Electronic-State Manipulation of Surface Titanium Activates Dephosphorylation Over TiO2 Near Room Temperature

Article abstract of DOI:10.1002/anie.202104397

Dephosphorylation that removes a phosphate group from substrates is an important reaction for living organisms and environmental protection. Although CeO₂ has been shown to catalyze this reaction, cerium is low in natural abundance and has a na

Full text of DOI:10.1002/anie.202104397

A Journal of the Gesellschaft Deutscher Chemiker
Angewandte
Chemie
International Edition
www.angewandte.org
Accepted Article
Title: Electronic State Manipulation of Surface Titanium Activates
Dephosphorylation over TiO2 Near Room Temperature
Authors: Quan Wang, Xianfeng Yi, Yu-Cheng Chen, Yao Xiao, Anmin
Zheng, Jian Lin Chen, and Yung-Kang Peng
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: Angew. Chem. Int. Ed. 10.1002/anie.202104397
Link to VoR: https://doi.org/10.1002/anie.202104397

10.1002/anie.202104397
Angewandte Chemie International Edition
RESEARCH ARTICLE
Electronic State Manipulation of Surface Titanium Activates
Dephosphorylation over TiO2 Near Room Temperature
Quan Wang,[a] Xianfeng Yi,[b] Yu-Cheng Chen,[c] Yao Xiao,[b] Anmin Zheng,*[b] Jian Lin Chen[d] and
Yung-Kang Peng*[a,e]
[a]
[b]
Quan Wang, Prof. Yung-Kang Peng
Department of Chemistry, City University of Hong Kong, Hong Kong SAR.
E-mail: ykpeng@cityu.edu.hk
Dr. Xianfeng Yi, Yao Xiao, Prof. Anmin Zheng
State Key Laboratory of Magnetic Resonance and Atomic and Molecular Physics, National Center for Magnetic Resonance in Wuhan, Key Laboratory of
Magnetic Resonance in Biological Systems, Wuhan Institute of Physics and Mathematics, Innovation Academy for Precision Measurement Science and
Technology, Chinese Academy of Sciences, Wuhan 430071, China.
E-mail: zhenganm@wipm.ac.cn
[c]
[d]
[e]
Yu-Cheng Chen
Department of Mechanical Engineering, City University of Hong Kong, Hong Kong SAR, China.
Prof. Jian Lin Chen
Department of Science, School of Science and Technology, The Open University of Hong Kong, Hong Kong SAR, China.
Prof. Yung-Kang Peng
City University of Hong Kong Shenzhen Research Institute, Shenzhen 518057, China.
Supporting information for this article is given via a link at the end of the document.
Abstract: Dephosphorylation that removes a phosphate group from
substrates is an important reaction for not only living organisms but
also environmental protection. Although CeO2 has been shown to
catalyze this reaction, as one of the rare earth elements, cerium is
low in natural abundance and has a narrow global distribution (>
90% of these reserves are located within six countries). It is thus
imperative to find another element/material with high worldwide
abundance that can also efficiently extract the phosphate out of
agricultural wastes for phosphorus recycle. Using para-nitrophenyl
phosphate (p-NPP) as the model compound, we demonstrate herein
that TiO2 with F-modified (001) surface can activate p-NPP
dephosphorylation at temperature as low as 40 °C. By probe-
assisted nuclear magnetic resonance (NMR), it was revealed that
the strong electron withdrawing effect of fluorine makes Ti atoms
(i.e., the active sites) on (001) surface very acidic. The bidentate
adsorption of p-NPP on this surface further promotes its subsequent
activation with barrier ~20 kJ/mol lower than that of pristine (001)
and (101) surfaces, allowing the activation of this reaction at nearly
room temperature (from > 80 °C).
It is thus imperative to develop robust catalysts that can extract
the phosphorus out of phosphorylated wastes in an ecologically
sustainable manner.
Scheme 1. Schematic illustration of catalytic p-NPP dephosphorylation over
CeO2.
Pioneered by Tan et al in 2008, CeO2 nanoparticles (NPs)
were adopted to mimic phosphatases catalyzing the
dephosphorylation
of
various
phosphopeptides[5].
The
phosphatase-like activity of CeO2 NPs was attributed to the
activation of targets’ phosphoryl oxygen by surface Ce for later
SN2 hydrolysis (Scheme 1). The phosphate anions can thus be
extracted from animal/agriculture wastes and recycled for further
applications to achieve phosphorus sustainability[1-4]. Using para-
nitrophenyl phosphate (p-NPP) as a model molecule, Tan et al.[5]
firstly suggested a surface Ce4+-mediated pathway while later
Kuchma et al.[6] found the activity is proportional to the
concentration of surface Ce3+. A balanced surface Ce4+/Ce3+
ratio[7] and surface oxygen vacancy[8] were also proposed the
key factor facilitating this CeO2-catalyzed dephosphorylation.
