TiO2-photocatalyzed reactions of some benzylic donors

Article abstract of DOI:10.1139/v03-048

TiO₂-photocatalyzed oxidation of toluene (1a), benzyltrimethylsilane (1b), and 4-methoxybenzyltrimethylsilane (1c) has been carried out in acetonitrile under oxygen, under nitrogen, and in the presence of electrophilic alkenes under various conditions (using Ag₂SO ₄ as electron acceptor, adding 2.5% H₂O, changing solvent to CH₂Cl₂). Benzyl radicals, formed via electron transfer and fragmentation, are trapped. A good material balance is often obtained. The overall efficiency of the process depends on the donor E_ox, on the rate of fragmentation of the radical cation, and on the acceptor present (Ag⁺ is an efficient oxidant, an electrophilic alkene a poor one, O₂ is intermediate). With ring-unsubstituted benzyl derivatives 1a and 1b, oxidative fragmentation occurs mainly close to the catalyst surface. The benzyl radicals form at a high local concentration and give benzaldehyde under O₂, bibenzyl under N₂ and dibenzylated derivatives by attack on the alkenes (acrylonitrile, fumaronitrile, maleic acid). In this case, using CH₂Cl₂-O₂ enhances the yield of benzaldehyde. With methoxylated 1c, however, the radical cation migrates into the solution before fragmentation and, therefore, the free benzyl radical is formed. This radical in part is oxidized to the cation, giving a considerable amount of benzylacetamide (or of the alcohol with water), and in part is trapped by the alkenes. The last reaction is less efficient in this case and yields monobenzyl derivatives.

Full text of DOI:10.1139/v03-048

560
TiO2-photocatalyzed reactions of some benzylic
donors
Laura Cermenati, Maurizio Fagnoni, and Angelo Albini
Abstract: TiO2-photocatalyzed oxidation of toluene (1a), benzyltrimethylsilane (1b), and 4-
methoxybenzyltrimethylsilane (1c) has been carried out in acetonitrile under oxygen, under nitrogen, and in the pres-
ence of electrophilic alkenes under various conditions (using Ag2SO4 as electron acceptor, adding 2.5% H2O, changing
solvent to CH2Cl2). Benzyl radicals, formed via electron transfer and fragmentation, are trapped. A good material bal-
ance is often obtained. The overall efficiency of the process depends on the donor Eox, on the rate of fragmentation of
the radical cation, and on the acceptor present (Ag+ is an efficient oxidant, an electrophilic alkene a poor one, O2 is in-
termediate). With ring-unsubstituted benzyl derivatives 1a and 1b, oxidative fragmentation occurs mainly close to the
catalyst surface. The benzyl radicals form at a high local concentration and give benzaldehyde under O2, bibenzyl un-
der N2 and dibenzylated derivatives by attack on the alkenes (acrylonitrile, fumaronitrile, maleic acid). In this case, us-
ing CH2Cl2–O2 enhances the yield of benzaldehyde. With methoxylated 1c, however, the radical cation migrates into
the solution before fragmentation and, therefore, the free benzyl radical is formed. This radical in part is oxidized to
the cation, giving a considerable amount of benzylacetamide (or of the alcohol with water), and in part is trapped by
the alkenes. The last reaction is less efficient in this case and yields monobenzyl derivatives.
Key words: photocatalysis, oxidation, alkylation.
Résumé : On a effectué l’oxydation photocatalysée du toluène (1a), du benzyltriméthylsilane (1b) et du 4-méthoxy-
benzyltriméthylsilane (1c) par le TiO2 dans l’acétonitrile, sous atmosphère d’azote ou d’oxygène, et en présence
d’alcènes électrophiles et sous diverses conditions (faisant appel au Ag2SO4 comme accepteur d’électron, avec addition
de 2,5% d’eau et en changeant le solvant pour le CH2Cl2). On a piégé les radicaux benzyles qui se forment par trans-
fert d’électron et fragmentation. On a souvent obtenu une bonne balance pour les produits ayant réagi. L’efficacité glo-
bale du processus dépend du Eox du donneur, de la vitesse de fragmentation du cation radical et de la présence d’un
accepteur (Ag+ est un oxydant efficace, un alcène électrophile est mauvais alors que O2 est intermédiaire). Avec les dé-
rivés 1a et 1b comportant un noyau benzyle non substitué, la fragmentation oxydante se produit relativement près de la
surface du catalyseur. Les radicaux benzyles se forment à une concentration locale élevée et conduisent à la formation
de benzaldéhyde en présence de O2, en bibenzyle sous N2 et en dérivés dibenzylés par attaque sur les alcènes (acrylo-
nitrile, fumaronitrile, acide maléique). Dans ce cas, l’utilisation de CH2Cl2–O2 conduit à une augmentation du rende-
ment en benzaldéhyde. Toutefois, avec le dérivé méthoxy 1c, le cation radical migre en solution avant la fragmentation
et on observe la formation d’un radical libre. Ce radical est partiellement oxydé en cation conduisant à la formation de
benzaldéhyde et à une quantité considérable de benzylacétamide (ou d’alcool avec l’eau); il est aussi piégé par les al-
cènes. La dernière réaction est moins efficace dans ce cas et elle conduit aux dérivés monobenzyliques.
