Determination of the Quantum Yield for the Photochemical Generation of Hydroxyl Radicals in TiO2 Suspensions

Article abstract of DOI:10.1021/jp9505800

The generation of hydroxyl ((.)OH) radicals plays a key role in the heterogeneous photocatalytic degradation of organic pollutants in aqueous suspensions of TiO2.The quantum yield of this process is thus an important parameter; however, it is not easy to measure in a particulate system arising from problems caused by light scattering from the particles.In this work, a reliable method for the determination of the quantum yield of hydroxyl radical production in heterogeneous systems has been developed, based on measurements of (.)OH radical generation rates and the photon flux absorbed by TiO2 suspensions.In this procedure, a modified integrating sphere method was used to determine the true fraction of light absorbed by TiO2 suspensions.A ferrioxalate chemical actinometer was used to measure the incident photon flux.As a check on the quantum yield method, good agreement with known literature values was obtained for quantum yield measurements of the photochemical generation of the p-benzosemiquinone (BQ(.-)) radical in the photolysis of p-benzoquinone and of the (.)OH radical generation in the photolysis of hydrogen peroxide, respectively.Accordingly, the quantum yield for the (.)OH radical production in the TiO2 suspension was determined to be 0.040 +/- 0.003 at pH 7.Effects on the quantum yield of suspension loading, photon flux, and electron-acceptor addition (H2O2 and O2) were explored.

Full text of DOI:10.1021/jp9505800

J. Phys. Chem. 1996, 100, 4127-4134
4127
Determination of the Quantum Yield for the Photochemical Generation of Hydroxyl
Radicals in TiO2 Suspensions
†
Lizhong Sun and James R. Bolton*
Photochemistry Unit, Department of Chemistry, The UniVersity of Western Ontario,
London, Ontario, Canada N6A 5B7
X
ReceiVed: March 1, 1995
•
The generation of hydroxyl ( OH) radicals plays a key role in the heterogeneous photocatalytic degradation
2
of organic pollutants in aqueous suspensions of TiO . The quantum yield of this process is thus an important
parameter; however, it is not easy to measure in a particulate system arising from problems caused by light
scattering from the particles. In this work, a reliable method for the determination of the quantum yield of
•
hydroxyl radical production in heterogeneous systems has been developed, based on measurements of OH
radical generation rates and the photon flux absorbed by TiO
integrating sphere method was used to determine the true fraction of light absorbed by TiO
2
suspensions. In this procedure, a modified
suspensions. A
2
ferrioxalate chemical actinometer was used to measure the incident photon flux. As a check on the quantum
yield method, good agreement with known literature values was obtained for quantum yield measurements of
•-
the photochemical generation of the p-benzosemiquinone (BQ ) radical in the photolysis of p-benzoquinone
•
and of the OH radical generation in the photolysis of hydrogen peroxide, respectively. Accordingly, the
•
quantum yield for the OH radical production in the TiO
2
suspension was determined to be 0.040 ( 0.003 at
pH 7. Effects on the quantum yield of suspension loading, photon flux, and electron-acceptor addition (H
and O ) were explored.
2 2
O
2
1
. Introduction
The photocatalytic degradation of pollutants in aqueous
for a very high concentration of propanal. Valladares and
Bolton used a special cell, in which an aluminum foil jacket
had a small aperture to let in the light, which was then scattered
and reflected by the grains, to measure the true quantum yield
9
solutions is attracting considerable attention for application to
environmental problems.1
-3
This is a process in which pho-
(0.05) for the bleaching of methylene blue. None of these earlier
toinduced holes in semiconductor particles (such as TiO2)
oxidize hydroxide ions (or water molecules) adsorbed on the
surface of the particles to produce highly oxidizing OH radicals,
which subsequently attack adsorbed pollutant molecules. This
primary step initiates a series of degradation reactions that
ultimately lead to mineralization of the pollutants.
studies involved the direct determination of the true quantum
yield of hydroxyl radical generation but rather focused on the
removal of a specific contaminant.
•
•
In this work, we have employed methanol to trap OH radicals
produced in the photocatalytic process in TiO2 suspensions. A
modified integrating sphere method was utilized to measure the
true fraction of light absorbed in the TiO2 suspensions. The
incident photon flux was measured using the ferrioxalate
actinometer. Confirmations of the reliability of the method and
the system for quantum yield measurements were carried out
for the p-benzoquinone (BQ) system and the H2O2 system,
respectively, and good agreement was obtained with literature
values for the quantum yields for both the p-benzosemiquinone
•
Obviously the quantum yield for the generation of OH
radicals plays a very important role in determining the efficiency
of photodegradation of the pollutants. Nevertheless, it has
proved difficult to measure this quantum yield in heterogeneous
photocatalytic processes because of the problem of light
scattering. Perhaps this is why many studies on photocatalysis
in heterogeneous systems either do not report quantum yields
or just give “apparent” quantum yields based on the incident
•-
•
(
BQ ) radical generation and the OH radical production,
4
-6
photon flux.
respectively. The effects of suspension loading, light flux and
7
Schiavello et al. developed an experimental method for the
detection of the quantity of photons absorbed by aqueous TiO2
dispersions, in which two parameters, the photon flux reflected
and the photon flux absorbed by particle dispersions, were
evaluated by measuring the photon flux entering the photoreactor
and the photon flux transmitted by the photoreactor. Lepore et
•
electron acceptors (H2O2 and oxygen) on the OH radical
quantum yield in TiO2 were examined.
