Anal. Chem. 1966, 58,3115-3121 Waiters, J. P. Science (Washington, D . C . ) 1977, 198, 787-797. Sacks, R. D.; Goldberg, J. M.; Collins, R. J.; Suh, S. Y. Prog. Anal. At. Spectrosc. 1982,5 , 111-154. Scott, R. H. Spectrochim. Acta, Part 8 1978,338, 123-125. Ishizuka, T.; Uwamino, Y. Spectrochim. Acta, Part 8 1983, 388, 519-527. Ishizuka. T.; Uwamino, Y. Anal. Chem. 1980,52, 125-129. Bennett, W. H. Phys. Rev. 1934,4 5 , 890-897. Anderson, 0. A.; Baker, W. R.; Colgate, S. A,; Ise, J., Jr.; Pyie, R. V. Phys.Rev.1958, 110,1375-1387. Chen, F. F. Introduction to Plasma Physics; Plenum: New York, 1974. Shiioh. J.; Fisher, A.; Rostoker, N. Phys. Rev. Lett. 1978, 4 0 , 515-518. Dennen, R. S.;Wilson, L. N. I n Exploding Wires; Chase, W. G., Moore, H. K.. Eds.; Plenum: New York, 1962; Vol. 2. Vitkovsky, I.M.; Bey, P. P.; Faust, W. R.; Fulper. R.. Jr.; Leavitt, G. E.; Shipman, J. D., Jr. I n Exploding Wires; Chase, W. G . , Moore, H. K., Eds.; Plenum: New York, 1962; Vol. 2. Nash, C. P.; DeSieno, R. P.;Olsen, C. W. I n Exploding Wires; Chase, W. G., Moore. H. K., Eds.; Plenum: New York, 1964; Vol. 3. Schenk, G.; Linhart, J. G. I n Exploding Wires; Chase, W. G., Moore, H. K., Eds.; Plenum: New York, 1964; Vol. 3. Stallings. C.; Nieisen, K.; Schneider, R. Appi. Phys. Lett. 1978, 2 9 , 404-406. Turchi, P. J.; Baker, W. L. J . Appi. Phys. 1973,4 4 , 4936-4945. Baker, W. L.; Clark, M. C.; Degnan, J. H.; Kiuttu, G. F.; McClenahan, C. R.; Reinovsky, R. E. J . Appi. Phys. 1978. 4 9 , 4694-4706. Lau, J. H.; Gupta, R. P.; Kekez, M. M.; Lougheed, G. D. I n Proceedings of the 1982 I€€€ International Conference on Plasma Science; IEEE: New York, 1982. Kekez. M. M.; Gupta, R. P.; Lau, J. H.; Lougheed, G. D. I n Proceedings of the 1982 I€€€ International Conference on Plasma Science; IEEE: New York, 1982. Cnare, E. C. J . Appl. Phys. 1966,3 7 , 3812-3816. Freeman, J. R.; Cnare. E. C.; Waag, R. C. Appi. Phys. Lett. 1967, IO, 111-113. Kachiiia, D.; Herlach, F.; Erber, T. Rev. Sci. Instrum. 1970,4 1 , 1-7. Clark, E. M.; Sacks, R. D. Spectrochim. Acta, Part 8 1980, 358, 471-488. Goldberg, J.; Sacks, R. Anal. Chem. 1982,5 4 , 2179-2188. Swan, J. M.; Sacks, R. D. Anal. Chem. 1985,57, 1261-1264.
