You are on page 1of 26

Article

Photocatalytic Hydrogenation of Carbon


Dioxide with High Selectivity to Methanol at
Atmospheric Pressure
Lu Wang, Mireille Ghoussoub,
Hong Wang, ..., Le He,
Xiaohong Zhang, Geoffrey A.
Ozin
lehe@suda.edu.cn (L.H.)
xiaohong_zhang@suda.edu.cn (X.Z.)
gozin@chem.utoronto.ca (G.A.O.)

HIGHLIGHTS
Photocatalytic hydrogenation of
CO2 to solar methanol at
atmospheric pressure

One-pot methanol synthesis with


50% selectivity and excellent
stability

Rod-like In2O3 x(OH)y nanocrystal


superstructure enhanced
photocatalytic performance

Rod-like In2O3 x(OH)y nanocrystal superstructure enhanced solar methanol


synthesis with a remarkable production rate (0.06 mmol gcat 1 h 1) and selectivity
(50%) at atmospheric pressure.

Wang et al., Joule 2, 1369–1381


July 18, 2018 ª 2018 Elsevier Inc.
https://doi.org/10.1016/j.joule.2018.03.007
Article
Photocatalytic Hydrogenation of Carbon
Dioxide with High Selectivity
to Methanol at Atmospheric Pressure
Lu Wang,1,2,4 Mireille Ghoussoub,2,4 Hong Wang,2 Yue Shao,2 Wei Sun,2 Athanasios A. Tountas,2
Thomas E. Wood,2 Hai Li,1 Joel Yi Yang Loh,3 Yuchan Dong,2 Meikun Xia,2 Young Li,2
Shenghua Wang,1 Jia Jia,2 Chenyue Qiu,2 Chenxi Qian,2 Nazir P. Kherani,3 Le He,1,*
Xiaohong Zhang,1,* and Geoffrey A. Ozin2,5,*

SUMMARY Context & Scale


The production of solar methanol, directly from gaseous CO2 and H2, is impor- The current fashion for
tant for the development of a sustainable energy economy. Despite growing synthesizing methanol continues
activity in the field, very few photocatalysts exist that can efficiently and stably to be the high-pressure, thermally
hydrogenate gaseous CO2 to methanol at ambient pressure with high selec- driven heterogeneous catalytic
tivity. Here, we report that a defect-laden indium oxide, In2O3 x(OH)y, with a conversion of synthesis gas
rod-like nanocrystal superstructure, can photocatalyze the hydrogenation of (CO-H2) over an alumina-
CO2 to methanol with 50% selectivity under simulated solar irradiation. supported copper-zinc oxide
Notably, the solar methanol production of the In2O3 x(OH)y nanocrystal super- catalyst and fossil fuel to power
structures can be stabilized at a rate of 0.06 mmol gcat 1 h 1 at atmospheric the process. It is an energy-
pressure. This is 120 times higher than that of the best-known photocatalysts. intensive process with a large
This discovery bodes well for the development of a low-pressure solar methanol greenhouse gas footprint. Thus, it
process using CO2 and renewable H2 feedstocks. would be highly desirable to
produce it in a sustainable way
and use CO2 as feedstock and
INTRODUCTION solar energy to drive the synthesis.
The development of methanol synthesis from the hydrogenation of CO2 is important
for achieving a greener chemical industry. Methanol is consumed at a scale of Solar technologies that facilitate
65 million tons/year (2013) and is considered to be a future energy carrier due to the efficient conversion of CO2
its high volume-specific energy density.1 It is also a convenient medium for the and H2 into methanol offer a
safe storage of H2 and an important renewable feedstock for the chemical industry, sustainable path to produce
as around 30% of known chemicals can be derived from methanol.2,3 Furthermore, renewable fuels. Here, we present
the desire of the transportation industry to replace diesel by dimethyl ether or its a rod-shaped In2O3 x(OH)y
higher homolog oxy-methylene ether (OME-1), as a cleaner burning fuel has re- nanocrystal superstructure that
focused interest on its production from the catalytic dehydration of methanol can efficiently hydrogenate CO2
made from CO2. to methanol at atmospheric
pressure with a selectivity of more
There have been extensive research efforts over many decades directed to the than 50%. The remarkable
synthesis of methanol by the thermal-catalytic hydrogenation of CO2, mostly by production rate of 0.06 mmol
Cu/ZnO-based nanostructures, to promote selectivity for methanol production by gcat 1 h 1 and excellent long-term
inhibiting CO generation through the competing reverse water gas shift (RWGS) re- stability of this catalyst in solar
action.4–14 Due to the competing RWGS reaction, the exothermic nature of meth- methanol synthesis make it an
anol synthesis, equilibrium-limited conversion, and very high activation energy bar- interesting candidate for
rier of the CO2 hydrogenation to methanol reaction, the process must balance converting CO2 to methanol at an
operation temperature and megapascal pressures to achieve a high selectivity industrial scale.
and production rate.15 It remains a big challenge to achieve effective methanol

Joule 2, 1369–1381, July 18, 2018 ª 2018 Elsevier Inc. 1369


production from CO2 hydrogenation with a reasonably high rate and selectivity at
atmospheric pressure.

It would be of considerable scientific and technological interest if this reaction could


be photocatalytic because, together with renewable solar energy, methanol can be
produced without generating greenhouse gas emissions, which would make it an
important feedstock for the chemical industry as well as an energy carrier for trans-
portation and fuel cells.16–22 We emphasize that the source of H2 remains key to the
environmental viability of photocatalysis, and therefore we envision a process
whereby the H2 would be sourced via electrolysis driven by renewable electricity.
With the emergence of technologies targeting the issues of efficiency, scalability,
and costs associated with renewable methods of H2 production, alternative routes
are anticipated to become competitive options to the industrially ubiquitous steam
reforming of natural gas.23–25 Prior attempts for photo-assisted hydrogenation of
CO2 have mainly produced CH4 and CO.26–33 The photocatalytic conversion of
gaseous CO2 and H2 to methanol would be of pivotal interest and importance in
the quest for a sustainable CO2-based economy.34–40 Despite the growing interest
in this photocatalytic reaction, examples of efficacious photocatalysts are limited,
and the best photocatalytic activity at atmospheric pressure reported to date is
below mmol gcat 1 h 1, far from practical applicability.41–44

Here, we report that defect-laden indium oxide, In2O3 x(OH)y, with a rod-like nano-
crystal superstructure, can effectively catalyze the hydrogenation of CO2 to meth-
anol under illumination at atmospheric pressure at a champion rate of around
0.06 mmol gcat 1 h 1 for long-term reaction, impressive 50% selectivity, and notable
long-term operational stability. Success in this endeavor is enabled by recognition of
the photocatalytic activity of surface frustrated Lewis pairs (SFLP), comprising Lewis
acidic coordinately unsaturated In surface sites proximal to Lewis basic In hydroxide
surface sites in non-stoichiometric metal oxides, exemplified by archetypal nano-
crystal forms of In2O3 x(OH)y.45–48 SFLP have been shown to increase their Lewis
acidity and Lewis basicity respectively in the excited state compared with the ground
state, thereby facilitating the photochemical RWGS reaction, with a decrease in acti-
vation energy by 20 kJ mol 1 under illumination.49,50

