Angewandte
Research Articles
Chemie
materials, the initial CO formation rates for all materials are
microscopic methods. For example, M-O-Cu paths in EXAFS
similar and nearly identical to those for Cu/SiO2 (2.95– fitting (Supporting Information, Section S4) were not fit, but
3.47 mmol(gCu s)À1 for Cu/M@SiO2; 3.8 mmol(gCu s)À1 for Cu/
SiO2). These results suggest that these reactions occur on the
same types of sites and are consistent with previous observa-
tions that (reverse) water-gas-shift reactions occur on Cu
surfaces.[37,40–42] Consequently, initial CH3OH molar selectiv-
ities are greater for Cu/M@SiO2 materials (77–85%) com-
pared to Cu/SiO2 (49%). Molar selectivity is greatest for Cu/
Ti@SiO2 (85%) and decreases slightly as Ti is replaced with
Nb, Ta, Zr, and Hf, respectively (from 82% to 77%). These
trends reflect the same promotional effects of M@SiO2 on
CH3OH formation rates.
the small scattering features at greater distances could arise
from Cu nanoparticles. Nevertheless, the use of SOMC to
generate isolated metal sites and/or metal nanoparticles
involves the grafting of metal precursors onto isolated
silanols, which are likely statistically distributed ( ꢀ 1 OH/
nmÀ2).[22] SOMC generates highly dispersed metal sites or
metal nanoparticles upon thermal treatments, as evidenced by
EXAFS (Supporting Information, Section S4) and microsco-
py (Figure 1c,d and Supporting Information, Figure S3),
respectively. Consequently, the random distribution of M
sites and Cu nanoparticles across the SiO2 surface results in
each Cu nanoparticle being near at least one M site. The
proximity of these M sites and Cu nanoparticles is further
corroborated by the observed promotional effects on CH3OH
formation rates and selectivities for these catalysts, while
physical mixtures of M@SiO2 with Cu/SiO2 behave as Cu/SiO2
alone (vide supra). These findings are also consistent with
reaction pathways determined using density functional theory
calculations for Cu-based materials on crystalline ZrO2 and
Al2O3 supports.[17,46] Sites at the interface of Cu nanoparticles
and the Lewis acid support were shown to not only promote
the activation of CO2 but also to stabilize reaction intermedi-
ates such as formate and methoxy surface species, compared
to pathways catalyzed by Cu surfaces alone. Additionally,
although CO2 could be activated by the Lewis acid sites of the
support alone, methanol formation does not occur in the
absence of Cu nanoparticles.
The increases in CH3OH rates and selectivities arising
from these isolated group IVand V metal centers on SiO2 can
therefore be attributed to their role as Lewis acid sites, which
have previously been demonstrated to stabilize electron-rich
formate and methoxy surface intermediates.[17–20] These
interfacial sites here result in the observed decreases in
CH3OH formation rate with increasing residence time
because products formed during these reactions (i.e., H2O
or CH3OH) competitively adsorb onto these Lewis acid sites.
Next, we measure Lewis acid strength for M@SiO2 and
examine the stabilization of surface intermediates.
These promotional effects are explored further by exam-
ining the effects of reactor residence time on individual
product formation rates. CO and CH3OH formation rates on
Cu/SiO2 (Figure S8f) are nearly independent of residence
time. For Cu/SiO2, the support can be considered innocent
and Cu nanoparticles are likely the dominant active site for
catalysis. CO formation rates for all Cu/M@SiO2 materials,
normalized by mass of Cu, are similar in value
( ꢀ 3.2 mmol(gCu s)À1) and nearly constant, irrespective of
both residence time and catalyst identity (Figure 2; Fig-
ure S8). Furthermore, their values are similar to those for Cu/
SiO2 (3.78 mmol(gCu s)À1), suggesting that these reactions
occur on the same active site as those for Cu/SiO2 and that
CO is not an intermediate to form CH3OH (consistent with
previous reports[36–42]). In contrast, CH3OH formation rates
on Cu/M@SiO2 decrease precipitously (by 20–30% as resi-
dence time increases 2-fold) with increasing residence time
(Figure S8), in spite of the low value of CO2 conversion
(< 7%). These trends cannot be explained by invoking the
same active site as for Cu/SiO2, because the rates of reactions
catalyzed by such active sites (i.e., both CO and CH3OH
formation rates over Cu/SiO2 and CO formation rates over
Cu/M@SiO2) are essentially independent of residence time.
These residence time effects on CH3OH formation rates for
Cu/M@SiO2 suggest the presence of two different active sites
for these materials: the first is likely the same as for Cu/SiO2
and is responsible for the formation of CO, and the second is
unique to Cu/M@SiO2 and is the active site for the formation
of CH3OH.
To further investigate the nature of this latter active site,
the catalytic performance of M@SiO2 and physical mixtures
of M@SiO2 with Cu/SiO2 were assessed (Table S13). Concen-
trations of products were below detection limits for all
M@SiO2, indicating that these dispersed metal sites alone are
not capable of catalyzing CO2 hydrogenation. Each physical
mixture of M@SiO2 and Cu/SiO2 yielded the same product
formation rates and selectivities as Cu/SiO2 alone. These
results suggest a requirement for site proximity between the
M sites and the Cu nanoparticles. Consequently, the active
sites (or regions) for CH3OH formation over Cu/M@SiO2 are
likely composed of adjacent Cu and M atoms, at the periphery
of Cu nanoparticles. Similar active sites that interface metal
nanoparticles and metal oxide supports have been invoked in
a variety of catalytic reactions.[13–15,43,44]
Measures of Lewis Acid Strength
The enthalpy of adsorption for pyridine, a molecule with
a lone pair of electrons that interact with electrophilic Lewis
acid sites and whose interaction has been widely stud-
ied,[26,28,45] was measured. Pyridine adsorption enthalpies
(DHads,pyridine
)
were determined from pyridine isobars
(0.1 kPa, 483–523 K) using IR spectroscopy (spectra at
483 K included in the Supporting Information, Figure S10a,
for M@SiO2 materials) by evaluating the area of the pyridine
vibrational band at 1450 cmÀ1 and regressing these values to
the vanꢀt Hoff equation (isobars in the Supporting Informa-
tion, Figure S10b). Heats of adsorption were assessed using
M@SiO2 because of the presence of an additional vibration
band at 1444 cmÀ1 for Cu/M@SiO2; the area and the FWHM
of the 1450 cmÀ1 bands are nearly identical for each M@SiO2
compared to its respective Cu/M@SiO2 material after sub-
Noteworthy is the challenge in assessing the proximity
between M atoms and Cu nanoparticles by spectroscopic or
ꢀ 2021 Wiley-VCH GmbH
Angew. Chem. Int. Ed. 2021, 60, 9650 –9659