Differential reactivity and structure of mono- and dialkoxides: The reactions of ethylene glycol on Mo(110)

Article abstract of DOI:10.1021/ja953987r

The reactions of ethylene glycol on Mo(110) were studied using temperature-programmed reaction, infrared reflection absorption, high-resolution electron energy loss, and X-ray photoelectron spectroscopies. The major reaction pathway is double C-O bond scission to evolve gas-phase ethylene at 350 and 390 K. Both X-ray photoelectron and infrared spectra demonstrate the existence of two surface intermediates, a bidentate (-OCH₂CH₂O-) and a monodentate (-OCH₂CH₂OH) species, at saturation coverage of ethylene glycol. We demonstrate that all ethylene glycol in the mixed overlayer of mono- and bidentate species reacts via a bidentate surface intermediate. Furthermore, in contrast to previous studies on other surfaces, the dialkoxide ethylene glycol intermediate is shown to be more reactive than similar monoalkoxides on Mo(110). Finally, analysis of the infrared spectra demonstrates that the bidentate species adsorbs with C₂ (or lower) symmetry at 300 K.

Full text of DOI:10.1021/ja953987r

3896
J. Am. Chem. Soc. 1996, 118, 3896-3904
Differential Reactivity and Structure of Mono- and
Dialkoxides: The Reactions of Ethylene Glycol on Mo(110)
K. T. Queeney,† C. R. Arumainayagam,‡ M. K. Weldon,†,§ C. M. Friend,*,† and
M. Q. Blumberg‡
Contribution from the Departments of Chemistry, HarVard UniVersity,
Cambridge, Massachusetts 02138, and Wellesley College, Wellesley, Massachusetts 02181
ReceiVed NoVember 20, 1995X
Abstract: The reactions of ethylene glycol on Mo(110) were studied using temperature-programmed reaction, infrared
reflection absorption, high-resolution electron energy loss, and X-ray photoelectron spectroscopies. The major reaction
pathway is double C-O bond scission to evolve gas-phase ethylene at 350 and 390 K. Both X-ray photoelectron
and infrared spectra demonstrate the existence of two surface intermediates, a bidentate (-OCH2CH2O-) and a
monodentate (-OCH2CH2OH) species, at saturation coverage of ethylene glycol. We demonstrate that all ethylene
glycol in the mixed overlayer of mono- and bidentate species reacts via a bidentate surface intermediate. Furthermore,
in contrast to previous studies on other surfaces, the dialkoxide ethylene glycol intermediate is shown to be more
reactive than similar monoalkoxides on Mo(110). Finally, analysis of the infrared spectra demonstrates that the
bidentate species adsorbs with C2 (or lower) symmetry at 300 K.
Introduction
(111)5 and Ni(100),6 the C-C bond in the ethylenedioxy
intermediate cleaves, liberating CO and H2; no evidence was
found for C-O bond scission on Rh(111). However, the
dominant reaction channel on Ni(100) is nonselective decom-
position to atomic oxygen and carbon.
The lower reactivity of the bidentate ethylenedioxy species
compared to a related monodentate methoxy (CH3O-) species
on Ag(110) was attributed to the negative entropy of activation
for the decomposition of the dialkoxide intermediate.2 Specif-
ically, it was proposed that during transition state formation the
bidentate species loses the rotational freedom associated with
chelating to a single Ag atom, while the methoxide transition
state has the same entropy as the adsorbed state, presumably
bound to a high coordination site. The dialkoxide intermediate
is expected to be more constrained than its monoalkoxide
analogue, hindering the access of the â-hydrogens to the surface
and thus contributing to the decrease in the rate of â-hydrogen
elimination in the dialkoxide.
The reactions of oxygen-containing compounds on metal
surfaces have been extensively explored due, in part, to the
prevalence of oxygenates in metal-bonding adhesives and liquid
feedstocks derived from coal, shale, and tar sands. In addition,
the reactions of alcohols on metal surfaces have also been widely
studied because they are both primary reactants and products
of catalytic partial oxidation reactions as well as reactants in
deoxygenation processes. For example, the industrial conver-
sion of methanol to formaldehyde is accomplished by catalytic
partial oxidation reactions over silver surfaces.1 The surface
chemistry of diols is of particular interest because the presence
of two identical alcohol groups could, in principle, permit
reaction channels involving the interaction of one or both OH
groups with the surface. As the simplest diol, ethylene glycol
(HOCH2CH2OH) affords an excellent opportunity to compare
the stability of alkoxide (monodentate) and dioxometallacyclic
(bidentate) intermediates on extended surfaces.
The high strength of the Mo-O bond compared to other
metal-oxygen bonds suggests that ethylene glycol will exhibit
markedly different reactivity on Mo(110) compared to Ag(110)
and, indeed, to all other surfaces investigated to date. For
example, carbon-oxygen bond scission occurs in all mono-
dentate alkoxides studied on Mo(110) at temperatures between
350 and 400 K.7-10
In this study, we have used temperature-programmed reaction,
X-ray photoelectron, infrared reflection absorption, and high-
resolution electron energy loss spectroscopies to probe the
structure, bonding, and reactivity of ethylene glycol on Mo-
(110). We demonstrate the existence of a direct intramolecular
elimination pathway involving selective C-O bond scission,
resulting in evolution of gaseous ethylene (C2H4) with ∼85%
On most surfaces studied to date, the C-O bond in ethylene
glycol is retained. For example, ethylene glycol adsorbs
reversibly on clean Ag(110) but undergoes O-H bond activation
on a partially oxidized Ag(110) surface to form a cyclic
ethylenedioxy (-OCH2CH2O-) intermediate and water (H2O)
below 300 K.2 The ethylenedioxy intermediate decomposes via
C-H bond activation, yielding glyoxal ((CHO)2) at 380 K.
Similarly, ethylene glycol dehydrogenation to form glyoxal also
occurs on Cu(110), presumably via the same dialkoxy inter-
mediate, although no vibrational data were reported for this
surface.3 Excess atomic oxygen on Ag(110) induces C-C bond
scission in the cyclic ethylenedioxy, yielding water, formalde-
hyde (H2CO), and surface formate (HCO2).4 On both clean Rh-
* To whom correspondence should be addressed.
† Harvard University.
(5) Brown, N. F.; Barteau, M. A. J. Phys. Chem. 1994, 98, 12737.
(6) Madix, R. J.; Yamada, T.; Johnson, S. W. Appl. Surf. Sci. 1984, 19,
43.
(7) Wiegand, B. C.; Uvdal, P.; Serafin, J. G.; Friend, C. M. J. Phys.
Chem. 1992, 96, 5063.
(8) Wiegand, B. C.; Uvdal, P. E.; Serafin, J. G.; Friend, C. M. J. Am.
Chem. Soc. 1991, 113(17), 6686.
‡ Wellesley College.
