This is an open access article published under an ACS AuthorChoice License, which permits

copying and redistribution of the article or any adaptations for non-commercial purposes.

Research Article

Cite This: ACS Catal. 2017, 7, 7695-7703 pubs.acs.org/acscatalysis

Cobalt Spinel Nanocubes on N‑Doped Graphene: A Synergistic

Hybrid Electrocatalyst for the Highly Selective Reduction of Carbon
Dioxide to Formic Acid
Pandiaraj Sekar,† Laura Calvillo,† Cristina Tubaro,† Marco Baron,† Anuj Pokle,‡ Francesco Carraro,†
Alex Martucci,§ and Stefano Agnoli*,†
†
Department of Chemical Science, University of Padova, Via F. Marzolo 1 35131 Padova, Italy
‡
School of Physics and CRANN, Trinity College Dublin, Dublin 2, Ireland
§
Downloaded via NATL INST TECHLGY TIRUCHIRAPPALLI on August 27, 2018 at 11:33:39 (UTC).

Department of Industrial Engineering, University of Padova, via F. Marzolo 9, 35131 Padova, Italy
See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles.

*
S Supporting Information

ABSTRACT: Carbon dioxide reduction into useful chemical

products is a key technology to address urgent climate and energy
challenges. In this study, a nanohybrid made by Co3O4 and graphene
is proposed as an efficient electrocatalyst for the selective reduction of
CO2 to formate at low overpotential. A comparison between samples
with different metal oxide to carbon ratios and with or without
doping of the graphene moiety indicates that the most active catalyst
is formed by highly dispersed and crystalline nanocubes exposing
{001} oriented surfaces, whereas the nitrogen doping is critical to
obtain a controlled morphology and to facilitate a topotactic
transformation during electrocatalytic conditions to CoO, which
results in the true active phase. The nanohybrid made up by
intermediate loading of Co3O4 supported on nitrogen-doped
graphene is the most active catalyst, being able to produce 3.14
mmol of formate in 8 h at −0.95 V vs SCE with a Faradaic efficiency of 83%.
KEYWORDS: CO2 reduction, graphene, nanocomposites, metal oxide, electrocatalysts

1. INTRODUCTION Recently, nanostructured metal oxide based electrocatalysts

Due to its high energy efficiency and environmental have displayed quite interesting performances for the CO2RR,
in terms of selectivity, durability, and fast kinetics. Among the
sustainability, the electrochemical CO2 reduction reaction
various metal oxides, Co3O4-based catalysts have been
(CO2RR) is a promising method for converting readily
thoroughly studied as electrocatalysts for water oxidation,
available CO2 into added-value feedstocks such as CO, CH4,
oxygen reduction reaction, hydrogen evolution reaction,
HCOOH, and other small organic molecules.1−5 This process
etc.20−25 The origin of the electrochemical activity in
has the 2-fold benefit of consuming CO2, whose emission in the nanostructured spinel Co3O4 mainly stems from the presence
environment is posing serious concerns, and producing of mixed oxidation states (Co2+ in tetrahedral and Co3+ in
different types of fuels, possibly using renewable energy sources octahedral sites) and by the presence of highly active surface
such as solar radiation.6−8 Unfortunately, at the present state of sites.26 Gao et al. reported that ultrathin layers of Co3O4 show
technology, the sluggish kinetics and formation of multiple high CO2 reduction activity and are very selective for the
products in CO2RR hinder its large-scale application.9,10 formation of formic acid.27 The special activity was connected
Hence, great research efforts are currently being made to not only to the increased number of surface sites typical of low-
develop improved electrocatalysts with high product selectivity dimensional systems, but also to the peculiar electronic
at low overpotential.11 Among metals, Cu-, Au-, Pd-, and Ag- structure of the ultrathin films that endows the material with
based electrocatalysts have shown the highest conversion rates easy electronic transfer, fast carrier transport, and low corrosion
for the CO2RR.12−19 Unfortunately, these materials generally rate. Recently, the same group has reported that thin layers of
suffer from poor product selectivity, activity loss due to catalyst partially oxidized cobalt exhibit an excellent catalytic activity
poisoning, and relatively high cost and limited availability.
Therefore, the development of new electrocatalysts with high Received: July 3, 2017
selectivity and durability made up of cheap and abundant Revised: September 23, 2017
materials still remains a highly challenging task. Published: September 28, 2017

