H and D atom addition to ethylene on Cu(100): Absence of ethyl H/D shift and decomposition

Article abstract of DOI:10.1021/jp952271f

The addition of gas phase H and D atoms to unsaturated hydrocarbons physisorbed on metal surfaces is a viable synthetic route to partially-deuterated alkyl groups on the surface (Jenks, C. J.; Xi, M.; Yang, M. X.; Bent, B. E. J. Phys. Chem. 1994, 98, 2152-2157). Because these processes are exothermic by ～60 kcal/ mol, the possibility of alkyl decomposition and/or rearrangement prior to thermal accommodation with the surface (a possibility not explicitly addressed in prior studies) should be considered. In the studies here, these decomposition and rearrangement possibilities have been investigated by studying H and D atom addition to variously-deuterated ethylenes physisorbed on a Cu(100) surface. Ethyl decomposition by C-H, C-D, or C-C bond scission has been addressed and shown not to occur by comparison with results from previous studies of the surface species that would be formed by these bond scission processes. H/D shift between the two carbons of the ethyl groups has been addressed by heating the surface to induce β-hydrogen or β-deuterium elimination. The resulting alkene product ratios are compared with those for β-elimination from selectively-deuterated ethyl groups formed by an independent route, i.e., the dissociative adsorption of a labeled bromoethane. The results show that the extent of H/D shift, if it occurs at all, is <5%. On the basis of this finding and the product ratios, it is determined that the kinetic isotope effect (k_H/k_D) for β-hydrogen/β-deuterium elimination is 9.5 ± 0.4 at ～260 K on Cu(100). No secondary isotope effect of D for H substitution at the α-carbon is detected to within the experimental uncertainty. These results demonstrate, at least for a Cu-(100) surface, the feasibility of synthesizing selectively-labeled surface alkyl groups from H and D atom addition to alkenes.

Full text of DOI:10.1021/jp952271f

822
J. Phys. Chem. 1996, 100, 822-832
H and D Atom Addition to Ethylene on Cu(100): Absence of Ethyl H/D Shift and
Decomposition
Michael X. Yang and Brian E. Bent*
Department of Chemistry, Columbia UniVersity, New York, New York 10027
ReceiVed: August 7, 1995; In Final Form: October 11, 1995X
The addition of gas phase H and D atoms to unsaturated hydrocarbons physisorbed on metal surfaces is a
viable synthetic route to partially-deuterated alkyl groups on the surface (Jenks, C. J.; Xi, M.; Yang, M. X.;
Bent, B. E. J. Phys. Chem. 1994, 98, 2152-2157). Because these processes are exothermic by ∼60 kcal/
mol, the possibility of alkyl decomposition and/or rearrangement prior to thermal accommodation with the
surface (a possibility not explicitly addressed in prior studies) should be considered. In the studies here,
these decomposition and rearrangement possibilities have been investigated by studying H and D atom addition
to variously-deuterated ethylenes physisorbed on a Cu(100) surface. Ethyl decomposition by C-H, C-D,
or C-C bond scission has been addressed and shown not to occur by comparison with results from previous
studies of the surface species that would be formed by these bond scission processes. H/D shift between the
two carbons of the ethyl groups has been addressed by heating the surface to induce â-hydrogen or â-deuterium
elimination. The resulting alkene product ratios are compared with those for â-elimination from selectively-
deuterated ethyl groups formed by an independent route, i.e., the dissociative adsorption of a labeled
bromoethane. The results show that the extent of H/D shift, if it occurs at all, is <5%. On the basis of this
finding and the product ratios, it is determined that the kinetic isotope effect (kH/kD) for â-hydrogen/â-deuterium
elimination is 9.5 ( 0.4 at ∼260 K on Cu(100). No secondary isotope effect of D for H substitution at the
R-carbon is detected to within the experimental uncertainty. These results demonstrate, at least for a Cu-
(100) surface, the feasibility of synthesizing selectively-labeled surface alkyl groups from H and D atom
addition to alkenes.
1. Introduction
Scheme 1, two extremes (both very exothermic) may be
envisioned for the nascent product of this reaction: i.e. hot ethyl
radicals or hot surface ethyl groups. The thermochemical cycle
presented in Scheme 2 shows that the enthalpy change for
formation of ethyl radicals (neglecting any radical-surface
interaction) is given by
Extensive studies have been carried out on the addition of
atomic hydrogen to unsaturated π systems in the gas phase.1
More recently, it has also been reported that gas phase hydrogen
atoms add to unsaturated molecules adsorbed on surfaces.2-8
The addition of gas phase H atoms to unsaturated hydrocar-
bons on surfaces2,4 is of interest not only for generating and
isolating adsorbed hydrocarbon fragments which are of catalytic
relevance and which are difficult to prepare by other methods9
but also as a means to generate, via deuterated hydrocarbons
and/or deuterium addition, selectively-deuterated surface frag-
ments. A common feature of these addition reactions is that
they are highly exothermic and that they occur by direct or
quasi-direct mechanisms (Eley-Rideal mechanisms) in which
the H or D atoms react with the unsaturated hydrocarbon prior
to thermally accommodating with the surface.2,4 As a result of
the reactions’ exothermicities and the direct nature of these
process, the possibility of product rearrangement and/or de-
composition prior to accommodation to the surface must also
be considered.10 These rearrangement/decomposition possibili-
ties have important implications with respect to the generation
and isolation of selectively-labeled surface fragments.
∆H1 ) Edes + ∆Hgas
(1)
where Edes is the desorption activation energy of ethylene from
Cu(100) and ∆Hgas is the enthalpy change in the gas phase for
H addition to ethylene under standard conditions. Given that
the activation energy for ethylene desorption from Cu(100), Edes
,
is 8 ( 2 kcal/mol3 and that the gas phase reaction enthalpy is
-36.5 kcal/mol on the basis of the heats of formation of ethyl
radicals (28.0 kcal/mol),11 ethylene (12.5 kcal/mol),12 and H
atoms (52 kcal/mol),13 the enthalpy change for formation of
surface ethyl radicals, ∆H1, is ∼ -28 kcal/mol. In fact, the
actual energy that the nascent ethyl radicals possess could be
even larger, since the H atoms in our studies have a translational
energy of 5-6 kcal/mol.
The other extreme in the D + physisorbed ethylene reaction,
i.e. direct formation of surface ethyl groups, is even more
exothermic than formation of ethyl radicals, since the carbon-
metal bond energy for ethyl groups adsorbed on Cu(100) is 33
( 6 kcal/mol.3 Thus, as indicated in Scheme 2, formation of
surface ethyl groups is accompanied by an enthalpy change of
∆H2 ) ∆H1 - EM-C ) -61 kcal/mol.
Since the exothermicity of this addition reaction is signifi-
cantly larger than the activation energies for a number of copper-
catalyzed ethyl decomposition reactions, one might expect not
only ethyl rearrangement by H/D shifts but also ethyl decom-
position by C-H and C-C bond scission processes. These
In this paper, the issue of product rearrangement and/or
decomposition is addressed for the addition of H and D atoms
to an adsorbed ethylene monolayer on copper surfaces at 110
K. Previous studies2,3 have documented that it is H (and D)
atoms that have not thermally equilibrated with the surface that
react with adsorbed ethylene to form ethyl groups. It is also
clear that this is a highly exothermic process. As shown in
* To whom correspondence should be addressed. Phone: (212)854-
3041. FAX: (212)932-1289. E-mail: Bent@chem.columbia.edu.
X Abstract published in AdVance ACS Abstracts, December 15, 1995.
0022-3654/96/20100-0822$12.00/0 © 1996 American Chemical Society

H and D Atom Addition to Ethylene on Cu(100)
J. Phys. Chem., Vol. 100, No. 2, 1996 823
SCHEME 1
correspond to the exposures of H2 (D2) molecules to the
filament. For reference, an exposure of ∼30 L of H2 or ∼50 L
of D2 is required to saturate a clean Cu(100) surface with H or
D atoms. Also for reference, the average thermal energy for a
Maxwellian distribution at 1800 K is 5.4 kcal/mol. However,
previous time-of-flight studies indicate that a beam of H atoms
generated from thermal dissociation of hydrogen molecules in
a tungsten tube at 2300 K has an energy ∼8.2 kcal/mol, which
is slightly higher than the expected thermal energy of 6.9 kcal/
mol,15 so the average thermal energy on our hot wire source at
1800 K may be slightly higher than expected for a Maxwell
distribution.
