Microscale Molecular Tennis' L.J.F. Hermans Huygens Laboratory, Leiden University, P.O. Box 9504, 2300RA Leiden, The Netherlands Abstract. Most of the cutting-edge research on molecule-surface interactions has been taken over by molecular-beam experiments. Even so, work on transport properties—much in the spirit of Lloyd Thomas—has been yielding subtle information on the role of the internal state that is difficult to obtain by other methods. Two such transport experiments are highlighted. One is the influence of an external magnetic field on a Knudsen flow. It gives information on nonequilibrium angular momentum polarizations produced by nonspherical molecule-surface interactions. The other is Surface Light-Induced Drift. It gives information on the rotational and vibrational-state dependence of accommodation coefficients. This includes the role of the direction of the angular momentum with respect to the surface ("helicopter" vs. "cartwheel"). INTRODUCTION Our earliest understanding of the behavior of molecules at solid surfaces was derived largely from experiments on transport properties. The pioneering work of Lloyd B. Thomas, R.G. Lord and many others has yielded a wealth of data on energy and momentum accommodation since the 1940s [1,2]. Much of this work has been reviewed by H.Y. Wackman [3]. A drawback of this type of experiments is the difficult control over surface quality, since the data are obtained under non-UHV conditions by definition. The introduction of molecular-beam techniques in combination with Ultra High Vacuum equipment opened up possibilities to study molecule-surface interactions under well-defined surface conditions. In addition, beam data are much more detailed since they employ selected velocities rather than a near-Maxwellian. The advent of narrow-band tunable lasers has added state-specificity to many beam data; one may say that they provide information on scattering kernels rather than on the scattering operator. A consequence is that a direct comparison between beam data and transport data is no longer possible: molecular beams produce so-called "beam accommodation coefficients" rather than the traditional "global" accommodation coefficients, which are obtainable only by integration over different directions and energies. It is fortunate that the "unified kinetic approach" developed by Borman, Krylov and coworkers does build a bridge between the two types of data sources (see, e.g., [4, 5, 6]). Although much of the cutting-edge research on molecule-surface interactions has been taken over by molecularbeam experiments, two transport-based experimental techniques were developed over the last two decades which yield rather unique information on the role of the internal state on the scattering process. One is the influence of an external field on Knudsen transport properties. It employs precession of the molecules in an external field. The other is Surface Light-Induced Drift. It employs velocity-selective excitation of molecules by a narrow-band tunable IR-laser. It is the purpose of this paper to briefly discuss both types of experiments, and to highlight the information that they have provided. MAGNETC FIELD EFFECT ON KNUDSEN FLOW In any Knudsen transport situation, the nonequilibrium velocity distribution will, by nonspherical molecule-surface interaction, also cause the angular momentum distribution to be anisotropic. This coupling between v -space and Jspace will have its impact on the transport. This can be made visibe by changing the anisotropy in J-space using the 1 Lloyd Thomas Memorial Lecture CP663, Rarefied Gas Dynamics: 23rd International Symposium, edited by A. D. Ketsdever and E. P. Muntz © 2003 American Institute of Physics 0-7354-0124-1/03/$20.00 10 (a) Without field____________ (b) With field FIGURE 1. Elementary picture of the field effect on a Knudsen flow, using the analogy of a ball bouncing from a rough surface, (a) The spin produced by tangential forces during the first collision yields a drawback for the forward motion in the second, (b) Precession of the rotational axis between two collisions, caused by an external magnetic field, removes the drawback. For the gas as a whole this enhances the flow. precession of the molecules around an external field: this will, in turn, change the velocity distribution, and thus the transport. The theory of such effects was worked out by Borman et al. [7], by Knaap and Kuscer [8] and by Cercignani [9]. The field effect on a heat flux in a Knudsen gas was first observed by Borman et al. [10]. Here we will discuss the magnetic field effect on a Knudsen particle flow through a flat channel. An elementary picture of the effect is given in Fig. 1. To achieve precession of the molecules around the field, use is made of the (small) magnetic moment resulting from the rotation of the molecule: JJL = g j-fiN, where g is the molecular Lande-factor and p,jy the nuclear magneton. In an external field, this produces a (J-independent) precession frequency u = (gp>N/h)B, with B the field strength. The relevant parameter for the field dependence is the precession angle CJT, where r is a typical flight time between the two parallel surfaces forming the channel: r — b/(2kT/m)1^2 with b the spacing between the surfaces. The field dependence of the change in the Knudsen flow, £$/<!