Surface Structure and Reaction Dynamics in Catalysis - American

20 Surface Structure and Reaction Dynamics in Catalysis 1

K. Christmann and G. Ertl

Downloaded by UNIV LAVAL on July 3, 2014 | http://pubs.acs.org Publication Date: October 16, 1985 | doi: 10.1021/bk-1985-0288.ch020

Institut für Physikalische Chemie der Universität München, D-8000 München 2, Federal Republic of Germany

The interaction of H and D with Ni (110) and Ni (111) was investigated using molecular beam, LEED, and thermal desorption (TD) techniques to elucidate the interaction dynamics and the surface structure. Elastic H scattering leads to marked diffraction only with Ni (110), whereas Ni (111) shows almost no corrugation. The difference in the elastically scattered H and D intensity reveals that the energy exchange is dominated by phonon and not by electronic excitations. The differences in the sticking probabilities s between the two faces suggest the absence of an activation barrier for Ni (110), but the existence of such a barrier (height ~0.1 eV) for Ni (111). Increasing H (D) coverages Θ induce various surface structures on Ni (110) including two reconstructed phases. A one-dimensionally ordered 'streak' phase is the only stable phase above 250 K. With Ni (111) no reconstruction occurs and the H (D) atoms are disordered at 300 K. It is shown that the functions s(Θ) are influenced by the formation of the reconstructed phases; the implications for surface reactivity are discussed. 2

2

2

2

2

o

H(D)

In the recent past much experimental and t h e o r e t i c a l e f f o r t has been undertaken t o understand the microscopic steps o f heterogeneous surface reactions. The main problem c o n s i s t s o f evaluating the t o t a l energy o f the r e a c t i n g components (including the surface atoms!) as a function o f a l l nuclear coordinates a t any reaction time. The s o l u t i o n o f t h i s problem i s extremely d i f f i c u l t . Detailed studies w i t h model systems, however, can shed some l i g h t upon the various steps o f the i n t e r a c t i o n pattern. Even the f i r s t step, the c a l c u l a t i o n o f the multidimensional ground s t a t e energy o f a s i n g l e p a r t i c l e i n t e r a c t i n g w i t h a surface implies a l o t o f d i f f i c u l t i e s . I t i s , f o r example, p o s s i b l e that the spacings 'Current address: Institut fur Physikalische Chemie der FU Berlin D-1000 Berlin 33, Takustr. 3, Federal Republic of Germany

0097-6156/85/0288-O222S06.00/0 © 1985 American Chemical Society In Catalyst Characterization Science; Deviney, M., et al.; ACS Symposium Series; American Chemical Society: Washington, DC, 1985.

20. CHRISTMANN AND ERTL

223

Surface Structure and Reaction Dynamics

of varioxos adjacent surface atoms change upon interaction with an adsorbate atom (surface relaxation or reconstruction). It i s , e.g., well-known that chemisorbed hydrogen tends to reconstruct a number of metal surfaces, e.g., Ni, Pd, Mo, or W (1). The second step has to include the multi-particle effects, and the total energy as a function of a l l geometrical configurations of the particles has to be evaluated.


The treatment of the reaction dynamics represents a further d i f f i c u l t barrier i n the understanding of the heterogeneous reaction. I t re quires the knowledge of the cross sections determining the transiti ons between different states which i n turn are correlated with the energy transfer into the various degrees of freedom. In this article, we shall concentrate only on very simple surface reactions, namely, the adsorption of H~ and D and the H /D exchange reaction on the differently structured wi( 1107 and (111) surfaces. Our experimental techniques comprise static techniques such as LEED, thermal desorption spectroscopy (TDS) and work function measurements (Δρ) and dynamic techniques like scattering of H and D molecular beams. Details of the experimental methods are given elsewhere (2,3). 2

2

2

2

2

The goal of this paper i s to demonstrate the internal correlation between the surface structure and the reaction dynamics which, i n ex treme cases (Pt(100)l) can give rise even to kinetic surface o s c i l l a tions (4). The Ni(110)/H system i s another example where not only se veral lattice gas phases are formed by hydrogen but also reconstruc tion can occur depending on the temperature and the chemical potenti a l of the adsorbed hydrogen. Phenomena like 'surface explosions have been reported previously with this surface i n the course of surface reactions, e.g., the decomposition of formic acid (5). We argue that structural changes of the surface during a heterogeneous chemical re action may be responsible for effects of this kind. 1

