Ensemble effects in nanostructured TiO2 used in the gas-phase photooxidation of trichloroethylene

Article abstract of DOI:10.1021/jp0131121

The effects of crystal size, structure, and crystallinity on the photocatalytic activity of nanostructured TiO₂ with different crystal sizes (3, 5, and 6 nm) for the gas-phase heterogeneous photocatalytic oxidation of trichloroethylene (TCE) wa

Full text of DOI:10.1021/jp0131121

4608
J. Phys. Chem. B 2002, 106, 4608-4616
Ensemble Effects in Nanostructured TiO2 Used in the Gas-Phase Photooxidation of
Trichloroethylene
,
†
‡
†
†
§
King Lun Yeung,* A. Javier Maira, Jennifer Stolz, Eppie Hung, Nick Ka-Chun Ho,
|
‡
|
†
An Chyi Wei, Javier Soria, Kuei-Jung Chao, and Po Lock Yue
Department of Chemical Engineering, The Hong Kong UniVersity of Science and Technology,
Clear Water Bay, Kowloon, Hong Kong, P.R. China, Instituto de Cat a´ lisis y Petroleoqu ´ı mica, CSIC,
Campus UAM, Cantoblanco, Madrid 28047, Spain, Material Preparation and Characterization Facility,
The Hong Kong UniVersity of Science and Technology, Clear Water Bay, Kowloon, Hong Kong, P.R. China,
and Department of Chemistry, National Tsinghua UniVersity, Hsinchu, Taiwan, R.O. China
ReceiVed: August 10, 2001; In Final Form: January 18, 2002
Nanostructured TiO
2
with different crystal sizes (i.e., 3, 5, and 6 nm) but identical aggregate size (i.e., 100
catalysts with crystal
nm) and shape (sphere) were successfully made using a modified sol-gel method. TiO
2
size of 5 nm and aggregate size of 100 nm were also prepared. This batch of catalysts has similar phase
structure, surface area, band-gap energy, and degree of surface hydroxylation, and differs mainly in their
crystallinity. Synchrotron XRD, XANES and EXAFS, Raman spectroscopy and transmission electron
microscopy (TEM) were used to characterize the structural properties of the nanometer sized TiO , whereas
2
ultraviolet, X-ray photoelectron and electron paramagnetic resonance spectroscopy provided important
information on the electronic properties of these nanoparticles. The catalytic performance of the nanostructured
TiO
2
for gas-phase photocatalytic oxidation of trichloroethylene exhibited strong dependency on both crystal
size and crystallinity.
Introduction
affected by the structural and electronic properties of the catalyst
1
1-17
18-20
that depend on particle size
and preparation method.
Heterogeneous photocatalytic oxidation (PCO) is a promising
technology for degradation of volatile organic compounds
Recent reports have shown that nanostructured anatase TiO2
particles exhibit higher activity11,12,18,21 and selectivity12 com-
VOCs) in process air stream.1-3 Using a semiconductor catalyst
e.g., TiO2) and a UV light source, this technology can oxidize
(
(
pared with the commercial TiO2 (P25, Degussa) for gas-phase
photocatalytic oxidation of volatile organic compounds. Al-
though there is a significant interest in the use of nanostructured
TiO2 in photocatalysis, the exact role of structure, electronic
and surface chemistry on the photocatalytic properties of the
nanometer sized TiO2 has not yet been fully resolved. Smaller
TiO2 crystals possess larger surface area for adsorption and
degradation of organic pollutants. As the size of TiO2 catalyst
is decreased to nanometer range, its electronic and structural
properties display a significant deviation from that of the bulk
material. Discretization of the energy band structure occurs,22
accompanied by a significant change in the optical and chemical
airborne VOCs into carbon dioxide, water, and mineral com-
pounds (e.g., HX, where X ) Cl, Br) at room temperature. The
UV irradiated catalyst absorbs photons having energy greater
than its band-gap, exciting electrons (e ) from the valence to
the conduction band and generating holes (h ). The resulting
-
+
electron-hole pairs migrate toward the catalyst surface and
initiate redox reactions that oxidize the adsorbed organic
molecules. Titanium dioxide has been widely used as a
photocatalyst for the oxidation of organic pollutants due to its
photostability, commercial availability, low cost, and relatively
good activity. Among commercial samples, Degussa TiO2 P25
has been the most commonly used photocatalyst for organic
pollutant photooxidation. Most commercial catalysts have
moderate quantum yield and total mineralization of certain
substrates is difficult.4 Catalyst deactivation is also a problem
for PCO of a number of pollutants.8-10 The efficiency of the
TiO2 photocatalyst depends on many factors including the
number and energy of absorbed photons during irradiation,
electron-hole pair formation and recombination rates, reactants
and products adsorption and desorption rates, and charge transfer
rate to the adsorbed species. Most of these factors are strongly
2
3
properties of TiO2. It also produces a net shift in the
11
semiconductor band-gap to higher energies, which means that
for the same light source, there are fewer photons with the
-7
-
+
required energy to generate the e /h pairs needed for the
reaction resulting in slower reaction rate. The altered band gap
-
+
structure also modifies the redox potential of the e /h pairs
2
4
participating in the reaction process. Ensemble effects also
become important as the particle size decreases. Small crystals
have larger surface curvature and contain greater proportions
of edge and corner sites, while the large crystals consist mainly
of planar sites. This directly affects the types of adsorption and
catalytic sites available for the reaction. In addition, the local
structural environment of the atoms in nanometer sized crystal
*
Author to whom correspondence should be addressed. Tel: 852-2358-
7
123. Fax: 852-2358-0054. E-mail: kekyeung@ust.hk.
†
Department of Chemical Engineering, The Hong Kong University of
2
5
Science and Technology.
is distinctly different from the bulk material. This work
attempts to elucidate the effects of crystal size, structure, and
crystallinity on the photocatalytic activity of nanostructured TiO2
for gas-phase PCO of TCE.
‡
Instituto de Cat a´ lisis y Petroleoqu ´ı mica, CSIC.
Material Preparation and Characterization Facility, The Hong Kong
§
University of Science and Technology.
|
National Tsinghua University.
1
0.1021/jp0131121 CCC: $22.00 © 2002 American Chemical Society
Published on Web 04/17/2002

Nanostructured TiO2 Used in the Photooxidation of TCE
TABLE 1: Preparation of Nanostructured TiO Catalysts
J. Phys. Chem. B, Vol. 106, No. 18, 2002 4609
2
titania
catalysts
hydrothermal treatment composition
(mole ratios)
crystallization
temperature (K)
crystallization
time (h)
P3
18 mL H
15 mL H
17 mL H
18 mL H
10 mL H
2
O + 7 mL 2-propanol
2
O + 10 mL 2-propanol
2
O + 8 mL 2-propanol
2
O + 7 mL 2-propanol
2
O + 15 mL 2-propanol
373
423
423
423
423
9
8
8
8
7
P5a
P5b
P5c
P6
Experimental Section
imaging, the sample powder was dispersed in 2-propanol and
placed onto a carbon-coated copper grid. The excess liquid was
removed using a paper wick and the deposit was dried in air
prior to imaging. The deposited TiO2 particles were examined
with a Philips CM20 transmission electron microscope at an
accelerating voltage of 200 kV. Routine chemical analysis using
an in-situ energy-dispersive X-ray spectrometer (EDXS, Philips
XL30) detected only the energy signatures corresponding to the
Ti and O elements. The average TiO2 crystal and aggregate
sizes, as well as their morphologies were obtained from the TEM
micrographs. The BET surface area of the TiO2 photocatalyst
was measured using the N2 physi-adsorption technique (Mi-
cromeritics ASAP 2010). The dried powder was outgassed under
vacuum at 523 K for 4 h prior to the analysis.
Synthesis and Characterization of Nanostructured TiO2.
In this study, a modified sol-gel method was employed in the
synthesis of nanostructured TiO2 photocatalysts of controlled
crystal and aggregate sizes.11 The procedure involves the
precipitation of amorphous titanium oxide gel spheres followed
by the controlled crystallization of the anatase TiO2 nanocrystals.
The concentrations of the precursor (i.e., titanium isopropoxide)
and water have direct effects on the hydrolysis, condensation,
and polymerization reactions that dictate the size of precipitated
gel spheres. In this study, the synthesis conditions were adjusted
to give 100 nm gel spheres of uniform size. Table 1 summarizes
the treatment conditions used to crystallize the anatase TiO2.
During the hydrothermal treatment, the amorphous gel spheres
were transformed into aggregates of TiO2 nanocrystals with no
significant change in size and shape. Thus, TiO2 photocatalysts
with different crystal sizes but fixed aggregate size (i.e., 100
nm) and shape (i.e., spheres) were obtained. This preparation
procedure was also adopted for the preparation of TiO2 with
similar crystal size but different crystallinity.
The crystal structure, particle size, morphology, and surface
area of TiO2 photocatalysts were characterized using X-ray
diffraction (XRD), X-ray absorption spectroscopy (XAS), micro-
Raman, transmission electron microscopy (TEM), and N2 physi-
adsorption. X-ray diffraction and X-ray absorption analyses of
the TiO2 powder were conducted on beamline BL17A at the
Synchrotron Radiation Research Center (SRRC) in Taiwan. The
X-ray radiation (λ ) 1.3271 Å) with a beam current of 120-
The surface composition and chemical state of the catalyst
samples were determined by X-ray photoelectron spectroscopy
(XPS, Physical Electronics PHI 5600) using a monochromatic
aluminum X-ray source. The catalyst powder was pressed onto
a double-sided carbon tape to make a flat surface and then
introduced into the instrument for analysis. The change in the
band-gap energy of the TiO2 as a result of the material
preparation was also measured using a Philips/Unicam Pu8700
UV/VIS spectrometer with diffuse and specular reflection
accessories to provide information on the electronic band gap.
A Ba2SO4 standard white plate was used as background. The
spectra were recorded at a scan speed of 125 nm/min and a
bandwidth of 2 nm.
Evaluation of Photocatalytic Activity of TiO2. Gas-phase,
photocatalytic oxidation of trichloroethylene (TCE) was used
as a probe reaction to study the catalytic activity of nanostruc-
tured TiO2. The reaction was conducted in a flat-plate photo-
2
00 mA was supplied from a 1.5 GeV storage ring. The XRD
patterns were recorded for 15° < 2θ < 55° by step-scanning at
.05° increments. The phase structure, crystallinity, and grain
0
1
1
reactor using TiO2-coated stainless steel plates. Catalyst
powder (30 mg) was uniformly coated onto the 25 mm × 25
mm plate. A metered amount of TCE vapor (0.5 µL/min) was
mixed with flowing oxygen (200 mL/min) before entering the
photoreactor to give a TCE partial pressure of 70 Pa (4.5 ppmv).
The reactant mixture then enters the reactor and flows over the
catalyst plate through a narrow rectangular channel (2 mm deep
size of the TiO2 were determined from the X-ray diffraction
patterns. The X-ray absorption spectra of the TiO2 samples were
measured in transmission mode at room temperature. To
eliminate the effects of sample thickness, the TiO2 powder was
rubbed uniformly onto a Scotch tape and folded to get the
desired thickness that satisfies ∆µx e 1, where ∆µx is the edge
step. The spectra were collected using a gas ionization detector.
The ion chambers used for measuring the incident (I0) and
transmitted (I) synchrotron beam intensities were filled with a
He/N2 gas mixture and N2 gas, respectively. Data were collected
from 200 eV below the Ti K-edge (4966 eV) to 1100 eV above
the edge. Standard titanium metal foil and commercial anatase
TiO2 were used as reference standards. Information on the bond
structure (i.e., bond lengths, numbers, and types of neighboring
atoms) could be obtained from the X-ray absorption data.
