Dynamics of the formation of CD4 from the direct reaction of incident D atoms with CD3/Cu(111)

Article abstract of DOI:10.1063/1.472840

Using molecular beam techniques we find that incident D atoms can abstract CD₃ from a Cu(111) surface to yield CD₄ in a direct (Eley-Rideal) gas-surface reaction with a cross section of ～10^-16 cm²/D atom. Dynamical evidence for a direct reaction includes the observation of an extremely sharp angular distribution that is clearly displaced from the surface normal, and the determination of a very high translational energy of the product, E_f, which is ～2 eV. For a 0.25 eV D-atom beam incident at 45° on a 95 K surface, this energy varies with the detection angle, θ_f, as E_f(θ_f) = (1.8+θ_f/45) eV, where θ_f0° in the "backscattering'' direction. For these conditions, the angular distribution approximately follows the function cos⁷⁰(θ_f-5.5), being peaked 5.5° from the normal with a full width at half maximum of <17°. Lowering the beam energy to 0.07 eV gives a broader angular distribution peaked at about 1.5° from the normal, consistent with cos⁶⁰(θ_f-1.5). The reaction with 0.25 eV H incident at 45° gives a similar distribution peaked at ～3.5° from the normal. The shifts in the angular distributions are approximately consistent with parallel momentum conservation. The CD³/Cu(111) surface was prepared by thermal dissociation of CD₃I on the surface or by adsorbing CD₃ directly from a CD₃ beam produced by the pyrolvsis of azomethane.

Full text of DOI:10.1063/1.472840

Dynamics of the formation of CD4 from the direct reaction of incident D atoms with
CD3/Cu(111)
C. T. Rettner, D. J. Auerbach, and J. Lee
Citation: The Journal of Chemical Physics 105, 10115 (1996); doi: 10.1063/1.472840
View online: http://dx.doi.org/10.1063/1.472840
View Table of Contents: http://scitation.aip.org/content/aip/journal/jcp/105/22?ver=pdfcov
Published by the AIP Publishing
Articles you may be interested in
Alignment of D2(v,J) desorbed from Cu(111): Low sensitivity of activated dissociative chemisorption to approach
geometry
J. Chem. Phys. 105, 9702 (1996); 10.1063/1.472979
Dynamics of the displacement of CO from Cu(111) by H atoms incident from the gas phase
J. Chem. Phys. 105, 8842 (1996); 10.1063/1.472613
Dissociative chemisorption of H2 on Cu(100): A fourdimensional study of the effect of parallel translational
motion on the reaction dynamics
J. Chem. Phys. 105, 5979 (1996); 10.1063/1.472450
Si/XeF2 etching: Reaction layer dynamics and surface roughening
J. Vac. Sci. Technol. A 14, 2780 (1996); 10.1116/1.580200
Modification of Ni(111) reactivity toward CH4, CO, and D2 by twodimensional alloying
J. Chem. Phys. 104, 7289 (1996); 10.1063/1.471395
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
130.18.123.11 On: Tue, 23 Dec 2014 18:42:08

Dynamics of the formation of CD4 from the direct reaction of incident D
atoms with CD3/Cu(111)
C. T. Rettner, D. J. Auerbach, and J. Leea)
IBM Almaden Research Center, 650 Harry Road, San Jose, California 95120-6099
͑Received 9 August 1996; accepted 9 September 1996͒
Using molecular beam techniques we find that incident D atoms can abstract CD3 from a Cu͑111͒
surface to yield CD4 in a direct ͑Eley–Rideal͒ gas–surface reaction with a cross section of ϳ10Ϫ16
cm2/D atom. Dynamical evidence for a direct reaction includes the observation of an extremely
sharp angular distribution that is clearly displaced from the surface normal, and the determination
of a very high translational energy of the product, Ef , which is ϳ2 eV. For a 0.25 eV D-atom beam
incident at 45° on a 95 K surface, this energy varies with the detection angle, ␪f , as
Ef(␪f)ϭ(1.8ϩ␪f/45) eV, where ␪fϽ0° in the ‘‘backscattering’’ direction. For these conditions, the
angular distribution approximately follows the function cos70͑␪fϪ5.5͒, being peaked 5.5° from the
normal with a full width at half maximum of Ͻ17°. Lowering the beam energy to 0.07 eV gives a
broader angular distribution peaked at about 1.5° from the normal, consistent with cos60͑␪fϪ1.5͒.
The reaction with 0.25 eV H incident at 45° gives a similar distribution peaked at ϳ3.5° from the
normal. The shifts in the angular distributions are approximately consistent with parallel momentum
conservation. The CD3/Cu͑111͒ surface was prepared by thermal dissociation of CD3I on the surface
or by adsorbing CD3 directly from a CD3 beam produced by the pyrolysis of azomethane. © 1996
American Institute of Physics. ͓S0021-9606͑96͒03746-4͔
I. INTRODUCTION
In order to assess the overall importance of such direct
gas–surface reactions and to understand the generality of
their role in gas–surface chemistry, information is needed for
more systems. In particular, since gas–surface reactions usu-
ally involve relatively large adsorbates, it is important to
determine whether direct reactions can occur for more com-
plex systems involving polyatomic adsorbates. Polyatomic
reagents clearly have more modes to store the energy release,
which could prevent the product from leaving the surface.
While the bonding to an atomic adsorbate may be relatively
straightforward, the additional atoms in a polyatomic may
hinder access to the reaction site or block the necessary ap-
proach direction. In an earlier study19 of the reaction of O
atoms which CO, for example, the CO2 product emerged
symmetrically about the surface normal with a velocity close
to that observed ͑under different conditions͒ for the LH oxi-
dation of CO. More recently, it has been observed20 that
exposing O2/Pt͑111͒ to O, H, and N atoms leads to desorp-
tion of the O2, without formation of a triatomic gas–phase
product. Jachilowsku and Weinberg,9 have observed a tri-
atomic product, in the reaction of H atoms with CO/Ru͑100͒,
but the HCO product remains on the surface. Similarly, Bent
and co-workers6 have reported the hydrogenation of a num-
ber of species where the products remain on the surface.
