You are on page 1of 15

Article

pubs.acs.org/JPCC

Self-Sensitized Photocatalytic Degradation of Colorless Organic


Pollutants Attached to Rutile NanorodsExperimental and
Theoretical DFT+D Studies
Kezhen Qi,† Filip Zasada,*,† Witold Piskorz,† Paulina Indyka,† Joanna Gryboś,† Mateusz Trochowski,†
Marta Buchalska,† Marcin Kobielusz,† Wojciech Macyk,*,† and Zbigniew Sojka†

Faculty of Chemistry, Jagiellonian University, ul. Ingardena 3, 30-060 Kraków, Poland
*
S Supporting Information
See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles.

ABSTRACT: Utilization of visible light by photosensitization


Downloaded via QUEEN'S UNIV OF BELFAST on October 28, 2023 at 11:52:38 (UTC).

of semiconductor photocatalysts via surface attachment of


small colorless organic pollutants (COP) is an effective way to
stimulate their photocatalytic degradation. Herein, by means of
the spectroscopic, photoelectrochemical, spectroelectrochem-
ical, and photocatalytic studies combined with the DFT+D
molecular modeling, we show how disubstituted benzene
derivatives, like catechol (CAT), salicylic acid (SAL), phthalic
acid (PTA), and terephthalic acid (TPA), can tune the
photocatalytic properties of rutile nanorods with the dominant
(110) termination. We elucidated in a systematic way the COP
ligand-binding configurations, the alignment of energy levels, and the charge-transfer pathways from the organic admolecules to
the titania substrate. The pDOS structures of the COP@r-TiO2(110) assemblies were interpreted in terms of electronic
interactions between the titania photocatalyst and the COP adspecies. It was shown that the appearance of additional states
within the band gap and in the conduction band allows for a one-step HOMO → CB ligand to metal charge transfer and a two-
step HOMO → LUMO → CB sensitization. Screening of the photocatalytic performance of the COP@r-TiO2 samples revealed
that the self-degradation efficiency gauged by the initial rate constant varies in the following order: catechol (−OH, −OH; 0.024
min−1) > salicylate (−OH, −COOH; 0.013 min−1) > phthalates (−COOH, −COOH; 0.005 and 0.004 min−1 for PTA and TPA,
respectively), showing a beneficial role of hydroxyl functionalities at an early stage of degradation process. It was found that the
higher activity of the OH-bearing catechol and salicylate adspecies was associated with the direct HOMO → CB electron-transfer
pathway operating in the visible light. The two-step HOMO → LUMO → CB mechanism (requiring UV light) characteristic of
carboxyl-bearing functionalities, despite favorable energy level alignment and coupling, is less efficient due to low density of the
electronic states at the top of the conduction band, and low flux of the solar radiation in that energy region. The in situ diffuse
reflectance spectroscopic (DRS) measurements revealed that at early stages of the photocatalytic degradation the aromatic rings
of the COP moieties are readily photohydroxylated, fostering the visible light utilization via the HOMO → CB electron transfer
route. Such latent autocatalytic hydroxylation processes are relevant for photocatalytic degradation of those pollutants that
originally do not exhibit hydroxyl functionalities provided that a photogenerated hole is localized at the organic moiety.

■ INTRODUCTION
Wide-band-gap semiconductors, such as titania, are promising
photocatalysts. Band gap modifications, resulting from either
tuning the valence and conduction edge levels or by
materials for a wide range of applications in environmental introducing the empty or occupied states within the band
photocatalysis1−4 and energy conversion.5,6 Their importance gap, are of a primary importance in chemical systems that
results from the unique electronic properties that depend utilize light for various photocatalytic purposes. Another astute
mostly on the favorable energies of the band gap edges. Yet, the approach involves fabrication of appropriate heterostructures
photoactivity of TiO2 is limited by its rather large intrinsic band that play a crucial role in the titania photocatalyst amelioration,
gap of ∼ 3.0 eV, making this oxide capable of absorbing the as it has been recently reviewed.14 The resultant variants of
ultraviolet light only. This fact lowers the efficiency of the solar band alignment at the interface include a type I scheme, where
light harvesting and disfavors photocatalytic degradation of the band energy levels of one material straddle those of the
pollutants such as volatile organic compounds (VOC) or water- other resulting in the transfer of both holes and electrons to the
soluble organic compounds (WSOC). Band gap engineering
and appropriate sensitization by bulk doping7−9 and surface Received: November 9, 2015
functionalization10−13 are then the key issues for more effective Revised: February 15, 2016
application of various titania polymorphs as the proficient Published: February 16, 2016

© 2016 American Chemical Society 5442 DOI: 10.1021/acs.jpcc.5b10983


J. Phys. Chem. C 2016, 120, 5442−5456
The Journal of Physical Chemistry C Article

narrower band gap material. In the type II heterostructures, the The interaction between small benzene derivatives and the
band energy levels are staggered (or broken−type III), making titania surface depends not only on the structure of the exposed
the charge separation energetically more favorable.15 facets but also on the chemical nature of the functional groups
Functionalization of semiconductor surfaces with organic acting as anchoring moieties. Among them, hydroxyl (−OH)
dyes has attracted much attention because the ensuing hybrid and carboxyl (−COOH) groups belong to the most common
interfaces allow us to control more effectively not only the light anchors.38 Herein, we combine photocurrent, spectroelectro-
harvesting but also the electron injection from the photoexcited chemical, and photodegradation measurements with DFT
dye LUMO state to the semiconductor conduction band and modeling to study the self-sensitization of r-TiO2 nanorods
the charge recombination as well.16,17 Despite the undeniable with selected simple COP benzene derivatives, such as catechol,
progress made in this field recently, understanding the detailed salicylic acid, phthalic acid, and terephthalic acid. These
mechanism of those processes at the molecular level is still a molecules contain hydroxyl and carboxyl functionalities in
subject of extensive experimental and theoretical investiga- different positions of the aromatic ring, which are commonly
tions.18 Typically, the efforts toward improving the sunlight- formed during oxidation of aromatic pollutants. The goal was
induced activity are focused at increasing dye extinction to elucidate the impact of the anchoring diversity imposed by
coefficient, shifting its absorption toward longer wavelengths the −COOH and −OH groups on the electronic structure
to improve the compatibility with the solar light and a better modification of the rutile nanorod photocatalyst, alignment of
energetic matching between the electron donor and acceptor energy levels, and the photodegradation pathways of these
levels. prototypical colorless pollutants.
In contrast to organic dyes, which are intentionally designed
to exhibit good light-harvesting properties due to their high
extinction coefficient, small aromatic pollutants are usually
■ EXPERIMENTAL SECTION
Sample Preparation and Characterization. Synthesis.
colorless. Their efficient photodegradation depends essentially For synthesis of rutile nanorods, 0.2 mL butyl titanate (TBOT)
on a direct excitation of the photocatalyst (UV-light) or a self- was dropped into 16 mL of hydrochloric acid solution
sensitization effect,19 made possible even in the visible region (composed of 4 mL of 36.5% HCl, 12 mL of water, and 2
by favorable band alignment and appropriate frontier orbital mg of cetyltrimethylammonium bromide). Then the mixture
coupling to the conduction band states of the photocatalyst.20 was transferred to a 20 mL Teflon-lined steel autoclave and
It has been argued that formation of surface charge-transfer heated at 180 °C for 24 h. Then the autoclave was cooled to
complexes plays a beneficial role in pushing the photocatalytic room temperature. The resulting product was separated by
degradation of colorless organic pollutants (COP) toward the centrifugation, washed several times with absolute ethanol and
visible region.21 However, until now, this issue has been rather distilled water, and finally dried at 60 °C in air for 5 h. The
scarcely studied. synthesized material was additionally calcined at 450 °C for 3 h
Among various wide-band-gap semiconductor materials, to produce a well-crystalline r-TiO2 material. The surface
rutile TiO2 (r-TiO2) is a common substrate for the design of modification of the rutile samples with organic molecules was
tailored heterointerfaces.22−24 The main exposed (110) facet of achieved by impregnation of r-TiO2 (50 mg) with 5 mL of the
the r-TiO2 crystals is not only very stable but also an excellent 0.1 mol·dm−3 of COP solution in methanol. Catechol (CAT),
template with protruding dangling bonds apt for interacting salicylic acid (SAL), phthalic acid (PTA), and terephthalic acid
efficiently with organic compounds such as alcohols (e.g., (TPA) were used as exemplary COP molecules for surface
methanol),25 phenols (e.g., 4-chlorophenol),26 or aromatic modification. The suspension of TiO2 in the COP solution was
diols (e.g., catechol).27,28 These organic molecules with treated with ultrasound for 5 min and left overnight. Then the
different functional groups were used to exploit the site-specific powders were centrifuged, washed 3 times with water, and
adsorption chemistry of the r-TiO2(110) surface.29 dried in the oven at 50 °C.
The density functional theory (DFT) modeling has been Characterization of Materials. The phase composition of
widely applied to explore the interactions between organic the samples was verified by X-ray diffraction (XRD) (Rigaku
admolecules and oxide surfaces.30−32 For instance, Selloni et al. D/max 2500 V/PC, Cu Kα radiation, λ = 1.5406 Å). UV−vis
studied self-organization of catechol on rutile (110)29 and diffuse reflectance spectra of the synthesized materials were
anatase (101)26,33 using DFT calculations supported by STM recorded using PerkinElmer Lambda 12 spectrophotometer
imaging. They found that the most energetically favorable equipped with an integrating sphere of 5 cm in diameter.
bonding takes place in a bridging fashion involving two Raman spectra were recorded using a Renishaw, in Via
titanium atoms. DFT calculations with inclusion of dispersion instrument, operating with the 785 nm excitation laser source.
forces revealed, in turn, the key importance of the van der Transmission electron microscopy (TEM) imaging was carried
Waals interactions for ordering aromatic admolecules on the out by means of a Tecnai Osiris microscope (FEI) operating at
rutile surface.34,35 Calzolari et al. have used DFT calculations to 200 kV. Prior to TEM analysis the samples were ultrasonically
study sensitization of the ZnO (10−10) surface by adsorbed dispersed in methanol on a holey carbon film supported on a
catechol and found that the alignment of energy levels copper grid (400 mesh). The grid was dried for 45 min, and
originates from the simultaneous interplay between the then surface contaminations were removed by plasma-cleaning
conjugation of the adsorbed molecule and the electron (Solarus Gatan 950). The simulations of TEM images were
donor/acceptor capability of the specific anchoring groups of carried out using JEMS simulation package.39 According to the
the adsorbed molecule.36 Thomas et al. have investigated by experimental conditions of imaging, the Cs and Cc coefficients,
means of DFT modeling p-aminobenzoic acid (pABA) equal to 1.2 μm, were used. Two perpendicular families of
interaction with the anatase-TiO2(101) surface. The observed diffraction spots, located on the diffraction pattern 3.26 nm−1
red shift after pABA adsorption has been attributed to the and 3.02 nm−1 away from the central spot, were assigned to the
presence of the highest occupied molecular orbitals within the (001) and (01̅1) planes, respectively, revealing a [110]
TiO2 band gap region.37 orientation of the specimen. In order to determine the values
5443 DOI: 10.1021/acs.jpcc.5b10983
J. Phys. Chem. C 2016, 120, 5442−5456
The Journal of Physical Chemistry C Article