© 2017 American Chemical Society 7695 DOI: 10.1021/acscatal.7b02166

ACS Catal. 2017, 7, 7695−7703
ACS Catalysis Research Article

toward the CO2RR. The origin of such activity is not Scheme 1. Various Methods Involved in the Synthesis of
completely understood, but it seems to be a combination of NG-Co3O4 and RG-Co3O4 Electrocatalysts
nanoconfinement of metal atoms that are able to promote a
first electronic transfer to CO2 and the presence of an oxide
phase that favors the adsorption of CO2 and additionally
catalyzes the H+ transfer.28 The analysis of these works
therefore indicates that Co3O4 is a very promising catalyst, but
its properties such as poor electron conductivity, limited surface
area, and inefficient electron transfer have to be ameliorated by
a careful nanodesign. The same goal, however, can be achieved
in a simple and effective way by creating a composite material.
Following this approach, the electrical conductivity and surface
area problems could be overcome by coupling Co3O4 with
different carbon-based materials such as graphene, carbon
nanotubes, fibers, etc.29 Owing to its high surface area, electrical
conductivity, mechanical stability, and easily controllable The NPs preferentially expose the {100} surface plane, as may
surface chemistry, graphene is the best candidate to support be deduced from the lattice fringes with a periodicity of 0.289
metal oxide nanoparticles (NPs).30 In addition, the introduc- nm observed in Figure 1g, corresponding to {220} crystal
tion of heteroatoms (N, B, and S) into the graphene backbone planes, which are cutting at 45° the surface planes of the
could create additional catalytically active sites and provide a nanocubes. The origin of the cubic morphology can be ascribed
large amount of nucleation centers for the NPs.31−35 Moreover, to the use of cobalt acetate as a precursor without the addition
graphene can facilitate electron transport and improve the NP of any base. This leads to a synthesis environment with a low
electronic connection to the electrode. Actually, some of the concentration of hydroxyl ions, which favors the faster growth
best-performing catalysts for different electrochemical reactions of {111} over {100} facets, resulting in the formation of
such as the OER and ORR are based on graphene/Co3O4 cubes.45,46
nanocomposites.29 Quite interestingly, in a recent work, Wu et The TEM analysis also provides information about the effect
al.36 have reported that N-doped graphene (NG) shows some of the Co3O4 loading on the morphology of the Co3O4 NPs.
activity for the conversion of CO2 into CO due to the change in Figure 1b,e shows an increase in the average particle size (35−
the density of states created by doping.37 Moreover, the 40 nm) for the NG-Co3O4-40 sample, in comparison with NG-
combination of N-doped graphene with cobalt-based com- Co3O4-30, although the cubic shape is maintained. In this
pounds is particularly interesting for catalytic application, since sample, some agglomerates are also observed due to the higher
it mimics the active centers that are found in several natural loading of Co3O4. In high-magnification images, it is possible to
catalysts38 and, similarly, Co-porphyrin-based metal−organic distinguish clearly the presence of lattice fringes evidencing a
frameworks have demonstrated to be quite efficient for the lattice spacing value (0.282 nm), confirming that also in this
CO2RR.39−44 By considering all the aforementioned facts, case the nanocubes are terminated by {100} surfaces. The RG-
herein we have developed a simple and highly scalable Co3O4 sample, however, shows a more heterogeneous
procedure for the synthesis of a special nanocomposite that morphology, since both shapeless spherical and cubic Co3O4
comprises cubic spinel Co3O4 NPs and NGr (NG-Co3O4). NPs are observed in Figure 1c,f. In addition to the morphology,
This nanocomposite was optimized with the aim of combining another important difference deduced from the lattice fringe
the high catalytic activity of Co3O4 together with the excellent analysis of the RG-Co3O4 is the nature of the exposed surface.
electronic properties of NGr. Eventually, this approach has From high-resolution images, a lattice spacing value of 0.47 nm
produced a hybrid material that has shown a very high activity can be determined, indicating the presence of {111} terminated
toward the reduction of CO2 to formic acid with high selectivity NPs.47 This may be attributed to the different surface chemistry
at very low overpotential. of the supporting graphene sheets and in particular to the
absence of N functionalities, which has been reported in the
2. RESULTS literature to play an important role in the early stage of
Scheme 1 illustrates the synthesis protocols for the preparation nucleation and control of the NP shape.34,35,48 In fact,
of NG-Co3O4 and reduced graphene (RG)-Co3O4 nano- photoemission measurements have demonstrated that the
composites. This route takes advantage of a low-temperature pyridinic functionalities in NG may coordinate Co centers,
hydrothermal step for the preparation of intimately connected which therefore may act as nucleation centers for the growth of
cobalt oxide/graphene composites, which are characterized by the nanocubes.49 Such a strong interaction between the two
highly crystalline oxide NPs supported on highly dispersed materials is also expected to affect the final morphology of the
single- or few-layer graphene sheets. nanocomposite, given the possible modification in the
The morphology, size, and crystallinity of the Co3O4 NPs on interfacial stability between the two materials.35
NG were studied by TEM. Figure 1 shows representative TEM The NG-Co3O4 nanocomposites were also investigated by
images of the NG-Co3O4-30, NG-Co3O4-40, and RG-Co3O4-30 SEM (Figure S3 in the Supporting Information). Figure S3
samples, whereas the particle size distribution is reported in confirms the formation of Co3O4 nanocubes on a large scale,
Figure S1 in the Supporting Information. Figure 1a,d shows which are quite homogeneously dispersed on the graphene
that the Co3O4 NPs of NG-Co3O4-30 have a cubic shape with sheets. Some aggregates can be identified, especially in the case
an average size of 25−30 nm (Figure S1) and are uniformly of higher Co3O4 loading. The large-scale morphology of
distributed on the NG sheets. The Co3O4 nanocubes are highly graphene is the same for both the N-doped and RG samples.
crystalline, as confirmed by the selected area electron diffraction Figure 2 reports the Raman spectra of the graphene-Co3O4
(SAED) pattern (Figure S2b in the Supporting Information). nanocomposites. All of the samples show bands that are
7696 DOI: 10.1021/acscatal.7b02166
ACS Catal. 2017, 7, 7695−7703
ACS Catalysis Research Article

Figure 1. Low (a−c)- and high-magnification (d−f) TEM images as well as HR-TEM images (g−i) for the NG-Co3O4-30 (a, d, g), NG-Co3O4-40
(b, e, h) and RG-Co3O4-30 (c, f, i) samples. The inset in image (e) shows the presence of agglomerated NPs in NG-Co3O4-40.

characteristics of the Co3O4 spinel (at 193, 481, 617, and 690 (1.08) and is mainly due to the presence of N defects. When
cm−1; see the red stars), confirming the TEM results.50 No Co3O4 is introduced, the ID/IG ratio further increases up to
additional bands corresponding to different oxides or oxo- 1.35, indicating that the formation of Co3O4 creates additional
hydroxides of cobalt are detected, indicating that the synthesis defects in the carbon matrix, suggesting a strong interaction
procedure is very selective toward the production of pure spinel between the two moieties.
phase, in agreement with previous reports.51 The successful doping of graphene oxide (GO) with nitrogen
In addition, the samples containing graphene exhibit two and the chemical composition of the nanocomposites were
bands at 1319 and 1592 cm−1, which are attributed to the in-
investigated by XPS. The photoemission measurements of the
plane vibrations of the defective (D) and graphitic (G) domains
of the graphene matrix, respectively. The ratio between the NG-Co3O4-30 sample are reported in Figure 3 as an example,
intensity of the D and G bands (ID/IG) provides information while the data obtained for the other graphene-Co3O4 samples
about the degree of defectivity of the carbon material. The are shown in Figures S4−S6 in the Supporting Information.
positions of the D and G bands for NG are 1310 and 1592 The survey spectra show peaks corresponding only to the
cm−1 (black stars in Figure 2), respectively, and the intensity presence of elements such as C, N, O, and Co, indicating that
ratio (ID/IG) is 1.2, which is slightly higher than that of RG the nanohybrids do not contain impurities.
7697 DOI: 10.1021/acscatal.7b02166
ACS Catal. 2017, 7, 7695−7703
ACS Catalysis Research Article

