Active agents in heterogeneous photocatalysis: Atomic oxygen species vs. OH. radicals: Related quantum yields

Article abstract of DOI:10.1002/1522-2675(20010919)84:9<2731::AID-HLCA2731>3.0.CO;2-L

After a general survey of the fundamental characteristics of heterogeneous photocatalysis, the present article classifies the ensemble of reactions into three major categories: i) mild oxidations, ii) total oxidations, and iii) reactions involving hydrogen. Depending on the presence or absence of H₂O, the active oxidizing species will be either a dissociated neutral oxygen species, denoted as O*, which is present in anhydrous systems and responsible for selective mild oxidation reactions, or an OH^. radical, formed in the presence of H₂O and responsible for totally degradative oxidation reactions. The existence of O* species is substantiated by photoconductivity measurements, oxygen-isotope exchange, and reactions in which oxygen-free NO is the oxidizing agent. The influence of the five main parameters that govern kinetics experiments i) the mass of the catalysts, ii) the wavelength, iii) the concentrations or partial pressures of reactants, iv) the temperature, and v) the radiant flux is examined to determine the best conditions for obtaining the optimum photocatalytic quantum yield (PQY), the definition of which is based on the quantum yield given in [1] for photochemistry.

Full text of DOI:10.1002/1522-2675(20010919)84:9<2731::AID-HLCA2731>3.0.CO;2-L

Helvetica Chimica Acta ± Vol. 84 (2001)
2731
Active Agents in Heterogeneous Photocatalysis:
.
Atomic Oxygen Species vs. OH Radicals: Related Quantum Yields
by Jean-Marie Herrmann
Laboratoire de Photocatalyse, Catalyse et Environnement, IFOS (UMR CNRS no 5621);
Â
Ecole Centrale de Lyon, BP 163, F-69131 Ecully Cedex
(phone: (33) 472186493; fax: (33) 478330337; e-mail: jean-marie.herrmann@ec-lyon.fr)
Â
Dedicated to Professor Andre M. Braun on the occasion of his 60th birthday
After a general survey of the fundamental characteristics of heterogeneous photocatalysis, the present
article classifies the ensemble of reactions into three major categories: i) mild oxidations, ii) total oxidations,
and iii) reactions involving hydrogen. Depending on the presence or absence of H2O, the active oxidizing species
will be either a dissociated neutral oxygen species, denoted as O*, which is present in anhydrous systems and
.
responsible for selective mild oxidation reactions, or an OH radical, formed in the presence of H2O and
responsible for totally degradative oxidation reactions. The existence of O* species is substantiated by
photoconductivity measurements, oxygen-isotope exchange, and reactions in which oxygen-free NO is the
oxidizing agent. The influence of the five main parameters that govern kinetics experiments i) the mass of the
catalysts, ii) the wavelength, iii) the concentrations or partial pressures of reactants, iv) the temperature, and v)
the radiant flux is examined to determine the best conditions for obtaining the optimum photocatalytic quantum
yield (PQY), the definition of which is based on the quantum yield given in [1] for photochemistry.
1. Introduction. ± Heterogeneous photocatalysis is a discipline that includes a large
variety of reactions: selective mild oxidations, total oxidations, dehydrogenation, hydrogen
transfer, 16O2 18O2 and alkane-deuterium isotopic exchange, noble metal deposition
for separation and recovery, water detoxification, gaseous pollutant removal, etc. In line
with the two latter points, photocatalysis can be considered one of the new ꢀAdvanced
Oxidation Technologiesꢁ (AOT) for air and water purification treatment. Several books
and reviews have been recently devoted to this problem [2± 7]. A recent review reported
more than 1700 references on the subject [8] and constitutes the basic reference for the
best exhaustive report of global activity on heterogeneous photocatalysis.
Heterogeneous photocatalysis can be carried out in various media: gas phase, pure
organic liquid phase, or aqueous solution. As for classical heterogeneous catalysis, the
overall process can be broken down into five independent steps: 1) transfer of the
reactants from the fluid phase to the surface, 2) adsorption of at least one of the
reactants, 3) reaction in the adsorbed phase, 4) desorption of the product(s), or 5)
removal of the products in the fluid phase.
The photocatalytic reaction occurs in the adsorbed phase during Step 3). The only
difference with conventional catalysis is the mode of activation of the catalyst, in which
the thermal activation is replaced by a photonic activation as developed in the next
section. The activation mode is not concerned with the remaining steps, although
photo-adsorption and photo-desorption of some reactants, mainly oxygen [9] and
oxygen-free NO [10], does occur.

2732
Helvetica Chimica Acta ± Vol. 84 (2001)
2. Fundamentals of Heterogeneous Photocatalysis. ± 2.1. Principle of Heteroge-
neous Photocatalysis. When a semiconductor catalyst (SC) of the chalcogenide type
(oxides (TiO2, ZnO, ZrO2, CeO2, ...), or sulfides (CdS, ZnS, ...)) is illuminated with
photons whose energy is equal to or greater than their band-gap energy EG (hn ꢀ EG),
the absorption of these photons by the solid occurs and is followed by the creation,
within the bulk, of electron-hole pairs, which dissociate into free photoelectrons in the
conduction band and photoholes in the valence band, the photoholes being virtual
positive-charge carriers corresponding to an electron vacancy in the pool of lattice
anions (O2 in oxides and S2 in sulfides).
Simultaneously, in the presence of a fluid phase (gas or liquid), a spontaneous
adsorption occurs and, according to the redox potential (or energy level) of each
adsorbate, electron transfer proceeds towards acceptor molecules, whereas positive
photoholes are transferred to donor molecules (actually the hole transfer corresponds
to the cession of an electron by the donor to the solid).
‡
hn ‡ (SC) ! e ‡ p
A (ads) ‡ e ! A (ads)
‡
‡
D (ads) ‡ p ! D (ads)
Each ion formed subsequently reacts to form the intermediates and final products.
Gaseous acceptor species can be O-species and/or O2 molecules [9] and O2-free NO
[10]. In H2O, H2O2 and/or S2O28 can be considered electrophilic reactants or electrons
scavengers [11]. Donor species can be either negatively charged ionosorbates such as
O , O2 , NO , and OH , or organic molecules such as alcohols and carboxylic acids,
both dissociated in their reactive adsorbed states:
.
‡
R
O ‡ p ! R
O
.
‡
R
COO ‡ p ! R ‡ CO2
the last reaction being known as the ꢀphoto-Kolbeꢁ reaction.
As a consequence of the three reactions above, the photonic excitation of the
catalyst appears to be the initial step of the activation of the whole catalytic system.
Thence, the efficient photon has to be considered a reactant and the photon flux a
special fluid phase, the so called ꢀelectromagnetic phaseꢁ [12]. The photon energy is
adapted to the absorption of the catalyst, not to that of the reactants. The activation of
the process proceeds via excitation of the solid rather than that of the reactants: there is
no photochemical process in the absorbed phase but only a true heterogeneous
photocatalytic regime, as demonstrated further.
The photoefficiency can be reduced by electron-hole recombination, which
corresponds to loss of the photoelectronic energy as heat.
‡
e ‡ p ! N ‡ E
(N: neutral center; E: dissipated energy (light hn' ꢁ hn or heat)

