Faraday Discuss., 1993,96,245-254 Dynamics of the Gas/Liquid Interface from Laser Molecular Beam Scattering Anthony J. Kenyon, Anthony J. McCaffery," Cristina M. Quintella and Mohammed D. Zidan School of Molecular Sciences, University of Sussex, Brighton, UK BNI 9QJ We report the results of a laser molecular beam scattering study of liquid surfaces. Molecular iodine is used as a probe molecule and its number density and quantum-state distribution is determined before and after scat- tering from the surface of a liquid by laser-induced fluorescence (LIF). Time resolution of the LIF signal allows us to separate processes of direct inelas- tic scatter from trapping followed by desorption. Here we focus mainly on the latter process and obtain data on the kinetics of desorption and the activation energy of desorption together with insights into the dynamical processes that lead to trapping and desorption. The gas/liquid interface is of considerable importance in many areas of science and of technology.Although liquids have been the subject of numerous studies' there are still many unanswered questions relating to the structures and dynamics of liquid surfaces. Much experimental work has centred on the determination of liquid surface structures, using X-ray, neutron or optical scattering methods. Only recently have experimental methods been developed that probe the dynamics of this important interface. This con- tribution describes the results of a state-resolved scattering experiment through which the kinetics and the dynamics of adsorptive and desorptive processes at liquid surfaces are simultaneously probed.Molecular beam scattering methods are now widely used in the investigation of solid surfaces.2 These techniques have been central to the study of gas-phase reaction and collision dynamics for several decades and their application to the gas/liquid interface is a relatively straightforward extension of established methods. The liquid investigated must be of sufficiently low vapour pressure that the molecular beam interacts principally with the liquid surface i.e. collisions with the vapour in equilibrium with the liquid do not give a dominant contribution. Liquids having vapour pressures below lop4 Torrt will generally fulfil this criterion and there are many atomic and molecular liquids that possess vapour pressures, at accessible temperature, lower than this value.Problems such as surface contamination, atomic and molecular motions at the surface etc. have features in common with solid surface studies. The methods used to obtain ultra-clean solid surfaces are not readily transferred to liquids, however, techniques have been devised that ensure a fresh liquid surface is constantly presented for e~periment,~.~ thus permitting the study of uncontaminated surface molecules. Two recent studies of liquid surface dynamics using atomic and molecular beam scattering indicate that valuable data on the intermolecular forces at the gas/liquid inter- face may be obtained by these methods.Nathanson and co-~orkers~.~ have used time- of-flight (TOF) velocity resolution techniques, including angular dependence, to analyse t 1 Torr = (101 325/760) Pa. 245 Laser Molecular Beam Scattering the scattering of rare gas atoms and inert molecules off the surfaces of organic liquids. The liquids used were 2,6,10,15,19,23-hexamethyltetracosane (squalane) and per-fluorinated polyether (PFPE), the incident atoms predominantly confronting the F and H atoms of C-F and C-H bonds. This work established that the scattering pattern comprised two main signals, one a sharp, fast peak and the second a slower, broader component. The positions and relative sizes change with incident and with exit angle and show a distinctive dependence on both probe atom and liquid.In analogy to similar experiments on solid surfaces the fast signal was identified with an inelastic scattering (IS) process whilst the slower signal reflects trapping followed by desorption (TD). Time and quantum-state resolved scattering of diatomic molecules confirms this In these experiments, I, molecules from a supersonic jet in Ar carrier gas were scattered off the surface of polydimethylsiloxane (PDMS), squalane and PFPE. Quantum-state populations of the I, molecules were determined, as a function of time, before and after interaction with the surface by means of LIF. The most recent results of this study are the subject of this contribution but it is relevant to the context of this section to describe briefly some features of earlier work.As with the TOF ~tudies~.~ two distinct signals are seen which have very different spectral and temporal signatures. Time resolution in these spectroscopic experiments is by delay of the LIF laser probe relative to the opening pulse of the supersonic nozzle. With short delay times, the scattered molecules show little rotational or vibrational excitation. These are inelastically scattered molecules interacting with the surface in a manner analogous to gas-phase collisions in which the repulsive intermolecular forces dominate. As time delay is increased, molecules that have undergone more extensive interactions with the surface are ~bserved.~~~ These molecules have been rotationally and vibrationally warmed whilst trapped on the surface. The desorbing species were found invariably to have internal temperatures well below that of the ~urface.~3~ There are striking similarities in the behaviour of scattered species from liquid and from solid2 surfaces.Furthermore there are significant differences from (as well as simi- larities to) patterns observed in rotationally and vibrationally inelastic gas-phase scat- teringg Both TOF data and state-resolved data indicate that scattering experiments are sensitive to the chemical nature of the surface and to the probe species. We anticipate therefore that carefully designed scattering experiments can throw new light on the dynamics of the gas/liquid interface. Experimental A pulsed nozzle valve (initially fuel injector, later superceded by Newport BV100) is mounted vertically in the top of a stainless-steel chamber, the interior of which holds a translatable stage on which a container of liquid stands.The stage is equipped with heaters and a thermocouple in the liquid is used to maintain temperatures to within 0.3 K. The liquid may be placed directly below the nozzle or may be moved completely out of the expanding jet stream. The chamber is equipped with several ports in the horizontal plane. The laser enters through one of these and intersects the gas jet just above the surface of the liquid. The laser used for this work was an Nd : YAG pumped dye laser with Rhodamine 6G or Rhodamine B as active media. Neutral density filters were used to attenuate the incident beam and prevent saturation. The LIF signal was detected by a photomultiplier situated in one of the ports and the fluorescence was collected and imaged on the photocathode using a combination of lenses and filters.The signal was amplified by means of a gated photon counter. Probe laser and detection gate were triggered by a delay generator with timing initiated by the opening pulse of the nozzle valve. In the work reported here, this permitted complete spectral scans of the probe molecule LIF at 20 ps intervals during and after the gas pulse. The probe molecule throughout this work was I,. It has a number of advantages A. J. Kenyon et al. in this role, for example, the spectroscopy of this species is very well known.It has low rotational and vibrational constants and thus only a very short spectral scan will give accurate vibrational and rotational temperatures. A third useful feature is that in the region chosen here, there is a fortuitous enhancement of the Franck-Condon factors from urdquo; = 1, 2 which permits very accurate determination of vibrational temperatures for the probe species. It is also helpful that much is known concerning the quantum- resolved collision dynamics of this species.rsquo; The probe molecule enters the vacuum chamber seeded (0.3 of total gas) in argon. Chamber pressures down to lop7 Torr were achievable in the presence of PDMS, the liquid used in the study reported here. Care was taken to ensure surface cleanliness for the duration of the experiment.Tests made before and after a lengthy series of measure- ments indicated that the surface tension of the liquid was unchanged. This is a sensitive measure of surface contamination. The liquid was heated 30 K above that temperature to be used in the experiment for 30 min and then allowed to stabilise for a further 30 min prior to data acquisition. Trapped molecules desorb on a ps timescale and will be pumped away between pulses of the nozzle which operates at around 4 Hz. Pulse duration is around 1 ms and thus the duty cycle is less than 1 of the time of the experiment. Results and Discussion As briefly described above, the TD LIF signal contains information on probe molecule warming through contact with the surface and here we focus on this aspect of the gas/ liquid interaction.The state-resolved scattering experiment we report is less informative about the process of trapping than are TOF measurements. The energetics of desorption however are very clearly revealed, as we demonstrate. Both trapping and desorption will be dependent upon energy distribution within the impinging iodine molecules and it is essential to characterise the incident beam in order to make a full interpretation of the data. A careful study was made of the time evolution of iodine number density and vibra- tional temperature over the duration of the gas pulse. Number density was determined by fluorescence intensity and internal temperature from the intensity ratios of