Although a wide range of Ce-based catalysts have been
reported to catalyze this reaction[9-12], the key surface Ce species
and the underlying mechanism are still unclear. Using probe-
assisted nuclear magnetic resonance (NMR), we recently
demonstrated that the electronic state (i.e., Lewis acidity) of
surface Ce shifts “continuously” with its local structure and
cannot be ascribed to the “discrete” oxidation state (e.g., 3+ and
Introduction
Phosphorus is the crucial element in living organisms that
involves many biologically important processes such as plant
growth, energy storage, photosynthesis, cell regulation and
signaling etc[1]. Phosphatase, a natural enzyme widely found in
living systems, can remove phosphate groups from organic
substrates into inorganic phosphates and hence play
a
significant role in the recycle of phosphorus[2]. The majority of
the environmental phosphorus came from animal wastes before
the 18th century. Nowadays, more than 80% phosphorus used in
fertilizers relies on the non-renewable phosphate rocks[3]. Given
that the demand for phosphorus fertilizers continues to grow with
population, the global reserve of phosphate rocks is estimated to
be depleted within this century[4]. However, it is not practical to
use natural enzymes (i.e., phosphatase) for the recycle of
environmental phosphorus since they suffer from easy
denaturation and high costs in both preparation and purification.
1
This article is protected by copyright. All rights reserved.

10.1002/anie.202104397
Angewandte Chemie International Edition
RESEARCH ARTICLE
4+ for Ce) as in literatures above[13]. The surface Ce with strong
Lewis acidity was found to facilitate not only the adsorption of p-
NPP but also its subsequent activation.
and (101) terminal facets have been prepared by varying the
amount of HF added during hydrothermal synthesis for the study
of facet activity in a wide range of applications[19, 20]. The extra
addition of alcohols was recently found can not only make them
well-defined but also fine-tune the (001)/(101) ratio by affecting
Although cerium is the most abundant rare earth element,
it comprises only 0.0046 wt% of the Earth's crust. Most
importantly, the global distribution of rare earth elements is
extremely heterogeneous. According to estimates by the U.S.
Geological Survey, more than 90% of these reserves are located
within six countries: China (36.6%), Brazil (18.3%), Vietnam
the hydrolysis-condensation rate of titanium(IV) butoxide[21-24]
.
(18.3%), Russia (10%), India (5.75%) and Australia (2.75%)[14]
.
From sustainable point of view, it is crucial to find another
element with high worldwide abundance that can also extract the
phosphate out of agricultural wastes for phosphorus recycle.
Titanium is the ninth abundant element comprising about 0.565
wt% of the Earth's crust (i.e., more than 100 times of cerium). It
is much cheaper and can be found in nearly all rocks/sediments
among the crusts of Earth. Over the past decades, its oxide form,
TiO2, has received a tremendous attention due to many
important applications ranging from the conventional areas (e.g.,
paint, cosmetic, toothpaste, etc.) to the advanced photo-electro
areas[15]. The acid-base applications of TiO2 NPs are equally
important however less explored presumably due to the
moderate Lewis acidity of surface Ti atoms among transition
metal oxides.
Figure 1. TEM images of TiO2 (a) F-NS-1, (b) F-NS-10, (c) and (d) are their F
removed samples denoted as NS-1 and NS-10.
Given that each terminal facet of a crystallite possesses
distinct surface energy, the Lewis acidity of surface cations is
believed to vary with host facets and hence influence the
Herein, TiO2 samples with different (001)/(101) ratio were
prepared with a fixed amount of HF (i.e., 0.8 mL) in the presence
of 1 mL (denoted as F-NS-1) and 10 mL (denoted as F-NS-10)
methanol (see experimental section for details). F-NS-1 (Figure
1a) and F-NS-10 (Figure 1b) indeed reveal different
morphologies as evidenced by their corresponding TEM images.
The anatase structure of both samples was confirmed by X-ray
diffraction (XRD) analysis in Figure S1. This result can also be
supported by the lattice spacings of 0.235 nm found in these two
samples (Figure S2), corresponding to the (001) crystallographic
plane of anatase TiO2. The percentage of (001) and (101) facet
exposed in both samples was further estimated by the Wulff
construction model (Figure S3)[25, 26]. As summarized in Table S1,
more than 70 % of F-NS-1 surface is enclosed by (001) facet
while F-NS-10 preferentially exposes (101) facet (i.e., 64.7 %).
Note in this table that the percentage of the dominant facet for
each sample is more than doubled than the minor facet of the
other sample. Although the same amount of HF was used for
their preparation (i.e., 0.8 mL), the resulting fluorine
concentration is believed proportional to the area of their (001)
surface. Indeed, F-NS-1 with dominant (001) surface shows a
stronger F signal than F-NS-10 in the corresponding X-ray
photoelectron spectroscopy (XPS) F1S spectra (Figure S4).
NaOH wash was used herein to remove F from the (001)
adsorption/activation of coming reactant[15,16]
adsorbed species (e.g., surfactant)
.
