“This document is the Accepted Manuscript version of a Published Work that appeared in final form in
Chemistry of Materials, copyright © American Chemical Society after peer review and technical editing by the
publisher. To access the final edited and published work see [insert ACS Articles on Request author- directed link
to Published Work, see
http://pubs.acs.org/doi/abs/10.1021/cm503258z

In-situ Formation of Heterojunctions in Modified Graphitic Carbon
Nitride: Synthesis and Noble Metal Free Photo-catalysis
Menny Shalom†§*, Miguel Guttentag Ɨ§, Christian Fettkenhauer†, Sahika Inal±, Dieter Neher±,
Antoni Llobet Ɨ, Markus Antonietti†
†

Max-Planck Institute of Colloids and Interfaces, Department of Colloid Chemistry, Research

Campus Golm, 14424 Potsdam, Germany
Ɨ

Institut Català d’Investigació Química, Paisos Catalans 16, Tarragona, 43007, Spain

±

Institute of Physics and Astronomy, University of Potsdam, Karl-Liebknecht-Str. 24-25,

Potsdam, 14476, Germany
E-mail address: menny.shalom@ mpikg.mpg.de
Herein we report the facile synthesis of an efficient roll-like carbon nitride (C3N4)
photocatalyst for hydrogen production using a supramolecular complex composed of
cyanuric acid, melamine and barbituric acid as the starting monomers. Optical and
photocatalytic investigations show, along with the known red shift of absorption into the
visible region, that the insertion of barbituric acid results in the in situ formation of in-plane
heterojuctions, which enhance the charge separation process under illumination. Moreover,
platinum as the standard co-catalyst in photocatalysis could be successfully replaced with
first row transition metal-salts and complexes under retention of 50 % of the catalytic
activity. Their mode of deposition and interaction with the semiconductor was studied in
detail. Utilization of the supramolecular approach opens new opportunities to manipulate
charge transfer process within carbon nitride toward the design of more efficient carbon
nitride photocatalyst with controlled morphology and optical properties.
1

1. Introduction
The production of solar fuels through water splitting is highly desired due to the increasing
demands for clean and renewable energy.1 In general, the semiconductor photocatalyst
should possess a suitable band gap (< 1.23eV) with the right position of conduction and
valence band.2 In addition, the photocatalytic performance is strongly dependent on the
charge transfer processes of the semiconductor under illumination (electrons and holes
separation and internal recombination rate). One of the ways to manipulate the charge
separation of excitons in many light-driven reactions (e.g. in solar cells,3 light-emitting
diodes4 or for photocatalysis5) is the formation of a heterojunction which can be carefully
designed in order to direct the charge flow in a given direction. While most of the research in
this field was focused on metal-based semiconductors (metal oxides,6 sulfides7 and nitrides8)
as photocatalysts, in the last years metal-free graphitic carbon nitride (C3N4) materials have
attracted widespread attention due to their outstanding (electro)catalytic and photocatalytic
activity.9 C3N4 materials demonstrate high activity in many energy-related fields: as catalyst
in heterogeneous catalysis and as metal-free photocatalysts and electrocatalysts in watersplitting reactions,10,

11

in the degradation of pollutants,12 and many more. The growing

interest in C3N4 materials as photo- and electro-catalyst is mainly due to its excellent
chemical and photophysical properties. Apart of its high chemical and thermal stability (up
to 400°C), energy level positions of C3N4 materials are suitable for water splitting under
illumination (both proton reduction and water oxidation). The photocatalytic activity of C3N4
is strongly dependent on its morphology, size, surface area, defects, and energy states along
with its electronic properties (band gap, exciton lifetime etc.). In the last year, many
modifications of the common synthesis were developed, including the introduction of new
heteroatoms (sulfur, phosphor)13 into the C3N4 structure, in order to manipulate its
electronic, optical and catalytic properties. However, despite great progress in C3N4
synthesis, the traditional solid-state synthesis of C3N4 results in disorganized texture, with
small grain size and surface defects, which leads to low catalytic activity.
2

