Energy and angular dependence of H - (D - ) ions produced by dissociative electron attachment to H 2 O(D 2

Transcription

1 Journal of Physics B: Atomic and Molecular Physics Related content Energy and angular dependence of H - (D - ) ions produced by dissociative electron attachment to H 2 O(D 2 O) To cite this article: D S Belic et al 1981 J. Phys. B: At. Mol. Phys Dissociative electron attachment to metastable oxygen (a 1 g ) D S Belic and R I Hall - Resonances in dissociative electron attachment to water N Bhargava Ram, Vaibhav S Prabhudesai and E Krishnakumar - Dissociative electron attachment on H 2 S: energy and angular distributions of H - ions R Azria, Y Le Coat, G Lefevre et al. View the article online for updates and enhancements. Recent citations - Anion emission from water molecules colliding with positive ions: Identification of binary and many-body processes J.-Y. Chesnel et al - Ion-momentum imaging of resonant dissociative-electron-attachment dynamics in methanol D. S. Slaughter et al - A practical approach to temperature effects in dissociative electron attachment cross sections using local complex potential theory Yuji Sugioka and Toshiyuki Takayanagi This content was downloaded from IP address on 23/08/2018 at 05:13

2 J. Phys. B: At. Mol. Phys. 14 (1981) Printed in Great Britain Energy and angular dependence of H-(D-) ions produced by dissociative electron attachment to H20(D20) D S BeliCi, M Landau and R I Hall Groupe de Spectroscopie par Impact Electronique et Ioniquef, Laboratoire de Physique et Optique Corpusculaires, UniversitC Pierre et Marie Curie, 4 Place Jussieu, Tour 12, E 5, Paris CCdex 05, France Received 14 July 1980, in final form 15 September 1980 Abstract. Energy and angular distributions of H-(D-) ions formed by dissociative attachment in H20(D20) have been observed for the processes situated in the region of 6.5,8.6 and 11.8 ev. The angular behaviour would indicate that the symmetries of the resonant states responsible for these processes are 'B1, 'Al and 'B2, respectively. The energy distributions of the H-(D-) ions for all three processes show that most of the dissociation energy is in the form of translation and furthermore the distributions for each process and isotope are similar, This similarity would indicate that the potential surfaces and the autodetachment widths of the HzO- states are also about the same which would support the proposals and calculations of Jungen, Vogt and Staemmler who predict that these three states are formed by two electrons in 3sal orbitals bound to an HzO+ core with a vacancy in the lbl, 3al and lb;? orbitals, respectively. 1. Introduction There have been many observations of negative ion formation by electron impact on H20 (see, for example, Melton 1972). Most of them studied the integral cross sections and were concerned with the energy location of the different structures and the identification of the negative ion species produced. These experiments show that three principal processes exist, occurring with maxima at 6*5,8*6 and 11.8 ev. The first two have been reported to yield H-, 0- and OH- ions, whereas only 0- and OH- ions have been detected for the 11.8 ev process. Further information on these processes can be obtained from a knowledge of the energy and spatial distributions of the fragments. When dissociative attachment takes place in a triatomic molecule the excess energy is distributed among the different degrees of freedom of the fragments in a way that depends on the conformation of the resonant state. Furthermore, the symmetry of the resonance influences the spatial distribution of the fragments. Hence the observation of the kinetic energy and angular behaviour of, for instance, the H- ions formed in HzO yields information on the electronic, vibrational and rotational excitation of the OH radical as well as on the symmetry of the dissociating states. t Permanent address: Faculty of Natural Sciences and Institute of Physics, PO Box 57, Beograd, Yugoslavia. $ Equipe AssociCe au CNRS (ERA 703) /81/ The Institute of Physics 175

