (67.8%) 60Ni (26.16%) (jlNi (1.2§%) 62Ni (3.66%) and 64Ni (1.16%).A minor isotope of Fe (58Fe; 0.33% natural abundance) generally precludes measurement of the major isotope of Ni in biological samples even when isobaric corrections are applied. The expected concentration of Fe in human serum is about 1 mg l-l or 1 x 10J-l x lo4 times that of Ni. Similarly 64Ni is obscured by the major isotope ofZn (64Zn; 48.89O/o) also present at about 1 mg I-'. This leaves 6oNi as the isotope of choice for total Ni determinations in serum with 61Ni and 62Ni available for isotope dilution and tracer studies. Spectral intensities from a typical analysis of pooled serum in the range of the Ni mlz values studied (Table 3) indicate additional polyatomic interferences at the inter- mediate Ni values and from previous study of urinel0 it seems likely that these arise from the calcium species 44Ca160+ 44Ca16QH+ and 46Ca160+.An additional inten- sity peak at mlz 62 could only be reproduced with a solution containing both Na and K; 23Naz160+ and 23Na39K+ might be contributing polyatomic ions. Cryogenic desolvation of the sample aerosol has been reported to decrease dramati- cally solvent loading of the plasma and thereby suppress oxide formafion.l6 Zhu and Browner17 studied oxide and hydroxide formation of Ba and Ce as a function of spray chamber temperature using both the SGIEX Elan 250 ICP mass spectrometer and a comparable instrument from VG Elemental. In both instruments higher temperatures from uncooled chambers increased the amount of the polyatornic ions.However even at lower temperatures the decrease in these species was only fractional and would not circumvent the need for correction in high-calcium matrices. Further- more in contrast to the VG instrument greatly improved signal intensity of monatomic ions was seen at slightly higher temperatures using the Elan 250 and this was particularly true of those elements closest in relative atomic mass 'SO Ni namely Co and Cr." Therefore following previous experience PCA was applied to the measurement of Ni in serum. The approach consistently required three components to describe the set of ion intensities at mlz 59 60 6 1 and 62 and effective data remodelling was achieved with target solutions of pure Ca and Ni standards in addition to a combined solution of Na and K. In order to assess the imprecision of the method a sample of pooled human serum was analysed 40 times on three consecutive days using standard additions and a Kh internal standard and were found to contain 0.46 2 0.06 pg 1-1 of Ni.This was spiked with an additional 1 .OO pg 1 - I of Wi. Within-run and between-run imprecision in this serum matrix were found to be about 6 and 1% (RSD),JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY APRIL 1993 VOL. 8 447 Table 4 Effect of position of target solution analysis in an analytical sequence. A serum sample spiked to contain 1.46 pg I-' of Ni was analysed by 3-component PCA using pure Ca (100 mg 1-I) and Ni (100 pg 1-l) standards as target solutions and a solution of 1500 mg I-' of Na and 500 mg I-' of K.The target solutions were analysed at the beginning of the run and after every fifth sample. The total data set of 15 repeats of the sample was analysed separately against each of the sets of target solutions to determine if the characteristic mass spectra of the pure components changed during the analytical run. The experiment was repeated on consecutive days. Values given as mean +_ SD Nilpg I-' No. of sample repeats prior to target solutions Day 1 Day 2 0 1.56 t-0.09 1.57 +O. 14 5 1.251t0.07 1.36k0.12 10 1.58 t-0.09 1.39 k0.12 15 1.49 It0.09 1.35 +O. 12 respectively at 1.46 pg 1-I. Recovery of the spike contain- ing 1 pg 1-' of Ni was 11229%. Based on repeated measurements of a blank nitric acid digest a detection limit (blank+30) of 0.37 was obtained.Natural levels of Ni in the serum and plasma of healthy unexposed adults are expected to be in the range 0.1-1 pg 1-l and the analyst is generally hampered in assessing the accuracy of a technique by the lack of suitable reference sera with values of Ni certified at this level. However Seronorm Trace Elements in Serum with a preliminary recommended value of 5.0 2 0.5 pg 1 - I of Ni was analysed and a value of 4.08 -t 0.17 pg 1-l was obtained. Analysis of multiple samples of biological origin leads to some salt deposition at the ICP-MS interface and on the ion optics which could conceivably change the formation and response of the polyatomic species. To evaluate the irnpor- tance of the position of the target solutions in the sample sequence the pooled serum spiked to 1.46 pg 1 - I of Ni was analysed 15 times consecutively with Ni Ca and Na plus K target solutions measured initially and again after 5 10 and 15 sample determinations.Nickel was calculated by PCA from targets run in each position. The experiment was repeated on a second day. The position of the target vectors at least in this short analytical run was unimportant (Table 4). Nickel in Urine The successful determination of Ni in urine by PCA using a National Institute of Standards and Technology (NIST) (formerly National Bureau of Standards) Standard Refer- ence Material (SRM) 2670 Toxic Metals in Freeze-Dried Urine with a Ni level certified at 70 pg 1-L has been reported previously.IO However at the expected levels of 1-5 pg 1-l in healthy adult^,^ the method was not sufficiently robust for routine analysis.The Ca content of human urine (approximately 5 mmol 1-I) was thought to contribute to changing sampling characteristics throughout a prolonged analytical run and the Ca target was not adequately modelled at the end of such a sequence. Therefore the precipitation of Ca as the oxalate salt was considered. Duplicate samples of human urine were spiked with either 0.3 pCi of 45Ca (original activity 6.74 Ci g-I) or 0.01 6 pCi of 63Ni (8.19 Ci g-l) prior to treatment with oxalate. Calcium was effectively removed while Ni remained in solution (Table 5). The reason for the low recovery of 45Ca in the pellet (9 1%) is not known but may represent a change in the efficiency of scintillation counting in the presence of the precipitate.The spectral intensities arising from a solution containing 180 mg 1-l of Ca are compared with those of Table 5 Retention of Ni after removal of Ca from human urine with oxalic acid. Fresh urine was spiked with either 45Ca or 63Ni prior to treatment with oxalic acid. Initially radioactivity in the original sample (counts min-l) was compared with that in the supernatant and washed pellet after treatment. Added Ni was about 7 pg 1 - l ; the amount of Ca added was negligible with respect to the concentration of Ca pre-existing in the urine. Counting efficiency for 45Ca might be decreased in the precipitate giving an apparent recovery of < 100°/o 63Ni 45Ca counts counts Sample min-' Yo min-' Oh Supernatant 668590 96 28 0.1 Pellet 27805 4 32548 91 Original 697900 - 35827 - urine in Fig.1. Only at mlz 62 is the Ca signal insufficient to account for a significant proportion of the urine signal. After treatment with 25 mmol 1 - I oxalic acid (approxi- mately a five-fold excess relative to urinary Ca) the urine sample intensities at rnlz 59 60 and 61 are substantially decreased while the intensity at rnlz 62 is unaffected. Increasing the concentration of oxalic acid to 50 mmol l-l has no further effect at the Ca-dependent masses. The removal of the Ca by precipitation caused an observ- able improvement in the signal monitored with the 100 pg 1-l of Rh internal standard from 54 520 k 500 ions s-l to 56 290 f 320 ions s-l although this slight improvement in signal suppression is not expected to be significant with respect to the analytical performance of the method.A random sample of human urine was then analysed at 40% dilution in 0.1 O/o HN03 after calcium oxalate precipita- tion and measured to contain 4.53k0.33 pg 1-l of Ni against an external calibration graph with a Rh internal standard. The sample was then spiked with an additional 1.00 pg 1-l of Ni; recovery of added Ni was 1.20kO. 11 pg 1-I. A detection limit of 0.09 pg 1-l (blank+3a) was estimated from a nitric acid blank treated with oxalate. Within- and between-run imprecision (n= 15) were found to be 6.7 and 7.3% (RSD) respectively at 4.53 pg 1-l of Ni. Analysis of the reference materials by this method gave 41.922.7 (Seronorm; recommended value 40 pg 1 - I ) and 0.29k0.01 mg 1-l (NIST SRM 2670; information value 0.30 mg 1-l).0.10 > v) a * .- * .- a 0.05 - a CT 0 58 59 60 61 62 mlz Fig. 1 Effect of oxalic acid precipitation on the mass spectrum of urine in the region of the Ni isotopes. Ion intensity on the ordinate is expressed relative to that of a 100 pg 1-' Rh internal standard measured at mlz= 103. Solid bars 180 mg 1-' of Ca standard solution; shaded bars human urine diluted to 40% in 0.1 To HNO,; stippled bars 40% urine in HNOJ after precipitation with 25 mmol I-' oxalate; and open bars 40% urine in HNO after precipitation with 50 mmol 1-L oxalate448 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY APRIL 1993 VOL. 8 50 40 2 30 20 10 r I - rn - .- 0 1 2 3 4 Solution number 5 6 Fig. 2 Separation by PCA of signals from a mixture of naturally occurring Ni and Ni sources enriched in 61Ni and 62Ni.Solutions of the three Ni sources were mixed to give a total concentration of Ni of 50 pg 1-I in six different combinations the enriched sources being omitted from the first. In each pair of bars the first represents the total amount of Ni added and the second the measured concentration after 3-component PCA. Solid bars natural abun- dance; shaded bars 62Ni-enriched source; and open bars 'jlNi- enriched source. Regression lines of Ni measured versus Ni added were calculated. For natural abundance y= - 0.1 1 1 +0.938x; for 61Ni y= -0.092+0.951x (r= 1.00); and for 62Ni y= -0.042 +0.946x (r= 1.00) Choice of an Isotope for Tracer Studies Preliminary stable isotope tracer studies with 62Ni in rats and 61Ni in a human volunteer have been reported elsewhere.I* The 'jlNi is available at reasonable cost and approximately 90% enrichment and seems to have the advantage over 'j2Ni in that the only identifiable interfer- ence is from Ca species.The absence of Na- and K- dependent interferences at this mass means that 61Ni can be detected in urine directly after oxalate precipitation. How- ever both isotopes are available for dual isotope tracer studies. Because PCA separates components with distinct spectral patterns an element with isotopes present at natural abundance can be separated from an isotopically enriched source. The use of PCA to distinguish enriched sources of 61Ni and 62Ni in the presence of naturally occurring Ni was evaluated.Mixtures of the three Ni sources were analysed in aqueous solution at different proportions at a total concen- tration of 50 pg 1-l. Both tracers could be determined to within 7% of their added concentrations in the presence of an approximately 1 0-fold excess of naturally abundant Ni (Fig. 2). It can be concluded that PCA allows the simulta- neous determination of the two tracers at levels likely to be encountered in tracer studies.l13l2 However there is a slight negative analytical bias apparent in the method. Plots of Ni recovered versus Ni added have correlation coefficients for linear least squares fits of 1.00 for all three sources of Ni but the slopes of the lines are 0.94-0.95 for each; a negative bias of about 5%. In summary methods have been presented that allow the determination of Ni in human serum and urine by ICP-MS at concentrations expected to be present in reference populations.An advantage over the more widely used technique of Zeeman-corrected electrothermal atomic ab- sorption spectrometry is the isotope-specific information obtained. The direct analysis of Ni in urine after precipita- tion of calcium oxalate and the ability of PCA to separate signals from naturally abundant Ni and sources enriched in 61Ni or 62Ni should facilitate more detailed single and dual isotope tracer studies of Ni biokinetics in human subjects. Supported by an operating grant from the Ni Producers Environmental Research Association (NIPERA). The con- structive comments of Professors John Savory and Sam Houk during the course of this work are gratefully acknow- ledged.References 1 Sunderman F. W. Jr. Crisostomo M. C. Reid M. C. Hopfer S. M. and Nomoto S. Ann. Clin. Lab. Sci. 1984 14 232. 2 Nixon D. E. Moyer T. P. Squillace D. P. and McCarthy J. T. Analyst 1989 114 1671. 3 Sunderman F. W. Jr. Hopfer S. M. Crisostomo M. C. and Stoeppler M. Ann. Clin. Lab. Sci. 1986 16 219. 4 Paschal D. C. and Bailey G. G. Sci. Total Environ. 1989,89 305. 5 Sunderman F. W. Jr. in Handbook on Toxicity of Inorganic Compounds Nickel eds. Seiler H. G. and Sigel H. Marcel Dekker New York 1988 p. 453. 6 Angerer J. and Lehnert G. Int. Arch. Occup. Environ. Health. 1990 62 7. 7 Templeton D. M. and Sunderman F. W. Jr. in preparation. 8 Templeton D. M. Paudyn A. and Baines A. Biol. Trace Elem. Res. 1989 22 17. 9 Hieftje G. M. and Vickers 6. H. Anal. Chim. Acta 1989 216 1. I0 Vaughan M. A. and Templeton D. M. Appl. Spectrosc. 1990 10 1685. 1 1 Sunderman F. W. Jr. Hopfer S. M. Sweeney K. R. Marcus A. H. and Most B. M. Proc. SOC. Exp. Biol. Med. 1989 191 5. 12 Templeton D. M. Xu S. X. and Stuhne-Sekalec L. Sci. Total Environ. 1993 in the press. 13 Templeton D. M. and Vaughan M.-A. in Applications of Plasma Source Mass Spectrometry eds. Holland G. Eaton E. A. The Royal Society of Chemistry Cambridge 199 1 p. 101. 14 Wirsz D. F. and Blades M. W. Anal. Chem. 1986 58 51. 15 Malinowski E. R. and Howery D. G. Factor Analysis in Chemistry Wiley-Interscience New York 1980. 16 Alves L. C. Wiederin D. R. and Houk R. S. Anal. Chem. 1992 64 1164. 17 Zhu G. and Browner R. F. J. Anal. At. Spectrom. 1988 3 781. Paper 2/05866I Received November 2 1992 Accepted December 2 1992

ISSN:0267-9477    

DOI:10.1039/JA9930800445

出版商:RSC    

年代:1993

数据来源: RSC

摘要:

JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY APRIL 1993 VOL. 8 449 Determination of Lead Isotope Ratios and Concentrations in Sea-water by Inductively Coupled Plasma Mass Spectrometry After Preconcentration Using Chelex-I00 Akira Miyazaki and Raul A. Reimer* National Institute for Resources and Environment 16-3 Onoga wa Tsukuba lbaraki 305 Japan The determination of lead isotope ratios and concentrations in open and coastal sea-water by a combination of chelating resin (Chelex-100) preconcentration and inductively coupled plasma mass spectrometry is described. The 206Pb:204Pb 208Pb:204 207Pb:208Pb and 20sPb:207Pb isotope ratios were measured with relative standard deviations of 1.2 0.9 0.1 and 0.3% respectively. The accuracy of the isotopic data was checked with reference materials National Research Council Canada NASS-3 Open Ocean Sea-water and National Institute of Standards and Technology (NIST) Standard Reference Material (SRM) 981 Lead Common Isotopic and compared with those in other work.The results revealed that the proposed method was applicable to the measurement of lead isotope ratios in sea-water with precision adequate to be able to discuss the origin of lead. The detection limit was 0.8 ng I-'. Keywords Inductively coupled plasma mass spectrometry; lead isotope ratio; lead concentration; sea-water; Chelex- 100 The concentration of lead in sea-water is increasing owing to industrial activity and gasoline combustion.' To solve the problem of lead pollution of sea-water it is important to clarify the origin of the lead present by measuring the lead isotope ratios2 Hitherto thermal ionization mass spectro- metry (TIMS) has been used for this purpose.Although it allows the isotope ratios and concentrations to be deter- mined very accurately and precisely it is time consuming because each sample droplet is placed on a filament and after evacuation of the ionization chamber the filament is gradually heated to the temperature at which lead ions are stably emitted. A faster method would be advantageous for environmental studies where many samples have to be analysed. Inductively coupled plasma mass spectrometry (ICP-MS) seems to be the best alternative method. Al- though many papers3-14 have been published on the measurement of lead isotope ratios by ICP-MS most of them are related to airborne particulates pond sediments blood tooth and milk powder etc.in which lead csncen- trations range from ng ml-l (or ng g-l) to sub-pg r n l - I (or sub-pg g-l) levels. As the lead concentration in open sea- water is below several tens of ng l-l l s and sea-water contains large amounts of salts some preconcentration and matrix separation procedure is essential in order to deter- mine accurately the isotope ratios and concentrations of lead. Beauchemin et al. successfully measured lead isotope ratios and concentrations in the reference material Na- tional Research Council Canada (NRCC) NASS-2 Open Ocean Sea-water by ICP-MS after preconcentration on silica-immobilized 8-hydroxyquinoline. However to the best of our knowledge there has been no report of the detection of isotope ratio differences in real sea-water samples by using ICP-MS.The object of this work is to demonstrate the applicability of ICP-MS to lead isotope ratios measurements in sea-water with a precision adequate to permit the investigation of the provenance of the lead. In addition lead concentrations in sea-water were measured and analytical features such as accuracy and precision were evaluated. *On leave from Yacimientos Petroliferos Fiscales SA GGAT Av. Calchaqui Km 23.5 1888 Fcio Varela Buenos Aires Argen- tina. Experimental Apparatus A Yokogawa (Tokyo Japan) PMS 2000 ICP-MS system was used; the operating conditions are given in Table 1. Sample was introduced into the water-cooled spray chamber by using a peristaltic pump. Sea-water Samples and Reagents Coastal sea-water samples were taken from the surface of the sea with a pre-cleaned polyethylene beaker connected to a nylon rope and stored in an acid-washed 10 1 polyethylene tank.Open ocean sea-water samples were likewise collected during the cruise of the Hakurei-maru Japan from August to September 199 1. All solutions were prepared with de-ionized water ( 1 8 MQ em) purified using a Mi11iQ-I1 system (Millipore Bedford MA USA). A solution of 1 mol 1-l ammonium acetate (Wako Osaka Japan) was prepared and its pH was adjusted to 6 with ammonia solution or acetic acid and then purified by passing it through a Chelex-100 column. Nitric Table 1 Operating conditions Plasma R.f. power Plasma Ar flow rate Auxiliary Ar flow rate Nebulizer Ar flow rate Sample flow rate Spray chamber temperature Mass spectrometer Lenses Sampler Skimmer Operating pressures Dwell time Number of scans Repetition 1.3 kW 14 1 min-I 0.5 1 min-' 0.45 1 min-' 0.3 ml min-I 0 "C Optimized on 10 pg I-' 1.0 mm orifice copper 0.5 mm orifice copper Interface region 1 Torr; mass spectrometer chamber 1 x Torr Pb standard 100 ms 20 (for concentration); 300 (for isotope ratios) 2 (for concentration); 10 (for isotope ratios)450 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY APRIL 1993 VOL.8 acid hydrochloric acid and ammonia solution were of Merk ultra-pure quality. Chelex- 100 chelating resin (50- 100 mesh) was provided by Bio-Rad Laboratories (Richmond CA USA). Lead stock standard solution (1 00 mg 1-l) was obtained from Wako and intermediate working standard solutions were prepared by appropriate dilution.A portion (0.1 g) of National Institute of Standards and Technology (NIST) Standard Reference Material (SRM) 981 Lead Common Isotopic was dissolved in ultra-pure HN03 (1 + 1 v/v) with heating before dilution to the desired lead concentration with 0.5 mol 1-l HN03. Procedure The column system used and the column preparation were the same as described previously16 and are briefly summar- ized here. To avoid contamination the column system was designed to be closed from the atmosphere. After cleaning the resin with 5 moll-' HCl water and 2 moll-' FINO3 the resin (mass 2.5 g) was packed into the column and preconditioned with ammonia solution water and ammon- ium acetate buffer (pH 6) prior to each use. The inner diameter of the column was 10 mm and the length of the preconditioned resin bed in the column was about 30 mm.A 1 1 sea-water sample was filtered through a 0.45 pm Millipore filter and 10 ml of 1 mol 1-l ammonium acetate buffer solution (pH 6) were added. The pI-I was adjusted to pH 6 with 2 moll-' ammonia solution and 1 mol 1-1 acetic acid. The solution was passed through the Chelex-100 column using a peristaltic pump at a flow rate of 2.0 ml mine'. The resin was then washed with 50 ml of water and for removal of matrix salts the column was treated with 80 ml of 1 mol 1-1 ammonium acetate buffer (pH 6) and 80 ml of water after each sea-water preconcentration. The trace metal preconcentrated on the resin was eluted with 10 ml of 5 mol 1-l HN03 followed by 10 rnl of water into a PTFE beaker.Then 6 ml of concentrated HCl were added and the sample was evaporated to dryness on a hot-plate in a flow of clean air. Residual ammonium nitrate was volatilized in this step. The residue was dissolved in 5 ml of 0.2 mol 1-l HN03 placed in a 10 ml calibrated flask and diluted to volume with water. The 1 00-fold concentrated solution was diluted 10-fold in order to minimize the salt content and was then ready for ICP-MS measurement. All procedures were carried out on a clean bench or in a laminar flow cabinet. Results and Discussion Isotope Ratio Measurements The isotope ratio of lead varies because of the radioactive decay of 238U to 206Pb 235U to 207Pb and 232Th to 20sPb. One istope 204Pb has no long-lived radioactive parents.Hence differences in the isotopes of lead arise primarily from differences in the ages and parent:daughter ratios in geochemical formations. These differences are reflected in oceanic waters which receive lead inputs from geological formations via weathering processes. Therefore the mea- surement of isotope ratios of lead in sea-water is useful when considering the origin of the lead. Istope ratios were measured under the conditions