JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY SEPTEMBER 199 1 VOL. 6 Analytical Performance Evaluation of a 40.68 MHz Inductively Coupled Plasma Mass Spectrometer 43 1 Chang J. Park and Kwang W. Lee Inorganic Analytical Laboratory Korea Standards Research Institute P. 0. Box 3 Taedok Science Town Taejon Korea The instrumentation of a new 40.68 MHz inductively coupled plasma (ICP) mass spectrometer is described. An optimization study has been carried out for the instrument. Of the many operating parameters the carrier argon flow rate is the most sensitive and the optimum carrier argon flow rate is found to be much lower than the literature values for a 27.12 MHz ICP mass spectrometer. The analytical performance of the instrument i.e. the detection limits background mass spectrum and formation of oxides and doubly charged ions is also evaluated.Keywords Inductively coupled plasma mass spectrometry; 40.68 MHz plasma; optimization and analytical performance; detection limits; background spectrum Since the first commercially available inductively coupled plasma mass spectrometer was introduced at the 1983 Pittsburgh Conference inductively coupled plasma mass spectrometry (ICP-MS) has received a lot of interest in many analytical fields such as geochemical e~ploration,~-~ environ- mental pr~tection,~ biomedical research5 and quality control in nuclear fuel6 and materials7 industries. In Korea ICP-MS is being used increasingly because of the important advan- tages of simple spectra high sensitivity and isotope abun- dance information. During 1989- 1990 about ten ICP-MS instruments were installed in Korean national research institutes and quality control laboratories.It is however reported in Korea that the down time of the instruments is much higher than that of other well-established analytical instruments such as ICP emission spectrometers and atomic absorption spectrometers probably because ICP-MS is a relatively young technique and most operators do not fully understand the basic principles.8 The Korea Standards Research Institute as a national centre for measurement of standards and also as an educational centre for analytical instrumentation developed an ICP-MS instrument to be used for the certification of reference materials and fundamental studies of ICP-MS. This paper presents the instrumentation and evaluation of the analytical performance of a laboratory-built ICP-MS system. The evaluated analytical performance includes the back- ground mass spectrum doubly charged and oxide ion formation detection limits and isotope ratio measurements of the National Institute of Standards and Technology (NIST) Standard Reference Material (SRM) isotopic standards.Experimental Reagents All laboratory ware was washed with de-ionized water from a Milli-Q system (Millipore Bedford MA USA). Standard solutions were prepared using sub-boiling quartz-distilled water. All chemicals employed were of analytical-reagent grade. Instrumentation The ICP source consists of an r.f. generator impedance matching unit gas panel load coil torch and sample introduction unit as shown in Fig.1. The plasma is generated by a 40.68 MHz r.f. generator (Model ICP-l6L RF Plasma Products Marlton NJ USA) which uses a field effect transistor (FET) instead of the conventional power tube. It is therefore very compact (482.6 mm rack mount) and requires no warm-up time. For sample introduction a Meinhard concentric nebulizer (TR-30-K3) and Scott-type spray chamber are used and a peristaltic pump (Minipuls 3 Gilson Medical Electronics Middleton WI USA) is em- ployed to control the sample uptake rate. The carrier argon gas flow is precisely controlled by a mass flow controller (Vacuum General San Diego CA USA). The torch box housing a standard short torch (Superior ICP Denver CO USA) nebulizer spray chamber and impedance matching unit sit on a laboratory-built xyz translator. Typical ICP- MS operating conditions are given in Table 1.A schematic diagram of the interface and quadrupole mass spectrometer is shown in Fig. 2. The interface is pumped by a 7.5 I s-' rotary pump (SD450 Varian Lexington MA USA). The sampler and skimmer orifices are 1 and 0.7 mm in diameter