3115
(26) Goode, S. R.; Pipes, D. T. Spectrochim. Acta, Part 8 1981,368, 925-929. (27) Kamia, 0. J.; Scheeline, A. Anal. Chem. 1986,5 8 , 923-932. (28) Kamla, G. J.; Scheeline, A. Anal. Chem. 1986,58, 932-939. (29) Albers, D.; Johnson, E.: Tisack, M.; Sacks, R . Appi. Spectrosc. 1986, 4 0 , 60-70. (30) Albers, D.; Sacks, R. Spectrochim. Acta, Part 8 1986, 4 1 8 , 39 1-402. (31) Coleman, D. A.; Sainz, M. A.; Butler, H. T. Anal. Chem. 1980,52, 746-753. (32) Carney, K. P.; Goidberg, J. M. Anal. Chem., following paper in this issue. (33) Sugden, S. J . Chem. SOC.1933,768-776. (34) Suh, S.Y. Ph.D. Thesis, University of Michigan, Ann Arbor, MI, 1980. (35) Suh. S. Y.; Collins, R. J.; Sacks, R. D. Appi. Spectrosc. 1981, 35, 42-52. (36) Salmon, S.G.; Hoicombe, J. A. Anal. Chem. 1978,5 0 , 1714-1716. (37) Swan, J. M.; Sacks, R. D. Spectrochim. Acta, Part 8 1985, 4 0 8 , 1239-1254. (38) Suh, S. Y.; Sacks, R. D. Spectrochim. Acta, Part 8 1981, 368, 1081-1 098. (39) Collins, R. J.; Sacks, R. D. Anal. Chem. 1983,55, 2036-2043. (40) Pearlstein. F. I n Modern Electroplating, 3rd ed.; Lowenheim, F. A,, Ed.; Wiley: New York, 1974. (41) Englehard Data Sheet Pure Metal Resinates; Englehard Corp.: East Newark, NJ. (42) Carney, K. P.; Goidberg, J. M., Department of Chemistry, University of Vermont, Burlington, VT, unpublished research.
RECEIVED for review June 25,1986. Accepted August 7,1986. We gratefully acknowledge financial support from a Research Corporation Cottrell Research Grant, the donors of the Petroleum Research Fund, administered by the American Chemical Society, and the University of Vermont Committee on Research and Scholarship. We also are grateful to BASF Wyandotte Corporation for donation of the microphotometer and to IBM for donation of the spectrograph.
Analytical Characterization of an Imploding Thin-Film Plasma Using Spatially and Temporally Resolved Spectrometry Kevin P. Carney and Joel M. Goldberg* Department of Chemistry, Uniuersity of Vermont, Burlington, Vermont 05405 The analytlcal utility of emission from an Imploding thin-fllm plasma atom cell for the dlrect atomlc spectrochemlcal analysls of solid samples Is evaluated. Time-resolved emlsslon profiles of spatlally Integrated radiation from the plasma reveal a very intense continuum background as well as reversal of analyte Ion and neutral atom llnes early in time. Timegating late in the discharge Improves analyte Ilne-tobackground ratlos. Spatially resolved spectrographic detection of plasma emission shows that the hlgh continuum background and analyte line reversal are localized In the center of the discharge tube, while vapor expelled from the tube produces a well-defined analyte spectrum In regions outslde of the confined plasma. Line-to-background ratios obtained by temporally resolved observation of emlsslon from outside of the plasma discharge tube were much improved over spatially Integrated values; however, the IrreproduclMlIty and lack of mass dependence of the analyte emisslon from thls outslde region preclude any analytical use. Temporally resolved measurement of the degree of analyte line reversal In the center of the plasma discharge tube is found to be the most analytically useful of all of the measurement techniques investigated wlth this Imploding plasma source.
Interest in direct solid sampling for atomic spectrochemical analysis has stimulated the investigation of novel high power
density plasmas as potential solid sample atom cells (1-7). In a companion report (8),we have presented the results of initial studies of the characteristics of an imploding thin silver film plasma suitable for direct solid sampling methods. Peak power densities in the megawatt per cubic centimeter range were reported, and sampling of solid powders deposited inside the tube prior to initiation of the plasma discharge was demonstrated. Initial spectroscopic studies, however, demonstrated poor powers of detection due to an extremely intense continuum background emission as well as significant selfabsorption of analyte emission lines. The spatial and temporal heterogeneity of emission from the plasma, though, indicate that powers of detection may be enhanced through the use of time-gating and spatial-masking techniques. This report, then, presents the results of a series of spectroscopic investigations of the analytical utility of emission from imploding thin-film plasma atom cells. Both spatially and temporally resolved spectroscopic techniques were used in order to better understand the emission characteristics of these plasmas.