Recently, other groups and ours used the supramolecular chemistry approach to synthesize
ordered structures of C3N4 such as hollow boxes, spheres and spherical macroscopic
assemblies.12,

14

In addition, this approach can also be applied to grow C3N4 on different

substrates.10 Supramolecular chemistry provides a great opportunity for the synthesis of
nanostructured materials without any further templating techniques. The supramolecular
approach includes the use of non-covalent interactions such as hydrogen bonding to form
order between building blocks for the desired synthesis. Hydrogen bonds are very useful for
controlling molecular self-assembly thanks to the reversibility, specificity, and directionality
of this class of interactions.15,

16

The structure of the final products can be controlled by

choosing the appropriate monomers and solvents for the synthesis. The starting monomers
will organize in different structures according to their ability to form hydrogen bonds in the
given solvent and form ordered and stable aggregates which consecutively define the
resulting materials.
Here we use the supramolecular approach in order to improve the efficiency of the C3N4
photocatalyst for hydrogen evolution. The modified carbon nitride was prepared by the
pyrolysis of the supramolecular complex of cyanuric acid (C), melamine (M), and barbituric
acid (B) as precursors in different molar ratios. We demonstrate that changes in the
hydrogen bond network and the replacement of nitrogen atoms with carbon strongly
influenced the electronic and the chemical properties of C3N4. Moreover, lifetime
measurements indicated the in situ generation of heterojunctions, which are favorable for
photocatalytic applications. The chemical structure, morphology and optical properties of
the resulting carbon nitrides were characterized by X-ray diffraction (XRD), Fourier
transformed infrared (FT-IR) spectroscopy, scanning electron microscopy (SEM), Elemental
Analysis, UV-vis absorption, steady-state and time-resolved fluorescence spectroscopy. The
photocatalytic activities of the different modified C3N4’s were tested by measuring hydrogen
evolution in combination with Pt, first row transition metal salts and known water reduction
catalysts as co-catalysts.
2. Results and discussion
Supramolecular CM1BX complexes were prepared by mixing different molar ratio of the three
precursors in water (equimolar amounts of melamine and cyanuric acid = CM1, x = 0.0033,
0.05, 0.1, 0.5 molar ratio of barbituric acid with respect to CM1. CM:B = 1:x). Although the
3

solubility of the three monomers in water is generally low, it was sufficient to allow for the
formation of organized aggregates after 4h of shaking (Figure S1). In the absence or in low
amounts of barbituric acid (< 0.1) the formation of rod-like aggregates of CMB complex was
observed, while in high amount the aggregates in water exhibited a disk-like structure. The
change in morphology with higher barbituric acid concentration was probably due to the
removal of one hydrogen bond structural unit (when comparing barbituric acid with cyanuric
acid), which allowed for more flexibility during the supramolecular complex formation.
Evidence for the formation of high crystallinity supramolecular complexes is given by the FTIR spectroscopy and XRD analysis (Figure S2).16 The formation of hydrogen bonds was
reflected in the FT-IR spectra by the shift of the stretching vibration of the C=O groups in the
cyanuric and barbituric acid from 1691 to 1730 cm-1 and by the shift of the triazine ring
vibration of melamine to lower wavenumbers (from 812 to 761 cm-1). In addition, the
stretching vibration of the N-H group in melamine disappeared. The appearance of new XRD
peaks for CMB compared with the starting monomers is strong evidence for the creation of
new arrangements (Figure S2b). The indication for an in-plane pattern in CM1Bx is shown by
the well-resolved peaks at 10.7, 19.8 and 21.75°, which can be indexed as (100), (110) and
(200) of the in-planar packing, respectively. Photoactive carbon nitride materials were
obtained by heating the CMB complexes to 550 °C for 4 hours under nitrogen atmosphere
with a heating ramp of 2.3 K/min. The effect of barbituric acid on texture is clearly observed
in the SEM images of all CM1BX-C3N4 samples (Figure 1).