respectively and the distance between the two orifices is 8 mm. Both aluminium and nickel orifices were successfully employed in this instrument but throughout this work the aluminium orifices were used. The orifices are mounted on water-cooled flanges without the use of bolts they are simply screwed into the flanges. With this mount- / * Copper . braided I wire Tesla P coil cooling water pressure Gas 23-96 PF R.f. Spray chamber II Nebulizer Mass flow controller 11 40.68 MHz r.f. generator cooling water - R.f.inter-lock Gas panel Fig. 1 ICP generation system432 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY SEPTEMBER 199 1 VOL. 6 Einzel lens Fig. 2 Schematic diagram of interface and quadrupole mass spectrometer .Il....m......._r_..rrl..LI.I...m_..._ Fig. 3 Calculated ion trajectories in the electrostatic lens system Table 1 Optimized ICP-MS operating conditions Forward power Reflected power Coolant flow rate Auxiliary flow rate Carrier gas flow rate Nebulizer pressure Sample uptake rate Sampling depth Interface pressure Quadrupole chamber pressure 1.2 kW <3 w 12 1 min-' 0.4 1 min-' 0.37 1 min-I 2.1 x los Pa 0.8 ml min-I 10 mm 2.1 x lo2 Pa 1.1 x Pa ing a vacuum leak which was often observed for the bolt- fastened orifices did not occur for the 6 months the spectrometer was in operation.The vacuum chamber which houses an electrostatic lens assembly and a quadru- pole mass filter is isolated from the interface by a laboratory-built gate valve (36 mm thick) when the plasma is not sampled. Between the skimmer cone and gate valve is a cylindrical lens biased to about - 200 V to attract positive ions towards the electrostatic lens assembly. The vacuum chamber is separated into two parts by the exit plate (hole of 4 mm diameter) of the Bessel box. The first chamber houses the front part of the electrostatic lens assembly and is pumped by a 1200 1 s-I diffusion pump (VHS-4 Varian). The second chamber contains the rear part of the lens assembly and a quadrupole mass filter and is pumped by a 285 1 s-' diffusion pump (HS-2 Varian).The chamber and vacuum line pressures are monitored by two vacuum controllers (Varian 843 and Balzers TPG 300) and vacuum valves are automatically controlled by a labora- tory-built valve controller. The ion optics in this instrument were designed by utilizing a computer program (SIMION Idaho National Engineering Laboratory Idaho Falls ID USA). Fig. 3 shows the calculated ion trajectories assuming a 10 eV ion energy. The quadrupole mass filter (SX300 VG Quadru- poles Winsford Cheshire UK) is composed of four (12 mm i.d. 230 mm long) molybdenum rods and has a mass range of 1-300 u. The quadrupole mass control is accom- plished by the 10 V analogue output of a 12 bit digital to analogue converter (DAC) card (WB-AVO-B2 Omega Stamford CT USA).Pulse counting mode is used for signal detection. Negative pulses are generated when ions hit an electron multiplier (CEM4870 Galileo Electro-Optics Sturbridge MA USA). They are then converted into transistor-transistor logic (TTL) pulses through a charge sensitive preamplifier-discriminator (A 1 1 1 AMPTEK Bedford MA USA). The discriminator threshold was set to approximately 1 x lo6 electrons in order to remove theJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY SEPTEMBER 199 1 VOL. 6 433 quadrupole r.f. noise at the highest mass detected (238U160+). The TTL pulses are counted displayed and manipulated by a multichannel scaler card (ACE-MCS EG&G ORTEC Oak Ridge TN USA) plugged into an IBM PC-AT computer. In mass scanning each DAC output change advances one channel in the multichannel scaler and thus for the 12 bit DAC 13.65 channels on average are allocated to a particular mass.Results and Discussion Background Mass Spectra The background mass spectrum observed when distilled de-ionized water is nebulized into the plasma is shown in Fig. 4(a) and was obtained by accumulating 50 scans from m/z= 10-85. Five hundred and twenty channels were assigned to cover the 76 masses with a dwell time per channel of 2 ms and hence a complete spectrum was obtained in 52 s. The integrated count rate over 100 ms in each channel was converted into a count rate per second. The detector bias potential was substantially reduced so that the maximum peak count rate [Fig. 4(a)] could not exceed 1 x lo6 counts.The background spectrum in Fig. 4(a) shows typical major peaks at mlz values of 16 (O+) 17 (OH+) 18 (OH2+) 32 (02+) 40 (Ar+) and 4 1 (ArH+). Among the major peaks the doubly charged argon ion (Ar2+) is unusual indicating that this instrument has a serious discharge at the sampler orifice. The secondary discharge was also confirmed by visual inspection of the sampler orifice. The discharge at the sampler orifice did not enlarge the orifice diameter but produced widely eroded areas around the orifice. The aluminium sampler orifice was less affected by erosion than the nickel orifice with no noticeable change in sensitivity. Minor background peaks at full sensitivity from 20 to 30 and from mlz 43 to 85 were separately obtained with an optimum bias potential appiied to the detector and are shown in Fig.4(b). It can be seen that the intensity of the A1 background peak at m/z = 27 from the orifice material is about 1600 counts s-l which is equivalent to a standard signal for approximately 2 ppb. Fig. 4(b) also shows that the distilled de-ionized water used in this work is contami- nated with Ca Sc and Zn. In general however background ion-free regions of the spectrum are fairly clean with a noise level of about 10 counts s-l. Optimization Horlick et aL9 published the effect of the operating parameters for the Sciex Elan 250 ICP-MS instrument. Long and Brownlo subsequently reported their work on the optimization of the operating parameters for a VG Plasma- Quad ICP-MS instrument. Both groups concluded that the most sensitive parameters were the aerosol carrier gas flow rate and r.f.forward power. In this work optimization of three operating parameters for the 40.68 MHz ICP-MS instrument was studied. The three parameters were carrier gas flow rate r.f. forward power and sampling depth. In Figs. 5-10 the data points plotted for each mass are the means of 60 channel values which are obtained by the accumulation of 500 sweeps in the single ion monitoring mode with a dwell time per channel of 2 ms. Typical relative standard deviation (RSD) values for the ions of interest are given in Table 2. Fig. 5 shows the effect of the carrier gas flow rate for six different ions at a 1.2 kW r.f. forward power and 10 mm sampling depth. The response curves show that the opti- mum carrier gas flow rate is between 0.35 and 0.4 1 rnin-l and varies little for different ions.The optimum carrier gas flow rate of this instrument is found to be much lower than those reported for ICP-MS instruments with 27.12 kW r.f. 1 .o 0.9 0.8 v) 0.7 3 0.6 2 0.5 g 0.4 = 0.3 0.2 0.1 0 v) 4- 8 IL) \ U > .- al + 10 30 50 70 90 40 30 v) 3 0 4- 2 20 1 + > v) al C .- 4- - 10 20 30 40 60 mlz 80 Fig. 4 background peaks from mlz 20 to 30 and 43 to 85 (a) Background mass spectrum from mlz 1 to 85; (b) minor ~ ~~ Table 2 Typical RSD values of 60 data points with an integration time of 1 s per data point when nebulizing 100 ppb solutions Ion Mass number RSD (%) ArO+ cu+ Ar + &+ Ba+ La+ Ce+ Pb+ U+ 56 63 80 107 138 139 140 208 238 3.3 0.9 0.7 0.7 0.9 0.8 0.9 0.8 0.7 generators (0.69,' 0.6-0.8,ll and 1.0612 1 min-l) which implies that the plasma backpressure of the 40.68 MHz ICP is lower than that of the 27.12 MHz ICP owing to the reduced skin depth. Because of the low optimun? carrier gas flow rate the TR-30-C1 Meinhard nebulizer initially employed was replaced by a TR-30-K3 nebulizer which was specially designed for low (0.7 1 min-') flow.The two types of Meinhard nebulizer gave almost the same optimum carrier gas flow rates and sensitivities. The effect of r.f. forward power is presented in Fig. 6 for five analytes with the carrier gas flow rate and sampling depth held constant at 0.37 1 min-l and 10 mm respec-434 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY SEPTEMBER 1991 VOL. 6 626 0.3 0.34 0.38 0.42 0.46 Carrier gas flow rate/t min-' Fig.5 Effect of carrier gas flow rate on 100 ppb analyte signals 1 I I 1 I J 0.9 1.0 1.1 1.2 1.3 1.4 1.5 R.f. forward power/kW Fig. 6 Effect of r.f. power on 100 ppb analyte signals tively. The optimum r.f. forward power of 