High rotational excitation of molecular hydrogen in plasmas

Transcription

1 Chemical Physics Letters 400 (004) High rotational excitation of molecular hydrogen in plasmas P. Vankan *, D.C. Schram, R. Engeln Department of Applied Physics, Eindhoven University of Technology, P.O. Box 513, 5600 MB Eindhoven, The Netherlands Received 7 October 004; in final form 6 October 004 Available online 11 November 004 Abstract Highly rotationally excited H molecules are observed in a hydrogen plasma. The rotational distribution, which is measured up to J = 17 in the vibrational state v = is highly non-thermal. Whereas the lower rotational levels (J < 5) are distributed at the ambient temperature, the high rotational levels (J > 7) are distributed at a much higher temperature, close to the apparent vibrational temperature. The ro-vibrational distribution is compared to distributions observed in other plasma sources and all of the rotational distributions show strikingly similar non-thermal distributions. Of all the possible production mechanisms surface processes are strongly suggested to be responsible for the observed high rotational excitation. Ó 004 Elsevier B.V. All rights reserved. The rotational and vibrational populations of molecular hydrogen in plasmas have been the subject of a wide range of experimental studies [1 8]. Most of these studies were aimed at a better understanding of the vibrational kinetics, which are now rather well established, even though perhaps not every detail is known. Especially the volume reactions leading to vibrational excitation and involving vibrationally excited molecules are well understood [0]. On the subject of surface reactions and its influence on vibrational excitation, extensive theoretical studies have been performed as well as a number of experimental studies [4,9,10]. In contrast to the work on vibrational kinetics, reactions involving rotationally excited H are still poorly understood. Although the H molecules often are highly rotationally excited [1,,7,8], the influence of the rotational excitation is not well known. Consider, for example the charge exchange process between H + and H which plays a crucial role in the volume recombination in the divertor region of fusion plasmas [11]. * Corresponding author. address: p.j.w.vankan@tue.nl (P. Vankan). This process is greatly enhanced by vibrational energy of the initial H molecule [1]. Cross-sections have been calculated as function of the initial vibrational state [1], but the dependence on rotational excitation is unknown, even though it could well have a considerable influence. The process of dissociative attachment of electrons to H leading to the formation of negative H ions, for example, is known to be enhanced five orders of magnitude by increasing the internal energy of the H molecules to.5 ev [13]. Rotational excitation is almost as effective in enhancing the process as vibrational excitation. By considering both the rotational and vibrational energy in the dissociative attachment reaction, instead of solely the vibrational energy, we calculate an increase in the source function of a factor of more than 0. Rotational populations of H in plasmas are known to be non-thermal [1 3,7]. The low rotational levels (J < 5) are populated according to a Boltzmann distribution at ambient temperature, but the higher rotational levels are extremely overpopulated compared to the Boltzmann population given by the lower levels. We measured, for example, overpopulations of more than seven orders of magnitude for the high rotational levels of the vibrational state v = /$ - see front matter Ó 004 Elsevier B.V. All rights reserved. doi: /j.cplett

2 P. Vankan et al. / Chemical Physics Letters 400 (004) Rotational populations have been measured in various plasmas, using several techniques. In the so-called multicusp source, rotational populations of the lower vibrational states have been measured by Pealat et al. [1] using coherent anti-stokes Raman scattering (CARS) and by Stutzin et al. [,5] using vacuum-uv absorption spectroscopy. Mosbach et al. [3] measured rotational populations in the higher (v = 11 13) vibrational levels of molecular hydrogen in a comparable multipole source using laser induced fluorescence (LIF). All of these studies were conducted using different plasma parameters. Ro-vibrational populations are furthermore measured in our group [6,7] in the expansion from a thermal arc using both CARS and LIF. All of these rotational populations showed the same non-boltzmann character. This raises the question whether this similarity points at a generic rotational excitation mechanism. Apart from in plasmas, ro-vibrational populations have also been measured in studies of the recombinative desorption of hydrogen molecules from a copper surface [10,14]. These rotational populations also showed very similar non-boltzmann distributions. In this Letter, we will first present rotational distributions in several vibrationally excited states, that were measured in the expansion from a thermal arc. We will show that the measured distributions approximately reflect the distributions, as they were actually produced. We will then compare these ro-vibrational distributions to the distributions, measured