The crystal size was also measured from Raman line
broadening and compared with the results of the X-ray diffrac-
tion experiments. The Raman spectra were measured using a
Renishaw 3000 micro-Raman system with an Olympus BH-2
microscope. The objective lenses with 20× and 50× magnifica-
tions were selected. The excitation source used was an argon
laser operating at 514.5 nm with an output power of 25 mW.
Raman spectra were taken of the powders on a glass microscope
×
112 mm wide). After the feed concentration had stabilized (t
1 h), the photocatalyst was illuminated by five fluorescent
∼
black lamps (6 W, BLB Sankio Denki) located 7 mm above
the reactor window. The outlet gases were separated using a
GS-GASPRO capillary column (0.32 mm × 30 m) and analyzed
using a gas chromatograph (HP 6890) equipped with thermal
conductivity and flame ionization detectors. The reactant and
product concentrations were monitored continuously with time.
The photocatalytic reaction was also investigated using an
in-situ infrared photoreaction cell. The cell consists of two KBr
windows and a sample holder for the catalyst wafer. The
reactants and products enter and exit the cell through a set of
inlet and outlet ports. The compressed air used in the reaction
was metered (800 mL/min) and pretreated through a series of
moisture and hydrocarbon traps that included a liquid N2 trap,
a drying column packed with desiccant, and molecular sieves.
Trichloroethylene was fed using a syringe pump (kdScientific
1000) to a mixing tee where it was vaporized and mixed with
-1
slide. The spectral resolution was set at approximately 1.0 cm
and the spot size was about 2 µm in diameter. For the TEM

4610 J. Phys. Chem. B, Vol. 106, No. 18, 2002
Yeung et al.
Figure 1. (a) Tapping-mode atomic force microscopy image of P5a dispersed onto mica substrate (inset: individual 100 nm aggregates of 5 nm
crystals), (b) synchrotron X-ray diffraction pattern, and (c) XANES spectra of the TiO samples (inset: magnified preedge region of commercial
anatase TiO ).
2
2
the dry air. The reactant mixture then flowed through the in-
situ infrared photoreaction cell and allowed to equilibrate at
room temperature (298 K). Once the reactant concentration had
stabilized, the inlet and outlet ports were shut-off and the UV
lamp was turned on to allow the photoreaction to proceed under
batch conditions. The infrared spectra were continuously col-
lected by a Fourier transform infrared spectrometer (FTS 6000,
Bio-Rad Laboratories) during the reaction until the intensity of
the characteristic peaks of adsorbed TCE molecule was de-
creased to less than 0.2 of the original. The UV lamp was then
turned off and the cell purged with flowing dry air. The catalyst
wafers used in the reaction experiments were prepared by mixing
reference standard for the synchrotron X-ray experiments.
Standard FTIR-grade potassium bromide (Aldrich) was used
as catalyst diluent in the in-situ IR experiments. The reactants
used for catalyst testing were trichloroethylene (99.5%, Aldrich)
and oxygen (HP, HKO). Gases used in the GC were helium
(UHP, CW), hydrogen (UHP, HKO), and synthetic air (HP,
HKSP).
Results and Discussion
Characterization of Nanostructured TiO2. Figure 1a dis-
plays typical TiO2 aggregate clusters (P5a) prepared by the
modified sol-gel process. The atomic force microscope image
reveals that the clusters are made up of individual nanocrystals
with 5 nm diameter (Figure 1a inset). It is clear from the
presence of crystal facets that the TiO2 are crystalline. Indeed,
this is confirmed by the X-ray diffraction patterns (Figure 1b)
which show that the primary crystalline phase of the nanostruc-
tured TiO2 is anatase. Brookite is also present in small amount
in all the samples. The average crystal size calculated from the
broadening of the anatase (101) peak is displayed in Table 2. It
is also clear from the diffraction patterns that although the P5a,
P5b, and P5c have comparable crystal size, they exhibit different
crystallinity. The Raman spectra of 5 and 6 nm TiO2 samples
0.08 g of TiO2 with 0.70 g of KBr. Three wafers were prepared
in each batch, each weighing 0.20 g. The wafer thickness as
measured by scanning electron microscopy (JEOL 6300) was
2
2
2
80 ( 22 µm, thus giving the wafers an average density of
3
.28 g/cm . The wafers were pretreated in flowing dry air for
h at room temperature prior to the reaction to remove adsorbed
CO2.
Materials. The chemicals used in the synthesis of nanostruc-
tured TiO2 photocatalysts included titanium isopropoxide (97%,
Aldrich), 2-propanol (99.5%, Aldrich), and distilled water.
Anatase TiO2 (99.9+ %) from Aldrich Chemicals was used as

Nanostructured TiO2 Used in the Photooxidation of TCE
J. Phys. Chem. B, Vol. 106, No. 18, 2002 4611
TABLE 2: Structural Properties of Nanostructured TiO
2
Prepared by Modified Sol-Gel Method
Ti K-edge EXAFSd
primarily
secondary
BET
2
-3
titania
catalysts
particle size
particle size
σ ×10
surface area
a
b
crystallinityc
- -
2
2
(nm)
(nm)
shell
R(Å)
C. N.
(Å )
(m /g)
P3
2.9
100
Ti-O
Ti-Ti
Ti-O
Ti-O
Ti-Ti
Ti-Ti
Ti-O
Ti-O
Ti-Ti
Ti-Ti
Ti-O
Ti-O
Ti-Ti
Ti-Ti
Ti-O
Ti-O
Ti-Ti
Ti-Ti
1.93
3.05
1.94
1.99
3.05
3.79
1.93
1.99
3.05
3.79
1.94
1.99
3.05
3.79
1.95
2.00
3.07
3.81
4.5
1.5
3.0
2.8
1.9
0.7
3.0
3.0
2.1
1.2
2.8
3.2
2.2
1.0
4.2
2.1
2.4
2.9
15
330
9.9
10.3
10.5
6.9
7.1
9.4
9.6
6.3
6.4
8.9
9.1
5.9
6.1
9.8
10.1
6.6
6.8
P5a
4.6
100
1.0e,f
0.8e
0.5e
1.0e
275
P5b
P5c
P6
5.0
5.1
5.7
100
100
100
240
220
175
a
The crystal size was calculated from XRD line broadening and confirmed by micro-Raman and TEM analyses. b The average aggregate size
c
was obtained from 50 direct measurements of TiO
XRD peak, i.e., crystallinity ) I/I , using P5a as the reference . EXAFS values were obtained from model fitting using a standard software
0
package (UWXAFS 3.0).
