water samples. The technique seems to provide a solution to the problem of rapid-routine quantitative analysis (approximately 1400 determinations/day) in a high-pro- ductivity laboratory environment. Data presented in this paper show that the accuracy of the method compares favorably with established and recognized techniques, (i.e., atomic absorption spectrometry and ultraviolet-vis- ible absorption spectrophotometry), in terms of both sensitivity and precision. In addition, no special sample pretreatment or preparation is required other than rou- tine field filtration and preservation techniques specified for previous methods.

Other applications of water quality analysis utilizing induction-coupled plasma methodology have been re- ported in the literature. 4' 2~-28 However, the methodology described in this paper represents a significant improve- ment in efficiency and ease of analysis compared to other techniques (for example, that described by Taylor, 27 who recommends careful matching of the matrix of the stand- ards to each sample, to provide accurate results of low- concentration analyte elements in the presence of varia- ble major-element concentrations). In addition, the abil- ity to perform highly quantitative analysis of trace ele- ments at low ambient-concentration levels at a very high sample-production rate is unique.

1. E. Brown, M. W. Skougstad, and M. J. Fishman, Methods for Collection and Analysis of Water Samples for Dissolved Minerals and Gases (USGS-TWRI 1970), Book 5, Chap. A1, 1970.

2. M. J. Fishman and D. E. Erdmann, Anal. Chem. 49, 139R (1977). 3. M. W. Skougstad, M. J. Fishman, D. E. Erdmann, and S. S. Duncan, USGS-

Open File Report #78-679 (1978), pp. 1-800. 4. R. K. Winge, V. A. Fassel, R. N. Kniseley, E. Dekalb and W. J. Haas, Jr.,

Spectrochim. Acta 32B, 327 (1977). 5. F. Brech and R. C. Crawford, The Jarrell-Ash Plasma AtomComp (Jarrell-

Ash Division, Fisher Scientific Co., Waltham, MA (1975), pp. 1-20. 6. C. Allemand, ICP Inf. News 2, 1 (1976). 7. A. F. Ward, Paper No. 310 presented at Pittsburgh Conference on Analytical

Chemistry and Applied Spectroscopy, Cleveland (1977). 8. J. R. Garbarino and H. E. Taylor, Paper No. 28 presented at 4th FACSS

Meeting, Detroit (1977). 9. S. E. Valente and W. B. Schrenk, Appl. Spectrosc. 24, 197 (1970).

10. R. H. Scott, V. A. Fassel, R. N. Kniseley, and D. E. Nixon, Anal. Chem. 46, 75 (1974).

11. P. W. J. M. Boumans and F. J. deBoer, Spectrochim. Acta 27B, 391 (1972). 12. S. Greenfield, H. McD. McGeachin, and P. B. Smith, Talanta 23, 1 {1976). 13. S. Greenfield and P. B. Smith, Anal. Chim. Acta 57, 209 (1971}. 14. S. S. Berman and J. W. McLaren, Appl. Spectrosc. 32, 372 (1978). 15. L deGalan, Spectrosc. Letters 3, 123 (1970). 16. R. K. Skogerboe and C. L. Grant, Spectrosc. Letters 3, 215 (1970). 17. K. S. Subramanian, C. L. Chakrabarti, J. E. Sueiras, and I. S. Maines, Anal.

Chem. 50, 444 (1978L 18. R. K. Skogerboe, J. J. Lamothe, G. J. Bastians, S. J. Freeland, and G. N.

Coleman, Appl. Spectrosc. 30, 495 (1976). 19. W. Snellman, T. C. Rains, K. W. Yee, H. D. Cook, and O. Menis, Anal. Chem.

42, 394 (1970). 20. G. F. Larsen, V. A. Fassel, R. K. Winge, and R. N. Kniseley, Appl. Spectrosc.

