Upload
yoshio-miura
View
213
Download
1
Embed Size (px)
Citation preview
Effects of the kinetic energy on the hydrogenabstraction dynamics on Cu(110)
Yoshio Miura a,b,*, Hideaki Kasai b,*, Wilson Agerico Di~nno b
a Japan Science and Technology Corporation, Kawaguchi, Saitama 332-0012, Japanb Department of Applied Physics, Osaka University, 2-1 Yamadaoka, Suita, Osaka 565-0871, Japan
Abstract
We investigate and discuss the effects of the kinetic energy (Et) of the incident hydrogen atom on the hydrogenabstraction dynamics on Cu(1 1 0) by performing quantum dynamics calculations on an ab initio potential energy
surface. Our calculation results show that the hydrogen abstraction probabilities on Cu(1 1 0) initially increase, then
decrease, with increasing Et, in the energy range Et ¼ 0:05–1.0 [eV]. Furthermore, we show that the mean vibrationalenergy of the product hydrogen molecules increases with increasing Et, while the mean surface parallel translationalenergy of the product hydrogen molecules decreases with increasing Et. From these results, we can conclude that in-creasing Et mainly enhances the excitation of the vibrational motion, rather than the surface parallel motion of producthydrogen molecules.
� 2003 Elsevier Science B.V. All rights reserved.
Keywords: Models of surface chemical reactions; Molecular dynamics; Chemisorption; Atom–solid interactions; Diffusion and
migration; Quantum effects; Copper; Hydrogen atom; Hydrogen molecule
1. Introduction
Investigations of hydrogen-surface reactions are
essential for obtaining a fundamental under-
standing of dynamical processes occurring on solid
surfaces, and performing efficient control of in-dustrially important gas-surface reactions. One of
the most challenging topics in the study of
hydrogen-surface reactions, both experimentally
[1–9] and theoretically [10–17], is hydrogen
abstraction on metal and semiconductor surfaces.
When a gas-phase hydrogen atom impinges a hy-
drogen pre-adsorbed surface, it abstracts an ad-
sorbed hydrogen atom in a single collision, then a
hydrogen molecule desorbs. This process corre-
sponds to �direct abstraction�. Otherwise, it isscattered back to the gas-phase, or trapped on the
surface. The trapped hydrogen atom diffuses along
the surface, and reacts with other adsorbed hy-
drogen atoms, then a hydrogen molecule desorbs.
This process corresponds to �indirect abstraction�.In a previous work [18], we investigated and
discussed the dynamics of hydrogen abstraction on
Cu(1 1 1) by performing quantum dynamics cal-culations. We introduced a new theoretical ap-
proach that quantum mechanically considers both
the direct abstraction and the indirect abstraction.
*Corresponding authors. Address: Department of Applied
Physics, Osaka University, 2-1 Yamadaoka, Suita, Osaka 565-
0871, Japan. Tel.: +81-6-6879-7857; fax: +81-6-6879-7859.
E-mail addresses: [email protected]
(Y. Miura), [email protected] (H. Kasai).
0039-6028/03/$ - see front matter � 2003 Elsevier Science B.V. All rights reserved.doi:10.1016/S0039-6028(03)00429-1
Surface Science 532–535 (2003) 148–153
www.elsevier.com/locate/susc
We discussed how the coverage of initially ad-
sorbed hydrogen atoms on Cu(1 1 1) affects the
vibrational and surface-parallel translational en-
ergy distributions of product hydrogen molecules.
Our results showed that the mean vibrational en-
ergy of product hydrogen molecules hEvibi de-creases with decreasing coverage, while the mean
surface parallel translation energy of product hy-
drogen molecules hEparai increases with decreasingcoverage. We concluded that for the low coverage
case, hydrogen abstraction occurs mainly through
the indirect process, where the vibrational excita-
tion of product hydrogen molecules is suppressed,
and the excitation of the surface parallel transla-tional motion is enhanced, as compared with those
for the high coverage case, where the hydrogen
abstraction occurs through both direct and indi-
rect processes.
Here, we report on the effects of the kinetic en-
ergy of the incident hydrogen atom on hydrogen
abstraction on Cu(1 1 0). Due to the attractive in-
teraction between a hydrogen atom and Cu(1 1 0),the incident hydrogen atoms have large kinetic
energies at the surfaces (�1.5 [eV]). This energy istransferred to the translational and other internal
degrees of freedom of the product hydrogen mol-
ecules. Therefore, to demonstrate energy flow and
energy disposal in the hydrogen abstraction on
Cu(1 1 0), it is essential to obtain a microscopic
understanding of the hydrogen abstraction mech-anisms. We perform quantum dynamics calcula-
tions on an ab initio potential energy surface
(PES), using the time-independent coupled-chan-
nel method. We calculate the hydrogen abstraction
probabilities, the mean vibrational energy hEvibi,and the mean surface parallel translational energy
hEparai, as a function of the kinetic energy of inci-dent hydrogen atoms.
