A. F. Lee et al.
to a palladium/palladium oxide mixture reveals the presence
of a nearest-neighbor PdÀO coordination shell at 2.01 ꢁ, but
no additional PdÀO scattering pairs and multiple PdÀPd coor-
dination shells characteristic of face-centered-cubic metallic
palladium (Figure 5, and also see Figure S10 and Table S1).
These fits are best rationalized by a metallic palladium core en-
capsulated by an oxide overlayer, as proposed for Pd/Al2O3 cat-
alysts prepared in an identical fashion. Although XPS is intrinsi-
cally surface sensitive and XAS provides an average measure-
ment across all bulk and surface chemical environments, the
inelastic mean free path of Pd3d5/2 photoelectrons (1.31 nm
using AlKa X-rays)[62] is comparable to the diameter of palladi-
um nanoparticles (ꢀ1 nm) in the 0.05 wt% Pd/SBA-15/16 ma-
terials; hence, similar surface and bulk oxide compositions are
anticipated for such small metal core–oxide shell nanostruc-
tures. Particle size estimates, derived from the palladium–palla-
dium first shell coordination number adopting the method of
Jentys[63] and assuming a spherical morphology, predict diame-
ters around 1.4 and 0.8 nm for 0.05 wt% Pd/SBA-15 and
0.05 wt% Pd/SBA-16, respectively, which are close to those ob-
tained through CO chemisorption and TEM.
aside, external mass transport to Pd/SBA-16, Pd/KIT-6, and Pd/
SiO2 catalysts is notably enhanced over Pd/SBA-15, with a maxi-
mum constant activity attained at lower stirring speeds (
ꢀ400–600 rpm for the former versus 800–900 rpm for the
latter) for both crotyl alcohol and cinnamyl alcohol selox. The
origin of this is not yet understood and may arise from differ-
ing surface polarity or roughness. The confirmation that our re-
action kinetics were measured in the absence of bulk diffusion
limitations was obtained by varying the catalyst/substrate ratio
and the oxygen flow rate through the reactor: normalized ini-
tial rates were independent of the catalyst amount for crotyl
alcohol and cinnamyl alcohol selox over 0.89 wt% Pd/SBA-15,
and the oxygen flow for cinnamyl alcohol selox over 0.46 wt%
Pd/KIT-6 (see Figures S12 and S13). Potential metal leaching
was assessed by removing a catalyst during a reaction through
hot filtration (see Figure S14).[68] The remaining filtrate showed
no activity, and atomic absorption analysis confirmed no solu-
ble palladium within the detection limit (ꢀ5 ppm). Further evi-
dence for the heterogeneous nature of our catalysis was pro-
vided by recycle tests in which negligible change in the initial
rate or final conversion/selectivity was observed over three
consecutive reactions (see Figure S15), also discounting irrever-
sible deactivation.
Selective oxidation of allylic alcohols
All four Pd/silicas show a dramatic inverse dependence of
normalized initial selox rates on total metal loading (see Fig-
ure S16), similar to that reported for alumina-supported palla-
dium nanoparticles in the selox of crotyl alcohol, cinnamyl al-
cohol, and benzyl alcohol.[23,25,26,29,34] The magnitude by which
selox activity can be systematically tuned simply by varying
palladium loading is at first glance surprising in light of the
small concomitant changes in the particle size so induced
within each Pd/silica family (Table 2). However, recall that the
electronic and geometric properties of metal nanoparticles
evolve most rapidly over precisely the size range spanned in
this work. For example, increasing the palladium loading from
0.05 to 2.17 wt% over the SBA-15 support increases the mea-
sured particle size from 1.2 to 2.3 nm, which, for an unrelaxed,
spherical face-centered cubic palladium metal cluster, repre-
sents an increase from 43 to 429 constituent atoms—of which
the number of low-coordinate sites (ꢁ6 nearest neighbors)
falls from 55 to 11%. We firmly believe that the observed palla-
dium loading (size) dependence arises from the associated
change in the surface oxidation state, and we will return to
this point shortly.
The preceding families of Pd/silicas were screened in the aero-
bic selox of crotyl alcohol and cinnamyl alcohol to quantify
structure–function relations and optimize reaction conditions.
To measure intrinsic reaction kinetics in the absence of mass-
transfer limitations (arising from reactant/product diffusion
across either the gas–liquid interface or the liquid–solid boun-
dary layer of silica particles), the effect of mixing speed on ac-
tivity was initially studied for representative 1 and 2.5 wt% Pd/
silicas to determine the optimum stirrer speed for the subse-
quent quantitative evaluation of support effects (see Fig-
ure S11).[64] In all cases, mixing speeds greater than 1000 rpm
were sufficient to eliminate external mass transport, which is
likely dominated by oxygen solubilization under our mild con-
ditions (atmospheric pressure). Having established an efficient
mixing regime, striking differences emerge between the inher-
ent activities of the four distinct Pd/silica catalysts, with Pd/
SBA-16 and Pd/KIT-6 giving maximal rates 2–3 times those of
Pd/SBA-15, which in turn demonstrates a similar magnitude
enhancement over the amorphous Pd/SiO2 catalyst. The rela-
tive selox activities for both alcohols are thus intimately linked
to the degree of mesopore connectivity (and corresponding
palladium dispersion/oxidation state), which is in line with pre-
vious studies demonstrating the benefits of using intercon-
nected pore architectures in heterogeneous catalysis.[20,39,65–67]
The slower oxidation of cinnamyl alcohol versus crotyl alcohol
is likewise observed over Pd/Al2O3 and Au–Pd/TiO2 catalysts,[23]
and it may reflect either more sluggish in-pore molecular diffu-
sion owing to its heavier molecular mass, a greater adsorption
“footprint” requiring a larger (more scarce) active site ensem-
ble, or a higher activation barrier to the rate-determining OÀH/
CÀH cleavage. In the case of cinnamyl alcohol selox, Pd/KIT-6
slightly outperforms Pd/SBA-16, which may reflect the narrow-
er ink bottle-shaped pore opening of the latter.[40,41] As an
Selox activity is also a strong function of silica support,
which increases from 6000 mmol gPdÀ1 hÀ1 over the best com-
mercial low-area Pd/SiO2 to 14000 mmol gPdÀ1 hÀ1 over the
analogous high-surface area mesoporous Pd/SBA-15 and
reaches 24000 mmol gPdÀ1 hÀ1 for the interconnected mesopo-
rous Pd/SBA-16 and Pd/KIT-6 variants. Hence, support surface
area and mesopore architecture both play an important role in
controlling selox performance, with (interconnected) meso-
pores promoting crotyl alcohol and cinnamyl alcohol conver-
sion either through improved in-pore diffusion to the active
site or by increasing the number of such sites. To explore
which of these factors is most influential and to shed insight
on the nature of the active palladium species responsible for
&6
&
ꢀ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemCatChem 0000, 00, 1 – 13
ÝÝ
These are not the final page numbers!