2
clusters imaged by TEM. The crystallinity was based on the intensity of the anatase (101)
e
f d
display a distinct shift in their spectral lines when compared
with coarser TiO2 crystals (i.e., 11 nm P11). The peak intensity
decreases and the spectral line broadens with decreasing particle
size. A strong correlation between Raman line broadening and
samples with different crystal size. The 4-, 5-, 6-fold coordinated
Ti can be identified according to the position and intensity of
the preedge peaks, which are located at 4969.5 (A1), 4970.5
(A2), and 4971.5 (A3) eV, respectively. As the TiO crystal
2
26
26
TiO2 particle size has been reported. Iida and co-workers
size decreases, the interfacial region comprising titanium atoms
claim that this method is more sensitive for nanometer sized
crystals than X-ray line broadening. The average crystal size
calculated from the Raman peak broadening is in good agree-
ment with the synchrotron XRD results. Although TEM was
used mainly to measure the size and shape of the TiO2
aggregates, the crystal size could also be determined from the
high-resolution images. Their values usually agree well with
that calculated from X-ray and Raman peak broadening. The
BET surface area of the TiO2 measured by N2 physi-adsorption
displays a monotonic increase with decreasing crystal size (Table
with low coordination number increases. Thus, the titanium
atoms in smaller crystals will exhibit lower coordination
numbers. Indeed, the 3 nm TiO2 (P3) has the most intense A2
peak (4970.75 eV) that is consistent with 5-fold coordinated
Ti. Similar structural geometry has been reported for 1.9 nm
27
TiO2. Larger crystals, which have smaller interfacial area,
exhibit a weaker A2 peak but a more intense A3 peak (4971.75
eV) that corresponds to 6-fold coordinated Ti. The preedge fea-
tures of the 5 nm TiO2 with different crystallinity are identical.
Extended X-ray absorption fine structure (EXAFS) analysis
of the TiO2 samples provides detailed local structural informa-
tion not available from the X-ray diffraction data. The analysis
was done using UWXAFS 3.0 software package and FEFF8
program and the results are tabulated in Table 2. The magnitude
2
).
Figure 1c displays the normalized Ti K-edge X-ray absorption
near-edge spectra (XANES) of the TiO2 samples listed in Table
, including that of a commercial crystalline anatase TiO2 being
2
used as a standard. The XANES spectrum of the commercial
TiO2 (Figure 1c) indicates a somewhat distorted TiO6 octahedron
atomic arrangement with four short Ti-O (1.93 Å) and two
long Ti-O bonds (1.98 Å). The preedge region of the anatase
Ti K-edge XANES usually consist of four preedge features
labeled A1, A2, A3, and B (cf. Figure 1c inset). The feature
labeled A2 is sometimes observed as a weak shoulder on the
low energy side of the A3 peak. The intensity and position of
the preedge transitions are sensitive to the symmetry of the
surrounding atoms, being dipole forbidden but quadrupole
allowed, are weak in symmetrical environments and display an
increase in intensity as the environment is distorted. The
postedge region of the Ti K-edge XANES from 4980 to 5020
eV contains a number of resolved and understood intense 3s f
np dipole-allowed transitions. The normalized Ti K-edge
XANES spectra of TiO2 samples with different crystal size (P3,
P5a, and P6) and crystallinity (P5a, P5b, and P5c) are also
shown in the figure. It is clear from the spectra that as the crystal
size of TiO2 samples decreases, the features in the postedge
region become poorly resolved. A pronounced difference in the
feature of the preedge region is evident in the spectra of TiO2
3
of Fourier transform of Ti K-edge k -weighted EXAFS data
(uncorrected for phase shift) for the nanostructured TiO2 exhibit
a general increase in amplitude of the FT peaks with increasing
particle size, in particular the relative intensity of the peaks
corresponding to the higher shells. The fitting results for the 3
nm TiO2, P3 (Table 2) show that the Ti-O bond length is 1.93
Å and the coordination number is 4.5, which is consistent with
the XANES result. This catalyst, which is comprised of 3800
TiO2 units, has a surface region (Ap) that is comparable to the
bulk (Vp) with an Ap/Vp ratio of 1. In the other TiO2 samples,
the first peak must be fitted with two shells of Ti-O. The P5a,
P5b, and P5c samples (Ap/Vp ) 0.6) have the same results
showing the presence of three short bonds (1.93 Å) and three
long Ti-O bonds (1.98 Å) in the first shell despite the difference
in their crystallinity. Although the Ti in these samples are 6-fold
coordinated, they are not like the octahedron of the anatase TiO2
that is characterized by four short and two long Ti-O bonds.
This could be due to the contribution from the interfacial atoms
or the possible presence of structural defects. The P6 (Ap/Vp )
0.5) has the same structure as the commercial anatase TiO2 (four

4612 J. Phys. Chem. B, Vol. 106, No. 18, 2002
Yeung et al.
signals disappeared when annealing the TiO2 at high temper-
atures eliminates the low coordinated surface sites. Subsurface
4+
defects containing low coordinated Ti were detected indirectly
from the formation of subsurface O species (O-b) when the
nanocrystals were irradiated in the presence of adsorbed oxygen.
