Article

pubs.acs.org/JPCC

Impregnation versus Bulk Synthesis: How the Synthetic Route

Aects the Photocatalytic Eciency of Nb/Ta:N Codoped TiO2
Nanomaterials
Luca Rimoldi, Claudia Ambrosi, Giovanni Di Liberto, Leonardo Lo Presti,*,,, Michele Ceotto,,
Cesare Oliva, Daniela Meroni,, Serena Cappelli, Giuseppe Cappelletti,, Guido Soliveri,,
and Silvia Ardizzone,

Universita degli Studi di Milano, Department of Chemistry, Via Golgi 19, 20133 Milano, Italy

CNR-ISTM, Via Golgi 19, 20133 Milano, Italy

Center for Materials Crystallography, Aarhus University, Langelandsgade 140, DK-8000, Aarhus, Denmark

Consorzio INSTM, Via Giusti 9, 50121 Firenze, Italy

*
S Supporting Information

ABSTRACT: Selecting the proper doping strategy is essential

to controlling the photocatalytic activity of TiO2-based nano-
materials. In this work, we compare impregnation and bulk
synthesis methods for feeding titania nanocrystals with either Nb
or Ta as transition metal dopant and N as nonmetal codopant.
The resulting photocatalytic eciency was tested toward ethanol
degradation under either UV or simulated solar irradiation.
Microstructure, morphology, and electronic properties at various
length scales were deeply investigated and compared with DFT
simulations. Instead, under UV irradiation, impregnated samples
performed better than bulk synthesis ones, with Ta-doped
powders being more ecient than Nb-doped and undoped
TiO2. Under simulated solar irradiation, bulk synthesis Nb-
doped materials were the most active ones, while all the impregnated samples were even less performing than the undoped TiO2
reference. On the basis of XPS, EPR, DRS, and XRPD results, such dierences were attributed to the bulk synthesis approach
producing a more homogeneous distribution of guest dopants within the grains, in conjunction with a higher amount of intrinsic
defects (such as O vacancies). Implications of these ndings on the engineering of ecient titania photocatalysts are discussed.

1. INTRODUCTION studies have proposed Niobium and Tantalum as ecient

Titanium dioxide is one of the most studied photocatalysts in
environmental remediation, owing to its stability, low cost, and
high activity.1,2 However, its wide band gap (>3.0 eV)
corresponds to light absorption in the UV region and prevents
this material from being employed in several practical
applications, including indoor usages, for which activity under
solar light would be preferable. In recent years, considerable
eort has been devoted to promoting the photocatalytic activity
of TiO2 under solar radiation by visible light sensitization.
Anion doping with p-block elements, such as nitrogen,3 has
been widely studied in this context.4,5 Nitrogen doping creates
midgap states that strongly enhance visible light absorption.68
However, dopants and doping-induced oxygen vacancies can
also act as recombination centers and eventually reduce the
lifetime of electronhole (e-h+) pairs, resulting in a neat
decrease of quantum yields.9
TiO2 modication with transition metals (TMs) has also
been proposed to produce defects benecial for visible light
harvesting. In recent years, theoretical10 and experimental1114
dopants, thanks to the favorable location of the electronic states
created by their 4d and 5d orbitals. For instance, Wang et al.12
prepared Ta-doped macroporous titania photocatalysts with
superior performance in the degradation of 4-nitrophenol at
high Ta/Ti molar ratios.
Very recently, codoping by metal and nonmetal species has
attracted much attention as a possible solution to optimize light
absorption while preserving the quantum yield of titania. The
simultaneous presence of electron donors (e.g., a pentavalent
metal like Nb or Ta) and acceptors (e.g., N) might lead to
charge compensation phenomena,15 possibly benecial to the
photocatalytic activity.1618 The theoretical study by Zhu et
al.19 underlined the positive eects of compensated codoping in
Nb,N- and Ta,N-copromoted titania systems, such as an
increased amount of nitrogen species entering the TiO2 lattice.
Received: July 15, 2015
Revised: September 9, 2015
Published: October 8, 2015