>, would have a simple (1 — COSCJT) behavior if r would be uniform throughout the gas. Taking the Maxwell velocity distribution into account, one expects a behavior of the form 6$ f°° — ~F(LJT) = I [(1 — COSUJT)/V]Vexp(—v )dv (1) ^ Jo where v represents the dimensionless z-velocity of a molecule, with z along the surface normal. However, for a complete description of the effect, one has to take into account that various types of J-polarization can be produced, which can give different contributions to <5<!>/<I>. It should be stressed at this point that, according to the Curie principle, a particle flow, having a vector character, can generate only polarizations (in the combined velocity-angular-momentum space) which are true vectors. In the kinetic description of ref. [8], only the four simplest polarizations were considered, viz., two of first rank in J ("orientation") and two of second rank ("alignment"). These are listed in Table 1. Note that orientation and alignment give contributions with opposite signs, and that alignment—due to its symmetry—can produce a "double-frequency behavior", i.e., F(GJT). The experiments were performed in a capillary Wheatstone bridge arrangement, with one of the capillaries replaced by a flat channel in a magnetic field. The field effect is then detected as an off- balance, and measured in terms of a small pressure difference. For a description of the experiment see, e.g., [11,12, 13] and references therein. The role of the molecular symmetry Examples of experimental results are shown in Fig. 2 for two different molecules and the same LiF(OOl) surface. It is seen that the two gases have markedly different behavior: CH4 behaves almost perfectly according to the Jy polarization, i.e., like the elementary tennisball picture in Fig. 1. This seems to reflect the "rough-sphere"-like behavior of CH4. By contrast, for H2 we see (both from the sign of the effect and the position along the c<;r-axis) a clear dominance of alignment types of polarization, viz., the lower two in Table 1. This seems to reflect the "dumbbell'Mike behavior of diatomic molecules, since these alignment types require torques along the z-axis during the collision, which are absent, by symmetry, for a spherical molecule. 11 TABLE 1. Four types of polarization which can be produced in a Knudsen flow (along z-axis) between parallel plates (surface normal n along z). The behavior of their contributions to the field effect 63>/$ is given for three field orientations (cf. Eq. (1)). Note that the polarizations are actually x-components of true vectors (J itself is a pseudovector). Example: Jy = (J x n) x , which is the elementary "tennis ball" type shown in Fig. 1.________________________________ Type of polarization_____B||rr___ vyJz F(UJT) -F(2vT) JxJz -J 0.2 (a) CH4/LiF(001) 0.1 O -0.1 10 0.1 U)T FIGURE 2. The change in flow £<$/$ measured as a function of precession angle ur for CHi (field along z) and H2 (field along y). The surface is LiF(OOl) with the flow along the [100] direction. The curves through the data points are theoretical curves (see Eq. (1) and Table 1) with the vertical scale as the only free parameter. The role of surface corrugation Tangential forces acting during the molecule-surface collision are obviously essential for this phenomenon: no such effects can arise for a perfectly smooth surface. The question then arises what role surface corrugation plays. To investigate this, experiments were done using a LiF(OOl) surface with two orientations with respect to the crystal axes. This makes it possible to vary the corrugation for essentially the same surface. Details of the set-up are given in [13]. Data obtained from this experiment are shown in Fig. 3 for H 2 , HD and N2. When comparing the left-hand panel with the righ-hand panel for each gas, it is seen that the role of corrugation is remarkably large for H2, but much smaller forHDandN 2 . The explanation of this behavior may be given in the framework of the unified theory discussed in Refs. [4] and [5]. For H2 the large rotational level splitting (H2/2Ik = 85 K) combined with the selection rule A J = 0, ±2 requires a large energy exchange with the phonon bath if the