A structural effect in direct reactions: Kinetics of D abstraction from Pt(110) 1×2 surfaces with gaseous H atoms

Article abstract of DOI:10.1016/S0009-2614(98)00075-X

Deuterium-covered Pt(110) surfaces were subjected to a flux of thermal H atoms at 100 K. The rates of product formation, HD and D₂, were measured simultaneously with H atom exposure as a function of D coverage. D₂ as a product and the kinetics of HD formation contradict the operation of Eley-Rideal type mechanisms in these reactions. More appropriate seems the assumption that a hot-atom type mechanism operates in abstraction of adsorbed D by gaseous H. The H/D abstraction kinetics at Pt(110) surfaces are completely different from that observed previously at Pt(111) surfaces. This structural effect puts a further question mark on the validity of the Eley-Rideal mechanisms in abstraction reactions and is interpreted in the present work as a consequence of the interrelation between a hot-atom mechanism and the surface structure of the reconstructed substrate.

Full text of DOI:10.1016/S0009-2614(98)00075-X

3 April 1998
Ž
.
Chemical Physics Letters 286 1998 15–20
A structural effect in direct reactions: kinetics of D abstraction
ž
/
from Pt 110 1=2 surfaces with gaseous H atoms
a
a
a,b
Th. Biederer , Th. Kammler , J. Kuppers
¨
a
Experimentalphysik III, UniÕersitat Bayreuth, 95440 Bayreuth, Germany
¨
b
Max-Planck-Institut fur Plasmaphysik, 85748 Garching, Germany
¨
Received 8 December 1997; in final form 9 January 1998
Abstract
Ž
.
Deuterium-covered Pt 110 surfaces were subjected to a flux of thermal H atoms at 100 K. The rates of product
formation, HD and D2, were measured simultaneously with H atom exposure as a function of D coverage. D2 as a product
and the kinetics of HD formation contradict the operation of Eley–Rideal type mechanisms in these reactions. More
appropriate seems the assumption that a hot-atom type mechanism operates in abstraction of adsorbed D by gaseous H. The
Ž
.
Ž
.
HrD abstraction kinetics at Pt 110 surfaces are completely different from that observed previously at Pt 111 surfaces. This
structural effect puts a further question mark on the validity of the Eley–Rideal mechanisms in abstraction reactions and is
interpreted in the present work as a consequence of the interrelation between a hot-atom mechanism and the surface structure
of the reconstructed substrate. q 1998 Elsevier Science B.V.
1. Introduction
The kinetics of abstraction of D by H expressed
Ž
.
as an Eley–Rideal ER process via
Reactions of adsorbates with gaseous H atoms
have attracted a growing interest recently. In particu-
lar, the simplest feasible reaction, abstraction of D
adsorbed on metal surfaces by H towards a gaseous
HD product molecule was studied for various sys-
d HD rdts D sF
1
Ž
.
Ž .
reveals as the rate of formation of the product
d HD rdts D sF exp ysF t ,
Ž .
2
Ž .
Ž
.
Ž .0
Ž
.
where s is the reaction cross-section, F is the H
Ž
. w .
x
Ž
. w x
Ž
. w x
Ž
Ž .
tems, Cu 111 1,2 , Ni 110 3 , Ni 100 4 , Al 100
atom flux and D is the D coverage at the start of
0
w x
Ž
. w x
Ž .
5 and Pt 111 6 . As the incoming atom is not
accommodated at the surface prior to reaction, the
HD product molecule is expected to carry significant
energy. This was confirmed in laser spectroscopy–
molecular beam studies by Rettner and coworkers
the reaction. To obtain the solution 2 it is necessary
to assume that the atom flux is directed at the surface
as a step function located at ts0. It is clear from
the rate expression
Ž .
2
that the rate assumes its
maximum at ts0 and decreases exponentially
w
x
Ž
.
1,2 on abstraction from Cu 111 surfaces. They
thereafter.
observed that the reaction energy of ;2.5 eV is
distributed in translational, rotational and vibrational
degrees of freedom of the product.
Kinetic studies performed in this laboratory on the
abstraction of D from Ni 100 4 and Pt 111 6
surfaces revealed that the time fluence dependen-
Ž
. w x . w x
Ž
Ž
.
0009-2614r98r$19.00 q 1998 Elsevier Science B.V. All rights reserved.
Ž
.
PII S0009-2614 98 00075-X

(
)
16
Th. Biederer et al.rChemical Physics Letters 286 1998 15–20
Ž .
Ž .
Ž
.
Fig. 1. HD a and D2 b reaction rates measured upon subjecting D-covered Pt 110 surfaces to a flux of H atoms. H atom flux 0.21 Ml
sy1, substrate temperature 100 K. The D coverages given in the figure are those at reaction start times0 .
Ž
.