2015 American Chemical Society 24104 DOI: 10.1021/acs.jpcc.5b06827

J. Phys. Chem. C 2015, 119, 2410424115
The Journal of Physical Chemistry C
However, Ta- and Ta,N-codoped TiO2 was investigated to a
lesser extent than Nb- and Nb,N-codoped20 titania. A very
recent theoretical work by Liu and co-workers21 showed how
Ta ion is promoted as a lattice guest thanks to the presence of a
nonmetal species (sulfur). Obata et al.22 rst synthesized Ta,N-
codoped lms by magnetron sputtering, which showed
enhanced photocatalytic activity under visible light. Further-
more, Wang et al.23 demonstrated that even tridoped TaNF
systems could be promising systems for visible-light applica-
tions. Based on EPR and XPS outcomes, they also found charge
compensation eects and set an optimal Ta/Ti molar ratio as
low as 1%. A further test of the photocatalytic activity was
provided by Gong et al.24 for Ta,B-codoped TiO2, showing the
benecial eects of codoping with respect to single-doping in
dye degradation tests. Consistently, other authors observed a
better performance of copromoted samples with respect to
single-doped ones for photoelectrochemical water oxidation.25
Given their proximity in the periodic table, a comparison
between Nb- and Ta-containing TiO2 nanomaterials is crucial
to clarify what structureproperty relationships eectively
contribute to determine a given photocatalytic performance.
To the best of our knowledge, no previous experimental
comparison between Nb,N- and Ta,N-codoped TiO2 photo-
catalysts have been reported to date. The few works that
compare Nb,N- and Ta,N-codoped titania are either theoreti-
cal26,27 or involve single-doping strategies for electrochemical
and DSSC applications.2830
Our approach not only combines experimental character-
izations and Density Functional Theory (DFT) simulations,
but also investigates the role played by the synthetic route on
the nal photocatalytic performance. This issue is of paramount
importance, although it is seldom discussed in the literature.
The synthetic procedure controls phase composition, crystallo-
graphic lattice distortions, nature and concentration of bulk
defects, electronic band structure, and surface properties, whose
interplay determines, a fortiori, crucial macroscopic properties
such as the photocatalytic performance. Among other synthetic
approaches, we will focus on bulk doping by classical solgel
(sy) routes and postsynthesis impregnation (im), since
these two techniques are among the most frequently employed
in modern catalytic applications.31 While bulk doping generally
leads to more homogeneous systems, impregnation is known to
be a very eective method for the production of high
performing catalysts.32
In this work, we compare the photocatalytic performance of
TiO2 single-doped and codoped nanomaterials as a function of
the doping method (either sy or im). Niobium and
tantalum were considered as TM dopants, while nitrogen was
selected as a possible nonmetal codopant. The photocatalytic
eciency was tested toward the degradation of ethanol, a
pollutant model system, under UV and simulated solar light
irradiation in the gas phase. The photodegradation ability of the
various materials was related to their structural, morphological,
and electronic properties at various length scales, as
investigated by XRPD, BET, XPS, DRS, and EPR character-
izations. Experimental results were also complemented by bulk
plane-wave DFT simulations. A tentative structural/electronic
model was then proposed to account for the observed
dierences in the photocatalytic performance between the
im and sy sample series. Eventually, implications on the
suitability of the im and sy synthetic routes on the
engineering of more ecient titania photocatalysts are
discussed.
24105
Article
2. EXPERIMENTAL AND THEORETICAL METHODS
2.1. Samples Preparation. All the reactants were
purchased from Sigma-Aldrich. No further purication
procedures were applied. Doubly distilled water passed through
a Milli-Q apparatus was used to prepare solutions and
suspensions. Two series of titania samples were synthesized,
diering in the doping procedure: either addition during the
solgel synthesis (sy) or impregnation (im). Each series
consisted of ve samples: N-doped (TN), Nb-doped (TNb),
Ta-doped (TTa), Nb,N-codoped (TNbN), and Ta,N-codoped
(TTaN) TiO2. The nominal dopant/Ti molar ratio was xed at
1%. Solutions of NH4OH, NbCl5, and TaCl5 were employed as
dopant precursors. For the TNbN specimens, the two dopants
were provided at the same time by NH4[NbO(C2O4)2(H2O)]
nH2O, whereas NH4Cl and TaCl5 were employed as precursors
for the TTaN samples. (sy) or (im) symbols are added to
the various sample labels to specify whether each specic
material was respectively obtained either from the sy or the
im procedure. An undoped titania sample (T) was also
synthesized as a reference for both series.
The same solgel procedure was employed throughout as
the basic synthetic route. More in detail, 0.063 mol of
titanium(IV) isopropoxide was mixed with 24 mL of 2-
propanol in order to obtain a molar ratio of 1:5. The solution
was stirred for 10 min at 300 rpm. Then, 113 mL of a HCl
aqueous solution at pH 3 was added dropwise while stirring
vigorously to obtain a nal Ti/H2O molar ratio of 1:100. The
mixture was stirred continuously for 90 min to complete the
hydrolysis. Then, the solvent was removed drying in oven at
80 C.
The rst series (sy) of samples was prepared by adding the
dopant precursors directly during the solgel synthesis in the
2-propanol solution. A NH4OH solution was mixed to the
aqueous solution for the preparation of the N-doped (TN(sy))
sample. The powders were then dried in an oven at 80 C and
calcined at 400 C for 6 h in O2 ux (9 NL h1).
The second series (im) of photocatalysts was prepared by
suspending 1.5 g of the uncalcined powder in 15 mL of an
ethanol/water (2:1) solution (or 15 mL of an aqueous solution
for the TN(im) sample) containing the dopant precursors.
After 90 min, the material was dried in an oven at 90 C and
calcined at 400 C for 6 h in O2 ux (9 NL h1).
2.2. Materials Characterization. 2.2.1. X-ray Powder
Diraction. All samples were characterized by in-home X-ray
powder diraction (XRPD) experiments using a Philips PW
3710 BraggBrentano goniometer equipped with a scintillation
counter and 1 divergence slit, 0.2 mm receiving slit, and 0.04
Soller slit systems. We employed graphite-monochromated Cu
K radiation at 40 kV 40 mA nominal X-ray power. :2
scans were performed between 20 and 90. The interested
reader can nd a detailed description on the XRPD
experiments within the SI (section 1).
2.2.2. Specic Surface Area and Porosity Determinations.
Specic surface areas were determined by the Brunauer
EmmettTeller (BET) method using a Coulter SA3100
apparatus. Pore size distributions were determined from the
desorption isotherms by the BarrettJoynerHalenda (BJH)
method.
2.2.3. Spectroscopic Characterizations. Electron para-
magnetic resonance (EPR) spectra were measured at room
temperature and at 77 K using an X-band EPR spectrometer
(Bruker Elexsys) at the working frequency of 9.4 GHz. The
DOI: 10.1021/acs.jpcc.5b06827
J. Phys. Chem. C 2015, 119, 2410424115
The Journal of Physical Chemistry C
Figure 1. Collected XRPD patterns (blue dots). The least-squares tting curves are in red, the yobs ycalc point-by-point dierences in green and the
rened background function in yellow. Intensities are drawn on a common arbitrary scale. The diractions for the anatase and brookite phases are
shown as red and blue markers, respectively: (a) TTaN(sy); (b) TTaN(im).
Bruker SimFonia program was used to perform the spectral
simulations.
Diuse reectance spectroscopy (DRS) measurements were
carried out on a Shimadzu UV-2600 UVvis spectropho-
tometer equipped with an integrating sphere, collecting data
from 250 to 700 nm. Band gap, Eg, determinations were
performed applying the KubelkaMunk equation.
X-ray photoelectron spectroscopy (XPS) spectra were
obtained using an M-probe apparatus (Surface Science
Intruments). The source was monochromatic Al K radiation
(1.486 keV). A spot size of 200 m 750 m and pass energy
of 25 eV were used. The 1s level hydrocarbon-contaminant
carbon was taken as the internal reference at 284.6 eV. The
atomic ratios reported in Table 3 are obtained by elaboration of
survey spectra.
2.2.4. Photocatalytic Activity Tests. All samples were tested
under both UV (iron halogenide lamp Jelosil HG500, eective
power density 17 mW cm2 between 280 and 400 nm,
measured using a Thorlabs S314C radiometer) and simulated
solar irradiation (halogen lamp by Lot Oriel, eective power
density 1 mW cm2 in the range 280400 nm and 14 mW
cm2 in the range 400800 nm) in the gas phase photocatalytic
degradation of ethanol, as a pollutant model for volatile organic
compounds (VOCs). Photocatalytic tests were carried out in a
previously described apparatus.33 A total of 50 mg TiO2
