JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY OCTOBER 1991 VOL. 6 545 Preliminary Investigations of a Helium Alternating Current Plasma for the Determination of Metals by Atomic Emission Spectrometry Luis A. Colon* and Eugene F. Barry Department of Chemistry University of Lowell One University A venue Lowell MA 01 854 USA The construction and operation of an alternating current plasma are described. The plasma is generated across two copper electrodes utilizing helium as the plasma and nebulizer gas. The plasma operates at a frequency of 60 Hz. Aqueous solutions containing elements of interest are introduced into the plasma by two different devices a glass frit nebulizer and a thermospray interface. Improved detection limits are associated with the latter. Analytical characteristics for 14 elements are reported including detection limits (30) for the metals investigated at the ppb level.Linear ranges of 2-4 orders of magnitude are observed. Signal precision at the level of 10 times the detection limit ranges from 1.9 to 10% relative standard deviation. Keywords Alternating current plasma; glass frit nebulizer; thermospray nebulizer; atomic emission spectro- metry; aqueous metal determination Atomic emission spectrometry (AES) using plasmas as excitation sources has become one of the most commonly used analytical techniques for trace element determination. The widely used emission sources are inductively coupled plasma (ICP); direct current plasma; and microwave- induced plasma (MIP). All systems offer low detection limits reliable precision and a large linear dynamic range for many elements the ICP being the most widely investi- gated and used.'** Unfortunately the original investment and the cost of operation can be considered as limiting factors.Possible approaches to minimizing these factors have been ~ t u d i e d . ~ ? ~ The MIP is the least expensive and least complicated of the three emission source^,^-^ particularly when low power (approximately 200 W) is used with minimal gas consump- tion. It is common practice to introduce the sample into the MIP in the vapour phase making this type of plasma source a suitable detector with gas chromatography (GC).8>9 With the development of the Beenakker cavity,lOJ1 liquid aero- sols can be introduced at atmospheric pressure although this process requires large power levels and a high gas ons sump ti on;^^-^^ these requirements may lead to complex- ity and an increase in operating cost.Few reports have appeared on the use of low-powered MIPS for the direct introduction of an aqueous solution.1sJ6 The detection limits found are considerably higher (ppm level) in compar- ison with other plasma systems. Lower detection limits have been reported with the use of a low power high efficiency MIP." A helium alternating current plasma (ACP) developed in this laboratory is described in this paper as an inexpensive alternative for analysis by AES. The ACP is based on the same principle as the micro-arc originally developed as a sample introduction device for the MIP.'* The micro-arc has also been employed as an emission source in atomic spectroscopy for samples in the gaseous phase.19920 Never- theless analysis by AES of direct liquid samples introduced via nebulization into the micro-arc has not been reported. The ACP is generated by a high voltage low a.c. step- up transformer. The plasma is not extinguished by the introduction of liquid sample aerosols and utilizes helium as the plasma support gas simultaneously serving as the nebulizer gas. The helium species in the plasma produce sufficient energy to excite other atomic species producing characteristic elemental emission. The ACP is probably the simplest and most inexpensive system to construct and operate that can be used as an emission source. The ACP has been successfully employed as a specific element detector for GC2'J2 and high-performance liquid chromatography paper.23 Thus this paper describes the construction operation and analytical performance of the ACP for the detection of 14 elements.Liquid aerosols generated via a glass frit nebulizer (GFN) are directly Table 1 Instrumental components Component Power supply (ax.) Monochromator (slit-width 30 pm and slit-height 5 