2
. Principles
The quantum yield for a photochemical reaction can be
expressed as
8
al. determined an “apparent” quantum yield of 0.012 for the
removal of 4-chlorophenol in a TiO2 suspension. Using
optically transparent preparations of very small TiO2 particles,
they determined a true quantum yield of 0.020 at 313 nm for
the oxidation of iodide. For the same TiO2 suspension they
found true quantum yields of 0.2-0.6 for the removal of
propanal and speculated that the limiting quantum yield is unity
rate of reaction induced by photon absorption
φ )
(1)
flux of absorbed photons
When photons with an energy greater than the bandgap are
absorbed by TiO2 particles, electrons are promoted from the
valence band to the conduction band leaving holes in the valence
band. The photoinduced holes migrate to the surfaces of the
particles and there react with adsorbed hydroxide ions (or water
†
Current address: School of the Environment, Duke University, A141
Science Research Center, Research Drive, Durham, NC 20078.
*
To whom correspondence should be addressed.
Abstract published in AdVance ACS Abstracts, February 1, 1996.
X
•
molecules), producing OH radicals. At the same time, the
0022-3654/96/20100-4127$12.00/0 © 1996 American Chemical Society

4
128 J. Phys. Chem., Vol. 100, No. 10, 1996
Sun and Bolton
(8)
conduction band electrons reduce oxygen adsorbed on the
surfaces to form O2• radicals:
I ) I F
a o S
-
which can be determined photochemically and spectrophoto-
metrically (see Appendix 1).
Final Expression for the Quantum Yield. From eqs 3 and
8 and the derivation in Appendix 1, the final expression for the
quantum yield is
hν
-
+
TiO2
9
8 e + hvb
(2a)
(2b)
(2c)
cb
hvb+ + OH f OH
-
•
ecb- + O f O
•-
ROH
2
2
φS )
F φ 2+
Fe
(9)
S
RFe2+
•
The OH radical quantum yield φOH can be determined from
the generation rate ROH of OH radicals and the flux Ia of
•
where
absorbed photons:
FS ) FFe(III)/FS
(10)
φOH ) ROH/Ia
(3)
where FFe(III) is the fraction of light absorbed in the ferrioxalate
actinometer solution.
•
Determination of the OH Radical Generation Rate.
Hydroxyl radicals are known to react with aliphatic alcohols
Modified Integrating Sphere Method. The fraction of light
absorbed by a sample undergoing a photochemical reaction is
an important parameter in a quantum yield measurement. It
can easily be calculated from the spectrophotometric absorbance
in a homogeneous system. But the determination of the fraction
of light absorbed for a heterogeneous suspension is not so easily
determined due to light scattering by the particles. In our
experiments, we used a modified integrating sphere to determine
the fraction of light absorbed in a light-scattering suspension.
The sample, in an EPR flat cell (path length 0.33 mm) was
placed at the sample position of the integrating sphere with air
as the reference. The detailed method and theory of the
modified integrating sphere are described in Appendix 2.
1
0,11
through abstraction of a hydrogen atom from a C-H bond.•
The method we have used for determining the absolute OH
radical generation rate is based on R-H atom abstraction by
OH radicals from methanol (in excess), followed by moni-
toring of the formation rate of the principal stable product,
•
1
2-15
formaldehyde.
The reaction is
•
•
OH + CH OH f CH OH + H O
(4)
3
2
2
In the presence of oxygen, formaldehyde is formed as the
dominant stable product in a quantitative reaction (k ) 9.6 ×
9
-1 -1 16
1
0 M s )
•
•
CH OH + O f HCHO + HO
(5)
3. Experimental Section
2
2
2
Materials. The TiO2 powder (anatase) was from Aldrich
Chemical Co., Inc. The suspension was prepared by sonication,
centrifugation, and evaporation to narrow the particle size
The formaldehyde generated by reactions 4 and 5 can be
measured by HPLC with UV absorbance detection at λ ) 360
nm, after forming its 2,4-dinitrophenylhydrazine (DNPH)
2
1,22
distribution and to increase the loading of the suspension.
1
7
derivative:
The loading was measured using a weight method (see ref 23
for details). The TiO2 particle diameters, measured using
scanning electron microscopy (SEM), ranged from 100 to 210
nm. Most suspensions were made in 1.0 mM phosphate buffer
solution (BDH).
O2N
NH NH2 + HCHO
NO2
A 2,4-dinitrophenylhydrazine (DNPH) solution, used for the
derivatization of the formaldehyde produced in reaction 6, was
made by dissolving 0.475 g of DNPH (Kodak Laboratory &
Research Products) in 100 mL of a solution containing
concentrated HCl (12 M), water and acetonitrile in the ratio
2:5:1 (v/v/v).
O2N
NH
NO2
N
CH2 + H2O
(6)
•
The hydrogen abstraction of aliphatic alcohols by OH radicals
had been considered to have a 100% yield;
and Ginns found that R-H abstraction by OH radicals occurs
to only 90% and 86% from ethanol and 2-propanol, respectively.
Asmus et al. showed that the efficiency of the reaction of
OH radicals with methanol by R-H abstraction is 93%. The
18,19
however, Burchill
2
0
•
A 0.0398 mM solution of the HCHO derivative of DNPH
was used with and without BaSO4 (BDH) particles (loading of
1
2
-1
the BaSO4 particles was 1.36 g L ), as a test of the modified
•
integrating sphere method.
remaining 7% is accounted for by methoxy radicals formed in
the reaction
A solution of 1.0 mM p-benzoquinone (BQ, Fisher Scientific
Co.) was used for the measurement of the quantum yield for
the p-benzosemiquinone (BQ ) generation.
The oxygen concentrations in the suspension were controlled
by bubbling a mixture of oxygen and nitrogen at a certain
proportion into the suspension until adsorption equilibrium of
the oxygen on particles in the suspension was reached.
Double-distilled water was used for the preparation of all the
solutions and suspensions.