30, 384 (1976). 21. Jarrell-Ash Model 750 AtomComp Instruction Manual, No. 90-750, Waltham,

MA (1973). 22. P. W. J. M. Boumans, Spectrochim. Acta 31B, 147 {1976). 23. M. W. Skougstad and M. J. Fishman, AWWA Technical Conference Pro-

ceedings, XIX-1 (1974). 24. R. L. Dahlquist and J. W. Knoll, Appl. Spectrosc. 32, 1 (1978). 25. C: T. Apel, T. M. Bieniewski, L. E. Cox, and D. W. Steinhaus, Paper Presented

at 20th Annual ORNL Gatlinburg Conference on Analytical Chemistry in Energy and Environmental Technology, Gatlinburg, TN (1976}.

26. R. W. Pongar and G. R. Thompson, Ja~rell-Ash Newsletter 1, 5 (1978). 27. C. E. Taylor, Environmental Protection Technical Series, EPA-600/2-77-113

(1977). 28. R. J. Ronan, Paper No. 361 presented at Pittsburgh Conference on Analytical

Chemistry and Applied Spectroscopy, Cleveland (1976).

Comparison of Interference Effects for Manganese in Constant Temperature vs Pulse-type Electrothermal Atomization

LYNN HAGEMAN, AHMED MUBARAK, and RAY WOODRIFF Montana State University, Bozeman, Montana 59717

The appl icat ion of e l ec tro thermal a tomiza t ion t echn iques for a tomic absorpt ion s p e c t r o s c o p y has increased m a n y fold in recent years . A l o n g wi th this w i d e s p r e a d appl icat ion, matr ix in ter ferences in pu l se - type e l ec trothermal a tomizers have been observed . Unt i l recent ly , v e r y f ew in ter ference s tud ies have i n v o l v e d cons tant t emperature e l ec tro thermal atomizers . In this paper a c o m p a r i s o n is made o f in ter ferences in pu l se - type vs cons tant t emperature a tmoz iers for var ious meta l ch lor ides on the Mn atomic absorpt ion s ignal . The resu l t s s h o w no interfer- e n c e s in the cons tant t emperature s y s t e m whi l e s igni f icant m a t r i x in ter ferences are observed in the pulse type unit. It is proposed that the observed d i f ferences are due m a i n l y to the rate at w h i c h vo lat i l i zed ana ly t e a t ta ins a tomizat ion tempera- ture and the res idence t ime in the hot env ironment .

Index Headings: E lec tro thermal atomizat ion; A t o m i c absorpt ion spec troscopy; Matr ix interferences , m a n g a n e s e .

226 Volume 33, Number 3, 1979

INTRODUCTION

In the past few years, matrix effects in electrothermal atomic absorption atomizers have been investigated by several authors. 1-5 Significant matrix interferences on manganese absorption have been reported with pulse- type flameless atomizers of the Massman design. 3 Mas- sart et al. reported interferences of various salts on manganese and copper atomic absorption using a Perkin- Elmer graphite furnace HGA 72. However, much work still remains to be done to explain the observations that have been made. Previous interference studies have in- volved very few observations under isothermal condi- tions. A study in this laboratory with the Varian Mini- Massman CRA confirms the reported matrix effects in

APPLIED SPECTROSCOPY

the pulse-type atomizers. The observations from the pulse-type atomizers are compared with those made with the same matrices in a constant t empera ture graphite e lect rothermal atomizer. Although considerable interfer- ences are observed in the pulse-type atomizer, similar interferences are not observed in the constant tempera- ture atomizer.

I. E X P E R I M E N T A L

A. I n s t r u m e n t a t i o n . Two different e lectrothermal atomizers were employed in collecting the data: a con- s tant t empera ture furnace (CTF) developed at Montana State Universi ty s-9 used in conjuction with a Varian AA- 6 spect rometer and a Varian carbon rod atomizer (CRA) used in conjunction with ei ther a Varian AA-5 or a Varian 1150. The Varian AA-5 spect rophotometer was equipped with an AA-6 conversion module, BC-6 simul- taneous background corrector, model 63 carbon rod at- omizer and a Beckman model 1005 strip chart recorder. A Varian AA-1150, also with simultaneous background corrector and a carbon rod atomizer, was used for confir- mat ion of the analyses. The Varian AA-6 spectrophotom- eter used for constant tempera ture measurements is also equipped with a BC-6 simultaneous background correc- tor and a He a th Servo-Recorder, model EUW-20A. In- s t rument conditions used are shown in Table I.