Experimental The experiments were carried out i n two different UHV* chambers equip ped with the standard f a c i l i t i e s to clean and to characterize a metal single crystal surface. One apparatus was used for the 'static' inve stigations and contained a Video-LEED system including data processor, a mass spectrometer for TDS, and a Kelvin probe for measuring contact potential differences. The other apparatus consisted of a molecular beam source connected to a UHV scattering chamber with a rotatable mass spectrometer as well as LEED and Auger f a c i l i t i e s . In the mole cular beam (1VB) source, the hydrogen was expanded through a 0.07 mm hole at pressures between 200 and 500 torr. The beam diameter and an gular divergence were determined by an aperture i n the final pumping stage of the source. At the surface, beam diameters of 1.5 - 2 mm and angular divergence of 0.06° were obtained. Owing to the comparatively mild expansion conditions of the nozzle the rotational distribution of the molecules remains essentially unaltered. Therefore the nozzle temperature determines directly the translational kinetic energy of the pure H or D beams. Kinetic energies of 26 meV and 64 meV beams corresponded to nozzle temperatures of 120 Κ and 300 K, respectively. Separation of molecules scattered directly from those being trapped 2

2

In Catalyst Characterization Science; Deviney, M., et al.; ACS Symposium Series; American Chemical Society: Washington, DC, 1985.

CATALYST CHARACTERIZATION SCIENCE

224

in the chemisorption potential was possible by nodulating the i n c i dent beam at 63 Hz and using a lock-in technique.


The Ni (110) and (111) samples were prepared and cleaned i n the usual manner (6). In the M3 experiments, the azirauthal orientation was cho sen such that the most corrugated direction was directed to the i n c i dent beam: The [001] direction for Ni(110) and the [112 ] direction for Ni(111 ). Since both surfaces chemisorb hydrogen readily the ex perimental conditions i n the experiments were adjusted to assure scattering from surfaces covered with less than 0.01 monolayers of H at a l l times. The sample orientations as well as the perfection of the surfaces were checked by LEED and by monitoring the relative i n tensities of the specularly reflected He beams. Results The interaction of hydrogen (deuterium) molecules with a transition metal surface can be conveniently described i n terms of a Lennard-Jones potential energy diagram (Fig. 1 ). I t consists of a shallow molecular precursor well followed by a deep atomic chemisorption po tential. Depending on their relative depths and positions the^wells may or may not be separated by an activation energy barrier Ε as schematically indicated by the dotted curve i n Fig. 1. Here we shall be concerned with the interaction of incaning diatomic molecules (Η-, DO with either types of potential energy wells: The molecular interaction (responsible for elastic and direct-inelastic scattering with extremely short residence times of the impinging mo lecules i n the potential) and the chemisorptive interaction (leading to dissociative adsorption and associative desorption, respectively, and associated with H (D) atoms trapped i n the chemisorption potenti a l for an appreciable time). The elastic scattering. The elastic scattering of a diatomic mole cule obeys the rules for conservation of energy and momentum; there i s no energy exchange between the surface and the incident particle. However, the possibility for internal transformation of translational into rotational energy states has to be taken into account particular l y withron-synimtriciiulecniLes like HD as was, e.g., shown by Oowin et a l . (7) for a Pt(111) surface. Translation - vibration energy transformation requires too much energy to occur under our experimen t a l conditions. The modulated beam technique conveniently allows a separation of the elastically scattered molecules from those which have been formed by recombination from chemisorbed atoms. The angular distribution of the scattered molecules i s a direct probe to distin guish elastically scattered from inelastically scattered particles (8) . In Fig. 2a,b we display the angular distributions obtained for H molecules scattered from Ni(110) and N i ( H l ) i n the zero-coverage limit. The azirauthal orientation of the surfaces i s indicated i n the figure. 2

For Ni (110) the angular distribution exhibits pronounced diffraction peaks, i n addition to an intense specularly reflected beam. The po sition of the diffraction maxima agrees with the prediction from the known lattice constant and the i n i t i a l velocity of the particle. With Ni (111)

there i s practically no

diffraction visible which clear-



20.

