Journal of Photochemistry and Photobiology A: Chemistry 129 (1999) 8999

Photocatalytic properties of ZrO2 and Fe/ZrO2 semiconductors

prepared by a solgel technique
Silvia G. Botta a , Jos A. Navo b , Mara C. Hidalgo b , Gloria M. Restrepo b , Marta I. Litter a,
a

Unidad de Actividad Qumica, Comisin Nacional de Energa Atmica, Centro Atmico Constituyentes,
Av. Gral. Paz 1499, (1650) San Martn, Prov. de Buenos Aires, Argentina
Instituto de Ciencia de Materiales de Sevilla, Centro Mixto CSIC-Universidad de Sevilla, Av. Amrico Vespucio, s/n,
Isla de Ia Cartuja, 41092-Sevilla, Spain
Received 5 July 1999; accepted 12 July 1999

Abstract
The photocatalytic efficiency of pure and 0.520 wt.% Fe-containing ZrO2 samples, prepared by a solgel technique, was tested in the
transformation of environmentally important substrates, such as nitrite, EDTA and Cr(VI) and compared with that of TiO2 (Degussa P-25).
All samples were active, although the efficiency was lower than that of TiO2 . Nevertheless, the presence of iron increased the activity for
nitrite photooxidation, with the maximum efficiency at 5 wt.% Fe content. At higher Fe content, the efficiency decreased but it was always
higher than that of the undoped sample. For EDTA oxidation and Cr(VI) reduction, the same trend was observed. Remarkably, Cr(VI)
reduction in the presence of EDTA was strongly accelerated, all zirconia samples being as active as P-25, although initial rates were lower
and somewhat detrimentally affected by the presence of iron. The photocatalytic activity depends on the structural, surface and optical
properties of the sample, on the preparation conditions and on the nature of the photocatalytic reaction. 1999 Elsevier Science S.A. All
rights reserved.
Keywords: Heterogeneous photocatalysis; Zirconia; Titania; Nitrite; EDTA; Cr(VI)

1. Introduction
Zirconium dioxide (ZrO2 ) is an important material widely
used in ceramics technology [1] and in heterogeneous catalysis [2,3]. Due to its nature as n-type semiconductor, it has
been considered recently as a photocatalyst in photochemical
heterogeneous reactions. The reported values of the energy
of the bandgap (Eg ) of this oxide range between 3.25 and
5.1 eV [412], depending on the preparation technique of the
sample. From these, the most frequent and accepted value is
5.0 eV, with a conduction band potential of 1.0 V vs. NHE
at pH 0 [13]. Accordingly, the corresponding value of the valence band potential is +4.0 V vs. NHE. The relatively wide
Eg value and the high negative value of the conduction band
potential allowed its use as a photocatalyst in the production of hydrogen through water decomposition [10,13,14].
Other photocatalytic reactions performed with zirconia were
the oxidation of 2-propanol to acetone [12,15], the oxidation

Corresponding author
E-mail address: [email protected] (M.I. Litter)

of propene [4,16] and ethane [9], the photodegradation of

4-chlorophenol [17], 4-nitrophenol [18] and 1,4-pentanediol
[19], the oxidation of CO [9,20] and the reduction of HCO3
to CO [13,14]. Also, it was found that ZrO2 can photocatalyze the oxygen isotopic exchange [9,21]. Kohno et al.
used ZrO2 to photocatalytically reduce CO2 with hydrogen
[22,23] and with methane [24]. In photooxidative reactions,
the activity of ZrO2 samples is generally much lower than
that of TiO2 . Although ZrO2 presents an absorption maximum around 250 nm, some samples show a non-negligible
absorption in the near UV range (290390 nm) [9,11,12,18];
moreover, some photocatalytic reactions could be performed
under irradiation in this range, without the participation of
light of higher energy [12,15,16,18,21,25]. Sato and Kadowaki attributed the photocatalytic activity of the ZrO2 sample used in their work to an absorption band arising of an
impurity level at 320 nm [9].
It is well known that some dopants included in a semiconductor matrix, such as Fe(III), inhibit the recombination of
e /h+ pairs, enhancing the photocatalytic efficiency of the
processes, an effect sensitive to the amount of dopant [25].
Although the systems Fe-O/TiO2 have been object of studies

1010-6030/99/$ see front matter 1999 Elsevier Science S.A. All rights reserved.
PII: S 1 0 1 0 - 6 0 3 0 ( 9 9 ) 0 0 1 5 0 - 1

90

S.G. Botta et al. / Journal of Photochemistry and Photobiology A: Chemistry 129 (1999) 8999

Table 1
Properties of the oxide samples
Sample
Sample

Fe
(wt.%)

Calcination
time (h)

SBET
(m2 g1 )

Crystal phases

IEP

monoclinic
tetragonal (tr.)
monoclinic
tetragonal (tr.)
monoclinic (0.7)
tetragonal (0.3)
tetragonal
tetragonal
tetragonal
tetragonal

5.3

5 to 50

3.7

NDc

5 to 80

3.8

NDc

5 to 50

3.1

5.3
NDc
NDc
NDc

5 to 50
l0 to 200
5 to 90
5 to 60

2.4
2.3
2.3
2.2

0.5

500

24

30

0600

600

0.5500

0.5

0.5

500

24

15

0.5
3
3
3

500
600
500
500

24
3
24
24

15
22
47
56

5
5
10
20

Eg b
(eV)

Calcination
temperature ( C)

0500

5500
5600
10500
20500

Particle sizea
(m)

Digestion
time (h)