given in Table 1. It took about 20 min to measure the isotope ratios of one sample. The relative standard deviations of the 206Pb:204Pb 208Pb:2o4Pb 207Pb:208Pb and zo6Pb:207Pb ratios were 1.2 0.9 0.1 and 0.3% respectively. To confirm the accuracy of isotopic data NRCC NASS-3 Open Ocean Sea- water and NIST SRM 981 were analysed. The final concentrations for ICP-MS measurement of SRM 981 and NASS-3 were 2 and 0.4 pg l - l respectively.As shown Table 2 Lead isotope ratios of SRM 981 and NASS-3 NIST SRM 981 Isotope ratio This work Certified value* 204Pb:206Pb 0.05882 f 0.0001 5t 0.059042 f 0.000037 207pb:206pb 0.9174-tO.0018t 0.914641+0.00033 208pb:206pb 2.1701 k0.0012t 2.168 1 2 0.0008 NRCC NASS-3 Previous data This work (pH 5.5)9 204pb:208pb 0.0284 f 0.0004t 0.0323 _t 0.0040 206pb:ZOSpb 0.479 f 0.005t 0.494 k 0.009 207pb:ZOSpb 0.4 I06 f 0.002t 0.425 f 0.020 *The certified values were measured by TIMS and show a higher ?Standard deviation of three determinations through the whole precision than those measured by ICP-MS. procedure. in Table 2 our results were in good agreement with those obtained in other work or with the certified value.The standard deviations in this work were based on sub-samples carried through the procedure. In Table 2 previous data for NASS-2 were obtained from ref. 9. However because NASS-3 was prepared from the same batch of sea-water the data were used here for comparison. Although the drift of all isotope ratios during the measurements was less than 0.lo/o NASS-3 was occasionally analysed to check and compensate for the drift. Isotope data are plotted in Figs. 1-4 for coastal sea-water and open ocean water samples. As can be seen from Figs. 1 and 2 the data for open ocean water samples have a spread overlapping those for the coastal waters most of which were collected at locations far from human habitation. This is also true for the data in Figs.3 and 4. However in Fig. 1 the data for Chiba New Port Tsushima Omodahama Tsushima Izuhara Tomakomai Port Tsushima Tanohama and Tomakomai Beach are apparently different from those for the other samples. In Fig. 3 samples from Sendai Port Chiba New Port Tokyo Bay St. 22 and Tsushima Omoda- Chiba New Port 4 Omodahamaw lzuhara Tomakomai Port 37 t 0 Tanohama Tomakomai Beach 0 36 t =. ' Samples far from human habitation 34 m I . Fig. 1 Isotope ratio for coastal sea-water 6 and 8 sea-water sample; 0 NASS-3; and * SRM 981. The bars in the figure indicate the standard deviation of the dataJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY APRIL 1993 VOL. 8 0 0 0 0 0 0 i I 45 1 36 P Q N. . a 2 35 n Q 34 Fig. 2 Isotope ratio for open ocean sea-water. See Fig.1 for details 1.17 1.16 1.15 n a 6 N.. 1.14 n a : 1.13 1.11 1.1 1 OSendai Port Chiba New Port 0 Tokyo Bay St. 22 lmoda ha ma Samples far from H human habitation a Tanohama 0 lzuhara 0 OH H Fig. 3 Isotope ratio for coastal sea-water. See Fig. 1 for details 1.15 1.14 n 1.13 n a 2 a N. . 0 1.12 1.11 0 +i O0 0 0 * I I d 0.41 0.42 207pb:208pb Fig. 4 Isotope ratio for open ocean sea-water. See Fig. 1 for details hama have different isotopic data compared with those for the other samples. These results suggest that the lead in the Chiba New Port and Tsushima Omodahama samples which show distinct isotope data in both Figs. 1 and 3 has different origins from those for other the samples. Although further measurements of lead isotopes are necessary to elucidate the origin of the lead in sea-water samples the above results demonstrate that ICP-MS is useful for the measurement of variations in lead isotope ratios not only in coastal sea-water samples but also open ocean water samples.Concentration Measurements The lead concentrations in open ocean and coastal sea- water samples are summarized in Table 3. The concentra- tions were measured using the external calibration graph method with simple acid standards. The ICP-MS method gave simultaneously four concentrations which were mea- sured at four masses. As the four values agreed within an Table 3 Analytical results for sea-water samples Sample Pb concentrationlpg 1-I Open Ocean*- 48"N 175"E 44"N 175"E 38"N 175"E 30"N 175"E 27"N 175"E SON 175"E 0"N 175"E Coastal sea- water Sendai Tokyo Bay Sendai Port Chiba New Port 17 m depth St.22 22 m depth *Pacific Ocean. 0.046 0.044 0.040 0.043 0.044 0.840 0.043 0.087 0.203 0.167 Sample Coastal sea-water- Okinawa Naha Tomari Port Naha Tomari Port Naha near Tomari Port Awazure Hama Motobu Bize Cape Motobu Bize Cape Henoko Cape Tomakomai Port Tomakomai Beach Omodahama Izuhara Port Tanohama Miutahama Hokkaido Tsushima Pb concentratiodpg I-' 0.131 0.097 0.06 1 0.050 0.070 0.043 0.054 0.300 0.200 0.1 11 0. I30 0.080 0.078452 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY APRIL 1993. VOL. 8 Table 4 Blank detection limit accuracy and precision Parameter Pb concentrationlpg I-' Accepted Pb valuelpg I-' - 0.003 f 0.00045* - Blank Detection limit 0.0008 Okinawa Henoko Cape 0.04 1 4 0.005t 0.054 4 Cl.OO5t *Standard deviation of five determinations through the whole procedure.tStandard deviation of three determinations through the whole procedure. 0.039 f 0.006 - NASS-3 error of O.8% the concentrations given in Table 3 are the values measured at mlz 208. The lead concentration in open ocean sea-water was found to be lower than that in coastal sea-water. The results demonstrate that the precon- centration method using a chelating resin combined with ICP-MS is useful for the determination of lead in sea-water. Blank Detection Limit Accuracy and Precision The blank value detection limit accuracy and precision are shown in Table 4. The blank value was obtained by performing the same preconcentration procedure five times using 100 ml of water. The analytical detection limit (30) was calculated from the signal intensity of the fluid after the preconcentration of 0.1 pg 1-l Pb solution (final concentra- tion for ICP-MS measurement 1 pg 1-l) and the blank solution. TQ check the accuracy of the method NASS-3 was analysed.The standard deviations of blank NASS-3 and Okinawa Henoko Cape samples measured in our laboratory are based on the values for the complete procedure. The result for NASS-3 agrees well with the accepted value. The coastal sea-water (Okinawa) containing 54 ng 1-I of lead was analysed with a relative standard deviation of 9.3%. The detection limit obtained was 7.5 times better than that obtained using chelating resin concentration followed by hydride generation ICP atomic emission spectrornetry.l6 Conclusion The lead isotope ratios in sea-water were measured by ICP- MS with a precision adequate to be able to discuss the origin of the lead.Although the precision was worse than that of TIMS ICP-MS has the advantage of a higher sample throughput. Therefore the proposed method would be able at least to supplement TIMS for the purpose of measure- ment of lead isotope ratios in sea-water. In addition the method proved to be sensitive accurate and precise for the measurement of lead in sea-water. The authors thank Mr. K. Ishikawa and Dr. J. Ishizaka of this Institute for the sampling of open ocean water. References 1 Settle M. R. and Patterson C. C. J. Geophys. Res. 1982 87 8857. 2 Flegal A. R. and Stukas V. J. Mar. Chem. 1987 22 163. 3 Sturges W. T. and Barrie L. A. Atmos. Environ. 1989 23 25 13. 4 Hinners T. A. Heithmer E. M. Splitter T. M. and Henshaw J. M. Anal. Chem. 1987 59 2658. 5 Longerich H. P. Fryer B. J. and Strong D. F. Spectrochim. Acta Part B 1987 42 39. 6 Russ P. G. 111 and Bazan J. M. Spectruchim. Acta Part B 1987 42 49. 7 Dean J. R. Ebdon L. and Massey R. J. Anal. At. Spectrom. 1987 2 369. 8 Wang X. Viczian M. Lasztity A. and Barnes R. M. J. Anal. At. Spectrum. 1988 3 821. 9 Beauchemin D. McLaren J. W. Mykytiuk A. P. and Berman S. S. J. Anal. At. Spectrom. 1988 3 305. 10 Campbell M. J. and Delves H. T. J. Anal. At. Spectrum. 1989 4 235. 1 1 Mukai H. and Ambe Y. Bunseki Kugaku 1990 39 177. 12 Lang S. J. and Rosman K. J. R. Anal. Chim. Acta 1990,235 367. 13 Furuta N. J. Anal. At. Spectrorn. 1991 6 199. 14 Furuta N. Anal. Sci. 1991 7 823. 15 Murozumi M. Bunseki Kagaku 198 1,30 S30. 16 Reimer R. and Miyazaki A. J. Anal. At. Spectrom. 1992 7 1239. Paper 2/05319E Received October 5 1992 Accepted November 13 1992