RESULTS
Synthesis and Characterization of In2O3 x(OH)y Nanocrystal Superstructures
In2O3 x(OH)y superstructures composed of nanocrystal were prepared through a 1Institute
of Functional Nano & Soft Materials
two-step process with slight modifications from the previous study.47 Figure S1 (FUNSOM), Jiangsu Key Laboratory for
shows the transmission electron microscopy (TEM) image of precursor In(OH)3 Carbon-Based Functional Materials & Devices,
Soochow University, 199 Ren’ai Road, Suzhou,
nanorods prepared from InCl3 and urea with the reaction time of 14 hr at 80 C. Jiangsu 215123, People’s Republic of China
This sample was then treated at 250 C in air to obtain a rod-like In2O3 x(OH)y nano- 2MaterialsChemistry and Nanochemistry
crystal superstructure with an average length of 2.6 mm, denoted as NR-14h Research Group, Solar Fuels Cluster,
Departments of Chemistry, University of Toronto,
(Figure 1A). The nanoporous feature of the materials obtained, composed of nano-
80 St. George Street, Toronto, ON M5S 3H6,
crystal superstructures, was confirmed by TEM studies (Figures 1B and 1C). The Canada
nanocrystals inside each rod-like superstructure have similar orientations, as shown 3Department of Materials Science and
by the selected area electron diffraction (SAED) pattern from a single rod (Figure 1D). Engineering, University of Toronto, 184 College
Street, Suite 140, Toronto, ON M5S 3E4, Canada
The powder X-ray diffraction (PXRD) pattern of the products can be assigned to the 4These authors contributed equally
phase pure bixbyite structure type of In2O3 x(OH)y (Figure 1E). The grain size calcu- 5Lead Contact
lated from the width of the strongest PXRD peak at 30.6 was 11 nm. X-ray photo-
*Correspondence: lehe@suda.edu.cn (L.H.),
electron spectroscopy (XPS) provided information regarding element oxidation xiaohong_zhang@suda.edu.cn (X.Z.),
states (Figure S2). The spin-orbit doublet in the In 3d core level XPS spectra at gozin@chem.utoronto.ca (G.A.O.)
444.1 and 451.6 eV confirmed the oxidation state of In(III) in In2O3 x(OH)y. The O https://doi.org/10.1016/j.joule.2018.03.007

1370 Joule 2, 1369–1381, July 18, 2018


Figure 1. Characterization of In2O3 x(OH)y Nanocrystal Superstructures
(A) SEM image of prepared superstructures. Inset in (A) shows the statistic histogram of the rod length.
(B) TEM image of a single rod-like nanocrystal superstructure.
(C) High-resolution TEM image of the prepared nanocrystal superstructure.
(D) SAED pattern of the obtained In 2 O 3 x (OH) y superstructure.
(E) PXRD patterns of the precursor In(OH) 3 and as-prepared In 2 O3 x (OH) y .
(F) O 1s core level XPS spectra of the as-prepared rod-like In 2 O 3 x (OH) y nanocrystal superstructure. De-convolution of the O1s peak exhibits three
distinct peaks at 529.6 eV, 531.1 eV, and 532.1 eV. The peaks at 529.6 eV (O I ) and 532.1 eV (OIII ) correspond to native oxide and surface hydroxyl groups,
respectively. The peak at 531.1 eV (O II ) can be attributed to the presence of oxygen vacancies.

1s core level XPS spectra could be fitted into three peaks at 529.6 eV, 531.1 eV, and
532.1 eV, corresponding to oxide, oxygen vacancies, and hydroxyl groups, respec-
tively (Figure 1F).45,47 The O/OH ratio in In2O3 x(OH)y for NR-14h was found to be
7:1 from the XPS results, which is consistent with the value from thermogravimetric
analysis (TGA) studies (Figure S3). The In2O3 x(OH)y nanocrystal superstructures ex-
hibited nanoporous morphology comprising a three-dimensional network of nano-
crystals as confirmed by a N2 absorption-desorption study (Figure S4). The specific
surface area of the In2O3 x(OH)y nanocrystal superstructure was found to be reason-
ably high at 161.21 m2/g with a narrow pore size distribution from 3 nm to 7 nm.

Photocatalytic Activity of In2O3 x(OH)y Nanocrystal Superstructures


The photocatalytic activity of the In2O3 x(OH)y nanocrystal superstructures for the
hydrogenation of CO2 was tested at three different temperatures: 200 C, 250 C,
and 300 C, in a flow reactor system working at atmospheric pressure. The ratio of
H2 to CO2 was maintained at 3:1 throughout the reaction. At 200 C, CO was found
to be the only product under both dark and light conditions (Figure 2A). The CO pro-
duction rate increased from 6.35 to 14.54 mmol gcat 1 h 1 on switching from dark to

Joule 2, 1369–1381, July 18, 2018 1371


Figure 2. Methanol Production Rate and Selectivity at Different Temperatures and Atmospheric Pressure
(A–C) Carbon monoxide rate (A), methanol rate (B), and methanol selectivity (C) of the In 2 O3 x (OH) y nanocrystal superstructure (NR-14h) in catalyzing
hydrogenation of CO 2 with and without solar irradiation.

light operation conditions. The rate increase in the light can be explained by the
lower activation energy of the photocatalytic process relative to the thermocatalytic
process.49 In2O3 x(OH)y nanocrystal superstructures showed effective CO produc-
tion under illumination at 150 C in a batch reactor with a 1:1 ratio of H2 to CO2.
These results suggest that below 200 C, the In2O3 x(OH)y nanocrystal superstruc-
tures can only catalyze the RWGS reaction. The production of methanol is most likely
kinetically inhibited at low temperatures and pressures.15,51

On raising the temperature to 250 C, the CO production rate increased to 39.79 and
86.17 mmol gcat 1 h 1 for dark and light reaction conditions, respectively. Methanol
production was also observed at 250 C with a selectively of over 50% at 97.3 and
54.1 mmol gcat 1 h 1 with and without illumination, respectively (Figures 2B and 2C).
The methanol production rates observed at atmospheric pressure are over 200 times
higher than the highest rate with simulated solar irradiation reported thus far in the
literature.41–43 As discussed later, it is most likely that the function of light is to activate
the SFLP and further decrease the activation energy of the reaction.

On raising the reaction temperature to 300 C, the production rate of methanol drop-
ped to 37.83 and 59.46 mmol gcat 1 h 1 for dark and light reaction conditions with
21% selectivity, respectively. This could be explained by the exothermic nature of
the methanol production reaction (CO2 + 3H2 # CH3OH + H2O) shifting the equi-
librium toward the reactant side at higher temperatures according to Le Chatelier’s
principle. In other words, the thermodynamics of the two competing reactions favors
RWGS over methanol synthesis at 300 C. A preliminary temperature dependence of
solar methanol production over In2O3 x(OH)y nanocrystal is shown in Figure S5
and confirmed that 250 C was the optimal temperature for the solar methanol pro-
duction over the In2O3-x(OH)y catalyst, which is also the temperature used for ther-
mocatalytic methanol synthesis. Furthermore, the quantum efficiency of the best
sample NR-14h was estimated to be 0.19% at 250 C, which could potentially in-
crease by 100–500 times (2%–10%) with improved photocatalyst and photoreactor
architectures and optimization of the solar light optics that maximizes light adsorp-
tion (Supplemental Information).

