A low-power oxygen inductively coupled plasma for spectrochemical analysis—IV. Analytical features

A low-power oxygen inductively coupled plasma for spectrochemical analysis--IV. Analytical features PENGYUANYANG and RAMON M. BARNES* Department of Chemistry, Universrty of Massachusetts, GRC Towers, Amherst, MA 01003-0035,U.S.A. (Receiued

1December 1988, in revtsed

form15 June 1989)

Abstract-The detection limits for 38 elements in a low-power (2 kWf oxygen mductively coupied plasma (ICP) are found to be a factor of 10 Inferior to those in an argon ICP as reported by WINGE et al. [A@ Spectrosc. 33,206 (197911,but better than in an air or nitrogen ICP. A I4-times improvement of the signal-to-background ratio has been achieved in the determination of barium in a CaO powder sample in the oxygen ICP compared to that in the argon ICP. 1. ~NTR~OUC~ON PURE OXYGEN and argon-oxygen mixed gas inductively coupled plasma (ICP) discharges have been described previously [l-5]. The combination of argon and oxygen gas generally requires no modification of either the torch configuration or plasma generator. However, pure oxygen or other pure molecular gas ICP discharges often caused damage to the torch in earlier studies. To sustain a pure oxygen ICP, originally a demountable torch was developed by MEYER and BARNES[6]. Their design alleviated the torch damage but did not eliminate it totally. Recently, a computer simulation method was employed [7] to predict the possible torch geometry for an oxygen ICP. The results of the simulation indicated that the gas velocity in the central channel plays an important role in the discharge stability. By carefully controlling the central gas flow, a stable low-power (2 kW) oxygen discharge should be sustained in a conventional torch with no modification. The detection limits for 19 elements in a low-power (2 kW) oxygen ICP were reported by MEYER and THOMSON [l]. Under operating conditions identical to those employed in an air ICP, the oxygen ICP gave inferior detection limits compared to the results from the air discharge. Our studies [7-g], however, indicate that the oxygen ICP is hotter than either the air ICP [lo] or nitrogen ICP [l l] owing to its unique thermal properties. In contrast to results in Ref. Cl], the detection limits should be lower in the oxygen ICP than in the air ICP, because the discharge temperature is higher and molecular band intensities, especially from nitrogen-containing species, lower. Previous studies [IO, 12-141 showed that an air ICP is suitable for the analysis of powders and organic samples. This experience provides good reason to believe the oxygen ICP should also be suitable for these analyses. The analytical figures of merit of the oxygen ICP, therefore, are expected to be better than those of other molecular gases [7, 15]. However, experimental verification is necessary. This paper describes a modified procedure to generate an oxygen ICP discharge at low power (2 kW) and high frequency (40.68 MHz). The detection limits for 38 elements are reported. A comparison of detection limits obtained in the oxygen ICP with those in argon and air ICP discharges is given. The capability to determine the Ba impurity in a CaO powder sample is also illustrated, and compared with that of an argon ICP.

2. EXPERIMENTAL 2.1. Apparatus The essential instrumentation is listed in Table 1. The torch housing and match box were placed on a computer-controlled translation stage. The plasma image was formed with unit magni~cation on the entrance slit of the monochromator. Data were collected either by an analog-to-digital converter at the output of a picoammeter or with a recorder. A conventional ICP torch was used with an 18 mm inner

*Author to whom the correspondence should be sent. 167

168

P.

YANG

and R. M. BARNES

Table 1. Instrumentation Inductively coupled plasma

HFS 5ooOD, 40.68MHz, Plasma Therm

Scanning monochromator

l*

SPEX-1700 II, 1200 grooves mm-‘, 0.75 m focal length, 1.11 nmmm- ’ linear reciprocal dispersion, slit width 0.02 mm

Scanning monochromator

27

Heath EU-700,0.35 m focal length, 1180 grooves mm- 1

Photomultiplier

lP28A, -750V

Optics

Quartz, 2 in diameter, 200 mm focal length, 1: 1 image

* Used in measurements of detection limits. t Used in plasma background measurements. diameter outer tube, 16mm i.d. intermediate tube, and 1.2 mm i.d. central tube. A Meinhard concentric nebulizer and the Scott spray chamber were used for all experiments.

