Vous êtes sur la page 1sur 14

Pergamon

PII:

Solar Energy Vol. 66, No. 2, pp. 169182, 1999


1999 Elsevier Science Ltd
S 0 0 3 8 0 9 2 X ( 9 8 ) 0 0 1 2 0 0 All rights reserved. Printed in Great Britain
0038-092X / 99 / $ - see front matter

www.elsevier.com / locate / solener

SOLAR PHOTOCATALYTIC DEGRADATION OF WATER AND AIR


POLLUTANTS: CHALLENGES AND PERSPECTIVES
BLANCO, BENIGNO SANCHEZ,

MANUEL ROMERO, JULIAN


ALFONSO VIDAL, SIXTO
MALATO, ANA I. CARDONA and ELISA GARCIA
CIEMAT, Avda. Complutense 22, E-28040, Madrid, Spain
Revised version accepted 25 September 1998

AbstractSolar photocatalytic oxidation processes (PCO) for degradation of water and air pollutants have
recently received increasing attention. Some field-scale experiments have demonstrated the feasibility of using
a semiconductor (TiO 2 ) in solar collectors and concentrators to completely mineralize organic contaminants in
water and air. Although successful pre-industrial solar tests have been carried out, there are still discrepancies
and doubt concerning process fundamentals such as the roles of active components, appropriate modelling of
reaction kinetics or quantification of photoefficiency. Challenges to development are catalyst deactivation, slow
kinetics, low photoefficiency and unpredictable mechanisms. The development of specific non-concentrating
collectors for detoxification and the use of additives such as peroxydisulfate have made competitive use of
solar PCO possible. The challenges and perspectives of solar driven PCO as illustrated in the literature and our
own results in large solar field loops at the Plataforma Solar de Almeria and CIEMAT laboratories are
described. 1999 Elsevier Science Ltd. All rights reserved.

1. INTRODUCTION

Detoxification has recently become the most


successful solar photochemical application, with
some relevant field installations in operation.
Since Carey et al. (1976) published their results
on PCB decomposition by illuminating aqueous
TiO 2 suspensions more than 20 years ago, and
Ibusuki and Takeuchi (1986) reported photodegradation of toluene in air by irradiating TiO 2
at ambient temperature 10 years ago, references
and related patents on heterogeneous photocatalytic removal of toxic and hazardous compounds from water and air have multiplied by
thousands, on a wide variety of applications and
target compounds (Blake, 1997). Reports on solar
photocatalytic detoxification processes (Goswami,
1995) and related subjects, such as fundamental
chemistry (Bahnemann et al., 1994), photoefficiencies (Serpone, 1997), target contaminants
(Legrini et al., 1993) and catalysts (Hermann,
1995) are found in the literature.
But even with this dramatic increase in interest
and successful photocatalytic oxidation (PCO)
demonstration projects, the truth is that pre-industrial applications still involve unsolved fundamental questions concerning their chemistry. A
recent publication by Parent et al. (1996) points

Author to whom correspondence should be addressed.

out that low hydroxyl radical production efficiency and slow kinetics are important barriers
to marketing the technology, since they may limit
economic feasibility of the process. Unless the
cost of solar components and / or catalyst efficiencies change drastically, applications should be
restricted to certain specific processes like the
removal of low-ppm concentrations of such highly toxic compounds as pesticides in water (micropollutants) (Vidal et al., 1997), or the elimination of specific organometallic compounds like
phenylmercury or EDTA 1 Ag in which the rate
of destruction is improved by the organic 1
metal combination (Prairie et al., 1993). Although
development is difficult, application to industrial
wastewater is feasible, with the addition of a
sacrificial reagent like S 2 O 85 . Organic concentrations of several hundred ppm have been eliminated from water with this method (Malato et al.,
1996). Economics are far from definitive, but may
be considered at least pre-competitive (Blanco et
al., 1998). Gas-phase degradation is even more
unclear, since there are many conventional
catalysts available on the market that can work at
temperatures slightly over 1008C. Considering
that solar technologies can easily achieve such
temperatures, solar thermal catalysis might be a
more efficient degradation process than pure PCO
at ambient temperatures. Finding a niche for gasphase solar PCO for air purification is therefore
not a trivial matter. Significant photoeffect has

169

170

M. Romero et al.

been demonstrated in a few chlorinated compounds, however results with most typical VOCs
are not encouraging. Although, as in water,
residence times are high, some air purification
applications are under development (Watanabe et
al., 1993).
Some of the critical challenges to eventual
deployment of solar PCO are described here in
detail with reference to own results as related to
the literature.
2. MECHANISMS AND KINETICS: A
PERMANENT SUBJECT OF STUDY

Much of the literature analyzing mechanisms


contains hypotheses on the involvement of photoproduced holes (h 1 ) or assumptions on surfacetrapped hydroxyl radicals ( ? OH) (Hoffmann et al.,
1995). The complex primary events affecting
band-gap irradiation of TiO 2 particles have been
studied with detailed laser-flash photolysis measurements (Bahnemann et al., 1997). These
events, which yield ? OH radicals, are usually
summarized in three steps in which oxygen is
often the electron acceptor and available OH 2 and
H 2 O are electron donors.
1
TiO 2 1 hn e 2
cb 1 h vb

(1)

2
?
h1
vb 1 OH OH

(2)

2?
e2
cb 1 O 2 O 2

(3)

Chemical analysis shows the formation of hydroxylated intermediates which, in many cases,
coincide with those found in other reactions
produced by the attack of hydroxyl radicals in
homogeneous phase, as in photo-Fenton reactions
in water or the hydroxyl radical attack on organic
contaminants photocatalyzed by smog, as described in tropospheric studies (Killus and Whitten, 1982).
During our tests with the degradation of toluene
in air we have detected only benzaldehyde and

benzoic acid, with 2,5-furandione and 1,3-isobenzofurandione intermediates (Blanco et al.,


