689
2005, 109, 689-691 Published on Web 12/23/2004
Inclined N2 Desorption in a Steady-State N2O + CO Reaction on Pd(110) Yunsheng Ma,† Ivan Kobal,‡ and Tatsuo Matsushima*,† Catalysis Research Center, Hokkaido UniVersity, Sapporo 001-0021 Japan, and J. Stefan Institute, 1000 Ljubljana, SloVenia ReceiVed: NoVember 12, 2004; In Final Form: December 11, 2004
The angular and velocity distributions of desorbing products were analyzed in the course of a catalyzed N2O + CO reaction on Pd(110). The reaction proceeded steadily above 450 K, and the N2 desorption merely collimated sharply along 45° off the surface normal toward the [001] direction. It is proposed that this peculiar N2 desorption is induced by the decomposition of adsorbed N2O oriented along the [001] direction. On the basis of the observation of similar inclined N2 desorption in both NO + CO and N2O + CO reactions, the N2 formation via the intermediate N2O(a) dissociation was confirmed in catalytic NO reduction.
I. Introduction. Nitrous oxide (N2O) reduction on palladium surfaces has received much attention because N2O is one of the undesired products in the catalytic NO reduction in automobile gas converters. However, knowledge of the reaction of adsorbed N2O is still limited.1 This letter is the first to deliver the angular and velocity distributions of desorbing products N2 and CO2 in a steady-state N2O + CO reaction on Pd(110) and assigns N2O(a) the key intermediate in the catalyzed NO reduction.2-5 The reaction steadily proceeded above 450 K and N2 desorption collimated sharply along 45° off the normal toward the [001] direction. The inclined N2 emission was commonly observed in NO + CO and NO + H2 reactions on Pd(110),6 indicating no direct contribution to N2 formation from a possible intermediate NCO(a), which was recently proposed from infrared reflection spectroscopy work.7 It has been difficult to analyze the decomposition dynamics of N2O under steadystate conditions because of the large fragmentation of N2O in a UHV chamber with an ion pump as well as in a mass spectrometer ionizer compared with the small N2 yield from a well-defined surface. In the present work, angle-resolved (AR) product desorption was successfully analyzed in a steady-state N2O + CO reaction. II. Experiments. The experiments were carried out in an ultrahigh-vacuum apparatus composed of three separately pumped chambers.1,4 The reaction chamber was equipped with low-energy electron diffraction (LEED) and X-ray photoelectron spectroscopy (XPS) optics, an Ar+ gun, and a mass spectrometer (MS) for angle-integrated (AI) desorption measurements. The chopper house with a pumping rate of about 7 m3 s-1 has a narrow slit facing the reaction chamber and a cross-correlation random chopper blade. The analyzer has another MS for ARproduct desorption and time-of-flight (TOF) analyses. The time resolution for TOF measurements was 20 µs. 15N O was introduced through a gas doser with a small orifice 2 (diameter; 0.1 mm) about 2 cm from a sample crystal while ordinary CO was backfilled. The product15N2 and 12CO2 signals * Corresponding author. Fax: (+81-11-706-9120). E-mail: tatmatsu@ cat.hokudai.ac.jp. † Hokkaido University. ‡ J. Stefan Institute.
10.1021/jp0448043 CCC: $30.25
were mostly monitored in the AR form. Hereafter, these are simply described as N2 and CO2 in the text. The desorption angle (polar angle; θ) was scanned in the plane along the [001] direction.1 III. Results and Discussion. A. Temperature Dependence. The AR signal was obtained by the MS in the analyzer as the difference between the signal at the desired angle and that acquired when the crystal was away from the line-of-sight position. The maximum flux of N2 and CO2 desorption in the steady-state N2O + CO reaction was located at θ ) 45 and 0°, respectively. The AR N2 and CO2 signals at their collimation angles are shown vs the surface temperature (TS) in Figure 1a. Both signals became noticeable at 450 K, increased quickly to the maximum at 530 K and then decreased above 540 K. On the other hand, the AR N2 signal at θ ) 0° remained negligibly small over the temperature range studied. The AI signals for N2 involved large experimental uncertainty because its maximum intensity was less than 10% of the total signal. Only the (1 × 1) LEED pattern was observed during the steady-state reaction (PN2O ) 3.3 × 10-8 Torr, PCO ) 0.5 × 10-8 Torr) between 450 and 700 K, indicating no surface reconstructions. B. Angular and Velocity Distributions. The N2 desorption sharply collimated along 45(2° off the normal. The AR N2 signal at 520 K was approximated as {cos28(θ + 45) + cos28(θ - 45)} (Figure 1). The distribution was broadened into a cos20(θ ( 45) form at 720 K. No normally directed desorption component was found even at 800 K. On the other hand, the CO2 desorption sharply collimated along the surface normal as approximated in a {0.73 cos13(θ) + 0.27cos θ} form at 520 K. This agrees with the results of CO2 in a steady-state CO + NO reaction on Pd(110).4 Thus, the ratio of the AR N2 signal at θ ) 45° to that of CO2 at θ ) 0° somewhat decreased above TS ) 600 K. It is interesting to find the absence of the cosine component, which is characteristic of the desorption after thermalization to the surface temperature,8 in the N2 desorption in contrast to the results in the NO + CO reaction.4,5 This was further examined by velocity distribution analysis. Typical velocity distribution curves at different desorption angles at TS ) 520 K are shown © 2005 American Chemical Society
690 J. Phys. Chem. B, Vol. 109, No. 2, 2005
Letters
Figure 1. (a) Surface temperature dependence of AR product signals at their collimation angles in a steady-state 15N2O + CO reaction at 3.3 × 10-6 Torr of 15N2O and 0.5 × 10-6 Torr of CO. Angular distribution of desorbing 15N2 at TS ) (b) 520 and (c) 720 K. (d) Angular distribution of desorbing CO2 at TS ) 520 K. A typical deconvolution was drawn by the broken curves.
