EPR investigation of persistent radicals produced from the photolysis of dibenzyl ketones adsorbed on ZSM-5 zeolites

Article abstract of DOI:10.1021/jo011047l

Photolysis of ketones (1, 1-oMe, 2, 2-oMe, 3, and 4) adsorbed on ZSM-5 zeolites produces persistent carbon-centered radicals that can be readily observed by conventional steady-state EPR spectroscopy. The radicals are persistent for time periods of seconds to many hours depending on the supramolecular structure of the initial radical@zeolite complex and the diffusion and reaction dynamics of radicals produced by photolysis. The structures of the persistent radicals responsible for the observed EPR spectra are determined by a combination of alternate methods of generation of the same radical, by deuterium substitution, and by spectral simulation. A clear requirement for persistence is that the radicals produced by photolysis must either separate and diffuse from the external to the internal surface or be generated within the internal surface and separate and diffuse apart. The persistence of radicals located on the internal surface is the result of inhibition of radical-radical reactions. Radicals that are produced on the external surface and whose molecular structure prevents diffusion into the internal surface are transient because radical-radical reactions occur rapidly on the external surface. The reactions of the persistent radicals with oxygen and nitric oxide were directly studied in situ by EPR analysis. In the case of reaction with oxygen, persistent peroxy radicals are formed in high yield. The addition of nitric oxide scavenges persistent radicals and leads initially to a diamagnetic nitroso compound, which is transformed into a persistent nitroxide radical by further photolysis. The influence of variation of radical structure on transience/persistence is discussed and correlated with supramolecular structure and reactivity of the radicals and their parent ketones.

Full text of DOI:10.1021/jo011047l

2606
J . Org. Chem. 2002, 67, 2606-2618
EP R In vestiga tion of P er sisten t Ra d ica ls P r od u ced fr om th e
P h otolysis of Diben zyl Keton es Ad sor bed on ZSM-5 Zeolites
Nicholas J . Turro,*,† Xue-gong Lei,† Steffen J ockusch,† Wei Li,† Zhiqiang Liu,†
Lloyd Abrams,§ and M. Francesca Ottaviani‡
Department of Chemistry, Columbia University, New York, New York 10027,
Institute of Chemical Sciences, University of Urbino, Piazza Rinascimento, 6 Urbino, Italy 61029,
and E. I. Dupont, Central Research, Experimental Station, Wilmington, Delaware 19880
turro@chem.columbia.edu
Received November 5, 2001
Photolysis of ketones (1, 1-oMe, 2, 2-oMe, 3, and 4) adsorbed on ZSM-5 zeolites produces persistent
carbon-centered radicals that can be readily observed by conventional steady-state EPR spectroscopy.
The radicals are persistent for time periods of seconds to many hours depending on the
supramolecular structure of the initial radical@zeolite complex and the diffusion and reaction
dynamics of radicals produced by photolysis. The structures of the persistent radicals responsible
for the observed EPR spectra are determined by a combination of alternate methods of generation
of the same radical, by deuterium substitution, and by spectral simulation. A clear requirement
for persistence is that the radicals produced by photolysis must either separate and diffuse from
the external to the internal surface or be generated within the internal surface and separate and
diffuse apart. The persistence of radicals located on the internal surface is the result of inhibition
of radical-radical reactions. Radicals that are produced on the external surface and whose molecular
structure prevents diffusion into the internal surface are transient because radical-radical reactions
occur rapidly on the external surface. The reactions of the persistent radicals with oxygen and
nitric oxide were directly studied in situ by EPR analysis. In the case of reaction with oxygen,
persistent peroxy radicals are formed in high yield. The addition of nitric oxide scavenges persistent
radicals and leads initially to a diamagnetic nitroso compound, which is transformed into a persistent
nitroxide radical by further photolysis. The influence of variation of radical structure on transience/
persistence is discussed and correlated with supramolecular structure and reactivity of the radicals
and their parent ketones.
In tr od u ction
or longer will be persistent and detectable by CW EPR.