SCHEME 2
Dosing was achieved by backfilling the chamber via precision
sapphire leak valves. Exposures are reported in units of
Langmuirs (L), where 1 L equals to 1 × 10-6 Torr‚s. H2
(Matheson, 99.9995%), D2 (Matheson, 99.5%), C2D4 (MSD
isotopes, 99 atom % D) and trans-CDHdCDH (from Prof. R.
Bersohn’s group; unknown original source) were used without
further purification. Bromoethane-1,1,2,2-d4 and iodoethane-
2,2,2-d3 (Cambridge Isotope Laboratory, 98 atom % D) were
purified by several freeze-pump-thaw cycles. Sample purities
were verified in situ by mass spectrometry.
possibilities along with a preview of the conclusions from the
present work are shown in Scheme 1. In prior studies2,3 of this
system, it was assumed that the rearrangement and decomposi-
tion pathways in Scheme 1 did not occur, and one of the
objectives of the current study is to determine whether or not
this is a valid assumption. The second objective is to provide
highly accurate determinations of the deuterium kinetic isotope
effect for â-hydride elimination by ethyl groups on Cu(100).
The focus of the studies here is H and D atom addition to
variously-deuterated ethylenes physisorbed on a Cu(100) surface
at 100 K. The isotope distribution in the resulting ethyl groups
has been determined by measuring the isotopic composition of
the ethylene produced by â-H or â-D elimination from the
surface ethyl groups. To access kinetic isotope effects, the
results are compared with the ethylene product ratio from â-H
(â-D) elimination from labeled ethyl groups formed by dis-
sociative adsorption of BrCD2CD2H. The results indicate that
little or no rearrangement occurs in ethyl groups upon their
formation from H or D addition to ethylene on Cu(100). The
results also substantiate previous conclusions concerning the
kinetic isotope effect (kH,kD) for â-elimination by ethyl groups
on Cu(100).3
Since quantification of isotopic purity is critical for the kinetic
isotope effect results presented in section 3, several comments
on the in situ mass spectra of the variously deuterated
compounds are warranted. In the case of iodoethane-2,2,2-d3
(Cambridge Isotope Laboratory, 98 atom % D), the detected
m/e ) 158 signal is ∼2.5% of that for m/e ) 159 (molecular
ion of C2D3H2I). While cracking of the molecular ion could in
principle account for part of the m/e ) 158 intensity, for the 15
eV electron impact ionization energy used in these studies, no
cracking contribution is expected at this mass. More probably,
the m/e ) 158 intensity is due to impurity C2D2H3I, and the
2.5% relative intensity of this ion places an upper limit on the
amount of this species. For bromoethane-1,1,2,2-d4 (C/D/N,
99.3 atom % D), m/e ) 111 (probably C2D3H2Br) has an
intensity ∼0.6% of that for m/e ) 112 (parent C2D4HBr). For
C2D4 (MSD isotopes, 99 atom % D) small amounts of m/e )
31 (probably C2D3H), m/e ) 30 (probably C2D2H2), and m/e )
28 (probably C2H4 or N2) are observed. The latter two masses
are not an issue for the study here, but the C2D3H (m/e ) 31)
impurity is significant for the experiments. The intensity of
this ion is 4.5% of that for m/e ) 32. In order to confirm that
m/e ) 31 is due to ethylene-d3, TPD studies of m/e ) 31 and
m/e ) 32 evolution after C2D4 adsorption were performed. In
these studies, both m/e ) 31 and 32 showed the same peak
temperatures, and the ion ratio was the same as that measured
in the mass spectra of the dosing gas. On the basis of these
results we can safely conclude that all m/e ) 31 intensity is
due to C2D3H. trans-CDHdCDH (m/e ) 30) has small
intensities at m/e ) 26, 27, 28, and 29. The level of m/e ) 29
(probably ethylene-d1) is ∼5.0% of that of C2D2H2, and this
factor is taken into account in the analysis of the results
presented in section 3.3.
2. Experimental Section
Details of the ultrahigh vacuum (UHV) chamber and the
experimental procedures have been described previously.14 The
Cu(100) crystal (Monocrystals, 99.999%) was mounted on a
molybdenum heating button that can be heated resistively to
1100 K and cooled by liquid nitrogen to 110 K. The sample
temperatures were measured by a chromel-alumel thermo-
couple inserted into a hole drilled into the side of the crystal.
The surface was cleaned by cycles of Ar+ sputtering and
annealing, and the cleanliness of the surface was confirmed by
Auger electron spectroscopy.
Temperature-programmed desorption (TPD) and temperature-
programmed reaction (TPR) studies were performed using a
shielded and differentially-pumped quadrupole mass spectrom-
eter with an electron impact ionization energy of 15 eV to
eliminate most of the fragmentation of the molecular ions.3 The
heating rates were 3 K/s, and up to three ion intensities were
monitored simultaneously by computer-controlled multiplexing
of the mass spectrometer.
3. Results and Interpretation
The experimental results are presented as follows. Section
3.1 briefly reviews the results of D atom addition to ethylene-
d0 to exclude the possibility of C-C and/or C-H bond
dissociation concurrent with D addition. The thermal chemistry
of BrCD2CD2H on the surface is then used in section 3.2 to
determine the kinetic isotope effect for â-hydride elimination
by surface ethyl groups. Finally, this isotope effect is used,
together with the results of D atom addition to trans-
CHDdCHD and H atom addition to CD2dCD2, to determine
the extent of H/D shift between the two carbon atoms during
the H/D addition reaction.
Hydrogen (deuterium) atoms were produced by dissociation
of hydrogen molecules on a hot (∼1800 K) tungsten filament.
During dosing, the crystal was positioned line-of-sight ∼2 cm
in front of the filament. The H (D) atom exposures indicated

824 J. Phys. Chem., Vol. 100, No. 2, 1996
Yang and Bent
3.1. D Atom Addition to Ethylene: Absence of C-H(D)
and C-C Bond Scission. As mentioned in the Introduction
and shown in Scheme 1, possible reactions for excited ethyl
radicals (groups) formed by D addition to ethylene on Cu(100)
include copper-catalyzed C-H (D) and C-C bond scission. On
the basis of previous studies, it is known that the activation
energies for these various processes are comparable to or less
than the exothermicity of the ethyl formation, which is at least
30 kcal/mol (see Introduction). For example, previous studies
have shown the surface ethyls undergo â-hydride elimination
on Cu(100) to yield ethylene and a chemisorbed H atom with
a reaction barrier of only 14.5 ( 2 kcal/mol.16,17 By contrast,
the activation energy for R-hydrogen elimination by methyls is
27 ( 2 kcal/mol on copper surfaces.14,18 No C-C bond
dissociation by alkyl groups has been reported on Cu(100), but
previous studies have indicated an energy barrier of 39 ( 14
kcal/mol for ethyl group C-C bond cleavage on Cu(110).14
The possibility of these various copper-catalyzed C-H and
C-C bond scission reactions after H addition to ethylene but
prior to thermal accommodation of the ethyl product with the
Cu(100) surface at the 100 K surface temperature of reaction
has been addressed by TPR studies of the D + C2H4 reaction.
For example, if â-hydride elimination by the nascent ethyl
intermediate were to occur at the D addition reaction temperature
of 110 K, then both the kinetic isotope effect for elimination
(kH/kD ∼ 10, see sections 3.2 and 3.3) and the H:D ratio of 2:1
at the â-carbon [This ratio assumes that there is no rearrange-
ment by 1,2 H or D shift of the ethyl species formed by D
additionsan assumption which is confirmed by studies presented
in section 3.3. Note that if rearrangement were to occur, the
average H:D ratio at the â-carbon would be g2.] would favor
â-H elimination over â-D elimination. As a result, some
ethylene-d1 would be generated on the surface at the addition
reaction temperature of 110 K. In subsequent heating of the
surface, ethylene-d1 evolution would be observed at the ethylene
molecular desorption temperature of <140 K. Note that we
treat the possibility that the D atom added to the CdC bond is
preferentially eliminated (because the CsD bond formed is
vibrationally excited) as equivalent to a lack of D addition.
