Structure–Activity Correlations for Br?nsted Acid, Lewis Acid, and Photocatalyzed Reactions of Exfoliated Crystalline Niobium Oxides

Article abstract of DOI:10.1002/cctc.201601131

Exfoliated crystalline niobium oxides that contain exposed but interconnected NbO₆ octahedra with different degrees of structural distortion and defects are known to catalyze Br?nsted acid (BA), Lewis acid (LA), and photocatalytic (PC) reactions efficiently but their structure–activity relationships are far from clear. Here, three exfoliated niobium oxides, namely, HSr₂Nb₃O₁₀, HCa₂Nb₃O₁₀, and HNb₃O₈, are synthesized, characterized extensively, and tested for selected BA, LA, and PC reactions. The structural origin for BA is associated mainly with acidic hydroxyl groups of edge-shared NbO₆ octahedra as proton donors; that of LA is associated with the vacant band position of Nb⁵⁺ to receive electron pairs from substrate; and that of PC is associated with the terminal Nb=O of NbO₆ octahedra for photon capture and charge transfer to long-lived surface adsorbed substrate complex through associated oxygen vacancies in close proximity. It is believed that an understanding of the structure–activity relationships could lead to the tailored design of NbO_x catalysts for industrially important reactions.

Full text of DOI:10.1002/cctc.201601131

Accepted Article
Title: Structure- Activity Correlations for Brønsted Acid, Lewis Acid and
Photo- Catalysed Reactions of Exfoliated Crystalline Niobium
Oxide Layers
Authors: Edman Shik Chi Tsang, Yusuke Koito, Gregory J. Rees, John
V. Hanna, Molly M.J. Li, Yung-Kang Peng, Tim Puchtler,
Robert Taylor, Tong Wang, Hisayoshi Kobayashi, Ivo. F.
Teixeira, M. Abdullah Khan, and Hannah Kreissl
This manuscript has been accepted after peer review and appears as an
Accepted Article online prior to editing, proofing, and formal publication
of the final Version of Record (VoR). This work is currently citable by
using the Digital Object Identifier (DOI) given below. The VoR will be
published online in Early View as soon as possible and may be different
to this Accepted Article as a result of editing. Readers should obtain
the VoR from the journal website shown below when it is published
to ensure accuracy of information. The authors are responsible for the
content of this Accepted Article.
To be cited as: ChemCatChem 10.1002/cctc.201601131
Link to VoR: http://dx.doi.org/10.1002/cctc.201601131
A Journal of
www.chemcatchem.org

10.1002/cctc.201601131
ChemCatChem
FULL PAPER
Structure- Activity Correlations for Brønsted Acid, Lewis Acid
and Photo- Catalysed Reactions of Exfoliated Crystalline Niobium
Oxides
Yusuke Koito,[a] Gregory J. Rees,[b] John V. Hanna,[b] Molly M.J. Li,[a] Yung-Kang Peng,[a] Tim
Puchtler,[c] Robert Taylor,[c] Tong Wang,[c] Hisayoshi Kobayashi,[d] Ivo F. Teixeira,[a] M. Abdullah
Khan,[a] Hannah T. Kreissl[a] and S.C. Edman Tsang[a]
[a] Inorganic Chemistry Laboratory, University of Oxford, South Parks Road, Oxford, Oxfordshire, OX1 3QR, UK
[b] Department of Physics, University of Warwick, Gibbet Hill Road, Coventry, CV4 7AL, UK
[c] Department of Physics, Clarendon Laboratory, University of Oxford, Oxford, OX1 3PU, UK
[d] Department of Chemistry and Materials Technology, Kyoto Institute of Technology, Matsugasaki, Sakyo-ku, Kyoto, 606-8585,
Japan
Abstract: Exfoliated crystalline niobium oxides containing exposed
but interconnected NbO6 octahedra with different degrees of
structural distortions and defects are well-known to catalyse
Brønsted acid (BA), Lewis acid (LA) and Photo catalysed (PC)
reactions efficiently but their structure-activity relationships are far
from clear. Here, three exfoliated niobium oxides, namely
HSr2Nb3O10, HCa2Nb3O10 and HNb3O8 are synthesised, extensively
characterized and tested for selected BA, LA and PC reactions. It is
revealed that the structural origin for BA is found to associate mainly
with acidic hydroxyl groups of edge shared NbO6 octahedra as
proton donors; LA to vacant band position of Nb5+ to receive electron
pairs from substrate; PC to terminal Nb=O like of NbO6 octahedra for
photon capture and charge transfer to long-lived surface adsorbed
substrate complex via associated oxygen-vacancy in close proximity.
It is believed that such understanding of the structure-activity
relationships could lead to tailored designs of NbOx catalysts for
industrial important reactions.
catalyse many hydrolysis and hydration reactions whilst Lewis
acidity catalyses typical Meerwein-Ponndorf-Verley (MPV),[8] Baeyer-
Villiger oxidation[9] and glucose isomerisation[10] reactions. Niobium
oxides possess both Brønsted and Lewis acid properties, which act
as versatile catalysts to activate multistep reactions involving both
acid functions. For instance, water-tolerant H3PO4-treated Nb2O5 can
catalyse glucose transformation into 5-(hydroxylmethyl) furfural
under mild reaction conditions.[11] But, there is still poor
understanding on the chemical nature of Brønsted sites and the
structural definition of Lewis acidity. Additionally, structural
correlation with photocatalytic properties of niobium oxides is
scarcely represented in the literature. It is generally thought to relate
to electronic band structure, degree of oxygen-vacancy and
efficiency of electron and hole-pairs generation under light irradiation,
etc.[12] In some instances, these features and surface acidity of
niobium oxide are somehow linked to achieve desirable redox
reactions but no structural model is yet given. A substrate-Lewis acid
sites surface complex structure has recently been proposed during
the photo-oxidation reaction of benzylic alcohol on thin layer
HNb3O8.[13] But no discussion is yet given how the photon is
captured on a specific site of niobium oxide structure and how it
subsequently leads to the formation of products.
Introduction
In this paper, highly crystalline thin layers of related structures of
exfoliated (ex) HSr2Nb3O10, exHCa2Nb3O10 and exHNb3O8 were
carefully synthesized through an exfoliation process (as shown in the
Experimental section) from their respective bulk materials. These
exfoliated sheets maintain the 2D crystal structures, forming
precisely defined atomic structures. This can facilitate examination of
the origins of solid Brønsted and Lewis acidity and more importantly
photocatalysis in relationships with these acidities in comparative
and comparative manner based on an atomic scale and quantum
chemical levels, which is not well understood for most poorly defined
solid acid materials. Thus, their structural integrities, degrees of
distortion of octahedra and defects, acidities and band structures of
these thin layered structures have been extensively characterized
with XRD, multi-nuclear solid state NMR, EPR, XPS, CV, UV-vis,
TRPL, etc in comparative way. Model reactions including hydrolysis
of ethyl acetate (Brønsted acid catalysed reaction); conversion of
pyruvic aldehyde to lactic acid (Lewis acid catalysed reaction) and
air oxidation of benzyl alcohol to benzylaldehyde and hydrogen
production from water (photocatalysis) by niobium oxides have been
examined, respectively. The structural aspects of the unique
photocatalytic properties of these high surface 2-D layered niobium
oxide structures are particularly discussed.
Niobium oxides possess many interesting physiochemical
properties such as high water-tolerant Brønsted and Lewis acidities,
semiconducting and photo and electrochromic properties which
enable the materials for a wide variety of applications such as for
gas sensing, catalysis, electronics, optical devices and displays.[1-4]
The structures and morphologies of niobium oxides such as HNb3O8,
H4Nb6O17, HSr2Nb3O10, HCa2Nb3O10 and HTiNbO5 can also be easily
altered, which provide structural means to achieve tunable
properties for these exciting applications.[5,6] Recently, it has been
observed that exfoliated 2D niobium oxide structures with mostly
exposed cation and anion sites exhibit more superior acid and photo
catalytic performance than bulk materials.[7] However, there is at
present,
a lack of understanding on their structure-activity
relationships of these well-defined niobium oxides in correlating the
above properties in systematic and comparative manner.
In general, solid acids are broadly classified into two types,
Brønsted and Lewis acids. They exhibit different functions in acid
catalytic reactions. The Brønsted acid is capable of transferring
protons to substrate whereas Lewis acid facilitates the formation of a
Lewis acid-substrate complex intermediate by the transfer of
electron pairs in electrophile-nucleophile interaction, resulting in the
polarization of the substrate. Consequently, Brønsted acidity can
This article is protected by copyright. All rights reserved.

