Photoreduction of 1,1,2-trichlorotrifluoroethane initiated by TiO2 particles

Article abstract of DOI:10.1021/jp963070u

Reductive dehalogenation of 1,1,2-trichlorotrifluoroethane (CFC 113) takes place upon illumination of air-free suspensions of TiO₂ particles containing formate ions. Chloride ions and 1,2-dichlorotrifluoroethane (HCFC 123a) are the main reaction products. Reaction rates increase with increasing concentration of CFC even beyond the solubility limit of the Freon in water. Evidence is presented that a chain transformation of the Freon takes place in formate solutions. The results are explained assuming that reduction of the Freon proceeds via a radical chain mechanism, which involves participation of CFC and formate species located on as well as next to the oxide surface.

Full text of DOI:10.1021/jp963070u

J. Phys. Chem. B 1997, 101, 3769-3775
3769
Photoreduction of 1,1,2-Trichlorotrifluoroethane Initiated by TiO2 Particles
S. Weaver and G. Mills*
Department of Chemistry, Auburn UniVersity, Auburn, Alabama 36849
ReceiVed: October 7, 1996; In Final Form: February 27, 1997X
Reductive dehalogenation of 1,1,2-trichlorotrifluoroethane (CFC 113) takes place upon illumination of air-
free suspensions of TiO2 particles containing formate ions. Chloride ions and 1,2-dichlorotrifluoroethane
(HCFC 123a) are the main reaction products. Reaction rates increase with increasing concentration of CFC
even beyond the solubility limit of the Freon in water. Evidence is presented that a chain transformation of
the Freon takes place in formate solutions. The results are explained assuming that reduction of the Freon
proceeds via a radical chain mechanism, which involves participation of CFC and formate species located on
as well as next to the oxide surface.
Introduction
Experimental Section
Chemicals were reagent grade (Aldrich, Fisher or PCR) and
Reactions of organic compounds initiated by illumination of
semiconductor particles in solution continue to receive consider-
able interest.1,2 Oxidation processes have been frequently
studied, since they may offer a simple method for the elimination
of undesirable chemicals.3-8 Reduction reactions have received
less attention; most of these studies have examined the dynamics
of interfacial electron transfer.1,2 Photoinitiated transformations
usually involve radical reactions. Radicals form when substrates
present next to or on the particle surface are attacked by holes,
electrons, or other reactive species formed upon exposure of
the particles to light.9 Except for chain processes,5,10-12 the
photoreactions proceed with low quantum efficiencies.1-4
Earlier studies have shown that efficient reductive dehalo-
genations of organic halides take place in deoxygenated alcohols
through free radical chain reactions.13-15 Chain reductions of
fluoro halides in homogeneous solution have been observed
occasionally.14,16-18 Only a few heterogeneous photoreductive
dehalogenations have been investigated. Reductions of CCl4
and CHBr3 occurred in suspensions of TiO2-containing electron
donors;19,20 halothane (2-bromo-2-chloro-1,1,1-trifluoroethane)
was photoreduced in aqueous solutions of alcohols containing
platinized TiO2 particles.21 The efficiencies of these transfor-
mations are limited because of fast competing processes such
as charge carrier recombination reactions. However, higher
efficiencies seem achievable in cases where illumination of the
semiconductors may initiate chain dehalogenation reactions.
The photocatalyzed reduction of fluoro halides was used to
test this concept. Light-initiated reductions are interesting as
possible ways to modify chlorofluorocarbons (CFCs). These
chemicals interfere with the natural stratospheric cycle of
ozone22 and are currently being replaced by their hydrogen-
substituted analogs (HCFCs).23 The photoreduction of 1,1,2-
trichlorotrifluoroethane (CFC 113) was examined in this study.
Air-free aqueous suspensions of TiO2 particles were utilized to
initiate the reactions, and efforts were made to improve the
efficiency of the reduction process. Formate ions were selected
as electron donors because the Freon was efficiently reduced
in acetonitrile solutions containing HCO2- ions.16 In the present
report kinetic evidence is presented that the photoreduction of
CFC 113 proceeds via a radical chain mechanism involving the
electron donor.
were not further purified. The water used in all preparations
was obtained from a Milli-Q/RO system (Millipore). TiO2
particles from Degussa (P-25, 30 nm average diameter) were
used in this study. Aqueous suspensions of the semiconducting
material containing sodium formate were prepared in the
photochemical reactors and then degassed with N2 for at least
20 min. Liquid CFC 113 was deaerated using the freeze-
pump-thaw method and subsequently injected into the degassed
TiO2 suspensions. Prior to illumination the mixtures were
equilibrated by stirring in the dark for several minutes.
Irradiations were performed at room temperature using
borosilicate glass vessels (internal volume 168 mL) as photo-
chemical reactors. The vessels contained 130 mL of the
suspensions, which were stirred continuously during illumina-
tion. A PTI 1010 S system with a 150 W Xe arc lamp was
utilized as a light source. Infrared radiation was filtered with
a 10 cm water filter and an IR-absorbing filter. For irradiations
with multiple wavelengths, the light was filtered with a Kopp
GS-7-60 filter (320 e λ e 385 nm). The effect of light intensity
(I0) on the reaction rates was investigated with an Oriel 340
nm narrow-band filter for I0 < 3 × 10-7 M (hν) s-1 and with
the GS-7-60 filter for higher intensities. Neutral density filters
were employed to adjust I0; actinometry was performed with
Aberchrome 540.24 Efficiencies of the photoreactions are given
in terms of photonic efficiency (ú), which corresponds to the
rate of product formation divided by the rate of incident
photons.4b The reported values of ú were not corrected for
photon losses due to scattering of light in the TiO2 suspensions.