1996). Benzaldehyde from oxidation of toluene
by UV-irradiation of TiO 2 had already been
reported by Ibusuki and Takeuchi (1986) and it
has recently been suggested that benzoic acid
promotes deactivation of the catalyst with this
contaminant (Luo and Ollis, 1996). This mechanism, based on proposals by tropospheric air
pollution researchers, accounts for the high yields
of benzaldehyde formed during the photocatalytic
reaction. First benzyl radicals are produced by
abstraction of H atoms from the methyl group by
?
OH radicals. The final transformation of benzyl
radicals into benzaldehyde strongly suggests the
presence of benzyloxyl radicals (Akimoto et al.,
1978), as depicted in Fig. 1. This transformation
must originate with oxygen and an additional
redox reaction. Sephson et al. (1984); Killus and
Whitten (1982) report that in smog chamber tests,
substituted furans are produced by the intramolecular H-atom transfer reaction in the initial
radicals, leading to ring cleavage.
We have reported the same kind of hydroxylated intermediates for another aromatic compound (ethylbenzene) in water (Vidal et al.,
1994a). Ethybenzene photo-oxidation leads to
intermediates like 4-ethylphenol, acetophenone,
2-methylbenzyl alcohol, 2-ethylphenol and 3ethylphenol. Again H-atom abstraction by ? OH
radical attack on the ethyl group is presumed to
be the first step in the formation of hydroxylated
species and further ring cleavage (Fig. 2).
We were also able to demonstrate the intervention of ? OH radicals in more complex structures,
including heteroatoms. During the degradation of
compounds containing sulfur-like thiocarbamates
(EPTC), the formation of -OH, -CHO and a C=O
group containing thiocarbamate derivatives, was
observed in water. One of these formyl EPTC
derivatives detected, S-ethyl-N-formyl-N-propyl
thiocarbamate, (Vidal et al., 1994b), has been
reported in the literature on atmospheric OH
radical studies.

Fig. 1. Mechanism proposed for benzaldehyde and benzoic acid formation during gas-phase photocatalytic oxidation of toluene
with a monolithic catalyst based on sepiolite / TiO 2 . First benzyl radicals are produced by abstraction of H atoms from the methyl
group by ? OH radicals. The final transformation of benzyl radicals into benzaldehyde strongly suggests the presence of
benzyloxyl radicals.

Solar photocatalytic degradation of water and air pollutants: challenges and perspectives

171

Fig. 2. Hydroxylated cyclic intermediates detected during photocatalytic degradation of ethylbenzene in water with TiO 2 Degussa
P-25. H-atom abstraction by ? OH radical attack on the ethyl group leads to 4-ethylphenol, acetophenone, 2-methylbenzyl alcohol,
2-ethylphenol and 3-ethylphenol.

Even with the general evidence of the formation of hydroxylated by-products, the fundamentals of the roles of the oxidizing holes, hydroxyl
radicals or spectators such as chlorine, continue to
be controversial. Some authors claim that in their
experiments, the substrates were directly oxidized
by holes. Goldstein et al. (1994) suggest that at
high concentrations of phenol (.0.5 M) photooxidation is directly conducted by holes. This
dependence on substrate concentration is supported by others (Sun and Bolton, 1996). However, it is in gas-phase where there is greater
controversy on the role of the oxidants due to the
contradictory results obtained by different authors,
depending on the operating parameters, type of
contaminant and their concentrations. Involved in
this controversy are the typical VOC chlorine
radicals, as for example, the contradictory results
in the degradation of the relatively well-known
TCE (trichloroethylene). Nimlos et al. (1993)
argue that the high quantum efficiency of TCE
degradation in air is promoted by chlorine radicals. Tests with physisorbed chloromethane and
oxygen suggest the dioxygen anion as another key
active species (Lu et al., 1995). A study of the
mechanistical correlations in the photocatalytic

oxidation of air contaminants by chlorine radicals,


hydroxyl radicals and holes conducted by dHennezel and Ollis (1996) concludes that in the
presence of chlorine atoms, it is the chlorine
radical that appears to be active on the surface.
Their results reveal that in the absence of chlorine, hydroxyl radicals are not the reaction site, but
most probably, either a hole and oxygen vacancy
or a dioxygen anion.
However, other authors think that both the
mechanistical ? Cl and ? OH approaches are essentially correct and that the eventual pathway is
dependent on the characteristics of the catalyst,
e.g., internal surface (YamazakiNishida et al.,
1996). According to this second explanation,
when the inner catalyst surface is large enough,
the process continues with ? OH, since the porosity
of the catalyst is sufficient to trap chlorine radicals and avoid the formation of chlorinated byproducts ( ? Cl-initiated reactions seldom occur on
the catalyst surface), but when the inner surface is
small, the chlorine radicals can escape and react
with the organics in the bulk gas.
Our experience with TCE in air streams has
shown chloroform and pentachloroethane to be

the main by-products (Sanchez


et al., 1997). In

172

M. Romero et al.

some situations (very low residence times and


highly concentrated analytical samples), traces of
phosgene and dichloroacetyl chloride could also
be identified. The mechanism might be as described in Fig. 3, where chlorine radicals take the
lead. The by-products may be explained by the
proposal of Sanhueza and Heicklen (1975), in
which homogeneous-phase attack by chlorine
radicals follows a chain reaction. The mechanism
and by-products in Fig. 3 are similar to those
(1997).
presented by Hung and Marinas
All of the above gives an impression of the
complexity of the process and how, even compounds studied in depth such as TCE, are a source
of controversy and contradictory results.
Regarding kinetics, a rigorous treatment by
Turchi and Ollis (1990) and further commented
by Pelizzetti and Minero (1993), usually leads to
a saturation-type reaction rate expression similar
to the LangmuirHinshelwood model (LH).
Other authors suggest the use of Freundlich
et
isotherms at moderate concentrations (Bekbolet
al., 1996). The LH isotherm has been rather
useful for process modeling, and it is generally
agreed that, although rate constants and orders are
only apparent, they describe the rate of degradation and may be used for a particular case of
reactor optimization, but they have no physical
meaning, and may not be used to identify the
surface process (Serpone et al., 1993; Minero et
al., 1996b). The use of the unmodified LH
model, through common understanding, seems to

have become a useful tool for engineers and solar


designers, provided the same kind of collector and
photoreactor are used.
Normal laboratory procedure for obtaining rate
constants is to fit the organic substrate concentration plotted over time. Our experience with
large solar fields has demonstrated the need to use
the photon flux ( l ,387 nm) incident on the
receiver as the independent variable instead of
time (Malato et al., 1996). With this procedure,
conclusions are possible which aid in extrapolation for scaling up results and obviate the problem
stemming from the use of an intermittent radiation
source. Figs. 46 show examples of solar transients and their effect on the estimation of rate
constants. Fig. 4 shows the results of conversion
for two different tests (A and B). Changing UV
radiation during the day and the appearance of
clouds produce different illumination conditions.
Fig. 5 shows PCP (pentachlorophenol) degradation for both tests and the estimate of the initial
reaction rate (r o ). r o differs substantially in these
two tests, but surprisingly, the k o constant, assuming first order, is quite similar. This may be
explained by the differences between residence
time and real time and the effect of solar radiation
transients during the day. The same experiments
are depicted in Fig. 6, but this time versus the
accumulated photons per volume of photoreactor
(Einstein.l 21 ). The initial reaction rate is more
congruent when expressed in mg.Einstein 21 than
in the usual mg.l 21 .min 21 . In this way acceptable