10.1002/cctc.201601131
ChemCatChem
FULL PAPER
Results and Discussion
(a)
(i)
(ii)
(iii)
Fig. 1(a) shows the bulk Nb(IV) containing structures, HSr2Nb3O10,
HCa2Nb3O10 and HNb3O8. The first two belong to the class of
layered perovskite structures, AB2Nb3O10 containing 12 corner
shared NbO6 units with Sr2+ or Ca2+ at the B sites. These
compounds consist of negatively charged perovskite sheets that
stack to form a two-dimensional layered structure interleaved with
H+ in the A sites for compensation of the negative charge of sheets.
It is noted from Figure 1(i) that Sr2+ ion (1.58 Å in 12 coordination)
at the B site fits very well to the layered perovskite structure.
However, the occupancy of smaller Ca2+ cations (ionic radius of
1.48 Å with higher charge density clearly reduce the dimensions of
perovskite layer, decrease the symmetry (see Fig. 1(ii)) cause a
substantial distortion to individual NbO6 units than that of larger Sr2+
ions.[14] On the other hand, apart from corner-sharing NbO6 HNb3O8,
the structure also contains edge-sharing NbO6[15] and the H+ cations
are emplaced between the 2D anion sheets. The XRD patterns of
bulk and exfoliated HSr2Nb3O10, HCa2Nb3O10 and HNb3O8 samples
were collected and are shown in Fig. 2(a). Notice that the
intensities of (004) and (006) peaks due to layer stacking along z-
axis for exHSr2Nb3O10 and exHCa2Nb3O10 significantly decrease
after exfoliation although much broader and weaker (002)
diffraction peaks are still observed even after the exfoliation
process. This indicates that there was no long range layer stacking
of the materials but thin exfoliated layers are produced. The TEM
images of the exfoliated samples shown in Fig. 2(b) indeed depict
the thin nanosheets morphology, where flake-like 2D nanosheets
structure ranging from 1-8 layers is clearly evident for the samples.
(b)
(b)
(i)
(ii)
(iii)
(iv)
(v)
(vi)
Figure 2. (a) XRD patterns of (i) HSr2Nb3O10, (ii) HCa2Nb3O10 and (iii)
HNb3O8, :(I) hydrated crystal precursor, (II) formed from (I) by calcination
at 150 °C for 3 h. (III) synthesized by exfoliation from (II) and (IV) formed
from (3) by calcination at 150 °C for 3 h. (b) TEM and HRTEM images of
exHSr2Nb3O10 ((i) and (ii)), exHCa2Nb3O10 ((iii) and (iv)) and exHNb3O8
((v) and (vi)).
(a)
(ii)
(iii)
(i)
(b)
(iii)
(ii)
(i)
Figure 1. Two views (a, b) of crystal structures of niobium oxides of (i)
HSr2Nb3O10 (ii) HCa2Nb3O10 and (iii) HNb3O8 according to ISCD
,
Figure 3. The 1H MAS (30 kHz) NMR at 14.1 T for (a) HSr2Nb3O10 , (b)
exHSr2Nb3O10 (c) HCa2Nb3O10, (d) exHCa2Nb3O10, (e) HNb3O8 and (f)
exHNb3O8. The isotropic shift (δiso) are given above each resonances.
database. GaussView5.0 was used to reproduce the structures. Atoms are
colour labelled as follows: Nb (green), Sr (dark blue) and Ca (light blue).
Proton sites on shared NbO6 units and Brønsted acidity (BA)
The solid state 1H MAS NMR (ssNMR) spectra shown in Fig. 3 (a),
(c), (e) reflect the different chemical environment of protons in
HSr2Nb3O10, HCa2Nb3O10 and HNb3O8, respectively, which are
resident between the three bulk negatively charged crystalline
niobium layer oxides of strongest BA in the range between 9.8 to
11.2 ppm. They have clearly shifted to lower frequencies along with
weaker and broadened peak intensity, as observed in their exfoliated
counterparts due to the collapse of the layer structures, as
previously discussed by A. Takagaki et al.[16] As a result, analysis of
1H SSNMR spectroscopy can reflect the new chemical environment
of surface attached protons on different degrees of connectivity of
the NbO6 octahedra over these different forms of exfoliated niobium
oxides. For the surface exposed sheets, the 1H chemical shift
indicates acidity, and hydrogen species with a high downfield shift
are expected to possess stronger Brønsted acidity. Clearly, the
layered structure of exHSr2Nb3O10 contains stacked NbO6 octahedra
in highest symmetry with primarily shared corners giving the lowest
average proton chemical shift value (the distribution curve with the
most probable peak at iso = 4.6 ppm as shown in Fig. 3b) with the
weakest BA (Fig. 3b). Comparatively, the structural distorted
octahedra in the case exHCa2Nb3O10 can give stronger BA sites
(assigned peak atso= 6.3 ppm as shown in Fig. 3d). Conversely,
the layered structure of HNb3O8 contains stacked NbO6 octahedra
with shared edges, giving the comparatively strong BA (assigned as
iso= 7.0 6.5 and 5.9 ppm in Fig. 3f). It is to note that the strongly
acidic BA site on Group IV-VI oxides is associated with exposed
bridging hydroxyl groups between octahedra with shared edges or
faces while the weaker BA corresponds to a proton localized on a
terminal oxygen.[16] We realized that the complex relationship
between 1H chemical shift of ssNMR and acid strength so
quantitative analysis cannot be done easily.[16] Conventionally NH3-
TPD technique was attempted, however, the strong adsorption and
This article is protected by copyright. All rights reserved.