-
+
2-
The trapping of photogenerated h at these subsurface O
anions was shown to be mediated by the surface hydroxyl groups
2-
4+
-
4+
(
-O -Ti -OH + h+ f -O -Ti -OH) that were abun-
dant on the surface of nanostructured TiO2.
1
7
Only titanium and oxygen were detected by in-situ EDXS
analysis of the TiO2 particles during TEM imaging. The XPS
measurements also gave similar results for the surface composi-
tion of the crystalline samples with adsorbed atmospheric
carbons being the main surface contaminant. Due to the larger
mean free path of photoelectrons from the valence band, the
valence band spectra will contain signals from both the surface
and bulk, and can be used to probe the electronic structure of
2
Figure 2. UV spectra of the nanometer sized TiO .
short Ti-O bonds of 1.93 Å and two long Ti-O bonds of 1.98
Å).
TiO2. The XPS results indicated that the valence bands of P6
and P5a TiO2 were identical, but the energy distribution of the
valence band of P3 displayed a distinct difference from the larger
TiO2. The crystallinity has little effect on the structure of the
valence band. Figure 2 shows that there is a clear shift in the
semiconductor band-gap toward higher energy when the TiO2
crystal size is decreased from 11 to 6 and 3 nm, but there is
only a slight difference between the 5 and 6 nm TiO2. The
occurrence of the blue-shift means that for the same light source,
there are fewer photons with the required energy to generate
The high concentrations of low coordinated Ti4 cations, as
well as the distorted anatase crystal structure observed by
XANES and EXAFS, are strongly suggestive of the presence
of amorphous materials in the nanometer sized TiO2. However,
a recent EPR study conducted on the same set of nanostructured
TiO217 was unable to detect the broad Ti signal that is
characteristic of UV irradiated amorphous TiO2 in a vacuum,
+
3+
4
+
although it did confirm the presence of low coordinated Ti
cations. The EPR spectra for the nanostructured TiO2 contained
-
+
the e /h pairs needed for a reaction resulting in a lower
utilization of photons.
3+
mainly the signals corresponding to Ti cations formed by the
trapping of photogenerated electrons at Ti4 sites on the crystal
surface (i.e., corners, steps, edges, kinks and defects). These
+
Photocatalytic Activity of Nanostructured TiO2. Crystal
Size and Structure. Figure 3a plots the areal conversion rate of
Figure 3. (a) Areal TCE conversion rates as a function of time for different crystal sizes, (b) areal CO
different crystal sizes, (c) areal TCE conversion rates as a function of time for different crystallinity, (d) areal CO
of time for different crystallinity.
2
production rates as a function of time for
production rates as a function
2

Nanostructured TiO2 Used in the Photooxidation of TCE
J. Phys. Chem. B, Vol. 106, No. 18, 2002 4613
Figure 4. (a) Transient in situ FTIR transmittance spectra of TCE photooxidation on P5a and (b) FTIR transmittance spectra collected immediately
after turning the UV off [1] and after 32 h of purging in dry air [2].
TCE as a function of UV irradiation time for 5 and 6 nm TiO2
catalysts. The areal rate was used for convenience since the exact
nature and population of active sites were not known. The
photocatalytic oxidation of TCE over 5 nm TiO2 catalyst (P5a).
The spectra were corrected using the clean TiO2 as the
background. Prior to UV illumination (t ) 0), the spectrum
displays the characteristic TCE bands at 850 (δC-H), 950 (νC-Cl),
1
1
experiments were conducted in the flat-plate photoreactor at
room temperature and under steady-state flow conditions. The
photocatalytic oxidation reaction was run under dry condition.
Both catalysts exhibit activation behavior, reaching their
maximum conversion rates after 20 min of irradiation. Similar
reaction behavior has been reported for photocatalytic oxidation
of TCE over commercial anatase TiO2 powder.28 After activa-
tion, P6 has a stable activity but the 5 nm TiO2 (i.e., P5a)
displays a slow deactivation that was accompanied by a color
change from white to light yellow. It is clear from Figure 3a
that the 6 nm TiO2 is more active than the 5 nm catalyst. It
also exhibits a higher mineralization rate as shown by the plot
of CO2 production rate versus irradiation time (Figure 3b). The
TCE conversion over commercial Degussa P25 catalyst is six
-
1
1255 (δCH-Cl), 1550 (νCdC), 1580 (νCdC), and 3070 cm
3
0
(νC-H). Upon irradiation, the intensity of TCE bands (e.g.,
-
1
850, 950, 1255, and 3070 cm ) display a monotonic decrease
with time, while the signals corresponding to CO2 (i.e., 2340
-
1
and 2360 cm ) and dichloroacetaldehyde (i.e., 1230 (δC-H),
-1
29
1400 (δCHdO), and 1740 cm (νCdO) ) increase as the reaction
progresses. It is evident from the time-resolved spectra that both
associated and isolated hydroxyl groups native to the catalyst
-
1
-1
(i.e., the broad band (3200-3500 cm ) centered at 3400 cm
-
1
19
and the band at 3687 cm , respectively ) were either being
consumed or desorbed during the reaction. In contrast, water
associated with the bands at 1630 and 3100 cm was produced,
as is expected for the photooxidation of TCE.
-1
-
10
1
-1
times slower than P6 with an areal rate of 3 × 10
m . The corresponding CO2 production rate over P25 is 1 ×
mol s m .
