State-selective studies of the associative desorption of hydrogen from Pd(100) and Cu(100)

摘要

Faraday Discuss., 1993,96, 55-65 State-selective Studies of the Associative Desorption of Hydrogen from Pd(100) and Cu(100) L. Schroter, Chr. Trame, J. Gauer and H. Zacharias Fachbereich Physik, Universitat Essen, 0-4300 Essen I, Germany R. David IG V,Forschungszentrum Jiilich, D-5170Jiilich, Germany W. Brenig Physik Department, T U Miinchen, 0-8046 Garching, Germany Vibrational-state populations and velocity distributions of H, , HD, and D, desorbing from Pd( 100) are measured with rotational state selectivity over a wide temperature range from T, = 325 to 825 K. At all surface temperatures the vibrational populations, increasing exponentially with T,, are found to be significantly higher than those expected for thermally equilibrated mol- ecules. The slopes of Boltzmann plots are considerably lower than expected for a thermal excitation mechanism of the vibrational states at the corre- sponding gas-phase energies.They also show a non-trivial isotope effect. The velocity distributions from clean Pd(100) are found to be Maxwell- Boltzmann-like. The translational energies of H, molecules are accommo- dated at the surface temperatures, whereas those of D, are higher than kT, by ca. 10-30 meV. Both the vibrational excitation and the isotope effect in translation can be understood with a quantum-mechanical model calcu- lation on a two-dimensional potential-energy surface. The vibrational-state selective angular distributions for desorption from Cu( 100) display a differ- ent behaviour for ground and excited states.The distributions for the vibra- tionally excited states are broader than that of ground-state molecules. They also show an isotope effect. The reaction dynamics of hydrogen on various metal surfaces has been the subject of many theoretical and experimental investigations. Detailed information has been gath- ered, particularly for two systems, hydrogen on copper and on palladium surfaces.'.' While copper shows a high thermodynamic barrier to dissociative adsorption and associative de~orption,~.~ Thethe adsorption on palladium is mainly n~n-activated.~.~ internal state-specific measurements of the flux of associatively desorbing molecules from Cu(ll1) and Cu(ll0) by Kubiak et aL7 revealed, for the vibrational degree of freedom, a 50-100 times higher population than expected thermally.A corresponding study for desorption from poly-Pd also showed an overpopulation of the vibration by an order of magnitude.' A basic difference between the two systems arises from the electronic structure of the metals. In particular, for the s metal, theoretical calculations show for dissociative adsorption of hydrogen a high barrier late in the adsorption channel.' Classical trajectory and quantum-mechanical calculations on two-dimensional reaction potentials confirmed the enhanced vibrational population observed in desorp- tion, and hence predicted a greatly enhanced sticking for vibrationally excited hydrogen 55 Associative Desorption of Hydrogen on copper.’ 0-14 In subsequent adsorption studies several groups investigated the influ- ence of the incident kinetic energy and angle and the contribution of vibrational excita- tion to the sticking probability in copper.15-18 Vibrational energy of molecules incident on the copper surface strongly enhances the dissociative sticking probability. The coup- ling of translational and vibrational degrees of freedom deduced from the adsorption measurements has been confirmed in desorption. The translational energy decreases by ca. 0.2 eV for each increase of the vibrational quantum in the desorbing D,. The rotational-state resolved kinetic energy varies also for each vibration in a non-monotonic fashion, becoming significant for J” 8.” This shows that the rotation of a molecule might also enhance the sticking in dissociative hydrogen adsorption.The adsorption of hydrogen on clean palladium shows mainly non-activated behav- iour without a significant thermodynamic barrier.6 Earlier state-integral measurements of the desorption flux from palladium’ yielded Maxwellian velocity distributions with TkinT,, in accord with the recent adsorption data. In this paper we report results of x internal state-specific measurements for the vibrational and translational degrees of freedom for the associative desorption of hydrogen and its isotopes from Pd( 100). The angular distribution of the flux of molecules being adsorbed or desorbed is commonly related to the dynamics of the interaction between the molecule and the surface.The existence of energetic barriers on the pathway from the adsorption state to the gas phase can be related to an angular distribution peaked in the direction of the surface normal.20 For the recombination reaction of hydrogen atoms on metal surfaces angular distributions strongly peaked in the normal direction have been observed for copper surfaces in de~orption~.