in other plasmas, and we will discuss the possible production mechanisms leading to the observed high rotational excitation. Finally we will show that surface recombination is most likely responsible for the production of the high rotationally excited hydrogen molecules in plasmas, and that inelastic collisions with surfaces could also have an influence. The rotational and vibrational distributions are measured in an expanding hydrogen plasma. The plasma and the measurement system are described extensively elsewhere [7]. In short, a 1 ev plasma is created in a high current cascaded arc, of which a cross-section is shown schematically in Fig. 1. The arc channel with a diameter of 4 mm and a length of 50 mm is formed by the central holes in four insulated cascade plates and is connected to a vacuum vessel via a nozzle with a constant diameter of 4 mm. The plasma expands supersonically from the source where the pressure is 1 kpa, into a vacuum vessel, which is kept at 0 Pa. The expansion is adiabatic, so the temperature decreases to values below 0.1 ev. The ro-vibrationally excited H molecules are measured in the expansion using multiplexed laser induced fluorescence (m-lif) in the vacuum-uv spectral range. A tunable, frequency doubled dye-laser beam of around 30 nm is used to pump a Raman cell in which the so-called stimulated anti-stokes Raman scattering process takes place. In this process up to 10 anti-stokes laser beams are produced and each subsequent laser beam is shifted Fig. 1. Cross-section of the cascaded arc plasma source. The dotted lines represent current flow lines ð~jþ. In the nozzle, there is no power dissipation anymore, leading to efficient production of H rv. down in wavelength. In this way, tunable, narrow-band laser radiation can be produced down to 115 nm. This laser radiation is used to excite hydrogen from a ro-vibrational level in the electronic ground state to the B 1 R þ u state. Using the resulting fluorescence spectrum the density of the initial level can be determined. The measurements have been corrected for all of the experimental parameters (laser power, optical transmission, detector efficiency), the linearity has been checked, and the densities have been calibrated using CARS measurements [6]. The distributions are measured on the expansion axis at a distance of 50 mm from the plasma source. The number of momentum exchange collisions from the source to the detection volume is 0. This number of collisions can partly relax the lower rotational levels (J < 5), but it is insufficient to alter the vibrational distribution or the distribution of the higher rotational levels. Therefore, the measured distribution reflects the population distribution leaving the source. More details on the evolution of the rotational and vibrational populations in the expansion can be found in a foregoing paper [8]. In this letter we will only use the fact that the distribution at 50 mm approximately reflects the distribution emerging from the plasma source, especially for the higher rotational levels (J > 5) and we will focus on the production mechanism of this high rotational excitation. Two typical rotational populations are shown in Fig., where the rotational population in the v = and v =3 at 50 mm from the source are plotted. These populations can not be described by a normal Boltzmann distribution. The low rotational levels (J < 5) are distributed according to temperatures of 850 K for the v =, which is approximately the ambient temperature and 600 K for the v = 3. However, the density in the high rotational levels (J P 7) can not be described by this temperature. The density of the level v =, J = 17 for example is seven orders of magnitude higher than expected from a

3 198 P. Vankan et al. / Chemical Physics Letters 400 (004) Fig.. Rotational distributions of the vibrational states v = and v = 3 in the hydrogen plasma expansion at 50 mm from the source. For both distributions, the density per statistical weight is plotted as function of the rotational energy. The background pressure was 0 Pa. Fig. 3. The distribution of H rv as function of the internal energy. The data is reproduced from [1,], and this work. Boltzmann distribution at 850 K. The densities in the high J levels are distributed at a temperature of 900 K, which is close to the vibrational temperature of 600 K. The non-boltzmann rotational distributions have also been measured by others [1 3,6]. Studies have been conducted in different plasmas and using different techniques. In Fig. 3 the population distributions measured by Pealat et al. [1] and Stutzin et al. [] are compared to the population distributions that are presented here. The most striking feature of the figure is perhaps the similarity of the distributions. Although the population distributions have been measured in completely different plasmas, using different settings for example for the pressure and temperature, the shape of the rotational distributions has, in all cases, the same non-boltzmann character. The low rotational levels are distributed at the low ambient temperature, whereas the high J levels are distributed at a much higher temperature. Furthermore, not only the shape of the distributions is very similar, also