Here we report the results of a model study of the reac-
tion of D atoms with CD3 adsorbed on Cu͑111͒. We have
found that not only does the incident atom abstract the me-
thyl radical efficiently, but that the methane molecule leaves
the surface in a narrow cone of angles with almost all of the
available energy as translational motion. This observation
has important consequences for this class of gas–surface
chemistry. For example, the result indicates that attempts to
Recent experiments1–11 and calculations12–17 have
shown that incident H atoms can react directly with atoms
adsorbed on surfaces. These studies were primarily con-
cerned with establishing that such direct gas–surface reac-
tions can indeed occur. The strongest evidence for this, so-
called Eley–Rideal ͑ER͒ mechanism, has come from reactive
scattering measurements that focused on the channel leading
to a gas–phase product.5,7,8,10,11 Earlier work by Kuipers
et al.,18 had provided the first proof that ER reactions occur
at all, by studying the velocity of product species following
proton abstraction by hyperthermal molecules. In the H-atom
cases,5,7,8,10,11 the measurements demonstrated, for example,
that the angular distribution of the product is asymmetric
about the surface normal, implying that the reaction must
occur before the parallel momentum of the incident species
has been accommodated.
So far, however, such dynamical results have only been
reported for two systems, involving the abstraction of
5,10,11
H͑D͒
and Cl7,8 from Cu͑111͒ and Au͑111͒ surfaces,
respectively. Even for the maximum surface coverages, there
is evidence that a large fraction of the incident atoms do not
react directly. For the H(g)ϩCl/Au͑111͒→HCl(g) reaction,
it is clear that reaction also occurs between fully accommo-
dated species, by a so-called Langmuir–Hinschelwood ͑LH͒
mechanism. For the H(g)ϩD/Cu͑111͒→HD(g) reaction,
most incident atoms actually trap, rather than react ͑and
would react by an LH process at higher surface temperature,
Ts͒. These systems involve the abstraction of a single atomic
adsorbate.
a͒
Permanent address: Department of Chemical Technology, Seoul National
University, Sinrim-dong San 56-1, Kwanak-gu, Seoul 151-742, Korea.
J. Chem. Phys. 105 (22), 8 December 1996
0021-9606/96/105(22)/10115/8/$10.00
© 1996 American Institute of Physics
10115
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
130.18.123.11 On: Tue, 23 Dec 2014 18:42:08

10116
Rettner, Auerbach, and Lee: Formation of CD4
produce radical species on the surface by hydrogenation may
be limited by the rate at which such radicals are abstracted.
rowed to Ͻ1 mm about 5 mm upstream from the heated
region, to reduce reactions of the radicals upstream, as sug-
gested by Peng et al.24
We synthesized azomethane from 1,2-dimethylhydrazine
dihydrochloride ͑Aldrich͒, which was treated with NaOH to
yield 1,2-dimethylhydrazine,25 and then oxidized with yel-
low HgO following the procedure of Renaud and Leitch.26
About 5 g of azomethane were transferred to a ϳ100 ml Cu
container. We pumped off residual gases while emersing the
base of this container in liquid nitrogen. The vapor pressure
of the liquid was close to two atmospheres at room tempera-
ture. Although azomethane can explode at elevated
temperatures,27,28 it is stable at room temperature provided
air is excluded. The azomethane container was cooled in an
ice–water bath to reduce reflux of the liquid into the gas line
connecting it to the source. A remotely operated shutoff
valve was used to seal the container except during dosing.
Flow to the source was regulated with a variable leak valve
͑Granville Phillips, 203͒
II. EXPERIMENT
The molecular beam apparatus employed in this study
has been described in detail previously.8,21,22 We direct
atomic beams at a Cu͑111͒ single crystal contained in an
ultrahigh vacuum scattering chamber mounted on a manipu-
lator that provides accurate control of incidence angle, ␪i ,
and surface temperature, Ts . The crystal surface is within
Ϯ0.2° of the ͑111͒ plane and contamination levels are deter-
mined to be below the Ӎ1% limit of Auger spectroscopy.
¯
The scattering plane was within 5° of the ͓101͔ azimuth. The
surface was cleaned by Arϩ sputtering ͑1 ␮A/cm2, 500 V͒,
followed by annealing to 1100 K. Iodine was removed by
flashing to 950 K. We also noted that the atomic H ͑D͒
beams removed O, S, I, and C from the hot ͑800 K͒ surface.
A. D-atom beam sources
C. CD3I dosing
The measurements reported here are made using one of
two different atomic beam sources. Relatively low energy
͑ϳ0.06 eV͒ D-atom beams are generated using a microwave
discharge source consisting of a quartz tube with a ϳ2 mm
diameter hole.21,22 This source yields a beam that is about
80% dissociated and has an atomic flux of about 1 ML/s at
the crystal position, which is Ӎ30 cm from the orifice. To
obtain this high degree of dissociation, part of the hydrogen
gas is bubbled through water, introducing ϳ0.1% water to
the beam. No additional O is detected by Auger after expo-
sure to this beam. Higher energy beams ͑ϳ0.25 eV͒ are
made by thermal dissociation in a tungsten tube beam
source23 operated at about 2400 K. Using a 2 mm diameter
hole and a source pressure of ϳ15 Torr gives an atomic flux
of about 1 ML/s, together with a molecular flux of ϳ5 ML/s
͑about 10% dissociation͒. D-atom beams from either source
are chopped by a high-speed chopper, which reduces the flux
by a factor of 100 and yields 10 ␮s pulses for TOF measure-
ments with a repetition rate of ϳ1000 pulses/s.