§ Present address: AT&T Bell Laboratories, 600 Mountain Ave., Murray
Hill, NJ 07974.
X Abstract published in AdVance ACS Abstracts, April 15, 1996.
(1) Heterogeneous Catalysis: Principles and Applications, 2nd ed.; Bond,
G. C., Ed.; Clarendon Press: Oxford, 1987.
(2) Capote, A. J.; Madix, R. J. J. Am. Chem. Soc. 1989, 111, 3570.
(3) Bowker, M.; Madix, R. J. Surf. Sci. 1982, 116, 549.
(4) Capote, A. J.; Madix, R. J. Surf. Sci. 1989, 214, 276.
(9) Serafin, J. G.; Friend, C. M. Surf. Sci. 1989, 209, L163.
(10) Weldon, M. K.; Uvdal, P.; Friend, C. M.; Serafin, J. G. J. Chem.
Phys. 1995, 103, 5075.
0002-7863/96/1518-3896$12.00/0 © 1996 American Chemical Society

Differential ReactiVity of Mono- and Dialkoxides
J. Am. Chem. Soc., Vol. 118, No. 16, 1996 3897
selectivity. For the first time, we identify both the mono- and
bidentate moieties on the same surface, making use of the high
resolution attainable with surface infrared spectroscopy. In
contrast to findings on other transition metal surfaces, we find
that the bidentate species at saturation coverage on Mo(110) is
more, not less, reactive than both the co-existing monodentate
species and similar monoalkoxide surface intermediates. Fi-
nally, by analysis of the C-H stretching region of the infrared
spectrum, we deduce that the bidentate moiety adsorbs with C2
symmetry or less, i.e. the ethylene dialkoxy intermediate has
no mirror plane of symmetry on Mo(110).
Experimental Section
Temperature-programmed reaction experiments were performed in
a stainless steel ultrahigh vacuum chamber with a base pressure of 5
× 10-11 Torr at Wellesley College. The UTI Model 100C quadrupole
mass spectrometer used in these experiments was enclosed by a shield
possessing a small (0.125-in. diameter) aperture in line-of-sight with
the crystal surface, in order to optimize the detection of molecules
desorbing from the center of the Mo(110) crystal. The crystal could
be cooled to 85 K with liquid nitrogen and heated to 800 K radiatively
or to 2200 K by electron bombardment. The temperature was measured
by a W-5% Re-vs-W-26% Re thermocouple spot-welded to the edge
of the crystal. As the surface was radiatively heated from 85 to 650 K
at a rate of ∼5 K s-1 several masses were sampled as a function of
temperature during a single temperature-programmed experiment under
a multiplexing arrangement.11 Since the crystal is close to the mass
spectrometer filament during these experiments, the crystal was
negatively biased (-100 V) prior to each experiment to prevent
electron-induced reactions. The temperature-programmed reaction
experiments were repeated in conjunction with all spectroscopic
measurements to confirm that the spectra probed intermediates impor-
tant in the chemical processes.
X-ray photoelectron experiments were performed in an ultrahigh
vacuum chamber, described elsewhere,12,13 which is equipped with a
Physical Electronics ESCA 5300 X-ray photoelectron spectroscopy
system. The spectra were collected using a Mg KR (1253.6 eV) anode
and a bandpass energy of 17.9 eV. All binding energies are referenced
to the Mo(3d5/2) peak at 228.3 eV. Spectra were recorded at 100 K
after annealing the ethylene glycol overlayer to the indicated temper-
ature, using the same heating rate as for the temperature-programmed
reaction experiments. The O(1s), Mo(3d), and C(1s) regions were
collected in a single experiment consisting of 75, 5, and 100 scans,
respectively, resulting in a total collection time of 15 min. The spectra
did not change as a function of time, indicating that there was no
significant X-ray induced decomposition of the adsorbate during data
acquisition. This conclusion was further verified by the correspondence
of temperature-programmed reaction data obtained prior to and after a
15-min exposure of the adsorbate layer to the X-ray source.
Figure 1. Temperature-programmed reaction spectra of ethylene glycol
on Mo(110) for (A) ethylene glycol (m/e 31), (B) ethylene (C2H4, m/e
26), and (C) dihydrogen (H2, m/e 2) as a function of ethylene glycol
exposure. The relative exposures are (i) 0.11, (ii) 0.15, (iii) 0.38, (iv)
0.81, and (v) 1 (saturation) ) 0.14 ML.18 The spectra are uncorrected
for fragmentation, and the m/e 26 trace contains a contribution from
the parent molecule. All masses shown are the most intense for that
species, except for ethylene, where m/e 26 was chosen over the most
intense fragment, m/e 28, to avoid contributions from background CO.
A nearly constant heating rate of ∼5 K s-1 was used.
The long-range order and surface cleanliness of the Mo(110) crystals
were verified with low-energy electron diffraction and Auger electron
spectroscopy, respectively, in ultrahigh vacuum chambers at Harvard
University. One Mo(110) crystal was transferred to the chamber at
Wellesley and subsequently cleaned in situ by oxidation in oxygen (2
× 10-9 Torr at a surface temperature of 1200 K for 5 min) followed
by heating briefly to 2200 K. The carbon coverage on the surface
after the cleaning procedure described above was estimated to be 0.002
ML (1 monolayer (ML) ) 1.4 × 1015 atoms/cm2), using temperature-
programmed reaction spectroscopy to quantify the evolution of CO at
high temperature from the reaction of C(a) and O(a) following oxygen
adsorption at 85 K.
Both ethylene glycol samples, HOCH2CH2OH (Aldrich, 99.8%) and
HOCD2CD2OH (Cambridge Isotope, 98% D enrichment), were stored
over baked molecular sieves, distilled under vacuum, and degassed by
3 freeze-pump-thaw cycles before use each day. Oxygen (Wesco,
99.994%) was used without further purification. The purity of the gases
introduced into the ultrahigh vacuum chamber was confirmed by mass
spectrometry. Direct dosers with precision leak valves allowed for
controlled deposition of the ethylene glycol onto the clean Mo(110)
crystal surface at a surface temperature of 85 K while the chamber
pressure remained below 3 × 10-10 Torr.
Surface infrared experiments were performed in a third ultrahigh
vacuum chamber, also described in detail elsewhere.10,14 All infrared
reflection absorption spectra were collected at a surface temperature
of 100 K using a single-beam, clean-air-purged Fourier transform
infrared spectrometer (Nicolet, Series 800) and averaged over 1000
scans using an MCT-A detector at 2-cm-1 resolution, resulting in a
scan time of approximately 15 min. Sample spectra were ratioed against
a background taken immediately after the sample scan by flashing the
crystal to 700 K. The background scan was initiated after the crystal
temperature had returned to 100 K. Temperature-programmed reaction
investigations of isotopically mixed ethylene glycol overlayers were
also performed in this chamber, using the same type of mass
spectrometer and experimental setup as in the chamber at Wellesley
College. The high-resolution electron energy loss spectrum of ethylene
glycol was recorded in a separate ultrahigh vacuum chamber equipped
with an LK2000-14-R spectrometer operated at a primary beam energy
of 3 eV. Both the chamber and the experimental procedure are
described in detail elsewhere.15,16
Results
Temperature-Programmed Reaction Spectroscopy. Eth-
ylene is the major gas-phase product of the ethylene glycol
temperature-programmed reaction on clean Mo(110), accounting
for ∼85% of the reaction of a saturated monolayer of ethylene
glycol. Gaseous dihydrogen and surface atomic oxygen are also
formed in this pathway. Nonselective decomposition to ad-
sorbed atomic carbon and oxygen and gaseous dihydrogen is a
secondary reaction channel. A trace amount of water desorption
(∼5 × 10-4 ML) was also detected at ∼350 K. Ethylene glycol
evolution is also observed at ∼195 K for exposures above ∼0.8
of saturation and is attributed to ethylene glycol multilayer de-
sorption, since this peak increases in intensity without limit as
the ethylene glycol exposure is increased (Figure 1A(iv-v)).17
(15) Wiegand, B. C.; Friend, C. M.; Roberts, J. T. Langmuir 1989, 5,
1292.
(11) Liu, A. C.; Friend, C. M. ReV. Sci. Instrum. 1986, 57, 1519.
(12) Roberts, J. T.; Friend, C. M. J. Am. Chem. Soc. 1986, 108, 7204.
(13) Xu, X.; Friend, C. M. J. Am. Chem. Soc. 1991, 113, 6779.
(14) Weldon, M. K.; Friend, C. M. Surf. Sci. 1994, 310, 95.
(16) Bol, C. W. J.; Friend, C. M. J. Am. Chem. Soc. 1995, 117, 5351.
(17) Intensity in this peak prior to saturation of the ethylene glycol
monolayer is attributed to three-dimensional clustering, as has been
previously observed for ethylene glycol on Ag(110).2

3898 J. Am. Chem. Soc., Vol. 118, No. 16, 1996
Queeney et al.
All masses between 2 and 150 amu were monitored during these
experiments, and no other gaseous reaction products were
detected.11
conclusion, as well as the assignment of the â1 tail at 500 K to
reaction-limited dihydrogen evolution, is confirmed by temper-
ature-programmed reaction of HOCD2CD2OH. The observed
predominance of H2 in the â2 state indicates that this peak
originates from the recombination of hydroxyl protons and that,
therefore, facile O-H bond scission occurs below 390 K. In
contrast, the reaction-limited â1 tail at 500 K consists mainly
of D2, resulting from recombination of surface deuterium formed
from dehydrogenation of a surface hydrocarbon intermediate.
Finally, the saturation coverage for ethylene glycol on Mo(110)
is estimated to be 0.14 ML, based on the relative dihydrogen
yields observed during temperature-programmed reaction of
ethylene glycol and a known coverage of methanol on Mo-
(110).10
Ethylene is formed in two distinct states with peak temper-
atures of ∼350 and ∼390 K during temperature-programmed
reaction of a saturation exposure of ethylene glycol on Mo-
(110) (Figure 1B(v)).18,19 The evolution of ethylene in both
states is limited by the rate of the reaction, not desorption, since
ethylene desorbs below 300 K when adsorbed on both clean
and O-covered Mo(110).20 Ethylene production is a sensitive
function of the initial ethylene glycol coverage; at the lowest
exposure studied (0.05 of saturation), nonselective decomposi-
tion is the sole reaction pathway, and no ethylene evolution is
detected. However, as the coverage is increased to 0.11 of
saturation, ethylene is produced in a single state with a peak
maximum at ∼365 K (Figure 1B(i)). The two distinct ethylene
states are evident at ∼0.5 of saturation and above (Figure
1B(iv,v)). A small high-temperature ethylene tail at ∼510 K
also grows in at the highest ethylene glycol exposures (Figure
1B(iv,v)). Importantly, data obtained using HOCD2CD2OH
demonstrate that there is no reversible C-H bond scission along
the path to ethylene formation, since C2D4 is formed exclusively;
no C2D3H, C2D2H2, C2DH3, or C2H4 are detected (data not
shown).21 These data suggest that ethylene is evolved directly
into the gas phase at ∼350 K following facile C-O bond
scission. Further evidence against ethylene interaction with the
surface is provided by previous studies of the reactions of
ethylene on clean and oxygen-precovered Mo(110), which
demonstrate that any residual ethylene on these surfaces above
250 K undergoes decomposition.20 We estimate that at satura-
tion coverage ethylene is eliminated with ∼85% selectivity, with
the remaining ∼15% of the ethylene glycol nonselectively
decomposing, based on the ratio of the high-temperature CO
produced during temperature-programmed reaction of ethylene
glycol and methanol.22
Dihydrogen is produced in two states, the primary one with
a peak temperature of 390 K (â2), with a pronounced shoulder
at 500 K (â1), for temperature-programmed reaction of a satur-
ation exposure of ethylene glycol on Mo(110) (Figure 1C(v)).
There is almost no change in the peak temperature of the â2
state over the entire coverage range studied (Figure 1C(i-v)),
other than a slight broadening on the low-temperature side at
the highest ethylene glycol coverage. We assign the â2
dihydrogen peak to the recombination of surface hydrogen
originating from the hydroxyl hydrogens of the parent glycol
molecule, by reference to previous studies of hydrogen as well
as alkoxides on Mo(110), which indicate that this â2 hydrogen
peak is limited by desorption, not reaction kinetics.7,10,23 This
X-ray Photoelectron Spectroscopy. X-ray photoelectron
spectroscopy indicates that there is a single surface intermediate
with an intact C-O bond after annealing low coverages of
ethylene glycol (θ < 0.4θsat) to 300 K.24 The X-ray photo-
electron peak with an O(1s) binding energy of 532.6 eV and
resolution-limited width is assigned to oxygen in an alkoxide
group by comparison with X-ray photoelectron spectra of other
alcohols on Mo(110) (Figure 2A(i)).7,10,23 Importantly, the
existence of only a single C-O bond environment and the
complete absence of an X-ray photoelectron feature due to intact
C-O-H groups near 533.7 eV (see below) is strongly sugges-
tive of the existence of a single bidentate-type intermediate at
these coverages. In addition, a small amount of decomposition
also occurs at low coverage, based on the peak at 530.8 eV,
which is ascribed to atomic oxygen by comparison to X-ray
photoelectron spectra of oxygen adsorbed on Mo(110). Notably,
no atomic oxygen is detected in X-ray photoelectron spectra
recorded at 100 K (data not shown), showing the absence of
any C-O bond scission at these lower temperatures. In
agreement with the data for the O(1s) region, there is a single
carbon environment at 300 K, signified by a single C(1s) peak
at 286.7 eV, which is similar to that measured for other
alkoxides on Mo(110) (Figure 2B(i)).7,10,23 The presence of
atomic oxygen and the concomitant lack of atomic carbon
detected in the X-ray photoelectron experiments at 300 K are
consistent with temperature-programmed reaction spectra in
which ethylene evolution begins below 300 K at these coverages
(Figure 1B(iii)).
