Ordered macroporous BiVO4 architectures with controllable dual porosity for efficient solar water splitting

Article abstract of DOI:10.1002/anie.201302680

Holes in holes: By using a modified colloidal crystal templating method, periodically ordered macroporous architectures with controllable dual porosity are fabricated (see picture). Taking BiVO₄ as an example, these unique structures provide a platform for a better understanding of the correlation between material geometrical features and charge migration for the photoelectrochemical water splitting process. Copyright

Full text of DOI:10.1002/anie.201302680

Angewandte
Chemie
DOI: 10.1002/anie.201302680
Solar Water Splitting
Ordered Macroporous BiVO4 Architectures with Controllable Dual
Porosity for Efficient Solar Water Splitting**
Min Zhou, Hao Bin Wu, Jian Bao, Lin Liang, Xiong Wen (David) Lou,* and Yi Xie*
Photoelectrochemical (PEC) water splitting based on the
Fujishima–Honda effect[1] has been demonstrated as an
attractive method for solar energy conversion and storage.[2]
For this technology to be competitive, a crucial factor for the
design of efficient photoelectrodes is to improve the utiliza-
tion of charges.[3] In addition to the intrinsic nature of
materials, morphology and structure also play a vital role on
charge migration.[4] Hence, studies on relationships between
geometrical characteristics and charge migration are very
important for solar water splitting.
Among many architectures for photoelectrodes, periodi-
cally ordered macroporous structures (POMS) have attracted
increasing attention for many interesting properties derived
from their special structural features.[5] As shown in Sche-
me 1a, the geometrical characteristics of periodically ordered
macroporous structures are mainly determined by two
parameters. One is the diameter of macropores surrounded
by the final skeletal walls (denoted as D1). The other is the
diameter of pores between neighboring macropores (denoted
as D2). Both D1 and D2 will exert their influence on various
processes in PEC water splitting.[6] Many reports have
confirmed that the relative sizes of D1 and D2 have full
influence on light absorption, deriving from the shift of the
photonic stop band.[6a,7a] Lee et al. also mentioned that the
proper D1 size would lead to efficient electron drift mobi-
lity.[7b] However, the influence of dual porosity on charge
migration, which originates from, for example, different
charge transport channels, grain size, surface states, has long
been neglected. A possible reason for this deficiency is the
lack of synthetic control over dual pore sizes. The difficulties
mainly stem from the requirements of meeting the following
three factors simultaneously. First, controllable templates
with plasticity of the diameter and the interconnected area
between neighboring spheres are required to adjust D1 and
D2. Second, a proper infiltration method with higher
infiltration fraction of precursors is needed to maintain
topological transformation from the template. Lastly, the
method should be versatile for various attractive materials.
Inspired by above considerations, we highlight herein
a general solution approach to tackle the challenge via
a modified colloidal crystal templating method. Different
from previous strategies, the combination of a sandwich
solution infiltration strategy and a post-heating treatment of
template greatly benefits both the structural and composi-
tional controllability, and therefore, provides a platform for
understanding correlations between geometrical character-
istics and PEC properties. As a proof of concept, we choose
BiVO4 as a model material to study the relationship between
dual pore sizes in periodically ordered macroporous struc-
tures and charge migration process during PEC water
splitting. BiVO4 has been shown to be highly efficient for
PEC water splitting, but still suffers primarily from some
problems of inefficient charge migration according to theo-
retical and experimental studies.[8]
Fabrication of such architectures with designed features is
achievable as outlined in Scheme 1a. Owing to the high
degree of periodicity in three dimensions, the template can be
translated into the replica periodically ordered macroporous
structure topologically. During this transformation, if the
diameter of the spheres can be controlled and the contact area
between neighboring spheres, it is in principle feasible to
control the sizes of D1 and D2 in the final structure. We
choose monodispersed polystyrene (PS) spheres as the unit in