(
)
Th. Biederer et al.rChemical Physics Letters 286 1998 15–20
17
cies of experimental HD rates differ substantially
chamber. This aperture can be closed by a shutter
and at the opening of the shutter the sample surface
is exposed to a flux of H atoms from the atom
source. The atom source consists of a heated W tube
Ž
from the predicted ones. At the reaction start H flux
.
on the rates jump to specific coverage-dependent
values R0, but increase later on. After reaching
w
x
maximum values at tstm, the rates drop exponen-
according to the design by Bischler and Bertel 15 ;
however, the electron beam heating of the tube was
replaced by ohmic heating. The atom flux from the
source was determined from the tube front end tem-
tially controlled by cross-sections, which are of the
2
˚
order of A . The magnitude of R0 does not increase
linearly with the D coverage. The lengths of the
induction periods tm, decrease with increasing D
coverage. If H is coadsorbed at saturation, tm is
independent of coverage and the initial rates R0
scale linearly with the partial D coverage.
Ž
.
perature 2000 K and the hydrogen gas throughput
through the tube. The flux in monolayerrs is ex-
pressed with respect to the areal atom density on the
unreconstructed Pt 110 surface, 1 Ml s s0.92=
y1
Ž
.
1015 cmy2 sy1
.
In addition to HD, homonuclear reaction product
molecules D2 were observed in both studies, as were
After proper cleaning, hydrogen desorption spec-
Ž
.
reported earlier in abstraction experiments at Ni 110
tra were found in excellent agreement with the spec-
w x
w
x
surfaces 3 .
tra published by Anger et al. 16 . These authors
reported 1.7 Ml as the H saturation coverage, which
was confirmed from the present spectra within an
error margin of 10%.
According to these experimental findings, ER
mechanisms do not provide a correct description of
the abstraction kinetics for these reactions. As an
alternative description of direct reactions, the hot-
w x
atom 7 reaction pathway has been suggested. Reac-
tion trajectories calculated in recent molecular dy-
3. Results and discussion
namics simulations of abstraction of H by H from
Ž
. w
x
Ž
. w
x
Si 110 8,9 and Cu 111 10–14 surfaces confirm
the existence of hot-atom reactive trajectories and
homonuclear reaction products. These theoretical
studies also reveal that the products carry the reac-
HD and D2 were recorded as the reaction prod-
ucts. The fluence dependence of the rates of these
products are shown in Fig. 1a and Fig. 1b as a
function of D coverage prior to reaction at a constant
H flux of 0.21 Ml sy1 and 100 K sample tempera-
ture. After reaction completion, thermal desorption
spectra confirmed that no D was left at the surface,
indicating complete abstraction. As all preadsorbed
D have to appear in HD or D2 products, the D
saturation coverage could be used to convert the HD
and D2 signals measured by the quadrupole into
units of Ml sy1, as was done in Fig. 1.
w
x
tion energy in internal excitations 8–14 .
The present investigation was performed in order
to explore whether a structural effect exists in ab-
straction. As within the Eley–Rideal picture the sur-
face geometry is irrelevant for the rate of a product,
a structural effect in the rates would raise further
questions on the validity of this mechanism for
atomradsorbate reactions.
Up to a D coverage of ;1 Ml the HD rate curves
in Fig. 1a exhibit a common behavior. At the open-
ing of the shutter they jump to a finite value, de-
crease and increase again. After having reached an
absolute or relative maximum at the end of an
induction period, the rates decay exponentially, con-
firmed by a logarithmic plot of the HD rate curves.
At D coverages of 1.2 and 1.55 Ml the rates increase
immediately after the rate jump and at a coverage of
1.7 Ml the rate decreases right after the jump. Evalu-
ation of the late rate fluence dependence of all rate
2. Experimental
The experiments were carried out in the UHV
system used in the previous study on Ni 100 sur-
Ž
.
w x
faces 4 . In this system, the H atom source and the
mass spectrometer for monitoring the reaction prod-
uct are contained in a separately pumped UHV sys-
Ž
.
tem source chamber attached to a main chamber
which houses the sample and surface analytical tools.
During the reaction, the sample is placed in front of
an aperture which connects the main and source
Ž
.
Ž .
curves shown beyond the induction period in terms
of an exponential law as expressed in Eq. 2 reveals