powder, drop casted on a 64 cm2 Petri dish, was employed in
conjunction with a starting ethanol concentration in the range
198238 ppm. The disappearance of the pollutant molecule,
the formation of the main intermediate (acetaldehyde), and its
complete mineralization to CO2 were monitored during
photocatalytic tests by means of a gas-chromatographic system
(Agilent 7890 equipped with DB-VAX and Porapak columns,
two FID detectors and a methanator).
2.3. DFT Modeling. To simulate doped TiO2 anatase, we
performed DFT+U calculations34 using a VASP suite of
codes.3540 The projector augmented wave (PAW) pseudopo-
tentials were employed,36,38 considering the spin-polarization.
The Generalized Gradient Approximation (GGA) was used41
with the PerdewBurkeErnzerhof parametrization.39 The
supercell doping arrangements and other DFT parameters are
described in the SI.
3. RESULTS
3.1. Microstructure. Figure 1 shows typical recorded X-ray
diractograms, together with the corresponding least-squares
ttings, for the codoped samples TTaN(im) andTTaN(sy).
The interested reader can nd full information on the whole set
of collected diractograms, the nal least-squares agreement
factors and the rened crystallographic and microstructural
24106
Article
parameters in the SI (Figure S1, Tables S1S4). Details
concerning the outcome of the tting procedure can be also
found in section 1, SI.
As expected on the basis of the core synthetic proce-
dure8,15,42,43 (see also section 2.1), all the materials here
investigated are biphasic, with anatase TiO2 (space group: I41/
amd) being always the major polymorph in the mixture and
brookite (space group: Pbca) the minority one. When the two
im and sy sample series are compared (Figure 2), their
Figure 2. Brookite content for the im (circles, dotted line) and sy
samples (triangles, full line), as estimated from the Rietveld renement
against the XRPD data (see section 1, SI). The lines serve just as
guides for the eye. The size of the symbols are roughly equal to the
estimated standard deviation (e.s.d.).
trends as a function of the dopant are qualitatively similar.
Nevertheless, im materials invariably bear a higher brookite
content, while the sy ones are richer in anatase and, in
general, more similar to the undoped T reference.
On average, the brookite content turns out to be as high as
43(2)% in impregnated specimens and 36(2)% in the solgel
ones. Interestingly, Nb,N-codoping seems to somewhat
(weakly) favor the anatase phase with respect to single-doped
powders,15 while the contemporary presence of Ta and N in
the lattice has just the opposite eect. This holds true no matter
the dopant feeding method, even though changes in the
brookite content are more evident in the sy nanostructured
powders.
Also, for what concerns the crystallite dimensions of the
anatase phase (Figure 3, left axis), the im samples are
associated with a bit larger Dv values (34(2) ) than the sy
ones (31(1) ). However, there are neither striking nor
systematic deviations with respect to the reference T material
and dierences in Dv between the two series are scarcely
signicant from a statistical viewpoint. This is even more true
when the WilliamsonHall microstrain is taken into account
(Figure 3, right axis). Rather, sy and im samples always bear
very similar values and just a tendency toward higher lattice
DOI: 10.1021/acs.jpcc.5b06827
J. Phys. Chem. C 2015, 119, 2410424115
The Journal of Physical Chemistry C
Figure 3. Average volume-weighted crystallite dimensions of the
anatase phase (left axis, full black markers) and lattice microstrain
values (right axis, open red markers), as retrieved by the Williamson
Hall method from integral breadth of h0l reections up to 2 = 55:
Circles, dotted lines: im samples; Triangles, full lines: sy samples;
Error bars correspond to 1 e.s.d.
strain in codoped samples, no matter the doping strategy, could
be inferred from Figure 3. In turn, this likely implies that both
Nb- and Ta-based doping somewhat increases the average
concentration of defects in the nanomaterials here studied. It
should be also noted that Ta-containing materials invariably
bear higher WilliamsonHall microstrain parameters than the
Nb-containing ones. This could be likely related to the longer
ionic radius of tantalum (vide infra), which implies greater
lattice distortions.
In general, microstructural results for Nb- and Nb,N-
(co)doped anatase are similar to previous investigations on
similar materials,15 as the presence of both dopants discourages
the brookite formation and leads to smaller anatase crystallites.
Probably, such a behavior can be traced back to the strong
thermodynamic driving force toward the I41/amd phase, which
results in smaller critical radii of the crystal nuclei and speeds
up nucleation over the growth process.44 However, this is no
longer true when Tantalum is considered: Ta,N-codoping is
associated with larger anatase crystallites and invariably imply a
+5% increment in the brookite content (Figures 2 and 3).
Assuming the validity of the simple nucleation model above
sketched, such a dierence might be due to a lower ability of
the tetragonal lattice in hosting the larger Ta5+ ions (for the
sake of comparison, ionic radii of Ti4+, Nb5+, and Ta5+ in 6-fold
coordination45 are 0.605, 0.640, and 0.780 ).
More quantitative information on the amount of distortion
to which the crystallographic long-range structure is subject
upon doping can be obtained by following the changes of the
anatase cell edges as a function of the dopant (Figure 4). The
corresponding graphical entries for the brookite phase, together
with individual cell edge length values, can be found in the SI
(Figure S2 and Table S2).
Figure 4. Relative changes of the |c| (a) and |a|,|b| (b) cell axis modules throughout the im (circles, dotted line) and sy (triangles, full line) sample
series with respect to the reference material T as a function of the dopant nature. Error bars correspond to 1 e.s.d.
24107
Article
At the quite low nominal dopant concentration here
employed (vide supra), the relative changes of the crystallo-
graphic anatase cell edges are generally immaterial in terms of
their estimated standard deviations. More in details, both the c
and the symmetry-equivalent a,b cell edges have a length equal
to those shown by the reference sample T within 0.1%. The
two codoped impregnated TNbN(im) and TTaN(im) samples
are the only patent outliers, as they show a signicant
lengthening (up to +0.3%) of the |c| vector module with
respect to the reference compound.
In conclusion, the following take-home messages can be
derived by our microstructural XRD analysis. First, the dopant
feeding strategy mainly aects the phase composition, with im
materials showing a higher brookite content than the sy ones.
Second, other average microstructural parameters, such as
crystallite dimensions and microstrain, are very similar between
the two sample series, with the im one possibly showing a
small tendency toward larger Dv. Moreover, Ta-containing
materials show higher microstrain, , than Nb-containing ones.
Third, the crystallographic lattice parameters are scarcely
aected by the dopant nature and the synthetic route, with
the only exception of impregnated codoped materials, which
undergo a +0.2, 0.3% increment of the cell along the unique
axis. In general, structural distortions and compositional
dierences are enhanced by both Nb,N- and Ta,N-codoping,
no matter how the dopants are added to the material, a fact that
mirrors metal/nonmetal synergic eects already detected in
previous works on similar systems.15
3.2. Morphological Features. The adsorptiondesorption
isotherms of N2 in subcritical conditions of single- and codoped
samples are reported in Figure S3, SI. All the samples show
hysteresis loops characteristic of a mesoporosity. The two
doping approaches have markedly dierent eects on the
morphological features of the resulting materials (Figure S3a,b,
SI).
As for sy materials, both doped and codoped samples
generally show lower surface areas, together with a relevant
decrease of the total pore volume (Table 1 and Figure S4, SI),
with respect to the reference T. Furthermore, the presence of
either Nb or Ta species induces a shift of the hysteresis loop to
lower pressures (Figure S3a, SI), indicating a decrease in the
average pore size. The shape of the hysteresis loop is also
modied upon metal doping, becoming more typical of H2-
type (bottleneck shape). On the contrary, N-doping has an
almost negligible eect on the morphological features of TiO2,
in agreement with previous reports.15
The presence of both dopants has a dierent eect with
respect to both single dopant. N,Nb-codoping results in an
appreciable increase in surface area while the total pore volume
DOI: 10.1021/acs.jpcc.5b06827
J. Phys. Chem. C 2015, 119, 2410424115
The Journal of Physical Chemistry C Article

