single crystal

View Article Online / Journal Homepage / Table of Contents for this issue

Faraday Discuss., 1997, 107, 435È444

InÑuence of acoustic wave excitation on CO oxidation over a Pt{110} single crystal S. Kelling, T. Mitrelias, Y. Matsumoto, V. P. Ostanin and D. A. King*

Published on 01 January 1997. Downloaded on 21/02/2015 18:39:29.

Department of Chemistry, University of Cambridge, L ensÐeld Road, Cambridge, UK CB2 1EW

A unique ultra-high vacuum (UHV) compatible excitation system combined with an advanced ultra-high amplitude and frequency resolution acoustic spectrometer has been designed and constructed to permit accurate studies of the fundamental mechanism by which acoustic excitation inÑuences heterogeneous catalytic reactions. A clean PtM110N thin Ðlm single-crystal catalyst was excited with low-energy acoustic waves (Rayleigh waves) under high vacuum and UHV conditions to increase the reaction rate for carbon monoxide oxidation. A remarkable six-fold increase in the chemical activity was observed. By using a new, very accurate method to monitor the sample temperature using high-resolution acoustic wave resonance spectroscopy (HRAWRS), a non-thermal acoustic-wave-induced enhancement of the reaction rate is clearly demonstrated. The pressure and temperature dependences of the enhancement provide some insight into the mechanism by which acoustic waves enhance catalytic reactions on solid surfaces.

The cleaning e†ect of ultrasound and its surface activation play an important role in the sonochemical enhancement of reactivity in chemical processes involving solid and liquid phases. There have been very few studies on the e†ects of acoustic waves (AWs) on surface chemical reactions under high-vacuum conditions by the application of piezoelectric surface AW transducers, with no practical consequences to date. In liquid/ solid systems, the chemical reactivity with ultrasound is improved by cleansing of the surface, the sweeping of intermediates or products away from the interface, and the creation of surface defects that act as reactive centres, induced by high-intensity acoustic Ðelds.1h6 Over the past seven years Inoue and co-workers have reported that ultrasound can signiÐcantly increase the rate of some catalytic reactions at solid surfaces.7h10 In their work, polycrystalline metal Ðlms were deposited between interdigital transducer (IDT) electrodes on an LiNbO substrate. Experiments were performed in a gas-circulating 3 apparatus under high-pressure conditions. The results demonstrate that the state of the catalyst surface has a major inÑuence on the efficiency of activation by AWs.9 For example, acoustic excitation is much more efficient for an oxide surface than for a metallic one.10 Furthermore, the inÑuence of AWs depends upon the kind of catalytic reaction7 and the AW displacement mode has no big inÑuence on the AW e†ect.11 These results are remarkable since, for reactions at the gas/solid interface, the physical interaction between sound and matter cannot be the same as in liquid/liquid or liquid/ solid systems. The phenomenon of increasing rate of reactions taking place at the gas/solid interface by exciting the catalyst acoustically is theoretically challenging, since the quantum of energy associated with acoustic waves is ca. 107 times smaller than the energy barrier associated with a surface reaction. However, to understand the underlying physical 435

View Article Online

436

Acoustic wave enhancement of catalytic reactions

mechanism a basis of data measured on well deÐned surfaces under controlled conditions is needed. Here, we present studies which were performed in a high-vacuum reactor cell and an ultra-high vacuum (UHV) chamber using Pt thin-Ðlm single crystals as catalysts. Early results were obtained using a reactor with only limited cleaning and analysing facilities, at a gas pressure during reaction of ca. 2 ] 10~5 Torr. A six-fold increase in the CO oxidation reaction rate during acoustic excitation of a PtM110N catalyst surface was observed.12 Subsequent experiments were carried out in a UHV chamber which enabled us to clean the PtM110N thin Ðlm single crystal in vacuo and to characterise its surface at di†erent stages of the experiment.13 To throw some light onto the mechanism of the acoustic e†ect, its pressure and temperature dependence was studied. A new method for accurate sample temperature measurements, based on HRAWRS,14 has been successfully applied and this helped to exclude the observed e†ect as being due to thermal heating.