Helvetica Chimica Acta ± Vol. 84 (2001)
2733
2.2. Photocatalysts. Various chalcogenides (oxides and sulfides) have been used:
TiO2, ZnO, CeO2, CdS, ZnS, etc. As generally observed, the best photocatalytic
performance with maximum quantum yields are always obtained with TiO2, since it
better combines its adsorptive properties with respect to reactants with its absorptive
properties with respect to photons [6]. In addition, anatase is the most active allotropic
form of TiO2 among the various ones available, either natural (rutile and brookite) or
artificial (TiO2 B, TiO2 H). Anatase is thermodynamically less stable than rutile, but
its formation is kinetically favored at lower temperature (<6008). This lower synthesis
temperature could account for a higher surface area and a higher surface density of
active sites for adsorption and for catalysis. It is generally admitted that the surface sites
of titania, especially in aqueous medium, can be considered as surface OH groups,
identified and titrated by the pioneer work of Böhm [13]. In most of the systems
described in the literature, the catalyst used was mainly anatase and, in particular,
titania (Degussa TiO2 P-25, 50 m2/g, 70 ± 80% anatase). The systems described below
will concern this photocatalyst, unless otherwise stated.
3. Main Types of Photocatalytic Reactions. ± 3.1. Mild Oxidation Reactions. The
gas-phase oxidations in which O2 in air is the oxidizing agent mainly concern the mild
oxidation of alkanes, alkenes, and alcohols to aldehydes and ketones [14]. Inorganic
gases such as CO, NH3 or H2S could also be oxidized: NH3 was oxidized mainly to N2
[15]. NO could be photocatalytically decomposed into N2 at low pressures and into N2O
at higher pressures, whereas the O2 generated could be used to oxidize butanols to
butanal and butanone [10].
Liquid-phase reactions concerned the selective mild oxidation of liquid hydro-
carbons (alkanes, alkenes, cycloalkanes, aromatics) to aldehydes and ketones [16a]. For
instance, cyclohexane and decahydronaphthalene were oxidized to cyclohexanone and
octahydronaphthalen-2-one, respectively, with identical selectivities of 86% [16b].
Aromatic hydrocarbons [17] such as alkyltoluenes or o-xylenes were selectively
oxidized on the Me group into alkylbenzaldehydes:
R
C6H4 CH3 ‡ O2 ! R C6H4 CHO ‡ H2O
Pure liquid alcohols were also oxidized to their corresponding aldehydes or ketones.
In particular, the oxidation of i-PrOH to acetone was chosen as a photocatalytic test for
measuring the efficiency of passivation of TiO2- or ZnO-based pigments in painting
against weathering.
The high selectivity was ascribed to a photoactive neutral, atomic oxygen species
[14 ± 18]
‡
O (ads) ‡ p ! O*
The existence of such a dissociated activated species was put into evidence by various
means, mainly based on photoconductivity measurements.
The creation of photo-electrons and holes can be directly detected by photo-
conductivity s measurements. The variation in s as a function of the incident photonꢁs
wavelength was found to be parallel to the absorbance of the catalyst. Under O2, s was

2734
Helvetica Chimica Acta ± Vol. 84 (2001)
found to vary as a function of PO 1/2 at low (decreasing) pressures and as PO 1 at higher
2
PO . This has been ascribed to dissociative and associative O2 ionosorption 2
2
‡
(TiO2) ‡ hn ! e ‡ h
O2 (g) > 2 O (ads)
O (ads) ‡ e
O2 (g) > O2 (ads)
O2 (ads) ‡ e O2 (ads)
>
O (ads)
>
In a pure organic (gaseous or liquid) phase, O2 ionosorbates are the only negatively
charged species able to react with holes.
‡
O2 (ads) ‡ h ! O2* (ads)
‡
O (ads) ‡ h ! O* (ads)
The reaction heat is partially dissipated by O2 desorption, which is endothermic.
However, what is the active one in partial mild oxidation?
In situ photoconductivity s measurements were performed in a photoreactor-cell,
specially designed to simultaneously follow s and the reaction rates as a function of the
partial pressures of reactants, in the partial oxidation of isobutane to acetone.
The results were:
1
s ˆ k PO Pi0
2
rˆ k' PO0 Pi0.35
2
The partial orders of both reactants with respect to s clearly indicated that the alkane
has no electronic interactions with titania (s / Pi0), whereas O2 is mainly adsorbed as
the O2 species. By contrast, isobutane has a partial order (0.35) with respect to the
reaction rate. This is in line with a Langmuir-Hinshelwood mechanism with r ˆ k qi ˆ
k K Pi/(1 ‡ K Pi) ꢂ k K' Pi0.35
.
Oxygen is zero order with respect to r, whereas its order with respect to s is 1,
indicative of the presence of O2 (ads) species. The conclusion was reached that,
although O2 species are present, they are not the active species in mild oxidation
reactions; these are instead assumed to be dissociated activated O* species, as has been
corroborated by various observations: 1) Oxygen isotopic exchange (OIE) followed
the ꢀR' mechanismꢁ based on the existence of a dissociated oxygen species (16OS) at the
surface:
18O2 (g) ‡ 16OS
!
18O 16O (g) ‡ 18OS
2) The same OIE performed in the presence of isobutane initiated only when isobutane
had totally reacted, clearly indicating the concurrence for the same species. 3) Mild
oxidation reactions could be performed with O2-free NO acting as a source of O-atoms
according to:

Helvetica Chimica Acta ± Vol. 84 (2001)
2735
NO ‡ e ! NO
.
‡
NO ‡ h ! NO* ! N ‡ O*
.
.
N ‡ N ! N2(g)
O* could be used in situ as the oxidizing agent i) of BuOH to butanone [10]
C4H9 OH ‡ O* ! C4H8O ‡ H2O
and ii) also of CO to CO2 [19] as illustrated in Fig. 1.
2 NO ‡ 2 CO ! 2 CO2 ‡ N2
Besides the experimental contribution to proof the existence of dissociated oxygen
species, this reaction indicates that two toxic gases could be eliminated by conversion to
two innocuous ones, CO2 and N2, in an anaerobic gas phase. 4) All the photoactive
oxide catalysts, besides titania, viz. ZrO2, ZnO, CeO2, Sb2O4, exhibited dependence of
s on PO 1/2 [9].
2
Fig. 1. Photocatalytic oxidation of CO to CO2 by NO in contact with UV-irradiated titania (Degussa P-25).
Partial pressures 1 Torr; CO (^), NO (&), CO2 (~), N2 (Â), N2O (n).
3.2. Total Oxidation Reaction in Aqueous Phase. The selective mild oxidation
reaction could be obtained in pure gaseous or liquid organic phases. By contrast, as
soon as H2O is present, the selectivity changes to favor total oxidative degradation. This
was ascribed to the photogeneration of stronger, unselective, oxidant species, namely
.
OH radicals originating from H2O via the OH groups on titaniaꢁs surface. In aqueous
media, because of the high concentration of H2O (55.55 mol/l) and of the highly
hydrophilic character of titania, its surface can be assumed to be saturated not only in
chemisorbed H2O but especially in OH groups, whose coverage at saturation is

2736
Helvetica Chimica Acta ± Vol. 84 (2001)
estimated to be equal to 5 Â 1018/m2 [13]. These OH groups are continually renewed
by the ionic dissociation of H2O, and constitute the main negatively charged species
that react with holes:
‡
H2O > H ‡ OH
OH ‡ p ! OH
.
‡
substrate
.
OH
Intermediates ! Final Products (CO2, H2O, X , A ...)
ƒƒƒƒ!
This system is the most promising issue for application of heterogeneous photo-
catalysis, since it is directly connected to water detoxification and to pollutant removal
in aqueous effluents.
The most spectacular results were obtained in the field of organic-pollutants
.
removal [2 ± 8]. OH Radicals are especially efficient in the destruction of aromatic
contaminants since they are able to substitute heteroatoms such as halides on an
aromatic ring to give the corresponding phenolic compounds. An example is given for
the degradation of pesticide tetrachlorvinphos in Scheme 1.
Similarly, various toxic anions can be oxidized into harmless or less toxic
compounds with TiO2 as the photocatalyst. For instance, nitrite is oxidized to nitrate
[20][21], phosphite to phosphate [22], and sulfide to sulfite [23], and thiosulfate [24] to
sulfate, whereas cyanide is converted either to isocyanide [25], N2 [26], or nitrate [27].
Heavy metals are generally toxic and can be removed from industrial waste
effluents [24][28 ± 31] as small crystallites deposited on the photocatalyst according to:
n
hn
.
‡
Mn‡ ‡ H2O
M ‡ n H ‡ O2
ƒƒƒƒƒƒ!
semiconductor
4
provided the redox potential of the cation metal couple is higher than the flat-band
potential of the semiconductor. Under identical conditions, the following reactivity
pattern was found: Ag > Pd > Au > Pt ꢃ Rh ꢃ Irꢃ CuˆNiˆFe ˆ 0. Whereas photo-
electrons were consumed to reduce the noble metals, photoholes oxidized H2O to O2
via the neutralization of OH by holes.
2.3. Photocatalytic Reactions Involving Hydrogen. In photocatalytic reactions
involving hydrogen, either as a reactant (deuterium-alkane isotopic exchange (DAIE)
[32]) or as a product (alcohol dehydrogenation [33]), the system requires the presence
of a metal to act as a cocatalyst to i) dissociate the reactant (D2) and ii) recombine H
and D into H2, HD, or D2. Additionally, the metal i) attracts electrons by photoinduced
metal-support interaction (PMSI), ii) decreases the electron-hole recombination, and
iii) makes the reaction run catalytically as exemplified by the two cyclic mechanisms in
Schemes 2 and 3. In Scheme 2, it appears clearly that DAIE occurs via deuterated
.
hydroxyl radicals (OD ), whereas in Scheme 3, alcohols react directly with holes
because of their dissociative adsorption as alcoholate plus a proton.
.
4. O* vs. OH Reactivities. ± From the above considerations, two oxidizing active
species can be generated at the surface of UV-irradiated titania, depending on the
reaction medium.
In anhydrous media, mild oxidation reactions can be obtained that lead to very high
selectivities. By contrast, in the presence of H2O (or of humid air), selectivities change

Helvetica Chimica Acta ± Vol. 84 (2001)
2737
Scheme 1. Tetrachlorvinphos Degradation before the Opening of the Aromatic Ring
from partial to total oxidation and, generally, final products can reach their upper ±
often harmless ± oxidation states. This is true for both organic and inorganic compounds.
For instance, dry gaseous NH3 was found to oxidize mainly to N2 (selectivity of
‡
80%) [14], whereas NH4 is oxidized to nitrate via nitrite ions in aqueous solution.
Similarly, dry o-xylene is 100% oxidized to 2-methylbenzaldehyde, whereas, with