1.2-1.35 kW is slightly different for different analytes. Comparison of Figs. 5 and 6 indicates that the effect of r.f. forward power is not as great as that of the carrier gas flow rate. Since the lowest sampling depth (separation distance between sampler orifice and the grounded end of the load coil) is 10 mm for this instrument the effect of varying the sampling depth from 10 to 19 mm with 1.5 mm intervals was studied. The results are presented in Fig. 7 for three analytes argon oxide and an argon dimer. For Figs. 7-10 the values of r.f.power carrier gas flow rate and sampling depth listed in Table 1 were used unless otherwise stated. Fig. 7 shows that the analyte and argon oxide signals decrease continuously as the sampling depth increases while the argon dimer signal increases. This could indicate that the zone at which the argon dimer ions form is further along in the ICP than that at which analyte and argon oxide ions form. The effect of sample uptake rate was also studied. However little variation was observed in the analyte signals as the sample solution uptake rate was varied from 0.8 to 1.5 ml min-'. Formation of Oxides and Doubly Charged Ions Vickers et al.13 compared the analytical characteristics of a 27.12 MHz with those of a 40.68 MHz ICP-MS instrument. They found that interference effects were independent of frequency but that the formation of oxides and doubly charged ions was affected by a more significant orifice- linked discharge with the 40.68 MHz ICP.Since the 160 A 7 120 La Ba 8 10 12 14 16 18 20 Sampling depth/mm Fig. 7 Effect of sampling depth on 100 ppb of analyte ArO+ and AT2+ signals instrument employed in this work used a 40.68 MHz generator with its load coil grounded at the end near the sampler orifice the orifice-linked discharge gave a large number of doubly charged ions (e.g. Ba2+:Ba+=2.2). Wilson et a1.I' reported that the orifice-linked discharge was drastically reduced by running a short strap between the end of the load coil and the sampler orifice flange. Thus the same connection was adopted in this instrument (Fig.l) which reduced the Ba2+:Ba+ ratio to 0.3. Abdallah et al.14 compared the temperatures of the 5 and 40 MHz ICPs and found that excitation and ionization temperatures of the 40 MHz ICP were lower than those of the 5 MHz ICP. Barnes and Schlei~her~~ applied a com- puter model of the ICP discharge to the calculation of the possible effects of frequency on the discharge properties. The computations predicted that in the higher frequency discharge the maximum temperature was lower than in the low frequency discharge. Thus the ICP mass spectrometer equipped with the 40.68 MHz ICP source is expected to give higher oxide signals compared with the conventional 27.12 MHz instrument. Since oxides and doubly charged ions can present a spectral interference problem the ratios of oxides and doubly charged ions to singly ionized species (MO+:M+ and M2+:M+) are often reported in the literature as an impor- tant rating of analytical performance.Thus in this work the effect of carrier gas flow rate and sampling depth on MO+:M+ and M2+:M+ ratios was investigated for five elements whose oxide bond strengths and second ioniza- tions potentials are given in Table 3. Fig. 8 shows the MO+:M+ ratios increasing slowly as the carrier gas flow rate is increased. The primary reason for this trend is that for all five elements the MO+ counts showed a steady increase as the carrier gas flow rate increased while the M+ counts showed maxima at opti- mum flow rates. The above trend is not consistent with that reported by Long and Brownlo except for Ba.They observed minimum MO+:M+ ratios at near optimum flow rates for Pb Cs and Sm. For the M2+:M+ ratios M2+ counts in addition to M+ counts were observed to increase owing to the formation of a central channel in the plasma as the carrier gas flow rate is increased to an optimum however the M2+ counts tend to increase more rapidly up to the maximum. When the flow rate was further increased both counts dropped owing to the decreased residence time in the plasma. Thus M2+:M+ ratios plotted in Fig. 9 show an increasing rise with increasing carrier gas flow rate. Figs. 