As a mat te r of convenience, a schematic drawing of the constant tempera ture furnace is shown in Fig. 1. The

TABLE I. Instrumental conditions. CRA Constant temp furnace

Spectral line (nm) 279.48 279.48 Spectral band pass (nm) 0.2 0.2 Mn lamp current (mA) 4 9 H2 lamp current (mA) 8 20 Inert gas flow (L/min) 6 N2 0.3 Ar CH4 flow during ashing and 0.5 ...

atomization (L/min) Drying 300 K" 28 s Drying is achieved Ashing 1100 K h 10 s under a heat lamp Atomizing 2900 K b 3 s before sample

enters; atomizer 2220°K; no ashing necessary

"Estimated maximum temperature obtained during dry cycle. h Maximum temperatures obtained during ashing and atomization

stages as measured by optical pyrometer.

45~4 : -- . \

FIG. 1. Constant temperature furnace {schematically). 1, heater tube; 2, side tube; 3, pedestal; 4, crucible; 5, heater tube holder; 6, thermal insulation; 7, argon inlet.

heater tube (1) is kept at high tempera ture continuously by resistance heating. Samples are pipeted into the sam- ple crucible (4) at room temperature . The sample-con- taining graphite crucible (4) is introduced into the fur- nace by means of the pedestal (3). Th e crucible is flushed with argon as it enters the furnace. The crucible is hea ted by radiation and conduction rapidly enough tha t a short evaporat ion t ime compared to the residence t ime is ob- tained. Th e pedestal makes a seal with the constriction in the side tube, and sample vapor diffuses through the end of the heater tube (if impermeable material is used for hea ter tubes and pedestals). Graphite felt insulation (6) reduces the radiation and conduction losses facilitat- ing the use of a relatively small power supply. 1°

For the purpose of integration of absorption signal for both atomizers, an integrator was constructed wi th an operational amplifer, voltage frequency converter, and digital counter. Tempera tu re measurements were made with a Pyro optical pyrometer . Exper imentat ion was carried out with various atomization tempera tures in an effort to minimize interference in the carbon rod atom- izer. Various ashing tempera tures and times were also used in an effort to eliminate the interferences.

B. R e a g e n t s . Trace metal s tandards were prepared by diluting appropriately a certified reference s tandard solution from Fisher Scientific Company. The concentra- t ion of the s tandard solution is 1 mg of Mn metal in 1 ml of dilute nitric acid. Th e interferent solutions containing 1000 ~g/ml of salt were prepared by dissolving 1 g of analytical reagent grade salt in 1000 ml of deionized water. These solutions were diluted to obtain other con- centrat ions used. Th e solutions were analyzed within a few hours after their preparat ion and no preservatives were added.

C. P r o c e d u r e . In order to obtain the Mn absorbance with no interferent present, 2.5 ftl of 0.1 ftg/ml of Mn solution was pipeted into the CRA tube as described by the manufac turer with an adjustable 10 ~l Unimetrics Teflon-t ipped syringe. The interference absorption val- ues are obtained by pipeting the same amount of a solution containing both the analyte and the specified amount of interfering salt. Th e same solutions and tech- niques were used to load the sample crucible of the CFT as were used for the CRA.

II. R E S U L T S A N D D I S C U S S I O N

The results obtained with the 100 and 1000 ~g/ml amounts of all added salts investigated are shown in Table II. It should be noted tha t these high amounts of interfering salts are not unrealistic. Soil solution obtained from semiarid soils and dissolved biological materials are two examples of low analyte concentrat ion and very high ionic strength. The interferences are expressed as percent change of the absorption signal caused by added inter- ferents. Since the precision of e lect rothermal atomization atomic absorption spectroscopy is about 5 to 8%, 3 ob- served interferences of less than 10% have been consid- ered insignificant in Table II.

A more detailed investigation was carried out with metal chlorides because they caused significant interfer- ence. Th e interference is plot ted as a function of the

APPLIED SPECTROSCOPY 227

T A B L E II. I n t e r f e r e n c e s o b s e r v e d i n e l e c t r o t h e r m a l a t o m i z a - t i o n A A S d e t e r m i n a t i o n o f M n ( e x p r e s s e d i n %).