10.1002/cctc.201601131
ChemCatChem
FULL PAPER
the facile nature of the single layers induced a significant degree of
structural collapse so no clear desorption profiles could be used for
the quantitative analysis. Nevertheless, as according to the BET
analysis (Fig. S1), exHSr2Nb3O10, exHCa2Nb3O10 and ex HNb3O8
give surface areas of 53, 56 and 112 m2/g, respectively. Thus, the
adduct that consequently, enhances LA catalytic reactions.[17] The
LUMO energy can correspond to the conduction band edge in the
solid catalyst, inferring that Lewis acids with lower energy of the
conduction band edge are advantageous for activating Lewis acid
catalyzed reaction.
quantity and strength of Brønsted acidity follow exHNb3O8
>
Fig. 5 shows (a) UV-vis spectra for layered and exfoliated
HSr2Nb3O10, HCa2Nb3O10 and HNb3O8 and (b) corresponding
Kubelka-Munk plots against the energy of light. The estimated band
gap energies of exHSr2Nb3O10, exHCa2Nb3O10 and exHNb3O8, are
3.30, 3.39, 3.29 eV, respectively.[18] Further, using the equation of
ECB ≈ 1.23−Eg/2,[19] the lowest conduction band potential of each
niobium oxides; exHSr2Nb3O10, exHCa2Nb3O10 and exHNb3O8 are
estimated to −0.42, −0.47 and −0.42 eV. These results show that
exHCa2Nb3O10 has a more negative CB edge potential, presumably
reflecting the degree of structural distortion of the NbO6 units due to
the polarization of the Ca2+ in B site. This observation matches with
our EXAFS data (Table S1) which shows significant distortion of
NbO6 units to longer Nb-O bonds (2.51 Å) when facing the high
charge density Ca2+ in B site in exHCa2Nb3O10 (to maintain charge
neutrality shorter Nb-O bonds of 1.66 Å are simultaneously created).
The degrees of distortion for both exHSr2Nb3O10 and exHNb3O8 are
similar and minor. Blasse et al. reported the relationship between
crystal structures and energy delocalization of perovskite structures
consisting of corner-sharing NbO6 octahedral units.[20] According to
exHCa2Nb3O10 > exHSr2Nb3O10. It is noted that the activity for the BA
catalyzed hydrolysis of ethyl acetate reaction[16] also gives the same
order as above. The measured rates for ethanol production from
ethyl acetate over exHSr2Nb3O10, exHCa2Nb3O10 and exHNb3O8 are
0.10, 0.12 and 0.20 μmol/min, respectively (Fig. 4).
(a)
(iii)
(ii)
(i)
Figure 4. Catalytic performance of hydrolysis of ethyl acetate. A mixture of
1.9 mL ethyl acetate, 0.1 ml water and 50 mg catalyst was heated in a
vacuum vial with fluoro-rubber cap at 333 K for 20 h.
(b)
(a)
(b)
(i)
(ii)
(iii)
Figure 6. (a) XPS spectra of (i) exHSr2Nb3O10, (ii) exHCa2Nb3O10 and (iii)
exHNb3O8. The spectra fitting was conducted by Casa XPS. (b) Cyclic
voltammograms of exHSr2Nb3O8, exHCa2Nb3O10 and (c) exHNb3O8.
Figure 5. (a) UV-vis absorption spectra of (i) exHSr2Nb3O10 (ii)
exHCa2Nb3O10 and (iii) exHNb3O8; (b) Transformed Kubelka-Muck plot
versus energy of light of (i) exHSr2Nb3O10 (ii) exHCa2Nb3O10 and (iii)
exHNb3O8
their study, the relatively stronger interaction between NbO6
octahedra with B cations results in localization of the excited state
energy, because a bond angle of Nb–O–Nb is far from the ideal
angle of 180o, i.e., the NbO6 octahedral units are tilted. The
localization of excited state energy clearly causes the position of
absorption band edge to be shifted to shorter wavelength with
weaker Lewis acidity.
Distorted NbO6 units and Lewis acidity (LA)
In general, the position of LUMO of Lewis acid sites (cationic sites)
are responsible for the formation of stable LA-nucleophilic reactant
This article is protected by copyright. All rights reserved.

10.1002/cctc.201601131
ChemCatChem
FULL PAPER
The chemical nature of niobium species has been further
investigated by XPS as shown in Fig. 6(a). Note that the main
signals of Nb3d5/2 and Nb3d3/2 of exHSr2Nb3O10 and exHNb3O8
appear at slightly higher binding energy (BE) than that of
exHCa2Nb3O10. In case of exHSr2Nb3O10 and exHNb3O8, BE of
Nb3d3/2 were seen at 210.35 eV and 210.43 eV, respectively and
however, for exHCa2Nb3O10, this peak was observed at 209.52 eV
Distorted NbO6, Nb=O, lattice vacancies and photocatalysis
As stated, photocatalysis is a complex phenomenon and there is so
far very little knowledge how the structures of exfoliated niobium
oxides correlate to their photo-activity. However, it is believed that
the samples must contain ‘chromophore’ groups that can capture
appropriate photonic energy and transfer it to substrate-solid
indicative of the more distorted and reduced nature of exHCa2Nb3O10. complex to catalyse chemical reactions at longer timescale. Thus,
Therefore, the conduction band level of Nb5+ in exHCa2Nb3O10 is
apparently higher compared to the other two exfoliated niobium
oxide samples due to the polarization distortion of NbO6 by the Ca2+.
Cyclic Voltammetry (CV) has also been used to probe the reduction
potential of Nb5+ sites in exHCa2Nb3O10. As shown in Fig. 6(b), the
reduction potential of exHNb3O8 was −0.82 V (V vs VAgCl), which is
slightly higher than that of −0.86 V of exHSr2Nb3O10. However, for
the exHCa2Nb3O10, it again shows that the significant lowest voltage
of −1.10 V is required to reduce Nb5+ site in exHCa2Nb3O10. As a
result, from the correlation of conduction band energy level with
Lewis acidity of the three exfoliated samples, the Lewis acidity order
is expected to be exHNb3O8≥ exHSr2Nb3O10 >> exHCa2Nb3O10. This
result is good agreement with the results of UV-vis, EXAFS and XPS,
etc.
elucidations on the chemical nature of the ‘chromophore’ groups and
the sites for substrate-solid complex are crucially important in
understanding their structure-activity relationship in photocatalysis. It
is also noted that this lifetime for this substrate-solid complex for
(b)
(a)
Figure 8. (a) ESR spectra of (i) exHSr2Nb3O10, (ii) exHCa2Nb3O10 and (iii)
exHNb3O8 measured at ambient temperature. (b) Raman scattering spectra
of (i) HSr2Nb3O10 (ii) exHSr2Nb3O10, (iii) HCa2Nb3O10 (iv) exHCa2Nb3O10
,
(v) HNb3O8, (vi) exHNb3O8.
energy transfer in redox manner in photocatalysis should be long
enough to initiate typical chemical reactions (10-10 to 10-5 s).[21]
Therefore, the formation of stable substrate-niobium oxide
complexes via Lewis acid-base interaction would be considered to
be preferable for an effective electron migration under UV irradiation.
In addition, the contribution of oxygen-vacancy sites cannot be
underestimated since trapping substrate onto oxygen-vacancy site
favours binding stabilization.
First, ESR was used to demonstrate the formation of oxygen-
vacancy sites in these exfoliated samples. As shown in Fig. 8(a),
exHNb3O8 give a clear signal at g=2.01, which is assigned to free
unpaired electron trapped in oxygen-vacancy site (Ov).[22] Only very
weak or marginal signal is observed in exHCa2Nb3O10 followed by
exHSr2Nb3O10. Therefore, oxygen-vacancy Ov sites were mainly
produced in the exHNb3O8 structure with highest density of these
three samples above. Interestingly, another small signal at g=1.98 is
noted in exHCa2Nb3O10, which could be assigned to the Nb4+. The
Nb4+ species generally give a very short spin relaxation time
according to previous work.[23] This signal can only be confirmed at
super-cold temperature (~4K).[24]
Figure 7. The catalytic yield of lactic acid from after 2h reaction over
three exfoliated samples is shown. A mixture of catalyst (50 mg) and
2.0 mL aqueous pyruvic aldehyde solution (0.1 M) were heated in a
vacuum vial with fluoro-rubber cap at 383 K. Comparison can be made
with normalized quantity of catalysts (in mg or in mmol marked in red).
It is well-accepted that Lewis acids can promote the conversion of
pyruvic aldehyde to lactic acid.[17] Their catalytic performance (yield
and rate) depends on the LUMO energy position in homogeneous
Lewis acid site or the lowest conduction band position in solid Lewis
acids.[17] Thus, this reaction can be used as a chemical probe to
illustrate the effectiveness of the exfoliated niobium oxide samples.
Fig. 7 shows the conversion of pyruvic aldehyde and yield of lactic
acid over exHSr2Nb3O10, exHCa2Nb3O10 and exHNb3O8. Despite the
small difference in catalytic activity of exHSr2Nb3O10 and exHNb3O8,
they consistently exhibit higher yield of lactic acid than
exHCa2Nb3O10 per gram or mole of sample per unit time. The
apparent low catalytic performance of exHCa2Nb3O10 can clearly be
attributed to its higher conduction band energy due to structural
distortion, which weakens the binding of pyruvic aldehyde to lead to
lactic acid. Thus, the polarization distortion of NbO6 by B site cation
appears to play an important role in determining the Lewis acidity of
the layered Nb oxide structures.
Fig. 8(b) shows the Raman spectra of bulk crystalline and
exfoliated HSr2Nb3O10, HCa2Nb3O10 and HNb3O8. It is interesting
to see the energetic lattice vibration of 950-990 cm-1, which can
be assigned to terminal Nb=O bond.[25] Thus, the signal of 961
cm-1 for exHCa2Nb3O10 is apparently higher in wavenumber than
that of exHSr2Nb3O10 (941 cm-1), presumably due to polarization
distortion NbO6 by the smaller size Ca2+ (B site) in the layer
perovskite structure, as discussed. However, it is interesting to
reveal that both bulk HNb3O8 and exHNb3O8 give the most
intense and sharp signal at 981 cm-1 which clearly suggest the
ighest content of Nb=O. It is not yet known the reason why
HNb3O8 structure can give rise to a higher degree of Nb=O
(reflected by Raman) and Ov (reflected by ESR) defects
This article is protected by copyright. All rights reserved.