This is a very long time scale compared to the lifetimes
(e.g., microseconds to milliseconds) of carbon-centered
radicals in nonviscous fluid solutions.2
In organic chemistry, carbon-centered free radicals are
generally considered to be transient reactive intermedi-
ates, as the result of the inherent fast rates of reactions
of radicals with molecules and of reactions of radicals
with other radicals.1 Even if “inert “ solvents are em-
ployed to remove radical-molecule reactions, radical-
radical reactions (combination and disproportionation)
that typically occur at or near the diffusion rate for simple
carbon-centered radicals will limit the radical lifetime to
the microsecond to millisecond domain.2 The term “per-
sistence” has been suggested3 to describe a radical that
has a significant lifetime compared to some standard,
“transient” radical, under similar conditions in nonvis-
cous solvents. In this report, we operationally define
radicals that are produced in zeolites and that are readily
detected by conventional steady-state CW EPR at room
temperature as “persistent” and radicals that are pro-
duced but cannot be detected by steady-state EPR as
“transient”. This operational definition implies that the
radicals that possess “lifetimes” on the order of seconds
Carbon-centered radicals in inert solvents can be
rendered persistent through intramolecular steric factors
that slow radical-radical reactions.3 For example, in the
case of the classic triphenyl methyl radical,4 the three
phenyl groups form a twisted propeller environment
which sterically inhibits radical-radical reaction at the
trivalent carbon atom. Similar steric effects are respon-
sible for the persistence3,5 of other carbon-centered
radicals in solution. The idea of molecular steric stabi-
lization of radicals has been adapted to design supramo-
lecular steric stabilization to render radicals persistent.6
These studies have shown that benzyl and related
carbon-centered radicals, which are transient and un-
dergo radical-radical reactions at close to the diffusion-
controlled rate in fluid solutions,2 can be rendered
persistent when they are created by the photolysis of
ketones adsorbed on zeolites.6,7 However, not all benzyl
(4) (a) Gomberg, M. Ber. 1900, 33, 3150-3163; J . Am. Chem. Soc.
1900, 22, 757-771. (b) McBride, J . M. Tetrahedron 1974, 30, 2009-
2022.
(5) Lankamp, H.; Nauta, W. T.; MacLean, C. Tetrahedron Lett. 1968,
249-254.
(6) (a) Hirano, T.; Li, W.; Abrams, L.; Krusic, P. J .; Ottaviani, M.
F.; Turro, N. J . J . Am. Chem. Soc. 1999, 121, 7170-7171. (b) Turro,
N. J . Acc. Chem. Res. 2000, 33, 637-646.
† Columbia University.
‡ University of Urbino.
§ E. I. Dupont. Contribution No. 8198.
(1) Fischer, H.; Radom, L. Angew. Chem. 2001, 40, 1340-1371, and
references therein.
(2) Schuh, H.-H.; Fischer, H. Helv. Chim. Acta 1978, 61, 2130-2164.
(3) Griller, D.; Ingold, K. U. Acc. Chem. Res. 1976, 9, 13-19.
10.1021/jo011047l CCC: $22.00 © 2002 American Chemical Society
Published on Web 03/22/2002

Photolysis of Dibenzyl Ketones
J . Org. Chem., Vol. 67, No. 8, 2002 2607
Ch a r t 1. Str u ctu r es of Keton es In vestiga ted in
Th is Rep or t
Su p r a m olecu la r P h otoch em istr y of Or ga n ic Mol-
ecu les Ad sor bed on Zeolites. In molecular organic
chemistry, covalent bonds are critical in determining
molecular chemical structure and reactivity, whereas in
supramolecular organic chemistry noncovalent bonds are
critical in determining supramolecular structure and
reactivity. The “guest@host” paradigm8-10 inspired by the
chemistry of enzymes provides a useful working model
for supramolecular chemistry. In this paradigm the
concept of the molecular “solvent cage” is translated to
the supramolecular “supercage” which holds together
guest molecules (or reactive intermediates) by providing
a cage with “walls” of varying degrees of hardness or
softness through noncovalent bonds.