The ‘‘solar advantage’’ for methanol production in the light compared with the dark
can be readily ascertained by gas chromatography, which shows different diagnostic
peak heights for photochemical and thermochemical methanol product (Figure S6).
Additional confirmation for methanol production stems from 1H and 13C solution-
phase nuclear magnetic resonance (NMR) studies of the hydrogenation product of

1372 Joule 2, 1369–1381, July 18, 2018


Table 1. Summary of Properties of Different In2O3 x(OH)y Samples
Sample tia Lb Dc Ad Rde Sd (%)f R Lg SL (%)h
NC – – 12 115 8.26 40 9.98 39
NR-3h 3 950 12 146 9.39 60 14.09 60
NR-9h 9 1685 11 167 33.65 57 54.45 57
NR-14h 14 2580 11 162 54.06 57 97.30 53
a
Aging time for the synthesis of In(OH)3 nanorods (hr).
b
The length of rod-like nanocrystal superstructures (nm).
c
Grain size calculated from PXRD patterns (nm).
d
Specific surface area obtained from BET measurement (m2/g).
e
Methanol rate without solar irradiation (mmol gcat 1 h 1).
f
Methanol selectivity without solar irradiation.
g
Methanol rate with solar irradiation (mmol gcat 1 h 1).
h
Methanol selectivity with solar irradiation. Methanol rates obtained at 250 C.

gaseous 13CO2, when employed as the CO2 source (Figure S7). In these experi-
ments, D2O was the solvent for collecting methanol product. Diagnostic NMR chem-
ical shift peaks at 3.36 and 3.07 for 1H NMR and at 48.77 for 13C NMR confirmed the
formation of 13C methanol by the light-assisted hydrogenation of 13CO2 on
In2O3 x(OH)y.52

A megapascal reaction pressure is typically required for effective methanol produc-


tion to improve methanol selectivity.15,16,51 Izumi and co-workers reported that
layered double hydroxides could enable the photoconversion of CO2 and H2 into
methanol with the highest rate of 2.7 mmol gcat 1 h 1 at 0.4 MPa.44 Thus, our study
represents a breakthrough in effective photocatalytic methanol production from
CO2 hydrogenation at atmospheric pressure.16,36

Effect of Nanocrystal Superstructures


It is known that nanocrystal superstructures can promote photocatalytic processes
by elongating the lifetime of photoexcited charge carriers.47 To understand whether
the same effect contributed to the high rate of methanol production, two
In2O3 x(OH)y nanocrystal superstructure samples, denoted as NR-3h and NR-9h,
with shorter lengths were prepared from In(OH)3 precursors made with aging times
of 3 and 9 hr, respectively (Figures S8A and S8B). Meanwhile, an In2O3 x(OH)y
nanocrystal sample, denoted as NC, was also prepared as the reference material
(Figure S8C).45 PXRD patterns were found to be similar for all samples and could
be assigned solely to the phase pure bixbyite structure type of In2O3 x(OH)y
(Figure S9). The grain sizes obtained from the PXRD results are very close among
the four samples (Table 1). The specific surface areas of the nanocrystal superstruc-
ture samples were similar but slightly larger than that of nanocrystals (Table 1 and
Figure S4). All samples showed the same In(III) oxidation state, and the total concen-
tration of oxygen vacancy and hydroxide defects that were obtained from XPS are
similar for all samples (Figures S2B and S2C). The concentration of oxygen vacancies
in the nanocrystal superstructures are higher than that in the nanocrystal samples
(Figure S2C). An obvious slight red shift of the absorption edge in the diffuse reflec-
tance spectra occurs from nanocrystals to nanocrystal superstructures (Figure S10).
To further confirm the prolonged lifetime of the photoexcited charge carrier,
photoconductivity measurements were performed under the same conditions
(H2:CO2 = 3:1, 250 C) at atmospheric pressure for NR-3h, NR-9h, and NR-14h.
The photocurrents of NR-3h, NR-9h, and NR-14h with increasing lengths were re-
corded as 11.74 mA, 19.48 mA, and 87.97 mA, respectively, at a bias of 5 V

Joule 2, 1369–1381, July 18, 2018 1373


Figure 3. Effect of Nanocrystal Superstructures
(A and B) Methanol rate (A) and methanol selectivity (B) for nanocrystal (NC) and superstructure
samples in catalyzing hydrogenation of CO 2 at 250  C with and without solar irradiation at
atmospheric pressure.

(Figure S11). The photocurrent was observed to increase with increasing length of
the nanocrystal superstructure. The enhancement is caused by efficient charge car-
rier separation enabling a longer lifetime of the charge carriers.53 This result showed
good agreement with our previous transient absorption spectroscopy study on
In2O3 x(OH)y superstructures.47

The catalytic activity of these samples for CO2 hydrogenation to methanol and CO at
250 C with and without illumination are shown in Figures 3 and S12, and summarized
in Table 1. The methanol production rate was the lowest for the nanocrystals and
increased with the length of the superstructures under both dark and light condi-
tions. Since these samples exhibited similar specific surface areas and grain sizes,
the difference in the methanol production rate between nanocrystals and nanocrys-
tal superstructures under solar illumination could be attributed to the prolonged life-
time of excited charge carriers in nanocrystal superstructures.47 The increase in rate
parallel with the length of rod-like superstructures under illumination may also be ex-
plained by the increased lifetime of photoexcited charge carriers in longer rods.
Indeed, the methanol production rate was found to increase by 20%, 50%, 62%,
and 80% from dark to light for NC, NR-3h, NR-9h, and NR-14h, respectively. The
methanol production rate was higher in the light than in the dark, while the selectivity
was similar for dark and light, for all four samples. In other words, light could pro-
mote the conversion rates for both CO and methanol rather than just one of them.
This implies that the formation of CO and methanol may follow a similar reaction
pathway, with reduced activation energies in the light for both cases. Furthermore,
methanol selectivity increased from 40% for nanocrystals to above 50% for nanocrys-
tal superstructure samples. This reveals that the nanocrystal superstructures could
boost both the methanol production rate and selectivity.

Photocatalytic Stability of In2O3 x(OH)y Nanocrystal Superstructure


The photocatalytic stability of the In2O3 x(OH)y nanocrystal superstructure sample
was demonstrated for NR-14h. This sample was tested continuously for 20 hr at
250 C in the light, which was the optimum condition for methanol production. After
an initial drop by 40% in the first hour, the methanol rate stabilized during the next
15 hr, followed by a noticeable increase in the 16th hour (Figure 4A). The final stabi-
lized methanol production rate is about 0.064 mmol gcat 1 h 1, which is about 120
times higher performance than prior work. The methanol production rate of other
photocatalysts and comparable techniques are shown in Table S1. The selectivity
for methanol production exhibited very small changes and was above 50% during
the entire 20-hr measurement (Figure 4B). The PXRD pattern of sample NR-14h after

1374 Joule 2, 1369–1381, July 18, 2018


Figure 4. Stability of Catalytic Performance
(A and B) Methanol rate (A) and methanol selectivity (B) of NR-14h in catalyzing hydrogenation of
CO2 under illumination during a continuous 20-hr test at atmospheric pressure.

being tested for 20 hr exhibited very small new peaks at 33 and 39.3 , indicating
that toward the end of the photocatalytic testing, In2O3 x(OH)y is very slightly
reduced to metallic indium (Figure S13). The grain size of indium was about 40 nm
as estimated by the Scherrer equation. Metallic indium most likely existed on the sur-
face of rods. TEM and SAED studies notably confirmed that the nanocrystal super-
structures remained essentially intact and elemental mapping indicated the uniform
distribution of In and O (Figures S14 and S15). Corresponding In 3d core level XPS
spectra for the used and unused NR-14h are essentially the same, implying the fast
oxidization of metallic In to In(III) during the sample preparation process.

One can surmise that the small amount of an In@In2O3-x(OH)y heterostructured


nanocomposite with a metal/n-type semiconductor Mott-Schottky junction could
serve to enhance the photo-generation of electrons and holes, facilitate their trans-
fer to SFLP sites, and boost the production of CO and CH3OH compared with pris-
tine In2O3 x(OH)y.54 The results reveal that the In2O3 x(OH)y nanocrystal superstruc-
ture catalyst could achieve long-term methanol production with a high rate and
stable selectivity at ambient pressure.