Optical determinations were carried out using a Hitachi U-2000
spectrophotometer.
Concentrations of Cl- and F- ions were monitored in situ
using a Radiometer K601 mercurous sulfate reference electrode
and ion-selective electrodes (Orion). Calibration curves were
obtained before each kinetic run with suspensions identical to
those used during the illuminations. Using this procedure
ensured that halide ions adsorbed on the surface of the particles
were accounted for. Each experiment was performed at least
twice, and in most cases, the results deviated by less than 10%.
Measurements of pH were accomplished with a Radiometer GK
273920S electrode containing a saturated K2SO4 solution.
Radiometer pH meters models PHM 84 and PHM 95 were used
in the potentiometric determinations. The data were acquired
after stopping the irradiation to avoid artifacts generated by
* Corresponding author. e-mail address: millsge@mail.auburn.edu.
X Abstract published in AdVance ACS Abstracts, April 1, 1997.
S1089-5647(96)03070-2 CCC: $14.00 © 1997 American Chemical Society

3770 J. Phys. Chem. B, Vol. 101, No. 19, 1997
Weaver and Mills
Figure 1. Plot of [Cl-] vs time in degassed suspensions with (a) 0.5,
(b) 0.2, and (c) 0.1 g L-1 TiO2; I0 ) 1 × 10-6 M (hν) s-1, pH ) 7.1,
0.077 M HCO2-, and 2 mL of CFC 113.
Figure 2. Variation of k0 as a function of Freon concentration in
systems at pH ) 7.5 with 0.5 g L-1 TiO2 and 0.5 M HCO2-; I0 ) 9.4
× 10-7 M (hν) s-1. The inset corresponds to experiments where the
volume of Freon added exceeded the solubility limit.
direct illumination of the electrodes. Immersion of the elec-
trodes into the suspensions was achieved via glass joints located
on the top of the reactors. The electrodes were tightly fitted
through perforated septa, and degassing of the suspensions
proceeded after sealing the joints with these septa.
Identification of volatile products was carried out by analysis
of CH2Cl2 extracts of the aqueous phase and by headspace
analysis; both sampling methods gave the same results. GC-
MS measurements were conducted on a Finnigan 9500 GC
equipped with either a Porapaq-Q or a 5% Fluorocol column
(Supelco), in conjunction with a Finnigan 3300 MS instrument.
Results
Illumination of degassed suspensions of TiO2 containing
formate ions and CFC 113 generated Cl- ions and small
amounts of F- ions. Figure 1 shows the variation in [Cl-]
during irradiation (320 e λ e 385 nm) of suspensions saturated
Figure 3. Dependence of the zero-order rate constant for Cl- ion
-
with the CFC that contained 7.7 × 10-2 M HCO2 ions and
formation on [HCO2-] in suspensions with 0.5 g L-1 TiO2 and 2 mL
several concentrations of semiconductor particles. A linear
increase in the concentration of Cl- ions with increasing time
was noticed after an initial period of 5-8 min. Formation of
chloride ions was slow during this initial period and obeyed no
simple rate law; the initial formation rate was less than 10% of
the rate observed at longer times. Very low concentrations of
F- ions were measured at short times, with a rate smaller by a
factor of 10 than the formation rate of Cl- ions. The length of
the initial slow process remained unaffected by increases in the
degassing time. Similar two-step processes occurred in all
experiments with the Freon; the kinetic results presented in this
report correspond to data collected at times longer than the initial
period of slow reaction. During the second step, formation of
Cl- ions always followed zero-order kinetics, and the slopes of
the straight lines in the plots of [Cl-] vs time were used to obtain
zero-order rate constants (k0) for the light-induced reduction
process. While F- ions were also detected at longer reaction
times, their concentrations were less than 1% of the [Cl-]. In
addition, formation of H+ ions occurred during the photoreac-
tions, but only small decreases in the pH of the suspensions
were measured due to the buffering effect exerted by the formate
ions. Since it was not possible to fit the fluoride and H+ data
to any simple kinetic model, these results were not further
analyzed. Irradiation of suspensions containing acetate instead
of CFC 113; I0 ) 9.1 × 10-7 M (hν) s-1
.
order rate constant increased slowly with increasing [Freon] at
concentrations lower than 7 × 10-4 M , followed by a steep
rise at the higher concentration. Results from experiments where
larger amounts of CFC 113 were injected into the suspensions
are presented in the inset. The limiting solubility of this Freon
in water is 9 × 10-4 M (22 µL of Freon in 130 mL solution).25
Formation of two phases occurred when the amounts of CFC
added exceeded the solubility limit, and a large number of small
Freon droplets were generated upon stirring of the suspensions.
In spite of this, k0 increased rapidly with increasing volume of
CFC even beyond the limit of solubility. Increases in the
reaction rate leveled off at 2 mL of Freon added, and k0 was
constant thereafter. The TiO2 particles remained dispersed in
the aqueous phase for systems where two phases were formed,
and negligible reduction of CFC 113 occurred in the absence
of the semiconductor. Consequently, formation of chloride ions
at the [Freon] of the inset was due to reactions taking place in
the aqueous phase containing the oxide particles.
Figure 3 illustrates the variation of k0 as a function of formate
ion concentration. The zero-order rate constant for the formation
of Cl- ions increased with increasing concentration of electron
donor until [HCO2-] ) 0.5 M and remained constant thereafter.