2. Theory
We consider three coordinates (r; Z;X ) as dy-namical variables to describe the direct and in-
direct abstraction. r is the relative distance of thetwo hydrogen atoms, Z is the center-of-mass co-ordinate of the two hydrogen atoms normal to
the surface, and X is the center-of-mass coordi-
nate of the two hydrogen atoms parallel to the
surface, along the [1 �11 0] direction on Cu(1 1 0).X ¼ 0:0 [�AA] denotes the position of a gas-phasehydrogen atom when it is just above the adsorbed
hydrogen atom on Cu(1 1 0). We consider the
reaction between an incident H atom and a targetD atom initially adsorbed on a hollow short-
bridge (HL-SB) site of Cu(1 1 0) [19]. The incident
H either directly or indirectly abstracts the ad-
sorbed D, and the resulting HD desorbs from the
surface. In our calculations, the position of the
adsorbed D is fixed during the abstraction. This
means that the motion of the desorbing HD
parallel to the surface is suppressed in our cal-culations. The hydrogen abstraction on Cu(1 1 0)
is about 2.5 [eV] exothermic reaction, and most
of this energy is transferred to the translational
energy normal to the surface of the desorbing
HD. Therefore, the approximation that the center
of mass position of the desorbing HD parallel to
the surface hardly changes, is reasonable to de-
scribe the motion of desorbing HD in the ab-straction on Cu(1 1 0).
In order to include both the direct and indirect
processes in the quantum dynamics calculations, it
is advantageous to transform these Cartesian co-
ordinates ðr; ZÞ into reaction path coordinatesðs; vÞ [20,21]. s corresponds to the reaction pathcoordinate along the potential minimum on a PES,
and v corresponds to the vibrational coordinateperpendicular to the reaction path, respectively. In
this paper, we leave the details of the transfor-
mation to Ref. [18], and show only the final form
of the Hamiltonian, which is given by
H ¼ � �h2
2lg�1 o
osg�1 o
os
�þ g�1 o
ovgo
ov
�
� �h2
2Mo2
oX 2þ V ðs; v;X Þ: ð1Þ
Here, g is the Jacobian of the transformationgðs; vÞ ¼ 1� v CðsÞ. The reaction path curvatureCðsÞ ¼ Cmax= coshðs� s0Þ, where Cmax ¼ 5:3 [�AA�1]
and s0 ¼ 0:994 [�AA].V ðs; v;X Þ is the PES in the reaction path coor-
dinate system, and we can separate the v- andX -dependencies in the form
Y. Miura et al. / Surface Science 532–535 (2003) 148–153 149
V ðs; v;X Þ ¼ Vcorðs;X Þ þ Vvibðs; vÞ: ð2ÞTo evaluate the Vcorðs;X Þ, we perform total en-ergy calculations for various configurations of theincident atom H on the D adsorbed Cu(1 1 0)
based on the density functional theory and the
generalized gradient approximation (PW91) for
the exchange correlation energy [22]. The total
energies are computed in a supercell geometry,
using plane waves and pseudopotentials, with the
ab initio density functional code DACAPO [23].
We include three Cu(1 1 0) layers and three vac-uum layers in the supercell. Each layer includes a
(3 2) unit cell, where 3 and 2 Cu atoms areplaced along the [1 �11 0] and [0 0 1] directions, re-spectively. The plane-wave basis set cut off kinetic
energy is 350 [eV]. The surface Brillouin zone
integration is performed using the special point
sampling technique of Monkhorst and Pack [24].
In the V ðs;X Þ, s ¼ þ1 corresponds to the con-figuration of a gas-phase H and a D adsorbed on
a HL-SB site of Cu(1 1 0), s ¼ 0 corresponds tothe configuration of two hydrogen atoms (H and
D) on Cu(1 1 0), and s ¼ �1 corresponds to the
configuration of two hydrogen atoms (H and D)
in the gas-phase. Then, we fit the ab initio PES
results to an analytical functional form. We as-
sume a cosine function and a Gaussian functionfor describing the interaction between a gas-phase
H and a D adsorbed on Cu(1 1 0).
