Both H2O and O2 are needed to form active hydroperoxy
species required for PO synthesis. Neither C3H6–O2 nor
C3H6–H2O reactants formed detectable products on
Au/TiO2 at 350 K. PO synthesis also requires the presence
of both Au and Ti sites. We did not detect PO from
C3H6–O2–H2 or C3H6–O2–H2O reactants on either TiO2 or
Au/Al2O3 catalysts. Au and Ti sites must reside in reasonable
proximity, since PO was not detected when C3H6–O2–H2 or
C3H6–O2–H2O reactants were contacted with a physical mix-
ture Au/Al2O3 and TiO2 (mass ratio 1 : 1, 0.125–0.250 mm
aggregates). Water inhibits combustion during C3H6–O2 reac-
tions on Ag-based catalysts, but does not increase PO syn-
thesis rates.19–21 We find instead that the role of H2O
on Au/TiO2 is to significantly increase PO synthesis rates.
Propene combustion rates were undetectable on Au/TiO2 with
or without added H2O.
The rate of PO formation (extrapolated to zero time-
on-stream) increased slightly from 0.5 to 0.7 mol hÀ1
(g-at Au)À1 when the H2O partial pressure in equimolecular
C3H6–O2 mixtures (4 kPa) increased from 1 to 2 kPa (Fig. 1);
higher H2O pressures (up to 12 kPa), however, did not
influence reaction rates.
Fig. 2 Selectivity (carbon basis) to propylene oxide obtained with
Au/TiO2 at 350 K at different propene conversion levels changed by
catalyst deactivation (4 kPa C3H6; 4 kPa O2; 4 kPa H2; 1–12 kPa H2O).
conversion decreased (Fig. 2). These data, taken together with
the intermediate selectivites observed as data are extrapolated
to zero conversion, indicate that both PO and acetone form as
primary products and that deactivation occurs by blocking of
sites without concomitant changes in the relative rates of PO
and acetone synthesis.
The PO synthesis rates reported here are inconsistent with
the use of gaseous H2O2 reactants as intermediates and
indicate that PO is formed instead from propene and adsorbed
hydroperoxide species. H2O2 pressures in equilibrium with
Fig. 1 and 2 provide evidence for the previously unrecognized
ability of Au/TiO2 to catalyze propene epoxidation with O2 and
H2O (instead of H2) as co-reactant. We have carried out
experiments for extended periods of time (ESIw), during which
we carry out more than 10 epoxidation turnovers, calculated on
the basis of surface Au atoms (from TEM cluster diameters).
These turnover numbers represent a lower bound, because they
assume that all exposed Au atoms act as active sites, irrespective
of their location with respect to Ti centers. These data confirm
that the propene epoxidation rates reported here with H2O–O2
co-reactants are catalytic on Au/TiO2.
O2/H2O (4 kPa/12 kPa) are 7 Â 10À18 kPa (Keq
=
1.48 Â 10À19 kPaÀ1/2; 350 K). At this pressure, the fre-
quency of H2O2 collisions with Au clusters (ESIw) would be
5 Â 10À6 mol hÀ1 (g-at Au)À1, a value much lower than
required to maintain the observed epoxidation rates
(B0.7 mol hÀ1 (g-at Au)À1), consistent with an inadequate
supply of H2O2(g) as the reactive species and with PO synth-
esis via propene reactions with hydroperoxy surface species
instead of H2O2. The involvement of bound hydroperoxy
species requires, in turn, atomic proximity between the sites
that form *OOH (Au) and those that consume it via reactions
with propene (Ti), possibly at Au–TiO2 interfaces, as proposed
earlier4 and consistent with the absence of epoxidation turn-
overs on physical mixtures of Au/Al2O3 and TiO2.
The synthesis of PO during water electrolysis, probably
via in situ generation of H2O2 or OOH species, has been
reported.14 In contrast, catalytic epoxidation of propene or
other substrates with H2O–O2 reactants have not been
previously reported. PO synthesis rates with H2O–O2 reactants
are significantly lower than with H2–O2, but the former avoid
significant losses of costly H2 co-reactant via its unproductive
pathways to form H2O instead of OOH species (hydrogen
efficiency B29%). O2–H2O mixtures may form HOO* species
via the microscopic reverse of elementary steps for H2O2
decomposition.15,16 These steps can occur, in spite of their
unfavorable thermodynamics, because of their kinetic cou-
pling with propene epoxidation steps that scavenge HOO*
intermediates to form PO. OH* groups formed in these steps
must recombine to form H2O* and O*, since H2O is not
consumed during reaction. O* is not predominantly removed
via unselective scavenging with PO to form other products
(e.g., acetone, propanal, acrolein), because measured PO
selectivities are 450% (Fig. 2). Thus, O* species must act as
epoxidation reactants, desorb as O2 (2O* - O2 + 2*), or
form ozone molecules (O* + O2 - O3*), which are then used
in epoxidation turnovers.17,18
Density functional theory (DFT)22 suggests that OOH
species can form from H2O and O2 on Au8 clusters to form
[O2ÁH2O] complexes in which protons are shared between H2O
and O2 to ultimately form adsorbed (HOO*) complexes. We
propose that these species can act as effective oxidants in
reactions of CO to CO2.11 and, in the present study, for
propene epoxidation to PO.
C3H6–O2–H2O reactants form predominantly PO and
acetone on Au/TiO2 at 350 K. Fig. 2 shows PO selectivities at
various H2O inlet pressures. PO selectivities increased slightly
from 65 to 70% as H2O pressure increased from 1 to 2 kPa, but
then decreased markedly (70 to 20%) at higher H2O pressures
(2–12 kPa). with a concurrent increase in acetone selectivity.
The selectivities to PO and acetone depend only weakly on
propene conversion, suggesting that both can form via primary
pathways. The strong effects of H2O pressure on PO selectivity
and the slight increase in acetone selectivity with increasing
conversion show, however, that secondary reactions of PO to
form acetone are favored by H2O. PO synthesis rates, however,
ꢀc
This journal is The Royal Society of Chemistry 2009
Chem. Commun., 2009, 352–354 | 353
Ojeda, Manuel