2738
Helvetica Chimica Acta ± Vol. 84 (2001)
Scheme 2. Eight-Step Cyclic Mechanism of Deuterium-Cyclopentane Isotopic Exchange on UV-Irradiated
Pt/TiO2 Catalyst
humid air, it is oxidized to CO2 via the formation of phenolic intermediates [34]. This
reaction has been selected as a model reaction to illustrate the fate of aromatic
hydrocarbons in the (humid) atmosphere in the presence of solid aerosols originating
from fly ash [34].
5. Examination of Common Physical Parameters Governing the Kinetics of
Photocatalytic Reactions. ± Whichever reaction is considered among the very different
ones described above, all follow common physicochemical laws, whose expressions are
summarized in the five diagrams of Fig. 2.
5.1. Mass of the Catalyst. Either in static, or in slurry, or in dynamic-flow
photoreactors, the initial rates of reaction (r) were found to be directly proportional to
the mass (m) of catalyst (Fig. 2,a). This is indicative of a true heterogeneous catalytic
regime in which r increases proportionally with the total surface area exposed (ST ˆ
SBET Â m), i.e., with the total number of surface sites. However, above a certain value of
m, r levels off and becomes independent of m. This limit depends on the geometry and
on the working conditions of the photoreactor (catalytic bed, volume, light flux, ...),
and was found to be equal to 2.5 mg TiO2 per cm3 of suspension (2.5 g/l) in the static
photoreactor described in Fig. 3. This optimum value was found to be equal to 1.3 mg
TiO2/cm2 deposited on a plate of the fixed bed [35] and to 0.2 g/l in the solar
photoreactor at Plataforma Solar de Almeria (Spain) [11][36]. These limits correspond

Helvetica Chimica Acta ± Vol. 84 (2001)
2739
Scheme 3. Eight-Step Cyclic Mechanism of Alcohol Dehydrogenation on UV-Irradiated Pt/TiO2 Catalyst
to the maximum amount of TiO2 that can be totally illuminated. For higher quantities
of catalyst, a screening effect of excess particles occurs, which masks part of the
photosensitive surface. In some cases, a high excess of mass of titania has even been
found to decrease the reaction rate beyond the plateau of Fig. 2,a. This has been
thought to be related to increased scattering of photons out of the reactor as catalyst
concentration near the reaction wall is increased, as has already been reported [37].
For any study or application, this optimum mass of catalyst has to be chosen i) to
avoid an useless excess of catalyst and ii) to ensure total absorption of efficient photons.
This value can vary over a rather wide range, e.g., from 2.5 g TiO2/l in laboratory
experiments, with a batch photoreactor, down to 0.2 g/l for the CPC solar reactor at
PSA [11][36], which is a recirculation plug flow reactor corresponding to a batch
reactor. Actually, this optimum varies with the optical pathway within the suspension.
5.2. Wavelength. The variation in r as a function of the wavelength (l) follows the
absorption spectrum of the catalyst (Fig. 2,b), with a threshold corresponding to its
band-gap energy. For TiO2 having EG ˆ 3.02eV, this requires l ꢁ 400 nm, i.e., near-UV
wavelength (UV-A). In addition, it must be ensured that the reactants do not absorb
light so that exclusively the catalyst undergoes photoactivation for a true heteroge-
neous catalytic regime (no homogeneous catalysis nor any photochemistry in the
absorbed phase). Actually, the variations rˆ f(l) can only be obtained with sensitive
reactions such as 16O2 18O2 isotopic exchange [10] or facile reactions having a high
quantum yield (defined further), such as MeOH dehydrogenation [33].

2740
Helvetica Chimica Acta ± Vol. 84 (2001)
Fig. 2. Variation in photocatalytic activity of a solid, as estimated by the reaction rate (r) as a function of the five
basic parameters that govern the kinetics. a) Mass of catalyst, b) wavelength, c) concentration (or partial
pressure), d) temperature (expressed as its reciprocal in an Arrhenius plot), and e) radiant flux (in W/m2).
5.3. Concentrations and/or Partial Pressures. Generally, the kinetics formally follows
a Langmuir-Hinshelwood mechanism, confirming the heterogeneous-catalytic charac-
ter of the system with the rate r varying proportionally with the coverage q as:
rˆ k q ˆ k (K C/(1 ‡ K C))
where k is the true rate constant and K is the adsorption constant. Two limiting cases
are often encountered. For dilute solution, when the product KC becomes ꢄ1, the
reaction becomes apparently first order.
r ꢂ k K C ˆ kapp
C

Helvetica Chimica Acta ± Vol. 84 (2001)
2741
Fig. 3. Schematic diagram of a laboratory-scale batch slurry photoreactor used for mild oxidation of organics,
oxidative water decontamination, and alcohol dehydrogenation
On the contrary, for high concentrations when KC becomes ꢃ1, the surface coverage
becomes saturated. Correspondingly, the reaction appears to be zero order and r is
maximum and equal to k (Fig. 2,c). In between, the reaction order with respect to
concentration varies between 1 and 0. Over a narrow enough concentration range, the
homographic function q ˆ k (KC/(1 ‡ KC)) can be assimilated to a power function q ˆ
k Ca, where a is constant and 0 < a < 1.
For gas-phase reactions, similar Langmuir-Hinshelwood expressions are found,
which are expressed with partial pressures P instead of C. Kinetic studies can provide
information about the mode of adsorption of the reactant. In some cases, such as in
liquid alcohol dehydrogenation [33], r follows variations including the square root of
the concentration:
rˆ k [K1/2 C1/2/(1 ‡ K1/2 C1/2)]
This indicates that the reactant is in a dissociated adsorbed state. For maximum yield,
reactions should be performed with initial concentrations equal to or higher than the
threshold of the plateau (C0 ꢀ ca. 5 Â 10 3 m). In reactions to purify and detoxify water,
pollutant concentrations are generally (and fortunately) low, and r is expected to be
substantially lower than the optimum value at surface saturation.