Determined by SEM.
Determined from the diffuse reflectance spectra.
c ND = Not determined.
b

in the field of the heterogeneous photocatalysis [2634],

the photocatalytic properties of the Fe-O/ZrO2 systems have
been not described.
Nitrite is a common pollutant in the environment, and the
oxidation to NO3 can take place photocatalytically in the
troposphere by the action of semiconductor particles suspended in cloud water. In this way, nitrite can be one of the
species responsible for the acid rain problem. Some results
on the heterogeneous photocatalytic oxidation of nitrite using different semiconductors, especially TiO2 , are reported
in the literature (see Refs. [28,31,32] and references therein).
Chromium(VI) is also a common contaminant in industrial wastewaters coming from electroplating, leather tanning or paint-making; it is toxic, carcinogenic and has been
regulated in many countries. Remediation of this pollutant
consists generally in its transformation to the less harmful
Cr(III), which can be eliminated then by precipitation in
neutral or alkaline solutions. The photocatalytic reduction of
Cr(VI) with semiconductors such as TiO2 , ZnO, CdS, ZnS
and WO3 has been widely studied (see Refs. [31,32,34,35]
and references therein). EDTA is a common pollutant coming from industrial and domestic wastes, extensively used in
detergents, fertilizers, herbicides, etc. Due to its effectiveness as a metallic complexant, it takes part in decontaminating and cleaning mixtures of boilers and components of
nuclear reactors. As EDTA is not easily biodegradable or
degradable by chlorine, its elimination has been attempted
using activated carbon filters, ozonization and UV/H2 O2 ,
with variable results (see Ref. [34] and references therein).
The photocatalytic degradation of EDTA has been tested by
us with TiO2 and Fe-doped TiO2 samples [26,29,30,32,34].
Generally, the presence of a donor accelerates the heterogeneous photocatalytic reduction of Cr(VI), as previously reported by us and other authors (see Refs. [34] and [35] and
references therein).
In a previous work, some ZrO2 -based samples, pure
and doped with Fe(III), have been prepared by a solgel
technique, and results of the structural, morphological and
surface characterization as well as some spectroscopic

and adsorption properties of these specimens have been

presented [3638]. In this work, photocatalytic experiments
using these samples were performed in the oxidative and
reductive systems mentioned above, and the relationship between the activity of the samples and their physicochemical
properties were compared.

2. Experimental
2.1. Chemicals
ZrO2 and Fe/ZrO2 samples (named hereafter X-Y,
where X is the weight percentage of iron in the ZrO2
matrix and Y is the calcination temperature in C) were
prepared by hydrolysis of aqueous suspensions of commercial ZrOCl2 8H2 O (Merck) containing different amounts
of Fe(NO3 )3 8H2 O (Panreac). Aqueous ammonia (Merck,
25 wt.%) was added dropwise to the mixture with continuous stirring at pH 910. Two different digestion times
(30 min and 3 h) were applied to the suspensions. After
gelation, the solids were filtered and repeatedly washed until negative AgNO3 test, to ensure the absence of chloride
ions in the samples, as confirmed by XPS analysis. The
samples were dried at 100 C for 24 h and the gel precursors
were annealed at several temperatures, without a heating
rate. Pure ZrO2 samples were prepared by the same procedure but in the absence of iron. TiO2 (Degussa P-25) was
a commercial sample, kindly supplied by the manufacturer
(Degussa A.G., Germany) and used as provided. The main
characteristics of the oxide samples are listed in Table 1,
taken from Refs. [36] and [37]. More detailed information
concerning the preparation and characterization of these
samples will be reported hereafter [38].
NaNO2 (Merck), Na2 EDTA (Carlo Erba) and K2 Cr2 O7
(Carlo Erba) were of quality grade and used as provided.
All other reagents were at least of reagent grade, and used
without further purification. Water was double-distilled in a

S.G. Botta et al. / Journal of Photochemistry and Photobiology A: Chemistry 129 (1999) 8999

quartz apparatus. Diluted HClO4 and NaOH were used for

pH adjustments.

2.2. Photocatalytic studies

Irradiations were performed using a high-pressure xenon
arc lamp (Osram XBO, 150 W) with a bandpass filter (Schott
catalog No. BG 1, thickness 3 mm; 270 nm < < 510 nm;
maximum transmission (87%) at 360 nm). The IR fraction
of the incident light was removed by a suitable filter (Schott
catalog No. KG 5). Actinometric measurements were performed by the ferrioxalate method [39]. A photon flow per
unit volume of 1.5 105 Einstein dm3 s1 was calculated.
UVvisible absorption measurements were performed on
a Shimadzu 210A spectrophotometer.
Photocatalytic runs were done in a thermostatted cylindrical Pyrex cell. In all cases, a fresh solution (10 cm3 ) of
the substrate at a known concentration was adjusted to the
desired pH, and the catalyst was suspended in the solution
immediately before the irradiation. The concentration of all
ZrO2 -based samples in the photocatalytic experiments was
4.0 g dm3 . Comparative tests with 1.0 g dm3 TiO2 (P-25)
were carried out. All oxide concentrations guaranteed total
light absorption.
Prior to irradiation, suspensions were kept in the dark and
stirred at 25 C a time enough to assure substrate-surface
equilibrium. The extent of adsorption of the substrate on the
catalyst was determined by measuring concentrations before
and after stirring.
Irradiations for a fixed period of time were performed under magnetic stirring. Samples were periodically withdrawn
and filtered through a 0.22 m Millipore filter. At least,
duplicated runs were carried out for each condition, averaging the results. In all cases, the adsorption in the dark was
discounted.
Nitrite experiments were performed at 1.0 mmol dm3
and different pH conditions. The suspensions were stirred
for 45 min until equilibration; a water-saturated oxygen
stream was bubbled in the suspension at a constant rate
(1.4 cm3 min1 ) during the whole experiment. The final
nitrite concentration was determined in the filtered solution
by spectrophotometry at 520 nm using the sulphanilic acid
method [40]. No nitrite conversion was observed in the
absence of oxide.
The oxidative degradation of EDTA was performed using 1.0 mmol dm3 suspensions at pH 2 with the reactor
open to air. Suspensions were stirred in the dark for 30 min
before irradiation. EDTA concentration was evaluated by
spectrophotometric analysis of the Co(III) complex [41]. No
reaction was observed in the absence of oxide.
In the case of Cr(VI) experiments, 0.4 mmol dm3
K2 Cr2 O7 solutions at pH 2 were used and suspensions were
stirred for 30 min before irradiation. Experiments were performed with the cell open to air. Changes in Cr(VI) concentration were followed by UV spectrophotometry at 349 nm

91

[42]. These experiments were repeated in the presence of

1.0 mmol dm3 EDTA.
2.3. Photocorrosion of the catalyst
Fe and Zr were analyzed in the filtered solution after irradiation in order to study the photostability of the ZrO2 -based
materials. Dissolved zirconium was evaluated by the alizarin
method [43]. For iron-containing oxides, total iron in solution was determined by the thioglycolate method [44].

