JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, FEBRUARY 1987, VOL. 2 21 Effect of Torch Size on a 148-MHz Inductively Coupled Plasma Bryan D. Webb* and M. Bonner Denton Department of Chemistry, Faculty of Natural Sciences, University of Arizona, Tucson, A2 84721, USA Physical parameters and analytical performance are presented for three analytical lCPs operated at 148 MHz. The torch size has been varied in these three systems in order to investigate more closely the influence of the ratio of plasma radius ( r ) to skin depth (s) on the plasma characteristics. The electron number density appears to be directly related to the r/s ratio, while the excitation temperatures and ion to atom intensity ratios only follow a general trend. A 10 mm i.d. torch a t 148 MHz provides conditions most similar to a standard 18 mm i.d.torch at 27 MHz. An intermediate-size torch of 13 mm i.d. provides a good balance between the increased ease of sample handling of a large r/s ratio torch and the improved sensitivity of a small r/s ratio system. The r/s ratio is shown to be a convenient means of understanding the effects of changes in the plasma operating frequency and in the torch size. Keywords: Inductively coupled plasma; torch size; plasma characteristics When investigating the effect of torch size on inductively coupled plasma (ICP) operating characteristics, it is useful to think of the energy deposition region in terms of the skin depth, s. An electromagnetic field is induced in the discharge by the oscillating current flowing in the load coil.The skin depth is defined as the distance required for this induced field strength to fall off to l l e of its value at the surface. A previous investigation into plasma characteristics1 studied the effect of decreasing the skin depth by increasing the operating fre- quency to 148 MHz, and electron density was found to change proportionally with skin depth at a fixed torch size. It was not clear from this study whether the skin depth alone or the ratio of plasma radius, r, to skin depth (r/s) was the controlling parameter. The ratio of plasma radius to skin depth may be adjusted at a fixed frequency by changing the torch size. An “optimum” rls ratio of 2.25 was stated in the literature,z which is approximately obtained with an 18 mm i.d. torch at 27 MHz. Several workers have investigated smaller torches at 27 MHz,3-5 which have a smaller r/s, but no electron densities were measured, and excitation temperatures were determined only at single observation heights .3,4 These observations appear to have been made in the normal analytical zone (NAZ) as defined by Koirtyohann and co-workers.697 The advantage of a larger rls ratio is increased ease of sample introduction, as the analyte, which is injected along the central channel of the plasma, is farther from the primary energy deposition region, and hence less perturbing of the energy deposition process.The disadvantage is a lower energy environment experienced by the analyte. Detection limits were found to degrade at the larger r/s ratio due to the lower energy environment, but organic samples were capable of being aspirated with no change in operating parameters.1 An intermediate rls ratio may be expected to provide a more optimum trade-off between sensitivity and sample handling ability. At a fixed frequency, the rls ratio may be reduced simply by reducing the torch size. The lower size limit is set where the analyte begins to interact with and perturb the energy deposition region. This paper describes the results of such an investigation, in which three different torch sizes have been evaluated at the operating frequency of 148 MHz. The physical parameters of excitation temperature, electron num- ber density and ion to atom line intensity ratios have been * Present address: Unocal Science and Technology Division, 376 South Valencia, Brea, CA 92621, USA. determined for the three torch systems used in this study.Analytical performance has also been evaluated in regard to signal to background ratios, detection limits, calibration graphs, interferences and ease of organic sample introduction. Apparatus and Techniques Instrumentation The 148-MHz generator, matching network and plasma diagnostic techniques have been described previously. 