As the coverage of ethylene glycol on Mo(110) increases
beyond 0.4θsat, both the O(1s) and C(1s) alkoxide peaks develop
higher-binding energy shoulders after annealing to 300 K,
suggesting the existence of a second surface species (Figures
2A(ii) and 2B(ii)). As the coverage is increased to saturation
the new feature in the O(1s) region, centered at ∼533.7 eV,
has approximately 0.2 times the integrated area of the 532.6-
eV alkoxide peak (Figure 2A(ii)).25 Similarly, the C(1s)
spectrum becomes highly asymmetric, displaying a shoulder
centered at ∼287.9 eV. While the intensity of this peak is
sensitive to the fit, the data are well fit when the shoulder
accounts for ∼0.2 times that of the main alkoxide peak at 286.7
eV, in agreement with the O(1s) region (Figure 2B(ii)). By
comparison with the X-ray photoelectron spectra for multilayers
of ethylene glycol (data not shown), these new features are
assigned to the oxygen and carbon of an intact C-O-H group.
(18) Saturation coverage, θsat, is defined as the exposure beyond which
the reaction product yields do not increase any further and the multilayer
desorption peak at 195 K increases without bound.
(19) Identification of products was verified by comparison of cracking
patterns observed during reaction with that obtained using a genuine sample
of ethylene. Importantly, the entire m/e 28 signal during temperature-
programmed reaction of ethylene glycol was accounted for by the observed
ratio of m/e 28:m/e 27:m/e 26 in the genuine ethylene sample, confirming
that no CO was produced by the reaction of ethylene glycol on Mo(110).
(20) Serafin, J. G.; Friend, C. M. J. Am. Chem. Soc. 1989, 111, 6019.
(21) If reversible C-H bond scission occurred, recombination of the
alcoholic hydrogens with the deuterated alkyl fragments would take place,
incorporating H into some of the ethylene evolved.
(22) The selectivity estimated was based on the saturation coverage of
methanol on Mo(110), previously found to be 0.25 ML.10 Over 95% of the
methanol undergoes nonselective decomposition to form gaseous dihydrogen
along with adsorbed carbon and oxygen on Mo(110). The carbon and oxygen
subsequently recombine to form CO ∼1200 K. Comparison of the integrated
CO intensity for ethylene glycol and methanol on Mo(110) is used to
estimate the amount of nonselective reaction.
(24) Approximately the same heating rate was used to anneal the surface
prior to recording X-ray photoelectron and infrared spectra as was used for
the temperature-programmed reaction experiments.
(25) The intensity due to the new feature is distinct from the intensity
always seen as a tail to the higher-binding energy side of the O(1s) signal
due to final state effects.26 Intensity due to the tail is estimated by comparison
to lower-coverage spectra, where no shoulder due to a second species
appears.
(23) Uvdal, P.; Wiegand, B. C.; Serafin, J. G.; Friend, C. M. J. Chem.
Phys. 1992, 97(11), 8727.

Differential ReactiVity of Mono- and Dialkoxides
J. Am. Chem. Soc., Vol. 118, No. 16, 1996 3899
Table 1. Vibrational Assignments for Ethylene Glycol
A
liquid phase
-OCH2CH2O- on -OCH2CH2OH on assigned
HOCH2CH2OH27 Mo(110)a (cm-1
)
Mo(110)a (cm-1
)
mode
3150-3400
2935
2875
1459
1087
NOb
2913
2857
1455
1031
896
3440, 3220
2931
2840
(1455)
1080
910
ν(OH)
νas(CH)
νs(CH)
δ(CH2)
νs(CO)
ν(CC)
864
a Frequencies for surface species correspond to a saturated monolayer
of ethylene glycol at 300 K. b NO ) not observed.
X-ray photoelectron studies of water on Mo(110).26 Moreover,
the formation of surface hydroxyl groups would not account
for the appearance of the new feature in the C(1s) region of the
spectrum. Rather, these peaks suggest the existence of a second
surface species with an intact (molecular) hydroxyl group at
these higher ethylene glycol coverages, a conclusion cor-
roborated by our infrared studies (see below).
Analysis of the thermal evolution of the X-ray photoelectron
spectra of a saturation exposure of ethylene glycol on Mo(110)
indicates that the two alkoxide species are observed by 250 K
and one alkoxide persists to above 400 K (data not shown).
Upon annealing the surface to 400 K, by which temperature
ethylene evolution is almost complete, a dramatic increase in
the intensity of the atomic oxygen peak is observed, together
with a shift in its binding energy to 530.5 eV, at the expense of
the intensity in the other two O(1s) peaks. Importantly, the
533.7 eV peak attributed to the C-O-H moiety disappears
completely, while the alkoxide peak at 532.6 eV persists,
indicating that a single species with an intact C-O bond is
present on the surface at 400 K, as was observed in the spectra
obtained by annealing a low coverage of ethylene glycol to 300
K. In addition, a peak due to atomic carbon at 283.6 eV is
observed together with the alkoxide C(1s) peak at 286.6 eV.
Finally, only atomic oxygen (530.5 eV) and atomic carbon
(283.5 eV) remain on the surface after annealing the saturated
overlayer to 750 K (data not shown). Comparison of the total
C(1s) intensity after annealing to 750 K, when all the ethylene
glycol has reacted, to the intensity at 250 K (at which
temperature the multilayer has desorbed to leave a saturation
coverage of ethylene glycol), yields an estimate of 17 ( 5%
for nonselective decomposition, in good agreement with the
estimate based on temperature-programmed reaction data.
Vibrational Spectroscopies. Infrared spectra indicate that
a single intermediate, ethylenedioxy (-OCH2CH2O-), is
formed below 300 K on Mo(110) for exposures of ethylene
glycol corresponding to less than 0.6 of saturation (Figures
3A(i) and 3B(i)). The complete absence at these exposures of
the O-H stretching mode at ∼3300-3600 cm-1, the dominant
feature in the infrared spectrum of multilayers of ethylene glycol
at 100 K (data not shown), indicates that O-H bond scission
has occurred by 300 K. There is essentially no change in the
infrared spectra at these exposures upon annealing to 350 K, a
temperature at which recombinant dihydrogen evolution from
the alcoholic protons has commenced, supporting the assertion
from the X-ray photoelectron data that the O-H bonds break
below 300 K. The infrared spectrum obtained after annealing
low ethylene glycol exposures on Mo(110) to 300 K is assigned
in terms of an intact -O-CH2-CH2-O- skeleton (Table 1).