ISSN:0267-9477    

DOI:10.1039/JA9930800449

出版商:RSC    

年代:1993

数据来源: RSC

摘要:

JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY APRIL 1993 VOL. 8 453 Multi-element and Isotopic Analyses of Iron Meteorites Using a Glow Discharge Mass Spectrometer Tadas hi S hi ma mu ra School of Hygienic Sciences Kitasato University 1 - 15- 1 Kitasato Sagamihara Kanaga wa 228 Japan Takako Takahashi Analytical Research Laboratory Marubun Corporation 3-3-4 Minamisuna Kotoku Tokyo 136 Japan Masatake Honda and Hisao Nagai Department of Chemistry College of Humanities and Science Nihon University Sakurajoshui 3 Setagayaku Tokyo 156 Japan Multi-element and isotopic analyses were performed of eight iron meteorites using a glow discharge mass spectrometer. Virtually all the elements can be determined. Major components (Fe Co and Ni) through to ultra- trace constituents at the sub ng g-’ level (Sc etc.) are measured directly with minimum chemical treatment within the same analytical cycle.Extremely high relative concentrations of Nil Cu Pd In Sn Sb Te Pb and Bi were found in the meteorite Yamato 791694; the concentration of C was also very high. The isotopic composition of Pb in this meteorite was close to primordial. Low concentrations of C N and 0 were found in the meteorite Gibeon. Very low S and Zn and high Cr and V were found in a third meteorite Chinga. An excess of 53Cr was observed in Yamato 75031 which was consistent with the cosmic ray production level expected from the concentration of 45Sc. Keywords Iron meteorite; glow discharge mass spectrometry; multi-element analysis Glow discharge mass spectrometry (GDMS) is now ac- cepted as a useful analytical method for the determination of trace level impurities in high purity metals alloys or semiconducting materials.For example manufacturers of high purity metals [such as Ga Al In or rare earth elements (REE)] use GDMS to certify the impurity level of metallic elements in their products. For the semiconductor industry GDMS is used not only for trace metallic element analysis but for gas components like oxygen or ~ a r b o n . * ~ ~ ~ ~ The advantages of GDMS are that (i) solid samples can be directly analysed with minimum chemical treatment (ii) virtually all the elements can be determined (iii) the wide dynamic range of the detector allows the determination of both major components and trace constituents within the same analytical cycle and (iv) decoupling of the atomization and ionization processes results in uniform sensitivities for many elements and also minimum matrix effects.A major disadvantage of GDMS is that the sample must have electrical conductivity. Iron meteorites are unique objects which are natural Fe Ni and Co alloys containing many other minor and trace elements. Many reports are available for the elemental analysis of iron meteorites and the data are used to classify the meteorites and interpret their origin. Because most of the data were obtained by neutron activation analysis (NAA) the number of elements determined was rather limited. Since iron meteorites have good electrical conduc- tivity GDMS would be an ideal analytical method for those objects however only a few reports where GDMS has been used have been published4-’ because availability of the instruments is still limited in the geochemical field.The feasibility of applying GDMS to the elemental and isotopic analysis of iron meteorites is investigated in the present study. Experimental Instrumentation A glow discharge mass spectrometer (VG9000 VG Elemen- tal UK) was used for both elemental and isotopic analyses. The basic principles of GDMS are described in ref. 1. Typical analytical conditions are summarized in Table 1. The discharge cell was cooled by liquid N2 to reduce background ions. This is particularly effective for the determination of C N and 0. Ionized atoms are accelerated to 8 keV and injected into an inverse geometry double focused mass analyser. Standard mass resolution was set to 5000. Ions are detected by a combination of Faraday cup and Daly ion-counting system which has a dynamic range of 1 1 orders of magnitude.Generally the main components are detected by the Faraday cup and minor and trace elements are detected by the Daly detector. The counting efficiency of the Daly detector was calibrated using the 47Ti ion beam. Samples Eight iron meteorites were analysed Yamato 79 1694 (IAB? anomalous) Yamato 7 503 1 (anomalous) Canyon Diablo (IA) Odessa (IA) Tocopilla (IIA) Guin (anomalous) Gibeon (IVA) and Chinga (IVB) where IA IIA etc. denote the iron meteorite group. The samples were sawed to be a 2 x 2 ~ 15 mm3 pin shape etched by HN03- HC1-H20( 1 + 3 + 4) solution for 1- 10 min and rinsed with H 2 0 and acetone. Because GD can remove surface contam- ination no further cleaning was required.Elemental Analysis For the elemental analysis a suitable isotope of each element was selected. General rules of selection of the isotope(s) for each element are (i) most abundant (ii) free of Table 1 GDMS operating conditions Discharge current/mA 3 Discharge voltage/kV 1 Pre-sputter current/mA 5 Pre-sputter time/min 50 Matrix ion current/nA 2-5 Mass resolution 5000454 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY APRIL 1993 VOL. 8 isobaric interferences and (iii) least polyatomic or multiple charged ion interferences. The ions detected for the selected isotope peak were integrated to give total ion counts. For a major component the ions were detected by a Faraday cup as an ion current.In this case integrated ion current was also converted to number of ion counts. The ion beam ratio (IBR) of an element X is defined as follows where Xi is total ion counts of the selected isotope i of the element x cxi is isotopic abundance of the isotope i A k iS total ion counts of the selected isotope k of the matrix element A cAk is isotopic abundance of the isotope k. Since sputtering rate ionization efficiency and detection effici- ency of the ions are similar for most of the elements the IBR gives an approximate atomic concentration of the element X relative to the element A. For quantitative analysis however relative sensitivity factors (RSFs) must be known. For convenience RSF is defined as RSF(X) = IBR(X)/m(X) (2) where m(X) is mass concentration of the element X (g g-l).To obtain RSFs National Institute of Standards and Technology (NIST) Standard Reference Materials (SRMs) 662 LA Steel 663 LA Steel Cr-V(mod.) and 665 LA Steel Electrolytic Iron were analysed. All the RSFs were deter- mined relative to Fe. Thus a value for m(X) of elements in the meteorites relative to Fe can be obtained. Because Fe+Ni+Co exceeds 99% of the total mass the mass concentration of the element X in the sample relative to Fe + Ni + Co was calculated. The RSFs in some cases from NAA were also estimated. The measured RSFs are sum- marized in Table 2. Generally the RSFs are consistent with each other but for P S Ge As Se and Hf the RSF values are not in good agreement. These disagreements are considered to be caused mainly by sample heterogeneity.In Table 2 RSFs from Mykytiuk et al.z are also listed. Although their values were obtained from a different type of discharge cell which has a smaller cell volume agree- ment between the RSF values is fairly good. On comparing values obtained by other workers using NAA5v8*9 the values of the RSFs for Ni Cu Ge As and Hf were modified. An RSF of 0 was obtained from Japanese Steel Standards (JSS) Certified Reference Material (CRM) GS lc 2b 3b and 5a. The RSF values for Sc Ga Ge Ru Rh Pd In W Re Os Ir Pt and Au were obtained from NAA data from Honda et ~ 1 . ~ The RSFs of the following elements are not available yet; Li Be N F Na C1 K I Cs Ba REE (except La Ce and Pr) Hg T1 Th and U. Data accumulations were performed as follows. Prior to data acquisition mass calibration was carried out for the whole mass range using several dominant or obvious mass peaks such as 40Ar3+ 40Ar2+ *(jFe+ 40Arz+ 56Fe2+ 194Pt+ and 18*Ta40Ar+.The magnetic field was set to the lower mass side of the selected isotope then an accelerating voltage scan was performed (peak jump and scan); 1-10 scans were carried