CHRISTMANN AND ERTL


225

Figure 1. One-dimensional Lennard-Jones potential energy diagram for adsorption of a caatomic molecule (hydrogen). ρ denotes the reaction coordinate.

e

-10 0 10 20 30 40 50 60 70 v}[J

•10 0

10 20 30 40 50 60 70 ^ [°] f

Figure 2. Angular distribution of the relative scattered intensi ty of hydrogen molécules scattered from a) Ni(110); b) Ni(111). The angle of incidence of the molecular beam and the azimuthal orientation of the surface are indicated i n the figure. In Catalyst Characterization Science; Deviney, M., et al.; ACS Symposium Series; American Chemical Society: Washington, DC, 1985.

226


ly demonstrates the much smoother nature of this densely packed sur face. With both surfaces H- scattering leads only to a very small background with a flat maximum in the specular direction. Apparently, most of the incident molecules suffer true elastic scattering. Angular distributions for D scattering from the same crystal planes are presented in Fig. 3a, b. Again, pronounced diffraction features are apparent only with Ni (110). Interestingly, however, there i s now a much larger fraction of diffuse background intensity observed which peaks strongly i n the specular direction. This background i n tensity clearly arises from the so-called direct-inelastic scattering which w i l l be described next.


2

The direct-inelastic scattering. This type of scattering originates from energy exchange processes between the incident molecule and the surface during a single collision event. Multiple collisions are, to a f i r s t approximation, negligible owing to the small mass ratio m/M where m denotes the mass of the incident particle and M stands for the mass of the surface atom. The energy exchange can occur through either electron-hole pair excitation (whereby electrons are excited from just below to just above the Fermi level, a process also refer red to as 'electronic f r i c t i o n ) , or through excitation of metal phonons which couple to resonance states of the incoming particle. Accor dingly, the f i r s t process should depend on the electronic configurati on of the particle and the density of states at the Fermi level of the substrate, but not on the mass of the incident particle, whereas the second energy dissipation mechanism, the phonon interaction, should be a function of the mass ratio m/M, that i s , of the mass m of the incident particle. A comparison of the intensities of the elastically scattered beams of hydrogen and deuterium from the same sur face reveals that the elastically scattered intensity (for particles with equal velocity, i.e., equal residence time near the surface) i s appreciably larger for hydrogen than for the heavier deuterium mole cule. I t i s thus indicated that the acxxmitiodation or dissipation of the kinetic energy of the incident particle i s dominated by excita tion and annihilation of phonons and not by electron-hole pair exci tation. This conclusion i s in agreement with recent theoretical cal culations (9) which showed that the latter process occurs only with very small probability. 1

The dissociative adsorption. From the MB studies i t appears that a great fraction of incident molecules dissociates. The H atoms are held so tightly on the surface that, even around 300 K, their rate of recombination and thermal desorption i s quite small, at least at low surface concentrations Θ · The stickiJig_probabil.ity_s^ We refer to the sticking probability as to the relative probability for the chemisorption process to occur with respect to the total number of molecules incident on the surface, s i s determined from the area under a TDS curve as described elsewnere (10). For Ni (110), we obtain s = 0.96 for equal surface and gas phase temperatures Τ = Τ = 300 Κ. Our value agrees well with data re ported i n the literature (11); we also point out that, within our l i mits of accuracy, no difference was found between s(H) and s(D). I t is also important to note that for Ni (110) s does not depend on Τ between 100 Κ and 800 K. In contrast to this°face, s i s much lower for Ni(111), namely s =0.1, which i s in agreement with numbers reΗ

9

Q


20. CHRISTMANN AND ERTL

227


ported previously (12). Unlike Ni(110)/H, we find with Ni(111)/H a pronounced dependence of S on the angle of incidence, θ^: Q

