ent instrumental conditions as information essential for establishing thresholds.

For all three measures of similarity, the better repeat- ability of RTF is consistently shown.

Of all three measures considered, the correlation coef- ficient is to be preferred because of its small variability; in addition, it proved to be at least sensitive to raw or normalized spectral comparisons.

The superiority of LTL technique, was indicated by its more accurate discriminating capability as exhibited by properly differentiating between m-xylene and o-xylene.

Based on the findings of this study, corrected reference spectra of selected hazardous chemicals have been incor- porated into our computer library of LTL and RTF spectra. As the number increases, it becomes more im- portant to have a library search based upon a suitably chosen feature set. Unlike other techniques (such as IR) where information on peak-position alone is sufficient for identification, for LTL and RTF where many of the spectra are broad, peak location may not be a reliable identification.

The next phase of investigation of fluorescence/lumi- nescence spectra will involve a comparison of several feature sets--from the intuitively appealing to the math- ematically abstract--in order to evaluate their conveni- ence and reliability.

ACKNOWLEDGMENTS

The authors gratefully acknowledge E. Hunt, A. Hanna, and J. Gilbert for their assistance with the many facets of computer interfacing and data processing.

1. P. C. Jurs and T. L. Isenhour, ChemicalApplications of Pattern Recognition (Wiley-Interscience, New York, 1975).

2. M. F. Delaney and P. C. Uden, Anal. Chem. 51, 1242 (1979). 3. S. L. Morgan and C. A, Jacques, Anal. Chem. 53, (1981). 4. T. C. Miller and L. R. Faulkner, Anal. Chem. 48, 2083 (1976). 5. K. W. K. Yim, T. C. Miller, andL. R. Faulkner, Anal. Chem. 49, 2069, (1977). 6. J. S. Lyons, P. J. Hardesty, C. S. Baer, and L. R. Faulkner, in Fluorescence

Spectroscopy, E. L. Wehry, Ed. (Plenum, New York, 1981), vol. 3. 7. J. T. Brownrigg, D. A. Busch, and L. P. Giering, A Luminescence Survey of

Hazardous Materials. Baird Corporation for U.S. Coast Guard Research and Development Center, Rept. CG-D-53-79 (May 1979}. (Available to the U.S. public through the National Technical Information Services, Springfield, VA ADA #073-828).

8. C. A. Parker, Photoluminescence of Solutions (Elsevier, New York, 1968). 9. R. S. Becker, Theory and Interpretation of Fluorescence and Phosphores-

cence (Wiley-Iuterscience, New York, 1969). 10. S. H. Fortier and D. Eastwood, Anal. Chem. 50, 334 (1978). 11. D. Eastwood, S. H. Fortier, and M. S. Hendrick, Am. Lab. March (1978). 12. J. S. Bendat and A. G. Piersol, Random Data: Analysis and Measurement

Procedures (Wiley-Interscience, New York, 1971) pp. 23, 77. 13. R. N. Bracewell, The Fourier Transform and Its Applications, (McGraw-Hill

New York, 1978) pp. 113-115. 14. M. K. Anderberg, Cluster Analysis for Application (Academic Press, New

York, 1973) pp, 70-130. 15. J. T. Tou and R. C. Gonzales, Pattern Recognition Principals (Addison-

Wesley, Reading, MA, 1974) pp. 87-86. 16. B. R. Kowalski, Chemometries: Theory and Application (American Chemical

Society, Washington, DC, 1977) pp. 34-35.

Effects of Aerosol Introduction on the Analyt ica l Uti l i ty of dc P lasma Jets

GEOFFREY N. COLEMAN* and ALEXANDER M. ALLEN t Department of Chemistry, University of Georgia, Athens, Georgia 30602

Large aeroso l in troduct ion tubes are s h o w n to affect a d v e r s e l y ana ly t i ca l ca l ibrat ion curves and apparent ion- to -atom inten- s i ty rat ios in two-e l ec t rode dc p lasma jets . With large ch imneys , the aeroso l that skirts the p l a s m a resul t s in subs tant ia l ground state a tom concentrat ions a long the opt ical ax i s b e t w e e n the source and detector. The use o f exc i ted ion emis s ion l ines fo r analys i s c i rcumvents mos t o f the adverse effects .

Index Headings: E m i s s i o n spec troscopy; Sample introduct ions .