the Co 2p photoemission line that is representative of all

samples. All of the Co 2p spectra show the same shape with
two peaks at 781 and 795 eV that are due to the spin−orbit
coupling. The multiplet structure and relative intensity of the
satellites of the Co 2p3/2 lines is characteristic of Co3O4 spinel,
as the analysis depicted in Figure 3c evidences.56,57 The
photoemission peak can be fitted with five components at
779.4, 780.9, 782.2, 785.5, and 789.8 eV, which are in good
agreement with the values reported in the literature for the
spinel type Co3O4 phase. As an example in Figure 3d we show
the O 1s spectrum of the NG-Co3O4-30 sample. Four
components can be clearly identified: the peak at 529.9 eV is
due to the O2− anions in Co3O4, the feature at 531.2 eV is
connected both to hydroxyl species on the oxides and carbonyl
groups present on NG, the component at 532.1 eV is indicative
of aliphatic alcohols (C(sp3)−OH), and finally the peak at
533.2 eV may be associated with phenol groups (C(sp2)-OH)
Figure 2. Raman spectra of RG-Co3O4-30 and NG-Co3O4 nano- and adsorbed water (see Table S1).54 Moreover, from XPS
composites with different oxide loadings. The Raman spectra for NG, measurements, the surface composition of the different samples
RG, and Co3O4 spinel are also shown as reference. was obtained. The results are given in Table S2 in the
Supporting Information. Hence, the XPS results strongly
suggest the formation of Co3O4 spinel and confirm the N
doping of graphene.
2.1. Electrochemical Analysis. The electrochemical
activity of the graphene-Co3O4 nanocomposites toward the
CO2RR was investigated by linear sweep voltammetry (LSV).
Figure 4a shows the LSV for the different electrocatalysts. At
potentials more negative than −1.1 V, H2 evolution takes place
under both N2- and CO2-saturated conditions. When the
electrolyte solution is saturated with CO2, a clear peak related
to the electroreduction of CO2 is observed at more positive
potentials (in comparison with the potential at which HER
occurs). For NG-Co3O4-20, the onset potential and the peak
potential are observed at −0.89 and −0.95 V, respectively. For
NG-Co3O4-30, the onset potential is shifted to −0.82 V and a
significant increase in the CO2 reduction current is obtained, in
comparison to NG-Co3O4-20. These values are comparable
with those for other advanced cobalt-based catalysts27 and are
even better (i.e., more positive) with respect to other non-
Figure 3. C 1s (a), N 1s (b), Co 2p (c), and O 1s (d) photoemission precious-metal catalysts.31,58,59 The peak current density of
lines for the NG-Co3O4-30 sample. The whole set of photoemission NG-Co3O4-30 (10.50 mA cm−2) is 2.5 times higher than that
measurements for the graphene-Co3O4 samples is reported in Figures obtained for NG-Co3O4-20 (4.1 mA/cm2). The better
S4−S6 in the Supporting Information. performance of this sample could be ascribed to the higher
loading of Co3O4 and, therefore, to the higher number of active
The deconvolution of the C 1s spectrum into chemically centers. However, a further increment in the loading of Co3O4
shifted components shows the presence of CC (284.5 eV), has a negative effect on the electrochemical activity, since a
C−N (285.6 eV), C−OH (286.3 eV) and CO (287.8 eV) decrease in the current density (6.30 mA cm−2) is observed for
species in all the NG-based samples, confirming the N-doping the NG-Co3O4-40 sample. This result may be attributed to the
(see Table S1 in the Supporting Information).52−54 The agglomeration of Co3O4 nanocubes observed by TEM, which
analysis of the N 1s photoemission line (Figure 3b) suggests may reduce the exposed active sites.
the presence of pyridinic, pyrrolic, and graphitic nitrogen Another important tool to evaluate the kinetics of the
species at BE values of 397.5, 399.1, and 400.7 eV, respectively, electrocatalytic reaction is the Tafel plot derived from the linear
as well as nitrogen oxides at 401.9 eV (see Table S1).52 The sweep voltammograms. Figure S7a in the Supporting
nitrogen doping percentage was calculated from the intensity of Information shows the Tafel plots for the NG-Co3O4 and
the N 1s and C 1s photoemission lines, taking into account the RG-Co3O4-30 catalysts. The lowest Tafel slope is displayed by
corresponding sensitivity factors, and resulted to be around NG-Co3O4-30 (97 mV dec−1), demonstrating its high intrinsic
3.9−4.9 atom %. The nature of the N species was determined catalytic activity with respect to NG-Co3O4-40 (121 mV dec−1),
as 49% pyridinic, 34% pyrrolic, and 12% graphitic N and 4.5% NG-Co3O4-20 (107 mV dec−1), and RG-Co3O4 (210 mV
nitrogen oxides. Among them, pyridinic groups are expected to dec−1).
act as anchoring sites for the growth of Co3O4 due to their A value of 118 mV/dec (as typical of bulk Co3O4) indicates
coordination capabilities.34 In addition, it has been reported in that the rate-determining step of the reaction is the initial
the literature that graphitic nitrogen is responsible for the transfer of one electron to CO2, whereas a Tafel slope of 59
electronic n-doping of the graphene sheets.55 Figure 3c shows mV/dec suggests a reaction path characterized by a reversible
7698 DOI: 10.1021/acscatal.7b02166
ACS Catal. 2017, 7, 7695−7703
ACS Catalysis Research Article

other samples such as NG-Co3O4-40 (3.85 cm2), NG-Co3O4-

20 (2.40 cm2), and RG-Co3O4 (1.56 cm2) (detailed ECSA
calculations are given in the Supporting Information, Table S3,
and Figure S9). Moreover, in order to get more precise
information on the specific activity, we normalized the current
observed at −0.95 V by the mass of cobalt determined by
photoemission measurements. Figure S8 in the Supporting
Information shows the comparative mass activity analysis for
the NG-Co3O4 and RG-Co3O4 samples with respect to the
loading of Co3O4. Among the catalysts, the highest mass
activity is observed for NG-Co3O4-30 (44.20 mA/mg of
Co3O4) in comparison to other catalysts such as NG-Co3O4-40
(20.05 mA/mg of Co3O4), NG-Co3O4-20 (29.50 mA/mg of
Co3O4), and RG-Co3O4 (9.2 mA/mg of Co3O4). Surprisingly,
NG-Co3O4-20 shows higher activity than NG-Co3O4-40,
demonstrating that it is possible to achieve a high catalytic
performance even with minimum catalyst loading and that an
excessive loading promotes the aggregation of NPs rather than
an effective increase in activity.
Electrolysis tests for CO2RR at different potentials in the
range between −0.9 and −1.2 V were conducted to quantify the
reaction byproducts and determine the stability of the catalysts
(Figure S7b in the Supporting Information). The analysis of the
electrolytic solution at different times of electrolysis by 1H
NMR has evidenced the formation of formate ions only, and no
other species bearing hydrogen groups (methanol or form-
aldehyde) (see Table 1 and Figure S10 in the Supporting
Information). The possible presence of volatile byproducts
produced by the CO2 reaction has been verified by gas
Figure 4. (a) Linear sweep voltammograms in N2 (dotted line)- and chromatography. Surprisingly, no other species bearing carbon
CO2-saturated (solid line) 0.1 M KHCO3 at room temperature and at such as CO and CH4 were detected, proving the high selectivity
a scan rate of 5 mV s−1. (b) Faradaic efficiency to formate at different
potentials for the graphene-Co3O4 nanocomposites. The currents
of the graphene-Co3O4 nanocomposites toward the production
shown in (a) are normalized by the geometric area. The corresponding of formate. These data are in good agreement with previous
data normalized by the oxide mass are shown in Figure S8 in the works from the literature62 showing that cobalt oxide based
Supporting Information. materials and, more in general, metal oxides (e.g., Ga2O3,
SnO2) may achieve good catalytic activity but are able to reduce
transfer of on electron to CO2 followed by a chemical rate- selectively CO2 only to formate, whereas other products such as
determining step, which may consist of the protonation of CO, methanol, and alkanes are not usually produced. In
CO2•− or migration to a different site on the electrode general, it is reported that the reaction path toward formate is
surface.60 favored by an adsorption of the CO2•− intermediates through
The lower Tafel slope values obtained for NG-Co3O4-30 and the oxygen atoms, whereas the formation of CO and other
NG-Co3O4-20 suggest that, as the nanocube dimensions more reduced products requires binding through the C atom.63
decrease, the formation of CO2•− is progressively facilitated. The i−t curves at −0.95 V are shown in Figure S7b in the
This is likely connected to the change of the electronic Supporting Information, whereas the calculated Faradaic
properties of nanocubes, which as they get smaller are more efficiency to formate at different applied potentials is reported
and more affected by the interaction with the substrate and may in Figure 4b. NG-Co3O4-30 displays a maximum Faradaic
present additional electronic states, as demonstrated for 1.72 efficiency of 83% at −0.95 V, which is comparatively higher
nm thick Co3O4 nanosheets27 and ultrathin CoOx layers.61 than those of NG-Co3O4-40 (72%) and NG-Co3O4-20 (64%).
The enhanced performance displayed by the NG-Co3O4 Quite interestingly, the Faradaic efficiency is only weakly
samples can be further explained by electrochemical active dependent on the applied potential, differently from other
surface area analysis. Among the prepared catalysts, NG-Co3O4- materials such as Pd64 and other metals65 where, on the other
30 shows the highest ECSA of 5.8 cm2 in comparison to the hand, the efficiency for the CO2RR quickly decreases at more