4

Figure 1. Graphic representation of the CM1Bx supramolecular complex (see also scheme S1) and SEM images
of the hydrogen-bonded complexes CM1BX -C3N4 in different molar ratios after condensation at 550 °C for 4
hours. Scale bars are 1 µM.

The morphology changed from sheet-like to a roll-like structure, even with low amounts of
barbituric acid, due to the local removal of one hydrogen bond. We speculate that the
removal of one hydrogen bond introduced more adaptability to the complex by adding more
degrees of freedom, which resulted in the ability to create ordered 3D structures. At higher
barbituric acid concentration in the starting complex, the roll-like morphology was partially
destroyed, and a less regular nanostructured morphology was obtained. This change was
also reflected by the BET surface area of all the CM1BX-C3N4s materials (Figure S3). In the
absence and with a low amount of barbituric acid in the starting complex the surface area
was relatively high (60-70 m2 g-1), while its further increase resulted in a surface area
decrease. The formation of carbon nitride after heating was confirmed by FT-IR and XRD
analyses (Figure 2).17 The vibrational spectra of the CM1BX-C3N4 materials were compared to
the one that condensed solely from dicyandiamide (DCDA). The strong stretching modes of
CN heterocycles from 1200 to 1600 cm-1 alongside the characteristic breathing mode of the
triazine units at 800 cm-1 were observed, confirming the formation of C3N4. Additional proof
for the C3N4 formation is given by the XRD patterns (Figure 2b). The pattern of the carbon
5

nitride which was condensed from CM1BX is similar to the one which was prepared by DCDA
condensation. The strong interplanar stacking peak of aromatic systems around 27.2°
(indexed as 002) was shifted to lower degrees and became less pronounced and broader
with additional barbituric acid. The shift of the interplanar stacking peak can be explained by
higher defect rate in the CM1BX-C3N4s and by the polymer-like growth of CM1BX-C3N4
compared with the grain-growth in the DCDA route.

Figure 2. FT-IR spectra (a) and X-ray diffraction pattern (b) of CM1Bx after condensation at 550 °C for 4 hours,
indicating the formation of C3N4.

The formation of C3N4-like materials was also confirmed by elemental analysis (Table S1). For
all the CM1BX-C3N4 materials, the C/N molar ratio varied from 0.7-0.77 (the increase of
carbon content was due to a larger amount of barbituric acid in the starting complex) with
~2% hydrogen only, meaning that most of the NH2 groups reacted during the condensation
process. In addition, the total weight of all the elements clearly indicated the absence of
oxygen, implying that most if not all the oxygen from the C=O groups in the cyanuric and
barbituric acid were eliminated.
In order to evaluate the changes in the electronic properties of the modified C3N4s, we
measured their photophysical properties. In general, different morphologies, composition,
surface states, and defect sites can directly influence the electronic properties of C3N4. The
UV-vis diffuse reflectance absorbance spectra of all the CM1BX-C3N4 materials are shown in
Figure 3a. For all the modified CM1BX-C3N4 materials, the absorption edges were red-shifted
to lower energies along with an increase of optical densities at longer wavelengths. The red
shift in absorption spectra indicates an alteration of the electronic band structure due to the
replacement of nitrogen atoms by carbon atoms, as was previously described by Wang et
al18. Further changes in the electronic properties are shown in the photoluminescence (PL)
spectra of the CM1BX-C3N4 (Figure 3b). As observed in the absorption spectra, the emission
6

peak of all the CM1BX-C3N4 materials shifted to lower energy compared to that of the CMC3N4. In addition, the emission intensity is strongly quenched for almost all the CM1BX-C3N4
materials, probably due to different charge transfer process as will be discussed later.

Figure 3. Diffuse reflectance absorption spectra (a) and emission spectra (b) of CM1Bx-C3N4 obtained after
condensation at 550 °C for 4 hours.