mm) PMT power supply PMT (1000 V) Picoammeter Nebulizers Lens Discharge tube Data acquisition Computer Support gas Low pass filter Model/T ype Webster ignition EU-700 0.35 m transformer 12-8AB7 7640 R446 4 14-s Glass frit Thermospray Fused silica biconvex 1 mm id. quartz LabCalc AT compatible Helium ultra-high purity Time constant 0.2 s 25.4 mm diameter 101 mm focal length Manufacturer STA-Rite Frankfort KY USA GCA McPherson Acton MA USA McPherson Acton MA USA Hamamatsu Middlesex NJ USA Keithley Instruments Cleveland OH USA Laboratory constructed Laboratory constructed Oriel Stratford CT USA Laboratory constructed Galactic Industries Salem NH USA Zenith Data Systems St.Joseph MI USA Northeast Airgas Manchester NH USA Laboratory constructed * Present address Department of Chemistry Stanford University Stanford CA 94305-5080 USA.546 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY OCTOBER 199 1 VOL. 6 introduced into the plasma. The results are compared with those given by a thermospray sample introduction system. Linear responses detection limits and precision are discussed. Experimental Instrumentation The instrumental components utilized in the construction of the ACP emission system are listed in Table 1.A diagram of the ACP is shown in Fig. 1. The a.c. discharge was generated across two copper or tungsten electrodes (3 mm 0.d.) and maintained in a controlled helium atmosphere by means of the quartz discharge assembly illustrated in Fig. 2. The airtight assembly was constructed in the laboratory using two pieces of quartz tubing (4 mm i d . x 6 mm o.d. 5 D 0 r' Fig. 1 Schematic representation of the ACP A a.c. power supply; B PMT power supply; C quartz discharge assembly (see Fig. 2); D focusing lens; E monochromator; F PMT in housing; G picoam- meter; H sample introduction system; I flow meter controller; J helium supply; and K data acquisition system * 0 10 E Fig. 2 Schematic representation of the quartz discharge assembly with thc clcctrodes in place A copper electrodes B 2 cm long ( 1 mm i.d.x 6 mm 0.d.) discharge tube C transfer and electrode container tubing ( 5 cm long 4 mm i.d. x 6 mm 0.d.); D electrode holder; and E sample from nebulizer cm long) attached to a piece of quartz capillary tubing (1 mm i.d. x 6 mm o.d. 2 cm long). The discharge was constrained into the capillary tube of the discharge assembly. One of the electrodes was kept inside the horizontal arm of the assembly (see Fig. 2) while the second electrode was positioned outside of the assembly at the exit end of the capillary tube. The sample was introduced into the plasma through the arm that is perpendicular to the capillary tubing of the discharge assembly. The furnace ignition transformer used as the ax.power supply was operated at its maximum capability (14000 V 20 mA) with the aid of a Powerstat variable autotransformer which was fed with a line supply of 120 V at a frequency of 60 Hz. The ignition transformer was water-cooled to avoid over- heating. Sample Introduction The GFN was constructed from a modified Pyrex sintered glass filter-funnel (15 ml) and has been described else- where.24 An Ismatec peristaltic pump (Model 76 14-30 Cole-Parmer Chicago IL USA) delivered the sample of interest to the GFN at a rate of 0.5 ml min-l. The helium supply was maintained at a pressure of 60 psi. The thermospray interface was utilized in the flow injection mode. The thermospray probe was constructed of stainless-steel capillary tubing (k in 0.d. x 0.004 in id.30 cm long) (Alltech Associates Deerfield IL USA). The probe was introduced into a flexible metal tube (4 in id.) wrapped with heating tape and glass wool then with aluminium foil. Chromel-ahmel thermocouples attached to the flexible metal tube were employed in order to Capillary probe (in Swagelok fitting) Coolant out t Helium Liebig I Coolant in Waste condenser 35 cm r Fig. 3 Spray chamber-desolvation system .............. .... I t Helium ........ ............. \ S P W chamber Fig. 4 Tip of the capillary probe in place with a Swagelok fittingJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY OCTOBER 1991 VOL. 6 547 Table 2 Detection limits (cL) and linear response for Cd and Zn using two plasma viewing positions Element Viewing position cL (ppb) Linearity (order of magnitude) Cd End-on 24 Side-on 46 Zn End-on 5 Side-on 29 2 3 2 3.5 monitor the temperature of the interface.A heated spray chamber (1 5 cm long 5 cm 0.d. x 4.6 cm id.) connected to a water-cooled Liebig condenser provided partial desolvation and solvent removal respectively. The system resembles one which has been used in combination with the ICP.25 The spray chamber-desolvating system is depicted in Fig. 3 and the tip of the capillary held in place with a Swagelok fitting is illustrated in Fig. 4. Condensation in the transfer tube (1 5 cm long 4 mm i.d. x 6 mm o.d.) from the condenser to the plasma was prevented by maintaining the temperature above 70 "C with heating tape. A Spectra-Physics solvent delivery system (Model SP8700 San Jose CA USA) delivered the sample carrier stream (water) at a rate of 1 ml min-l to a Rheodyne-type injector (Cotati CA USA) equipped with a sample loop of 200 pl which was connected to the capillary probe.A silica-based column (150~4.6 mm i.d. 5 pm C8 Alltech Associates) was placed immediately before the sample injector in order to establish the required minimum pump pressure of 600 psi for proper flow regulation. Reagents Reference solutions of all metals were prepared from 1000 ppm stock solutions (Fisher Scientific Fair Lawn NJ USA) using distilled water and dilute hydrochloric or nitric acid (1%) (Ultrex J. T. Baker Phillipsburg NJ USA) matching the matrix of the original stock solution as required. Procedure The nebulized solutions were transported to the plasma by using helium as the carrier gas which at the same time served as the plasma supporting gas. Optimization was performed by means of univariate searches using a signal- to-background noise ratio as the optimization criterion.The thermospray probe and spray chamber were optimized for Cr and Zn metals and the resulting optima were maintained throughout. For the comparative line intensity experiments solutions of 100-200 ppm of the elements studied were nebulized into the plasma. The plasma was focused at the entrance slit of the monochromator with a magnification factor of 1.5 throughout. The net analyte emission resulted from the difference between the apparent analyte emission and the blank emission. Data collection was achieved by means of the data acquisition system at a sampling rate of 3 Hz (Le.three data points were sampled every second). Random noise was reduced by using the Savitzky-Golay algorithm (a moving-window smoothing function),26 yielding improved signal-to-noise (S/N) ratios. The reported results were obtained by measuring the peak height of the response. Detection limits were calculated according to the method recommended by IUPAC (30 n = 201.27 Results and Discussion The response of the ACP is dependent on the ax. voltage output with the maximum signal response at the upper limit of the power supply. Thus the power supply was operated at its maximum capability producing an output 7 10 15 20 25 Distance/mm Fig. 5 Effect of the electrode distance on the ACP response A Zn response side-on viewing using GFN; and B Cr response end-on viewing using a thermospray interface voltage of approximately 14000 V a.c.(20 mA). The distance between the electrodes also has a pronounced effect on the plasma response as shown in Fig. 5 for Cr and Zn. Although the measurements were obtained at two plasma viewing positions (Table 2) with two sample introduction devices the optimum signal and S/N ratio is associated with a 20 mm gap in each instance. The design of the discharge tube shown in Fig. 2 allows two possible views of the plasma; side-on (transversal) and end-on (axial). In the transversal confi- guration the plasma is positioned vertically and viewed through the walls of the discharge tube. In the end-on position the centre of the fireball of the horizontally positioned plasma is observed.The differences in plasma response between the axial and transversal viewing in terms of detection limits (c,) and linearity for Zn and Cd are compiled in Table 2. The lower c in the axial position can be attributed to the fact that the measure- ment of emitted radiation is performed on the total analyte species emitting along the plasma axis at a parti- cular time interval. Alternatively