JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, FEBRUARY 1987, VOL. 2 I I PI p2 + s 1 59 PAR X- Yplotter O- -4 162 Population Distribution of Atomic Uranium in the Afterglow of a Pulsed Hollow-cathode Discharge 0 0 PAR PAR 164 164 Yves Demers, Jean-Marie Gagne and Piero Pianarosa Laboratoire d'Optique et de Spectroscopie, Departernent de Genie Physique, Ecole Polytechnique, Case Postale 6079, Succursale "A", Montreal H3C 3A7, Canada From laser absorption measurements we have deduced the time evolution of the population distribution of atomic uranium in the afterglow of a pulsed hollow-cathode type discharge. The vapour generator operates with xenon as the discharge sustaining gas a t a pressure of 280 Pa (2.1 Torr). The current pulse characteristics are width 250 ps and height 1.5 A.The pulse repetition frequency is 100 Hz. It is shown that the populations in the three metastable levels at 6249, 3868 and 3800 cm-1 decrease almost exponentially in a time interval between 150 and 300 ps. From 400 ps onwards in the afterglow, the atom population is essentially shared between the ground and the first metastable (620 cm-1) levels. Furthermore, starting from 9 ms in the afterglow more than 80% of the U atoms are found in the ground level. Keywords: Pulsed hollow-cathode discharge; afterglow; uranium; population distribution Hollow-cathode type discharges, widely used in the past essentially as sources of narrow spectral lines for spectroscopic investigations, are now finding applications as generators of metal atoms in the vapour phase.'-10 Our laboratory has been using d.c.and pulsed vapour generators of a hollow-cathode design for a number of years. They were used for spectroscopic studies of uranium,ll-15 for the isotope analysis of uranium samples16 and, more recently, to investigate the production of atomic vapours of Th,17 Mo18 and Zr.19 Despite the wide range of applications cathode sputtering affords, many of the physical phenomena and parameters occurring in and characterising the discharge are still not fully understood. Among the latter, the temperature and popula- tion distributions in the various atomic levels are important for the characterisation of this type of vapour generator. While excitation or spectroscopic temperatures and the ensuing population distribution in the atomic level apparent in d.c.powered hollow-cathode discharges have been exten- sively studied,2@-23 the same parameters and the correspond- ing population distribution in the afterglow of pulsed dis- charges are much less well understood. In the afterglow of a pulsed discharge, excitation of atoms by the discharge and their interaction with the radiation in the hollow-cathode cavity no longer exist. As a consequence, at r 1 the end of the current pulse, radiative levels decay rapidly to the ground state or to any one of the low-lying metastable levels. These metastable states decay mainly by collisions with low-energy electrons or foreign gas atoms and their lifetime can be considerable indeed. It would then be quite feasible to have an appreciable fraction of sputtered atoms in the low-lying levels well after the end of the current pulses.To improve the characterisation of our pulsed discharge tubes we therefore deemed it important to investigate the distributian of U atoms between the ground and the first few excited levels using laser absorption spectroscopy. Experimental The reservoir of U atoms used in this work, a laboratory-made hollow-cathode discharge cell, has already been described elsewhere (see, for example, references 13-15). The pulsed power supply built in our laboratory generates current pulses with intensity and width fixed specifically for this investiga- tion, at 1.5 A and 250 ps, respectively. Pulse rise and fall times are less than 5 ps. The pulse repetition rate is fixed at 100 Hz. For this study the discharge sustaining gas is xenon [Xe pressure 280 Pa (2.1 Torr)].These experimental conditions have been found to correspond to a maximum U I ground state density. 