of defocus (Δf) and the specimen thickness (t), a wide range of the band gap, we employed the DFT+U approach46 with the
possible settings, from 55 to 72 nm for Δf and 2÷20 nm for t, Hubbard parameter set to U = 3.0 eV for all Ti ions. For
was tested. Detailed conditions used for image simulations are modeling the interaction between the hydrocarbon admolecules
presented in Table S1 in Supporting Information (SI). and the oxide substrate, a semiempirical dispersion term,
Photoelectrochemical Measurements. Photocurrents were parametrized by Grimme,47 was added to the quantum-
measured using a three-electrode setup. The electrolyte (0.1 mechanical energies and gradients (DFT+D).
mol·dm−3 KNO3) was purged with argon for at least 5 min A number of tests of the TiO2 bulk properties were initially
prior to and during the measurements. A platinum wire and performed to verify the accuracy of the applied calculation
Ag/AgCl were used as the counter and reference electrodes, scheme (variation of the cutoff energy and the k-point set). The
respectively. A 150 W xenon lamp (XBO-150), equipped with a standard Monkhorst−Pack48 grid (4 × 4 × 3 sampling mesh for
water-cooled housing and an automatically controlled mono- bulk calculations and 3 × 3 × 2 for slab calculations) with the
chromator (Instytut Fotonowy), was used as the light source. cutoff energy of 450 eV and the Methfessel−Paxton49 smearing
The working electrodes (ITO-coated transparent foil of 60 Ω/ parameter σ = 0.1 eV were used. For solving the Kohn−Sham
sq resistance, Sigma-Aldrich, with the casted and dried equations, the SCF convergence criterion was set to energy
material) were irradiated from the backside through the ITO change of 10−5 eV between two successive iterations. Geometry
layer in order to minimize the influence of the film thickness on optimization was performed until the changes in the forces
the measured photocurrents. The measurements were con- acting upon the ions were smaller than 0.001 eV/Å per atom.
trolled with the electrochemical analyzer (Autolab, Bulk titania unit cell was obtained by optimization of the
PGSTAT302N). experimental tetragonal (D4h14-P42/mnm) rutile (2 × 2 × 4)
Photocatalytic Decomposition of COP Molecules. The r- unit cell (a = b = 9.16 Å, and c = 11.80 Å) containing 96 ions
TiO2 powder (4 mg) and COP (0.1 mg) were suspended in (Ti32O64). The optimal cell volume was calculated from the E/
0.01 mol·dm−3 NaOH solution (in order to capture the V fit (Birch−Murnaghan equation of state50) with full
evolving CO2 for further analysis) and placed in a quartz optimization of all internal degrees of freedom (with an error
cylindrical cuvette (5 cm dia., 1 cm optical path, 17 mL <10−4 eV).
volume). A 150 W XBO lamp equipped with the 10 cm water Initial surface geometry was constructed by cleaving the
filter (with 0.1 mol·dm−3 CuSO4 solution) was used as the light optimized rutile structure in the [110] direction. For the
source with the fluence of 50−60 mW cm−2. The reaction calculations, we used a (4 × 2) slab supercell (a = 12.17 Å, b =
progress was monitored by measurements of the total organic 13.36 Å), which consisted of 9 atomic layers with the thickness
carbon (TOC) decay (Shimadzu, TOC-V series). In order to of 9.01 Å and a vacuum layer of 20 Å, located above the
elucidate the mechanism of self-sensitized photodegradation, modeled surface in the [110] direction. Both the stoichiometry
the SAL@r-TiO2 photocatalyst was irradiated , and the changes of the bulk TiO2 (Ti48O96), as well as the top and the bottom
in its chemical state were followed by diffuse reflectance slab terminations, was preserved. The four outermost atomic
spectroscopy. The recorded spectra were converted into the layers were allowed to relax, whereas the remaining deeper
Kubelka−Munk function. The powder samples of SAL@r-TiO2 layers were kept fixed at the optimized atomic bulk positions,
were ground in an agate mortar with BaSO4 (1:50 wt. ratio) see Figure S1 and Table S2 in SI. In the aqueous environment,
and pressed into pellets of ca. 2 cm in diameter. The catalyst the surface dangling bonds will be obviously replaced by the
pellets were irradiated with 150 W xenon lamp (XBO-150) hydroxyl groups; their influence on electronic structure,
equipped with a NIR filter (0.1 mol dm−3 aqueous CuSO4 however, should be of minor importance, so they are not
solution) and 420 nm cutoff filter. The DRS spectra were taken into account. Due to their high electronic hardness, the
collected every 10 min in the range of 250−800 nm by means hydroxyls should contribute to the DOS regions far from the
of a Shimadzu UV-3600 spectrophotometer, equipped with an band gap region; therefore, for calculation of relative values,
integrating sphere (15 cm diameter). BaSO4 was used as a their contribution cancels out to the large extend.
blank reference. Adsorption of the COP molecules was performed with the
Spectroelectrochemical Measurements. The measurements assumption that only one molecule per slab unit cell is present,
were carried out in a three-electrode cell (quartz cuvette), with which gives the theoretical surface coverage of Θ = 0.61
platinum wire and Ag/AgCl as auxiliary and reference molecule per square nm. This value is in accordance with
electrodes, respectively. The cuvette was placed in front of experimental assessment of the Θ value (vide inf ra) and
the integrating sphere, facing the working electrode (platinum corresponds to the intermolecular distance between the
plate with deposited r-TiO2) toward the light beam. The periodically repeated admolecules of at least 7 Å in each
potential control was provided by the electrochemical analyzer direction. The adsorption energy (ΔEads) was defined as ΔEads
(BioLogic SP-150) with the scan rate of 0.5 mV·s−1. Relative = ECOP@slab − (Eslab + ECOP), where ECOP@slab is the energy of
reflectance changes were collected by PerkinElmer UV−vis the optimized model of the COP/TiO2 assemble, Eslab denotes
Lambda 12 spectrometer equipped with a 5 cm diameter the energy of the bare surface, and ECOP stands for the energy
integrating sphere. of the isolated organic molecule. The atomic charges were
Computational Details. For all calculations, the DFT calculated using the Bader population analysis,51 and the bond
theory implemented in the Vienna Ab initio Simulation Package orders within the crystal orbital overlap population (COOP)
(VASP)40 was employed. We used a projector augmented plane approach.
wave (PAW)41 method for describing electron−ion interactions To estimate the injection time between the COP adspecies
together with the PW91 GGA exchange−correlation functional and the semiconductor surface, the Newns−Anderson52 model
as parametrized by Perdrew et al.,42 which produce reliable with the formulation of Persson et al.53,54 was employed.
results in describing electronic structure of oxide systems like According to this approximation, the effect of adsorption on the
titania.43−45 In accounting for the electron correlation effects electronic structure of the sensitizer is characterized by the
and the Coulomb on-site repulsion, for proper reproduction of energy shift, relative to the dye level, and the level broadening
5444 DOI: 10.1021/acs.jpcc.5b10983
J. Phys. Chem. C 2016, 120, 5442−5456
The Journal of Physical Chemistry C Article

(Γ), which provides a reasonable assessment of the lifetime of the near Scherzer focus conditions) together with the pristine
the excited state, τ = 1/Γ. The level broadening was estimated and FFT filtered images are shown in Figure 1b1 and 1b2. The
as the width of the fitted Lorentzian distribution of the density indicated interplanar distances, d(001) = 3.056 Å and d(1−10) =
of states. The partial DOS (pDOS) was calculated and 3.305 Å, corresponding to the (001) and (1−10) planes,
projected as a fraction of the total DOS (tDOS) to obtain respectively, are in a very good agreement with the values
the dye’s contribution (rDOS). The projection coefficients, pi, reported in literature.56
of rDOS are defined as follows:53,54 The DFT-optimized atomic structure of the rutile (110)
A plane (see below), shown in Figure 1c1 (Ti − blue, O − red),
∑ j (cijA )2 was used for simulation of the observed TEM pictures. The
n
∑ j (cij)2 resultant simulated images without noise and with inclusion of
the 40% noise (to match better the experimental conditions)
where i iterates bands and j iterates basis functions of dye are shown in Figure 1c2 and 1c3, respectively. They match the
atoms A. The rDOS was then fitted to the Lorentzian experiment very well. It should be noted, however, that there is
distributions by means of the Levenberg−Marquardt opti- no simple, one-to-one correspondence between the observed
mization algorithm.55 The values of full width at half-maximum TEM spots and the atomic structure (cf. Figure 1c1 with b1 and
(fwhm) of the obtained Lorentzians were used as input for the c3). In the applied TEM imaging conditions, the bright spots
Newns−Anderson model. are equivalent to the projected atomic columns, composed of