(
)
18
Th. Biederer et al.rChemical Physics Letters 286 1998 15–20
2
˚
;1 A as the reaction cross-section, irrespective of
D coverage.
The rate curves at small D coverages in Fig. 1a
exhibit in the induction period, corresponding to the
first few Ml of atom fluence, a fine structure which
was reproduced several times. These structures are
not artifacts. As seen in Fig. 1a, the induction period
is longer at smaller D coverages.
Ž
.
The D2 rates measured almost simultaneously
with the HD rates shown in Fig. 1b exhibit similar
behavior to the HD rates: a rate jump at the start of
the reaction, a rate increase in the induction period,
followed by a rate decay. The D2 rate induction
period has the same length as the respective HD rate
induction period. The late rate decay is twice as fast
as the HD rate decay, which can be verified by a
Ž
logarithmic plot no perfect line as opposed to the
.
HD rate plot . A ‘reaction cross-section’ towards D2
of twice the HD formation cross-section was ob-
tained. The quotation marks are used in order to
emphasize that the cross-section towards D2 is only
indicative for the efficiency of the process and not a
real cross-section as D2 formation is a second-order
process.
The bottom curve in Fig. 1a was measured with-
out preadsorbed D in order to confirm that back-
ground effects were negligible. Likewise, control
experiments with a cold tube revealed no products at
all, indicating that H adsorption via D displacement
can be neglected at 100 K. At higher temperatures
T)150 K, this is not the case, as expected.
The yields towards HD and D2 and corresponding
rate steps deduced from Fig. 1 are shown in Fig. 2.
The D2 yields can be fitted excellently with a square
dependence on the D coverage, in accordance with a
second-order process. This phenomenon was also
observed for the D2 rates in the previous studies on
Fig. 2. HD and D2 yields and HD and D2 rate steps at reaction
start deduced from Fig. 1.
can be drawn from the HD yields measured earlier at
Ž
.
Ž
.
Ni 100 and Pt 111 surfaces.
The rate jumps shown in Fig. 2 do not exhibit a
simple functional dependence. Neither is the HD rate
jump linear in the D coverage nor does the D2 rate
jump depend on the square of the D coverage.
Leaving aside the formation of D2 , the kinetics of
HD formation does not exhibit the expectations drawn
from the operation of an ER reaction mechanism, as
was already concluded from the kinetics measured
Ž
.
previously in the abstraction reactions on Ni 100
Ž
.
and Pt 111 surfaces. Only in the late reaction
Ž .
regimes are exponential decays according to Eq. 2
Ž
.
Ž
.
Ž
.
observed on Ni 100 , Pt 111 and the present Pt 110
Ž
. w x .
Ž
the abstraction kinetics on Ni 100 4 and Pt 111
w x
surface. The results reported for abstraction from
Ž
. w x
Ž
. w x
6 . The HD yields seem to be approximated well by
a linear dependence, of the D coverage, which is,
however, misleading. The HD yields have to exceed
the magnitudes deduced from a linear coverage de-
pendence as the yield of the competing D2 product
scales with the square of the D coverage. As at most
11% of the adsorbed D appear in the D2 reaction
channel, the deviation of the HD yield from a linear
relationship is not very significant but the data shown
in Fig. 2 are in line with this. A similar conclusion
Ni 110 3 and Al 111 5 surfaces do not provide
detailed information on the initial reaction period
and data on coverage dependence, both of which are
critical for an analysis of the reaction mechanism.
The kinetics of HD and D2 formation in the
induction period measured in the present study at
Ž
.
Pt 110 surfaces is completely different from that
Ž
.
Ž
.
measured earlier at Pt 111 surfaces. At Pt 111 sur-
faces the HD and D2 rates jump at the opening of
the shutter to significantly smaller values than re-