10.1002/cctc.201601131
ChemCatChem
FULL PAPER
presumably their formation may relate to the electronic and
structural effects caused by the edge sharing octahedra. Such
simultaneous formation of terminal Nb=O and Ov of NbO6 in
HNb3O8 structure could maintain the oxidation state of Nb (+5).
Static photo-luminescence (PL) and time resolved photo-
luminescence (TRPL) analyses have therefore been conducted to
reveal the fundamental photo-electron excitation and energy transfer
processes in the time domain. Fig. 9(a) shows the PL spectra of
exHNb3O8, exHSr2Nb3O10 and exHCa2Nb3O10. Clearly, exHNb3O8
with the highest content of Nb=O terminal bonds gives the highest
PL signal 400-700 nm followed by exHCa2Nb3O10 and then
exHSr2Nb3O10 upon 266 nm excitation. Thus, it is evident that Nb=O
can act as a ‘chromophore’ to take up this photon energy which can
then be transferred to radiative processes from 400- 700 nm through
lattice relaxation. TRPL decay curves recorded during the UV
irradiation are shown in Fig. 9(b). Table 1 summarizes the decay
constants and amplitudes of each decay components. There is a fast
decay component of the radiative recombination of the excited
electrons and holes due to relaxation of the excited Nb=O structure.
The relaxation times of excited Nb=O bond in exHNb3O8,
exHCa2Nb3O10 and exHSr2Nb3O10 are inversely related to their
Raman vibration frequencies (Nb=O bond strength) as: (165 ps, 981
cm-1), (343 ps, 961 cm-1) and (552 ps, 941 cm-1), respectively. On
the other hand, there is also a slow decay observed, which is a
component ascribed to the recombination of excited electrons and
holes of Nb-O. The decay constant of the fast component is
measured to be 165 ps of exHNb3O8, which is substantially faster
than that of exHSr2Nb3O10 (552 ps) and exHCa2Nb3O10 (343 ps). In
addition, its amplitude in the exHNb3O8 case is more than 7-fold
higher than that of exHSr2Nb3O8 and exHCa2Nb3O10, reflecting the
largest amount of Nb=O chromophore. In contrast, the slow
components (~1.5- 2.0 ns) show no pronounced difference in terms
of lifetime and amplitude for all the three exfoliated samples. It is
thought that the co-existence of Ov in a close proximity in exHNb3O8
may offer a recombination centre or intermediate pathway for the
rapid relaxation of excited Nb=O after capturing the photon energy,
as shown in Scheme 1.
Table 1. Experimental measured TRPL parameters of exHNb3O8,
exHSr2Nb3O10 and exHCa2Nb3O10
.
Catalyst
Fast Component
Slow Component
Decay
Decay
Constant
(ps)
Amplitude
Constant
(ps)
Amplitude
exHSr2Nb3O10
exHCa2Nb3O10
exHNb3O8
552
343
165
1312
1382
9861
2033
1796
1532
1654
2098
3408
(a)
(b)
(i)
(ii)
(iii)
Interestingly, we note from Fig. 8(b) that there is no significant
change of the Raman Nb=O peak of exHNb3O8 at 981 cm-1 in
comparison to the bulk crystalline (982 cm-1) although exHSr2Nb3O10
and exHCa2Nb3O10 show clear red shifts by 16 cm-1 and 23 cm-1,
respectively after exfoliation. The 93Nb NMR spectra shown in Fig.
10 also provide similar structural information with that observed in
Raman spectra. The exfoliation of the samples causes a decrease in
the local symmetry; this gives a corresponding broadening in the
93Nb spectra. The chemical shift of 93Nb has been shown to depend
on the oxygen coordination of the niobium site.[26] The large positive
shift observed (~300 ppm) during exfoliation of the HSr2Nb3O10 and
HCa2Nb3O10 complexes is attributed to the relaxation of NbO6 units
during the exfoliation process. In contrast, the exHNb3O8 spectra
show no obvious shift. This result implies the formation of Nb=O
terminal bonds stabilise the niobium site and prevent structural
relaxation. It is noted from Table 2 that the broadening of the spectra
after exfoliation for the exHSr2Nb3O10 and exHCa2Nb3O10 mean they
can no longer be reliably deconvoluted to the substituted crystal
sites. This is expected as the symmetry has been dramatically
reduced due the distortion of the NbO6 units. Conversely, despite the
broadening of the exHNb3O8, the resonance can still be decovoluted
into two niobium sites: this is expected as the formation of Nb=O
terminal bonds would maintain some of the local symmetry. The
ESR data (Fig 8(a)) shows an appreciable increase in oxygen
vacancies for the exHNb3O8 sample, when compared to the lack of
shift in the corresponding SSNMR and Raman data, it can be
presumed that the oxygen vacancies also alleviate distortions of the
NbO6 units. As stated, many physiochemical properties of niobium
Figure 9. (a) PL spectra of (i) exHSr2Nb3O10, (ii) exHCa2Nb3O10 and (iii)
exHNb3O8 and (b) TRPL spectra of (i) exHSr2Nb3O10, (ii) exHCa2Nb3O10
and (iii) exHNb3O8. The decay curve is fitted by a red line according to the
+A2exp ( - x-x
)
-
x-x
0
0
following equation; y=y0+A1 exp
t1
t2
Catalyst
Fast Component
Decay
Slow Component
Decay
Constant
(ps)
Amplitude
Constant
(ps)
Amplitude
exHSr2Nb3O10
exHCa2Nb3O10
exHNb3O8
552
343
165
1312
1382
9861
2033
1796
1532
1654
2098
3408
Scheme 1. Schematic of an energy diagram depicting CB edge (mainly
contributed by vacant 4d band of Nb), VB edge (mainly contributed by
filled O2p band) and oxygen-vacancies with electron transfer under UV
irradiation and their relationship with HOMO levels of benzyl alcohol and
pyruvic aldehyde.
This article is protected by copyright. All rights reserved.

10.1002/cctc.201601131
ChemCatChem
FULL PAPER
oxides are derived from their structural distortion and presence of
defect sites. The exposure of [NbO] sites by exfoliation is useful to
enhance these properties provided that the structure is still
maintained. Notably, the structure rigidity of HNb3O8 structure is
clearly retained after the exfoliation process.
(exHNb3O8≥ exHSr2Nb3O10 >> exHCa2Nb3O10) is not consistent with
the order of catalytic performance in this photo-oxidation reaction.
Similarly, photocatalytic hydrogen evolution over Pt loaded
exfoliated samples has also been evaluated under UV excitation
using sacrificial reagent (methanol).[27] No H2 production was
collected over the exfoliated samples without Pt. Fig. 11(b) clearly
shows the results of hydrogen evolution rates in the presence of the
three exfoliated samples upon irradiation. The catalytic rate order for
the Pt loaded sample are exHNb3O8> exHCa2Nb3O10> exHSr2Nb3O10,
which also corroborates well with that of photocatalytic oxidation of
benzyl alcohol. We note that the photocatalysis of the two reactions
clearly reflects the same order as their relative content of Nb=O and
Ov. It is believed that the presence of Nb=O as ‘chromophore’ is
responsible to capture the photon while the presence of Ov can offer
trapping of oxygenated substrate.
Figure 10. The 93Nb static NMR for (a), (b) HNb3O8 crystal at 14.1 and
19.9 T, (c) exHNb3O8, (d), (e) HSr2Nb3O10 crystal at 14.1 and 19.9 T, (f)
exHSr2Nb3O10, (g), (h) HCa2Nb3O10 crystal at 14.1 and 19.9 T and (i)
exHCa2Nb3O10
.
The simulation of the individual deconvoluted
resonances is shown below each spectra and their sum is given in red.
Figure 11. (a) Catalytic yields of photooxidation of benzyl alcohol into
benzaldehyde after 20 h. A mixture of benzyl alcohol (0.0258 ml) and
powder catalyst (20 mg) were added to benzotrifluoride (BTF) (7.5 mL)
and stirred under UV irradiation at ambient temperature. Comparison
can be made with normalized quantity of catalysts (in mg or in mmol
marked in red), (b) Photocatalytic hydrogen evolution over various
catalysts. A mixture of co-solvents (water: 90 mL and methanol: 10 mL)
and catalyst (20 mg) were agitated under UV irradiation for 2 h at room
temperature.
Table 2. Experimentally measured 93Nb isotropic shift (δiso) data for the
crystal and exfoliated samples. To determine the δiso the line broadening
and shift parameters were allowed to iterate in the Topspin SOLA function,
all other NMR interactions were kept constant. Where no δiso could be
reasonably measured ‘na’ has been added to the table.