1 Two different demountable torch bases were used in this study with quartz tube sizes and gas flow-rates as shown in Table 1. These tubing sizes were not specifical!y optimised for operation at extremely low powers or gas flow-rates, but were selected from standard sizes for ease of construction. Reducing the diameter of the upper portions of the quartz tubing allowed the smallest “micro-torch” to share the same acrylic plastic torch base and PTFE tube spacers as the standard torch.8 The medium-size “midi-torch” was placed in a base constructed from stainless-steel Ultra-Torr fittings (Cajon Co., Macedo- nia, OH, USA), which were bored out to accept the metric size quartz tubing.The outer gas flow was adjusted in each torch to just keep the plasma off the walls of the outer tube at 1500-W forward power. The nebuliser gas flow was adjusted so that the tip of the initial radiation zone (IRZ) fell 15 mm above the load coil, so that similar plasma zones were observed at each observation height independent of torch size. Table 1. Torch dimensions (mm i.d. x mm 0.d.) Torch Tube Standard Midi Micro Outer- Upper . . . .. . 18 X 20.5 13 x 15 10 x 12 Lower . . . . . . 18 X 20.5 13 X 15 18 x 20.5 lmin-‘ . . . . . . 12.2 12.5 12.8 Upper . . . . . . 1 3 x 15 8 x 10 7 x 9 Lower . . . . . . l o x 12 6 x 8 10 x 12 1min-l . . . . . . 0 0 0 Opening . . . . 1.5mm l.Omm Lower . . . . . . 4 x 6 2 x 4 4 x 6 1min-l . . . . . . 0.7 0.6 0.5 Gas flow-rate/ Intermediate- Gas flow-rate/ Injector- l.Omm Upper . . . . . . 4 x 6 2 x 4 2 x 4 Gas flow-rate/22 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, FEBRUARY 1987, VOL. 2 z F I 4- 5000 C 0 m V u1 .- c 4- .- 4000 Skin Depths The skin depth, s, may be calculated from the equation s(mm) =50*3(aClv)-+ . . . . . . (1) where u is the specific electrical conductivity and p is the relative permeability of the absorbing medium, and Y is the frequency in MHz.2 The relative permeability of a gas can be taken as unity, and the dependence of CJ on temperature at atmospheric pressure is plotted as Fig.1 of reference 2. The actual argon temperature in the energy deposition region to be used in the estimation of u is unknown, but the value of u = 10 S2-1 cm-1 used by Scott et aZ.2 is consistent with the 7000-8000 K found by Uchida et aZ.9 The value of s = 3.1 mm calculated at 27 MHz scales to s = 1.3 mm at 148 MHz due to the change in frequency. Ratios of rls for the three different torches used in this study are 6.2 for the 18 mm i.d. standard torch, 4.6 for the 13 mm i.d. midi-torch and 3.5 for the 10 mm i.d. micro-torch. In contrast, the rls ratio for a standard size torch at 27 MHz is 2.7. - - Results and Discussion Excitation Temperatures Iron I excitation temperatures determined for the three different torches at three power levels at 148 MHz are shown in Fig.1. The highest temperatures are obtained in the 5000 - 2 E c Q ; c 4000 0 m V X UJ - .- c c .- 3000 I 10 20 30 smallest torch at the highest power, which is in agreement with an intuitive picture of the situation. A higher energy environ- ment prevails in the analyte channel when the torch size is decreased, because the analyte channel becomes closer to the energy deposition region. This is in contrast to previous results at 27 MHz, where lower temperatures were found in the NAZ as torch size was decreased.3.4 As rls was less than 2.25 for these torches, it may be reasonable to assume that the maximum temperature is obtained near the previously dis- cussed “optimum” rls = 2.25.2 A different vertical temperature profile is seen for each torch size.That of the largest torch is smoothly decreasing, while the smallest torch exhibits a maximum temperature ca. 20 mm above the load coil. This is 4-5 mm above the bottom of the NAZ. A similar vertical spatial behaviour was also seen for the standard size torch at 27 MHz.lJOJ1 The midi-torch provides the most uniform temperature environment over a range of observation heights. In Fig. 2, excitation temperatures are compared at three different observation heights for the range of rls ratios covered by this study. Values at 27 MHz are included from the previous work.1 The general trend of increasing temperature with decreasing rls can be seen, but a strict linear dependence is not maintained at all observation heights because of the different vertical temperature profiles of the three torch systems, The rls ratio does provide a means to account for the effects of changing both the torch size and the generator