Surface pre-
with different
electronegativity can further modify this property on a given
facet[17]. Accordingly, we manipulate the electronic state of
surface Ti in this study by tuning its local structure from anatase
TiO2(101) to TiO2(001) with different surface adsorbed species
(O, F). Fluorine was selected here as the only element more
electronegative (χ = 3.98) than oxygen (χ = 3.44) in the periodic
table. Interestingly, the activation energy for p-NPP
dephosphorylation is considerably reduced for TiO2 samples
with F-modified (001) (i.e., F-(001)) surface. Probe-assisted
NMR reveals that the Lewis acidity of Ti atoms on this surface is
comparable to super solid acids and hence activates p-NPP
dephosphorylation at temperature as low as 40 °C. Moreover,
the reaction rate was found to only associate with the area of F-
(001) surface for a given TiO2 crystallite while the rest surfaces
such as pristine (101) and (001) are both silent at temperature <
80 °C. In addition to NMR, techniques such as infrared (IR) and
transmission electron microscopy-electron energy loss
spectroscopy (STEM−EELS) were further adopted to unravel the
adsorption and activation structures of p-NPP on this particular
F-(001) surface.
surface of both samples (denoted as NS-1 and NS-10) as often
20, 27,
adopted in literatures[19,
28]. This post-treatment is very
effective on F removal that no apparent XPS F1S signal can be
observed for NS-1 and NS-10 (Figure S4) while keeping their
morphology (Figure 1c,d), crystallinity (Figure S1), Raman
structural fingerprint (Figure S5) and surface area (Figure S6
and Table S1) nearly unchanged.
Results and Discussion
Although anatase TiO2 are often covered with
thermodynamically stable (101) facet (0.44 J m-2) in nature,
Yang et al. (2008) found that fluorine can specifically
bind/stabilize the high-energy (001) facet (0.90 J m-2) and hence
reverse the relative thermodynamic stability of these two
facets[18]. Since then, anatase TiO2 with different ratio of (001)
The dephosphorylation activity of F-NS-1, F-NS-10 and
their F removed samples (i.e., NS-1 and NS-10) was evaluated
2
This article is protected by copyright. All rights reserved.

10.1002/anie.202104397
Angewandte Chemie International Edition
RESEARCH ARTICLE
using para-nitrophenyl phosphate (p-NPP) as
a
model
Figure 2. (a) The UV-vis spectra collected for the F-NS-1 sample at 60 ℃ as a
function of time. (b) The time-dependent p-NP yield (defined as Ct/C0, where
Ct refers concentration of p-NP at a given time, and C0 refers to initial p-NPP
concentration) of TiO2 samples at 60 ℃ and (c) the corresponding kinetics
analysis. (d) Comparison of k constant obtained at 60 ℃ and (101)/(001)
surface area among TiO2 samples. (e) Arrhenius plots for the activation energy
of TiO2 samples and (f) the comparison with their corresponding (101)/(001)
surface area.
molecule[5-8,13]. Since the characteristic absorption peaks of p-
NPP (313 nm) and its hydrolyzed form para-nitrophenol (p-NP,
405 nm) are well-separated, the progress of dephosphorylation
can thus be tracked by UV–vis spectroscopy. Figure 2a shows
the time-dependent UV–vis spectra of p-NPP dephosphorylation
collected over F-NS-1 at 60 °C. As the reaction proceeds, the
signal of p-NP at 403 nm enhances pronouncedly at the cost of
p-NPP at 313 nm due to the difference in their absorption
coefficient. Since the adsorption of p-NPP varies a lot between
TiO2 samples (vide infra) while the product (i.e., p-NP) desorbs
easily from their surfaces (by water solvation), the generation of
p-NP is thus used for activity comparison and later kinetic study
at different temperatures. The yield of p-NP can be obtained by
a calibration plot with a series of p-NP solutions with known
concentration (Figure S7). As shown in Figure 2b, F-NS-1
provides the highest activity followed by F-NS-10 with the yield
of p-NP as 39.03% and 28.30% by 6 hours at 60 °C. While for
the F removed samples (i.e., NS-1 and NS-10) their p-NP yield
maintains at around 5% at this temperature and below (Figure
S8a,b). Note that the amount of p-NP produced also decreases
the pH of the solution as reaction proceeds (see Figure S9 for
detail). To test whether fluoride alone or fluoride with clean TiO2
can trigger the dephosphorylation of p-NPP, two sets of control
experiments were conducted with reaction solutions containing
(1) NaF (13.86 mg) alone and (2) NaF (13.86 mg) and NS-1 (6
mg). The amount of NaF added here was determined by the
concentration of HF used in the synthesis of F-NS-1. As shown
in Figure S10, the introduction of fluoride has a very minor effect
on the dephosphorylation of p-NPP compared to that of blank.
This can also be evidenced by the tiny change of p-NP yield
after adding NaF to solution containing NS-1. Since no leakage
of fluorine was detected by 19F NMR during the cyclic catalytic
testing of F-NS-1 (see Figure S11 for detail), the high k constant
obtained for this sample (also for F-NS-10) in Figure 2c thus
suggests that the activation of p-NPP over Ti atoms is due to the
presence of surface attached fluorine.