■ RESULTS AND DISCUSSION


Sample Characteristics. The XRD pattern and Raman
the Ti cations only, which are located at the corners of the
rectangular unit cell (shown in yellow in Figure 1c1, c2, and b1),
and separated by the d(100) and d(1−10) distances. The Ti
spectra (see SI, Figure S2) confirmed a high quality and phase columns located at the edges of the unit cell and separated by
purity of the synthesized rutile (P42/mnm) samples. The results half of the d(001) period remain together with the O2− anions
of microscopic HR-TEM and electron diffraction characteristics practically invisible. The resultant characteristic pattern with the
of the samples are presented in Figure 1. As revealed by Figure 4-fold symmetry, consistent with the inserted FFT image, is
then well reproduced in the simulated images.
The rutile nanorods functionalized with catechol [o-
Ph(OH)2], salicylic [o-Ph(OH)(COOH)], phthalic [o-Ph-
(COOH)2], and terephthalic [p-Ph(COOH)2] acids of differ-
ent −OH/−COOH ratio were next examined by IR spectros-
copy. The presence of the COP admolecules on the r-TiO2
nanorods was confirmed by appearance of their diagnostic
vibrational bands (see Figure S3 in SI). The surface loading in
the range of 0.12 to 0.57 molecules·nm−2 was estimated from
the amount of CO2 produced upon the temperature-
programmed oxidation of the samples, complemented by
BET surface measurements (Table S3, SI). These results were
used to build up the corresponding computational models.
Electronic Structure and Photocatalytic Activity
Studies. Following the literature,57 dye-sensitized TiO2
photocatalysts can be categorized into class-A and class-B. In
the case of the class-A systems, the electrons are injected from
the photoexcited dye admolecules to the titania substrate in a
Figure 1. High-resolution microscopic analysis of r-TiO2 nanorods. A two-step process (Scheme 1). For the class-B photocatalysts,
general view of the nanorods morphology (a1) together with electrons are injected not only along the pathway A but also by
corresponding experimental and theoretical diffraction patterns (a2 a direct one-step electron injection process via photoexcitation
and a3, respectively). Pristine (b1) and filtered HR-TEM images of the
r-TiO2 structure (b2) with the FFT pattern (inset) viewed along the Scheme 1. Mechanistic Pathways of Photoinduced Charge-
[110] direction. The simulated TEM images with the imposed atomic Transfer Processes in the COP@TiO2 Assembliesa
structure (c1 Ti; blue, O: red) without (c2) and with the white noise
inclusion (c3).

1a1, the r-TiO2 photocatalyst consists of well-developed single


crystal nanorods of high aspect ratio ∼100 nm × ∼10 nm, and
the estimated thickness of 10−15 nm. The [110] orientation of
the nanorod was determined by the analysis of the
corresponding electron diffraction pattern (Figure 1a2). The
observed pattern is fully consistent with the simulated
diffraction for the P42/mnm structure using both the kine-
matical (green dots) and dynamical (red dots) approaches,
confirming definitely the rutile phase identity and the single-
crystal nature of the synthesized nanorods at the nanoscale
(Figure 1a3). The high-resolution TEM image of the r-TiO2 a
Class A: HOMO → LUMO → CB (left); class B: HOMO → CB
nanorod oriented along the [110] zone axis (obtained within (right).

5445 DOI: 10.1021/acs.jpcc.5b10983


J. Phys. Chem. C 2016, 120, 5442−5456
The Journal of Physical Chemistry C Article

of the dye-to-TiO2 charge-transfer bands. As a result, the nature the two hydroxyl groups (Figure 2a). The catechol oxygen
of the anchoring group as well as the energy positioning, shape, atoms are then connected to two adjacent surface Ti5c ions with
and symmetry of the HOMO and LUMO orbitals of the the Ti5c−OCAT bond length of 1.90 Å, giving rise to a local C2v
organic molecules to be photodecomposed play an important symmetry of the adsorption complex. The released protons are
role in their efficient self-sensitization. As discussed else- attached to the adjacent bridging O2c(a) atoms with the H−
where,58 an anchor that connects the admolecule to the catalyst O2c(a) bond length equal to 0.97 Å, without any significant
surface should exhibit a weak coupling of HOMO to the titania interaction with the catechol admolecule. The main axis of the
CB manifold to prevent the charge recombination, yet strong molecule is parallel to the surface normal [110] direction (α =
coupling of LUMO to foster the photoelectron injection rate. 0°, Figure 2a), whereas the short axis is oriented along the
Such a situation is preferred when LUMO is extensively mixed [001] direction (β = 0°, Figure 2a1).
with CB, whereas the HOMO orbital with possibly negligible Salicylic Acid Adsorption. The highest adsorption energy
orbital density on the linking groups is left unchanged (class A). (Eads = −2.63 eV with the dispersion part of −0.30 eV) for the
On the other hand, a strong interaction of HOMO with CB SAL admolecule is related with the bidentate bridging
gives rise to efficient sensitization of the titania photocatalyst attachment via the −COO− anchor with the hydroxyl group
owing to the development of a pronounced charge-transfer remaining intact (Figure 2b). The both carboxyl oxygen atoms
band in the visible region (class B). are ligated to two adjacent surface Ti5c ions with the Ti5c−OSAL
For a preliminary insight into electronic and charge-transfer bond length of 2.07 Å. The released proton is stabilized on the
properties of the investigated COP molecules, we explored nearby bridging O2c(a) atom (dH−O2c = 0.95 Å), with only a
briefly their molecular structure in the parent states. The energy negligible interaction with the SAL admolecule. Such bidentate
level manifolds and the shapes of the frontier orbitals for the bridging geometry of a local Cs symmetry has already been
catechol, salicylic, phthalic, and terephthalic molecules are reported for r-TiO2(110) surface,60−62 and is typical for
shown in Figure S4, whereas their ionization potentials (IP), adsorbed monocarboxylic acids, such as formic, acetic, or
electron affinities (EA), and the HOMO−LUMO energy gaps benzoic ones. The orientation of the SAL molecule with respect
(Eg) are collated in Table S4 in SI. The potential relevance of to the substrate, defined by the α = 0° and β = 53.5° angles
these parameters for sensitization of titania is also discussed (Figure 2b1), corresponds to a rotated vertical conformation,
therein. In the next step, we calculated electronic structure of which results from the trade-off between the considerable
the COP@TiO2 assemblies in a distal and a proximal attraction, involving the −OH group and the bridging oxygen,
configuration. For DFT calculations, it was assumed that a and the steric repulsion of the aromatic ring with the closed
single organic admolecule occupies the r-TiO2(110)(4 × 2) shell surface O2c2− ions.
slab, which corresponds to the coverage of 0.61 admolecule/ Phthalic Acid Adsorption. In the case of the PTA
nm2, in line with the experimental estimate of the surface COP admolecule a tilted binding mode (α = 24.5°) involving both
loading (cf. SI, Table S3). For such coverage, the periodically carboxyl functionalities was found to give rise to the most stable
repeated molecules are mutually separated by at least 7.5 Å for Cs-geometry with dTi−O equal to 1.91 Å (Figures 2c,c1). The
the top-on CAT, SAL, PTA adspecies, and by 6.1 Å in the case two protons released upon the attachment produce surface
of the plane-on adsorbed TPA molecules. In such arrangements hydroxyls with the bond lengths equal to dH−O = 0.95 Å. They
the calculated dispersion energies of the interaction of a given are involved in quite strong hydrogen bonds with the adjacent
admolecule with its four nearest neighbors were equal to 0.05, carboxylic oxygen atoms (dO···HO = 1.45 Å). The adsorption
0.04, 0.06, and 0.02 eV only, for CAT, SAL, PTA, and TPA, energy for this conformation is equal to Eads = −2.99 eV with
respectively. Following previous literature,59 all calculations the dispersion part of −0.46 eV.
were performed within an undefected titania surface model. Terephthalic Acid Adsorption. Contrary to the previous
Several starting adsorption geometries of the organic molecules admolecules, in the case of TPA the highest adsorption energy
placed on the TiO2(110) slab were explored. They include a (Eads = −3.92 eV) was found to be associated with the plane-on
plane-on (horizontal) and top-on (vertical) attachment, a geometry (α = 90°) shown in Figure 2d. The binding occurs
mono- or bidentate bonding to the surface, as well as several now through the carbonyl oxygen atoms of both carboxylic
dissociative and nondissociative adsorption modes (see SI, groups and the surface Ti5c ions, resulting in a local C2
Table S5 and Figure S5 for more details). Only those with the symmetry (Figure 2d1). The dispersion energy of −0.72 eV is
highest adsorption energies were subjected to further analysis, roughly twice bigger than for the vertically adsorbed species
and the obtained results are summarized in Table 1 and in (Table 1), which may be associated with the larger interaction
Figure 2. area between the substrate and the flat-lying admolecule. The
Catechol Adsorption. The most stable adsorption mode of binding entails rotation of the carboxylic group by ∼80° along
this molecule (Eads = −2.90 eV with the dispersion part equal to the Ph−COOH axis, as it can be inferred from Figure 2d, and
−0.43 eV) is a top-on conformation entailing deprotonation of the resultant Ti5c−OTPA bond length is equal to 1.97 Å. The
discharged protons are accommodated on the nearest bridging
Table 1. Principal Adsorption Parameters of the COP oxygen atoms (O2c(a)), giving rise to two surface hydroxyls with
Admolecules dH−O2c(a) = 0.99 Å and the OTPA···HO2c(a) hydrogen bond
length of 2.15 Å. The long axis of the TPA molecule is rotated
Eads/eV by the angle of 19.1° with respect to the [001] direction,
delineated by the rows of the bridging oxygen anions (Figure
adsorption mode total dispersion dO−Ti α, β
2d1). Such adsorption mode allows the carboxylic groups of the
CAT top-on −2.90 −0.43 1.90 0°, 0° TPA admolecule to better interact with the surface Ti5c ions.
SAL top-on −2.63 −0.30 2.07 0°, 53.5° For all adsorbed organic molecules, no significant mod-
PTA top-on −2.99 −0.46 1.91 24.5°, 5.1° ification of the aromatic ring was observed. The only relevant
TPA plane-on −3.92 −0.72 1.97 90°, 19.1° changes include the already mentioned rotation of the
5446 DOI: 10.1021/acs.jpcc.5b10983
J. Phys. Chem. C 2016, 120, 5442−5456
The Journal of Physical Chemistry C Article

Figure 2. Most stable adsorption geometries of COP molecules on the r-TiO2(110) surface. Perspective views (a-d) of the topmost trilayers of
titanium dioxide and adsorbed molecules, together with top views (a1-d1). The surface atoms are colored as follows: oxygen−red, titanium−light
gray, hydrogen−yellow, carbon−black, bridging oxygen ions of the substrate−violet, ligand oxygen ions attached to the surface−green.