~~ as well as in adsorption.6 For clean palladium surfaces, on the other hand, mainly cosine distributions were obtainede6 Berger and Rendulic recently found that the angular distribution of the sticking coefficient of H, on Cu(ll0) broadens with increasing temperature of the nozzle of the molecular beam source.22 This observation is interpreted to be due to the contribution of vibrationally excited molecules to the sticking.The number of vibrationally excited molecules increases with the temperature of the beam nozzle. Theoretical calculations of Kuchenhoff and Bre~~ig,~also predict, at a given kinetic energy, a considerably broader angular distribu- tion for the sticking of (u” = 1) molecules compared to ground-state molecules. In this paper we report first results on vibrational-state selectively measured angular distribu- tions for H, and D, desorption from Cu(100). Experimental A detailed description of the experimental method has been given previou~ly.~~ Briefly, the experiments are carried out in an ultra-high-vacuum (UHV) chamber with a base pressure of 3 x lo-’’ mbar (see Fig. 1). Order and cleanliness of the surface are deter- mined by low-energy electron diffraction (LEED) and Auger spectroscopy.Adsorbates are removed when necessary by soft Ar+ ion sputtering at low energy and current density. After sputtering the surface is annealed before an experimental run. Hydrogen atoms are supplied to the surface of the Pd(100) and Cu(100) crystals by permeation through the bulk of the radiatively heated sample. The thicknesses of the samples are 1 mm and 0.5 mm for Pd(100) and Cu(lOO), respectively. The experiments are performed at constant temperatures which range from 325 to 825 K for palladium, and at 885 K for copper. During desorption from palladium the pressure in the chamber is kept in the (6-30) x lo-* mbar range by adjusting the stagnation pressure at each temperature accordingly. Because of the much lower permeation rate of hydrogen through copper the corresponding chamber pressure rises only up to 3 x mbar at 500 mbar stag- nation pressure.This average hydrogen pressure in the chamber gives rise to a back- ground signal which has to be taken into account. L. Schriiter et al. 57 SHG Nd YAG laser 3I vuv detect ion generation nm 106-108 np Cu(100) Fig. 1 Schematic diagram of the experimental set-up For state-selective detection of desorbing hydrogen molecules resonantly enhanced two-photon ionization [(l + 1’) REMPI] is empl~yed.~’ The first excitation step from the electronic ground state XICS+(u”,J”) to the intermediate B’C:(u’, J’) provides the vibrational- and rotational-state selectivity.This Lyman band transition is excited by vacuum ultraviolet (VUV) laser radiation tunable in the 106-108 nm spectral range. The VUV radiation is generated by frequency tripling, in xenon gas, the frequency doubled output of a tunable dye laser. In the interaction region with the desorption flux about 5 x lo9 photons per pulse are available with a spectral bandwidth of 0.8 cm-’. Ioniza- tion of the electronically excited molecules is carried out with a second UV laser beam, either at a wavelength around 320 nm, the fundamental of the VUV radiation, or at 266 nm, the fourth harmonic of the primary Nd :YAG pumping laser. In either case a laser pulse energy of ca. 5 mJ in a 5 ns pulse is provided for the ionization step.The laser beams interact with the desorption flux at a distance of 20 mm in front of and in a plane parallel to the surface. For the measurement of the state-selected angular distributions the laser beams and the detector are located in a differentially pumped section of the apparatus (see Fig. 1). At a distance of 9 mm in front of the crystal a 2 mm diameter hole in the separating wall selects a 5O-wide section of the desorption flux. Angular distribu- tions are obtained by rotating the sample around a vertical axis. In this part of the chamber the base pressure