bands originating on urdquo; = 0, 1 and 2.The pulse characteristics may be summarised as follows. Pulse duration was 1.2 ms. Number density of I, molecules rises sharply to a maximum in 200 ps, remains approximately constant for 660 ps and then closes sharply in less than 100 ps. Measurement of vibrational temperature Kib at different points in the nozzle cycle indicate two separate regimes. On the leading edge of the pulse, the nozzle is only partially open and collision conditions for supersonic expansion are not met. In this region amp;, values are high whilst translational temperatures (Trans)will be low. In the centre of the pulse, conditions of nozzle diameter and mean-free-path are appropri- ate for supersonic expansion, hence qib is low and Transwill be high.The lagging edge of the pulse will also contain a component of effusive expansion since as the nozzle closes, mean-free-path and nozzle diameter will become similar in magnitude. The results of our study confirm this behaviour and there is a significant fraction of hot I, species in the final 100 ps of the expansion. This group of warm molecules at the end of the pulse can be expected to add complexity to the signal. Results of the LIF scans in the presence of a liquid surface are shown in Plate 1. The liquid in this instance in PDMS. Spectra taken at 20 ps intervals were amalgamated into 3D plots which show the spectral intensities as a function of time with zero set by the electrical pulse that opens the nozzle. Of particular relevance here is the desorption region after nozzle closure.The variation of signal intensity with time represents the desorption kinetics whilst the spectral distributions in this time region give a measure of the dynamics of the trapping and desorption process. Plate 1A shows the results of LIF Laser Molecular Beam Scattering Plate 1 3D plot showing time evolution of LIF spectra during and after nozzle closure. A Shows spectra obtained with no liquid and B displays time-dependent LIF of I, scattered from PDMS. Turf= 300 K. (a) Inelastic scattering and (b)trapped/desorbed species. scans in the absence of liquid and Plate 1B displays data in the presence of PDMS. Vibrational (and rotational) warming can clearly be seen in the 3D spectral representa- tions.Kinetic Analysis Fig. 1 shows the time evolution of the iodine number density as a function of time, with and without liquid surface present. Fig. l(a) shows a view of the complete pulse cycle and in Fig. l(b) and (c) the region at the end of the pulse is amplified to show more A. J. Kenyon et al. Plate 1 continued clearly the decay of signal from molecules desorbing from the surface. Fig. l(a) and (c) show data from PDMS at 300 and 330 K, respectively. The increased rate of desorption at the higher temperature may clearly be seen. The decay signal is quite lumpy in both cases though more markedly so at the lower surface temperature. As can be seen, there is no one simple exponential decay, though the process of desorption is effectively over 120 ps after closure of the nozzle for Turf= 300 K and 40 ps after closure at the higher surface temperature.Laser Molecular Beam Scattering N,( 1 1 I I I I I-20 lsquo; 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 30 I I I I 1 2025 I 1 I I I-5 1 1.15 1.20 1.25 1.30 1.35 30 I I I I 20 25 1 I I I I I 1.14 1.16 1.18 1.20 1.22 1.24 time/ms Fig. 1 Plot of I, number density us. time, with (broken line) and without surface present. (a) Shows the complete nozzle pulse cycle, (b)an expanded version of the region after pulse closure for Turf= 300 K, (c) displays this same region for Turf= 330 K. Some comment on the structure of the kinetic plot is appropriate. At present the origin is not completely understood.It is very likely the result of essentially two groups of I, molecules interacting with the surface. The first is the lsquo;supersonicrsquo; group (low Kib, high Trans)from the central part of the pulse and the second an lsquo;effusiversquo; group from the trailing edge of the pulse. The former group begins a slow transformation into the latter at ca. 900 ps from the start of the pulse. The time evolution of effusive I, number density from the nozzle will be complex and difficult to model with certainty. It is likely that these species with higher internal temperatures will desorb more rapidly than those from the supersonic regime which have lower values of Tib.Using arguments of this kind we have been able to model approximately the shapes of the decay curves shown in Fig.l(b) and (c) and to obtain kinetic parameters using the assumptions outlined above. We anticipate that experimental improvements will lead to removal of the pulse tail in future experiments. Kinetic data for two sets of I, molecules may be evaluated and we obtain residence time 115 ps on the 300 K surface and 80 ps on the 330 K surface. Analysis