Table 1. Specic Surface Area (SBET), Total Pore Volume

(Vpores), and Average Grain Diameter Calculated from SBET
Values (DBET) for the Doped and Undoped Samples of the
Two Series
sample SBET (m2 g1) Vpores (mL g1) DBET (nm)
T 149 0.277 10.2
Bulk Synthesis
TN(sy) 132 0.291 11.6
TNb(sy) 119a 0.166a 12.8
TNbN(sy) 161a 0.223a 9.5
TTa(sy) 135 0.187 11.3
TTaN(sy) 111 0.169 13.8
Impregnation
TN(im) 165 0.390 9.3
Figure 5. Total spin-up (black continuous line) and spin-down
TNb(im) 133 0.328 11.5
(dashed red lines) DOSs. In green, PDOS for the Ta 5d or Nb 4d
TNbN(im) 150 0.268 10.2 states.
TTa(im) 124 0.281 12.3
TTaN(im) 123 0.325 12.4
a DOSs. In the case of substitutional doping (Figure 5a), the
See ref 15. Fermi energy of the system lies at the bottom of the conduction
band and the 5d states of the dopant are located deep into the
remains unchanged, a fact that might be related to the smaller conduction band. This opens the possibility for enhanced UV
crystallites (Table S3, columns 23) and smaller pores (Figure absorption. Instead, when Ta is located interstitially (Figure
S4a, SI). N,Ta-codoping gives rise instead to a decrease in the 5b), midgap states are injected into the band gap roughly at the
surface area, which may be traced back to the larger average bottom of the conduction band, as shown by the Ta 5d
pore size and bigger crystallites. projected DOS. Considering that the Fermi energy is in both
With respect to bulk doping, the impregnation method leads cases just below the conduction band, we can deduce that Ta
to higher total pore volumes and to larger average pore sizes acts in any case as an electron-donor. Taking into account that
(Figure S4b, SI), although no clear trends can be appreciated titania is highly defective,4749 we simulated oxygen-defective
by comparing the specic surface areas of the samples obtained and Ta-doped supercells. Clearly, the vacancies act as electron
by the two doping approaches. Furthermore, impregnation donors and the Fermi energy is pinned at the bottom of the
does not give rise to appreciable eects on the shape of the conduction band (Figure 5c,d). A comparison between these
hysteresis loop of the adsorption/desorption isotherms (Figure four panels shows that the Ta orbital location is left unchanged
S3b, SI). The im samples doped with Nb and Ta have lower upon Oxygen vacancy generation.
surface areas with respect to the undoped reference, but have a To better rationalize the experimental results, we report on
higher total pore volume and larger pores, as appreciable from the two lower panels (Figure 5e,f) the DOSs relative to Oxygen
the shift of the hysteresis loop to higher pressures (Figure S3b, defective Nb substitutional and interstitial doping. In the case
SI). Finally, the addition of nitrogen, both in single- and of Nb dopant, Nb 4d orbitals act as electron donors as Ta 5d
codoped samples, seems to increase the surface area. ones, but Nb states are just below the conduction band and
Figure S5, SI, reports the trend between the experimental potentially available for visible light excitation.
surface area and the average crystallite size (Dv) obtained by We have reported for the sake of documentation both
elaboration of X-ray powder diractograms. As expected, a substitutional and interstitial TMs doping. However, in the
roughly inverse correlation between SBET and Dv can be present case, only substitutional doping is considered
envisaged. The only exception is the sample TN(im), which plausible.15 DFT+U calculations show that both Nb15,47 and
presents a large surface area together with the largest crystallite Ta are preferentially accommodated in substitutional Ti sites.
size. However, this specimen also bears the largest total pore The comparison of the crystallographic strains obtained by the
volume (0.390 mL g1), which determines a signicant XRD renements and the DFT+U simulations (Figure S6, SI)
increment of the corresponding surface area. excludes the possibility for an interstitial TMs doping site.
According to the equation46 SBET = (6 104)/( DBET), Finally, we present the codoping scenarios to complete the
the BET surface area can be written as the total surface area of description of the experimental samples. To avoid rare
the powder population divided by the powder mass, where is congurations, we placed the TM doping site away from the
the density of the sample (3.9 g cm3, as calculated by XRPD nitrogen one.15 We have considered all possible substitutional
results for the anatase phase). The factor 6 applies for spherical and interstitial codoping combinations, even though, as stated
as well as for cubic particles. DBET (Table 1, fourth column) above, the TM interstitial location is quite improbable. Figure 6
could be therefore taken as a rough estimate of the average shows an electron transfer from the 5d Ta electronic states to
particle size in the assumption that both porosity and surface the partially empty N orbitals located above the valence band,
roughness are negligible. It can be observed that for all samples as it was reported also for the 4d Nb states (Figure 5 of ref 15).
the DBET value is larger than the crystallite data obtained from This electron transfer closes the electronic shell and shifts the
X-ray elaboration (Figure 3), possibly indicating the occurrence Fermi energy toward midgap values. The occurrence of this
of aggregation phenomena among the crystallites. phenomenon is conrmed by EPR experiments (see section
3.3. Electronic Structure. Figure 5 reports the electronic 3.6). Clearly, this charge compensation eect generates extra
density of states (DOSs) of TM single-doped TiO2. Both stability for the doping process and possibly enhances the
substitutional and interstitial Ta-doping lead to quite dierent photocatalytic performances (vide infra).
24108 DOI: 10.1021/acs.jpcc.5b06827
J. Phys. Chem. C 2015, 119, 2410424115
The Journal of Physical Chemistry C
Figure 6. Total spin-up (continuous black line) and spin-down
(dashed red line) DOSs. Green lines for transition metal PDOS and
pink for N PDOS.
3.4. Visible Absorption Properties. DRS results of single-
doped and codoped samples provide direct insight into the
mutual role of the guest species in modifying the photo-
absorption properties of TiO2.
The undoped reference T has the characteristic absorption
behavior of anatase TiO2, showing a band gap value of 3.2 eV.
Bulk doping with either N or metal species (Figure 7a) gives
Figure 7. Diuse reectance spectra (DRS) of (a) bulk doped samples
and (b) impregnated samples. [a] See ref 15.
rise to both an absorption in the UV region and to a visible
sensitization of TiO2. N-doping causes a visible light absorption
at 450500 nm, in agreement with previous reports about TiO2
doped with ammonia salts.7,15,18 Such a localized visible light
absorption can be traced back, on the grounds of DFT
calculations, to the formation of shallow intragap states located
above the valence band of TiO2.42,50 Bulk doping with either
Nb or Ta leads instead to a red shift of the absorption edge,
which is comparable for both samples and results in a reduction
of the observed band gap of about 0.1 eV (Table S5).51 Similar
red shift eects have previously been reported for Nb-doped
24109
Article
TiO215,18,52,53 and can be attributed to the formation of donor
levels located close to the conduction band of TiO215,19 (see
Figure 5e). With respect to Nb-doping, the Ta-doped sample
shows also a broader absorption in the bluegreen region,
which can be attributed to the presence of the 5d state localized
into the conduction band (Figures 5a and 6a). Interestingly,
codoped samples display the characteristic features of the
absorption spectra of both N- and Nb/Ta-doped samples. As
reported previously,15 N,Nb-codoping enhances the visible
absorption at 450 nm with respect to nitrogen single-doped
samples. Such an eect is even more marked in the case of the
N,Ta-codoped sample. Theoretical calculations19 have shown
that transition metals (Nb, Ta) might facilitate the insertion of
p-type dopants (N and C) in the TiO2 lattice. Therefore, the
observed enhanced visible absorption at 450 nm might be
related to a higher amount of N entering the TiO2 lattice in
codoped samples.
The adopted doping strategy remarkably aects the optical
properties of the resulting materials. DRS experiments show
that impregnation with TMs (Nb, Ta) does not give rise to
appreciable eects on the absorption features of TiO2 (Figure
7b). In particular, no shifts of the absorption edge are observed
upon either Nb- or Ta-doping with respect to the undoped
reference. On the contrary, impregnated N atoms lead to a
visible light absorption around 450 nm, which is fully
comparable to what is observed for bulk-doped samples. It is
noteworthy that impregnated codoped samples do not shift the
absorption edge and the visible absorption localized around 450
nm has a lower intensity with respect to the N-doped sample.
3.5. Photocatalytic Activity. The photocatalytic activity of
the doped and codoped samples was tested toward the gas
phase degradation of ethanol as a model volatile organic
compound (VOC). Ethanol and its main degradation
intermediate, acetaldehyde, are indeed important atmospheric
and indoor pollutants.5457 During photocatalytic tests, the
concentrations of ethanol, acetaldehyde, and of the product of
complete mineralization (CO2) were monitored by a gas-
chromatographic system (Figure S7, SI). Besides acetaldehyde,
no other intermediate compounds were determined due to the
diculty in quantifying trace amounts. The carbon balance was
always complete at the end of the reaction. Photocatalytic tests
were carried out under both UV and simulated solar light
irradiation (lamps emission spectra are shown in Figure S8, SI).
The emission spectrum of the UV lamp (Figure S8a, SI)
contains small visible tails, but these are not expected to
signicantly contribute to the generation of e-h+ pairs with
respect to the main UV peaks. Actually, the latter are able to dig
electrons from deeper valence states, while the visible tail might
involve at most near-edge regions, where the density of states is
much lower. On the contrary, the solar lamp contains an UV
component analogously to the natural solar emission spectrum
(Figure S8b, SI). Photolysis tests under both irradiation sources
showed negligible ethanol degradation.
Ethanol degradation was completed for all samples within
120 min of UV irradiation. Acetaldehyde showed the typical
bell-shaped curve of a reaction intermediate (Figure S7, SI) and
a signicant degree of mineralization was obtained for all
samples. Figure 8 compares the CO2 formation for the bulk
doped (Figure 8a) and impregnated samples (Figure 8b) under
UV irradiation. Bulk doping with either Nb or Ta has a clear
detrimental eect on the TiO2 photocatalytic behavior (Figure
8a). Conversely, nitrogen bulk doping gives rise to a very
limited enhancement of the photocatalytic activity with respect
DOI: 10.1021/acs.jpcc.5b06827
J. Phys. Chem. C 2015, 119, 2410424115
The Journal of Physical Chemistry C Article

formation percentage, although mineralization trends are, as

expected, more evident.
The situation is quite dierent for simulated solar tests. For
all samples, ethanol degradation was not completed under
simulated solar irradiation during the reaction time (Figure 9)

Figure 8. Mineralization curves of (a) bulk doped samples and (b)

impregnated samples, under UV light irradiation.

