Available online at www.sciencedirect.

com

Applied Catalysis A: General 340 (2008) 153–161

www.elsevier.com/locate/apcata

Low temperature synthesis, characterization and substrate-dependent

photocatalytic activity of nanocrystalline TiO2 with
tailor-made rutile to anatase ratio
Zoltán Ambrus, Károly Mogyorósi, Ágnes Szalai, Tünde Alapi,
Kata Demeter, András Dombi, Pál Sipos *
Department of Inorganic and Analytical Chemistry, Research Group of Environmental Chemistry, ReGECh,
University of Szeged, PO Box 440, H-6701 Szeged, Hungary
Received 8 October 2007; received in revised form 8 February 2008; accepted 12 February 2008
Available online 17 February 2008

Abstract
TiO2 photocatalysts with predetermined phase composition (anatase-to-rutile ratio) have been prepared from strongly acidic aqueous HCl/TiCl4
mixtures with 0.1 M  [HCl]TOT  5 M and 0.05 M  [TiCl4]TOT  3 M (where the subscript TOT denotes total or analytical concentration). During
the syntheses, TiCl4 was added to a cold-HCl solution. The initially precipitate free reaction mixture was boiled, resulting in the formation of large
amount of solid, nanocrystalline (particle size 4–12 nm) TiO2 particles with large specific surface area (up to 240 m2/g). We found from the XRD
patterns that at constant [HCl]TOT, the anatase content of the crystallized solid systematically increases with decreasing [TiCl4]TOT, while at constant
[TiCl4]TOT, the rutile content was found to systematically increase with increasing [HCl]TOT. Thus the phase composition of the photocatalysts
prepared was possible to be fine-tuned with the aid of these two simple initial parameters. The photocatalytic activity of the as-prepared catalysts was
found to strongly correlate with the phase composition. In phenol decomposition the samples’ photocatalytic performance gradually and significantly
increases with their anatase content: photocatalysts containing only anatase as crystalline phase were up to three times more efficient than rutile ones.
Interestingly, in salicylic acid decomposition, rutile-only catalysts were found to show no activity at all, but some of our catalysts (both anatase-only
and rutile–anatase mixtures) at pH 3 (but not at pH 7) displayed photocatalytic activity commensurable to that of Degussa P25. This can be
explained in terms of the efficient surface chemisorption of salicylate ion on large specific surface area TiO2 particles in acidic solutions.
# 2008 Elsevier B.V. All rights reserved.

Keywords: Titania; Rutile; Anatase; Titanium tetrachloride; Hydrolysis; X-ray diffraction; Heterogeneous photocatalysis; Photocatalytic activity; Nanocrystals