Since those Ti atoms are hosted by various surfaces, the
obtained k constant was further compared with the facet
distribution for a given TiO2 sample to identify their location
(Figure 2d). As expected, the k constant for the as-prepared
TiO2 samples (i.e., F-NS-1 and F-NS-10) was found proportional
to the area of F-(001) surface while no such correlation can be
observed for F removed samples (i.e., NS-1 and NS-10). For
example, the F-NS-1 with doubled area of F-(001) surface (cf. F-
NS-10, Table S1) provides k constant nearly two times higher
than that of F-NS-10 at 60 °C (Figure 2c). In fact, the F-(001)
surface of a given TiO2 sample can activate this reaction at
temperature as low as 40 °C (Figure S8b), which is in stark
contrast to pristine (001) and rest (101) surfaces catalyzing this
reaction at temperature higher than 80 °C (Figure S8e,f). This
can also be evidenced by their k constant in Table S2 since the
rate of an activated reaction should in theory double for every
10 °C rise in temperature. Note that this reaction without catalyst
becomes active at 95 °C with activity even comparable to clean
TiO2 samples (i.e., NS-1 and NS-10, Figure S8f). The k constant
of F-NS-1 and F-NS-10 at each temperature (Figure S12a) was
further normalized by their area of F-(001) surface in Figure
S12b. A similar ks constant (h-1·g·m-2) can be obtained for these
two samples at temperatures below 80 °C, confirming again that
the area of F-(001) surface determines the activity for TiO2
samples at low temperature. F-NS-10 provides a higher ks
constant above this temperature presumably due to its higher
area of (101) surface. The activation energy (Ea) for TiO2
samples was analyzed by Arrhenius plot (Figure 2e) and
compared with their facet distribution in Figure 2f. As expected,
F-NS-1 and F-NS-10 with different facet distribution show similar
Ea around 60 kJ/mol while their F removed samples both exhibit
20 kJ/mol higher (i.e., 80 kJ/mol). This result further indicates
that this reaction is facet-insensitive for pristine (101) and (001)
surfaces and the ~20 kJ/mol difference in Ea can be exclusively
attributed to the presence of F-(001) surface. This again
explains why the area of F-(001) surface determines the activity
for the as-prepared TiO2 samples at temperature higher than
40 °C while not much difference in activity can be observed for F
removed samples at all tested temperatures (Figure S8).
Accordingly, we can conclude that the presence of F-(001)
surface on a TiO2 catalyst significantly lowers the Ea for this
reaction and its area determines the observed activity.
It is believed that the electronic property (or Lewis acidity)
of surface Ti atoms affects the adsorption/activation of p-NPP
and hence the obtained activity. To gain more insight into the
electronic effect imposed by fluorine to surface Ti atoms,
TiO2(001) slabs w/o surface F were firstly analyzed in Figure 3.
Note that the relative electron density (ΔDelectron) for each surface
5-coordinated Ti (Ti5C) atom in this figure was obtained by
comparing to their 6-coordinated counterpart in the bulk (i.e.,
ΔDelectron = Delectron(Ti5C) - Delectron(Ti6C)). For pristine (001) surface
(Figure 3a), all Ti5C atoms exhibit similar ΔDelectron with a
magnitude of 6.79*10-4. Smaller magnitudes can be found for
3
This article is protected by copyright. All rights reserved.

10.1002/anie.202104397
Angewandte Chemie International Edition
RESEARCH ARTICLE
Ti5C atoms on F-(001) surface (Figure 3b), suggesting that the
between TiO2 samples w/o surface F because the pyridine is too
Ti5C atoms on this surface are more acidic (i.e., electron deficient) basic to distinguish those Ti5C atoms[29]. The evolution of BA
than the pristine (001) surface. The F-attached Ti6C indeed
shows the lowest ΔDelectron among Ti atoms on F-(001) surface
due to the formation of strong Ti-F bond (i.e., the only bond
stronger than Ti-O). However, the F-attached Ti6C should not
directly participate in this reaction since it has no vacant orbital
to interact with the coming p-NPP. Accordingly, the facilitated
adsorption/activation of p-NPP observed over F-NS-1 and F-NS-
10 should closely associate with those acidic Ti5C atoms in a
proximity to F on F-(001) surface.
sites on the as-prepared TiO2 samples (i.e., F-NS-1 and F-NS-
10) with a strong IR signal at 1490 cm-1 is due to the hydrogen
bonding stabilization of protons by surface fluorine (Figure 4c)
as evidenced by the decrease of this signal for F removed
samples (i.e., NS-1 and NS-10, Figure 4b). Although the charge
analysis in Figure 3 shows the strong electronic withdrawing
effect imposed by surface F to Ti5C atoms (i.e., the LA sites),
many researchers may mistakenly attribute the enhanced
activity to the surface F-induced BA sites since this is the only
difference among TiO2 samples they can observe using
conventional surface techniques. However, a control experiment
using conventional BA catalysts (i.e., HY zeolite) shows no
enhance in activity compared to that of F-NS-1 (Figure S13).
Figure 3. Side view and enlarged top view of the anatase TiO2(001) slabs
without (a) and with (b) surface fluorine (accompanied by an acidic proton).
The relative electron density (ΔDelectron) for each surface 5-coordinated Ti (Ti5C
)
atom was obtained by comparing to their 6-coordinated counterpart in the bulk
(i.e., ΔDelectron = Delectron(Ti5C) - Delectron(Ti6C)). The position of the slicing plane is
blue highlighted. Note that the coverage of F (25%) used here does not
represent the F coverage for both F-NS-1 and F-NS-10 samples (see below).
Conventional surface tools such as Ti2p XPS (Figure 4a)
and pyridine-infrared (Py-IR, Figure 4b) were adopted to unravel
the nature of Ti5C atoms predicted above on F-(001) surface.