to undoped TiO2. Interestingly, codoping with nitrogen

signicantly improves the performance of Nb- and Ta-doped
materials, although they still remain less active than the Figure 9. Ethanol disappearance curves of (a) bulk doped samples and
(b) impregnated samples under simulated solar light irradiation.
nitrogen single-doped sample. Impregnated samples largely
outperform their bulk doped counterparts under UV light. All
impregnated samples are indeed more active than the undoped and mineralization percentages were extremely low (Table 2a,b,
reference and give rise to complete mineralization of the fth column). Acetaldehyde was the main reaction product for
pollutant in less than 120 min (Figure 8b), with the only all the tested samples. Figure 9 reports the ethanol
exception of TNb(im) whose photocatalytic activity is disappearance curves for bulk doped (Figure 9a) and
comparable to that of the undoped reference. The most active impregnated samples (Figure 9b) under simulated solar
sample is TTa(im), followed by TN(im), TTaN(im), TNbN- irradiation. As for the UV light tests, nitrogen bulk doped
(im), and TNb(im). Table 2b (columns 23) reports the initial TiO2 has a behavior similar to the undoped reference, while Ta-
rates of ethanol degradation under UV light determined using doping plays a detrimental role on the TiO2 photocatalytic
the method of initial decay7 and the CO2 percentage after 60 activity (Figure 9a). On the contrary, Nb-doped and codoped
min reaction time. The initial rates of ethanol degradation show samples exhibit an enhanced photocatalytic performance with
the same trends in photocatalytic activity than the CO2 respect to the reference, with the codoped sample being the
most active. These results are in agreement with the electronic
Table 2. Kinetic Constants and Percentage of Mineralization structure calculations in Figures 5 and 6 and also with previous
in UV and Simulated Solar Tests for (a) Bulk Doped reports,17 showing improved photocatalytic activity of Nb,N-
Samples and (b) Impregnated Samples codoped samples under visible-light irradiation.
Strikingly, all the impregnated samples are less active than
UV solar the undoped reference (Figure 9b), with the only exception of
kpseudo first order UV % CO2 kpseudo first order solar % CO2 TNb(im) which still presents a comparable behavior with
sample (s1) (t = 60 min) (s1) (t = 210 min)
respect to the undoped TiO2. It is noteworthy that Ta-doped
(a)
and codoped samples are the least active under solar irradiation.
T 0.053 37.7 0.0021 0.6 To fully rationalize the above highlighted results is important
TN(sy) 0.057 47.6 0.0031 1.1 to investigate the amount of paramagnetic defects in our
TNb(sy) 0.039 14.8 0.0051 2.5 samples (section 3.6). A tentative model to understand the
TNbN(sy) 0.056 44.4 0.0057 2.6 behavior of the materials here investigated under photoactivity
TTa(sy) 0.052 23.2 0.0009 0.3 tests is given in section 4 below.
TTaN(sy) 0.052 36.5 0.0007 0.3 3.6. EPR Analyses. Several paramagnetic features, intro-
(b) duced by the dopants, can be appreciated in EPR spectra
T 0.053 37.7 0.0021 0.6 obtained at 77 K. For the attribution of the various EPR peaks,
TN(im) 0.060 80.6 0.0020 0.0 see section 6, SI.
TNb(im) 0.051 37.9 0.0032 0.9 TN(sy) and TN(im) have the same chemical nominal
TNbN(im) 0.063 54.6 0.0023 0.7 composition and indeed show the same EPR features.
TTa(im) 0.060 88.5 0.0018 0.5 However, the same lines have dierent intensities in the two
TTaN(im) 0.066 74.6 0.0009 0.2 EPR spectra (Figure S9, SI). Specically, we can compare the
24110 DOI: 10.1021/acs.jpcc.5b06827
J. Phys. Chem. C 2015, 119, 2410424115
The Journal of Physical Chemistry C
intensity ratio, R, between the most intense Nb (star-labeled)
line and that of B, which is the most intense line due to oxygen
radical species. This ratio is very dierent for the two spectra,
being R(TN(sy)) 0.44 and R(TN(im)) 2.1. Furthermore,
the E line attributed to Ti3+ is observed only in the case of
TN(sy), but not in the case of N(im). The two adopted doping
procedures apparently give rise to a dierent balance between
paramagnetic defects.
The role played by the metal (Nb, Ta) in modifying N-
induced defects can be now examined, starting from the
codoped samples obtained by impregnation.
Only oxygen radicals, as well as F+ and Ti3+ species (except
for the one producing the E line), are appreciable in the
TNbN(im) codoped sample (red in Figure 10a). Several
authors reported the occurrence of bulk Ti3+ species also in Nb-
doped TiO2,15,52,53 whose formation was attributed to a
mechanism of valence induction.17,19
Figure 10. Experimental EPR spectra collected at 77 K of (a)
TNbN(im) and (b) TTaN(im) (red) with respect to TN(im) (blue).
The disappearance of Nb paramagnetic species in the
TNbN(im) sample supports the occurrence of charge
compensation phenomena between Nb (acting as an electron
donor) and N (acting as an electron acceptor), as already
reported elsewhere.15,19
In the case, instead, of TTaN(im), the compensation
mechanism is not as complete as for sample TNbN(im),
since low intensity N b lines add to the F + and Ti 3+
paramagnetic species in the spectrum (Figure 10b). The
intensity ratio R pertaining to TTaN(im) is about 0.21, to be
compared with R 0 for N,Nb-codoped, impregnated sample.
In the case of the codoped samples obtained by bulk
synthesis (sy), the compensation eects are less signicant than
for the impregnated samples (Figure S10, SI). As for
TNbN(sy), the only relevant eect due to the presence of
24111
Article
the metal is the disappearance of the E line attributed to Ti3+
species, while no other signicant dierence can be appreciated.
The R value is around 0.47.
The picture is slightly dierent in the case of TTaN(sy)
(Figure S11, SI). The R(TTaN(sy)) 0.25 ratio is almost half
with respect to the value of TN(sy) and suggests that the
presence of Ta in the synthesis modies the balance between
Nb and of the oxygen-based radical species.
4. DISCUSSION
XPS results are aected by a large noise, possibly due to the low
amount of the provided dopants. However, as a general trend,
XPS results point out that im samples invariably bear an
apparent higher amount of dopants with respect to the sy
ones, as reported in Table 3. Figure S11, SI, shows an assay of
representative XPS spectra for the two guest metals.
Table 3. Apparent Concentrations of Transition Metal (TM)
Dopants within the im and sy Nanostructured Powders,
as Retrieved from the Area Ratios of the Nb/Ti and Ta/Ti
XPS Peaks and Expressed as TM/Ti Atomic Ratios (See
Section 2.2.3)
im sy
TNb 0.14 0.03
TTa 0.06 0.01
TNbN 0.05 0.01
TTaN 0.02 0.01
This was quite an unexpected result, as the reaction mixtures
were always fed with the same amount of precursor reagents
within the experimental error. The two sample series were
calcined under identical thermodynamic control conditions and
nonvolatile TM species should have been granted the same
chances of entering the nanoparticles. Besides, no evidence of
segregated Nb- and Ta-based phases were found by the XRD
analysis (see section 3.1 and Figure S1, SI). To explain the
detected dierences (Table 3), we hypothesize that im
samples are less homogeneous than the sy ones. Actually, it is
reasonable to assume that the distribution of point defects
diers depending on the adopted doping procedure. In the sy
case, guest ions are supplied when the TiO2 crystalline phase is
growing, and it is probable that they spread out quite uniformly
throughout most of the anatase lattice structure.15 On the
contrary, in im materials, the dopants are supplied after the
bulk phases are formed and prior to the calcination step. With
this procedure, the dopants must rst diuse through grain
pores and interstitial voids to enter the crystallites. This
diusion needs to occur when the amorphous-to-crystalline
transition is taking place. Consequently, a radial concentration
gradient is set up within large aggregates of crystallites (grains),
and the extrinsic impurities are more concentrated within
crystallites near the boundaries and less at the core of each
grain (Figure 11). This defect modeling could explain the
dierences that we have detected in the XPS signals (Table 3).
Indeed, higher concentrations of dopants in crystallites at the
grain surface produce more intense signals, as less photo-
electrons are available from inner crystallites. Actually,
photoelectrons generated from regions buried in depth within
the grain have a higher probability of being inelastically
scattered prior to escaping from the surface.
The inelastic mean free path of the emitted photoelectrons
associated with the characteristic kinetic energy of the Al K
DOI: 10.1021/acs.jpcc.5b06827
J. Phys. Chem. C 2015, 119, 2410424115
The Journal of Physical Chemistry C
Figure 11. Sketch of the mesoscale defect model proposed to explain
compositional dierences detected by XPS spectroscopy (see text).
The ellipsoids represent TiO2 grains, small blue boxes (possibly)
dierently oriented crystallites and darker regions imply a higher
concentration of extrinsic defects. Particles have the same amount of
extrinsic defects, but with dierent distributions.
radiation (1.487 keV) here employed is 11.5 nm.58
Therefore, the expected sampling depth in our samples is
34.5 nm, which is comparable with the average crystallite
diameters retrieved by powder XRD experiments for both the
im and sy series (see Figure 3 above and Table S3, SI). In
turn, this implies that, on average, each crystallite makes
available its whole volume as a photoelectron source. However,
crystallites farther than 4.5 nm from the grain surface are
essentially silent. Hence, the fact that im materials produce
more intense Ta and Nb XPS peaks might imply that
impregnated nanoparticles are more near-surface crystallites
rich in extrinsic defects (Figure 11).
It should be stressed, however, that it is the overall amount of