Salt

Varian model 63 C R A ef- fect on 0.1 p p m M n of

Cons t an t t emp furnace el- feet on 0.1 p p m Mn of

100 ~g /ml 1000 tzg/ral t00 ~g /ml 1000 ~g /ml sal t sa l t sal t sal t

CaC12 - 9 0 - 1 0 0 ±<10 +_<10 Ca(NO3)2 ±<10 ± < 1 0 +_<10 _+<10 Caa(PO~)2 ±<10 - 1 0 ± < 1 0 ± < 1 0 MgC12 - 1 0 - 7 6 ± < 1 0 ±<10 Mg(NO3)2 ±<10 ± < 1 0 ±<10 ± < 1 0 MgSO4 ±<10 ± < 1 0 ±<10 ±<10 NaC1 - 1 0 - 5 0 ± < 1 0 ±<10 NaNO3 ±<10 +10 ±<10 ±<10 Na2SO4 ±<10 ± < 1 0 ±<10 ±<10 Na2HPO4 ±<10 ± < 1 0 ± < 1 0 ±<10 KC1 ±<10 ± < 1 0 ± < 1 0 ±<10 KNO3 ±<10 ±<10 ± < 1 0 ± < 1 0 K2SO4 ±<10 ± < 1 0 ±<10 ± < 1 0 K2HPO, ±<10 ± < 1 0 ± < 1 0 ± < 1 0 ZnC12 - 1 5 - 7 0 ± < 1 0 _+<10

z

I I I

0 . 8 -

JJ /,, O, 4 11 / / I/

0.2 I I1̀11 /

1300 1800 2300 2800 3300

(TEMPERATURE IlK

FIG. 2. Absorpt ion vs t empera tu re for 2.5 ng of M n in Varian carbon rod atomizer (A- - -A, peak he igh t - - -, peak area) and 5 ng of M n in the cons tan t t empera tu re furnace ( O , peak height; - - peak area).

I l I I I +iO - -

o~'~ 0 ~ ...... ~m .................. o~ -e~ 2 m

ca <

z

-50 z <

- t o o _ _ - , I I ,,k I00 250 500 750 I0~0

~lglml SALTS

FIG. 3. Percent interference of various meta l chlorides on the M n absolute signal as a funct ion of concentrat ion in the model 63 CRA ( ) and cons tan t t empera tu re furnace ( - - - ) O, CaCl2; A, MgC12; BB, NaC1.

concentration of matrix elements. The effects of CaC12, MgC12, and NaCl on the manganese atomic absorption signal are illustrated in Fig. 3.

228 Volume 33, Number 3, 1979

The efforts made to decrease interference in the CRA revealed that no significant interferences changes were noted with variation in atomization temperature. A tem- perature of about 1670 K was necessary to atomize the Mn solutions completely (Fig. 2) in the CRA; therefore, atomization temperatures studied ranged from about 1700 to 3000 K. Also, in the effort to minimize interfer- ence of chlorides on manganese in the CRA it was found that by increasing the ashing temperature the magnitude of the interference was decreased. At a voltage of 4.5 for 10 s the magnitude of the observed interferences was about 20% greater than at 6.5 V for 10 s. At a voltage of 8.5 for 15 s the magnitude of the interference was about 20% less than at 6.5 V for 10 s, but at the 8.5 V ashing there was about a 20% loss of the Mn. In order to achieve the least amount of interference without significant loss of Mn the 6.5 V for 10 s ashing was used to collect the data in Table II and Fig. 3. According to the Varian manual this setting should provide a maximum temper- ature of about 1000 K but an optical pyrometer estimate indicated that a maximum temperature of 1100 ± 50°K was obtained. It should be noted that both the ash and atomization temperatures discussed here are the maxi- mum attained during that cycle and the carbon rod is at that temperature for only a very small part of the cycle.