13 Fig. 1. Laser spectrometer, see text for details60 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, FEBRUARY 1987, VOL. 2 The laser spectrometer used for this investigation is shown in Fig. 1. The laser, a Coherent Radiation 599, is monomode and frequency stabilised; its line width is less than 7 MHz. The dye is Rhodamine 6G. A 0.5-m monochromator is used to tune the laser wavelength to the atomic transitions under study. The incident laser beam is divided by a beam splitter M1, whose transmission efficiency is such that the beam intensity incident at the generator is less than 6 pW cm-2. Such a low value is needed in order to have an absorption that is independent of the incident intensity,13 hence saturation effects will not plague the measurements.Two photodiodes, P1 and P2, (rise time ca. 20 ns) provide signals proportional to incident, Io(v), and transmitted, Itr(v,t), light intensities, respectively. These signals are fed into the sampling heads of a boxcar integrator (Princeton Allied Research Models 164 and 162). The boxcar is triggered by the leading edge of the current pulse. Its output provides a voltage signal that is proportional to -log [It,(v,t)/I0(v)] = absorbance and drives the Y axis of a chart recorder. The Y axis has been calibrated to give directly the value of the integral of the absorption coefficient over the optical path, i.e., With this experimental configuration two different types of measurements are possible.1. We can fix the sampling windows of the boxcar at a definite time ti measured from the end of the current pulses. By scanning the laser frequency synchronously with successive current pulses we obtain the profile of the absorption coefficient over the optical path at the predetermined time ti. In this mode of operation the frequency resolution is approximately 15 MHz and the temporal resolution is equal to the width of the sampling windows (5 p). 2. Alternatively, we can fix the laser frequency at the centre of an absorption line (YO) and, with each current pulse, shift the boxcar sampling windows an amount At. We then obtain the time evolution of the absorption coefficient at the line centre Jb k,,(t, <)dc.In this mode of operation the frequency resolution is ca. 50 MHz and the temporal resolution 50 ys. Results We investigated the atom density in the 5L:.ground state and in four low-lying metastable levels of uranium. These levels are the 620 cm-1 (5K3, the 3800 cm-1 (5L:), the 3868 cm-1 (53 belonging to the Pds2 configuration and the 6249 cm-1 (7 ) level belonging to the f3d2s configuration. The para- meters associated with the transitions that were studied are listed in Table 1. Measurements of Iik(v, t = ti, c)d< The integral k(v, t = ti, 5)dt; over the vapour phase length was measured at different ti between 0 and 400 ps. We assumed that in this time interval there is no diffusion of the atomic vapour outside the cavity of the cathode.The absorbing path length is then equal to the cathode length Z (Z = 3 cm). Assuming the vapour to be homogeneous, we have Zk(v, t = ti) = JLk(v, t = t i , c)dc . . . . (2) The population level can then be evaluated using24 N = - 8ng1 $” k(v,t=ti)dv . . . . (3) ho2Ag2 -” In this expression gl and g2 are the statistical weights of the lower and upper level of the transition, A is the Einstein transition probability and the wavelength of the transition at the line centre. The total absorption coefficient Jwk(v, t = tJdv used in equation (3) is obtained by simply taking the area under the experimentally measured curve of k(v, t = ti) versus frequency. Fig. 2 is an example of the profile of an absorption coefficient k(v, t = ti) versus v. The profile deviates slightly from a purely Gaussian shape.This could be due to the presence, in the early afterglow, of spatial inhomogeneities in the kinetic temperature values and, as pointed out by Leblanc et aZ.,25 of motion of the vapour as a whole. The experimental results are summarised in Table 2. Fig. 3 is a plot of atom density in the three levels at 3800, 3868 and 6249 cm-1 as a function of the time in the afterglow. There are two main sources of error in the density values measured. The first one is the uncertainty in the published values of the transition probabilities. The second one is the difficulty in estimating the length Z of the volume occupied by the vapour inside the cavity of the cathode. We estimate that the population densities given here are accurate to within a factor of two.