when the plasma is observed in the transversal position the measurement is obtained for the analyte at a particular point of the plasma channel. Thus the observed signal in the axial mode may be expected to be higher than the signal observed in the transversal position owing to a summa- tion behaviour of the signal along the plasma axis. How- ever the longer emitting path offered by the end-on approach permits increased analyte residence times where self-absorption is more likely to occur as the concentration of the emitting species increases thereby causing non-linearity in profiles of response versus con- centration.Evidently the lower cL for the end-on approach com- promises the linear response of the ACP. On the other hand the discharge tube which may endure approximately548 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY OCTOBER 1991 VOL. 6 20 h of use is continually undergoing changes owing to the deposition of the metallic copper sputtered from the electrodes and other metallic oxides from the analytes. In addition as observed with MIP source^,^ devitrification is likely to occur generating a reproducibility problem. Therefore the end-on approach was maintained for the remaining studies.Thermospray Optimization The temperature of the thermospray probe and the spray chamber has an influence on the analyte response. The probe temperature affects the degree of vaporization of the carrier liquid and is dependent on its composition and flow rate;28~29 therefore optimum temperatures may vary with the nature of the vaporizer.30 The effect of the probe and spray chamber temperature on the S/N ratio was systemati- cally studied for sample solutions containing 2 ppm of Cr and 0.5 ppm of Zn. The optimum temperatures were found to be 245 and 240 "C for the probe and spray chamber respectively. These values were maintained for the remain- ing thermospray-ACP studies. Plasma Characteristics The excitation of the analyte species can be attributed to highly energetic helium species present in the plasma.A characteristic pink-purple colour was observed in the well defined and stable plasma. A characteristic background spectrum from the helium ACP with de-ionized water being nebulized into the plasma (GFN 0.5 1 min-I) is presented in Fig. 6. Molecular emission from the OH NH and N2 bands is shown. Other elemental species present in the plasma are also shown (H He 0). Although it is not shown in Fig. 6 emission from the electrodes was observed but at a relatively low intensity. However Si emission from the discharge tube was not observed and can be attributed to the fact that the plasma arc was concentrated in the centre of the capillary tube as a thin plasma jet being more diffuse towards the edges.Apparently the edges of the plasma do not have sufficient energy to atomize and excite Si from the discharge tube. An analyte can undergo various excitation processes depending on the plasma and its characteristic^.^^ This situation may produce different relative intensities for a given element depending upon the excitation source. Results from a comparative line intensity study for the most intense lines of the elements are given in Table 3. All the measurements were performed using a slit- width and height of 30 pm and 5 mm respectively. The H H H OH (second order) \ 200 400 600 Wavelengt h/n m I Fig. 6 Characteristic background spectrum of the helium ACP K Li Mn Na Ni Pb Table 3 Comparison of line intensity for several elements Element Wavelength (and line)/nm Relative intensity Ba 455.40 (11) 1 .oo 493.41 (11) 0.64 553.55 (I) 0.079 Ca 393.37 (11) 0.24 396.85 (11) 0.17 422.67 (I) 1 .oo Cd 214.44 (11) 0.12 226.50 (11) 0.28 228.80 (I) 1 .oo c o 240.72 (I) 1 .oo 241.16 (I) 0.44 242.49 (I) 0.78 345.35 (I) 0.53 359.35 (I) 0.70 360.53 (I) 0.49 248.33 (I) 1 .oo 371.99 (I) 0.2 1 766.49 (I) 1 .oo 670.70 (I) 1 .oo 257.61 (11) 0.99 259.37 (11) 1 .oo 260.57 (11) 0.86 403.08 (I) 0.73 330.14 (11) 0.28 589.00 (I) 1 .oo 589.59 (I) 0.49 341.48 (I) 1 .oo 346.17 (I) 0.60 351.51 (I) 0.53 352.46 (I) 0.99 283.31 (I) 0.93 363.57 (I) 0.56 368.35 (I) 1 .oo 405.78 (I) 0.94 Sr 407.77 (11) 1 .oo 421.55 (11) 0.77 460.77 (I) 0.84 Cr 357.87 (11) 1 .oo Fe 238.20 (11) 0.084 248.81 (I) 0.58 769.90 (I) 0.76 Zn 