The absorption vs temperature curves for Mn in the constant temperature furnace is also shown in Fig. 2. All data were obtained near the temperature of maximum sensitivity (above 2220 K). The conditions under which the carbon rod atomizer and constant temperature fur- nace were operated are shown in Table I. Absorbances of 0.30 and 0.60 were obtained for 2.5/B of 0.1 #g/ml of Mn in the constant temperature furnace and the CRA, re- spectively. The background correction was found to be satisfactory for both electrothermal atomizers since no signal was observed for solution of interfering elements without analyte.

The large interferences observed in the case of the pulse type carbon rod atomizer for the chlorides agree with those previously reported for the pulse type atom- izer of the Massman design, 3 but a significant difference is noted in the case of the constant temperature furnace. The most pronounced example is that due to the CaC12 interference. The Mn atomic absorption signal is vir- tually eliminated in the pulse type atomization, at 100 ~g/ml of CaC12 or higher, but in the constant temperature furnace no interference on the Mn absorption signal is noted. The results plotted in Fig. 3 are based on peak height absorption values. The recorded peak area results have not been included here because of the similarity to the peak height data.

Many authors have emphasized that interferences in pulse-type atomizers are often complex?' u It is, there- fore, necessary when studying interferences to consider the effect of the interferent at various concentrations. It is also necessary to consider the effect of both the cation and the anion, as is demonstrated by the effects of the various chloride compounds on the Mn atomic absorp- tion signal. Comparison of the percent suppressions for constant chloride level shows that the cation with which the chloride is associated has a significant effect on the extent of interference (Table III). The data suggest that the interference taking place in the pulse-type atomizers

T A B L E III . Compar i son of in ter ference magn i tude in pulse type a tomizer due to var ious chloride sa l ts w i t h propert ies of these salts .

Chloride Interference M--C1 bond dis- from 750 ng sociation energy

C1 (%) (Kcal/molF

Boiling point (°K)"

1780 {sublimes) 1686 1685

>1900 1000

K 0 101.3 Na -30 97.5 Mg -45 89.0 Ca -100 81.0 Zn -50 50.0

" Ref. 16.

is dependent upon the bond dissociation energy and the boiling point of the interfering chloride. The magnitude of interference correlates with bond dissociation energy for high boiling interferents; the boiling point becomes the important factor with low boiling interferents such as ZnC12. A lower bond dissociation energy should cause more chloride to be available for interference, except in the case of ZnC12 where the low boiling point would cause much of the chloride to boil off and escape before the Mn vaporizes. Therefore, both the bond dissociation energy and the boiling point appear to be important in interference mechanisms.

There have been many studies on the mechanism of atomization in carbon furnaces. 12-1~ Both the thermody- namic and kinetic approaches have considered the im- portance of the hot carbon in the atomization process based on reaction (1).

MO(s) + C(s) --~ CO(g) + M(g) (1)

Aggett and Sprott 11 concluded that Reaction 1 is not the case for Mn, based upon free energy considerations and comparison of appearance temperatures on carbon and tantalum strips. Since their appearance temperature of about 1470 K agrees well with the appearance tempera- tures seen in this study (Fig. 2), it is believed that the carbon does not enter into the Mn atomization process and the chloride interference is probably involved in the vapor phase. For the pulse-type atomization it is obvious that some chloride compounds interfere in such a manner as to reduce the absorption signal. There are several processes by which this type interference could occur as has been reported previously by Woodriff. 6

It appears that the interferences seen in the pulse furnaces are due to the formation of intermediate chlo- rides. This conclusion was reached after considering the relationship of bond dissociation energies and boiling points to interference and the fact that no interference is present in the constant temperature furnace. These chlo- rides escape before they have time to decompose into atoms. In the case of the CRA the small size could easily allow the manganese chloride to escape out of the ends before sufficient temperature has been attained for atom- ization. Similarly, in the Massman furnace the chlorides could escape out of the exit hole or condense in the cooler portions of the tube before decomposition. However, in the large volume constant temperature furnace two im- portant factors are responsible for the minimization of interference. First, in CTF, as the sample approaches the hot carbon tube and attains vaporization temperature, the vaporized molecules are immediately in a very high temperature environment in the optical path. This is in

T A B L E IV. Mn interference correct ions for mode l 63 CRA.