In order to evaluate the photo-activity of the CM1BX-C3N4 materials with respect to the bulk
and porous C3N4, we measured the hydrogen evolution in a water/triethanolamine (TEOA)
solution and with Pt as co-catalyst under white light illumination. Considering the low
abundance and high cost of platinum, we further conducted a detailed investigation of
CM1B0.05 in combination with different known molecular water reduction catalysts (WRCs)
and a variety of first row transition metal salts as co-catalysts. In particular detail was
investigated the interaction of CM1B0.05 with Co(BF4)2 and the nature of the catalytic active
species in this system.
All of the CM1BX-C3N4 materials demonstrated higher catalytic activity with Pt as co-catalyst
compared to the bulk and CM- C3N4. Moreover, the best photocatalyst (CM1B0.05) shows
more than 60 % higher turnover frequency (TOF) compared to mesoporous C3N4, which is
one of the most photo-active C3N4 modifications reported up to date. In addition, the
photocatalyst demonstrated high stability in the hydrogen evolution reaction, as shown in
Figure S4 and S6. The enhanced photocatalytic activity can originate from several factors: (1)
better light harvesting which can lead to more charge carriers that can be used in the
catalytic process (2) improvement of the charge separation process (with appropriate energy
band positioning), and (3) high surface area. From the absorption spectra and the hydrogen
production rate it can be estimated that there is no direct correlation between the light
harvesting and the photocatalytic activity (e.g. the hydrogen productions of CM1B0.1/0.5 are
lower compared to CM1B0.05/0.01 although they absorb higher amount of light). In order to
7

further study the charge separation process, we employed time-resolved fluorescence
spectroscopy with a single photon counting set-up19. The intensity-averaged fluorescence
lifetimes of all the C3N4 materials are shown in Figure 4b. In accordance with the trend in the
steady-state PL spectra, the fluorescence lifetime decreased for all the CM1Bx- C3N4s
compared to CM. The quenching of the emission intensity and its lifetime indicate that the
relaxation of a fraction of CM1BX-C3N4 excited states occurs via non-radiative paths,
presumably through charge transfer of electrons and holes to new localized/surface states.
Furthermore, for the most active materials, when the monitoring wavelength was changed
from 475 nm to 525 nm, the average fluorescence lifetime varied as well, which might point
to the presence of more than one type of emitters. The variation of the lifetime suggests
that at low barbituric acid concentration the C3N4 materials was composed of a distribution
of different semiconducting domains with different band gaps (probably due to the low
amount of barbituric acid in the starting complex). The in situ creation of such domains can
form heterojunctions at the grain boundaries, which facilitate the charge transfer processes
and consequently enhance the photocatalytic activity (as illustrated in Figure S5).

Figure 4. (a) TOFs for different CM1Bx-C3N4s, bulk-C3N4 and mesoporous graphitic (mpg)-C3N4 under
photocatalytic conditions (50 mg g-C3N4, 0.21 mM H2PtCl6, 10 vol% TEOA, H2O, total volume 38 ml). (b)
Intensity-averaged fluorescence lifetime of CM1Bx-C3N4s; the samples were excited at 405 nm and their
emission was monitored at 475, 500 and 525nm. (c+d) TOF versus the surface area and TOF versus the average
lifetime. Results are summarized in Table S2.