3. Results
3.1. Photocatalytic nitrite oxidation
Fig. 1 shows the time course of nitrite oxidation under
irradiation using the pure 0500 and 0600 samples as photocatalysts. Results in the absence of light are also shown.
Normalized concentration was plotted against reaction time
to circumvent slight differences in concentration between
experiments. All data have been corrected discounting the
adsorption after 45 min stirring in the dark. Both pure zirconia samples were active; the reaction cannot be attributed to
a thermal effect since in the dark the decrease in NO2 concentration was almost negligible after 4 h irradiation. It can
be observed that the sample calcined at 600 C was slightly
more active.
The effect of the thermal treatment on the photocatalytic
behavior of the samples is shown in Fig. 2. As it can be

Fig. 1. Normalized concentration vs. time for NO

2 oxidation over different
oxides in the presence and absence of light. Conditions: [oxide] = 4 g dm3 ;
[NO2 ] 1.0 mmol dm3 ; pH 4; T = 25 C; I0 = 1.5 105 Einstein dm3 s1 ;
O2 bubbling.

92

S.G. Botta et al. / Journal of Photochemistry and Photobiology A: Chemistry 129 (1999) 8999

Fig. 2. Normalized concentration vs. time for NO2 oxidation over different ZrO2 samples with different thermal treatments. Inset: % nitrite
oxidation at two irradiation times. Same conditions as in Fig. 1.

Fig. 4. Normalized concentration vs. time for NO2 oxidation over

Fe-doped zirconia with different iron loadings. Inset: % nitrite oxidation
at different pH at two irradiation times. Same conditions as in Fig. 1.
Table 2
pHin , pHfin and % initial (dark) adsorption of NO
2 in the photocatalytic
experiments

Fig. 3. Normalized concentration vs. time for NO2 oxidation over 0500
at different pH. Inset: % nitrite oxidation at different pHin after 60 min
irradiation. Same conditions as in Fig. 1.

seen, the activity increased with the calcination temperature.

Samples not calcined (00) or calcined at a lower temperature (0500) gave rise to a constant reaction rate, while for
the sample calcined at 600 C a slight deceleration occurred
at long irradiation times (180 min).
The effect of pH was studied with pure 0500 as the catalyst. In Fig. 3, the concentration profiles and the photooxidation degree after 1 h irradiation are presented. It can be
observed that both the photocatalytic efficiency and the re-

Photocatalyst

pH

pHfin

% Initial NO2 adsorption

00
0500
0500
0500
0500
0500
0.5500
5500
10500
20500
0600
TiO2 (P-25)

4.0 0.1
3.0 0.1
3.5 0.1
4.0 01
5.0 0.1
6.0 0.1
4.0 0.1
4.0 0.1
4.0 0.1
4.0 0.1
4.0 0.1
4.0 0.1

5.7 0.1
3.3 0.1
4.0 0.1
5.6 01
5.9 0.1
6.3 0.1
5.0 0.1
4.6 0.1
4.6 0.1
4.9 0.1
5.5 0.1
6.1 0.1

3.5
35.5
14.6
5.4
1.4
1.1
6.9
14.6
10.2
8.9
7.8
7.9

action rate decreased as the pH of the suspension increased.

From pH 4 and up, the reaction rate was almost similar.
Decelerating effects in concentration profiles were not observed, at least up to 60 min irradiation.
The effect of Fe-doping was studied by irradiating suspensions at pH 4 over zirconia samples with different iron
contents (0, 0.5, 5, 10 and 20%). Fig. 4 shows the results.
At low Fe content, an increase in the photocatalytic activity
was observed with a maximum at 5 wt.% Fe after 240 min
irradiation. For iron contents higher than 5 wt.%, the activity decreased, but it was still higher than that of the undoped sample. As expected, results over TiO2 (P-25) show
that this oxide is the most active material for this reaction.
Concentration profiles as a function of time for 0500 and
0.5500 showed a linear kinetic behavior, while in the case
of a higher Fe loading a decelerating effect was observed.

S.G. Botta et al. / Journal of Photochemistry and Photobiology A: Chemistry 129 (1999) 8999

Fig. 5. Photodegradation of EDTA over different photocatalysts.

Conditions: [oxide] = 4 g dm3 ; [TiO2 ] = 1 g dm3 ; pH 2; irradiation
time = 120 min; T = 25 C; I0 = 1.5 105 Einstein dm3 s1 ; open to air.
Table 3
% Initial (dark) adsorption of EDTA in the photocatalytic experiments
Photocatalyst

% Initial EDTA adsorption

0500
0.5500
5500
TiO2 (P-25)

29.3
24.4
26.2
0

93

Fig. 6. Normalized concentration vs. time for Cr(VI) reduction over different catalysts. Inset: % Cr(VI) reduction at two irradiation times. Conditions: [oxide] = 4 g dm3 ; [TiO2 ] = 1 g dm3 ; [Cr(VI)] = 0.4 mmol dm3 ;
pH 2; T = 25 C; I0 = 1.5 105 Einstein dm3 s1 ; open to air.

In Table 2, initial (pHin ) and final (pHfin ) pH values are

presented as well as the adsorption of the substrate in the
dark for each photocatalytic condition, taken from Refs. [36]
and [38]. In all cases, an increase in pH after 240 min irradiation was observed, slightly lower at a low pH.
3.2. EDTA degradation
Fig. 5 shows the degree of EDTA degradation after 2 h irradiation over different samples (0500, 0.5500 and 5500)
as catalysts. For comparative purposes, a similar experiment
over TiO2 (P-25) is also shown. The photocatalytic efficiencies of all ZrO2 -based catalysts were remarkably lower than
that of P-25, but the conversion degree increased slightly
with the amount of iron.
No changes in pH (at pHin 2) were observed after irradiation. The adsorption of the substrate in the dark is presented
in Table 3, taken from Refs. [36] and [38].
3.3. Cr(VI) photocatalytic reduction
The effect of irradiation on Cr(VI) suspensions at pH 2
containing different photocatalysts (0500, 5500 and TiO2

Fig. 7. % Cr(VI) reduction at two irradiation times in the presence of

EDTA over different catalysts. [EDTA] = 1.0 mmol dm3 . Same conditions
as in Fig. 6.