2.2. Background

measurement Observation heights for the background emission measurements were - 5,2,4,8,

13 mm relative to the top of the three-turn induction coil. The oxygen ICP background was recorded from 200 to 700 nm where the most prominent analytical lines can be found. All molecular bands and atomic lines of oxygen were identified according to Refs. [16] and [17]. 2.3. Detection limits measurement For detection limit measurements the translation stage was moved vertically along the axis (4 to 10 mm above the induction coil) until the best signal-to-background ratio (SBR) was found. To avoid possible spectral interference during the measurement, a single element solution was used. Solutions with concentrations between 1 and 1OOpgml were nebulized depending on line sensitivities. The calculation of detection limit was based on the SBR method proposed by BOUMANS and DE BOER[ 183, which assumes that the standard deviation of the background (S,) is at about 1% level of the background signal (I,). Whether the assumption applies or not depends primarily on the flicker noise in the ICP and secondarily on the shot noise contribution [19-211. However, this assumption has been applied to evaluate detection limits [22, 233 as: C, = 3(&,/I,) (C/SBR). 2.4. Solid sample analysis Sampling devices used to introduce a powder sample directly into the ICP discharge have been reviewed [24-271. The device described by NG et al. [26] employed pulse sampling. The pulse introduction approach has several advantages over the continuous introduction technique, such as fast response, low contamination, and high analysis speed. Unfortunately, with the procedure used by NG et al., the plasma must be turned off before sample loading and be reignited for analysis [26]. To surmount this limitation, a sample delivery system was designed in the present investigation which is independent of the plasma system. The schematic diagram is given in Fig. 1. a

TO

SILICON

RUaaER

‘2

ICP

COVER

SAMPLING

CARRIER ,!,

IN

~&f=&NJEcToR

GAS

IN

CHAMaER

Fig. 1. Cross sectional view of powder sample delivery system. (a) Sample receiver, (b) powder injector, (c) injector chamber.

Low-power

oxygen

ICP-IV

169

As illustrated in Fig. 1, a tee connector (LEE Instrument, part no. TFMA 06291022) was used as the powder receiver. One end was sealed with a small silicone rubber septum. The gas flows from the inlet through a right angle in the middle of the tube, and then the powder aerosol is carried out of the other end. The powder sample was introduced through the powder injector. The injector was made from a modified hypodermic needle (Fig. 1). The size of the needle could be selected according to the powder size (e.g., a 0.23 mm i.d. was suitable for a powder sample with diameter < 100 pm). The sampling gas inlet tube was connected tangentially to the wall of the sample chamber [26]. A plastic cover was used to close the sampling system after sample loading. The powder sample was delivered as follows. A 0.5 to l.Omg CaO powder sample (200 mesh) was weighed and loaded into the injector chamber. The needle was inserted into the tee tube through the rubber septum (see Fig. l), and then the sample carrier gas flow was started. All of the powder was transferred to the powder receiver and carried into the plasma. When a measurement was finished the injector was withdrawn from the septum to prepare for the next sample load. No prepurging of air introduced during the sample loading was necessary. 3. RESULTSAND DISCUSSION 3.1. Generation of a low-power oxygen ICP discharge The successful generation of an oxygen ICP was started in argon first and converted to an oxygen discharge through sequential steps. A conversion procedure was outlined previously by MERYERand BARNES[6, 281. In this earlier work a boron nitride aerosol tube in a demountable torch was employed to prevent overheating from the hot plasma during the conversion. To improve the conversion procedure, the effect of three gas flows during the conversion was examined by computer simulation [7]. The results showed that the hot plasma core expands close to the gas inlet tube if the aerosol gas how is absent. Consequently, the sample injection orifice is overheated, resulting in an unsuccessful conversion. Stopping the aerosol gas flow was a necessary step in the previous scheme [28]. However, to prevent overheating of the sample injection tube from the plasma core in a conventional torch, the central gas flow must not be stopped during argon-to-oxygen conversion procedure. With this condition a conventional quartz torch can be used to generate an oxygen ICP [7]. Thus, in the present study, the central gas flow remained on all of the time. Neither modification of the plasma torch nor alteration of the plasma generator is considered necessary. A modified two-step conversion procedure is summarized in Table 2. The conversion starts by the gradual addition of oxygen gas into the outer argon gas (step 1). At this time the plasma volume begins to contract, depending on the amount of oxygen added. The rf generator power level is then adjusted to 1.5 kW, and the reflected power is fine tuned to a minimum. This step takes 1 to 2min. The next step (step 2) is to switch the intermediate and then aerosol gases quickly within a few seconds from argon to oxygen. A stable oxygen ICP discharge is eventually produced at 2 kW. In this way the plasma core stays within the coil region rather than shifting downstream. When starting, care must be taken to mix the two gases in step 1. The reflected power in the beginning step shows a large change owing to the change of plasma impedance, so that monitoring the reflected power is important in the first step. Recently, MEYER improved the conversion procedure for an air ICP discharge [13]. An air-cooled argon ICP was generated in one step. The intermediate argon gas was then quickly switched to air to form a pure air ICP. This modified scheme was not successful with the oxygen ICP. Table 2. ICP conversion Power Step 0 1 2