•-
•
•
OH + CH OH f H O + CH O
(7)
3
2
3
The concentration of the HCHO derivative, divided by a
•
factor of 0.93, is thus the corresponding OH radical concentra-
tion produced in the reaction of the H2O2 system. The fraction
•
of OH radicals that react with methanol to produce HCHO in
the TiO2 system is not available; we make the arbitrary
assumption that it has the same value as that found in the
homogeneous system (0.93).
Flux of Absorbed Photons. The photon flux Ia absorbed
by a sample is the product of the incident photon flux Io, and
for a sample S, the integrated absorption fraction FS over the
wavelength range used in the experiment:
Photolyses. Photolyses were carried out using a water-
filtered 200 W xenon/mercury UV lamp (Oriel Corp. of
America) with a UV transmitting black glass filter (365 ( 50
nm, Melles Griot). The transmittance of the UV transmitting
black glass filter in the range 300-400 nm was measured by a
UV/vis spectrophotometer (Figure 1). The photolysis, both for
•
the OH radical measurement with methanol and for the photon

Photochemical Generation of Hydroxyl Radicals
J. Phys. Chem., Vol. 100, No. 10, 1996 4129
Figure 2. Absorbance of a HCHO derivative solution with and without
Figure 1. Transmittance spectrum of the UV black transmitting glass
filter.
-5
BaSO
4
particles: (9) absorbance of a 3.98 × 10 M HCHO-DNPH
particles, BaSO loading ) 1.36
g L ; (0) absorbances of the same solution without BaSO particles;
*) absorbance of the solution with BaSO particles, as determined using
the modified integrating sphere method.
derivative in the presence of BaSO
4
4
-1
4
flux determinations, took place in a WG-812 EPR flat sample
cell (Wilmad Glass Company, Inc.) at the same light intensity,
which are the same conditions as used in our previous work
linked with the measurement of the hydroxyl radical trapping
efficiency by a spin trap.23 The cell path length was determined
to be 0.33 mm from the absorbance measurement of a
(
4
Model ESP300 electron paramagnetic resonance spectrometer.
The details concerning the EPR measurements can be found in
2
3
our previous work.
5
,5-dimethyl-1-pyrroline N-oxide solution (DMPO) of known
concentration.
4
. Results and Discussion
Absorption measurements for the TiO2 suspensions were
determined using an integrating sphere assembly (Shimadzu
Corp.) linked with a UV-260 spectrophotometer (Shimadzu
Corp.).
Fraction of Light Absorbed in the TiO2 Suspensions.
Check of the Modified Integrating Sphere Method. As a check
of the modified integrating sphere method, the absorbance of a
0.0398 mM solution of the HCHO derivative of DNPH was
determined with and without BaSO4 particles (loading of BaSO4
HPLC Measurements. The photolyzed solution (1.0 mL)
in an EPR flat cell was washed out and diluted to 3 mL final
volume and then derivatized by adding 100 µL of the DNPH
solution. After sitting for 1.5 h at room temperature (25 °C),
an aliquot of the sample was injected into a Waters Model 501
solvent delivery system connected with a µ Bondapak C18
column and measured by Waters 486 HPLC tunable absorbance
detector set at 360 nm. The derivative was eluted isocratically
with 45% acetonitrile aqueous solution as the mobile phase.
All the samples for HPLC measurement were filtered using
Millipore films (0.1 µm) before measurement. The data for
HCHO derivative measurements are averages of three measure-
ments. The reliability of the HCHO derivative as a measure of
-1
was 1.36 g L ). Figure 2 shows the marked difference for the
suspension with the BaSO4 particles between a normal absorp-
tion spectrum (i.e., the absorption spectrum measured with a
normal spectrophotometer without using an integrating sphere),
and one determined using the modified integrating sphere
method. The latter spectrum, which has been corrected for
the scattering, is close to that for the solution without BaSO4
suspended particles. This suggests that the modified integrating
method effectively eliminates the interference, arising from the
scattering light, on the absorption measurement and that the
absorption spectrum measured using this method is an accurate
representation of the true absorption spectrum of the suspension.
True Absorbance Spectrum of TiO2. Figure 3 shows a
comparison of the apparent absorbance of TiO2 suspensions
using a normal UV spectrophotometer and the true absorbance
from the modified integrating sphere method, where eq A11
has been utilized.
•
the OH radical generation in the system has been proven in
2
3
our previous work.
Measurement of the Photon Flux. The photon flux was
measured immediately after a photolysis run using potassium
ferrioxalate actinometry. The EPR flat cell (0.33 mm path
length), containing a 0.15 M ferrioxalate solution, prepared from
K3Fe(C2O4)3 (Fisher) recrystallized in the dark, was put into
the EPR cavity and irradiated with the UV lamp. The rate of
production of ferrous ions RFe2+ was determined by a UV/vis
The fraction of light absorbed at a given wavelength λ is
then obtained from eq A2 in Appendix 1.
Hydroxyl Radical Formation Rates. Figure 4 shows the
change of the HCHO derivative concentration as a function of
irradiation time in the TiO2 suspensions. A linear relation of
the HCHO derivative concentration with irradiation time was
obtained. From this figure the growth rate of HCHO derivative
spectrophotometer (Hewlett-Packard Model 8450A) at 510 nm
1
h after addition of 0.1% of 1,10-phenathroline (Aldrich) to
the photolyzed solution. The weighted average of the quantum
yield of the ferrous ion production from ferrioxalate (0.15 M)
over the bandwidth of the black transmitting glass filter is known
-7
-1
•
was found to be (3.16 ( 0.13) × 10 M s , and thus the OH
2
2,24
•
-7
to be 1.19.
radical generation rate is d[ OH]/dt ) (3.40 ( 0.13) × 10
M
-
1
Electron Paramagnetic Resonance (EPR) Measurements.