P-25) is shown in Fig. 6. ZrO2 -based samples presented a

very low photocatalytic activity compared with that of P-25;
however, the efficiency increased with Fe-doping as it was
observed in the previous cases.
The degree of Cr(VI) photocatalytic reduction after 30
and 120 min over different photocatalysts in the presence of
1.0 mmol dm3 EDTA is presented in Fig. 7. For comparative

94

S.G. Botta et al. / Journal of Photochemistry and Photobiology A: Chemistry 129 (1999) 8999

Zr nor Fe was detected in solution, it was concluded that the

samples were photostable in the present conditions.
4. Discussion
4.1. Mechanism
The heterogeneous photocatalytic process is a complex
sequence of reactions that can be expressed by the following
set of simplified equations:
hv

SC ecb + hvb + recombination

(1)

hvb + + H2 Oads HOads + H+

(2)

EEg
+

hvb + HOads HOads

+

hvb + Dads Dads

HO

ecb + Aads Aads

Table 4
% Cr(VI) initial (dark) adsorption in the different experiments
Photocatalyst % Initial Cr(VI) adsorption % Initial Cr(VI) adsorption
(without EDTA)
(with EDTA)
0500
0.5500
5500
TiO2 (P-25)
a

40.5
NDa
44.4
13.6

20.4
27.7
30.4
17.1

(3)
(4)

+ Dads Doxid

Fig. 8. Normalized concentration vs. time for Cr(VI) reduction(VI) in the

presence of EDTA over different catalysts. Same conditions as in Fig. 6.

(5)

(6)

where D and A are respectively donors and acceptors.

The oxidative pathway can be performed by direct hole

attack or mediated by HO radicals, in their free or adsorbed

form, depending on the substrate. In many cases, complete
mineralization of an organic substrate to CO2 and H2 O occurs. Generally, A is dissolved O2 , which is transformed
in superoxide radical anion.(O2 ) and can lead to the ad
ditional formation of HO :
e

ecb + O2ads O2 HO

(7)

ND = Not determined.

purposes, the results in the absence of photocatalyst and

over 0500 in the absence of EDTA are included. It can
be observed that the extent of Cr(VI) reduction with EDTA
under irradiation in the absence of catalyst, i.e. the homogeneous photochemical reaction, is significant. The same
reaction in the dark was negligible (at least up to 2 h). After 120 min irradiation, all oxides presented a similar photocatalytic efficiency, comparable with that of P-25. However, after shorter irradiation times, a deleterious effect of
Fe-doping on the photocatalytic activity could be observed.
Concentration profiles show a dramatic acceleration of the
Cr(VI) reduction by addition of EDTA, with a slower initial
rate in the most loaded samples (cf. Figs. 6 and 8).
Either in the presence or in the absence of EDTA, no
changes in pH were observed at the end of the experiment.
The amount of Cr(VI) adsorbed on zirconia samples after
equilibration in the dark (with and without EDTA) is presented in Table 4, and is taken from Refs. [36] and [38].
3.4. Deactivation of the catalysts
At the end of irradiation, Zr and Fe (for Fe-containing
samples) were determined in the filtered solution. As neither

A can be also a metal ion species transformed to a

different oxidation state species:
Mn+ + zecb M(nz)+

(8)

4.1.1. NO2 oxidation

In a heterogeneous photocatalytic system, nitrite is anodically oxidized to nitrate while O2 is reduced:
NO2 + H2 O NO3 + 2H+ + 2e
O2 + 4H+ + 4e 2H2 O

(9)
(10)

The net reaction is therefore:

2NO2 + O2 2NO3

(11)

In previous papers we have proposed the following mechanism for the photocatalytic oxidation of nitrite [31,32]
NO2 + HO HNO3 (NO3 )2 + H+

(12)

HNO3 NO2 + OH

(13)

(NO3 )2 + H2 O NO2 + 2OH

(14)

2NO2 N2 O4

(15)

S.G. Botta et al. / Journal of Photochemistry and Photobiology A: Chemistry 129 (1999) 8999

N2 O4 + H2 O NO3 + NO2 + 2H+

(16)

NO2 + HO HO2 NO

(17)

HO2 NO NO3 + H

(NO3 )2 + O2 NO3 + O2

(18)
(19)

The reductive pathway is expressed by Eq. (7), and con

tributes also to HO generation and nitrite oxidation. Other
reactions related to the photochemistry of nitrogen species
might also take place, such as the reaction of NO2 with
ecb (provided the redox potential of the conduction band
of the ZrO2 samples was thermodynamically enough), but
they would end equally in nitrate formation [45].
4.1.2. Oxidative degradation of EDTA
In the case of EDTA, direct oxidation by holes or attack by

HO produces hydrogen abstraction, leading to a carboxyl

radical as the primary oxidation product:
RCOOH + h+ (HO ) RCOO + H+ (H2 O) CO2
+ other products

(20)

The cathodic process is the reduction of oxygen (Eq. (7)),

which contributes also to EDTA oxidation.
4.1.3. Reduction of Cr(VI)
The cathodic photocatalytic reaction in this system is
the three electron-reduction of Cr(VI) to Cr(III) whereas
the conjugate anodic reaction is the oxidation of water to
oxygen. Successive one-electron reducing steps have been
proposed [31,32,34,35], ending in Cr(III), the stable final
product:
e

Cr VI Cr V Cr IV Cr III

(21)

+
1
2 H2 O + h

(22)

41 O2 + H+

The net reaction at pH 2 is:

2Cr 2 O7 2 + 16H+ 4Cr 3+ + 8H2 O + 3O2

(23)

The oxidative pathway, i.e. oxidation of water by holes

(Eq. (22)), is kinetically slow. When no hole or hydroxyl
radical scavengers are present, recombination of electrons
and holes is a competing reaction that lowers the efficiency
of Cr(VI) reduction. In the presence of a donor such as
EDTA, a cooperative action of both substrates was observed,
leading to an increased rate of photocatalytic transformation
([34, 35] and references therein]).
4.2. Photocatalytic efficiency
The efficiency of a catalyst in heterogeneous photocatalytic systems under polychromatic irradiation is evaluated
generally through the photonic efficiency ( %) [46]:
% =

dC/dt
100
P0

95

Table 5
Calculated photonic efficiencies for the photocatalytic oxidation of nitrite
in different conditions
Photocatalyst

pHin

00
0500
0500
0500
0500
0500
0.5500
5500
10500
20500
0600
TiO2 (P-25)

4.0 0.1
3.0 0.1
3.5 0.1
4.0 0.1
5.0 0.1
6.0 0.1
4.0 01
4.0 01
4.0 0.1
4.0 01
4.0 01
4.0 0.1