of this data A. J. Kenyon et al. 251 for two sets of molecules and two values of Turfyields an estimate of the activation energy of desorption Ed, namely 30 kJ mol-l. The pre-exponential factor obtained from this data is vd = 3 x 10rdquo;. This value of vd is rather low for a surface process and such a value is often regarded as characteristic of surface diffusion prior to desorption.lo Dynamical Behaviour The internal-state distributions of desorbing I, molecules are summarised in 3D plots of spectral intensities as depicted in Plate 1. There is a qualitative difference in the quantum-state distributions for molecules detected within the gas pulse and those inter- rogated after nozzle closure. The former show little change in qib and Totfrom the incident molecules. The latter show extenstive heating and population redistribution which is most clearly seen at the end of the pulse cycle. This is more strongly emphasised in the lsquo;differencersquo; spectra in Fig. 2 obtained by computer subtraction of data obtained with no liquid present from that taken in the presence of PDMS. Fig. 2(a) shows the signal up to nozzle closure and portrays mainly the IS process.Fig. 2(b) shows develop- ment of quantum-state populations after nozzle closure and gives an indication of the time evolution of probe molecule warming by the liquid surface. Note that these spectra are normalised and the signal-strength axis is a relative one for each time interval. As can be seen in the figures, the early part of the pulse cycle is dominated by molecules scattered with very little vibrational or rotational excitation although signifi- cant population shifts into u = 1 are seen as the nozzle begins to close. After nozzle closure, desorbing I, molecules rapidly increase both qiband cotuntil near the end of the process, as the last few molecules are desorbing, the population has shifted strongly to higher vibrational and rotational states.A more quantitative plot of data is shown in Fig. 3 which shows values of qib at selected points across the pulse cycle calculated from spectra taken with and without liquid surface present. Fig. 3(a) displays values of qib with and without surface present, during and after the opening of the nozzle. Fig. 3(b)is an expanded version of the region following nozzle closure for Turf= 300 K and Fig. 3(c) shows the same region for Turf= 330 K. The desorption qib values show interesting structure for both surface temperatures. In the case of the 300 K surface, there is evidence that molecules desorbing just as the nozzle is closing emerge slightly colder than the control group of I, molecules direct from the nozzle.This region is likely to be dominated by effusive I, species and the result suggests that some of the energy of desorption for this group of molecules comes from the internal energy of the adsorbed species. The predominant group of cold I, molecules show only small changes in Kjb as interaction time with the surface increases before finally rising rather sharply as the last few molecules emerge. For Turf= 330 K, warming of desorbing species is more rapid, although here also there is evidence of a cooling mechanism in operation for the group of molecules emerging just as the nozzle closes. General Trends Although there is little experimental data available on liquid surface dynamics it is inter- esting to speculate on the possible origin of differences that are observed both in TOF studies4,rsquo; and in LIF data6*rsquo; between the liquids investigated.The former experiments indicate that squalane is more effective at trapping species than is PFPE and that PDMS is also a highly efficient molecular trap. In addition, the polysiloxane is the most efficient of this group in ejecting or desorbing trapped molecules. Saecker et aL4 attrib-ute variations in trapping efficiency to different values of effective collision mass of liquid molecules. Laser Molecular Beam Scattering Fig. 2 3D plot of normalised difference LIF spectra (data with surface minus data without surface) showing time evolution of I, quantum-state populations during and after the nozzle pulse. (a)Shows the time interval during which the nozzle is open (mainly IS) and (b)displays data in the TD regime after nozzle closure.Turf= 300 K. The collision dynamics of the process require that the I, translational energy (EtranS) be lost to the surface to become trapped in the potential well of the surface-molecule potential. Clearly the relative magnitudes of well depth and E,,,,, will be of importance in this respect. Also of relevance will be the efficiency of individual molecules at convert- ing incident I, translational energy into internal motions of the liquid molecules. Evi- 253 1 I I I I 1 1-300 -(a)-250 -00 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 240 I I 220 I2 lsquo;BQ1 180 ,rsquo;:A-/ 160 140 k1 I I I