Fig. 3. Scheme of gas-phase TCE degradation mechanism obtained by illuminating a monolithic catalyst based on sepiolite /
TiO 2 . Chlorine radicals are more likely adding to C1. Afterwards a second ? Cl is added to C 2 forming penthachloro ethane (n88).
With the intervention of O 2 it is possible to produce a peroxy radical (n83) which dimerizes and decomposes into two oxyradicals
(n84) Then the cleavage of CC bond leads to either dichloroacetyl chloride (n85) or phosgene (n86) and a CHCl 2 radical (the
base for chloroform production n87).

Solar photocatalytic degradation of water and air pollutants: challenges and perspectives

173

Fig. 4. Degradation of PCP (pentachlorophenol) in water at different hours the same day, using a large solar field at Plataforma
Solar de Almeria and 200 mg.l 21 of TiO 2 Degussa P-25.

reproducibility was obtained in solar field tests


using the same solar system.
3. THROUGHPUTS AND
PHOTOEFFICIENCIES

Even considering the important cost reductions


from simplifying the solar technology, and the
success in the complete mineralization of many
substrates, the physicalchemical limitations of
PCO are of great concern. The slow kinetics and
associated mediocre photoefficiencies require a
large amount of energy from the original source

(Parent et al., 1996). Any detoxification technology involving consumption of energy must use it
efficiently, whatever the source. Solar energy
should be used efficiently both for economic and
environmental reasons. The few studies on PCO
costs demonstrate its sensitivity to photoefficiency
(Turchi and Miller, 1994). However, by mimicking conventional treatment, authors have traditionally reported throughputs of m 3 of water treated
per m 2 of solar collector and time. Table 1
summarizes a comparison of our results with
micropollutants in water versus other references
in the literature. We were able to obtain through-

Fig. 5. Decomposition of PCP in the experiments of Fig. 4, versus residence time (t R ).

174

M. Romero et al.

Fig. 6. Same case as Fig. 5 but represented versus accumulated number of photons inside the reactor (EE ).

puts of 0.7 l.m 22 .min 21 , which might be comparable to other results. This shows the production
capacity of the collectors, but is restricted to water
with the same characteristics used in the experiment. The value depends strongly on initial and
final concentrations of the contaminant and provides no information on the energy efficiency of
the process. The same is true for gas-phase, where
gas results are usually given in spatial velocities,
as for instance, in conventional thermal-catalysis,
which refers to mol / g.s (mols of contaminant
treated per second and grams of catalyst) (Fu et
al., 1995), or area velocities in the case of
monoliths (volume of gas treated per unit of time
and surface area of catalyst channels) (Blanco et
al., 1996). Again spatial velocity merely describes
treatment capacity without any information on
energy consumption.
It has recently been suggested that a figure of
merit be used, like the Electrical Energy per
Order (EEO) defined as the kWh of electrical
energy required to reduce the concentration of a
pollutant by an order of magnitude in 1000 l of
contaminated water with UV lamps (Bolton et al.,
1996a) or in solar collectors, the Collector Area
per Order (CAO) defined as the collector area
(m 2 ) required to reduce the concentration of a
given pollutant by one order of magnitude in 1

hour in 1000 l of water when the incident solar


irradiance is 1000 W/ m 2 (Bolton et al., 1996b).
The CAO tries to rationalize the previous value of
throughput, including the contribution of solar
light and the initial and final concentration of
contaminant. But, in any case the CAO parameter
continues to be dependent on the specific contaminant and the initial concentration. The reaction rate constant changes depending on CO , as
illustrated in Fig. 7. According to this performance, the reaction rate constant is faster at lower
concentrations. Therefore, an order of degradation
with CO makes sense as a reference only under
similar conditions. In addition, CAO requires a
detailed description of the solar system, since
solar collector and reaction efficiencies are combined. It will surely have to be complemented
with information on catalyst performance with
catalytic and photonic efficiencies.
Other parameters, like Turnover Frequency,
have hardly been used in PCO tests, even though
they are very important for forthcoming studies
on catalyst deactivation, since the number of real
active sites and true illuminated surface area are
unknown. Some authors approximate the surface
site density as 515.10 14 sites.cm 22 for their
estimations (Peral and Ollis, 1997).
However, use of the quantum yield (f ) as an

Table 1. System throughputs of solar field experiences of water detoxification


Solar

Reference

Concentration

Conditions

Throughput

collector

compound

Cinitial (pbb)

Cfinal (pbb)

[TiO 2 ] g / l

(l / min m 2 )

EPTC
TCE
TCE

20500
10 4
200

0.1
10 3
5

0.5
1
1

0.7
1.7
0.4

CPC
CPC
PTC

Reference
Vidal et al., 1997
Pacheco et al., 1993
Mehos and Turchi, 1992

Solar photocatalytic degradation of water and air pollutants: challenges and perspectives

175

Fig. 7. Degradation of ethylbenzene in water by PCO with TiO 2 Degussa P-25 for three different initial concentrations.