0.03
0.52
0.32
0.05
0.05
0.04
0.09
0.15
0.13
0.10
0.10
2.48

where dC/dt is the reaction rate and P0 is the incident

photonic flow per unit volume.
The photonic efficiency allows to compare experiments
performed in similar conditions of substrate and oxide concentration, geometry, incident light intensity and is only
valid for each experimental condition. Photonic efficiencies
calculated at initial times allow the evaluation of the true
initial activity, intrinsic of each sample, and independent of
any type of inactivation, inhibition or other secondary effect
that could appear at long irradiation times.
4.2.1. NO2 oxidation
In this case, the efficiencies were calculated from initial
rates (determined graphically), and are shown in Table 5. As
expected, the photonic efficiency of P-25 is largely higher
than that of ZrO2 -based catalysts. From values in Table 5
and the corresponding concentration profiles, the following
conclusions can be withdrawn:
4.2.1.1. Calcination temperature. Photocatalysts calcined
at different temperatures present differences in initial photonic efficiencies: pure ZrO2 calcined at 500 C is 35% more
efficient than the amorphous oxide, but a 100 C increase
in the calcination temperature doubles the activity. This
tendency is also maintained at long irradiation times (see
Fig. 2). Different factors could account for these results.
Regarding differences in initial substrate adsorption (see
Table 2), the higher activity of the semiconductor treated at
the highest temperature could be attributed to the higher adsorption. However, this adsorption cannot be correlated with
the SBET of the sample, since our results show a decrease in
the specific surface with the increase of the calcination temperature (see Table 1). Consequently, the specific surface
area does not seem to be a determining factor. The weak
correlation between activity and specific surface is a common fact in heterogeneous photocatalysis [46]. In addition,
the IR spectra of the samples [3638] show that the bands of
adsorbed water and adsorbed hydroxyl groups are less
intense in the samples calcined at 600 C, suggesting a
lower activity for these samples. However, this is not the

96

S.G. Botta et al. / Journal of Photochemistry and Photobiology A: Chemistry 129 (1999) 8999

result of the photocatalytic experiments. The higher activity

of 0600 could be attributed then to a higher crystallinity
degree resulting from the modification of the parameters
during the preparation procedure (3 h of digestion time and
3 h of annealing at 600 C), reinforced by the fact that the
non calcined (amorphous) sample shows the lowest activity. Without a thermal treatment or after a calcination at
low temperature, crystal defects could remain in the sample, acting then as recombination centers for e /h+ pairs.
Besides, the calcination procedure could lead to surface
changes, affecting the type of active sites. The combination
of all these factors could be responsible for the differences
in the activity of ZrO2 samples prepared in slightly different
conditions and calcined at different temperatures.
4.2.1.2. pH. Table 5 shows a decrease in the initial photonic efficiency of 0500 with increasing pH, reaching a constant value from pH 4 and up. This behavior is kept at long
reaction times (see Fig. 3). This effect could be attributed
to a decrease in the initial nitrite adsorption as long as pH
increases.
The increase in pH observed at the end of the reaction
in all experiments (see Table 2) cannot be assigned to the
global photochemical process (Eq. (11)), but to the different
trends of the oxide samples to adsorb H+ or OH depending
on pHin . Experiments performed in the dark showed an increase in pH after 4 h stirring, which reinforces the previous
hypothesis.

lead to a decrease in the recombination rate and to an increase in the photocatalytic efficiency [26,46]. On the other
side, if the Fe content is high, it can act as a recombination
site, influencing negatively the activity. It has been noticed
that this is a determining effect in the case of Ti oxides
doped with Fe [31].
It is worthwhile to remember that the incorporation of
iron into a ZrO2 matrix leads to a transformation of phase
from monoclinic to tetragonal with a distortion of the lattice
[47]; these factors can affect the substrate-surface affinity
and the we of the sample and, in consequence, could modify
the photocatalytic efficiency.
It should be taken into account that an increase in Fe
content increases the absorption in the visible range of the
semiconductor sample [36,38]. Therefore, provided that the
bands in the visible range are photochemically active, a more
efficient use of the incident radiation could lead to an increase of the efficiency. However, the 20500 sample, which
shows the highest absorption, is not the most active specimen. We conclude that the spectroscopic properties are not
a determining factor for the photocatalytic activity. This fact
has been previously observed in the case of Fe/Ti oxides
[31,32,34].
In an isolated experiment, the photocatalytic activity of
5600 for NO2 oxidation was tested. The efficiency of this
sample was found to be lower than that of 0600, suggesting
that the effect of Fe content on zirconia depends also on the
calcination temperature.
The IR spectra of the samples [3638] do not reveal special differences concerning the hydration and hydroxylation
degree of the Fe-containing samples compared with the pure
specimens. Therefore, differences in activity cannot be explained by this factor.

4.2.1.3. Fe content. Table 5 shows that the initial photonic efficiency increases with the Fe content, attaining a
maximum at 5 wt.% iron. After 240 min, these differences
in relative efficiency are maintained (see Fig. 4). These results agree with those previously obtained by us regarding
oxygen photoadsorption, which go from a very slight ability
in the undoped sample to a more pronounced value in the
case of doped samples. However, this effect is reported to
be maximum for the 0.5 wt.% Fe sample [37,38].
From Table 1, it is observed that SBET values increase with
the Fe content and that the initial adsorption of the substrate
follows the same trend until 5 wt.% (Table 2). Although both
factors could lead to an increase in efficiency, this shows
again that SBET is not a determining factor, since 5-500 is the
most efficient sample compared with more loaded samples,
in spite of its lower surface. The observed variation could
be thus explained by the effect of Fe(III) ions in the ZrO2
matrix that, depending on the iron content, can be positive
or negative. On one side, Fe can act as a trap of e or h+

(NO3 )2 reacts through reaction (14) yielding NO2 ,

which can react with superoxide radical anion formed
through reactions (7) and (19), regenerating nitrite and
oxygen (reaction (28)):

Fe3+ + e Fe2+

(25)

NO2 + O2 NO2 + O2

Fe3+ + h+ Fe4+

(26)

If the conduction and valence band potentials of the semiconductor samples are thermodynamically favorable for both
trapping processes, the presence of a doping agent would

4.2.1.4. Inhibition: possible reasons. As mentioned previously, nitrite oxidation to form nitrate is a slow process at
room temperature. O2 is essential in the photocatalytic reaction, because it acts in the conjugate reduction reaction and
decreases the recombination rate. In its absence, the process
is very slow. When NO3 produced in the reaction reaches
a considerable concentration, it can compete with O2 for the
photogenerated e [31]:
NO3 + e (NO3 )2

(27)

This effect could decrease the photocatalytic efficiency and

could explain the deceleration observed with the most active
photocatalysts (5500, 10500 and 20500) (see Fig. 4) and
also with TiO2 (not shown). Inhibition by competitive adsorption of NO3 is also possible [31].