The azomethane pyrolysis source was not used to dose
CD3 as we could not obtain the required quantity of the
deuterated compound. We therefore dosed the surface with
CD3I, which dissociates to give CD3 and I.29 CD3I ͑Aldrich,
99.5% D͒ was purified only by freeze–pump–thaw cycles.
The surface was held at 170 K during the dose, which is high
enough to ensure complete dissociation, yet sufficiently low
to suppress radical–radical reactions. The dose gas emerged
from a 6 mm diameter hole about 1 cm from the source.
Exposures were calibrated relative background exposures
͑with the crystal far from the aperture͒, using temperature
programmed desorption ͑TPD͒. With the crystal in position,
the exposure was about three times higher than from back-
ground alone. Gas was admitted via a variable leak valve
͑Varian, 951-5100͒. In order to obtain better reproducibility,
this valve was left open at a fixed setting and the gas flow
turned on by admitting the CD3I to a small volume behind
the valve and turned off by evacuating the gas.
D. Angular and velocity distributions
B. CH3-radical beam source
Angular and velocity distributions were obtained using
time-of-flight ͑TOF͒ techniques, similar to those described
previously.8,30 Pulses of methane product were detected us-
ing a differentially-pumped quadrupole mass spectrometer
that can be rotated about an axis passing through and parallel
to the crystal face. The background level at m/eϭ20 for CD4
was ϳ200 counts/s, which is more than an order of magni-
tude smaller than for m/eϭ19–15. Hence the best signal-to-
noise was obtained with incident D and CD3I dosing. Angu-
lar distributions were obtained by integrating the TOF
signals for each angle.
In order to obtain accurate information on the velocity of
the product, we first determine the time zero of the chopper
opening and the form of the chopper opening function. These
are measured optically by detecting light from the tungsten
source on a photomultiplier attached to a telescope looking
In order to dose the Cu͑111͒ surface with CH3, we con-
structed a quartz pyrolysis source similar to that described by
Peng et al.24 and Chiang and Bent.25 We obtain a beam of
CH3 radicals from this source by the pyrolysis of azomethane
at 1100 K. The primary contaminants of the CH3 beam are
N2 from the pyrolysis and C2H6 from recombination reac-
tions in the source. Neither molecule adsorbs on the Cu͑111͒
surface at 170 K. The quartz pyrolysis tube ͑6 mm o.d., 3
mm id͒ was heated along the final 25 mm with 0.5 mm
diameter Ta wire wound directly on the tube. A silica-based
high-temperature adhesive ͑Sauereisen, No. 1 Paste͒ was
used to ensure good thermal contact. Two type-K thermo-
couples were spot-welded to small pieces of Ta sheet and
secured in the adhesive. The beam orifice was narrowed to 2
mm to match the beam collimation. The tube was also nar-
J. Chem. Phys., Vol. 105, No. 22, 8 December 1996
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
130.18.123.11 On: Tue, 23 Dec 2014 18:42:08

Rettner, Auerbach, and Lee: Formation of CD4
10117
down the beam axis. Zero time was taken as the peak of the
chopper opening. We additionally allow for the spread in
arrival times of the incident D͑H͒ atoms, which is deter-
mined by separate TOF measurements on the H͑D͒ beams.
Finally we subtract the TOF of the ions between the
electron-impact ionizer and the electron multiplier detector.
This time is calculated as the sum of the flight times through
the nine separate regions of the ion flight path. The energy in
each region is taken as the sum of the potential in that region
plus the approximate energy of the neutral. This method was
checked against an Ar beam, measuring TOF’s for m/eϭ40,
20, and 40/3. The calculated ion TOF was approximately
1/2
given by: 3.45 m[35/(35ϩE)]
␮s, where 35 is the ap-
͕
͖
proximate mean ion potential and E is the energy of the
neutral, which is determined iteratively. The TOF analysis
͑which involves a nonlinear least-squares fit to a model that
allows for convolution over the arrival times of the incident
beam and chopper function͒ was similar to that described
previously for energy transfer studies.30 In these studies,
however, the velocity after the surface encounter is assumed
to be independent to the velocity of the incident species.
FIG. 1. Time-of-flight distribution of CD4 formed by the reaction of D
atoms with CD3/Cu͑111͒. This distribution was obtained for Tsϭ95 K de-
tecting in the direction of the surface normal. The incident D atoms have a
mean kinetic energy of 0.07 eV and incidence angle of 45°. The solid line
shows a fit to the data as discussed in the text. The overall TOF is the time
it takes the beam to travel the 10.7 cm from the chopper to the surface plus
the time it takes the product to travel the 10.0 cm to the detector. The broken
line shows the equivalent TOF for the beam, where the beam travels the full
20.7 cm from the chopper to the detector. It is apparent that the CD4 product
is much faster than the incident D.
III. RESULTS
Exposing the CD3͑CH3͒/Cu͑111͒ surface to an incident
D͑H͒-atom beam leads to prompt formation of methane. For
the unchopped beam, with a flux ϳ1 ML/s, the signal decays
to zero in a few seconds as the methyl species is depleted.
From an analysis of TOF distributions, we can set an upper
limit on the surface residence time of Ͻ5 ␮s.
the TOF for the D atoms, when the detector looks directly at
the beam. It is clear that the CD4 arrives earlier, implying
that the CD4 product is much faster than the incident D at-
oms. In fact, the CD4 product is so much faster than the
A. D؉CD3/Cu(111)˜CD4
In order to exploit the favorably low background count
rate at m/eϭ20, most studies were performed with D atoms
incident on CD3/Cu͑111͒, to give the CD4 product. Working
with incident D also has the advantage of leading to poten-
tially larger displacements of the product angular distribution
from the normal than for incident H, since it has a larger
momentum for a given energy. A potential disadvantage of
this combination is that we needed to use CD3I as a source of
adsorbed CD3, as opposed to dosing directly with a CD3
source. We shall see, however, that our results for dosing
with CH3I and CH3 radicals are very similar. This finding is
consistent with the work of Chiang and Bent25 who have
studied the effect of co-adsorbed I on the electron energy
loss and TPD spectra of the CH3 adsorbed on Cu͑111͒. They
concluded that the predominant effect of co-adsorbed I is to
block surface sites, thereby reducing the maximum methyl
coverage.