1. Introduction Of the seven polymorphs of TiO2, three occur in nature: rutile

Since the first observation of water splitting with sunlight on
irradiated titania coated electrodes [1], the study of TiO2-based
semiconductor photocatalysts is one of the most active areas of
materials science, both at fundamental and applied levels [2–4].
The recently attained popularity of titania is mainly due to its
environmentally friendly (i.e., non-toxic) nature, low cost, high
natural abundance and versatile potential applications. Indeed,
the most important use of titania-based photocatalysts is
concerned with water- and air-purification, i.e., photocatalytic
destruction of hazardous organic pollutant molecules [5].
* Corresponding author. Tel.: +36 62 544 338; fax: +36 62 420 505.
E-mail address: sipos@chem.u-szeged.hu (P. Sipos).
0926-860X/$ – see front matter # 2008 Elsevier B.V. All rights reserved.
doi:10.1016/j.apcata.2008.02.010
and anatase (both tetragonal) and brookite (orthorhombic). In
photocatalysts studies, anatase and rutile have been studied most
widely, as pure brookite is cumbersome to prepare and
amorphous TiO2 has no photocatalytic activity at all. Anatase
is known to be the low temperature (300–550 8C) form of titania,
and it transforms to rutile upon heating. The transformation takes
place (depending on the primary particle size and the way of
preparation) from 600 to 1100 8C. For the photocatalytic activity,
the band gap between the valence and conducting levels is of key
importance. The band gap of rutile is 3.0 eVand that of anatase
is 3.2 eV. Rutile is known to have better visible-light response,
while anatase is usually of better photocatalytic activity. Phase
pure forms of the two common polymorphs are seldom obtained,
and the syntheses often yield various mixtures of anatase and
rutile. In photocatalysis research, for comparative purposes the
154 Z. Ambrus et al. / Applied Catalysis A: General 340 (2008) 153–161
mass produced Degussa P25 TiO2 is used (and perhaps
considered as the benchmark of photocatalysts), which is
composed of rutile and anatase from 1:4 to 1:9 ratio.
The use and performance of TiO2 for a given application
almost always depend not only on the phase composition but
also on the morphology, the size and porosity of the particles,
their specific surface area, hydrophilicity, concentration of
surface hydroxyl groups, etc. Accordingly, it is of importance to
develop synthetic methods by which these properties (or at least
some of them) can simultaneously be controlled (for a recent
review see Ref. [6]). Plenty of synthetic methods are already
published in the literature. The most common routes include the
sol–gel process using titanium alkoxides [7,8], hydrolysis of
inorganic salts such as titanium-sulphate [9], titanium
trichloride [10] or titanium tetrachloride [11–15], combustion
synthesis [16] and reverse micelle method [17]. The hydrolytic
syntheses are often coupled with hydrothermal treatment for
transforming the amorphous titania to various crystalline
products [8–10,18,19]. The majority of these syntheses require
very strict control of the experimental parameters to achieve the
desired titania properties. Furthermore, they are usually too
costly for mass-producing TiO2, due to the special instrumental
requirements or processing conditions. Recently, an inexpen-
sive and simple synthesis has been described by Chu et al. [15].
The method is based on the very good solubility of Ti(IV) in
ice-cold concentrated aqueous HCl solutions, which rapidly
decreases with the increasing temperature, resulting in the
precipitation of crystalline TiO2 upon heating. According to
Chu et al. [15], from solutions of [HCl]TOT  0.1 M and
[TiCl4]TOT  0.1 M phase pure rutile was obtained and the
particle size was found to systematically increase with
increasing [HCl]TOT. Given, that the photocatalytic activity
of the nanocrystals thus prepared has not been characterized in
Ref. [15], and rutile is known to behave as reasonably good
photocatalyst in certain conditions [19], we repeated these
syntheses in our laboratories with the aim of establishing
correlation between the photocatalytic activity and the
structural properties (phase composition, particle size, mor-
phology, specific surface area) of the photocatalysts thus
prepared. The composition of the reaction mixture was
significantly broader than was in Ref. [15] and covered
0.1 M  [HCl]TOT  5 M and 0.05 M  [TiCl4]TOT  3 M.
Somewhat surprisingly, in this concentration range not only
rutile, but also anatase and rutile anatase mixtures were isolated
and we found that with the composition of the reaction mixture
it is possible to manipulate the anatase-to-rutile ratio of the
photocatalysts. This made it possible to use this inexpensive
preparative method for producing phase-mixtures with pre-
determined composition and thus to systematically investigate
the relationship between the phase composition and the
photocatalytic activity of titania specimens. In the current
study, two model compounds, i.e., phenol and salicylic acid
were used as substrates. Phenol was chosen, because it is indeed
the most frequently studied model compound for such studies,
and it possesses several chemical properties of anthropogenic
organic pollutants in our environment. Salicylic acid, a
substrate with adsorption properties superior to phenol, was
also investigated. This made possible to compare photocatalytic
performances of our photocatalysts against ligands with
different sorption properties.
2. Experimental
2.1. Syntheses
For the syntheses, analytical grade HClcc (ca. 37 wt.%,
Merck) was used. TiCl4 (>99%) precursor was supplied by
Sigma–Aldrich and was stored at 15 8C in a refrigerator, to
minimize fuming, which is observed when TiCl4 reacts with
airborne moisture. Titanium dioxide powders (anatase and
rutile, 99.9%, Sigma–Aldrich) were used as XRD calibration
standards, and Degussa P25 (90 wt.% anatase and 10 wt.%
rutile) was employed as photocatalytic standard, all of them as
received. For the experiments Millipore Milli-Q water was used
throughout.
During the preparation, 200 mL of HCl solution with the
desired concentration was placed in a three-necked bottle. The
TiCl4 was kept in a dropping funnel equipped with a CaCl2 pipe
to exclude moisture. All the quickfit connections were
lubricated with H2SO4cc. The reaction mixture was kept in
an ice-water bath, and was vigorously stirred during the drop
wise (ca. 0.5 mL min1) addition of the calculated amount of
TiCl4. At the place of dropping, a yellowish precipitate was
formed, which then instantaneously dissolved. At this point of
the preparation, all the systems in the concentration range
investigated were transparent and free of any precipitate. The
solution was stirred for 15 min. The dropping funnel was
changed to a condenser and the mixture was heated to 95–
100 8C and then refluxed for 3 h. By the end of the heating a
white suspension was formed which was then allowed to cool.
From this point the synthesis resolves to four slightly different
sub-routes (Scheme 1) Synthesis 1 involved ageing of the
reaction mixture under vigorous stirring for 24 h at room
temperature (unless otherwise stated), followed by evaporation.
In the other methods, after decantation, washing with 0.1 M
HCl and centrifugation, the suspension was exhaustively
dialyzed in a 12,000 Da cut-off limit dialysis sack (Medicell
International Ltd., London) against distilled water (until the pH
of the dialyzing solution reached 5.5) and either evaporated
(Synthesis 2, which is practically identical to that described in
Ref. [15] or freeze-dried at 52 8C and 60 Pa (Synthesis 3).
Synthesis 4. is identical to Synthesis 3, except that aging of the
suspension was carried out in a Teflon lined, closed autoclave,
at 110 8C and for 24 h.
The solid photocatalysts thus obtained were milled in an
agate mortar and kept in closed screw top containers.
2.2. Photocatalyst characterization
X-ray diffraction (XRD) measurements were performed on a
Philips PW 1820/1830 diffractometer (Cu Ka = 0.15406 nm,
40 kV, and 30 mA, in the 208  2Q  408 regime for solid
powder samples). The average diameters of the particles, (d)
were obtained by means of the Scherrer equation [20] [d = ka/
 Z. Ambrus et al. / Applied Catalysis A: General 340 (2008) 153–161
Scheme 1. Schematic representation of the preparative pathways used for synthesizing TiO2 photocatalysts.
(b cos Q), where b = line broadening (b = bS  b0, where bS
and b0 are the half-widths of the XRD peak of the sample and of
the silicon standard), k is related to the crystallite shape (k = 0.9
for spherical particles), and a and Q are the radiation
wavelength and Bragg angle, respectively]. A single-crystal
silicon standard (b0 = 0.105 2Q8) was used for calibration. In
order to characterize the phase composition of the powders,
calibration curves of the anatase and rutile phases were
determined. Mixtures of rutile and anatase were prepared at
different (0–100, 20–80, 40–60, 60–40, 80–20 and 100–
0 wt.%) ratios. The diffraction peaks recorded (2Q = 25.38 for
anatase (1 0 1) and at 2Q = 27.58 for rutile (1 1 0)) were
integrated and their area values were plotted versus the anatase
and rutile content. The anatase and rutile contents were
calculated from the area values of the deconvoluted peaks of the
samples. The anatase peaks were corrected for the overlapping
brookite contribution (where appropriate) on the basis of Ref.
[21].
155
Transmission electron microscopic (TEM) measurements
were performed to observe the morphology of the photo-
catalysts. TEM micrographs were recorded on a Phillips CM 10
instrument using formvar coated copper grids.
A MOM Derivatograph Q-1500D instrument was used for
simultaneous thermal analysis combining thermogravimetry
(TG), and differential thermoanalysis (DTA).
The specific surface area of the catalysts was determined by
nitrogen adsorption at 77 K by a Micromeritics gas adsorption
analyzer (Gemini Type 2375), after vacuum drying of the
samples at 100 8C for 12 h. The specific surface area was
calculated using the BET method.
The performance of the catalysts prepared was characterized
by using the photocatalytic decomposition of two model
compounds: phenol and salicylic acid. During these experi-
ments, the photoreactor (100 mL) was an open tube with double
glass walls, surrounded by a thermostating jacket
(25.0  0.1 8C). The continuously stirred reactor was sur-
156 Z. Ambrus et al. / Applied Catalysis A: General 340 (2008) 153–161