Table 1. Summary of the Electrochemical Performance of the NG-Co3O4-Based Electrocatalysts toward the CO2 Reduction
Reaction

onset peak current density

potential potential (mA/cm2 at peak overpotential quantity of formic acid at −0.95 V Faradaic efficiency of formate at −0.95 V
sample (V vs SCE) (V vs SCE) potential) (V) for 8 h (mmol) (%)
NG-Co3O4-20 −0.890 −0.950 4.1 0.295 1.59 64
NG-Co3O4-30 −0.820 −0.958 10.50 0.235 3.14 83
NG-Co3O4-40 −0.885 −0.962 6.30 0.284 2.16 72

7699 DOI: 10.1021/acscatal.7b02166

ACS Catal. 2017, 7, 7695−7703
ACS Catalysis Research Article

reducing potential because of competition with the hydrogen satellites characteristic of the Co3O4 spinel structure. After the
evolution reaction. synthesis, the sample was transferred to the electrochemical cell
Moreover, as seen in Figure S7b in the Supporting through the UHV-EC transfer system and polarized at −1.0 V
Information, NG-Co3O4-30 displays a stable performance in CO2-saturated 0.1 M KHCO3. Subsequently, the chemical
throughout the electrolysis without any variation in the current changes induced under reaction conditions were investigated by
density, while the Faradaic efficiency of 83% is maintained XPS. The analysis of the Co 2p3/2 peak of the sample polarized
(Figure 4b). NG-Co3O4-40 and NG-Co3O4-20 give lower at −1.0 V indicates a clear reduction of Co3O4 to CoO.
current densities, in comparison to NG-Co3O4-30. However, Therefore, for the first time, it is demonstrated that Co3O4 is
NG-Co3O4-40 shows a gradual increase in the current density not stable under CO2RR conditions, being reduced to CoO,
with time, which could be attributed to an initial activation which is the catalytically active phase for Co3O4-based systems.
(reduction of Co3O4 to CoO as seen below) under the action The electrochemical reduction of Co3O4 to CoO/Co was also
of a constant reducing potential. confirmed for the powders by performing CVs in the absence
To identify the real catalytic phase and determine the of CO2 (Figure S11). The reduction of Co3+ to Co2+ is attained
stability of Co3O4 under reaction conditions, we completed a in the potential range where the CO2 reduction takes place.
study performed on the powders with the investigation of the This finding is consistent with other works that clearly
Co3O4 model system in UHV following a previous work by demonstrated how the catalytic activity of several electro-
some of us.61 With this aim, we combined in situ photo- catalitic systems based on oxides is due to the “in situ”
emission measurements with electrochemistry to determine the formation of new phases.66,67
chemical changes of Co3O4 under CO2RR conditions. Forty This is in agreement with recent work of Gao et al.,28 which
monolayers (i.e., a film with a thickness of 8 nm) of Co3O4 shows that the CoO/Co intermixed phase is responsible for a
were grown on Pd(100) to avoid the exposure of the metal very high activity toward the ORR. In the present case the role
substrate to the electrolyte. The presence of a Co3O4 phase was of the metal phase is adopted by NG: as a matter of fact,
confirmed by low-energy electron diffraction and XPS nitrogen-doped carbon materials present metal-like behavior
measurements.61 Figure 5 shows the Co 2p region for the “as and are able to donate charge to supported nanoparticles,
prepared” sample. As seen above for the powders, the Co 2p3/2 favoring their reduction and stabilizing a low oxidation state.
peak exhibits a multiplet structure and relative intensity of the This tendency has been experimentally demonstrated for the
Co3O4/NG system where, by means of spatially resolved
spectroscopic measurements, a clear correlation was observed
between the presence of nitrogen atoms and the transformation
of Co3+ to Co2+.68 Previous works on other easily oxidizable
metals62,63 indicate that a subtle balance between oxide and
metal component is responsible for a highly efficient CO2
electroreduction,69 where the oxide is needed to reduce activity
for direct H+ reduction to H2 but nonetheless acts as a reservoir
of protons for the intermediate steps of the CO2 reduction.
In order to confirm the formation of such a mixed metal/
metal oxide phase also on powder samples, the NG-Co3O4-30
catalyst was investigated by Raman spectroscopy, XPS, GIXRD,
and SEM, comparing the results before and after electrolysis
(Figures S12−S14 in the Supporting Information). The cubic
morphology of the NPs is maintained after the electrochemical
treatment; however, photoemission measurements indicate that
the surface of the Co3O4 nanocubes, which is in contact with
the electrolyte, is reduced under reaction conditions to CoO
and even a small metal component is observed. Nonetheless,
the Co3O4 in the core remains stable as suggested by Raman
spectra. The same indication can be observed on comparison of
the GIXRD spectra before and after the electrochemical tests.
In the pristine material only the diffraction peaks of Co3O4 can
be observed, whereas after the electrochemical work, two new
peaks at 35.3 and 42.6°, attributed to CoO, become visible as
well as a faint shoulder at 44°, which may be ascribed to a
nascent metal phase.70
The transformation from Co3O4 to CoO provides an
indication of the key role played by the nanoparticle shape:
Co3O4 can be topotactically converted to CoO, since both
structures share the same oxygen sublattice and the octahedral
cations. From a morphological point of view, the {100} surfaces
Figure 5. Co 2p photoemission lines and corresponding deconvolu-
tion into chemical shifted components, for the Co3O4/Pd(100) are rather stable on both oxides, whereas {111} facets are stable
sample before (top) and after (bottom) the polarization at −1.0 V in only in the spinel phase, while they represent a highly unstable
CO2-saturated 0.1 M KHCO3 at room temperature. Yellow curves are polar termination for CoO. This suggests than an easy
the components related to the spinel phases whereas the green curves topotactic conversion can be observed for {100} oriented
are the components related to CoO. nanocubes, whereas for other shapes involving {111} facets, a
7700 DOI: 10.1021/acscatal.7b02166
ACS Catal. 2017, 7, 7695−7703
ACS Catalysis Research Article

more complicated morphological rearrangement would be cold water (500 mL) and H2O2 (3 mL). The resulting yellow
necessary. solid was washed several times with deionized water and
Overall these data confirm previous results on a cobalt-based subsequently separated by centrifugation (5000 rpm). Finally,
catalyst that has proven to be extremely selective for the the obtained suspension was dialyzed for 7 days to remove all
reduction of CO2 to formate and indicate that N-doped impurities and lyophilized to give GO as a solid.
graphene plays a crucial stabilization role, but without providing An 100 mg amount of GO was mixed with 500 mg of
an intrinsic reactivity (as also evidenced directly by the LSV of melamine and ground until a uniform powder was obtained.
bare NG reported in Figure 4a). Actually, it has been recently This mixture was transferred into an alumina crucible and
demonstrated that, on N-doped graphene foams, pyridinic- heated at 900 °C with a rate of 15 °C min−1 under an argon
based defects are able to electrochemically convert CO2 to CO atmosphere for 3 h in a tubular furnace. Subsequently, the
at rather low overpotential.36 Similarly, cobalt-based porphyrins crucible was cooled to room temperature. The resulting black
and porphyrin-like systems mostly selectively reduce CO2 to solid was dispersed in deionized water. Several washing and
CO.1,44 Because we are not observing any production of CO, centrifugation cycles were used to remove any impurities.
we may deduce that the real catalytic phase is represented by Finally, the resulting precipitate was dried at 60 °C overnight to
CoOx, and not some Co−N−C species. This demonstrates that give solid NG. Similarly, reduced graphene (RG) was obtained
the effect of nitrogen-doped graphene is indirect and affects by the same thermal treatment method without adding the
only the morphological and electronic properties of the melamine.
supported metal oxide, without a direct involvement in the 4.2. Synthesis of NG-Co3O4 Electrocatalysts. The
catalytic cycle. Interestingly, the sample NG-Co3O4-30 provides nanohybrid made up of Co3O4 NPs supported on NG sheets
a rather high Faradaic efficiency at a very low overpotential with was prepared by following the procedure reported in the
respect to other non-precious-metal-based catalysts such as literature with some modifications.49 Initially, 40 mg of NG and
ultrathin Co3O4 nanosheets,27 mixed metal oxides,71 copper 60 mg of cobalt acetate were dispersed in 15 mL of an ethanol/
nanofoams,9 copper oxides,59 or metal alloys.72 Therefore, its water (3/2 v/v) solution and sonicated for 30 min. The
catalytic performance is very promising in terms of both resulting dispersion was transferred into a Teflon-lined
selectivity and stability in comparison with those reported in hydrothermal reactor and heated at 140 °C for 12 h. The
the literature for other metal oxides and noble-metal NPs.58 resulting product was purified by cycles of washing with
deionized water and centrifugation and subsequently dried at
3. CONCLUSION 60 °C. The obtained sample was denoted as NG-Co3O4-30 (30
refers to the weight percent of cobalt determined from
In conclusion, we have developed a simple and potentially
photoemission analysis). Similarly, NG-Co3O4-20 and NG-
highly scalable synthetic protocol for the preparation of Co3O4/
Co3O4-40 samples were synthesized by using 30 mg of NG and
graphene nanohybrids that can be used as electrocatalysts for
40 mg of cobalt acetate and 20 mg of NG and 75 mg of cobalt
the CO2 reduction reaction. When nitrogen-doped graphene is
acetate, respectively. In addition, to understand the effect of N
used as a support for the metal oxide, a higher catalytic activity
doping, we prepared a nanohybrid using pure GO, namely RG-
is observed with respect to undoped graphene, indicating that
Co3O4-30, which was synthesized by using 40 mg of RG and 60
the doping has an important, though indirect, role. Actually, in
mg of cobalt acetate.
situ photoemission measurements demonstrate that the
4.3. Physical Characterization. X-ray photoelectron
nitrogen-doped graphene helps the transformation toward
spectroscopy (XPS) measurements were carried out in a
reduced CoO, which is the true active phase under electro-
custom-made UHV system equipped with an Omicron EA150
catalytic conditions.
electron analyzer and an Omicron DAR 400 X-ray source with
By a comparison of the activity of catalysts with different
a dual Al−Mg anode, working at a base pressure of 10−9 mbar.
cobalt oxide loading, it turns out that there is an optimal value
Core level photoemission spectra (C 1s, N 1s, O 1s, and Co 2p
of the ratio between active phase and support due to the
tradeoff between the maximum number of exposed cobalt sites regions) were collected at room temperature with a non-
and minimization of NP aggregation. The NG-Co3O4-30 monochromated Mg Kα X-ray source (1253.6 eV) and using an
catalyst shows the highest activity with an onset potential of energy step of 0.1 eV, 0.5 s integration time, and 20 eV pass
−0.82 V vs SCE with a peak current density of 10.50 mA/cm2. energy. The samples were dispersed in water and drop-casted
It also shows high selectivity toward formate formation with a onto a copper foil. Raman measurements were performed with
Faradaic efficiency of 83%. The combination of low over- a ThermoFisher DXR Raman microscope using a laser with an
potential and high selectivity makes this Co3O4/nitrogen-doped excitation wavelength of 780 nm (1 mW), with a 50× objective
graphene nanohybrid one of the best catalysts among non- (Olympus). Scanning electron microscopy (SEM) images were
precious metals for the CO2 reduction. obtained with a GEMINI column (Zeiss Supra VP35)
microscope, equipped with an EDX detector for elemental
analysis, operating at an acceleration voltage between 5 and 10
4. EXPERIMENTAL SECTION kV. High-resolution transmission electron microscopy (HR-
4.1. Synthesis of Nitrogen-Doped Graphene (NG). TEM) images were obtained using a TITAN (80-300) field
Graphene oxide (GO) was synthesized by following the emission microscope operated at an acceleration voltage of 300
modified Hummer method as reported in detail elsewhere.73 kV.
Initially, graphite and KMnO4 were mixed together by 4.4. Electrochemical Characterization. The electro-
mechanical grinding. The resulting powder mixture was slowly chemical measurements were carried out in a custom-made
added to 150 mL of a H2SO4/H3PO4 (9/1 v/v) solution kept two-compartment cell, where the anode and the cathode were
in an ice bath (0 to 5 °C) with stirring for 6 h. Then, the separated by a glass frit. A saturated calomel electrode (SCE)
reaction mixture was heated and maintained at 55 °C for 12 h. and a Pt wire were used as reference and counter electrodes,
The reaction was quenched by adding slowly a mixture of ice- respectively. All of the potentials in the text are referred to the
7701 DOI: 10.1021/acscatal.7b02166
ACS Catal. 2017, 7, 7695−7703
ACS Catalysis Research Article

SCE. The working electrode consisted of 20 μL of catalyst ink pumped into the EC cell through a tubing system at 1 mL
drop-casted onto a glassy-carbon (GC) electrode and dried in min−1 by a syringe pump (N-101 Pump System Inc.).
air. The catalyst inks were prepared by dispersing 2 mg of active The ultrahigh-vacuum (UHV) system and the electro-
phase in 0.5 mL of Milli-Q water and 10 μL of Nafion (8 vol % chemical cell used for this study are described in more detail
in alcohol/water solution, Sigma-Aldrich). Before ink deposi- in the Supporting Information.
tion, the GC electrode was polished with 1 and 0.25 μm
diamond paste and sonicated in isopropyl alcohol. A 0.1 M
KHCO3 solution, saturated with N2 or CO2 gases, was used as
■
*
ASSOCIATED CONTENT
S Supporting Information
electrolyte. During the electrochemical measurements, the cell The Supporting Information is available free of charge on the
was kept under a CO2 atmosphere with constant stirring. The ACS Publications website at DOI: 10.1021/acscatal.7b02166.
measurements were performed using a PGSTAT101 Autolab Additional characterizations by TEM, Raman spectros-
potentiostat running with NOVA. Cyclic (CV) and linear copy, and XPS, NMR spectra of the reaction product,
sweepvoltammetry (LSV) measurements were carried out and capacitance data for determining the surface area
between 0 and −1.1 V with a scan rate of 0.005 V s−1 in (PDF)

■
order to determine the activity of the samples toward the
CO2RR. The electrochemical active surface area (ECSA) was
calculated by measuring the roughness factor Rf of the working AUTHOR INFORMATION
electrode in the non-Faradaic region (a detailed description of Corresponding Author
the procedure is given in the Supporting Information). *E-mail for S.A.: [email protected].
4.5. Stability Tests and Product Analysis and ORCID
Quantification. The long-term stability of the catalysts and Laura Calvillo: 0000-0001-9256-0133
product analysis were determined by electrolysis experiments at Cristina Tubaro: 0000-0001-7724-735X
different potentials in the range between −0.9 to −1.2 V for 8 Francesco Carraro: 0000-0001-8485-4676
h. The liquid products were analyzed and quantified by 1H Stefano Agnoli: 0000-0001-5204-5460
NMR spectroscopy (Bruker Avance 300 MHz). The samples
for 1H NMR analysis were prepared by mixing an aliquot of the Notes
The authors declare no competing financial interest.

■
electrolytic solution (2.0 mL) with deuterated water (D2O, 0.4
mL) and using dimethyl sulfoxide (1 μL) as an internal
standard. The 1H NMR spectra were acquired with the water ACKNOWLEDGMENTS
presaturation method on a portion (0.7 mL) of the obtained This work was partially supported by the Italian MIUR through
mixture. The gaseous products (CO and CH4 ) were the national grant Futuro in Ricerca 2012 RBFR128BEC
determined by gas chromatography (Agilent Technologies “Beyond graphene: tailored C-layers for novel catalytic
7890A with a 5975C VL MSD mass spectrometer) using a materials and green chemistry” and PRIN project SMART-
flame ionization detector (FID) and thermal conductivity NESS (PRIN 2015K7FZLH) and by the University of Padova
detector (TID) and He as a carrier gas. The measurements funded project CPDA128318/12, “Study of the catalytic
were carried out by collecting with a syringe 100 μL of the activity of complex graphene nanoarchitectures from ideal to
gases from the headspace of the airtight electrochemical cell real conditions”. The MAECI (Ministero degli Affari Esteri e
during electrolysis and directly injecting it into the gas della Cooperazione Internazionale) is gratefully acknowledged
chromatograph. for the bilateral Italy−China GRAPE-MAT project. P.S.
4.6. Synthesis and Characterization of Co3O4/Pd(100) acknowledges Fondazione Cariparo for financial support. We
Model System. Thin films of Co3O4 on Pd(100) were thank Prof. Andrea Sartorel for the gas chromatography
measurements.

■
prepared by physical vapor deposition, by direct evaporation of
metal Co on Pd(100) under an oxygen atmosphere (PO2 = 2 ×
10−6 mbar) at 660 K using an evaporation rate of 0.35 ML REFERENCES
min−1 (1 ML corresponds to a thickness of 2.02 Å). (1) Lin, S.; Diercks, C. S.; Zhang, Y.-B.; Kornienko, N.; Nichols, E.
Subsequently, the sample was annealed under 10−3 mbar M.; Zhao, Y.; Paris, A. R.; Kim, D.; Yang, P.; Yaghi, O. M.; Chang, C. J.
oxygen partial pressure at the same temperature in order to Science 2015, 349, 1208−1213.
obtain a full oxidation. Previously to the Co deposition, the (2) Kuhl, K. P.; Cave, E. R.; Abram, D. N.; Jaramillo, T. F. Energy
Environ. Sci. 2012, 5, 7050.
Pd(100) substrate was cleaned by repeated cycles of sputtering (3) Costentin, C.; Passard, G.; Robert, M.; Saveant, J. M. J. Am.
with 2 keV Ar+ ions at 300 K and annealing at 900 K. The Chem. Soc. 2014, 136, 11821−11829.
residual carbon was removed in 2.0 × 10−7 mbar of oxygen at (4) Quadrelli, E. A.; Centi, G.; Duplan, J.-L.; Perathoner, S.
700 K and, subsequently, the adsorbed oxygen was removed by ChemSusChem 2011, 4, 1194−1215.
annealing at 900 K. (5) Kuhl, K. P.; Hatsukade, T.; Cave, E. R.; Abram, D. N.; Kibsgaard,
The chemical composition and chemical changes induced by J.; Jaramillo, T. F. J. Am. Chem. Soc. 2014, 136, 14107−14113.
the electrochemical measurements were analyzed by XPS, as (6) Zhu, D. D.; Liu, J. L.; Qiao, S. Z. Adv. Mater. 2016, 28, 3423−
described above for the powders. 3452.
The electrochemical measurements were performed in a (7) Yang, N.; Waldvogel, S. R.; Jiang, X. ACS Appl. Mater. Interfaces
2016, 8, 28357−28371.
home-built PEEK microcell. A Pt wire and an Ag/AgCl(3 M
(8) Lee, S. Y.; Lim, S. Y.; Seo, D.; Lee, J.-Y.; Chung, T. D. Adv. Energy
KCl) electrode placed in a Luggin capillary were used as Mater. 2016, 6, 1502207.
counter and reference electrodes, respectively. The EC (9) Sen, S.; Liu, D.; Palmore, G. T. R. ACS Catal. 2014, 4, 3091−
experiments were carried out in CO2-saturated 0.1 M 3095.
KHCO3 solution. The sample was polarized at −1.0 V vs (10) Li, F.; Zhao, S.-F.; Chen, L.; Khan, A.; MacFarlane, D. R.;
SCE for 600 s. During the measurement, the electrolyte was Zhang, J. Energy Environ. Sci. 2016, 9, 216−223.

7702 DOI: 10.1021/acscatal.7b02166

ACS Catal. 2017, 7, 7695−7703
ACS Catalysis Research Article

(11) Rosen, B. A.; Salehi-Khojin, A.; Thorson, M. R.; Zhu, W.; (44) Morlanés, N.; Takanabe, K.; Rodionov, V. ACS Catal. 2016, 6,
Whipple, D. T.; Kenis, P. J. A.; Masel, R. I. Science 2011, 334, 643− 3092−3095.
644. (45) Yu, X.-Y.; Meng, Q.-Q.; Luo, T.; Jia, Y.; Sun, B.; Li, Q.-X.; Liu,
(12) Kim, B.; Hillman, F.; Ariyoshi, M.; Fujikawa, S.; Kenis, P. J. A. J. J.-L.; Huang, X.-J. Sci. Rep. 2013, 3, 2886.
Power Sources 2016, 312, 192−198. (46) Geng, B.; Zhan, F.; Fang, C.; Yu, N. J. Mater. Chem. 2008, 18,
(13) Verma, S.; Lu, X.; Ma, S.; Masel, R. I.; Kenis, P. J. Phys. Chem. 4977−4984.
Chem. Phys. 2016, 18, 7075−7084. (47) Liang, H.; Raitano, J. M.; Zhang, L.; Chan, S.-W. Chem.
(14) Kortlever, R.; Peters, I.; Koper, S.; Koper, M. T. M. ACS Catal. Commun. 2009, 7569−7571.
2015, 5, 3916−3923. (48) Carraro, F.; Calvillo, L.; Cattelan, M.; Favaro, M.; Righetto, M.;
(15) Zhu, W.; Michalsky, R.; Metin, O.; Lv, H.; Guo, S.; Wright, C. J.; Nappini, S.; Píš, I.; Celorrio, V.; Fermín, D. J.; Martucci, A.; Agnoli, S.;
Sun, X.; Peterson, A. A.; Sun, S. J. Am. Chem. Soc. 2013, 135, 16833− Granozzi, G. ACS Appl. Mater. Interfaces 2015, 7, 25685−25692.
16836. (49) Singh, S. K.; Dhavale, V. M.; Kurungot, S. ACS Appl. Mater.
(16) Gao, D.; Zhou, H.; Wang, J.; Miao, S.; Yang, F.; Wang, G.; Interfaces 2015, 7, 442−451.
Wang, J.; Bao, X. J. Am. Chem. Soc. 2015, 137, 4288−4291. (50) Hadjiev, V.; Iliev, M.; Vergilov, I. J. Phys. C: Solid State Phys.
(17) Hall, A. S.; Yoon, Y.; Wuttig, A.; Surendranath, Y. J. Am. Chem. 1988, 21, L199.
Soc. 2015, 137, 14834−14837. (51) Yang, J.; Liu, H.; Martens, W. N.; Frost, R. L. J. Phys. Chem. C
(18) Feng, X.; Jiang, K.; Fan, S.; Kanan, M. W. J. Am. Chem. Soc. 2010, 114, 111−119.
2015, 137, 4606−4609. (52) Zhang, C.; Wang, B.; Shen, X.; Liu, J.; Kong, X.; Chuang, S. S.
(19) Oh, S.; Gallagher, J. R.; Miller, J. T.; Surendranath, Y. J. Am. C.; Yang, D.; Dong, A.; Peng, Z. Nano Energy 2016, 30, 503−510.
Chem. Soc. 2016, 138, 1820−1823. (53) Favaro, M.; Agnoli, S.; Di Valentin, C.; Mattevi, C.; Cattelan,
(20) Passard, G.; Ullman, A. M.; Brodsky, C. N.; Nocera, D. G. J. Am. M.; Artiglia, L.; Magnano, E.; Bondino, F.; Nappini, S.; Granozzi, G.
Chem. Soc. 2016, 138, 2925−2928. Carbon 2014, 68, 319−329.
(21) Manna, P.; Debgupta, J.; Bose, S.; Das, S. K. Angew. Chem., Int. (54) Ganguly, A.; Sharma, S.; Papakonstantinou, P.; Hamilton, J. J.
Ed. 2016, 55, 2425−2430. Phys. Chem. C 2011, 115, 17009−17019.
(22) Singh, S. K.; Dhavale, V. M.; Kurungot, S. ACS Appl. Mater. (55) Xing, Z.; Ju, Z.; Zhao, Y.; Wan, J.; Zhu, Y.; Qiang, Y.; Qian, Y.
Interfaces 2015, 7, 21138−21149. Sci. Rep. 2016, 6, 26146.
(23) Ramsundar, R. M.; Debgupta, J.; Pillai, V. K.; Joy, P. A. (56) Biesinger, M. C.; Payne, B. P.; Grosvenor, A. P.; Lau, L. W.;
Electrocatalysis 2015, 6, 331−340. Gerson, A. R.; Smart, R. S. C. Appl. Surf. Sci. 2011, 257, 2717−2730.
(24) Jin, H.; Wang, J.; Su, D.; Wei, Z.; Pang, Z.; Wang, Y. J. Am. (57) Tripathy, S. K.; Christy, M.; Park, N.-H.; Suh, E.-K.; Anand, S.;
Chem. Soc. 2015, 137, 2688−2694. Yu, Y.-T. Mater. Lett. 2008, 62, 1006−1009.
(25) Hamdani, M.; Singh, R.; Chartier, P. Int. J. Electrochem. Sci. (58) Wang, Z.-L.; Li, C.; Yamauchi, Y. Nano Today 2016, 11, 373−
2010, 5, 556. 391.
(26) Xie, X.; Shen, W. Nanoscale 2009, 1, 50−60. (59) Gupta, K.; Bersani, M.; Darr, J. A. J. Mater. Chem. A 2016, 4,
(27) Gao, S.; Jiao, X.; Sun, Z.; Zhang, W.; Sun, Y.; Wang, C.; Hu, Q.; 13786−13794.