We have obtained TOF and angular distributions for the
CD4 product for D beams from both the microwave dis-
charge source and the thermal dissociation source. The pri-
mary difference between the two beams is that the transla-
tional energy for the thermal dissociation source is ϳ0.25
eV, compared to ϳ0.07 eV from the microwave source. Fig-
ure 1 shows a TOF distribution for the CD4 product in the
case of D atoms from the microwave source. This result was
obtained for ␪iϭ45° and Tsϭ95 K. The dotted line indicates
FIG. 2. Polar plot of the angular distributions of the CD4 product of the
reaction of D atoms with CD3/Cu͑111͒ for Eiϭ0.07 eV and Tsϭ95 K.
͑Signal intensities are proportional to the distance from the origin͒. The
angle of the incidence beam ͑45°͒ is indicated graphically by the arrow. The
line corresponds to cos60͑␪fϪ1.5͒. The tics are placed at 10° intervals.
J. Chem. Phys., Vol. 105, No. 22, 8 December 1996
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
130.18.123.11 On: Tue, 23 Dec 2014 18:42:08

10118
Rettner, Auerbach, and Lee: Formation of CD4
FIG. 4. Energy distribution for the CD4 product of the reaction of D atoms
with CD3/Cu͑111͒ obtained from an analysis of the TOF distribution in Fig.
3. The broken line shows the energy distribution of the incident D-atom
beam.
FIG. 3. Time-of-flight distribution of CD4 formed by the reaction of D
atoms with CD3/Cu͑111͒. This distribution was obtained for Tsϭ95 K de-
tecting in the direction of the surface normal. The incident D atoms have a
mean kinetic energy of 0.25 eV and incidence angle of 45°. The solid line
shows a fit to the data as discussed in the text. The resulting energy distri-
bution is displayed in Fig. 4. The broken line shows the equivalent TOF for
the beam, where the beam travels the full distance from the chopper to the
detector. It is apparent that the velocity of the CD4 product is close to that of
the incident D.
CD4 is ten times higher, we can immediately deduce that the
CD4 energy must be a little less than ten times the 0.25 eV of
the incident D atoms. The full analysis yields a translational
energy of 1.8Ϯ0.3 eV. This analysis consists of fitting the
TOF to a velocity distribution of the form
f u͒ϰu3exp Ϫ uϪu ͒2/␣2
,
͔
͑1͒
͑
͓
͑
0
incident beam that the form of the TOF is dominated by the
time it takes for the D to reach the surface, with the conse-
quence that the TOF distribution is consistent with a wide
range of final energy distributions. ͑Recall that the overall
TOF consists of two parts, the TOF for the incident beam to
the sample and the TOF of the product to the detector͒. From
fits to these results, we estimate the mean translational en-
ergy of the CD4 to be between 2 and 5 eV. The solid line on
Fig. 1 shows a typical fit.
By integrating the TOF distributions for different detec-
tion angles we have obtained the angular distribution dis-
played in Fig. 2 in polar form. The line added to the data has
the form: f(␪f)ϭcos60͑␪fϪ1.5͒. This function has a full
width at half maximum ͑FWHM͒ of about 17.4° and is
peaked 1.5° from the surface normal. Since we cannot reli-
ably analyze these TOF distributions to yield velocity distri-
butions, this angular distribution refers to integrated TOF
signals, not fluxes. We note, however, that the peaks of the
TOF’s at different angles are the same within the uncertain-
as discussed in Sec. II D. The solid line in Fig. 3 is a fit that
yields u0ϭ3700 m/s and ␣ϭ1005 m/s. The corresponding
energy distribution associated with this form is shown in Fig.
4. While the fitting procedure is sensitive to the mean en-
ergy, it is less sensitive to the u0/␣ ratio. Comparable fits can
be obtained by fixing u0 in the range 3200–3800 m/s, giving
␣’s in the range 950–1600 m/s. Analyzing TOF distributions
for different detection angles, we find a slight systematic
increase in translational energy as a function of increasing
angle from Ϫ10°–15°. This result is shown on Fig. 5. The
line is a linear fit, which yields Ef(␪f)ϭ(1.8ϩ␪f/45) eV.
Integrating the TOF’s as before, we obtain the polar an-
gular distribution shown in Fig. 6. The line added to the data
has the form: f(␪f)ϭcos70͑␪fϪ5.5͒, which has a FWHM of
16.1° and is peaked 5.5° from the surface normal. The bro-
ken line indicates the result of Fig. 2, for the lower energy
incident D. It is clear that increasing the D energy causes the
angular distribution to deviate further from the normal in the
specular direction.