(kw) 0.8 1.5 2.0

Outer gas (15 l/min) Ar Ar+O,

+02 0,

procedure

from argon

Intermediate (0.5 l/min) Ar Ar 0,

to oxygen Aerosol (0.7 I/min)

Time

Ar Ar 0,

2 min.
170

P.

YANG and R. M. BARNES

The swirl velocity of the outer gas flow was studied. To introduce the outer gas with a high initial tangential velocity, a nozzle injector with 1 mm i.d. replaced the original 3 mm tube inlet. High tangential velocity disturbed the stability of the oxygen ICP discharge, and the conversion was difficult if a 1 mm id. nozzle injector was used. The computer simulation [7] also predicted that a back-flow can be formed when the swirl velocity at the outer gas inlet is greater than 30m/s. As predicted by the computer simulation [7], the formation of the central channel in the oxygen ICP is also dependent on the central gas velocity. To test this prediction the internal diameter of the injection tube was varied. The tested parameters are listed in Table 3. A solution with lOOpg/ml Y or Na was nebulized into the oxygen ICP. The sample channel was clear in the normal analytical region above the coil where the plasma background decreases. This colorful channel was not visible within the core region owing to strong plasma background. The sample aerosol seems to be restricted to a narrow channel in the oxygen ICP compared to the argon ICP. With an orifice id. up to 1.4mm, forming a central channel was not difficult as long as the inner gas flow was greater than 0.4l/min. 3.2. Plasma background The spectra from 200 to 700 nm were recorded for the oxygen ICP at observation heights of - 5,2,4,8 and 13 mm relative to the top of the induction coil. The background spectrum is very similar to that reported in Ref. [ 1J. As MEYERand THOMPSON[l] and LIU er al. [29] indicated, the oxygen ICP background differs significantly with the selected observation height. In general, the &human-Runge band system of molecular oxygen is strong, and the atomic oxygen line emissions are weak in the upper plasma region (e.g. 13 mm above the load coil). The opposite is found in the hot plasma region (e.g. - 5 mm below the top of the induction coil). In the normal analytical zone (2 to 1Omm above the top of the induction coil), the magnitude of both molecular band and atomic line emission are comparable [8]. The plasma background in the visible region is relatively low compared to that in an argon ICP [8]. 3.3. Detection limits in the oxygen ICP Based on results of a study of excitation processes in the oxygen ICP [30], 84 lines of 38 elements were selected for measurement of detection limits. The results are listed in Table 4. For comparison the detection limits in an argon ICP reported by WINGEet al. [22] and in oxygen and air ICP discharges measured by MEYERand THOMPSONCl] are also listed. Recently, BOUMANS and VRAKKING [313 studied 100 prominent lines in the OH band region using a 50 MHz ICP with a high-resolution echelle monochromator with predisperser. The detection limits were found to be much better than those reported by WINGEet al. [22]. In the comparison of the detection limits, the values of WINGE et al. for an argon ICP are considered, however. In general, the detection limits for most elements measured are 10 times poorer in the oxygen ICP than in the argon ICP of WINGE et al., but they are better than or comparable to those for an air ICP. The detection limits in the present oxygen ICP are better than those measured by MEYERand THOMPSON[l] under similar operation conditions. For alkali elements the detection limits are competitive with those in the argon ICP, and some of them, such as Na I 589.0nm line, are even better. This feature results from the high sensitivity of the line and the relatively low plasma background. Therefore the SBR is Table 3. Tested parameters for the central tube diameter

i.d. (mm) 0.5 0.8 1.0 1.4

Flow rate (I/min)