the absolute densities are comparable to within a factor of 3. This similarity between the measured rotational distribution both in shape and in absolute density may point to identical rotational production mechanisms. Another feature of the population distribution is that similar non-boltzmann rotational distributions are observed in all of the vibrational levels, in which a rotational distribution was measured, i.e. v = 0 6 [1,,7] and v = 11 1 [3] (the only exception forms the highlying state v =13 [3], that may be too close to the dissociation limit). Using these measured population distributions and more particular the rotational distributions we will underpin the idea that surface recombination is the main process in the formation of the non-boltzmann rotational distributions in hydrogen plasmas. Three different possible production mechanisms for the high rotational excitation of H can be distinguished: the first possible process to be considered is the rotational excitation by electrons. Although vibrational excitation via electrons is very well possible (E-V and e-v processes), rotational excitation by electrons is very inefficient. Since the nuclear spin is conserved, rotational transitions occur in multiples of DJ = ±, and typical cross-sections are smaller than 10 3 m, even for the most efficient rotational transitions with DJ = ± for low J [15]. Using an electron density of 10 1 m 3 which is the electron density encountered in the center of the arc channel, and an electron temperature of 0.8 ev, a typical time for the rotational excitation of DJ =+ s ej =(n e r ej v e ) ls can be calculated, whereas the typical transit time in the arc is less than 1 ls. This is by far not enough to produce the observed high rotational excitation, certainly not when taking into account that multiple excitation steps are needed. In the other discussed plasmas, the electron density is even much lower than 10 1 m 3, in the order of m 3, leading to much longer excitation times, whereas the loss times by diffusion are smaller than 10 ls. Therefore, we conclude that electron excitation is not the main mechanism behind the high rotational excitation of H in plasmas. The second possible process is the dissociative recombination of H þ 3 e þ H þ 3! H ðv; JÞþH ð1þ This process has to compete with the dissociative recombination into three hydrogen atoms, which is.5 times as effective in the energy range under consideration [16]. For analysis, one should consider the production process of H þ 3 : atomic hydrogen ions charge exchange with H, forming H þ. which is followed by proton transfer from a H molecule, forming H þ 3. However Hþ 3 is only produced, if the electron density is low enough (<10 19 m 3 ) so that the proton transfer is able to compete with the dissociative recombination of H þ. In the

4 P. Vankan et al. / Chemical Physics Letters 400 (004) arc the ionisation degree is approximately 0.03 and this is even decreased in the nozzle, where the electron density is still high enough for the dissociative recombination of H þ to be efficient. This reduces the amount of ions available for H þ 3 production by at least a factor of 5. This would mean a H rv content of at maximum 0.%, which is considerably lower than the observed ratio. Furthermore, the first charge exchange reaction is endothermic and therefore only efficient if the initial H molecules posses around ev of internal energy. Given the branching ratio, this would mean a depletion of levels with internal energy around ev and an increase of molecules having higher internal energy, which has not been observed in any of the experiments. So we conclude that it is very unlikely, that H þ 3 -recombination is the main reaction mechanism leading to rotational excitation. The third process is surface association of hydrogen atoms H plasma þ H surf! H ðv; JÞ ðþ In this process, two hydrogen atoms recombine at a surface near the plasma, forming a hydrogen molecule. Depending on the exact reaction mechanism and the exact potential energy of the hydrogen atoms at the surface, 1 4 ev of reaction energy becomes available to excite the molecule rotationally or vibrationally. Several studies have been performed to determine the population distribution of the formed molecules. Hall et al. [9] and Eenshuistra et al. [17] showed that part of the formed molecules were highly vibrationally excited, which has been reproduced by several theoretical simulations, see for example [18]. Despite the limited amount of observed rotational levels, Eenshuistra et al. already hinted upon non-boltzmann distributions. This was confirmed by Rettner et al. [10] and Kubiak et al. [14], who measured that the rotational distribution of H recombinatively desorbing from a copper surface exhibited non-boltzmann characteristics. The shape of these rotational populations compares quantitatively with the rotational populations measured in plasmas. The higher rotational levels are populated according to the vibrational temperature whereas the lower rotational levels are approximately populated according to the surface temperature. Furthermore, the distribution starts to significantly deviate from Boltzmann at approximately the same rotational energy, E rot 0.5 ev