3 176 D S Belik, M Landau and R I Hall The first indications that the H- ions were formed with a large amount of kinetic energy were given by Schulz (1960) who came to the conclusion that the first two processes dissociated to OH and H- in their ground states. This was supported by Trajmar and Hall (1974) for the 6.5 ev process who showed that very little energy was contained in vibration and rotation of the OH fragment. These authors also observed the angular behaviour of H- which indicated that this process proceeds through a B1 state of H20-. An electron impact spectrometer (Hall et a1 1973) similar to that used by Trajmar and Hall (1974) has since been adapted more specifically to the observation of negative ions by the addition of a small mass filter (Hall et a1 1977, Schermann et a1 1978). In the present study the instrument has been further improved by the addition of a multicapilliary array for generating the HzO beam which increases the experimental resolution. This instrument has been used to observe in detail the energy and angular distribution of H- ions produced by the processes at 6.5 and 896eV. The energy partitioning among the vibrational and rotational states of OH is clearly observed. The angular distribution for H- at 8.6 ev indicates the presence of a A1 resonant state. In addition H- ions have been observed for the process at 11.8 ev and their intensity is approximately 600 times weaker than the 6-5 ev process. This resonance dissociates to the first electronic state of OH, A21;+, and the angular distribution of H- ions is compatible with Bz symmetry. Similar observations were also made for DzO. 2. Experimental The apparatus has been described in detail previously (Hall et a1 1977, Schermann et a1 1978). It consists of a crossed beam configuration and uses 127 electrostatic filters to produce the incident beam and analyse in energy and angle the negatively charged particles generated by the collision processes. The negative ions are then separated from the electrons by a small magnetic field before detection and storing with the usual channel multiplier and multichannel analyser assembly. In the experiments described here the instrument has been modified by the replacement of the metal tube which forms the gas beam by a multicapilliary array. This has the effect of improving the energy resolution with which the dissociation process can be observed. This is because the energy spread of the observed fragment distributions is dominated by broadening caused by the thermal motion of the target molecules. Chantry and Schulz (1967) have shown that this broadening in a gas cell is given by the expression AW = (11PkTW) (1) where W is the fragment energy, P the ratio of fragment to molecule mass, T the gas temperature and k Boltzmann s constant. This expression gives, in a typical case, A W = 480 mev for a homonuclear diatomic molecule at room temperature with fragment energy 1.5 ev. This compares with an instrument resolution which can be as low as 50 mev. We have studied the thermal broadening in detail by observing the energy distribution of 0- ions produced by dissociative attachment to 0 2 in the 6-5 ev region for the following cases: the background gas in the apparatus, a gas beam effusing from a 1 mm tube and a multicapilliary array. The equivalent temperatures seen by the

4 Dissociative attachment to H electrons in the collision plane obtained using the above formula are: 300 K (background), 50 K (tube) and 20 K (array). Hence the equivalent temperature is considerably reduced by replacing a gas cell by a beam and is further reduced in going from a tube to an array. But in view of the square root dependence on the temperature the reduction in spread is less spectacular for the case mentioned above, which is to say that for W = 1 a5 ev the thermal spread is A W = 190 mev for a tube and A W = 115 mev for an array compared to 480 mev for a cell. For the case in point the broadening of the H-(D-) distribution issued from dissociation of H20-(D20-) is lower due to its small mass compared with the molecule. With a beam formed by an array (T = 20 K) the H- ions have a spread given by the formula of 44 mev when their nominal engrgy is 2 ev. Therefore, the thermal broadening is no longer the dominant effect in determining the resolution of observation. In fact, total instrumental resolutions in the neighbourhood of 150 mev were employed to obtain a good signal and statistics and is sufficient to separate peaks in the H-(D-) distribution corresponding to vibrational excitation of OH or OD (i.e. 420 and 295 mev respectively). The array diameter was a in and its thickness 5 mm with 200 holes/"'. It was made of stainless steel and was vacuum annealed to eliminate residual magnetism. The array was very conveniently mounted in a non-magnetic Monel Swagelock union. The electron beam intersected the gas beam about 10 mm from the array surface; closer to the surface the equivalent temperature was observed to be nearer that of effusive flow from a tube. The energy scale was calibrated against the N2 resonance at ev observed in the v = 1 channel at 90". This gas was permanently admixed with the H20(D20) throughout these experiments. The H20 beam was formed from double-distilled water and dissolved gases were boiled off under vacuum. The heavy water used here was manufactured in Norway during the second world war. No trace of any other negative ions were detected during the observations. In particular, 0- from O2 if present would have shown up very clearly. 3. Results 3.1. First process at 6.5 e V Figure 1 shows the energy distributions for H- ions (not on the same scale) observed at 90" for 6.0, 6.5, 7.0 and 7.5 ev incident energies. The relationship between the H- energy (W), the incident electron energy (Ei) and the internal energy of the OH fragment (E,) is given by: W = (1 -P)(Ei- D -En) (2) where D is the dissociation energy of H20- to H- and OH and P = m/m where m and M are the H and HzO masses, respectively. The vertical lines represent the OH vibrational thresholds and were calculated using the spectroscopic constants given by Maillard etaz(l976) and the above relationship. The analyser optics were focused at an H- energy of 1.7 ev and were maintained for all the spectra of figure 1. Some distortion of the spectra is introduced by the variation of the transmission function with fragment energy. However, we estimate from observations on electrons that with the optics settings used, the transmission variation from 1 to 3 ev is less than 20%. As is indicated by the OH vibrational thresholds, the structure corresponds to vibrational excitation of