We have also looked for any dependence of the reaction
on Ts . Within the uncertainties ͑20%͒, TOF’s obtained for
Ts in the range 90–200 K are the same. Above this tempera-
ture, the surface coverage is depleted more rapidly during
exposure due to the LH reaction of D with CD3.
ties of the data, indicating that the velocities at different ␪
are similar.
f
Figure 3 shows a TOF distribution for the CD4 product
of the reaction with D atoms from the thermal dissociation
source for ␪iϭ45° and Tsϭ95 K and detection at ␪fϭ0°. As
for Fig. 1, the broken line indicates the TOF for the incident
atoms when detected directly into the mass spectrometer. In
this case, the TOF distribution of the beam is just slightly
earlier than that for the CD4 product. Since the mass of the
B. CH3 vs CH3I dosing
We have also performed experiments dosing directly
with CH3 radicals from the azomethane pyrolysis source. We
J. Chem. Phys., Vol. 105, No. 22, 8 December 1996
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
130.18.123.11 On: Tue, 23 Dec 2014 18:42:08

Rettner, Auerbach, and Lee: Formation of CD4
10119
FIG. 5. Variation of the translational energy with increasing angle for the
CD4 product of the reaction of D atoms with CD3/Cu͑111͒ for Eiϭ0.25 eV,
␪iϭ45°, and Tsϭ95 K. The line is
Ef(␪f)ϭ(1.8ϩ␪f/45) eV.
a linear fit, which yields
FIG. 7. Comparison of Time-of-flight distribution of CH3D formed by the
reaction of D atoms with CH3/Cu͑111͒ for ͑a͒ direct dosing with a CH3
beam, and ͑b͒ dosing with CH3I, which dissociates to CH3ϩI. These results
were obtained for Tsϭ95 K detecting at 5° from the surface normal. The
solid lines show fits to the data, which indicate that the product translational
energy is slightly higher for CH3 dosing.
find that the form of the TOF distributions are qualitatively
similar to those obtained with CH3I dosing. We note two
important differences, however. First, the distributions are
slightly earlier for CH3 dosing, indicating somewhat higher
product energy. Secondly, the overall signals are larger for
CH3 dosing than for CH3I dosing, consistent with the work
of Lin and Bent,29 who determined that the overall CH3 cov-
erage is higher for direct CH3 dosing, compared to CH3I
dosing. Figure 7 displays TOF distributions for the two dif-
ferent dosing methods for detection at ␪fϭ5°. Figures 7͑a͒
and 7͑b͒ were obtained with CH3 and CH3I dosing, respec-
tively. The broken lines have been added to guide the eye in
comparing the two distributions. The solid lines are the re-
sults of fits, yielding mean translational energies of 2.4Ϯ0.2
eV for CH3 dosing and 2.0Ϯ0.2 eV for CH3I dosing.
C. H؉CH3˜CH4
With the enhanced signals from CH3 dosing, we were
able to study the reaction with incident H, rather than D. In
this case, TOF analysis gave a mean translational energy of
2.2Ϯ0.3 eV. By integrating the TOF’s, we obtained the an-
gular distribution shown in Fig. 8. The curve added to the
plot has the form f(␪f)ϰcos60͑␪fϪ4͒. The broken line shows
the distribution for DϩCD3 for comparison ͑Fig. 6͒.
IV. DISCUSSION
We have seen that incident D͑H͒ atoms react with ad-
sorbed CD3͑CH3͒ to yield methane that is ejected from the
surface in a narrow range of angles and with high transla-
tional energy. The fact that the angular distribution is not
symmetric about the surface normal, but is displaced towards
the specular direction, indicates that the product retains a
‘‘memory’’ of the momentum of the incident atom. At least
part of the parallel momentum of the incident atom must be
retained by the product. This behavior implies that reaction
is direct, with the incident atom abstracting the methyl group
long before it has come to equilibrium with the surface. In
fact, we know that D͑H͒ atoms and CD3͑CH3͒ do not react at
all at equilibrium below about 250 K.25,31 We therefore con-
clude that the reaction of D͑H͒ atoms with adsorbed methyl
occurs by an Eley–Rideal mechanism.
FIG. 6. Polar plot of the angular distributions of the CD4 product of the
reaction of D atoms with CD3/Cu͑111͒ for Eiϭ0.25 eV and Tsϭ95 K. The
angle of the incidence beam ͑45°͒ is indicated graphically by the arrow. The
line corresponds to cos70͑␪fϪ5.5͒. The broken line shows the result for
Eiϭ0.07 eV ͑Fig. 2͒, for comparison.
J. Chem. Phys., Vol. 105, No. 22, 8 December 1996
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
130.18.123.11 On: Tue, 23 Dec 2014 18:42:08

10120
Rettner, Auerbach, and Lee: Formation of CD4
tive chemisorption of CH4 on Pt͑111͒, via the principle of
microscopic reversibility. While the exit channel for the
H(ad)ϩCH3(ad)/Pt͑111͒→CH4 reaction should be similar
to that for the present system, it is important to note that our
system involves an incident gas–phase atom, rather than an
adsorbed atom, providing energy well in excess of any de-
sorption barrier. Moreover, the microscopic reverse of the
present reaction involves an incident CD4 dissociating on the
Cu͑111͒ surface to liberate a gas–phase D atom. Note that
the assymetric angular distribution implies a preference for
off-normal incidence in the reverse process. This result is not
a general characteristic of the reaction but is a consequence
of having the D atom leave at 45°.
In all of these systems, we believe that the products are
accelerated away from the surface by a repulsive potential. A
narrow angular distribution will result if the repulsive inter-
action is relatively flat, forcing the product in a single direc-
tion. Why should the potential be especially flat for the
present system? One possibility is that the methyl reagent is
further out from the surface than other adsorbates, and there-
fore experiences lower corrugation. The adsorption energy
for CH3/Cu͑111͒ is at least 1 eV lower than that for
H/Cu͑111͒ and Cl/Au͑111͒. In the limit of a totally flat po-
tential, the width in the angular distribution will be deter-
mined by the spread in the parallel momentum of the system
prior to reaction. This spread will have contributions from ͑i͒
the incident atom, and ͑ii͒ the adsorbate motion. The spread
in parallel momentum due to the velocity spread in the D
beam can account for only a few degrees of the angular
width. We noted the peak D velocity of ϳ5000 m/s corre-
sponds to a displacement of about 5°. It follows that the
Ͻ3000 m/s spread in D velocities should give Ͻ3° spread in
displacement angles. We expect the energy associated with
the motion of the adsorbates parallel to the surface to be
ϽkTs , or Ͻ8 meV at 95 K. The spread in angles will then be
Ͻ8°, assuming a product energy of 1.8 eV. Although these
factors may play a role in determining the angular width, as
well as the ϳ3° angular spread of the incident beam, it seems
likely that the angular spread does indeed derive largely from
the surface corrugation.