2742
Helvetica Chimica Acta ± Vol. 84 (2001)
5.4. Temperature. Because of the photonic activation, the photocatalytic systems do
not require heating and are operating at room temperature. The true activation energy
Et is zero, whereas the apparent activation energy Ea is often very small (a few kJ/mol)
in the medium temperature range (208 ꢁ q ꢁ 808). However, at very low temperatures
(
408 ꢁ q ꢁ 08), the activity decreases and the apparent activation energy Ea becomes
positive (Fig. 2,d). By constrast, at ꢀhighꢁ temperatures (q ꢀ 70 ± 808) for various types
of photocatalytic reactions, the activity decreases and the apparent activation energy
becomes negative (Fig. 2,d). This behavior can be easily explained within the
framework of the Langmuir-Hinshelwood mechanism described above for rˆ f(C).
The decrease in temperature favors adsorption, which is a spontaneous exothermic
phenomenon. In the rate equation, q tends to unity, whereas KC becomes ꢃ 1. In
addition, lowering T also favors adsorption of the final reaction product P, whose
desorption tends to become the rate-limiting step. Therefore, r becomes:
rˆ k q ˆ k K C/(1 ‡ K C ‡ KP CP) ꢂ k/KP CP
where KP is the adsorption constant of final product P. Correspondingly, the apparent
activation energy Ea becomes equal to:
Ea ˆ Et DHP
i.e., to the sum of the true activation energy Et (theoretically equal to zero because of
the photonic-activation mode of the catalysts) and of the enthalpy of adsorption of
product P, which is negative (DHP ˆ QP, QP being the heat of adsorption counted
positive):
Ea ˆ Et ‡ QP ꢂ 0 ‡ QP ˆ QP
This relationship was demonstrated for reactions involving H2 (alcohol dehydro-
genation [33] or deuterium-alkane isotopic exchange [32]), which are carried out on
bifunctional Pt/TiO2 photocatalysts. Ea was found equal to ‡ 10 kcal/mol (‡42 kJ/
mol), which is just equal to the heat QH (ads) (or to the opposite of the enthalpy DHH (ads)
)
2
2
of the reversible adsorption of H2 on Pt measured by microcalorimetry [38]:
Ea ˆ Et DHH (ads) ˆ 0 ‡ QH (ads) ˆ ‡10 kcal/mol
2
2
On the contrary, when q increases above 808 and approaches the boiling point of
H2O, the exothermic adsorption of reactant A becomes disfavored and tends to become
the rate-limiting step, with r becoming:
rˆ k q ˆ k K C/(1 ‡ K C) ꢅ k K C
Correspondingly, the apparent activation energy becomes:
Ea ˆ Et ‡ DHA ˆ 0 QA < 0
This indicates that the adsorption of the reactant A becomes the rate-limiting step.

Helvetica Chimica Acta ± Vol. 84 (2001)
2743
Ea ˆ Et aQA ˆ 0 aQA < 0,
with a being a coefficient that varies progressively from 0 to 1 as the temperature
increases and as adsorption of reactant A becomes more and more the rate-limiting
step.
As a consequence, the optimum temperature is generally 20 ± 808, explaining why
previous solar devices that use sunlight concentrators including IR beams require coolers
[39]. That it is unnecessary to heat the catalyst to activate it is attractive for
photocatalytic reactions carried out in aqueous media and, in particular, for environ-
mental purposes (photocatalytic water purification). There is no nead to waste energy
in heating water, which possesses a high heat capacity, and explains why photocatalysis
is cheaper than incineration [40].
5.5. Radiant Flux. In laboratory experiments, the artificial light power used was
determined by measuring the efficient radiant flux F (in mW/cm2) with a radiometer,
and corresponded to 21% of the electrical power consumed. It has been shown for all
types of photocatalytic reactions that r is proportional to the radiant flux F (Fig. 2,e).
This confirms the photoinduced nature of the activation of the catalytic process, with
the participation of photoinduced electrical charges (electrons and holes) in the
reaction mechanism. However, above a certain value, which varies for each photo-
system used, r becomes proportional to F1/2. This can be demonstrated with the
following five basic equations:
‡
(TiO2) ‡ hn ! e ‡ p
(1)
(2)
(3)
(4)
(5)
‡
e ‡ p ! N ‡ energy
A ‡ e ! A
‡
‡
D ‡ p ! D
‡
A ‡ D ! Intermediates ! Products
and the rate-limiting step is the reaction in the adsorbed phase (Eqn. 5) according to a
Langmuir-Hinshelwood mechanism. Therefore
‡
rˆ r5 ˆ k5 [A ] [D ]
(6)
In an n-type semiconductor such as titania, the photoinduced-hole concentration is
much smaller than that of electrons, including both photoinduced electrons and n-type
‡
electrons originating from titaniaꢁs oxygen substoichiometry: [p ] ꢄ [e ]. Therefore
holes are the limiting active species. Thence:
‡
rˆ r5 ˆ r4 ˆ k4 [D] [p ]
(7)
(8)
At any instant, one has:
‡
‡
‡
d[p ]/dt ˆ r1 r2 r4 ˆ 0 ˆ k1 F k2 [e ] [p ] k4 [D] [p ]

2744
Helvetica Chimica Acta ± Vol. 84 (2001)
Thence:
k1 F
‡
[p ] ˆ
(9)
k2 ‰e Š ‡ k4 ‰DŠ
and
k1 k4 ‰DŠ F
k1 k4 ‰DŠ F
rˆ
ˆ
(10)
k2 ‰e Š ‡ k4 ‰DŠ K4 ‰DŠ ‡ k ‰e Š
From the Eqn. 10, it can be seen that r is directly proportional to the F.
‡
In the case of high fluxes, the instantaneous concentrations [e ] and [p ] become
much larger than k4 [D]. Therefore Eqn. 9 becomes:
k1 F
‡
‡
[p ] ꢅ
with [p ] ꢅ [e ]
(11)
(12)
k2 ‰e Š
Thence:
‡
[p ]2 ꢅ k1 F/k2
From Eqns. 7 and 12, r becomes:
rˆ r5 ˆ r4 ˆ k4 [D] (k1 F/k2)1/2
(13)
which means that r is proportional to F1/2 and that the rate of electron-hole formation
becomes greater than the photocatalytic rate. Since the electron-hole recombination is
a bimolecular reaction process, the second-order rate is favored by an increase in
concentrations. This means that r becomes dominated by the electron-hole recombi-
nation. Although r is still increasing with radiant flux, most of photons generate
electrical charges that recombine, thus dissipating noble UV-light energy into useless
heat. The optimal light-power utilization corresponds to the domain where r is
proportional to F, and it is of a paramount importance to determine it in any
photocatalytic device aimed at optimal use of photons.
In laboratory experiments, near-UV light was provided by a Philips UV-lamp
(HPK 125 watts), described in [1] and placed in front of the optical window of the
photoreactor as described in Fig. 3. IR Beams were removed with a circulating-water
cell. The wavelength domain was adjusted with optical filters (fused silica, or Pyrex, or
Corning glass filters), and the radiant flux was measured with a radiometer (Radio-
meter Technology Model 21A) calibrated against a calorimeter. Under these conditions,
a transition value of 25 mW/cm2 was found for F, e.g., in the oxidation of cyclohexane
in cyclohexanone [16a]. It should be noted that the rate is proportional to the efficient
photon flux, i.e., to the total number of incident photons able to create electron-hole
pairs. However, for a given light source, the efficient part of the spectrum remains
constant. This is why r is apparently proportional to the overall incident radiant flux FT.
5.6. Influence of O2 Pressure in Oxidation Reactions. The influence of O2 pressure is
easily determined in gas-phase reactions. For instance, in the mild photocatalytic
oxidation of alkanes to aldehydes and/or ketones, it was found that the reaction was
zero-order in O2, indicating a total surface coverage of this gas [18]. Photoconductivity