Interfering salts added

No inter- 300/ag/ 600/zg/ 1000 #g/ ml

ferent ml CaC12 MgC12 ml NaC1

No correction 0.75 0.05 0.18 0.40 AgNO3 correction 0.75 0.78 0.71 0.76 HNO~ correction 0.75 0.70 0.72 0.73

" A l l values are peak height absorbance values for 5/~1 of 0.1 #g/ml of Mn.

contrast to the pulse-type atomizers where a molecule such as manganese chloride could possibly start to vola- tilize very soon after the ashing stage. During the 1 to 3 s necessary for the atomizer to reach atomization tem- perature volatile molecules containing the analyte could have already been expelled from the atomizer. Second, the residence time is much longer in the constant tem- perature furnace. In the constant temperature furnace the sample must diffuse along the length or through the walls of a 30 cm carbon tube before leaving the optical path as opposed to the pulse-type atomizers from which the sample can escape readily after volatilization.

A comparative study such as this not only suggests the probable mechanism of the CRA interference, but the absence of matrix interferences in the constant temper- ature furnace eliminates certain mechanisms that would be common to both atomizers. Further work involving other elements and possible interferent mechanisms is in progress and will be reported in forthcoming publications.

It should also be noted that for analytical purposes there are methods of eliminating these interferences ob- served in the carbon rod atomizer. Removal of the chlo- rides by precipitation with silver and filtering of the precipitate works quite well. A second method involves incorporation of a strong acid (~20%) such as nitric acid into the sample plus metal chloride solution. The effect of the acid is to expel hydrochloric acid during the drying stage. The results using these methods on some known interfering salt concentrations are shown in Table IV.

In conclusion, matrix interference occurs when the pulse-type atomizer is used and does not take place with the constant temperature furnace. In the constant tem- perature furnace the sample cannot escape without going through the furnace which is above the dissociation temperature of the chloride. The interference results reported in this paper, it is hoped, provide a better understanding of the mechanisms by which matrix inter- ferences occur in pulse type atomizers.

ACKNOWLEDGMENTS

The Varian AA-6 used in this research was provided through the courtesy of the MSU Fisheries Bioassay Laboratory, NSF Grant Number OCE74-24317. The Varian AA-5 was provided by the Analytical Laboratory of the Montana State Department of Agriculture. We wish to thank the National Science Foundation for their support which enabled construction of constant temperature furnaces and equipment.

1. L. Hageman, L. Torma, and B. Ginther, J. Assoc. Offic. Anal. Chem. 58, 990 (1975).

2. R. Stone, Ph.D. Thesis, Montana State University (1974). 3. J. Smeyers-Verbeke, Y. Michotte, P. Vanden Winkel, and D. L. Massart,

Anal. Chem. 48, 125 (1976). 4. E. Czobik and J. Matousek, Anal. Chem. 50, 2 (1978). 5. J. Smeyers-Verbeke, Y. Michotte, and D. L. Massart, Anal. Chem. 50, 10

(1978). 6. R. Woodriff, Appl. Spectrosc. 98, 413 (1974).

APPLIED SPECTROSCOPY 2 2 9

7. M. Marinkovic and R. Woodriff, Appl. Spectrosc. 30, 458 (1976). 8. R. Woodriff and J. Lech, Anal. Chem. 44, 1323 (1972L 9. R. Woodriff, Appl. Spectrosc. 22, 207 (1968).

10. R. Woodriff, M. Marinkovic, R. A. Howald, and I. Eliezer, Anal. Chem. 49, 20O8 (1977}.

11. J. P. Matousek, Am. Lab. 3 (6), 45 {1971).

12. J. Aggett and A. J. Sprott, Anal. Chim. Acta 72, 49 (1974). 13. W. C. Campbell and J. M. Ottaway, Talanta 21, 837 (1974). 14. W. Frech and A. Cedergren, Anal. Chim. Acta 82, 83 (1976}. 15. C. W. Fuller, Analyst 99, 739 (1974). 16. Handbook of Chemistry and Physics, 55th ed., R. C. Weast, Ed. (CRC Press,

Cleveland, 1975).