213.86 (I) 1 .oo centre of the fireball of the axially viewed plasma was focused on the entrance slit of the monochromator for the atomic and ionic emission lines.At this stage of the investigation no experiments were conducted in order to observe if the intensity of the analyte emission line varied with the spatial region of the source. The line intensities were normalized to the most intense line of the element of interest to which a value of 1.00 was arbitrarily assigned. The strongest line was selected for the determination of a particular element. Analytical Performance The performance of the ACP was evaluated for 14 elements in aqueous solutions. The helium flow rate was optimized for each individual element studied using both sample introduction devices.Maximum S/N ratios were obtained by employing helium flow rates in the range from 0.5 to 1.8 1 min-l for the GFN and 0.8 to 1.2 1 min-l for the thermospray interface. The determinations of the elements under study were performed using the appropriate opti-JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY OCTOBER 199 1 VOL. 6 549 Table 4 Comparison of the detection limits (ppb) with other plasma sources ACP MIP Element GFN Thermospray Ar-LP* He-HE? NMIPS Ba 220 Ca 7.0 Cd 10 c o 54 Cr - Fe 130 K 18 Li 1.2 Mn 67 Na 0.74 Ni 180 Pb 120 Sr 29 Zn 29 * Low power Ar MIP ref. 16. t Helium high efficiency MIP ref. 17. 4 Nitrogen MIP ref. 34. 9 Ref. 35. 7 Ref. 36. 99 8.0 5.0 36 51 68 - 0.40 13 13 12 10 24 - 2 70 - 60 7.5 - - 2 700 75 12000 105 975 45 36 - 64 - 10 350 16.5 3 1.5 - - - - 37 17 675 12 15 28 1.2 - - - 5.4 0.22 0.29 - - - 16 120 IcPg 1.3 0.19 2.5 6.0 6.1 4.6 7.51 4.21 1.4 29 10 42 0.42 1.8 mum helium flow rates for the GFN and the thermospray nebulizer.By using the thermospray at a fixed helium flow rate (1 1 min-l) for all elements the results obtained at the optimum helium flow rates did not change significantly. However the GFN could not be operated at a fixed helium flow rate without having a significant impact on the signal response. This behaviour suggests that multi-element deter- minations are possible with one set of conditions when using the ACP-thermospray interface but are unlikely to be obtained with the ACP-GFN interface for the elements that require different helium flow rates. Although the GFN generates a very fine mist with a very small droplet size di~tribution,~~ which greatly enhances efficient sample introduction into the plasma,33 the thermo- spray nebulizer showed lower detection limits than the GFN by factors of 2-10 (see Table 4). It was believed that this performance might be attributed to the heated spray chamber and the partial removal of the solvent from the system permitting a highly desolvated analyte to reach the plasma instead of a wet aerosol.Thus in order to eliminate a possible ‘cooling’ effect of the aerosol on the plasma which may decrease the energy available for the excitation process a suitable desolvation system similar to that used with the thermospray interface was connected to the GFN. In this fashion partial desolvation of the analyte was facilitated; however the cL was improved for only a few elements (Co Ni Pb) and by no more than a factor of 2.This trend suggests that the application of heat to the expansion chamber further desolvates a fraction of the hot aerosol leaving the capillary probe since the analyte released by the thermospray is partially de~olvated.~~ This is supported by the fact that signal enhancement is observed when the expansion chamber is heated until an optimum is reached. On the contrary a desolvation chamber for the GFN might not achieve analyte desolvation to the same extent as the thermospray and the heated chamber because the aerosol leaving the GFN is not partially desolvated; therefore the only desolvation performed is by means of the heated chamber. Calibration graphs were constructed for analyte concen- trations ranging from near the cL to at least 100 ppm.Linear dynamic ranges were observed to be between 2 and 4 orders of magnitude. Detection limits acquired with the ACP for the metals studied and a comparison with data reported with other plasma sources are shown in Table 4. The detection limits obtained with other systems were not necessarily