view of their easily adjustable diameter and more importantly
plasticity resulting from the low glass transition temper-
ature.[9] With the increased post- heating time, PS spheres will
deform continuously so that the contact between neighboring
PS spheres can evolve from no contact to point contact, and
finally to quasi-face contact (Scheme 1b). This will give rise to
tunable geometrical features of the final architectures with
the help of high filling fraction to efficiently realize topo-
logical transformation. However, a high filling fraction of
precursor is difficult to achieve through common solution
infiltration methods.[6a] In view of the above, we propose
a novel sandwich-like configuration for infiltration as illus-
trated in Scheme 1b to realize higher filling fraction of
precursor solution. Specifically, the template is confined
between two substrates, one of which is hydrophilic and the
other is hydrophobic. A support (spacer) layer with a thick-
ness slightly larger than that of the template is inserted
between the two substrates. During the infiltration process,
the lower edge of the whole sandwich-like structure is
[*] Dr. M. Zhou, J. Bao, L. Liang, Prof. Y. Xie
Hefei National Laboratory for Physical Sciences at the Microscale
Department of Chemistry
University of Science & Technology of China
Hefei, Anhui 230026 (P.R. China)
E-mail: yxie@ustc.edu.cn
H. B. Wu, Prof. X. W. Lou
School of Chemical and Biomedical Engineering
Nanyang Technological University
62 Nanyang Drive, 637459 (Singapore)
E-mail: xwlou@ntu.edu.sg
Homepage: http://www.ntu.edu.sg/home/xwlou/
[**] This work was financially supported by National Basic Research
Program of China (2009CB939901), the Chinese Academy of
Sciences (XDB01020300), and National Nature Science Foundation
(11079004, 90922016).
Supporting information for this article is available on the WWW
under http://dx.doi.org/10.1002/anie.201302680.
Angew. Chem. Int. Ed. 2013, 52, 8579 –8583
ꢀ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
8579

.
Angewandte
Communications
Scheme 1. a) A schematic representation of the dual porosity in periodic ordered porous architectures. D1: diameter related to that of sphere. D2:
diameter related to interconnected point between spheres. b) Schematic illustration of the approach for controlling D2 in periodic ordered porous
architectures.
immersed into the precursor solution vertically, so that
solution can penetrate into the interstices of the template.[10]
The bottom substrate is hydrophilic so that the precursor
solution can infiltrate easily, especially around the interface
between the template and substrate; while the top substrate is
hydrophobic, therefore the intermediate composite can be
prevented from adhering to this substrate and the integrity of
the as-obtained architecture is maintained. Furthermore, the
support between two different substrates allows a local space
for the template spheres to suspend and disperse as much as
possible without structural collapse. With a short post-heating
time and hence less deformation of the PS spheres, the
extended distance between adjacent spheres enables more
solution to enter the interstitial space, and thus the filling
fraction of precursor solution is increased while maintaining
the structural integrity. Therefore, the two major parameters
of geometrical characteristics within the as-obtained archi-
tectures can be adjusted. D1 is mainly determined by the
diameter of PS spheres, and D2 is controlled by the contact
surface-to-volume ratio which is changed by the post-heating
time. Our goal of controllable dual porosity can be achieved
through the combination of the post-heating treatment and
the sandwich solution infiltration method.
To implement aforementioned concepts, we performed
the post-heating treatment on the colloidal crystal template
assembled from 200 nm PS spheres at 1108C for different
durations from S (short, 0 min) to M (medium, ca. 3–5 min)
and L (long, ca. 10–12 min). Clearly shown in Figure 1a–c and
Figure S1a in the Supporting Information, with continuous
deformation caused by heating, the contact between neigh-
boring PS spheres gradually changes from nearly no contact
to point contact to quasi-face contact, and finally to face
contact. This induced deformation by heating is general for PS
spheres with different diameters, such as 420 nm in Fig-
ure 1d–f and Figure S1b, 1000 nm in Figure 1g–i and Fig-
ure S1c. Then all the templates are infiltrated with proper
precursor solution of BiVO4 by the sandwich solution
infiltration method.