Mots clés : photocatalyse, oxydation, alkylation.
[Traduit par la Rédaction] Cermenati et al. 566
Introduction
superoxide anion. The alternative possibility that the organic
substrate is directly oxidized at the semiconductor surface is
of interest in two respects. First, because it has been shown
that, at least with reasonable donors, this mechanism partici-
pates in the TiO2 photocatalytic oxidation in water (2) and
second, because, particularly in nonaqueous solvents, such a
path may lead to the controlled oxidation of the substrate,
rather than to the complete oxidation to CO2 that is the tar-
get when photocatalysis is used for the recovery of polluted
Titanium dioxide is extensively used for the photoinduced
mineralization of organic contaminants in polluted water (1).
Under these conditions, photoinduced charge separation is
followed by hole and electron transfer to water and oxygen,
respectively, and the reaction of the organic molecule mainly
involves hydrogen abstraction by the thus formed hydroxyl
radicals; the organic radicals then react with oxygen or the
Received 7 January 2003. Published on the NRC Research Press Web site at http://canjchem.nrc.ca on 23 May 2003.
Dedicated to Professor Don R. Arnold, the pioneer of electron transfer photochemistry.
L. Cermenati, M. Fagnoni, and A. Albini.1 Dep. Organic Chemistry, University of Pavia, v. Taramelli 10, 27100 Pavia, Italy.
1Corresponding author (e-mail: albini@chifis.unipv.it).
Can. J. Chem. 81: 560–566 (2003)
doi: 10.1139/V03-048
© 2003 NRC Canada

Cermenati et al.
561
water. With reference to the second aspect, a number of pos-
itive results have been obtained. As an example, TiO2-
photocatalyzed oxygenation of cyclohexane neat or mixed
with organic solvents gives useful yields of products from
the intermediate oxidation, e.g., of cyclohexanone (3, 4).
Substituting a different electron acceptor for oxygen, e.g.,
the silver cation, adds flexibility (5). In this case different
radical reactions, e.g., radical dimerization, occur in nitro-
gen-flushed solution while controlled oxygenation takes
place in the presence of oxygen. Furthermore, the acceptor
can be an organic molecule, e.g., an electrophilic alkene,
which functions also as a radical trap and is alkylated (6).
More generally, photocatalysis in organic solvents may have
some synthetic interest (7), as opposed to the destructive
mineralization occurring in water.
Both controlled oxygenation and alkylation reactions have
been carried out successfully with benzylic derivatives. This
is not surprising because the pioneering work by Arnold has
shown that these compounds are both susceptible to photo-
induced single electron oxidation and prone to fragmentation
at the radical cation stage (8). In particular, benzylic oxygen-
ation and benzylic coupling in acetonitrile and in aqueous
media by TiO2–Ag+ have been described by Baciocchi and
Sebastiani and co-workers (5, 9) in several reports. We were
curious to compare such a reaction with the TiO2-photo-
catalyzed alkylation of alkenes (6, 10) through a systematic
examination. We surmised that this comparison may give in-
formation about the mode of reaction, in particular with re-
gard to the formation of the radicals either as solvated
species or close to the semiconductor surface, a difference
that is important for synthetic applications.
same time did not increase the rate of conversion, but en-
hanced the yield of the aldehyde (35–45%), accompanied, in
the latter case, by some benzyl alcohol. Carrying out the
photoreaction in oxygen-flushed dichloromethane rather
than in MeCN caused more extensive oxidation leading to
the aldehyde (59% at a 42% conversion of 1a), but also to a
large amount of benzoic acid (4a, 29%).