INTRODUCTION

Direct current plasma (DCP) jets have proven to be simple, cost-effective excitation sources for multielement emission spectrochemical analysis. 1-5 Several designs have been characterized but all show disappointing de- tection capabilities when compared with the inductively coupled plasma (ICP)) -6 High background from the ar- gon recombination continuum and tungsten electrodes,

Received 21 August 1981; 19 October 1981. * Author to whom correspondence should be addressed. 1" Present address: Mobil Chemical Co., P.O. Box 240, Edison, NJ 08817.

plasma instability (flicker), and poor sample introduction have been shown to be significant limitations with DCP's. 1-3, 5, 7

Instrumental background correction has been used to improve detection limits and to extend the dynamic range. 1 Johnson, Taylor, and Skogerboe 4 characterized a commercial three-electrode DCP (3-DCP) that utilizes two carbon anodes and a tungsten cathode; they found improved source stability and reduced background com- pared with earlier two-electrode designs (2-DCP). Urasa s has shown that a reduction of the internal diameter of the aerosol delivery tube (ADT) from 8 to 6 mm results in increased emission intensity. However, the plasma could not be sustained with smaller ADT's due to the gas flow required for efficient performance of the com- mercial, concentric nebulizer. In a recent publication we showed that reducing the angle between the electrodes of a commercial 2-DCP yields a more stable plasma and permits the use of ADT's as small as 1.2 mm when a low- flow (cross-flow) nebulizer is used. We suggested that the reduced noise resulted from less turbulent mixing of the gas flows near the optimum viewing zone. However, the

116 Volume 36, Number 2, 1982 APPLIED SPECTROSCOPY

improvement in detection capabilities substantially ex- ceeded what could be expected from reduced noise, thus we further suggested that the more colinear gas flows and smaller ADT's also improve aerosol entrainment.

In the present study we have systematically character- ized sample introduction in a 2-DCP and show the limi- tations of present designs. Spatial profiles of both free atoms and analyte emission have been used to study aerosol introduction.

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

A . A p p a r a t u s . Background-corrected emission mea- surements were made using the modified two-electrode dc plasma jet and optical system described previously. ° For absorption measurements, a hollow cathode lamp (HCL) was mounted on the optical axis and a chopper (Princeton Applied Research, model 222) was interposed between the HCL and the DCP. Since phase-sensitive detection (Princeton Applied Research, model 220) re-

T AEROSOL DELIVERY

STOPPERED FLASK

/ U U ~ R GAS

HEATING,/~2¢ "~) TEM BATH CONTRDL

FIG. 1. Mercury vapor generator.

+N I I I I I r

+3

+i ~E

~ O

~ -1 -

~ -2 -

- 3 -

-4

-5 [

-2 -1 +I +2 +3

DISTANCE FROM PLASMA CENTER, ~IM

FIG. 2. Spatial distribution of mercury absorption (253.6 nm) with 2.4- mm aerosol delivery tube and 75 ° electrode orientation. Gas flows only.

sponds only to the ac component, relative absorption data were obtained. A mercury vapor cell (Fig. 1) was used in gas flow studies and to deliver free atoms directly to the plasma. The concentration of mercury in the argon carrier was held constant through careful control of the bath temperature and the gas flow rate. A three-way valve permitted selection of Hg/Ar or Ar blank. In all cases the total carrier gas flow was held constant at 0.62 L/min at 0.38 MPa (55 psi). Quartz ADT's of 4.0, 2.4, and 1.2 mm i.d. were used as described below.

B. Reagents . Ultrapure nitric acid obtained through sub-boiling distillation s was used to dissolve high purity metals (Spex). Stock solutions were prepared with deion- ized (Barnstead) distilled water and brought to a final volume that was 3M in nitric acid. Standards for analyt- ical curves were prepared by serial dilution. Triply dis- tilled mercury was obtained from Bethlehem Instrument Co.

C. Procedures . For spatial profiles, the entrance slit was vignetted with a 0.2-ram aperture and a 1:1 image of the DCP was moved across the aperture in 0.2 mm increments by means of an xy stage. Measurements of relative free atom concentrations along the normal opti- cal axis, in front of, and behind the normal observation zone, were obtained by rotating the DCP 90 ° on the xy stage.

II. RESULTS AND D I S C U S S I O N

It has been suggested that a large portion of the aerosol leaving the ADT does not reach the high temperature core of the DCP. 1' 2, 7 One explanation is the existence of a plasma skin of thermal, electrical, and magnetic barriers which prevents low momentum droplets from penetrat- ing the core. 7' lo Additionally, the large diameter ADT's (greater than 4 ram) produce an aerosol stream which is

I --1 l l I - - - - - I + 4 -

+3- /

+2 --

° / ~ 0 -

-3

-4

_ I I I I J . _ _ _ l -2 -i 0 +i +2 +3

DISTANCE FROM PLASMA CENTER, MM

FIG. 3. Spatial distribution of mercury absorption (253.6 nm) with 2.4- m m aerosol delivery tube and 45 ° electrode orientation. Gas flows only.