is ca. 6 x lo-’’ mbar, which rises to ca. 5 x lo-’’ mbar during desorption. The generated hydrogen photoions are detected by a time-of-flight spectrometer using two-stage microchannel plate (MCP) amplification and a 50 R impedance-matched anode.The output of the MCP detector is processed by a counting electronic or a digital oscilloscope (LeCroy 9400) and transferred to a microcomputer (DEC, LSI 11/23). Results Vibrational Population In the desorption flux of hydrogen isotopes from the palladium surface single rovibra- tional states can be clearly resolved by (1 + 1’) REMPI spectroscopy.26 The relative heights of lines originating in X ‘c: (0’’= 1) being proportional to the population in Associative Desorption of Hydrogen (v” = l), to those from X ‘Xl (0’’= 0), representing the ground-state population, increases nearly exponentially with the surface temperature, T,. A Boltzmann plot of the logarithm of these intensity ratios for the three hydrogen isotopes H,, HD, and D, is shown in Fig.2 as a function of T,’. In this plot a nearly linear dependence on TSp1 over more than two decades of the population is observed. The behaviour of the vibrational degree of freedom during desorption can be described by the motion of the two atoms involved on a two-dimensional potential- energy surface (PES).’-I4 One coordinate of this PES is the reaction path, s, the other being the vibrational motion orthogonal to s. The dynamics of the recombination is determined by the curvature K(S) of the PES along the reaction coordinate s, and the diagonal potential V(s) along the reaction path. The vibrational frequency o(s) is assumed to vary along the reaction path.For the model calculation (see also ref. 13) we assume a smooth variation of the diagonal potential V(S)= E,/cosh2(h) (1) The maximum of this symmetrical potential at s = 0 is the nominal barrier height E,. II is a scaling parameter which describes the halfwidth of the curvature K(S) ~(s)= l/ro(s) = {ro cosh2[L(s -so)])-’ (2) where ro denotes the radius of curvature of the PES and so the position of the maximum curvature. The variation of the vibrational frequency along the reaction path is taken as W(S) = coo -A/cosh2(h) (3) where A is a parameter. Although different for all three isotopes, this parameter A shows the same relationship between isotopes as their vibrational frequencies in the gas phase = [IJ(pD2/PH2)lAD2 (4) where p denotes the reduced mass of the molecule: A,, = J2 AD, and similarly AH, = J(l.5) AD,Note that the parameters K and V of the PES are independent of the mass of the isotopes, while the vibrational frequency cu depends on the mass.Adsorbed deuterium atoms with a thermal distribution of energies at T,, i.e. a Maxwellian kinetic energy 1.0 2.0 3.0 1.0 2.0 3.0 1.0 2.0 3.0 103KIT, Fig. 2 Boltzmann plot of the relative vibrational excitation in hydrogen desorption from Pd(100) us. Ts-for (a)H, ,(b) HD and (c) D, L. Schruter et al. distribution and a Boltzmann population of the D-Pd bond, are allowed to recombine on this PES. The results of this model calculation are compared with the experimental data for D,. The parameters which fit the data best are shown in Table 1, yielding a height of the activation barrier of E, = 0.2 eV, and, for deuterium, A,, = 0.14 eV.The corresponding theoretical results for the other isotopes H, and HD are then calculated on the same PES, just changing the isotope specific values for the vibrational energy, (~(s),and the parameter A according to eqn. (4). A comparison of the vibrational popu- lation obtained from this model calculation, shown as lines in Fig. 2, with the experi- mental data reveals a very good agreement. Table 1 Parameters of the PES for the reaction of deuterium on Pd(100) ro/A A,,/eV EJeV E,,,/eV 2.75 0.5 0.25 0.14 0.2 0.13 The difference in zero-point energies of two adsorbed atoms and a desorbing mol- ecule leads to the interesting fact that part of this zero-point energy can be made avail- able for overcoming the barrier against recombination.Since these zero-point energies differ considerably for the different hydrogen isotopes, the effective barrier heights, Eeff, show also a significant variation. This effective barrier height can be calculated from Eeff= E, -$h[w(s= CO) -U(S = O)] (5) Eqn. (5) yields Eeff= 0.13 eV for D2,Eeff= 0.11 eV for HD, and Eeff= 0.10 eV for H, , The effective height of the barrier is now in the range of the thermal energy of the surface. The model