No appreciable reaction took place in nitrogen-flushed
acetonitrile (not reported), but it did in the presence of silver
sulfate, where the main product was bibenzyl (5a, 65% for a
16% conversion) accompanied by minor amounts of alde-
hyde 2a and nitrile 6a, as well as traces of 3a and of
benzylacetamide (7a). No significant reaction occurred in
the presence of 2.5% water in MeCN nor upon irradiation in
nitrogen-flushed CH2Cl2 (not reported; the other substrates
examined likewise did no react under these conditions).
Compared with 1a, benzyltrimethylsilane (1b) was more
easily photooxidized in MeCN–O2 (25% conversion, 66%
aldehyde, compare entries 2 and 1). Adding water had little
effect (except for the formation of a small amount of propio-
nitrile), while adding Ag2SO4 did not increase the rate of ox-
idation but improved selectivity for the formation of
benzaldehyde (92%, entry 8). In dichloromethane the reac-
tion was faster than in MeCN, with a similar product distri-
bution.
Again, no reaction took place under nitrogen in MeCN,
MeCN – 2.5% H2O or CH2Cl2. However, 1b was quite reac-
tive in deoxygenated MeCN in the presence of Ag+. The
conversion was larger than under oxygen (64 rather than
25%, entry 17) and bibenzyl was the main product (70%).
Also formed were small amounts of alcohol, aldehyde, and
propionitrile, as well as compounds resulting, as indicated
by mass spectrometry, from the benzylation of bibenzyl and
the starting benzylsilane. The experiment with both water
and Ag+ gave somewhat less of bibenzyl and more of benzyl
alcohol.
Therefore, we chose a small group of benzyl derivatives
and compared the effect of some key experimental parame-
ters as well as the effect of trapping by electrophilic alkenes.
As described in the following, this examination indeed re-
vealed new facets and demonstrated the versatility of the
system.
With 4-methoxybenzylsilane (1c) the rate of photooxi-
dation in MeCN further increased (85% conversion) and the
aldehyde (2c, 77%) was accompanied by a significant
amount of the alcohol (3c, 6%), which increased in the pres-
ence of water (12%, entry 6). The presence of the silver salt
made the reaction more selective for the aldehyde (83%, en-
try 9). This time, in dichloromethane the oxidation was
slower than in MeCN (55% rather than 85%, entry 15) and
led to both aldehyde and alcohol. Compound 1c was reactive
under nitrogen, though somewhat slower than under oxygen,
provided, as with the previous donors, that Ag+ was present;
the main product was the bibenzyl derivative (5c, 55%), but
4-methoxybenzylacetamide (7c, 25%) was also important
(entry 18). In the presence of both water and Ag+, 4-meth-
oxybenzyl alcohol (3c) was the main product (36%, entry
21). The solution became acidic during the irradiation and
was neutralized with triethylamine before work-up. If this
precaution was omitted, further benzylation reactions via the
benzylic cation formed from the alcohol led to the formation
of small amount of diphenylmethanes, a fact that has been
previously noted (9a), but is an artifact due to the acidity de-
veloped (compare to 11). Small amounts (0.5 to 1%) of 4-
methoxytoluene were detected in all the photocatalytic ex-
periments with 1c. The photooxidation of 1c was tested at a
somewhat large scale (0.78 g) in an immersion well appara-
Results and discussion
Photocatalysis of benzylic derivatives
Three benzylic derivatives were examined under standard
conditions (0.02 M solutions in, 3.5 mg mL–1 Degussa TiO2,
3 h irradiation) in tightly stoppered vessels, after flushing ei-
ther with oxygen or with nitrogen. Scheme 1 shows the
products formed and Table 1 reports the % conversion of
starting material and the % yield (calculated on the con-
verted substrate) of products formed.
The effects of adding silver sulfate and of a small percent-
age of water on the reaction in acetonitrile as well as of
changing the solvent to dichloromethane are compared. For
each condition, the results with the three substrates are listed
in sequence. No significant reaction took place over this
time when TiO2 was omitted.