As expected, periodically ordered macroporous structures
of phase-pure BiVO4 (Figure S2, see Supporting Information)
with controllable dual pore sizes are prepared as shown in
Figure 2. A careful observation reveals that D1 is mainly
8580 www.angewandte.org
ꢀ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2013, 52, 8579 –8583

Angewandte
Chemie
connect to the corresponding macropores in the
sub-layer through the necks. The diameter of
interconnected area (D2) is directly associated
with the post-heating time of templates. For
convenience, hereafter, we designate various
periodically ordered macroporous structures as
“X-DPS”, where X refers to the post heating
duration (S, M, or L) and DPS is the diameter of PS
spheres in nanometer. The positive effects of the
sandwich infiltration strategy are: If the top
surface is not hydrophobic, the surface of as-
obtained samples would be rugged (Figure S3)
because some PS spheres would adhere to the
substrate and then be separated from the template
when removing the top substrate. Moreover, if the
hydrophobic substrate and the support are not
used to make a local space, the template would
straggle easily during the infiltration of precursor
solution. As a result, the final structures obtained
will not be so periodically ordered (Figure S4).
Besides, the wall of the periodically ordered
macroporous structure becomes thinner with
increased post-heating time, showing another
direct consequence of various D2 derived from
the increased deformation of PS spheres and
higher filling fraction. All the above evidence
verifies the structural designs illustrated in
Scheme 1. The morphology and structure are
dependent more on the template rather than the
type of precursor solution. The premise of the
present approach lies in the fact that the wett-
ability is guaranteed, hence the precursor solution
could be arbitrarily chosen. Similar periodically
ordered macroporous structures with controllable
dual pore sizes can be easily fabricated for many
other materials, including binary materials, such
as Bi2O3 (Figure S5), ternary materials such as
LiV3O8 (Figure S6), Bi2WO6 (Figure S7). In addi-
tion, controllable doping for different composi-
tion and concentration can also be easily realized,
such as Mo-doped BiVO4 (Figure S8). From this
perspective, such a solution approach is of great
importance in providing structural and composi-
tional controllability and a better understanding
of the correlations between structure and perfor-
mance.
Figure 1. The deformation of the PS spheres template after post-heating treatment
for different durations. The diameter of the PS spheres is a)–c) 200 nm. d)–
f) 420 nm. g)–i) 1000 nm. Note: for S the post-heating time of the template is 0 min
for samples shown in (a), (d), and (g). For M the post-heating time of template is
3 min for the sample shown in (b) and 5 min for samples shown in (e) and (h). For
L the post-heating time of template is 10 min for samples shown in (c) and (f), and
12 min for the sample shown in (i).
To probe the resulting effects on charge
migration for water splitting, PEC cells were
fabricated using various samples from templates
of 200 nm and 1000 nm PS spheres with different
post-heating times. Because all the samples have
been optimized to share similar calculated max-
imum photocurrent densities (Jabs ꢀ 5.0 mAcmÀ2
under 1 sun AM 1.5 sunlight, details are provided
in Section S2), it could be asserted that the
difference in the photocurrent densities (J) can
Figure 2. Typical SEM images of corresponding BiVO4 periodically ordered porous
architectures with controllable dual porosity. a) S-200, b) M-200, c) L-200, d) S-420,
e) M-420, f) L-420, g) S-1000, h) M-1000, i) L-1000. The diameter of original PS
spheres is a)–c) 200 nm. d)–f) 420 nm. g)–i) 1000 nm. For S the post-heating time of
the template is 0 min for samples shown in (a), (d), and (g). For M, the post-heating
time of the template is 3 min for the sample shown in (b) and 5 min for samples
shown in (e) and (h). For L, the post-heating time of the template is 10 min for
samples shown in (c) and (f), and 12 min for the sample shown in (i).
related to the diameter of original PS spheres, which is still
valid when using templates with different post-heating times.