Methods The experiments were performed in a stainless-steel reactor chamber and a UHV vessel. The UHV chamber is equipped with a low-energy electron di†raction (LEED) facility, a quadrupole mass spectrometer (QMS), viewports, and a sputtering gun. The sample is held on a manipulator which is mounted on top of the cylindrical vessel and allows positioning in front of every facility mentioned above by rotation around the cylinder axis and by moving in the x-, y- and z-directions. Gas inlets for argon, carbon monoxide and oxygen are connected to the vessel. A base pressure of 2 ] 10~10 Torr is maintained by a turbomolecular pump backed by a rotary pump. The reactor has been specially constructed to carry out experiments under high-pressure conditions. It is attached to the gas line of the main UHV system pumped by a di†usion pump, and has a base pressure of 5 ] 10~8 Torr which is measured by an ionisation gauge. The reactor is connected to the main chamber via a leak valve. Analysis of the gases is performed using the QMS in the main chamber with the leak valve fully open. The di†erence between the pressures in the main analysis chamber and the reactor is 5 ] 103 Torr. AWs are generated and detected utilising an interdigital transducer coated with a 5000 Ó thin-Ðlm PtM110N single crystal cold-welded on the propagation path of the surface AW.15 The IDT consists of a poled 128¡ Y-cut LiNbO single-crystal substrate with an input and output aluminium electrode array capable of 3exciting Rayleigh waves. The centre frequency at room temperature (RT) is ca. 19.5 MHz. The IDT is clamped onto a 1 mm thick tantalum plate to support it, as shown in Fig. 1. A home-built, tunable, ultra-high amplitude and frequency resolution spectrometer (vector analyser),15 is used for narrow-band excitation and to measure the acoustic signal from the IDTÏs output electrode array. The RF power introduced to the sample during the experiments is of the order of 1 W. To reduce the induced RF inÑuence on the other instruments, ferrite sleeves and toroidal cores are used around the cables. The sample holder can be heated resistively using three heating wires mounted on the back of the supporting plate. The maximum temperature in our experiments is 600 K, which lies well below the Curie temperature of the LiNbO . For experiments carried 3 a K-type thermocouple out in the reactor cell the temperature was monitored using spot-welded onto the IDT mounting plate. It is known that even low-energy AWs cause local heating along their propagation path. As this temperature rise will only be detected indirectly on the mounting plate, owing to its high heat capacity, a second thermocouple clamped onto the crystal sample using the existing sample-grounding clamp was added for the series of experiments performed in the main chamber. Heat loss from the sample by conduction through the thermocouple was prevented by a heat barrier 5 mm from the clamping point. The measured di†erence between sample and backplate tem-

View Article Online


S. Kelling et al.

437

Fig. 1 IDT device with a thin-Ðlm single-crystal sample cold-welded to the top surface, is clamped onto a backplate which can be heated resistively. The temperature is measured using sample and backplate thermocouples.

perature, when heating the backplate, was found to be in very good agreement with theoretical calculations : this raises conÐdence in the measurement of the temperature of the thin sample Ðlm. During experiments in the reactor the temperature was stabilised using a EUROTHERM control unit. For experiments under low-pressure conditions in the main chamber we did not use this feedback mechanism and kept the temperature stable by applying a constant current to the heating wires. To exclude Ðrmly a purely thermal e†ect, an accurate measurement of the sample temperature is very important. Great care was taken to ensure that we were measuring the correct sample temperature at all times during the experiments. Owing to our sample-mounting system, the heating rate at high temperatures is too low to take temperature-programmed desorption (TPD) spectra. However, with a heating rate of 1.0 K s~1 the Ðrst CO desorption peak from an NiM110N surface lies at ca. 375 K,16 which can be measured with our set-up. We, therefore, equipped an IDT with an NiM110N crystal and took TPD spectra, using the sample thermocouple to monitor the temperature. The maximum of the desorption peak was measured at 373 K, in good agreement with the above. It is noted here that special care was taken to ensure that the heating rate was the same as that used in ref. 16. After verifying that the temperature of our thin-Ðlm sample can be determined using a thermocouple, the sample temperature increase due to AW propagation was measured. An IR thermometer with a working range from 600 to 900 K was used to detect the black body radiation from a 1 mm diameter spot on the sample surface during acoustic excitation. With continuous AW generation and the same parameters as used in our experiments, at 565 K the temperature rise measured with the sample thermocouple was ca. 20 K, which closely matched the temperature increase measured with the IR thermometer. As a Ðnal rigorous check of the AW-induced sample temperature increase we developed a new, very precise method to determine the temperature by taking advantage of our high-resolution acoustic spectrometer. A rise in the sample temperature causes a shift in resonant peaks of ca. 2 kHz K~1 towards lower frequencies. This shift can be measured with high accuracy by recording short scans of the acoustic spectrum. A special series of experiments was devoted to calibration of this thermal shift to the real sample temperature. The relation between frequency of a peak and sample temperature, shown in Fig. 2, was obtained with the PtM110N sample in thermal equilibrium at di†erent temperatures, established by applying a constant current to the heating wires. Short acoustic scans were performed with low power, for temperature determination : this