Vcorðs;X Þ ¼ V0ðsÞ þ V1ðsÞ cosð2pX=aÞþ ½V2ðsÞ þ V3ðsÞX 2 expð�X 2Þ; ð3Þ
where a ¼ 2:556 [�AA] is the lattice constant ofCu(1 1 0). The ViðsÞ ði ¼ 0–3Þ are determined sothat the difference to the ab initio PES results on
the average is smaller than 10 [meV]. Here, we giveViðsÞ ði ¼ 0–3Þ as follows:
V0ðsÞ¼a0=coshðb0ðs�c0ÞÞþd0 tanhðe0ðs�f0ÞÞþg0;
V1ðsÞ¼a1=coshðb1ðs�c1ÞÞþd1 coshðe1ðs�f1ÞÞþg1;
V2ðsÞ¼a2=coshðb2ðs�c2ÞÞþd2 tanhðe2ðs�f2ÞÞþg2;
V3ðsÞ¼a3=coshðb3ðs�c3ÞÞþd3 coshðe3ðs�f3ÞÞþg3:
ð4ÞThe fitting parameters ai to gi (i ¼ 0–3) are givenin Table 1.
For Vvibðs; vÞ, we assume a harmonic type po-tential,
Vvibðs; vÞ ¼1
2lxðsÞ2v2; ð5Þ
xðsÞ is given byxðsÞ ¼ �hxgas � dxðtanhðs� s0Þ þ 1Þ; ð6Þwhere �hxgas ¼ 0:443 [eV], and dx ¼ 0:126 [eV].xðsÞ takes on the value of the vibrational fre-quency between an adsorbed D atom and Cu atomat s ¼ þ1, and the vibrational frequency of adesorbing HD at s ¼ �1.The basis set used in the coupled channel cal-
culations is determined by the incident energy
Et ¼ 0:050–1.0 [eV] and the exothermicity of thisreaction (2.5 [eV]). In this case, the calculation
results converge with maximum quantum numbers
m ¼ 75 and n ¼ 8, where m and n are the quantumnumbers for the surface parallel translational
motion and the vibrational motion of product
HD. We carefully checked the convergence for
calculations with maximum quantum number
m ¼ 80, n ¼ 9, respectively.
3. Results and discussion
We calculate the hydrogen abstraction proba-
bilities on Cu(1 1 0) as a function of the surface
normal kinetic energy Et of the incident H. As can
Table 1
The fitting parameters of the analytical functional V ðs;X Þ that represents the ab initio PES resultsai [eV] bi [�AA�1] ci [�AA] di [eV] ei [�AA�1] fi [�AA] gi [eV]
i ¼ 0 )3.82 1.08 1.12 )1.10 0.87 2.26 3.52
i ¼ 1 )1.41 )1.22 0.99 1.35 1.29 1.23 0.00
i ¼ 2 2.03 1.09 1.03 2.37 0.64 0.90 )2.40i ¼ 3 0.90 0.50 )0.30 0.50 3.00 1.80 0.00
150 Y. Miura et al. / Surface Science 532–535 (2003) 148–153
be seen in Fig. 1, the abstraction probabilities
initially increase, then decrease with increasing Et,taking a maximum at Et ¼ 0:20 [eV]. We can saythat increasing Et initially enhances the abstrac-tion. However, an additional increase in Et pre-vents the incident H from abstracting the adsorbed
D. The initial increase in Et < 0:20 [eV] can beattributed to the amplitude of the surface corru-
gation Ecor, which corresponds to the diffusionbarrier for the H trapped on Cu(1 1 0). From the
ab initio PES, Ecor � 0:20 [eV]. When an incidentH collides with the surface, it loses its normal ki-
netic energy (Elose) due to energy transfer fromsurface normal motion to surface parallel motion
as a result of the diffraction. Thus, the incident His trapped normal to the surface, but can move
parallel to the surface with an energy Eloss. (Here,we neglect the energy dissipation to surface pho-
nons.) If Eloss is smaller than Ecor, the incident H isalso trapped parallel to the surface. However, if
Eloss is larger than Ecor, it can move parallel to thesurface. After some time, it may bond with an-
other D adsorbed on the surface, and leave thesurface as a product HD. Therefore, the abstrac-
tion probabilities increase with increasing Et, untilthe incident H has a kinetic energy corresponding
to Ecor. On the other hand, we attribute the de-crease in the abstraction probabilities with in-
creasing Et for Et > 0:20 [eV] to the decrease of theindirect abstraction at high Et. Since the scattering
probabilities increase with increasing Et, the initialtrapping of the incident H on Cu(1 1 0) occurs less
frequently at high Et. This leads to a decrease inthe hydrogen abstraction through the indirect
process. Therefore, we see a decrease in the ab-
straction probabilities with increasing Et in Et >0:2 [eV]. Experiments [4] report that the scatteringprobabilities of the incident H on Cu(1 1 1) do not
increase with increasing Et in the energy range ofEt ¼ 0:07–0.3 [eV]. This means that the abstractionreaction probabilities increase with increasing Et inthis energy range. This kinetic energy dependence