Table 2. Bond Orders, Bader Charges, and Surface Dipole Values for COP Molecules Adsorbed on the (110) Facet of r-TiO2
CAT SAL PTA TPA
total bond order 0.57 0.54 0.41 0.63
qB/|e| 0.50 0.37 0.31 0.20
surface dipole moment μ/Debye (μx, μy, μz), 18.6 (0.53, 0.12, 18.6) 13.8 (−0.97, −1.35, 13.6) 22.3 (0.87, −8.64, 20.6) 8.18 (−0.28, 0.07, 8.18)

carboxylic groups and a slight reduction of the C−O bond from the surface. The charge transfer and the surface dipole
length (by ∼2%), resulting from deprotonation. Relaxation of generate local fields that give rise to the shift in the band edge
the rutile surface upon the adsorption was found to also be positions of the titania substrate (vide inf ra). The dipole
rather small. Titanium ions involved in the binding moved component of the COP admolecules normal to the surface
slightly toward the admolecule by about ∼0.01 Å, whereas the gives rise to a shift in the titania CB.
positions of all other atoms remained essentially intact. For Electronic Structure of the Functionalized Rutile (110)
each investigated COP@TiO2 assembly the total bond order Surface. For an in-depth electronic level insight into the
(between the organic molecules and the rutile (110) surface),63 influence of the COP molecules on self-sensitization of the
the Bader charges (qB) on the admolecules produced upon the rutile nanorods, we discussed the energy level alignment and
attachment, and the resultant surface dipole values were the molecular coupling at the COP@r-TiO2 interface by
calculated. calculating the density of states (DOS) for distal and proximal
The results collected in Table 2 show that the bidentate contact configurations (Figures 3−6). These results were used
attachment via two hydroxyl functionalities (CAT) or via the to discuss possible implications of the different magnitude of
bidentate ligation through the carboxyl group (SAL) leads to a the electronic coupling between the pollutant molecules and
comparable bond orders (0.57 vs 0.54), whereas in the case of the TiO2 substrate in controlling the efficiency of forward and
COP molecules containing two carboxyls the bond order back electron-transfer processes. Additionally, the charge
obviously strongly depends on their mutual position in the density contours of the CB and VB edges and those related
phenyl ring. For the ortho-position (PTA) and the top-on to the relevant orbitals of the COP species (HOMO, LUMO)
attachment the bond order is much smaller (0.41) than for the were also shown. For a better clarity, the corresponding DOS
para-arrangement (TPA) and the plane-on ligation (0.63). The diagrams for the bulk and bare surface of rutile (Figure S7) are
Bader charge acquired by the adsorbed organic molecules briefly discussed in the SI. The calculated bulk band gap of 2.4
decreases in the sequence CAT > SAL > PTA > TPA, and is eV at Γ-point for r-TiO2, being very close to previous
well correlated with the EHOMO(COP) − EF(TiO2) energetic theoretical predictions (2.4 eV at GGA+U level),64 is yet
distance (SI, Figure S6). It shows that an extensive charge lower than the experimental one (3.01 eV),65,66 which is typical
rearrangement accompanies the COP-titania surface inter- for DFT calculations. Indeed, it is well-known that insufficient
action. The extent of the charge transfer is essentially controlled cancellation of self-interaction correction leads to under-
by the nature of the binding group, and is favored by the estimation of the energy gaps, as discussed elsewhere in more
hydroxyl functionalities. The resultant surface dipole is the detail.64 The (110) surface band gap of 1.1 eV is much smaller
highest for PTA, since the angular top-on configuration moves than the bulk value, which stems from a number of in-gap states
the carboxylic oxygen atoms not involved in the binding back due to the broken Ti−O bonds. It is similar to the previously
5447 DOI: 10.1021/acs.jpcc.5b10983
J. Phys. Chem. C 2016, 120, 5442−5456
The Journal of Physical Chemistry C Article

reported value of 1.57 eV obtained with cluster model of the molecular orbitals with a contribution to DOS from the organic
(110) rutile termination.67 The band edge energy levels of the admolecule larger than 3%. These are states located around −3
rutile support and the HOMO−LUMO energy levels of the eV (b1), −1 eV (b3), and 1 eV (b4). Both frontier orbitals of the
COP admolecules for all investigated samples are collected in CAT molecule were shifted upward in an uneven way (HOMO
Table 3. by 0.3 eV and LUMO by 0.6 eV). Because in the bound state
these orbitals formally are no longer frontier orbitals of the
Table 3. Band Edge Energy Levels (VBM and CBM) for COP molecule, in order to emphasize their derivation they are
Bare and COP Covered (110) Surface of r-TiO2, and the labeled “HOMO” and “LUMO” hereafter. Moreover, in the
Frontier Energy Levels of COP Admolecules (COPHOMO and middle of the conduction band, a new state with an appreciable
COPLUMO)a contribution of the catechol ligand appeared. The electron
density contours corresponding to the “HOMO” (b1) and
bare CAT@ SAL@ PTA@ TPA@ “LUMO” (b4) orbitals of the bare CAT are still strongly
eV TiO2 TiO2 TiO2 TiO2 TiO2
localized on the ligand and are very similar to their parent shape
VBM −3.17 −4.17 −4.38 −4.36 −4.76 at distal arrangement (Figure 3a). The electron density
CBM −2.01 −3.31 −3.46 −3.48 −3.63 contours at the very bottom of the conduction band consists
EFermi −2.85 −3.36 −3.82 −4.26 −4.38 essentially of titanium 3d states of the t2g manifold (b2),
COPHOMO − −3.43 −3.99 −4.46 −4.45 whereas the top of the valence band is dominated by the
COPLUMO − 0.92 −0.42 0.41 0.18 catechol “HOMO−1”. A strong π-type bonding between the
a
All values correspond to the Γ point. CAT admolecule and the substrate is clearly visible by
accumulation of the charge density between the catechol
CAT@r-TiO2. In Figure 3a and 3b, the partial density of states oxygen and the surface titanium atoms with the resultant bond
(pDOS) diagrams are presented, showing alignment of the order equal to 0.57.
adsorbed catechol energy levels with respect to the band The calculated sizable charge transfer between the
structure of the TiO2 (110) surface for distal and proximal admolecule and the substrate is equal to 0.50 |e|. It is
arrangements, respectively. In both diagrams, the green and the responsible for the appearance of a marked absorption band the
red lines denote the density of states for the rutile oxygen and in the UV−vis spectra at 450 nm (vide inf ra). On the other
titanium ions, respectively, whereas the navy blue contours hand, the “LUMO” states are significantly coupled to the
indicate the pDOS contribution of the catechol admolecule. continuum of the conduction band d-states, which shall favor
The gradient gray bars epitomize the width of the conduction electron injection form the photoexcited COP admolecule to
and valence bands of the TiO2 substrate. The positioning of the deep of the rutile substrate. However, a low density of titania
HOMO/LUMO energy levels of the catechol molecule with acceptor states in this region is rather inauspicious for the
respect to the edges of the rutile surface conduction and the efficient interfacial electron transfer (vide inf ra).
valence bands before attachment (Figure 3a), reveals a SAL@r-TiO2. In the case of the r-TiO2(110) surface
staggered (type II) band alignment. The HOMO level (a1) is functionalized with salicylic acid, the type II energy-level
situated in the band gap just below CB (0.1 eV), the HOMO− alignment (Figure 4a and 4b) is quite similar to that of the
1 at the top of VB, whereas the LUMO (a2) is placed ∼0.4 eV catechol system. In a distal configuration, the strongly localized
above the conduction band. Upon attachment, the type II highest occupied band (a1 in Figure 4a), distanced from the CB
alignment is essentially preserved. Considering the energetic minimum by 0.25 eV, has mainly the HOMO character of the
window of interest defined by the band gap and the conduction parent salicylic acid (cf. Figure S6b, SI). The band
band regions, following the literature, we report only the corresponding to the HOMO−1 orbital of the admolecule is

Figure 3. Partial DOS of the r-TiO2(110) surface slab model calculated for the catechol molecule at distal (a) and proximal (b) arrangements
showing the corresponding energy alignments. (a1) and (a2) are the charge density contours of the HOMO and LUMO orbitals of the catechol
molecule in the distal configuration, whereas (b1−b4) refer to the charge density contours corresponding to the most important states of the
proximal arrangement: (b1) “HOMO” of the admolecule, (b2) bottom of the titania CB, (b3) a CB state with an appreciable (>3%) contribution of
catechol, (b4) “LUMO” of the admolecule. On the DOS plots, the green and red lines refer to titanium and oxygen of TiO2 states, respectively,
whereas the blue color denotes the molecule contribution. Partial charge density contours are color coded to reveal if they are occupied (blue) or
empty (yellow).

5448 DOI: 10.1021/acs.jpcc.5b10983


J. Phys. Chem. C 2016, 120, 5442−5456
The Journal of Physical Chemistry C Article

Figure 4. Partial DOS of the r-TiO2(110) surface slab model calculated for the salicylic acid molecule at distal (a) and proximal (b) arrangements
showing the corresponding energy alignments. (a1) and (a2) are the charge density contours of the HOMO and LUMO orbitals of the salicylic acid
molecule in the distal configuration, whereas (b1−b4) refer to the charge density contours corresponding to the most important states of the
proximal arrangement: (b1) “HOMO” of the admolecule, (b2) bottom of the titania CB, (b3) “LUMO” of the admolecule. On the DOS plots, the
green and red lines refer to titanium and oxygen of TiO2 states, respectively, whereas the blue color denotes the molecule contribution. Partial charge
density contours are color coded to reveal if they are occupied (blue) or empty (yellow).