defined by the IUPAC ( 3 0 ) method.27 Thus in order to compare the ACP with the more established systems all detection limits (c,) shown in Table 4 were converted into the 30 criterion by multiplying the original values by the appropriate conversion factor where neces- sary. In this standardized approach the ACP results are comparable or superior to those reported for the MIP sources.Nevertheless a superior cL is achieved by the ICP which can be operated at one set of conditions for multi- element determinations. The feasibility of the ACP for the determination of halogens in an aqueous solution was studied. Unfortunately the results were not as encouraging as expected; thus further studies were not pursued with the present system. The repeatability of the peak height measurements was determined with solutions of the elements under considera- tion in this study. At least six repetitive measurements were performed for each analyte at a level of one order of magnitude higher than the estimated cL. The relative standard deviation (RSD) was <10% for both sample introduction approaches with average values of 5.4 and 7.0% for the GFN and the thermospray interface respectively.Interference Studies Interference studies were conducted by using two classical interference systems the Ca-phosphate system to study the depression of the Ca atom emission signal when refractory compounds are likely to be formed; and the Ca-Na system to observe the effect of an easily ionizable element on the Ca response. The Ca atomic signal was monitored at 422.7 nm. Solutions for each system were prepared containing 10 ppm of Ca and different concentra- tions of the corresponding interferent (phosphate as H3P04 and Na as NaC1). The effect of increasing the H3P04 and Na concentration on the Ca emission signal is shown in Figs. 7 and 8 respectively. The interference produced by increas- ing the H3P04 concentration becomes severe at an H3P04:Ca molar ratio above 0.12; at a molar ratio of 1.2 the signal is reduced to about lo% decreasing rapidly to zero above that molar ratio.On the other hand the interference produced by the addition of an easily ionizable element (Na) is associated with an enhancement of the Ca5 50 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY OCTOBER 199 1 VOL. 6 I % 0 2 0.6 a .s 0.4 m - 2 0.2 0 0.012 0.12 1.2 12 H,PO,:Ca molar ratio Fig. 7 Effect of increasing the H3P04 concentration on the Ca emission signal 1 I & 0.8 . al 2 0.6 . (d Q) U 2 - 0.4 0.2 I I 1 100 0 0.1 1 .o 10 “a1 ( ~ P m f Fig. 8 Effect of increasing the Na concentration on the Ca emission signal atomic signal for concentrations of up to 10 ppm of Na. A further increase of the Na concentration did not enhance the Ca emission response. Excitation Temperature The excitation temperature was determined from the spectral emission intensities of ten Fe atomic lines by using the slope method.37 The thermometric species was intro- duced into the ACP by means of the GFN with the monochromator slit-widths adjusted to 20 pm.Relative transition probabilities tabulated by Reips and Bridges and K ~ r n b l i t h ~ ~ and normalized to the Fe I 371.994 nm line were employed in the temperature calculations. Excita- tion temperatures obtained with the transition probabili- ties of Reif,38 5900 K k 7% yielded a slightly better correlation coefficient (ie. 0.98 in comparison with 0.97) and less relative error than those tabulated by Bridges and K ~ r n b l i t h ~ ~ 5640 K k 9%.Nevertheless no statistical dif- ference was found between the two results at the 95% confidence level. It should be noted that temperature gradients which may exist within the source have not been considered in these experiments. As observed by Vogel and K o l a ~ i n s k i ~ ~ in spectroscopic measurements of temperature distributions in short arcs higher temperature values observed for the ACP in these preliminary studies might correspond to a spatial region close to the electrodes. Although these values seem to be sufficient for excitation of the analytes and are comparable to other plasma sources,14J7941*42 an examination of the c and the inter- ference studies strongly suggests that the ACP is a ‘cooler’ and less robust plasma source than the ICP.Conclusion