8

In order to determinate whether the morphology or the electronic properties of the CM1BXC3N4 materials are responsible for the photo-activity, the hydrogen evolution rate was
plotted as a function of the surface area and the fluorescence lifetime, respectively (Figure
4c and 4d). Although the surface area is related to the materials photo-activity, presumably
due to the increase in number of active sites for the reaction, some of the CM1BX-C3N4
materials exhibit different photo-activity for similar surface areas. Lifetimes, however,
clearly correlate with the hydrogen evolution rate, suggesting that the photo-catalytic
properties are mainly governed by the electronic structure. This further supports the theory
of the formation of heterojunctions, since charge separated states were trapped more
efficiently and therefore photocatalytic activity increased with decreasing life time.
In order to replace the platinum co-catalyst, a series of cobalt based WRCs and first row
transition metals were tested with CM1B0.05. In a typical standard experiment 3 mL of a
degassed, aqueous solution of 10 vol% TEOA, containing 4 mg CM1B0.05 and 0.5 mM of a
metal salt or WRC were irradiated with 1 sun of visible light from a xenon lamp (> 400 nm
cut off filter) and the evolving hydrogen was detected by means of a Clark electrode. Apart
of TEOA other known sacrificial reducing agents were tested (namely ascorbic acid20, tris-(2carboxyethyl)phosphine/ascorbic acid21 and hydroquinone22), however, none of those
systems produced hydrogen. Figure 5a shows the hydrogen evolution traces for all active cocatalysts including the trace of a blank experiment, where a degased solution of only
CM1B0.05 and TEOA was irradiated. Turn over frequencies (TOFs), turnover numbers (TONs)
and initialization-time (time between beginning of irradiation and start of H2 production) for
all investigated co-catalysts are summarized in Table S3. While metal salts Co(BF4)2, NiAc2,
and the complex [CoDOH(H2O)2]TFMS2 (S1)23 showed catalytic activity, CuTFMS2, Fe(SO4)2,
ZnCl2, AlTFMS2, [CoIIICRCl2]Cl (S2)24 and [CoIITpyBr2] (S3)25 where inactive towards H2
production (Scheme S2). In the absence of a suitable co-catalyst the system did not produce
any hydrogen. Complex S1 showed the highest initial rate with an initialization of 2.5 hours,
while Co(BF4)2 performed less efficient but started H2 production already after 1 hour. NiAc2
was less active than both cobalt species, but started producing hydrogen almost
instantaneously after beginning of irradiation. This stands in contrast to a report of Sun et al.
When using traditional g-C3N4 with Acriflavine as co-photo-sensitizer, nickel salts performed
better than their cobalt analogues.26 The rather surprising complete inactivity of the two
well-known WRCs S224 and S325 gave a first indication on the mode of interaction between
9

CM1B0.05 and its co-catalysts (vide infra). Although reduction by the semiconductor should be
thermodynamically feasible, neither of the two catalysts produced hydrogen.

Figure 5. (a) H2 evolution traces for photo-catalytic experiments with different co-catalysts under standard
conditions (4 mg CM1B0.05, 0.5 mM metal salt/ WRC, 10 vol% TEOA, H2O, total volume 3 mL.) and blank
experiment without co-catalyst. (b) H2 evolution traces for recycle experiments with Co(BF4)2. Black: typical first
run under standard conditions (corresponds to red curve in figure 5a). Green: recycled CM1B0.05 from first run
with fresh TEOA solution (10 vol%). Red: recycled CM1B0.05 from first run with fresh TEOA solution (10 vol%) and
additional Co(BF4)2 (0.5 mM). Blue: reuse of the solution from first run after 21 hours of irradiation with fresh
CM1B0.05 (4 mg). Cyan: reuse of the solution from first run after 70 hours of irradiation with fresh CM1B0.05 (4
mg). (c) Red: H2 evolution trace for recycled CM1B0.05 with fresh TEOA solution (10 vol%) after first run was
performed in the dark. Black: standard experiment for comparison (corresponds to red curve in figure 5a). (d)
Amount of surface bound cobalt, in percentage of initial concentration (0.5 mM), for catalytic experiments
under standard conditions stopped after different irradiation times.