out depending on the expected concentration. In this case one scan took 30 s. The peak at 56Fe+ was measured at the beginning at 1 h intervals throughout and at the end in order to monitor the intensity then linear correction of the beam variation was made. Peak identification can be done by either calculating the exact mass difference between interfering peaks or checking against the other isotopes. Repeatabilities of IBR were usually constant within 1 O% except for lithophile elements (Na Mg Al Si and K) B Ag Hg halogens and sometimes In Sb or Pb.Because iron meteorites have texture element distributions are some- times heterogeneous and the sampling area of GDMS is Table 2 Relative sensitivity factor (RSF) (measured IBR/concentratiion in SRM by mass) RSF Element B C 0 Mg A1 Si P S V Cr Mn Fe c o Ni c u Zn Ge As Se Ag Sn Sb Te La Ce Pr Hf Pb Bi NIST SRM 662 NIST SRM 663 NIST SRM 665 0.668 1.30 0.846 0.245 0.173 0.143 0.7 1 0.7 1 - 1.06 - 0.906 0.339 0.52 0.58 0.245 0.36 0.156 0.273 0.6 1 0.342 2.08 2.38 2.50 0.594 0.553 0.597 0.935 0.835 0.930 1 .oo 1 .oo 1 .oo 1.1 1 0.958 0.957 0.688 0.581 0.6 15 0.192 0.170 0. I48 0.158 0.162 - 0.20 0.12 - 0.1 13 0.20 0.275 0.267 0.44 - 0.264 0.297 - 0.65 0.47 - 0.13 0.17 - 0.2 1 0.29 - 2.45 2.57 - 2.47 2.7 1 - 3.0 2.0 - 0.19 0.34 - 0.488 0.418 - 0.29 0.2 1 - - - - Average 0.94 0.16 0.013 0.7 1 0.98 0.48 0.25 0.41 2.3 0.57 0.90 1 .oo 1 .o 0.63(0.68)? 0.17(0.13)? 0.16 0.16(0.36)? 0.35 0.28 0.56 0.15 0.25 2.5 2.6 2.5 0.2 7(0.3)1.0.43 0.25 0.20(0.12)? Ref.* 1 .o 0.25 0.027 0.67 0.9 1 0.63 0.56 0.56 2.0 0.50 0.7 1 1 .oo 1.0 0.7 1 0.25 0.50 0.33 0.26 0.42 0.29 0.33 2.5 2.8 - - - 0.34 0.2 1 *Mykytiuk et ?Numbers in parentheses are modified by NAA data.JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY APRIL 1993 VOL. 8 45 5 Table 3 Interference and required mass resolution Isotope 1 6 0 19F 24Mg 28Si 32s 39K 40Ca 44Ca 45sc 46Ti 47Ti 48Ti 49Ti 50Ti 51v Y r 69Ga ?IGa 72Ge 74Ge 77Se 79Br 85Rb 86Sr 88Sr 89Y 91Zr 2 7 ~ 1 902r Interference Required resolution 36Ar3 + 1000 32s2+ 1800 38Ar2+ 1100 12c + 1 600 54Fe2+ 2 300 1560 1 6 0 + 1800 38A;l H + 5 700 40Ar+ 190 000 12C160 + 1300 13C160 + 1200 94M02+ 6 0000 98M02+ 11 000 I 02Ru2 + 6 200 40ArlZC+ 104Ru2 + 2400 600 31p38Ar+ 6 300 3 1 p40Ar+ 6 200 3aAr2+ 5 500 36Ar38Ar+ 8 100 36A140Ar* H +? 4 300 38A140Ar1H+? 5 400 54FeP+ 54 000 64 000 57FeP+.25 000 58NiP+ 27 000 59C~P+ 40 000 60NiP+ 83 500 12C160+ 56Fe2+ Y 31 500 57 500 9 10000 1 1 000 21 000 5 600 46Ca+ 92M02+ 48Ca 96M02+ 50V+ 1OOMo2+ 86=+ Interferences 62NiP+ FeAr+ CoAr+ FeAr+ CoAr+ NiAr+ CoAr+ NiAr+ CuAr+ NiAr+ CuAr+ Io2Pd+ NiAr+ CuAr+ CuAr+ ZnAr+? GaAr+ GaAr+ GeAr+ GaAr+ II5Sn+ Co$e+ NiFe+ Ar3+ Ni2+ Co2+ FeNi+ CoNi+ Ar3+ Ni2+ FeNi+ CoNi+ Ni2+ Ni2+ NiCu+ FeGa+ Ni2+ NiCu+ FeGa+ FeGa+ FeGe+ FeAr2+ FeAr2+ CoAr2+ NiAr + NiAr + Ni3+ 1920s40Ar 198P t40Ar Il2Sn+ 56Fe + l8lTal60 Required resolution 21 800 1 2 000- 1 8 000 10 000- 1 3 000 8 200- 1 1 300 7 600-9 900 8 100 7 500 6 900-7 600 7 100 7 000 6 400-8 300 6 200-7 900 6 500-7 000 54000 3400 2 16 000 3 300 3 500 8 400 3 300 3 400 3 000 8000 3000 2 900-3 000 2 800-3 200 2 800-3 100 2 900 3 300 3 000 2 900 3 000 3 000 1200 8 300 2 000 2 000 rather small (several tens of mm2) resulting in some variations of the element concentrations.This is also true for the NIST standards. The accuracies of the determina- tions were estimated to be 10-40% mainly caused by uncertainties in the RSFs and sample heterogeneity. Generally Li Be I Cs Ba REE Hf TI Th and U were present below the detection limits which were estimated using three times the standard deviation of the blank counts and the matrix current. Detection limits are dependent on isotopic abundance integration time background counts and matrix current.A matrix current of 2 x A was usually obtained and a background current of 1 x A. With an integration time of 300 s the practical detection limits were from 1 x to 1 x g g-l. Special caution must be used for the following elements. For C N and 0 surface contamination and instrumental background are significant therefore a longer pre-sputtering time is required. Usually after the general elemental analysis had been performed C N and 0 were determined. Consequently the effective pre-sputter time was 2-3 h. Concentrations of those elements decrease gradually and finally level off.The data of the final stage only were selected. However instrumental background which possibly comes from either Ar gas or the discharge cell is not fully understood yet. The estimated background for C and 0 from steel and semiconductor analyses was less than 0.1 pg g-' equivalent. Unfortunately the RSFfor N is not known so no clear idea of the background level of N is possible. For B Ag and Hg surface contamination seems to be significant. In many cases the concentration of these elements decreases steadily during runs and sometimes the signals disappear. It is possible that these elements tend to diffuse into the interior of the meteorites. Data were selected only when concentrations reached a constant level but still it is unclear whether the Ag concentration is sufficiently reliable.The elements R Rh Pd and Au suffer interferences from masses which are close to their isotopes. Thus the resolution is set to 6500. Increasing resolution sacrifices sensitivity but because the concentrations of these elements are relatively high sufficient ion intensities can be obtained. A typical mass spectrum near losPd with a resolution of 6500 is shown in Fig. 1; loSPd+ and a5C~40Ar+ are clearly resolved. Indium has two isotopes. Unfortunately both of the isotopes suffer isobaric interferences. Indium- 1 1 5 and II3In are overlapped by l15Sn and Ir3Cd respectively. The isotopic abundance of l131n is too small (4.2%) to have sufficient sensitivity so ll5In must be used. However the concentration of Sn in iron meteorites is usually 2 orders of magnitude higher than that of In.Thus the correction of llsSn on l151n was 5O-8O0/o. Interferences Although GDMS is a powerful tool for trace analysis of metal there are many sources of interference particularly 104.86 104.88 104.90 104.92 104.94 mlz Fig. 1 Typical mass spectrum near lo5Pd with a resolution of 6500 showing lo5Pd+ lines and 6SC~40Ar+ lines clearly resolved456 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY APRIL 1993 VOL. 8 with determinations below the ng g-' level therefore great care must be taken. Firstly isobars cannot be resolved with this instrument. Common interfering species are poly- atomic ions which associate with Ar (discharge gas) and matrix elements (Fe Co and Ni in this work) Namely Ar2+ FeNi+ ArFe+ ArCo+ etc.Multiple charged ions of those elements such as Fe2+ Ar2+ Ar3+ etc. are also common. Even a multiple charged ion or a combination of minor or trace level components like Mo2+ FeP+ or CuAr+ can cause problems for a constituent present at the sub-ng g-I level. Thus some particular elements cannot be determined. Possible interferences and required mass reso- lutions are summarized in Table 3. If the required resolu- tion exceeds 10 000 analysis will be difficult. Consequently Ti Rb Sr Y Zr Nb and Mo are difficult to determine. For Se discrepancies were sometimes observed between GDMS and NAA data. It is possible that there were some unknown interference(s) on 82Se.*o Because the discharge cell is made Table 4 Elemental concentrations in iron meteorites; numbers in parentheses are IBR values; values preceded by t are less than the detection limit (see text) Concentratiodg g-l Element (isotope) 7Li 9Be IlB 'ZC l 4N 1 6 0 19F 23Na 24Mg 28Si 31P 32s 35c1 39K 45sc 5'V S6Fe 60Ni 63Cu 66Zn 69Ga 73Ge 7 5 A ~ 82Se lo3Rh 2 7 ~ 1 5 9 c 0 101,102R~ 105,106pd 107J09Ag 1151n 118,119Sn Iz3Sb 2STe ' 3 3 0 137Ba 39La 140Ce 141Pr 157Gd 159Tb 169Tm 172Yb 182W lssRe 1271 178,179Hf 1890s 1911r 194Pt 1 9 7 A ~ 202Hg 208Pb 232Th 2 0 5 ~ 1 209Bi 238U Y 79 1694 (68) ((5.0 x lo-") ((6.0 x lo-") 2.3 x 10-9 3.2 x 10-4 (7.7 x 10-5) 3.2 x 10-5 (1.4 x (I.0x 10-6) 2.3 x 2 .4 ~ 10-7 3.8 x 9.0 x 10-4 1 . 3 ~ (1.1 x 10-5) ( 1 . 3 ~ 10-7) 1.9 x 10-lo 1 . 4 ~ 10-9 4 . 4 ~ 10-7 6.3 x 10-1 5 . 2 ~ 10-3 3.4 x 10-1 2.2 x 10-3 3 . 8 ~ 10-5 2.1 x 10-5 3 .6 ~ 10-5 1 . 