rounded and irradiated by six, fluorescent Hg-vapor lamps (6 W
power, radiation maximum at 365 nm, intensity of UV light was
I = 1.02  0.01  105 einstein dm3 s1 measured utilizing
ferrioxalate actinometry). During the reaction, air was
continuously bubbled through the reaction mixture at a
constant flow rate of 500 mL min1. The pH of the reaction
mixtures were measured before and after the reaction by using
glass electrode and was found to be 4.5–6.0 for phenol and 2.8–
3.3 for salicylic acid containing solutions. For salicylic acid
decomposition tests in neutral solutions, the pH was maintained
at 7 via using an HPO42/H2PO4 buffer (I = 0.02 M). The
concentration of phenol was measured with an HPLC system
consisting of a Merck–Hitachi L-7100 low-pressure gradient
pump equipped with a Merck–Hitachi L-4250 UV–vis detector
(l = 210 nm) and a Lichrospher RP 18 column applying
methanol/water (for phenol) and methanol/water–acetic acid
(for salicylic acid) mixture as eluent. The initial concentrations
of the model compounds were 5.0  104 M (phenol) and
103 M (salicylic acid) and 1 g dm3 photocatalyst concentra-
tion was used throughout. Samples were taken from the
reaction mixture in equal time intervals, for 100–180 min.
Further details of the experimental set-up were described
previously [22–24]. The initial rate of the photocatalytic
degradation of the model compounds, r0 was considered to be
the measure of the efficiency of a given photocatalyst. To
determine r0, a quadratic empirical function was fitted to the
experimentally observed c = f(t) data points at the conversion
range of <60% (in some cases <80%). If the fitting yielded
satisfactory description for the data (i.e., R2  0.99 coefficient
of correlation), then the slope of f(t) at t = 0 was considered to
represent the initial rate of the photocatalytic reaction. This
procedure was found to result in r0 values of comparability
superior to those obtained either from linear or from
exponential fitting of the c = f(t) function (the former almost
always overestimates, while latter underestimates the value of
r0 [25]).
3. Results and discussion
3.1. Optimization of synthesis parameters
We found, that in Syntheses 2–4, the several times repeated
decantation and centrifugation resulted in very low yields. To
avoid this, in Synthesis 1, the solid samples were obtained via
direct evaporation of the mother liquor (after only one
decantation, as TiO2 was found to settle rapidly from the
mother liquor after the 24 h ageing period). From the structural
parameters of the samples obtained during Syntheses 1–4 under
identical synthesis conditions before precipitation (i.e., samples
6, 16, 17 and 18 in Table 1), the crystalline phase is always a
mixture of anatase and rutile. (The rest is likely to be a mixture
of amorphous TiO2, some brookite and rutile/anatase crystal-
lites which are too small to be observed by X-ray
diffractometry.) Apparently, the anatase and rutile particle
sizes are also independent of the treatments after precipitation.
It is, however, striking, that the products from Syntheses 2–4
are significantly richer in rutile than in anatase, as opposed to
the product from Synthesis 1. As it is highly unlikely, that the
anatase-to-rutile ratio is changed in some sort of phase
transformation during the synthetic steps after precipitation,