Toluene (1a) reacted sluggishly in oxygen-flushed aceto-
nitrile and gave only traces of benzaldehyde 2a. Both con-
version and yield of the aldehyde increased in the presence
of 2.5% water, and further some benzyl alcohol (3a) and
phenylpropionitrile (6a) were formed. Adding 0.01 M
Ag2SO4 or having both this salt and water present at the
© 2003 NRC Canada

562
Can. J. Chem. Vol. 81, 2003
Scheme 1.
Table 1. TiO2 photocatalytic oxidation of benzylic donors.a
Conversion (%)
Products (%)b
Donor
Conditions
MeCN, O2
1
2
3
4
5
6
7
8
25
85
20
31
76
10
25
90
11
32
75
42
57
55
16
64
66
—
59
56
1
2
3
4
5
6
7
8
1a
1b
1cc
1a
1b
1c
1a
1b
1c
1a
1b
1c
1a
1b
1c
1a
1b
1cf
1a
1bf
1c
12
66
77
35
66
72
35
92
83
45
86
80
59
63
55
8
3
6
5
9
16
3
O2, H2Od
4
1
11
4
12
O2, Ag+e
4
2
3
3
9
10
11
12
13
14
15
16
17
18
19
20
21
O2, H2O, Ag+
CH2Cl2, O2
MeCN, N2, Ag+
N2, H2O, Ag+
11
8
6
3
2
2
5
36
1
29
1
1
65
70
53
12
tr
2
6
1
tr
25
6
2
1
8
58
7
6
36
1
aUpon irradiation of 0.02 M suspensions of the benzylic derivatives 1 (140 mg TiO2, 40 mL acetonitrile) for 3 h.
b% Yields on converted 1.
cSmall amounts (0.5–1%) of 4-methoxytoluene formed in all experiments with 1c.
dH2O (2.5%).
eAg2SO4 (0.01 M).
fSmall amounts (up to 3%) of (4-methoxybenzylbenzyl)trimethylsilane and 4-methoxybenzyltoluenes revealed by GC–MS.
tus and it gave 90% conversion (of which 91% was the alde-
hyde) after 8 h irradiation.
electrofugal cation Me3Si+ from the radical cation. These
compounds react much faster than 1a and give consistently a
good material balance. Fragmentation is no longer the rate-
limiting step. With these highly reactive radical cations, it
can be appreciated that Ag+ is a superior oxidant to O2 (with
1b, an increase in the rate of reaction by a factor of 2.5 in
MeCN/Ag+ in comparison with the experiments under oxy-
gen).
Fragmentation leads to benzyl radicals (Scheme 2). In ni-
trogen-flushed solution, these mainly dimerize to bibenzyls
5a, 5c. A minor process is hydrogen abstraction from the
solvent, which generates ·CH2CN radicals revealed by the
presence of arylpropionitriles 6a, 6c and is consistent with
the formation of a small amount of methylanisole from 1c.
The ratio of nitrile (6) – bibenzyl (5) is larger with toluene
(0.2) than with silane 1b (<0.02 in neat MeCN, though it
grows to 0.1 with 2.5% H2O), consistently with the fast
desilylation and thus the higher local concentration of radi-
cals in the latter case.
These results clearly support the role of radical cations
formed by hole transfer at the surface of the semiconductor
in the photooxidation of benzylic substrates (see Scheme 2),
adding to previous evidence by Baciocchi and Sebastiani
and co-workers (5, 9), showing that Ag+ can be a substitute
for oxygen in the role of electron acceptor. Favorable condi-
tions for the process are that the reagent is reasonably easily
oxidized and bears a good electrofugal group that makes
fragmentation of the radical cation competitive with back
electron transfer to the semiconductor. Thus, a moderate do-
nor such as toluene (Eox 2.2 V vs. SCE) with a poor leaving
group is slowly oxidized; in MeCN, cleavage of a proton
from the toluene radical cation (1a·+) is more than 100 times
slower than cleavage of a trimethylsilyl cation from 1b.+
(12). Introduction of a silyl group in α and of a methoxy
group on the ring both make the reagents better donors such
as 1b (Eox 1.78 V) and 1c (Eox 1.31V) (13) and offer a good
© 2003 NRC Canada

Cermenati et al.
563
Scheme 2.
Scheme 3.