EXPERIMENTAL SECTION All experimental conditions were as described in the accompanying paper (8) with the exception of the following. Optical and Electrical Monitoring. For the acquisition of spatially resolved but temporally integrated spectra, an overand-under mirror optical system was used to image the plasma
0003-2700/86/0358-3115$01.50/00 1986 American Chemical Society
3116
ANALYTICAL CHEMISTRY, VOL. 58, NO. 14, DECEMBER 1986
Table I. Effect of Time-Gate on Spatially Integrated Line-to-Background Ratios line-to-backgroundratios for time-gates mass of vanadium, p g
0-299 p s
0-70 p~
1 2 4
-0.030 0.006 -0.033
-0.085 --0.116 -0.163
70-299
ps
0.012 0.100 0.065
on the exit focal plane of the spectrograph. The optical system was designed so that the astigmatism generated by this mirror pair directly compensated for the astigmatism generated by the side-by-sidemirror pair in the spectrograph. Details of this optical system are similar to the spatially resolving photoelectric system described previously (8) except that 1000-mmfocal length mirrors separated by 676 mm were used with a 4' angle of off-axis illumination to produce an 8.6-mm separation between the tangential and sagittal foci. Design of such high-fidelity imaging systems is covered in detail by Goldstein and Walters (9, I O ) . Materials and Reagents. All powder samples were sized commercially by ATM Corporation, and only particles passing through a sieve with a 5-pm pore size were used in the studies presented in this report.
0-143 p~
143-299 p~
214-299 p s
-0.043 -0.051 -0.100
0.017 0.218 0.215
-0.026 0.471 0.619
I-
RESULTS AND DISCUSSION Temporally a n d Spatially Integrated Spectra. Photographic spectra from imploding thin-film plasmas have been presented in a companion paper (8). Weak analyte emission lines superimposed upon a very intense continuum background emission were observed for microgram quantities of solid powder samples of vanadium. Line-to-background ratios were not rigorously quantified but were significantly less than one for all ion and neutral atom analyte emission lines regardless of the electrical discharge conditions utilized. Similar emission spectra were observed regardless of the volatility of the vanadium compound (e.g., vanadium, VC, V205). While it is clear from these studies that refractory solids are sampled by the imploding thin-film plasma, the temporally and spatially integrated excitation characteristics demonstrated prohibit sensitive detection of analyte. Temporally Resolved b u t Spatially Integrated Spectra. Time-gated detection schemes have resulted in significant improvements in detection limits with exploding thin-film plasmas ( I I ) due to the temporal differences in the emission characteristics of analyte and continuum background. Temporally resolved emission profiles, then, might be useful for enhancement of the subunity line-to-background ratios observed with temporal integration of emission from imploding thin-film plasmas. Temporally resolved emission profiles for both vanadium ion and atom lines from imploding thin-film plasmas have already been presented in a companion report (8). These emission profiles show a net negative line emission (i.e., line reversal) during much of the first two current half-cycles of the discharge-only during the last two current half-cycles is a net positive line emission observed. Clearly, integration of emission from the plasma during the first half of the discharge adds only to the background signal and actually decreases the overall line-to-background ratio. In order to investigate the relationship between the analyte emission profile and the mass of analyte deposited inside the discharge tube, temporally resolved emission profiles were obtained for discharges performed with varying masses of vanadium carbide. Figure 1 shows emission profiles measured at the 311.84 nm V(I1) line from 6-kV, 50-pF discharges. Emission profiles obtained from VC powder samples containing 0, 1, 2, and 4 pg of vanadium are presented in Figure la-these profiles are not background-subtracted. Careful observation of these profiles shows significant differences during the first two current half-cycles. In general, the emission signals at the first two current maxima decrease with
0
100
200
300
Time, 1sec Flgure 1. Time-resolved but spatially integrated emission profiles from 6-kV, 50-pF discharges with (A) 0, (B) 1, (C) 2, and (D) 4 pg of vanadium as VC deposited in the discharge tube. Emission profiles (a) were recorded at V(I1) 31 1.84 nm, and background-corrected profiles (b) were generated by subtraction of profile A from each of the profiles B, C, and D.
increasing sample mass. As the current decreases to zero and reverses direction, all of the emission profiles reverse their relative orientations so that the emission intensities increase with increasing sample mass. This is more clearly seen in Figure l b where net line (Le., background-corrected) profiles are presented. Negative emission (absorption) is observed during the bulk of the first two current half-cycles, while positive emission is only observed late in the discharge and at the zero crossing of the discharge current between the first and second half-cycles. Line-to-background ratio profiles for this experiment show three major peaks at times corresponding to zero crossings of the discharge current. Additionally, the line-to-background ratio is consistently positive only a t times later than about 120 ps.