Gas velocity calculated (I/min)

0.2-0.7 0.3-l 0.4-l 0.4-l

16.8-58.8 10-32 8.4-21 5.4-l 1

Low-power Table 4. Detection

limits in a 2 kW oxygen

oxygen

ICP-IV

ICP. Elements groups

171

listed according

Detection wavelength Element

(nm)

Oxygen* (this work)

-

to their periodic

table

limits (ng/ml)

OxygenCl1

Air [l]

Argon

[22]

GroupI Li I Li I Na I Na I KI cu I

670.784 610.362 588.995 589.592 766.490 324.754 328.068 338.289 242.795 267.595

5 100 10 44 75 150 75 20 187 225

313.042 234.816 279.553 393.366 396.847 407.771 455.403 213.856 636.235 228.802 253.652

6 50 9 3 3 6 95 5000 21 126

BI Al I Al I Gal Ga I Ga 1 In II SC II SC II Y II Y II La II La II

249.772 396.152 309.273 294.364 403.296 417.206 230.606 361.384 363.075 371.030 437.494 394.910 408.677

32 272 163 375 600 136 600 19 45 20 25 9 100

Group IV Si I Sn 1 Sn 1 Pb I Pb I Pb II Ti II Ti II Zr II Zr II

251.611 286.333 303.412 368.348 283.307 220.353 334.941 336.121 257.139 267.863

150 600 500 1000 1300 2200 43 190 260 280

213.618 214.914 310.230 290.888 309.311 223.948 223.198 246.003 228.916

2100 1500 65 114 110 571 500 460 759

Ag Ag Au Au

I I I I

Group II Be I1 Be I Mg II Ca II Ca II Sr II Ba II Zn I Zn I Cd I Hg I

1

190

59

29 69 10 7 13 17 31

500

52

2500 21

40 8

26 1240

12 71

25400

6300

3 3 2 2 5 4 13 18 21 61

Group Ill

Group

PI PI v II v 11 v II Ta II Ta II Ta II Ta II

400

120

3000

2700

900

750

48 28 23 46 111 66 63 15 2 4 6 610t 10

214

110

384 142 42 8 5 10 15

V

16 76 64 8 50 31 31 28 52 (continued)

172

P. YANG and R. M. BARNES Table 4. (continued) Detection wavelength (nm)

Oxygen* (this work)

Cr II Cr II Cr II Cr I Cr I MO II MO II MO II MO II MO I W II W II W II

267.716 283.563 286.257 425.435 427.480 202.030 203.844 204.598 284.823 379.825 232.609 248.923 207.911

600 224 750 514 610 1870 2000 2560 789 900 1970 2800 5000

Group VII Mn II Mn II Re II Re II

257.610 259.373 221.426 227.525

5 12 267 750

Group VIII Fe II Fe II Fe II Fe II co II co II co II co I Ni I Ni I Ni I Pd I Pd I Pd I OS I OS II OS II Pt II Pt I Pt I Ce II Ce II Ce II Ce II Ce II Yb II Yb II

238.204 239.562 259.940 261.187 237.086 231.405 237.867 340.5 12 341.476 351.505 352.454 340.458 360.955 363.470 326.229 225.585 228.226 214.423 217.467 265.9 418.666 393.109 394.275 413.380 413.765 328.937 369.4

Element

Oxygen

limits (ng/ml)

Cl1

Air [l]

Argon

[22]

Group VI

500 476 187 500 182 221 130 140 60 100 60 115 150 300 1250 2400 3300 300 150 110 450 600 670 660 680 45 38

7 7 2

8 12 12 20 76 73 30

410

14 16 6 6

5 5 6 12 97 16 10 48

810

180

44 85 54 4 6 30 83 81 52 60 60 50 48 2 3

* Spectral bandwidth is about 23 pm for the slit width of 0.02 mm in this work. t Coincidence with Ar 394.10mm.