Niobate
System
Site
 /ppm
 /ppm
Considering the reaction scheme from energy perspective, it is
argued that the introduction of Ov in exHNb3O8 can compete with the
direct affinity for substrate of Lewis acidic Nb5+ to form a more stable
substrate-catalyst complex. Particularly, conjugated carbonyl group
from benzaldehyde is expected to give lower HOMO position than
the reactive pyruvic aldehyde (Scheme 1). The high HOMO energy
CRYSTAL
EXFOLIATED
Na
HSr2Nb3O10
A
B
C
−981
−996
Na
−1013
−794
Scheme 2. Proposed reaction mechanism of benzyl alcohol oxidation over
exfoliated niobium oxides under UV irradiation.
HCa2Nb3O8
HNb3O8
A
B
−1053
−1055
Na
−763
A
B
−1072
−1046
−1063
−1050
Fig. 11(a) shows the catalytic performance for the photo-oxidation
of benzylalcohol to benzaldehyde[12] as a test reaction for the three
exfoliated samples. The exHNb3O8 sample gives 41 % yield of
aldehyde from the photo-oxidation of benzyl alcohol, which is
significantly higher than that of exHSr2Nb3O10 (18 %) and
exHCa2Nb3O10 (29 %) with the same amount of catalyst used (0.05
mmol). It is generally thought that substrate adsorption takes place
mainly on a Lewis acid site in a photocatalyst.[12] However, Lewis
acid catalysed reactions can be fundamentally different from
photocatalysis. Interestingly, it is apparent that the order of Lewis
acidity of the three exfoliated samples previously derived
of pyruvic aldehyde is relatively favourable to interact with CB edge
of Nb Lewis acid site directly but only at excited state whereas
benzaldehyde is anticipated to interact relatively more easily with
oxygen vacancy site at compatible energy level to effective binding
of this substrate for sufficient time to induce photocatalysis.
This article is protected by copyright. All rights reserved.

10.1002/cctc.201601131
ChemCatChem
FULL PAPER
As a result, Scheme 2 summarises a proposed mechanism of
superior photocatalytic activity for benzyl alcohol oxidation over
exHNb3O8 layers from this present study. First, the oxygen vacant
site strongly binds the hydroxyl group in benzyl alcohol. This allows
the ‘chromophore’ group (Nb=O) in close proximity to transfer its
photon energy for chemical conversion. The UV photon energy
captured by Nb=O produces excited electrons on Nb5+ Lewis acid
site to interact with adsorbed benzyl alcohol specie via conjugated
phenolic ring and prolongs charge separation in this stable LA-
complex (photocatalytic reduction). Under this prolonged period of
time, electron deficient terminal and bridging oxygens can abstract
hydrogen atoms from the adsorbed benzyl alcohol as shown in the
scheme 2(iv). The activation of a di-oxygen molecule from the air by
the regenerated oxygen vacant site after desorption of benzaldehyde
can also take up the hydrogen atoms forming hydrogen peroxide as
As for the synthesis of the Pt nanoparticles,
a mixture of
hexachloroplatinic acid; H2PtCl6•6H2O (62.1 mg), PVP (Mw=45000)
(133 mg) and ethanol (180 ml) were refluxed for 3 h. The solvent
was evaporated and the residue was dispersed into water/ethanol
co-solvents (Vwater/Vethanol=1:1) (50 ml).[31] Pt loading on catalysts was
prepared by sonicating the mixture of niobium oxides and Pt colloidal
solution for
3 h. Following that this dispersed solution was
centrifuged at 5000 rpm for 5 min and the obtained precipitate was
washed by water and ethanol. Finally, the filtrated samples were
naturally dried at ambient temperature for a few days. In order to
remove the PVP residue the obtained samples were placed under
UV irradiation, which led to photo-decomposition of the PVP layer
that encapsulated the nanoparticles.[32]
a
side product (photocatalytic oxidation).[12] For the hydrogen
Catalytic reactions
evolution from water splitting, it is thought that for the capture of
photons by Nb=O in exHNb3O8 allows the excited electrons to
migrate to Pt surface (2 nm Pt particle, shown in Fig. S2) at the
materials interface for catalytic reduction of H+ in appropriate
timescale[27] while the hole (positively charge terminal oxygen) will
oxidise the adsorbed methanol at the prolonged relaxation time.
Hydrolysis of ethyl acetate into ethanol was carried out in a sealed
Pyrex tube containing the mixture of ethyl acetate (1.9 mL), water
(0.1 mL) and catalyst (0.12 mmol). After 20 h at 333 K, ethanol
product was extracted into water layer by adding water (5 mL) into
the product solution, followed by its analysis using high performance
liquid chromatography (HPLC). Catalytic activities of exfoliated
niobium oxides were examined for the conversion of pyruvic
aldehyde to lactic acid in water. This reaction involved heating the
mixture of catalyst (0.05 g) and 2.0 mL of aqueous pyruvic aldehyde
(0.1 M) in a sealed Pyrex tube at 383 K for 2 h. The product solution
was analyzed by HPLC. For photocatalytic oxidation of benzylic
alcohol to benzylaldehyde reaction, a mixture of benzyl alcohol
(0.0270 g), catalytst (20 mg) and benzotrifluoride (BTF) (7.5 mL)
were stirred in a tight closed flask (25 mL) under UV irradiation for
20 h after the solution was pre-saturated with di-oxygen molecule.
The resultant solution was then analyzed by 1H NMR (Bruker
Avance III HD nanobay 400MHz NMR).
Conclusions
Exfoliated crystalline niobium oxides containing exposed but
interconnected NbO6 octahedra with different degrees of structural
distortions and defects have been demonstrated to catalyse
Brønsted acid (BA), Lewis acid (LA) and Photo catalysed (PC)
reactions efficiently. It is shown that the catalytic performances of
these reactions are highly dependent on the structural connectivity of
the NbO6 octahedra, their distortions and defects. Their structure-
activity relationships are clearly presented. We believe that this
study could lead to understanding and rational optimization of Nb
oxides structures for more efficient catalysis in these reactions.
For the hydrogen evolution experiments, the reaction was
conducted by dispersed catalysts (20 mg) in a 100 mL co-solvent
((methanol (10 mL) and distilled water (90 mL)). Prior to that the
solution was bubbled with 5% CH4/Ar where CH4 was used as an
internal reference. This mixture was agitated under UV irradiation for
2 h and the obtained gas was analyzed by gas chromatography
(Agilent 7890A) fitted with a Flame Ionisation Detector (FID).
Experimental Section
Materials and Synthesis
KNb3O8, KSr2Nb3O10 and KCa2Nb3O10 as chemical precursors
were first synthesized by conventional solid state syntheses from
their physical mixture using stoichiometric amounts of K2CO3, Nb2O5,
Sr2CO3 and Ca2CO3. [28-30] The protonated forms were then produced
by agitation of these obtained powders in excess 4M HCl solution
repeatedly over 24 h. The preparation of exfoliated samples was
conducted by using an intercalator reagent, the TBAOH (tetra butyl
ammonium hydroxide) solution. The host crystals of HNb3O8,
HSr2Nb3O10 and HCa2Nb3O10 possessing layer structures were held
by acidic protons within their crystallographic planes. When they
were soaked in aqueous TBAOH solution (pH > 11.0), the amine
molecules were known to insert into each layers, resulting in layer
segregation by weakening the interlayer stacking.[5,6] After
centrifuging at 3,000 rpm for 5 min and the supernatant solution was
mixed with 1M HNO3, giving a white precipitate immediately. The
obtained precipitate was then washed with distilled water until the pH
of the filtrated solution became neutral and the sample was then
dried at 80 °C.
NMR measurements
All solid state 93Nb NMR spectra were recorded at 14.1 T (ν0
=
146.71 MHz) and 19.9 T (ν0 = 208.10 MHz) by Bruker Avance II+
and Avance III spectrometers, respectively. A Bruker static 5 mm
broadband probe was used at each field. The spectra were
referenced to the primary standard; saturated K[NbCl6] in dry
acetonitrile (δiso = 0 ppm).[33] A solid echo (π/4 – τ – π/4) pulse
sequence employing a 250 kHz non-selective pulse was utilised,
with a relaxation time of 500 ms determined and a minimum of 4096
transients were collected for each experiment. The resonances with
a width of over 250 kHz were collected via a variable offset
cumulative frequency stepped (VOCS) method.[34,35] Spectral
simulations were completed using the Topspin SOLA function. To fit
the exfoliated samples, the isotropic shift (δiso) and broadening
parameters where allowed to iterate whilst the CSA and quadrupolar
determined interactions were fixed to the previously determined bulk
This article is protected by copyright. All rights reserved.