View Article Online


438


Fig. 2 Calibration curve for sample temperature measurement using HRAWRS. (È) Measured before and (L) after a series of catalytic reaction experiment.

causes a maximum temperature rise of 4 K at RT and a temperature rise of 2.5 K at 600 K. Continuous excitation of AWs with an average power of 1 W as used to increase the reaction rate, caused a temperature rise of 60 K at RT. The acoustic heating decreases toward higher sample temperatures ; at 600 K continuous AW excitation led to a 25 K sample temperature rise. If not otherwise mentioned, all temperatures quoted refer to the real sample temperature established after acoustic heating. For experiments performed in the reactor, the platinum sample was cleaned by repeated oxidation (600 K, 2 ] 10~5 Torr) and annealing cycles, under conditions known to produce a good (1 ] 2) LEED pattern from the clean surface.17 In the main chamber, prior to reaction, the sample was cleaned using a standard cleaning procedure consisting of argon ion sputtering, oxygen treatment and high-temperature annealing. The surface condition was characterised using LEED, which showed the sharp (1 ] 2) pattern expected for the clean surface.

Results The inÑuence of AWs on the rate of the reaction 2CO ] O ] 2CO catalysed by a 2 2conditions in an PtM110N thin-Ðlm single crystal was studied under strictly isothermal HV/UHV environment. With the reaction at steady state, the RF generation is switched on, producing continuous scans from 18 to 20 MHz with a power of ca. 1 W. The reaction rate starts to increase slowly, and reaches a stable maximum. The time needed to stabilise the high-activity regime depends on the reactant partial pressures and the scanning parameters. A detailed description of the experimental results is given in the following two sections. Reactor experiments The partial pressures of CO and CO were monitored over a long period of time during 2 and o†. Prior to turning the RF on, the reaction which AW generation was switched on rate and temperature was kept stable for more than 1 h. The total pressure in the reactor cell was 4.9 ] 10~5 Torr, the CO partial pressure measured via a leak into the main chamber, 4 ] 10~12 Torr, and the2 backplate temperature 547 K. AWs were excited by continuously scanning from 18.8 to 19.8 MHz with a step width of 400 Hz and ca. 1 W input power. A single scan consisted of 250 steps and lasted for 75 s. Immediately after switching AW excitation on, the m/z 44 pressures started to rise. After ca. 100 s it

View Article Online


S. Kelling et al.

439

reached steady state at 1.6 ] 10~12 Torr, an increase by a factor of four, as shown in Fig. 3. A small temperature jump of 0.2 K was observed using the backplate thermocouple, but the temperature was quickly stabilised via the feedback mechanism. The higher reaction rate kept stable for as long as RF was applied. Switching the AW generation o† after 40 min of continuous operation caused a decrease in the CO partial pressure back 2 to the original value. The decrease is not abrupt but rather exponential owing to the limited pumping speed. The observed increase in the reaction rate is not attributable to artefacts, such as electromagnetic interference from the RF ampliÐer. The RF inÑuence was tested by performing control experiments with the leak valve connecting the reactor with the main chamber closed ; the variation in the QMS signal was less than 0.26 ] 10~12 Torr. In order to investigate whether or not the AW enhancement e†ect is associated with speciÐc acoustic frequencies (i.e. resonance modes), as has been claimed by Brezhnev et al.,18 the CO production was monitored while performing a wide frequency single scan 2 from 18.8 to 19.8 MHz, as shown in Fig. 4. The step width was 400 Hz and the total scan took 250 s. The periodic peaks visible in the acoustic spectrum correspond to superposition of reÑected waves on the fundamental Rayleigh wave. Following an induction period, the rate of production of carbon dioxide increased sharply, by a factor of ca. six, as a result of the AW excitation. Once the high-efficiency regime is established, the chemical activity remained high, independent of the amplitude of the acoustic signal, as