APPLIED SPECTROSCOPY 117

z

-2 -1 0 +i +2 +3 DISTANCE FROM PLASMA CENTER, MM

Fzo. 4. Spatial distribution of mercury absorption (253.6 nm) with 1.2- mm aerosol delivery tube and 45 o electrode orientation. Gas flows only.

much larger than the plasma cross section; l'~'s that a large portion skirts the plasma is visually apparent. Both the skin and the skirting contribute to less than optimum analyte introduction and to correspondingly disappoint- ing detection capabilities.

In an earlier report we suggested that reducing the angle between the electrodes to provide more colinear gas flows should improve aerosol entrainment as well as permitting the use of smaller ADT's and reducing mixing turbulenceP The use of a mercury vapor apparatus (Fig. 1) permits the characterization of gas flow effects whether or not a plasma is initiated. Moreover, it pre- cludes the effects of the preexcitation processes (desol- vation, vaporization, and dissocation) which have been shown to vary significantly with the wide range of droplet sizes produced by conventional nebulizers. 11-16 Figs. 2 and 3 clearly show that, in the absence of the discharge, a reduction of the angle between the electrodes from 75 ° to 45 ° yields improved aerosol entrainment; the corridor is narrower and extends further up into the electrode gas flows. Fig. 4 shows that even better entrainment is achieved with a 1.2-mm ADT compared with the 2.4-mm ADT (Figs. 2 and 3). With the plasma sustained and a 2.4-ram ADT, it is evident that more analyte penetrates nearer the plasma core at 45 ° (Fig. 5) than at 75 ° (Fig. 6) where the skin effect predominates. Measurements of ground state species near the high-temperature plasma core could not be made reliably and preclude additional inferences. It should be noted that the low aerosol flow rates required for use with the small diameter ADT's (2.4 and 1.2 ram), to prevent disruption of the plasma, cannot be achieved with conventional concentric nebulizers un- less flow splitting is used. The extent of the effect of the loss of aerosol under these conditions has not been re- ported.

It is important to identify any anomalies in the per-

118 Volume 36, Number 2, 1982

-3

[ I I I

25

3

2

1 j /

J 0

-1

I -2 ~--

I J

25

~ 150~/~ ?~---

-3 ~ 7

-4

-5 I _ . L _ I I I _ _ _ [ -2 -i O +i +2 +3

DISTANCE FROM PLASMA CENTER, MM

FIo. 5. Spatial distribution of mercury absorption (253.6 nm) with 2.4- mm aerosol delivery tube and 45 ° electrode orientation. Plasma sus- tained.

I I I I ..... ]-----

+3

+1 lO

2

~ - 2 -

-4

-2

T - -

J_ _ ] ] __1 ........ ] .]_.

-I 0 +1 +2 +3 DISTANCE FROM PLASMA CENTER, MM

+4 I I I I I I

+2

. . . . . . . I f

FIG. 6. Spatial distribution of mercury absorption (253.6 nm) with 2.4- mm aerosol delivery tube and 75 ° electrode orientation. Plasma sus- tained.

formance of the DCP which may result from the skirting of the plasma body by the aerosol. In order to determine relative concentrations of ground state atoms between the optimum viewing zone and the monochromator, the source was rotated 90 ° on its vertical axis. A horizontal scan at the height of the optimum viewing zone yields a measure of the relative ground state population along

the normal optical axis from the front to the back of the DCP. Fig. 7 shows that for chromium, with 4.0 mm, and even 2.4 mm ADT's, the preexcitation processes occur to a significant extent for aqueous aerosols and substantial ground state populations exist along the normal optical axis as far away as 6 mm from plasma center• As the ADT is reduced from 4.0 to 2.4 ram, and finally 1.2 ram, the significance of the skirting diminishes. Similar results were obtained for copper• Moreover, we have observed nearly identical phenomena with a 3-DCP. In this case, the construction of the plasma jet precludes a full hori- zontal scan thus the specifics are not presented here. However, since the mode of sample introduction in the 3-DCP is identical to that of the 2-DCP, the similarity of the results is to be expected.