in the experimental results [4] agrees with the ini-
tial increase we observe in Fig. 1 (in the energyrange Et ¼ 0:05–0.2 [eV]).We show the mean vibrational energy hEvibi and
the mean surface parallel translational energy
hEparai as a function of Et in Fig. 2. As can be seenin Fig. 2, for Et > 0:20 [eV], hEvibi and hEparai haveopposite Et dependence, i.e., hEvibi increases withincreasing Et, while hEparai decreases with increas-ing Et. When the hydrogen abstraction occurspredominantly through the indirect process, most
of the energy (large momentum parallel to the
surface) of the diffusing H will eventually be
transferred to the product HD. Therefore, the
decrease of hEparai with increasing Et indicates adecrease in hydrogen abstraction through the
Incident Energy Et [eV]
Abs
trac
tion
Pro
babi
litie
s
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0.00.0 0.20.1 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
Fig. 1. The hydrogen abstraction probabilities on Cu(1 1 0) as a
function of the surface normal kinetic energies Et of the incidentH.
Mea
n K
inet
ic E
nerg
y of
Pro
duct
HD
[eV
]
Incident Energy Et [eV]
Mean vibrational energyMean surface parallel translational energy
0.8 0.9 1.0
0.5
0.6
0.5
0.4
0.4
0.3
0.3
0.2
0.2
0.1
0.10.0
0.0 0.70.6
Fig. 2. The mean vibrational energy hEvibi (open-square lines-points) and the mean parallel translational energy hEparai (solid-triangle lines-points) of product HD as a function of the
incident energy Et.
Y. Miura et al. / Surface Science 532–535 (2003) 148–153 151
indirect process. On the other hand, when hydro-
gen abstraction occurs predominantly through the
direct process, the product HD will only have a
small momentum parallel to the surface. The ex-
cess energy coming from the exothermicity (2.5
[eV]) of the reaction is mainly used for the exci-
tation of the vibrational motion of product HD.
Therefore, the increase of hEvibi with increasing Etindicates that hydrogen abstraction preferentially
occurs through the direct process rather than the
indirect process at high Et.We show the vibrational distributions of prod-
uct HD desorbing from Cu(1 1 0) as a function of
Et in Fig. 3. We observe a strong Et dependence ofvibrational distributions. As can be seen in Fig. 3,
hydrogen abstraction probabilities for productHD in the vibrational ground state decrease with
increasing Et. However, hydrogen abstractionprobabilities for product HD in the vibrationally
excited states increase with increasing Et. Thismeans that in hydrogen abstraction, large amounts
of kinetic energies are transferred to the vibra-
tional motion of the product HD, thus exciting the
vibrational state.
4. Summary
In this paper, we investigate the effects of the
kinetic energy of the incident hydrogen atom on
the hydrogen abstraction dynamics on Cu(1 1 0) by
performing quantum dynamics calculations on the
ab initio PES. Our results show that increasing the
kinetic energy of the incident H initially enhances
the abstraction. However, additional increases in
the kinetic energy prevents the incident H from
abstracting the adsorbed D. Furthermore, themean vibrational energy of the product HD hEvibiincreases with increasing Et, while the mean sur-face parallel translational energy of the product
HD hEparai decreases with increasing Et. Theseresults indicate that additional increase in the ki-
netic energy of the incident H enhances the exci-
tation of the vibrational motion, rather than the
surface parallel motion of product HD. We alsoconclude that the hydrogen abstraction preferen-
tially occurs through the direct process, rather
than the indirect process, at high Et.In the present study, we employed an approxi-
mate treatment for the surface-parallel motion
and rotational motion of a hydrogen molecule.
For our future work, we will go beyond this and
also investigate the hydrogen abstraction dynam-ics on more industrially significant metal and
semiconductor surfaces, to obtain a more detailed
understanding for the selectivity of DIRECT/IN-
DIRECT reaction and the energy distribution of
product hydrogen molecules.