In these experiments the initial pH of the suspensions was
determined by each particular concentration of formate ions (the
“natural” pH values) and changed from 5.4 to 7.7 as [formate]
increased. A few additional experiments were performed at an
initial pH ) 7.3; significant increases in k0 occurred only at
[HCO2-] < 2 × 10-2 M. Attempts to fit the data of Figures 2
-
of HCO2 yielded Cl- ions as well, but the reactions were 30
-
times slower than those performed with HCO2 ions. Halide
ion generation via postirradiation reactions was not detected in
any of the experiments.
Shown in Figure 2 is the variation of k0 with molar
concentration of Freon at [CFC 113] e 9 × 10-4 M. The zero-

Photoreduction of 1,1,2-Trichlorotrifluoroethane
J. Phys. Chem. B, Vol. 101, No. 19, 1997 3771
Figure 5. Dependence of ú(Cl-) on the amount of dispersed TiO2 at
Figure 4. Changes in k0 as a function of pH in suspensions containing
0.1 mL CFC 113 and 0.5 g L-1 TiO2; I0 ) 7.8 × 10-7 M (hν) s-1. (O)
[HCO2-] + [HCO2H] ) 0.1 M; (2) [HCO2-] ) 0.1 M, pH adjusted
with buffers; (0) 0.025 M HCO2-, pH adjusted with HClO4 or NaOH.
-
pH ) 7.3 with 0.1 M HCO2 and 2 mL of CFC 113: curve a, I0 ) 6
× 10-8 M (hν) s-1; curve b, I0 ) 8.5 × 10-7 M (hν) s-1
.
and 3 with single adsorption site Langmuir-type kinetic models
were unsuccessful.4a Also, no linear correlations were obtained
from log-log plots of these results.
Figure 4 illustrates the variations of the zero-order rate
constant as a function of pH in suspensions with 0.5 g L-1 TiO2
and 0.1 mL of Freon. At pH > 9 high concentrations of formate
caused interferences in the potentiometric determinations of
halide ions. Experiments were therefore performed at pH e 9
and with [HCO2-] e 0.1 M, conditions that ensured a proper
operation of the ion-selective electrodes. A “natural” pH of
7.3 was typical for suspensions containing 0.1 M formate. In
the top curve (O), initial pH values lower than 7.3 were
established by mixing appropriate amounts of NaHCO2 with
formic acid so that [HCO2-] + [HCO2H] ) 0.1 M, whereas
pH values above 7.3 were obtained by addition of NaOH to
suspensions with 0.1 M NaHCO2. The rate of Cl- ion formation
increased sharply with increasing pH, reaching a maximum at
pH ) 6. Further increases in pH induced a decrease in k0 until
pH ) 8.1, after which k0 remained constant.
Figure 6. Light intensity effect on ú(Cl-) in suspensions at pH ) 7.3
that contained 0.1 M HCO2-, 0.5 g L-1 TiO2, and 2 mL of CFC 113.
The inset is a log-log plot of these results.
of suspended oxide and also on the light intensity. Therefore,
comparisons between results of different experiments were
simplified by using photonic efficiencies of Cl- ion generation
(ú(Cl-)) instead of rate constants. The data are presented in
Figure 5; curve a corresponds to experiments where the 340
nm narrow-band filter was utilized, with a light intensity of 6
× 10-8 M (hν) s-1. Low efficiencies were determined at 0.01
and 0.1 g L-1 TiO2 with ú(Cl-) ) 0.01 and 0.04, respectively.
The efficiencies increased substantially with increasing amounts
of dispersed oxide, reached a maximum of ú(Cl-) ) 2.9 in the
presence of 0.5 g L-1 TiO2, and dropped somewhat thereafter.
Less pronounced variations in ú(Cl-) with increasing amounts
of dispersed oxide took place at the higher light intensity of
curve b (I0 ) 8.5 × 10-7 M (hν) s-1). The efficiencies varied
between 0.1 e ú(Cl-) e 0.3; they increased slowly with
increasing amounts of TiO2 until a maximum was reached at
0.5 g/L. As in the case of curve a, ú(Cl-) decreased for
suspensions containing higher amounts of oxide particles.
Illustrated in Figure 6 is the influence of light intensity on
the yield of photoreduction. The highest photonic efficiency
of ú(Cl-) ) 2.9 occurred at I0 ) 6 × 10-8 M (hν) s-1 and
decreased continuously as the light intensity increased to a value
of ú(Cl-) ) 0.39 at I0 ) 9.1 × 10-7 M (hν) s-1. Shown in the
inset is a plot of log ú(Cl-) vs log I0, which yielded a straight
line with a slope of -0.64. The values of ú(Cl-) remained
unchanged when sealed vessels without electrodes were em-
ployed in the irradiations. This means that leakage of traces of
air into the suspensions was not significant in the experiments
The curve labeled with closed triangles (2) corresponds to
solutions in which the pH was fixed by buffers such as phthalate
-
(5 × 10-2 M + acid or base for 4 e pH e 5), H2PO4 (5 ×
10-2 M + base for 6 e pH e 8.1), and borax (1.3 × 10-2
M
+ acid for pH ) 9). Smaller rate constants were obtained in
these systems, and k0 increased slowly with increasing pH until
a maximum was reached at pH ) 7.3 (the pH in the absence of
buffers). Above this pH k0 dropped first and then remained
constant. With exception of the data measured at pH ) 7.3,
similar results were obtained when the pH of suspensions
containing 2.5 × 10-2 M formate was adjusted with HNO3 and
NaOH or with buffers. In the curve labeled with open squares
(0) the suspensions contained 2.5 × 10-2 M NaHCO2, and the
pH was adjusted with HClO4 and NaOH. The rate increased
with pH reaching a maximum at the “natural” pH (6.8); it
decreased until pH ) 8.1 and then remained constant. However,
below pH ) 6.8 the Freon photoreduction was about 2.5 times
-
faster than in systems with buffers. Evidence that ClO4 ions
affected the reduction process was obtained by irradiating
suspensions at pH ) 7.3 containing 0.5 g L-1 TiO2, 0.1 M
NaHCO2, and 0.1 mL of CFC 113. A rate constant of k0 ) 1.8
× 10-7 M s-1 was determined in the presence of 5 × 10-2
M
NaClO4, which is 50% of the value in the absence of perchlorate.