defects in the host (intrinsic + induced) and guest (extrinsic)
that determines the photocatalytic properties of these
materials.5 In this respect, EPR provides accurate estimates of
the ratios between paramagnetic host Oxygen species (O) and
guest Nb defects. When they are complemented with the
experimental Egs retrieved from DRS spectra (section 3.4), it is
possible to disentangle these two aspects, at least qualitatively
(Figure 12).
Figure 12. EPR-derived [Nb]/[O] ratios R for im (circles, dotted
line) and sy (triangles, full line) TN, TNbN, and TTaN materials.
For each point, the band gap width in eV, as estimated through the
KubelkaMunk equation, is also shown.
In Figure 12, the two sample series show very dierent
trends. For the im materials, the ratio among paramagnetic
defects is high in TN and then undergoes an abrupt reduction
when metal species are also present, while the apparent band
gap is invariant. Quantum mechanical calculations show that a
reduction of the band gap is mostly associated with shallow
midgap states created by oxygen vacancies.8 Otherwise, Eg is
expected to undergo a slight reduction in N (and TM,N co-)
doped samples, as occupied valence N states lie just at the
valence band edge15 (see also section 3.3).
Combined EPR, DRS, and DFT results provide compelling
evidence that the im method does not change the
concentration of host defects, that is, the denominator in the
24112
Article
[Nb]/[O] ratio. The abrupt reduction of this ratio (Figure 12,
circles) can be explained by the reduction of the amount of
paramagnetic Nb species that, in turn, can be related to the
occurrence of an internal electron transfer15 from TM to N
(see also sections 3.3 and 3.6 above). When considering the
sy samples, we observe a very dierent behavior (Figure 12,
triangles). First, the [Nb]/[O] ratio is always low and
constant, while Eg is systematically smaller than those of im
materials. This latter evidence suggests that in these cases the
concentration of host defects is always higher. This is also in
accordance with the increased absorption eciency detected
for these powders in the visible range (section 3.4 above).
We can now tentatively interpret these experimental results
at the light of the measured photocalytic activity (see section
3.5) of the two im and sy sample series. Under UV
irradiation, im samples perform better than sy ones, with
Ta-doped powders being signicantly more ecient than the
Nb-doped ones (Figure 9a,b). The rst observation can be
explained by invoking the anisotropic guest defect distribution
at the mesoscale we have hypothesized, as near-surface
crystallites should bear a higher amount of extrinsic guest
species and are therefore expected to be more ecient in light
harvesting. Also, the higher UV activity of Ta-doped specimens,
can be traced back to the electronic states injected by Ta ions
within the conduction band (see section 3.3), since these
orbitals improve the probability to allocate the photoexcited
carries.
When subjected to simulated solar light irradiation, sy
Nb,N-codoped and Nb-doped materials are the most active
ones, while all the impregnated samples are even less
performing than the undoped reference T (section 3.5). It
should be remarked that the spectrum of the solar lamp mostly
bears longer-wavelength components, which penetrate more
deeply than shorter UV ones. Therefore, it is reasonable that
larger portions of the grains are involved in the process of
creating photogenerated carriers. Consequently, dierences in
the defect distributions at the mesoscale (Figure 11) are
expected to be less important, and the electronic structure, as
resulting from the cooperative interplay between guest and host
defects, becomes determinant. In general, the less defective
im samples do not exploit any enhancement of the
photocatalytic activity with respect to the T reference. Instead,
the sy materials, which contain a higher amount of
paramagnetic host defects (vide supra), do enhance the
photocatalytic activity. This implies that a strict correlation
exists, as expected, between such defects and the visible-light
activity of the photocatalyst. At the same time, the synergic role
of Nb and N codopants15 is conrmed, as the TNbN material is
absolutely the best performing one. Also, sy Ta-doping
performs worse than Nb-doping (section 3.5, Figure 9a),
possibly because of kinetic eects. More specically, an
increased charge recombination is likely to occur since larger
Ta5+ ions induce a bit higher lattice strain than the
corresponding Nb5+-containing materials (Figure 3 and Table
S3, SI), that is, more lattice defects are present in Ta-doped
systems. A too high defect concentration is detrimental to
photocatalysis, as it favors recombination of the charge
carriers.5 It is also worth noting that the worst performing
TTaN specimen is associated with the highest (Table S3, SI)
and to the lowest Eg (Figure 12) estimates.
DOI: 10.1021/acs.jpcc.5b06827
J. Phys. Chem. C 2015, 119, 2410424115
The Journal of Physical Chemistry C
5. CONCLUSIONS
In this work, a thorough multidisciplinary experimental (XRPD,
BET, XPS, DRS, EPR) and theoretical (plane-wave DFT) study
was undertaken on both single-doped and codoped TM,N (TM
= Nb, Ta) nanostructured TiO2 materials. The photocatalytic
performances of the synthesized powders were tested toward
the gas phase degradation of ethanol under both UV and
simulated solar light irradiation. This work mainly focused on
relating the eciency of the photocatalytic process to the
synthetic route employed to feed titania nanocrystals with guest
species, i.e. impregnation (im) vs bulk synthesis (sy).
We found that the main absorption features of the im
samples are unchanged with respect to the undoped TiO2
reference, whereas the sy ones show a signicant apparent
reduction of the band gap. Accordingly, upon solar light
irradiation sy materials are the most active ones, while all the
im samples performing worse than the undoped TiO2
reference. These dierences can be understood on the grounds
of the dierent amount of host defects within the two sample
series. sy materials are more defective and the lattice
microstrain of Ta-containing nanoparticles are always higher
than the corresponding Nb-containing ones. In other words,
the macroscopic photocatalytic activity in this energy range is
mainly correlated with the presence of host defects. The more
defective Ta-doped materials show worse performances due to
a higher likelihood of charge recombination.
Under UV irradiation, the picture is very dierent. Here im
samples perform better than sy ones, with Ta-doped powders
being signicantly more ecient than the Nb-doped ones. To
account for this reversed behavior, we developed an anisotropic
defect distribution model at the mesoscale. In few words, we
observe that in im powders the crystallites near the grain
surface bear a higher amount of guest species. Therefore, they
are more ecient in light harvesting. Ta-doped specimens, on
the other hand, degrade ethanol more eciently than Nb-
doped ones as guest Ta ions inject 5d-like states within the
TiO2 conduction band, enhancing the number of e-h+ pairs
available to promote redox reactions at the grain surface.
As a possible take-home message, we would like to point out
that ne details of the selected synthetic procedure, such as just
the moment when the guest species are provided to the
reaction mixture, might have a dramatic eect on several crucial
macroscopic properties of the resulting material. A thoughtful
selection of the doping strategy should allow one to control the
photocatalytic response even without changing nature and
amount of the extrinsic dopants. In turn, this could help in
developing novel strategies to engineer nanomaterials for
tailored applications.
ASSOCIATED CONTENT
*
S Supporting Information
The Supporting Information is available free of charge on the
ACS Publications website at DOI: 10.1021/acs.jpcc.5b06827.
Section 1: Adopted experimental conditions and full
microstructure results from XRPD. Full list of the X-ray
powder diraction results. Section 2: BET adsorption
desorption isotherms, pore volume distribution, correla-
tion between specic surface area, and crystallite
dimensions for all the samples. Section 3: DFT modeling
conditions. Comparison between theoretical (DFT+U)
and experimental (XRPD renements) calculated strains.
Section 4: Band gap values from KubelkaMunk
24113
Article
elaboration of DRS for all the samples. Section 5:
Typical proceeding of ethanol degradation reaction.
Lamps emission spectra. Section 6: EPR spectra of
TN(im), TNbN(sy), and TTaN(sy) compared with
TN(sy). XPS spectra of TNb(im) and TTaN(im) in the
TM region (PDF).
AUTHOR INFORMATION
Corresponding Author
*E-mail: leonardo.lopresti@unimi.it. Phone: +39 0250314252.
Notes
The authors declare no competing nancial interest.
ACKNOWLEDGMENTS
We acknowledge the Italian supercomputing center CINECA
and the Regione Lombardia award under the LISA initiative
SURGREEN, for the availability of high performance
computing resources and support. We also thank Universita
degli Studi di Milano for further computational time at
CINECA and the Chemistry Department for funding through
the Development Plan of Athenaeum grant-line B1 (UNIAGI
17777). Partial funding by the Danish National Research
Foundation (DNRF) through the Center for Materials
Crystallography (CMC) has been also very much appreciated.
REFERENCES
(1) Fujishima, A.; Zhang, X.; Tryk, D. A. TiO2 Photocatalysis and
Related Surface Phenomena. Surf. Sci. Rep. 2008, 63, 515582.
(2) Serpone, N.; Emeline, A. V. Semiconductor Photocatalysis - Past,
Present, and Future Outlook. J. Phys. Chem. Lett. 2012, 3, 673677.
(3) Serpone, N. Is the Band Gap of Pristine TiO2 Narrowed by
Anion- and Cation-Doping of Titanium Dioxide in Second-Generation
Photocatalysts? J. Phys. Chem. B 2006, 110, 2428724293.
(4) Asahi, R.; Morikawa, T.; Irie, H.; Ohwaki, T. Nitrogen-Doped