Although the ACP is not as powerful and does not have the capabilities of the ICP for example it has been demon- strated that the helium ACP is a feasible inexpensive alternative for spectrochemical determinations at high ppb and low ppm levels. The ACP can easily be implemented in the laboratory at a relatively low cost. External initiation of the plasma is not required because the voltage used is above the breakdown voltage for striking the a.c. arc and direct nebulization of the aqueous solution does not extinguish the plasma. The authors acknowledge the Seed Money Research Program at the University of Lowell for providing financial support and L. A. C. thanks the Graduate School at the University of Lowell for providing a summer research fellowship in 1990.Also the authors are deeply apprecia- tive for the data acquisition system donated by Galactic Industries Corporation (Salem NH USA) and the capillary probe very generously donated by Alltech Associates (Deerfield IL USA). 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 References Inductively Coupled Plasma in Analytical Atomic Spectroscopy eds. Montaser A. and Golightly D. W. VCH New York 1987. Inductively Coupled Plasma Emission Spectroscopy ed. Boum- ans P. W. J. M. Wiley New York 1987 parts I and 11. Hieftje G. M. Spectrochim. Acta Part B 1983 38 1465. Boumans P. W. J. M. and Hieftje G. M. in Inductively Coupled Plasma Emission Spectroscopy ed. Boumans P. W. J. M. Wiley New York 1987 part I ch. 5. Skogerboe R. K. and Coleman G.N. Anal. Chem. 1976,48 61 1A. Zander A. T. and Hieftje G. M. Appl. Spectrosc. 1981 35 357. Goode S. R. and Baughman K. W. Appl. Spectrosc. 1984,38 755. McCormack A. J. Tong S. C. and Cooke W. D. Anal. Chem. 1965,37 1470. Estes S. A. Uden P. C. and Barnes R. M. Anal. Chem. 1981 53 1829. Beenakker C. I. M. Spectrochim. Acta Part B 1976 31 483. Beenakker C. I. M. Bosman B. and Boumans P. W. J. M. Spectrochim. Acta Part B 1978 33 373. Michlewicz K. G. and Carnahan J. W. Anal. Chem. 1985 57 1092. Hass D. L. and Caruso J. A. Anal. Chem. 1984 56 2014. Urh J. J. and Carnahan J. W. Anal. Chem. 1985 57 1253. Ng K. C. and Shen W. Anal. Chem. 1986 58 2084. Long G. L. and Perkins L. D. Appl. Spectrosc. 1987,41,980. Long G. L. and Perkins L. D. Appl. Spectrosc. 1989,43,499. Layman L.and Hieftje G. M. Anal. Chem. 1975 47 194. Churchwell M. E. Messman J. D. and Green R. B. Spectrosc. Lett. 1985 18 679. Green R. B. and Williams R. R. Anal. Chim. Acta 1986 187 301. Costanzo R. B. and Barry E. F. Anal. Chem. 1988,60 826. Costanzo R. B. and Barry E. F. J. Chromatogr. 1989 467 373. Costanzo R. B. and Barry E. F. J. High Resolut. Chromatogr. Chromatogr. Commun. 1989 12 522. Colon L. A. and Barry E. F. J. Chromatogr. 1990,513 159. Koropchak J. A. and Winn D. H. Anal. Chem. 1986 58 2258.JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY OCTOBER 199 1 VOL. 6 5 5 1 26 27 28 29 30 31 32 33 34 35 Savitzky A. and Golay M. J. E. Anal. Chem. 1964,36 1627. 36 Nomenclature Symbols Units and their Usage in Spectro- chemical Analysis 11 Data Interpretation Spectrochim. Acta 37 Part B 1978 33 241. Vestal M. L. and Fergusson G. J. Anal. Chem. 1985 57 38 2373. 39 Koropchak J. A. and Winn D. H. Appl. Spectrosc. 1987 41 131 1. 40 Roychowdhury S. B. and Koropchak J. A. Anal. Chem. 41 1990 62 484. Ingle J. D. Jr. and Crouch S. R. Spectrochemical Analysis 42 Prentice-Hall Englewood Cliffs NJ 1988 chs. 7 and 8. Layman L. R. and Lichte F. E. Anal. Chem. 1982 54 634. Nisamaneepong W. Hass D. L. and Caruso J. A. Spectro- chim. Acta Part B 1985 40 3. Deutsch R. D. Keilsonhn J. P. and Hieftje G. M. Appl. Spectrosc. 1985 39 531. Winge R. K. Peterson V. J. and Fassel V. A. Appl. Spectrosc. 1979 33 206. Fraley D. M. Yates D. and Manahan S. E. Anal. Chem. 1979 51,2225. Kalnicky D. J. Fassel V. A. and Kniseley R. N. Appl. Spectrosc. 1977 31 137. Reif I. Ph.D. Thesis Iowa State University 197 1. Bridges J. M. and Kornblith R. L. Astrophys. J. 1974 192 793. Vogel N. and Kolacinski Z. J. Phys. D 1987 20 545. Golightly D. W. Porrzapf A. F. and Thomas C. P. Spectrochim. Acta Part B 1977 32 3 13. Faires L. M. Palmer B. A. and Engleman R. Jr. Spectro- chim. Acta Part B 1984 39 819. Paper 1/01612A Received April 8th I991 Accepted June 20th 1991