The CFC photoreduction was also followed as a function of
the mass of TiO2 dispersed in the formate solutions. It turned
out that the rates depended in an unusual way on the amount

3772 J. Phys. Chem. B, Vol. 101, No. 19, 1997
Weaver and Mills
performed with electrodes. In another experiment, a degassed
suspension consisting of 0.5 g L-1 TiO2, 0.1 M NaHCO2, and
2 mL of CFC 113 was first exposed to a high light intensity (I0
) 9 × 10-7 M (hν) s-1) for 20 min. This procedure ensured
the photoreduction of traces of O2 that may remain adsorbed
on the surface of the particles even after degassing.26 The
preirradiated sample was then illuminated using a lower intensity
of I0 ) 6 × 10-8 M (hν) s-1. A substantial improvement was
achieved, because ú(Cl-) was 3.7 in the preilluminated sample
whereas ú(Cl-) ) 2.9 in a similar experiment but without
preirradiation (Figure 6).
Volatile products of the photoreduction were identified by
GC/MS analysis of suspensions containing 0.5 g/L TiO2, 0.1
M NaHCO2, and 2 mL of CFC 113, illuminated with I0 ) 1.2
× 10-6 M (hν) s-1. These determinations showed that 1,2-
dichlorotrifluoroethane (HCFC 123a) was generated in this
reaction. Traces of a byproduct, chlorotrifluoroethylene (CFC
1113), were also found, but the dimer (CClF-CF2Cl)2 was not
detected. As expected, no CO2 was detected during headspace
analysis because it is present mainly as bicarbonate in these
suspensions (pH ) 7.3), which chemisorbs on the oxide surface.3
Attempts made to quantify the yields of HCFC 123a and CFC
1113 failed because their Henry’s law constants and their
solubilities in the two phases (H2O-saturated CFC 113 and water
saturated with Freon) are unknown. The presence of CFC 1113
as one of the products indicated that CFC 113 was, in part,
reduced via a two-electron reaction since the olefin is the
expected product of this type of reaction.27 Furthermore, the
photoreduction of halothane in air-saturated systems containing
platinized TiO2 particles proceeds via two consecutive electron
transfer steps to form the olefin 2-chloro-1,1-difluoroethylene,
as well as Br- and F- ions.21 This process is in competition
with a one-electron reduction reaction that yields Br- and
trifluoroacetate ions.
Figure 7. Changes in [Cl-] and [F-] as a function of illumination
time for a suspension at pH ) 7.3 that contained 2 mL of CFC 113a,
0.1 M HCO2-, and 0.5 g L-1 TiO2, with I0 ) 6 × 10-8 M (hν) s-1
.
saturated (but sealed) suspensions the rates of chloride ion
formation were about 170 times lower than those of degassed
samples. Furthermore, [F-] increased linearly with time, and
formation of Cl- and F- ions occurred with the same rate.
However, in all air-saturated samples the rate of chloride ion
formation increased abruptly at about 90 min. After this time
the halide ion generation proceeded with the same rates as the
second step in degassed suspensions. This means that O2 was
consumed in the first 90 min during the decay of the Freon,
but the CFC was photoreduced as in air-free systems once the
O2 molecules were eliminated. Obviously, the transformation
of the Freon in the absence of oxygen is different from the
pathway involving O2, which enhances generation of F- ions
and suppresses formation of Cl- ions. While the decay of CFC
113 in the presence of oxygen was not further investigated, we
suspect that the faster release of F- in these samples results
from a process similar to the photoreduction of halothane with
air that yields trifluoroacetate ions.21
Efforts were therefore made to determine the importance of
the consecutive two-electron process using 1,1,1-trichlorotrifluo-
roethane (CFC 113a) instead of CFC 113. Reduction of CFC
113a via a two-electron process is expected to generate Cl-
and F- ions in a ratio of [Cl-]/[F-] ) 1/1 according to the steps
Discussion
Photoreactions of chlorofluorocarbons have not been observed
in TiO2 suspensions that are free of electron donors because
the oxidizing species photogenerated in the semiconductor
particles are unable to attack C-Cl or C-F bonds.28 The
information gathered in this study shows that CFC 113 is
dechlorinated in illuminated anaerobic TiO2 suspensions con-
taining HCO2- ions as electron donors through a photoreduction
process. A slow step takes place during the first 8 min, where
formation of chloride ions is 10 times faster than generation of
F- ions. At longer times a faster zero-order step is observed,
with rates of Cl- ion generation that are more than 100 times
higher than the rates of F- ion formation. The slow initial step
is probably related to the presence of small amounts of O2 in
the suspensions. Observations made with air-containing systems
support this assumption, since similar two-step processes occur
in these systems. Initially, the Cl- ion formation is slow in the
presence of air, but production of fluoride ions can reach rates
that are equal to those of Cl- ions. Once the O2 molecules are
consumed, the rates of halide ion formation are the same as
during the second step of air-free samples; that is, formation of
Cl- ions is fast whereas generation of F- ions is much slower.