S.G. Botta et al. / Journal of Photochemistry and Photobiology A: Chemistry 129 (1999) 8999
Table 6
Calculated photonic efficiencies for Cr(VI) reduction in the absence and
the presence of EDTA
Photocatalyst

% (without EDTA)

% (with EDTA)

No photocatalyst
0500
0.5500
5500
TiO2 (P-25)

ND
0.07
NDa
0.10
1.43

0.35
4.15
3.11
1.4
13.1

ND = Not determined.

4.2.2. Oxidative degradation of EDTA

In this work, kinetic curves for EDTA degradation were
not performed, precluding the calculation of photonic efficiencies. The evaluation of the photocatalyst efficiency was
made taking into account the results obtained after 120 min
of irradiation (see Fig. 5).
As in the case of nitrite oxidation, ZrO2 samples were
much less active than P-25. There is a slight increase in
ZrO2 activity with the increase on the Fe(III) content. The
amount of EDTA adsorbed on the different photocatalysts
was similar (Table 3); then, differences in efficiency cannot
be attributed to differences in substrate adsorption.
In the case of Fe-doped TiO2 photocatalysts, it has been
suggested that photolysis (or/and photocatalysis) of surface
or homogeneous complexes formed between EDTA and
dissolved or surface iron, can contribute additionally to
EDTA oxidation [26,34]. If the homogeneous complexes
were formed also in the Fe/ZrO2 systems, their photolysis
and photocatalysis would contribute to EDTA oxidation
according to reactions of the following type:
(>)FeIII EDTA + hv (>)FeII
+ oxidation products of EDTA

(29)

where > indicates surface species.

However, the production of Fe(III)EDTA complexes in
solution should be ruled out since no iron dissolution was observed. In addition, isoelectric point (IEP) data for (0500)
and (5500) suggest that there is no Fe(III) excess on the
oxide surface of these samples (see Table 1); moreover, XPS
data indicate a lower amount of iron on the surface compared with that in the bulk [36,37]. Therefore, a significant
contribution of surface complexes should be also ruled out,
avoiding contribution of Fe(III)EDTA photolytic reactions
to the degradation of the organic compound.
The improvement in the photocatalytic activity by the
presence of iron in the ZrO2 matrix can be then explained,
as in the NO2 case, by the effect of Fe(III) as a trapping
center for the photogenerated charges.
4.2.3. Cr(VI) reduction
Table 6 shows initial photonic efficiencies for Cr(VI)
reduction in the presence and in the absence of EDTA (initial rates calculated from Fig. 8). The profiles of Cr(VI)
reduction in the presence of EDTA could be fitted by first

97

order equations. Initial rates were calculated taking into

account the apparent first order constant (initial rate =
kapp [Cr(VI)]0 ).
Both in the presence and in the absence of the reductant,
P-25 showed the highest initial efficiency. In the absence
of EDTA, % increases with the Fe-loading (at least up
to 5 wt.% Fe). This trend is maintained at long irradiation
times (up to 120 min, see Fig. 8). The effect of iron can
be explained by the increase in the substrate adsorption at
higher Fe contents (see Table 4) and by the improvement of
the charge trapping rate vs. charge recombination rate ratio.
When EDTA was added to the system, a large increase
in the initial photonic efficiency (more than one order) was
promoted. However, contrary to that observed in the previous cases, the efficiency decreased with the Fe-doping, being
lower for the most loaded sample. So far, we do not have an
easy interpretation of this experimental observation. In the
absence of a photocatalyst, Cr(VI) reduction by EDTA (negligible in the dark) is significant. However, the addition of
the semiconductor increases noticeably the efficiency, proving that it acts as a true photocatalyst. At longer irradiation
times (30 min, the tendency of the decrease of the activity
with the Fe content is kept (see Fig. 7). It is worthwhile to remark that after 120 min irradiation all the samples presented
similar efficiencies, comparable with that of P-25, and that
Cr(VI) conversion was total with all the samples.
Although the Cr(VI) adsorption in the presence of EDTA
on ZrO2 -type catalysts is lower than in its absence (see Table 4), a significant increase in the photocatalytic activity is
observed. An increase in the efficiency for Cr(VI) reduction
in the presence of organic donors has already been observed
for TiO2 and Fe/TiO2 ([34, 35] and references therein). As

said before, EDTA is an efficient trap of h+ or HO and,

in its presence, recombination rate is surely lower. Analogously, an increase in EDTA oxidation efficiency should
have been observed by the addition of Cr(VI) (not verified
experimentally in this work).
The direct photolysis of Cr(VI)EDTA complexes might
be proposed as a contribution to the photocatalytic reduction of Cr(VI), but it has been suggested that the formation of these complexes is negligible [48]. However, for
the Cr(V)EDTA complex, a possible intermediate of the
reaction, a rapid intramolecular electron transfer process
leading to Cr(III) complexes has been postulated [49] and,
therefore, their contribution to the acceleration of the photocatalytic reaction should not be discarded. In a similar study
of the Cr(VI)/4-chlorophenol system, Fu et al. detected an
increased absorption in the visible range after a UV-induced
reaction [50], suggesting the formation of intermediate
active species.
4.3. Comparison between the photocatalytic systems
As the results suggest, the photocatalytic efficiency is
dependent on the nature of the redox reaction involved