(
)
Th. Biederer et al.rChemical Physics Letters 286 1998 15–20
19
ported here and they stay small throughout the induc-
tion period. Only late in the induction periods do the
rates increase and assume their maximum values.
Accordingly, the abstraction kinetics is sensitive
to the surface structure which contradicts the ER
reaction scheme.
In the previous work on D abstraction by H from
Ž
. w x
Ž
.
Ni 100 4 and more explicitly in the Pt 111 paper
w x
6 , it was argued that a hot-atom mechanism is
much more appropriate to explain the measured
product kinetics, in particular the occurrence of a D2
product and the observation that the HD product
rates at the opening of the shutter are smaller than
they could be at the given D coverage, i.e. the rates
do not assume their maxima right at the beginning of
the reaction.
A feasible hot-atom mechanism based scenario
starts with the assumption that H atoms impinging
from the gas phase stick at empty surface sites,
produce hot H atoms upon encountering H-occupied
sites and hot D or H atoms upon encountering
D-occupied sites, each event with a given probabil-
ity, pstick, pHHH, pHDH and pHDD. Hot H and D
atoms generated in these collisions travel on the
surface and pass along empty, H- and D-occupied
sites. With probabilities photstick and photreact they
stick at empty sites and react at occupied sites. If the
probabilities pstick and photstick are assumed to be
bigger than the reaction probability photreact, a com-
peting step to reaction is introduced which, by con-
struction, causes a coverage-dependent induction pe-
riod by reducing the product rates as long as the
surface exhibits empty sites. Only if the coverage has
approached saturation by the sticking of incoming
and hot atoms, do the product rates assume their
maximum values and decrease exponentially at later
times. Note that within this scenario the homonuclear
D2 reaction products are generated by the events
controlled by the probability pHDD. The second-
order kinetics for D2 formation follows automati-
cally, as well as the induction period in the D2
reaction channel.
Fig. 3. Interrupted flux reaction HD rate measurements measured
at D-covered Pt 110 surfaces at a H flux of 0.2 Ml sy1. Top
curve: substrate temperature 90K, bottom curves: reaction temper-
ature 175K. The shutter openrclosed periods are assigned to each
rate curve.
Ž
.
saturation value. The top rate curve in Fig. 3, mea-
sured at 90 K at a D coverage of 1.2, shows the
outcome of such an experiment. In accordance with
expectations from the above scheme, only after the
first opening of the shutter is a rate increaserdecay
sequence observed. At later opening events the rate
jumps to the value which was already reached at the
time of the previous shutter closing event.
The experiment was repeated at 175 K with vari-
ous shutter openrclosed periods. At 175 K isother-
mal desorption from a hydrogen-saturated surface
occurs. Therefore, after every ‘shutter closed’ period
a new induction period with a rate increase following
the rate jump should be observed. As seen in the
bottom curves of Fig. 3 this is indeed the case.
Moreover, with increasing shutter closed periods,
5–35 s, through isothermal desorption the hydrogen
The consequences of this scenario can be tested
experimentally. If the beam of H atoms is interrupted
in the late reaction period and switched on again at a
later time, the rate jump should immediately be
followed by a rate decay, rather than exhibit a new
induction period as the coverage remained at its
Ž
.
D plus H coverage decrease gets more pronounced
and, accordingly, the length of the induction period
increases. The rate curves in Fig. 3 suggest that after