Evidently, O2 inhibits the reduction of the Freon; the inhibition
period is short in degassed solutions due to the very small
amounts of O2 present in these samples.
CCl3-CF3 + e- f •CCl2-CF3 + Cl-
•CCl2-CF3 + e- f CCl2dCF2 + F-
(1)
(2)
Experiments with CFC 113a were carried out by illuminating
suspensions containing 2 mL of this Freon at I0 ) 6 × 10-8
M
(hν) s-1
. The variations of the chloride and fluoride ion
concentrations with time are presented in Figure 7. Both [Cl-]
and [F-] increased linearly with time, with formation photonic
efficiencies of ú(Cl-) ) 1.5 and ú(F-) ) 0.08. Hence, ú(Cl-)
- ú(F-) ) 1.42 is the yield of Cl- ions generated by a reaction
different from the consecutive two-electron-transfer process
represented by reactions 1 and 2. It follows that the two-electron
reaction channel contributed with only 5.3% to the overall
reduction process of the CFC. This conclusion is supported
by GC/MS data which identified 2,2-dichloro-1,1,1-trifluoro-
ethane (HCFC 123) as the main reduction product. Only very
small amounts of the product of reaction 2, the olefin 1,1-
dichloro-2,2-difluoroethylene (CFC 1112a), were found.
A few photoreductions of CFC 113 were performed in the
presence of air. The rate of Cl- ion formation decreased by
50% after injecting 3 mL of air into a degassed TiO2 suspension
prephotolyzed for 40 min. Also, generation of F- ions was 5
times faster after the addition of oxygen as compared with the
initial rate of air-free suspensions. On the other hand, in air-
Inhibitions induced by oxygen and photonic efficiencies
higher than 1 are typical of chain reactions, such as radical
polymerizations initiated by semiconductor particles.10 Also,

Photoreduction of 1,1,2-Trichlorotrifluoroethane
J. Phys. Chem. B, Vol. 101, No. 19, 1997 3773
the reduction of the Freon in acetonitrile solutions with formate
ions is believed to follow a chain mechanism,16 and as in that
system, the photoreduction of CFC 113 on the TiO2 particles
yields mainly HCFC 123a. Hence, it seems logical to assume
that light induces a chain reduction of the Freon in the oxide
suspensions. The results of Figure 2 are consistent with this
assumption. Although some Freon is probably adsorbed on the
Although formation of chain carriers via reactions 6-9 may
•
-
•
yield CO2 and CClF-CF2Cl radicals adsorbed on the oxide
surface, some of these species are expected to desorb from the
surface and propagate the chain reaction in solution. It is clear
that the chain length of this process is short since postirradia-
tion formation of Cl- ions was not detected and because the
values of ú(Cl-) were much smaller than the efficiencies of
polymerization.10c This means that the chain process is
restricted to the solution region that is within a few nanometers
of the TiO2 surface, and only molecules or radicals that are
adsorbed on the surface or close to it can participate in reactions
6-11. Hence, the chain reduction mainly involves CFC
molecules located within the interfacial region. Since Freon
molecules next to the oxide-solution interface also participate
in the chain process, the results of Figures 2 and 3 are not
consistent with Langmuir-type kinetic models which assume
that only adsorbed substrate is transformed under illumina-
tion.33
oxide surface, the data obtained at [CFC 113] e 9 × 10-4
M
suggest that the amounts of adsorbed Freon are very small,
because the solubility limit of CFC 113 does not seem to be
affected by the presence of the TiO2 particles in the aqueous
solutions. Slow reductions take place below the solubility limit
since, as in the case of chain photopolymerizations,10 efficient
chain reactions are not sustained at low [substrate]. The rate
of the photoreduction increases significantly above the solubility
limit. Phase separation occurs at [CFC 113] > 9 × 10-4 M,
and we observed formation of small Freon droplets during
stirring, some of which probably exist next to the oxide surface.
We speculate that a large number of CFC molecules in these
droplets are therefore available to participate in the chain
process, which induces a drastic increase in k0.
The reduction is more efficient above the solubility limit of
the Freon since CFC molecules from droplets existing next to
the TiO2 particles can diffuse to the interfacial region and
participate in the chain process. As in the CdS-initiated
photopolymerization of methyl methacrylate in ethylene glycol,
where the monomer is phase-separated from the alcoholic phase
containing the semiconductor particles,10c the reduction of CFC
113 above the solubility limit is another example of phase
transfer photocatalysis. In these processes the reactant is
primarily segregated in a liquid phase other than the liquid phase
containing the photocatalyst. Transformation of the reactant
occurs in two steps: (1) diffusion through the low-solubility
medium and (2) chemical reaction next to the semiconductor
surface.
The data of Figure 3 show that formate ions act as efficient
electron donors in the photoreductions. Higher values of k0
result upon increasing [HCO2-], suggesting that these ions
-
participate in the chain reduction of the Freon. HCO2 ions
adsorbed on the oxide surface, or located next to it, react with
•
holes or OH radicals to yield reducing radicals.29 Due to the
•
acidic nature of CO2H radicals (pKa ) 1.4),30 oxidation of
•
-
formate ions in our systems (pH g 3) mainly yields CO2
radicals. These radicals are strong reductants (E0 ) -1.8 V
for CO2/•CO2- 30 and have been used as reducing agents in the
TiO2-initiated photoreduction of metal ions.31 While CO2
)
•
-
radicals are capable of reducing CFC 113 (half-wave potential
) -1.21 V),27 the radicals formed by oxidation of acetate ions
do not seem to behave as reducing agents,32 which explains the
smaller rates of Freon reduction when acetate was the electron
donor. Thus, formate not only suppresses charge carrier
recombination by reacting with the oxidizing species but also
forms radicals that participate in the reduction of CFC 113.
Increases in the [HCO2-] yield higher concentrations of electron
donor available to scavenge •OH radicals and holes, as well as
faster reductions of the Freon.
Participation of the CO2 radicals in the transformation of
CFC 113 is consistent with results from an earlier study showing
that these radicals initiate the chain reduction of CF3-CHBrCl
in homogeneous solutions.17 As in the case of CFC 113, a
hydrogen-substituted derivative (CF3-CH2Cl) is the main
product of the reduction of halothane in formate solutions. We
therefore propose the following mechanism for the chain
reduction of the Freon on illuminated TiO2 particles:
The variation of k0 with pH shown in the top curve of Figure
4 is similar to the dependence of reaction rate on pH for the
reduction of CCl4 in TiO2 suspensions containing acetate ions
as electron donors.19a Increases in k0 are observed upon
increasing the pH from 3 to 6; analogous trends are seen in the
other curves. This behavior is related to changes in the amounts
of formate ions that are either adsorbed or next to the oxide
surface. Positively charged surfaces exist at pH < pzc (pzc )
pH at point of zero charge, 6.2 e pzc e 6.5 for TiO2 P-25).3,34
-
Since the concentration of HCO2 ions increases as [H+]
•
-
decreases (pKa ) 3.77 for HCO2H), larger amounts of the anions
are attracted to the surface at the higher pH values. This leads
to a faster step 6 and to more efficient chain reactions. On the
other hand, the TiO2 surface becomes negatively charged upon
raising the pH above the pzc. Repulsion between HCO2- anions
and the oxide particles decreases the concentration of electron
donor in the interfacial region, which accounts for the lower
rates of reaction at pH > 6 in the top curve. The fact that k0
remains constant at pH > 8 is an indication that the concentra-
tions of electron donor located either at or next to the TiO2
surface are high enough to sustain chain processes.
-
TiO2 + hν f hvb+ + ecb
hvb+ + ecb- f heat
(3)
(4)
(5)
(6)
(7)
(8)
Much lower rates of reaction are observed in curves where
the pH was adjusted with buffers (2) or HClO4 (0). Anions
such as phthalate, phosphate, borate, and even perchlorate inhibit
the photoreduction of CFC 113. Further evidence that these
anions affect the chain reaction is that the maximum values of
k0 in these curves occur at the “natural” pH, that is, in the
absence of the anions. However, perchlorate, nitrate, and borate
H2O + hvb+ f •OH + H+
hvb+/•OH + HCO2- f •CO2- + H+/H2O
TiO2 + •CO2- f ecb- + CO2
CCl2F-CF2Cl + ecb- f •CClF-CF2Cl + Cl-
•
CCl2F-CF2Cl + •CO2- f •CClF-CF2Cl + Cl- + CO2 (9)
•CClF-CF2Cl + HCO2- f •CO2- + HCClF-CF2Cl (10)
ions do not react with OH radicals, but phthalate ions are
•
attacked quickly by OH (k ) 5.9 × 109 M-1 s-1).35 An
analogous situation is expected in the case of scavenging of
holes, where phthalate ions are probably the only anions that
•CClF-CF2Cl + ecb- f CClFdCF2 + Cl-
(11)
-
+
can compete efficiently with HCO2 for hvb
. Thus, the

3774 J. Phys. Chem. B, Vol. 101, No. 19, 1997
Weaver and Mills
-
inhibiting effect is not related to the reactivity of the anions
toward the light-generated oxidizers. A logical explanation is
that addition of buffers or ClO4- ions induces a decrease in the
with a rate constant close to that of the reaction between •CO2
and halothane (k9 ≈ 7.6 × 107 M-1 s-1),17 then combination of
•
-
•
-
•
two CO2 radicals, or of CO2 with CClF-CF2Cl, via step
concentrations of formate ions and of Freon molecules located
14 (k14 ≈ 1 × 109 M-1 s-1 17
) is unable to compete with reaction
-
within the interfacial region. Lowering the interfacial [HCO2
]
9 at our light intensities and [CFC] concentrations.
reduces the rates of reactions 6 and 10. Decreases in the rate
of step 6 favor the charge carrier recombination step 4 and
inhibit propagation via step 9. On the other hand, reaction 8
and propagation step 9 become slower upon decreasing the CFC
concentration in the interfacial region. Consequently, the net
effect of adding buffers or anions such as ClO4- or NO3- is to
shorten the length of the chain reaction.
For chain photopolymerizations, light intensity exponents of
less than 0.5 are typical of termination reactions involving chain
carriers and primary radicals, the species that initiate the chain
transformation.10a We propose an analogous termination reac-
tion for the reduction of CFC 113, which is represented by
reaction 11. Supporting this idea are GC/MS results showing
that small amounts of the expected product of step 11, CFC
1113, were formed in illuminated suspensions. The data
obtained in the photoreduction of CFC 113a also support the
proposed mechanism. According to the results of Figure 7 and
of GC-MS experiments, this CFC is dehalogenated with a
photonic efficiency of ú(Cl-) ) 1.5 to yield the hydrogen-
substituted Freon HCFC 123. Interestingly, the contribution
of reaction 2 (the termination step in this system) to the overall
dehalogenation of CFC 113a can be estimated from ú(F-). As
anticipated for a termination step, the photonic efficiency for
fluoride ions is only 0.08, and very small amounts of CFC 1112a
are generated. This information indicates that photoreduction
of CFC 113a in formate-containing TiO2 suspensions proceeds
via a free radical chain dehalogenation similar to that of CFC
113, and reactions involving electron transfer to Freon radicals
act as chain termination steps.
According to the data of Figure 5, ú(Cl-) increases with in-
creasing amounts of oxide particles at both light intensities,
reaching maximum values at 0.5 g L-1 TiO2 followed by a
decrease thereafter. A comparison of curves a and b for
suspensions containing 0.1 g L-1 or less TiO2 indicates that
smaller values of ú(Cl-) are obtained at the lower light intensity.
This is because the probability of a photon striking an oxide
particle is low when both the number of suspended particles
and the number of photons entering the sample are small. On
the other hand, larger amounts of photons are available to excite
the oxide particles at the higher I0 of curve b, resulting in larger
photonic efficiencies. Increases in ú(Cl-) occur above 0.1 g
L-1 TiO2, but the reactions are about 9 times more efficient at
the lower light intensity. These results are in agreement with
the trend shown in Figure 6, where ú(Cl-) increases with
decreasing I0. Maximum rates of reaction similar to those of
Figure 5 occur at 0.5 g L-1 TiO2 for a variety of photoreactions.3
This effect has been rationalized under the assumption that
scattering of light by the particles is very significant above 0.5
g/L TiO2, which reduces the rate of the processes.
The present study has shown that efficient semiconductor-
initiated photoreductions of Freons to form HCFC compounds
can be achieved under conditions that favor radical chain
reactions. Similar strategies may be useful for the transforma-
tion of undesirable materials into valuable chemicals.
A
Substantial enhancements in the efficiency of photoinitiated
chain polymerizations take place in systems containing high
concentrations of semiconductor when the intensity of photons
is decreased.10 The data of Figure 6 indicate that an analogous
but less pronounced trend is followed in the photoreduction of
CFC 113. Slower charge carrier recombination and termination
steps (reactions 4 and 11) occur as I0 decreases, which results
in improvements in the chain length and values of ú(Cl-) larger
than 1. The rate expression for chloride ion formation is
remaining problem is the low efficiencies of Freon reduction
at high light intensities, which are about 5 times smaller than
the values of polymerization.10c Further investigations are
needed in order to address this problem, and to explore the
possible application of our strategy to other systems.
Acknowledgment. We thank Degussa for a gift of TiO2 P-25
samples and G. Goodloe for helping us with GC/MS measure-
ments. We are grateful to Auburn University for a fellowship
to S.W. through the PGOP program and to NTC for partial
support of this research.
d[Cl-]/dt ) R(Cl-) )
- x
k(I0)w[HCO2 ] [CFC 113]y[TiO2]z (12)
References and Notes
where R(Cl-) is the rate of Cl- ion formation. Since ú(Cl-) )
R(Cl)/I0, and because light intensity was the only changing
variable in the experiments of Figure 6, eq 12 can be written as
(1) (a) Kamat, P. V. Chem. ReV. 1993, 93, 267. (b) Fox, M. A. Chem.
ReV. 1993, 93, 341.
(2) Darwent, J. R.; Lepre, A. J. Chem. Soc., Faraday Trans. 2 1986,
82, 2323.
(3) Hoffmann, M. R.; Martin, S. T.; Choi, W.; Bahnemann, D. W.
Chem. ReV. 1995, 95, 69.
ú(Cl-) ) kobs(I0)w-1
(13)
(4) (a) Serpone, N.; Terzian, R.; Minero, C.; Pelizzetti, E. In Photo-
sensitiVe Metal-Organic Systems; Kutal, C., Serpone, N., Eds.; Advances
in Chemistry Series 238; American Chemical Society: Washington, DC,
1993; p 281. (b) Serpone, N.; Terzian, R.; Lawless, D.; Kennepohl, P.;
Suave, G. J. Photochem. Photobiol. A: Chem. 1993, 73, 11.
(5) Schwitzgebel, J.; Eckerdt, J. G.; Gerischer, H.; Heller, A. J. Phys.
Chem. 1995, 99, 5633.
where kobs ) k[HCO2-]x[CFC 113]y[TiO2]z. Hence, a plot of
log ú(Cl-) vs log(I0) should yield a straight line with a slope of
w - 1. Such a plot is shown in the inset of Figure 6; a value
of w ) 0.36 is calculated from the slope of the straight line.
This reaction order with respect to I0 is lower than the value of
0.5 predicted by simple mechanisms that assume dimerization
of the chain carriers as the main termination channel:10b
(6) Vinodgopal, K.; Stafford, U.; Gray, K.; Kamat, P. V. J. Phys. Chem.
1994, 98, 6797.
(7) Kesselman, J. M.; Shreve, G. A.; Hoffmann, M. R.; Lewis, N. S.
J. Phys. Chem. 1994, 98, 13385.
(8) Peterson, M. W.; Turner, J. A.; Nozik, A. J. J. Phys. Chem. 1991,
91, 221.
2R• f R-R
(14)
(9) (a) Micic, O. I.; Zhang, Y.; Cromak, K. R.; Trifunac, A. D.;
Thurnauer, M. C. J. Phys. Chem. 1993, 97, 7277. (b) Lawless, D.; Serpone,
N.; Meisel, D. J. Phys. Chem. 1991, 95, 5166.
(10) (a) Huang, Z.-Y.; Barber, T.; Mills, G.; Morris, M.-B. J. Phys.
Chem. 1994, 98, 12746. (b) Hoffman, A. J.; Yee, H.; Mills, G.; Hoffmann,
M. R. J. Phys. Chem. 1992, 96, 5540. (c) Hoffman, A. J.; Mills, G.; Yee,
H.; Hoffmann, M. R. J. Phys. Chem. 1992, 96, 5546.
•
-
•
where R• ) CO2 or CClF-CF2Cl and R-R are the dimer-
ization products. The fact that GC/MS analysis of illuminated
samples failed to detect the dimer (CClF-CF2Cl)2, one of the
expected products of reaction 14, is another indication that this
reaction is not an important termination step. If step 9 proceeds

Photoreduction of 1,1,2-Trichlorotrifluoroethane
J. Phys. Chem. B, Vol. 101, No. 19, 1997 3775
(11) Al-Ekabi, H.; de Mayo, P. J. Phys. Chem. 1986, 90, 4075.
(12) Okada, K.; Hisamitsu, K.; Takashi, Y.; Hanaoka, T.; Miyashi, T.;
Mukai, T. Tetrahedron Lett. 1984, 25, 5311.
(13) (a) Evans, R.; Nesyto, E.; Radlowski, C.; Sherman, W. V. J. Phys.
Chem. 1971, 75, 2762. (b) Bunnett, J. F.; Wamser, C. C. J. Am. Chem.
Soc. 1967, 89, 6712.
(24) Heller, H. G.; Langan, J. R. J. Chem. Soc., Perkin Trans. 2 1981,
341.
(25) Horvath, A. L. Halogenated Hydrocarbons: Solubility-Miscibility
with Water; Marcel Dekker: New York, 1982; pp 709-710.
(26) Lisenbigler, A. L.; Lu, G.; Yates, J. T., Jr. Chem. ReV. 1995, 95,
735.
(14) Horowitz, A. In The Chemistry of Functional Groups, Supplement
D, Part 1; Patai, S., Rappoport, Z., Eds.; J. Wiley: Chichester, 1983; p
369.
(15) (a) Hawari, J.; Demeter, A.; Samson, R. EnViron. Sci. Technol.
1992, 26, 2022. (b) Singh, A.; Kremers, W.; Smalley, P.; Bennett, G. S.
Radiat. Phys. Chem. 1985, 25, 11.
(16) Hu, C.-M.; Tu, M.-H. J. Fluorine Chem. 1991, 55, 105.
(17) Mo¨nig, J.; Asmus, K.-D. J. Chem. Soc., Perkin Trans. 2 1984, 2057.
(18) Alfassi, Z. B.; Heusinger, H. Radiat. Phys. Chem. 1983, 22, 995.
(19) (a) Choi, W.; Hoffmann, M. R. EnViron. Sci. Technol. 1995, 29,
1646. (b) Choi, W.; Hoffmann, M. R. J. Phys. Chem. 1996, 100, 2161. (c)
Hilgendorff, M.; Hilgendorff, M.; Bahnemann, D. W. J. AdV. Oxid. Technol.
1996, 1, 35.
(27) Hawley, M. D. In Encyclopedia of Electrochemistry of the Elements;
Bard, A. J., Lund, H. Eds.; Marcel Dekker: New York, 1980; Vol. XIV,
pp 35-40.
(28) Sabin, F.; Turk, T.; Vogler, A. J. Photochem. Photobiol. A: Chem.
1992, 63, 99.
(29) Augustynski, J. In Structure and Bonding. Solid Materials; Springer-
Verlag: Berlin, 1988; Vol. 69, p 1.
(30) Stanbury, D. M. AdV. Inorg. Chem. 1989, 33, 69.
(31) (a) Forouzan, F.; Richards, T. C.; Bard, A. J. J. Phys. Chem. 1996,
100, 18123. (b) Prairie, M. R.; Evans, L. R.; Stange, B. M. EnViron. Sci.
Technol. 1993, 27, 1776.
(32) Schuchmann, M. N.; Zegota, H.; von Sonntag, C. Z. Naturforsch.
1985, 40B, 215.
(20) Kuhler, R. J.; Santo, G. A.; Caudill, T. C.; Betterton, E. A.; Arnold,
R. G. EnViron. Sci. Technol. 1993, 27, 2104.
(33) Cunningham, J.; Al-Sayyed, G.; Srijarani, S. In Aquatic and Surface
Photochemistry; Helz, G. R., Zepp, R. G., Crosby, D. G., Eds.; CRC
Press: Boca Raton, FL, 1994; p 317.
(34) (a) Akratopulu, K. Ch.; Kordulis, C.; Lycourghiotis, A. J. Chem.
Soc., Faraday Trans. 1990, 96, 3437. (b) Janssen, M. J. G.; Stein, H. N. J.
Colloid Interface Sci. 1986, 109, 508. (c) Jaffrezic-Renault, N.; Pichat, P.;
Foissy, A.; Mercier, R. J. Phys. Chem. 1986, 90, 2733.
(35) Buxton, G. V.; Greenstock, C. L.; Helman, W. L.; Ross, A. B. J.
Phys. Chem. Ref. Data 1988, 17, 513.
(21) Bahnemann, D. W.; Mo¨nig, J.; Chapman, R. J. Phys. Chem. 1987,
91, 3782.
(22) Shen, T.-L.; Wooldridge, P. J.; Molina, M. J. In Composition,
Chemistry and Climate of the Atmosphere; Singh, H. B., Ed.; Van Nostrand
Reinhold: New York, 1995; Chapter 12.
(23) Ravishankara, A. R.; Lovejoy, E. R. J. Chem. Soc., Faraday Trans.
1994, 90, 2159.