In-situ infrared reaction study provides real-time monitoring
mol s
Figure 4b displays the postreaction infrared spectrum of TiO2
catalyst immediately after turning the UV illumination off
-2
-
10
1
-1
-2
3
1
0
(Figure 4b-1), and after 32 h of purging using dry air (800 cm /
min) at room temperature (Figure 4b-2). Both spectra used the
clean TiO2 wafer as the background reference. Spectrum 1
shows the presence of adsorbed TCE reactant as well as carbon
dioxide, dichloroacetaldehyde, and water. TCE and carbon
dioxide were removed within 10 minutes of purging, whereas
the water was desorbed after 80 min. An additional peak at 1600
of transient events that are occurring on the catalyst during the
reaction. The information on surface adsorbed species (i.e.,
reactants, intermediates, and products) obtained from the in-
situ study give important insight on the reaction mechanisms
and kinetics.29 Infrared spectroscopy is also an invaluable
technique for the characterization of surface hydroxyl groups
that populate the oxide surfaces (e.g., TiO2). Figure 4a displays
a set of infrared transmittance spectra obtained during the
-
1
cm was revealed after desorption of water (Figure 4b) which
had been assigned to νc)c of dichloroethylene by Phillips and
2
9
Raupp, however the accompanying signal for δCH-Cl (1302

4614 J. Phys. Chem. B, Vol. 106, No. 18, 2002
Yeung et al.
Figure 5. (a) Plots of the intensity of νC-Cl (950 cm-1) for TCE molecule and (b) Ln (I950) versus irradiation time for different TiO
] P3, O P5a, 0 P6), (c) plots of the intensity of I950, and (d) Ln (I950) versus irradiation time for different TiO crystallinity (O P5a, 0 P5b, ] P5c).
crystal sizes
2
(
2
-
1
-1
cm ) was missing. The signals at 1600 and 1360 cm had
also been assigned to adsorbed formic acid on TiO2. However,
gas chromatographic analyses indicate that dichloroethylene is
the more likely candidate. Continuous purgings with dry air were
unable to remove the adsorbed dichloroacetaldehyde (i.e., 1230,
and water increases steadily as the reaction progresses. The
presence of dichloroethylene cannot be ascertained due to the
signal interference from adsorbed water molecules. It is clear
from the figures that the product distributions are different for
the three catalysts. The 3 and 5 nm TiO2 catalysts produce more
dichloroacetaldehyde as compared to the 6 nm TiO2, which
generates more carbon dioxide during the photoreaction. Greater
quantities of water are also produced by the 3 and 5 nm TiO2
-
1
1
400, and 1740 cm ) and dichloroethylene. These signals
remained unchanged even after UV irradiation in flowing dry
air (i.e., room temperature and 6 h) and were only removed
when the sample was irradiated under humid condition (80 %
R. H.). It is plausible that these strongly adsorbed molecules
play a role in the observed catalyst deactivation.
-
1
at the expense of the weakly bound water (i.e., ∼3400 cm )
native to the catalyst. The signal for δH-O-H for P6 at 1630
-
1
cm is weak and there is no noticeable change in the intensity
-
1
of the signal centered at 3400 cm . Unlike the other two
catalysts, P6 displays little deactivation during TCE photocata-
lytic oxidation (up to 5 h) and the catalyst does not show any
discoloration after the reaction.
Figure 5a plots the intensity of the characteristic νC-Cl peak
950 cm ) for TCE as a function of irradiation time for the 3,
, and 6 nm TiO2 catalysts. The intensity of the νC-Cl peak
displays a monotonic decrease with time as TCE is consumed
by the photocatalytic reaction. It is clear from the plots that the
three catalysts differ in their activity. TCE conversion is highest
-
1
(
5
A closer inspection of Figure 6a indicates that dichloroac-
etaldehyde is produced first, before carbon dioxide. The latter
only becomes significant after 40 min of UV irradiation. This
suggests a stepwise oxidation of trichloroethylene toward the
final mineral products (i.e., CO2, H2O, Cl2, and HCl). This is
not inconsistent with the previous mechanism proposed in the
for P6 followed by the P5a with the 3 nm catalyst (P3) being
the least active. Several studies3
1-33
have reported that the
photocatalytic oxidation of TCE over anatase TiO2 fit to a
Langmuir-Hinshelwood expression where the reaction is first
order with respect to trichloroethylene at low TCE concentra-
tions. Figure 5b shows the results of fitting the intensity of the
νC-Cl peak to a first-order kinetic model. The agreement between
the experimental data and model is good, and the slopes (m) of
the lines are 0.006, 0.010, and 0.011 for 3, 5, and 6 nm TiO2,
respectively. Since the surface area of P6 is half that of the 5
nm TiO2 (cf. Table 2), it means that it has a higher areal rate
than the latter catalyst. Figure 6a-c plots the intensity of the
29,34
literature.
Water (δH-O-H) is detected immediately upon UV
irradiation and must have originated from the catalyst rather
than the product of photomineralization of TCE. Indeed, water
production is usually accompanied by a decrease in the signal
-
1
intensity of 3400 cm assigned to hydroxyl groups associated
by hydrogen bonding on the TiO2 catalyst surface. The most
noticeable difference between the 5 and 3 nm TiO2 is the
production of more carbon dioxide (Figure 6b). Figure 6c shows
a significant increase in carbon dioxide production for the 6
nm TiO2 at the expense of dichloroacetaldehyde, indicating the
high mineralization rate that could be obtained from this catalyst.
-
1
signals corresponding to δCHdO (1400 cm ) and νCdO (1740
-
1
-1
cm ) belonging to dichloroacetaldehyde, νCdO (2340 cm )
-1
of carbon dioxide and δH-O-H (1630 cm ) of strongly bounded
water for 3, 5, and 6 nm TiO2 catalysts, respectively. The plots
show that the amount of dichloroacetaldehyde, carbon dioxide,
-
1
The intensity of the water peak at 1640 cm is significantly
lower when compared to the other two catalysts. Qualitatively,

Nanostructured TiO2 Used in the Photooxidation of TCE
J. Phys. Chem. B, Vol. 106, No. 18, 2002 4615
2 2 2
Figure 6. Product distribution from TCE photocatalytic oxidation over (a) P3 (3 nm TiO ), (b) P5a (5 nm TiO ), (c) P6 (6 nm TiO ), (d) P5b
(crystallinity ) 0.8), and (e) P5c (crystallinity ) 0.5). (O I1640, b I2343, 0 I1400, 9 I1740).
the results of the flow reactor (i.e., flat-plate photoreactor) and
batch reactor (infrared cell) are in agreement that the 6 nm TiO2
is the better photocatalyst for TCE mineralization.