Acknowledgements
We thank Dr. Hiroshi Nakanishi for discussions
and comments on this work. This work is partly
supported by the Ministry of Education, Science,
Sports and Culture of Japan through their Grant-
in-Aid for COE Research (10CE2004), Scientific
Research(C) (11640375, 13650026) programs, the
New Energy and Industrial Technology Develop-ment Organisation (NEDO) through the Materials
and Nanotechnology program, the Japan Science
and Technology Corporation (JST) through their
Research and Development Applying Advanced
Computational Science and Technology program,
and the Toyota Motor Corporation through their
Cooperative Research in Advanced Science and
Technology program. Some of the calculationspresented here were done using the computer fa-
cilities of the JST, and the ISSP Supercomputer
Et=0.10 [eV]Et=0.25 [eV]Et=0.50 [eV]Et=0.75 [eV]Et=1.00 [eV]
Vibrational state of product HD
Abs
trac
tion
prob
abili
ties
0.35
0.3
0.25
0.2
0.15
0.1
0.05
0.00 1 2 3 4 5 6
Fig. 3. The vibrational distributions of product HD in the
hydrogen abstraction on Cu(1 1 0) as a function of the incident
energy Et.
152 Y. Miura et al. / Surface Science 532–535 (2003) 148–153
Center (University of Tokyo). Y.M. gratefully
acknowledges Fellowship grant from the JST
through their Research and Development Apply-
ing Advanced Computational Science and Tech-
nology program. W.A.D. acknowledges support
from the Marubun Research Promotion Founda-tion.
References
[1] C.T. Rettner, Phys. Rev. Lett. 69 (1992) 383.
[2] J. K€uuppers, S. Wehner, Th. Kammler, Surf. Sci. 339 (1995)125.
[3] G. Eilmsteiner, A. Winkler, Surf. Sci. 366 (1996) L750.
[4] C.T. Rettner, D.J. Auerbach, J. Chem. Phys. 104 (1996)
2732.
[5] S.A. Buntin, Chem. Phys. Lett. 278 (1997) 71.
[6] J. Boh, G. Eilmsteiner, K.D. Rendulic, A. Winkler, Surf.
Sci. 395 (1998) 98.
[7] Th. Kammler, D.K. Vellianitis, J. K€uuppers, Surf. Sci. 460
(2000) 91.
[8] M. Okada, K. Moritani, M. Nakamura, T. Kasai, Y.
Murata, Chem. Phys. Lett. 323 (2000) 586.
[9] J.Y. Kim, J. Lee, J. Chem. Phys. 113 (2000) 2856.
[10] M. Persson, B. Jackson, J. Chem. Phys. 102 (1995) 1078.
[11] J. Dai, J.C. Light, J. Chem. Phys. 110 (1999) 6511.
[12] S. Caratzoulas, B. Jackson, M. Persson, J. Chem. Phys. 107
(1997) 6420.
[13] M. Persson, J. Str€oomquist, L. Bengtsson, B. Jackson, D.V.Shalashilin, B. Hammer, J. Chem. Phys. 110 (1999) 2240.
[14] D.V. Shalashilin, B. Jackson, M. Persson, J. Chem. Phys.
110 (1999) 11038.
[15] B. Jackson, D. Lemoine, J. Chem. Phys. 114 (2001) 474.
[16] P. Kratzer, J. Chem. Phys. 106 (1997) 6752.
[17] M. Rutigliano, M. Cacciatore, G.D. Billing, Chem. Phys.
Lett. 340 (2001) 13.
[18] Y. Miura, H. Kasai, W.A. Di~nno, A. Okiji, J. Phys. Soc.Jpn. 71 (2002) 222.
[19] C. Bae, D.L. Freeman, J.D. Doll, G. Kresse, J. Hafner, J.
Chem. Phys. 113 (2000) 6926.
[20] Y. Miura, H. Kasai, W.A. Di~nno, A. Okiji, J. Phys. Soc.Jpn. 69 (2000) 3878.
[21] Y. Miura, H. Kasai, W.A. Di~nno, J. Phys. Soc. Jpn. 68
(1999) 887.
[22] J.P. Perdew, J.A. Chevary, S.H. Vosko, K.A. Jackson,
M.R. Pederson, D.J. Singh, C. Fiolhais, Phys. Rev. B 46
(1992) 6671.
[23] http://www.fysik.dtu.dk/CAMPOS/.
[24] H.J. Monkhorst, J.D. Pack, Phys. Rev. B 13 (1976) 5188.
Y. Miura et al. / Surface Science 532–535 (2003) 148–153 153