Also with 1c only traces of nitrile are formed, but there
another reaction competes. This is oxidation of the benzyl
radical to the cation (Scheme 2). This process is revealed
both by addition to acetonitrile, resulting after reaction with
adventitious water, in the formation of amide 7c, and, in the
presence of 2.5% water, in the formation of alcohol 3c. As
previously proposed (9), the radical is oxidized by silver cat-
ions (0.4 V in MeCN) (14). Accordingly, from 1c the yield
of amide is 25%, and with the better nucleophile H2O the
sum of the alcohol + amide is 42%, because oxidation of the
4-methoxybenzyl radical, with Eox 0.23 V (15), is fast. In
contrast, no amide is formed from 1b and only traces from
1a, since oxidation of the unsubstituted benzyl radical (Eox
0.73 V) (15) is slow.
In the presence of oxygen, benzyl radicals are trapped by
oxygen or the superoxide anion formed in the initial electron
transfer (Scheme 2). Decomposition of the hydroperoxides
or of the peroxyl radicals leads to the aldehydes (Ag+ when
present makes the decomposition faster). In the case of 1c,
the presence of the hydroperoxide was revealed by an en-
hancement of the yield of alcohol 3c (by a factor of 2) upon
treatment of the irradiated solution by triphenylphosphine.
When this process is efficient, secondary oxidation to ben-
zoic acids is less important. Thus the aldehyde:acid ratio
grows when the SET path is favored by the electron donat-
ing characteristics of the reagent (the ratio is 4 with 1b, 25
with 1c) or by the use of a better acceptor, such as Ag+,
which allows better transient charge separation at the semi-
conductor surface (the ratio is ca. 30 for both 1b and 1c).
Water is oxidized competitively with the aromatic substrate
in the case of poorly reacting toluene and the increased yield
of aldehyde and the formation of some propionitrile (en-
try 4) are indicative of the participation of OH· radicals.
However, competitive oxidation of water does not occur in
the case of silanes 1b and 1c. With these compounds, the re-
sult is not affected by the presence of 2.5% water, except for
the formation of more alcohol with the latter one, another in-
dication of the increased role of the benzyl cation in this
case.
Using dichloromethane in the place of acetonitrile obliter-
ates the difference in the rate of oxidation in the series 1a–
1c, enhancing the rate of the first two substrates and slightly
lowering the rate for the last one. This remarkable effect can
again be rationalized on the basis of the electron transfer
mechanism for the fragmentation. A less polar solvent in-
creases absorption of the substrate on the surface, thus facili-
tating oxidation. Furthermore, superoxide anion is expected
to be a stronger base in less polar CH2Cl2 than in MeCN
(compare to 16), and assists deprotonation of the radical cat-
ion (and apparently also desilylation, since some alcohol is
formed from the silanes), adding to the action of basic cen-
tres on TiO2. This offers an alternative mechanism of reac-
tion (path b in Scheme 3) for substrates for which
fragmentation of the free radical cation is unimportant (see
below).
Photocatalysis of benzylic derivatives in the presence of
electrophilic alkenes
Previous work had shown that TiO2 photocatalysis in the
presence of electrophilic alkenes leads to efficient alkylation
(6, 10). In that case, the alkene functions both as electron ac-
© 2003 NRC Canada

564
Can. J. Chem. Vol. 81, 2003
Scheme 4.
Table 2. TiO2 photocatalytic reactions of benzylic donors in the presence of alkenes 8 and 0.01 M silver sulfate.a
Conversion (%)
Products (%)b
Donor
Alkene
1
2
5
7
9
10
5
5
15
28
33
3
1
2
3
4
5
6
7
8
1b
1a
1b
1a
1b
1c
1b
1c
8x (0.02 M)
8x (0.1 M)
30
9
3
7
4
4
6
60
33
39
30
43
13
34
10
2
3
65
20
67
66
60
60
8y (0.02 M)
8z (0.02 M)
32
33
8
2
8
3
34
aUpon irradiation of 0.02 M suspensions of the benzylic derivatives 1 (140 mg TiO2, 40 mL acetonitrile) for 3 h.
b% Yield on converted 1.
ceptor and as radical trap, and we were curious to compare
the present system with Ag+ as the oxidant.
As compared with our previous data on the benzylation
reaction in the absence of the silver salt (6), a large differ-
ence in rate and product distribution is now observed. In the
absence of Ag+, the reaction of 1b with e.g., 8z required
20 h for a 15% conversion and gave 7% bibenzyl, 80%
monoadduct, and 5% bisadduct, rather than 60% conversion
in 3 h and 34, 2, and 34%, respectively, of the products in
the present experiment. In the presence of the alkene alone,
the conversion of 1b reached 35% after 60 h and that of 1c
100% after 30 h; it was 10–50 times slower than in the pres-
ence of Ag+ and the monoadducts were by far the major
products (50–80%). No useful result was obtained from 1a.