- W 0.5 1 0.6 0 1 2 3 4 5 6 dGHz Fig. 2. Profile of the absorption coefficient k(v) at ti = 0 for the transitions at 598.6 nm: continuous line, experimental profile; Gaussian line-shape of equivalent width ~~ Table 1. Wavelengths, upper and lower energies and classifications and transition probabilities Lower energy level Upper energy level WavelengtWnm Elcm-1 Classification Elcm-1 Classification All06 s-1 575.81 0 5L: 17361 395 7L6 0.21 f 0.04* 5Kz 17361 395 7L6 0.59 k 0.18.t 597.15 620.323 597.63 3800.829 5G 20528.898 ’M8 2.1 k 0.93 598.61 3868.486 5H: 20569.228 J = 4 1.1 k 0.4$ 599.73 6249.029 ’@ 22918.555 5L7 1.9 k 0.7$ * From reference 13. t From reference 14. t From reference 32.JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, FEBRUARY 1987, VOL.2 61 Table 2. Experimentally determined populations of the energy levels (1012 cm-3) tips 0 17 34 50 100 150 200 250 350 NO 24 - - 31 30 29 27 27 27 N620 N3800 21 4.7 - - - - 3.0 2.1 11 1.4 - 0.85 - 0.47 9 0.24 - - N3868 3.5 - - 2.1 1.5 0.95 0.60 0.35 0.21 N6249 4.4 2.3 1.8 1.4 0.75 0.35 - - - 0 100 200 300 400 U P Fig. 3. Time evolution of the population of the levels at 3800 cm-l (A), 3868 cm-1 (B) and 6249 cm-1 (C) at the beginning of the afterglow Time Evolution of 1; kv,(t, <)d& In a subsequent study we monitored the absorption coefficient time evolution at the line centre, i. e., the quantity sb kvo(t, l;) dl;. The measurements were performed for the transitions at 575.81 nm (0 cm-1 + 17361 cm-I), 597.1 nm (620 cm-1 -+ 17361 cm-1) and 597.6 nm (3800 cm-l-+ 20528 cm-1).The results obtained are presented in Figs. 4-6, respectively. At the end of the current pulse the atomic vapour starts diffusing out from the cathode cavity and, in the time span of a few milliseconds, it fills the entire volume of the generator. The quantity k(v, t, 0) becomes an unknown function of 5 and this fact makes the use of equation (3) impossible. As a consequence, the absolute population density in the different levels cannot be evaluated any longer. However, we can estimate the population ratio of two levels if we make the reasonable assumptions that this ratio is constant anywhere inside the generator and that the line shapes of the different transitions are the same. This ratio can be evaluated from Using equation (4) and the experimental values for the integrated absorption coefficients from Figs. 4 and 5 we evaluated the population ratios in the ground (5Lz) and first metastable (620 cm-1, 5K:) levels.The results are shown in Fig. 7. From the Doppler width of the h = 575.8-nm line we have deduced that the temperature of the U atoms, fort > 700 ps, is almost constant and equal to 380 K. For local thermodynamic equilibrium, for a temperature Tk = 380 K, the Boltzmann distribution would give a value 10.5 for the &&2dN62&0 ratio. As can be seen from Fig. 7 the exponential decay constant is ca. 4 ms, and at t = 9.75 ms the vapour has not yet reached the condition of equilibrium. 1 I I I 1 I 1 I I 0 1 2 3 4 5 6 7 8 9 10 tlms Fig. 4. Evolution of Jbkv,,(f, f)df for the h = 575.8-nm line Discussion and Conclusions From the values presented in Table 2 and from Figs.3-6 it appears that the time evolution of the metastable levels at 6249, 3868 and 3800 cm-1 is quite different from that of the 620 cm-1 level and of the ground level. The first three levels are characterised by a rapid decrease in population in the very early afterglow (WOO ps). This is followed, however, by a slight, but reproducible, population increase at t = 700 ps (see Fig. 6). This rapid destruction of the metastable levels is quite a common feature of gas discharges in general.2G28 Two facts lead us to believe that this rapid destruction of the metastable levels may proceed via superelastic collisions with electrons. Firstly, the slow decay (z = 0.5 ms) of the U I1 density in a similar discharge29 implies that the electron density is still rather high during the first few milliseconds in the afterglow.Secondly, in the afterglow of a pulsed discharge the electron gas temperature tends very rapidly (50 ps) to the carrier gas kinetic temperature.= In our case this corresponds to a drastic cooling of the electron gas from a temperature of the order of 3500 K30 to ca. 850 K in the first 50 ps of the afterglow regime.25 At this electron temperature, and for the levels under investigation, de-excitation phenomena are clearly predominating over excitation. The second trend typical of the time evolution of the population common to these three levels is the slight increase observed at ca. 700 p. This increase reaches a maximum at t = 1.2 ms (see Fig.6) then slowly diminishes and is zero again at t ca. 5 ms in the late afterglow. This population variation seems to proceed via ion recombination effects followed by radiative transitions to lower metastable levels.26 The 620 cm-1 and the ground levels, on the contrary, decrease more slowly in the discharge afterglow when compared with the previous three levels. One characteristic62 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, FEBRUARY 1987, VOL. 2 5 o 1 2 3 4 5 6 7 a 9 10 tlms Fig. 5. Evolution of $,kv0(t, 5)dc for the li = 597.1-nm line - 0 3 3 3 0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 