frequency.A B - C I I I I I 10 20 30 40 0 Observation heightim rn Fig. 1. B, midi- and C, standard-size torches Iron I excitation temperatures determined at 148 MHz for input powers of (a) 900, (b) 1200 and (c) 1500 W in the A, micro-, I (a’ 6ooo t 1 \ 1 I I I I I I 2 4 6 0 2 4 6 0 2 4 6 3000 0 Ratio of plasma radius to skin depth Fig. 2. Iron I excitation temperatures as a function of rls ratio for input powers of (a) 900, (b) 1200 and (c) 1500 W at observation heights of A, 13.0, B, 19.5 and C, 32.5 mm above the load coilJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, FEBRUARY 1987, VOL.2 23 I I I I Electron Number Densities like that seen earlier for the 27-MHz standard torch. The I I I I 1 I I Electron number densities determined from Stark broadening of the hydrogen HP line at 486.13 nm are presented in Fig. 3. For a given torch, the electron number density increases regularly with increasing power and decreases with observa- tion height. It is more interesting to plot electron densities as a function of the r/s ratio, as in Fig. 4. The electron number density was previously seen to be dependent on the skin depth at a fixed torch size.1 Here, it is more correctly seen to be dependent on the rls ratio. larger rls here seems to shift the position of maximum ion intensity higher in the plasma, which is consistent with the lower energy environment in the analyte channel in the larger torches.This is somewhat curious, however, because the IRZ visibly peaked 15 mm above the load coil for all three torches. The position of maximum ion intensity does not always seem to be correlated with the position of the IRZ. Increased power does not shift the position of the maximum ion to atom ratio, but does increase its value. Ion to Atom Intensity Ratios Fig. 5 shows magnesium ion (280.27 nm) to atom (285.21 nm) emission line intensity ratios determined for the three torches. Again the smallest torch exhibits a vertical spatial profile most Approach to Local Thermal Equilibrium The approach to local thermal equilibrium (LTE) can be estimated from the observed electron number densities and suitable ion to atom emission intensity ratios, (Z+/10).12J3 The Fig.3. B, midi- and C, standard-size torches Electron number densities determined from Stark broadening for input powers of (a) 900, (b) 1200 and (c) 1500 W in the A, micro-, Fig. 4. Electron number densities as a function of rls ratio for input powers of (a) 900, (b) 1200 and (c) 1500 W at observation heights of A, 13.0, B, 19.5 and C, 32.5 mm above the load coilJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, FEBRUARY 1987, VOL. 2 1 1 0 10 20 30 40 0 10 20 30 40 Observation heightimm 1 1 I I 10 20 30 40 Fig. 5. Magnesium ion A, micro-, B, midi- and nm) to atom (285.21 nm) emission intensity ratios for input powers of (a) 900, ( b ) 1200 and (c) 1500 W in the 0.30 2 0.20 >, 3 m 21 - 0.10 0 Fig.6. torches ( a ) //7r.\, /*- * - - 7 5 1 I I I I 1 I 1 10 20 30 40 0 10 20 30 40 0 10 20 30 40 0 bse rvat i on height/ m m Values of b, calculated for input powers of (a) 900, ( b ) 1200 and (c) 1500 W in the A, micro-, B, midi- and C, standard-size parameter b,, which indicates the deviation from LTE, is then obtained from Fig. 6 shows b, values calculated from the electron number densities and magnesium ion to atom intensity ratios for the three torches. As the electron number density falls off fairly smoothly with height, the vertical profile of the b, values closely resembles that of the magnesium intensity ratios. It should be noted that the theoretical basis for LTE here is the electron number density, rather than an excitation temperat- ure.Since excitation temperatures may be dependent on the element involved and the excitation energies of the levels used in the determination,14 the electron number density is felt to be a more suitable framework for LTE calculations.13 This leads to b, values of less than one, rather than of the order of 10-300. 