5 178 D S BeliC, M Landau and R I Hall H- energy (ev I Figure 1. Energy distribution of H- ions produced in H20 at the indicated incident electron energies (Ei) and at a 90" scattering angle. The energies corresponding to excitation of the vibrational levels (U) of OH are indicated. the OH fragment. At 6.0 ev all the excess energy is in the form of translational energy and the OH fragment is left in the o = 0 level. As the incident energy increases higher vibrational levels of OH are populated and at 7.5 ev excitation of the o = 6 level can be weakly observed. However, the larger part of the dissociation energy is always carried away by the fragments in the form of kinetic energy. The spectra of figure 1 were obtained with a resolution of about 140 mev. Other spectra were obtained with a resolution better than 100 mev and the structure was not resolved any better. Furthermore, deconvolution of a spectrum taken with an apparatus function of 100 mev did not improve the peak separation. Hence, the conclusion was reached that the observed broad peaks were physical and the reason was that rotational excitation of OH was also present. In addition the rotational population maximum was for J values above zero as the peaks are shifted away from the vibrational thresholds. Assuming that only o = 0 is populated at 6 ev we have taken this peak, which is about 350 mev wide, to represent the rotational distribution within each vibrational packet and have reproduced the experimental vibrational spectra with a superposition of these packets. In this way the contribution of each vibrational packet is obtained as shown in figure 2. This method is not really correct since the rotational distribution which makes up the vibrational packet varies with H- energy as will be shown below. However, the above assumption is sufficiently good compared with the experimental accuracy to give meaningful results. Thus, assuming that the angular behaviour is independent of the OH vibrational level and incident energy, the cross section for

6 Dissociative attachment to H Figure 2. The broken curve is the energy distribution of H- ions formed in H20 at 6.5 ev incident energy and 90". The full curves were obtained by a convolution procedure and correspond to vibrational levels of OH. The most intense peak is the U = 0 level. excitation of the vibrational levels of OH as a function of electron energy were evaluated by normalisation to the absolute integral cross sections for H- formation measured by Melton (1972) and are displayed in figure 3 on a semi-logarithmic scale. As mentioned above, rotational excitation of OH occurs as a result of the dissociation process and is indicated by the broad peaks and the shift of the peak positions from the vibrational thresholds. This can be seen more clearly in figure 4 which shows the H- distribution at 7.5 ev and 90". The thick vertical lines represent the J = 0 level for each vibrational state of OH and the thin lines the different rotational series. From this figure it is clear that the peak maxima do not occur at the same value of J and depend on the value of the H- energy. For high values of H- energy the maxima are located around J = 7 and as the energy decreases the peaks move to lower J values. It can also be seen that for the v = 0 level, the peak moves to higher J as the incident energy and consequently the H- energy increase. In order to obtain further information on the rotational distributions the convolution procedure was used to determine the relative contributions of each J level to the U = 0 level packet as obtained from the previous convolution of the 6.5 ev spectrum. The results are shown in figure 5. The apparatus function was determined by assuming that the form of the decaying edge of the H- distribution outside the group of the first four J levels is one half of the apparatus function; the complete function is then obtained by symmetry. This technique is obviously very approximate but gives a reasonable idea of the rotational level distribution. Here, for the U = 0 level, the maximum was found to occur at J = 7 where the H- energy is 1.9 ev and for the v = 4 level (see figure 4) it can be estimated to occur at J = 4 or 5 with a corresponding H- energy of 1.2 ev. The energy distribution of D- produced by dissociative attachment in DzO has also been observed in the 6.5 ev region. The distributions obtained at four energies at 90" are displayed in figure 6. The vibrational levels of OD shown in the figure were obtained from those given by Herzberg (1965). The important thing to notice from these spectra is that there is no spectacular structure compared with H20 but some