expression of catalyst efficiency referred to absorbed photons is more common in the literature.
With the exception of a few cases, in most
experiments, the researchers do not know the flux
absorbed in their system, and moreover, the
reaction rate depends on substrate concentration.
Therefore, f is complicated to obtain as a parameter. For TiO 2 slurries, light scattering in the
particles is very difficult to evaluate. More common in the literature is to find apparent quantum
yields, or more properly, photonic efficiencies ( z ), defined as the number of reactant
molecules transformed or product molecules
formed divided by the number of photons incident
inside the front window of the photoreactor
(Serpone, 1997). Photonic efficiencies for organic
substrates converted to CO 2 are low in real
photoreactors. The maximum zTOC seems to be on
the order of 1%: phenol 1% (Terzian et al., 1990),
4-chlorophenol 0.36% (Lindsebigler, 1995) and
1.1% (Mills and Morris, 1993), 2,4-dichlorophenol 0.4% (Serra et al., 1994) and 2-ethoxyethanol
1.2% (Brezova et al., 1991) Our experience with
large solar plants and P-25 suspensions shows
examples like zTOC 50.063% for atrazine (Minero
et al., 1996a), 1% for PCP (Malato, 1997), 2.7%
for 88 mg.l 21 of initial TOC in pharmaceutical
plant wastewater containing amino acids, alcohols
and phenols (Malato et al., 1996) and lindane
degradation tests with 13 mg.l 21 initial TOC had
a zTOC 50.5%. An assumption of 1% average
photoefficiency (TOC) seems to be realistic for
most potential applications of solar PCO in water,
and points out the poor yields to be expected.
Catalyst modification or addition of electron
scavengers are undoubtedly required to increase
those percentages.
Some chlorinated high-quantum compounds

such as TCE have been reported to achieve


photoefficiencies above 100% in air due to the
chlorine chain mechanisms initiated (Jacoby et
al., 1995). Fig. 8 shows the results from a lab
experiment we have performed with TCE using
TiO 2 on a solgel membrane. Reactor photonic
efficiencies of up to 90% for TCE disappearance
were confirmed, even at low residence times.
In contrast, for typical VOCs in air pollution
such as BTEX, isooctane or hexane photoefficiencies of about 10% are reported (Turchi and
Miller, 1994). Others like trioxane, carbon tetrachloride, methylene chloride, chloroform, methyl
chloroform and vinyl chloride are below 1%
(Berman and Dong, 1993). Conversions below
4% at initial concentrations of 116 ppm have been
published (Jacoby et al., 1996) for aromatics like
benzene with complete conversion possible by
only decreasing Co to less than 5 ppm (indoor
air). We found similar conversions with toluene
for concentrations of about 400 ppm with several
TiO 2 / sepiolite-based monoliths (Avila et al.,
1998). By comparing photo1thermal and pure
thermal tests, an estimated increase in conversion
of 35% could be estimated. This increase, which
is negligible considering the much higher contributions obtained with small additions of thermal
energy, is produced by the addition of UV photons
in the reaction.
These efficiencies are one order of magnitude
higher than for water detoxification but are low
compared to thermal gas catalysis. Only for some
high-quantum chlorinated olefins like TCE, perchloroethylene and trichloropropene might it become competitive versus existing control technologies. As a rough estimate, the available UV
solar radiation for a standard ASTM (AM51.5) is
14310 25 Einstein.m 22 .s 21 . Assuming typical

176

M. Romero et al.

Fig. 8. Conversion of TCE in an air stream (C0 5400 ppm) and Reactor Photonic efficiency at different photonic fluxes by using a
membrane coated with TiO 2 by solgel techniques. Flow rate51.1 l / min. Residence time55310 24 s.

solar collector efficiency of 75% (from radiation


available to the photoreactor window), an average
of 10.5310 25 Einstein.m 22 .s 21 or 0.38
Einstein.m 22 .h 21 incident photon flux may be
estimated. The photonic efficiencies mentioned
before for PCO in water (about 1%) lead to a
rough estimate of 0.0038 mols of contaminant
degraded.m 22 .h 21 in water phase and ten times
more in gas phase.
4. CATALYST MODIFICATIONS AND
ADDITIVES

Only 5% of the solar spectrum is of use for the


TiO 2 band-gap. A realistic assumption of 75%
solar collector efficiency and 1% catalyst efficiency means that only 0.04% of the original
solar photons are efficiently used in the detoxification process. From the standpoint of the solar
collector technology, this is a rather inefficient
process even for a high-added-value application.
In contrast to other Advanced Oxidation Technologies, PCO has the advantage of being solarizable and the main characteristic of solar PCO is
that it is a mild technology. The TiO 2 catalyst is
photostable and cheap, and the process may be
run at ambient temperature and pressure. Furthermore, the oxidant, molecular oxygen, (O 2 ) is the
mildest of all. Therefore, in principle, we have a
mild catalyst working at mild conditions with
mild oxidants. However, as the number and
concentration of contaminants increase, the process becomes more complicated, and such challenging problems as catalyst deactivation, slow
kinetics, low photoefficiencies and unpredictable
mechanisms need to be solved (Parent et al.,
1996). It is clear that TiO 2 alone cannot undertake

practical applications of industrial and environmental concern without extra processes and this
may cause solar PCO to lose its charm as a mild
operation.
According to Eqs. (1)(3), the redox process is
based on electron and hole migration to the
semiconductor surface, and two further oxidation
and reduction steps. It is widely argued that the
reduction step may be rate-limiting since electrons
react much slower than holes (Sun and Bolton,
1996). Two basic lines of R&D have been
working on balancing both half-reaction rates, one
modifying catalyst structure and composition and
the other adding electron acceptors. Both try to
promote competition for electrons and avoid
recombination of e 2 / h 1 pairs. A third approach
has focused not only on increasing quantum yield,
but finding new catalysts able to work with bandgaps that coincide with the solar spectrum better.
Many attempts have been made within the first
line of research, improving specific surface, doping with metal ions and by oxide deposition.
Exhaustive testing conducted by Magrini et al.
(1995), leads to the conclusion that an increase in
rate does necessarily mean an increase in cost,
durability is uncertain, and performance is quite
variable depending on the organic substrate. It is
postulated that metal depositions may diminish
e 2 / h 1 recombination, and there is much work
reported in the literature on depositing Pt over
TiO 2 (Hermann et al., 1986). Surprisingly, a
recent note by Sclafani et al. (1997), puts into
evidence contradictory behavior in the presence of
metals in dehydrogenation of 2-propanol: for
rutile, Ag and Pt deposits were beneficial, but for
anatase they were detrimental. According to the
authors, electron transfer to the metal, which was