6 ~ 10-5 7.2 x 10-7 5 . 3 ~ 10-7 2.9 x 10-7 1.4 x 10-5 2.5 x 10-7 1.2 x 10-7 9.0 x 10-5 5 . 6 ~ 10-7 ((7.9 x 10-11) ( ( 8 . 6 ~ 1O-I1) ((4.2 x 10-9) ( 8 . 0 ~ lO-*l (3.5 x 10-10 ( 6 . 0 ~ lo-" ( ( 2 . 4 ~ 10-lo) ( ( 5 . 8 ~ lo-") ( t 2 . 8 x 1O-l1) ( ( 3 . 0 ~ 10-lo) (5.3 x 10-10 4.3x 10-8 2 . 4 ~ 1 . 8 ~ 10-7 2.2 x 10-7 2 . 0 ~ 10-7 5.2 x 10-7 2.0 x 10-6 ( ( 1 . 8 ~ 10-lo) ( t 7 . 5 x 10-1') 8.6 x 2.8 x ( t 4 . 2 ~ lo-") ( t 4 . 2 ~ lo-") Y75031 (52) - - 5.4 x 10-8 8.8 x ( 1 . 8 ~ 1 . 9 ~ 10-4 ( 1 . 3 ~ 10-7) 2.8 x 10-7 3 . 5 ~ 10-7 2.3 x 10-3 - 5.5x 10-8 5.4 x 10-6 - - 2.3 x 10-9 8.7 x 10-9 4.2 x 10-7 8.5 x 10-1 6.1 x 10-3 1 . 4 ~ 10-l 4.8 x 10-4 4.4 x 10-6 3.4 x 10-5 2.3 x 10-4 1 . 5 ~ 10-5 2.1 x 10-7 1 . 2 ~ 10-5 5.7 x 10-7 7 . o ~ 10-9 8 . o ~ 10-7 7.8 x 1 .4 ~ 5.0 x 8 . 0 ~ - ( t 2 . 5 x lo-") (3.3 x 10-11 t 8 . 5 x 10-l' t 1 . 4 ~ ( t 5 . 9 x 10-11) ( t 9 . 0 ~ 10-l2) ((8.0 x 1 0-l2) ( t l . 1 x 10-10) t 2 . 1 x 10-9 1.2x 10-6 4.7 x 10-8 3.3 x 10-7 4.5 x 10-7 5.3 x 10-6 1 . 9 ~ (2.2 x 10-10) ( t 3 . 5 x 10-11) 1 . 8 ~ 10-9 t 1 . o x 10-10 ( t 2 . 0 x 10-11) ( t 2 . 0 x 10-11) ( ( 1 . 4 ~ 10-9) Canyon Diablo (H34-503 1) - 3.7 x 10-10 (6.5 x LOX 10-4 9.2 x 10-5 (9.7 x 10-9) (1.1 x 10-7) 4.4 x 10-9 1 . 4 ~ 10-3 1.8 x 3.5 x 10-8 1.1 x 10-6 ( 1 . 6 ~ (5.1 x 3.0 x 10-lo 5.6 x 9.3 x 10-1 6.9 x 1.8 x 10-9 5.1 x 10-3 2 . 3 ~ 10-4 2.5 x 10-5 2.1 x 10-4 4 . 2 ~ 1 0 - 4 7.5 x 10-6 2.3 x 4.4 x 10-6 1 . 3 ~ 4 . 0 ~ 3.6 x 7.9 x 10-6 2.1 x 10-7 6.8 x - ( t 2 . 3 ~ lo-") ( t 2 . 4 ~ 10-l1) ( t 2 . 0 ~ 10-9) t 4 .0 x t 7 . 7 x 10-11 t 1.4 x ( t 5 . 6 ~ 10-l1) ( t 8 . 5 x 10-I2) ( t 8 . 0 x 10-l2) ( t 1 . 2 x 10-10) t 2 . 9 x 1 0 - 9 9.2 x 10-7 3.3 x 10-7 2.2 x 10-6 2.4 x 6.4 x 1.3 x ( t 5 . 4 x 10-11) ((3.3 x 10-11) t 9 . 2 x 10-l1 ( ( 1 . 6 ~ lo-") ( ( 1 . 6 ~ lo-") 1.8 x Odessa (H9 1-202) - - - 2.6 x ( 1 . 4 ~ 10-7) 4.9 x 10-5 (5.1 x (9.8 x 2.3 x 10-7 2.2 x 10-5 1.9 x 3.0 x 10-3 2 . 4 ~ 10-5 ( 1 . 7 ~ 10-5) (1.9 x 10-7) 6.1 x 10-9 5.2 x 10-3 1 . 4 ~ 10-4 2.3 x 10-5 6.4 x 10-5 2.6 x 10-4 1 . 4 ~ 10-5 4.6 x 10-lo 8.2 x 9.2 x 1Q-l 7.4 x 10-2 2.0 x 10-8 4.3 x 10-6 1.5 x 4.0 x 3.1 x 3.8 x 6 . 0 ~ 7.8 x 10-9 3.0 x 10-7 ( t 3 . 3 x 10-11) ( t 3 . 3 x 10-11) ( ( 1 . 9 ~ 10-9) t 2 . 4 ~ lo-" t 6 . 5 x t 9 . 6 x 10-l2 ( t 7 . 8 x lo-") ( t 1 . 2 x 10-11) ( t 1 . 2 x 10-11) ( t 7 . 7 x 10-11) 4 .2 ~ 10-9 1.2 x 10-6 2.7 x 10-7 2.5 x 2.2 x 10-6 5.0 x 1.5 x ( 1 . 9 ~ 10-9) 1.1 x 10-7 (<1.2x 10-11) ( t 1 . 2 x 10-11) ( t 4 . 6 x 1 O-") < 1.3 x 10-lo Tocopilla (<3.4x 10-11) (<3.2 x 5.8 x 4.9 x 10-6 (6.8 x lo-" (2.7 x 10-7) (4.9 x 10-9) - t 2 . 7 x 1.3 x 10-8 t 7 . 9 x 10-10 5.2 x 10-3 1 . 6 ~ - - 3.6 x 10-lo 6.5 x 10-5 4.7 x 10-3 1 . 4 ~ 10-4 5.7 x 10-5 1.7 x 10-4 1 . 6 ~ 10-5 9.2 x 10-1 5.3x 10-2 1 . 8 ~ 5.3 x 10-6 1.2x 10-8 2.4 x 2.2 x 10-6 1.6 x - 1.2 x 10-7 5.5 x 10-8 1.5 x - ( t 5 . 3 x 10-11) ( ( 1 . 2 ~ 10-9) t 2 . 3 x t 4 . 6 x lo-" ( 1 . 3 ~ lo-" ( t 1 . 3 ~ 1O-lo) ( ( 1 . 9 ~ lo-") ( ( 1 . 7 ~ lo-") ( t 9 . 1 x lo-") 4 . 7 x 10-lo 2.7 x 2.8 x 10-7 1.2x 10-6 3.5 x 10-6 2.3 x 10-5 5.5 x 10-7 (t1.Ox 10-10) ( t 7 . 3 x 10-11) t 2 . 3 x (2.1 x 10-10 ( t 4 .1 x lo-") ( t 4 . 1 x 10-11) continued-JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY. APRIL 1993. VOL. 8 457 Table 1 conrinued- Element (isotope) 7Li 9Be IlB 12C 14N 1 6 0 19F 23Na 14Mg 28Si ,IP 32s 39K 45sc 51v 52Cr 56Fe 60N i 63Cu 66Zn 69Ga 73Ge 7 s A ~ 82Se Io3Rh 2 7 ~ 1 3 5 c 1 59c0 lOl.lO2R~ 105,106pd 107*'09Ag ll51n 123Sb lzsTe I33Cs L37Ba 139La lroCe I4IPr IS7Gd lS9Tb la9Tm 17*Yb 182W IssRe I 18,119Sn 1271 178,179Hf 1890s 1911r 194Pt 1 9 7 A ~ 202Hg 2osPb 232Th 2 0 5 ~ 1 209Bi 238U Guin (< 1.9 x 10-I1) ((1.8X 10-11) 3.5 x 10-'0 4.4 x 10-6 ( 1 . 8 ~ ( 1 . 6 ~ 2 . 6 ~ - - 5 . 8 ~ 10-9 3 . 3 ~ 10-3 - 9.5x 10-6 - - 1.5 x 10-lo 1 . 3 ~ 10-9 1.8x 1 0 - 6 8.9 x 10-I 6 . 3 ~ 10-3 1.ox 10-1 2.5 x 10-4 4.1 x 4.3 x 10-5 1 . 4 ~ 10-4 1 .3 ~ 10-5 2.0 x 10-8 LOX 10-5 2 . 6 ~ 5.0 x 1.7 x 10-7 7.5 x 10-7 1.3 x 10-7 - 2.4 x - - ( ( 1 . 6 ~ 10-9) (3.4 x 10-11 (8.8 x 10-11 (1.2X lo-" ((7.1 x 10-Ii) ((1.1 x lo-") (<8.5x 10-I2) ( t 7 . 5 x lo-") 4 . 2 ~ 10-9 2.0 x 10-6 1.ox 10-6 9.0 x 7.1 x 1.2 x 10-5 1.2x 1 0 - 6 ( 4 . 0 ~ I O - I O ) ((4.1 x 1 . 3 ~ 10-9 < 1.2 x 10-10 - - Gibeon ( t 3 . 2 x I O - I I ) ((2.9 x < 1.6 x 1 O - I l 5.2 x 10-7 ( 1 . 7 ~ 2.2 x 10-7 ((1 .6 x ( t 1 . 2 x 10-10) t 2 . 0 x 10-'0 2 . 9 ~ 1.1 x 10-7 3.5 x 1 0 - 4 (4.0 x 10-9) 6.5 x I O - I I 3.7 x 10-7 3.0 x 10-4 4. I x 10-3 1.6 x 10-4 1.1 x 10-7 2.4 x ((3.0 x 1 O-IO) 9.2 x 10-I 7.5 x 10-2 1.9 x 8.9 x 2.6 x 5.7 x 10-9 4.4x 10-6 1.1 x 10-6 3.7 x 10-6 (8.6 x 10-Io - 7 . 7 ~ 10-9 t 7 . 3 x 10-'0 t 1 . 6 ~ 10-9 ( t 4 .9 x lo-") ((4.9 x 10-11) ((2.2 x 10-9) (2.2 x lo-" t 6 . 5 x I O - I I <1.2x lo-" - ( t 1 . 2 x 10-'0) ((1.8X 10-11) ( t l . 1 x 10-10) ( 1 . 4 ~ 10-9 5.6 x 10-7 3.7 x 10-7 ( ( 1 . 8 ~ IO-II) 2.8 x 2.2x 10-6 4.4 x 10-6 7.6 x 10-7 ( ( 1 . 6 ~ 10-lo) ((6.9 x 1 O-") t 2 . 2 x 10-10 (2.0 x 10-'0 ( ( 1 . 5 ~ 10-l') ( ( 1 . 5 ~ *Obtained from NAA. ?Modified after comparison with NAA data. Chinga - - (3.3 x 10-10 2.5 x (2.3 x 4.3 x 10-7 - - - - - 9.6 x 10-4 9 . 6 ~ 10-9 - - 3.3 x 10-10 7.4 x 10-6 8. I x 10-4 8.4 x 10-I 6.3 x 10-3 1 . 5 ~ 10-I ISX 10-5 4.6 x 2.1 x 10-7 6.9 x 2.8 x 6.1 x 2.0x 10-6 9.0 x 6.1 x - - (3.2 x 10-9 ( 2 . 4 ~ 10-9 ( 5 . 2 ~ 10-9 - - - - - - - - - - - 7.1 x 10-7 1.5 x 1.2 x 10-5 5.1 x 1 . 1 x 10-5 4.8 x 10-7 (<3.3x 10-10) ( t 2 . 3 x 1O-Io) t 4 .2 x 10-lo (3.9 x 10-10 - - RSF - - 0.94 0.16 0.01 3 - - - 0.7 1 0.98 0.48 0.25 0.4 1 - - 2.0* 2.3 0.57 1 .o 1 .o 0.68t 0.13t 0.16 0.28* 0.36* 0. I 2 t 0.35 I .8* 0.66* 0.3* 0.28 0.5* 0.56 0.15 0.25 - - - 2.5 2.6 2.5 - - - - 0.37 0.48* 0.6* 0.4* 0.55* 0.36* 0.29* - - 0.43 0.25 - - from Ta a Ta blank is too high to measure indigenous Ta. It is also difficult to measure 55Mn because of the high base line from 56Fe and 54Fe tailing. Isotopic Analysis The glow discharge mass spectrometer is not really a suitable instrument for precise isotopic analysis. This is because the ion source is not stable enough owing to flickering of the glow discharge. The inhomogeneous distri- bution of an element in a sample alos directly affects variations in ion intensity. Hence sufficient precision cannot be achieved. However since chemical separation is not necessary GDMS is useful to obtain preliminary results or for samples in which large isotopic variations are expected.Lead isotopic analysis was performed on Yamato 79 1694 expecting the primordial composition of Pb. Chromium isotopes were also analysed on Yamato 75031 in which components produced by a cosmic ray were expected.458 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY APRIL 1993 VOL. 8 Pb composition At the beginning resolution was set to 50000 to check possible interferences then reset to 1300 to give sufficient was Of the Same number of channels were integrated. No dead time correc- tion was made because the maximum count rate on 208Pb was less than 30 000 counts s-l and the resolving time of the Daly counter was of the order of 1 X amount of information available it appears that Sn and In concentrations of this meteorite are the highest.12.13 Also perhaps this meteorite may be the only example of a meteorite containing significant amounts of pb and Bi in data since Pb and Bi are difficult elements to determine by NAA. width in the top of the flat peak- A jump and Scan procedure the metal phase. This might be simply because of lack of adopted and Only the flat top s.Mercury-202 Yamato 75032 (AN) was-monitored for 204Hg correction on 204Pb. No significant interferences were observed at mlz 206 207 and 208 and a small correction was made on 204 (from 2 to 0.3%) from 204Hg which seemed to result from surface contamination. Beam variation was monitored by the 56Fe peak and a linear correction was made.Lead isotopes were also measured in the NIST SRM 465 Spectrographic Ingot Iron E which contains approximately the same amount of Pb (400 ng g-I) as a reference of 'common' Pb to check possible detector bias. No significant detector bias was observed. The data acquisition was repeated twice. At the first measurement acquisition times for 204Pb 206Pb 207Pb and 208Pb were set to 600 150 150 and 