Figure 1. Temperature-programmed reaction spectra of m/e ) 18
(CH2D2), 28 (C2H4), 29 (C2H3D), and 30 (C2H2D2) after impinging 5
langmuirs (L) of D atoms onto a surface precovered with a 1.0 L
exposure of C2H4. The spectra were obtained with a surface heating
rate of 3 K/s and an electron impact ionization energy of 15 eV.
elimination before thermal accommodation to the substrate and
still be consistent with the data here is 2%.
Given that â-H elimination is the most facile copper-catalyzed
ethyl reaction pathway, the absence of â-hydride elimination
suggests that significant R-hydride elimination and CsC bond
dissociation before ethyl thermal accommodation to the surface
are unlikely. These predictions are confirmed by experiments.
For example, suppose that CRsH bond dissociation occurs to
form ethylidene-2-d1 (CH2DCHd) on the surface. Studies
presented elsewhere19 have shown that the excess surface D
atoms (generated from D atom dosing) would scavenge these
ethylidene groups to form ethyl-1,2-d2 (CH2DCHDs) groups,
which would then form ethylene-1,2-d2 (m/e ) 30) via â-hydride
elimination at ∼255 K. As shown in Figure 1, the absence of
an ethylene-d2 (m/e ) 30) peak at ∼255 K excludes the
possibility of R-hydride elimination by the ethyl groups formed
at 110 K on Cu(100).
The experimental results regarding the possibility of â-H
elimination at the D addition temperature of 110 K are presented
in Figure 1, which dislays TPD spectra taken after dosing “5
L” D (10% of clean surface saturation) onto a Cu(100) surface
precovered by 1 L of C2H4 (∼40% of monolayer saturation).
As shown, unreacted ethylene (m/e ) 28) has a molecular
desorption peak only at <140 K. Ethylene-d1 (m/e ) 29),
however, shows a peak at ∼258 K, which, as previously
documented, is due to â-hyrdide elimination by surface ethyl
groups (CH2CH2D) formed in the D addition reaction. The lack
of an appreciable ethylene (m/e ) 28) peak at 250 K reflects
the absence of detectable â-D elimination by surface ethyl-d1
groups, which can be attributed to a combination of the large
kinetic isotope effect for reaction and the stoichiometry in the
ethyl group (H:D ) 2:1 in the â position [This ratio assumes
that there is no rearrangement by 1,2 H or D shift of the ethyl
species formed by D additionsan assumption which is con-
firmed by studies presented in section 3.3. Note that if
rearrangement were to occur, the average H:D ratio at the
â-carbon would be g2.]). The key feature of these studies with
respect to â-H elimination during D addition at 110 K is that
no ethylene-d1 evolution is detected at <140 K. The intensity
of the very small m/e ) 29 peak at 132 K in Figure 1 is 2.4 (
0.4% of that of m/e ) 28 peak and is entirely attributable to
the 2.2% natural abundance of 13C in C2H4. An upper limit
for the percent of ethyl groups that could undergo â-hydride
Finally, the possibility of CsC bond cleavage can be
addressed on the basis of what is known about the surface
reactions of methylene (CH2d) and methyl-d1 (CH2Ds) groups
on Cu(100). Specifically, it is known from prior studies that
at ∼240 K methylene (CH2d) reacts by one of three pathways
on Cu(100): coupling with other CH2 groups to form ethylene-
d0, coupling with surface methyl groups to form ethyl-d1 (CH2-
DCH2s), and/or coupling with surface D atoms to form methyl-
d1 (CH2Ds) groups.20 Mass balance implies that, unless all of
the CH2 groups recombine with CH2D groups to reform ethyl-
d1 (highly unlikely in the presence of surface D14), then
regardless of the relative yields of the three possible processes,
there must be methyl-d1 (CH2Ds) groups left on the surface at
temperatures >250 K. Furthermore, it is known that surface
methyl groups readily combine with surface D atoms at ∼350
K to form methane, and in this case the product would be
methane-d2 (CH2D2, m/e ) 18).14 As shown in Figure 1, the
absence of both methane-d2 (m/e ) 18) evolution at ∼350 K
and ethylene-d0 (m/e ) 28) evolution from CH2 coupling at
∼240 K indicate that no detectable CsC bond cleavage occurs
during D addition to physisorbed ethylene at 110 K on Cu-
(100).
3.2. Thermal Chemistry of Bromoethane-1,1,2,2-d4: De-
termination of the Kinetic Isotope Effect for â-Elimination
from Surface Ethyl Groups. As will be elaborated in section

H and D Atom Addition to Ethylene on Cu(100)
J. Phys. Chem., Vol. 100, No. 2, 1996 825
Figure 3. Temperature dependence of the kinetic isotope effect (kH/
kD) for â elimination for CD2HCD2- groups on Cu(100). The isotope
effect is given by twice the ratio of the H and D elimination rates in
the TPR spectra shown in Figure 2. Different symbols represent data
from different experiments. The solid and dotted lines (which are
essentially indistinguisable because of overlap) represent curve-fitting
results assuming that the kH/kD ratio is of the form A exp(-E/RT). The
fitting parameters are A ) 1.17, E ) 1.08 kcal/mol for the solid line
and A ) 1.00, E ) 1.16 kcal/mol for the dotted line, as discussed in
the text.
Figure 2. Evolution of C2D4 (m/e ) 32) and C2D3H (m/e ) 31) after
adsorption of 1.0 L of CD2HCD2Br on Cu(100) at 110 K. The spectra
were obtained with a surface heating rate of 3 K/s. The inset schematics
show the origin of the observed ion intensities.
3.3, â-hydride elimination together with isotope labeling has
been applied to investigate the extent of H/D shift between the
two carbon atoms in ethyl groups formed by H/D addition to
physisorbed ethylene. An important value required in these
studies is the deuterium kinetic isotope effect for â-hydride
elimination. In the present section, we present a determination
of this isotope effect from studies of the thermal chemistry of
bromoethane-1,1,2,2-d4. The idea is that bromoethane-1,1,2,2-
d4 dissociates on Cu(100) to generate labeled ethyl groups (CD2-
CD2H) and that the relative rates at which CD2CD2 and
CD2CDH are formed by â-H/â-D elimination from this species
in a TPR experiment provide the kinetic isotope effect.
Bromoethane monolayers on Cu(100) partially dissociate to
yield ethyl groups and bromine atoms on the surface. In the
absence of H or D shift during this dissociative adsorption
reaction (an assumption to be verified below), bromoethane-
1,1,2,2-d4 generates ethyl-1,1,2,2-d4 groups on the surface. The
results of â-hyrdide elimination by these ethyl-1,1,2,2-d4 groups
are presented in Figure 2, which shows the evolution of
CD2dCD2 and CD2dCDH as a result of â-H and â-D
elimination, respectively. (Previous studies have established that
the coadsorbed halogen (bromine) atoms do not have a
substantial effect on the â-hydride elimination rate.3) The
isotope effect for â-hydride elimination (kH/kD) is given by twice
the ratio of the H and D elimination rates, where the factor 2
accounts for the D/H ratio in the â-position of the ethyl group.
Seven measurements were conducted in which both m/e ) 32
(C2D4) and m/e ) 31 (C2D3H) evolution were monitored
simultaneously. (It is worth emphasizing again that the mass
spectrometer is operated with an electron impact ionization
energy of 15 eV throughout the studies reported here, so the
molecular ion intensities represent more than 98% of the total
ion intensity in the cracking patterns.) Neglecting for a moment
the temperature dependence of kH/kD, an average value (deter-
mined by integrating the TPR peaks) of 4.7 ( 0.1 is found for
the ratio of the H-elimination and D-elimination yields. After
correcting for the 0.6% impurity of bromoethane-d3 in bromo-
ethane-1,1,2,2-d4 (see section 2) and taking into account the
2:1 D:H ratio at the â-carbon, this result corresponds to a kinetic
isotope effect of 9.5 ( 0.4 for â-hydride elimination over the
247-265 K temperature range of the TPR peak.
dide elimination. Since the TPR peak heights are directly
proportional to the rates of product evolution, if we make the
reasonable assumption that CD2CD2 and CD2CDH have the
same detection sensitivities in the mass spectrometer, than the
temperature dependence of the isotope effect can be determined
from twice the ratio of the CD2CD2 and the CD2CDH peak
heights. After correcting for the 0.6% impurity of bromoethane-
d3 in bromoethane-1,1,2,2-d4 (see section 2), the results are
shown in Figure 3, where different symbols are used to represent
data from independent experiments. The temperature range is
limited to 247-265 K on the basis of the signal-to-noise ratio
in the data. Although the temperature range is limited, there
appears to be a temperature dependence to the kH/kD ratio. The
kH/kD ratio decreases with increasing temperature, indicating that
H elimination has a lower energy barrier. Further discussion
of the temperature dependence is deferred to section 4.2.