Although XPS has been widely adopted as a surface-sensitive
tool to atom oxidation states, no shift of Ti2P signal can be
observed among TiO2 samples w/o surface F (Figure 4a). From
our point of view, the reactant molecule, p-NPP, interacts only
with Ti5C atoms on the topmost surface of TiO2 samples.
However, the sampling depth of XPS is depending on the
penetration length of photoelectrons and hence it often averages
signals > 10 atomic layers from the surface[16], resulting a very
similar XPS Ti2P spectra among TiO2 samples. Unfortunately,
many reports simply draw a conclusion that no change in the
Figure 4. (a) Ti2p XPS spectra of TiO2 samples. (b) IR spectra of TiO2 samples
before (dash line) and after (solid line) pyridine adsorption. (c) Schematic
illustration of the interaction between probe molecules (Py and TMP) with
different acidic sites on TiO2 surface. (d) TMP-assisted 31P NMR spectra of
TiO2 samples. (e) A correlation plot showing the linear relationship between
the δ31P of TMP-adsorbed TiO2(001) facet w/o F and the corresponding
calculated adsorption energy.
electronic state of surface Ti atoms before/after F removal based
19,
on their XPS result[15,
20]. This may explain why different
The use of 31P NMR with trimethylphosphine (TMP) as the
probe molecule has been demonstrated by our group a
promising method for the surface study of various metal
oxides[30-33]. In general, the 31P signal with chemical shift (δ31P)
between 0 and -5 ppm can be attributed to the formation of
TMPH+ species due to the surface BA site, while the
coordination of TMP to a surface cation can generate δ31P in a
wider range from -20 to -60 ppm depending on its LA strength
(or electron density)[34]. Herein, the degree of this Lewis acid-
base interaction (i.e., TMP↔Ti5C) determines the electron
density of 31P nuclei and hence the corresponding δ31P in 31P
NMR. A stronger TMP→Ti5C bond formation over Ti5C with high
interpretations and even disagreement often found among
researchers using F-capped TiO2 as catalyst. On the other hand,
pyridine, a Lewis base molecule, has been widely used with IR
for acid sites characterization of catalysts. In general, the
interaction between pyridine and surface Lewis acid (LA) sites
generates two dominant IR signals at 1450 cm-1 and 1600 cm-1
with a minor IR signal at 1490 cm-1 overlapped with surface
Brønsted acid (BA) sites[29]. As shown in Figure 4b, the presence
of surface LA sites (i.e., Ti5C atoms) for all TiO2 samples can be
confirmed by both dominant IR signals at 1448 cm-1 and 1608
cm-1. However, no shift in these two IR peaks can be observed
4
This article is protected by copyright. All rights reserved.

10.1002/anie.202104397
Angewandte Chemie International Edition
RESEARCH ARTICLE
LA strength would thus push the δ31P downfield (i.e., shift
towards positive ppm). Figure 4d shows the TMP-31P NMR
spectra of F-NS-1, F-NS-10 and their F removed samples. It is
clear that TMP-31P NMR can provide more surface details
compared with conventional XPS (Figure 4a) and IR (Figure 4b)
since the peak shift and change in intensity are both
distinguished among TiO2 samples w/o surface F. For the BA
range (i.e., δ31P at 0 ~ -5 ppm), F-NS-1 with higher F-(001)
surface area indeed shows a stronger signal than that of F-NS-
10, while no BA signal in this range can be observed for their F
removed samples (Figure 4d). This result again confirms above
observation that (1) surface F can induce BA sites on F-(001)
and (2) its removal by NaOH wash is very efficient with no BA
sites left. Since the electronic state of surface Ti5C atoms is
believed to vary with their host surfaces, one would expect to
see two distinct 31P signals in the LA range (i.e., δ31P at -20 ~ -
60 ppm) for each TiO2 sample. For example, the NS-1 sample
enclosed by 73.1%/26.9% of (001)/(101) surface shows a
dominant δ31P at -41.6 ppm and a minor δ31P at -31.6 ppm
(Figure 4d). These two δ31P signals can be individually assigned
to Ti5C atoms on pristine (001) and (101) surfaces. Considering
activated either synergistically by a BA (i.e., acidic proton) and a
LA site (i.e., Ti5C) or by two BA/LA sites.
that pristine (001) consists of 100% Ti5C while (101) provides
[15]
only 50% Ti5C
, this explains why NS-10 sample with
35.3%/64.7% of (001)/(101) surface displays similar intensity at
both δ31P (Figure 4d). Since F binds specifically to the (001)
surface, the deconvoluted δ31P at -28.5 ppm for F-NS-1 and F-
NS-10 samples (Figure 4d, with a fixed δ31P at -31.6 ppm) can
be attributed to the remaining Ti5c atoms on F-(001) surface with
F-Ti6c species nearby. The 13 ppm downshift for Ti5C atoms from
-41.6 ppm of pristine (001) to -28.5 ppm of F-(001) confirms the
electron withdrawing effect imposed by surface F to surrounding
Ti5C atoms as suggested in Figure 3. Since the adsorption
energy (Ead) of TMP on a given facet is believed to proportional
to its corresponding δ31P, the Ead of TMP was further calculated
as -1.73 eV for F-(001) and -1.2 eV for pristine (001) (Figure
S14), which indeed displays a good linear relationship with the
experimental δ31P value (Figure 4e). Although F-NS-1 has a
doubled area of F-(001) surface (cf. F-NS-10, Table S1), similar
F coverage (~40%) was obtained for both shapes as expected
(see SI for detailed calculation).