Two narrow, symmetric absorption bands are observed at 881
and 1008 cm-1 in the 800-1500-cm-1 region of the infrared
spec trum of the lowest exposure of ethylene glycol (Figure
3B(i)). The 881-cm-1 band is assigned to a mode which has
predominantly C-C character, by reference to the known
B
Figure 2. X-ray photoelectron spectra obtained for (A) O(1s) and (B)
C(1s) regions after heating ethylene glycol to 300 K at ∼5 K s-1
.
Ethylene glycol was adsorbed at ∼100 K, and the exposures shown
are (i) 0.06, (ii) 0.08, and (iii) 0.14 ML (saturation). Noise below the
baseline in all spectra is cut off by the Perkin Elmer curve-fitting
algorithm.
Importantly, the new O(1s) feature is not consistent with the
appearance of surface hydroxyl groups, by reference to previous
(26) Serafin, J. G.; Friend, C. M. J. Am. Chem. Soc. 1989, 111, 4233.

3900 J. Am. Chem. Soc., Vol. 118, No. 16, 1996
Queeney et al.
associated with this mode, by analogy with the parent molecule27
and by reference to a recent study in which it was demonstrated
that a mode at 875-880 cm-1 in the (qualitatively similar)
ethoxide species (CH3CH2O-) adsorbed on Mo(110) consists
of approximately equal amounts of C-C and C-O motion.28
Importantly, given the above normal mode description, the
adsorbate symmetry cannot be trivially deduced based on this
mode alone, since it is surface-dipole allowed for all C-C bond
orientations.
A
The feature at 1008 cm-1 in the infrared spectrum of the
lowest exposure of ethylene glycol is assigned (predominantly)
to the symmetric C-O stretching mode of an ethylenedioxy
intermediate (Figure 3B(i)). There are two possible assignments
for the 1008-cm-1 feature: either the symmetric (in-phase) or
the asymmetric (out-of-phase) C-O stretching modes, observed
at 1087 and 1038 cm-1, respectively, in the condensed glycol.
On first inspection, one might assign this band to νa(C-O), by
direct analogy with the condensed-phase molecule. However,
in the aforementioned study of ethoxide adsorbed on Mo(110),
a pronounced perturbation of the C-O stretching mode was
observed upon formation of the alkoxide. Specifically, the mode
assigned predominantly to ν(C-O) at 1068 cm-1 in the parent
molecule was observed to shift down to 1020 cm-1 for CH3-
CH2O-Mo. Furthermore, assignment of the 1008-cm-1 feature
to the out-of-phase C-O stretch is inconsistent with the rest of
the infrared data, since the exclusive observation of this mode
would be indicative of a geometry wherein the C-C bond was
normal to the surface, which is inconsistent with the bidentate
ethylenedioxy geometry suggested by analysis of the O-H
stretching region and supported by X-ray photoelectron data
(Figure 4a). Conversely, the in-phase C-O stretching mode
would predominate when the C-C bond axis is parallel to the
Mo(110) plane, since the asymmetric C-O mode would be
effectively screened out by the metal image dipole.
B
Further support for these assignments is provided by analysis
of the C-H stretching region of the infrared spectra (Figure
3A(i)). The features at 2869 and 2942 cm-1 are assigned to
the in-phase symmetric and out-of-phase asymmetric C-H
stretching modes, respectively, given the exact correspondence
with the condensed-phase data (Table 1). Importantly, the
symmetric C-H mode has a dipole moment that is perpendicular
to the C-C bond axis, so that the observation of this mode is
consistent with a predominantly parallel geometry of the
dialkoxy intermediate, as suggested above. In contrast, the
asymmetric C-H mode is infrared inactive for C2V adsorbate
symmetry (the highest possible adsorbate symmetry group with
the C-C axis parallel to the surface plane), because the dipole
moment is zero when the -O-C-C-O- plane is normal to
the metal surface. However, if no mirror plane exists in the
adsorbate, i.e. the symmetry is C2 or lower, this mode becomes
surface dipole allowed (Figure 4b). Interestingly, the above
description is consistent with the known geometry of ethylene
glycol in the condensed phase.27
As the ethylene glycol exposure is increased to 0.6 of
saturation, there is little change in the infrared spectrum obtained
after annealing to 300 K,24 other than a uniform increase in the
intensity of all features observed at lower exposures, together
with a slight upshift in the frequency of νs(C-O) and ν(C-C)
and concomitant decrease in the frequencies of the C-H
stretching modes (Figures 3A(ii) and 3B(ii)). All infrared
features persist after heating to 300 K for all ethylene glycol
coverages above 0.6 of saturation (Figures 3A(iii) and 3B(iii),
Figure 3. Infrared spectra of ethylene glycol on Mo(110) obtained
after annealing to 300 K for (A) the C-H and O-H stretching region
and (B) the C-O and C-C stretching region. The coverages correspond
to (i) 0.04, (ii) 0.08, and (iii) 0.14 ML (saturation). The inset labeled
(iv) is the C-H and O-H stretching region of the electron energy loss
spectrum of a surface prepared identically to that in parts a and b(iii).
infrared spectrum of ethylene glycol (Table 1).27 However, we
suggest that a substantial component of C-O motion is also
(27) Sawodny, W.; Niedenzu, K.; Dawson, J. W. Spectrochim. Acta 1967,
23A, 799.
(28) Uvdal, P.; MacKerell, A. D., Jr.; Wiegand, B. C. J. Electron
Spectrosc. Rel. Phenom. 1993, 64/65, 193.

Differential ReactiVity of Mono- and Dialkoxides
J. Am. Chem. Soc., Vol. 118, No. 16, 1996 3901
data indicate that the OH groups belong to a molecular species.
The possibility that the ν(OH) features belong to a species with
two intact OH groups can be eliminated based on the persistence
of these infrared peaks up to 300 K. Intact ethylene glycol
desorbs below 200 K, which is consistent with the low
desorption temperature observed for species containing intact
OH bonds on Mo(110), including molecular water26 as well as
other intact alcohols.23,29 Furthermore, assignment of the ν-
(OH) features to a monoalkoxide species is consistent with the
presence of both hydrogen-bonded and non-interacting OH
groups, since the length and lack of rigidity of the monoalkoxide
surface moiety would allow some OH groups on adjacent
monodentate species to interact with each other, giving rise to
the lower-frequency (hydrogen-bonded) ν(OH) peak. The fact
that some intact OH groups do not interact with neighboring
OH groups, as indicated by the presence of a higher-frequency
ν(OH) peak, is supported by the relatively low concentration
of these species as compared to the bidentate species.
Conclusive support for the identification of the ν(OH) modes
is given by the electron energy loss spectrum in Figure 3A(iv).