increased. For the La II line at 394.9 nm, a good detection limit of 10 ng/ml is obtained which is comparable to that in the argon ICP with an alternative wavelength, since the La II line (394.91 nm) is overlapped with an argon atomic line (394.90nm). The alkaline earth elements, as in argon ICP, are also sensitive in the oxygen ICP. Their lines are usually 10 to 100 times as sensitive as the prominent lines of other elements. The

Low-power

oxygen ICP-IV

173

detection limits for alkaline earth elements are even lower in the oxygen ICP than in the air ICP El]. comparing the SC family elements with the Cr family, the detection limits are 10 times better in the former than in the latter. As discussed previously 1301, the total energy required for ionization and excitation for the SC family elements is much Iower than the ionization potential of oxygen. By charge transfer reaction with oxygen ions, the population of the excited state for these ions is enhanced over the LTE values. The non-metal and metalloid elements normally exhibit the same behavior as in the argon ICP. They are not sensitive in the oxygen ICP either. Similarly, the Cu and Zn family elements exhibit a detection limit more than 10 times poorer in the oxygen ICP than the argon ICP. A noteworthy line is the Zn I line at 636.2 nm. This line is very sensitive in arc emission, but extremely insensitive in the oxygen ICP, which indicates their different excitation mechanisms. The seriousness of OH band interferences in an argon ICP was reported by BOUMANSand VRAKKING[31]. In the oxygen ICP the OH bands seem to be stronger compared to the argon ICP. Therefore, some lines listed in Table 4 are not recommended as analytical lines because of the OH band interferences. The Be II 3 13.042 nm and V II 310.230 nm lines are seriously overlapped by the OW 313.02’7, 313.056nm, 310.214 and 3i~.236nm bands, respectively. The Al I 309.273 nm, Cr II 283.563 nm, and V II 290.88 am lines are also interfered with by the OH bands. The coincidence of a prominent line with a band component (oxygen line, or 0, and OH band) worsens the detection limits considerably, not only because the SBR is increased, but even more because the selectivity is decreased [32, 33j. The oxygen ICP exhibits detection limits better than those reported in the air ICP for many elements, although it cannot be concluded that the oxygen ICP is superior to the air ICP from the few reports available in the literature. The detection limits measured in this study may be improved by using a high-resolution spectrometer [19] and by further optimizing the operating conditions. For example, the power level was not varied in this investigation. The applied power (2 kW) may not be optimum for some elements.

Some preliminary results of CaO powder sample analysis in the oxygen ICP will be discussed. The results obtained were not optimized values, because this exploratory investigation is intended only to demonstrate the potential of powder sample analysis. 3.4.1. Evaluation of sample delivery system. With the powder delivery arrangement described (Fig. I}, the system is capable of introducing a powder sample in a simple and fast way without seriously disturbing the plasma. The tangential transport gas is important. Gas flow con~gurations of different types were tested, and the present one gives a high velocity at the wall side of the sample chamber, which efficiently removes all the powder from the injector system. The sample receiver allows the normal carrier gas flow to pass through, while the powder sample can be introduced at any time. For both argon and oxygen ICP discharges, removing air from the injector system is not necessary. In the argon ICP, introduction of small amounts of air will alter the central channel temperature and cause some molecular background emission. This effect is found to be negligible in the oxygen ICP. 3.4.2. Measurement of Ba II in CaO powder sample. To examine the potential of the oxygen ICP in the analysis of a powder sample, CaO powder (40-60 pm) was introduced into both argon and oxygen ICP discharges. The barium impurity with unknown concentration was monitored by emission spectroscopy. The Ba II 455.4nm line was selected as a probe line owing to its high sensitivity in both discharges. The power levels (1 kW for the argon ICP and 2 kW for the oxygen ICP) were not optimized in either case, since they are the normal operating conditions and are comparable with each other. The results are ~llust~ted in Fig. 2. An improved analytical figure of merit has been derived from these results. The background recorded at 455.4 nm from the argon ICP is twice as large as in the oxygen ICP. The SBR based on peak height is 0.25 for the argon ICP and 3.5 for the oxygen ICP, thus an advantage of a factor of 14 is obtained for the oxygen ICP. The improvement in the detection

174

P. YANGand R. M. BARNES

2 ; .z 5 z

0.8

-

0.6

.