In order to understand the nature of the catalytically active species when using Co(BF4)2 as a
co-catalyst, a series of experiments were performed, in which either the solution or the
solids where reused for a second photo-catalytic experiment (Figure 5b/c). At first, a
catalytic standard solution was centrifuged after one day of irradiation, and the solid was
washed and dried. Then, upon addition to a fresh solution of 10 vol% TEOA only, catalytic
activity was restored to about 50 % of the original initial rate (Fig. 5b, green). When
repeating the procedure, full catalytic activity could be restored when adding additional
Co(BF4)2 to the second run (Fig. 5b, red). When filtering the catalytic solution after one day
of irradiation and reusing the filtrate with fresh CM1B0.05, catalytic activity was only restored
10

to a very small extent (Fig. 5b, dark blue). These results strongly indicate a surface bound
catalytically active cobalt species. This was further underlined by an experiment, where the
first cycle was performed in the dark (1 night, no H2 detected). After separation from the
first catalytic solution and subsequent washing and drying, the solid was used with a fresh
solution of TEOA. The material still showed catalytic activity, suggesting the deposition or
incorporation of cobalt into the semiconductor even in the absence of photo-induced
reducing equivalents (Figure 5c, red).
Since reproducibility of the second runs were in all cases very dependent on the duration of
the first run, a sequence of standard experiments were stopped at different reaction times,
and the cobalt concentrations of the filtered solutions were determined by ICP
measurements. The differences between the initial cobalt concentration and the results
from the ICP quantification were assumed to be the amount of metal deposited in some
manner on the surface (Figure 5d). Until about 25 hours after start of the reaction, surface
cobalt concentration steadily increased up to the point, where all cobalt was bound to the
semiconductor. This explained the low but still detectable activity in the second run of the
recycle experiment, when fresh CM1B0.05 was added to the filtered catalytic solution, and the
not fully restored activity of the recycled CM1B0.05 sample, since in both experiments the
catalysis was stopped shortly before reaching complete deposition of the cobalt (small
amount of cobalt were still in solution). Therefore, when the first cycle was stopped after 70
hours of irradiation (to ensure complete deposition of the metal) and then reusing the
solution with a fresh sample of CM1B0.05, no hydrogen was detected in 1 day of irradiation
(Fig. 5b, light blue).
For the other metal salts investigated, the amount of deposited material was determined in
the same manner with which the surface cobalt content was established. With the exception
of AlTFMS2, all metal ions were deposited on the surface of the semiconductor or somehow
incorporated into its framework. The Zn-salt and the Cu-salt almost completely disappeared
from the solution, the iron-salt to about 18 %. Nickel was deposited to around 28 % after
one day of catalysis. Similar to Co(BF4)2, it is probably gradually accumulated on CM1B0.05.
The same applies for S1, which showed around 32 % less cobalt in solution after the same
irradiation period (Table S3). The fact that the two highly stable WRCs S2 and S3 were
inactive towards catalysis, strongly suggests, that mostly deposited material is acting as
11

active species and that a significant portion of the activity of S1 is due to its deposited
decomposition products undergoing a similar fate as Co(BF4)2. However, considering the high
activity of S1 and reports of conventional g-C3N4 used with similar WRCs the possibility of a
homogeneous catalytic contribution cannot be discarded completely.27, 28
To further corroborate the presence of cobalt in the solids and as well to determine the
nature of the deposition (either in form of nanoparticles or isolated, into the framework of
the semiconductor incorporated cobalt ions), SEM-EDX and TEM experiments were
performed with the solids before and after catalysis (Figure 6). Although cobalt was found by
EDX analysis in the samples from after the catalysis (up to 0.2 mol%, roughly consistent with
ICP data), it was not possible to pinpoint the location of the metal. Distribution maps
showed evenly dispersed cobalt over the whole of several of the investigated areas. In
samples of fresh CM1B0.05 no cobalt was detected. In agreement with the lack of discrete
cobalt sites on the surface, no large agglomerates (>5nm) of particles could be found in TEM
analysis either, suggesting a mono-atomic distribution of cobalt on or in the semiconductor.
In addition, the absence of crystalline cobalt aggregates is also evidenced in the XRD analysis
(Figure S7) in which only the typical reflection peaks of C3N4 are obtained.12, 29, 30

Figure 6. Top: SEM picture of CM1B0.05 after catalysis (left, scale bar 30 µm) and mapping of cobalt according to
EDX (right, scale bar 30 µm). Bottom: TEM pictures of CM1B0.05 before (left, scale bar 100 nm) and after (right,
scale bar 10 nm) catalysis.