The difference is due to the double role of 1,2-electron-
withdrawing substituted alkenes such as 8y and 8z. These
act both as electron acceptors (8y, Ered –1.63 V vs. Ag/Ag+)
(17) and as radical traps. In the absence of other acceptors,
the alkene radical anions build up and either directly trap the
radicals or reduce the radical adducts, thus leading to the
monobenzylated alkenes (path a, a′, Scheme 4).
As shown in Table 2 and Scheme 4, with TiO2–Ag+ the
percentage conversion of donors 1a–1c was similar to that
observed in the absence of the alkene, but even a poor trap
such as acrylonitrile (8x, 0.1 M) led to the formation of a
certain amount of adduct. As an example, from 1b 15% of
the bibenzylated propionitrile 10ax was formed at the ex-
pense of bibenzyl (dropping from 70% to 39%, entry 3). A
lower amount of 10ax (5%) was formed from 1b with a
lower trap concentration (0.02 M 8x, entry 1) as well as
from a poorer radical precursor such as toluene and 0.1 M
8x (entry 2). However, a better trap such as fumaronitrile
(8y) was effective at the 0.02 M concentration with both 1a
and 1b, making the yield of bibenzylated 10ay (28–33%,
mixture of the two diastereoisomers) comparable to that of
bibenzyl (30–43%, entries 4, 5). Bibenzylation was obtained
in a similar yield (34%) also from 1b and maleic acid (8z,
entry 7).
The result was different with the methoxybenzylsilane 1c.
In this case, the yield of bibenzyl was much reduced when
the alkenes 8 were present (while the yield of amide 7c, if
anything, slightly increased), but trapping by the alkene was
inefficient, giving a small amount of the mono derivatives
9cy, 9cz, and with 8y only traces of bisadduct 10cy (entries
6, 8).
With the silver cation the two functions are separated, the
inorganic cation being the main oxidant and the alkene func-
tioning as radical trap because Ag+ is a better acceptor. Un-
der these conditions photoinduced fragmentation of the
donors is much faster with 1b and 1c, and even with 1a
some reaction occurs, while none was observed in the ab-
sence of Ag+. Furthermore, the radical anion 8.– does not
© 2003 NRC Canada

Cermenati et al.
565
build up in this case, and the relatively stable (α-cyano or
carboxyl) adduct radicals have no easy path for reduction.
Thus, these couple with a benzyl radical to give the
bisadduct (path b, Scheme 4, see below). This path requires
a high concentration of benzyl radicals, a situation that ap-
plies in the vicinity of the photocatalyst surface with donors
1a, 1b. The nonstabilized radical cations of these substrates
do not migrate and fragment (with different efficiency) close
to the surface (path a in Scheme 3), generating a high local
concentration of the radicals.
cosolvent is competitively oxidized. This process is faster in
CH2Cl2, probably via a O2 assisted reaction (path b).
.–
Experimental
Reagents
Degussa P25 titanium dioxide was dried in a oven at
110°C for 24 h and acetonitrile was distilled over calcium
hydride and stored over molecular sieves. The chemicals
were commercially available or prepared through conven-
tional procedures for the case of 1c.
On the contrary, migration of the longer lived stabilized
radical cation from 1c precedes fragmentation; thus, the free
benzyl radical is formed from the solvated radical cation
(path c in Scheme 3). The solvated radical is in part oxidized
by Ag+ (soluble in MeCN), and thus, is not available for
trapping. A part is trapped, as revealed by the strong de-
crease of bibenzyl in the presence of alkenes, but the radical
adduct has little opportunity both for coupling with benzyl
radicals (because of the low concentration in solution) and
for reduction (since no reducing 8.– builds up). Presumably
higher molecular weight products are formed and the yield
of isolated products (monoadducts 9 and traces of bisadducts
10) is far from balancing the amount of radicals trapped.
Thus, trapping by alkenes gives an indication about the
mode of α-cleavage, in solution or close to the surface. On
the other hand, the photoinduced reaction becomes mush
faster with Ag+, but the yield of the useful process,
alkylation of the electron-withrawing substituted alkene is
decreased by the competing oxidation of the radical, and
from the synthetic point of view the reaction in the absence
of silver salts (6) is preferable.