Helvetica Chimica Acta ± Vol. 84 (2001)
2745
measurements indicated that O2 was photoadsorbed as O2 [18]. For liquid-phase
reactions, it was difficult to study the influence of PO because the reaction is polyphasic.
2
It is generally assumed that O2 is adsorbed on titania from the liquid phase, where it is
dissolved according to Henryꢁs law. If O2 is continually supplied to the aqueous
suspension, it can be assumed that its coverage at the surface of titania is constant and
can therefore be integrated into the apparent rate constant:
A ‡ O2 ! P
rA ˆ d [A]/dt ˆ k qA q ˆ kapp qA
O2
Actually, the apparent rate constant is a function of the power flux and of the O2
coverage. No examples in the literature have yet been found on the use of a
photoreactor working under high O2 pressure, which would substantially increase the
reaction rate.
5.7. Stoichiometric Threshold. The stoichiometric threshold is defined as the number
of moles converted above which a photo-assisted reaction can be considered truely
catalytic. The first observation was of the photocatalytic dehydrogenation of aliphatic
C1 ± C4 alcohols performed on titania-deposited noble metals [33]. In that reaction, it
was demonstrated that titania was the photoactive agent but required a noble metal
additive in order to i) attract photogenerated electrons, ii) neutralize protons to H-
atoms, iii) recombine the H-atoms to H2, and iv) desorb H2 in the gas phase.
Without Pt, titania was able to produce a certain amount of H2, but, after a period of
ca. 1 h depending on the conditions, the evolution of H2 stopped. The number of moles
evolved was considered the stoichiometric threshold, corresponding to the exhaustion
of nonrenewable active sites. In presence of Pt, the evolution of H2 could be performed
over long UV-irradiation periods, enabling us to reach conversions 1000 Â greater than
the photocatalytic threshold without any loss of activity, warranting the catalytic nature
of the reactions. Similarly, the liquid-phase oxidation of 4-tert-butyltoluene in
4-tert-butylbenzaldehyde [17] and cyclohexane in cyclohexanone [16a] gave conver-
sions that exceeded the photocatalytic threshold by factors of 1000 and 100,
respectively.
In aqueous media to be detoxified, pollutant concentrations are generally rather
.
low. Since OH radicals, originating from the neutralization of OH surface groups by
photoholes, are generally considered the degrading oxidative species, the stoichio-
metric threshold has to be determined by calculating the initial number no of OHS
groups:
no ˆ m ´ SBET ´ d
where m is the mass of catalyst used, SBET is its surface area and d is the surface OH
group density at saturation, which can be taken to be 5 Â 1018/m2 according to Böhm
[13]. For instance, in the solar pilot plant at PSA, with 50 g of Degussa P-25 at 50 m2/g,
one obtains a maximum initial number of 2.075 Â 10 2 mol of OH groups. The
total degradation of 1.4 Â 10 4 mol/l of herbicide 2,4-D in 250 l corresponded to 3.5 Â
10 2 mol [41]. The stoichiometric threshold has been exceeded by a factor of 1.7 for
one experiment in dilute medium. Similarly, the photocatalytic degradation of

2746
Helvetica Chimica Acta ± Vol. 84 (2001)
4-chlorophenol by the titania-activated carbon system reused several times in the same
pilot reactor has converted a number of moles 5-fold higher than the stoichiometric
threshold without any loss of activity [41].
When it exceeds unity, this parameter ensures that the reaction is truly catalytic. It
should be systematically determined, especially when using modified titania, to check
whether any additives or modifications really contribute to the catalytic process of the
reaction or artificially to parallel side stoichiometric processes.
6. Photocatalytic Quantum Yield. ± The concept of photocatalytic quantum yield
(PQY, 1) must be clearly defined from the two parent disciplines, photochemistry and
catalysis. In photochemistry, the quantum yield is defined as the number of molecules
transformed per adsorbed photon [1]. In a given heterogeneous catalytic reaction,
A ! B ‡ C ‡ ...
the yield 1B in reaction product B is defined as the ratio of the partial conversion tB of
reactant A to product B to the overall conversion t of reactant A [44].
1B ˆ tB/t ˆ [nB/nA ]/[(nA nA)/nA ] ˆ nB/nA nA
0
0
0
0
Presently, for 1, we have chosen the dynamic definition given by Braun et al. [1], with 1
defined as the ratio of the photocatalytic reaction rate to the incident efficient photon
flux fe:
1 ˆ r/fe ˆ ( dn/dt)/fe
with r in molecules/s and f in photons/s or with r in mol/s and f in Einstein/s (1
Einstein ˆ 1 mol of photons). Since a photocatalytic system is particularly heteroge-
neous, including i) gas and liquid fluid phases, ii) a solid catalyst, and iii) the
ꢀelectromagnetic phaseꢁ, i.e., the photon flux, the efficient photon flux has to be clearly
determined.
Since the reaction rate parallels the absorption spectrum of the photocatalyst,
indicating the transition of some electrons from the valence band into the conduction
band of the semiconductor catalyst, only efficient photons have to be considered, i.e.,
those whose energy hn is ꢀ EG (EG ˆ 3.2 eV for titania, i.e., l ꢁ 380 nm).
The determination of efficient photon flux is based on various chromatic
parameters: i) the spectral energetic distribution of the UV-lamp in the various light
bands, ii) the transmittance of the optical window of the photoreactor, or that of an
additional filter when a special wavelength range has to be selected to avoid a
concurrent photochemical reaction, and iii) the absorbance of titania.
The calculations have been made for the UV lamp (Philips Model HPK 125)
described in [1]. For each UV-ray of the medium-pressure mercury lamp, the relative
light power has been recorded from the data provided by Philips and multiplied by the
transmittance of the UV-filter. This new monochromatic relative light power produced
by transmittance through the filter is then normalized for a total polychromatic power
light of 1 W. Then, for each ray of the lamp, a monochromatic photon flux fi is
calculated by dividing the monochromatic normalized light power Pi by the