Fig. 3 InÑuence of continuous AW excitation of the PtM110N catalyst surface on the CO oxidation rate. The thick and dotted lines represent the CO and CO partial pressures, respectively, as 2 measured in the main chamber. The thin line corresponds to the measured temperature of the backplate.

View Article Online


440


Fig. 4 m/z 44 signal (thick line) during one AW scan from 18.8 to 19.8 MHz, recorded with a resolution of 100 Hz. This experiment was carried out in the gas reactor at a total pressure of 4.9 ] 10~5 Torr, an O /CO partial pressure ratio of ca. 10, and a backplate temperature 547 K. 2

can be seen in Fig. 4. Moreover, when scanning slightly di†erent frequency windows (e.g. 18.85È19.85 MHz) after a similar induction period the same enhancing e†ect could be observed but with the sharp rate increase appearing at di†erent frequencies. Reducing the scanning step width from 400 to 30 Hz (slower scanning speed) led to a longer induction period. The enhancement e†ect was found to be especially pronounced at speciÐc frequency windows corresponding to a minimum propagation loss of the AW, where the signal to the receiver is maximum. This is, therefore, simply a direct consequence of the amount of acoustic power that is absorbed by the catalyst over di†erent frequency ranges. UHV chamber experiments For all experiments carried out in the main UHV chamber a thermal e†ect was eliminated by monitoring the thermal shift of resonance peaks after each reaction rate measurement. Since the frequency shift of a given peak is proportional to the true sample temperature (Fig. 2) a plot of the reaction rates, measured with acoustic excitation on and o†, vs. frequency gives directly the enhancement factor due to AW generation, excluding the thermal e†ect. This “ true Ï acoustic e†ect was found to depend, amongst other things, on the reactant partial pressures and the catalyst temperature. Pressure and temperature dependence curves were measured. The diagram in Fig. 5 shows the dependence of the AW enhancement e†ect on the O /CO ratio at constant 2 temperatures. For three di†erent sample temperatures the CO production (%) with AW 2 on (Ðlled data marker) and AW o† (open data marker) is plotted vs. CO (%) in the reactant mixture, at constant oxygen partial pressure of 1.2 ] 10~7 Torr. The CO partial pressure was slowly increased stepwise to 8 ] 10~8 Torr. To ensure constant sample conditions throughout the experiment, the Ðrst measuring point was always veri-

View Article Online

441


S. Kelling et al.

Fig. 5 CO partial pressure dependence of the reaction rate enhancement on AW excitation measured at sample temperatures of 528, 445 and 387 K at a constant oxygen pressure of 1.2 ] 10~7 Torr. (…) AW on, (L) AW o†.

Ðed after reaching the highest CO pressure. From these plots we can draw the following conclusions for all three temperatures. Acoustic excitation does not inÑuence the reactivity at low CO partial pressures. With the AW on, the peak in reaction shifts toward higher CO/O ratios. At CO pressures just above that corresponding to the peak in the absence of the2 AW, the enhancement is always observed. In fact, the AW enhancement has precisely the appearance of a substrate temperature increase. The temperature dependence of the acoustic e†ect was measured in the range from 345 to 545 K at di†erent CO partial pressures. The plots for 70% and 26.6% CO partial pressure at an oxygen pressure of 1.2 ] 10~7 Torr are shown in Fig. 6. In both diagrams the curves with Ðlled data points were taken with AW generation on ; the reaction rate is shown as a function of the surface temperature, determined from the AW resonance