Since the adsorption (step 1) and activation (step 2) of
reactants on catalysts’ surface are both critical steps in
heterogeneous catalysis, the surface information revealed by
TMP-31P NMR can further provide insights into the interplay
between TiO2 samples and p-NPP. F-NS-1 and NS-1 with same
morphology, surface area and the (001)/(101) ratio (Table S1)
were thus picked for an in-depth comparison. The facilitated
adsorption of p-NPP on F-(001) surface was visualized by
Figure 5. (a) TEM-EELS Ti/O/F/P mapping of p-NPP adsorbed F-NS-1 and
NS-1 samples. (b) IR spectra of F-NS-1 and NS-1 samples before (dash line)
and after (solid line) p-NPP adsorption. (c) 31P NMR spectra of p-NPP
adsorbed F-NS-1 and NS-1 samples (the spectra of Na3PO4 and p-NPP
solution were included for comparison). (d) The relationship between 31P
chemical shift and the configuration of phosphate group with various chemical
environments.
31P NMR was further adopted to investigate the atomic
configuration of adsorbed p-NPP on F-(001) surface. As shown
in Figure 5c, the p-NPP adsorbed F-NS-1 shows a dominant
δ31P signal at -14.70 ppm and a minor one at -6.03 ppm, while
no apparent signals at these two positions can be observed for
NS-1. This result is in line with STEM-EELS (Figure 5a), IR
(Figure 5b) and Raman (Figure S15) studies above suggesting
the facilitated adsorption of p-NPP on F-NS-1. To evaluate the
electronic effect of BA (or acidc proton) to p-NPP phosphorus,
the 31P NMR spectra of phosphate compounds with different
degree of protonation was firstly recorded (Figure S16). A
continuous upshift of δ31P can be observed with the increased
3-
2-
degree of protonation from PO4 (Q0, 2.77 ppm), HPO4 (Q1,
scanning
TEM-electron
energy
loss
spectroscopy
2.59 ppm), H2PO4 (Q2, 0.18 ppm) to H3PO4 (Q3, 0 ppm). Note
-
(STEM−EELS). As shown in Figure 5a, F-NS-1 displays a strong
phosphorus signal from surface adsorbed p-NPP while this
signal is very weak for NS-1 with pristine (001) surface. The
considerable difference in p-NPP adsorption also reflects to their
IR (Figure 5b) and Raman (Figure S15) spectra. Both IR and
Raman fingerprint of p-NPP (black line) can be identified on F-
NS-1 (red line) while not for NS-1 (green line) after p-NPP
adsorption. The upshift of P=O vibration frequency from 1098
cm-1 to 1122 cm-1 (Figure 5b) further suggests the subsequent p-
NPP activation on F-(001) surface. Since the rest phosphoryl
oxygens of p-NPP can interact with up to two surface cations at
the same time, this allows surface-adsorbed p-NPP being
that the "i" of Qi represents the number of bridged oxygens in a
given phosphate compound. Each P-O→H bond formation
disturbs the electron structure of center phosphorus, making it
electron rich with an upshifted δ31P. Since the sequential
3-
protonation of PO4 only shifts the δ31P towards upfield within 3
ppm (Figure S16), this cannot explain the considerable shift of
δ31P from -0.47 ppm of p-NPP (Q1) to -6.03 ppm and -14.70 ppm
(Figure 5c). The possibility of the activation of p-NPP by BA sites
alone on F-(001) is therefore eliminated. This also explains why
HY zeolite is inactive at 60 °C (Figure S13). In stark contrast, a
considerable shift in δ31P has been reported when the oxygen
3-
atoms of PO4 is bonded with transition metal cations[35, 36]. The
5
This article is protected by copyright. All rights reserved.

10.1002/anie.202104397
Angewandte Chemie International Edition
RESEARCH ARTICLE
phosphorus of PO43- also becomes more and more shielded with
increasing number of surrounding cations. For example,
Knowles et al. incorporated TiO2 into the P2O5-based glass
samples and found the formation of each P-O→Ti bond shifts
the δ31P to upfield about 7 ppm[36]. Accordingly, the δ31P of the
adsorbed p-NPP at -6.03 ppm and -14.70 ppm (Figure 5c) can
be attributed to the formation of Q2 (monodentate) and Q3
(bidentate) configurations on F-(001) surface (Figure 5d). The
dominant δ31P signal at -14.70 ppm also suggests that p-NPP
preferentially adsorbs on this surface with Q3 configuration,
which weakens the phosphate ester bond for subsequent
dephosphorylation. We can thus conclude that the strong Lewis
acidity of Ti5C atoms on F-(001) surface not only promotes the
adsorption of p-NPP, the formation of Q3 configuration on this
surface further lowers the Ea for its subsequent activation. The
Ead of p-NPP with bidentate configuration on both surfaces was
also calculated (Figure S19). However, the result cannot explain
the considerable difference in the amount of surface p-NPP
observed between these two surfaces (Figure 5a-c and Figure
S15), presumably due to the involvement of solvent (see SI for
detailed discussion). Scheme 2 shows the proposed mechanism
for p-NPP adsorption/activation over TiO2(001) w/o surface F.