12

Conclusions
In summary, we demonstrate the synthesis of a highly efficient carbon nitride photocatalyst
for hydrogen production by careful design of supramolecular complexes. The resulting
CM1Bx –C3N4 demonstrates roll-like morphology with different chemical properties. In
addition, the insertion of barbituric acid leads to an increase in the optical density along with
a red shift into the visible region. The in situ formation of heterojunctions (composed from
different semiconducting domains) is indirectly evident by life time measurements. The
formation of heterojunction improves the charge separation process, resulting in high
photocatalytic activity at low barbituric acid concentration. In addition, we also show the
successful replacement of platinum as co-catalyst in photo-catalytic hydrogen production by
more sustainable transition metals. The system is visible light responsive and performs well
without an additional photo-sensitizer or use of UV-light. Co- and Ni-ions and complex S1 act
as co-catalysts for CM1B0.05 for photo-catalytic hydrogen generation. S1 shows the highest
initial rate while Co(BF4)2 performed better than NiAc2. Recycle experiments and ICPmeasurements with Co(BF4)2 and CM1B0.05 suggest a deposited catalytically active species
rather than a homogeneous catalyst. SEM-EDX, TEM and XRD analysis show no evidence for
particle formation above 5 nm diameter, further suggesting very small aggregates or a
molecular catalytically active species. This work opens the possibility for facile synthetic
modification and optimization of C3N4 type semiconductors and semiconductor heteroassemblies, using only earth abundant and low cost starting materials. In addition, the
substitution of platinum as co-catalyst with first row transition metals corroborates the
importance of the further study of these systems. Lowering the absorption edge further into
the visible and improving the interaction between heterogeneous co-catalysts and the
material should allow for progressive improvement of photocatalytic efficiencies.

Supporting Information: Experimental details, SEM images, Nitrogen sorption isotherms,
FTIR ,XRD and Elemental analysis data along with the structure of the catalysts investigated
in this study are given in the supporting information.
AUTHOR INFORMATION
13

Corresponding Author
Menny.Shalom@mpikg.mpg.de
Author Contributions
§

These authors contributed equally to this work.

Notes
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
M.S. would like to thank “Minerva Fellowship” for financial support. M.G. is grateful for a
post-doctoral grant from the Swiss National Foundation. A.L. thanks MINECO (CTQ-201349075-R) and “La Caixa Foundation” for financial support.

14

REFERENCES
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
21.
22.
23.
24.
25.
26.
27.
28.
29.
30.