FIG. 8. Polar plot of the angular distributions of the CH4 product of the
reaction of H atoms with CH3/Cu͑111͒ for Eiϭ0.25 eV and Tsϭ95 K. The
angle of the incidence beam ͑45°͒ is indicated graphically by the arrow. The
line corresponds to cos60͑␪fϪ3.5͒. The broken line shows the result for
incident D ͑Fig. 6͒, for comparison.
The displacement of the angular distributions from the
surface normal depends both on the energy of the incident
atom ͑Fig. 2 vs Fig. 6͒ and with its mass ͑Fig. 6 vs Fig. 8͒.
The qualitatively trend is clearly that the displacement angle
increases with increasing parallel momentum, as we should
expect for a direct reaction. For D atoms incident at 0.07 eV
and 45°, parallel momentum conservation would lead to a
displacement of 2.4° for a CD4 molecule leaving the surface
at 1.8 eV. Increasing this energy to 0.25 eV should increase
the displacement to 4.8°. Changing to H atoms incident at
0.25 eV should give a displacement of 3.6° for a CH4 prod-
uct. The displacements of 2.4°, 3.6°, and 4.8° are in rela-
tively good agreement with the peaks in the angular distri-
butions for the respective cases, of 1.5°, 3.5°, and 5.5°.
Perhaps the most striking result of the present study is
the extreme sharpness of the angular distributions. For the
0.25 eV incident D, the distribution has a FWHM of about
16°. We are not aware of any other desorption system with
such a narrow angular spread. The angular distributions for
the reactions OϩCO/Pt͑111͒→CO2,19 HϩD/Cu͑111͒→HD,5
and HϩCl/Au͑111͒→HCl7,8 are about 40°, 44°, and 35°, re-
spectively, i.e., more than a factor of 2 wider. Interestingly,
one of the narrowest distributions reported to date for a sur-
face reaction is for the reaction of CH3 with H on Pt͑111͒.32
In this study, CH3 was produced at the surface by photolyz-
ing CH3Br and the H was present as chemisorbed atoms
͑bound by about 2.6 eV͒. The angular distribution of the CH4
produced upon heating the coadsorbate system follows a
Another striking feature of the present results is the very
high energy of the product. The CD4 energy exceeds 2 eV
for ␪fϾ10°. This is about a factor of 2 higher than for HϩD/
Cu͑111͒ and HϩCl/Au͑111͒. Part of the reason for the
higher energy is that the reaction is more exothermic than
either of these systems. The CD3–D bond energy is ϳ4.5
eV33 while the CD3–Cu͑111͒ desorption energy is ϳ1.3
eV,29 making the reaction about 3.2 eV exothermic. For
11
8
comparison, the HϩD/Cu͑111͒ and HϩCl/Au͑111͒ sys-
tems are both about 2.3 eV exothermic. The heat of reaction
is disposed into four channels, translational, vibrational, and
rotational energy of the product, plus energy transfer to
the surface, Etrans , Evib , Erot , and Esurf , respectively,
where Etransϭ E . For HϩD/Cu͑111͒, we have shown11
͗
͘
f
that these energies are approximately in the ratio:
Etrans :Evib :Erot :Esurfϭ0.43:0.30:0.17:Ͻ0.10. For HϩCl/
cos37 ␪ function, consistent with a FWHM of about 22°. The
f
authors noted that this angular distribution is consistent with
the energy and angular dependence of the activated dissocia-
Au͑111͒, this ratio is 0.27:0.14:0.05:0.54. For the present
system, we do not have information on the energy disposal
J. Chem. Phys., Vol. 105, No. 22, 8 December 1996
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
130.18.123.11 On: Tue, 23 Dec 2014 18:42:08

Rettner, Auerbach, and Lee: Formation of CD4
10121
into the CD4 internal modes. However, we do know that
transϭ1.8 eV, or 0.56 of the available energy. Why should
action potential. In particular, it depends on where the energy
is released along the reaction coordinate. There are two ex-
tremes, attractive energy release—where an appreciable frac-
tion of the energy is released as the reagents approach—and
repulsive energy release—where energy is primarily released
as the products separate. Although we certainly expect en-
ergy to be released as the D atom approaches, we also be-
lieve that there should be considerable repulsive energy re-
lease. ͑1͒ Attractive energy release: Bringing a D atom up to
an isolated CD3 radical leads to the release of the bond en-
ergy of 4.5 eV. Bringing a D atom up to an adsorbed CD3
radical should also lead to the release of a large amount of
energy since the CD3 interacts only weakly with the Cu͑111͒
surface. ͑2͒ Repulsive energy release: Since the D atom is so
much lighter than the CD3, the bond will form before the
CD3 has moved away from the surface, leaving the CD4
molecule close to the position of the original CD3. This po-
sition will be high on the repulsive wall of the CD4/Cu͑111͒
interaction, ͑as the bonding of the CD4 is much weaker than
for CD3͒, giving rise to repulsive energy release. We believe,
therefore, that the potential for this system may be of the
form leading to so-called ‘‘mixed’’ energy release.