The effect of integration period on line-to-background ratio for the three sample masses used in this study is presented in Table I. As expected, the line-to-background ratios increase as the integration start times are stepped through the first two current half-cycles. It is also clear that integration of emission over the entire discharge degrades not only powers of detection but also any analytically useful relationship be-
ANALYTICAL CHEMISTRY. VOL. 58, NO. 14. DECEMBER 1986 * 3117
i A
i
B
-
1
C
Figure 2. SpaHally resolved but tempaaliy integrated specvum from an EkV. 24-pF discharge wim a 1-pg sample of vanadium as VC: (A) AH]). 328.07 nm: (B) V(1). 318.34 nm. 318.40 nm. 318.54 nm: IC) V(II), 309.31 nm. 310.23 nm. 311.07. 311.84 nm.
tween analyte mass and net line emission intensity. In fact, ~ when integrated over the entire dwharge ( W 2 9 9 - p timegate), there is very little difference observed between the signals obtained with any of the three masses of analyte used in this experiment. Maximum lineto-background ratios are recorded for an integration period beginning a t the start of the fourth current half-cycle (214-299-ps time-gate) and represent a significant improvement over nongated integration. The line-to-background ratios are still quite poor (below a value of one), however, and although they increase with increasing analyte mass, the increase is decidedly nonlinear. Furthermore, although the reproducibilities of the background and line plus background signals are reasonable (about 5-10% relative standard deviation (RSD)), the resulting net line reproducibilities (generally around 50% RSD) are unacceptable for quantitative work. It is apparent that temporal resolution alone will not improve powers of detection to a degree sufficient for accurate quantitation of analyte introduced into the imploding discharge. Temporally Integrated b u t Spatially Resolved Spectra. Knowledge of the spatial distribution of emission from the plasma might afford some insight into the use of spatial-masking techniques for the improvement of analyte detection. The high image fidelity spectrographic system, then, wa8 used to obtain spatially resolved but temporally integrated spectra from imploding thin-film plasmas. A portion of the spectrum obtained from an 8-kV, 24-pF discharge with a VC sample containing 1pg of vanadium is shown in Figure 2. An extremely intense continuum background emission (disrupted only by broad self-reversal of the silver neutral resonance line) is evident in all spatial regions corresponding to the interior of the plasma discharge tube. What is most striking is the observation of significant emission from regions outside of the discharge tube-this represents emission from material that has been expelled from the confined plasma. The spectrum from regions outside of the discharge tube is characterized by intense line emission from both silver and vanadium species superimposed on a relatively low intensity continuum background. Spatial masking of the intense continuum emitted from the center of the discharge, then, should reduce the background emission signal significantly and improve lineto-background ratios. Temporally a n d Spatially Resolved Spectra: Outside Discharge Tube. Based on the results of the spatially resolved spectrographic studies, the spatially resolving photoelectric system was used to investigate the emission profiles of analyte from regions outside of the discharge tube. For this study, emission from 6-kV, 50-pF discharges with 0, 1,2, and 4 pg of vanadium as VC were monitored from a 250-pm-wide region located about 3 mm helow (vertical displacement) the
o
30
60
90
im
150
lWt?C.I(SP(
Flgure 3. Tempwaliy and spatially resolved emission profiles from 6-kV. 50-pF discharges with (A) 0. (B) 1. (C) 2, and (D) 4 pg of vanadium as VC deposited in me discharge tube. Emission profileswere recorded a1 V(I) 318.40 nm and are not background-corrected. A
250-@m-wideemission slice located approximately 3 mm below the outer edge of the discharge tube was monitored outer edge of the discharge tube. Figure 3 shows the emission profiles recorded with the monochromator set to monitor the V(I) 318.40-nm line. In a qualitative sense, these emission profiles are consistent with the spatially resolved photographic results presented in the previous section: the background emission is considerably reduced relative to the analyte emission. Furthermore, all emission profiles from discharges containing analyte are greater in intensity than the continuum background emission profile; no analyte self-reversal is observed. Line-to-background ratios recorded in this study were much improved over the spatially