H. B. Gray, Abstr. Pap. Am. Chem. Soc., 2013, 246, 1.
K. Maeda and K. Domen, J. Phys. Chem. Lett., 2010, 1, 2655-2661.
M. C. Scharber, D. Wuhlbacher, M. Koppe, P. Denk, C. Waldauf, A. J. Heeger and C. L. Brabec,
Adv. Mater., 2006, 18, 789-+.
R. Konenkamp, R. C. Word and M. Godinez, Nano Lett., 2005, 5, 2005-2008.
H. Xu, Y. G. Xu, H. M. Li, J. X. Xia, J. Xiong, S. Yin, C. J. Huang and H. L. Wan, Dalton T., 2012,
41, 3387-3394.
M. Batzill, Energy Environ. Sci., 2011, 4, 3275-3286.
H. J. Yan, J. H. Yang, G. J. Ma, G. P. Wu, X. Zong, Z. B. Lei, J. Y. Shi and C. Li, J. Catal., 2009,
266, 165-168.
A. Kudo and Y. Miseki, Chem. Soc. Rev., 2009, 38, 253-278.
X. C. Wang, K. Maeda, A. Thomas, K. Takanabe, G. Xin, J. M. Carlsson, K. Domen and M.
Antonietti, Nat. Mater., 2009, 8, 76-80.
M. Shalom, S. Gimenez, F. Schipper, I. Herraiz-Cardona, J. Bisquert and M. Antonietti, Angew.
Chem.-Int. Edit., 2014, 53, 3654-3658.
Y. Zheng, Y. Jiao, J. Chen, J. Liu, J. Liang, A. Du, W. M. Zhang, Z. H. Zhu, S. C. Smith, M.
Jaroniec, G. Q. Lu and S. Z. Qiao, J. Am. Chem. Soc., 2011, 133, 20116-20119.
M. Shalom, S. Inal, C. Fettkenhauer, D. Neher and M. Antonietti, J. Am. Chem. Soc., 2013,
135, 7118-7121.
Y. J. Zhang, T. Mori, J. H. Ye and M. Antonietti, J. Am. Chem. Soc., 2010, 132, 6294-+.
J. Young-Si, L. Eun Zoo, W. Xinchen, H. Won Hi, G. D. Stucky and A. Thomas, Adv. Funct.
Mater., 2013, 23, 3661-3667.
T. Aida, E. W. Meijer and S. I. Stupp, Science, 2012, 335, 813-817.
G. Arrachart, C. Carcel, P. Trens, J. J. E. Moreau and M. W. C. Man, Chem.-Eur. J., 2009, 15,
6279-6288.
B. Jurgens, E. Irran, J. Senker, P. Kroll, H. Muller and W. Schnick, J. Am. Chem. Soc., 2003, 125,
10288-10300.
J. S. Zhang, X. F. Chen, K. Takanabe, K. Maeda, K. Domen, J. D. Epping, X. Z. Fu, M. Antonietti
and X. C. Wang, Angew. Chem.-Int. Edit., 2010, 49, 441-444.
J. R. Lakowica, Springer, 2009, 3rd ed
M. Guttentag, A. Rodenberg, R. Kopelent, B. Probst, C. Buchwalder, M. Brandstätter, P.
Hamm and R. Alberto, Eur. J. Inorg. Chem., 2012, 59-64.
C. Bachmann, B. Probst, M. Guttentag and R. Alberto, Chem. Commun., 2014, 50, 6737-6739.
B. Rausch, M. D. Symes and L. Cronin, J. Am. Chem. Soc., 2013, 135, 13656-13659.
B. Probst, M. Guttentag, A. Rodenberg, P. Hamm and R. Alberto, Inorg. Chem., 2011, 50,
3404-3412.
S. Varma, C. E. Castillo, T. Stoll, J. Fortage, A. G. Blackman, F. Molton, A. Deronzier and M. N.
Collomb, Phys. Chem. Chem. Phys., 2013, 15, 17544-17552.
M. Guttentag, A. Rodenberg, C. Bachmann, A. Senn, P. Hamm and R. Alberto, Dalton T., 2013,
42, 334-337.
J. F. Dong, M. Wang, X. Q. Li, L. Chen, Y. He and L. C. Sun, ChemSusChem, 2012, 5, 2133-2138.
S. W. Cao, X. F. Liu, Y. P. Yuan, Z. Y. Zhang, J. Fang, S. C. J. Loo, J. Barber, T. C. Sum and C. Xue,
Phys. Chem. Chem. Phys., 2013, 15, 18363-18366.
L. X., A. J. Ward, A. F. Masters and T. Maschmeyer, Chem. Eur. J., 2014, 20, 1-7.
J. S. Zhang, M. Grzelczak, Y. D. Hou, K. Maeda, K. Domen, X. Z. Fu, M. Antonietti and X. C.
Wang, Chem. Sci., 2012, 3, 443-446.
J. T. Jin, X. G. Fu, Q. Liu and J. Y. Zhang, J. Mater. Chem., 2013, 1, 10538-10545.

15