DISCUSSION
Earlier investigations of the solar-powered RWGS reaction CO2 + H2 / CO + H2O
on In2O3 x(OH)y, identified SFLP InOH/In as the active sites responsible for the
excited state conversion of CO2 to CO.46,49,50 Given that light was not observed
to affect the selectivity toward methanol over carbon monoxide, we hypothesized
the reduction of CO2 to CH3OH to share a similar reaction pathway (Figure S16);
that is, one that proceeds via the heterolytic splitting of molecular hydrogen over
the SFLP site. To explore this idea further, density functional theory (DFT) calcula-
tions were performed to investigate the possible CO2 to CH3OH pathways on the
defective hydroxylated indium oxide surface. Previous computational studies re-
vealed the SFLP site, formed by a surface hydroxide being near an oxygen vacancy,
to be particularly effective in activating molecular hydrogen, and consequently in
reducing CO2 to produce CO.49 Therefore, we chose to investigate the possibility
of methanol formation via the SFLP site.

Figure 5 shows the full energy profile for the reaction 3H2 + CO2 / H2O + CH3OH
over the SFLP site. All intermediate state energies were computed relative to the
initial surface configuration, depicted in panel (I). In the first step, hydrogen is het-
erolytically split by the Lewis base surface hydroxide InOH and the Lewis acid sur-
face In. This process results in a hydride bound to the surface In as In-H, and a

Joule 2, 1369–1381, July 18, 2018 1375


Figure 5. Molecular-Level Mechanism for CO2 Hydrogenation into Methanol
Reaction energy profile from DFT calculations for methanol production via CO 2 hydrogenation over In 2 O3 x (OH) y (111). Gray atoms denote In, red atoms
denote O, green atoms denote C, and blue atoms denote H.

proton bound to the surface OH group as InOH2, shown in panel (II). Next, we
tested the possibility of CO2 interacting to form a surface bound formate
(HCO2*). Both monodentate and bidentate formate configurations were investi-
gated, the latter was further tested with all possible neighboring In atoms, and
the lowest energy configuration is shown in panel (III). The exergonic nature of
this step agrees with previously reported interactions between CO2 and the hydro-
genated SFLP site.49 The addition of a hydride produces an acetal intermediate
(H2CO2*), as shown in panel (IV); again, all possible configurations with neighboring
In were probed and the lowest energy case was selected as the most likely interme-
diate. The addition of molecular hydrogen results in the C-O bond splitting, with
the proton binding to the carbon fragment to form a methoxide (CH3O*) interme-
diate bound to a neighboring surface In. The remaining hydride binds to O on the
Lewis acid In and they are joined by a proton from the Lewis base site via a hopping
mechanism to form surface bound water (H2O*), as shown in panel (V). The
addition of a final proton completes the mechanism in panel (VI), resulting in des-
orbed methanol and water, with the overall process being slightly exergonic
( 209 kJ/mol). While this result highlights the role of the unique surface configura-
tion in driving the reaction, the effect of light further contributes to the catalytic
effect by enhancing the reactivity of the SFLP site for hydrogen activation, as
confirmed by previous time-dependent DFT results.50 Further studies, such as in-
elastic neutron scattering spectroscopy, are necessary to fully elucidate the reaction
mechanism.55

Undoubtedly, the most surprising discovery of our study is the production of solar
methanol with 50% selectivity through CO2 hydrogenation at ambient pressure.
In the high-pressure mechanism of CO2 hydrogenation to methanol catalyzed by
Cu/ZnO nanostructures, H2 is homolytically dissociated into H atoms.10,13 In
contrast, the SFLPs in defect-laden indium oxide, In2O3 x(OH)y, drives the splitting
of H2 heterolytically to form a reducing indium hydride and an acidic protonated in-
dium hydroxide group.46,49,50 It is the unique reactivity of the SFLP, in which the
Lewis acid becomes more acidic and the Lewis base site more basic in the excited
state, that enables the increase in both methanol and carbon monoxide production

1376 Joule 2, 1369–1381, July 18, 2018


rates under light conditions.50 This distinctive surface chemistry, in conjunction with
the longer-carrier lifetimes that arise from the nanocrystal superstructure nature of
the material, enable the remarkably high methanol formation rates observed under
ambient conditions.47

Conclusion
It has been demonstrated for the first time that a defect-laden indium oxide,
In2O3 x(OH)y, with a rod-like nanocrystal superstructure, under simulated solar
irradiation in the presence of H2, can reduce gaseous CO2 to methanol at atmo-
spheric pressure, with high selectivity and long-term stability. While this discovery
speaks well for the development of a low-pressure solar methanol process, much
research to understand the surface chemistry responsible for the production of
solar methanol from CO2 and H2 remains. This is a long-term project necessi-
tating the development of a range of operando spectroscopy techniques. These
include inelastic neutron scattering spectroscopy, IR, Raman, NMR, gas chroma-
tography-mass spectrometry (GC-MS), and TGA, which can simultaneously probe
catalyst structure and surface chemistry under reaction conditions. With the aid
of 2H and 13C isotope substitution and time domain DFT, it should prove
possible to elucidate the underlying mechanism of solar methanol formation on
nanostructured In2O3 x(OH)y, knowledge of which could enable the rational
design, engineering, and optimization of improved performance methanol
photocatalysts.

EXPERIMENTAL PROCEDURES
Synthesis of In2O3 x(OH)y Nanocrystals
In2O3 x(OH)y nanocrystals were synthesized through a reported recipe.45 In brief,
0.397 g of InCl3 was dissolved in 6 mL of ethanol. A mixed solution of 20% ammo-
nium (2.5 mL), 62.5% ethanol (7.5 mL), and 17.5% deionized water (2 mL) was then
added to the previous ethanol solution. The suspension with white precipitate was
immediately immersed into a pre-heated oil bath at 80 C with magnetic stirring
for 10 min. The suspension was then removed from the oil bath and allowed to
cool to room temperature. The precipitate was separated via centrifugation and
washed three times with deionized water. The white powder collected was dried
overnight at 80 C in a vacuum oven and calcined for 4 hr at 250 C to form
In2O3 x(OH)y nanocrystals.

Synthesis of In2O3 x(OH)y Nanocrystal Superstructures


In2O3 x(OH)y nanocrystal superstructures were synthesized through a modified re-
ported recipe.47 Four grams of urea and 0.5 g of InCl3 were dissolved in 45 mL of
deionized water. The aqueous solution was heated at 80 C under magnetic stirring
for different times. The precipitate was separated via centrifugation and washed
three times with deionized water. The white powder collected was dried overnight
at 80 C in a vacuum oven and calcined for 4 hr at 250 C to form a rod-like
In2O3 x(OH)y nanocrystal superstructure.

Characterization
PXRD was performed on a Bruker D2-Phaser X-ray diffractometer, using Cu Ka radi-
ation at 30 kV. The morphology of as-prepared samples was observed by scanning
electron microscopy (Zeiss, Supra 55). The detailed morphology of the In2O3 x(OH)y
nanocrystal and superstructures were collected with a Tecnai G20, FEI transmission
electron microscope, operating at 200 kV. The TEM samples were prepared by trans-
ferring one drop of sample dispersion onto a carbon-coated copper grid and then
drying in air. Nitrogen Brunauer-Emmet-Teller (BET) adsorption isotherms were

Joule 2, 1369–1381, July 18, 2018 1377


obtained at 77 K using a Quantachrome Autosorb-1-C. Diffuse reflectance of the
samples was measured using a Lambda 1050 UV/visible/near-infrared spectrometer
from PerkinElmer and an integrating sphere with a diameter of 150 mm. XPS was per-
formed using a PerkinElmer Phi 5500 ESCA spectrometer in an ultrahigh vacuum
chamber with base pressure of 1 3 10 9 Torr. The spectrometer uses an Al Ka
X-ray source operating at 15 kV and 27 A. The samples were coated onto carbon
tape, and all results were calibrated to C1s 284.5 eV. Photoconductivity measure-
ments utilized a white light source on a pressed powder film of nanocrystal super-
structures NR-3h, NR-9h, and NR-14h, weighing 1 mg, with electrical contact
made between two sputtered 500-nm-thick silver electrodes on a glass substrate
separated by 0.8 mm. The I-V curve of the nanorods in the reactor was measured un-
der the same conditions as the reaction (CO2:H2 = 1:3, 250 C) at atmospheric
pressure.