7 180 D S Belic', M Landau and R I Hall x10-'8 0.01' 60 I Incident electron energy lev) Figure 3. The broken curve is the total cross section for H- formation in HzO and the full curves are the partial cross sections for excitation of the vibrational levels (U) of the OH fragment. weak structure corresponding to OD vibration can be seen particularly at 7.5 ev. However, the envelopes are very similar to those for H20 and show that vibrational excitation is again inexistent at 6 ev but increase with incident energy. The instrumental resolution is sufficiently good to separate the OD vibrational levels. The thermal broadening is slightly larger for D- but the overall resolution is still of the order of 150 mev compared with an OD separation on the D- energy scale of 294 mev. Hence, rotational excitation is again present and is more important for OD than for OH. If we assume that the peak at 6-0 ev is a pure o = 0 level then the maximum occurs for about J = 12 with an energy of 190 mev compared to J = 7 and 70 mev for H Second process at 8.6eV This process for H- formation is about five times smaller than the one at 6.5 ev. The energy distribution of H-(D-) fragments has been observed and the results obtained at 8.5eV and 50" for H20 and 90" for D20 are shown in figures 7(a) and 7(b), respectively. In this case the analyser optics were focused for H- energies of 3 ev. The broken curves represent an estimation of the contribution from the first process which is still weakly present at this energy. The contribution to the D20 spectrum at 90" is larger as can be understood from the angular dependencies discussed below. In the same way as for the first process most of the dissociation energy is in the form of fragment kinetic energy. The envelopes are smooth for both isotopes and what is more, have almost exactly the same form. If anything the D- curve maximum occurs at

8 Dissociative attachment to H H- energy lev1 Figure 4. Energy distribution of H- ions produced in H20 at 7.5 ev and 90" indicating the energies corresponding to excitation of vibrational (U) and rotational (J) levels of OH. ON rotol~onal energy ievl Figure 5. The chain curve is the energy distribution of H- ions formed in H20 at 90" corresponding to the OH, U = 0 level obtained from the convolution procedure of figure 2. The full curve is the convolution of the apparatus function with the indicated rotational level (J) intensities. a slightly lower value of the fraction of total possible energy (U = 0) indicating a slightly larger fraction of rovibrational (i.e. rotational and vibrational) excitation for OD than for OH. Again the absence of vibrational structure indicates the presence of rotational excitation and this would appear to be stronger than for the 6.5 ev process.

9 182 D S BeliC, M Landau and R I Hall D- energy lev1 0 Figure 6. Energy distribution of D- ions produced in DzO at the indicated incident electron energies (El) and at a 90" scattering angle. The energies corresponding to excitation of the OD vibrational levels (0) are indicated Third process at 11-8 ev H- ions produced by this process have not been observed previously and the reason for this is the weakness of the cross section. Here it has been estimated to be about 600 times smaller than that at 6-5 ev. These ions have been detected here because firstly, the instrument is very sensitive and secondly, which is the more important, the fragment energy is analysed. This process can be observed in the presence of the two others in the spectrum presented in figure 8. This spectrum was obtained by fixing the analysis energy at 3 ev, which corresponds to the distribution maximum for this process, and by sweeping the incident energy. The factor of 600 mentioned above was obtained from such a spectrum taking into account the different angular behaviour of the processes and the fact that the first process peak is at 7.5 ev and has to be related to its maximum at 6.5 ev. The energy distribution of H- and D- from the two isotopes at 12 ev incident energy and 90" are shown in figures 9(a) and 9(b), respectively. This process dissociates to the first excited state of OH (A 'Z+) and a ground-state H- ion with a limit at 8.38 ev. Thus, the vibrational levels shown in the figure are those corresponding to this state and were obtained from Herzberg (1965). Here also the major part of the excess energy is in the form of kinetic energy although the distributions are much narrower than in the two previous cases. No vibrational structure can be seen indicating the presence of rotational excitation of the OH(0D) radical. The distributions are very similar for the two isotopes and it would appear from the peak positions that the proportion of rovibrational energy contained in the diatomic fragment is the same for each isotope although it does not correspond to the same vibrational level in each case.

10 Dissociative attachment to HzO 183 Ib) D-/ 40 e I goo E,: 8.5 ev 1 I I 1 I\\ L Ion energy lev1 Figure 7. Energy distribution of H- ions and D- ions produced in (a) H20 at 50" and (b) D20 at go", for an incident electron energy of 8.5 ev. I e ui - C?? e c 9 - x - VI a e E! I 3.4. Angular distributions Electron energy lev) Figure 8. Spectrum obtained at 90" by fixing the H- analysis energy ( W) at 3.0 ev and sweeping the incident electron energy. The angular behaviour of the H-(D-) intensity taken at the distribution'maximum for an incident energy of 6.5 ev is shown in figure 10. These intensities have been corrected to compensate for distortion by the instrumental angular transmission by calibration against known elastic differential cross sections in helium (Andrick and Bitsch 1975); as can be seen these observations are in excellent agreement with those of Trajmar and Hall (1974). The angular distribution of H- for the second process is shown in figure 1 l(a). This was obtained at 8.8 ev and corresponds to the v = 0 level of OH and not to the distribution maximum. This level was chosen to reduce the contribution of the first process; however, a slight contribution is still present and is indicated by the small hump near 90".