Note that the well-defined macropores in the top layer
be fully ascribed to the difference in photon utilization, which
directly depends on charge migration in bulk and on surface.
As summarized in Table 1 (also in Figure S9a, S10a), analysis
Angew. Chem. Int. Ed. 2013, 52, 8579 –8583
ꢀ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
8581

.
Angewandte
Communications
Table 1: Relationship between PEC performance and dual porosity.
in an increased rate of surface recombination,
effective suppression on bulk recombination con-
Sample D1 [nm] D2 [nm] J @ 0.4 V/ J @ 1.2 V/ Rct [W] Photoelectron life-
mAcmÀ2 [a] mAcmÀ2 [a]
time @ 0.6 V/s[b]
tributes more to the enhanced charge migration and
hence PEC performance.[14] On the other hand,
S-200
M-200
l-200
S-1000
M-1000
l-1000
180
170
155
850
780
710
35Æ5
85Æ5
0.38
0.30
105Æ5 0.29
105Æ5 0.25
150Æ5 0.22
560Æ5 0.13
1.35
0.84
0.64
0.80
0.65
0.55
268.7
350.0
762.3
1187
1621
2815
2.01
1.82
1.62
1.70
1.54
1.31
under illumination, the steady-state charge-carrier
concentrations at the junction set the position of the
quasi-Fermi levels, which in turn determines Voc.
Conversely, under transient conditions when the
illumination is stopped, more inner defects and
boundaries become preferable sites where electrons
[a] The voltage is referenced to Ag/AgCl. [b] The voltage is referenced to RHE.
and holes are easy to recombine.[15] As such, a faster
Voc decay would be observed immediately after the
of the photocurrent densities on various samples allows us to
make the following two observations. 1) All the photoelectr-
odes achieve much higher photocurrent density over unor-
dered porous film (only 0.3 mAcmÀ2 at 1.2 V vs. Ag/AgCl;
Figure S11). This result implies that periodically ordered
macroporous skeleton is indeed advantageous as a photo-
electrode. 2) Carefully analyzing the curves of samples from
templates with PS spheres of the same diameter, we find that
the enhancement under a higher voltage bias is more
pronounced with shorter post-heating times. It is established
that bulk charge recombination is the main limiting factor
with the increased bias.[8a,11] Therefore, this apparent en-
hancement suggests that a smaller D2 is beneficial for
suppressing charge recombination, and hence easier charge
migration.
In essence, the hypothesis we want to verify is that the
existence of smaller D2 within the BiVO4 periodically
ordered porous architectures can facilitate charge migration.
At least two types of measurements can be conducted to study
the effects of D2. The first one is based on the assumption that
the interface charge transfer kinetics can be measured by
electrochemical impedance spectroscopy (EIS). As presented
in Figure S9b and S10b, the Nyquist plots consist of one
dominant semicircle, whose diameter is related to charge
transfer resistance at BiVO4/electrolyte interface.[12] Compar-
ing samples with identical D1 but different D2, samples with
smaller D2 lead to a smaller diameter of the semicircle. To
further understand the details, we interpret the plots in terms
of the equivalent circuit using the Randles–Ershler model
displayed in the inset of Figure S9b and S10b, where Rct
represents the charge transfer resistance across the interface.
As listed in Table 1, the fitted values of Rct clearly show that
the interface charge transfer in samples of smaller D2 is much
easier.
removal of illumination. With the help of the methodology by
Bisquert et al.,[16] we find that the Voc decay in L-DPS is faster
than in S-DPS (insets of Figure S9c, S10c). Furthermore, the
Voc decay rate is related to the photoelectron lifetime.[16b] With
shortened post-heating time (Table 1), the photoelectron
lifetime is correspondingly prolonged (Section S3, see Sup-
porting Information). Slower decay kinetics and longer
photoelectron lifetime are indicative of less charge trapping
when illuminated, hence, more effective charge separation for
water splitting.[15]
Meanwhile, when comparing the PEC performance of L-
200 and S-1000 which have similar D2 values, the tendency of
enhanced photocurrent density with increased D1 is very
different under low and high bias voltages. As for L-200,
a slightly higher photocurrent at the lower bias indicates that
smaller D1 enables suppression of surface recombination and
easy interface charge transfer, which is evidenced by smaller
Rct. The S-1000 sample exhibits a higher photocurrent density
at the higher bias. This observation together with higher
photoelectron lifetime verifies that increased D1 benefits the
elimination of bulk charge recombination.