Table 1
Structural data and photocatalytic activity of nanocrystalline TiO2 photocatalysts prepared at various [HCl]TOT/[TiCl4]TOT reactant ratios
Sample Synth route a [HCl]TOT/ A (wt.%) R dA dR Bb r0c aSBET d
[TiCl4]TOT (M/M) (wt.%) (nm) (nm) (108 M s1) (m2 g1)
1 1 0.1/0.05 61 0 4.5  + 2.54 225
2 1 0.1/0.1 45 0 4.5  + 2.41 211
3 1 0.1/0.5 49 9 4.5 7.3 + 1.92 238
4 1 0.1/1.5 0 51  5.0  0.96 77
5 1 1.0/0.05 27 17 11.8 10.4 + 1.64 61
6 1 1.0/0.1 45 30 7.7 9.9 + 2.48 73
7 1 1.0/0.5 23 46 5.0 6.0 + 1.70 125
8 1 1.5/1.5 0 48  4.7  0.68
9 1 2.0/0.05 41 40 6.0 7.0 + 1.00
10 1 2.0/0.1 8 36 7.8 6.7 + 0.63 80
11 1 2.0/0.5e 9 67 7.1 5.6 + 1.94
12 1 2.0/1.5 0 51  5.1  0.75
13 1 2.0/3.0 0 49  4.5  0.67
14 1 4.0/0.1 0 80  4.8  0.55 46
15 1 5.0/1.5 0 54  4.3  0.96
16 2 1.0/0.1 19 40 6.7 7.8 + 1.76
17 3 1.0/0.1 6 55 8.7 11.5 + 1.36
18 4 1.0/0.1 18 38 7.9 8.4 + 1.76
A: anatase, R: rutile; B: brookite; dA and dR are the average diameters of anatase and rutile particles, respectively; estimated uncertainties: A/R contribution: 10%;
particle size: 10%; r0  10%; aSBET : 5%.
a
Synthetic route applied, according to Scheme 1.
b
Brookite is detectable (+) or not detectable () in the photocatalyst.
c
Initial decomposition rate of phenol.
d
Specific surface area.
e
Conditions similar to those used in Ref. [26].
 Z. Ambrus et al. / Applied Catalysis A: General 340 (2008) 153–161 157

therefore the most plausible explanation is, that mostly anatase
is lost during decantation/centrifugation (and, less probably,
during dialysis). This effect might well be associated with the
difference of the densities of anatase (3.84 g cm3) and rutile
(4.26 g cm3). This observation could also explain why rutile
was the major phase found by Chu et al. [15]. Regarding the
photocatalytic activity of these samples, the product from
Synthesis 1 was found to be significantly better in phenol
decomposition, than the other samples. In the rest of the current
work, for preparing the TiO2 photocatalysts the simplest
preparative way (Synthesis 1) was used throughout.
3.2. Structural characterization of the products
The relative contributions of rutile and anatase to the X-ray
detectable (crystalline) fraction (Table 1) appears to strongly
depend on the [HCl]TOT and [TiCl4]TOT of the reaction mixture.
High HCl and TiCl4 concentrations promoted the formation of
rutile: from XRD, only rutile was detected in samples obtained
from [HCl]TOT  4 M [TiCl4]TOT  1.5 M solutions. Conver-
sely, at low concentration of HCl (0.1 M) and TiCl4 (0.1 M)
the formation of anatase is more favored. Perhaps more
importantly, mixed phases were obtained from solutions of
intermediate concentrations (Fig. 1). At a given [HCl]TOT,
increasing the [TiCl4]TOT resulted in gradually decreasing
anatase and increasing rutile content. On the other hand, at a
given [TiCl4]TOT and with increasing [HCl]TOT, the rutile
content was found to systematically increase at the expense of
anatase.
This effect has recently been noticed by Testino et al. [26]
under experimental conditions very similar to those used by us.
They speculated, that initially (the thermodynamically less
stable) anatase crystals are formed in a process that is
independent of [TiCl4]TOT or [HCl]TOT. However, in strongly
acidic pH, the solution species Ti(OH)22+ and Ti(OH)3+
predominate, which are suggested by these authors to transform
preexisting anatase to rutile. Testino et al. [26] also argues that
the evolution of phase composition is strongly dependent not
only on the pH, but also on the final temperature and reaction
time. In our experimental protocol, latter two parameters were
kept constant, therefore the phase compositions are affected
only by the pH.
Regarding rutile formation at high acid concentrations, our
findings confirm those of Chu et al. [15] and Wu et al. [27].
However, formation of anatase was not observed in Ref. [15]
under apparently identical experimental conditions to ours