11 184 D S BeliC, M Landau and R I Hall H-/ HzO E, =12 0 ev e = 90" Io1 i D-/ D20 E, 2 12 OeV e = 90" 6 L 2 O+v IIII II I L Ion energy lev1 Figure 9. Energy distribution of H- ions and D- ions produced in (a) H20 and (b) DzO, at 12 ev and 90". The energies corresponding to excitation of OH and OD vibrational levels (0) are shown. The angular behaviour of the third process at 12 ev is shown in figure 1 l(b). This is the behaviour of the H- intensity taken at the distribution maximum. 4. Discussion 4.1. H20- state symmetries and configurations The angular behaviour in resonant scattering processes is determined directly by the state symmetries and the number of partial waves. As the number of partial waves is usually limited to one or two at the most due to the almost atomic nature of the resonant electron orbitals, the differential cross section often takes on simple forms and can be interpreted to give information about the symmetry of the resonant state. The symmetry of the first resonance at 6.5 ev was proposed to be 2B1 by Trajmar and Hall (1974) on the basis of the angular behaviour of the H- ion and simple symmetry arguments first put forward by Dunn (1962) for diatomic molecules extended to the case of triatomic molecules. These arguments only give indications on the cross sections at specific angles relative to the incident beam. More complete information can be obtained from the theory contained in the paper by Azria et a1 (1979) who made observations similar to those described here for the analogous H2S molecule. These authors took the theory of O'Malley and Taylor (1968) and developed it using group theory to account for angular behaviour for dissociative attachment in HIS. They obtained the characteristic shapes of the angular distribution for all four CZv symmetries of the resonance formed from the AI ground state for each of the first two allowed partial waves. The theory of Azria et a1 (1979) confirms the 'B1 symmetry and furthermore indicates that the process is purely p wave, the first allowed partial wave. Theory gives a

12 Dissociative attachment to H * 1 Scattering angle Idegl Figure 10. Angular distribution of H-(D-) ions formed in HzO(D20) at 6.5 ev: +, HzO; A, DzO; 0, Trajmar and Hall (1974). c 5 1 lbl,=i2 0 ev bell-shaped curve centred at 90" similar to that observed. The slight shift of the maximum of the experimental curve to 100" can be explained by the presence of direct potential scattering (mostly in the elastic channel) which distorts the incident electron wave from the plane wave considered in the theory (see Tronc et a1 1977).

13 186 D S Belic', M Landau and R I Hall A survey of the excited states and configuration of H20 leads to the same conclusions as above. The first excited state, 3B1 lies at 7.0 ev (Chutjian et all975) and has a (1bl)-'3sal configuration. By analogy with atoms and diatomic molecules (see the review of Schulz 1973) the 6.5 ev resonance would then be a Feshbach resonance with the 3B1 state as parent and would be formed by the capture of the incident electron into the 3sal orbital to form a (lbl)-1(3sa1)22b1 state. In other words, the resonance is the H20+ 2B1 ground state (grandparent) with two 3sal Rydberg electrons. This result also came out of the H20- state calculations of Weiss and Kraus (1970) andn Winter (1975, private communication). The two states which correlate to H- and OH in their ground states are 'A2 or 'B1 and 'A1 or 'B2. As 2B1 was assigned to the 6-5 ev process this leaves either 2A1 or 'B2 for the process at 8.6 ev. A survey of the theoretical curves indicates that of these two states, only 2A1 can lead to an angular behaviour similar to the one observed. Also, the contributing partial waves would be s and d, the d wave yielding the double maxima at 30" and presumably 150". There is little sign of a p-wave contribution although no fitting procedure was attempted. Fitting three partial waves (that is to say five parameters) to the experimental curves would not lead to unique and meaningful results, especially as we cannot precisely estimate the contribution of the 6.5 ev process in the 90" region. Weiss and Krauss (1970) also gave 'A1 symmetry for this resonance with a (3al)-'(3sal)' configuration, that is, the resonance is formed by adding two 3sal electrons to the first excited state of H20+ 'A1. Hence the parent state is g3a1 (3aJ13sal which is situated at 9-3 ev (Chutjian et a1 1975) and leads to a binding energy of about 0.7 ev for the resonant electron. The angular behaviour of H- ions obtained for the 11-8 ev process which dissociates to the first excited A 'Z+ state of OH is characteristic of a 'B2 resonance. The theoretical curves of Azria et a1 (1979) indicate that the main contribution comes from the d wave which produces the 90" peak and not from the first allowed p wave. As the second excited state of H20+ is 'B2 then, by analogy with the other two processes for H- formation, we can assume that it is the grandparent of the 11-8 ev resonance formed by again adding two 3sal electrons. This would then give 'Bz (lb2)-1(3sa1)2 for the resonant state. The 3B2(lb2)-1(3sal) parent state has been located at 1103 ev (see Chutjian etal 1975) but would be too low if the above resonant configuration is correct. Instead, if a reasonable electron affinity of 0.5 ev is taken, it should be situated in the 12.3 ev region. Jungen et a1 (1979) also interpreted these three dissociative attachment processes as being due to Feshbach resonances formed by the binding of two 3sal electrons to the 2 2 BI, A and 'B2 states of H20+ formed by ejecting lbl, 3al and lb2 electrons, 2 respectively. These authors calculated the energies of the 'B1, Al and 'B2 states thus formed and obtained reasonable agreement with the experimentally observed energies. In the analogous HzS molecule, Azria et a1 (1979) obtained similar angular dependencies for the first two processes leading to H- formation. These authors came to the conclusion that the resonant states were 'B1 and 'A1 respectively, which is the same situation as that occurring here in H20. However, in their observations there is an interesting and intriguing feature which has not yet received an explanation. Azria et a1 (1979) observed in the case of the first process in HzS at 6 ev that the peaks in the H- spectrum corresponding to HS in the v = 0 and 1 levels had different angular behaviour, the 'B1 symmetry mentioned above being obtained from that of the v = 0 level. This phenomenon was carefully looked for in these experiments but the H- energy distribution at 6.5 ev remained unchanged with observation angle.

14 Dissociative attachment to H Energy distribution of H- The complexity involved in interpreting fragment energy distributions resulting from dissociative attachment in triatomic molecules can be understood from the calculations of the Orsay group. This group performed classical trajectory calculations firstly on a repulsive surface (surface with a saddle point) and were able to describe the mechanisms which produce energetic H- ions in HzO at 6.5 ev (Goursaud et a1 1976) and 0- formation in COz at 8.2 ev (Sizun and Goursaud 1979) qualitatively. Secondly, Goursaud et a1 (1978) have studied dissociative attachment on an attractive surface (surface with a potential well) in comparison with a repulsive surface. The physical ideas contained in the above calculations can be used to give a schematic description of the dissociative attachmeat process. The fragment energy distribution can be considered to depend on the form of the potential surface of the resonant state and its lifetime with respect to autodetachment. In a classical description the potential surface determines the different trajectories which lead to dissociation and the time they take to get there. These trajectories also determine the energy partitioning between translation, vibration and rotation of the fragments. The lifetime then specifies which trajectories survive to dissociation by imposing a sort of time limit, eliminating those which take too long. Hence, the kinetic energy distribution depends on the characteristics of those trajectories which survive to dissociation Firstgrocess at 6.5 ek As described above this resonance has 'B1 symmetry and contains electrons in excited 3sal Rydberg orbitals. These orbitals are strongly correlated with 4al orbitals which are strongly O-H antibonding and produce the repulsive nature of this state. No complete calculations of the surface have been done, however Weiss and Krauss (1970) indicate that it should be similar to the analogous 3s1B1 states which are the parents of this resonance. Calculations exist for the 3B1 state (Hosteny et a1 1970) and show a strongly repulsive surface with a saddle point minimum in the symmetric stretch coordinate at an OH separation 20% greater than that of the ground state. This minimum was roughly estimated to be about 0.5 ev below the potential at the point corresponding to the ground-state minimum which is at about 6-5 ev. Thus the gradient at this point is low in the symmetric stretch direction whereas it should be steeper in the asymmetric coordinate as the surface goes to the H--OH limit more than 2 ev lower at 4.36 ev. The potential energy is almost flat with bending, decreasing slightly by about 0.08 ev when going from 105 to 180". No vibrational structure is seen when the 321B1 states are excited by electron impact (Chutjian et a1 1975) indicating that the 3sal-4al mixing leads to dissociation to OH and H before vibration can occur in the other modes. Information on the autodetachment width (r) has been obtained from observations of the isotope effect on the total cross section for this process by Tronc (1973). The effect was small (v(h-/hzo): v(d-/dzo) = 1.25 : 1) indicating quite a long lifetime and these authors derived a value of r = 0.15 ev from a simple model of dissociative attachment. This width corresponds to a lifetime of 2 X s which compares with a symmetric stretch vibrational period of 1 x s for the HzO ground state. Thus, the lifetime and the dissociation time in the resonant state can be expected to be of the same order. In fact the calculations of Goursaud et a1 (1976) indicate that the probability of HzO- surviving to dissociation is about 15%. The experimental observations can now be discussed with reference to the available knowledge about the 2B~ state presented above. The energy distributions of figure 1 can be accounted for in the following way. The spectrum at 6 ev corresponds to the