As seen from above characterizations, the BiVO4 periodi-
cally ordered macroporous structures with different dual
porosity offer multiple effects. Specifically, smaller D2 is
helpful for charge migration both in bulk and on surface,
while smaller D1 facilitates surface charge migration, but
impedes bulk charge migration. They can be further ration-
alized: 1) A more compact space occupation of samples with
smaller D1 and D2 results in larger electrode/electrolyte
junction area, which ensures higher electrode/electrolyte area
and more active sites to reduce resistance at interface/charge
transfer resistance and open circuit voltage. It can be expected
that there are more holes capable of surviving the recombi-
nation and then reaching the electrode/electrolyte interface
for further utilization. Accordingly, more electrons can be
collected by the current collector. 2) With respect to samples
with smaller D2 and larger D1, the thickness of the wall
increases, leading to more interfacial contact between elec-
trode materials and the current collector. So the contact
resistance can be minimized and charge transfer at the
electrode/current collector interface is facilitated.[4a] 3) It can
be imagined that the thin wall formed within the interstitial
spaces generated by smaller PS spheres is much more fragile
during the calcination process. With D1 being constant,
a more compact packing or smaller grain size of BiVO4 grains
in the interstitial space can be achieved by introducing
a smaller D2. The opposite situation occurs while keeping D2
The second technique to appreciate the advantages of
smaller D2 relies on how it affects bulk charge migration
behavior. Shown in Figure S9c and S10c is the open circuit
voltage decay among the samples with different D2 values.
On one hand, the observation of the slightly negative shift of
the open circuit voltage (Voc) with less post-heating time
indicates that the D2 size has an important impact on the
process occurring at the interface, that is, a smaller D2 means
larger junction area.[13] If the flux of photo-generated charges
is instead distributed uniformly over a much larger junction
area, the bulk recombination limit would be reduced,
associated with improved surface reactions and hole interface
transfer. Although increased junction area would also result
8582
www.angewandte.org
ꢀ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2013, 52, 8579 –8583

Angewandte
Chemie
constant and introducing smaller D1. Therefore, there will be
less inner defects and grain boundaries, which usually serve as
charge recombination centers, by having small D1 and large
D2 in the periodically ordered macroporous structures. This
situation will benefit charge transfer in the bulk of the
photoelectrode.[7b] Overall, the above evidence and analysis
support the hypothesis that the periodically ordered macro-
porous structures with different dual porosity play a vital role
in charge migration. Being able to apply a similar level of
synthetic control over dual porosity to various materials
offers the potential to ease charge migration problems in PEC
water splitting. Although only one Jabs is considered, the
observed effects of dual pore size should be general. Hence, it
is believed that our findings will shed some light for further
development of efficient photoelectrodes for PEC cells.
In summary, we have demonstrated a general solution
approach to fabricate periodically ordered porous architec-
tures with control over dual porosity based on a modified
colloidal crystal templating method. As a proof of concept,
BiVO4 was used as a model material to investigate the
influences of geometrical features on charge migration for
PEC water splitting. A series of measurements, including
PEC characterizations, electrochemical impedance measure-
ments, and open-circuit voltage decay, have verified that
charge migration in the BiVO4 periodically ordered macro-
porous architectures shows strong dependence on the relative
size of dual porosity. Under the conditions of approximately
5.0 mAcmÀ2 (Jabs) under 1 sun AM 1.5 sunlight, a smaller D2
value is beneficial for charge migration both in the bulk and
on the surface, while a smaller D1 value facilitates surface
charge migration but impedes the bulk charge migration. This
finding can serve as a guideline for further design of
photoelectrodes with high efficiency.