Solar photocatalytic degradation of water and air pollutants: challenges and perspectives

supposed to avoid recombinations and subsequently benefit photoefficiency, is detrimental


in this case, because fewer electrons are available
for the reduction in Eq. (3) and furthermore, some
holes might also be attracted to the surface of the
metal, generating recombinations there.
This is a good example of how complex
catalyst performance can become. Therefore, an
innovative catalyst composition has not been
employed in solar plants where P-25 suspensions
continue to be the more reliable solution.
Catalysts immobilized by different techniques
perform two and three times less efficiently than
suspensions. Confusion has also arisen from the
attempt to modify the naked titania catalyst, again
without conclusive results. In comparisons between P-25 and Hombikat UV-100, two titania
catalysts with different surface areas and percentages of anatase, leading to degradation of
dichloroacetic acid (Lindner et al., 1995; Goslich
et al., 1997), demonstrate that the degradation rate
is 23 times faster with Hombikat, but in contrast,
other tests with phenol as the substrate showed
four times more efficiency for P-25 (Tahiri et al.,
1996). This kind of behavior is symptomatic of
highly specific catalystsubstrate performance,
due to internal events within the semiconductor
that influence the kinetics and subsequent external
events are influenced by the kind of substrate.
Experiments in the Plataforma Solar de Almeria
(PSA) large solar plant produced very similar
results with both catalysts. Operation under real
solar conditions, using real multicomponent
wastewaters and electron oxidant additives seems
to dampen the effect of catalyst properties (Fig. 9)
(Malato et al., 1996).

177

The experience in large solar facility tests with


different contaminants has shown the use of
electron scavengers to be the most versatile and
useful way of improving reaction rates, providing
an opportunity to extend application of heterogeneous photocatalysis to hundreds-of-ppm concentrations. H 2 O 2 has been used as an additive in
lab-scale suspensions in which it has been able to
increment quantum yield of ? OH production from
4% to 22% (Sun and Bolton, 1996). H 2 O 2
competes with O 2 for conduction-band electrons:
?
2
e2
cb 1 H 2 O 2 OH 1 OH

(4)

But again, these results are not directly applicable


to any situation, and the contribution of H 2 O 2 is
reportedly more complex. Both H 2 O 2 concentration (Pichat et al., 1995) and the procedure by
which it was added (Enzweiler et al., 1994) may
make its role beneficial or detrimental. A recent
paper (Dillert et al., 1996) describes the detrimental role of H 2 O 2 in the degradation of TNT at
different pH, due to competition between the
compounds for conduction-band electrons. Although at solar field scale we have been able to
demonstrate a two-fold increase in the reaction
rate in PCP degradation tests with H 2 O 2 , the
oxidant had to be added every 15 minutes to
maintain a stable concentration since H 2 O 2 decomposed at a rate of 7mM.h 21 (Malato, 1997).
By contrast, peroxydisulfate has been shown to be
a more reliable and cheaper electron acceptor for
a wide variety of contaminants tested in the PSA
solar loops (Blanco et al., 1997; Malato et al.,
1996).
2
S 2 O 822 1 e CB
SO 42? 1 SO 422

(5)

Fig. 9. Influence of peroxydisulphate (S 2 O 22


8 ) addition on the degradation rate of residual waste water (distillation effluent from a
pharmaceutical company containing basically amino acids, alcohols and phenols). Two types of TiO 2 : Degussa P-25 and
Hombikat. Collector Helioman type (see Fig. 11). Catalyst slurry (200 mg.l 21 ).

178

M. Romero et al.
?
22
1
SO 2?
4 1 H 2 O OH 1 SO 4 1 H

(6)

Complex multicomponent water containing intermediate loads of pesticides and industrial / pharmaceutical effluents have been successfully
treated with TiO 2 1S 2 O 5
8 . Table 2 shows the
reaction rates for different contaminants using
naked TiO 2 and TiO 2 with peroxydisulfate.
In gas-phase PCO, much less has been done
with regard to additives. The use of TCE to
promote photocatalytic oxidation of air pollutants
by taking advantage of chlorine radical formation
has been proposed by dHennezel and Ollis
(1997). The oxidation of mixtures of TCE1
organics in some cases has shown photoenhancement, but only at high TCE / Organic ratios,
otherwise they act as chlorine scavengers and may
block the process. With acetone or methylene
chloride, however, TCE contribution was detrimental. An exhaustive work with organic binary
mixtures has revealed dissimilar performances
under photocatalytic conditions (Lichtin et al.,
1996). Anyhow, it seems that addition of chlorinated compounds such as TCE to a gas-phase
PCO process would be not practical in real-life
situations, but only in those cases where chlorine
is already present.
Deactivation of catalysts has also been insufficiently treated. In water phase, the extensive use
of batch reactors has not been very helpful in
discriminating possible deactivation, since that
effect might be masked by changes in adsorption

during the process. Recently some work in continuous gas-phase flow has verified significant
deactivation (Peral and Ollis, 1997; Sauer and
Ollis, 1996), after 110 equivalent monolayers of
contaminant have been converted. This work
reports irreversible deactivation in the presence of
such heteroatoms as N and Si, but not sulfur. Our
results in water under batch regimes with sulfurcontaining compounds such as thiocarbamates
(e.g., EPTC and molinate) did not show any
deactivation of the catalyst (Vidal et al., 1994b),
however a dithiocarbamate (metham sodium) did
interrupt TOC degradation. A careful analysis of
intermediates confirms that the action of metham
is similar to thiocarbamates (Fig. 10), but a
refractory intermediate (methyl isothiocyanate)
caused incomplete degradation. XPS analyses
reveal much larger sulfur deposits on the catalyst
surface in tests with metham and methyl isothiocyanate than EPTC and other thiocarbamates.
5. SOLAR TECHNOLOGY: THE USE OF
DIFFUSE RADIATION

Field tests conducted since 1990 have contributed to clarify what kind of solar technology is
best for detoxification. The parabolic troughs
initially used for water treatment and the dishes or
furnaces used for gas phase have since evolved
into lower-flux systems. One-sun systems for
water treatment are firmly based on two factors:
first, the high percentage of UV photons in the

Table 2. Solar photocatalytic degradation of contaminants in the CPC system at PSA comparing TiO 2 only and TiO 2 1S 2 O 822
(0.01 M), (250 l total treated volume, 200 mg / l TiO 2 , UV solar radiation about 30 W/ m 2 in all the cases)
Substrate

TOC 0

r 90,TOC
(mg.m 22 .l 21 )

r 0,TOC
(mg.l 21 )

t 95%,TOC
(min)

(mg.l 21 )