EPR spectra of the p-benzosemiquinone produced during
photolysis of p-benzoquinone were obtained using a Bruker
s , after applying the 0.93 correction factor.
Confirmation of the Quantum Yield Measurement Method.
To confirm the reliability of the method for quantum yield

4
130 J. Phys. Chem., Vol. 100, No. 10, 1996
Sun and Bolton
(12)
RBQφFe2+FBQ
RFe2+
φBQ
)
where RBQ is the generation rate of BQ•- radicals.
RBQ was obtained from the slope of the growth curve of the
radicals, when the field was fixed at the peak of the center line
of the BQ•- EPR spectrum, using a method similar to that
2
3
described in our previous work. The coefficient FBQ for the
BQ solution (cf. eq 10) in the range 300-400 nm was found to
be 289 ( 8. The corresponding parameters used for the
calculation of the quantum yield are summarized in Table 1.
•
-
According to the results, the quantum yield of BQ radical
generation was found to be 0.44 ( 0.05, which is very close to
2
5
the literature value (0.47 ( 0.04).
•
Quantum Yield of the OH Radical Generation in the
Homogeneous H2O2 System. In the H2O2 system, OH radicals
•
are produced homogeneously by the photolysis of hydrogen
peroxide:
hν
•
H O
9
8 2 OH
(13)
2
2
2
Figure 3. Absorbance spectra of a TiO suspension. Loading of the
-1
•
suspension is 67 mg L : (0) uncorrected spectrum in the normal UV/
vis measurement; (9) corrected absorption spectrum using the modified
integrating sphere method.
The generation rate ROH of OH radicals in 5.0 mM H2O2
solution was measured by using the same method as that used
-7
in the TiO2 suspensions and found to be (3.60 ( 0.11) × 10
-1
M s . The absorption fraction FH O (eq A1 of Appendix 1)
2
2
of the H2O2 solution, based on the absorbance of the solution,
-
4
was calculated to be 7.79 × 10 . Thus, the coefficient FH O
2
2
of the solution (cf. eq 10) was found to be 1285 ( 11.
Corresponding parameters for the calculation of the quantum
yield in eq 9 are listed in Table 1. Accordingly, the quantum
•
yield for the OH radical generation in the homogeneous H2O2
system was calculated to be 0.96 ( 0.09, which is consistent
with the literature values (0.98 ( 0.03 at 308 nm and 0.96 (
0
.04 at 351 nm) in the wavelength range this experiment
2
6
involved.
The excellent agreement with literature values for the
quantum yields of the two systems examined confirmed that
the chosen method for quantum yield measurements is feasible
and reliable.
•
Quantum Yield of OH Radical Production in TiO2
Suspensions. The intermediate data for the quantum yield
•
measurement, including the growth rate ROH of OH radicals,
the incident photon flux Io and the absorption fractions FS, both
-1
for TiO2 suspension (loading of 100 mg L ) with 30 mM
methanol and for the ferrioxalate actinometry solution, ROH,
RFe2+, and F are listed in Table 1. The calculation, using eq 9,
Figure 4. Effect of irradiation time on the concentration of the HCHO
•
shows that the quantum yield of the OH radical generation in
derivative in the TiO
2
suspension. Loading of the suspension is 100
-
1
mg L ; [CH
3
OH] ) 30.0 mM.
the TiO2 system is 0.040 ( 0.003.
This low value explains why many studies of the photodeg-
radation of pollutants in TiO2 suspensions show very low
quantum yield values for the disappearance of the pollutants or
the formation of degradation products, which are processes
measurement method, two systems, whose quantum yields are
well-known, were tested: the p-benzoquinone (BQ) system for
•-
the generation of the p-benzosemiquinone (BQ ) radial and
•
•
the H2O2 system for the homogeneous generation of OH
following the primary generation of OH radicals. For example,
2
7
radicals.
Matthews reported that the “apparent” quantum yield for the
disappearance of phenol from a 100 mM solution in 0.1% TiO2
Quantum Yield of the p-Benzosemiquinone Radical Genera-
2
8
suspension is ∼0.6%; Bahnemann et al. determined that the
tion in the BQ System. In an aqueous solution, p-benzosemi-
•-
“apparent” quantum yield of H2 formation in the illumination
quinone (BQ ) radicals can be generated through the photolysis
2
9
of p-benzoquinone (BQ):25
of a colloidal TiO2 solution was about 0.01. Anpo et al.
obtained ∼0.12% maximum “apparent” quantum yield for the
photocatalytic hydrogenolysis of methylacetylene (CH3CtCH)
with H2O in the TiO2 (anatase) particulate system.
hν
•-
BQ
9
8 BQ
(11)
Effects of Several Factors on the Quantum Yield. Loading
of the TiO2 Suspensions. Table 2 shows quantum yields
measured in suspensions with different TiO2 loadings. The
average value of the quantum yield is 0.039 ( 0.003. The
results indicate that variation of the loading in a fairly wide
The BQ•- radical has a five-line electron paramagnetic
resonance (EPR) spectrum. In a similar manner to eq 9, the
quantum yield φBQ for the BQ• radical generation can be also
expressed as
-

Photochemical Generation of Hydroxyl Radicals
J. Phys. Chem., Vol. 100, No. 10, 1996 4131
TABLE 1: Intermediate Data in the Quantum Yield Calculations
/M s-
1
R
Fe2+/M s-1
F
a
F
a
ρb
φ
system
R
s
Fe2+
s
-
-
-
7
7
7
-4
-4
-4
BQ
(5.50 ( 0.38) × 10
(3.35 ( 0.11) × 10
(3.16 ( 0.13) × 10
(4.30 ( 0.11) × 10
(4.56 ( 0.09) × 10
(4.56 ( 0.09) × 10
0.9977
0.9977
0.9977
0.0034
289 ( 8
1285 ( 11
58 ( 1
0.44 ( 0.05
0.94 ( 0.05
0.040 ( 0.003
-4
H
2
O
2
7.79 × 10
TiO
2
0.0172
a
b
Calculated using eq A1 of the Appendix for the wavelength range 300-400 nm. Calculated using eq 10.