calculation also predicts the vibrational population at a given T,. At T, = 677 K a value of 1.3% population in U” = 1 is obtained for D,, which can be compared with the experimental value2 determined by VUV laser-induced fluorescence to be (1.5 & 0.3)%.Velocity Distributions State-selective velocity distributions can be determined with REMPI by measuring the time-of-flight (TOF) from the ionization volume to a fast ion detector.28 Because of the short ionizing laser pulse duration of only 5 ns short flight distances of only a few cm are sufficient for an adequate velocity resolution for thermal molecules. In general, with pulsed REMPI, the measured TOF spectra, nl(t),are density-like, because in a short time interval a stationary volume-distributed part of the flux is probed.’ The distribution n,(t) has thus to be transformed from the density-time domain into the flux-velocity domain, Fflux(v). This can be done by the transformation where I dt/du I represents the Jacobi determinant.The average kinetic energy of a desorb- ing particle with mass rn is then obtained from The summation is carried out over all velocities ui. For Maxwellian velocity distribu- tions the experimental TOF data are more easily fitted directly in the time domain by a Associative Desorption of Hydrogen reverse-transformed Maxwell velocity distribution n,(t) =constant x t-4 exp[ -mL?/2kT,j, t2] (8) Gindenotes the temperature parameter of the distribtion, L is the length of the flight path, and k the Boltzmann constant. The mean kinetic energy of the desorbing mol- ecules is then related to the temperature parameter Kinby (Ekin)=2kT,,, . In order to check the measurement set-up Fig.3 shows a typical TOF spectrum of H, where the chamber is backfilled with hydrogen at a pressure of 9 x mbar. In this case a density-like Maxwell velocity distribution has to be fitted to the data. As can be seen the H, (u” =0, J” = 1) ion signal is well fitted with a velocity distribution with Tkin=300 K as temperature parameter. The additional ion signal at a flight time of ca. 9.5 ps can be assigned to H+ ions. These atomic hydrogen ions are produced from hydrogen molecules by multiphoton dissociation and ionization according to the follow- I I I I 1 I I I I (a) 0 2 4 6 8 10 12 14 16 18 20 TOF/ps Fig. 3 (a) Velocity distribution of a bulk sample of H, (d’=0,J” = 1) at room temperature; (b) enlarged portion of the TOF spectrum around the H+ atom peak.The solid line gives the expected TOF distribution for H atoms that dissociate from thermal H, molecules with Ekin= 0.433 eV. The data are fitted only when the velocity distribution of the H, molecules is also taken into account. T =300 K. L. Schrb;ter et al. 61 ing excitation pathway :29 H, X ‘Zgf(u”,J”)+ hv,,, -P H2 B ‘Z:(v’, J’) (14 H, B ‘Z;(v’, J’) + hvuv-+ HlX ’Z+ + e-(Ekin) (Ib) -+ H, Rydberg(v, J) (Ib’) H, Rydberg(v, J) + H,fX ’Z+ + e-(Ekin) (14 +HC(ls), ’s112)I+ HC(2s), 2S1/29 (2P), 2P1,2.3p)I (14 HC(2s), ’Slj2 (2P)2P1/~,7 3/2)I hvuv +H+ + e-(Ekin) (14 After the first state-selecting excitation step (Iu) the absorption of a UV photon (A 320 nm) may lead to the production of H; ions via direct ionization (Ib) or autoionization from an excited Rydberg state (Ic).For hydrogen, this last process has a significant pr~bability.~~,~’The highly excited Rydberg state may also dissociate into two neutral hydrogen atoms, one in the (Is), ,S1,, ground state, the other in the n = 2 excited state (Ic’). Absorption of a second UV photon from the laser beam now leads to ionization of only the n = 2 hydrogen atoms (Id). The processes (Ia, b’, c’, d) thus produce atomic hydrogen ions via (VUV + 2UV) REMPI. The dissociation of the hydrogen molecule from the Rydberg state (Id) is accompained by a transformation of electronic energy in excess of the dissociation threshold into kinetic energy of the two neutral atoms. From the spectroscopy of hydrogen this excess energy is well known for a given UV photon energy.In the present case for H, XIX,f(u”= 0, J” = 1) excited via the P-branch into €3 ‘E:(v’ = 3, J’ = 0) and a UV photon energy of 3.9 eV (R = 319.35 nm) this excess energy amounts to 0.866 eV, distributed equally over the two hydrogen atoms. The H+ ion TOF signal is thus a convolution of a sharp 0.433 eV kinetic energy from the disso- ciation and the velocity distribution of the parent H, ground-state molecules, which in this case is a Maxwellian velocity distribution with Tkjn = 300 K. It is