Angewandte
Chemie
ing Information, transmission electron microscopy (TEM)
images of the catalysts show sheet-like structures, as is the
the results obtained by 48 h reaction. Bare TiO (entry 1)
2
scarcely produces H O (less than detection limit, 0.2 mmol).
2
2
[
15]
[6f]
[6d]
case for g-C N . X-ray diffraction (XRD) of g-C N (Fig-
TiO loaded with Au or modified with fluoride, which
3
4
3
4
2
ure S2) shows a peak at 2q = 27.4 (d = 0.325 nm) assigned to
have been proposed for H O2 production, are inactive
2
[16]
(
002) packing of the melem sheets. Increasing the amounts
(entries 2 and 3). g-C N (entry 4) does not produce H O .
3
4
2
2
of PDI units decreases its intensity, along with a formation of
new two peaks at 2q = 19.0 (d = 0.467 nm) and 29.6 (d =
Mesoporous g-C N (mpg-C N ) with a large surface area
3 4 3 4
2
ꢀ1
[21]
(190 m g ) prepared by a silica-templated polymerization
is also inactive (entry 5). In contrast, g-C N /PDI produce
0
.302 nm). These are assigned to p,p-stacking of PDI
3
4
x
[17]
units and donor–acceptor interaction between melem and
very large amounts of H O (entries 6–8); the amount of H O
2 2 2 2
formed on g-C N /PDI (50.6 mmol) is more than 250-fold of
3 4 51
[18]
PDI units, respectively. These data suggest that, as shown in
Scheme 2b, the PDI units are randomly incorporated within
the melem sheet and the sheets are layered multiply. As
shown in Figure S3, X-ray photoelectron spectroscopy (XPS)
that obtained with g-C N (< 0.2 mmol). As shown by entries 9
3
4
and 10, N,N’-dipropylPDI (Scheme 2c), a reference com-
pound for PDI, and a mixture of g-C N and N,N’-dipro-
3
4
2
of g-C N (N1s level) shows three peaks assigned to sp -
pylPDI scarcely produces H O . These data suggest that
2 2
3
4
hybridized N atoms of melem at 398.5 eV (red, Scheme 2a),
trigonal N atoms of the melem center at 399.4 eV (blue,
Scheme 2a), and terminal amine N atoms at 401.0 eV (green,
incorporating PDI units within the g-C N network facilitates
3 4
H O production from water and O .
2
2
2
Table S1 shows the results of photoreaction on g-C N /
3
4
[
19]
Scheme 2a), respectively. Incorporating PDI units creates
a new peak at 400.1 eV, assigned to imide N atoms of the PDI
PDIx with 2-PrOH as a sacrificial electron donor. The
selectivity for the amount of H O2 formed relative to the
2
[20]
units (purple, Scheme 2). Integrating these signals there-
fore allows rough determination of the mole fraction of PDI
units (x) within g-C N /PDI .
amounts of photooxidation products (acetone and CO ) is
2
about 90%, which is similar to that obtained with g-C N . This
3
4
suggests that g-C N /PDI selectively promote two-electron
3
4
x
3
4
x
[
9]
Diffuse-reflectance UV/Vis spectra of g-C N /PDI (Fig-
reduction of O as does g-C N . Figure S6 shows the change
2 3 4
3
4
x
ure S4) show absorption in the visible region, as is the case for
g-C N ; their band-gap energies are 2.4–2.8 eV. Electrochem-
in the amounts of H O formed during reaction in a water/O
2 2 2
system with g-C N /PDI . The rate of H O evolution is
3
4
3
4
x
2
2
ical Mott–Schottky plots of g-C N /PDI (Figure S5) show
almost constant even after prolonged irradiation, indicating
that the catalysts produce H O without loss of activity. The g-
3
4
x
typical n-type character. The obtained flat band potentials
and band-gap energies provide the band structures of g-C N /
2
2
C N /PDI catalyst, when reused for further reaction
3
4
3
4
51
PDI (Figure 1). Both conduction band (CB) and VB levels
(entry 7, Table 1), shows almost the same activity as the
fresh sample. In addition, the recovered catalyst shows similar
X-ray diffraction (XRD) pattern to that of the fresh one
x
(
Figure S2). These data suggest that the catalyst is stable and
reusable for further reaction.
Figure S7 shows the action spectrum for H O forma-
[22]
2
2
tion on g-C N /PDI . The apparent quantum yields (FAQY)
3
4
51
agree with the absorption spectrum of the catalyst, indicating
that its band-gap excitation promotes water oxidation and O2
reduction. Note that FAQY at 420 nm is 2.6%, which is similar
to that for overall water splitting on a highly active solid-
[23]
solution catalyst (ca. 2.5% at 420 nm). Ab initio calcula-
tions based on the density functional theory (DFT) were
performed within the Gaussian03 program to clarify the
effect of PDI unit on the electronic structure of g-C N with
3
4
simple melem and melem-PDI models (Figure 2). Their main
electronic transitions (S !S ) are HOMO!LUMO and
0
1
HOMO!LUMO + 2, respectively. Incorporating PDI units
Figure 1. Electronic band structures of g-C N and g-C N /PDI . Photo-
graphs show the corresponding powders of the samples.
3
4
3
4
x
decreases both S and S levels; this agrees with the observed
0
1
VB and CB levels (Figure 1). The electrons on both HOMO
and LUMO + 2 of the melem-PDI model are located mainly
at the melem unit with partial distribution to the PDI units.
become more positive by the incorporation of PDI units
[
13]
[13]
owing to their high electron affinity, with the CB levels still
This is reflected by the high electron affinity of PDI unit.
[11]
more negative than the reduction potential of O (0.08 V).
The electrons on HOMO are located at the N2 and N6
positions of the melem unit, and those on LUMO + 2 are at
the C1 and N4 units. These data suggest that these respective
atoms on the melem units behave as the oxidation (N2 and
N6) and reduction (C1 and N4) sites, as is the case for g-
2
These data suggest that g-C N /PDI may possess enhanced
3
4
x
capability for water oxidation with sufficient potentials for O2
reduction.
Photocatalytic reactions were performed by photoirradia-
tion of pure water (30 mL) containing respective catalysts
[
9]
C N4 (Scheme 1).
3
(
50 mg) by a Xe lamp (l > 420 nm) with magnetic stirring
The mechanism for H O formation on the photoexcited
2
2
[
24]
under O atmosphere (1 atm) at 298 K. Table 1 summarizes
g-C N /PDI can be explained as Scheme 3. Photoexcitation
2
3
4
Angew. Chem. Int. Ed. 2014, 53, 1 – 7
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
3
These are not the final page numbers!