8 and 9 generally show that elements of higher oxide bond strength give higher MO+:M+ ratios and that those ofJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY SEPTEMBER 199 1 VOL. 6 43 5 Table 3 Oxide bond strength and second ionization potential of the elements of interest Oxide bond strength/ Second ionization potential/ Element kJ mol-l eV Ba 563 La - Ce 795 Pb 378 U 76 1 10.00 11.06 10.85 15.03 - 0.28 0.32 0.36 0.4 Carrier gas flow rate/! min-' Fig.8 MO+:M+ ratios versus carrier gas flow rate 0 1 -5 0.28 0.32 0.36 0.4 Carrier gas flow rate/I min-' Fig. 9 M2+:M+ ratios versus carrier gas flow rate lower second ionization potential give higher M2+:M+ ratios as expected. Comparison of the ratios at an optimum carrier gas flow rate with the literature suggests that this instrument should generate slightly higher oxide and doubly charged ion levels. The higher M2+:M+ ratios are believed to be due to the enhanced secondary discharge at the orifice from the 40.68 MHz ICP and the higher MO+:M+ ratios due to the lower temperature of the higher frequency ICP.Lim et all8 measured the potential of a supersonic jet expanding into the interface at various plasma sampling positions. They found that the jet poten- tial increased suddenly at a sampling depth of about 32 mm. Further increase in the sampling depth gradually reduced the jet potential. Their work suggests that the secondary discharge in the spectrometer should be some- how affected by the sampling depth. Thus the effect of sampling depth on the M2+:M+ and MO+:M+ ratios was studied with this instrument. The results are plotted in Fig. I I x - x I -1 .o -5.0 + PbO' P b' 8 10 12 14 16 18 20 Sampling de pt h/m m Fig. 10 MO+:M+ and M2+:M+ ratios versus sampling depth 10 and show an obvious trend i.e.MO+:M+ ratios fall as the sampling depth is increased while M2+:M+ ratios rise. The M2+ counts show a maximum at a sampling depth of about 12 mm which is 2 mm longer than the lowest sampling depth at which M+ counts show a maximum. However M+ counts of the three elements in Fig. 10 were observed to drop more rapidly than M2+ counts as the sampling depth was increased further. It is obvious that M+ M2+ and MO+ counts decrease owing to dilution as the sampling depth is increased. However there seems to be a mechanism which affects the order of the rate of decrease (ie. MO+ counts drop most rapidly and M2+ counts least rapidly). A strong discharge between the sampler orifice and the plasma which was not visible at the normal sampling depth was observed visually with this instrument when the sampling depth was very high (about 30 mm).Thus the secondary discharge at the orifice can get stronger as the sampling depth is increased from 10 to 19 mm which could account for the rising M2+:M+ ratios and falling MO+:M+ ratios in Fig. 10. Detection Limits The operating conditions listed in Table 1 were obtained through an optimization study and used in determining detection limits and isotope ratios. In order to estimate detection limits the background noise was obtained by aspirating distilled de-ionized water at each mass and the signal counts from nebulizing 100 ppb standard solutions. Both noise and signal counts were acquired by accumulat- ing 100 sweeps of 60 channels in the single-ion mode with a dwell time of 10 ms per channel.Detection limits were determined by calculating the concentration of the element which would give a signal three times the standard deviation (SD) of the background noise. Detection limits of six elements together with the SD of the background noise and 100 ppb standard signals are presented in Table 4 and show that this instrument can offer almost the same levels of detection limits as the commercially available instruments. Isotope Ratio Measurements One of the advantages of ICP-MS is the rapid and convenient determination of isotope ratios with moderate precision (0.1- 1 Yo) and high sensitivity. The rapid determi- nation of isotope ratios enables analysts to use the isotope dilution method which is one of the few absolute methods offering high accuracy.l 9 The isotope ratio measurement capability of this instru- ment was evaluated by analysing four NIST