Most importantly, the activity of F-NS-1 was found even
comparable to CeO2 cube at 60 °C (Figure S13 and Figure S17).
See SI for a more detailed discussion on the activity comparison
with various CeO2 surfaces in literature (Figure S18 and Table
S3).
reactants and hence the obtained activity. However,
conventional probe-assisted IR provides very limited
information[29] for both catalyst’ surface (Py-IR, Figure 4b) and its
interplay with the reactant molecules (e.g., p-NPP, Figure 5b).
Taking p-NPP dephosphorylation over TiO2 as an example here,
we strongly suggest the adoption of advanced surface
characterizations such as probe-assisted NMR to ensure a solid
structure-activity correlation is unambiguously built.
Conclusion
In summary, we demonstrate that TiO2 with F-modified
(001) surface can activate p-NPP dephosphorylation at nearly
room temperature. As predicted by model analysis, the
electronic withdrawing effect of fluorine imposed on TiO2(001)
surface strongly manipulates the electronic state of surrounding
Ti5C atoms by making them very acidic with 13 ppm shift in TMP-
31P NMR. Those acidic Ti5C atoms on F-(001) surface were
found to facilitate the bidentate adsorption of p-NPP and its later
activation with an Ea of 60 kJ/mol, which is ~20 kJ/mol lower
than that of pristine (001) and (101) surfaces. The as-prepared
TiO2 samples with F-(001) surface can thus activate this reaction
at temperature as low as 40 °C (cf. F removed counterparts at >
80 °C). This also explains why the F-NS-1 sample with doubled
area of F-(001) surface provides k constant nearly two times
higher than that of F-NS-10. Our study highlights the importance
of unraveling the electronic effect imposed by surface pre-
adsorbed species (e.g., surfactant) as it can significantly affect
the physiochemical properties of active sites on catalysts’
surface and hence the observed activity. The successful
implementation of this will not only resolve the disagreement
found among literatures but also provide a solid guideline for the
design (or seeking) of catalysts with high activity.
Acknowledgements
Scheme 2. Proposed mechanism for p-NPP dephosphorylation over TiO2(001)
w/o surface F.
We thank National Natural Science Foundation of China
(21902138, 21802164, 22032005 and U1832148), the Hong
Kong Research Grants Council (CityU 21301719 and CityU
11300020) for funding support. The authors acknowledge Dr. Jin
Shang for the BET measurement.
Since the first report by Yang et al. in 2008[18], anatase
TiO2 with high coverage of F-capped (001) surface has been
adopted in many literatures for a wide range of catalytic
reactions[15,18-20]
.
The surface pre-adsorbed fluorine (i.e.,
surfactant) is often removed by either calcination or NaOH wash
before catalytic testing. Many literatures (even for those
published in high-impact journals) simply concluded no change
in the electronic state of surface Ti atoms after F removal based
on XPS results[15,18-20]. However, for commercial XPS equipped
with Al as X-ray source (1486.6 eV), Ti2p photoelectrons with
energy about 1000 eV can penetrate few nanometers from TiO2
surface and hence provide an averaged oxidation state of Ti
cations. Since the thickness of F-NS-1 sample is 4.89 nm,
commercial XPS is undoubted a bulk technique for not only TiO2
samples here (Figure S2) but also many nanocatalysts used in
literatures. Given that heterogeneous catalysis mainly involves
active sites on the topmost surface of the catalyst, those
structure-activity correlations established in literatures based on
XPS are thus questionable. Some researchers may realize the
importance of obtaining the electronic state of active sites as this
property can significantly affect the adsorption/activation of
Keywords: Electronic state manipulation • Dephosphorylation •
Titanium dioxide • Surface characterization • Nuclear magnetic
resonance
[1]
[2]
N. N. Greenwood, A. Earnshaw, Chemistry of the Elements. 2012.
J. Ali, D. C. Sharma, A. Bano, A. Gupta, S. Sharma, P. Bajpai, N.
Pathak, Enzymes in Food Biotechnology 2019, 503-519.
D. Cordell, S. White, Sustainability 2011, 3, 2027-2049.
D. Cordell, J.-O. Drangert, S. White, Glob. Environ. Change 2009, 19,
292-305.
[3]
[4]
[5]
[6]
F. Tan, Y. Zhang, J. Wang, J. Wei, Y. Cai, X. Qian, J. Mass Spectrom
2008, 43, 628-32.
M. H. Kuchma, C. B. Komanski, J. Colon, A. Teblum, A. E. Masunov, B.
Alvarado, S. Babu, S. Seal, J. Summy, C. H. Baker, Nanomedicine
2010, 6, 738-44.
[7]
A. A. Vernekar, T. Das, G. Mugesh, Angew. Chem. Int. Ed. 2016, 55,
1412-1416.
[8]
[9]
M. J. Manto, P. Xie, C. Wang, ACS Catal. 2017, 7, 1931-1938.
G. C. Shearer, J. G. Vitillo, S. Bordiga, S. Svelle, U. Olsbye, K. P.
Lillerud, Chem. Mater. 2016, 28, 7190-7193.
6
This article is protected by copyright. All rights reserved.