10.1002/cctc.201601131
ChemCatChem
FULL PAPER
crystal parameters (Table S2). If it was no longer viable to
distinguish the resonance (due to the dramatic broadening caused
by exfoliation), then that fit was removed from the deconvolution.
For CV measurements (Ivium, CompactStat), the scan was
recorded by measuring the potential across the membrane hosting
catalysts on the tip of electrode, which was prepared by drying the
mixture of catalysts (10 mg), Nafion (Nafion D-521 dispersion, 5%
w/w in water and 1-propanol, >0.92 meq/g exchange capacity) (10
μl) and milliQ water (1 ml) in order to stabilize the membrane.
Electrolyte used was 0.1 M aqueous KCl solution. A sweeping
voltage in the range of 0.3 V to -1.4 V at 50 mV s-1 was used.
The 1H MAS solid state NMR data was collected at 14.1 T (ν0
=
599.44 MHz) by a Bruker Avance II+ spectrometer. A 30 kHz MAS
frequency was facilitated by a Bruker 2.5 mm double air bearing HX
probe. The spectra were referenced to the secondary standard
alanine which gives three resonances at δiso = 1.1 (CH3), 3.5 (CH)
and 8.5 (COOH) ppm with respect to TMS (δiso = 0 ppm).
Acknowledgements
TRPL measurements
We are grateful to the EPSRC for supporting the work at the
University of Oxford. YK would like to acknowledge the postdoctoral
funding from the Uehara Memorial Foundation Research Fellowship
to enable him to work at Oxford. J.V.H. thanks the University of
Warwick, EPSRC and the Birmingham Science City for access to the
The sample was excited by a frequency-trippled Ti:Sapphire laser
(266 nm) giving ~150 fs pulses with a repetition frequency of ~7.57
MHz. This beam was passed through a reflecting x36 objective lens
(0.5 NA) onto the sample. The emitted signal was passed back
through the same lens to an Andor Shamrock 303i spectrometer
using an iDus 420 CCD. For time-resolved PL, the beam was
directed to an output slit in the spectrometer (selecting a window of
~10 nm) and passed to a PMT detector (Becker and Hickl PMH-100,
with an IRF ~ 200 ps) connected to a PicoQuant TimeHarp 260
TCSPC card (resolution 25 ps).
14.1
T solid state MAS NMR instrumentation used in this
research. The latter was funding obtained through the Birmingham
Science City Advanced Materials Project 1: Creating and
Characterising Next generation Advanced Materials project, with
support from Advantage West Midlands (AWM) and the European
Regional Development Fund (ERDF).
In addition, J.V.H
acknowledges the UK 850 MHz Solid State NMR National Facility
also used in this research which was funded by EPSRC, BBSRC
(contract reference PR140003) and the University of Warwick, which
included partial funding through Birmingham Science City Advanced
Materials Projects 1 and 2 supported by Advantage West Midlands
(AWM) and the European Regional Development Fund (ERDF).
Collaborative assistance from the UK 850 MHz Facility Manager
(Dinu Iuga, University of Warwick) is also acknowledged.
XPS measurements
XPS measurements were recorded on a Thermo Scientific K-Alfa
XPS instrument equipped with micro-focused monochromated Al X-
ray source. The source was operated at 12 keV and a 400 micron
spot size was used. The analyzer operated at the analyzer energy
(CAE) of 200 eV for survey scans and 50 eV for detailed scans.
Charge neutralization was applied using a combined low energy/ ion
flood source. The data acquisition and analysis were conducted with
CasaXPS (Casa software Ltd.). The peak position was referenced to
C1s peak of the carbon tape at 285.00 eV[36] and peak fitting was
applied by using Lorentzian / Gaussian (L/G) 30 % curve.
Keywords: Niobium oxide • Lewis and Brønsted acidity •Photo
oxidation • oxygen vacancy
[1]
[2]
[3]
[4]
[5]
[6]
[7]
Y. Zhao, C. Eley, J. Hu, J. S. Foord, L. Ye, H. He and S. C. Tsang,
Angew. Chem., 2012, 51, 3846-3849.
UV-vis, XRD, BET surface area analyzer, Raman spectra,
XPS, CV, EPR and TEM measurements
B. Varghese, S. C. Haur and C. T. Lim, J. Phys. Chem. C, 2008, 112,
10008-10012.
D. D. Yao, R. A. Rani, A. P. O’Mullane, K. Kalantar-zadeh and J. Z.
Ou, J. Phys. Chem. C, 2014, 118, 476-481.
Crystalline phase of the samples were identified by powder XRD; X’
pert Pro (PANalytical) operating in Bragg-Brentano focusing
geometry and using Cuka radiation (λ= 1.5418 Å) from a generator
operating at 40 kV and 40 mA. TEM images of the samples were
recorded by JEOL 2010 electron microscope with a high resolution
pole piece. The UV-vis adsorption spectra were recorded with UV-
vis spectrometer (Lambda 750S, Perkin Elmer) with potassium
bromide as the reference.
R. A. Rani, A. S. Zoolfakar, J. Z. Ou, M. R. Field, M. Austin and K.
Kalantar-zadeh, Sens. Actuators B: Chem., 2013, 176, 149-156.
A. Dias, S. Lima, D. Carriazo, V. Rives, M. Pillinger and A. Valente, J.
Catal., 2006, 244, 230-237.
C. Tagusagawa, A. Takagaki, S. Hayashi and K. Domen, J. Phys.
Chem. C, 2009, 113, 7831-7837.
(a) Q. Wu, Y. Yan, Q. Zhang, J. Lu, Z. Yang, Y. Zhang, Y. Tang
ChemSusChem, 2013, 6, 820-825. (b) Z-J. Yang, Y-F. Li, Q-B. Wu, N.
Ren, Y-H. Zhang, Z-P. Liu, Y. Tang, J. Catal., 2011, 280, 247-254.
A. Corma, M.E. Domine and S.J. Valencia, J. Catal., 2003, 215, 294-
304.
Specific surface area was derived from the corresponding
adsorption isotherms with a conventional BET nitrogen adsorption
apparatus (micromeritics 3flex). The samples were dried at 150 °C
under vacuum for 24 h prior to the experiments.
[8]
[9]
M. Renz, T. Blasco, A. Corma, V. Fornés, R. Jensen and L. Nemeth,
Chem. Eur. J., 2002, 8, 4708-4717.
[10] Y. Román-Leshkov, M. Moliner, J. A. Labinger and M. E. Davis, Angew.
Chem., 2010, 49, 8954-8957.
Raman spectra were measured with via Raman Microscope
(Renishaw) with a laser excitation wavelength of 532 nm. Exposure
time of 10 sec and 8 number scans were adopted for each
experiment. These samples were also dried at 150 °C before the
measurements.
[11] K. Nakajima, Y. Baba, R. Noma, M. Kitano, J. N. Kondo, S. Hayashi
and M. Hara, J. Am. Chem. Soc., 2011, 133, 4224-4227.
[12] G. Niu, X. Guo and L. Wang, J. Mater. Chem. A, 2015, 3, 8970-8980.
[13] S. Liang, L. Wen, S. Lin, J. Bi, P. Feng, X. Fu and L. Wu, Angew.
Chem., 2014, 53, 2951-2955.
[14] K. Maeda and T. E. Mallouk, J. Mater. Chem., 2009, 19, 4813.
This article is protected by copyright. All rights reserved.