in Figure 2. The distribution did not involve the component expected by the Maxwell distribution at the surface temperature (the dotted curve), supporting no cosine component. The translational temperature calculated from the average kinetic energy (〈E〉) as T〈E〉 ) 〈E〉/2k, is shown in brackets in the figure, where k is the Boltzmann constant. The value was maximized to 3250 K at the collimation angle, consistent with the inclined desorption. The velocity curve is wide even at the collimation angle, yielding 1.1 for the speed ratio defined as (〈V2〉/〈V〉2 1)1/2/[(32/9π) - 1]1/2, where V is the velocity of the molecule, 〈V〉 is the mean velocity and 〈V2〉 is the mean square velocity. The speed ratio describes the distribution width and is usually below unity for a hyperthermal component around the collimation position.8 Hence, the distribution curve was deconvoluted into two components of the modified Maxwellian distribution, f(V) ) V3 exp {-(V - V0)2/R2}, where V0 is the stream velocity and R is the width parameter. Here, we simply assumed a common width parameter.4 The resultant deconvolutions are shown by broken curves in Figure 2a-c. The faster component showed 5200 K at θ ) 45° and the slower one yielded 1800 K. These values decreased with increasing shift from the collimation position. These results are very close to those in the NO + CO reaction, supporting the N2 emission model through the N2O intermediate dissociation in the latter.4,5 These fast components were proposed to be due to the different vibrational states.3 Experiments with higher energy resolutions are highly desired.1 On the other hand, the velocity distribution of desorbing CO2 involved the component from the Maxwell distribution at the surface temperature (Figure 2d). The apparent translational temperature reached 1380 K at the normal direction. The velocity distribution curves after subtraction of the thermalized
component yielded the translational temperature of 1780 K, which is close to the values reported in NO + CO and CO + O2 reactions.1,4 C. Comparison with the NO + CO Reaction. The N2 yield in the N2O + CO reaction was about 20% of that in the NO + CO reaction under the optimum conditions at the same NO and N2O pressure. This small yield is due to the small N2O dissociation or adsorption probability. In fact, the sticking coefficient of N2O is about 0.8 at 85 K and decreases quickly above 400 K,9 whereas that of NO is kept at about 0.5 on the clean surface to about 550 K.10 Fortunately, the N2 desorption is concentrated in a narrow desorption angle range, yielding observable AR signals. On the other hand, the AI signal disappeared in the background even at the use of the doser with a fine orifice.4 The high yield of N2O including the inclined N2 desorption in the NO + CO reaction indicates that the reaction of NO(a) with N(a) is fast enough to keep large vacant areas on the surface. In the N2O + CO reaction, the N2 desorption merely collimated in the inclined way. Neither the normally directed desorption nor the thermalized component was found even at high temperatures. On the other hand, in a steady-state NO + CO reaction, three removal pathways of surface nitrogen, i.e., (i) 2N(a) f N2(g), (ii) NO(a) + N(a) f N2O(a) f N2(g) + O(a), and (iii) N2O(a) f N2O(g), are operative.4,5 The thermalized component of desorbing N2 was suggested to be due to process i. The present work supports this speculation. The absence of the thermalized component suggests that no interaction results from leaving N2 on the surface. The structure of N2O on Pd(110) was studied by infrared absorption spectroscopy, showing the terminal nitrogen atom
Letters
J. Phys. Chem. B, Vol. 109, No. 2, 2005 691
Figure 2. Velocity distributions of desorbing (a-c) 15N2 at different angles and (d) CO2 at the normal direction at TS ) 520 K under the conditions in Figure 1. The average kinetic energy is indicated in brackets in temperature units. The dotted curve shows the position of a Maxwell distribution at the surface temperature. Typical deconvolutions are shown by broken curves. The solid line indicates their summation.