(
)
20
Th. Biederer et al.rChemical Physics Letters 286 1998 15–20
35 s isothermal desorption has established an equilib-
rium coverage, as no changes in the induction peri-
ods and rates are observed at the even longer shutter
closed period of 95 s.
From the above model construction it is immedi-
ately clear that reactions at surfaces, on which a
fractional D coverage and H coadsorption at satura-
tion were established, exhibit induction periods of
length which do not depend of the D coverage, as
was observed experimentally in the present and pre-
vious studies from this laboratory.
reducing the induction period effect. This phe-
nomenon could be even amplified by the structure of
the adsorbed D layer to be abstracted. If D gets
predominantly adsorbed in chains along the chan-
nels, the D reactants would be arranged and the H
reactants move in a restricted geometry. As these
Ž
.
restrictions are absent on Pt 111 surfaces the ab-
straction kinetics on that surface is expected to differ
Ž
.
from the present Pt 110 surface.
The general features of the rate curves in Fig. 1
4. Summary
Ž
.
for abstraction from Pt 110 surfaces, as well as the
Ž
.
rates measured in the previous studies with Ni 100
In summary, we have shown that the kinetics of
abstraction reactions exhibit surface structure sensi-
tivity. This observation contradicts the operation of
Eley–Rideal mechanisms in these reactions. Hot-
atom type reaction scenarios seem to provide a much
more appropriate description.
w x
Ž
. w x
4 and Pt 111 6 surfaces, are explained within the
hot-atom scenario and can be rationalized by model
calculations of the kinetics based on a random walk
w
x
scheme 17 with the parameter settings psticks1,
photsticks1 and preacts0.01–0.1. However, as
the HD rates in Fig. 1a illustrate, in the initial
Ž
.
reaction period at Pt 110 , there is an almost ER-type
Ž
.
behavior, which is absent at Pt 111 surfaces: a rate
References
jump followed by a fast rate decrease. It seems as if
Ž
.
on Pt 110 surfaces a small ER contribution without
an induction period effect dominates initially at small
coverages, at later times and higher coverages over-
w x
Ž .
C.T. Rettner, Phys. Rev. Lett. 69 1992 383.
C.T. Rettner, D.J. Auerbach, J. Chem. Phys. 104 1996
2732.
1
w x
2
Ž
.
Ž
.
ruled by the hot-atom scenario. At Pt 111 this initial
jumprdecrease phase is absent, only a pronounced
hot-atom induction period exists, more pronounced
than at Ni 100 surfaces.
The structural difference between the Pt 111 and
Pt 110 surfaces is the corrugation of these surfaces.
The Pt 111 surface is flat and the Pt 110 surface is
highly corrugated due to its 1=2 reconstruction
with tilted 111 oriented plane segments. This corru-
gation introduces a directional anisotropy which can
cause the movement of hot atoms across the surface
to exhibit directional restrictions. It might be that
they move predominantly along the channels pro-
vided by the reconstruction. Due to this effect, hot
atoms on Pt 110 surfaces might have a limited
chance to stick at empty sites, as only the sites in
those channels are available for sticking, thereby
w x
3
G. Eilmsteiner, W. Walkner, A. Winkler, Surf. Sci. 352r354
Ž
.
1996 263.
w x
4
Ž
.
Th. Kammler, J. Lee, J. Kuppers, J. Chem. Phys. 106 1997
¨
7362.
Ž
.
w x
5
J. Boh, G. Eilmsteiner, K.D. Rendulic, A. Winkler, Surf. Sci.
Ž
.
Ž
.
395 1998 98.
Ž
.
Ž
w x
Ž .
S. Wehner, J. Kuppers, J. Chem. Phys. in press .
¨
6
.
Ž
.
w x Ž .
7 J. Harris, B. Kasemo, Surf. Sci. 105 1981 L281.
w x
Ž .
P. Kratzer, J. Chem. Phys. 106 1997 6752.
8
w x
Ž .
9
P. Kratzer, W. Brenig, Surf. Sci. 254 1991 275.
Ž
.
w
w
w
w
w
x
Ž
.
10 B. Jackson, M. Persson, J. Chem. Phys. 96 1992 2378.
x
Ž
.
11 M. Persson, B. Jackson, J. Chem. Phys. 102 1995 1078.
x
Ž
.
12 M. Persson, B. Jackson, Chem. Phys. Lett. 237 1995 468.
x
Ž
.
13 B. Jackson, M. Persson, J. Chem. Phys. 103 1995 6257.
x
14 S. Caratzoulas, B. Jackson, M. Persson, J. Chem. Phys. 107
Ž
.
1997 6420.
w
w
x
Ž
.
15 U. Bischler, E. Bertel, J. Vac. Sci. Technol. A5 1993 485.
Ž
.
x
16 G. Anger, H.F. Berger, M. Luger, S. Feistritzer, A. Winkler,
Ž
¨
.
K.D. Rendulic, Surf. Sci. 219 1989 L583.
w
x
Ž
.
17 Th. Kammler, S. Wehner, J. Kuppers to be published .