I I 1.10 1.15 1.20 1.25 1.30 1.35 1.40 I I I I I 1 I100 1 1.10 1.12 1.14 1.16 1.18 1.20 1.22 1.24 time/ms Fig.3 Plot of I, qibvalues us. time with (broken line) and without surface present. (a) Displays the complete time region investigated (b) is an amplification of the final part of the pulse and desorption region Turf= 300 K, (c)is as for (b)but Turf= 330 K. dence obtained from these studies indicates that liquid molecules in which C-H bonds form the outermost protruding groups are more effective traps than those with C-F protruding bonds. All liquid species examined have been large molecules with many vibrational states, many of which will be low-frequency torsional modes. Squalane and PDMS differ from PFPE in that they have a number of high- frequency modes arising from the C-H bonds. It is tempting to suggest that effective trapping results from conversion of Etransto internal liquid molecule energy through the excitation of C-H vibrational modes.This suggestion is at variance with known rates of translational to vibrational (T-V) energy transfer in gas-phase species, evidence from which is that this is a very slow process. However, there are indications that the rate of T-V transfer increases rapidly as molecular complexity increases. In the case of PDMS and squalane, the C-H stretch region will be one of exceptionally high density of states from the many low-frequency modes of these molecules. These would enhance intramolecular relaxation and conversion of C-H mode excitation into torsional or Laser Molecular Beam Scattering other low-frequency modes.Accommodation could be the result of multiple excitation of these low-frequency modes or by single excitation of higher-frequency modes followed by rapid intramolecular relaxation. In this sense then the C-H bonds could be seen as providing acceptor vibrational modes which enhance the trapping efficiency of liquid- state molecules. The extent of warming of desorbing I, molecules follows the sequence PDMS squalane PFPE.6*7No measure of Etransof the desorbing species was avail- able in this LIF experiment. PDMS was found to be particularly efficient at energising trapped species and this also may be the result of internal modes of the liquid molecules. If so, these might be termed promoter modes and careful study of the dependence of T,,, and Kib on chemical nature of the surface may help identify particular molecular motions as responsible.It is well known that the Si-0-Si torsional mode is important in raising the glass-transition temperature in polysiloxanes'2 and this may play a role in activating adsorbed species prior to desorption. Conclusions This contribution has highlighted recent results of studies of molecular scattering from the surface of some molecular liquids. The kinetics and dynamics of desorption from PDMS have been investigated in detail and we have obtained values of the activation energy of desorption of I, from PDMS together with the frequency factor. Comparison of data obtained on several different liquid surfaces using molecular scattering and LIF dete~tion~.~and atomic scattering together with TOF techniques indicates that liquids vary widely in their trapping efficiencies and their effectiveness at energising adsorbed molecules prior to desorption. Polydimethylsiloxane is particularly effective at both.We thank SERC for financial support and for a Studentship. We also wish to thank CNPq(Brasi1) and AEC(Syria) for Studentships. References 1 D. J. Beaglehole, in Fluid Interface Phenomena, ed. C. A. Croxton, Wiley, New York, 1986, ch. 11. 2 D. J. Auerbach, in Atomic and Molecular Beam Methods, ed. G. Scoles, Oxford University Press, New York, 1992, vol. 2, p. 444. 3 M. Faubel and T. H. Kisters, Nature (London), 1989,389, 527.4 M. E. Saecker, S. T. Govoni, D. V. Kowalski, M. E. King and G. M. Nathanson, Science, 1991, 252, 1421. 5 M. E. King, G. M. Nathanson, M. A. Hanning-Lee and T. K. Minton, Phys. Rev. Lett., 1993,70, 1026. 6 A. J. Kenyon, A. J. McCaffery, C. M. Quintella and M. D. Zidan, Chem. Phys. Lett., 1992, 190,45. 7 A. J. Kenyon, A. J. McCaffery, C. M. Quintella and M. D. Zidan, J. Chem. Soc., Faraday Trans., 1993, 90,3877. 8 C. M. Quintella, A. J. McCaffery and M. D. Zidan, Chem. Phys. Lett., in the press. 9 J. I. Steinfeld and P. P. Ruttenberg, Scaling lawsfor Inelastic Collision Processes in Diatomic Halogen Molecules, JILA Information Center Report No 23, Boulder, Colorado, 1983. 10 M. Balooch, W. J. Siekhaus and D. R. Olander, J. Phys. Chem., 1984, 88, 3521; R. C. Baetzold and G. A. Somarjai, J. Catal., 1976,45, 94. 11 R. D. Levine and R. B. Bernstein, Molecular Reaction Dynamics and Chemical Reactivity, Oxford Uni- versity Press, New York, 1987. 12 Comprehensive Organometallic Chemistry, ed. G. Wilkinson, F. G. A. Stone and E. W. Abel, 1982, vol. 2, p. 334. Paper 3/03009A; Received 24th May, 1993
展开▼