150 s respectively. The main contribution to the fluctuation of the data however is not due to counting statistics but to intensity fluctuations of the ion beam during the measurements. Hence for the second run acquisition times were set to 210 60 60 and 30 s respectively for the isotopes mentioned above.Since the isotopic standard of Pb was not measured a mass discrimi- nation factor is not available. Cr composition Because 50Cr and 54Cr were overlapped by sOV SoTi and 54Fe only 52Cr and 53Cr were measured. Since both of the isotopes were interfered with by Arc mass resolution was set to 2030. With this resolution a flat top peak was still obtained. Acquisition times for and 53Cr were 150 and 600 s respectively. Results and Discussion Elemental Analysis The results of elemental analysis of eight iron meteorites are presented in Table 4. The values in parentheses are IBRs since RSFs for these elements are not available. However deviations from true concentrations are thought to be within a factor of ten. It is evident from Table 4 that more than 50 elements can be routinely analysed using GDMS.Yamato 791694 (IAB? anomalous) Trace element concentrations of this meteorite are highly anomalous. Remarkable points are as follows. (i) Very high Ni (34%). (ii) Very high concentrations of Cu (2200pg g-l) In (0.12 pg g-l) Sn (90 pg g-l) Sb (2.4 pg g-l) Te (0.56 pg g-l) Pb (0.86 pg g-l) and Bi (28 ng g-l). Most of those elements are chalcophile but the concentrations of some other chalcophile elements like Zn or As are not particularly enriched. Tellurium and Hg are below the detection limit. The concentration of S (1.3 pg g-I) is rather low indicating that trace chalcophile elements do not reside in the sulfide phase. The value for Cu is consistent with the results of Wasson et al.ll (1943 pg g-l) but for Sb is somewhat lower than their value (4 pg g-l).(iii) High concentrations of C (320 pg g-l) and N (7.7 x IBR). (iv) Low Cr (0.44 pg g-' and V (1.4 ng g-l); Cr concentration is far lower than the value reported by Wasson et al." (8 pg g-l) which will be discussed later. ( v ) High Pd (1 4 pg g-l) low Rh (0.29 pg g-l) and Ru (0.53 pg g-l); Rh:Pd=0.021 Ru:Pd=0.038 are distinctly lower than other meteorites. With the limited The concentration of B (54 ng g-l) is more than one order of magnitude higher than for other meteorites. Also the content of 0 is high (190 pg g-I) but N is low (1.8 x IBR). A Sc concentration of 2.3 ng g-l which could be attributed totally to a cosmic ray p r o d ~ c t ~ was found. Considering a Cr concentration of 0.42 pg g-l about a 50% excess of s3Cr produced by cosmic ray would be expected as is discussed later.Wasson et al." claimed that they confirmed pairing of Yamato 75031 and Yamato 791076. The data obtained here are consistent with the data of Wasson et al." for Yamato 791076 except for Cr (15 pg g-l). The value for Pd is two times higher (12 pug g-l) than that obtained using NAA (5.3 pg g-l).s Canyon Diablo (IA) Lithophile element contents in Canyon Diablo are more variable than other meteorites analysed. Carbon ( 100 pg g-l) is much higher than in Odessa (2.6 pg g-l). The Ga level is three times higher than that found by other ~ o r k e r s . ~ J ~ The reason is not clear but present values seem to be in error. Agreement between the data obtained by Scott14 and those obtained here is rather poor. Odessa (IA) Lithophile elements are very variable in Odessa.Concentra- tions of Pb and S are rather high but very variable. This suggests that Pb may reside in micro-sulfide inclusions. Tocopilla (IIA) Very high levels of Ru (1 6 pg g-l) were found in Tocopilla and Ru:Pd (7.3) and Rh:Pd (1.1) are also very high. Data agreement between Willis and Wasson,15 Smales et all3 and the results obtained in the present study is excellent. Guin (AN) Ruthenium (1 0 pg g-I) and Rh (2.6 pg g-I) levels are high in Guin. Data agreement with Wasson et al.II is very good except for Cr. Gibeon (IVA) Trace element concentrations are generally low in Gibeon especially C (0.52 pg g-l) and 0 (0.32 pug g-l). Zinc (0.1 1 pg g-l) and Sn (9.9 ng g-l) are also low and variable. Indium Sb and Te are below the detection limits.However Cr (300 pg g-l) and V (0.37 pg g-l) are rather high. The concentration of Cr is higher than that found by Willis and WassonIS (192 pg g-l) and Smales et all3 (201 pg g-l) but results for other elements which they determined are consistent with the data obtained here. Chinga (IVB) Carbon N and 0 levels are low in Chinga. Concentrations of S (9.6 ng g-l) and Zn (46 ng g-l) are very low. Copper Sn Sb and Te levels are also very low but Cr (810 pg g-I)JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY APRIL 1993 VOL. 8 459 and V (7.4 pg g-') are very high. The 0s:Pt ratio (1.1) is relatively high. Cosmochemical Consideration of Yamato 791694 Detailed discussion of the cosmochemical aspects of Yam- ato 79 1694 will be described in a future paper.However the origin of the unique iron meteorite Yamato 791694 is considered briefly. In summary the specific properties of the meteorite are as follows (i) rich in Ni; (ii) Cu In Sn Sb Te Pb and Bi are highly enriched; (iii) rather poor in S; (iv) highly volatile elements Zn Hg and T1 are not particularly enriched; and (v) platinum group elements except Pd are rather depleted. Smales et all3 measured 11 elements in Santa Catharina and San Cristobal. These two meteorites have similar elemental distribution to Yamato 79 1694. They found that the S concentration of those meteorites was very high and variable (236-1339 ppm for San Cristobal and 740-5596 ppm for Santa Catharina). They pointed out that the high concentrations of Cu Sb and As in those meteorites could be atributed to the presence of small troilite inclusions.In the case of Yamato 791694 the presence of troilites is not expected because of the low concentration of S. Considering the above the following assumptions can be made. (a) The original material which was poor in S was in a moderately reducing environment; lithophile elements including part of the Fe separated from the other non-lithophile elements. (b) During the fractionation processes (condensation or melting) from other iron meteorites refractory siderophile elements were extracted or condensed earlier; partial removal of sidero- phile elements occurred. (c) At the final stage of fractiona- tion chalcophile elements were still in the residual metallic phase since those elements could not be removed in sulfides because of the lack of S thus most of the chalcophile elements remain in this meteorite whereas the removal of these elements coupled with S took place in ordinary iron meteorites.(4 Zinc Hg and T1 were still not condensed yet or vaporized during the re-heating process which happened later highly volatile elements did not join with other chalcophile elements. On the whole it can be speculated that decoupling of S and enrichment of the chalcophile elements took place by some unknown mechanism in the early stage of the history of this meteorite. The hypothesis can be extended at least qualitatively as follows. The so- called chalcophile nature of these elements might not be observed at the earlier higher temperature stage. Many elements could be present in the form of an alloy with Fe and Ni rather than combined with S.