Angewandte
Chemie
amount of butanol produced as a secondary product would be
below the detection limit of our apparatus. This data gives
additional support to the conclusion that butanol is formed by
a concerted process. Further catalytic tests were performed
using the primary products (crotyl alcohol and butyralde-
hyde) as reactants and are presented in the Supporting
Information. These tests help to further elucidate the
mechanism we propose.
increases the number of O-vacancy sites. The nonresonant
component of the SFG spectra is largely the result of free-
carrier motion at the surface of a sample. As more O vacan-
cies are formed on the TiO2 surface the free-carrier concen-
tration increases, as evidenced by increased surface conduc-
tivity,[25] and subsequently the nonresonant component of the
sum-frequency signal is larger. This increased signal serves as
a marker for the chemical and electronic structure of the
oxide surface and supports our hypothesis that O vacancies
are the active catalytic sites. The increased nonresonant
contribution can serve to elevate the baseline level in SFG
spectra and also enhance the intensity of resonant features
because of the relationship between the nonresonant and
resonant components of c(2) and the SFG intensity. The
resonant features observed are out of
To better understand the mechanism by which TiO2
=
enhances the rate of C O bond hydrogenation SFG vibra-
tional spectroscopy is employed to determine surface reaction
intermediates. Figure 3 shows the SFG spectra for Pt/SiO2 and
Pt/TiO2 under hydrogenation reaction conditions for croton-
aldehyde. The proposed adsorption geometries and reaction
phase with this nonresonant back-
ground and thus appear as negative
peaks.
The reported spectrum in Fig-
ure 3B is remarkably similar to the
same region in the gas-phase IR
spectrum for crotyl alcohol, and is
indicative of a crotyl-oxy intermedi-
ate bound to the TiO2 surface through
the aldehyde oxygen atom. The fea-
tures at 2880 and 2930 cmÀ1 are
assigned to the CH3 symmetric and
asymmetric stretching modes, respec-
tively.[26] The stretch at 2965 cmÀ1 is
assigned to the CH2 asymmetric
stretching mode of the O-bound
carbon atom.[27] Comparison of the
Pt/TiO2 spectrum with that of TiO2
alone (see Figure 3 in the Supporting
Information) shows that the aldehyde
stretch disappears and a CH2 stretch
appears in the presence of Pt, thus
indicating hydrogenation of the car-
Figure 3. SFG spectra for A) Pt/SiO2 and B) Pt/TiO2 under reaction conditions of 1 Torr crotonalde-
hyde, 100 Torr hydrogen, and 669 Torr argon at a catalyst temperature of 1208C. Assignment of
these spectra, explained in the text, yields the accompanying surface intermediates.
intermediates are also shown. The Pt/SiO2 spectrum shows
two features at 2860 and 2915 cmÀ1, which are consistent with
previous work from Kliewer et al. on Pt single-crystal
surfaces.[19] The peak at 2915 cmÀ1 is assigned to a CH3 stretch
and the peak at 2860 cmÀ1 is assigned to an aldehyde CH
stretch. It has been shown previously that aldehyde stretches,
usually near 2800 cmÀ1, can be shifted into this range when
adsorbed onto a surface.[24] This assignment is in agreement
with the proposed h2 adsorption mode, which is believed to be
the intermediate responsible for the formation of butyralde-
hyde. The clear similarities between the SFG spectra taken on
the Pt/SiO2 and Pt (111) single crystal further support the
claim that the SiO2 support plays no active role in the
hydrogenation of crotonaldehyde on Pt.
The spectrum of Pt/TiO2 (Figure 3B) shows several
intense negative features at 2880, 2930, and 2965 cmÀ1. Both
the strength and sign of these peaks are due to the increased
nonresonant component of the signal from Pt/TiO2 in the
presence of hydrogen. Hydrogen spillover from Pt to TiO2
serves to reduce the surface of the TiO2 to TiO2–x, and in turn
makes it more metallic. Additionally, this surface reduction
bonyl carbon atom by H spillover from the Pt to the TiO2. The
result of this initial hydrogenation step is a crotyl-oxy surface
intermediate (Figure 3B), which is the precursor to the crotyl
alcohol product.
As stated previously, it is believed that the Ti3+ atoms at
O-vacancy sites bind strongly to the aldehyde oxygen atom
and activate the aldehyde carbon atom for hydrogenation by
a charge-transfer process, thus drawing an analogy to acid-
base catalysis. The charge-transfer process described previ-
ously,[4] wherein the excess electron at the Ti3+ site forms
À
a covalent Ti O bond with the aldehyde oxygen atom, results
=
in a charge on the aldehyde carbon atom as the C O bond
À
becomes a C O s bond. This charged carbon atom is highly
reactive and quickly reacts with the atomic hydrogen present
on the surface. In the previous case of furfural, the adsorption
of the aldehyde group at this Ti3+ site resulted in an energy
decrease of the system by 1.35 eV as determined by DFT
calculations. Addition of the hydrogen at the carbonyl carbon
atom lowered the energy of the system by an additional
0.2 eV. The adsorption and initial single hydrogen addition
are believed to be very fast and result in a fairly stable surface
Angew. Chem. Int. Ed. 2014, 53, 3405 –3408
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim