Isotope studies of photocatalysis: Dual mechanisms in the conversion of anisole to phenol

Article abstract of DOI:10.1021/ja994030h

The partial TiO₂-mediated photocatalytic degradation of anisole has been studied using isotopically labeled substrates and comparisons to Fenton and hv/H₂O₂ chemistry. An H/D isotope selectivity of 2.7 is observed for hydrogen abstraction from the methyl group in all cases. An ¹⁸O tracer study shows that anisole is simultaneously converted to phenol by both ipso attack and degradation of the methyl. The isotope selectivity for hydroxylation of anisole is reported to be about 1.3 under all conditions. This is attributed to an O₂-dependent competition along the reaction pathway that can be avoided by use of benzoquinone instead of O₂ as an electron acceptor with TiO₂. Under those conditions, the H/D isotope selectivities are between 0.9 and 1.0, as would be expected for inverse secondary kanetic isotope effects.

Full text of DOI:10.1021/ja994030h

11864
J. Am. Chem. Soc. 2000, 122, 11864-11870
Isotope Studies of Photocatalysis: Dual Mechanisms in the
Conversion of Anisole to Phenol
Xiaojing Li and William S. Jenks*
Contribution from the Department of Chemistry, Iowa State UniVersity, Ames, Iowa 50011-3111
ReceiVed NoVember 15, 1999. ReVised Manuscript ReceiVed October 3, 2000
Abstract: The partial TiO2-mediated photocatalytic degradation of anisole has been studied using isotopically
labeled substrates and comparisons to Fenton and hν/H2O2 chemistry. An H/D isotope selectivity of 2.7 is
observed for hydrogen abstraction from the methyl group in all cases. An 18O tracer study shows that anisole
is simultaneously converted to phenol by both ipso attack and degradation of the methyl. The isotope selectivity
for hydroxylation of anisole is reported to be about 1.3 under all conditions. This is attributed to an O2-
dependent competition along the reaction pathway that can be avoided by use of benzoquinone instead of O2
as an electron acceptor with TiO2. Under those conditions, the H/D isotope selectivities are between 0.9 and
1.0, as would be expected for inverse secondary kinetic isotope effects.
Introduction
been concerned with the chemical mechanisms by which
aromatic molecules are degraded. Ring opening of benzene
moieties appears to be driven by SET-derived chemistry, but
only after two or three oxygenations of the ring. Other reactions,
such as degradation of alkyl substituent moieties, hydroxylation
of unsubstituted aryl positions, and aryl substitution reactions
are all processes that occur in competition “early” in the
degradations and are usually attributed to hydroxyl radical
chemistry.
A mechanistic issue of interest is the manner in which
substituents are lost: by attack on the arene or by degradation
of the exterior alkyl substituent, i.e., is degradation of the alkyl
or related stubstituents typically faster than, slower than, or
competitive with reactions at the arene portion. The first steps
of both types of process, addition to aromatic rings and hydrogen
abstraction from alkyl positions, are well-known reactions of
HO•. Conversion of chloroarenes to phenols occurs by ipso
attack of HO• 22 on the aromatic ring, but the situation is less
clear for other substituents.23 For example, alkoxy substituents
in particular could be “substituted for” or could have the alkyl
portion degraded away; both paths lead to phenols.
Titanium dioxide mediated photocatalytic degradation of
organic compounds in aerated water leads to their complete
mineralization, that is, oxidation to CO2, H2O, and other
inorganic ions.1-7 It is thus an attractive technology for certain
water purification applications. The initial event in photocatalytic
degradation is charge separation within the semiconductor
particle, induced by absorption of near-UV light. Superoxide
(O2•-) is formed by rapid electron transfer to molecular oxygen,
which prevents rapid charge recombination within the TiO2
particle. The remaining valence band hole can oxidize water to
give an adsorbed hydroxyl radical and a proton or oxidize an
adsorbed organic substrate directly by single electron transfer
(SET).
Much of the degradative chemistry has been attributed to
reactions of surface bound hydroxyl radicals with the organics,
but some products appear to derive from reactions of the SET-
oxidized organic compound.8-21 In particular, we have recently
(1) Pelizzetti, E.; Serpone, N. Homogeneous and Heterogeneous Photo-
catalysis; D. Reidel Publishing Company: Boston, MA, 1986.
(2) Serpone, N.; Pelizzetti, E. Photocatalysis: Fundamentals and Ap-
plications; John Wiley & Sons: New York, 1989.
(3) Helz, G. R.; Zepp, R. G.; Crosby, D. G. Aquatic and Surface
Photochemistry; Lewis Publishers: Boca Raton, FL, 1994.
(4) Fox, M. A.; Dulay, M. T. Chem. ReV. 1993, 93, 341-357.
(5) Linsebigler, A. L.; Lu, G.; Yates, J. T., Jr. Chem. ReV. 1995, 95,
735-758.
(6) Stafford, U.; Gray, K. A.; Kamat, P. V. Heterog. Chem. ReV. 1996,
3, 77-104.
(7) Schiavello, M. Heterogeneous Photocatalysis; Wiley: New York,
1997.
We report here a study of the initial steps of photocatalytic
degradation of anisole, examining both its hydroxylation and
its conversion to phenol. It was chosen as the simplest model
with which reactions of alkoxyarenes could be examined, and
one in which HO•-mediated chemistry was expected to dominate
over SET-induced chemistry based on our previous experi-
(15) Piccinini, P.; Minero, C.; Vincenti, M.; Pelizzetti, E. Catal. Today
1997, 39, 187-195.
(8) Li, X.; Cubbage, J. W.; Tetzlaff, T. A.; Jenks, W. S. J. Org. Chem.
1999, 64, 8509-8524.
(16) Minero, C.; Mariella, G.; Maurino, V.; Pelizzetti, E. Langmuir 2000,
16, 2632-2641.
(9) Li, X.; Cubbage, J. W.; Jenks, W. S. J. Org. Chem. 1999, 64, 8525-
(17) Sahyun, M. R. V.; Serpone, N. J. Photochem. Photobiol. A 1998,
115, 231-238.
8536.
(10) Cermenati, L.; Pichat, P.; Guillard, C.; Albini, A. J. Phys. Chem. B
1997, 101, 2650-2658.
(11) Amalric, L.; Guillard, C.; Pichat, P. Res. Chem. Intermed. 1994,
20, 579-594.
(12) Theurich, J.; Bahnemann, D. W.; Vogel, R.; Ehamed, F. E.;
Alhakimi, G.; Rajab, I. Res. Chem. Intermed. 1997, 23, 247-274.
(13) Maillard-Dupuy, C.; Guillard, C.; Courbon, H.; Pichat, P. EnViron.
Sci. Technol. 1994, 28, 2176-2183.
(14) Maillard-Dupuy, C.; Guillard, C.; Pichat, P. New J. Chem. 1994,
18, 941-948.
(18) Galindo, C.; Jacques, P.; Kalt, A. J. Photochem. Photobiol. A 2000,
130, 35-47.
(19) Guillard, C. J. Photochem. Photobiol. A 2000, 135, 65-75.
(20) Soana, F.; Sturini, M.; Cermenati, L.; Albini, A. J. Chem. Soc.,
Perkin Trans. 2 2000, 699-704.
(21) Cermenati, L.; Albini, A.; Pichat, P.; Guillard, C. Res. Chem.
Intermed. 2000, 26, 221-234.
(22) The use of the term “HO•” is elaborated in the Discussion section.
(23) Stafford, U.; Gray, K. A.; Kamat, P. V. J. Phys. Chem. 1994, 98,
6343-51.
10.1021/ja994030h CCC: $19.00 © 2000 American Chemical Society
Published on Web 11/11/2000

Isotope Studies of Photocatalysis
J. Am. Chem. Soc., Vol. 122, No. 48, 2000 11865
ence.8,9 We show that phenol is formed competitively by both
(a) the substitution reaction of HO for CH3O beginning with
ipso attack and (b) the degradation of the methyl group
beginning with hydrogen abstraction. Additionally, we use
isotopic labels to examine the hydroxylation and hydrogen
abstraction reactions that occur in various positions.
This finding of a substitution mechanism for the conversion
of RO to HO stands in contrast to previous results for cyanuric
acid derivatives. The common herbicide atrazine (6-chloro-N-
ethyl-N′-isopropyl[1,3,5]triazine-2,4-diamine) is degraded to
cyanuric acid but no further.24-26 We have recently shown that
model alkoxytriazines are converted to cyanuric acid without
attack at the ring and that TiO2-mediated exchange of the oxygen
atoms is very slow if it occurs at all.27
Table 1. Relative Yields of Products Formed by TiO2-Mediated
Photocatalytic Degradation, H2O2 Photolysis, and Fenton Reactionsa
relative yield
entry
conditions
pH
2
3
(18O)b
4
5
6
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
hv + TiO2 + O2
1.0 28 68
2.0 27 43
4.0 16 30
(31)
3
18
1
9
4
1
1
(19)
(17)
27 26
29 40
32 44
34 13
62 23
53 36
7.0
8.7
10.8
1.0
7.0
7.0
6
1
0
0
0
0
24
23
44
15
8
(29)
(19)
9
0
3
1
2
1
hv + TiO2 + BQ
hv + TiO2
hv + H2O2 + O2
90
7
4
2
5
1.0 69 21
7.0 22
1.0 69 36
7.0 24
1.0 62 33
7.0 22
1.0 33 66
7.0 18
3
(14)
(16)
48 23
2
55 17
5
hv + H2O2
Experimental Section
0
4
5
4
Materials. Anisole (1), anisole-d3 (C6H5OCD3), and anisole-d8 were
used as received from Aldrich. Anisole-d5 (C6D5OCH3) was prepared
by the reaction of phenol-d6 with CH3I in the presence of K2CO3 and
acetone. 18O-enriched anisole was prepared by the reaction of CH3I
with 18O-enriched phenol, which was prepared as described by Winkel.28
Phenyl formate was prepared according to the method described by
Stevens.29 The water employed was purified by a Milli-Q UV plus
Fe2+ + H2O2 + O2
Fe2+ + H2O2
1
45 27
0
46 32
a All entries were from at least duplicate runs, except the 18O
b
abundance of phenol. 18O percentage in 3 when starting material 1
contains 71.6% 18O.
system (Millipore) resulting in a resistivity more than 18 MΩ cm-1
.
TiO2 was Degussa P-25, which consists of 75% anatase and 25% rutile
with a specific BET surface area of 50 m2 g-1 and a primary size of 20
nm.30 Other chemicals were obtained from Aldrich as the highest grade
available and purified as necessary.
array UV/vis detection and an ODS Hypersil C18 reverse phase column
(5 µm loop, 200 × 2.1 mm). Substances were routinely quantified from
their absorbance at 220 nm. The eluent was 15% aqueous methanol
and the flow rate was 0.5 mL min-1. First-order decays were obtained
from plots of concentration of anisole vs time. Relative standard errors
of the fits were all <5% and absolute time constant values from
independent runs were reproducible to within 10%. The relative rates
are of the greatest significance, since the absolute rates are functions
of light intensity, sample geometry, etc.
Photocatalytic Degradations. Samples containing the desired
concentration of anisole and 50 mg of suspended TiO2 were prepared
in water (80 mL). Three types of runs were carried out. Kinetics runs
were analyzed by HPLC and the initial concentration of anisole was
2.0 mM. Product distribution runs were analyzed by GC, and the initial
concentration of anisole was 4.0 mM. For isotope selectivity runs using
a pair of anisole isotopomers, total initial concentrations were 1-5 mM,
while the ratio of one anisole isotopomer to the other was varied from
run to run. Products were again analyzed by GC. When used (as noted
in the table entries), the initial concentration of benzoquinone (BQ)
was 10 mM.
GC-MS of Product Distributions and Isotope Selectivity Runs.
After reactions were stopped, the solutions were acidified to pH 2.5
and filtered if necessary, then extracted with ether four times. The ether
solutions were dried with MgSO4 and concentrated to 1 mL. The
concentrated sample was directly injected. 4-Methylbenzyl alcohol was
used as an internal standard for integration by addition after the reaction.
The GC-MS analyses were carried out on a Finnigan Magnum ion trap
mass spectrometer equipped with a 30 m DB-5 column. A HP 5890
series II Gas Chromatograph with 30 m ZB-5 and FID detection was
used for routine work.
Degradations were carried out at a variety of pH values. Between
pH 4.0 and 8.7, 1 mM phosphate buffer was used, and pH 10.8 was
maintained with 25 mM phosphate buffer. The pH 1 and 2 solutions
were obtained by adding perchloric acid.
Prior to irradiation, each mixture was treated in an ultrasonic bath
for 5 min to disperse larger aggregates of TiO2 and then purged with
O2 (or Ar as noted) for 20 min in the dark. The samples were sealed
with a rubber septum and the headspace (∼90 cm3) was filled with O2
(or Ar, as appropriate). Irradiations were carried out at 25 ( 2 °C in a
Rayonet photochemical minireactor equipped with a magnetic stirring
device and a fan. Light was provided by 8 × 4 W “black light” bulbs
whose emission was centered at 360 nm. Control experiments showed
that anisole did not degrade on the time scale of the experiments if
any one of the components of the mixtures or light was not present.
HPLC Analysis of Kinetic Runs. Samples of approximately 0.5
mL were taken through the septum by syringe before photolysis and
at regular intervals during the irradiation. They were filtered using
syringe-mounted 0.2 µm Millipore filters before HPLC analysis. The
concentrations of anisole, phenol, 2-methoxyphenol, and 4-methoxy-
phenol were measured by HPLC using a HP 1050 HPLC with diode
Entries in Table 1 are based on at least two independent runs and
multiple GC shots for each run. Run-to-run reproducibility was better
than 10% relative error. Some of the 18O abundance runs were only
carried out once.
Entries in Table 2 are based on integration of the separated GC peaks,
including product isotopomers, which were also separable by GC.
Response factors for the commercial isotopomers of anisole were all
within 1% of each other and thus it was assumed that all isotopomer
pairs would also have identical response factors. All reported values
in Table 2 represent at least duplicate shots. The total range of product
ratios for several GC shots never exceeded 5% of the reported value.
Seven independent runs of different ratios of 1 and 1-d8 were performed
at pH 7, all yielding nearly identical selectivity values. For other
conditions, fewer runs were carried out. A complete table of all runs is
available as part of the Supporting Information.
(24) Minero, C.; Maurino, V.; Pelizzetti, E. Res. Chem. Intermed. 1997,
23, 291-310.
H2O2 Photolysis. Solutions were prepared as above, leaving out the
TiO2. Immediately before photolysis 1.0 mL of H2O2 (30% in water)
was added. The mixtures were irradiated with 8 × 4 W lamps whose
emission is centered at 300 nm. Analyses were done as previously
described.
(25) Pelizzetti, E.; Maurino, V.; Minero, C.; Carlin, V.; Pramauro, E.;
Zerbinati, O.; Tosato, M. L. EnViron. Sci. Technol. 1990, 24, 1559-1565.
(26) Yue, P. L.; Allen, D. Photocatalytic Degradation of Atrazine. In
Photocatalytic Purification and Treatment of Water and Air; Ollis, D. F.,
Al-Ekabi, H., Eds.; Elsevier: New York, 1993; Vol. 3, pp 607-611.
(27) Tetzlaff, T.; Jenks, W. S. Org. Lett. 1999, 1, 463-465.
(28) Winkel, C.; Aarts, M. W. M. M.; Van der Heide, F. R.; Buitenhuis,
E. G.; Lugtenburg, J. Recl. TraV. Chim. Pays-Bas 1989, 108, 139-146.
(29) Sevens, W.; Van Es, A. Recl. TraV. Chim. 1964, 83, 1294-1298.
(30) DeGussa DeGussa Technol. Bull. 1984, 56, 8.
Fenton Reactions. Reactions were conducted at room temperature.
Normal conditions were 4 mM anisole, 8 mM FeSO4, and 80 mM H2O2.
The pH of the solution was adjusted with H2SO4 for runs at pH 1, and
with phosphate buffer (0.1 M) for runs at pH 7. Analyses were done
as previously described.

11866 J. Am. Chem. Soc., Vol. 122, No. 48, 2000
Li and Jenks
Table 2. the Isotope Selectivity Ratios Observed on Degradations
of Isotopomeric Mixtures of 1
isotope selectivity
starting
entry materials
conditions
pH
2
3
4
5
1
2
3
4
5
6
7
8
9
1, 1-d8
hν, TiO2, O2
1.0 2.48 1.35 1.27 1.30
2.0 2.54 1.43 1.17 1.46
4.0 2.62 1.27 1.20 1.07
7.0a 2.73 1.30 1.31 1.17
8.5 2.80 1.34 1.35 1.19
hν, TiO2, BQb
7.0
1.31 0.94 0.98
Fe2+, H2O2, O2 7.0 2.72 1.15 1.11 1.08
b
b
hν, H2O2, O2
7.0 2.70 1.14 1.30 1.04
7.0 1.34 0.95 0.94
7.0 1.04 0.98 1.30
b
hν, H2O2
10 1, 1-d5
11
hν, TiO2, O2
b
Fe2+, H2O2, O2 7.0 1.01 0.96 1.09
12 1-d3, 1-d5 hν, TiO2, O2
13
Fe2+, H2O2, O2 7.0 2.39c 1.29
14 1-d3, 1-d8 hν, TiO2, O2 7.0 0.96 1.06
15
Fe2+, H2O2, O2 7.0 1.01 0.99
7.0 2.71 1.31
b
Figure 1. Rates of degradation of 1 as a function of pH. The initial
concentrations were ca. 2.0 mM. See text for details of buffers.
a Average of seven independent runs under different ratios of 1/1-
d8 of 0.44 to 3.03. Standard deviations of all ratios are (0.03. b Average
finally, mixtures of 1 and the illustrated isotopomers were
degraded to determine H/D selectivities for the formation of
products 2-6.
c
of two runs. Subject to larger error because of small quantities of
product.
Results
Kinetics Runs. To design conditions with which to study
the initial product distributions of the degradations, kinetics runs
were carried out. The rate of degradation of a given organic
substrate is expected to vary with pH because of the changing
surface charge on the TiO2, the change in adsorption charac-
teristics of the organics, and other factors. Usually, degradation
is faster at higher pH.6 Because photocatalytic degradations
inherently create acid, buffers are generally required to study
pH effects, and were used here except at the lowest pH values.
The rate of decay of 1 and the buildup and decay of several
products were followed by HPLC analyses of photocatalytic
degradations. As illustrated in Figure 1, the disappearance of 1
was exponential over at least 2-3 half-lives. The rate of decay
increased almost linearly with pH until a dip from pH 8.8 to
10.8. Similar correlations have been observed previously for
hydroquinone and other aromatic systems.6,8,38
The oxidation of anisole has been studied previously under
pulse radiolysis,31-33 Fenton,34 and other conditions,35-37 but
not by photocatalysis. Here, partial degradations of anisole and
several isotopomers were carried out using TiO2-mediated
photocatalysis, the Fenton reaction, and hydrogen peroxide
photolysis. The latter two sets of conditions were used as control
studies to characterize the TiO2 chemistry. In general, the results
of all three sets of degradation conditions were very similar
(vide infra), and this was taken as evidence that the TiO2-
mediated degradations were initiated mainly by adsorbed HO•,
not hole oxidation. This result is consistent with our previous
and continuing work with similar substrates.8,9
In addition to decay of 1, the buildup and loss of several
degradation intermediates were monitored. By HPLC, only
compounds 3-5 were uniquely identified as products. (In GC-
detected runs described below, primary degradation products 2
and 6 were also uniquely identified.) Additional small, fast-
running (polar) HPLC peaks that have been shown8,9 to include
multiple acyclic acids and/or polyhydroxylated aromatics were
also observed, but were taken to represent downstream degrada-
tion. The concentrations of 3-5 as a function of time for a pH
7 solution are given in Figure 2.
Determination of Primary Product Distributions. On the
basis of data in Figures 1 and 2, 30 min was chosen as a standard
degradation time that would yield a representative initial primary
product distribution. The results of these reactions are given in
Table 1, which only lists the major primary products. Analyses
carried out by GC and GC-MS showed other trace compounds
in the mixtures: di- and trihydroxylated species and ring-opened
compounds.8,9 Phenyl formate (2) has not been reported previ-
ously as an anisole degradation product, but its analogue has
been observed in the degradation of more highly substituted
cases.39 It should be noted that 2 is subject to hydrolysis or
secondary photocatalytic degradation to 3. It is thus assumed
that the values in Table 1 are lower limits on the actual
proportion of 2 formed.
Photocatalytic degradations were carried out in O2-saturated
aqueous solutions at several pH values. Three different types
of measurements were made. Kinetics of degradation as a
function of pH were obtained using periodic sampling and
HPLC detection. Product distributions were determined by GC
analysis of runs designed from the kinetic information, and
(31) Fang, X.; Pan, X.; Rahmann, A.; Schuchmann, H.-P.; von Sonntag,
C. Chem. Eur. J. 1995, 1, 423-429.
(32) Steenken, S.; Raghavan, N. V. J. Phys. Chem. 1979, 83, 3101-
3107.
(33) Eberhardt, M. K. J. Phys. Chem. 1977, 81, 1051-1057.
(34) Kurata, T.; Watanabe, Y.; Katoh, M.; Sawaki, Y. J. Am. Chem.
Soc. 1988, 110, 7472-7478.
(35) Jefcoate, C. R. E.; Norman, R. O. C. J. Chem. Soc. B 1968, 48-
53.
(36) Perry, R. A.; Atkinson, R.; Pitts, J. N., Jr. J. Phys. Chem. 1977, 81,
1607-1611.
(37) O’Neill, P.; Schulte-Frohlinde, D.; Steenken, S. J. Chem. Soc.,
Faraday Discuss. 1977, 63, 141-148.
(38) Djebbar, K.; Sehili, T. Pestic. Sci. 1998, 54, 269-276.
(39) Guillard, C.; Amalric, L.; D’Oliveira, J.-C.; Delprat, H.; Hoang-
Van, C.; Pichat, P. Aquat. Surf. Photochem. 1994, 369-386.

Isotope Studies of Photocatalysis
J. Am. Chem. Soc., Vol. 122, No. 48, 2000 11867
Figure 2. Concentrations of intermediate degradation products from
photocatalytic degradation of anisole. Initial conditions: 2.0 mM
anisole, 5 mM phosphate buffer, pH 7, 50 mg of TiO2 in 80 mL. The
decay rate of 1 is 0.0128 min-1
.
Figure 3. Ion relative abundance for determination of 18O content.
Normalized relative abundances were from a single GC-MS run after
partial photocatalytic degradation at pH 7. The phenol fragment (base
peak) was used for phenyl formate.
The product distributions were sensitive to pH. Phenol and
its formate ester predominate at low pH, but the proportion of
ortho and para hydroxylation to give 4 and 5 rises until pH
8.7. Concomitant with the drop in the reaction rate, the
proportion of 4 and 5 drops again at the highest pH.
Discussion
In a few runs, benzoquinone (BQ) was used as an electron
acceptor in lieu of O2, an idea taken from the result that BQ is
reduced to hydroquinone in the absence of O2 under photocata-
lytic conditions.40,41 In good accord with radiolysis experiments
in the presence of BQ,32 hydroxylation dominates the product
distribution, even at low pH.
Degradations using Other Conditions. To compare with
the TiO2-mediated photocatalytic degradation, photolysis of O2-
saturated solutions containing H2O2 were carried out, as were
dark Fenton reactions with excess H2O2. Both were kept to low
conversion of anisole, and the results of these experiments are
shown in Table 1 as entries 10-13 and 14-17, respectively.
The product distributions for the Fenton chemistry as a function
of pH were in good conformity with a previous report.35
Isotope Selectivities. Isotope selectivities were viewed as a
method to examine the mechanism(s) of the reactions. Reactions
were carried out at approximately 4 mM total concentration of
mixtures of the H/D isotopomers of 1, using various ratios of
the labeled and unlabeled materials. Most of the product
isotopomers were separable by GC.42
The quantities of the products 2-5 that derived from each
isotopomer were determined from the GC runs. Selectivities,
shown in Table 2, were obtained by dividing the observed
isotopomer ratios of the products by the initial isotopomeric
ratio of the anisole. A version of Table 2 that lists each run
with its individual ratio of anisole isotopomers is available in
the Supporting Information.
In several degradation runs, anisole that had been enriched
in 18O (1-18O) was used. Its initial isotopic enrichment was 72%
as determined by mass spectrometry. The isotopic enrichment
of the products under various conditions was examined. The
enrichment of 2 and 4-6 remained at 72% within a small
experimental error under all conditions. The enrichment for
phenol, however, varied and is reported in Table 1. Representa-
tive data are shown in Figure 3.
Absorption of UV light by TiO2 results in charge separation
in the particle. To avoid rapid charge recombination and energy
wasting, an electron acceptor is necessary; this role is usually
filled by O2. The remaining “hole” causes single electron
oxidation of adsorbed substrates. Radical cations can be formed
from organic substrates or surface-bound hydroxyl radicals may
be formed. Because anisole is not strongly adsorbed to TiO2
and because of the similarity of product distributions obtained
from photocatalytic degradation, the Fenton reactions, and the
hydrogen peroxide photolyses, the assumption is made through-
out this discussion that all the photocatalytic degradation
reactions considered here begin with species related to the
adsorbed hydroxyl radical, rather than SET. Further, just as it
has been argued that the oxidant in TiO2-mediated photocatalytic
degradation is not a free-floating hydroxyl radical, it has recently
been argued that Fenton chemistry is distinct from free hydroxyl
radical chemistry.43 It is not our purpose here to argue the exact
nature of the oxidizing agents in either of these conditions. The
important notion to read from this is that the mode of reactivity
in the photocatalytic conditions is that of the hydroxyl radical
type, as opposed to the single electron-transfer type. For
purposes of clarity, we will write the species as HO•, though it
must be clear that this may include surface- or metal-bound
species.
Multiple Mechanisms for Phenol Formation by Photo-
catalysis. Hydroxyl radical has a well-deserved reputation for
addition reactions to aromatic rings, particularly with electron-
rich systems, such as anisole. Indeed, the degradation of anisole
with hydroxyl radicals produced by radiolysis seems to have
been rationalized without consideration that HO• might have
reacted with the methyl group.32 In contrast, a gas-phase study
of anisole showed that at room temperature approximately 20%
of the reactivity of HO• with anisole was expected at the methyl
position by extrapolation of the data.36 The current results for
photocatalytic degradation are more consistent with the latter
interpretations, and lead to the conclusion that there are at least
(40) Richard, C. New J. Chem. 1994, 18, 443-445.
(41) Richard, C.; Boule, P. New J. Chem. 1994, 18, 547-52.
(42) The methoxyphenol selectivities are not reported for 1 vs 1-d3 and
1-d5 vs 1-d8 because of overlap problems in the GC runs.
(43) Sawyer, D. T.; Sobkowiak, A.; Matsushita, T. Acc. Chem. Res. 1996,
29, 409-416.

11868 J. Am. Chem. Soc., Vol. 122, No. 48, 2000
Li and Jenks
Scheme 1
retention of 18O does not exceed 60%. Attack of HO• on the
ipso position accounts for loss of 18O if the adduct 9 loses the
elements of methanol. The formation of methanol in nearly equal
proportion to phenoxyl radicals has been observed in radiolysis
conditions.37 The final hydrogen atom may be scavenged either
directly from a source such as HOO• or by stepwise electron
and proton transfer.
two mechanisms that lead to phenol formation. As illustrated
in Scheme 1, we envision pathways that begin with attack at
the methyl and with ipso attack on the aromatic ring.
Entry 8 in Table 1 indicates that there are also two
mechanisms for formation of phenol when BQ is used as the
electron sink instead of O2. While path b does not invoke O2,
we have postulated its involvement in path a. Yet just over a
quarter of the phenol appears to be formed by attack at methyl,
based on the partial retention of 18O. There is no specific
evidence for the mechanism of 3-18O formation. However, it is
reasonable to hypothesize that 7 is oxidized to the cation, which
is highly susceptible to hydrolysis to phenol. The greater yield
of 3 under acidic conditions may be consistent with the greater
oxidizing power of BQH+, compared to BQ itself. Furthermore,
the 19% abundance of 18O in 3, combined with an estimate of
kH/kD for abstraction of the methyl hydrogens of 2.7 predicts
an observed isotope selectivity of 1.3 for formation of 3, in
reasonable agreement with the observed value.
Anomalous Isotope Effects for 4 and 5. On first inspection,
the H/D selectivity values for 3-5, particularly those for 4 and
5, are surprising. It is well-known that hydroxyl adds to aromatic
rings, and an inVerse secondary isotope effect is expected. In a
gas-phase study, no kinetic isotope effect for benzene and
benzene-d6 was observed near room temperature,53 though it is
unclear whether the precision of these experiments would allow
detection of an effect of just a few percent.54
We consider first path a), and suggest that phenyl formate
(2) as well as phenol derives from it. Consistent with the
formation of 2 beginning with attack at the methyl, there is no
significant isotope effect for its formation from 1 vs 1-d5 (entries
10 and 11 in Table 2). Moreover, when 1-18O is used as starting
material, the phenyl formate contains the same degree of
enrichment as the starting material. Thus, postulating that
formation of radical 7 is the initial step, trapping by O2 provides
the peroxyl radical 8. The formate is obtained by dispropor-
•
tionation or reaction with HO2 according to the Russell
mechanism.44,45 Consistent with this reaction sequence, 2 is not
observed when benzoquinone is used as an alternate electron
acceptor for O2 in the TiO2-mediated degradation (entries 7 and
8 in Table 1). The formation of 2 under the Fenton and H2O2
photolysis conditions without additional oxygen presumably
comes about because of the formation of O2 as a natural product
of the conditions,43,46 though other mechanisms cannot be ruled
out.
The present selectivity values, which are greater than unity,
require explanation. One reasonable rationalization is that there
is an isotope-dependent branching point in the mechanism of
product formation after the initial addition step. The isotope-
dependent branch leads to products unaccounted for by observa-
tion of 4 and 5 and affects the apparent selectivity.
The H/D selectivity for formation of 2 is approximately 2.7,
as shown by the various entries in Table 2. This selectivity is
taken to reflect the H/D kinetic isotope effect in formation of
radical 7.47 It is quite comparable for all three oxidation methods
used here and implies that the transition states bear qualitative
resemblance to one another (though there may be metal or
semiconductor association of the hydroxyl species). This value
of 2.7 is somewhat smaller than the 3.3 obtained for trimethyl
cyanurate in comparable experiments,27 but within the fairly
wide range of 2-5 expected from other reported studies.48-52
Having established the existence of path a in Scheme 1, we
consider path b. From Table 1, it can be seen that the 18O-
enrichment of phenol (3) varied, but was consistently below
the 72% enrichment of the starting material 1-18O. Even
including 2, in which 18O was fully retained, as “pre-3”, the
A related observation was made in radiolysis of benzene and
benzene-d6 in oxygenated water.55 The ratio of G-values
(efficiencies) of phenol formation from benzene and benzene-
d6 was 1.35. This was interpreted as outlined in Scheme 2.55
After formation of the initial adduct and trapping by O2 to give
11, the elimination of HOO• to give phenol has a (unmeasured)
primary isotope effect, but the pathway in which another oxygen
molecule adds to the cyclized species to degrade to other
compounds does not. (Two of the four possible isomers of 11
cannot undergo the elimination, but the addition of O2 is
reversible.56) Since the conversion of 11 to 3 is faster in the
proteo case than in the deuterio case, less of the 1H-11 is diverted
to the other products. This type of branching would distort the
competition experiments reported here just as it affects the
G-values in the radiolysis experiments. Dividing the H/D
(44) Russell, G. A. J. Am. Chem. Soc. 1957, 79, 3871-3877.
(45) Chuang, C.-C.; Chen, C.-C.; Lin, J.-L. J. Phys. Chem. B 1999, 103,
2439-2444.
(46) Walling, C. Acc. Chem. Res. 1975, 8, 125-131.
(47) It should also be noted that there may be pathways from 7 to 3 that
bypass 2. We have no specific evidence on this point.
(48) McCaulley, J. A.; Kelly, N.; Kaufman, F. J. Phys. Chem. 1989, 93,
1014-1018.
(49) Tully, F. P.; Droege, A. T. J. Phys. Chem. 1986, 90, 1949-1954.
(50) Tully, F. P.; Droege, A. T. J. Phys. Chem. 1986, 90, 5937-5941.
(51) Tully, F. P.; Droege, A. T. J. Phys. Chem. 1987, 91, 1222-1225.
(52) Tully, F. P.; Hess, W. P. J. Phys. Chem. 1989, 93, 1944-47.
(53) Lorenz, K.; Zellner, R. Ber. Bunsen-Ges. Phys. Chem. 1983, 87,
629-636.
(54) Above 420 K, an isotope effect was observed because hydrogen
abstraction from benzene becomes a kinetically important route.
(55) Pan, X.-M.; Schuchmann, M. N.; von Sonntag, C. J. Chem. Soc.,
Perkin Trans. 2 1993, 289-297.
(56) Pan, X.-M.; von Sonntag, C. A. Naturforsch. 1990, 45b, 1337-
1340.

Isotope Studies of Photocatalysis
J. Am. Chem. Soc., Vol. 122, No. 48, 2000 11869
Scheme 2
Scheme 3
a Compound 11 is representative of both ortho and para trapping
products.
selectivities of 3-5 in the oxygenated experiments (entries 1-5
and 7-11) by the factor of 1.35 would produce adjusted product
ratios in line with expectations for an inverse secondary isotope
effect.
The experiments using BQ instead of O2 as an electron sink
can be used to test this hypothesis. Noting that the isotope
selectivities we had observed for photocatalytic degradation in
the presence of oxygen for the formation of 4 and 5 were
roughly the same as the reported G(C6H6)/G(C6D6) ratio,55 the
experiments using BQ as a substitute for O2 were carried out
to eliminate the competition between an isotope-dependent and
isotope-independent reaction of the initial hydoxyl adduct of
anisole. Obviously, the mechanism for hydroxylation of anisole
must change from that shown in Scheme 2 because of the lack
of oxygen, but undoubtedly the first step is the same: formation
of the HO• adduct on the aryl ring. After adduct formation, we
presume that BQ oxidizes the radical to the cation, from which
the methoxyphenol is formed.
radiolysis literature.32,35,57,58 Protonation and dehydration of 12
gives the relatively stable radical cation 1•+, which recovers an
electron from other easily oxidizable species. This serves as an
inhibition of formation of the hydroxylation products because
it competes with the oxygenation route to hydroxylation
products. By contrast, acid-promoted decomposition of 9 (the
ipso-attack precursor to phenol formation) would likely lead
mainly to 10 because of the easier decomposition of 14 as
compared to 15. Thus, the ratio of 3-16O to 4 should increase
at low pH, as it does. At pH 1 (entry 1, Table 1), the quantity
of phenol attributable to ipso attack is approximately 10 times
greater than the sum of 4 and 5, compared to a ratio of about
1:4 at pH 7 (entry 4). It seems unreasonable to attribute this to
changing selectivity of the initial attack. However, it is also
clear that this scheme is a simplification of the entire suite of
reactions that occur. The falloff in 4 and 5 does not appear to
be linear with [H+]. Indeed, the overall rate of decomposition
of 1 slows approximately linearly with pH (not [H+]!).
Nonetheless these reactions provide a framework for under-
standing the observed selectivities and changes thereof.
The observed isotopic selectivities of 0.94 and 0.98 for
formation of 4 and 5 (entry 6, Table 2) now reflect the expected
inverse secondary isotope effect, or at worst, no isotope effect.
This result not only makes sense within our own system, but
supports the hypothesis of von Sonntag even under his distinctly
different conditions.55 It also gives us a basis for understanding
the slightly different selectivities for 4 and 5 under normal
photocatalytic conditions, i.e., that the competition between the
HOO• elimination and further oxygen trapping is slightly altered
by the position of the substituents.
Product Distributions. The predominance of ortho and para
hydroxylation over meta is readily understood from the electron
donating ability of the methoxy group and the electrophilicity
of HO•.33,34,37 The same predomination of ortho and para attack
should be observed over ipso, but phenol is observed in high
proportion at low pH. In fact, the overall pH dependence of the
product distribution is dramatic. Qualitatively similar pH trends
are observed for all three oxidation methods, in that products 2
and 3 predominate at low pH, but hydroxylation products are
more important at higher pH.
Conclusions
The gross product distributions in the photocatalytic degrada-
tion of anisole follow trends observed in both radiolysis and
Fenton conditions. This is taken to show that the major reaction
channel for anisole under photocatalytic degradation is hydroxyl-
like chemistry rather than electron transfer.
The use of 18O-labeled anisole (1-18O) has revealed that the
TiO2-mediated conversion of anisole to phenol occurs com-
petitively by two major classes of mechanisms involving ipso
attack and degradation of the methyl group. This implies that
photocatalytic degradation of related aromatic species may result
in the release of the alkoxy portion of aryl alkyl ethers before
complete degradation of the alkyl group. Similarly, both Fenton
and H2O2/hν show both modes of attack.
Scheme 3 shows a plausible explanation for the dramatic
change in (2 + 3)/(4 + 5 + 6) with pH. Acid-promoted
dehydration of adducts such as 12 has been proposed in the
(57) O’Neill, P.; Steenken, S.; Schulte-Frohilinde, D. J. Phys. Chem.
1975, 79, 2773-2779.
(58) Land, E. J.; Ebert, M. Trans. Faraday Soc. 1967, 63, 1181-1190.

11870 J. Am. Chem. Soc., Vol. 122, No. 48, 2000
Li and Jenks
Isotopic selectivities (H vs D) of about 2.7 were obtained
for the conversion of anisole to phenyl formate under photo-
catalytic, Fenton, and oxygenated H2O2 photolysis conditions.
This was attributed to a primary kinetic isotope effect on the
abstraction of a methyl hydrogen. A more complex explanation
is required for the selectivities observed for hydroxylation of
anisole. Any measurable kinetic isotope effect on addition of
hydroxyl to the aromatic ring ought to be less than 1.00; the
observation of selectivities of the order of 1.3 can be rationalized
if there is a branch point in the mechanism in which the
competing pathways have different kinetic isotope effects. The
one proposed here, and supported by experiments in which BQ
replaces O2 in photocatalytic degradation, pits an isotope-
dependent loss of HOO• against isotope independent trapping
of the same intermediate with a second molecule of O2.
Acknowledgment. The financial support of the NSF in the
form of a CAREER grant is gratefully acknowledged.
Supporting Information Available: A table of degradation
rates vs pH, a table showing all the concentration ratios used
and reaction times for the isotopic selectivities, and an explana-
tion of how the anticipated H/D selectivity of 1.3 in the TiO2
degradation with BQ is obtained (PDF). This material is
available free of charge via the Internet at http://pubs.acs.org.
JA994030H