Helvetica Chimica Acta ± Vol. 84 (2001)
2747
corresponding quantum energy hni. The sum of these monochromatic fluxes gives the
overall photon flux (in photons/s) for an overall light power of 1 W. Subsequently, the
total light power received by the photocatalyst the radiant flux F has to be measured,
with a radiometer. Since the response of the radiometer depends on the wavelength
range used, we have calibrated the radiometer against a microcalorimeter. The total
power entering the photoreactor PT is given by: PT ˆ F  S where S is the section area
of the photoreactor window.
Such a photon flux should not be used to determine P. Only the efficient photon flux
should be taken into account and determined for each catalyst. Therefore, each
monochromatic photon flux is multiplied by the absorbance (in %) of the catalyst. The
sum of all these monochromatic fluxes gives the total efficient photon flux fe for a given
catalyst. For instance, when using a Philips HPK 125 UV-lamp and a Pyrex optical
window that transmits wavelengths l ꢁ 290 nm, one gets an efficient photon flux per
watt fW for titania of 8.36 Â 1017 efficient photons/s/W, and 1 is:
ꢆ dnA=dt†
1 ˆ r/fe ˆ
fW F S
Since one absorbed photon creates one electron-hole pair that can generate redox
processes when they totally dissociate and do not recombine, and, since no radical chain
reactions are involved, 1 will be necessarily smaller than unity.
Optimal Conditions for a Maximum PQY. For a given illumination system, the
efficient photon flux fe being constant, 1 can be only improved by considering the
reaction rate r, which varies as a function of the five main parameters described in
Fig. 2, i.e., the mass of catalyst, the wavelength, the partial pressure (in gas-phase
reactions) or the concentration (in liquid-phase reactions), the temperature (only in
ꢀextremeꢁ conditions for photocatalysis, i.e., either T< 08 or T> 808), and the radiant
flux (in W/m2). Therefore, to obtain maximum PQY, one is required to use: i) a mass of
catalyst at least equal to the optimum mass mopt at the beginning of the plateau of curve
rˆ f(m) (Fig. 2,a); ii) in a gas-phase reaction, a partial pressure corresponding to a full
coverage of the surface to give the maximum reaction rate in a Langmuir-Hinshelwood
mechanism, or, similarly in a liquid-phase reaction, a concentration high enough to
provide full surface coverage; iii) a temperature close to the optimum of the Arrhenius
plot and far from ꢀextremeꢁ values, i.e. below 108 where the reaction becomes limited
by the product desorption or above 858 where the reaction becomes limited by the
adsorption of the reactant; and iv) a radiant flux lower than the value at which r
becomes proportional to F1/2, indicating that the electron-hole recombination rate
prevails over the reaction rate. In the case of the mild oxidation of cyclohexane
into cyclohexanone, the limit was found equal to 25 mW/cm2 in the photoreactor in
Fig. 3.
The kinetic nature of PQY, defined as the ratio of the reaction rate to the efficient
photon flux [1] enables one to check whether a reaction is truly catalytic or not. In
Fig. 4, which shows H2 production from alcohols with titania and titania-deposited Pt
photocatalysts, the PQY of naked titania is zero after a period of one h, corresponding
to exhaustion of the stoichiometric threshold defined above. A mean quantum yield,
which would have been based on the number of molecules converted to the number of

2748
Helvetica Chimica Acta ± Vol. 84 (2001)
Fig. 4. Hydrogen production from aq. MeOH solution (50 :50) with naked titania and by Pt/TiO2. The dashed
line represents the mean rate for naked titania under ca. 5 h of UV-irradiation.
incident photons, would appear permanently positive as indicated by the dashed line in
Fig. 4, although the solid is no more active at tUV > 1 h.
The Usefulness of PQY. The knowledge of PQY of a given photocatalytic system is
particularly interesting for various purposes, enabling one to i) compare the catalytic
activities of various catalysts in a given reaction under identical conditions, ii) establish
the relative feasibility of various reactions when performed on the same photocatalyst,
and iii) calculate the energetic yield involved in a given reaction and the corresponding
cost when using artificial light provided by electrical lamps.
7. Conclusions. ± Photocatalytic reactions cover a large domain in both organic and
inorganic chemistry. Although the basic activation is identical for all systems