The effect of the ground state populations on analytical calibration curves is shown in Fig. 8. Reducing the ADT from 4.0 to 2.4 mm and 1.2 mm increases the upper limit of linearity from 700 to 1200 ppm and 1500 ppm Cu, respectively• Moreover, the smaller ADT's achieve the log-log slope of unity predicted by theory, whereas the 4.0-ram ADT yields a slope somewhat less. In contrast, using the ion line for copper (224.7 nm) and a 4.0-mm ADT, the analytical calibration curve exhibited a log-log slope of unity over the range 0.01 to 50 000 ppm. That self-absorption by ground state ions is insignificant is not surprising since the population of ground state ions out- side the high temperature core should be minimal•

One other aspect of Fig. 8 which merits discussion is the fact that the curves do not roll over to a slope of lh as predicted by theory. ~7' is Rather, they roll to a slope of near zero which is what theory predicts for continuum- excited atomic fluorescence• This suggests that the in- tense argon recombination continuum may enhance the populations of excited neutral atom species• A further characterization of this phenomenon is presently under- way.

160

120

I I I I

. °

~ ~ °"

°

. °

I I I -8 -6 -4

I I l

f\

I

I , I . /

/ !

o" • *~

I I ___L_ -2 0 +2

DISTANCE FROM PLASMA CENTER, MM

I _ _ £ +4 ÷6

Fro. 7. Relative absorption by Cr I (357.9 nm, 100 ppm) along the normal optical axis at the normal viewing height showing the effect of aerosol skirting with 4 mm ( . . . ) , 2.4 mm (---), and 1.2 mm ( ) aerosol delivery tubes. The skew is a result of difficulty in alignment.

The effects of concommitant easily ionized elements (EIE's) on excitation conditions in plasmas have been the subject of many reports• The majority of workers dealing with DCP's have reported EIE-induced enhance- ments of both neutral atom and ion lines similar to those reported for the inductively coupled plasma, G' 19 whereas others have reported varying degrees of enhancement, and even suppression. 2°-22 We find that the addition of an EIE enhances both neutral atoms and ions (Fig. 9) but that the degree of enhancement varies with the ADT size as well as the concentration of the EIE. Fig. 10 shows that with the 1.2-mm ADT, the ratio Ca II/Ca I increases smoothly but with the larger ADT's a threshold level exists. Part of the explanation lies in the previous discus- sion: the ground state population found along the optical

1000

i0o

I I I I i

,S,% .-~J ~

, /~ B~''(°/I°I°

I I I I I 1 i0 i00 lO00 i0000

LOG CONCENTRATION (MS/L) Fro. 8. Analytical calibration curves for Cu I (324.8 nm) obtained with different sized aerosol delivery tubes. (C)) 4 mm, rolloff at 700 ppm; (A) 2.4 mm, rolloff at 1200 ppm; (O) 1.2 mm, rolloff at 1500 ppm.

4,00

3,00

2,00

1,00

3,00

2,00

1,00

I I I I

4.0 MM ~ ~ ADT . o

2,0 MM ADT _ ~ a

2,003'00 ~ I 1,0 MM ADT ~ ~

1,oo o e ~ - - ~ ~ - - ~ - - - - - °

I I I I

i 10 100 1000

LOG LITHIUM CONCENTRATION (MG/L)

FIG. 9. Relative emission intensities for Ca I (422.6 nm, (Q)) and Ca II (393.3 nm, (A)) as a function of the concentration of an easily ionized element (Li) with different aerosol delivery tubes. To facilitate com- parisons, the signal from Ca I and Ca II in the absence of Li was normalized to unity; the calcium concentration was 100 ppm.

APPLIED SPECTROSCOPY 119

I

I, 50

1.00

I I F

o

I I I i0 100 1000

LOG LITHIUM CONCENTRATION (MG/L)

Fro. 10. Ratio of Ca II (393.3 nm) to Ca I (422.6 nm) emission intensity as a function of the concentration of an easily ionizable element (Li) with different aerosol delivery tubes: 4 mm (O), 2.4 m m (A) and 1.2 mm (©). To facilitate comparisons, the ratio was normalized to unity in the absence of any Li; it is ordinarily approximately 300.

axis with larger ADT's reduces the apparent Ca I inten- sity and makes the ratio appear larger than it really is. The remainder of the explanation lies in the elucidation of the excitation mechanism; several studies are presently underway. In any case, it is clear that conflicting reports regarding the effects of EIE's in DCP's may be resolved if aerosol introduction effects can be accounted for.