TABLE 2: Effect of Suspension Loading on Quantum Yield of OH Radical Generationa
•
loading/mg L-
1
R
OH/M s-1
R
Fe2+/M s-1
F
F
ρb
φ
Fe2+
s
-
-
-
-
-
-
-
-
8
8
7
7
7
7
7
7
-4
-4
-4
-4
-4
-4
-4
-4
1
2
4
6
1
2
4
7
(3.59 ( 0.18) × 10
(7.40 ( 0.35) × 10
(1.37 ( 0.03) × 10
(2.15 ( 0.09) × 10
(3.16 ( 0.13) × 10
(3.32 ( 0.16) × 10
(3.42 ( 0.16) × 10
(3.49 ( 0.17) × 10
(4.56 ( 0.09) × 10
(4.56 ( 0.09) × 10
(4.56 ( 0.09) × 10
(4.56 ( 0.09) × 10
(4.56 ( 0.09) × 10
(4.56 ( 0.09) × 10
(4.56 ( 0.09) × 10
(4.56 ( 0.09) × 10
0.9977
0.9977
0.9977
0.9977
0.9977
0.9977
0.9977
0.9977
0.0021
0.0042
0.0082
0.0119
0.0172
0.0184
0.0192
0.0196
480 ( 6
238 ( 4
121 ( 2
84 ( 2
58 ( 1
54 ( 2
52 ( 1
51 ( 1
0.038 ( 0.003
0.039 ( 0.003
0.036 ( 0.002
0.034 ( 0.003
0.040 ( 0.003
0.039 ( 0.004
0.039 ( 0.003
0.039 ( 0.003
1
1
2
3
00
50
25
25
a
See footnotes to previous table.
•
Figure 5. Dependence of the quantum yield on the photon flux. TiO
2
Figure 6. Effect of H
2
O
2
addition on the OH radical generation rate:
loading is 100 mg L-
1
.
(a) TiO suspension + H O ; TiO loading is 100 mg L-1; (c) H O
2
2
2
2
2
2
only; (b) difference of curves a and c.
-
1
range (from 11 to 325 mg L ) does not alter the measured
quantum yield. This seems reasonable; according to eq 9, when
the loading of a suspension is high, the flux of absorbed photons
Reduction Process. The basic redox reactions that take place
on the surfaces of TiO2 particles are given in eqs 2b and 2c. As
a redox process, the consumption rates of holes and electrons
should be equal. But usually, on reaching the surface of the
particles, holes react with species adsorbed there much faster
than electrons. Therefore, kinetically eq 2c is considered as
a rate-controlling process. Accordingly, any methods that can
influence the rate of eq 2c, such as competing for electrons with
oxygen, that is, by adding electron acceptors into the suspension
•
is larger, which generates more OH radicals, that is, the
•
generation rate of OH radicals ROH increases. It is notable that
in Table 2 the increase of ROH and the decrease of the coefficient
FOH of the TiO2 suspensions are in about same proportion,
resulting in an invariant quantum yield.
3
2
Photon Flux. Figure 5 shows the relation of the quantum
yield with incident photon flux Io. It indicates that the quantum
yield decreases with increasing incident photon flux. This is
because increasing the incident photon flux level increases the
production rate of electron-hole pairs, which increases the
opportunity for electron-hole recombination, that is, it decreases
the relative number of the photoinduced carriers taking part in
the redox reaction on the surface. This results in a decrease of
(e.g., hydrogen peroxide) or by increasing the concentration of
oxygen, can promote the reduction process, avoid an accumula-
tion of the electrons on the particles (i.e., reduce the recombina-
tion rate), and thus increase the rate of the whole redox process,
which increases the quantum yield.
Addition of H2O2 as an Electron Acceptor. Figure 6 shows
3
0
the quantum yield when the photon flux is high.
The
the effect of different concentrations of H2O2 on the generation
•
measurements indicate that the quantum yield is inversely
proportional to the square root of the incident photon flux level,
rate of OH radicals in the TiO2 suspension. In the figure, curve
a is the case of the TiO2 suspension with H2O2; curve c is the
case of a homogeneous solution with same concentrations of
H2O2 as in curve a but without TiO2 particles; and curve b is
the difference between curves a and c, which represents the net
3
1
which is consistent with the result reported by Kormann et al.
in a study of the photocatalytic degradation of chloroform in
aqueous suspensions of TiO2. This suggests that the incident
photon flux level used in the experiment was in the limiting
range for the induction of carriers in the particles.
•
effect on the generation rate of the OH radicals by the addition
of H2O2 to the suspension. Accordingly, the effect of H2O2

4
132 J. Phys. Chem., Vol. 100, No. 10, 1996
Sun and Bolton
Figure 7. Effect of H
2
O
2
addition on the quantum yield; TiO
2
loading
Figure 8. Effect of oxygen concentration on the quantum yield; TiO
2
loading is 100 mg L .
-
1
-1
is 100 mg L
.