rotational state is to be changed. The interaction range effective for this process is limited to the small region of closest approach, where the interaction is strong. It is in this part of the interaction range that the corrugation makes itself felt. Thus, an appreciable influence of the corrugation is to be expected. For a molecule like HD, however, the smaller level spacing and the possibility of A J = 1 collisions makes rotational transitions occur over a much larger region of the trajectory. Therefore the molecules, while receding from the surface, can continue to interact with the surface up to distances where the influence of the corrugation is effectively washed out. This argument is corroborated by the fact that H2—although rather spherical in its interaction—showed larger field effects than did HD or N2 : it "benefits" more from the corrugation than do heavier molecules having 12 i i mi———i—i—r i i i ni———i—i—r H2/LiF(001) Flow/[100] H2/LiF(OOl) Flow/[110] 1CT 1.5 I HD/LiF(001) Flow/[100] 1.0 I I I ll|————I——I—I I I III]—————I——I—! I I I I l| HD/LiF(001) F!ow/[110] 0.5 ^P -0.5 -1.0 ' I I Itit ———I———t I I I I MI 10" 3 do- ) 1 I CJJT l i f t . 10- 10 N2/LiF(001) Flow/[100] 10 N2/LiF(001) Flow/[110] U)t 10 FIGURE 3. Data for H2, HD and N2 flowing along a LiF(OOl) surface, with two crystal orientations (see inset). Each panel shows data for two field orientations. Left-hand panels: flow along [100] (higher corrugation). Right-hand panels: flow along [110] (lower corrugation). smaller level spacing. SURFACE LIGHT-INDUCED DRIFT Consider a Knudsen gas at rest in a tube which is sealed off by optical windows. Let us shine a narrow-band tunable laser down the tube, tuned into the wing of the Doppler-broadened absorption profile of some ro-vibrational transition. This will produce velocity-selective excition (see Fig. 4). If the tangential momentum accommodation coefficient is state dependent, a net tangential momentum transfer to the surface will take place, and the gas will be set into motion. In a closed tube this will give rise to a pressure difference, which can be related to the difference in accommodation coefficient (see e.g., Ref. [14] for details). Even if ro-vibrational excitation involves a change in both vibrational and rotational quantum number, it has 13 vL(=Aco/k) excited state v k ground state Laser pressure difference FIGURE 4. The principle of Surface Light Induced Drift. Shown is the case of laser detuning in the blue wing which excites molecules moving to the right. Here, excited molecules are assumed to have larger accommodation. x a < -1 -2 DR(J-1) OP(J) * average -3 -4 FIGURE 5. The change in accommodation coefficient for tangential momentum upon excitation, Aa, as a mention of rotational quantum number J. Results are shown for P( J) and R( J - 1) transitions in the fundamental vibrational band v — 0 -> 1. Note that, for each J, the two data points correspond to opposite J transitions. been shown [15] that the two contributions can often be separated into a rotational and a vibrational part: Aa = Aa roi + Aayi&. This is achieved by combining the results for P( J) and R( J - 1) transitions. For such pairs, the changes in J are equal but opposite, while the change in v (v = 0 -)• 1) is the same. An example for HF interacting with a stearic acid coated surface is shown in Fig. 5. For each pair of data points, the average can be interrpreted as Aa v i6, while half the difference corresponds to Aa roi . From the data it is found that a decreases slightly upon vibrational excitation (by almost 1 x 10~3 in the above case). This may be qualitatively explained by vibrational-to-translational energy transfer arguments [15]. By contrast, a is found to increase with increasing J for the case studied. A qualitative explanation along the lines of the unified approach [4] is under way [5, 6, 16]. 14 0.4 ;OCS:P(5) n e /n 0 1 II I I II I I -0.2 10' P (Pa) 10* FIGURE 6. Results of SLID experiments in a plane parallel geometry for OCS interacting with a glass surface. Shown is the normalized observed pressrue difference for ro-vibrational excitation ^ = 0 — » 2 , J = 5-»4. Black circles denote laser polarization parallel to the surface, open squares perpendicular. The difference between the data points corresponds to "helicopters" having the lower accommodation. Helicopters vs. Cartwheels We will now address the question how the accommodation coefficient depends on the orientation of the rotational axis with respect to the surface. This can sensitively be studied by using a flat channel in combination with the excitation by linearly polarized light. Details of this experiment can be found in [17]. Results for the linear molecule OCS interaction with a glass surface are shown in Fig. 6. It