Irradiations
Irradiations were carried out on magnetically stirred solu-
tions contained in 4.5 cm diameter serum-capped Pyrex cy-
lindrical reactors. Suspensions (40 mL) containing the
benzylic derivatives (0.02 M), the appropriate additives (see
Tables 1 and 2) and 140 mg TiO2 were used. These were
sonnicated for 1 min and then stirred and flushed with either
oxygen or purified nitrogen (through needles) in the dark for
20 min. The tubes were put in the centre of four 15 W phos-
phor-coated lamps (centre of emission, 360 nm) and irradi-
ated for 3 h while maintaining slow gas flow and stirring.
The suspension was then filtered over a 0.2 µm porosity fil-
ter under vacuum and either directly analyzed or rotary
evaporated and taken up to a fixed volume.
The products were identified by comparison of their GC–
MS spectra with authentic samples (compounds 9cy, 9az,
9cz and diastereoisomeric 10ax, 10ay, 10cy, 10az were pre-
viously reported) (6) and quantified by GC on the basis of
calibration curves using dodecane or cyclododecane as the
internal standard. Gas chromatography analyses were carried
out by using an HP 5890 apparatus with a 0.3 mm × 30 m
capillary column with a flame ionization detector. Gas chro-
matography – mass spectrometry determination was per-
formed using an HP 5970B instrument operating at a
ionizing voltage of 70 eV, connected to an HP 5890 instru-
ment equipped with the same column as above.
Conclusions
The TiO2 photocatalytic oxidation of benzylic donors
gives products resulting from the trapping of benzyl radicals
(benzaldehydes, bibenzyl or alkylated derivatives), some-
times with satisfactory material balance. The efficiency of
the electron transfer oxidation of these substrates depends on
the oxidation potential of the donor, on the presence of an
electrofugal group in the α position, and on the presence of a
good electron acceptor that allows exploitation of the tran-
sient charge separation on the semiconductor surface.
Among these, Ag+ is more efficient than O2, and both of
these are much better than an electrophilic alkene. On the
other hand, the presence of a strong oxidant such as silver
sulfate causes partial further oxidation of the radical to the
cation and limits the benzylation of electrophilic alkenes
since it hinders a required step, reduction of the adduct radi-
cal.
Comparing photocatalysis for the substrates 1a–1c sug-
gests that the reaction occurs via two main mechanisms. The
stabilized radical cation 1c.+ is solvated and fragments in so-
lution (path c, Scheme 3). The free radical formed can be
trapped. Further oxidation of the radical to the cation also
has an important role. The shorter lived, less stable ana-
logues 1a.+, 1b.+ do not migrate and mainly fragment close
to the surface of the catalyst (path a), e.g., yielding bibenzyl
or bibenzylation of added electrophilic alkenes.
Deprotonation of 1a.+ is slow, so that with 2.5% water the
A larger scale irradiation was carried out with 1c (0.78 g,
0.02 M) in a suspension of TiO2 (0.7 g) in MeCN (200 mL)
in an immersion well apparatus fitted with a Pyrex-filtered,
water-cooled medium pressure mercury arc (150 W). The
suspension was magnetically stirred and flushed with argon
during irradiation (8h).
Acknowledgement
Partial support of this work by MIUR (Ministerio
dell’Istruzione Universitaria e Ricerca), Rome, is gratefully
acknowledged.
References
1. (a) A. Mills and S. Le Hunte. J. Photochem. Photobiol. A.
108, 1 (1997); (b) P. Pichat. In Handbook of heterogeneous ca-
talysis. Edited by G. Ertl, H. Knoezinger, and J. Weitkamp.
Academic Press, New York. 1997. p. 2111; (c) D. Bahnemann,
J. Cunningham, M.A. Fox, E. Pelizzetti, P. Pichat, and N.
Serpone. In Aquatic and surface photochemistry. Edited by
G.R. Helz, R.G. Zepp, and D.G. Crosby. Lewis Publ., Orlando.
1994. p. 261; (d) M.R. Hoffmann, S.T. Martin, W. Choi, and
D.W. Bahnemann. Chem. Rev. 93, 671 (1993); (e) O. Legrini,
© 2003 NRC Canada

566
Can. J. Chem. Vol. 81, 2003
E. Oliveros, and A.M. Braun. Chem. Rev. 93, 671 (1993);
(f) P. Pichat. Catal. Today, 19, 313 (1994); (g) N. Serpone. In
Kirk-Othmer encyclopedia of chemical technology. Vol. 18.