View Article Online


442


Fig. 6 Temperature dependence of the acoustic e†ect between 340 and 545 K sample temperature for a constant O pressure of 1.2 ] 10~7 Torr and CO pressures of 8.4 ] 10~8 Torr and 2 Torr (70% and 26.6% CO, respectively). 2 (…) AW on, (L) AW o†. 3.2 ] 10~8

peak shift, which unequivocally demonstrates that there is a true non-thermal AWinduced enhancement in the reaction rate. With the AW on, the reaction starts at lower temperatures. For higher CO pressures, where CO island formation occurs and O adsorption is the reaction rate-limiting step, the peak CO production with the AW on2 2 at high temperatures is still larger than without acoustic excitation. For the following PtM110N surface preparations no change was observed in the LEED pattern after 6 min of continuously scanning the acoustic spectrum from 19 to 20 MHz with a power of ca. 1 W clean (1 ] 2) at RT ; CO covered (1 ] 1) at RT ; oxygen (3 ] 10~8 Torr) following preadsorbed CO (5 ] 10~9 Torr), high background (1 ] 1) at 470 K ; and oxygen covered (1 ] 2) at RT and at 430 K.

Discussion From the high-pressure results we conclude that the AW reaction rate enhancement cannot be associated with speciÐc acoustic resonance frequencies, as has been claimed in the past by Brezhnev et al.18 From InoueÏs work on polycrystalline surfaces, the excitation frequency range does appear to play an important role when results are compared from di†erent transducers operating at 10 and 20 MHz, but not over the narrow range studied in the present work. Inoue attributes his results to a frequency-dependent change in the magnitude of the lattice displacement.19

View Article Online


S. Kelling et al.

443

An Arrhenius plot from the temperature dependence of the data is presented in Fig. 7. The diagram shows two parallel lines, for AW on and o†, respectively, leading to the conclusion that the activation energy for the reaction is una†ected by acoustic excitation. However, the curve for “ AW on Ï demonstrates a larger pre-exponential factor. The pressure dependence (Fig. 5) demonstrates that an acoustic e†ect can only be observed at higher CO pressures. In this range the catalyst surface is CO poisoned, CO islands are formed and O adsorption is the rate-limiting step. One possible reason for the 2 enhancing e†ect might be that, by acoustic excitation, the oxygen dissociative sticking probability is increased. The observed induction period is roughly similar to that observed by Ladas et al. in the development of rate oscillations on PtM110N where in situ LEED experiments revealed it to be due to a faceting process.20 It was shown that the formation of microfacets during the reaction is accompanied by a continuous increase in the catalytic activity. This increase was traced back to an enhanced oxygen sticking coefficient (under conditions of high CO coverage) on the faceted surface. In our experiments no change in the LEED pattern consistent with faceting could be seen, but AW generation might enhance some other morphological change in the surface. Stimulated by these results, we are currently developing a theoretical model for coupling low-energy acoustic waves into modes which can enhance the rate of catalytic reactions. The model is closely related to the phenomenon of superlubricity Ðrst predicted by Hirano and Shinjo in 1990.21 In the superlubricity regime, two contacting solid surfaces can slide without resistance. Shinjo and Hirano conclude in ref. 22 that the superlubric state appears when the system satisÐes the following two conditions : the sliding velocity is so slow that the atoms follow their equilibrium positions adiabatically, and the two solid surfaces contact incommensurably.21 In such a contact, the ratio between the lattice units of the surfaces is irrational along the sliding direction, so each individual atom experiences di†erent amounts of lateral force along di†erent directions. These forces consequently o†set each other, resulting in zero friction. This o†setting of forces is made possible by the continuous motion of atoms, which is the basic principle behind superlubricity. As Shinjo and Hirano point out, high dimensionality of systems is crucial for atoms to move continuously.22 Experimental evidence of the state of vanishing friction was recently obtained by Hirano et al. by sliding atomically clean surfaces across each other using a UHV scanning tunnelling microscope (STM).23 Superlubricity exists in realistic systems and appears for a wide class of (strong or weak) adhesion, such as metallic bonding and van der Waals interaction.