Keywords: charge migration · colloidal crystal templates ·
dual periodic porosity · photoelectrochemistry
.
[1] A. Fujishima, K. Honda, Nature 1972, 238, 37.
[2] a) M. Grꢀtzel, Nature 2001, 414, 338; b) M. G. Walter, E. L.
Warren, J. R. McKone, E. A. Santori, N. S. Lewis, Chem. Rev.
2010, 110, 6446.
[3] C. Liu, Y. J. Hwang, H. E. Jeong, P. Yang, Nano Lett. 2011, 11,
3755.
[4] a) M. Zhou, J. Bao, Y. Xie, Chem. Commun. 2012, 48, 3362; b) K.
Sivula, F. Le Formal, M. Grꢀtzel, ChemSusChem 2011, 4, 432.
[5] a) G. von Freymann, V. Kitaev, B. V. Lotsch, G. A. Ozin, Chem.
Soc. Rev. 2013, 42, 2528; b) K. A. Arpin, A. Mihi, H. T. Johnson,
A. J. Baca, J. A. Rogers, J. A. Lewis, P. V. Braun, Adv. Mater.
2010, 22, 1084.
[6] a) L. Liu, S. K. Karuturi, A. I. Y. Tok, Energy Environ. Sci. 2011,
4, 209; b) J. Du, X. Lai, N. Yang, J. Zhai, D. Kisailus, F. Su, D.
Wang, L. Jiang, ACS Nano 2011, 5, 590; c) J. Du, X. Lai, J. E.
Halpert, Y. Yang, D. Wang, Sci. China Chem. 2011, 54, 930.
[7] a) C. Cheng, S. K. Karuturi, L. Liu, J. Liu, H. Li, L. T. Su, A. I. Y.
Tok, H. J. Fan, Small 2012, 8, 37; b) E. S. Kwak, W. Lee, N. G.
Park, J. Kim, H. Lee, Adv. Funct. Mater. 2009, 19, 1093.
[8] a) M. Zhou, J. Bao, W. Bi, Y. Zeng, R. Zhu, M. Tao, Y. Xie,
ChemSusChem 2012, 5, 1420; b) W. Luo, Z. Yang, Z. Li, J.
Zhang, J. Liu, S. Yan, T. Yu, Z. Zou, Energy Environ. Sci. 2011, 4,
4046.
[9] Y. Li, W. Cai, B. Cao, G. Duan, C. Li, F. Sun, H. Zeng, J. Mater.
Chem. 2006, 16, 609.
[10] J. H. Moon, S. Yang, Chem. Rev. 2010, 110, 547.
[11] H. Gao, C. Liu, H. E. Jeong, P. Yang, ACS Nano 2012, 6, 234.
[12] B. Klahr, S. Gimenez, F. Fabregat-Santiago, T. Hamann, J.
Bisquert, J. Am. Chem. Soc. 2012, 134, 4294.
[13] J. R. Maiolo, H. A. Atwater, N. S. Lewis, J. Phys. Chem. C 2008,
112, 6194.
[14] B. M. Kayes, H. A. Atwater, N. S. Lewis, J. Appl. Phys. 2005, 97,
114302.
[15] Y. Lin, Y. Xu, M. T. Mayer, Z. I. Simpson, G. McMahon, S.
Zhou, D. Wang, J. Am. Chem. Soc. 2012, 134, 5508.
[16] a) A. Zaban, M. Greenshtein, J. Bisquert, ChemPhysChem 2003,
4, 859; b) Z. Zhang, P. Wang, Energy Environ. Sci. 2012, 5, 6506.
Received: April 1, 2013
Published online: June 14, 2013
Angew. Chem. Int. Ed. 2013, 52, 8579 –8583
ꢀ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
8583