evident from Fig. 3(b) that the fast ion signal can be fitted well by taking both the thermal energy of the parent H, molecules and the energy in excess of the dissociation threshold into account.The fast H+ ion signal can thus always serve as a cross-check to the fit of the velocity distribution of the slower H; ion signal. The relative intensity of the H+ and the HZ signals depends on (i) the probability of exciting a Rydberg state [process (Ib’) versus (Ib)], and (ii) the dissociation probability of this Rydberg state compared to its autoioni- zation probability [process (Ic’) versus (Ic)]. The relative height thus varies strongly with the intermediate B-state level excited and the wavelength of the UV laser beam. Fig. 4 shows a typical velocity distribution measured for D2 (u” = 1, J” = 2) mol-ecules desorbing from Pd(100) at T, = 683 K. It is evident that this distribution cannot be described by a Maxwellian distribution at T,, shown as a dashed line, but has to be fitted with a considerably higher temperature parameter Tkin.A best fit is obtained for Gin= 775 K. Such velocity distributions have been measured for many surface tem- peratures between T, = 440 K and 825 K. H, molecules desorbing from Pd(100) in the vibrational ground state with J” = 0 to 5, thereby spanning an internal energy range E,,, = 0-1740.2 cm-’ (0,216 eV), always show velocity distributions with Gin= T, within the experimental uncertainty. D, molecules also show a linear increase of (Ekin)/2kwith T,, however, with energy offsets which vary slightly for different internal states. Linear least-squares fits to the data assuming a constant slope of one and having as free parameter the intercept of the fits with the kinetic energy scale at T, = 0 K support this conclusion. The intercepts obtained from the least-squares fits are shown versus the internal energy E(v”,J”) in Fig.5 for each internal state. The dashed line at an intercept value I = 0 meV represents the case of thermal equilibrium at the surface temperature T,. The velocity distributions of €3, molecules (filled symbols) can well be Associative Desorption of Hydrogen (7111111111 20 40 60 80 100 120 140 160 180 TO F/ps Fig. 4 TOF distribution for D, (u” = 1, J” = 2) desorbing at T, = 683 K from Pd(100) I I I I I I I I 40 -””= 0 V’‘ = 1 30 -0 -20 -00 10-062-0. om---e----A ---------E B-10--.--20 --30 -I I I I I I I I I described as equilibrium distributions at T,? independent of the internal rotational state. For D,, however, intercepts of ca.10 f7 meV for D, (d’= 0, J” = 2 and 3) and 28 & 5 meV for D, (v” = 0, J” = 4 and 5) and the vibrationally excited D2 (v” = 1, J = 2) are obtained. The correlations of the fits with linear regressions of slope one are rather good (r = 0.85 to 0.99). Angular Distributions Fig. 6 shows in polar coordinates rovibrationally state-resolved angular distributions of the hydrogen and deuterium desorption flux from a Cu(100) surface. In all cases the Cu(100) surface was kept at a constant temperature of T, = 885 K. In Fig. 6(a) are plotted results for H, desorption in the vibrational ground state (v” = 0, J” = 5) (filled symbols) and for desorption in the vibrationally excited state (0’’ = 1, J” = 1) (open symbols).In Fig 6b the corresponding results for D2 desorption in (v” = 0, J” = 5) (filled symbols) and for D, (u” = 1, J” = 0 and 2) (open symbols) are shown. Least-squares fits L. Schriiter et al. 20 10 0 10 20 I‘ 111’ 20 I0 0 10 20 20 10 0 10 20 Fig. 6 Angular distributions of H, and D, desorbing in the vibrational ground and excited state from Cu(100) at T, = 885 K. (a) .,H, (u” = 0, J” = 5); n = 6.2; 0,H, (u” = 1, J” = 1); n = 5.0. (b)., D, (u” = 1, J” = 0) and 0,D, (0’’ = 0, J” = 5); n = 7.5; 0, D, (u” = 1, J” = 2); n = 5.7. of the data points to cos” 0 functions are shown as lines in the figures. For the ground- state molecules we obtain exponents for the cosine function of n = 6.2 for H, (J” = 5) and n = 7.5 for D, (J” = 5).These exponents are in agreement with those obtained earlier in state-integral measurement^.