Gas-Phase Photocatalytic Measurements


The flow experiments were conducted in a plug flow capillary reactor, which con-
sisted of catalyst packed into a quartz tube. The quartz tube had an inner diameter
of 2 mm and was positioned on a groove carved out to accommodate the tube. A
capillary flow reactor was filled with 15 mg of sample and then introduced into
the test system. The sample bed of length 0.7 cm was fully irradiated with an
unfiltered 130 W Xe lamp. An OMEGA temperature controller was attached to a
heating cartridge inserted into the copper block along with a thermocouple in-
serted into the quartz tube in contact with the catalyst bed for control of the cata-
lyst temperature. A Newport 130 W Xe arc lamp was used to illuminate the catalyst
without any filter. The diameter of the light spot was about 2 cm, with an area of
about 3.14 cm2, which could fully cover the sample. CO2 and H2 were flowed
through at a ratio of 1:3 (2 mL/min and 6 mL/min) controlled by Alicat Scientific
digital flow controllers. The amounts of CO and methanol produced were deter-
mined using GC-MS (7890B and 5977A, He carrier, Agilent) and gas chromatog-
raphy (Agilent 7820A). In order to further confirm the production of methanol,
13
CO2 was used, and the product was collected in D2O. After 6 hr of collection,
the resulting D2O solution was measured by 1H NMR and 13C NMR to confirm
the formation of 13C methanol.

Computational Methods
DFT calculations were conducted using the Plane-Wave basis in the Quantum
Espresso software.56 All calculations were spin polarized and implemented using
the Perdew-Burke-Ernzerhof (PBE) exchange correlation functional, together with
the Rappe-Rabe-Kaxiras-Joannopoulos (RRKJ) ultrasoft pseudopotentials.57,58
Kinetic energy cutoffs were 50 Ry and 400 Ry for the wavefunctions and charge den-
sity, respectively, and the criterion for self-consistent field convergence was set to
1 3 10 6 Ry. Relaxation was performed using the conjugate gradient minimization
algorithm until the magnitude of the residual Hellman-Feynman force on each
atom was less than 13 10 3 Ry/Bohr. The In2O3 x(OH)y(111) surface was modeled
by a continuous four-layer slab with 161 atoms, roughly 11.5 Å in thickness to cap-
ture the effect of non-edge nanocrystal regions. The (111) surface has been shown
to be the most thermodynamically stable59 and hence the most reliably abundant
surface on nanocrystalline indium oxide. The choice of supercell is based on the re-
sults from a previously conducted DFT study in which the locations of oxygen defects
and hydroxyl groups in the In2O3(111) supercell were rigorously tested at all possible
sites to determine the most energetically favorable configuration.49 The location of
the SFLP site, characterized by the presence of a sterically hindered Lewis acid/base
pair, was confirmed using Bader charge analysis.60 We emphasize that the surface

1378 Joule 2, 1369–1381, July 18, 2018


model describes point defects, specifically, an O-vacancy and a terminal hydroxide
group, so as to focus on the local electronic effects resulting from the presence of the
SFLP site. While previous studies have used alternative computational methods
(e.g., embedded QM/MM) to capture a broader range of defects that can exist,
such as edge, step, or kink sites, this is not within the scope of the present study.61
Brillouin zone integrations were performed over the gamma point due to the large
size of the supercell. The bottom two layers were frozen in all calculations, while
the top two layers and adsorbates were allowed to relax. The enthalpy of
adsorption (Eads) was defined by Eads = Eadsorbate/surface (Eadsorbate + Esurface), where
Eadsorbate/surface is the total energy of the surface containing the adsorbed
species, Eadsorbate is the total energy of the free adsorbate, and Esurface is the total
energy the bare surface. The energetics of the overall reaction were computed by
DG = Esurface/H2O/CH3OH (Esurface + 3EH2 + ECO2).

SUPPLEMENTAL INFORMATION
Supplemental Information includes Supplemental Experimental Procedures,
16 figures, and 1 table and can be found with this article online at https://doi.org/
10.1016/j.joule.2018.03.007.

ACKNOWLEDGMENTS
We acknowledge financial support from the National Basic Research Program of
China (973 Program, grant no. 2013CB933500), the National Natural Science Foun-
dation of China (grant no. 51373188, 21401135), the Major Research Plan of the Na-
tional Natural Science Foundation of China (grant no. 91333208), the Natural Sci-
ence Foundation of Jiangsu Province (BK20160309, 1601042C), 111 project, the
Collaborative Innovation Center of Suzhou Nano Science & Technology, and the Pri-
ority Academic Program Development of Jiangsu Higher Education Institutions
(PAPD) and Qing Lan Project. G.A.O. is a Government of Canada Research Chair
in Materials Chemistry and Nanochemistry. Financial support for this work was pro-
vided by the Ontario Ministry of Research Innovation (MRI), Ministry of Economic
Development, Employment and Infrastructure (MEDI), Ministry of the Environment
and Climate Change, Connaught Innovation Fund, Connaught Global Challenge
Fund, Best in Science, and the Natural Sciences and Engineering Research Council
of Canada (NSERC).

AUTHOR CONTRIBUTIONS
L.W. and M.G. contributed equally to this work. G.A.O., X.Z., L.H., and L.W.
conceived and designed the experiments. L.W. and H.W. carried out the synthesis
of the materials and catalytic testing. L.W., Y.D., Y.S., H.L., M.G., J.Y.Y.L., W.S.,
M.X., Y.L., S.W., C. Qiu, T.W., C. Qian, and A.T. performed materials characteriza-
tions. M.G. performed DFT calculations. L.W., J.Y.Y.L., N.P.K., M.G., H.W., J.J.,
A.A.T., and T.E.W. contributed to data analysis. G.A.O., X.Z., and L.H. supervised
the project. G.A.O., X.Z., L.H., and L.W. wrote the paper. All authors discussed
the results and commented on the manuscript.

DECLARATION OF INTERESTS
The authors declare no competing interests.

Received: December 6, 2017


Revised: February 9, 2018
Accepted: March 9, 2018
Published: April 3, 2018; corrected online: April 13, 2018