The high-frequency region of this spectrum clearly demonstrates
the existence of two peaks in the ν(OH) region, one at 3572
cm-1 and one at 3298 cm-1. The increased intensity of these
peaks relative to the ν(CH) peak (one peak at 2930 cm-1), in
comparison with the infrared data, may be attributed to
contributions from both impact and resonance scattering, which
may add intensity to such high-frequency adsorbate modes.
Indeed, the fact that the electron energy loss peaks appear at
slightly different frequencies than the corresponding infrared
peaks, though still in regions corresponding to unassociated and
hydrogen-bonded molecular O-H bonds, is also consistent with
the enhancement of a subset of the overlayer hydroxyl modes
by such non-dipolar scattering mechanisms.30
Figure 4. The molecular motions used to deduce the orientation of
adsorbates derived from ethylene glycol on Mo(110) at 300 K. The
following comparisons are illustrated: (a) the in-phase (observed) and
out-of-phase (not observed) C-O stretches in the bidentate species,
which allows for assignment of the observed νs(CO) peak consistent
with a C-C bond axis parallel to the surface; (b) the out-of-phase
asymmetric C-H stretch in the bidentate species for a planar O-C-
C-O skeleton with eclipsed hydrogens (infrared inactive) and a gauche
conformation (infrared active), showing that the appearance of this mode
in the infrared spectrum indicates a gauche conformation of the bidentate
species; and (c) the symmetric C-H stretch in a monodentate species
with its C-C axis normal to the surface (infrared inactive) and inclined
away from the surface normal (infrared active), demonstrating that the
appearance of this peak in the monodentate infrared spectrum indicates
a tilted C-C bond axis.
The appearance of a new νs(CH) mode in the infrared
spectrum, attributable to the monodentate species, requires that
the C-C bond of this moiety be inclined with respect to the
surface normal in order for this mode to be experimentally
observed (Figure 4c). Importantly, both the in-phase and out-
of-phase components of the C-O stretching modes of such a
(tilted) monodentate species are surface-dipole allowed and
should therefore in principle be observable. However, it is
widely recognized that strong dipole-dipole coupling can occur
between co-adsorbates, resulting in an upshift in the frequency
of the absorbance band (for a predominantly perpendicular
dipole).31 In addition, there may be appreciable intensity bor-
rowing by high-frequency oscillators from the lower frequency
sites in an inhomogeneous overlayer, resulting in a further
upshift in the absorbance maximum and a decrease in the band
width, since essentially all the intensity appears in the highest
frequency modes in a given spectral region. Indeed, this is
precisely what is observed for all alkoxides studied to date on
Mo(110). For example, on a Mo(110) surface populated with
both CD3O- and CH3O- species, the two ν(CO) peaks are
indistinguishable at all but the lowest coverages, despite being
separated by more than 30 cm-1 in the condensed-phase parent
molecules.32 By analogy, we propose that similar dipole-dipole
coupling occurs at ethylene glycol coverages approaching
Table 1). In addition, a new infrared absorption band is apparent
at 1455 cm-1, which is assigned to the in-phase methylene
scissors mode that is observed at 1459 cm-1 in condensed
ethylene glycol (Table 1). Once again, the observation of this
feature is indicative of a predominantly parallel orientation of
the C-C bond with respect to the surface plane.
While the infrared peaks corresponding to the bidentate
species remain relatively unperturbed as the ethylene glycol
coverage is increased, at saturation coverage a second species
is detected in the infrared spectrum (Figures 3A(iii) and 3B(iii),
Table 1). The new features are assigned to a monodentate,
acyclic 2-hydroxyethoxide moiety (-OCH2CH2OH) in which
the C-C bond is inclined with respect to the surface plane.
The “chemical similarity” between the bidentate species, the
sole intermediate at low coverage, and this second species is
suggested by the observation of a second set of modes at high
coverage, shifted by ∼10-20 cm-1 in frequency from the
corresponding bidentate modes. However, an important dif-
ference is apparent on inspection of the 3100-3700-cm-1 region
of the infrared spectrum at saturation coverage; two bands are
observed at 3220 and 3436 cm-1 that are attributed to the O-H
stretching vibrations of associated and (largely) unassociated
OH moieties, respectively, by reference to the wealth of
condensed-phase literature of such functional groups. In
general, such ν(OH) features may arise either from surface
hydroxyl groups or from intact OH groups within molecular
(e.g. alcoholic) species. However, the C(1s) X-ray photoelectron
(29) Serafin, J. G. Ph.D. Thesis, Harvard University, 1989.
(30) Slight differences in the base temperatures used for initial adsorption
and/or spectral collection in infrared and electron energy loss experiments
might also contribute to the differences in the O-H stretch frequencies in
the two types of experiments. The O-H stretch frequency is known to be
sensitive to chemical environment; thus, small differences in surface
concentration or temperature might lead to a shift.
(31) Persson, B. N. J.; Ryberg, R. Phys. ReV. B 1981, 24(12), 6954.
(32) Uvdal, P.; Weldon, M. K.; Friend, C. M. Phys. ReV. B In press.

3902 J. Am. Chem. Soc., Vol. 118, No. 16, 1996
Queeney et al.
saturation, so that only a single intense ν(CO) band is resolved
for the monodentate species despite the tilted C-C axis.
Complementary temperature-programmed reaction studies of
isotopically mixed layers allow investigation of the relative
reactivities of the mono- and bidentate species. Infrared studies
of ethylene glycol-d0 were used to establish the conditions for
saturation of the bidentate population. Temperature-pro-
grammed reaction experiments were then performed on an
overlayer in which the bidentate state was selectively populated
with ethylene glycol-d4 and the monodentate state with ethylene
glycol-d0 using the conditions established by the infrared
experiments (data not shown).33 Both C2D4 and C2H4 were
produced in both the 350 and 390 K states previously identified
for the isotopically pure sample. Therefore, both the mono-
and bidentate surface species contribute to both states of ethylene
evolution. However, the ratio for low- to high-temperature
ethylene from the bidentate (deuterated) intermediate is ∼1.3:
1, as compared to a ratio of ∼1.1:1 for the monodentate
(undeuterated) intermediate, indicating a slight preference for
reaction of initially bidentate moieties in the early stages of
ethylene evolution.
A
Additional information on the mechanism of ethylene produc-
tion is given by the temperature evolution of the infrared spectra,
which demonstrates conversion of the monodentate intermediate
to bidentate as the reaction of ethylene glycol on Mo(110)
proceeds. Specifically, as the temperature is raised to 325 K
in the midst of ethylene evolution, the intensity of the mono-
dentate modes is severely attenuated, while the intensity of the
bidentate modes remains essentially constant (Figures 5A(ii)
and 5B(ii)). In conjunction with the above temperature-
programmed reaction data, which demonstrate that both the
mono- and bidentate moieties give rise to similar ethylene
formation features, the infrared data suggest that the monoden-
tate species reacts via a bidentate intermediate to produce
ethylene. The bidentate species is thus repopulated over the
early part of the reaction. Furthermore, as the annealing
temperature is raised to 350 K, only the bidentate species is
apparent in the infrared spectrum, despite the observation of
continued ethylene evolution from (initially) monodentate
species during temperature-programmed reaction (Figures
5A(iii) and 5B(iii)). These data are consistent with the
conversion of mono- to bidentate, concomitant with the progress
of the reaction.
B
Discussion
Ethylene glycol predominantly reacts on Mo(110) to evolve
gas-phase ethylene; the selectivity for this reaction over
nonselective decomposition increases as a function of ethylene
glycol coverage, reaching ∼85% at saturation. Our X-ray
photoelectron and infrared studies show that, at low ethylene
glycol coverage, the sole surface intermediate for ethylene
production is a bidentate, dialkoxide species adsorbed with C2
(or lower) symmetry. At coverages greater than 0.4 of satura-
tion, a second surface intermediate is formed, identified as a
monoalkoxide species with one intact O-H bond. The two
species coexist at 300 K such that the monoalkoxide:dialkoxide
ratio is ∼2:3 at saturation coverage.34 As the surface temper-
ature is raised above 300 K, the bidentate species begins to react,
(33) Infrared spectroscopy was used to calibrate the exact dose required
just to saturate the bidentate state, and this dose was used to populate that
state fully with only HOCD2CD2OH. Sufficient nondeuterated ethylene
glycol was then dosed to saturate the monodentate state. Infrared data for
the mixed overlayer were not collected.
(34) This ratio was calculated from the relative intensities of the 533.7-
and the 532.6-eV X-ray photoelectron peaks in Figure 2A(iii), taking into
account the contribution of the monoalkoxide species to the intensity in
the 532.6-eV alkoxide peak, assumed to be equal to the total intensity of
the 533.7-eV peak.
Figure 5. Infrared spectra of the temperature evolution of a saturation
coverage of ethylene glycol on Mo(110) for (A) the C-H and O-H
stretching region and (B) the C-O and C-C stretching region. The
spectra shown were recorded at ∼120 K after annealing a multilayer
dose to (i) 300, (ii) 325, and (iii) 350 K. Peaks labeled “M” and “B”
are those attributed to the monodentate and the bidentate species,
respectively; their frequencies are listed in Table 1.

Differential ReactiVity of Mono- and Dialkoxides
J. Am. Chem. Soc., Vol. 118, No. 16, 1996 3903
model by demonstrating the depletion of the monodentate state
over the course of the reaction to leave only one alkoxide
intermediate, spectroscopically indistinguishable from the bi-
dentate species formed at low coverages.
Since conversion of the monodentate to a bidentate species
is favored over all other reaction pathways involving direct C-O
bond scission, C-O bond scission must be more facile for the
dialkoxide than for the monoalkoxide. Simple C-O bond
scission in the monoalkoxide intermediate would yield the
•
transient radical product, H2C-CH2OH. Such a species is
proposed to rearrange to form acetaldehyde on Rh(111);5 the
lack of any production of acetaldehyde or its subsequent
decomposition products in our study eliminates this as a pathway
for monodentate ethylene glycol reaction on Mo(110). Alter-
natively, the hydroxyethyl radical could undergo a reaction
analogous to the formation of ethylene from adsorbed ethoxide,35
eliminating one hydrogen to form CH2dCHOH. However, in
addition to the intrinsic instability of its vinyl alcohol product,
this pathway is also disfavored due to the higher temperature
of the analogous reaction for ethoxide on Mo(110) (400 K)
relative to the temperature of ethylene elimination from ethylene
glycol. Likewise, hydrogenation of the alkoxide carbon of the
2-hydroxyethoxide to form gas-phase ethanol does not occur,
consistent with the absence of hydrogenation to form ethane
from ethoxide on Mo(110).35 Hence, none of these reactions
is more favorable for the monodentate species than formation
of the bidentate alkoxide, with subsequent double C-O bond
scission to eliminate ethylene.
In fact, since the monodentate ethylene glycol intermediate
does not undergo any direct C-O bond scission, no direct
comparison can be made between the temperatures for C-O
bond scission in the two species. However, a useful comparison
can be made between the dialkoxide ethylene glycol intermediate
and the monoalkoxide methoxide and ethoxide species on Mo-
(110); decomposition of the ethylene dialkoxide species by C-O
bond scission to form ethylene begins below 300 K and peaks
at ∼350 K, while C-O bond scission in both methoxide and
ethoxide on Mo(110) does not occur until ∼400 K.10 The
enhanced reactivity of the dialkoxide species suggests that
bonding via both oxygens may lead to additional C-O bond
weakening or less structural rearrangement in the transition state
for ethylene elimination. Furthermore, formation of a C-C π
bond adds to the thermodynamic favorability of ethylene
elimination in the dialkoxide pathway. Electronic structure
calculations comparing mono- and bidentate species are needed
to evaluate the contributions of these structural and bonding
effects to the relative ease of deoxygenation in mono- and
dialkoxides.
The reactivity of ethylene glycol is distinctly different on this
surface than on all other surfaces studied, consistent with facile
C-O bond scission of other alcohols on Mo(110). In all
previous studies of ethylene glycol on transition metal surfaces,
the C-O bonds were retained, and the bidentate alkoxide was
found to have comparable or decreased reactivity compared to
similar monoalkoxide species. However, in both the surface
and organometallic complexes of ethylene dialkoxide studied
previously, the increased reactivity of the monodentate over the
bidentate moiety refers specifically to â-hydrogen elimination.
For example, on Rh(111) the temperature for reaction of the
proposed bidentate ethylene glycol intermediate is approximately
the same as that for ethanol. Decarbonylation of the dialkoxide
has begun by 260 K and decomposition is complete by just
over 300 K, with the exact sequence of decarbonylation/
dehydrogenation unresolved.5 In comparison, ethanol decar-
Figure 6. Proposed reaction scheme for ethylene glycol on Mo(110).
which we propose reduces steric crowding on the surface and
thus allows for conversion of the monodentate to the bidentate
(Figure 6). Importantly, this study is the first to utilize the high
resolution of infrared spectroscopy to identify both the mono-
and bidentate species on the same surface.
The presence of two distinct surface species at saturation
coverage of ethylene glycol is indicated by both X-ray photo-
electron and infrared reflection absorption spectroscopies. At
low ethylene glycol coverages the O(1s) and C(1s) regions of
the X-ray photoelectron spectra are consistent with the presence
of a surface dialkoxide. The low-coverage infrared spectrum
can also be assigned completely to a single bidentate species.