0.4

-

0.0

J 0.0

2.4

4.8

7.2

Transient

time,

9.6

12.0

S

Fig. 2. Signal response of the Ba II 455.4nm line. The argon ICP was operated at 1 kW, and the observation height was 16 mm above the induction coil where a high SBR was found. The oxygen ICP was run at 2 kW, and the observation height was 12 mm above the induction coil for the best SBR. A total of OSmg powder sample was injected. (a) Argon ICP f! kW), (b) oxygen ICP (2 kW).

0.80

0.64

0.32

0.18 -

1

Distance

above

load coil, mm

Fig. 3. Effect of observation height on signal response. The oxygen ICP was operated at 2 kW; 1 mg powder of CaO was injected, and Ba II 455.4 nm line was monitored.

of limit is estimated from peak heights of the Ba II and background signals for both discharges assuming a background RSD (relative standard deviation) to be 1%. The detection limit in the oxygen ICP is found to be l/10 of that in the argon ICP for the Ba II determination. The thermal pinch effect in the oxygen ICP, resulting from the high thermal conductivity of oxygen, is strong in addition to the magnetic pinch effect [S, 303, which supplies sufficient energy for sample decomposition. As a consequence, the oxygen ICP seems to be a good source for sample decomposition.

175

Low-power oxygen ICP-IV 0.60

0.48

0.36

0.24

0.8

1.2

Mass of powder,

1.6

2.0

mg

Fig. 4. Signal response as a function of mass of powder introduced. The oxygen ICP was operated at 2 kW, 1 mg of CaO powder sample was introduced, and Ba II 4554nm line was monitored. Duplicate or triplicate samples are indicated by data points.

3.43. Efict of observation height and amount powder. The influence of observation height in the oxygen ICP was examined. The results for 1 mg of sample injected are plotted in Fig. 3. The background levels were relatively low over a wide range from 10 to 20mm above the induction coil. The sensitivities of analyte emission were also fairly good in this region. The optimum observation location is around 12 to 14 mm. In contrast, the observation height in the analysis of an aqueous solution is around 4 to 10 mm above load coil in the oxygen ICP. The time required for a complete powder decomposition shifts the observation point upstream above the induction coil. This period would determine the total distance traveled by a particle in the plasma region. The Ba signal response as a function of the amount powder injected is shown in Fig. 4. A limit appears to exist on the amount of injected powder. When the amount of powder is less than 0.8 mg a linear regression of Ba emission signal against the sample weight was obtained (the solid line in Fig. 4). The correlation coefficient of this linear regression is 0.99, showing a farily good linearity between the amount of sample introduced and the corresponding emission signal. Higher loads of sample do not correspond to an increased signal. The diameter of the injector needle may restrict the powder delivery rate, so that the maximum peak height might be limited as the amount of powder is increased. Possibly not all of the particles are decomposed as the total number of particles increases. The particle decomposition and velocity of particles in the oxygen ICP still remain to be studied. 4. C0NcLUsr0~ No modification is necessary for the generation of an oxygen ICP discharge for a conventional torch used routinely in the argon ICP. The conversion from an argon ICP to an oxygen ICP takes about 2 min in a two-step procedure. The central sample channel is easily formed, depending on the aerosol gas velocity. The detection limits for many elements measured in the oxygen ICP are generally a factor of 10 poorer than those reported by WINCE et al. [223 in an argon ICP, but are better than or comparable to those ;n the air ICP. The oxygen ICP exhibits its potential application in solid sample analysis. Particles are found to undergo better decomposition in the oxygen ICP than in the argon ICP. Thus, the SBR and detection limits can be expected to improve

176

P. YANG and R. M. BARNES

significantly in the oxygen ICP. The results obtained in this study should encourage further investigation and application of the oxygen ICP for spectrochemical analysis. Acknowledgment-Research

sponsord by the ICP Information Newsletter.

REFERENCES Cl] G. A. Meyer and M. D. Thompson, Spectrochim. Acta 4OB, 195 (1985). [Z] S. Greenfield and D. J. Smith, Anal. Chim. Acta 59, 341 (1972). [3] K. Liu, N. Kovacic, I. Bletsos and R. M. Barnes, 29th Annual Conference Spectroscopy Society of Canada, Paper 11, 1982. [4] A. Montaser and R. L. Van Hoven, CRC Crit. Rev. Anal. Chem. 18, 45 (1987). [S] K. D. Ohls, D. W. Golightly and A. Montaser, Mixed-gas, molecular-gas, and helium inductively coupled plasmas operated at atmospheric reduced pressures, in: Inductively Coupled Plasmas in Analytical Atomic Spectrometry, Eds D. W. Golightly and A. Montaser, Ch. 15, p. 563. VCH Publishers, New York (1987). [6] G. A. Meyer and R. M. Barnes U.S. Patent 4,482,246, (November 13, 1984). [7] P. Yang and R. M. Barnes, Spectrochim. Acra 44B, 000 (1989). [8] P. Y. Yang, Ph.D. dissertation, University of Massachusetts, Amherst, MA (1987). [9] P. Y. Yang and R. M. Barnes, (1986) Winter Conference on Plasma Spectrochemical Analysis, Paper 125, Kailua-Kona, Hawaii, 1986. [lo] N. Kovacic, G. A. Meyer, K. Liu and R. M. Barnes, Spectrochim. Acta 4OB, 943 (1985). [11] R. M. Barnes and S. Nikdel, Appl. Spectrosc. 30, 310 (1976). [12] G. A. Meyer, Spectrochim. Acta, 42B, 201 (1987). [13] G. A. Meyer, 1986 Winter Conference on Plasma Spectrochemical Analysis, Paper 59, HawaII, 1986. 1141 R. M. Barnes and G. A. Meyer, Spectrochim. Acta 4OB, 893 (1985). [15] P. Y. Yang and R. M. Barnes, 1987 Winter Conference on Plasma and Laser Spectrochemistry, Paper 44, Lyon, France (1987). [16] G. R. Harrison, MIT Wavelength Tables, 2nd edn, MIT Press, Cambridge, MA (1969). [17] R. W. B. Pearse and A. G. Gaydon, The IdentiJication ofMolecular Spectra, 4th Edn, Wiley, New York (1976). [18] P. W. J. M. Boumans and F. J. De Boer, Spectrochim. Acta 32B, 365 (1977). Cl93 P. W. J. M. Boumans and J. J. A. M. Vrakking, Spectrochim. Acta 42B, 553 (1987). [20] P. W. J. M. Boumans, R. J. McKenna and M. Bosved, Spectrochim. Acta 36B, 1031 (1981). [21] P. W. J. M. Boumans and J. J. A. M. Vrakking, Spectrochim. Acta 39B, 1261 (1984). [22] R. K. Winge, V. J. Peterson and V. A. Fassel, Appl. Spectrosc. 33, 206 (1979). [23] P. W. J. M. Boumans and M. Bosveld, Spectrochim. Acta 34B, 59 (1979). [24] R. M. Dagnall, D. J. Smith, T. S. West and S. Greenfield, Anal. Chim. Acta 54, 397 (1971). [25] H. C. Hoare and R. A. Mostyn, Anal. Chem. 39, 1153 (1967). [26] K. C. Ng, M. Zerezghi and J. A. Caruso, Anal. Chem. 56, 417 (1984). [27] M. W. Routh and M. W. Tikkanen, Introduction of Solids into Plasmas, in: Inductively Coupled Plasmas in Analytical Atomic Spectrometry, Eds D. W. Golightly and A. Montaser, Ch. 15, p. 431. VCH Publishers, New York (1987). [28] G. A. Meyer, Ph.D. dissertation, University of Massachusetts, Amherst, MA (1982). [29] K. Liu, N. Kovacic and R. M. Barnes, Spectrochim. Acta 45B, 145 (1990). [30] P. Y. Yang and R. M. Barnes, Spectrochim. Acta 4SB, 157 (1990). [31] P. W. J. M. Boumans and J. J. A. M. Vrakking, Spectrochim. Acta SOB, 1423 (1985). [32] P. W. J. M. Boumans and J. J. A. M. Vrakking, Spectrochim. Acta 42B, 819 (1987). [33] P. W. J. M. Boumans and J. J. A. M. Vrakking, J. Anal. Atom. Spectrom. 2, 513 (1987)