Joule 2, 1369–1381, July 18, 2018 1379


REFERENCES
1. Olah, G.A. (2013). Towards oil independence 15. Porosoff, M.D., Yan, B., and Chen, J.G. (2016). 29. Albero, J., Garcia, H., and Corma, A. (2016).
through renewable methanol chemistry. Catalytic reduction of CO2 by H2 for synthesis Temperature dependence of solar light
Angew. Chem. Int. Ed. 52, 104–107. of CO, methanol and hydrocarbons: assisted CO2 reduction on Ni based
challenges and opportunities. Energy Environ. photocatalyst. Top. Catal. 59, 787–791.
2. Chang, C.D. (1983). Hydrocarbons from Sci. 9, 62–73.
methanol. Catal. Rev. 25, 1–118. 30. Jia, J., O’Brien, P.G., He, L., Qiao, Q., Fei, T.,
16. Wang, W., Wang, S., Ma, X., and Gong, J. Reyes, L.M., Burrow, T.E., Dong, Y., Liao, K.,
3. Haw, J.F., Song, W., Marcus, D.M., and (2011). Recent advances in catalytic Varela, M., et al. (2016). Visible and near-
Nicholas, J.B. (2003). The mechanism of hydrogenation of carbon dioxide. Chem. Soc. infrared photothermal catalyzed
methanol to hydrocarbon catalysis. Acc. Chem. Rev. 40, 3703–3727. hydrogenation of gaseous CO2 over
Res. 36, 317–326. nanostructured Pd@Nb2O5. Adv. Sci. (Weinh)
17. Bensaid, S., Centi, G., Garrone, E., Perathoner, 3, 1600189.
4. Kasatkin, I., Kurr, P., Kniep, B., Trunschke, A., S., and Saracco, G. (2012). Towards artificial
and Schlögl, R. (2007). Role of lattice strain and leaves for solar hydrogen and fuels from 31. Albero, J., Dominguez, E., Corma, A., and
defects in copper particles on the activity of carbon dioxide. ChemSusChem 5, 500–521. Garcia, H. (2017). Continuous flow
Cu/ZnO/Al2O3 catalysts for methanol
photoassisted CO2 methanation. Sustain.
synthesis. Angew. Chem. Int. Ed. 46, 7324– 18. Centi, G., Quadrelli, E.A., and Perathoner, S. Energy Fuels 1, 1303–1307.
7327. (2013). Catalysis for CO2 conversion: a key
technology for rapid introduction of renewable 32. Jia, J., Qian, C., Dong, Y., Li, Y.F., Wang, H.,
5. Behrens, M., Studt, F., Kasatkin, I., Kühl, S.,
energy in the value chain of chemical Ghoussoub, M., Butler, K.T., Walsh, A., and
Hävecker, M., Abild-Pedersen, F., Zander, S.,
industries. Energy Environ. Sci. 6, 1711–1731. Ozin, G.A. (2017). Heterogeneous catalytic
Girgsdies, F., Kurr, P., Kniep, B.L., et al. (2012).
The active site of methanol synthesis over Cu/ hydrogenation of CO2 by metal oxides: defect
19. Rongé, J., Bosserez, T., Martel, D., Nervi, C., engineering - perfecting imperfection. Chem.
ZnO/Al2O3 industrial catalysts. Science 336, Boarino, L., Taulelle, F., Decher, G., Bordiga, S.,
893–897. Soc. Rev. 46, 4631–4644.
and Martens, J.A. (2014). Monolithic cells for
solar fuels. Chem. Soc. Rev. 43, 7963–7981. 33. Jia, J., Wang, H., Lu, Z., O’Brien, P.G.,
6. Liao, F., Zeng, Z., Eley, C., Lu , Q., Hong, X., and
Tsang, S.C.E. (2012). Electronic modulation of a Ghoussoub, M., Duchesne, P., Zheng, Z., Li, P.,
20. Kim, D., Sakimoto, K.K., Hong, D., and Yang, P. Qiao, Q., Wang, L., et al. (2017). Photothermal
copper/zinc oxide catalyst by a heterojunction (2015). Artificial photosynthesis for sustainable
for selective hydrogenation of carbon dioxide catalyst engineering: hydrogenation of
fuel and chemical production. Angew. Chem. gaseous CO2 with high activity and tailored
to methanol. Angew. Chem. Int. Ed. 51, 5832– Int. Ed. 54, 3259–3266.
5836. selectivity. Adv. Sci. (Weinh) 4, 1700252.
21. Han, N., Wang, Y., Ma, L., Wen, J., Li, J., Zheng,
7. Kuld, S., Conradsen, C., Moses, P.G., 34. Yan, S.C., Ouyang, S.X., Gao, J., Yang, M.,
H., Nie, K., Wang, X., Zhao, F., Li, Y., et al.
Chorkendorff, I., and Sehested, J. (2014). Feng, J.Y., Fan, X.X., Wan, L.J., Li, Z.S., Ye, J.H.,
(2017). Supported cobalt polyphthalocyanine
Quantification of zinc atoms in a surface alloy Zhou, Y., et al. (2010). A room-temperature
for high-performance electrocatalytic CO2
on copper in an industrial-type methanol reactive-template route to mesoporous
reduction. Chem 3, 652–664.
synthesis catalyst. Angew. Chem. Int. Ed. 53, ZnGa2O4 with improved photocatalytic activity
5941–5945. 22. Yoon Suk Lee, L., and Wong, K.Y. (2017). in reduction of CO2. Angew. Chem. Int. Ed. 49,
Electrocatalytic reduction of carbon dioxide. 6400–6404.
8. Studt, F., Sharafutdinov, I., Abild-Pedersen, F.,
Chem 3, 717–718.
Elkjær, C.F., Hummelshøj, J.S., Dahl, S., 35. Dhakshinamoorthy, A., Navalon, S., Corma, A.,
Chorkendorff, I., and Nørskov, J.K. (2014). 23. Luo, J., Im, J.H., Mayer, M.T., Schreier, M., and Garcia, H. (2012). Photocatalytic CO2
Discovery of a Ni-Ga catalyst for carbon Nazeeruddin, M.K., Park, N.G., Tilley, S.D., Fan, reduction by TiO2 and related titanium
dioxide reduction to methanol. Nat. Chem. 6, H.J., and Grätzel, M. (2014). Water photolysis at containing solids. Energy Environ. Sci. 5, 9217–
320–324. 12.3% efficiency via perovskite photovoltaics 9233.
9. Schlögl, R. (2015). Heterogeneous catalysis. and Earth-abundant catalysts. Science 345,
1593–1596. 36. Izumi, Y. (2013). Recent advances in the
Angew. Chem. Int. Ed. 54, 3465–3520. photocatalytic conversion of carbon dioxide to
24. Wang, M., Wang, Z., Gong, X., and Guo, Z. fuels with water and/or hydrogen using solar
10. Kuld, S., Thorhauge, M., Falsig, H., Elkjær, C.F., energy and beyond. Coord. Chem. Rev. 257,
Helveg, S., Chorkendorff, I., and Sehested, J. (2014). The intensification technologies to
water electrolysis for hydrogen production – a 171–186.
(2016). Quantifying the promotion of Cu
catalysts by ZnO for methanol synthesis. review. Renew. Sustain. Energy Rev. 29,
573–588. 37. Li, R., Hu, J., Deng, M., Wang, H., Wang, X., Hu,
Science 352, 969–974. Y., Jiang, H.-L., Jiang, J., Zhang, Q., Xie, Y.,
11. Martin, O., Martı́n, A.J., Mondelli, C., Mitchell, 25. Dincer, I., and Acar, C. (2015). Review and et al. (2014). Integration of an inorganic
S., Segawa, T.F., Hauert, R., Drouilly, C., evaluation of hydrogen production methods semiconductor with a metal–organic
Curulla-Ferré, D., and Pérez-Ramı́rez, J. (2016). for better sustainability. Int. J. Hydrogen framework: a platform for enhanced gaseous
Indium oxide as a superior catalyst for Energy 40, 11094–11111. photocatalytic reactions. Adv. Mater. 26, 4783–
methanol synthesis by CO2 hydrogenation. 4788.
Angew. Chem. Int. Ed. 55, 6261–6265. 26. Meng, X., Wang, T., Liu, L., Ouyang, S., Li, P.,
Hu, H., Kako, T., Iwai, H., Tanaka, A., and Ye, J. 38. Zhao, Y., Chen, G., Bian, T., Zhou, C.,
12. An, B., Zhang, J., Cheng, K., Ji, P., Wang, C., (2014). Photothermal conversion of CO2 into Waterhouse, G.I.N., Wu, L.Z., Tung, C.H.,
and Lin, W. (2017). Confinement of ultrasmall CH4 with H2 over group VIII nanocatalysts: an Smith, L.J., O’Hare, D., and Zhang, T. (2015).
Cu/ZnOx nanoparticles in metal–organic alternative approach for solar fuel production. Defect-rich ultrathin ZnAl-layered double
frameworks for selective methanol synthesis Angew. Chem. 126, 11662–11666. hydroxide nanosheets for efficient
from catalytic hydrogenation of CO2. J. Am. photoreduction of CO2 to CO with water. Adv.
Chem. Soc. 139, 3834–3840. 27. O’Brien, P.G., Sandhel, A., Wood, T.E., Jelle, Mater. 27, 7824–7831.
A.A., Hoch, L.B., Perovic, D.D., Mims, C.A., and
13. Kattel, S., Ramı́rez, P.J., Chen, J.G., Rodriguez, Ozin, G.A. (2014). Photomethanation of 39. Zhao, Y., Jia, X., Waterhouse, G.I.N., Wu, L.Z.,
J.A., and Liu, P. (2017). Active sites for CO2 gaseous CO2 over Ru/silicon nanowire Tung, C.H., O’Hare, D., and Zhang, T. (2016).
hydrogenation to methanol on Cu/ZnO catalysts with visible and near-infrared Layered double hydroxide nanostructured
catalysts. Science 355, 1296–1299. photons. Adv. Sci. (Weinh) 1, 1400001. photocatalysts for renewable energy
production. Adv. Energy Mater. 6, 1501974.
14. Hansen, P.L., Wagner, J.B., Helveg, S., Rostrup- 28. Sastre, F., Puga, A.V., Liu, L., Corma, A., and
Nielsen, J.R., Clausen, B.S., and Topsøe, H. Garcı́a, H. (2014). Complete photocatalytic 40. Navalón, S., Dhakshinamoorthy, A., Álvaro, M.,
(2002). Atom-resolved imaging of dynamic reduction of CO2 to methane by H2 under solar and Garcia, H. (2013). Photocatalytic CO2
shape changes in supported copper light irradiation. J. Am. Chem. Soc. 136, 6798– reduction using non-titanium metal oxides and
nanocrystals. Science 295, 2053–2055. 6801. sulfides. ChemSusChem 6, 562–577.