^-^' The corresponding exponents for the vibra- tionally excited states are n = 5.0 for H2 (u” = 1, J” = 1) and n = 5.7 for D, (u” = 1, J”’ = 0 and 2). It is evident that the angular distributions for molecules desorbing in the vibrationally excited state are considerably broader than for those in the ground state. As may be noticed in Fig. 6(b),the angular distributions for the different rotational states (v” = 1, J” = 0) and (u” = 1, J” = 2) of D, are fairly similar within the experimen- tal uncertainties. Under the assumption that the angular distributions do not change significantly with J” in both the ground and the vibrationally excited state, which might not be justified for weakly populated high-l” states, a correction factor for the total state population can be derived by integrating the angular distributions over 8.For H, (0’’ = 1) this integral is ca. a factor of 1.14 larger than for H, (d‘ = 0). The same numeri- cal value of 1.14 for this ratio D, (u” = 1)/D2 (u” = 0) is derived for deuterium. These factors should be applied as a correction when vibrational-state populations are deter- mined from measurements at @ = 0’. It should be noticed that a rotationally symmetric dependence of the desorption flux around the azimuth angle was assumed. Discussion For a molecule which shows a cosine angular distribution in desorption from palladium and a nearly Maxwellian translational energy distribution5 it might be expected that the internal degrees of freedom would also be in equilibrium with the surface.However, static thermal models were not able to explain the main observations: the temperature Associative Desorption of Hydrogen dependence of the vibrational population, the non-trivial isotope effect of the slopes of the Arrhenius plots, and the absolute magnitude of the vibrational p~pulation.~~.~~ Thus, a dynamical treatment of the H atom recombination is required. Without a realis- tic ab initio PES the calculations described above are performed on a model PES with a curved reaction path. Along this reaction path an energetic barrier against the recombi- nation of the hydrogen atoms with a Gaussian shape [eqn.(l)] is assumed. The width and the height of this barrier are fitted to the experimental data for D, desorption. The velocity distribution of the adsorbed atoms is taken to be thermal at T,, thus modelling the thermal character of the reaction. The other important ingredient in the calculation is the variation of the hydrogen vibrational frequency along the reaction path, account- ing for the fact that the atoms experience attractive forces after surmounting the maximum of the barrier. With these assumptions and an adjustment to the D, data the temperature dependence of the vibrational population in H, and HD can be predicted quite well, including the non-trivial isotope effect.From this model calculation it can be concluded that excess zero-point energy can be made available to overcome the energetic barrier. This leads to different effective barrier heights with Eeff= 100 meV for H, and 130 meV for D,, i.e. lower for hydrogen than for deuterium. It is then expected that deuterium molecules desorb with a higher kinetic energy than hydrogen moleules. The translational energy measurements confirm this expectation. For H, complete accommodation of the kinetic energy to T, is observed, whereas for D, an additional energy offset of 10 to 30 meV compared to Ekin= E, is found. This supports the existence of a small barrier against recombination also on a palladium substrate, although the absolute value of the kinetic energy is over- estimated by this model PES even when a van der Waals attraction of the hydrogen molecules by the surface is included in a discussion of kinetic energies.Recent results for desorption from Cu( 1 1 1) show also that D, molecules have systematically higher kinetic energies than H, molecules in the corresponding internal state^.^ For desorption from Pd(100) neither molecule shows a significant dependence of the translational energy on the rotational or vibrational state. This is in constrast to recent measurements for the desorption from Cu(l1 1),19931 where a strong decrease in &in with increasing vibra- tional state, and in addition with increasing J” for J” 8, especially in the ground vibrational state, has been observed.For the palladium substrate the coupling between the internal degrees of freedom and the translation seems thus to be much weaker than for copper, which, in view of the small barrier, is not unexpected. Measurements of the state selective-angular distributions from Cu(100) support this strong coupling between internal and translational degree of freedom. The angular dis- tributions for (u” = 1) states are significantly broader than those for ground-state mol- ecules. Berger and Rendulic22 measured recently the sticking coefficient