Fig. 7 CO production rate vs. inverse temperature in the range from 390 to 495 K measured with 2 AW generation on (=) and o† (K). O /CO \ 10/7. P(O ) \ 1.2 ] 10~7 Torr. 2 2

View Article Online


444


Our model incorporates the presence of partially incommensurate mesoscopic-scale surface structures. Islands of the required size have been observed for example, on a PtM110N surface by Rotermund24 using photoemission electron microscopy (PEEM). Such structures may consist of domains of relatively clean (1 ] 2) reconstructed areas within a sea of CO-covered PtM110N (1 ] 1). Depending on the degree of lattice mismatch, a more or less efficient coupling of AWs to these mesoscopic scale domains can provide the energy needed to cause cooperative island di†usion and/or fragmentation. This would provide an efficient non-thermal route to enhance the catalytic reaction rate ; for example, the access of dissociated oxygen adatoms to the boundaries of CO islands may be improved. This model provides, at least qualitatively, a means of explaining the e†ects observed to date, and we are currently planning experiments to test the validity of this approach. Note that thermal excitation should induce the same e†ect which, at a given temperature, would be simply enhanced by coupling to AWs. The model is readily applied to the M100N surface of Pt, since the clean surface reconstructs to form a stable hexagonal top layer which is almost incommensurate with the second layer, which retains the square symmetry of the ideal M100N surface. An adsorbate lifting the reconstruction to form the ideal (1 ] 1) structure will leave large areas of clean Pt hex surface. By exciting the surface acoustically it is expected that mesoscopic areas of the hex surface will be induced to slide across the surface, or to break up, increasing the chemical reactivity. We would like to acknowledge Professor Inoue of Nagoaka University of Technology, Japan, for providing the IDT crystals ; and J. Chevallier from Aarhus University, Denmark, for preparing the thin single crystals. The EU is acknowledged for Ðnancial support to S.K. and to T.M.

References 1 For a review see : M. Gruyters, T. Mitrelias and D. A. King, Appl. Phys. A, 1995, 61, 243. 2 T. J. Mason and J. P. Lorimer, Practical Sonochemistry, Ellis Horwood, New York, 1991. 3 B. P. Barber, R. Hiller, K. Arisaka, H. Fetterman and S. J. Putterman, J. Acoust. Soc. Am., 1992, 91, 3061. 4 R. J. Zanetti, Chem. Eng., 1992, 99, 37. 5 K. S. Suslick, Science, 1990, 247, 1439. 6 N. A. Maksimenko and M. A. Margulis, Russ. J. Phys. Chem., 1992, 66, 396. 7 Y. Inoue, Y. Matsukawa and K. Sato, J. Chem. Phys., 1992, 96, 2222. 8 Y. Inoue and Y. Matsukawa, Chem. Phys. L ett., 1992, 198, 246. 9 Y. Inoue, J. Chem. Soc., Faraday T rans., 1994, 90, 815. 10 Y. Inoue, Y. Watanabe and T. Noguchi, J. Phys. Chem., 1995, 99, 9898. 11 Y. Inoue, M. Matsukawa and H. Kawaguchi, J. Chem. Soc., Faraday T rans., 1992, 88, 2923. 12 T. Mitrelias, S. Kelling, R. I. Kvon, V. P. Ostanin and D. A. King, Surf. Sci., submitted. 13 S. Kelling, T. Mitrelias, Y. Matsumoto, V. P. Ostanin and D. A. King, J. Chem. Phys., in press. 14 T. Mitrelias, S. Kelling, M. Gruyters and D. A. King, Appl. Phys. L ett., 1996, 69, 3677. 15 T. Mitrelias, V. P. Ostanin, M. Gruyters and D. A. King, Appl. Surf. Sci., 1996, 101, 105. 16 J. L. Falconer and R. J. Madix, Surf. Sci., 1985, 160, 393. 17 Y. Y. Yeo, C. E. Wartnaby and D. A. King, Science, 1995, 268, 1731. 18 V. N. Brezhnev, A. I. Boronin, V. P. Ostanin, V. S. Tupikov and A. N. Belyaev, Chem. Phys. L ett., 1992, 191, 379. 19 Y. Inoue, personal communication. 20 S. Ladas, R. Imbihl and G. Ertl, Surf Sci., 1988, 197, 153. 21 M. Hirano and K. Shinjo, Phys. Rev. B, 1990, 41, 11837. 22 K. Shinjo and M. Hirano, Surf. Sci., 1993, 283, 473. 23 M. Hirano, K. Shinjo, R. Kaneko and Y. Murata, Phys. Rev. L ett., 1997, 78, 1448. 24 H. H. Rotermund, Surf. Sci., 1993, 283, 87. Paper 7/03246C ; Received 12th May, 1997