1 5 7 1 6 An under population of ions compared with LTE predictions is indicated instead of an over population as has been generally stated. This is entirely due to the choice of LTE framework. The trend with the rls ratios shown in Fig. 7 is not as well defined as the electron number density dependence, but LTE is more closely approached in the NAZ as rls is decreased and as power is increased.This is consistent with the increased electron number densities and temperatures obtained under these conditions. Analytical Performance Signal to background ratios were determined for ten lines of seven elements at an observation height of 19.5 mm at 1200 W. Although several different concentrations and electrometer settings were employed, the results in Table 2 are all corrected to the same electrometer scale and equivalent concentration; detection limits are also included. A uniform increase in analytical performance is seen for all these elements when torch size is reduced, with a relatively greater increase for the ion emission lines compared with the atom lines. This behaviour is consistent with the increased temperature and electron number density obtained in the smaller torch.However, the magnitude of the improvement does not seem to be strictly related to either of these factors. The smallest torchJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, FEBRUARY 1987, VOL. 2 25 0.50 0.40 0.30 v) f m 4 - ', 0.20 0.10 I I I b) \ \ s i C) 0 2 4 6 0 2 4 6 0 2 4 6 Ratio of plasma radius to skin depth Fig. 7. Values of b, calculated for input powers of (a) 900, (b) 1200 and (c) 1500 at observation heights of A, 13.0, B, 19.5 and C, 32.5 mm above the load coil ~~ ~ Table 2. Signal to background ratios (SBR) and detection limits (DL) Standard torch Midi-torch Micro-torch DLt/ DL/ DW Ca I . . . . . . . . 422.67 22 0.1 31 0.08 145 0.01 Ca I1 . . . . . . . . 393.37 22 0.1 717 0.007 86 1 0.002 Cu I . . . . . . . . 324.75 8.7 0.5 23 0.2 36 0.1 Fe I .. . . . . . . 371.99 4.7 0.5 6.2 0.4 7.6 0.4 Fe I1 . . . . . . . . 259.94 1.9 2 4.5 1 9.6 0.5 Mg I . . . . . . . . 285.21 7 0.5 33 0.09 43 0.07 Ni I . . . . . . . . 341.48 2.2 3 4.5 2 6.5 1 VII . . . . . . . . 309.31 3.8 0.5 4.5 0.4 5.9 0.3 Zn I . . . . . . . . 213.86 2.3 2 8.5 1 18 0.4 Element Wnm SBR* pgml-1 SBR pgml-1 SBR pgml-l Mg I1 . . . . . . . . 279.55 11 0.3 95 0.03 140 0.02 * Corrected to 2 x 10-7 A V-1 scale and 100 pg ml-l equivalent. 7 Concentration of analyte in pg ml-1 that gave an average signal equal to three times the standard deviation of the blank, actually evaluated at that concentration level. comes close, within a factor of 2-8, to the figures of merit determined for the standard torch at 27 MHz using the same optical system.' Calibration graphs were linear over four or more orders of magnitude in all the torches for all elements in this study. Representative graphs for Ca I1 and Cu I are shown in Fig.8 for the 18- and 10-mm torches. The calibration graph for Ca I1 is starting to become non-linear at 1000 g ml-1 in the micro-torch, due to self reversal of this intense emission line,l7 but not nearly as much as in the standard torch at 27 MHz. The greater sensitivity obtained in the micro-torch is also seen in this figure. The interference of phosphate on the Ca I1 393.37-nm emission from a 3 g ml-1 calcium solution in the NAZ was studied up to a molar ratio of 1300: 1 for P043- : Ca. Behaviour essentially similar to that already publishedlJ8 was observed in this study, with a slight decrease in calcium emission at molar ratios greater than 700 : 1.The emission signal decreases less in the 10-mm torch compared with the standard torch, as shown in Fig. 9. Savage and HieftJe19 reported a somewhat greater level of interference in their micro-torch than is seen here. The refractory compounds postulated as a mechanism for this interference2O-22 are apparently adequately vaporised in the NAZ of the reduced- size torches used in this study. In the standard-size torch at 148 MHz, many common organic solvents (including acetone, benzene, hexane, methanol, toluene, chloroform and xylene) were capable of being aspirated at 1200 W with no change in either operating conditions or matching network tuning from aqueous solu- tions. This is a distinct advantage when a combination of aqueous and organic solutions are to be analysed, or when the composition of the sample solution is 'not known. In the medium-size torch, the more volatile solvents began to create a problem, with benzene quenching the discharge after ca.1.5 min. No re-tuning was required for the other solvents. The smallest torch would not accept organic solvents at 1200 W. The discharge decreased in height and expanded in diameter, such that it could not be kept away from the walls of the coolant tube. This progression in sample handling ability is a direct reflection of the change in the r/s ratio, with the larger torches being most immune to sample introduction effects. The progression in detection limits and freedom from vapod- sation interferences are also correlated with the r/s ratio.26 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, FEBRUARY 1987, VOL.2 106 105 104 - m 103 .