Figure 5. Partial DOS of the r-TiO2(110) surface slab model calculated for the phthalic acid molecule at distal (a) and proximal (b) arrangements
showing the corresponding energy alignments. (a1) and (a2) are the charge density contours of the HOMO and LUMO orbitals of the phthalic acid
molecule in the distal configuration, whereas (b1−b4) refer to the charge density contours corresponding to the most important states of the
proximal arrangement: (b1) “HOMO” of the admolecule, (b2) bottom of the titania CB, (b3) a CB state with an appreciable (>3%) contribution of
phthalic acid, (b4) “LUMO” of the admolecule. On the DOS plots, the green and red lines refer to titanium and oxygen of TiO2 states, respectively,
whereas the blue color denotes the molecule contribution. Partial charge density contours are color coded to reveal if they are occupied (blue) or
empty (yellow).

positioned 0.2 eV above the VB edge, whereas the LUMO state the oxidized dye can be expected, in comparison to deeply
(a2) is situated on the top of the conduction band. Binding of localized bulk states.69 On the other hand, localization of the
the SAL molecule moves the “HOMO” level (b1) by 0.2 eV “HOMO” orbital in the band gap, just below the CB edge,
below the CB minimum (Figure 4b) and shifts the LUMO facilitates significant charge transfer between the SAL
orbital (b3) down by 0.15 eV. As expected, the bottom of the admolecule and the rutile substrate, revealed by the appearance
CB is composed mainly of Ti d states (b2), whereas due to of a sizable light absorption in the visible region (Figure 8, vide
stabilization of the HOMO−1 state upon ligation, the VB inf ra). The calculated partial charge on the SAL adspecies equal
maximum is dominated by the oxygen 2p states of rutile. The to 0.37 |e| and the total bond order of 0.54 are in line with such
mixing of “LUMO” and “HOMO” with the substrate states is significant charge transfer.
rather small, implying a weak coupling. Although it hinders PTA@r-TiO2. The electronic structure of the phthalic acid
undesirable charge recombination with a hole center localized adsorbed on the r-TiO2(110) in a distal setting with HOMO
on “HOMO”, such situation is less favorable for photoelectron (a1) and LUMO (a2) embracing the titania conduction band
injection via “LUMO” channel, suggesting a nonadiabatic (Figure 5a) resembles those previously found for the CAT or
mechanism of the interfacial electron transfer. Following de SAL molecules. Yet, in contrast to CAT and SAL, upon the
Angelis,68 it can be therefore conceived that the photoelectron ligation the HOMO state (b1) is significantly stabilized (by 0.9
injection should correspond to surface CB states localized close eV) and placed on the top of VB (0.05 eV below the edge),
to “LUMO”, for which a faster undesirable recombination with whereas there are no PTA related states at the bottom of CB
5449 DOI: 10.1021/acs.jpcc.5b10983
J. Phys. Chem. C 2016, 120, 5442−5456
The Journal of Physical Chemistry C Article

Figure 6. Partial DOS of the r-TiO2(110) surface slab model calculated for the terephthalic acid molecule at distal (a) and proximal (b)
arrangements showing the corresponding energy alignments. (a1) and (a2) are the charge density contours of the HOMO and LUMO orbitals of the
phthalic acid molecule in the distal configuration, whereas (b1−b4) refer to the charge density contours corresponding to the most important states
of the proximal arrangement: (b1) “HOMO” of the admolecule, (b2) bottom of the titania CB, (b3) a CB state with an appreciable (>3%)
contribution of terephthalic acid, (b4) “LUMO” of the admolecule. On the DOS plots, the green and red lines refer to titanium and oxygen of TiO2
states, respectively, whereas the blue color denotes the molecule contribution. Partial charge density contours are color coded to reveal if they are
occupied (blue) or empty (yellow).

(b2). An extended pDOS features that can be traced back to anchoring groups of the COP admolecules has more influences
“LUMO” (b4) appeared in the upper part of the conduction on the relative position of the “HOMO” than on the “LUMO”
band, together with localized state with a sizable PTA states, with respect to the CB and VB edges. The “HOMO”
contribution (b3) placed lower by 0.45 eV with respect to b4. position can be systematically shifted from the CB to VB edge
As a result, there are no new localized states in the band gap. through replacement of the −OH anchors by the −COOH
The charge transfer and the total bond order of the PTA-TiO2 ones.
assembly are equal to 0.31 |e| and 0.41, respectively. Charge Transfer. The calculated DOS structures for the
TPA@r-TiO2. The DOS structure of the r-TiO2 covered by COP@r-TiO2 photocatalysts can be used to assess the
the plane-on adsorbed TPA molecules in distal placement is efficiency of possible electron transfer along the A and B
presented in Figure 6a. The HOMO state is located on the top pathways. In order to favor a fast injection of the photoelectron
of VB (a1), and the LUMO above the middle of the conduction along the two-step pathway A, the COP donor level should be
band (a2). There are also the same localized states with TPA well-immersed in the semiconductor manifold of the
contribution at the upper part of CB. The binding of TPA does unoccupied CB acceptor states, preferably in the region of
not influence the position of “HOMO” appreciably (Figure 6b), the high density of states.70,71 Following the literature, two
and its shape reveals rather weak coupling with the titania VB properties related to the molecular structure of the COP
orbitals (b1). Together with the titanium t2g states (b2), they admolecules, namely, the electron injection energy, ELUMO, and
constitute the VB and CB edges. The TPA related “LUMO” the percentage of “LUMO” localized on the anchoring moiety,
state, b4, is shifted to the upper part of the conduction band. are crucial.67,68 For all investigated molecules the high
There are also empty states with significant contribution of “LUMO”-CB gap is adequate for a pronounced thermodynamic
TPA, located in the upper part of the CB (exemplified by b3). driving force for the fast electron transfer. The dependence of
The electron density repartition shows that they are quite well the injection efficiency on the “LUMO” energy with respect to
coupled with the corresponding acceptor levels of the CB can be accounted for within the Newns−Anderson
conduction band. The side-on attachment results in a model.52,53 The center of the projected rDOS distribution
considerably smaller charge transfer (0.20 |e|) as a result of corresponds to the energy of the effective “LUMO” of the COP
the multiplicity of the TPA-TiO2 bonds in the largest bond adspecies, which epitomizes the excited electron injection state,
order (0.63), in comparison to other examined COP molecules was evaluated as the position of the fitted Lorentzian peak
(Table 3). (Figure 7). Within this model, the “LUMO” broadening, Γ,
In summary, the results of electronic structure analysis gives an estimation of the electron transfer: τ = 1/Γ. The
showed that the “HOMO” levels are shifted to lower energies calculated values, equal to 2.0, 2.7, 2.9, and 3.6 fs and the peak
with the increasing −COOH to −OH ratio, changing thereby positions with respect to EF are 4.67, 4.86, 4.48, and 4.35 eV,
their positions dramatically from the level close to the CB for CAT, PTA, TPA, and SAL adspecies, respectively, show that
minimum to the top of VB. The surface band gap narrows in all COP molecules exhibit rather similar rates of the electron
the opposite direction, from 1.13 eV for TPA@r-TiO2 (nearly transfer. The only remarkable difference is a more favorable
the same as in bare TiO2) to 0.78 eV for CAT@r-TiO2. This immersion of the PTA “LUMO” in the conduction band of the
accounts quite well for the observed enhancement of the light rutile nanorods in comparison to the other COP admolecules.
absorption of the visible region (vide inf ra). The difference It may be thus expected that for all investigated COP
between the energies of the “HOMO” and the CB levels is molecules in their adsorbed state the “LUMO” energetic
lower for CAT@r-TiO2 than for SAL@r-TiO2, and it implies a position with respect to the CB manifold is favorable for a fast
bathochromic shift of the absorption onset observed with the photoelectron injection. There is also no substantial influence
decreasing −COOH to −OH ratio. The nature of the of the anchoring groups (−OH, −COOH) on the electron
5450 DOI: 10.1021/acs.jpcc.5b10983
J. Phys. Chem. C 2016, 120, 5442−5456
The Journal of Physical Chemistry C Article

Figure 8. Diffuse reflectance spectra of bare and modified rutile


nanorods together with the picture of the samples suspended in water
(inset).

role in the photosensitization of the rutile nanorods in the


visible region. The highest effect was observed in the presence
of two −OH groups (CAT), a medium one for the mixed
−COOH/−OH case (SAL), whereas for dicarboxylic mole-
cules (PTA and TPA), no effect was observed essentially.
Figure 7. Lorentzian fitting of the pDOS corresponding to the Spectroelectrochemical Measurements. Determination of
approximated “LUMO” states of the COP admolecules, showing their the redox properties of the COP@r-TiO2 photocatalysts was
position with respect to schematically outlined DOS profile of the performed according to the method developed recently.74 The
COP@TiO2 assembly. platinum electrode covered with the tested material was
subjected to a slow potential sweep (0.5 mV·s−1). The
simultaneously measured changes of the reflectance signal at
transfer rate. Furthermore, for all the COP species the VB is 780 nm, transformed into the Kubelka−Munk (K−M)
dominated by the “HOMO” orbital of the admolecule, which is function, enabled determination of the potential at which
beneficial for charge separation. As a result, we can expect some TiO2 was reduced (Figure 9). In this way, it is possible to probe
UV sensitization effect along the two-step pathway A, but its
overall efficiency should be reduced by low intensity of the
solar light in the energy window above 3 eV, where this process
is expected to take place. However, formation of the new donor
states in the band gap associated with the “HOMO” level in the
case of the CAT and SAL admolecules (Figure 3 and 4) opens
the route for a parallel visible light sensitization via a direct
injection of the photoelectron according to ligand (“HOMO”
of COP) to metal (CB of TiO2) electron transfer (one step
pathway B). The corresponding absorption bands were indeed
observed in UV−vis spectra (vide inf ra). Inspection of Figure
3b shows that for CAT species, a unique localization of
“HOMO” just below the CB and HOMO−1 above the VB
levels are especially beneficial for this photosensitization
channel.
UV−vis Absorption. In order to study the light absorption of
the functionalized r-TiO2 nanorods, the UV−vis spectra were Figure 9. Normalized spectral changes at 780 nm (represented by
measured (Figure 8). The bare nanorods show a significant changes of the Kubelka−Munk function) recorded upon the potential
absorption of UV light (λ < 400 nm), due to a direct VB to CB sweep from −0.6 to −1.5 V vs Ag/AgCl electrode.
excitation of TiO2. The CAT@r-TiO2 sample shows the highest
absorption of the visible light, extending from 400 nm to ca. the density of states of the photocatalyst, and the higher it is,
600 nm, and the dark orange color of this sample is the most the steeper slope of the K−M curve is observed. An increase of
intense (see inset in Figure 8). Absorption of visible light by the the light absorption below −1.0 V results from the reduction of
SAL@r-TiO2 sample (light yellow) is less pronounced as the the energy states localized at the bottom of the conduction
tail of the absorption extends to ca. 550 nm. All other samples band. They can essentially be identified with the position of the
remained practically white upon the functionalization. These b2 states in Figures 3−6, which are crucial for the driving the
results are in agreement with the calculated DOS structure and dioxygen reduction. For SAL@r-TiO2 and PTA@r-TiO2, the
with those reported previously.72,73 A brief inspection of the edge of the conduction band is shifted toward lower energies,
Figure 8 reveals that the −COOH/−OH ratio plays a crucial implying the lower efficiency of O2•− formation.75 CAT@r-
5451 DOI: 10.1021/acs.jpcc.5b10983
J. Phys. Chem. C 2016, 120, 5442−5456
The Journal of Physical Chemistry C Article

Figure 10. Photocurrent as a function of the photoelectrode potential (vs Ag/AgCl) and incident light wavelength recorded for photoelectrodes
covered with (a) bare r-TiO2, (b) CAT@r-TiO2, (c) SAL@r-TiO2, (d) PTA@r-TiO2, and (e) TPA@r-TiO2. The red and blue areas correspond to
the anodic and cathodic photocurrents, respectively, whereas the white areas represent a zero net photocurrent.