98

S.G. Botta et al. / Journal of Photochemistry and Photobiology A: Chemistry 129 (1999) 8999

in the process, which is a common fact in heterogeneous

photocatalysis.
It has been proposed that the photocatalytic oxidation of

nitrite is a fast process of HO attack, that may take place

probably at the interface [31]. Therefore, surface properties,
as the affinity of the substrate for the surface, have in this
reaction a determining role. The influence of this factor is
reflected in the variation of NO2 adsorption with the increase of the Fe content (see Table 2) and affects notably the
reaction rate. For example, comparing 0500 and 5500 at
pH 4, while the adsorption degree is doubled, the photonic
efficiency triplicates in the doped sample.
This effect is less marked in the slower reactions, i.e., the
oxidation of EDTA and the Cr(VI) reduction. The relative
differences in adsorption between the pure and the doped
samples are not large, and the same trend is observed in the
reaction rates.
According to the results of characterization of the samples, in the ZrO2 -based catalysts containing up to 5% Fe,
iron is not preferentially located at the surface, suggesting
the formation of an external layer of zirconia [36,37]. In
the case of the samples containing a higher Fe content, iron
oxide is not segregated at the surface and a solid solution
is observed [38]. In any case, there is no Fe excess in the
surface in any sample. However, the presence of iron could
modify the surface properties of the material regarding the
type of active sites, presence of defects, etc., which could
increase the adsorption and favor the interfacial reactions.
In slower reactions, it is probable that the influence of
bulk properties on the photocatalytic rate overcome that of
surface properties.
The case of the complex Cr(VI)/EDTA system is different,
because electron transfer processes in possible intermediate
Cr(V)EDTA species could trigger the reaction. This is the
only system, among the ones studied here, in which iron
exerts a deleterious effect on the initial rate, and should be
object of a more detailed and complete study.

carried out with reused samples are adequate to demonstrate

the stability of the photocatalysts. Experiments in this sense
are in progress.
4.5. Comparison with other photocatalysts
Although the calculated bandgap energies of the ZrO2
samples prepared by the solgel technique are close to that
of P-25 (3.1 eV [52]), the photocatalytic efficiencies of both
types of photocatalysts in the range of wavelengths used in
this work are significantly different.
It must be noticed that, to warrant a total incident light absorption, a 4-fold amount of ZrO2 -based catalysts is needed
in comparison to P-25. The higher activity of P-25 could be
assigned to a higher efficiency in the separation of the photogenerated charges (less e /h+ recombination rate) due to
the special structure of this material [52]. On the other hand,
ZrO2 samples present a very low absorbance in the near UV
range that can be attributed to intrabandgap surface states.
Therefore, the low activity of ZrO2 samples can arise from
the contribution of these surface states. The new samples,
as well as the commercial Degussa ZrO2 , present, in contrast, intense absorption bands in the short UV range, around
230 nm [3638]. It would be interesting, then, the comparison of the photoactivity of the samples with that of P-25
under short UV irradiation. Experiments in this direction are
underway. Also, differences in back-scattering between the
ZrO2 and TiO2 samples, which originate important losses in
the absorbed light, can account for differences in the activity. Other favorable factors of P-25 are the somewhat higher
specific surface area (49 m2 g1 ), a smaller particle size and
a lower amount of defects in the structure. Nevertheless, important advantages of the new FeZrO2 samples compared
with Fe-doped titania are the enhancement of the activity
compared with the pure specimens, the absence of photodissolution after illumination [26,2932,34] and the higher particle size that facilitates their recovery from the reaction
medium by fast precipitation.

4.4. Stability of the photocatalysts

5. Conclusions
The results indicate that the samples synthesized by the
solgel method do not suffer photodissolution in the irradiation conditions. Due to this advantageous property, it should
be possible, in principle, their reuse.
No thermodynamic data for Zr dissolution is available
in the literature [51], and we assume that photodissolution
is not possible in these conditions. Concerning Fe dissolution, as previously mentioned, IEP and XPS data suggest that there is no Fe(III) excess on the surface and, in
consequence, the particles would be more resistant to Fe
photocorrosion.
However, the lack of dissolved Zr and Fe detection only
indicates the absence of photodissolution but it does not
eliminate the possibility of photocorrosion without dissolution, due to surface transformation. Only kinetic experiments

The photocatalytic activity in the near UV range of zirconia and Fe/zirconia samples prepared by a solgel technique
was tested in systems of environmental interest. All samples
were found active, although its efficiency was lower compared to that of TiO2 P-25. The incorporation of Fe(III),
acting as an electron or hole trapping center in zirconia,
improved the photocatalytic activity. This behavior was observed in the oxidations of NO2 and EDTA and in the
reduction of Cr(VI). In contrast, in the case of the Cr(VI)
reduction in the presence of EDTA, the photocatalytic efficiency decreased slightly with the iron content.
The different activities of the samples have been associated not only to differences in structural, optical and surface
properties, affected by the conditions of synthesis, but also

S.G. Botta et al. / Journal of Photochemistry and Photobiology A: Chemistry 129 (1999) 8999

to the nature and conditions of the photocatalytic reaction.

In the case of Cr(VI) reduction in the presence of EDTA,
the yield of the ZrO2 -based catalysts was similar to that of
P-25. These results are promising, as photocatalysis emerges
an adequate procedure for the treatment of real wastewaters,
where oxidants and reductants are simultaneously present.
In addition, the new zirconia samples do not suffer photocorrosion after the irradiation tests and, owing to their large
particle size (see Table 1), are easily separable from the reaction medium.
The results of this work are meaningful for future studies on the improvement of the TiO2 photocatalytic efficiency. Binary Ti/Zr oxides have been reported to present
an enhanced photocatalytic performance [53]. Further complementary investigations with ternary Fe/Zr/Ti oxides are
needed to test the photocatalytic activity of these materials
and are in progress.

Acknowledgements
Work performed as part of Comisin Nacional de Energa
Atmica CNEA-CAC-UAQ project #95-Q-03-05. J.A.N.
thanks Direccin General de Investigacin Cientfica y
Tcnica (DGICYT-Spain), PB96-1346 for supporting part
of this work. S.G.B. thanks CNEA for a fellowship to perform this work. M.I.L. wish to thank Fundacin Antorchas
(Argentina) a grant for travel expenses to Sevilla. M.I.L. is
a member from CONICET (Argentina).
References
[1] D.A. Ward, E.I. Ko, Chem. Mater. 956 (1993).
[2] H.H. Kung, Transition Metal Oxides. Surface Chemistry and
Catalysis, Studies Surface Sci. Catal. 45 (1989).
[3] A. Corma, Chem. Rev. 95 (1995) 559.
[4] J.G. Bendoraitis, R.E. Salomon, J. Phys. Chem. 69 (1965) 3666.
[5] P. Clechet, J.-R. Martin, R. Ollier, C. Valloy, C.R. Acad. Sci. Paris
282 Srie C (1976) 887.
[6] J.-M. Herrman, J. Disdier, P. Pichat, J. Chem. Soc. Faraday Trans.
1 (1981) 77.
[7] A.R. Newmark, U. Stimming, Langmuir 3 (1987) 905.
[8] S. Preusser, U. Stimming, K. Wippermann, Electrochim. Acta 39
(1994) 1273.
[9] S. Sato, T. Kadowaki, J. Catal. 106 (1987) 295.
[10] H. Wiemhfer, U. Voher, Ber. Bunsenges Phys. Chem. 96 (1992)
1646.
[11] K. Ganguly, S. Sarkar, S. Bhattacharyya, J. Chem. Soc. Chem.
Comm. (1993) 683.
[12] J.A. Navo, G. Coln, Stud. Surf. Sci. Catal. 82 (1994) 721.
[13] K. Sayama, H. Arakawa, J. Phys. Chem. 97 (1993) 531.
[14] K. Sayama, H. Arakawa, J. Photochem. Photobiol. A: Chem. 94
(1996) 67.
[15] J.A. Navo, G. Coln, J.-M. Herrmann, J. Photochem. Photobiol. A:
Chem. 108 (1997) 179.
[16] P. Pichat, J.-M. Herrmann, J. Disdier, M.-N. Mozzanega, J. Phys.
Chem. 83 (1979) 3122.
[17] G. AlSayyed, J.C. DOliveira, P. Pichat, J. Photochem. Photobiol.
A: Chem. 58 (1991) 99.
[18] J.A. Navo, G. Coln, M. Macas, P.J. Snchez-Soto, V. Augugliaro,
L. Palmisano, J. Molec. Cat. A: Chem. 109 (1996) 239.