•
Effect of the Oxygen Concentration. As mentioned above,
in redox eqs 2b and 2c, the reduction reaction is a rate-
controlling step. Therefore, oxygen, as an electron acceptor,
can affect the efficiency of spontaneous photoelectrochemical
processes on semiconductor particles. Figure 8 shows the effect
of oxygen concentration in suspensions on the quantum yield
addition on the quantum yield of OH radical production was
obtained. Because the reduction of H2O2 by a conduction band
•
electron generates one additional OH radical for every one
produced by oxidation of water in reaction 2b, the measured
OH radical generation in the presence of the H2O2 radical
generation was divided by a factor of 2 to obtain the real
•
•
•
of the OH radical generation. The experimental results indicate
generation rate of the OH radicals produced by photoinduced
that the higher the content of dissolved oxygen in the suspension,
which is in equilibrium with the oxygen adsorbed on the
particles in the suspension, the larger the quantum yield. This
means that the redox reaction on the surface of the particles is
moderately influenced by the quantity of O2 adsorbed on the
holes from the semiconductor particles. The quantum yields
in the case of the addition of the H2O2 in Figure 7 were
calculated in this manner. Note that these quantum yields in
the presence of H2O2 are lower limits because we have assumed
that all of the incident UV light is available for the direct
photolysis of H2O2; in fact, some is absorbed by the TiO2.
Figure 7 indicates that the addition of the H2O2 to the
suspension can dramatically increase the quantum yield of
39
TiO2 particle surfaces. Okamoto et al. found a similar increase
of the initial reaction rate of phenol in a TiO2 suspension by
increasing the partial pressure of O2 in the bubbled gases.
•
•
Reliability of the Quantum Yield of OH Radical Genera-
OH radical generation (from 0.040 to 0.22). When the H2O2
concentration in the suspension is larger than 18 mM, a plateau
is reached.
tion in TiO2. The major uncertainty in our measurement of
•
the quantum yield of OH radical generation in TiO2 is the
•
efficiency of reaction 4 for reaction of OH radicals on the
An explanation of the above results is that the presence of
the H2O2 in the suspension can influence the redox reactions
happening on the surfaces of the particles. In the process, the
added H2O2 acts as an electron acceptor, which competes with
surface of TiO2. If this efficiency is less than 93%, our quantum
yield could be underestimated. We feel that this is unlikely for
three reasons:
(1) In ref 23, we have shown that when the concentration of
methanol is increased, the yield of the DMPO-OH spin adduct
reaches a plateau above 10 mM. We have used a concentration
of 30 mM which is well into the plateau region. Thus there is
enough methanol present to saturate reaction 4.
•
O2 in the redox reactions and produces additional OH radicals
at the same time:
-
•
-
H O + e f OH + OH
(14)
2
2
cb
(
2) Also in ref 23, we find that the trapping efficiency of
OH radicals by the spin trap DMPO is almost the same between
the homogeneous H O system and the heterogeneous TiO
•
The competition of the H2O2 with O2 for conduction band elec-
trons speeds up the reaction process and the whole redox
2
2
2
•
•
process, so as to increase the generation rate of OH radicals.
system. This means that the OH radicals on the surface of
TiO2 have a similar access to a trapping agent as in the
homogeneous solution.
•
Hence, the quantum yield of OH radical generation is greatly
increased.
We can now understand the reason for the increases of
degradation rates of organic pollutants by the addition of H2O2,
(3) The addition of H O caused our measured quantum yield
2
2
to increase to 0.22. If one can believe that this quantum yield
is really 1.0, then our values would be underestimated by a factor
of at most 4.5, which we find hard to believe. The findings of
which were reported by some researchers in several TiO2
suspension systems.3
6,37
Our work now confirms that this is
•
40
the result of the increase of the quantum yield for OH radical
production in these systems.
Brezova et al. are in agreement with this picture.
There has been some discussion that substrates can be
oxidized directly by valence-band holes rather than by hydroxyl
radicals generated by reaction 2b. We contend that direct
reaction of valence-band holes with organic substrates is only
The plateau in the high H2O2 concentration region (>18 mM)
perhaps indicates that a limiting reaction rate has been reached
in the system, perhaps due to reaction of hydroxyl radicals with
3
8
8
H2O2.
important at very high substrate concentrations. Lepore et al.

Photochemical Generation of Hydroxyl Radicals
J. Phys. Chem., Vol. 100, No. 10, 1996 4133
found a limiting quantum yield of about unity at very high
concentrations of propanal in optically transparent suspension
of very small TiO2 particles. They found that the propanal is
strongly adsorbed and the high quantum yields are likely due
to direct hole oxidation. Goldstein et al.4 suggested that at
low concentration of phenol, hydroxyl radicals are the primary
oxidant, whereas at high concentrations (>0.5 M), photogener-
ated holes directly oxidize phenol. At the concentration of
methanol employed in this work (30 mM), we are convinced
that almost all the oxidation of methanol occurs via reaction 4.
1
Figure 9. Integrating sphere assembly for the direct measurement of
the absorbance of a particle suspension.
5
. Conclusions
In this study, a method to measure the quantum yield of OH
S
is the fraction of light absorbed at wavelength λ, where Aλ is
absorbance of the sample S at wavelength λ.
•
radical generation in heterogeneous systems was developed. In
the method, a modified integrating sphere, which avoids the
interference of scattering light from the particulate system, was
used to measure the absorbance of the TiO2 suspensions, and a
ferrioxalate actinometer was used to measure the incident light
flux. Accordingly, the flux of absorbed photons could be
determined.
The incident photon flux Io can be determined by a standard
actinometer method, based on the photochemical conversion of
the ferrioxalate ion into Fe(II), as described by Calvert and
23
Pitts:
hν
3-
2+
2-
2
Fe(C O )
9
8 2Fe + 5C O4 + 2CO2 (A3)
2
4 3
2
The confirmation experiments of the method for the quantum
yield measurement were carried out in the p-benzoquinone (BQ)
and hydrogen peroxide (H2O2) systems, whose quantum yields
The generation rate of Fe2+ ions RFe2+ (M s ) can be
-1
determined spectrophotometrically at 510 nm after forming a
complex with 1,10-phenathroline. The incident photon flux Io
•-
•
for the p-benzosemiquinone (BQ ) and OH radical generation
are well-known. Good agreement with the literature values was
obtained, which shows that the method for the quantum yield
measurement is reliable.