10.1002/anie.202104397
Angewandte Chemie International Edition
RESEARCH ARTICLE
[10] K. Khulbe, P. Roy, A. Radhakrishnan, G. Mugesh, ChemCatChem
2018, 10, 4826-4831.
[11] K. Khulbe, G. Mugesh, Polyhedron 2019, 172, 198-204.
[12] J. Yang, K. Li, C. Li, J. Gu, Angew. Chem. Int. Ed. 2020, 59, 22952–
22956.
[13] Z. Tan, T.-S. Wu, Y.-L. Soo, Y.-K. Peng, Appl. Catal B. 2020, 264,
118508.
[14] U. G. Survey, MINERAL COMMODITY SUMMARIES 2020. 2020, 137.
[15] G. Liu, H. G. Yang, J. Pan, Y. Q. Yang, G. Q. Lu, H. M. Cheng, Chem.
Rev. 2014, 114, 9559-9612.
[16] Y.-K. Peng, S. C. E. Tsang, Nano Today 2018, 18, 15-34.
[17] M. A. Boles, D. Ling, T. Hyeon, D. V. Talapin, Nat. Mater. 2016, 15,
141-153.
[18] H. G. Yang, C. H. Sun, S. Z. Qiao, J. Zou, G. Liu, S. C. Smith, H. M.
Cheng, G. Q. Lu, Nature 2008, 453, 638-41.
[19] S. Liu, J. Yu, M. Jaroniec, Chem. Mater. 2011, 23, 4085-4093.
[20] W. Q. Fang, X.-Q. Gong, H. G. Yang, J. Phys. Chem. Lett. 2011, 2,
725-734.
[21] H. G. Yang, G. Liu, S. Z. Qiao, C. H. Sun, Y. G. Jin, S. C. Smith, J. Zou,
H. M. Cheng, G. Q. Lu, J Am Chem Soc 2009, 131, 4078–4083.
[22] C. Z. Wen, J. Z. Zhou, H. B. Jiang, Q. H. Hu, S. Z. Qiao, H. G. Yang,
Chem. Commun. 2011, 47, 4400-4402.
[23] M. Li, Y. Chen, W. Li, X. Li, H. Tian, X. Wei, Z. Ren, G. Han, Small
2017, 13, 1604115.
[24] L. Ruan, X. Wang, T. Wang, Z. Ren, Y. Chen, R. Zhao, D. Zhou, G. Fu,
S. Li, L. Gao, Y. Lu, Z. Wang, H. Tian, X. Kong, G. Han, ACS Appl.
Mater. Interfaces 2019, 11, 37256-37262.
[25] U. Diebold, Surf. Sci. Rep. 2003, 48, 53–229.
[26] A. S. Barnard, L. A. Curtiss, Nano Lett. 2005, 5, 1261-1266
[27] X. Han, Q. Kuang, M. Jin, Z. Xie, L. Zheng, J. Am. Chem. Soc. 2009,
131, 3152-3153.
[28] Y. K. Peng, Y. Hu, H. L. Chou, Y. Fu, I. F. Teixeira, L. Zhang, H. He, S.
C. E. Tsang, Nat. Commun. 2017, 8, 675.
[29] A. Zheng, S. B. Liu, F. Deng, Chem Rev 2017, 117, 12475-12531.
[30] Y. K. Peng, L. Ye, J. Qu, L. Zhang, Y. Fu, I. F. Teixeira, I. J. McPherson,
H. He, S. C. E. Tsang, J. Am. Chem. Soc. 2016, 138, 2225-34.
[31] Y. K. Peng, H. L. Chou, S. C. E. Tsang, Chem. Sci. 2018, 9, 2493-2500.
[32] Y. K. Peng, B. Keeling, Y. Li, J. Zheng, T. Chen, H. L. Chou, T. J.
Puchtler, R. A. Taylor, S. C. E. Tsang, Chem. Commun. 2019, 55,
4415-4418.
[33] Z. Tan, G. Li, H.-L. Chou, Y. Li, X. Yi, A. H. Mahadi, A. Zheng, S. C. E.
Tsang, Y.-K. Peng, ACS Catal. 2020, 10, 4003-4011.
[34] X. Yi, H. H. Ko, F. Deng, S. B. Liu, A. Zheng, Nat Protoc 2020, 15,
3527- 3555.
[35] A. Kiani, L. S. Cahill, E. A. Abou Neel, J. V. Hanna, M. E. Smith, J. C.
Knowles, Mater. Chem. Phys. 2010, 120, 68-74.
[36] A. Kiani, J. V. Hanna, S. P. King, G. J. Rees, M. E. Smith, N. Roohpour,
V. Salih, J. C. Knowles, Acta Biomater. 2012, 8, 333-340.
7
This article is protected by copyright. All rights reserved.

10.1002/anie.202104397
Angewandte Chemie International Edition
RESEARCH ARTICLE
Entry for the Table of Contents
We demonstrate that TiO2 with F-modified (001) surface can activate p-NPP dephosphorylation for the first time at nearly room
temperature. The electronic withdrawing effect of fluorine imposed on TiO2(001) surface strongly manipulates the electronic state of
surrounding Ti5C atoms by making them very acidic, facilitating not only the bidentate adsorption of p-NPP but also its subsequent
activation.
8
This article is protected by copyright. All rights reserved.