is seen that in the data points with laser polarization perpendicular to the surface are consistently higher than those with the light polarization parallel. This corresponds to lower accommodation for the molecules having their rotational axis perpendicular to the surface ("helicopters"). This result may seem at odds with molecular-beam data for NO interacting with Pt( 111), obtained by Jacobs et al. [ 18, 19]. They found that NO molecules desorb preferentially as "helicopters" for high J- values (J > 30). Given the relation between various accommodation trapping coefficients (as different moments of the same relaxation probability function [4]) one may conclude that NO would also have higher accommodation for tangential momentum in the "helicopter" mode. An explanation may be given by considering the role of rotational (de)excitation in the collision dynamics. We will illustrate this for the model of a linear "dumbbell"-like molecule approaching a smooth surface. Let us first consider an incoming molecule having a low J (J « J thermal)- This molecule will tend to suffer rotational excitation, thereby losing part of its "normal energy". The lower value of the outgoing velocityu vz will increase the effective flight time t and thus also a. However, this mechanism fails to increase a for molecules in the helicopter mode: since these have their internuclear axes always parallel to the surface, the absence of a torque will leave the rotational state of the molecule unaffected (model calculations elucidating this point may be found, e.g., in Refs. [20, 21]. Thus, for low-J molecules, this mechanism gives the cartwheel mode a higher a than the helicopter mode. Let us now consider incoming molecules with J » J thermal- Here, for molecules in the cartwheel mode, the opposite will take place: rotational de- excitation will tend to increase the outgoing "normal energy" and thus decrease a. Again, incoming molecules in the helicopter mode will be unaffected. Thus, for high-J molecules, this mechanism will give the helicopter mode the higher a, in agreement with experiment. ACKNOWLEDGMENTS The author pays tribute to the memory of Jan J.M. Beenakker and Ivan Kuscer, who were closely involved in this field of research. Furthermore he is indebted to Sergey Yu. Krylov and Anneke Aschoff for many stimulating discussions, and to the graduate students who carried this research: J.J.G.M. van der Tol, R. Home, R.W.M. Hoogeveen, G.J. van der Meer, B. Broers and EJ. van Duijn. 15 REFERENCES 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. See, e.g., Thomas, L.B., and Olmer, F.G., J. Am. Chem. Soc. 65, 1036-1043 (1943). See, e.g., Lord, R.G., "Tangential momentum accommodation coefficients of rare gases on polycrystalline metal surfaces", in Proceedings of the 10th International Symposium on Rarefied Gas Dynamics, edited by J. Leith Potter, Progress in Astronautics and Aeronautics 51, 1977, pp. 531-538. Wachman, Harold Y., "Thermal Accommodation Coefficient: The Contribution of Lloyd Brewster Thomas", in Proceedings of the 18th International Symposium on Rarefied Gas Dynamics, edited by Bernie D. Shizgal and David P. Weaver, Progress in Astronautics and Aeronautics 158, 1993, pp. 461-478. Borman, YD., Krylov, S.Yu, and Prosyanov, A.V., Sov. Phys. JETP 67, 2110- 2121 (1988). Hoogeveen, R.W.M., Hermans, L.J.F., Borman, V.D., and Krylov, S.Yu., Phys. Rev. A 42, 6480-6485 (1990). Pankov, A.Yu, Krylov, S.Yu., Van Duijn, E.J., and Hermans, L.J.R, J. Chem. Phys. 112, 8680-8686 (2000). Borman, V.D., Krylov, S.Yu., Nikolaev, B.I., Ryabov, V.A., and Trojan, V.I., Sov. Phys. JETP 52, 904 (1980). Knaap, H.F.P., and Kuscer, I., Physica A 104, 95-114 (1980). Cercignani, C., and Lampis, M, Physica A 99, 305 (1979). Borman, V.D., Laz'ko, VS., and Nikolaev, B.I., Sov. Phys. JETP 39, 657 (1974). Home, R., and Hermans, L.J.R, Phys. Rev. Lett. 60, 2777-2780 (1988). Home, R., and Hermans, L.J.R, J. Chem. Phys. 91, 1261-1267 (1989). Home, R., and Hermans, L.J.R, Chem. Phys. Lett. 179, 58-62 (1991). Hoogeveen, R.W.M., Meer, Van der Meer, G.J., and Hermans, L.J.R, Phys. Rev. A 42, 6471-6479 (1990). Van Duijn, E.J., Nokhai, R., Hermans, L.J.R, Pankov, A.Yu., and Krylov, S.Yu., J. Chem. Phys. 107, 3999-4005 (1997). Pankov, A.Yu., and Krylov, S.Yu., to be published. Broers, B., Van der Meer, G.J., Hoogeveen, R.W.M., and Hermans, L.J.R,,/ Chem. Phys. 95, 648-655 (1991). Jacobs, D.C., Kolasinski, K.W., Madix, R.J., and Zare, R.N., J. Chem. Phys. 87, 5038-5039 (1987). Jacobs, D.C., Kolasinski, K.W., Shane, S.R, and Zare, R.N., J. Chem. Phys. 91, 3182-3195 (1989). Wolf, R.J., Collins, D.C., and Mayne, H.R., Chem. Phys. Lett. 119, 533-537 (1985). Jacobs, D.C., and Zare, R.N., J. Chem. Phys. 91, 3196-3207 (1989). 16
© Copyright 2025 Paperzz