15 188 D S BeliC, M Landau and R I Hall survival of those few trajectories which go straight to dissociation without any excursion in the symmetric stretch coordinate. All the other slower trajectories are eliminated by autodetachment. Here the H20- state is formed at a point which is not very steep and is probably not far from the saddle point minimum. As the incident energy increases to 6-5 ev the resonance is formed higher up at a steeper part of the surface and more energy is available in the dissociating system. Here, not only the direct trajectories survive. Some trajectories which make a slight excursion in the stretching coordinate are fast enough to escape the electron decay. These latter trajectories leading to a small amount of OH vibrational excitation. Similarly, at 7.0 and 7.5 ev, the nuclei have more and more energy available as they are formed higher up on the surface and those trajectories which make large symmetric stretch motions are now fast enough to escape to dissociation. Thus the OH vibration further increases at these incident energies. The D- distributions of figure 6 show the same evolution with incident energy as those of H- and the same simple qualitative explanations as above can be invoked. However, this picture runs into difficulties when the two isotopes are compared. In the above reasoning one would expect less vibrational excitation for OD than for OH if one makes the reasonable assumption that does not alter. As the D- moves more slowly on the surface only the more direct trajectories yielding little or no vibrational excitation should survive. In fact the OD vibrational energy is as high if not slightly higher than that of OH. The reason for this effect is not obvious and can only be given by complete calculations. The experimental observations also indicate that the OH fragment is rotationally excited. This excitation appeared to depend on the H- kinetic energy: as the latter increases so does the rotational energy. However, this rotational energy appeared to be larger for OD than for OH. At first sight rotational excitation can be expected to result from dissociation of H20-. The OH radical has its centre of mass slightly displaced from the oxygen atom so if one assumes that dissociation is along the OH bond then in a simple classical reasoning the dissociating H- exerts a force on the oxygen atom and sets the OH rotating. Also, this rotation would increase with dissociation energy as the more energy there is available the more energy is shared with rotation. Furthermore, this effect would be more important for D20 as the OD centre of mass is further from the oxygen atom than that of OH. One can also imagine in this scheme that a tendency to change angle during the dissociation process will lead to rotational excitation and this excitation would be more intense the steeper the surface in the bending coordinate. As discussed above the B1 surface probably only has a weak slope towards 180 and does not play an important role in the rotational excitation of OH. However, it should be again pointed out that the above discussion stems from a very simplified picture. The understanding of the mechanisms which lead to rotational excitation of OH represents a complex dynamical problem and only complete calculations on accurate potential surfaces can give the true physical origins of the observed effects Second process at 8.6eV. No theoretical surface calculations of this 2 A1(3al)-1(3sa1)2 state of H20- are available. However, the presence of the two 3sal electrons would again indicate that the surface is strongly repulsive. This surface should be the analogue of the 3 A1(3al)-13sal states. While little is known about the triplet, the singlet state is observed in photon absorption (Herzberg 1967) and the spectrum indicates strong dissociation to H + OH but some weak vibrational structure can be seen and is attributed to bending vibrations. The AI and B1 states are part of a Renner

16 Dissociative attachment to H surface and have the same energy at 180". As described above the B1 surface is expected to be flat with angle thus the A1 surface would be quite steep with a variation of about 2 ev when going from the bent ground state to the linear geometry. This steep surface could account for the observed bending vibration. The autodetachment width of this resonance is probably bigger than that of the 6.5 ev process as the isotope effect for dissociative attachment is larger for this process. This effect was observed by Tronc (1973) who obtained the ratio c(h-/h20) : c(d-/d20) = 2.2 : 1 compared with 1.25 : 1 at 6.5 ev. This higher width possibly also explains the lower cross section although it would still correspond to a lifetime of the same order as a vibrational period. As for the first process the energy distribution of the H-(D-) ions (figure 7) are compatible with a repulsive surface as most of the dissociation energy is in the form of fragment kinetic energy. However, the smooth envelopes of the distributions with the absence of vibrational structure indicate strong OH(0D) rotational excitation. Whether vibrational excitation is present as well is not clear but the sharp onset at high fragment energy would indicate that this is so. If the distribution corresponded to pure rotational excitation a sharp onset would not be expected. Consequently, with regard to vibrational excitation, the situation is similar to the 6.5 ev process and the same discussion of the mechanisms can be employed. If vibrational excitation is in fact occurring then another similarity would be that both isotopes have about the same proportion of dissociation energy contained in this mode. As pointed out above, this is the opposite to what one would expect from simple reasoning. The main difference with the lower process, however, is in the large amount of rotational excitation. This phenomenon can be correlated to the high gradient of the surface in the bending coordinate and corresponds to a large bending motion being imposed on the fragments.during dissociation leading to OH(0D) rotational excitation Third process at 11.8 ev. The experimental H-(D-) distributions shown in figure 10 are very similar although somewhat narrower than those obtained for the two previous processes. This infers that this 2B2(lb2)-1(3sa1)2 state of H20- has a surface similar to the other two dissociating states and that this is a consequence of the two electrons in antibonding 3sal orbitals. Furthermore, the rotational excitation of OH(0D) which prevents the vibrational structure from being resolved would indicate that some bending occurs during the dissociation. No other information is available on the potential surface of either this state or its parents as they have not been observed by other means. 5. Conclusion The observation of H- ions produced by the process of dissociative attachment in H20 has led to the identification of the resonant state symmetries corresponding to the processes situated in the region of 6.5 and 8.6 ev as well as to that of a third process 2 occurring near 11.8eV. These states have 'B1, Al and *B2 symmetries and thus confirm the predictions and calculations of Jungen eta1 (1979). These three states are composed of two 3sal electron and an H20t core formed by the ejection of an electron from each of the first three valence orbitals of H20, that is to say lbl, 3al and 1b2 respectively.