(Even at a later stage some elements are still partly siderophiles.) The sulfides could be separated mechanically because of having a lower density than metal under a gravitational field or an equivalent field. This separation could occur by the mecha- nism described above. In space in most cases such density separation might not occur and sulfides might coexist with the metal accompanying many 'chalcophile' elements. In the last stage the troilite bubbles were separated and removed from the Fe thus extracting most of chalcophile elements from the metal phase. Isotopic Analysis Lead During the course of elemental analysis very high concen- trations of Pb were found in Yamato 791694. Because the concentrations of U and Th which are parent nuclei of Pb isotopes are very low Pb isotopic composition must be primordial. Tatsumoto et af.reported Pb isotopic compo- sition in Canyon Diablo troilite. According to their report the concentration of Pb in the troilite phase was 8 pg g-' Table 5 Pb isotopic ratios of Yamato 791694; errors quoted are standard deviation (1 a) 206pb:204pb 207pb:204pb 2OSPb:204pb Run 1 9.20 k 0.39 10.23 k 0.40 29.95 * I .O 2 9.39k0.29 10.32 k0.27 29.82 k0.92 NIST SRM 465 18.34k0.54 15.57k0.41 38.57&0,93 Tatsumoto* 9.304 k 0.003 10.294 f 0.003 29.479 f 0.009 *Tatsumoto et a1.16 and the concentrations of Th and U were less than 0.1 ng g-l. As discussed previously Pb might reside not in the sulfide phase but in the metal phase.The results for Th and U contents are less than 0.04 ng g-' but in the metal phase much lower concentrations of Th and U might be expected than in a sulfide phase hence higher Pb:Th and Pb:U ratios in metal. The Pb isotopic composition of Yamato 791694 was determined. Results are presented in Table 5 together with values obtained by Tatsumoto et all6 The first run is the average of seven sets of ratios and the second run is the average of 22 ratios. It is evident that the Pb isotopic composition in Yamato 791694 is indeed very close to primordial. Further discussions cannot be made because the precision of the data is not sufficient. Chromium For the following discussions it is assumed that Yamato 7503 1 and Yamato 79 1694 are paired as described earlier.'* The concentration of 45Sc in Yamato 75031 is 2.3 ng g-' which can be considered to have been produced by a cosmic ray.The calculated exposure age of a cosmic ray is 700 million years considering that cosmic ray produced 4He* concentration is 2.63 x cm3 g-I [standard temperature and pressure (STP)] and 2'Ne* concentration is 10.2 x cm3 g-l (STP).I7J8 If Yamato 75031 is a medium-sized meteorite estimated from 4He:21Ne and also from the data for 'OBe 36C1 and 53Mn concentrations in Yamato 75031 and Yamato 79107619 the expected cosmic ray produced 53Cr* would be 25 ng g-'. A Cr concentration of 0.42 pg g-I as in Table 4 is calculated from 52Cr assuming terrestrial isotopic abundance of Cr. If this value is correct which is 35 times lower than the value given by Wasson et al.," the isotopic ratio of 53Cr:52Cr can be expected to be (53Cr:52Cr)observed=(53Cr + 53Cr*)/(52Cr + s2Cr*) (3) where Cr represents native Cr and Cr* represents cosmic ray produced Cr.It is assumed the isotopic composition of native Cr is the same as terrestrial and the following values are used:20 50Cr (4.35%); 52Crn (83.79%); 53Crn (9.50%); and 54Crn (2.36%). Production rates of Cr isotopes by cosmic ray irradiation [p(Cr*)] in an iron meteorite are assumed to be Eqn. (5) is obtained from spallation reaction systematics.z1 If the differences in the atomic mass of Cr isotopes is neglected and using S3Cr*=25 ng g-' then 52Cr*= 25/1.4= 17.9 ng g-l 52Cr,=0.42 ~0.8379-~~Cr* =0.334 pg g-' 53Cr,=0.334 x 0.095/0.8379=0.0379 ng g-I and finally (53Cr:52Cr)observed=0. 188 is obtained as the expected value.The results are summarized in Table 6. Considering the uncertainty of the RSF of Cr and the precision of the data the observed isotopic ratio is what is expected. Hence it can be concluded that the concentration of Cr was correct and also that the estimation of exposure age and/or460 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY APRIL 1993 VOL. 8 Table 6 Cr isotopic ratio in Yamato 75031 Run 1 2 3 4 Average Reference? Excess 52Cr counts 53Cr counts 53Cr:52Cr 9403 1722 0.183 11061 2098 0.190 13713 21 12 0.154 12409 21 16 0.170 0.174 rt 0.014* 0.1 1338k0.00014 54% *Standard deviation given as lo. ?Shields et al.*O production ratio of 53Cr* and 52Cr* were reasonable. The same estimation can be made for 51V produced by cosmic rays using the same systematics.The expected concentra- tion is 10-12 ng g-l. From Table 4 a value of 8.7 ng g-' is found for V. This value is somewhat lower than expected however considering again the uncertainty of the RSF and expected value the agreement is reasonable. Thus it can be concluded that V in Yamato 75031 is almost totally produced by cosmic ray irradiations. It is often observed that if the Cr content is low a discrepancy between the results by NAA and GDMS is found (NAA data are higher than GDMS) for example Yamato 791694 NAA (8 pg g-l Wasson et a/.") versus GDMS (0.44 pg g-l) Guin NAA (13 pg g-I Wasson et al. I f ) versus GDMS ( 1.8 pg g-*). Chromium isotope analysis suggests that NAA data for low level Cr might have suffered from interferences.Willis and Wassonl5 suggested that interference from 316 keV 1921r on 320 keV 51Cr would be significant if the Cr concentration is less than 100 pg g-l and Ir concentration is more than 0.5 pg g-l0 Wasson and WangZ2 discussed an interfering reaction of S4Fe(n a ) T r which is equivalent to 3 pg g-I Cr in an iron sample. Thus if the concentration of Cr is less than 1 pg g-l the correction must be enormous. Conclusion Glow discharge MS was applied to the analysis of natural metallic iron meteorite samples. Up to 53 elements were determined without lengthy chemical treatments. Phospho- rus S Zn Te Pb and Bi which are difficult to determine by NAA can be determined with higher sensitivities. Boron C N 0 Ag and Hg can also be determined but need longer pre-sputtering time.Typical detection limits were of the order of 1 x 10-lo-1 x lo-" g g-l with an integration time of 5- 10 min for one isotope. Low level Cr determination in iron by NAA seems to suffer from a serious interference reaction. Isotopic analysis can be utilized for the samples which have large isotopic variations. Anomalous elemental compositions were found in Yamato 791694 which has extremely high Ni and some chalcophile elements. The isotopic composition of Pb in this meteorite is close to primordial. Chromium isotopic composition in Yamato 7503 1 shows about a 50% excess of 53Cr which is consistent with a cosmic ray origin expected from 45Sc 4He:21Ne IOBe "(21 and 53Mn measurements. We thank the National Institute of Polar Research for providing the meteorites Yamato 79 1694 and Yamato 75031.We also thank anonymous reviewers for helpful suggestions to improve the manuscript. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 References Harrison W. W. Hess K. R. Marcus R. K. and King F. L. Anal. Chem. 1986 58 341A. Mykytiuk A. P. Semeniuk P. and Berman S. Spectrochim. Acta Rev. 1990 13 1 . Milton D. M. P. Hutton R. C. and Ronan G. A. Fresenius' Z. Anal. Chem. 1992 in the press. Shimamura T. Nagai H. and Honda M. Lunar Planet. Sci. 1986 XVII 795. Honda M. Nagai H. and Shimamura T. Proc. Znst. Nut. Sci. Nihon Univ. 1991 26 123. Shimamura T. Takahashi T. Honda M. and Nagai H. 16th Symposium on Antarctic meteorites abstract National Insti- tute of Polar Research Japan 199 1 73. Smith S. P. and Huneke J. C. Meteoritics 1991 26 396. Kracher A. Willis J. and Wasson J. T. Geochim. Cosmo- chim. Acta 1980 44 773. Malvin D. J. Wang D. and Wasson J. T. Geochim. Cosmochim. Acta 1984 48 785. Honda M. personal communication. Wasson J. T. Ouyang X. Wang J. and Jerde E. Geochim. Cosmochim. Acta 1989 53 735. De Laeter J. R. and Jeffery P. M. Geochim. Cosmochim. Acta 1967 31 969. Smales A. A. Mapper D. and Fouche K. F. Geochim. Cosrnochim. Acta 1967 31 673. Scott E. R. D. Earth Planet. Sci. Lett. 1977 37 273. Willis J. and Wasson J. T. Radiochim. Acta 1981 29 45. Tatsumoto M. Knight R. J. and Allegre C. J. Science 1973 180 1279. Nagao K. Takaoka N. and Saito K. 8th Symposium on Antarctic Meteorites abstract National Institute of Polar Research Japan 1983 83. Honda M. Nagai H. Takaoka N. Yamamuro K. Kume T. Takahashi A. and Akizawa S. Proc. of the National Institute of Polar Research Symposium on Antarctic Meteorites 1988 1 197. Nishiizumi K. Data Compilation Nucl. Tracks Radiat. and Meas. 1987 13 209. Shields W. R. Murphy T. J. Catanzaro E. J. andGarner E. L. J. Res. Nat. Bur. Stand. 1966 70A 193. Honda M. and Arnold J. R. Handbuch der Physik 1967 4612 613. Wasson J. T. and Wang J. Geochim. Cosmochim. Acta 1986 50 725. Paper 2/01 173E Received March 4 1992 Accevted November 27. 1992

ISSN:0267-9477    

DOI:10.1039/JA9930800453

出版商:RSC    

年代:1993

数据来源: RSC