Wiley, New York. 1996. p. 820; (h) M. Sturini, F. Soana, and
A. Albini. Tetrahedron, 58, 2943 (2002).
9. (a) E. Baciocchi, C. Rol, G.V. Sebastiani, and L. Taglieri. J.
Org. Chem. 59, 5272 (1994); (b) E. Baciocchi, M. Bietti, M.I.
Ferrero, C. Rol, and G.V. Sebastiani. Acta Chem. Scand. 52,
160 (1998); (c) E. Baciocchi, G.C. Rosato, C. Rol, and G.V.
Sebastiani. Tetrahedron Lett. 33, 5437 (1992); (d) L. Amori, T.
Del Giacco, C. Rol, and G.V. Sebastastiani. J. Chem. Res. 644
(1998).
10. (a) L. Cermenati, C. Richter, and A. Albini. Chem. Commun.
85 (1998); (b) L. Cermenati and A. Albini. J. Adv. Oxid.
Techol. 5, 58 (2002).
2. (a) L. Cermenati, P. Pichat, C. Guillard, and A. Albini. J. Phys.
Chem. 101, 2650 (1998); (b) M. Ranchella, C. Rol, and G.V.
Sebastiani. J. Chem. Soc., Perkin Trans. 2, 311 (2000); (c) F.
Soana, M. Sturini, L. Cermenati, and A. Albini. J. Chem. Soc.,
Perkin Trans. 2, 699 (2000).
3. P. Boarini, V. Carassiti, A. Maldotti, and R. Amadelli.
Langmuir, 14, 2081 (1998).
4. C.B. Almquist and P. Biswas. Appl. Catal. A, 214, 252 (2001).
5. E. Baciocchi, C. Rol, G.C. Rosato, and V. Sebastiani. J. Chem.
Soc., Chem. Commun. 59 (1992).
11. (a) E.F. Pratt and H.J.E. Seagrave. J. Am. Chem. Soc. 81, 5369
(1981); (b) K.A. Rutheford, O.A. Mamer, J.M. Prokipcak, and
R.A. Jobin. Can. J. Chem. 44, 2337 (1966).
12. (a) M. Freccero, A.C. Pratt, A. Albini, and C. Long. J. Am.
Chem. Soc. 120, 284 (1998); (b) N. d’Alessandro, A. Albini,
and P. S. Mariano. J. Org. Chem. 58, 937 (1993).
13. J.P. Dinnocenzo, S. Farid, J.L. Goodman, I.R. Gould, W.P.
Todd, and S.L. Mattes. J. Am. Chem. Soc. 111, 8973 (1989).
14. C.K. Mann and K.K. Barnes. Electrochemical reactions in
nonaqueous systems. M. Dekker, New York. 1970. p. 479.
15. D.D.M. Wayner, B.A. Sim, and J.J. Dannnenberg. J. Org.
Chem. 56, 4853 (1991).
16. (a) C.P. Andrieux, P. Hapiot, and J.M. Savéant. J. Am. Chem.
Soc. 109, 3768 (1987); (b) D.T. Sawyer and M.J. Gibian. Tet-
rahedron, 35, 1471 (1979).
6. L. Cermenati, M. Mella, and A. Albini. Tetrahedron, 54, 2575
(1998).
7. (a) H. Kisch and M. Hopfner. In Electron transfer in chemis-
try. Vol. 4. Edited by V. Balzani, Wiley-VCH, Weinheim.
2001. p. 232; (b) H. Kisch. Adv. Photochem. 26, 93 (2001);
(c) M.A. Fox and M.T. Dulay. Chem. Rev. 93, 341 (1993);
(d) A. Maldotti, A. Molinari, and R. Amadelli. Chem. Rev.
102, 3811 (2002).
8. (a) A.M.P. Nicholas and D.R. Arnold. Can. J. Chem. 60, 2165
(1982); (b) X. Du, D.R. Arnold, R.J. Boyd, and Z. Shi. Can. J.
Chem. 69, 1365 (1991).
17. D.R. Arnold and P.C. Wong. Can. J. Chem. 57, 1037 (1979).
© 2003 NRC Canada