TiO 2
only

TiO 2 1
S 2 O 22
8

TiO 2
only

TiO 2 1
S 2 O 822

TiO 2
only

TiO 2 1
S 2 O 22
8

Pesticides
Imidacloprid a
Acrinathrin a
Oxamyl a
2,4 dichlorofenoxiacetic acid

132
40
90
13

0.25
0.25
0.08
0.16

0.75
0.77
0.41

6.9
6.9
2.7
4.4

20.6
21.1
12.4

617
698
( b)
70

277
103
183

Phenols
Phenol
4-chlorophenol
2,4 dichlorophenol

38
72
88

0.19
0.12
0.09

0.90
0.40

5.2
3.3
2.6

24.7
11.0

( b)
600
450

60
120

Other contaminants
Benzofurane
Dichloroacetic acid
Olive oil mills

8
120
250

0.14
0.51
( b)

0.71
0.83

3.8
14.0
( b)

19.5
22.8

60
224
( b)

160
300

TOC 0 : initial concentration of contaminant; r o,TOC : TOC initial degradation rate; r9 o,TOC : initial degradation rate of TOC per m 2
of solar field.
t 95%,TOC : residence time required to degrade 95% of the existing total initial organic.
a
Commercial product.
b
Very long testing time required. Values not determined.
() Tests not performed.

Solar photocatalytic degradation of water and air pollutants: challenges and perspectives

179

Fig. 10. Degradation of metham and the intermediate methyl isothiocyanate compared to molinate. Water-phase tests with TiO 2
Degussa P-25 (slurry).

diffuse component of solar radiation and, second,


the low order rate-dependence on light intensity.
Experimental measurements show that above a
certain UV-photon flux, reaction rate dependence
on intensity changes from one to half an order
(Bahnemann et al., 1994; Sun and Bolton, 1996).
This change does not seem to occur at a specific
radiation intensity, as different researchers obtain
different results. This is presumably affected
significantly by experimental conditions. Some
authors (Kormann et al., 1991) attribute the
transition of r5f (I 1.0 ) to r5f (I 0.5 ) to the excess
of photogenerated species (e 2 , h 1 and ? OH). At
higher radiation intensities, another transition
from r5f (I 0.5 ) to r5f (I 0 ) is produced. At this
moment, the photocatalytic reaction becomes
independent of the radiation received, to depend
only on the mass transfer within the reaction. So,
although the radiation increases, the rate is constant. This could be for several reasons, such as
the lack of electron scavengers (i.e. O 2 ), or
organic molecules in the proximity of the TiO 2
surface and / or excess of products occupying
active centres of the catalyst, etc. These phenomena really appear more frequently when working
with supported catalysts, and / or with slow agitation, which implies that catalyst surface in contact
with the liquid is small and there is less turbulence. This does not favor reactant contact with
the catalyst or diffusion of products from the
proximity of the catalyst to the liquid.
More important than dependence on intensity is
the fact that 50% of UV photons available at
AM51.5 (Iglobal 514310 25 einstein.m 22 .s 21 ) are
diffuse. This implies that non-concentrating technology could double the number of UV photons
incident in the photoreactor. Early engineeringscale tests with solar systems were based on

parabolic troughs with around 103 concentrations, but testing of a second generation of nonconcentrating thin films, flat plates and CPC
collectors started immediately thereafter (Gos
wami, 1995; Vidal et al., 1997; Bekbolet
et al.,
1996; Goslich et al., 1997; Pacheco et al., 1993).
Fig. 11 compares degradation endurance tests in
CPCs and parabolic troughs. CPC was five times
superior for dichloroacetic acid degradation under
analogous conditions (Richter et al., 1997) and
this ratio would be even higher if cloudy periods
were considered. A similar ratio has been reported
by Pacheco et al. (1993).
Of the one-sun collectors today, CPCs are the
most reliable and have the best performance for
catalyst suspensions, since it is easy to transfer
annular studies from the laboratory to the solar
field. Almost the same yields as CPCs may be
achieved with flat-plate designs with a better
potential for cost reduction, since cheaper materials can be used, but in contrast, preventing mass
transfer limitations in a flat plate is not a trivial
task. CPCs require a high-UV reflective surface
and the number of pyrex or teflon tubes becomes
costly, whereas flat plates may use cheap materials, but a high-UV transmittance glazing is necessary and the challenge of how to operate under
pressure must be solved. CPCs are a good option
for suspensions and flat plates are better adapted
for fixed catalysts.
Fig. 12 shows the schematic drawing of a water
treatment facility in batch mode operation. Engineering-scale tests at the PSA have shown
initial and final concentrations of 100 and 10 ppm
of TOC respectively to be a suitably practical
order of magnitude for a real solar detoxification
treatment plant. After pre-treatment (neutralization, filtration, etc.), the contaminated water is

180

M. Romero et al.

Fig. 11. Overall dichloroacetic acid degradation rate (mg of Total Organic Carbon mineralized per collector square meter and
hour) for the concentrating (Helioman, concentration ratio10) and nonconcentrating (Compound Parabolic Collector) systems.
The ratio of performance () and the mean value of global UV (? ? ?) during the initial reaction periods (zero order kinetic)
are represented on the right axis.

transferred to the closed-circuit solar detoxification loop. The drawing includes optional reuse of
water within the specific process which generates
the waste water and also proposes removal of the
last fraction of contaminants by carbon active
filter, which is practical and economical.

6. CONCLUSIONS

After almost ten years of field testing, it is now


possible to summarize the challenges remaining to
be solved for deployment of the solar detoxification technology:

Fig. 12. Conceptual diagram of a solar photocatalytic detoxification plant.