The nanostructured TiO2 catalyst displayed a progressive
decrease in their areal conversion rates despite the increase in
their surface area as the particle size diminishes. This phenom-
enon could be attributed to electronic and ensemble effects in
nanometer sized particles. It is clear from the UV spectra and
X-ray absorption spectroscopy (i.e., XANES and EXAFS) that
the decrease in TiO2 crystal size is accompanied by a blue shift
of different crystallinity (cf. Table 2 and Figure 1b) were
prepared using the modified sol-gel method. The TiO2 samples
also have comparable BET surface area (Table 2). Using P5a
as the reference, the crystallinity of P5b and P5c were assigned
the values of 0.8 and 0.5, respectively. Figure 3c,d displays the
plots of areal TCE conversion rate and CO2 production rate as
a function of UV irradiation time for TCE photocatalytic
oxidation in a flat-plate photoreactor. The plots show that the
three 5 nm photocatalysts have comparable activity for TCE
conversion (Figure 3c), but the least crystalline TiO2 (i.e.,
P5cS100) exhibits a significantly lower mineralization rate
(Figure 3d). All three catalysts experienced slow deactivation
and displayed a color change from white to light yellow at the
end of reaction. The results of the in-situ infrared reaction study
(Figure 5c,d) also show that the TCE conversions for the three
catalysts are comparable and unaffected by the difference in
TiO2 crystallinity. Figures 6b,d,e plot the intensity of the infrared
(Figure 2), and a significant change in the local structural
environment of the catalyst (Table 2). The number of defect
sites that stabilize the low coordinated Ti4 cations was also
observed by EPR spectroscopy to increase with decreasing
+
17
crystal size. The presence of subsurface defects directly affects
the photocatalytic properties of the TiO2 nanoparticles because
they act as trapping sites for photogenerated holes, preventing
them from reaching the surface region to participate in the
reaction. OH• radicals on TiO2 surface are considered to be
important active species in photocatalytic oxidation reactions.
They are formed from the interactions between photogenerated
holes and surface hydroxyl groups (i.e., basic OH). The trapping
of photogenerated holes in subsurface defect sites not only
hinders the formation of OH• radicals, but also lead to electron-
hole recombination. This explains the significant difference in
the performance of the 5 and 6 nm TiO2 (Figure 3a,b). It is
safe to attribute most of the observed activity and selectivity
behavior of 5 and 6 nm TiO2 to the catalyst structure (i.e.,
ensemble effects) since they differ in crystal size by less than
a nanometer (Table 2) and have comparable band-gap energies
-
1
signals corresponding to δCHdO (1400 cm ), δH-O-H (1630
-
1
-1
-1
cm ), νCdO (1740 cm ) and νCdO (2343 cm ) for P5a, P5b,
and P5c, respectively. The plots show that the amount of
dichloroacetaldehyde, carbon dioxide, and water increases
steadily with time, however the product distribution is different
for the three catalysts. The P5a and P5b produce the most carbon
dioxide, while the P5c produces the least. This is in general
agreement with the results of the flat-plate photoreactor experi-
ments (Figure 3d). It is interesting to note that the signal for
-
1
δH-O-H at 1630 cm is lower for less crystalline TiO2 (Figure
6e), although infrared study indicates that the surface of all three
catalysts is comparably hydroxylated.
XRD, XANES, and EXAFS analyses indicated that the three
(Figure 2).
TiO catalysts have comparable size, crystal phase, and atomic
2
Crystallinity. A set of TiO2 catalysts (i.e., P5a, P5b, and P5c)
structure (Table 2). The catalysts also have similar BET surface
with similar crystal size, structure, and band-gap energy, but
area and surface hydration as measured by N2 physi-adsorption

4616 J. Phys. Chem. B, Vol. 106, No. 18, 2002
Yeung et al.
experiment and infrared spectroscopy. UV spectroscopy and
XPS analyses of the catalysts also yield identical band gap
energy and valence band structure. The TiO2 differs mainly in
their crystallinity as measured by XRD. Crystallinity gave an
indirect qualitatiVe measure of the structural defects present in
the catalyst materials. However, the exact nature and number
of these defects are difficult if not impossible to characterize
and quantify. Their presence is known to affect the migration
of photogenerated charges and their recombination rate. This
is expected to directly impact the photocatalytic activity of the
nanostructured TiO2. It is clearly manifested by the increased
trapping of photogenerated holes at subsurface region of the
catalyst and concomitant decrease in OH• radicals as detected
and facility for conducting the XRD, XANES, and EXAFS
studies. We also acknowledge the help of Dr. Miguel Angel
Ba n˜ ares of Instituto de Cat a´ lisis y Petroleoqu ´ı mica for conduct-
ing the MicroRaman study, and Ms. Willma Lau Wai Ngar for
conducting some of the photoreaction experiments.
References and Notes
(1) Hoffmann, M. R.; Martin, S. T.; Choi, W.; Bahnemann, D. W.
Chem. ReV. 1995, 95, 69.
(2) Mills, A.; Davies, R. H.; Worsley, D. Chem. Soc. ReV. 1993, 417.
(
3) Fox, M. A.; Dulay, T. Chem. ReV. 1993, 93, 341.
(4) Nimlos, M. R.; Jacoby, W. A.; Blake, D. M.; Milne, T. A. EnViron.
Sci. Technol. 1993, 27, 732.
(5) Muggli, D. S.; Larson, S. A.; Falconer, J. L. J. Phys. Chem. 1996,
100, 15886.
1
7
by EPR. This suggests that the lower mineralization rate of
the less crystalline 5 nm TiO2 is directly related to the greater
number of defects in its structure.
(6) Maira, A. J.; Soria, J.; Augugliaro, V.; Loddo, V. Chem. Biochem.