1380 Joule 2, 1369–1381, July 18, 2018


41. Ahmed, N., Shibata, Y., Taniguchi, T., and 48. Hoch, L.B., Szymanski, P., Ghuman, K.K., He, L., Mott-Schottky heterojunctions for catalysis.
Izumi, Y. (2011). Photocatalytic conversion of Liao, K., Qiao, Q., Reyes, L.M., Zhu, Y., El- Chem. Soc. Rev. 42, 6593–6604.
carbon dioxide into methanol using zinc– Sayed, M.A., Singh, C.V., et al. (2016). Carrier
copper–M(III) (M=aluminum, gallium) layered dynamics and the role of surface defects: 55. O’Malley, A.J., Hitchcock, I., Sarwar, M.,
double hydroxides. J. Catal. 279, 123–135. designing a photocatalyst for gas-phase CO2 Silverwood, I.P., Hindocha, S., Catlow, C.R.A.,
reduction. Proc. Natl. Acad. Sci. USA 113, York, A.P.E., and Collier, P.J. (2016). Ammonia
42. Ahmed, N., Morikawa, M., and Izumi, Y. (2012). E8011–E8020. mobility in chabazite: insight into the diffusion
Photocatalytic conversion of carbon dioxide component of the NH3-SCR process. Phys.
into methanol using optimized layered double 49. Ghuman, K.K., Wood, T.E., Hoch, L.B., Mims, Chem. Chem. Phys. 18, 17159–17168.
hydroxide catalysts. Catal. Today 185, 263–269. C.A., Ozin, G.A., and Singh, C.V. (2015).
Illuminating CO2 reduction on frustrated Lewis 56. Paolo, G., Stefano, B., Nicola, B., Matteo, C.,
43. Kawamura, S., Zhang, H., Tamba, M., Kojima, pair surfaces: investigating the role of surface
T., Miyano, M., Yoshida, Y., Yoshiba, M., and Roberto, C., Carlo, C., Davide, C., Guido, L.C.,
hydroxides and oxygen vacancies on Matteo, C., Ismaila, D., et al. (2009). QUANTUM
Izumi, Y. (2017). Efficient volcano-type nanocrystalline In2O3-x(OH)y. Phys. Chem.
dependence of photocatalytic CO2 conversion ESPRESSO: a modular and open-source
Chem. Phys. 17, 14623–14635. software project for quantum simulations of
into methane using hydrogen at reaction
pressures up to 0.80MPa. J. Catal. 345, 39–52. 50. Ghuman, K.K., Hoch, L.B., Szymanski, P., Loh, materials. J. Phys. Condens. Matter 21, 395502.
J.Y.Y., Kherani, N.P., El-Sayed, M.A., Ozin,
44. Miyano, M., Zhang, H., Yoshiba, M., and Izumi, G.A., and Singh, C.V. (2016). Photoexcited 57. Perdew, J.P., Burke, K., and Ernzerhof, M.
Y. (2017). Selective photoconversion of carbon surface frustrated Lewis pairs for (1996). Generalized gradient approximation
dioxide into methanol using layered double heterogeneous photocatalytic CO2 reduction. made simple. Phys. Rev. Lett. 77, 3865–3868.
hydroxides at 0.40 MPa. Energy Technol. 5, J. Am. Chem. Soc. 138, 1206–1214.
892–900. 58. Kresse, G., and Joubert, D. (1999). From
51. Liu, X.M., Lu, G.Q., Yan, Z.F., and Beltramini, J. ultrasoft pseudopotentials to the projector
45. Hoch, L.B., Wood, T.E., O’Brien, P.G., Liao, K., (2003). Recent advances in catalysts for augmented-wave method. Phys. Rev. B 59,
Reyes, L.M., Mims, C.A., and Ozin, G.A. (2014). methanol synthesis via hydrogenation of CO 1758–1775.
The rational design of a single-component and CO2. Ind. Eng. Chem. Res. 42, 6518–6530.
photocatalyst for gas-phase CO2 reduction 59. Walsh, A., and Catlow, C.R.A. (2010). Structure,
using both UV and visible light. Adv. Sci. 52. Jia, Y., Xu, Y., Nie, R., Chen, F., Zhu, Z., Wang, stability and work functions of the low index
(Weinh) 1, 1400013. J., and Jing, H. (2017). Artificial photosynthesis surfaces of pure indium oxide and Sn-doped
of methanol from carbon dioxide and water via indium oxide (ITO) from density functional
46. Ghuman, K.K., Hoch, L.B., Wood, T.E., Mims, a Nile red-embedded TiO2 photocathode.
C., Singh, C.V., and Ozin, G.A. (2016). Surface theory. J. Mater. Chem. 20, 10438–10444.
J. Mater. Chem. A 5, 5495–5501.
analogues of molecular frustrated Lewis pairs
in heterogeneous CO2 hydrogenation 53. Yang, T.H., Huang, L.D., Harn, Y.W., Lin, C.C., 60. Henkelman, G., Arnaldsson, A., and Jónsson,
catalysis. ACS Catal. 6, 5764–5770. Chang, J.K., Wu, C.I., and Wu, J.M. (2013). High H. (2006). A fast and robust algorithm for Bader
density unaggregated Au nanoparticles on decomposition of charge density. Comput.
47. He, L., Wood, T.E., Wu, B., Dong, Y., Hoch, L.B., ZnO nanorod arrays function as efficient and Mater. Sci. 36, 354–360.
Reyes, L.M., Wang, D., Kübel, C., Qian, C., Jia, recyclable photocatalysts for environmental
J., et al. (2016). Spatial separation of charge purification. Small 9, 3169–3182. 61. French, S.A., Sokol, A.A., Bromley, S.T., Catlow,
carriers in In2O3–x(OH)y nanocrystal C.R.A., Rogers, S.C., King, F., and Sherwood, P.
superstructures for enhanced gas-phase 54. Li, X.H., and Antonietti, M. (2013). Metal (2001). From CO2 to methanol by hybrid
photocatalytic activity. ACS Nano 10, 5578– nanoparticles at mesoporous N-doped QM/MM embedding. Angew. Chem. Int. Ed.
5586. carbons and carbon nitrides: functional 40, 4437–4440.