As the exposure of ethylene glycol is increased, higher-binding
energy shoulders appear in both the O(1s) and C(1s) X-ray
photoelectron regions and are assigned to the oxygen and carbon
in intact C-O-H groups. The existence of an intact OH group
in a second intermediate at higher coverages of ethylene glycol
is verified by the appearance of two peaks in the ν(OH) region
of the infrared and electron energy loss spectra and by a second
set of peaks in the ν(CH), ν(CO), and ν(CC) regions of the
infrared spectrum. These and other infrared and X-ray photo-
electron spectroscopy features described above identify the two
surface species present at saturation coverage of ethylene glycol
as (1) a bidentate moiety (-OCH2CH2O-) of C2 (or lower)
symmetry with its C-C bond axis parallel to the surface and
(2) a monodentate species (-OCH2CH2OH) with its C-C bond
axis inclined away from the surface normal. Although prefer-
ential population of the bidentate species at low ethylene glycol
coverages is expected on Mo(110), based on the availability of
surface sites and the facile reactivity of alcoholic O-H bonds,
population of the monoalkoxide species occurs as the paired
surface sites required for bidentate bonding are depleted at
higher ethylene glycol coverages.
The conversion of the monodentate to the bidentate form prior
to ethylene elimination is clearly indicated by analysis of both
the reaction products formed and the evolution of surface
intermediates as the reaction progresses. Since no oxygen-
containing products are evolved, the intact OH group in the
monodentate alkoxide must interact with the surface to undergo
a second O-H bond scission before evolving ethylene. Our
temperature-programmed reaction studies of coadsorbed -OCD2-
CD2O- and -OCH2CH2OH clearly show that ethylene is
eliminated with virtually the same temperature profile from both
surface intermediates. This observation suggests that the
monodentate species reacts through the bidentate species.
Infrared and X-ray photoelectron data further corroborate this
(35) Chen, D. A.; Friend, C. M. Unpublished results.

3904 J. Am. Chem. Soc., Vol. 118, No. 16, 1996
Queeney et al.
bonylation and dehydrogenation is complete by 270 K on
Rh(111), though no alkoxide intermediate is observed in this
case.36 On oxygen-precovered Ag(110), entropic considerations
were invoked to rationalize the higher temperature, and thus
lower reactivity, of glyoxal formation from ethylene glycol
reaction relative to that for methoxy decomposition on the same
surface.2 Similarly, ethoxide undergoes C-H scission to form
acetaldehyde at significantly lower temperatures than that
observed for glyoxal formation from ethylene glycol on Ag-
(110).37 Likewise, the methoxide ligand in the organometallic
compound (DPPE)Pt(OCH3)2 loses a â-hydrogen to form
formaldehyde, whereas the analogous (DPPE)Pt(OCH2CH2O)
bidentate ethylene glycolate complex undergoes no such reac-
tion.38 The bidentate ligand is apparently less susceptible than
its monodentate counterpart to â-hydrogen elimination, since
the rigidity of the bidentate ligand hinders the access of its
â-hydrogens to the metal. However, on Mo(110) C-O bond
scission generally dominates over â-hydrogen elimination, as
in the case of 1-propanol, in which competition between C-O
bond breaking to form gaseous propene and â-hydrogen
elimination to initiate nonselective decomposition results in a
3:2 branching ratio in favor of propene elimination. Indeed,
since no C-H bonds are broken in the ethylene elimination
reaction of ethylene glycol on Mo(110), the inhibition of
â-dehydrogenation in the dialkoxide clearly increases the
selectivity for ethylene elimination over nonselective decom-
position involving â-C-H cleavage.7 The apparent lack of
â-dehydrogenation in the monoalkoxide is likely due to the
favorability of bidentate formation over direct monoalkoxide
reaction pathways, as discussed above.
For example, methoxide on clean Mo(110) nonselectively
decomposes at ∼400 K, while on the oxygen-covered surface
the majority of the reaction, to produce gaseous methyl radicals,
occurs at 500 K.39 A similar increase in reaction temperature
is observed for ethylene glycol on Mo(110) as oxygen is
deposited during the course of temperature-programmed reac-
tion. However, at ethylene glycol coverages near saturation,
the reactivity of the bidentate species increases, as evidenced
by the appearance and growth of the lower temperature (∼350
K) ethylene peak. This increase in reactivity is attributed to
intermolecular interactions, perhaps involving dipole-dipole
repulsion, which lead to a higher propensity toward C-O bond
scission in the bidentate species. Notably, ethylene is formed
at an intermediate temperature, ∼365 K, at the lowest coverage
(∼0.11 of saturation), where intermolecular and oxygen buildup
are both less significant, illustrating the importance of these
effects.
Conclusions
Ethylene glycol reacts on Mo(110) to form gas-phase ethylene
with ∼85% selectivity. At low coverages, ethylene glycol forms
two Mo-O bonds to adsorb in a bidentate (-OCH2CH2O-)
configuration below 300 K. At saturation coverage, two distinct
intermediates are present on the surface, both the bidentate and
a monodentate (-OCH2CH2OH) species. In contrast to previ-
ous studies of ethylene glycol reactivity on other metal surfaces,
we find that the reactivity of the bidentate species is not less
than that of the monodentate species. In fact, due to both the
stability of the bidentate intermediate and the greater predilection
for ethylene formation over potential monodentate reaction
pathways, the bidentate species is identified as the more reactive
species on a surface composed of both intermediates.
Since the two surface intermediates are shown to proceed
through the same reaction pathway, the two distinct states for
ethylene evolution at higher coverages are proposed to arise
from competition between steric and oxygen-induced effects
on the surface. The higher temperature of the ethylene evolution
peak at ∼390 K, with a peak maximum ∼25 K higher than the
single ethylene peak at the lowest coverage, is attributed to
inhibition of reaction by the oxygen deposited on the surface
in conjunction with initial ethylene elimination. Previous studies
have shown that alcohols on oxygen-covered Mo(110) react at
higher temperatures than they do on the clean surface.26,29,39
Acknowledgment. The group at Harvard gratefully ac-
knowledges the support of the Department of Energy under grant
DE-FG02-84-ER13289, as well as support for K. T. Queeney
from a National Science Foundation Graduate Fellowship. C.
R. Arumainayagam gratefully acknowledges partial support from
a Cottrell College Science Award of Research Corporation and
a Brachman Hoffman fellowship from Wellesley College.
(36) Papageorgopoulos, D. C.; Ge, Q.; King, D. A. J. Phys. Chem. 1995,
99, 17645.
(37) Wachs, I. E.; Madix, R. J. Appl. Surf. Sci. 1978, 1, 303.
(38) Bryndza, H. E.; Calabrese, J. C.; Marsi, M.; Roe, D. C.; Tam, W.;
Bercaw, J. E. J. Am. Chem. Soc. 1986, 108, 4805.
JA953987R
(39) Serafin, J. G.; Friend, C. M. J. Am. Chem. Soc. 1989, 111, 8967.