isotopic436 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY SEPTEMBER 1991 VOL. 6 Table 4 Detection limits for 100 ng ml-1 solutions Element Isotope c u 63 Ag 107 Ba 138 La 139 Pb 208 U 238 SD of background noisel counts s-l 8.2 7.8 8.5 9.9 19.1 7.4 Signall counts s-I 59 000 44 000 5 5 000 65 000 130 000 126 000 Detection limit/ ng ml-' 0.04 0.05 0.05 0.05 0.05 0.02 ~~ Table 5 Measured ratios of isotopic NIST SRMs Isotope ratio Run No. 1 2 3 4 5 Mean rt SD RSD (O/O) NIST certified value Error (O/O) NIST SRM 951 ( l0B IlB) 0.253 0.258 0.253 0.257 0.246 0.253 4 0.005 1.9 0.247 + 2.4 NIST SRM 976 2.194 2.167 2.155 2.189 2.139 2.169 k 0.020 1 .o (63Cu:Tu) 2.244 -3.3 NIST SRM 978a ( lo7.4g:Io9Ag) 1.04 1 1.034 1.045 1.047 1.023 0.9 1.038 k 0.009 1.076 - 3.5 Table 6 Measured isotope abundances of NIST SRM 981 Lead Metal Natural Isotope abundance (O/O) Run No.1 2 3 4 5 Mean k SD RSD (Yo) NIST certified value Error (Yo) 204Pb 1.32 1.38 1.42 1.53 1.45 1.42 k 0.08 5.5 1.4245 -0.3 206Pb 24.24 24.05 24.02 23.95 24.00 24.05 ? 0.1 1 0.5 24.1447 -0.4 207Pb 22.34 22.68 22.39 22.54 22.15 0.9 22.42 +- 0.2 22.0827 + 1.5 2osPb 52.10 5 1.89 52.17 5 1.98 52.40 52.1 1 +- 0.20 0.4 52.348 1 -0.5 standards. In Table 5 the measured isotope ratios of boron (NIST SRM Boric Acid) copper (NIST SRM 976 Copper Metal) and silver (NIST SRM 978a Silver Nitrate) are presented. The error between the determined ratios and the NIST certified data is less than 3.5% and RSD values for five measurements are also less than 1 O/o except for boron.In Table 6 the measured isotope abundances of a lead isotopic standard (NIST SRM 981 Lead Metal Natural) are shown. Table 6 shows the excellent agreement with the certified data and the good reproducibility except for the minor isotope 204Pb. The concentrations of the NIST isotopic standards were 200 ppb for Cu Ag and Pb and 500 ppb for B. For the data acquisition only minimum mass ranges were scanned 200 times with a dwell time of 10 ms per channel. Isotope ratios were then obtained by integrating the peak areas of the isotopes. Conclusion The new ICP-MS system built in this laboratory has almost the same detection power but gives slightly higher levels of doubly charged and oxide ions compared with the commer- cial instruments.Since a 40.68 MHz ICP source is employed in the system the optimum carrier gas flow rate is substantially lower than in the systems which utilize a 27.12 MHz source. The system shows great promise as a tool for quantitative work using isotope dilution and also for fundamental studies of ICP-MS. References 1 Date A. R. and Hutchinson D. Spectrochim. Acta Part B 1986 41 175. 2 Hall G. E. M. Park C. J. and Pelchat J. C. J. Anal. At. Spectrom. 1987 2 189. 3 Lichte F. E. Meier A. L. and Crock J. G. Anal. Chem. 1987 59 1150. 4 Boomer D. W. and Powell M. J.,Anal. Chem. 1987,59,2810. 5 Delves H. T. and Campbell M. J. J. Anal. At. Spectrom. 1988 3 343. 6 Beck G. L. and Farmer 0. T. J. Anal. At. Spectrom. 1988,3 771.JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY SEPTEMBER 199 1 VOL. 6 437 7 Stotesbury S. J. Pickering J . M. and Grifferty M. A. J. Anal. At. Spectrom. 1989 4 457. 8 Park C. J. and Lee K. W. paper presented at the 5th Annual Meeting of the Korean Society of Analytical Science Seoul 1990 paper No. PL3. 9 Horlick G. Tan S. H. Vaughan M. A. and Rose C. A. Spectrochim. Acta Part B 1985 40 1555. 10 Long S. E. and Brown R. M. Analyst 1986 111 901. 11 Wilson D. A. Vickers G. H. Hieftje G. M. and Zander A. T. Spectrochim. Acta Part B 1987 42 29. 12 Tan S. H. and Horlick G. Appl. Spectrosc. 1986 40 445. 13 Vickers G. H. Wilson D. A. and Hieftje G. M. J. ilnal. At. Spectrom. 1989 4 749. 14 Abdallah M. H. Diemiaszonek R. Jarosz J. Mermet. J.-M. Robin J. and Trassy C. Anal. Chim. Acta 1976 84 271. 15 Barnes R. M. and Schleicher R. G. Spectrochim. Acta Part B 1975 30 109. 16 Gray A. L. and Williams J. G. J. Anal. At. Spectrom. 1987 2 81. 17 Vaughan M. A. and Horlick G. Appl. Spectrosc. 1986 40 434. 18 Lim H. B. Houk R. S. and Crain J. S. Spectrochim. Acta Part B 1989 44 989. 19 Heumann K. G. TrAC. TrendsAnal. Chem. (Pers. Ed.) 1982 1 357. Paper I /006 I30 Received February 11 th I991 Accepted May I6th I991