Zu, X.; Yang, F.; Yang, S.; Liang, L.; Wu, J.; Xie, Y. Angew. Chem., Int. (60) Zhang, S.; Kang, P.; Meyer, T. J. J. Am. Chem. Soc. 2014, 136,
Ed. 2016, 55, 698−702. 1734−1737.
(28) Gao, S.; Lin, Y.; Jiao, X.; Sun, Y.; Luo, Q.; Zhang, W.; Li, D.; (61) Gragnaniello, L.; Agnoli, S.; Parteder, G.; Barolo, A.; Bondino,
Yang, J.; Xie, Y. Nature 2016, 529, 68−71. F.; Allegretti, F.; Surnev, S.; Granozzi, G.; Netzer, F. P. Surf. Sci. 2010,
(29) Liang, Y.; Li, Y.; Wang, H.; Zhou, J.; Wang, J.; Regier, T.; Dai, 604, 2002−2011.
H. Nat. Mater. 2011, 10, 780−786. (62) Zhang, L.; Zhao, Z.-J.; Gong, J. Angew. Chem., Int. Ed. 2017, 56,
(30) Shearer, C. J.; Cherevan, A.; Eder, A. D. Adv. Mater. 2014, 26, 11326.
2295−2318. (63) Kortlever, R.; Shen, J.; Schouten, J. J. P.; Calle-Vallejo, F.;
(31) Lei, F.; Liu, W.; Sun, Y.; Xu, J.; Liu, K.; Liang, L.; Yao, T.; Pan, Koper, M. T. M. J. Phys. Chem. Lett. 2015, 6, 4073−4082.
B.; Wei, S.; Xie, Y. Nat. Commun. 2016, 7, 12697. (64) Min, X.; Kanan, M. W. J. Am. Chem. Soc. 2015, 137, 4701−4708.
(32) Spilarewicz-Stanek, K.; Kisielewska, A.; Ginter, J.; Baluszynska, (65) Humphrey, J. J. L.; Plana, D.; Celorrio, V.; Sadasivan, S.; Tooze,
K.; Piwonski, I. RSC Adv. 2016, 6, 60056−60067. R. P.; Rodríguez, P.; Fermín, D. J. ChemCatChem 2016, 8, 952.
(33) Sreekanth, N.; Nazrulla, M. A.; Vineesh, T. V.; Sailaja, K.; Phani, (66) Celorrio, V.; Dann, E.; Calvillo, L.; Morgan, D. J.; Hall, S. R.;
K. L. Chem. Commun. 2015, 51, 16061−16064. Fermín, D. J. ChemElectroChem 2016, 3, 283.
(34) Wang, R.; Xu, C.; Sun, J.; Gao, L. Sci. Rep. 2015, 4, 7171. (67) Celorrio, V.; Calvillo, L.; Dann, E.; Granozzi, G.; Aguadero, A.;
(35) Duan, J.; Chen, S.; Dai, S.; Qiao, S. Z. Adv. Funct. Mater. 2014, Kramer, D.; Russel, A. E.; Fermín, D. J. Catal. Sci. Technol. 2016, 6,
24, 2072−2078. 7231.
(36) Wu, J.; Liu, M.; Sharma, P. P.; Yadav, R. M.; Ma, L.; Yang, Y.; (68) Wang, J.; Zhou, J.; Hu, Y.; Regier, T. Energy Environ. Sci. 2013,
Zou, X.; Zhou, X.-D.; Vajtai, R.; Yakobson, B. I.; Lou, J.; Ajayan, P. M. 6, 926−934.
(69) Lee, C. H.; Kanan, M. W. ACS Catal. 2015, 5, 465−469.
Nano Lett. 2016, 16, 466−470.
(70) Nam, K. M.; Shim, J. H.; Han, D. W.; Kwon, H. S.; Kang, Y. M.;
(37) Xu, J.; Kan, Y.; Huang, R.; Zhang, B.; Wang, B.; Wu, K. H.; Lin,
Li, Y.; Song, H.; Seo, W. S.; Park, J. T. Chem. Mater. 2010, 22, 4446−
Y.; Sun, X.; Li, Q.; Centi, G.; Su, D. ChemSusChem 2016, 9, 1085−
4454.
1089.
(71) Yadav, V. S. K.; Purkait, M. K. New J. Chem. 2015, 39, 7348−
(38) Breslow, R. Acc. Chem. Res. 1995, 28, 146−153.
7354.
(39) Hu, X.-M.; Rønne, M. H.; Pedersen, S. U.; Skrydstrup, T.;
(72) Choi, S. Y.; Jeong, S. K.; Kim, H. J.; Baek, I.-H.; Park, K. T. ACS
Daasbjerg, K. Angew. Chem., Int. Ed. 2017, 56, 6468.
Sustainable Chem. Eng. 2016, 4, 1311−1318.
(40) Behar, D.; Dhanasekaran, T.; Neta, P.; Hosten, C. M.; Ejeh, D.;
(73) Marcano, D. C.; Kosynkin, D. V.; Berlin, J. M.; Sinitskii, A.; Sun,
Hambright, P.; Fujita, E. J. Phys. Chem. A 1998, 102, 2870−2877.
Z.; Slesarev, A.; Alemany, L. B.; Lu, W.; Tour, J. M. ACS Nano 2010, 4,
(41) Kornienko, N.; Zhao, Y.; Kley, C. S.; Zhu, C.; Kim, D.; Lin, S.;
4806−4814.
Chang, C. J.; Yaghi, O. M.; Yang, P. J. Am. Chem. Soc. 2015, 137,
14129−14135.
(42) Cheng, M.-J.; Kwon, Y.; Head-Gordon, M.; Bell, A. T. J. Phys.
Chem. C 2015, 119, 21345−21352.
(43) Yao, S. A.; Ruther, R. E.; Zhang, L.; Franking, R. A.; Hamers, R.
J.; Berry, J. F. J. Am. Chem. Soc. 2012, 134, 15632−15635.

7703 DOI: 10.1021/acscatal.7b02166

ACS Catal. 2017, 7, 7695−7703