Joule 2, 1369–1381, July 18, 2018 1381


JOUL, Volume 2

Supplemental Information

Photocatalytic Hydrogenation of Carbon


Dioxide with High Selectivity
to Methanol at Atmospheric Pressure
Lu Wang, Mireille Ghoussoub, Hong Wang, Yue Shao, Wei Sun, Athanasios A.
Tountas, Thomas E. Wood, Hai Li, Joel Yi Yang Loh, Yuchan Dong, Meikun Xia, Young
Li, Shenghua Wang, Jia Jia, Chenyue Qiu, Chenxi Qian, Nazir P. Kherani, Le He, Xiaohong
Zhang, and Geoffrey A. Ozin
Calculation of quantum efficiency
In our study, the methanol production reaction also proceeds thermally under dark conditions. One
plausible way to probe the light effect is to consider the rate difference between dark and light as the
light-induced methanol production rate. Based on this assumption, we estimated the quantum yield as
described below.
By convoluting the optical reflectance spectrum with the input photon flux spectrum (300-2000 nm
range) incident on an estimated exposed surface area of 0.22 cm2, we determined an absorbed photon
flux of 8.02×1021/h, 7.73×1021/h, 1.19×1022/h, 1.38×1022/h, for NC, NR-3h, NR-9h and NR-14h,
respectively.
This yields an effective CH3OH production rate of 1.7, 4.7, 20.8 and 43.2 µmol gcat-1h-1 (Rlight-Rdark)
and quantum efficiency of 0.013%, 0.037%, 0.11% and 0.19% for NC, NR-3h, NR-9h and NR-14h,
respectively.
Note however that due to the geometry of the reactor and light illumination, the bulk of the catalyst
nanoparticle sample is not actually accessible to the incident light.
With the assumptions:
(i) Incident light penetration only reaches a depth of about 1 µm into a 1 mm diameter capillary tube
reactor,
(ii) Light is incident only on one side of the reactor,
(iii) Light scattering and reflection losses are not taken into account,
then the true quantum efficiency could potentially be 100-500 times greater, in the range of 2-10% at the
reported temperature of 250°C. With improved photocatalyst and photoreactor architectures and
optimization of the solar light optics that maximizes light adsorption, these impressive quantum
efficiencies are within our grasp.
Table S1. Methanol production rate and selectivity over different catalysts in different systems.
Catalyst Conditions Methanol rate Methanol Reference
-1
(µmol h gcat-1) selectivity
Photocatalysis
Ag/Zr3Ga LDH ~0.8 MPa 0.1 3.17% 40
H2(0.28 MPa)
CO2 (0.12 MPa)
500 W Xe lamp
Cu/TiO2 ~0.8 MPa 0.056 2.92% 40
H2(0.28 MPa)
CO2 (0.12 MPa)
500 W Xe lamp
Pd/TiO2 ~0.8 MPa 0.004 0.01% 40
H2(0.28 MPa)
CO2 (0.12 MPa)
500 W Xe lamp
Zn-Cu-Al LDH H2 (1.67 mmol) 0.2 25.64% 38
CO2 (0.177 mmol)
500 W Xe lamp
Zn-Cu-Ga LDH H2 (1.67 mmol) 0.17 68.27% 38
CO2 (0.177 mmol)
500 W Xe lamp
Zn–Ga–CO3 LDH 0.4 MPa 0.19 48.71% 41
H2(0.28 MPa)
CO2 (0.12 MPa)
500 W Xe lamp
+
[Zn1.5Cu1.5Ga(OH)8] 2 H2 (1.67 mmol) 0.49 87.50% 39
2-
[Cu(OH)4] CO2 (0.177 mmol)
423 K
500 W Xe lamp
In2O3-x(OH)y H2 (6 ml/min) 63.73 53.03% This work
nanocrystal CO2 (2 ml/min)
superstructure 523 K
140 W Xe lamp
Other techniques
Cu/ γAl2O3 Plasma 25.6 mmol/L 54% S1
H2:CO2 = 3:1
Lattice strain- Thermal-523 K 0.94 mmol h-1gcat-1 CO production 5
Cu/ZnO/Al2O3 60bar reach the
Syngas mixture equilibrium
Ni5Ga3 Thermal-523 K Close to CO:CH3OH=15: 8
H2:CO2 = 3:1 commercial 1, no more than
Cu/ZnO catalyst 6.25%
Figure S1. High resolution TEM image of rod-like In(OH)3.

Figure S2. XPS of (a) Survey scan, high resolution (b) In3d, (c) O1s for all samples. All samples showed
similar spectra, only In2O3-x(OH)y nanocrystal has a lower amount of oxygen vacancies (OII) and higher
amount of hydroxides (OIII).
100

Weight percentage (%)


95

90

85

80
0 40 80 120 160 200 240
Time (min)
Figure S3. (a) Thermogravimetric analysis (TGA) of In(OH)3-14h at 250 °C for 4 hours.

Figure S4. (a) N2 isothermal adsorption-desorption plot, and (b) pore size distribution calculated by
NLDFT of the In2O3-x(OH)y nanocrystal and nanorod samples.
Figure S5. Preliminary temperature dependence of solar methanol production over In 2O3-x(OH)y
nanocrystal.

Figure S6. Methanol gas chromatography signal obtained at 250 °C with and without light.
Figure S7. (a) 1H and (b) 13C NMR obtained for product 13C methanol in D2O. The peak shown at 3.21
indicates the presence of 12C methanol from residue CO2 in the system and peak at 4.64 represents D2O
as solvent.

Figure S8. SEM images of In2O3-x(OH)y nanorod superstructures aging with (a) 3 hours and (b) 9 hours.
(c) TEM image of In2O3-x(OH)y nanocrystals.
Figure S9. PXRD patterns of the as-prepared In2O3-x(OH)y nanocrystal and nanorod samples. PDF Card
for In2O3 (00-006-0416) was inserted.

Figure S10. Diffuse reflectance spectra of the prepared In2O3-x(OH)y nanocrystal and nanorods.
Figure S11. I-V curves of the NR-3h, NR-9h and NR-14h under Xe lamp irradiation. Condition: 250 °C,
H2 : CO2 = 3 : 1, atmospheric pressure.

Figure S12. CO rate at 250°C with and without light irradiation.


Figure S13. PXRD pattern for NR-14h and NR-14h-used. NR-14h-used was obtained by testing NR-
14h at 250 °C under illumination for 20 hours. The crystallite size of produced In0 was calculated to be
~40 nm in NR-14h-used.

Figure S14. (a) TEM and (b) HRTEM images, and (c) SAED pattern of NR-14h-used. This sample was
obtained by testing NR-14h at 250 °C under illumination for 20 hours.
Figure S15. (a) SEM image of used NR-14h nanocrystal superstructure. EDX elemental mapping of (b)
O and (c) In. (d) High-resolution XPS spectra of In 3d for used and fresh NR-14h nanocrystal
superstructure.

Figure S16. Proposed reaction pathway for both reverse water gas shift and CO 2 hydrogenation.
Reference
1. Wang, L., Yi, Y., Guo, H., Tu, X. (2018). Atmospheric Pressure and Room Temperature Synthesis of
Methanol through Plasma-Catalytic Hydrogenation of CO2. ACS Catal. 8, 90-100.

You might also like