Urns Fig. 6. Evolution of J$vo(t, 5)dC for the li = 597.6-nrn line 2t 0 2 4 6 8 1 0 t/ms Fig. 7. Time evolution of the ratio NOg62dN620g0 in the discharge afterglow.The broken line represents the equilibrium value of this ratio feature of the ground level is its increase in population in the first few instants of the afterglow regime (see Fig. 4). This increase follows from radiative and collisional de-excitation processes of the high lying levels populated by the glow discharge. By comparison, the increase in population of the 620 cm-1 level in the same time span is much less pronounced. This difference can be explained by the fact that, while ground- state atoms are destroyed in part by diffusion processes and in part by Penning type reactions,” the 620 cm-* atoms decay by collisional de-excitation to the ground state, further contribut- ing to this latter population, Note also that the small increase in lh k (Y, t , c)dc observed in Fig.5 could be due to a reduction in the afterglow regime of the Doppler line width. From our study it appears that, when the vapour generator operates in a pulsed mode, the populations of the levels at 6249, 3868 and 3800 cm-1 are, for t > 400 ps, approximately two orders of magnitude lower than the ground-state popula- tion. It stands to reason that the same temporal behaviour could be attributed to the other U metastable levels as they are higher in energy than the 3800 and 3868 cm-1 levels. Hence we conclude that in the late afterglow the majority of sputtered atoms are distributed between the ground level and the metastable state at 620 cm-1. Finally, in the very late afterglow ( t > 9 ms), at least 80% of the atoms are in the ground state.The technical assistance of P. A. Dion and Y. Lemire is greatly appreciated. This work has been supported in part by the Natural Sciences and Engineering Research Council of Canada and the Ministkre de 1’Education du Quebec. 1. 2. 3. 4. 5 . 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. 32. References Russell, B. J., and Walsh, A., Spectrochim. Acta, 1959, 15, 883. Gatehouse, B. M., and Walsh, A., Spectrochim. Acta, 1960, 16,602. Gough, D. S., Hannaford, P., and Walsh, A., Spectrochim. Acta, Part B, 1973, 28, 197. Gough, D. S., Anal. Chem., 1976,48, 1926. Gerstenberger, D. C., Solanki, R., Collins, G. J., ZEEE J. Quantum. Electron., 1980, QE16, 820 and references cited therein. Goleb, J.A., and Brody, J. K., Anal. Chim. Ada, 1963, 28, 457. Goleb, J. A,, Anal. Chem., 1963, 35, 1978. Goleb, J. A., Anal. Chim. Acta, 1966, 34, 135. Goleb, J. A,, and Yokoyama, Y., Anal. Chim. Acta, 1964,30, 213. Goleb, J. A., Anal. Chim. Acta, 1966, 36, 130. GagnC, J.-M., Mongeau, B., Leblanc, B., Saint-Dizier, J.-P., Pianarosa, P., and Bertrand, L., Appl. Opt., 1978, 17, 2507. GagnC, J.-M., Leblanc, B., Mongeau, B., Carleer, M., and Bertrand, L., Appl. Opt., 1979, 18, 1084. GagnC, J.-M., Carleer, M., Leblanc, B., Demers, Y., and Mongeau, B., Appl. Opt., 1979, 18,2107. GagnC, J.-M., Mongeau, B., Demers, Y., and Pianarosa, P., J. Opt. SOC. Am., 1981,71, 1140. Dr&ze, C., Demers, Y., and GagnC, J.-M., J. Opt. SOC. Am., 1982,72,912. Pianarosa, P., Demers, Y., and GagnC, J.-M., J. Opt. SOC. Am., 1984, B11, 704. Pianarosa, P., Demers, Y., and GagnC, J.-M., Spectrochim. Acta, Part B, 1984, 39, 761. Chevalier, G., unpublished results, Ecole Polytechnique, Montreal, Canada, 1983. GagnC, J.-M., and Chevalier, G., Internal Report, Ecole Polytechnique, Montreal, Canada, 1985. Mehs, D. M., and Niemczyk, T. M., Appl. Spectrosc., 1981,35, 66. Palmer, B. A., Keller, R. A., and Engleman, R., Jr., LA 8251 Report, Los Alamos National Laboratory, Los Alarnos, NM, USA, 1980. Palmer, B. A., and Engleman, R., Jr., LA 9615 Report, Los Alamos National Laboratory, Los Alamos, NM, USA, 1983. Keller, R. A., Englernan, R., Jr., and Zalewski, E. F., J . Opt. SOC. Am., 1979, 69, 738. Mitchell, A. C. G., and Zemanski, M. W., “Resonance Radiation and Excited Atoms,” Cambridge University Press, New York, 1971. Leblanc, B., Carleer, M., Demers, Y., and GagnC, J.-M., Appl. Opt., 1980, 19,463. Phelps, A. V., and Molnar, J. P., Phys. Rev., 1953,89, 1202. Phelps, A. V., Phys. Rev., 1955,99, 1307. Cherrington, B. E., “Gaseous Electronics and Gas Lasers,” Pergamon Press, New York, 1979. Bouchard, P., PhD Thesk, Ecole Polytechnique, 1985. Demers, Y., MSc Thesis, Ecole Polytechnique, 1980. GagnC, J.-M., Bouchard, P., and Gleizes, A., Can. J. Phys., 1985,63, 1506. Corliss, C. H., J. Res. Nutl. Bur. Stand., Sect. A, 1976, 80, 1. Paper JA.512 Received October 4th, I985 Accepted August 4th, 1986