The isotope effect determination above is based on the
assumption that no H or D shift occurs in ethyl groups during
thermal dissociation of the carbon-bromine bond, so that the
ethyl groups formed on the surface are exclusively ethyl-
1,1,2,2-d4. To investigate the possibility of H/D shift during
the thermal dissociation of carbon-halogen bonds in ethyl
halides, the dissociative adsorption of ethyl iodide-2,2,2-d3 has
been studied on Cu(100). In the absence of H/D shift, thermal
dissociation of the CsI bond generates ethyl-2,2,2-d3 (CD3-
CH2s) groups, which will form exclusively CD2dCH2 by â-D
elimination. However, either D shift upon CsI bond dissocia-
tion or reversible formation of ethyl from CH2dCD2 + D would
produce some ethyl-1,1,2-d3 (CH2DCD2s) groups on the
surface, which would eliminate a â-H to form CD2dCDH.
Figure 4 presents the results of CsI dissociation and â-elimina-
tion for a 1 L exposure ethyl iodide-2,2,2-d3. The evolution of
H-elimination (m/e ) 31) and D-elimination (m/e ) 30) products
are monitored simultaneously. The results show an m/e ) 31
(CD2dCDH) intensity, which is ∼5.0% of the m/e ) 30
(CD2dCH2) signal. Taking into account the 1.1% natural
abundance of 13C in CD2dCH2, we can conclude that the extent
of H/D shift upon CsI bond dissociation is less than 2.8%. In
fact, the degree of H/D shift is probably negligible, and the
2.8% of C2D3H is probably attributable to either some revers-
ibility in the â-elimination reaction (which has been previously
demonstrated for higher surface coverages3) or impurities (such
as ethyl iodide-1,2,2-d3 in the ethyl iodide-2,2,2-d3 sample.
The results in Figure 2 also allow determination of the
temperature dependence of the kinetic isotope effect for â-hyr-

826 J. Phys. Chem., Vol. 100, No. 2, 1996
Yang and Bent
Figure 4. Evolution of C2H2D2 (m/e ) 30) and C2D3H (m/e ) 31)
after adsorption of 1.0 L of CD3CH2I on Cu(100) at 110 K. The spectra
were obtained with a surface heating rate of 3 K/s and an electron
impact ionization energy of 15 eV.
Figure 5. Evolution of C2D4 (m/e ) 32) and C2D3H (m/e ) 31) after
impinging 5 L of H atoms onto a surface precovered with a 1.0 L
exposure of C2D4. The spectra were obtained with a surface heating
rate of 3 K/s and an electron impact ionization energy of 15 eV.
SCHEME 3
products with the yields determined in the control experiments
utilizing the labeled ethyl halides in section 3.2.
Figure 5 shows TPR spectra taken after exposing a Cu(100)
surface precovered with a partial monolayer of C2D4 (about 40%
of monolayer saturation) to 5 L of H atoms. As will be
discussed further in section 4.1, the extent of rearrangement is
expected to be independent of the absolute coverage of ethyl
groups formed in the addition reaction. The TPR experiments
in Figure 5 were carried out using an electron impact ionization
energy of 15 eV, so that the ion intensities at m/e ) 32 and 31
in Figure 5 accurately represent the evolution of ethylene-d4
and ethylene-d3, respectively. Notice that the â-hydride elimi-
nation peaks are observed at 258 K, which is the same to within
the (5 K experimental reproducibility as the peak temperature
of 261 K for â-elimination by the ethyl-1,1,2,2-d4 groups
generated from the thermal dissociation of ethyl bromide, as
presented in section 3.2 (Figure 2). (The higher temperature
of 273 K for â-elimination by ethyl-2,2,2-d3 groups in Figure
4 reflects the deuterium isotope effect for this reaction.) As in
the studies of Figure 2, the average H to D elimination reaction
yield ratio (yield-H/yield-D) over the 240-270 K temperature
range of the studies of Figure 5 can be determined by comparing
the m/e ) 32 (ethylene-d4) and m/e ) 31 (ethylene-d3) peak
areas. On the basis of three sets of experiments, the product
yield ratio is 4.2 ( 0.2. After correction is made for the 4.2%
C2D3H impurity in the C2D4 source (see section 2), the value is
5.1 ( 0.3. A complete discussion of the procedure for this
correction can be found in ref 3.
3.3. H Addition to Ethylene-d4 and D Addition to trans-
CHDdCHD: Extent of H/D Shift. As shown in Scheme 1, a
possible reaction for the nascent ethyl radical/group intermediate
is H or D atom shift between the two carbon atoms. While
calculations indicate that the barrier for a 1,2 H atom shift in
gas phase ethyl radicals is ∼84 kcal/mol,21,22 the possibility of
a surface-promoted H shift with a barrier less than the 30-60
kcal/mol exothermicity of the addition reaction cannot be ruled
out.
Using the results from sections 3.1 and 3.2, the possibility
of H/D atom shift upon ethyl radical/group formation from
hydrogen atom addition to ethylene on Cu(100) has been studied
by two different isotopic variations of the reaction. The first is
H addition to CD2dCD2, and the second is D addition to trans-
CDHdCDH. As shown in Scheme 3, in the absence of any
H/D shift, H addition to ethylene-d4 (CD2dCD2) yields ethyl-
1,1,2,2-d4 (CD2HCD2s), and D addition to ethylene-1,2-d2
(CDHdCDH) forms ethyl-1,2,2-d3 (CD2HCDHs). In other
words, both reactions produce ethyl groups with one hydrogen
and two deuterium atoms at the â position. On the other hand,
if H and/or D shifts occur, then the two nascent ethyl groups
generated by these addition reactions can convert to other
structural isomers. In the case of CD2HCD2s, only one other
structural isomer (CD3CDHs) is possible, while, for CD2-
HCDHs, both CDH2CD2s and CD3CH2s can be formed. As
a result, different H and D populations will be present at the
â-carbons in the ensemble of surface ethyl groups if H or D
shifts occurs. The extent of rearrangement can be quantified
by comparing the relative yields of the H and D elimination
Figure 6 presents TPR results from the reaction of “5 L” D
(∼10% of saturation) with a Cu(100) surface precovered by 1
L of trans-CHDdCHD (∼40% of monolayer saturation). The
ion intensities at m/e ) 31 and 30 represent the evolution of
ethylene-d3 and ethylene-d2 from â-H and â-D elimination.
Again, the relative yields of H and D elimination (yield-H
/
yield-D) can be determined from the results by comparing the
peak areas for ethylene-d3 and ethylene-d2 evolution. On the
basis of ten separate measurements, the yield ratio is 4.1 ( 0.2.
After a correction is made for the 4.0% ethylene-d1 impurity in
the ethylene-d2 source (see section 2), the value for the H to D
elimination reaction yield ratio (yield-H/yield-D) is 4.9 ( 0.3.
These results for H addition to ethylene-d4 and for D addition
to ethylene-1,2-d2 will be discussed in the following section,

H and D Atom Addition to Ethylene on Cu(100)
J. Phys. Chem., Vol. 100, No. 2, 1996 827
Figure 6. Evolution of C2D3H (m/e ) 31) and C2H2D2 (m/e ) 30)
after impinging 5 L of D atoms onto a surface precovered with a 1.0
L exposure of trans-CDHdCDH. The spectra were obtained with a
surface heating rate of 3 K/s and an electron impact ionization energy
of 15 eV.
Figure 7. Time evolution of the fractional surface coverages of the
two ethyl structural isomers (CD2HCD2- and CD3CDH-) possible for
C2D4H stoichiometry as a function of the number of H/D shifts between
the two carbon atoms for three different relative rates of deuterium
and hydrogen shift (rH/rD): (A) fast hydrogen shift (rH ) 10rD); (B)
equal rate for deuterium and hydrogen shift (rD ) rH); (C) fast deuterium
shift (rD ) 10rH). The initial coverages were taken as 1 for CD2HCD2-
and 0 for CD3CDH- to correspond to the experimental results for H
atom addition to C2D4 on Cu(100).
where it is shown that little or no H/D shift occurs during ethyl
group formation.
4. Discussion
4.1. Extent of H/D Shift. In order to quantify the extent of
H/D shift from the reaction kinetics data presented in section
3.3, we evaluate first the changes in the â-H(D) populations
that would be expected as a function of the extent of H/D shift.
In both of the experiments described (H + CD2dCD2 and D +
CHDdCHD), the nascent ethyl groups formed by addition have
two D atoms and one H atom at the â-carbon. Note that even
if the energy released upon H or D addition to the double bond
is not rapidly randomized within the ethyl fragment, so that
there is a preferential shift of the H or D atom that was just
added, this shift does not affect the isotope distribution in the
ethyl groups, since both ethylene molecules in these studies have
the same isotope composition at either end of the molecule. In
other words, immediate transfer of the added H or D to the
adjacent carbon is equivalent to initial addition at the adjacent
carbon, and the species formed by addition at either carbon are
equivalent with respect to isotope composition at the R and â
positions. For the purposes of the discussion that follows, we
assume that all â-hydrogen atoms have the same shift rate (rH)
and that all â-deuterium atoms have the same shift rate (rD).
We consider first the effects of H and D shift on H addition to
C2D4.
to the rate of change of coverage of either isomer by
dθ
dθ
CD2CD2H
CDHCD3
-
)
) 3rDθCDHCD - 2rDθCD CD H
3 2 2
dt
dt
) rD(3θC°D CD H - 5θCD CD H
)
2
2
2
2
where θ°
is the initial coverage of CD2CD2H formed
CD2CD2H
upon H atom addition to CD2dCD2 and
+ θ ) θ°
CD2CD2H
θ
CD2CD2H
CDHCD3
The time evolution of two isomers is therefore given by
3
5
2
5
θCD CD H ) θ°
+ e-5r t
D
CD CD H
(
)
2
2
2
2
2
5
2
θCDHCD ) θ°
- e-5r t
(3)
D
CD CD H
(
)
3
2
2
5
Figure 7 shows representative examples for the evolution of
the surface concentrations of the two isomers as a function of
the “number of H and D shifts” for three different relative rates
of deuterium and hydrogen shift (rD/rH). Because we have no
experimental measure of the relative rates of H and D transfer,
the plots in Figure 7 show the effects of an orders of magnitude
change in the relative rates. One “shift”, as indicated on the x
axis in Figure 7, is defined by the time constant 1/(2rD + rH).
For the case in Figure 7B, where there is no isotope effect for
H/D transfer and rD ) rH ) r, this time constant is 1/3r, and
this value reflects the average time per 1,2 shift event per ethyl
group, where a shift event is defined as a single exchange of
an H or D atom between the two carbon atoms. In the cases
where there is a deuterium kinetic isotope effect for the 1,2
(i) H Addition to CD2dCD2. The nascent ethyl group formed
by H atom addition to CD2dCD2 is ethyl-1,1,2,2-d4 (CD2-
HCD2s). Note that a 1,2 hydrogen shift maintains the lone
hydrogen atom at the â position, but a 1,2 deuterium shift
converts CD2HCD2s to its structural isomer CD3CDHs, which
has no hydrogen atom at the â position. These two structural
isomers, which are the only two possible without intermolecular
exchange, are interconverted only by D transfer between
carbons, and this interchange process can be written as
2rD
CD2CD2H y z CDHCD3
(2)
3rD
where rD is the rate constant for exchange of a single D atom.
The coverages of the two ethyl structural isomers can be related
shift (rD * rH, as in Figure 7A and C), the value of 1/(2rD +

828 J. Phys. Chem., Vol. 100, No. 2, 1996
Yang and Bent
rH) reflects the average time per shift event per ethyl only in
the limit of zero time, where there are one hydrogen and two
deuterium atoms at the â position of all surface ethyl groups.
As shown in Figure 7 and by eq 3, for small times (t f 0),
rates of population change for two of the three structural isomers
are given by
dθ
CDHCD2H
) 2rHθ
- rHθ
CDHCD2H
CD2CDH2
dt
dθ
CD2CD2H
) rH(2θC°DHCD H - 3θCDHCD H - 2θCH CD
)
) -2rDθ°
CD2CD2H
2
2
2
3
dt
dθ
dθ
CH2CD3
CDHCD3
) rDθ
- 3rDθ
) 2rDθC°D CD H
CD2CDH2
CH2CD3
2
2
dt
dt
) rD(θC°DHCD H - θCDHCD H - 4θCH CD )
3
and deuterium atom transfer in CD2HCD2- to form CD3CDH-
2
2
2
(6)
occurs at a rate of 2rDθC°D CD H, where θ°
is the initial
CD2CD2H
2
coverage of CD2CD2H assumin2 g that all CD2CD2H species are
formed simultaneously. On the other hand, for infinite times
(t f ∞), the two structural isomers reach equilibrium, with
where θCDHCD H + θCD CDH + θCH CD ) θC°DHCD H and θC°DHCD H
2
2
2
2
3
2
2
is the initial coverage of ethyl-1,2,2-d3 (CD2HCDH) formed by
D addition to CDHdCDH. Equations 6 above represent a
system of two linear first-order differential equations with two
variables for which the general solution has the form
3
5
∞
θ
) θ°
CD2CD2H
CD2CD2H
θCDHCD H ) C0 + C1e-r t + C2e-r t
1
2
2
5
θ∞CDHCD ) θ°
(4)
2
CD2CD2H
3
θCH CD ) C′0 + C′1e-r t + C′2e-r t
1
2
2
3
Note that this is the result expected for a statistical distribution
of one H and four D atoms between the two R and three â
positions in an ethyl group. The implication of the results in
Figure 7 for the experimentally-measured â elimination product
ratio is given by
Determination of the time constants r1 and r2 and the coefficients
C0, C1, C2, C′0, C′1, and C′2 from the constraints imposed by
eqs 6 gives the time evolution for each of the three structural
isomers as
Kiso
θ
θCDHCD H
)
CD2CD2H
2
yield-H
yield-D
2 + Kiso
)
(5)
5 rH - 2r2
5 (r1 - r2)
5 rH - 2r1
5 (r2 - r1)
3
5
e-r t
+
e-r t
2
1
2
θ°
+
θ
+ θ
CDHCD3
CDHCD2H
CD2CD2H
(
)
)
2 + Kiso
where yield-H/yield-D indicates the yield of the hydrogen
elimination product (CD2dCD2) relative to the yield of the
deuterium elimination product (CD2dCDH) and Kiso is the
isotope effect (kH/kD) for the â-hydride elimination reaction.
Notice from eqs 3 and 5 that the relative yields of H and D
elimination products are a function of only the shift rate constant
and the isotope effect for â-hydride elimination and are
independent of the initial coverage of CD2CD2H formed in the
H atom addition reaction. Using the results in Figure 7 and
the Kiso of 9.5 determined in section 3.2, the product yield ratio
is plotted as a function of number of shifts for various relative
H and D scrambling rates in Figure 9 (open circles). Further
comments on these results will be made after discussing the
analogous possibilities for the D + CHDdCHD experiment.