TiO2 and TPA@r-TiO2 samples exhibit the band edge position in the case of TPA@r-TiO2 is quite small (compare Figures 2
close to that of r-TiO2, making the photoelectron transfer and 9).
faster. Yet, these changes are not very significant in the case of Photocatalytic Tests. Photocatalytic degradation of colorless
rutile. The appearance of the shoulder for CAT@r-TiO2 at organic pollutants was tested by monitoring the total organic
−1.2 V corresponds to an empty acceptor state separated from carbon (TOC) content in the COP@TiO2 assembly irradiated
the conduction band by strong interaction with catechol ligand. with UV−vis light (full light of XBO lamp). The initial
Photocurrent Measurements. The UV−vis-induced activity decomposition rate constants were 0.024, 0.013, 0.005, and
of the samples was examined by photoelectrochemical 0.004 min−1 for CAT, SAL, PTA, and TPA, respectively, as
measurements. The results shown in Figures 10a−e reveal derived from the degradation curves shown in Figure 11. Two
significant differences among the investigated samples in the classes of the pollutant molecules can be distinguished here,
generated photocurrent as the function of the applied potential depending on the photodegradation rate constant. Molecules
and the wavelength of the incident light. A pronounced containing hydroxyl anchoring group (CAT, SAL) degrade
photocurrent induced by the ultraviolet light is observed for all much faster than those with the carboxyl attachment only
materials. In the case of r-TiO2 modified with organic
molecules a cathodic to anodic photocurrent switching appears
at various potential range. Photocurrents generated by the
incident visible light are higher for the electrodes covered with
modified r-TiO2 compared to that of bare r-TiO2. The most
pronounced sensitization effect was observed for CAT@r-TiO2
(Figure 10b); however, the visible-light-induced photoactivity
of SAL@r-TiO2 (Figure 10c) is also significant when compared
to its photoactivity induced by light of 400 nm.
This observation remains in a good agreement with the
results of the DFT calculations. Because the “HOMO” orbitals
of the catechol and salicylic acids are situated within the
bandgap of r-TiO2, these modifiers act as the photosensitizers
of r-TiO2. The photocurrent map recorded for TPA@r-TiO2
material (Figure 10e) resembles that of r-TiO2. Although the
“HOMO” of TPA is also localized within the bandgap, the
experimental coverage of this modifier at the (110) facet is
significantly lower than the adsorption of other molecules (vide
supra), and therefore, the influence of terephthalic acid on
photoelectrochemical properties of r-TiO2 is less significant.
Lower values of measured photocurrents are very likely the
consequence of a higher hydrophobicity of TPA@r-TiO2
material, which stems from the unique plane-on adsorption.
In this mode, both polar groups are anchored with the surface,
whereas the aromatic moiety is exposed to the solution. Also
the charge density flow between the organic molecule and the Figure 11. Self-decomposition of COP substrates: relative changes of
r-TiO2 crystal (ΔqB = 0.20 |e|) is less efficient than in the case total organic carbon content as the function of irradiation time
of other molecules, and therefore, the photosensitization effect (irradiation: full light of xenon lamp).

5452 DOI: 10.1021/acs.jpcc.5b10983


J. Phys. Chem. C 2016, 120, 5442−5456
The Journal of Physical Chemistry C Article

(PTA, TPA). These results show also that the photo- formed, similar to that of catechol that was found to be
sensitization of the rutile nanorod catalyst via the direct beneficial for the visible photosensitization of the system.
pathway B is much more efficient than the two step process Upon the prolonged irradiation, the photodegradation rates
that involves the pathway A. It should be noted that the become comparable (Figure 11), because the sensitization
influence of the sensitization on the photocatalytic decom- effect ceases due to gradual deterioration of the COP
position rate is most pronounced at the beginning (early stage) admolecules into smaller fragments absorbing UV only and
of the degradation process when all pollutant molecules are not finally to CO2 and H2O end-products. To the best of our
fragmented significantly yet and thus can act as the self- knowledge, the nature of such photocatalytic self-degradation in
photosensitizers. the case of the colorless organic compounds has not been
In order to elucidate the fate of the COP adspecies upon elucidated until now using jointly experimental and theoretical
irradiation, we investigated photoinduced chemical changes in approaches.
the SAL@r-TiO2 sample (taken as an example) by in situ DRS
measurements. The resultant differential spectra are presented
in Figure 12. As a reference, we added the differential spectra of
■ CONCLUSIONS
By combining several photoelectrochemical and spectroelec-
trochemical techniques with the DFT molecular modeling we
probed the electronic nature of self-sensitization of rutile
nanorods by small colorless organic pollutants, providing a
comprehensive background for mechanistic understanding of
their photodegradation. We selected disubstituted benzene
derivatives with −OH and −COOH groups (i.e., catechol,
salicylic acid, phthalic acid, and terephthalic acid) as model
colorless pollutants. The role of the anchoring groups (−OH
and −COOH) was explained in terms of one-step (HOMO →
CB) and two-step (HOMO → LUMO → CB) photo-
sensitization mechanisms. It was shown that the attachment
by hydroxyl groups triggers the one-step pathway via a direct
ligand to metal charge transfer (CAT, SAL). The presence of
carboxyl anchors, in turn, favors the two step sensitization
involving intramolecular excitation followed by the photo-
electron injection from LUMO to CB. The carboxyl anchors
locate the LUMO state at the top of the conduction band
making the two step pathway less efficient due to the low
Figure 12. Differential UV−vis spectra of SAL@r-TiO2 in the solid overlap with the solar spectrum (far UV), despite the fast
state recorded for different photoreaction times (up to 60 min) (a) electron transfer (2−4 fs). The efficiency of self-sensitized
together with reference differential spectra of r-TiO2 functionalized degradation of disubstituted benzene derivatives decreases in
with various organic molecules containing two hydroxyl groups: 2,3- the order: (−OH, −OH) > (−OH, −COOH) > (−COOH, −
dihydroxybenzoic acid (2,3-DHB), 2,5-dihydroxybenzoic acid (2,5- COOH), in an excellent agreement with the theoretical and
DHB), and catechol (CAT) with the subtracted SAL@r-TiO2 (for the experimental predictions. As the result the molecular under-
sake of a direct comparison) (b).
standing of photocatalytic degradation of small pollutants
showing a crucial role of the anchoring and photogenerated
hydroxyl groups in the self-sensitization process was accounted
r-TiO2 functionalized with various organic molecules contain- for with an unprecedented comprehensive way. It was shown
ing two hydroxyl groups, such as 2,3-dihydroxybenzoic acid that at early stages of the photocatalytic degradation process the
(2,3-DHB), 2,5-dihydroxybenzoic acid (2,5-DHB), and cat- aromatic rings of the COP moieties are readily photo-
echol (CAT) with the subtracted SAL@r-TiO2 (for the sake of hydroxylated, fostering the visible light utilization. Such
a direct comparison). Gradual development of a broad autocatalytic hydroxylation processes can also be relevant for
pronounced band in the range of 400−800 nm, characteristic photocatalytic degradation of those pollutants that originally do
for dihydroxybenzene derivatives, reveals a progressive not exhibit hydroxyl functionalities.


insertion of the photogenerated hydroxyl groups into the
aromatic ring of the SAL admolecules during the photocatalytic ASSOCIATED CONTENT
reaction. As it was discussed above, this explains the observed
bathochromic shift of the absorption with the increasing −OH/
*
S Supporting Information

−COOH ratio. As a result, the “HOMO” position is shifted The Supporting Information is available free of charge on the
toward the conduction band edge, which is beneficial for more ACS Publications website at DOI: 10.1021/acs.jpcc.5b10983.
effective utilization of the visible range of the solar light, via TEM image simulation conditions; relaxation of the bare
ligand (“HOMO” of COP) to metal (CB of TiO2) electron (110) surface of rutile TiO2; XRD, Raman, and FTIR
transfer. This observation reveals also an autocatalytic stage of spectra for unmodified and modified r-TiO2 nanorods;
COP self-degradation that results from specific tuning of the calculations of pollutants coverage; the details of the
band alignment between the photohydroxylated COP and the electronic structure for the set of gas phase COP
rutile moieties of the photocatalyst. It promotes localization of molecules; starting geometries for COP deposition;
the hole on the organic component facilitating immediate Bader charge (qB) vs the EHOMO(COP) − EF(TiO2)
reaction with water, [COP−H]•+ + OH− → COP−OH + H•. plot; DOS diagrams for bulk and bare surface titania
As a result, a doubly hydroxylated aromatic ring system is (PDF)
5453 DOI: 10.1021/acs.jpcc.5b10983
J. Phys. Chem. C 2016, 120, 5442−5456
The Journal of Physical Chemistry C Article

■ AUTHOR INFORMATION
Corresponding Authors
(9) Czoska, A.; Livraghi, S.; Chiesa, M.; Giamello, E.; Agnoli, S.;
Granozzi, G.; Finazzi, E.; Valentin, C. D.; Pacchioni, G. The Nature of
Defects in Fluorine-Doped TiO2. J. Phys. Chem. C 2008, 112, 8951−
*E-mail: zasada@chemia.uj.edu.pl. Tel.: +48 12 663 20 73. 8956.
*E-mail: macyk@chemia.uj.edu.pl. Tel.: +48 12 663 22 22. (10) Vijayan, B. K.; Dimitrijevic, N. M.; Finkelstein-Shapiro, D.; Wu,
Author Contributions J.; Gray, K. A. Coupling Titania Nanotubes and Carbon Nanotubes to
The manuscript was written through contributions of all Create Photocatalytic Nanocomposites. ACS Catal. 2012, 2, 223−229.
(11) Neubert, S.; Ramakrishnan, A.; Strunk, J.; Shi, H.; Mei, B.;
authors. All authors have given approval to the final version of Wang, L.; Bledowski, M.; Guschin, D. A.; Kauer, M.; Wang, Y.;
the manuscript. Muhler, M.; Beranek, R. Surface-Modified TiO2 Photocatalysts
Notes Prepared by a Photosynthetic Route: Mechanism, Enhancement, and
The authors declare no competing financial interest. Limits. ChemPlusChem 2014, 79, 163−170.