-1
(
einstein s ) is then obtained from
Io ) RFe2+/φFe2+FFe(III)
(A4)
Here φFe2+ is the quantum yield of Fe2 generation in the
ferrioxalate reaction, and FFe(III) is the integrated absorption
fraction of the ferrioxalate solution over the range of the
wavelengths involved in the experiment.
+
•
The primary quantum yield of OH radical generation in TiO2
suspensions was determined for the first time to be 0.040 (
0
.003, which provides a basic estimation of the efficiency of
the primary process in heterogeneous photocatalysis in this
system.
The photon flux Ia absorbed by a sample is
Several effects on the quantum yield were observed. Varia-
tion of the TiO2 loading in the experimental range does not
change the quantum yield significantly. The quantum yield
changes inversely with the square root of the light intensity.
The reduction process has a marked influence on the quantum
yield. The addition of the electron acceptor hydrogen peroxide
caused a dramatic change of the quantum yield from 0.04 to
I ) I F ) R 2+FS/φFe2+FFe(III)
(A5)
a
o
S
Fe
From eqs A1 and A5 for the wavelength range 300-400 nm
400
300 λ
S
RFe2+
∫
I T f dλ
λ λ
Ia )
(A6)
400
300 λ
Fe(III)
φFe2+
∫
I T f
dλ
λ λ
0
.22, which is the result of changing the reduction rate of the
reduction reaction (a rate-controlling process) and inhibiting the
recombination of photoinduced electrons and holes. Similar to
the addition of H2O2, the variation of the O2 concentration in
the solution also changed the rate of the reduction reaction;
however, the effect of O2 concentration on the quantum yield
is not as large as the addition of H2O2, due to the limitation of
adsorption of O2 on the surface and to its low solubility in
water.
According to eq A6, it is necessary, in addition to the
measurement of the incident photon flux Io, that the absorp-
tion fraction fλS of the sample in each wavelength band
be determined. The integrals in eq A6 were approximated
by a sum over the wavelength range 300-400 nm taking a
finite bandwidth of 5 nm for each term in the summa-
tions.
Appendix 2
Acknowledgment. This work was financially supported by
an operating grant from the Natural Sciences and Engineering
Research Council. We thank Dr. Aitken R. Hoy for his helpful
comments.
The integrating sphere is a useful device to determine the
4
2
fraction of light absorbed in particulate suspensions. The
principle of the integrating sphere is that when incident light,
of photon flux Io, passes through a sample, light, both from
that transmitted through the sample and that scattered from the
particles, can be completely reflected by an inner highly
reflecting surface, composed of very fine BaSO4 particles. The
light radiation level in the sphere can then be monitored by a
detector.
In a normal measurement using the integrating sphere, a
sample cell (0.33 mm path length) is put at position a (as
shown in Figure 9) (note that it is important to use a thin
cell to avoid losses from the sides). The incident light, of
photon flux Io, passes through the sample, and the transmitted
light, of photon flux It, and the scattered light from the parti-
cles, of photon flux IS, all influence the reading of the detec-
tor in the integrating sphere. The apparent absorbance reading
Appendix 1
The terms in eq 8 are determined as follows: In general, FS
is given by
λ2
λ2
f
S
f
FS )
∫
I T f dλ/
∫
I T dλ
(A1)
λ
λ
λ
λ λ
λ1
λ1
where Iλ is the relative incident photon flux in the wavelength
band dλ, Tλ is the transmittance of the UV transmitting black
f
filter at wavelength λ, and
S
-AλS
fλ ) 1 - 10
(A2)

4
134 J. Phys. Chem., Vol. 100, No. 10, 1996
Sun and Bolton
(
(
3) Kamat, P. V. Chem. ReV. 1993, 93, 267.
4) Cundall, R. B.; Rudham, R.; Salim, M. J. Chem. Soc., Faraday
Trans. 1 1976, 72, 1642.
5) Harvey, P.; Rudham, R.; Ward, S. J. Chem. Soc., Faraday Trans.
1983, 79, 1381.
6) Harvey, P.; Rudham, R.; Ward, S. J. Chem. Soc., Faraday Trans.
1983, 79, 2975.
(
1
1
3
(
(
32.
7) Schiavello, M.; Augugliaro, V.; Palmisano, L. J. Catal. 1991, 127,
(8) Lapore, G. P.; Langford, C. H.; Vichova, J.; Vleck, A. Jr. J.
Photochem. Photobiol. A: Chem. 1993, 75, 67.
9) Valladares, J. E.; Bolton, J. R. In Photocatalytic Purification and
(
Treatment of Water and Air; Ollis, D. F., Al-Ekabi, H., Eds.; Elsevier:
Amsterdam, 1993; pp 111-120.
Figure 10. Integrating sphere assembly used in the modified method
for the absorbance measurement of a suspension: (a) reference
measurement; (b) sample measurement.
(10) Gibson, J. F.; Ingram, D. J. E.; Symons, M. C. R.; Townsend, M.
G. Trans. Faraday Soc. 1957, 53, 914.
(
11) Matheson, M. S.; Dorfman, L. M. Pulse Radiolysis; MIT Press:
Cambridge, MA, 1969.
12) Asmus, K.-D.; Mockel, H.; Henglein, A. J. Phys. Chem. 1973, 77,
218.
A′ of the instrument is
I - I - I
(
1
o
s
t
(13) Walling, C.; Kato, S. J. Am. Chem. Soc. 1971, 93, 4275.
A′ ) -log
(A7)
(
)
(14) Mopper, K.; Zhou, X. Science 1990, 250, 661.
Io
(
15) Zhou, X.; Mopper, K. Mar. Chem. 1990, 30, 71.