Titanium Dioxide as Visible-Light-Sensitive Photocatalyst: Designs,
Developments, and Prospects. Chem. Rev. 2014, 114, 98249852.
(5) Spadavecchia, F.; Ceotto, M.; Presti, L.; Aieta, C.; Biraghi, I.;
Meroni, D.; Ardizzone, S.; Cappelletti, G. Second Generation
Nitrogen Doped Titania Nanoparticles: A Comprehensive Electronic
and Microstructural Picture. Chin. J. Chem. 2014, 32, 11951213.
(6) Asahi, R. Visible-Light Photocatalysis in Nitrogen-Doped
Titanium Oxides. Science 2001, 293, 269271.
(7) Meroni, D.; Ardizzone, S.; Cappelletti, G.; Oliva, C.; Ceotto, M.;
Poelman, D.; Poelman, H. Photocatalytic Removal of Ethanol and
Acetaldehyde by N-Promoted TiO2 Films: The Role of the Different
Nitrogen Sources. Catal. Today 2011, 161, 169174.
(8) Ceotto, M.; Lo Presti, L.; Cappelletti, G.; Meroni, D.;
Spadavecchia, F.; Zecca, R.; Leoni, M.; Scardi, P.; Bianchi, C. L.;
Ardizzone, S. About the Nitrogen Location in Nanocrystalline N-
Doped TiO2: Combined DFT and EXAFS Approach. J. Phys. Chem. C
2012, 116, 17641771.
(9) Zhang, J.; Wu, Y.; Xing, M.; Leghari, S. A. K.; Sajjad, S.
Development of Modified N Doped TiO2 Photocatalyst with Metals,
Nonmetals and Metal Oxides. Energy Environ. Sci. 2010, 3, 715.
(10) Long, R.; English, N. J. Band Gap Engineering of (N,Ta)-
Codoped TiO2: A First-Principles Calculation. Chem. Phys. Lett. 2009,
478, 175179.
(11) Sheppard, L. R.; Hager, S.; Holik, J.; Liu, R.; Macartney, S.;
Wuhrer, R. Tantalum Segregation in Ta-Doped TiO2 and the Related
Impact on Charge Separation during Illumination. J. Phys. Chem. C
2015, 119, 392400.
(12) Wang, C.; Geng, A.; Guo, Y.; Jiang, S.; Qu, X. Three-
Dimensionally Ordered Macroporous Ti1xTaxO2+x/2 (x = 0.025,
0.05, and 0.075) Nanoparticles: Preparation and Enhanced Photo-
catalytic Activity. Mater. Lett. 2006, 60, 27112714.
DOI: 10.1021/acs.jpcc.5b06827
J. Phys. Chem. C 2015, 119, 2410424115
The Journal of Physical Chemistry C
(13) Altomare, M.; Lee, K.; Killian, M. S.; Selli, E.; Schmuki, P. Ta-
Doped TiO2 Nanotubes for Enhanced Solar-Light Photoelectrochem-
ical Water Splitting. Chem. - Eur. J. 2013, 19, 58415844.
(14) Bawaked, S. M.; Sathasivam, S.; Bhachu, D. S.; Chadwick, N.;
Obaid, A. Y.; Al-Thabaiti, S.; Basahel, S. N.; Carmalt, C. J.; Parkin, I. P.
Aerosol Assisted Chemical Vapor Deposition of Conductive and
Photocatalytically Active Tantalum Doped Titanium Dioxide Films. J.
Mater. Chem. A 2014, 2, 12849.
(15) Marchiori, C.; Di Liberto, G.; Soliveri, G.; Loconte, L.; Lo
Presti, L.; Meroni, D.; Ceotto, M.; Oliva, C.; Cappelli, S.; Cappelletti,
G.; et al. Unraveling the Cooperative Mechanism of Visible-Light
Absorption in Bulk N,Nb Codoped TiO2 Powders of Nanomaterials. J.
Phys. Chem. C 2014, 118, 2415224164.
(16) Breault, T. M.; Bartlett, B. M. Lowering the Band Gap of
Anatase-Structured TiO2 by Coalloying with Nb and N: Electronic
Structure and Photocatalytic Degradation of Methylene Blue Dye. J.
Phys. Chem. C 2012, 116, 59865994.
(17) Lim, J.; Murugan, P.; Lakshminarasimhan, N.; Kim, J. Y.; Lee, J.
S.; Lee, S. H.; Choi, W. Synergic Photocatalytic Effects of Nitrogen
and Niobium Codoping in TiO2 for the Redox Conversion of Aquatic
Pollutants under Visible Light. J. Catal. 2014, 310, 9199.
(18) Breault, T.; Bartlett, B. Composition Dependence of TiO2:(Nb,
N)-X Compounds on the Rate of Photocatalytic Methylene Blue Dye
Degradation. J. Phys. Chem. C 2013, 117, 86118618.
(19) Zhu, Y.; Wu, Y.; Lu, Y.; Xu, J.; Wang, Y.; Ma, X. Effect of
Compensated Codoping on the Photoelectrochemical Properties of
Anatase TiO2 Photocatalyst Effect of Compensated Codoping on the
Photoelectrochemical Properties of Anatase TiO2 Photocatalyst. J.
Phys. Chem. C 2011, 115, 1696316969.
(20) Cottineau, T.; Bealu, N.; Gross, P.-A.; Pronkin, S. N.; Keller, N.;
Savinova, E. R.; Keller, V. One Step Synthesis of Niobium Doped
Titania Nanotube Arrays to Form (N,Nb) Codoped TiO2 with High
Visible Light Photoelectrochemical Activity. J. Mater. Chem. A 2013, 1,
2151.
(21) Liu, Q.-L.; Zhao, Z.-Y.; Liu, Q.-J. Impact of Sulfur-, Tantalum-,
or Codoping on the Electronic Structure of Anatase Titanium Dioxide:
A Systematic Density Functional Theory Investigation. Mater. Sci.
Semicond. Process. 2015, 33, 94102.
(22) Obata, K.; Irie, H.; Hashimoto, K. Enhanced Photocatalytic
Activities of Ta, N Codoped TiO2 Thin Films under Visible Light.
Chem. Phys. 2007, 339, 124132.
(23) Wang, W.; Lu, C.; Ni, Y.; Su, M.; Huang, W.; Xu, Z. Preparation
and Characterization of Visible-Light-Driven N-F-Ta Tri-Doped TiO2
Photocatalysts. Appl. Surf. Sci. 2012, 258, 86968703.
(24) Gong, Y.; Fu, C.; Ting, L.; Chenu, J.; Zhao, Q.; Li, C. Exploring
the Effect of Boron and Tantalum Codoping on the Enhanced
Photocatalytic Activity of TiO2. Appl. Surf. Sci. 2015, 351, 746752.
(25) Hoang, S.; Guo, S.; Mullins, C. B. Coincorporation of N and Ta
into TiO2 Nanowires for Visible Light Driven Photoelectrochemical
Water Oxidation. J. Phys. Chem. C 2012, 116, 2328323290.
(26) Huang, J.; Wen, S.; Liu, J.; He, G. Band Gap Narrowing of TiO2
by Compensated Codoping for Enhanced Photocatalytic Activity. J.
Nat. Gas Chem. 2012, 21, 302307.
(27) Yin, W. J.; Tang, H.; Wei, S. H.; Al-Jassim, M. M.; Turner, J.;
Yan, Y. Band Structure Engineering of Semiconductors for Enhanced
Photoelectrochemical Water Splitting: The Case of TiO2. Phys. Rev. B:
Condens. Matter Mater. Phys. 2010, 82, 16.
(28) Liu, J.; Duan, Y.; Zhou, X.; Lin, Y. Influence of VB Group
Doped TiO2 on Photovoltaic Performance of Dye-Sensitized Solar
Cells. Appl. Surf. Sci. 2013, 277, 231236.
(29) Liu, X.; Wu, X.; Scott, K. Study of Niobium and Tantalum
Doped Titania-Supported Pt Electrocatalysts for Methanol Oxidation
and Oxygen Reduction Reactions. Catal. Sci. Technol. 2014, 4, 3891
3898.
(30) Mazzolini, P.; Gondoni, P.; Russo, V.; Chrastina, D.; Casari, C.
S.; Li Bassi, A. Tuning of Electrical and Optical Properties of Highly
Conducting and Transparent Ta-Doped TiO2 Polycrystalline Films. J.
Phys. Chem. C 2015, 119, 69886997.
24114
Article
(31) Ay, H.; U ner, D. Dry Reforming of Methane over CeO2
Supported Ni, Co and NiCo Catalysts. Appl. Catal., B 2015, 179,
128138.
(32) Minguzzi, A.; Locatelli, C.; Cappelletti, G.; Bianchi, C. L.;
Vertova, A.; Ardizzone, S.; Rondinini, S. Designing Materials by Means
of the Cavity-Microelectrode: The Introduction of the Quantitative
Rapid Screening toward a Highly Efficient Catalyst for Water
Oxidation. J. Mater. Chem. 2012, 22, 8896.
(33) Antonello, A.; Soliveri, G.; Meroni, D.; Cappelletti, G.;
Ardizzone, S. Photocatalytic Remediation of Indoor Pollution by
Transparent TiO2 Films. Catal. Today 2014, 230, 3540.
(34) Dudarev, S. L.; Savrasov, S. Y.; Humphreys, C. J.; Sutton, A. P.
Electron-Energy-Loss Spectra and the Structural Stability of Nickel
Oxide: An LSDA+U Study. Phys. Rev. B: Condens. Matter Mater. Phys.
1998, 57, 15051509.
(35) Kresse, G.; Hafner, J. Ab Initio Molecular-Dynamics Simulation
of the Liquid-Metalamorphous-Semiconductor Transition in Germa-
nium. Phys. Rev. B: Condens. Matter Mater. Phys. 1994, 49, 14251
14269.
(36) Kresse, G. Efficient Iterative Schemes for Ab Initio Total-Energy
Calculations Using a Plane-Wave Basis Set. Phys. Rev. B: Condens.
Matter Mater. Phys. 1996, 54, 1116911186.
(37) Kresse, G. From Ultrasoft Pseudopotentials to the Projector
Augmented-Wave Method. Phys. Rev. B: Condens. Matter Mater. Phys.
1999, 59, 17581775.
(38) Kresse, G.; Hafner, J. Ab Initio Molecular Dynamics for Liquid
Metals. Phys. Rev. B: Condens. Matter Mater. Phys. 1993, 47, 558561.
(39) Perdew, J. P.; Burke, K.; Ernzerhof, M. Generalized Gradient
Approximation Made Simple. Phys. Rev. Lett. 1996, 77, 38653868.
(40) Blochl, P. E. Projector Augmented-Wave Method. Phys. Rev. B:
Condens. Matter Mater. Phys. 1994, 50, 1795317979.
(41) Perdew, J. P.; Wang, Y. Accurate and Simple Analytic
Representation of the Electron-Gas Correlation Energy. Phys. Rev. B:
Condens. Matter Mater. Phys. 1992, 45, 1324413249.
(42) Lo Presti, L.; Ceotto, M.; Spadavecchia, F.; Cappelletti, G.;
Meroni, D.; Acres, R. G.; Ardizzone, S. Role of the Nitrogen Source in
Determining Structure and Morphology of N-Doped Nanocrystalline
TiO2. J. Phys. Chem. C 2014, 118, 47974807.
(43) Spadavecchia, F.; Cappelletti, G.; Ardizzone, S.; Ceotto, M.;
Azzola, M. S.; Presti, L.; Cerrato, G.; Falciola, L. Role of Pr on the
Semiconductor Properties of Nanotitania. An Experimental and First-
Principles Investigation. J. Phys. Chem. C 2012, 116, 2308323093.
(44) Mullin, J. W. Crystallization; Butterworth-Heinemann: Oxford,
U.K., 2001.
(45) Shannon, R. Revised Effective Ionic Radii and Systematic
Studies of Interatomic Distances in Halides and Chalcogenides. Acta
Crystallogr., Sect. A: Cryst. Phys., Diffr., Theor. Gen. Crystallogr. 1976, 32,
751767.
(46) Ardizzone, S.; Bianchi, C. L.; Cappelletti, G.; Gialanella, S.;
Pirola, C.; Ragaini, V. Tailored Anatase/Brookite Nanocrystalline
TiO2. J. Phys. Chem. C 2007, 111, 1322213231.
(47) Zhao, Z.; Liu, Q. Designed Highly Effective Photocatalyst of
Anatase TiO2 Codoped with Nitrogen and Vanadium under Visible-
Light Irradiation Using First-Principles. Catal. Lett. 2008, 124, 111
117.
(48) Spadavecchia, F.; Cappelletti, G.; Ardizzone, S.; Ceotto, M.;
Falciola, L. Electronic Structure of Pure and N-Doped TiO2
Nanocrystals by Electrochemical Experiments and First Principles
Calculations. J. Phys. Chem. C 2011, 115, 63816391.
(49) Irie, H.; Watanabe, Y.; Hashimoto, K. Nitrogen-Concentration
Dependence on Photocatalytic Activity of TiO2-xNx Powders. J. Phys.
Chem. B 2003, 107, 54835486.
(50) Livraghi, S.; Paganini, M. C.; Giamello, E.; Selloni, A.; Di
Valentin, C.; Pacchioni, G. Origin of Photoactivity of Nitrogen-Doped
Titanium Dioxide under Visible Light. J. Am. Chem. Soc. 2006, 128,
1566615671.
(51) Kuznetsov, V. N.; Serpone, N. On the Origin of the Spectral
Bands in the Visible Absorption Spectra of Visible-Light-Active TiO2
DOI: 10.1021/acs.jpcc.5b06827
J. Phys. Chem. C 2015, 119, 2410424115
The Journal of Physical Chemistry C Article