(12) Conesa, J. C. Modeling with Hybrid Density Functional Theory
ACKNOWLEDGMENTS the Electronic Band Alignment at the Zinc Oxide−Anatase Interface. J.
Phys. Chem. C 2012, 116, 18884−18890.
The support from the Foundation for Polish Science within the (13) Hamraoui, K.; Cristol, S.; Payen, E.; Paul, J.-F. Computational
TEAM/2012-9/4 project, cofinanced by the EU European Investigation of TiO2-Supported Isolated Oxomolybdenum Species. J.
Regional Development Fund, is highly acknowledged. A part of Phys. Chem. C 2007, 111, 3963−3972.
the work (photocatalytic tests) was financed by the National (14) Wang, Y.; Wang, Q.; Zhan, X.; Wang, F.; Safdar, M.; He, J.
Science Centre, Poland, within the grant number 2011/01/B/ Visible Light Driven Type II Heterostructures and Their Enhanced
ST5/00920. M.B. acknowledges the financial support by FP7 Photocatalysis Properties: A Review. Nanoscale 2013, 5, 8326−8339.
EU project 4G-PHOTOCAT (309636). The authors want to (15) Selinsky, R. S.; Ding, Q.; Faber, M. S.; Wright, J. C.; Jin, S.
dedicate this paper to professor Elio Giamello from University Quantum Dot Nanoscale Heterostructures for Solar Energy
Conversion. Chem. Soc. Rev. 2013, 42, 2963−2985.
of Torino to honor his 65th anniversary.


(16) Yin, Y.; Alivisatos, A. P. Colloidal Nanocrystal Synthesis and the
Organic−Inorganic Interface. Nature 2005, 437, 664−670.
ABBREVIATIONS (17) Macyk, W.; Szaciłowski, K.; Stochel, G.; Buchalska, M.;
BG, band gap; CAT, catechol; CB, conduction band; COP, Kuncewicz, J.; Łabuz, P. Titanium(IV) Complexes as Direct TiO2
colorless organic pollutants; DFT, density functionals theory; Photosensitizers. Coord. Chem. Rev. 2010, 254, 2687−2701.
(18) Calzolari, A.; Ruini, A.; Catellani, A. Surface Effects on
DFT+D, density functionals theory with inclusion of dispersion
Catechol/Semiconductor Interfaces. J. Phys. Chem. C 2012, 116,
forces; DFT+U, density functionals theory with Hubbard 17158−17163.
correction; EF, Fermi Energy; SI, Supporting Information; (19) Kim, S.; Choi, W. Visible-Light-Induced Photocatalytic
GGA, General Gradient Approximation; ITO, indium tin oxide; Degradation of 4-Chlorophenol and Phenolic Compounds in Aqueous
PAW, Plane Augmented Wave; PTA, phthalic acid; r-TiO2, Suspension of Pure Titania: Demonstrating the Existence of a Surface-
rutile phase of TiO2; SAL, salicylic acid; TEM, transmission Complex-Mediated Path. J. Phys. Chem. B 2005, 109, 5143−5149.
electron microscopy; TPA, terephthalic acid; VB, valence band; (20) Oprea, C. I.; Panait, P.; Gîrţu, M. A. DFT Study of Binding and
XC, eXchange-Correlation Electron Transfer from Colorless Aromatic Pollutants to a TiO2


Nanocluster: Application to Photocatalytic Degradation under Visible
REFERENCES Light Irradiation. Beilstein J. Nanotechnol. 2014, 5, 1016−1030.
(21) Agrios, A. G.; Gray, K. A.; Weitz, E. Photocatalytic
(1) Kisch, H. Semiconductor Photocatalysis: Principles and Applications; Transformation of 2,4,5-Trichlorophenol on TiO2 under Sub-Band-
John Wiley & Sons: Hoboken, NJ, 2015. Gap Illumination. Langmuir 2003, 19, 1402−1409.
(2) Shiraishi, Y.; Togawa, Y.; Tsukamoto, D.; Tanaka, S.; Hirai, T. (22) Sánchez-Sánchez, C.; Martínez, J. I.; Lanzilotto, V.; Méndez, J.;
Highly Efficient and Selective Hydrogenation of Nitroaromatics on Martín-Gago, J. A.; López, M. F. Antiphase Boundaries Accumulation
Photoactivated Rutile Titanium Dioxide. ACS Catal. 2012, 2, 2475−
Forming a New C60 Decoupled Crystallographic Phase on the Rutile
2481.
TiO2 (110)-(1 × 1) Surface. J. Phys. Chem. C 2014, 118, 27318−
(3) Schneider, J.; Matsuoka, M.; Takeuchi, M.; Zhang, J.; Horiuchi,
27324.
Y.; Anpo, M.; Bahnemann, D. W. Understanding TiO2 Photocatalysis:
(23) Voïtchovsky, K.; Ashari-Astani, N.; Tavernelli, I.; Tétreault, N.;
Mechanisms and Materials. Chem. Rev. 2014, 114, 9919−9986.
(4) Prieto-Mahaney, O.-O.; Murakami, N.; Abe, R.; Ohtani, B. Rothlisberger, U.; Stellacci, F.; Grätzel, M.; Harms, H. A. In Situ
Correlation between Photocatalytic Activities and Structural and Mapping of the Molecular Arrangement of Amphiphilic Dye
Physical Properties of Titanium(IV) Oxide Powders. Chem. Lett. 2009, Molecules at the TiO2 Surface of Dye-Sensitized Solar Cells. ACS
38, 238−239. Appl. Mater. Interfaces 2015, 7, 10834−10842.
(5) Bai, Y.; Mora-Seró, I.; De Angelis, F.; Bisquert, J.; Wang, P. (24) Muñoz-Batista, M. J.; Gómez-Cerezo, M. N.; Kubacka, A.;
Titanium Dioxide Nanomaterials for Photovoltaic Applications. Chem. Tudela, D.; Fernández-García, M. Role of Interface Contact in CeO2−
Rev. 2014, 114, 10095−10130. TiO2 Photocatalytic Composite Materials. ACS Catal. 2013, 4, 63−72.
(6) Ampelli, C.; Passalacqua, R.; Genovese, C.; Perathoner, S.; Centi, (25) Kim, K.; Barteau, M.; Farneth, W. Adsorption and
G.; Montini, T.; Gombac, V.; Jaen, J. J. D.; Fornasiero, P. H2 Decomposition of Aliphatic Alcohols on Titania. Langmuir 1988, 4,
Production by Selective Photo-Dehydrogenation of Ethanol in Gas 533−543.
and Liquid Phase on CuOx/TiO2 Nanocomposites. RSC Adv. 2013, 3, (26) Tunesi, S.; Anderson, M. Influence of Chemisorption on the
21776−21788. Photodecomposition of Salicylic Acid and Related Compounds Using
(7) Devaraji, P.; Sathu, N. K.; Gopinath, C. S. Ambient Oxidation of Suspended Titania Ceramic Membranes. J. Phys. Chem. 1991, 95,
Benzene to Phenol by Photocatalysis on Au/Ti0.98V0.02O2: Role of 3399−3405.
Holes. ACS Catal. 2014, 4, 2844−2853. (27) Moser, J.; Punchihewa, S.; Infelta, P. P.; Graetzel, M. Surface
(8) Di Valentin, C.; Finazzi, E.; Pacchioni, G.; Selloni, A.; Livraghi, S.; Complexation of Colloidal Semiconductors Strongly Enhances
Czoska, A.; Paganini, M.; Giamello, E. Density Functional Theory and Interfacial Electron-Transfer Rates. Langmuir 1991, 7, 3012−3018.
Electron Paramagnetic Resonance Study on the Effect of N-F (28) Rodriguez, R.; Blesa, M. A.; Regazzoni, A. E. Surface
Codoping of TiO2. Chem. Mater. 2008, 20, 3706−3714. Complexation at the TiO2 (Anatase)/Aqueous Solution Interface:

5454 DOI: 10.1021/acs.jpcc.5b10983


J. Phys. Chem. C 2016, 120, 5442−5456
The Journal of Physical Chemistry C Article

Chemisorption of Catechol. J. Colloid Interface Sci. 1996, 177, 122− (48) Monkhorst, H. J.; Pack, J. D. Special Points for Brillouin-Zone
131. Integrations. Phys. Rev. B 1976, 13, 5188.
(29) Li, S.-C.; Wang, J.-G.; Jacobson, P.; Gong, X.-Q.; Selloni, A.; (49) Methfessel, M.; Paxton, A. High-Precision Sampling for
Diebold, U. Correlation between Bonding Geometry and Band Gap Brillouin-Zone Integration in Metals. Phys. Rev. B: Condens. Matter
States at Organic−Inorganic Interfaces: Catechol on Rutile Tio2 (110). Mater. Phys. 1989, 40, 3616.
J. Am. Chem. Soc. 2009, 131, 980−984. (50) Murnaghan, F. The Compressibility of Media under Extreme
(30) Liu, L.-M.; Li, S.-C.; Cheng, H.; Diebold, U.; Selloni, A. Growth Pressures. Proc. Natl. Acad. Sci. U. S. A. 1944, 30, 244−247.
and Organization of an Organic Molecular Monolayer on TiO2: (51) Bader, R. F. A Quantum Theory of Molecular Structure and Its
Catechol on Anatase (101). J. Am. Chem. Soc. 2011, 133, 7816−7823. Applications. Chem. Rev. 1991, 91, 893−928.
(31) Fernández-Torre, D.; Carrasco, J.; Ganduglia-Pirovano, M. V.; (52) Muscat, J.; Newns, D. Chemisorption on Metals. Prog. Surf. Sci.
Pérez, R. Hydrogen Activation, Diffusion, and Clustering on CeO2 1978, 9, 1−43.
(111): A DFT+U Study. J. Chem. Phys. 2014, 141, 014703. (53) Persson, P.; Lundqvist, M. J.; Ernstorfer, R.; Goddard, W.;
(32) Li, H.; Ratcliff, E. L.; Sigdel, A. K.; Giordano, A. J.; Marder, S. Willig, F. Quantum Chemical Calculations of the Influence of Anchor-
R.; Berry, J. J.; Brédas, J. L. Modification of the Gallium-Doped Zinc Cum-Spacer Groups on Femtosecond Electron Transfer Times in
Oxide Surface with Self-Assembled Monolayers of Phosphonic Acids: Dye-Sensitized Semiconductor Nanocrystals. J. Chem. Theory Comput.
A Joint Theoretical and Experimental Study. Adv. Funct. Mater. 2014, 2006, 2, 441−451.
24, 3593−3603. (54) Lundqvist, M. J.; Nilsing, M.; Lunell, S.; Åkermark, B.; Persson,
(33) Duncan, W. R.; Prezhdo, O. V. Theoretical Studies of P. Spacer and Anchor Effects on the Electronic Coupling in
Photoinduced Electron Transfer in Dye-Sensitized TiO2. Annu. Rev. Ruthenium-Bis-Terpyridine Dye-Sensitized TiO2 Nanocrystals Studied
Phys. Chem. 2007, 58, 143−184. by DFT. J. Phys. Chem. B 2006, 110, 20513−20525.
(34) Godlewski, S.; Tekiel, A.; Piskorz, W.; Zasada, F.; Prauzner- (55) Wojdyr, M. Fityk: A General-Purpose Peak Fitting Program. J.
Bechcicki, J. S.; Sojka, Z.; Szymonski, M. Supramolecular Ordering of Appl. Crystallogr. 2010, 43, 1126−1128.
Ptcda Molecules: The Key Role of Dispersion Forces in an Unusual (56) Swope, R. J.; Smyth, J. R.; Larson, A. C. H in Rutile-Type
Transition from Physisorbed into Chemisorbed State. ACS Nano Compounds: I. Single-Crystal Neutron and X-Ray Diffraction Study of
2012, 6, 8536−8545. H in Rutile. Am. Mineral. 1995, 80, 448−453.
(35) Kolmer, M.; Ahmad Zebari, A. A.; Prauzner-Bechcicki, J. S.; (57) Tae, E. L.; Lee, S. H.; Lee, J. K.; Yoo, S. S.; Kang, E. J.; Yoon, K.
Piskorz, W.; Zasada, F.; Godlewski, S.; Such, B.; Sojka, Z.; Szymonski, B. A Strategy to Increase the Efficiency of the Dye-Sensitized TiO2
M. Polymerization of Polyanthrylene on a Titanium Dioxide (011)-(2 Solar Cells Operated by Photoexcitation of Dye-to-TiO2 Charge-
Transfer Bands. J. Phys. Chem. B 2005, 109, 22513−22522.
× 1) Surface. Angew. Chem. 2013, 125, 10490−10493.
(58) Maggio, E.; Martsinovich, N.; Troisi, A. Using Orbital Symmetry
(36) Calzolari, A.; Ruini, A.; Catellani, A. Anchor Group Versus
to Minimize Charge Recombination in Dye-Sensitized Solar Cells.
Conjugation: Toward the Gap-State Engineering of Functionalized
Angew. Chem., Int. Ed. 2013, 52, 973−975.
ZnO (1010) Surface for Optoelectronic Applications. J. Am. Chem. Soc.
(59) Lyubinetsky, I.; Yu, Z.; Henderson, M. A. Direct Observation of
2011, 133, 5893−5899.
Adsorption Evolution and Bonding Configuration of Tmaa on TiO2
(37) Thomas, A. G.; Jackman, M. J.; Wagstaffe, M.; Radtke, H.; Syres,
(110). J. Phys. Chem. C 2007, 111, 4342−4346.
K.; Adell, J.; Lévy, A.; Martsinovich, N. Adsorption Studies of P-
(60) Gutierrez-Sosa, A.; Martınez-Escolano, P.; Raza, H.; Lindsay, R.;
Aminobenzoic Acid on the Anatase TiO2 (101) Surface. Langmuir Wincott, P.; Thornton, G. Orientation of Carboxylates on TiO2 (110).
2014, 30, 12306−12314. Surf. Sci. 2001, 471, 163−169.
(38) Thomas, A. G.; Syres, K. L. Adsorption of Organic Molecules on (61) Chambers, S. A.; Thevuthasan, S.; Kim, Y.-J.; Herman, G. S.;
Rutile Tio2 and Anatase TiO2 Single Crystal Surfaces. Chem. Soc. Rev. Wang, Z.; Tober, E.; Ynzunza, R.; Morais, J.; Peden, C.; Ferris, K.;
2012, 41, 4207−4217. Fadley, C. S. Chemisorption Geometry of Formate on TiO2(110) by
(39) Stadelmann, P. JEMS−EMS Java Version; CIME-EPFL: Photoelectron Diffraction. Chem. Phys. Lett. 1997, 267, 51−57.
Lausanne, Switzerland, 2004. (62) Henderson, M. A.; White, J. M.; Uetsuka, H.; Onishi, H.
(40) Hafner, J. Ab-Initio Simulations of Materials Using Vasp: Photochemical Charge Transfer and Trapping at the Interface between
Density-Functional Theory and Beyond. J. Comput. Chem. 2008, 29, an Organic Adlayer and an Oxide Semiconductor. J. Am. Chem. Soc.
2044−2078. 2003, 125, 14974−14975.
(41) Kresse, G.; Joubert, D. From Ultrasoft Pseudopotentials to the (63) The actual values of the bond order are relatively small (like
Projector Augmented-Wave Method. Phys. Rev. B: Condens. Matter those of Mulliken bond orders for molecular systems) and should be
Mater. Phys. 1999, 59, 1758. compared to each other, rather than treated as an intuitive chemical
(42) Perdew, J. P.; Wang, Y. Accurate and Simple Analytic bond order based on a simple electron count.
Representation of the Electron-Gas Correlation Energy. Phys. Rev. B: (64) Arroyo-de Dompablo, M.; Morales-García, A.; Taravillo, M.
Condens. Matter Mater. Phys. 1992, 45, 13244. DFT+U Calculations of Crystal Lattice, Electronic Structure, and
(43) Di Valentin, C. Scanning Tunneling Microscopy Image Phase Stability under Pressure of TiO2 Polymorphs. J. Chem. Phys.
Simulation of the Rutile (110) TiO2 Surface with Hybrid Functionals 2011, 135, 054503.
and the Localized Basis Set Approach. J. Chem. Phys. 2007, 127, (65) Amtout, A.; Leonelli, R. Optical Properties of Rutile near Its
154705. Fundamental Band Gap. Phys. Rev. B: Condens. Matter Mater. Phys.
(44) Labat, F.; Baranek, P.; Domain, C.; Minot, C.; Adamo, C. 1995, 51, 6842.
Density Functional Theory Analysis of the Structural and Electronic (66) Pascual, J.; Camassel, J.; Mathieu, H. Fine Structure in the
Properties of TiO2 Rutile and Anatase Polytypes: Performances of Intrinsic Absorption Edge of TiO2. Phys. Rev. B: Condens. Matter
Different Exchange-Correlation Functionals. J. Chem. Phys. 2007, 126, Mater. Phys. 1978, 18, 5606.
154703. (67) Jones, D. R.; Troisi, A. A Method to Rapidly Predict the Charge
(45) Hammer, B.; Wendt, S.; Besenbacher, F. Water Adsorption on Injection Rate in Dye Sensitized Solar Cells. Phys. Chem. Chem. Phys.
TiO2. Top. Catal. 2010, 53, 423−430. 2010, 12, 4625−4634.
(46) Anisimov, V. I.; Zaanen, J.; Andersen, O. K. Band Theory and (68) Ronca, E.; Marotta, G.; Pastore, M.; De Angelis, F. Effect of
Mott Insulators: Hubbard U Instead of Stoner I. Phys. Rev. B: Condens. Sensitizer Structure and TiO2 Protonation on Charge Generation in
Matter Mater. Phys. 1991, 44, 943. Dye-Sensitized Solar Cells. J. Phys. Chem. C 2014, 118, 16927−16940.
(47) Grimme, S. Accurate Description of Van Der Waals Complexes (69) Pastore, M.; De Angelis, F. Computational Modelling of TiO2
by Density Functional Theory Including Empirical Corrections. J. Surfaces Sensitized by Organic Dyes with Different Anchoring
Comput. Chem. 2004, 25, 1463−1473. Groups: Adsorption Modes, Electronic Structure and Implication for

5455 DOI: 10.1021/acs.jpcc.5b10983


J. Phys. Chem. C 2016, 120, 5442−5456
The Journal of Physical Chemistry C Article

Electron Injection/Recombination. Phys. Chem. Chem. Phys. 2012, 14,


920−928.
(70) Martsinovich, N.; Troisi, A. How TiO2 Crystallographic Surfaces
Influence Charge Injection Rates from a Chemisorbed Dye Sensitiser.
Phys. Chem. Chem. Phys. 2012, 14, 13392−13401.
(71) Calbo, J.; Pastore, M.; Mosconi, E.; Ortí, E.; De Angelis, F.
Computational Modeling of Single-Versus Double-Anchoring Modes
in Di-Branched Organic Sensitizers on TiO2 Surfaces: Structural and
Electronic Properties. Phys. Chem. Chem. Phys. 2014, 16, 4709−4719.
(72) Pacia, M.; Warszyński, P.; Macyk, W. UV and Visible Light
Active Aqueous Titanium Dioxide Colloids Stabilized by Surfactants.
Dalton Trans. 2014, 43, 12480−12485.
(73) Łabuz, P.; Sadowski, R.; Stochel, G.; Macyk, W. Visible Light
Photoactive Titanium Dioxide Aqueous Colloids and Coatings. Chem.
Eng. J. 2013, 230, 188−194.
́
(74) Swiętek, E.; Pilarczyk, K.; Derdzińska, J.; Szaciłowski, K.; Macyk,
W. Redox Characterization of Semiconductors Based on Electro-
chemical Measurements Combined with UV-Vis Diffuse Reflectance
Spectroscopy. Phys. Chem. Chem. Phys. 2013, 15, 14256−14261.
(75) Buchalska, M.; Kobielusz, M.; Matuszek, A.; Pacia, M.; Wojtyła,
S.; Macyk, W. On Oxygen Activation at Rutile-and Anatase-TiO2. ACS
Catal. 2015, 5, 7424−7431.

5456 DOI: 10.1021/acs.jpcc.5b10983


J. Phys. Chem. C 2016, 120, 5442−5456

You might also like