of H, on Cu(ll0) for cold seeded hydrogen beams. At an incident kinetic energy of 0.2 eV the angular dependence of the initial sticking coefficient was found to be proportional to cos” 8, with n = 6.6 for a nozzle temperature of TN= 1200 K and n = 4.3 for TN= 1700 K.At the higher nozzle temperature the amount of vibrationally excited H, (u” = 1) in the beam is considerably larger (population ca. 2.9%) than at the lower TN(less than 0.7%), when a cooling of the vibrational degree of freedom during the beam expansion is neglected. From these experimental data it is qualitatively evident that vibrationally excited molecules show a broader angular distribution in sticking than ground-state molecules. Theoretical calculations by Kuchenhoff and Brenig23 show that at TN= 1700 K and &in = 0.15 eV the sticking of H, on copper is dominated by (u” = 1) molecules. A cosine distribution with n = 4.5 is obtained. A similar result is found for D, with an exponent of n = 6.5 for (d’= 1) and &in = 0.1 eV.Both the experimental and the theo- retical sticking coefficients show distributions in qualitative agreement with the observa- tions presented here for desorption. Also, the same isotopic behaviour is found in that the D2 distributions remain narrower than those of H, in the respective vibrational L. Schrater et al. state. Differences between the sticking measurements and calculations and the desorp- tion data arise from the distribution of kinetic energies present in the desorption flux. In this latter case the mean energies range from 0.31 eV (v” = 1) to 0.55 eV (0‘’ = 0) for H, and from 0.46 eV (v” = 1) to 0.62 eV (0’’ = 0) for D, ,31 These energies are thus consider- ably larger than the energies where tunnelling of H atoms through the barrier becomes important (below 0.1 eV), which, according to the theoretical calculations, generates a narrow H, angular distribution.We are grateful for financial support by the Deutsche Forschungsgemeinschaft (Za 110/2). References 1 G. Comsa and R. David, Surf: Sci.Rep., 1985,5, 145. 2 K. Christmann, Surf: Sci. Rep., 1988,9, 1. 3 G. Comsa and R. David, Surf: Sci., 1982,117,77. 4 G. Anger, R. Winkler and K. D. Rendulic, Surf Sci., 1988,220, 1. 5 G. Comsa, R. David and B. J. Schumacher, Surf: Sci., 1980,95,210. 6 K. D. Rendulic, G. Anger and A. Winkler, Surf Sci., 1989,208,404. 7 G. D. Kubiak, G. 0.Sitz and R. N. Zare, J. Chem. Phys., 1985,83, 2538. 8 H. Zacharias and R.David, Chem. Phys. Lett., 1985, 115, 205; L. Schroter, R. David and H. Zacharias, Appl. Phys. A, 1986,41,95. 9 J. Harris, T. S. Rahman and K. Yang, Surf: Sci., 1988, 198, 312. 10 J. Harris, S. Holloway, T. S. Rahman and K. Yang, J. Chem. Phys., 1988,89,4427. 11 J. Harris, Surf: Sci., 1989, 221, 335. 12 M. R. Hand and S. Holloway, J. Chem. Phys., 1989,91,7209; Surf: Sci., 1989,211/212,940. 13 W. Brenig, S. Kuchenhoff and H. Kasai, Appl. Phys. A, 1990,51, 115. 14 S. Kuchenhoff, W. Brenig and Y.Chiba, Surf: Sci., 1991,245, 389. 15 B. E. Hayden and C. L. A. Lamont, Phys. Rev. Lett., 1989,63, 1823. 16 B. E. Hayden and C. L. A. Lamont, Surf: Sci., 1991,243,31. 17 C. T. Rettner, D. J. Auerbach and H. A. Michelsen, Phys. Rev. Lett., 1992,68, 1164. 18 H.F. Berger, M. Leisch, A. Winkler and K. D. Rendulic, Chem. Phys. Lett., 1990,175,425. 19 H. A. Michelsen, C. T. Rettner and D. J. Auerbach, Phys. Rev. Lett., 1992,69,2678. 20 W. van Willigen, Phys. Lett. A, 1968,28, 80. 21 C. T. Rettner, H. A. Michelsen, D. J. Auerbach and C. B. Mullins, J. Chem. Phys., 1991,94,7499. 22 H. F. Berger and K. D. Rendulic, Surf: Sci., 1991,253,325; H. F. Berger, Ph.D. Thesis, Graz, 1992. 23 S. Kuchenhoff and W. Brenig, Surf. Sci., 1991,258, 302. 24 L. Schroter, R. David and H. Zacharias, Surf: Sci., 1991,258,259. 25 W. Meier, H. Rottke, H. Zacharias and K. H. Welge, J. Chem. Phys., 1985, 83, 4360; W. Meier, H. Rottke and H. Zacharias, Znst. Phys. Con$ Ser., 1988,94,93. 26 L. Schroter, S. Kuchenhoff, R. David, W. Brenig and H. Zacharias, Surf: Sci., 1992, 261,243. 27 L. Schroter, H. Zacharias and R. David, Phys. Rev. Lett., 1989,62, 571. 28 L. Schroter, G. Ahlers, H. Zacharias, R. David, J. Electron Spectrosc. Retat. Phenom., 1987,45,403. 29 H. Zacharias, Appl. Phys. A, 1988,47,37. 30 G. Herzberg and Ch. Jungen, J. Mol. Spectrosc., 1972, 41, 425; Ch. Jungen and 0. Atabek, J. Chem. Phys., 1977,66,5584. 31 C. T. Rettner, H. A. Michelsen and D. J. Auerbach, J. Vac.Sci. Technol.,in the press. Paper 3J031425; Received 28th May, 1993