Energy released as the D atom approaches should couple
efficiently into vibration of the new bond. As was first rec-
ognized by Evans and Polanyi,34 attractive potentials gener-
ally lead to a high degree of vibrational energy in the nascent
product. Far less vibrational excitation is expected for repul-
sive energy release, however, because of the low D-atom
mass. Although repulsive energy release can lead to vibra-
tional excitation, for the case of a light incident atom, little
vibrational excitation is expected. Low vibrational excitation
occurs because of the so-called ‘‘light-atom anomaly.’’ 35
Light incident atoms get close to the other reagent before
that species is appreciably accelerated by the repulsive
forces. These forces then serve to accelerate the whole new
molecule, leading to translational energy rather than vibra-
tional energy.
The reaction may also lead to considerable vibration ex-
citation of C–D ͑C–H͒ bonds of the original methyl group.
Two processes could contribute to this excitation. First, ex-
citation could result from the sudden change of geometry on
going from adsorbed methyl to methane. For example, if the
adsorbed methyl is closer to the planar structure of isolated
methyl than to the tetrahedral angles of methane, one would
expect to excite bending modes. If the bond length changes
significantly, stretching modes should be excited. Second,
the repulsive force between the surface and the departing
molecule could vibrationally excite the molecule, in a man-
ner similar to the vibrational excitation of NH3 in gas–
surface collisions.36 Only detailed dynamical calculations on
an accurate PES can provide quantitative estimates for the
probabilities of these processes.
E
such a large fraction of the available energy go into transla-
tion in this system? We shall argue that the kinematics of the
present system favor disposal into translational and vibra-
tional motions rather than to rotational motion or to energy
transfer to the surface.
Consider first rotational excitation. We have previously
argued that the low rotational excitation in the case of HϩCl/
Au͑111͒ is a consequence of the low H/Cl mass ratio.8 Simi-
lar arguments apply to the present system. When the incident
atom is light compared to the abstracted species, conserva-
tion of angular momentum limits the accessible product ro-
tational energy. For example, a D atom accelerated to 1 eV
by the gas–surface potential would have a velocity of ϳ104
m/s. A cross section of 10Ϫ16 cm2 gives a maximum impact
parameter ϳ0.6 Å, giving a maximum angular momentum of
equivalent to Jϭ19, which has a rotational energy of Ͻ0.15
eV, amounting to Ͻ0.05 of the available energy. The mean
rotational energy should be even less.
We can also estimate the fraction of energy transfer to
the lattice from the kinematics of the system. Let us model
the surface as a cube with effective mass Ms . As the product
accelerates away from the surface, conservation of linear
momentum demands that the translational energy of the sur-
face cube be Esurf ϭ Etrans(MCH /Ms) ͑neglecting the initial
4
momentum of the incident atom͒. In the limit that MsϭMCu
the mass of one Cu atom, ͑Esurf/Etrans͒ϳ0.3. Since we know
transϭ1.8 eV, EsurfϽ0.5 eV, or Ͻ0.16 of the available en-
,
E
ergy. From high resolution EELS studies, Chiang and Bent25
propose that methyl binds in a bridge site on Cu͑111͒, and
consider it unlikely that methyl binds to a top site. Thus, Ms
is unlikely to be as low as one Cu mass, lowering the esti-
mated energy transfer to recoil still further. In addition to
energy transfer to recoil, it is possible that energy could be
dissipated through adsorbate–adsorbate interactions. How-
ever, the fact that Etrans increases for direct CH3 dosing,
where the coverage is higher, indicates that such effects must
be small. In principle, energy transfer to electron-hole pairs
could adsorb a large fraction of the available energy, as was
discussed in the context of the HϩCl/Au͑111͒ system.8 Al-
though we are not able to estimate this coupling quantita-
tively, we will assume that such coupling is weak in the
present system.
From the above, we therefore have estimates for
ErotϽ0.15 eV and EsurfϽ0.5 eV. Given Etransϭ1.8 eV, we
obtain EvibϾ0.75 eV, or Ͼ0.23 of the available energy. This
level of vibrational excitation is comparable to that for
HϩD/Cu͑111͒ ͑Evibϭ0.7 eV͒, and much higher than for
HϩCl/Au͑111͒ ͑0.32 eV͒. Vibrational motions can be ex-
cited in the nascent CD4 molecule in two dynamically dis-
tinct ways, either through excitation of the newly formed
bond or through excitations of the other C–D bonds resulting
from the sudden change of shape. Consider first the newly
formed bond, which corresponds to the single diatomic bond
excited in HD and HCl ER products.
Finally, note that since the translational energy varies
with final angle, so must the energy disposal into the other
modes. A dependence of Ef on ␪ was also observed for the
f
HϩCl/Au͑111͒ system.8 In that case, it was suggested that
the changes in Ef were associated with changes in the level
of vibrational excitation with ␪f . In that case, the assignment
The degree of vibrational excitation of the newly formed
bond depends on the detailed form of the D–CH3–Cu inter-
J. Chem. Phys., Vol. 105, No. 22, 8 December 1996
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
130.18.123.11 On: Tue, 23 Dec 2014 18:42:08

10122
Rettner, Auerbach, and Lee: Formation of CD4
was based on the observation of bimodal TOF distributions
attributed to the HCl ϭ0 and уϭ1 product. The fact that
2 P. J. Eenshustra, J. H. M. Bonnie, J. Loss, and H. J. Hopmann, Phys. Rev.
Lett. 60, 341 ͑1988͒.
v
v
3 K. Lykke and B. D. Kay, in Laser Photoionization and Desorption Sur-
face Analysis Techniques, edited by N. S. Nogar, SPIE Proc. 1208, 18
͑1990͒.
Ef increases with increasing ␪ in the present system is con-
f
sistent with a fall in the level of vibrational excitation with
increasing ␪f . Again, we must await detailed calculations to
clarify these dynamics.
4 C. C. Cheng, S. R. Lucas, H. Gutleben, W. J. Choyke, and J. T. Yates, Jr.,
J. Am. Chem. Soc. 114, 1249 ͑1992͒.
5 C. T. Rettner, Phys. Rev. Lett. 69, 383 ͑1992͒.
6 M. Xi and B. E. Bent, J. Vac. Sci. Technol. B 10, 2440 ͑1992͒; J. Phys.
Chem. 97 4169 ͑1993͒.