Helvetica Chimica Acta ± Vol. 84 (2001)
2749
considered, very different results can be obtained concerning the selectivity of the
reaction. The origin of the differences must be related to the different natures of
photoinduced active species, and it has been clearly shown that the crucial parameter is
the presence or absence of H2O. In absence of H2O, i.e., of regenerable OH surface
groups, the generation of dissociated oxygen species prevails and is responsible for mild
and selective oxidation processes, especially with hydrocarbons. By contrast, the
presence of H2O saturates the surface of titania with OH groups, which can be
.
neutralized by positive photoholes into OH radicals, which are known to be strongly
oxidative and cracking agents. To each process corresponds a photocatalytic quantum
yield, which has been explicitely defined according to the various reaction conditions.
This parameter appears to be fundamental to establish whether a photoassisted
reaction is really of a photocatalytic nature and to determine the facile or demanding
character of such a reaction.
REFERENCES
[1] A. M. Braun, M. T. Maurette, E. Oliveros, ꢀTechnologie Photochimiqueꢁ, presses Polytechniques
Romandes, Lausanne, 1986.
[2] ꢀPhotocatalysis and Environmentꢁ, Ed. M. Schiavello, Kluwer Academic Publisher, Dordrecht, 1988.
[3] ꢀPhotocatalysis, Fundamentals and Applicationsꢁ, Ed. N. Serpone, E. Pelizzetti, Wiley, New York, 1989.
[4] ꢀPhotocatalytic Purification and Treatment of Water and Airꢁ, Ed. D. F. Ollis, H. Al-Ekabi, Elsevier,
Amsterdam, 1993.
[5] O. Legrini, E. Oliveros, A. M. Braun, Chem. Rev. 1993, 93, 671.
[6] J.-M. Herrmann, Catal. Today 1999, 53, 115.
[7] D. W. Bahnemann, J. Cunningham, M. A. Fox, E. Pelizzetti, P. Pichat, N. Serpone, in ꢀAquatic Surface
Photochemistryꢁ, Ed. R. G. Zepp, G. R. Helz, D. G. Crosby, F. L. Lewis Publishers, Boca Raton, 1994,
p. 261.
[8] D. M. Blake, ꢀBibliography of Work on Photocatalytic Removal of Hazardous Compounds from Water and
Airꢁ, NREL/TP-430-22197, National Renewable Energy Laboratory, Golden, 1997 and 1999.
[9] J.-M. Herrmann, J. Disdier, P. Pichat, J. Chem. Soc., Faraday Trans. 1 1981, 77, 2815.
[10] P. Pichat, J.-M. Herrmann, H. Courbon, J. Disdier, M. N. Mozzanega, Can. J. Chem. Eng. 1982, 60,
27.
[11] a) S. Malato, Ph. D. Thesis, Almeria University, Spain, 1997; b) S. Malato, in ꢀSolar Photocatalytic
Degradation of Pentachlorophenol Dissolved in Waterꢁ, Coleccion Documentos Ciemat, Madrid, 1999.
[12] J.-M. Herrmann, Catal. Today 1995, 24, 157.
[13] H. P. Böhm, Adv. Catal. 1976, 16, 179.
[14] J.-M. Herrmann, H. Courbon, J. Disider, M. N. Mozzanega, P. Pichat, Stud. Surf. Sci. Catal. 1990, 59, 675.
[15] H. Mozzanega, J.-M. Herrmann, P. Pichat, J. Phys. Chem. 1979, 83, 2251.
[16] a) W. Mu, J.-M. Herrmann, P. Pichat, Catal. Lett. 1989, 3, 73; b) J.-M. Herrmann, W. Mu, P. Pichat, Stud.
Surf. Sci. Catal. 1990, 55, 405.
[17] P. Pichat, J. Disdier, J.-M. Herrmann, P. Vaudano, Nouv. J. Chim. 1986, 10, 545.
[18] J.-M. Herrmann, J. Disdier, M. N. Mozzanega, P. Pichat, J. Catal. 1979, 60, 369.
[19] H. Courbon, J.-M. Herrmann, unpublished results.
[20] A. Zafra, J. Garcia, A. Milis, X. Domenech, J. Mol. Catal. 1991, 70, 343.
[21] Y. Hori, A. Nakatsu, S. Susuki, Chem. Lett. 1985, 1429.
[22] J.-M. Herrmann, unpublished results.
[23] S. N. Frank, A. J. Bard, J. Phys. Chem. 1977, 81, 1484.
[24] J.-M. Herrmann, J. Disdier, P. Pichat, J. Catal. 1988, 113, 72.
[25] S. N. Frank, A. J. Bard, J. Am. Chem. Soc. 1977, 99, 303.
[26] H. Hidaka, T. Nakamura, A. Ishizaha, M. Tsuchiya, J. Zhao, J. Photochem. Photobiol., A 1992, 66, 367.
[27] C. H. Pollema, J. L. Hendrix, E. B. Milosavljevic, L. Solujic, J. H. Nelson, J. Photochem. Photobiol., A 1992,
66, 235.
[28] J.-M. Herrmann, J. Disdier, P. Pichat, J. Phys. Chem. 1986, 90, 6028.

2750
Helvetica Chimica Acta ± Vol. 84 (2001)
[29] E. Borgarello, N. Serpone, G. Eno, R. Harris, E. Pelizzetti, C. Minero, Inorg. Chem. 1986, 25, 4499.
[30] E. Borgarello, R. Harris, N. Serpone, Nouv. J. Chim. 1985, 9, 743.
[31] N. Serpone, E. Borgarello, M. Barbeni, E. Pelizzetti, P. Pichat, J.-M. Herrmann, M. A. Fox, J. Photochem.
1987, 36, 373.
[32] H. Courbon, J.-M. Herrmann, P. Pichat, J. Catal. 1985, 95, 539.
[33] P. Pichat, J.-M. Herrmann, J. Disdier, H. Courbon, M. N. Mozzanega, Nouv. J. Chim. 1981, 5, 27; P. Pichat,
J.-M. Herrmann, J. Disdier, H. Courbon, M. N. Mozzanega, Nouv. J. Chem. 1982, 6, 53.
[34] J. Casado, J.-M. Herrmann, P. Pichat, in ꢀTransport and Transformation of Pollutants in the Troposphereꢁ,
Ed. P. Borrell, P. M. Borrell, W. Seiler, SPB Academic Publishing, The Hague, 1991, p. 307.
Â
[35] F. Juillet, F. Lecomte, H. Mozzanega, S. J. Teichner, A. Thevenet, P. Vergnon, Faraday Discuss., Symp.
Chem. Soc. 1973, 7, 57.
Â
[36] S. Malato Rodríguez, J. Blanco Galvez, J.-M. Herrmann, Catal. Today 1999, 54, 191 (Introduction by the
Guest Editors to the Special Issue on Solar Photocatalysis).
[37] C. S. Turchi, D. F. Ollis, J. Catal. 1989, 119, 483.
[38] J.-M. Herrmann, M. Gravelle-Rumeau-Mailleau, P. C. Gravelle, J. Catal. 1987, 104, 136.
[39] C. S. Turchi, M. Mehos, J. Pacheco, in ꢀPhotocatalytic Purification and Treatment of Water and Airꢁ, Ed.
D. F. Ollis, H. Al-Ekabi, Elsevier, Amsterdam, 1993.
[40] R. Miller, R. Fox, in ꢀPhotocatalytic Purification and Treatment of Water and Airꢁ, Ed. D. F. Ollis, H. Al-
Ekabi, Elsevier, Amsterdam, 1993, p. 573.
[41] J.-M. Herrmann, J. Matos, J. Disdier, C. Guillard, J. Laine, S. Malato, J. Blanco, Catal. Today 1999, 54, 255.
Â
Â
[42] J. E. Germain, in ꢀLa Catalyse de Contactꢁ, les Techniques de lꢁIngenieur, Serie J 1181, 1979, p. 11.
Received June 2, 2001