Specimens Analysis and Assignments. J. Phys. Chem. C 2009, 113,

1511015123.
(52) Biedrzycki, J.; Livraghi, S.; Giamello, E.; Agnoli, S.; Granozzi, G.
Fluorine- and Niobium-Doped TiO2: Chemical and Spectroscopic
Properties of Polycrystalline N-Type-Doped Anatase. J. Phys. Chem. C
2014, 118, 84628473.
(53) De Trizio, L.; Buonsanti, R.; Schimpf, A. M.; Llordes, A.;
Gamelin, D. R.; Simonutti, R.; Milliron, D. J. Nb-Doped Colloidal
TiO2 Nanocrystals with Tunable Infrared Absorption. Chem. Mater.
2013, 25, 33833390.
(54) Nguyen, H. T. H.; Takenaka, N.; Bandow, H.; Maeda, Y.; De
Oliva, S. T.; Botelho, M. M. F.; Tavares, T. M. Atmospheric Alcohols
and Aldehydes Concentrations Measured in Osaka, Japan and in Sao
Paulo, Brazil. Atmos. Environ. 2001, 35, 30753083.
(55) Kwon, K. D.; Jo, W. K.; Lim, H. J.; Jeong, W. S. Volatile
Pollutants Emitted from Selected Liquid Household Products. Environ.
Sci. Pollut. Res. 2008, 15, 521526.
(56) Brickus, L. S. R.; Cardoso, J. N.; De Aquino Neto, F. R.
Distributions of Indoor and Outdoor Air Pollutants in Rio de Janeiro,
Brazil: Implications to Indoor Air Quality in Bayside Offices. Environ.
Sci. Technol. 1998, 32, 34853490.
(57) Gaffney, J. S.; Marley, N. A. The Impacts of Combustion
Emissions on Air Quality and Climate - From Coal to Biofuels and
beyond. Atmos. Environ. 2009, 43, 2336.
(58) Powell, C. J.; Jablonski, A. NIST Electron Inelastic-Mean-Free-
Path Database; National Institute of Standard and Technology:
Gaithersburg, MD, U.S.A., 2010.

24115 DOI: 10.1021/acs.jpcc.5b06827

J. Phys. Chem. C 2015, 119, 2410424115