Investigation into the Operating Characteristics of a "Microarc" Atmospheric-pressure Glow Discharge*

R. I. BYSTROFF, L. R. LAYMAN, t and G. M. HIEFTJE$ Department of Chemistry, Lawrence Livermore Laboratory, Livermore, California 94550

Details of the processes occurring during sample atomization from a "microarc" discharge have been studied photometrically, by use of high-speed color cinematography and through cur- rent-voltage waveforms. The microarc studied here is an at- mospheric-pressure inert-gas glow discharge supported be- tween 0.25 mm diameter tungsten wires; quiescent argonol% H2 provides reactive-sputtering conditions and improved behavior in the presence of oxygen impurities. Excitation temperatures of ca. 5000°K are measured for the argon glow. Samples of Na, A1, and Sr illustrate the influence of volatile, refractory, insu- lating, and electron-emitting sample properties on the tem- poral-spatial-electrical behavior of the discharge. The step-by- step events occurring in the discharge are described qualita- t ively and a variety of processes are invoked to explain sample volatilization, including sputtering, chemical reactions, and purely thermal effects. In the first stages of the discharge, instabilities are related to the placement and insulating char- acter of deposits. With heating, electron emission becomes im- portant in directing the discharge to or away from the sample; abnormal glow wandering and glow-to-arc transitions can en- sue. Improved stability is achieved by uniformly depositing multi-element samples along the electrode, which localizes the initial discharge and promotes ablative cooling of the sample and electrode. Index Headings: Microarc; Atomization; Glow discharge; Multi- element analysis.

INTRODUCTION

One of the most active current endeavors in analytical chemistry is the search for new methods of multi-element analysis. Improvements in many existing methods have been proposed and new ones developed. Perhaps the most significant of these advances has been the recent emergence of the inductively coupled plasma as a tool

Received 21 J u n e 1978. * Work performed under the auspices of the U.S. D e p a r t m e n t of Energy

by the Lawrence Livermore Labora tory under Contrac t W-7405- ENG-48.

t P e r m a n e n t address: D e p a r t m e n t of Chemist ry , Pacific Lu the ran Uni- versi ty Tacoma, WA 98447. P e r m a n e n t address: D e p a r t m e n t of Chemis t ry , Indiana Universi ty, Bloomington, IN 47401.

for multi-element determinations. 1 However, even this new, powerful source has been found to have limitations (e.g., high cost, bulkiness) and the search for alternative techniques continues.

Ideally, whatever approach is finally adopted in a multi-element scheme should be amenable to use with more than one kind of detection technique. That is, it should be possible to employ for atom detection an atomic fluorescence method, atomic emission, or even ion detection, 2 the choice being based upon the required sensitivity, selectivity, or operational convenience. The key to such flexibility is to develop a device for generating atoms which is useful under a variety of conditions and with any of the detection methods. One promising such atomizer is the so-called microarc . 3

The microarc, as originally configured, is essentially an atmospheric-pressure glow discharge operating in an ar- gon environment. The discharge is characterized by rel- atively high voltages, controlled currents, and a sputter- ing kind of atom formation process. Coupled with a microwave plasma excitation source, the microarc has been shown to yield high sensitivity, low susceptibility to matrix interferences, reproducibility of atom formation, low sample volume requirements, and the capability of operation at atmospheric pressure. However, this new discharge offers other potential advantages as well. The microarc could serve as an excitation source as well as sample atolnizer to possibly provide many of the advan- tages it previously exhibited but in a simpler configura- tion. Finally, the device is readily controlled and should be easily automated.

To realize the potential of this new kind of discharge, it must be studied and understood more completely. In the present investigation, a discharge similar to the mi- croarc was constructed and placed under careful control. A reproducible electrode arrangement, employing tung- sten wires, was configured in a chamber which enabled a controlled atmosphere (primarily argon) to be utilized at atmospheric and slightly lower pressures. With this new system, the character and stability of the resulting discharge, electrode erosion behavior, sample volatiliza- tion, and sample atomization were all studied as functions

2,30 Volume 33, Number 3, 1979 APPLIED SPECTROSCOPY