Solar photocatalytic degradation of water and air pollutants: challenges and perspectives

1. Destruction should lead to complete mineralization. Levels are not universal for all
compounds since it is demonstrated that mechanisms may vary. Some compounds also have
such slow kinetics that they may be classified
as not photoactive. Addition of radical promoters and modification of the surface of the
catalyst are some potential solutions.
2. Solar photons should be used efficiently. Quantum yields are still extremely low. Better
energy efficiency in the use of the primary
photons is needed, either through the use of
other catalysts working at longer wavelengths
(photo-Fenton, doping, structural modification,
etc.) or by including solar technologies with
maximum collection of the desired photons
(Compound parabolic collectors, flat plate,
double skin,...). The use of electron scavengers
like S 2 O 85 , H 2 O 2 or O 3 is recommended for
better use of solar radiation.
3. Throughputs should be reasonable. Spatial
velocities or throughputs and the consumption
of solar photons on the surface of the solar
collector should be acceptable. Reaction limitations, catalyst adsorptiondesorption steps
and mass-transfer restrictions in the photoreactor have to be improved to achieve larger
treatment capacities.
4. Technology should be low-cost and fully competitive. Information is still insufficient on this
point. Only some very specific applications
such as elimination of organometallic compounds, some pesticides at low-ppm concentrations or some effluents with addition of
S 2 O 85 , are almost economically comparable
with GAC (Granular Activated Carbon). No
applications with drastic cost reductions over
the existing technologies have been found to
date. However, even though this technology is
very recent, costs are not far from those of
others which have had many decades of development. On the other hand, the use of
activated carbon adsorption is feasible only so
long as legislation continues to allow storage
of this type of residue. In the European Union,
regulations with regard to generation of residues requiring storage are becoming stricter.
5. The process must be reliable (catalyst deactivation) : It is clear that water to be treated with
batch loads of contaminants should be recirculated in order to guarantee their complete
destruction. Continuous flow modes are necessary for gases. Therefore catalysts and collector components must be durable. To date,

181

yearly production has only been extrapolated


from limited-time tests.
REFERENCES
Akimoto H., Hoshino M., Inoue G., Okuda M. and Washida N.
(1978) Bull. Chem. Soc. Japan 51, 24962502.
Avila P., Bahamonde A., Blanco J., Sanchez B., Cardona A. I.
and Romero M. (1998) Appl. Cat. B: Env. 17, 7588.
Bahnemann D., Cunningham J., Fox M. A., Pelizzetti E.,
Pichat P. and Serpone N. (1994). In Aquat. Surf. Photochem., Zepp R. G., Helz G. R. and Crosby D. G. (Eds), pp.
261316. Lewis Publishers, Boca Raton, Florida.
Bahnemann D. W., Hilgendorff M. and Memming R. (1997) J.
Phys. Chem. B 101, 42654275.
M., Lindner M., Wechgrebe D. and Bahnemann D. W.
Bekbolet
(1996) Solar Energy 56, 455469.
Berman E. and Dong J. (1993). In: The third Int. Symp.
Chemical Oxidation: Technology for the Nineties, Eckenfelder W.W., Bowers A.R. and Roth J.A (Eds), pp. 183,
Technomic Publishers, Chicago.
Blake D.M. (1997). NREL / TP-430-22197, Golden, CO, National Renewable Energy Laboratory.
Blanco J., Avila P., Bahamonde A., Alvarez E., Sanchez B.
and Romero M. (1996) Catal. Today 29, 437442.
Blanco J., Malato S. and Richter C. (1997). In Proc. of the 8 th
Int. Symp. on Solar Thermal Concentrating Technologies,

Becker M. and Bohmer


M. (Eds), Vol. 3, pp. 14511469,

C.F. Muller
Verlag, Heidelberg.
Blanco J., Malato S., Milow B., Maldonado M.I., Fallmann H.,
Krutzler T., Bauer R. (1998). J. de Physique. In press.
(Special topic issue dedicated to the 9 th Int Symp. on Solar
Thermal Concentrating Technologies).
Bolton J. R., Bircher K. G., Tumas W. and Tolman C. A.
(1996) J. Adv. Oxid. Technol. 1, 13.
Bolton J.R., Ravel M., Cater S.R. and SafarzadehAmiri A.
(1996b). In Solar Engineering 1996 -ASME 1996, Davidson
J.H. and Chavez J. (Eds), pp. 5360, ASME.
Brezova V., Vodny S., Vesely M., Ceppan M. and Lapcik L.
(1991) J. Photochem. Photobiol. A: Chem 56, 125134.
Carey J. H., Lawrence J. and Tosine H. M. (1976) Bull. of
Environ. Contamination and Toxicol. 16, 697701.
dHennezel O. and Ollis D.F. (1996). In 11 th Int. Congress on
Catalysis 40 th Anniversary, Hightower J.W., Delgass W.N.,
Iglesia E. and Bell A.T. (Eds), Vol. 101, pp. 435442,
Elsevier.
dHennezel O. and Ollis D. F. (1997) J. Catal. 167, 118126.
Dillert R., Fornefett I., Siebers U. and Bahnemann D. (1996)
J. Photochem. Photobiol. A: Chem 94, 231236.
Enzweiler R.J., Mowery D.L., Wagg L.M. and Dong J.J.
(1994). In Solar Engineering 1994 ASME; Klett D.E.,
Hogan R.E and Tanaka T. (Eds), pp. 155162, ASME, N.Y.
Fu X., Zeltner W. A. and Anderson M. A. (1995) Appl. Catal.
B: Env. 6, 209224.
Goldstein S., Czapski G. and Rabani J. J. (1994) J. Phys.
Chem. 98, 6586.
Goslich R., Bahnemann D. and Schumacher H.W. (1997). In
Proc. of the 8 th Int. Symp. on Solar Thermal Concentrating

Technologies, Becker M. and Bohmer


M. (Eds), Vol. 3, pp.

13371353, C.F. Muller


Verlag, Heidelberg.

Goswami D.Y. (1995). In Advances in Solar Energy, Boer


K.W. (Ed.), Chap. 3., American Solar Energy Society,
Boulder, Colorado.
Hermann J. M., Disdier J. and Pichat P. (1986) J. Phys. Chem.
90, 6028.
Hermann J. M. (1995) Catal. Today 24(12), 157164.
Hoffmann M. R., Martin S. C., Choi W. and Bahnemann D. W.
(1995) Chem. Rev. 95, 6996.
B. J. (1997) Environ. Sci. Technol.
Hung C. H. and Marinas
31, 562568.