Fig. 1. XRD patterns of as-prepared TiO2 photocatalysts obtained from aqueous HCl/TiCl4 mixtures (using Synthesis 1 in Scheme 1): (a) and (b) constant [HCl]TOT
and gradually increasing [TiCl4]TOT; (c) and (d) constant [TiCl4]TOT and gradually increasing [HCl]TOT.
158 Z. Ambrus et al. / Applied Catalysis A: General 340 (2008) 153–161

(e.g., in reaction mixture of [HCl]TOT = 2 M and [TiCl4]-

TOT = 0.1 M).
In some samples, from the appearance of its (1 2 1)
reflection around 2Q  30.88, the presence of small amounts
of brookite was also detected. From the data of Table 1,
brookite is always formed together with anatase, and is never
detectable in rutile-only products.
The average particle sizes of rutile and anatase were
calculated from the Debye–Scherrer equation and were always
found in the range of 4–12 nm. In general, the particle sizes of
the as prepared samples appear to slightly depend on the
[TiCl4]TOT of the reaction mixture: larger precursor concentra-
tions result in smaller crystallite diameters. This effect is,
however, at the edge of detectability. It is to be noted here, that
Chu et al. [15] observations are opposite to ours : on their TEM
images the largest crystallites were observed at the highest
precursor concentrations.
Representative TEM images of the solid products obtained
are shown in Fig. 2. Predominantly, shuttle-like and
polyhedral (cuboid) morphologies have been observed. In
some exceptional instances, elongated (needle like) features
were also found. Given, that the Scherrer equation with
k = 0.9 can only be used for spherical particles, the particle
sizes obtained from the XRD line broadening can be
somewhat distorted. The shape of the particles, as seen
under the TEM, were found to be more or less independent of
the actual phase composition, i.e., both major crystal forms
might be present in any of these morphologies. In Ref. [15]
the shuttle-like morphology was mainly observed for rutile-
only samples, in our hands this was also seen in the anatase-
only samples’ TEM images too.

3.3. Photocatalytic activity of the products in phenol

decomposition

From the systematic investigation of the photocatalytic

performance of the samples prepared, their photocatalytic
performance in phenol decomposition and under the experi-
mental conditions employed (e.g., pH 5–6 and without added
supporting electrolyte) are inferior to that of Degussa P25 (for
which the initial rate of phenol decomposition from a large
number of independent measurements was found to be
r0 = 12  2  108 M s1). Systematic variations are seen
between r0 and the phase composition (Fig. 3 and Table 1).
With increasing anatase content, the photocatalytic activity
increases, and the largest r0 values were observed for anatase-
only, while the smallest ones for rutile-only photocatalysts. On
the other hand, the photocatalytic activity seems to be less for
samples with large X-ray amorphous fraction, and vice versa Fig. 2. TEM images of the photocatalyst nanocrystals with different shapes: (a)
(cf. r0 for samples 1 and 10). shuttle-like morphology (sample 6); (b) polyhedral morphology (sample 13);
No univariate correlation between the specific surface area (c) needle-like morphology (sample 11). Note that needle like morphology was
and photocatalytic performance in phenol decomposition was observed only in some exceptional cases, and predominantly shuttle-like and
polyhedral nanocrystals were seen.
observed. For example, r0 of 1 and 6 are identical, but have
widely different specific surface area, or samples 6 and 10 have The specimens which, in terms of anatase/rutile ratio,
identical specific surface area but quite different r0. This approach Degussa P25 (e.g., samples 3 and 5), are significantly
implies, that large surface area in itself does not yield enhanced less photoactive than any of the anatase-only samples (Table 1).
photocatalytic activity in phenol decomposition. It is suggested in the literature (and in some way supported by
 Z. Ambrus et al. / Applied Catalysis A: General 340 (2008) 153–161 159