- v, 102 10‘ 1 OU -10-2 10-1 100 101 102 103 104 Concentrationimg I-’ Fig. 8. Analytical calibration graphs for Ca I1 (393.37 nm) in the A, standard and B, micro-size torches, and Cu I (324.75 nm) in the C, standard- and D, micro-torches - .; , 10’ 1 02 103 104 g 60 1 00 Molar ratio of PO4- to Ca I .- - ; I I I I 10‘ 102 I? 6 0 ‘ Fig. 9. Effect of phosphate on calcium ion emission intensity from a 3 pg ml-1 solution in the A, micro-, B, midi- and C, standard-size torches Conclusion The ratio of plasma radius to skin depth is a useful parameter to consider when changing torch sizes or operating frequencies of ICPs. This provides a coherent means of organising and understanding the effects of such changes. Excitation temper- atures and ion to atom intensity ratios roughly follow the trend in rls, but the electron number density tracks rls quite closely. Because of the method of LTE determination, the approach to LTE of these torches also roughly follows the rls ratio.That these effects are not simply a function of the physical distance between the energy deposition region and the analyte channel (r-s) is demonstrated by the behaviour of the standard torch at 27 MHz, which has a net r-s distance nearly equal to that of the 13 mm midi-torch at 148 MHz. The physical characteristics and analytical performance of the standard 27-MHz ICP are most closely approached by the 10-mm micro-torch at 148 MHz, rather than by the midi-torch.This similarity of the standard and micro-torches does not extend to the case of introducing organic solvents at the low powers used for aqueous solutions. The midi-torch accepts organic solvents almost as well as the standard torch, however, and provides intermediate detection limits. For the analysis of organic solutions, a more optimum balance of sensitivity and sample handling is provided by the midi-torch, while freedom from vaporisation interferences is maintained. These effects are well correlated with the various rls ratios of the torches investigated in this study. This research was partially supported by the Office of Naval Research. 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21.22. References Webb, B. D., and Denton, M. B., Spectrochim. Acta, Part B, 1986,41, 361. Scott, R. H., Fassel, V. A., Kniseley, R. N., and Nixon, D. E., Anal. Chem., 1974,46,75. Savage, R. N., and Hieftje, G. M., Anal. Chem., 1979,51,408. Weiss, A. D., Savage, R. N., and Hieftje, G. M., Anal. Chim. Acta, 1981, 124, 245. Allemand, C. D., Barnes, R. M., and Wohlers, C. C., Anal. Chem., 1979,51, 2392. Koirtyohann, S. R., Jones, J. S., and Yates, D. A., Anal. Chem., 1980,52, 1965. Koirtyohann, S. R., Jones, J. S . , Jester, C. P., and Yates, D. A., Spectrochim. Acta, Part B, 1981, 36, 49. Windsor, D. L., Heine, D. R., and Denton, M. B., Appl. Spectrosc., 1979, 33, 56. Uchida, H., Tanabe, K., Nojiri, Y., Haraguchi, H., and Fuwa, K., Spectrochim. Acta, Part B, 1981, 36,711. Kawaguchi, H., Ito, T., and Mizuike, A., Spectrochim. Acta, Part B, 1981,36,615. Furuta, N., and Horlick, G., Spectrochim. Acta, Part B, 1982, 37,53. Caughlin, B. L., and Blades, M. W., Spectrochim Acta, Part B, 1984,39, 1583. Raaijmakers, I. J. M. M., Boumans, P. W. J. M., Van Der Sijde, B., and Schram, D. C., Spectrochim. Acta, Part B, 1983, 38,697. Alder, J. F., Bombelka, R. M., and Kirkbright, G. F., Spectrochim. Acta, Part B, 1980,35, 163. Boumans, P. W. J. M. , and De Boer, F. J., Spectrochim. Acta, Part B, 1977,32, 365. Furuta, N. , and Horlick, G., Spectrochim. Acta, Part B, 1982, Human, H. C. G., and Scott, R. H., Spectrochim. Acta, PartB, 1976, 31, 459. Larson, G. F., Fassel, V. A., Scott, R. H., and Kniseley, R. N. , Anal. Chem., 1975,47,238. Savage, R. N., and Hieftje, G. M., Anal. Chem., 1980, 52, 1267. Fassel, V. A., andBecker, D. A.,Anal. Chem., 1969,41,1522. Hermann, R., Alkemade, C. Th. J., and Gilbert, P. T., “Chemical Analysis by Flame Photometry,” Interscience, New York, 1963. Fukushima, S . , Mikrochim. Acta, 1959, 596. 37,53. Paper JA612 Received January 27th, 1986 Accepted September 19th, I986