III. CONCLUSIONS

Aerosol introduction (as opposed to generation and transport) plays a significant role in analyses by DCP emission spectrometry. The ground state populations due to the aerosol skirting the plasma body when large ADT's are used alter both the slope and the upper limit of linearity of analytical calibration curves. We have shown that better aerosol entrainment realized with a 45 ° angle between the electrodes and a 1.2-mm ADT is a major source of the improved detection capabilities reported previously ~ although in many cases the addi- tional benefits of the 1.2-mm ADT vs the 2.4-mm ADT may not be worth the attendent alignment problems. A

primary limitation is the availability of good, low gas flow nebulizers necessary when small ADT's are used. Be- cause the mode of sample introduction is identical for both the 2-DCP and the 3-DCP, similar results may be expected for the latter. Preliminary results have con- firmed our expectations, but the construction of the commercial 3-DCP has precluded a definitive study.

Many of the adverse effects of aerosol introduction can be surmounted if ion lines rather than neutral atom lines are used for analysis; we find that if the principal ion line is observed, it is usually more intense than the corre- sponding neutral atom. 2~ Even though the ion line is still affected by concommitant EIE's, standard additions and swamping methods should be successful.

ACKNOWLEDGMENT

The loan of certain equipment by SpectraMetrics, Inc. is greatly appreciated.

1. R. K. Skogerboe, I. T. Urasa, and G. N. Coleman, Appl. Spectrosc. 30, 500 (1976).

2. C. D. Kiers and T. J. Vickers, Appl. Spectrosc. 31, 273 (1977). 3. R. K. Skogerboe and I. T. Urasa, Appl. Spectrosc. 32, 527 (1978). 4. G. W. Johnson, H. E. Taylor, and R. K. Skogerboe, Spectrochim. Acta 34B,

197 (1979). 5. G. N. Coleman, W. P. Braun, and A. M. Allen, Appl. Spectrosc. 34, 24 (1980). 6. R. M. Barnes, Crit. Rev. Anal. Chem. 7,203 (1978). 7. P. Merchant and C. Veillon, Anal. Chim. Acta 70, 17 (1974). 8. I. T. Urasa, Ph.D. dissertation, Colorado State University, Fort Collins,

Colorado, 1977. 9. K. D. Burrhus and S. R. Hart, Anal. Chem. 44, 432 (1972).

10. I. Rief, V. A. Fassel, and R. N. Knisely, Spectrochim. Acta 23B, 105 (1973). 11. R. Herrmann, C. Th. J. Alkemade, Chemical Analysis by Flame Photometry,

(Wfley-Interscience, New York, 1963). 12. J. A. Dean and W. J. Carnes, Anal. Chem. 34, 192 (1962). 13. K. Fujiwana, H. Haraguchi, and K. Fuwa, Anal. Chem. 47, 743 (1975). 14. N. C. Clampitt and G. M. Hieftje, Anal. Chem. 46, 282 (1974). 15. R. K. Skogerboe and K. W. Olsen, Appl. Spectrosc. 32, 181 (1978). 16. B. D. Bleasden, E. P. Wittig, and G. M. Hieftje, Spectrochim. Acta 36B, 205

(1981). 17. P. J. T. Zeegers, R. Smith, and J. D. Winefordner, Anal. Chem. 40 (13), 26A

(1968). 18. J. D. Winefordner, V. Svoboda, and L. J. Cline-Love, Crit. Rev. Anal. Chem.

1, 233 (1970). 19. G. W. Johnson, Ph.D. dissertation, Colorado State University, Fort Collins,

Colorado, 1979. 20. D. W. Golightly and J. L. Harris, Appl. Spectrosc. 29, 233 (1975). 21. W. E. Rippetoe and T. J. Vickers, Anal. Chem. 17, 2082 (1975). 22. H. L. Felkel and H. L. Pardue, Anal. Chem. 50, 602 (1978). 23. R. R. Williams and G. N. Coleman, Appl. Spectrosc. 35, 312 (1981).

The Determinat ion of Moisture in Some Solid Materials by Near Infrared Photoacous t i c Spec troscopy

Q. JIN,* G. F. KIRKBRIGHT, and D. E. M. SPILLANE Department of Instrumentation and Analytical Science, U.M.I.S.T., Sackville Street, Manchester M60 1QD, England

A n improved technique for the determination of moisture in synthetic single-cell protein, paper, and powdered milk substi-

Received 26 May 1981; revision received 7 August 1981. * Permanent address: Depar tment of Chemistry, Jflin University,

Changchun, China.

120 Volume 36, Number 2, 1982

tute utilizing photoacoustic spectroscopy at 1.93 # m is desc r ibed . A conventional scanning dispersive system and a simple inter- ference f i l ter a s s e m b l y h a v e b e e n used. Index Headings: Analysis, for moisture; Instrumentation, photoa- coustie; Methods, analytical; Techniques, spectroscopic; Mois- ture determination; photoacoustic spectroscopy; Near infrared r eg ion .

APPLIED SPECTROSCOPY