integrated values-line-to-background ratios approaching unity were measured for an integration period spanning the entire discharge. It is important to note that time-gating would have little impact on the line-tobackground ratios in this situation due to the relatively even distribution of continuum throughout the discharge; in fact, time-gating late in the discharge would actually result in a decrease in line-to-background ratios here. Since this experiment measured emission from analyte expelled from the plasma tube, the analytical utility of the data is heavily dependent upon the expulsion process. In particular, the reproducibility of the emission from this region as well as its relationship to the mass of analyte deposited in the discharge tube must be evaluated. Unfortunately, this detection scheme failed in both regards. The emission measured from this spatial zone was extremely erratic-the RSD obtained from the four integrated emission intensities recorded for the blank in this experiment was over 40%; a RSD of ahout 5% is typical for spatially integrated continuum background emission measurements. Furthermore, a consistent relationship between the analyte emission intensity and the mass of analyte deposited in the discharge tube is not observed in Figure 3. Overall, then, while improved line-to-background ratios are observed outside of the discharge tube, the analyte emission intensities do not correlate well with analyte mass and the reproducibility of the analyte and background emission is quite poor. While clearly not of great analytical utility, physical information about the plasma may be obtained from emission measurements in this spatial zone. Surprisingly, significant analyte emission is observed outside of the plasma discharge tube as early as 30 ps into the discharge. In addition, the
3118
ANALYTICAL CHEMISTRY, VOL. 58, NO. 14, DECEMBER 1986 0-
-_J--
7
\
0-
:5-
25
I
'.l
50-
D
- A
( bl
%
c
x
+-A
Figure 5. Microphotometer traces from a spatially resolved but temporally integrated emission spectrum obtained with a 6-kV, 24-pF discharge using a I - p g sample of vanadium as VC. Emission from 2-mm regions located outside of the discharge tube (b) and in the center of the discharge tube {a) were integrated by the microphotometer slit: (A) V(II), 311.84 nm; (6) V(II), 311.07 nm; (C) V(II), 310.23 nm; (D) V(II), 309.31 nm.
most intense vanadium ion line outside the discharge tube (D) is similar to the decrease in emission intensity shown for the same line in the trace from the center of the discharge tube. Thus, the line reversals observed are actually quite significant. Line Reversal/Absorption Spectra. Since the prospects for successful quantification of analyte using emission techniques were shown to be quite poor even with the use of temporal and spatial resolution, investigations of the analytical utility of the severe line reversal observed with these plasmas were initiated. Of course, the effects of line reversal were evident in the spatially integrated but temporally resolved studies presented earlier in this paper-in those studies, a clear mass dependence of the degree of line reversal was observed (see Table I: 0-70 and 0-143 ps time-gates). Although the analytical utility of the line reversal was not rigorously evaluated, the intensity of the effect was not significantly greater than the intensity of the positive line emission, which was judged to be unsuitable for analytical work. The spatially resolved photographic and photoelectric studies presented earlier in this paper suggest that the degree of line reversal can be maximized by observation of radiation only from the center of the discharge tube. The spatially resolving photoelectric detection system, then, was used to study the quantitative characteristics of analyte line reversal in the center of the plasma discharge. A 2-mm-wide external slit was used in this optical system and was placed at a height so that only radiation from a central slice of the plasma was monitored. Our initial concern was with characterizing the temporal behavior of analyte line reversal from this region of the plasma discharge. Initial studies, then, simply involved recording continuum background emission profiles as well as analyte emission profiles. Since we expected the line reversal to be most severe for atomic species, these studies were performed with the monochromator set to monitor the V(1) 437.92-nm line. These preliminary investigations used microgram amounts of vanadium as VC powder and demonstrated that significant line reversal occurs in this spatial zone at almost
ANALYTICAL CHEMISTRY, VOL. 58, NO. 14, DECEMBER 1986
3119
U
0
2
4
8
6
10
1.