V. SUMMARY AND CONCLUSIONS
7 C. T. Rettner and D. J. Auerbach, Science 263, 365 ͑1994͒.
8 C. T. Rettner, J. Chem. Phys. 101, 1529 ͑1994͒.
9 T. A. Jachilowsku and H. Weinberg, J. Chem. Phys. 101, 10997 ͑1994͒.
10 C. T. Rettner and D. J. Auerbach, Phys. Rev. Lett. 74, 4551 ͑1995͒.
11 C. T. Rettner and D. J. Auerbach, J. Chem. Phys. 104, 2732 ͑1996͒.
12 A. B. Elkowitz, J. H. McCreery, and G. Wolken, Jr., Chem. Phys. 17, 423
͑1975͒.
We have shown that incident D͑H͒ atoms react with CD3
͑CH3͒ adsorbed on Cu͑111͒ by a direct abstraction, or Eley–
Rideal, mechanism. The primary evidence for this mecha-
nism is the absence of symmetry of the angular distribution
of the product about the surface normal. In addition we have
found the following:
13 J. C. Tully, J. Chem. Phys. 73, 6333 ͑1980͒.
14 P. Kratzer and W. Brenig, Surf. Sci. 254, 275 ͑1991͒.
15 B. Jackson and M. Persson, J. Chem. Phys. 96, 2378 ͑1992͒.
16 B. Jackson, M. Persson, and B. Kay, J. Chem. Phys. 100, 7687 ͑1994͒.
17 B. Jackson and M. Persson, J. Chem. Phys. 102, 1078 ͑1994͒.
18 E. W. Kuipers, A. Vardi, A. Danon, and A. Amirav, Phys. Rev. Lett. 66,
116 ͑1991͒.
͑1͒ The cross section for the reaction is ϳ10Ϫ16 cm2/D atom.
͑2͒ The peak of the angular distribution is displaced by an
amount that is approximately consistent with parallel
momentum conservation.
͑3͒ The angular distribution is remarkably sharp, Ͻ17°
FWHM for 0.25 eV D incident on CD3.
͑4͒ The product leaves the surface at very high energy, ϳ1.8
eV, or 56% of the exothermicity.
͑5͒ This energy varies with angle roughly as
Ef(␪f)ϭ1.8ϩ␪f/45 eV.
͑6͒ From kinematic considerations, we estimate ErotϽ0.15
eV and EsurfϽ0.5 eV.
͑7͒ We estimate EvibϾ0.75 eV, or Ͼ0.23 of the available
energy.
͑8͒ Results were found to be approximately independent of
Ts over the range 90–200 K.
19 C. B. Mullins, C. T. Rettner, and D. J. Auerbach, J. Chem. Phys. 95, 8649
͑1991͒.
20 C. T. Rettner and J. Lee, J. Chem. Phys. 101, 10185 ͑1994͒.
21 C. T. Rettner, L. A. DeLouise, and D. J. Auerbach, J. Chem. Phys. 85,
1131 ͑1986͒.
22 C. T. Rettner, H. A. Michelsen, D. J. Auerbach, and C. B. Mullins, J.
Chem. Phys. 94, 7499 ͑1991͒.
23 D. J. Auerbach and C. T. Rettner, Rev. Sci. Instrum. 63, 3939 ͑1992͒.
24 X.-D. Peng, R. Viswanathan, G. H. Smudde, and P. C. Stair, Rev. Sci.
Instrum. 63, 3930 ͑1992͒.
25 C.-M. Chiang and B. E. Bent, Surf. Sci. 279, 79 ͑1992͒.
26 R. Renaud and L. G. Leitch, Can. J. Chem. 32, 545 ͑1954͒.
27 H. C. Ramsperger, J. Am. Chem. Soc. 49, 913 ͑1927͒.
28 C.-A. O. Alen and O. K. Rice, J. Am. Chem. Soc. 57, 310 ͑1935͒.
29 J.-L. Lin and B. E. Bent, J. Phys. Chem. 97, 9713 ͑1993͒.
30 C. T. Rettner and C. B. Mullins, J. Chem. Phys. 94, 1626 ͑1991͒.
31 C.-M. Chiang, T. H. Wentzlaff, and B. E. Bent, J. Phys. Chem. 96, 1836
͑1992͒.
ACKNOWLEDGMENTS
32 V. A. Ukraintsev and Harrison, Surf. Sci. 286, L571 ͑1993͒.
33 CRC Handbook of Chemistry and Physics, 76th ed., edited by D. R. Lide
͑CRC, New York, 1995͒, pp. 9–64.
We are grateful to Dean Pearson for expert technical
assistance and Adel Nazzal for assistance in the synthesis of
the azomethane.
34 M. G. Evans and M. Polanyi, Trans. Faraday Soc. 35, 178 ͑1939͒.
35 P. J. Kuntz, E. M. Nemeth, J. C. Polanyi, S. D. Rosner, and C. E. Young,
J. Chem. Phys. 44, 1168 ͑1966͒.
1 R. I. Hall, I. Cadez, M. Landau, F. Pichou, and C. Schermann, Phys. Rev.
Lett. 60, 337 ͑1988͒; C. Schermann, F. Pichou, M. Landau, I. Cadez, and
R. I. Hall, J. Chem. Phys. 101, 8152 ͑1994͒.
36 B. D. Kay, T. D. Raymond, and M. E. Coltrin, Phys. Rev. Lett. 59, 2792
͑1987͒.
J. Chem. Phys., Vol. 105, No. 22, 8 December 1996
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
130.18.123.11 On: Tue, 23 Dec 2014 18:42:08