particle size and thus improve photocatalytic activity of the

products, further post-precipitation steps were included in the
preparation (Table 2). In the first, boiling of the reaction
mixture was continued for a longer period (24 and 72 h) than
described in the original recipe (3 h). This treatment did not
significantly increase the anatase or rutile content and their
ratios, however, it resulted in a small but significant increase in
particle size. This was found not to cause any major
improvement in the photocatalytic activity.
As expected, calcination (in air stream, 2 8C min1 heating
rate up to T = 400 8C, and the photocatalyst was held at this
temperature for 4 h) yielded a significant (20–40%) increase in
the particle size but left the crystalline phase content unaltered
for all of the specimens investigated. This is somewhat
surprising, as amorphous TiO2 would be expected to transform
Fig. 3. Photocatalytic decomposition curves of phenol in presence of various to anatase around this calcination temperature. The X-ray
photocatalysts: sample 4 (rutile-only); sample 1 (anatase-only) and sample 3 amorphous fraction of the samples (or at least part of it) might
(anatase–rutile mixture) in non-buffered solution (pH 4.5–5.0) and without
consist of X-ray amorphous anatase and rutile (i.e., nanocrys-
added supporting electrolyte.
tals with particle size that is too small to be detected by X-ray
diffractometry). However, from TG/DTA measurements
the mixed composition of Degussa P25), that a synergistic (Fig. 4) we observed more than 20 wt.% loss for sample 3
effect exists between rutile and anatase. It has been suggested during heating, which is likely to be due to presence of water in
by Ohno et al. [28], that anatase and rutile particles are in hydrated/amorphous TiO2. This step appears on the DTA curve
contact under the conditions of the photocatalytic reactions, as a strong endothermic peak between 30 and 250 8C, followed
which leads to a synergistic effect. This is why the by an exothermic step (anatase formation) in the range of 200–
photocatalytic activity was suggested to be more pronounced 500 8C. For determining the end temperature of the crystal-
in mixed phases than in phase-pure titania samples [12,29,30]. lization of anatase, the TG-DTA curves were recorded at
However, for the photocatalysts prepared by us, this seems not various heating rates (2.5, 5.0 and 10.0 8C min1). The
to be the case. exothermic DTA peak tailed to the baseline at about 475,
490 and 500 8C, respectively (Fig. 4 represents the last case).
3.4. Post-precipitation treatments of the products Therefore 400 8C calcination temperature is likely to be too low
to transform the total amorphous titania content in our
As it was shown in Section 3.1, during the synthesis the photocatalysts to anatase. If calcination is performed at
critical preparative step which determines the phase composi- 500 8C (sample 6B1C2 in Table 2), the particle size is further
tion is the precipitation. In order to modify crystallinity and increased. The contribution of anatase and rutile also larger, and
(somewhat surprisingly) the sum of their contribution is ca.
Table 2 70 wt.%. The rest is most probably brookite. The photocatalytic
Effect of post-synthetic treatments on the structure and photocatalytic activity activity in phenol decomposition was also found to be increased
TiO2 photocatalysts prepared at [HCl]TOT = 1.0 M and [TiCl4]TOT = 0.1 M (in fact, the highest r0 was found within the series studied, see
Synth. A R dA dR r0b Table 1).
route a (wt.%) (wt.%) (nm) (nm) (108 M s1)
6 1 27 30 7.7 9.9 2.48
6B1 1 36 11 9.8 8.3 3.30
6B2 1 37 13 10.2 10.7 2.46
6C1 1 29 35 13.8 19.2 2.69
6B1C1 1 34 7 12.9 17.1 3.52
6B1C2 1 58 11 17.4 29.2 4.24
16 2 19 40 6.7 7.8 1.76
16C1 2 20 44 10.5 11.0 3.14
17 3 6 55 8.7 11.5 1.36
17C1 3 7 62 9.1 13.9 2.35
18 4 18 38 7.9 8.4 1.76
18C1 4 20 42 10.4 11.3 1.64
a
Synthetic route applied, according to Scheme 1.
b
Initial decomposition rate of phenol. B1 and B2 denote boiling the reaction
mixture for 24 and 72 h after precipitation, respectively. C1 and C2 denotes
calcination at 400 and 500 8C, respectively (other shorthands are identical to
those used in Table 1). Specific surface areas of samples 6, 6C1 and 6B1C2 were Fig. 4. Thermoanalytical curves (TG and DTA) of sample 3 determined in still
found to be 73, 35 and 62 m2 g1, respectively. air at 10 8C min1 heating rate.
160 Z. Ambrus et al. / Applied Catalysis A: General 340 (2008) 153–161
3.5. Substrate dependence of the photocatalytic activity
The majority of the studies published in the literature
focuses on the characterization of a given type of photocatalyst
only for a single (or in some exceptional cases for a couple of)
model compounds. However, the direct comparison of the
photoactivity of a series of photocatalysts against a given
substrate could be somewhat misleading, as photoactivity is
strongly substrate dependent. If the photodecomposition
involves molecules that are adsorbed on the surface of the
photocatalyst, than TiO2 nanoparticles with large surface area
(i.e., large adsorption capacity) could be more suitable for the
photocatalytic degradation of well-adsorbing species (such as
salicylic acid [31,32]). On the other hand, larger crystallites are