;E Mass Vanadium, pg
-- - - -- -- - - - - -
P
3 U
0
1 ime,ps Figure 6. Time-resolved emission profiles (b) and corresponding absorbance profiles (a) from a 2-mm region in the center of the discharge tube measured at V(1) 437.92 nm for 6-kV, 24-pF discharges with (A) 0, (8) 1.0, (C) 4.7, (D)6.0, and (E) 12.3 pg of vanadium as VC deposited in the discharge tube.
all times during a 6-kV, 24-pF discharge. The mass dependence of the vanadium neutral atom line reversal was investigated by recording emission profiles from discharges having varying masses of vanadium deposited inside the discharge tubes. The results of this study are presented in Figure 6. Figure 6b shows the emission profiles obtained for discharges performed with 0, 1.0, 4.7, 6.0, and 12.3 pg of vanadium (as VC) present in the discharge tube. Throughout the first three current half-cycles, all analyte emission profiles are lower in intensity than the continuum background profile, indicating that the line is reversed for almost the entire discharge. For larger masses of analyte, line reversal is observed well into the fourth current half-cycle of the discharge. Note that significant line reversal is observed for even the lowest sample mass (1.0 kg). Furthermore, the degree of reversal of the line clearly depends, at least in a qualitative sense, on the mass of analyte. Since this experiment is, in essence, an atomic absorption experiment, quantification of the degree of line reversal would be best discussed in terms of absorbance (which should be proportional to the analyte mass). To a first approximation, then, we can view the continuum background emission profile as Io and convert the analyte emission profiles (I)to absorbance profiles. Absorbance profiles were generated in this fashion using the data presented in Figure 6b and are shown in Figure 6a. All four absorbance profiles show significant absorbance during the second and third current halfcycles-absorbance is first detected at about the first current peak and seems to oscillate approximately in phase with the discharge current. The quantitative relationship between absorbance and the mass of analyte deposited in the discharge tube was evaluated for a number of time-gates. For all time-gates, a reasonably linear relationship between integrated absorbance and anal@ mass was observed-changes in time-gate affected the slope of the analytical curves primarily. Analytical curves obtained for three time-gates are presented in Figure 7 . Integration over the entire discharge (curve A) gave the poorest analytical
a
60 90 TIM, ps
1MlW
Figure 7. Analytical curves for vanadium derived from absorba :e profiles shown in Figure 6a using three different time-gates: (A) 0-130 ps, (6) 15-1 10 ps, and (C) 75-1 10 ps.
-L 1
1
0
30
60 90 1 m e . ps
120
150
Flgure 8. Time-resolved ionic absorbance profiles measured at 310.23 nm for 6-kLI 24-pF discharges with (A) 1.0, (B) 2.0, (C) 4.7, and (D) 6.0 pg of vanadium as VC deposited in the discharge tube.
sensitivity due to the integration of some net positive line emission late in the discharge. Delaying the start of integration until 15 ps into the discharge (when significant analyte absorbance is first observed) as well as ending integration at 110 ps into the discharge (when significant positive line emission is first observed) results in the curve marked B. Note that this time-gate improves the analytical sensitivity significantly over the gate used in curve A. The analytical curve marked C was obtained with a delay in the start of integration of 75 pus. This curve is almost indistinguishable from curve B and is chsracteristic of most of the time-gates investigated that were within the integration period used for curve B. The precision associated with these measurements were quite reasonable (RSD typically between 10 and 20%). The analytical utility of the line-reversal technique was also investigated with ionic species. Emission intensity profiles at the V(I1) 310.2-nm line were recorded using analyte masses between 1.0 and 6.0 pg of vanadium as VC. The ionic ab-
3120
ANALYTICAL CHEMISTRY, VOL. 58, NO. 14, DECEMBER 1986
r
1O r
2 5 urn
0.8
-84
0
, 30
” 60
1
I
I
90
120
150
Time,ps
0
I
I
I
I
1
33
60
90
120
150
Ttme,ps
Flgure 9. Effect of initial charging voltage on time-resolved neutral atom absorbance profiles measured at 437.92 nm for 24-yF discharges with 5 yg of vanadium as VC deposited in the discharge tube. Initial charging voltages of (A) 4 kV, (6)5 kV, (C) 6 kV, and (0) 8 kV were utilized. Current waveform shown is for a 6-kV discharge.