182

M. Romero et al.

Ibusuki T. and Takeuchi K. (1986) Atmospher. Environ. 20,


1711.
Jacoby W. A., Blake D. M., Noble R. D. and Koval C. A.
(1995) J. Catal. 157, 8796.
Jacoby W. A., Blake D. M., Fennell J. A., Boulter J. E., Vargo
L. M., George M. C. and Dolberg S. K. (1996) J. Air and
Waste Manage. Assoc. 46, 891898.
Killus J. P. and Whitten G. Z. (1982) Atmospher. Environ. 16,
8.
Kormann C., Bahnemann D. W. and Hoffmann M. R. (1991)
Environ. Sci. Technol. 25, 494500.
Legrini O., Oliveros E. and Braun A. M. (1993) Chem. Rev.
93, 671698.
Lichtin N. N., Avudaithai M., Berman E. and Grayfer A.
(1996) Solar Energy 56, 377385.
Lindner M., Bahnemann D.W., Hirthe B. and Griebler W.D.
(1995). In Solar Engineering 1995 ASME, Stine W.B.,
Tanaka T. and Claridge D.E. (Eds), pp. 399408, ASME,
N.Y.
Lindsebigler A. L. (1995) Chem. Rev. 95, 735758.
Lu G., Lindsebigler A. and Yates J. J. T. (1995) J. Phys.
Chem. 99, 7626.
Luo Y. and Ollis D. (1996) J. Catal. 163, 111.
Magrini K.A., Goggin R.M., Watt A.S., Taylor A.M. and
Baker A.L. (1995). In Solar Engineering 1995 ASME,
Stine W.B., Tanaka T. and Claridge D.E. (Eds), pp. 415
420, ASME, N.Y.
Malato S., Richter C., Blanco J. and Vincent M. (1996) Solar
Energy 56, 401410.
Malato S. (1997). Ph.D. Dissertation, Universidad de Almeria
(Spain).
Mehos M. and Turchi C. (1992). In Proceedings of the 6 th Int.
Symp. on Solar Thermal Concentrating Tech., Mojacar
(Spain), Vol. II, pp. 11071122, CIEMAT, Madrid.
Mills A. and Morris S. (1993) J. Photochem. Photobiol. A:
Chem 71, 7583.
Minero C., Pelizzetti E., Malato S. and Blanco J. (1996) Solar
Energy 56, 411419.
Minero C., Pelizzetti E., Malato S. and Blanco J. (1996) Solar
Energy 56, 421428.
Nimlos M. R., Jacoby W. A., Blake D. M. and Milne T. A.
(1993) Environ. Sci. Technol. 27, 732740.
Pacheco K., Watt A.S. and Turchi C.S. (1993). In Solar
Engineering 1993 ASME, Kirkpatrick A. and Worek W.
(Eds), pp. 4349, ASME, NY.
Parent Y., Blake D., Magrini-Blair K., Lyons C., Turchi C.,
Watt A., Wolfrum E. and Prairie M. (1996) Solar Energy
56, 429438.
Pelizzetti E. and Minero C. (1993) Electrochimica Acta 38,
4755.
Peral J. and Ollis D. F. (1997) J. Molecular Cat. A: Chemical
115, 347354.
Pichat P., Guillard C., Amalric L., Renard A. C. and Plaidy O.
(1995) Solar Energy Mater. Sol. Cells 38, 391399.

Prairie M. R., Evans L. R., Stange B. M. and Martinez S. L.


(1993) Environ. Sci. Technol. 27, 17761782.
Richter C., Malato S. and Blanco J. (1997). In Proc. of the 8 th
Int. Symp. on Solar Thermal Concentrating Technologies,

Becker M. and Bohmer


M. (Eds), Vol. 3, pp. 15211526,

C.F. Muller
Verlag, Heidelberg.

Sanchez
B., Romero M., Cardona A.I., Fabrellas B., Avila P.
and Bahamonde A. (1997). In Proc. of the 8 th Int. Symp. on
Solar Thermal Concentrating Technologies, Becker M. and

Bohmer
M. (Eds), Vol. 3, pp. 14891507, C.F. Muller
Verlag, Heidelberg.
Sanhueza E. and Heicklen J. (1975) J. Phys. Chem. 79, 7.
Sauer M. L. and Ollis D. F. (1996) J. Catal. 163, 215217.
Sclafani A., Mozzanega M. N. and Hermann J. M. (1997) J.
Catal. 168, 117120.
Sephson P. B., Edney E. O. and Corse E. W. (1984) J. Phys.
Chem. 88, 41224126.
Serpone N., Pelizzetti E. and Hidaka H. (1993). In 9 th Int.
Conf. on Photochem. Transformation and Storage of Solar
Energy ( IPS-9), Tian Z.W. and Cao Y. (Eds), pp. 3373, Int.
Academic Publishers, Beijing, China.
Serpone N. (1997) J. Photochem. Photobiol. A: Chem. 104,
112.
J. and Domenech X. (1994) J.
Serra F., Trillas M., Garca
Environ. Sci. Health A 29, 14091421.
Sun L. and Bolton J. R. (1996) J. Phys. Chem. 100, 4127
4134.
Tahiri H., Serpone N. and Le van Mao R. (1996) J. Photochem. Photobiol. A: Chem 93, 199203.
Terzian R., Serpone N., Minero C., Pelizzetti E. and Hidaka H.
(1990) J. Photochem. Photobiol. A: Chem. 55, 243249.
Turchi C. S. and Ollis D. F. (1990) J. Catal. 122, 178192.
Turchi C.S. and Miller R.A. (1994). NREL / TP-471-7014;
November 1994, National Renewable Energy Laboratory,
Golden, CO.
Vidal A., Herrero J., Romero M., Sanchez B. and Sanchez M.
(1994) J. Photochem. Photobiol. A: Chem. 79, 213219.
Vidal A., Romero M., Sanchez B. and Mogyorodi F. (1994b).
In Solar Engineering 1994 ASME, Klett D.E., Hogan R.E
and Tanaka T. (Eds), pp. 117129, ASME, N.Y.
Vidal A., Ajona J.I., Blanco J., Romero M., Muguruza I., Diaz
A., Zallo M. and Gonzalez J. (1997). In Proc. of the 8 th Int.
Symp. on Solar Thermal Concentrating Technologies,

Becker M. and Bohmer


M. (Eds), Vol. 3, pp. 13651379,

C.F. Muller
Verlag, Heidelberg.
Watanabe T., Kitamura A., Kojima E., Nakayama C., Haskimoto K. and Fujishima A. (1993). In Photocatalytic
Purification and Treatment of Water and Air, Ollis, D.F.
and AlEkabi (Eds), pp. 747751, Elsevier.
YamazakiNishida S., Fu X., Anderson M. A. and Hori K.
(1996) J. Photochem. Photobiol. A: Chem. 97, 175179.

Vous aimerez peut-être aussi