(ii) D Addition to CHDdCHD. D atom addition to trans-
CHDdCHD forms initially ethyl-1,2,2-d3 (CD2HCDHs). In
this case, deuterium shift between carbons does not change (at
least initially) the ethyl structural isomer, but hydrogen shift
converts CD2HCDHs to CDH2CD2s. This new isomer can
then be either transformed back to CD2HCDHs through a
hydrogen shift or converted to a third species CD3CH2s via a
deuterium atom shift. The final possibility is conversion of CD3-
CH2s back to CD2HCDHs through a deuterium shift. The
entire reaction system can be summarized as
θCD CDH
)
2
2
3 r2 - 10rH
10 (r1 - r2)
3 r1 - 10rH
10 (r2 - r1)
3
10
e-r t
+
e-r t
1
2
θ°
+
CDHCD2H
(
θCH CD
)
2
3
r2
r1
1
10
1
2
θ°
+
e-r t
+
e-r t (7)
CDHCD2H
(
)
10(r1 - r2)
10(r2 - r1)
where
3rH + 4rD + [(3rH - 4rD)2 + 8rHrD]1/2
r1 )
r2 )
2
3rH + 4rD - [(3rH - 4rD)2 + 8rHrD]1/2
2
The curves in Figure 8 show the evolution of the surface
coverages of the three ethyl structural isomers as a function of
the extent of rearrangement for different relative rates of D and
H shifting. As for Figure 7, the “number” of shifts is defined
by 1/(2rD + rH), where a value of 1 corresponds to an average
of one shift per ethyl in the absence of any isotope effects.
As shown in Figure 8 and by eqs 6 and 7, as t f 0,
CDHCD2H yrHz CD2CDH2
2rH
dθ
CDHCD2H
) -rHθ°
CD2CDH2 yrDz CH2CD3
CDHCD2H
dt
3rD
dθ
CD2CDH2
) rHθ°
CDHCD2H
dt
where rH and rD are the rate constants for hydrogen and
deuterium shift per CâsH and CâsD bond, respectively. The

H and D Atom Addition to Ethylene on Cu(100)
J. Phys. Chem., Vol. 100, No. 2, 1996 829
Figure 8. Time evolution of fractional surface coverages of the three
ethyl structural isomers (CD2HCDH-, CDH2CD2-, and CD3CH2-)
possible for C2D3H2 stoichiometry as a function of number of H/D
shifts between the two carbon atoms for three different relative rates
of deuterium and hydrogen shift (rH/rD): (A) fast hydrogen shift (rH )
10rD); (B) equal rate for deuterium and hydrogen shift (rD ) rH); (C)
fast deuterium shift (rD ) 10rH). The initial coverages were taken as
1 for CD2HCDH- and 0 for CDH2CD2- and CD3CH2- to correspond
to the experimental results for D atom addition to CDHdCDH on Cu-
(100).
Figure 9. Relative â-H and â-D elimination yields as a function of
the extent of H/D shift for three different relative rates of deuterium
and hydrogen shift (rH/rD). Open circles represent the relative product
yields for H atom addition to C2D4. Solid circles represent the relative
product yields for D atom addition to CDHdCDH. The zero time value
of 4.7 in all cases reflects the isotope effect measured for H and D
elimination from CD2HCD2- groups formed by dissociative adsorption
of CD2HCD2Br.
Notice again that the relative yield of the two reaction products
is independent of the initial coverage of CDHCD2H formed by
D addition to CHDdCHD.
dθ
CH2CD3
) 0
dt
(iii) Implications of the Experimental Results. As illustrated
above, in both H addition to C2D4 and D addition to CDHdCDH,
H/D shift in the nascent product ethyl group will change the H
and D population at the â position and consequently the ratio
of H and D elimination reaction yields. Given the deuterium
kinetic isotope effect, Kiso ) kH/kD, of 9.5 determined in section
3.2, the relationship between the relative H and D elimination
reaction yields and the extent of H/D shift can be predicted
from eqs 5 and 9. Figure 9 presents these relationships for both
H addition to ethylene-d4 and D addition to ethylene-1,2-d2 for
three different relative rates of H and D shift (rD/rH).
In other words, 1,2-hydrogen shift initially transforms CD2-
HCDHs to CDH2CD2s, but the coverage of CD3CH2s
remains zero. At equilibrium (t f ∞), the populations of the
three structural isomers are given by
3
5
∞
θ
) θ°
CDHCD2H
CDHCD2H
3
10
∞
θ
)
θ°
CDHCD2H
CD2CDH2
The general characteristics of the plots in Figure 9 can be
understood as follows. In the case of H addition to ethylene-
d4, the nascent product is ethyl-1,1,2,2-d4 (CD2HCD2-). With
no D shift, one hydrogen and two deuterium atoms are present
at the â position, and on the basis of the bromoethane studies
in section 3.2, the expected yield ratio for H and D elimination,
yield-H/yield-D, is 4.7. This value is the point at x ) 0 for all
three plots in Figure 9. In the case of D shift, the CD2HCD2-
group coverage decreases, accompanied by an increase in the
CD3CDH- group coverage, as shown in Figure 7. Thus, the
H population at the â position steadily decreases, resulting in a
monotonic decrease in the H elimination yield, as shown by
the open circles in all three plots of Figure 9.
On the other hand, D addition to ethylene-1,2-d2 forms ethyl-
1,2,2-d3 (CD2HCDH-), and for short times (t f 0), the initial
decrease in the CD2HCDH- coverage as a result of H shift is
mirrored by an increase in the CDH2CD2- coverage while the
CD3CH2- coverage remains zero. Thus, the average H
population at the â position increases, which results in a slight
increase in the â-H elimination yield relative to the value of
1
10
θ∞CH CD
)
3
θ°
(8)
CDHCD2H
2
which is consistent with a statistical distribution of two H and
three D atoms between two R and three â positions. From an
ensemble of these three different structural isomers, two
â-hydride elimination products can be formed: ethylene-d3 and
ethylene-d2. The relative yield of hydrogen elimination product
(ethylene-d3) vs deuterium elimination product (ethylene-d2) is
given by
yieldC D H
2
3
)
yieldC D H
2 2
2
Kiso
2Kiso
θ
+
θ
CD2CDH2
CDHCD2H
2 + Kiso
1 + 2Kiso
1
(9)
2
θ
+
θ
+ θ
CH2CD3
CDHCD2H
CD2CDH2
2 + Kiso
1 + 2Kiso

830 J. Phys. Chem., Vol. 100, No. 2, 1996
Yang and Bent
4.7 for no rearrangement. This increase is, however, less
significant for a faster relative D shift rate, since D shift
transforms CDH2CD2- to CD3CH2-, which has no H popula-
tion at the â position. The limiting value for yield-H/yield-D
for a statistical distribution of the three structural isomers (see
Figure 8) and for a kH/kD ratio of 9.5 (see section 3.2) is 3.5
(see Figure 9).
of H and D elimination are given by
rate-H AH exp(-EH/RT)
A
)
)
H exp(-(EH - ED)/RT)
rate-D
AD
AD exp(-ED/RT)
where the A’s and E’s are the Arrhenius prefactors and activation
energies for H and D elimination. A fit of this function to the
results in Figure 3 gives the solid line shown and the following
fit parameters: AH/AD ) 1.17 and EH - ED ) -1.08 kcal/mol.
As discussed in refs 24 and 25, the issues with respect to the
possibility of tunneling are the extent to which the prefactor
ratio deviates from 1 and the magnitude of the action energy
difference with respect to the difference in zero-point energies
for the “reactive vibrational modes” in the reactants or products.
In the present system, the experimentally-determined zero-point
energy difference of 1.08 kcal/mol for â-H vs â-D elimination
is quite similar to the difference in zero-point energies for C-H
and C-D bond stretching, which are expected to be the reactive
coordinates. Specifically, the reported values of 2935 and 2153
cm-1 for C-H and C-D bond stretching in ethyl groups on
Cu(111) correspond to an energy difference of 1.12 kcal/mol.27
In the case of the prefactor ration, transition state theory
predicts the following:
The results in Figure 9 can be used to interpret the
experimental results as follows. As presented in section 3.3,
the ratio of H and D elimination product yields is 5.1 ( 0.3
after H addition to ethylene-d4 and 4.9 ( 0.3 after D addition
to trans-ethylene-1,2-d2. Comparing these results with the plots
presented in Figure 9, we can conclude that the extent of
rearrangement by 1,2 H or D shift in ethyl groups formed by H
or D addition to physisorbed ethylene on Cu(100) is immeasur-
ably small. On the basis of the experimental uncertainties, we
can place an upper limit on the extent of rearrangement at 5%.