sorbance profiles obtained are presented in Figure 8. In contrast with the neutral atom absorbance profiles, very little change in the ionic absorbance profiles is observed over this mass range even though there is a significant ionic absorbance signal. There are also qualitative differences between thg ionic and atomic absorbance profiles. First, while atomic absorption is observed as early in time as the first current peak, the first appearance of significant ionic absorption is delayed until the second current half-cycle. Furthermore, ionic absorption is not observed a t times later than the third current half-cycle with even the highest sample mass. This could mean that analyte ions are localized closer to the walls of the discharge tube and, thus, are out of the observation zone used for the absorption experiments. Overall, the line-reversal technique provides a unique window into ground-state atomic and ionic populations of analyte over the course of the imploding plasma discharge. In an analytical sense, reversal of atomic lines provides a mass-sensitive signal that can be used to probe analyte populations in a hot plasma environment not amenable to conventional emission and absorbance probes. The ionic absorption signal, however, is of limited analytical utility as it appears to be insensitive to the mass of analyte deposited in the discharge tube. The relatively intense absorption signals obtained for both atomic and ionic species, however, indicate that significant populations of analyte can be produced by direct atomization of very refractory solid powder samples from the discharge tube wall. The limitations of the line-reversal technique as an analytical tool with imploding thin-film plasmas become quite apparent when simple parametric studies are attempted. For example, in an effort to study the effects of discharge energy on atomization efficiency, neutral atom absorbance profiles were obtained for a constant mass of VC using four different initial charging voltages. The absorbance profiles obtained are presented in Figure 9. The four absorbance profiles are quite similar during the second current half-cycle only; later in the discharge, the magnitude of the absorbance signal is strongly related to the initial charging voltage. While these differences may be interpreted with respect to physical properties operative in the different plasma discharges, clearly the analytical relationship between sample mass and the ab-
Flgure 10. Effect of monochromator slit width on time-resolved neutral atom absorbance profiles measured at 437.92 nm for 6-kV, 24-yF discharges with 5 yg of vanadium as VC deposited in the discharge tube.
sorbance signal is shown to depend upon the electrical discharge parameters. The voltage study also raises questions about the validity of using the continuum background profile as I, in the conversion of analyte emission profiles to absorbance profiles. All of the absorbance profiles show a strong dependence upon the discharge current even though they have all been “normalized to the continuum background profile by the absorbance conversion. If the background emission profile is an accurate representation of Io, then the oscillations observed in the absorbance profiles actually reflect variations in the groundstate analyte populations in the plasma. Since ground-state atom populations should decrease with increasing discharge current due to ionization and excitation processes, this could imply that the atomization process occurs continuously throughout the entire discharge; this is highly unlikely as atomization efficiency studies with exploding thin-film plasmas indicate that sample atomization is completed during the first current half-cycle for such small particles (12). The absorbance oscillations can also be explained by more waveform. We careful consideration of the normalizing (lo) have assumed that the absorption process is due solely to ground-state analyte at the ends of the tube absorbing the intense continuum radiation emitted from the hot core of the plasma. Because of the extreme pressure broadening of the emission and absorption lines in the plasma, atomic and ionic absorption of continuum radiation may be sensitively detected with the moderate band-pass of our monochromator. It is also likely, however, that there is significant analyte line emission that is absorbed in addition to continuum radiation-if analyte line emission constitutes a significant portion of Io, then the use of the continuum background profile underestimates Io and will result in a net decrease in the calculated absorbance signal. Since one would not expect neutral atom line emission to oscillate as significantly as continuum background emission with the discharge current, the oscillations observed in the calculated absorbance profiles could be due to neglect of the line emission component of the true Io. The relative contributions of analyte line and continuum background emission of I , can be investigated by measuring absorbance profiles using different instrumental band-passes. An increase in the instrumental band-pass beyond the absorbing bandwidth should produce only a decrease in the sensitivity of the measured absorbance if Io is adequately approximated by the continuum background emission profile. If line emission is a significant component of Io, however, the degree of oscillation of the absorbance profile should be affected by the changes in the instrumental band-pass as well. Figure 10 shows absorbance profiles obtained using three
ANALYTICAL CHEMISTRY, VOL.
different slit widths: 25, 100, and 200 pm. As expected, the overall sensitivity decreases with increasing instrumental band-pass. In addition, the degree of oscillation of the absorbance profiles is seen to increase with decreasing instrumental band-pass. Recall that decreasing the instrumental band-pass results in a much greater dependence of the absorbance profile on the line emission component of Io;recall also that these absorbance profiles were calculated using an Io that ignores any line emission component. Thus, the increased oscillations observed with the smaller instrumental band-pass profiles can be attributed to neglect of the line emission component of Io in the calculation of these absorbance profiles. Analyte line emission from the center of the discharge tube, then, must be included in the Io profile in order to calculate absorbance profiles accurately. Furthermore, since the line emission will be mass dependent, a separate measurement is needed for each sample mass investigated. Unfortunately, it is not possible to measure the analyte line emission independently of the line-reversal profile; accurate absorbance measurements, then, cannot be made. While the extent of the influence of the line emission profile on line reversal is not known, the analytical limitations of the technique suggest that its greatest utility is as a plasma diagnostic method.
CONCLUSIONS The spatially and temporally resolved spectroscopic studies presented in this report paint a fairly complex picture of imploding thin-film plasma atom cells. The conditions for emission spectrometric determinations of analyte are extremely adverse: line-to-background ratios are very poor for samples of moderate size due to the very intense continuum background emission as well as significant self-reversal of analyte atom and ion lines. The high continuum background is confined primarily to the discharge tube itself and drops off in intensity during the last quarter of the discharge. Although some analyte is expelled from the discharge a t a fairly high initial velocity, this process is not very reproducible and is not observed to be mass dependent. Emission from the expelled analyte is first detected late in the first current half-cycle, almost coincident with the observation of line reversal in the center of the discharge tube. Analyte line reversal dominates the emission characteristics from within the plasma discharge and can actually be used as a sensitive probe of ground-state atom and ion populations. The degree of analyte neutral atom line reversal can be used analytically, but special care must be used in the interpretation of absorbance profiles as they are very dependent on electrical discharge conditions as well.
58, NO. 14, DECEMBER 1986
3121
All of the spectroscopic studies indicate that the imploding thin-film plasma is an effective atom cell which has the capability of rapidly converting even very refractory solid powder samples to an atomic/ionic vapor. The emission spectroscopic properties of these plasmas, however, are quite hostile to any sort of analytical determination even with the use of spatial and temporal discrimination. Successful utilization of these imploding plasmas for direct spectrochemical analyses of solid samples will depend upon the use of in situ reexcitation techniques for measurement of analyte in the postdischarge environment (13). Alternately, electrical and physical discharge conditions can be modified so as to produce a milder plasma environment exhibiting more satisfactory emission characteristics for analytical spectrometric measurements. Both of these avenues of investigation are presently being pursued.
ACKNOWLEDGMENT We thank Walter Weir for his help with the design and construction of the apparatus. Registry No. Ag, 7440-22-4; V, 7440-62-2; V', 14782-33-3.
LITERATURE CITED Laqua, K. I n Analytical Laser Spectroscopy; Omenetto. N..Ed.; Wiley: New York, 1979. Waiters, J. P. Science (Washington, D . C . ) 1977, 798, 787-797. Sacks, R. D.; Goldberg, J. M.; Collins, R. J.; Suh, S. Y. Prog. Anal. At. Spectrosc. 1982, 5 , 111-154. Goldberg, J.; Sacks, R. Anal. Chem. 1982, 5 4 , 2179-2186. Swan, J. M.;Sacks, R. D. Anal. Chem. 1985, 5 7 , 1261-1264. Kamla, G. J.; Scheeline, A. Anal. Chem. 1986, 5 8 , 923-932. Kamla, G. J.; Scheeline, A. Anal. Chem. 1986, 5 8 , 932-939. Carney, K. P.; Goldberg, J. M. Anal. Chem., preceding paper in this issue. Goldstein, S. A.; Walters, J. P. Spectrochim. Acta, Part8 1976, 3 1 8 , 20 1-220. Goldstein, S. A.; Walters. J. P. Spectrochim. Acta, Part 8 1976, 3 1 8 , 295-316. Suh, S. Y.; Collins, R. J.; Sacks, R. D. Appl. Spectrosc. 1981, 3 5 , 42-52. Clark, E . M.: Sacks, R. D. Spectrochim. Acfa, Part 8 1980, 3 5 8 , 471-460. Coleman, D. A.; Sainz, M. A.; Butler, H. T. Anal. Chem. 1980, 52, 746-753.
RECEIVED for review June 25,1986. Accepted August 7,1986. We gratefully acknowledge financial support from a Research Corporation Cottrell Research Grant, the donors of the Petroleum Research Fund, administered by the American Chemical Society, and the University of Vermont Committee on Research and Scholarship. We also are grateful to BASF Wyandotte Corporation for donation of the microphotometer and to IBM for donation of the spectrograph.