(16) Adams, G. E.; Michael, B. D.; Willson, R. L. Radiation Chemistry.
But this is not the true absorbance of the sample because the
back-scattered light, of photon flux Ib, from the particles is not
taken into account in the measurement. Thus a modified
method, which avoids this problem, was developed in this study.
In the modified method, a reference solution without particles
is put at position b between a reflection plate and the sphere
In AdV. Chem. Ser.; Gould, R. E., Ed.; American Chemical Society:
Washington, DC, 1968; Vol. 81, p 289.
(17) Mopper, K.; Stahovec, W. L. Mar. Chem. 1986, 19, 305.
(
18) Asmus, K.-D.; Wigger, A.; Henglein, A. Ber. Bunsen-Ges. Phys.
Chem. 1966, 70, 862.
19) Adams, G. E.; Willson, R. L. Trans. Faraday Soc. 1969, 65, 2981.
(20) Burchill, C. E.; Ginns, I. S. Can. J. Chem. 1970, 48, 1232, 2628.
21) Sun, L.; Schindler, K.-M.; Hoy, A. R.; Bolton, J. R. In Aquatic
(
(
(Figure 10). The instrumental reading A1 is
and Surface Photochemistry; Helz, G., Zepp, R., Crosby, D., Eds.; CRC
Press: Boca Raton, FL, 1994; pp 409-417.
Io - 2Ia
(22) Sun, L. Ph.D. Dissertation, The University of Western Ontario,
A1 = -log
(A8)
(
)
1994.
Io
(23) Sun, L.; Hoy, A. R.; Bolton, J. R. J. AdV. Oxidation Technol., in
press.
where Ia is the photon flux absorbed by the homogeneous
solution, and the coefficient 2 arises from the fact that the
incident light passes through the solution twice, once when the
incident light passes through the solution and second back again
after it reflects from the reflection plate. Equation A8 assumes
a small absorption fraction (Ia/Io), such that the beam Io is not
significantly attenuated between the incoming path and the
outgoing path. In the present case, the absorption fraction is
only about 1%, thus this approximation is reasonable. A sample
containing a suspension is then put at the same position as that
of the homogeneous solution (Figure 10b) and the instrumental
reading A2 is
(24) Calvert, J. G.; Pitts Jr., J. N. Photochemistry; Wiley: New York,
1967.
(25) Ononye, A. I.; Bolton, J. R. J. Phys. Chem. 1986, 90, 6270.
26) Zellner, R.; Exner, M.; Herrmann, H. J. Atmos. Chem. 1990, 10,
(
4
11.
27) Matthews, R. W.; McEvoy, S. R. J. Photochem. Photobiol. A:
Chem. 1992, 64, 93.
28) Bahnemann, D.; Henglein, A.; Spanhel, L. Faraday Discuss. Chem.
Soc. 1984, 78, 151.
29) Anpo, M.; Shima, T.; Kodama, S.; Kubokawa, Y. J. Phys. Chem.
1987, 91, 4305.
30) Lin, L.; Al-Thabaiti, S.; Kuntz, R. R. J. Photochem. Photobiol. A:
Chem. 1992, 64, 93.
31) Kormann, C.; Bahnemann, D. W.; Hoffmann, M. R. EnViron. Sci.
Technol. 1991, 25, 494.
32) This is the usual assumption that is made;
(
(
(
(
(
3
3,34
(
however, there is no
35
I - 2I - 2I
definitive evidence in the literature. For example, Gerischer, in his original
theoretical analysis of the kinetics of formation and decay of the transient
species in photolyzed TiO2, assumed that the rate constant for reaction 2b
is much faster than that for reaction 2c.
o
a
sa
A2 = -log
(A9)
(
)
Io
where Isa is the photon flux absorbed by particles in the
suspension.
(33) Gerischer, H.; Heller, A. J. Phys. Chem. 1991, 95, 5261.
(
34) Hidaka, M.; Kubota, H.; Gr a¨ tzel, M.; Serpone, N.; Pelizzetti, E.
NouV. J. Chim. 1985, 9, 67.
35) Gerischer, H. Electrochim. Acta 1993, 38, 3.
From eqs A8 and A9, the fraction fλ of light absorbed at
wavelength λ by the particles is
(
(36) Tanaka, K.; Hisanga, T.; Rivera, A. P. In Photocatalytic Purification
and Treatment of Water and Air; Ollis, D. F., Al-Ekabi, H., Eds.; Elsevier:
Amsterdam, 1993; pp 169-178.
(37) Wei, T. Y.; Wang, Y. Y.; Wan, C. C. J. Photochem. Photobiol. A:
Chem. 1990, 55, 115.
10-A1
-A2
Isa
- 10
2
fλ )
)
(A10)
(A11)
Io
and the true absorbance A is
(
38) Linden, L. A.; Rabek, J. F.; Kaminska, A.; Scoponi, M. Coord.
Chem. ReV. 1993, 125, 195.
39) Okamoto, K.; Yamamoto, Y.; Tanaka, H.; Tanaka, M.; Itaya, A.
(
Bull. Chem. Soc. Jpn. 1985, 58, 2015.
(40) Brezova, V.; Stasko, A.; Biskupic, S.; Blazkova, A.; Havlinova,
B. J. Phys. Chem. 1994, 98, 8977.
Isa
A ) -log 1 -
(
)
Io
(41) Goldstein, S.; Czapski, G.; Rabani, J. J. J. Phys. Chem. 1994, 98,
References and Notes
6586.
(42) Fischer, K. Beitr. Phys. Atmos. 1970, 43, 244.
(
(
1) Pelizzetti, E.; Minero, C. Electrochim. Acta 1993, 38, 47.
2) Fox, M. A.; Dulay, M. T. Chem. ReV. 1993, 93, 341.
JP9505800