likely to be more suitable for degrading even ill-adsorbing
substrates (such as phenol) through formation of reactive OH
radicals.
Given, that the photocatalysts described in the present work
(Table 1) are of significantly larger specific surface area than
Degussa P25 (aSBET ¼ 50 m2 g1 ), it was hoped, that, relative to
Fig. 5. Photocatalytic decomposition curves of salicylic acid in presence of
various photocatalysts (rutile-only: sample 4; anatase-only: sample 1 and
anatase–rutile mixture: sample 3), at various pH (a) in non-buffered solution
(pH 2.8–3.3); (b) in buffered solution (pH 7, phosphate buffer). In (a) rutile-
only curve, the initial increase in the equilibrium concentration of the substrate
is likely to be due to photodesorption.
Degussa P25, their photocatalytic performance will be better for
decomposition of salicylic acid. The results of these experiments
are summarized in Fig. 5. These experiments were performed
both on the natural (acidic) pH of the salicylic acid solutions (pH
2.8–3.3) and at pH 7 (using a Na2HPO4/NaH2PO4 buffer and
I = 0.02 M ionic strength). The initial analytical concentration of
the substrate was chosen to be 103 M (to compensate for the
adsorption occurring in acidic solutions, see below). Three
representative titania photocatalysts were selected for these
investigations, to elucidate the activity-phase composition
relationships: a rutile-only (sample 4), an anatase-only (sample
1) and an anatase–rutile mixture (sample 3).
It is seen from the decomposition curves obtained at pH 7,
that our photocatalysts are still inferior to Degussa P25, and the
photocatalytic activity of the samples is the same as for phenol
decomposition, i.e., rutile-only < rutile–anatase mixture -
< anatase-only Degussa P25. Therefore it is likely, that
the mechanism or the rate determining step of the photo-
catalytic decomposition of both substrates is very similar under
these experimental conditions. Moreover, as the initial
equilibrium concentration of both substrates is equal to the
analytical concentration at t = 0, therefore their adsorption on
the catalyst surfaces under the experimental conditions
employed is negligible.
From the decomposition curves of salicylic acid obtained in
acidic solutions, a radically different picture emerges. A
significant adsorption is seen at t = 0 in the dark for each
photocatalysts. 5–27% of salicylic acid was seen to adsorb on
the surface of the photocatalysts in the dark (note, that no
adsorption was seen for salicylic acid at pH 7 and for phenol
at pH 5). The extent of adsorption was found to strongly
correlate with the specific surface area of the photocatalysts
(Table 1) and follows the order of Degussa P25 < rutile-
only < anatase–rutile mixture  anatase-only. The adsorption
also manifests itself in the yellow discoloration of the
suspensions, which is due to the formation of the salicylato-
surface complexes. For the rutile-only photocatalyst (sample
4), the equilibrium concentration of the substrate seems to
increase initially in time, this is most likely to be due to the
photodesorption. After this initial increase, the substrate
concentration hardly decreases in time, demonstrating that
the rutile-only photocatalyst has practically no photocatalytic
activity in salicylic acid decomposition at acidic pH. Of the rest
of the catalysts, the degradation rate is the largest for
the anatase–rutile mixture, larger than that for Degussa P25.
The most adsorbing anatase-only sample also displayed
reasonable photocatalytic performance, commensurable with
that of Degussa P25.
These observations can be interpreted in the following way.
The photocatalytic decomposition of organic substrates in
aqueous solutions is thought to occur mainly in two parallel
processes: (i) the dissolved oxygen might work as an electron
acceptor generating OH radicals and these oxidize the
molecules in the proximity of the semiconductor surface or
(ii) the adsorbed molecules could be directly oxidized by the
photocatalysts’s photoexcited holes. There are two main factors
determining which one is being favored, i.e., the polarity of the
 Z. Ambrus et al. / Applied Catalysis A: General 340 (2008) 153–161
molecules or their adsorption affinity to the semiconductor
surface and the quality of the catalyst surface.
Degussa P25 TiO2 is one of the best commercially available
photocatalyst synthesized in flame-hydrolysis. Its performance
can hardly exceeded by sol–gel synthesized photocatalysts for
phenol decomposition. However in a non-buffered suspension
Degussa P25 had a much lower activity for the decomposition
of salicylic acid relative to its performance in neutral solution.
One can hypothesize, that the chemisorbed salicylic acid might
disable oxygen to adsorb onto the surface of titanium dioxide
and therefore it cannot play its important role as an electron
acceptor. Therefore at higher pH, at which the chemisorption is
not favored due to the negative surface charge of the
photocatalysts and the formation of salicylate anions, the
photocatalytic performance of Degussa P25 increases. This
effect is also noticed on titania nanoparticles, although their
activity was affected to a lower extent or even in reverse order
(see the ratio between r0–s sample 3 comparing the decay