surface.9
interacting with the However, this form is not the precursor for N2 emission because oxygen is released on the surface after N2O(a) dissociation. Density functional theory calculations in a generalized gradient approximation level predict two stable adsorption forms on Pd(110), i.e., one lying along the [001] direction and the other standing with the terminal nitrogen-metal bond.11 In recent near-edge X-ray absorption fine-structure work at around 60 K,12 N2O(a) was suggested to be either the standing form or that lying along the [001] direction from the remarkable anisotropy in the polarization dependence of the π resonance peaks. In fact, a lying form is inactive to the vibrational spectroscopy due to the surface selection rule. Thus, the inclined N2 emission was proposed to be induced in the decomposition of the parallel intermediate N2O(a) in NO reduction and was well reproduced in the N2O + CO reaction, confirming this intermediate mechanism.1 The collimation angle
was proposed to be controlled by the interaction of leaving N2 toward both deposited O(a) and the surface site atoms.2 The cosine component is frequently observed in thermal desorption of adsorbed species. On the other hand, sharply collimated desorption is limited in some associative processes, for example, 2H(a) f H2(g) and CO(a) + O(a) f CO2(g).1,8 The product molecule is commonly bulky and likely to be repelled by the surface at the moment of formation. When the repulsion is not strong enough, the molecules will be trapped on the surface by dispersion forces exerted from the surface. Thus, desorbing molecules may more or less involve the thermalized component. In fact, the component was found in desorbing N2 and CO2 products in TOF measurements in the NO + CO reaction.4,5 The product N2 from N2O(a) f N2(g) + O(a) is not trapped on the surface, and that from N(a) + N(a) f N2(g) is largely trapped. This can be explained by the fact that a large amount of energy is released due to the formation of the strong O-metal bond in N2O(a) dissociation, while the released energy is much less because of the strong N-metal bond in the N(a) associative process. Ikai et al. proposed that the reaction of N(a) + NO(a) is responsible for the off-normal N2 desorption by using an isotopelabeled experiment under temperature-programmed-desorption conditions.13 However, our results showed that no differences were found in the inclined N2 desorption dynamics between the reactions of NO + CO and N2O + CO. This suggests that the inclined N2 desorption is simply induced in the N2O(a) decomposition event rather than by the reaction of N(a) + NO(a) to give N2(g) directly. IV. Summary. The N2O + CO reaction on Pd(110) steadily proceeded above 450 K and the N2 desorption sharply collimated along 45° off the normal toward the [001] direction. Compared with the desorption of products in the NO + CO reaction, the N2O(a) intermediate in the latter was proposed to emit N2 in the inclined way. References and Notes (1) Matsushima, T. Surf. Sci. Rep. 2003, 52, 1. (2) Ohno, Y.; Kimura, K.; Bi, M.; Matsushima, T. J. Chem. Phys. 1999, 110, 8221. (3) Kobal, I.; Kimura, K.; Ohno, Y.; Matsushima, T. Surf. Sci. 2000, 445, 472. (4) Rzeznicka, I. I.; Ma, Y.-S.; Cao, G.; Matsushima. T. J. Phys. Chem. B 2004, 108, 14232. (5) Ma Y.-S.; Rzeznicka, I.; Matsushima, T. Chem. Phys. Lett. 2004, 388, 201. (6) Ma Y.; Matsushima, T. J. Phys. Chem. B 2004, 109, in press. (7) Ozensoy, E.; Goodman, D. W. Phys. Chem. Chem. Phys. 2004, 6, 3765. (8) Comsa, G.; David, D. Surf. Sci. Rep. 1985, 5, 145. (9) Haq, S.; Hodgson, A. Surf. Sci. 2000, 463, 1. (10) Sharpe, R. G.; Bowker, M. Surf. Sci. 1996, 360, 21. (11) Kokalj, A.; Kobal, I.; Matsushima, T. J. Phys. Chem. B 2003, 107, 2741. (12) Horino, H.; Rzeznicka, I.; Matsushima, T.; Takahashi, K.; Nakamura, E. UVSOR ActiVity Report 2002; Institute for Molecular Science: Okazaki, Japan, 2003; p 209. (13) Ikai, M.; Tanaka, K. I. J. Phys. Chem. B 1999, 103, 8277.