Eng. Q. 1997, 11 (2), 89.
(7) Djeghri, N.; Teichner, S. J. J. Catal. 1980, 6, 99.
(
(
8) Blake, N. R.; Griffin, G. L. J. Phys. Chem. 1988, 92, 5697.
9) Luo, Y.; Ollis, D. F. J. Catal. 1996, 163, 1.
Concluding Remarks
(10) Ameen, M. M.; Raupp, G. B. J. Catal. 1999, 184, 112.
It is important to elucidate the role of structure, chemistry
and electronic on catalytic activity of TiO2 nanoparticles in order
to further enhance their performance as an environmental
photocatalyst. The ability to engineer the phase structure, size
and crystallinity of the nanometer sized TiO2 enabled us to
separately probe the effects of structure (i.e., crystal and
(11) Maira, A. J.; Yeung, K. L.; Yan, C. Y.; Yue, P. L.; Chan, C. K. J.
Catal. 2000, 192, 185.
(12) Maira, A. J.; Yeung, K. L.; Soria, J.; Coronado, J. M.; Belver, C.;
Lee, C. Y.; Augugliaro, V. Appl. Catal. B: EnViron. 2001, 29, 327.
(13) Anpo, M.; Shima, T.; Kodama, S.; Kubokama, Y. J. Phys. Chem.
987, 91, 4305.
1
(14) Xu, N.; Shi, Z.; Fan, Y.; Dong, J.; Shi, J.; Hu, M. Z.-C. Ind. Eng.
Chem. Res. 1999, 38, 373.
(15) Wang, C.-C.; Zhang, Z.; Ying, Y. J. Nanostr. Mater. 1997, 9, 583.
(16) Zhang, Z.; Wang, C.-C.; Zakaria, R.; Ying, Y. J. Phys. Chem. B
aggregate size)1
1,12
9
and electronics,
12,17
as well as surface
1
hydroxyl group on the photocatalytic activity of the nano-
structured TiO2. This work demonstrated that the crystal size
and crystallinity of the nanometer sized TiO2 can be precisely
engineered using the modified sol-gel method. Reaction study
conducted using gas-phase photooxidation of TCE as a probe
reaction show that both crystal size and crystallinity affect the
catalyst activity. Although the influence of crystallinity on TCE
conversion rate is weak, it has a significant effect on the TCE
mineralization. The results of this study suggest that an effective
TiO2 photocatalyst must have large surface area populated by
low coordinated surface sites (i.e., corners, edges, kinks, and
defects) that are abundantly hydroxylated. This can be achieved
by decreasing the crystal size and by using hydrothermal
crystallization method. However, it is also necessary that the
nanometer sized TiO2 should be well crystallized with minimal
bulk defects that may trap migrating photogenerated charges.
This latter requirement is far more difficult to satisfy as size of
the TiO2 crystal diminishes. Further study will be conducted to
formulate new synthesis strategy for reducing the number of
bulk defect sites in nanostructured TiO2 (dp < 6 nm).
1
998, 102, 10871.
(17) Coronado, J. M.; Maira, A. J.; Conesa, J. C.; Augugliaro, V.; Soria,
J. Langmuir 2001, 17, 5368.
(18) Cao, L.; Huang, A.; Spiess, F. J.; Suib, S. L. J. Catal. 1999, 188,
48.
(
19) Maira, A. J.; Coronado, J. M.; Augugliaro, V.; Yeung, K. L.;
Conesa, J. C.; Soria, J. J. Catal. 2001, 202, 413.
20) Zaban, A.; Aruna, S. T.; Tirosh, S.; Gregg, B. A.; Mastai, Y. J.
Phys. Chem. B. 2000, 104, 4130.
21) Cao, L.; Gao, Z.; Suib, S. L.; Obee, T. N.; Hay, S. O.; Freihaut, J.
D. J. Catal. 2000, 196, 253.
22) Brus, L. J. Phys. Chem. 1986, 90, 2555.
(
(
(
(23) Kariyone, T.; Anpo, M.; Chiba, K.; Tomonari, M. Hyomen (surface)
991, 29, 156.
1
(
(
(
24) Dagn, G.; Sampath, S.; Lev, O. Chem. Mater. 1995, 7, 446.
25) Henglein, A. Chem. ReV. 1989, 89, 1861.
26) Iida, Y.; Furukawa, M.; Aoki, T.; Sakai, T. Appl. Spectrosc. 1998,
52, 673.
(27) Chen L. X.; Rajh, T.; J a¨ ger, W.; Nedeljkovic, J.; Thurnauer, M. C.
J. Synchrotron Radiat. 1999, 6, 445.
(28) D’Hennezel, O.; Ollis, D. F. J. Catal. 1997, 167, 118.
(29) Phillips, L. A.; Raupp, G. B. J. Mol. Catal. 1992, 77, 297.
(30) Dreiessen, M. D.; Goodman, A. L.; Miller, T. M.; Zaharoies, G.
A.; Grassian, V. H. J. Phys. Chem. B 1998, 102, 549.
(
31) Larson, S. A.; Falconer, J. L. Appl. Catal. B: EnViron. 1994, 4,
25.
(32) Wang, K.-H.; Tsai, H.-H.; Hsieh, Y.-H. Appl. Catal. B: EnViron.
1998, 17, 313.
3
Acknowledgment. The authors gratefully acknowledge the
following funding from the Hong Kong Research Grant Councils
(33) Yamazaki-Nishida, S.; Nagano, K. J.; Phillips, L. A.; Cervera-
(RGC-HKUST 6247/00P) and the Institute of Nano Science and
March, S.; Anderson, M. A. J. Photochem. Photobiol. A: Chem. 1993, 70,
Technology (INST). We also thank the Synchrotron Radiation
Research Center (SRRC) of Taiwan for providing the beam time
95.
(34) Pruden, A.; Ollis, D. J. Catal. 1983, 60, 404.