β

θ

Λ

< ι>

ο

1

s

= ~ (θ,= 0°) · c o s V Ο

1

(1) 1

This result indicates that s depends on the kinetic energy of the itomentum perpendicular to th§ surface, . By changing either the angle of incidence θ. or the nozzle temperature we could vary between 26 meV and lJo meV. We observe a significant increase of s from very small values (< 0.1 ) at the lowest up to about 0.4 at = 120 meV. The shape of the relation s () suggests that higher values would lead to even higher sticking coefficients so that eventually values similar to Ni(110)/H could be expected. We have also investigated the dependence of s on the temperature, for = 64 meV. As with Ni (110), s does apparently not depend on the temperature. This finding i s i n agreement with a report by Rendulic and Winkler (13). Our results can be rationalized i n terms of the po tential energy diagram of Fig. 1. They suggest the existence of an activation barrier Ε >0 for the Ni(111)/H system, whereas no such barrier i s present with H/Ni(l10). Obviously this barrier can be over come by hydrogen molecules with sufficiently high translational ener gy (as suggested by the dependence). An increase of the thermal energy of the solid, on the other hand, does not make the sticking process more efficient. Such an influence of the surface temperature would only be expected i f dissociation would occur through a molecug larly held precursor state. Not only would the activation barrier Ε be overcome by the thermal energy of the surface but also the phonon assisted trapping i n the molecular state should then increase with T . In contrast, our results demonstrate that direct collision from the gas phase i s the most efficient channel for dissociation and trapping i n the chemisorption potential. X


Q

g

By extending the TDS technique mentioned before to finite coverages, G^,the function s(6U) may be evaluated. As w i l l be shown i n the next section, there are various different surface phases formed with Ni (110) as the hydrogen coverage increases from 0 to one monolayer. At sufficiently high temperatures (T>250 K) chemisorbed hydrogen causes a reconstruction of the 1x1 surface to a phase ordered i n only one direction , the so-called 'streak phase (2,14). Fig. 4 shows the variation of the relative sticking coefficient, s/s , as a func tion of the relative coverage, Θ/Θ . Apparently, s remains constant up to about 40% of the saturation coverage and thereafter decreases to zero. The way i n which this data was obtained, namely by TDS, how ever, brings i t about that any fine structure i n s(©) maybe largely obscured. The situation changes somewhat i f the hydrogen i s adsorbed at low temperatures (120 K) and the TD program i s run at ve ry high, heating rates so that thermally activated phase transforma tions cannot gain importance. Adsorption at 120 Κ then leads to a s(0) function which i s shown in Fig. 4 as a dotted line. Again, there i s a constant sticking probability up to about 0.5 monolayers, there after s drops as expected. However, around 1 monolayer coverage there occurs a spontaneous reconstruction to a 1x2 phase which i s connected with the generation of 0.5 monolayers of additional H adsorption s i tes. In the s{@) curve this leads to a second plateau at €^ = 1 when 1



228

1

®

00

Io

[%]

Ni(110) D +

f\

2

°^ i l

0.2 3] = 57,6°

/

\

02 /

0.1


. ..······· -10 0 10 20 30 AO 50 60 70 ^[']

1

(D

00

Ni(111)*D

2

*=26°

-10

0

10 20 30 AO 50 60 70 & [°] f

Figure 3· Angular distribution of the relative scattered intensi ty of deuterium molecules scattered from a) Ni(110); b) Ni(111). The experimental conditions were the same as i n Figure 2. s/s1.0 0

Figure 4. Variation of the relative sticking probability, s/s , with Θ(Η); . for H on Ni (110) . Hie dashed line refers to experiments^fi which the hydrogen was adsorbed at 120 Κ and rapidly desorbed (see text for details). 2


20.

CHRISTMANN AND ERTL


229

the additional adsorption sites become occupied. With Ni(111), the coverage dependence of s differs markedly from that of Ni (110) as i s evident from Fig. 5. Not only i s s much lower for reasons explained before/ but s/s decreases quite sxeeply even at f a i r l y low co verages. Again,°this behavior has been reported previously (11,12). From LEED work (15) i t i s known that, below 270 K, an ordered c2x2 H phase i s formed at O. = 0.5 i n which the H atoms are arranged i n a honeycomb-like structure. The Ni(111) surface thereby does not re construct. Unlike the Ni(110)/H system, the formation of the c2x2 phase does not have any effect on the sticking probability coverage function. I t appears from our data that such an effect i s only to be expected i f the metal surface undergoes reconstruction and provides a different number of adsorption sites as compared to the unrecon structed surface. On the other hand, the sinple fact that a surface i s covered with an ordered or disordered overlayer i s apparently not sufficient to alter the s(9) relation noticeably.


Q

The hydrogen surface phases. 1. Ni (110). As described i n more detail elsewhere (2,6,14,16) hydrogsn adsorption at 120 Κ leads to a variety of surface phases, depending on the temperature and the local covera ge. In the submonolayer regime (0 )

P

*D

Surface Structure and Reaction Dynamics in Catalysis - American

Recommend Documents