99

[19] M.A. Fox, H. Ogawa, P. Pichat, J. Org. Chem. 54 (1989) 3847.

[20] F. Juillet, F. Lecomte, H. Mozzanega, S.J. Teichner, A. Thevenet, P.
Vergnon, Faraday Symp. Chem. Soc. 7 (1973) 57.
[21] H. Courbon, P. Pichat, C.R. Acad. Sci. Paris 285 Srie C (1977)
171.
[22] Y. Kohno, T. Tanaka, T. Funabiki, S. Yoshida, Chem. Commun.
(1997) 841.
[23] Y. Kohno, T. Tanaka, T. Funabiki, S. Yoshida, J. Chem. Soc. Faraday
Trans. 94 (1998) 1875.
[24] Y. Kohno, T. Tanaka, T. Funabiki, S. Yoshida, Chem. Lett. (1997)
993.
[25] M.A. Fox, M.T. Dulay, Chem. Rev. 93 (1993) 341.
[26] M.I. Litter, J.A. Navo, J. Photochem. Photobiol. A: Chem. 98 (1996)
171.
[27] J.A. Navo, M. Macas, M. Gonzlez-Cataln, A. Justo, J. Mater.
Sci. 27 (1992) 3036.
[28] A. Milis, I. Peral, X. Domnech, J.A. Navo, J. Molec. Catal. 87
(1994) 67.
[29] M.I. Litter, J.A. Navo, J. Photochem. Photobiol. A: Chem. 84 (1994)
183.
[30] J.A. Navo, G. Coln, M.I. Litter, G.N. Bianco, J. Molec. Catal. 106
(1996) 267.
[31] J.A. Navo, G. Coln, M. Trillas, J. Peral, X. Domnech, J.J. Testa,
D. Rodriguez, J.R. Padrn, M.I. Litter, Appl. Catal. B: Environ. 16
(1998) 187.
[32] E.A. San Romn, J.A. Navo, M.I. Litter, J. Adv. Oxid. Technol. 3
(1998) 261.
[33] J.A. Navo, G. Coln, M. Macas, C. Real, M.I. Litter, AppI. Catal.
A: General 177 (1998) 111.
[34] J.A. Navo, J.J. Testa, P. Djedjeian, J.R. Padrn, D. Rodrguez, M.I.
Litter, Appl. Catal. A: General 178 (1999) 191.
[35] M.I. Litter, Appl. Catal. B: Environ., in press.
[36] S. Botta, Master Thesis, Universidad de General San Martn, Buenos
Aires, 1998.
[37] J.A. Navo, M. Macas, P.J. Snchez-Soto, M.C. Hidalgo, G.M.
Restrepo, S. Botta, M. Litter, M.V. Tsodikov, Actas del XVI Simposio
Iberoamericano de Catlisis, A. Centeno, SA. Giraldo, E.A. Pez
Mozo (Eds.), Vol. III, Cartagena de Indias, 1998, pp. 18291834.
[38] J.A. Navo, M. Macas, P.J. Snchez-Soto, M.C. Hidalgo, G.M.
Restrepo, S. Botta, M. Litter, M.V. Tsodikov, submitted for
publication.
[39] C.G. Hatchard, C.A. Parker, Proc. R. Soc. A 235 (1956) 518.
[40] Colorimetric Determination of Non Metals, D.F. Boltz (Ed.),
Interscience, New York, 1958, p. 124.
[41] K.L.E. Kaiser, Water Res. 7 (1973) 1465.
[42] C. Wei, S. German, S. Basak, K. Rajeshwar, J. Electrochem. Soc.
140 (1993) L60.
[43] Colorimetric Determination of Traces of Metals, EB. Sandell (Ed.),
2nd ed., Interscience Publishers, New York, 1950, pp. 638642.
[44] D.L. Leussing, I.M. Kolthoff, J. Am. Chem. Soc. 75 (1953) 390.
[45] M.C. Gonzlez, A.M. Braun, Res. Chem. Intermed. 21 (1995)
837.
[46] M.R. Hoffmann, S.T. Martin, W. Choi, D. Bahnemann, Chem. Rev.
95 (1995) 69.
[47] M. Tsodikov, O. Bukhtenko, O. Ellert, V. Shcherbakov, D. Kochubey,
J. Mater. Sci. 30 (1995) 1087.
[48] M. Beck, D. Durham, J. Inorg. Chem. 33 (1971) 461.
[49] K.A. Easom, R. Bose, Inorg. Chem. 27 (1988) 2331.
[50] H. Fu, G. Lu, S. Li, J. Photochem. Photobiol. A: Chem. 114 (1998)
81.
[51] D. Tuck, Standard Potencials in Aqueous Solution, A. Bard, R.
Pearson, J. Jordan (Eds.), IUPAC, New York, 1985, pp. 547552.
[52] R.I. Bickley, T. Gonzlez-Carreo, J.S. Lees, L. Palmisano, R.J.D.
Tilley, J. Solid State Chem. 92 (1991) 178.
[53] X. Fu, L.A. Clark, Q. Yang, M.A. Anderson, Environ. Sci. Technol.
30 (1996) 647.