10.1002/cctc.201601131
ChemCatChem
FULL PAPER
[15]
K. Akatsuka, G. Takanashi, Y. Ebina, N. Sakai, M.-a. Haga and T.
Sasaki, J. Phys. Chem. Solids, 2008, 69, 1288-1291.
[16] (a) A. Takagaki, M. Sugisawa, D. Lu, N. J. Kondo, M. Hara, K. Domen
and S. Hayashi, J. Am. Chem. Soc., 2003, 125, 5479-5485. (b) A.
Takagaki, D. Lu, N. J. Kondo, M. Hara, S. Hayashi and K. Domen,
Chem. Mater., 2005, 17, 2487-2489.
[17]
Y. Koito, K. Nakajima, H. Kobayashi, R. Hasegawa, M. Kitano and M.
Hara, Chem. Eur. J., 2014, 20, 8068-8075.
[18] R. López and R. Gómez, J. Sol. Gel. Sci. Tech., 2012, 61, 1-7.
[19] Y. Matsumoto, J Solid State Chem., 1996, 126, 227-234.
[20] (a) M. Wiegel, M. H. J. Emond, E. R. Stobbe and G. J. Blasse, Phys. J.
Phys. Chem. Solids 1994, 55, 773-778 (b) G. Blasse and L. G. J. de
Haart, Mater. Chem. Phys. 1986, 14, 481-484
[21]
A. L. Linsebigler, G. Lu and J. T. Yates Jr, Chem. Rev. 1995, 95, 735
[22] F. Liao, Y. Huang, J. Ge, W. Zheng, K. Tedsree, P. Collier, X. Hong and
S. C. Tsang, Angew Chem., 2011, 50, 2162-2165.
[23] A. Folli, J. Z. Bloh, A. Lecaplain, R. Walker and D. E. Macphee, Phys.
Chem. Chem. Phys. 2015, 17, 4849-4853.
[24] J. Kiwi, J. T. Suss and S. Szapiro, Chem. Phys. Lett., 1984, 106, 135-
138.
[25]
L. J. Burcham, J. Datka and I. E. Wachs, J. Phys. Chem. B, 1999, 103,
6015-6024.
[26] J. V. Hanna, K. J. Pike, T. Charpentier, T. F. Kemp, M. E. Smith, B. E.
Lucier, R. W. Schurko and L. S. Cahill, Chem. Eur. J., 2010, 16, 3222-
3239.
[27]
J. Xiong, L. Wen, F. Jiang, Y. Liu, S. Liang and L. Wu, J. Mater. Chem.
A, 2015, 3, 20627-20632.
[28] T. Ban, S. Yoshikawa and Y. Ohya, J. Colloid Interf. Sci., 2011, 364, 85-
91.
[29]
Y.-S. Han, I. Park and J.-H. Choy, J. Mater. Chem., 2001, 11, 1277-
1282.
[30] K. Saruwatari, H. Sato, T. Idei, J. Kameda, A. Yamagushi, A. Takagaki
and K. Domen, J Phys. Chem. B, 2005, 109, 12410-12416.
[31]
R. M. Rioux, H. Song, J. D. Hoefelmeyer, P. Yang and G. A. Somorjai,
J Phys. Chem. B, 2005, 109, 2192-2202.
[32] L. R. Baker, G. Kennedy, M. Van Spronsen, A. Hervier, X. Cai, S. Chen,
L. W. Wang and G. A. Somorjai, J. Am. Chem. Soc., 2012, 134, 14208-
14216.
[33] R. K. Harris, E. D. Becker, S. M. Cabral de Menezes, R. Goodfellow and
P. Granger, Magn. Reson. Chem., 2002, 40, 489-505.
[34] K. Akatsuka, G. Takanashi, Y. Ebina, M.-a. Haga and T. Sasaki, J. Phys.
Chem. C, 2012, 116, 12426-12433.
[35] G. J. Rees, S. T. Orr, L. O. Barrett, J. M. Fisher, J. Houghton, G. H.
Spikes, B. R. Theobald, D. Thompsett, M. E. Smith and J. V. Hanna,
Phys. Chem. Chem. Phys. 2013, 15, 17195-17207.
[36] M. Descostes, F. Mercier, N. Thromat, C. Beaucaire and M. Gautier-
Soyer, Appl. Surf. Sci., 2000, 165, 288-302.
This article is protected by copyright. All rights reserved.

10.1002/cctc.201601131
ChemCatChem
FULL PAPER
This article is protected by copyright. All rights reserved.