as in the plasma distributions. This similarity in the population distributions strongly suggests that surface recombination of H atoms is the main production process for the high rotational excitation of the H molecules in hydrogen plasmas. In the case of the cascaded arc, the H rv molecules are produced at the surfaces of the nozzle, see Fig. 1. In the cascaded arc, an atomic hydrogen plasma channel is maintained by power dissipation. However, at the entrance of the grounded nozzle the power dissipation ends, and depending on the nozzle length, the H atoms collide a number of times with the nozzle surface. The high atomic fluxes towards the nozzle surfaces (>10 4 ML/s) together with the high surface recombination coefficient (c 0.1 1) are responsible for the production of the rotationally and vibrationally excited H. This argument is quantitatively supported by a comparison of the typical transit time in the nozzle and the typical diffusion time of hydrogen atoms towards the nozzle surface. The transit time is in the order of ls, whereas the typical diffusion time of H atoms towards the surface is in the order of 0.5 ls. So every H atoms collides several time with the walls, and has a chance of around 0.75 to be converted into ro-vibrationally excited H. Since at least 50% of the particles in the arc channel are H atoms, this could easily lead to the observed amounts of rotationally excited H. This is furthermore confirmed by the observed decrease in atomic flux of a factor 13, when increasing the nozzle length by a factor of, which is caused by the increased number of surface recombination reactions. An effect that should be considered in all of the aforementioned studies is, that the produced H rv has collided a number of times, both with a surface before being measured. This could alter both the rotational and vibrational distributions via inelastic collisions. Karo et al. [19] state that a few surface collisions suffice to transfer almost half of the vibrational energy to rotational and or translational degrees of freedom. We emphasize, that in the plasmas discussed here, even though the circumstances are very different, e.g. in terms of pressure and temperature, the observed distributions are strikingly similar. The only comparable experimental parameter is the Kundsen number, which is a measure for the number of surface collisions, compared to volume collisions. Therefore, in plasmas, where particles collide relatively frequently with the surfaces surrounding the plasma, wall processes dominate the excitation to higher rotational levels. This effect could furthermore shed some light on the production mechanism of negative hydrogen ions in volume sources. In conclusion, we have measured rotational populations in different vibrational states of molecular hydrogen in a hydrogen plasma, and compared them to populations measured in other plasmas. All of the rotational populations that are reported in literature show the same non-boltzmann character, in which the high rotational levels are highly overpopulated. We have compared the importance of the various processes that could possibly lead to the high rotational excitation, and of all the reaction mechanisms, surface recombination of H atoms contributes by far the most to the observed non-boltzmann rotational populations.

5 00 P. Vankan et al. / Chemical Physics Letters 400 (004) Acknowledgements This work is supported by the Dutch Foundation for Fundamental Research on Matter (FOM) and the Euratom Foundation. References [1] M. Péalat, J.-P.E. Taran, M. Bacal, F. Hillion, J. Chem. Phys. 8 (1985) [] G.C. Stutzin, A.T. Young, H.F. Döbele, A.S. Schlachter, K. Leung, W.B. Kunkel, Rev. Sci. Instrum. 61 (1990) 619. [3] T. Mosbach, H.-M. Katsch, H.F. Döbele, Phys. Rev. Lett. 85 (000) 340. [4] J.H.M. Bonnie, P. Eenshuistra, H.J. Hopman, Phys. Rev. A 37 (1988) 111. [5] G.C. Stutzin, A.T. Young, A.S. Schlachter, K. Leung, W.B. Kunkel, Chem. Phys. Lett. 155 (1989) 475. [6] R. Meulenbroeks, R.A.H. Engeln, J.A.M. van der Mullen, D.C. Schram, Phys. Rev. E 53 (1996) 507. [7] P. Vankan, S.B.S. Heil, S. Mazouffre, R. Engeln, D.C. Schram, H.F. Döbele, Rev. Sci. Instrum. 75 (004) 996. [8] P. Vankan, D.C. Schram, R. Engeln, J. Chem. Phys. 11 (004) [9] R.I. Hall, I. Čadež, M. Landau, F. Pichou, C. Schermann, Phys. Rev. Lett. 60 (1988) 337. [10] C.T. Rettner, H.A. Michelsen, D.J. Auerbach, J. Chem. Phys. 10 (1995) 465. [11] R.K. Janev, Contrib. Plasma Phys. 38 (1998) 307. [1] A. Ichihara, O. Iwamoto, R.K. Janev, J. Phys. B: At. Mol. Opt. Phys. 33 (000) [13] J.M. Wadhera, Phys. Rev. A 9 (1984) 106. [14] G.D. Kubiak, G.O. Sitz, R.N. Zare, J. Chem. Phys. 83 (1985) 538. [15] J.B. Hasted, Physics of Atomic Collisions, Butterworth, London, 197. [16] R.K. Janev, W.D. Langer, K. Evans, Elementary Processes in Hydrogen Helium Plasmas: Cross-sections and Reaction Rate Coefficients, Springer, Berlin, [17] P.J. Eenshuistra, J.H.M. Bonnie, J. Los, H.J. Hopman, Phys. Rev. Lett. 60 (1988) 641. [18] B. Jackson, M. Pearson, J. Chem. Phys. 96 (199) 378. [19] A.M. Karo, J.R. Hiskes, R.J. Hardy, J. Vac. Sci. Technol. A 3 (1985) 1. [0] M. Capitelli, R. Celiberta, F. Esposito, A. Laricchiuta, K. Hassouni, S. Longo, Plasma Sources Sci. Technol. 11 (00) A7.