4.2. Kinetic Isotope Effect for â-Hydride Elimination.
Having established that there is no detectable 1,2 H or D shift
in partially deuterated ethyl groups formed by H or D addition
to ethylene on Cu(100), we can use the relative rates measured
for â-H and â-D elimnination from these selectively labeled
alkyls to determine the kinetic isotope effect(s) for â elimination.
In the case of H addition to C2D4 to form CD2HCD2-, kH/kD
for â elimination is determined from the studies in section 3.3
to be 10.3 ( 0.7 over the temperature range 245-265 K. To
within the experimental uncertainty, this value is the same as
that of 9.5 ( 0.4 determined for â elimination from CD2HCD2-
groups generated by dissociative adsorption of bromoethane-
1,1,2,2-d4. Thus, while previous studies have shown that the
presence of coadsorbed bromine does not significantly affect
the rate of â-hydride elimination, the studies here show that
bromine also does not significantly affect the deuterium kinetic
isotope effect for this reason.
AH YH κH QqHQ0,D
)
q
AD YD κ
D QDQ0,H
where Y is a correction factor for tunneling, κ is the transmission
coefficient, Qq is the partition function for the transition state
(excluding the critical vibrational mode in the reaction
coordinate), and Q0 is the partition function for the reactant.
Since in the studies here, â-H and â-D elimination both occur
from the same reactant, Q0 is the same for the two reactions.
Furthermore the difference in Qq for the two reactions corre-
sponds (assuming that the reaction coordinate is C-H/C-D
stretching) simply to Qvib(C-D)/Qvib(C-H) ) 1 - e-hν /kT/(1 -
C-H
e-hν /kT). This ratio is a function of temperature, ranging from
The possibility of secondary kinetic isotope effects can be
addressed by comparing the results above with those for â
elimination from CD2HCDHs groups generated by D addition
to CHDdCHD. These latter studies show a kinetic isotope
effect of 9.8 ( 0.6 for the temperature range 245-265 K. To
within the experimental uncertainty, this isotope effect is the
same as that measured for CD2HCD2s groups, as described
above. We conclude that there is no measurable 2° isotope
effect for â elimination as a result of D for H substitution at
the R-carbon. This result is not surprosing given that studies
by Madix and Telford of methoxy decomposition by â elimina-
tion on Cu(110) have shown that there is no measurable 2°
isotope effect for this reaction despite H,D substitution at the
â-carbon.23
C-D
1.00 at 0 K to νC-H/νC-D at infinite temperature. At 260 K,
this ratio is 1.00. Assuming κH/κD ) 1, the nominal experi-
mental AH/AD ratio of 1.17 is >1 and indicative of tunneling
corrections. However, we estimate an experimental uncertainty
of at least (0.2 in the measured value of 1.17 on the basis of
the dashed line fit to the data in Figure 3 for a prefactor ratio
of 1.00 and an activation energy difference of 1.16 kcal/mol.
Thus, while tunneling cannot be ruled out, one need not invoke
tunneling to account for the experimental results. Similar
conclusions have been reached previously for other C-H bond
scission processes at metal surfaces.23,26
4.3. Competition between Alkyl Group Decomposition/
Rearrangement and Energy Dissipation to the Substrate. The
absence of any detectable ethyl group decomposition/rearrange-
ment in H atom addition to ethylene at 110 K on Cu(100)
indicates that the nascent ethyl radicals/groups dissipate sub-
stantial amounts of energy (30-60 kcal/mol) and accommodate
to the surface prior to the occurrence of these processes. In
the case of ethyl rearrangement by 1,2-hydrogen shift, the
absence of reaction is probably attributable to a high barrier
for reaction. Specifically, if one views the 1,2 shift as occurring
while the ethyl species exists as a free radical prior to forming
a metal-carbon bond, then the energetics of the process are
probably similar to those for a 1,2 shift in gas phase ethyl
groups, where a barrier of 84 kcal/mol is determined from the
potential energy surface calculated for the H + CH2dCH2
system.11,21 On the other hand, if the 1,2 shift were to occur
The possible role of tunneling in the ethyl â-hydride
elimination reaction on Cu(100) can be addressed using the
temperature-dependence measurements in Figure 3 for kH/kD
elimination from CD2HCD2-. As discussed in refs 23-26, two
diagnostics for tunneling are the relative activation energies and
the relative prefactors in the Arrhenius rate constants for the
isotopic variants of the reaction. These values can be deter-
mined from the results in Figure 3. In particular, if we make
the reasonable assumptions that Arrhenius expressions can be
used to represent the temperature dependences of the rate
constants and that the â-H and â-D elimination rates have the
same dependences on surface coverage, then the relative rates

H and D Atom Addition to Ethylene on Cu(100)
J. Phys. Chem., Vol. 100, No. 2, 1996 831
mol exothermicity for formation of adsorbed ethyl groups, then
the half-life for ethyl groups on the surface is ∼10 ps. On the
other hand, if only the 30 kcal/mol of energy released upon
formation of the copper-carbon bond is available for reaction,
then the half-life is >10 ns.
While no measurements have been made of energy transfer
between hot alkyls and surfaces, measurements of energy
relaxation for vibrationally excited carbon monoxide on metals
indicate energy transfer on the picosecond time scale. For
example, the lifetime for CO in V ) 1 has been determined to
be ∼3 ps on Pt(111)28 and ∼1.2 ps on Cu(100).29 The lifetime
for intra-adsorbate energy transfer in the case of the symmetric
CH3 stretching mode in CH3S on Ag(111) is also only ∼3 ps,
but adsorbate to surface energy transfer in this system occurs
on a much longer time scale with a time constant of ∼5 ns.29
In the case of ethyl formation from H + ethylene on Cu(100),
adsorbate-substrate energy transfer is probably significantly
faster than in the CH3S/Ag(111) system, since formation of the
copper-carbon bond places energy directly in the mode that
couples the adsorbate to the substrate. Thus, on the basis of
the expected adsorbate-substrate energy transfer on the order
of picoseconds and a statistical reaction rate of >10 ns for
internal energies <30 kcal/mol, the lack of â elimination by
ethyl groups during the H + ethylene reaction on Cu(100)
appears conceptually reasonable.
Despite the predominance of quenching over reaction in the
ethyl/Cu(100) system, there is evidence from other studies that
“hot” adsorbates can have chemically-significant lifetimes on
surfaces. For example, O atoms produced by photodissociation
of O2 on Pt(111) have been shown to react with coadsorbed
species prior to thermal accommodation with the surface.30-32
There are also reports of analogous reactions by hot atoms and
radicals produced by the thermal dissociation of adsorbates,33-39
and in the case of O2 dissociation on Al(111) there is evidence
from STM that hot O atoms produced by dissociative adsorption
of O2 diffuse ∼80 Å across the surfaces at 300 K with a lifetime
of >1 ps prior to thermal accommodation.40,41 While these
results provide evidence for a range of interesting phenomena,
further studies are needed to assess the relative rates of reaction
and quenching in excited adsorbate systems and to determine
the extent to which statistical theories are applicable for
predicting reactivity.
Figure 10. Unimolecular rate constant for ethyl â elimination of Cu-
(100) as a function of total energy available and the number of classical
oscillators, s, over which the energy is statistically distributed. As
described in the text, the reaction rate has been determined using RRK
theory for a unimolecular â elimination reaction with an energy barrier,
E0, of 15 kcal/mol.
after formation of the metal-carbon bond, then the shift process
would require metal-carbon scission, which has a barrier of
33 ( 6 kcal/mol. In either case, the expected energy barrier is
larger than that for a number of ethyl decomposition processes
which are also not observed.
The absence of ethyl decomposition is interesting, particularly
since the activation energy for â elimination is 14.5 ( 2 kcal/
mol on Cu(100),16,17 which is only one quarter of the overall
addition reaction exothermicity and less than half the energy
released by forming the copper-carbon bond. To address
whether or not this result is surprising given what is known
about the rates of energy dissipation from adsorbates to surfaces,
we consider first the rate expected for â elimination as a function
of the ethyl internal energy. Assuming that the internal energy
of the ethyl group is randomized and that the internal energy
available is significantly larger than the barrier for reaction (a
reasonable assumption in this case until significant energy has
been dissipated to the substrate), then the probability that the
critical oscillator along the reaction coordinate contains energy
greater than or equal to the energy required for reaction is given
by Rice-Ramsperger-Kassel (RRK) theory as
E - E0
s - 1
(
)
E
5. Conclusions
The addition, at 110 K, of gas phase H and D atoms to
physisorbed ethylenes on a Cu(100) surface has been studied.
The results show that, despite the 60 kcal/mol exothermicity of
this reaction, there is no detectable CsC or CsH bond
dissociation in the nascent ethyl products. Studies of H addition
to C2D4 and D atom addition to CDHdCDH also show that
the extent of H/D shift between two carbon atoms of the product
ethyl prior to thermal accommodation with the surface is
negligible. These results demonstrate the potential of such H
atom addition reactions for the synthesis of surface alkyl groups.
The results also demonstrate the potential of isotope labeling
for forming selectively-deuterated fragments on the surface, and
in the present studies, selectively-labeled ethyl groups have been
utilized to measure the deuterium kinetic isotope effect for
â-hydride elimination on Cu(100).
where E is the energy available, E0 is the barrier for reaction,
and s is the number of oscillators over which the energy is
statistically distributed. The unimolecular rate constant is simply
this probability multiplied by the vibrational frequency of the
critical oscillator. As an upper limit to the rate constant, we
choose the highest frequency mode associated with the Câ-H
bond being broken, i.e. the Câ-H stretch, as the critical
oscillator. Note that this is also the only mode which, in the
absence of tunneling, can account quasi∼classically for the large
kinetic isotope measured for â-hydride elimination (see Figure
3). This mode has a vibrational frequency of 9.0 × 1013 s-1
.
Given this value, Figure 10 presents the RRK unimolecular
reaction rate constant for â-hydride elimination by ethyl groups
on Cu(100) as a function of the total energy available and as a
function of the number of oscillators over which the energy is
distributed. If we neglect energy dissipation into the copper
lattice to determine an upper limit for the rate, then the number
of oscillators over which the energy is distributed for adsorbed
ethyl groups (note that adsorption converts rotations and
translations to hindered rotations and translations) is 21. In this
extreme case, if the total energy available is the full 60 kcal/
Acknowledgment. Financial support from the National
Science Foundation (Grant CHE 93-18625), from the Joint
Services Electronics Program through the Columbia Radiation
laboratory (Contract DAAH04-94-G-0057), and from the Dow
Chemical Company is gratefully acknowledged. We thank Dr.

832 J. Phys. Chem., Vol. 100, No. 2, 1996
Yang and Bent
(20) Lin, J.-L.; Chiang, C.-M.; Jenks, C. J.; Yang, M. X.; Wentzlaff, T.
H.; Bent, B. E. J. Catal. 1994, 147, 250.
(21) Sloane, C. S.; Hase, W. L. Faraday Discuss. Chem. Soc. 1977, 62,
210.
(22) Hase, W. L.; Mrowka, G.; Brudzynski, R. J. J. Chem. Phys. 1978,
69, 3548.
J. Tully, Prof. M. Gordon, and Prof. R. Bersohn for helpful
discussions in connection with this work, and we thank Dr. X.
Huang and Prof. R. Bersohn for loan of the CDHdCDH used
in these studies.
(23) Madix, R. J.; Telford, S. G. Surf. Sci. 1995, 328, 576.
(24) Melander, L. Isotope Effects on Reaction Rates; Ronald Press: New
York, 1960.
(25) Bell, R. P. The Tunnel Effect in Chemistry; Chapman and Hall:
New York, 1980.
(26) Yata, M.; Madix, R. J. Surf. Sci. 1995, 328, 171.
(27) Lin, J.-L.; Bent, B. E. Chem. Phys. Lett. 1992, 194, 208.
(28) Beckerle, J. D.; Casassa, M. P.; Heilweil, E. J.; Cavanagh, R. R.;
Stephenson, J. C. J. Electron Spectrosc. Relat. Phenom. 1990, 54/55, 17.
(29) Harris, A. L.; Levinos, N. J.; Rothberg, L.; Dubois, L. H.; Dhar,
L.; Shane, S. F.; Morin, M. J. Electron Spectrosc. Relat. Phenom. 1990,
54/55, 5.
(30) Relak, P. A.; Ho, W. Surf. Sci. 1994, 321, L233.
(31) Mieher, W. D.; Ho, W. J. Chem. Phys. 1993, 99, 9279.
(32) Ukraintsev, V. A.; Harrison, I. Surf. Sci. Lett. 1983, 286, L571.
(33) Wartnaby, C. E.; Stuck, A.; Yeo, Y. Y.; King, D. A. J. Chem. Phys.
1995, 102, 1855.
(34) Sun, Y.; Weinberg, W. H. J. Chem. Phys. 1991, 94, 4587.
(35) Peden, C. H. F.; Goodman, D. W.; Weisel, M. D.; Hoffman, F. M.
Surf. Sci. 1991, 253, 44.
(36) Matsushima, T. Surf. Sci. 1983, 127, 403.
(37) Allers, K.-H.; Pfnu¨r, H.; Feulner, P.; Menzel, D. J. Chem. Phys.
1994, 100, 3985.
(38) Au, C.; Carley, A. F.; Pashuski, A.; Read, S.; Roberts, M. W.
Springer Ser. Surf. Sci. 1993, 33, 241.
(39) Xi, M.; Bent, B. E. Langmuir 1994, 10, 505.
(40) Brune, H.; Wintterlin, J.; Behm, R. J.; Ertl, G. Phys. ReV. Lett.
1992, 68, 624.
References and Notes
(1) Jones, W. E.; MacKnight, S. D.; Teng, L. Chem. ReV. 1973, 73,
407 and references therein.
(2) Xi, M.; Bent, B. E. J. Vac. Sci. Technol., B 1992, 10, 2440.
(3) Jenks, C. J.; Xi, M.; Yang, M. X.; Bent, B. E. J. Phys. Chem. 1994,
98, 2152.
(4) Teplyakov, A. V.; Bent, B. E. J. Chem. Soc., Faraday Trans. 1995,
91, 3645.
(5) Son, K. A.; Mavrikakis, M.; Gland, J. L. J. Phys. Chem. 1995, 99,
6270.
(6) Son, K. A.; Gland, J. L. J. Am. Chem. Soc. 1995, 117, 5415.
(7) Horn, A.; Schnek, A.; Biener, J.; Winter, B.; Lutterloh, C.;
Wittmann, M.; Ku¨ppers, J. Chem. Phys. Lett. 1994, 231, 193.
(8) Xie, J.; Mitchell, W. J.; Lyons, K. J.; Weinberg, W. H. J. Chem.
Phys. 1994, 101, 9195.
(9) Zaera, F. Acc. Chem. Res. 1992, 25, 260.
(10) We are grateful to Dr. John Tully for calling this issue to our
attention.
(11) Castelhano, A. L.; Griller, D. J. J. Am. Chem. Soc. 1982, 104, 3655.
(12) Domalski, E. S.; Hearing, E. D. J. Phys. Chem. Ref. Data 1988,
17, 1637.
(13) Handbook of Chemistry and Physics, 71st ed.; CRC Press: West
Palm Beach, FL, 1990.
(14) Chiang, C.-M.; Wentzlaff, T. H.; Bent, B. E. J. Phys. Chem. 1992,
96, 1836.
(15) Rettner, C. T. Phys. ReV. Lett. 1992, 69, 383.
(16) Jenks, C. J.; Chiang, C.-M.; Bent, B. E. J. Am. Chem. Soc. 1991,
113, 6308.
(17) Jenks, C. J.; Bent, B. E.; Bernstein, N.; Zaera, F. J. Am. Chem.
Soc. 1993, 115, 308.
(18) Lin, J.-L.; Bent, B. E. J. Vac. Sci. Technol., A 1992, 10, 2202.
(19) Yang, M. X.; Bent, B. E.; Bare, S. R.; Holbrook, M. T. Manuscript
in preparation.
(41) Brune, H.; Wintterlin, J.; Trost, J.; Ertl, G.; Wiechers, J.; Behm,
R. J. J. Chem. Phys. 1993, 99, 2128.
JP952271F