17 190 D S BeliC. M Landau and R I Hall The 3sal orbitals are strongly OH antibonding and would appear to lead to similar potential surfaces for all three resonances, this being born out by the fact that the H- energy distributions are very similar for all three processes. That is to say, surfaces which have a saddle point whose minimum is located at an OH separation greater than that of the ground-state minimum and which are repulsive in the asymmetric coordinate allowing very little motion in the symmetric stretch mode. The three surfaces only differ from each other in some details such as more or less gradient in bending if this can be considered to correlate to the observed rotational excitation of the OH(0D) fragments. Also some other details of the surface or autoionisation width should account for the narrow H-(D-) energy distribution observed with a peak energy which does not correspond to the U = 0 level of OH A 'X+. An intriguing feature of these measurements which is common to all three processes is constituted by the observation that the proportion of rovibrational energy of OD is always as large or even larger than that of OH. The total D-/D20 cross section is inferior to that of H-/H20 and in the picture where dissociative attachment is considered to be a competition between autoionisation and dissociation one would expect the lower cross section for D-/D20 to result from the elimination compared with H20 of those trajectories which can be considered to be slow and which lead to fragment vibrational excitation. Instead it appears that the abundance in each trajectory is reduced by an equal amount when going from H20 to D20 or even perhaps those leading to vibration are slightly less attenuated in D20 compared with H20. References Andrick D and Bitsch A 1975 J. Phys. B: Atom. Molec. Phys Azria R, Le Coat Y, Lefevre G and Simon D 1979 J. Phys. B: Atom. Molec. Phys Chantry P J and Schulz G J 1967 Phys. Rev Chutjian A, Hall R I and Trajmar S 1975 J. Chem. Phys Dunn G H 1962 Phys. Rev. Lett Goursaud S, Sizun M and Fiquet-Fayard F 1976 J. Chem. Phys J. Chem. Phys Hall R I, eadei I, Schermann C and Tronc M 1977 Phys. Rev. A Hall R I, Joyez G, Mazeau J, Reinhardt J and Schermann C 1973 J. Physique Herzberg G 1965 Molecular Spectra and Molecular Structure vol 1 (Princeton, NJ: Van Nostrand) Electronic Spectra and Electronic Structure of Polyatomic Molecules (Princeton, NJ: Van Nostrand) Hosteny R P, Hinds A R, Wahl A C and Krauss M 1970 Chem. Phys. Lett Jungen M, Vogt J and Staemmler V 1979 Chem. Phys Mailland J P, Chauville J and Mantz A W 1976 J. Molec. Spectrosc Melton C E 1972 J. Chem. Phys O'Malley T F a!d Taylor H S 1968 Phys. Rev Schermann C, Cadei I, Delon P, Tronc M and Hall R I 1978 J. Phys. E: Sci. Instrum Schulz G J 1960 J. Chem. Phys Rev. Mod. Phys Sizun M and Goursaud S 1979 J. Chem. Phys Trajmar S and Hall R I 1974 J. Phys. B: Atom. Molec. Phys. 7 L Tronc M 1973 Thesis Universiti Paris XI, Orsay, France Tronc M, Fiquet-Fayard F, Schermann C and Hall R I 1977 J. Phys. B: Atom. Molec. Phys Weiss A M and Krauss M 1970 J. Chem. Phys