curves at pH 3 and 7, it decreased instead of increasing). The
most likely explanation is that these photocatalysts decompose
the adsorbed substrate in larger quantities by direct photo-
oxidation on their large surface. For that reason, the efficiency
of TiO2 nanoparticles exceeded the efficiency of Degussa P25
in non-buffered (pH 3) salicylic acid suspensions. From these
results it can be concluded that for efficient photocatalytic
decomposition of organic substrates, the sol–gel synthesized
nanoparticles can be recommended in practical situations
where the adsorption of the substrate is expected to be
significant.
4. Conclusion
TiO2 photocatalysts prepared from strongly acidic aqueous
HCl/TiCl4 mixtures with 0.1 M  [HCl]TOT  5 M and
0.05 M  [TiCl4]TOT  3 M were found to consist of anatase,
rutile or the mixture of the two, depending on the initial reaction
parameters. At a given [HCl]TOT (const.) the anatase content
systematically increased with decreasing [TiCl4]TOT. On the
other hand, at a given [TiCl4]TOT (const.), the rutile content was
found to systematically increase with increasing [HCl]TOT.
Particle sizes (assuming spherical particle shape) were
determined from the Debye–Scherrer equation and were found
to change between 4 and 12 nm, this is supposedly an
underestimation of the real sizes, as in some cases elongated
particles are seen on the TEM pictures.
The photocatalytic activity of the TiO2 catalysts thus
prepared strongly correlate with the phase composition. In
phenol decomposition the samples’ photocatalytic performance
correlates well with their anatase content: photocatalysts
containing only anatase as crystalline phase were up to three
times more efficient than rutile-only ones. The amorphous
phase content appeared to exert an adverse effect on
photoactivity. Various post-synthetic treatments were
employed to modify the photocatalysts’ structure and thus to
improve photocatalytic performance, calcination at 500 8C
appeared to be the most efficient. The photocatalytic activity
was proven to be substrate dependent. In salicylic acid
161
decomposition, unlike anatase-only photocatalysts, rutile-only
ones found to show no activity at all. However, some of our
catalysts displayed significantly better photocatalytic activity
than Degussa P25 at pH 3 (but not at pH 7). This can be
explained in terms of the efficient surface chemisorption of
salicylate ion on large specific surface area TiO2 particles under
acidic solutions.
Acknowledgements
This work was financially supported by grants from the
Hungarian National Office pf Research and Technology (RET-
07/2005) and from the Hungarian Research Foundation (OTKA
67559). KM thanks the Magyary Zoltán Foundation for
financial support.
References
[1] A. Fujishima, K. Honda, Nature 238 (1972) 37.
[2] A. Fujishima, T.N. Rao, D. Tryk, J. Photochem. Photobiol. C: Photochem.
Rev. 1 (2000) 1.
[3] B. Oregan, M. Grätzel, Nature 353 (1991) 737.
[4] W. Wang, B. Gu, L. Liang, W.A. Hamilton, D.J. Weselowski, J. Phys.
Chem. B 108 (2004) 14789.
[5] U. Diebold, Surf. Sci. Rep. 48 (2003) 53.
[6] O. Carp, C.L. Huisman, A. Reller, Prog. Solid State Chem. 32 (2004) 33.
[7] J. Slunecko, M. Kosec, J. Holc, G. Drazic, J. Am. Ceram. Soc. 81 (1998)
1121.
[8] S.S. Watson, D. Beydoun, J.A. Scott, R. Amal, Chem. Eng. J. 95 (2003)
213.
[9] Y. Zheng, E. Shi, S. Cui, W. Li, X. Hu, J. Am. Ceram. Soc. 83 (2000) 2634.
[10] E. Hosono, S. Fujihara, K. Kakiuchi, H. Imai, J. Am. Ceram. Soc. 126
(2004) 7791.
[11] K.M. Reddy, D. Guin, S.V. Manorama, J. Mater. Res. 19 (2004) 2567.
[12] Q. Zhang, L. Gao, J. Guo, Appl. Catal. B: Environ. 26 (2000) 207.
[13] Y. Li, Y. Fan, Y. Chen, J. Mater. Chem. 12 (2001) 1387.
[14] A. Pottier, C. Chaneac, E. Tronc, L. Mazerolles, J.-P. Jolivet, J. Mater.
Chem. 11 (2001) 1116.
[15] R. Chu, J. Jan, S. Lian, Y. Wang, F. Yan, D. Chen, Solid State Commun.
130 (2004) 789.
[16] R. Mueller, L. Madler, S.E. Pratsinis, Chem. Eng. Sci. 58 (2003) 1969.
[17] D. Zhang, L. Qi, J. Ma, H. Cheng, J. Mater. Chem. 12 (2002) 3677.
[18] W. Wang, B. Gu, L. Liang, W.A. hamilton, D.J. Weselowski, J. Phys.
Chem. B 108 (2004) 14789.
[19] S. Baldassari, S. Komarneni, E. Mariani, C. Villa, Mater. Res. Bull. 40
(2005) 2014.
[20] B.D. Cullity, Elements of X-ray Diffraction, Addison-Wesley, Reading,
1978.
[21] H. Zhang, J.F. Banfield, J. Phys. Chem. B 104 (2000) 3481.
[22] I. Ilisz, Z. László, A. Dombi, Appl. Catal. A: Gen. 180 (1999) 25.
[23] I. Ilisz, A. Dombi, Appl. Catal. A: Gen. 180 (1999) 35.
[24] A. Dombi, I. Ilisz, Z. László, G. Wittmann, Ozone Sci. Eng. 24 (2002) 49.
[25] P. Sipos, K. Mogyorósi, K. Demeter, unpublished observations.
[26] A. Testino, I.R. Bellobono, V. Buscaglia, C. Canevali, M. D’Arienzo, S.
Polizzi, R. Scotti, F. Morazzoni, J. Am. Chem. Soc. 129 (2007) 3564.
[27] M.M. Wu, J.B. Long, A.H. Huang, Y.J. Luo, S.H. Feng, R.R. Xu,
Langmuir 15 (1999) 8822.
[28] T. Ohno, K. Sarukawa, K. Tokieda, M. Matsumura, J. Catal. 203 (2001)
82.
[29] Q. Xiao, Z. Si, Z. Yu, G. Qiu, Mater. Sci. Eng. B 137 (2007) 189.
[30] Q. Xiao, Z. Si, Z. Yu, G. Qiu, J. Alloys Compd. 450 (2008) 426.
[31] G. Dagan, M. Tomkiewicz, J. Phys. Chem. 97 (1993) 12651.
[32] M. Tomkiewicz, S. Kelly, in: E. Pelizzetti (Ed.), Fine Particles Science and
Technology, Kluwer Academic, The Netherlands, 1996, p. 403.