In this report the guests are a series of dibenzyl
ketones (Chart 1), and the host is ZSM-5 crystals,
members of the MFI family of zeolites.11 The host was
selected because of its importance in the catalytic sci-
ences, which results from the size/shape selectivity of
ZSM-5 zeolites for adsorption of guests whose dimensions
are of the size of a benzene ring.12 MFI crystals possess-
ing Si/Al < 80 are termed MZSM-5, where M represents
an exchangeable cation. In this report, Si/Al ) 20, M )
Na, and the term ZSM-5 will be used to describe the
zeolite crystals. The adsorption of ketones onto a MFI
crystal produces supramolecular ketone@MFI complexes
on the external or internal surface, depending on the size/
shape relationships of the guest and the host void space.
In addition to the structural variations of the guest and
host just described, the variation of the loading (weight
of ketone adsorbed on a given weight of zeolite) provides
a means of varying the supramolecular structure of the
ketone@ZSM-5 complexes also. The reason for this is that
there are several different host binding sites which are
fixed in number depending on the surface area (external
or internal) of the zeolite, and as the loading of ketone is
varied, the occupancy or coverage of the binding sites will
change. Although the molecular structures of the host
and guest are not significantly changed by variation of
coverage of the host surface, we shall show that the
supramolecular structure and dynamics, and as a con-
sequence, the supramolecular reactivity of ketone@ZSM-
5, may change dramatically with coverage.
type radicals produced in this manner are persistent. The
proposed mechanism of radical persistence has been
attributed to supramolecular steric effects and to dy-
namic diffusional and/or structural constraints imposed
by the adsorption of radicals on the internal surface of
zeolites which inhibit radical-radical reactions.6
We report here an EPR investigation of the influence
of supramolecular structure and dynamics on the per-
sistence of a family of carbon-centered radicals produced
by photolysis of “guest” ketones 1, 1-oMe, 2, 2-oMe, 3,
and 4 (Chart 1) adsorbed on a “host” ZSM-5 zeolite. We
shall demonstrate that persistence and transience of
radicals adsorbed on zeolites can be predicted from
knowledge of the initial supramolecular structure and
dynamics of the guest@host complexes (the @ represents
noncovalent binding of the guest and host) that are
photolyzed. The results demonstrate that, to be persis-
tent, the radicals produced by photolysis must either
separate and diffuse from the external to the internal
surface or be generated within the internal surface and
separate and diffuse apart. Radicals that are produced
on the external surface and whose molecular structure
prevents diffusion into the internal surface are transient;
that is, they react too rapidly to be detected by steady-
state EPR analysis.
Size/shape characteristics are important factors in
determining molecular adsorption capacity and diffusion
(8) Lehn, J . M. Supramolecular Chemistry; VCH: New York, 1995.
(9) Herron, N. Chemtech 1989, 542-548.
(10) Parton, R.; De Vos, D.; J acobs, P. A. Enzyme Mimicking with
Zeolites. In Zeolite Microporous Solids: Synthesis, Structure and
Reactivity; Derouane, E. G., Lemos, F., Naccache, E., Ribeiro, F. R.,
Eds.; Kluwer Academic Publishers: Dordrecht, 1992; pp 555-578.
(11) (a) Meier, W. M.; Olson, D. H. Atlas of Zeolite Structure Types;
Butterworth-Heinemann: London, 1992. (b) Breck, D. W. Zeolite
Molecular Sieves: Structure, Chemistry, and Use; Wiley and Sons:
London, 1974. (c) Szostak, R. Molecular Sieves: Principle of Synthesis
and Identification; Van Nostrand Reinhold: New York, 1989. (d)
Flanigen, E. M.; Bennet, J . M.; Grose, R. W.; Cohen, J . P.; Patton, R.
L.; Kirchner, R. M.; Smith, J . V. Nature (London) 1978, 272, 512-
516. (e) Kokotailo, G. T.; Lawton, S. L.; Olson, D. H.; Meier, W. M.
Nature (London) 1978, 272, 437-438. (f) Olson, D. H.; Kokotailo, G.
T.; Lawton, S. L.; Meier, W. M. J . Phys. Chem. 1981, 85, 2238-2243.
(12) (a) Ruthven, D. M. Principles of Adsorption and Adsorption
Processes; Wiley: New York, 1984. (b) Choudhary, V. R.; Nayak, V.
S.; Choudhary, R. V. Ind. Eng. Chem. Res. 1997, 36, 1812-1818. (c)
Derouane, E. G. New Aspects of Molecular Shape-Selectivity: Catalysis
by Zeolite ZSM-5. In Catalysis by Zeolites; Imelik, B., Ed.; Elsevier:
Amsterdam, 1980; pp 5-18. (c) Derouane, E. G. Molecular Shape-
Selective Catalysis by Zeolites. In Zeolite: Science and Technology;
Ribeiro, F. R., Ed.; Martinus, Nijhoff Publ.: The Hague, 1984; pp 347-
370. (d) Derouane, E. G., Ed. Diffusion and Shape Selective Catalysis
in Zeolites; Academic Press: New York, 1982.
(7) (a) Turro, N. J .; Lei, X.; Li, W.; McDermott, A.; Abrams, L.;
Ottaviani, M. F.; Beard, H. S. Chem. Commun. 1998, 695-696. (b)
Turro, N. J .; McDermott, A.; Lei, X.; Li, W.; Abrams, L.; Ottaviani, M.
F.; Beard, H. S.; Houk, K. N.; Beno, B. R.; Lee, P. S. Chem. Commun.
1998, 697-698.

2608 J . Org. Chem., Vol. 67, No. 8, 2002
Turro et al.
Sch em e 1
rates12 of guests in zeolites. It has been demonstrated in
previous investigations that good mass balances of radi-
cal-radical products (ca. 80-90%) can be achieved in the
photolysis13,14 of ketones@ZSM-5, ensuring that the iso-
lated products are representative of the major portion of
the photochemistry and that the zeolite framework does
not react with radicals. Thus, the zeolite provides a
chemically inert host for the carbon-centered radicals
examined, ensuring that the persistence of the radicals
will be determined by the rate of radical-radical reac-
tions. However, we shall show that the radicals do
interact with the zeolite surface and exchangeable cations
and that these interactions influence the mobility and
therefore the reactivity of adsorbed radicals. Radical-
radical reactions on the external surface will be much
more rapid than radical-radical reactions on the internal
surface.6,15,16 Thus in this study, conditions were produced
that are decisive in rendering radicals persistent due to
a supramolecular inhibition of radical-radical reactions
on the internal surface of ZSM-5 zeolites.
surface of a MFI crystal consists11 of pores or “holes”
(approximately circular ca. 5.5 Å in diameter) and “solid”
framework between the holes (shown schematically in
Scheme 1). The external surface of a MFI crystal pos-
sesses a surface area consisting of ca. 20-30% “holes”
and ca. 70-80% “framework”. The internal void space of
MFI zeolites consists of channels that intersect to produce
roughly spherical supercages of ca. 9 Å in diameter. For
the guest ketones investigated (Chart 1), except for 1 and
1-d10 (adsorbed exclusively on the internal surface until
saturation of the internal surface occurs)12 and 1-o,o′DiMe
(adsorbed exclusively on the external surface),6a all the
guest molecules possess molecular cross sections that will
allow only partial adsorption or penetration of a phenyl
group into the holes of the external surface.
The working paradigm for the supramolecular struc-
ture of ketone@external surface complexes assumes6 that
for ortho-ring-substituted ketones (1-oMe and 2-oMe)
the first, more strongly bound ketone@external surface
complexes formed are those for which the ketone pref-
erentially adsorbs partially into the holes of the external
surface of MFI crystals, as shown in previous investiga-
tions.6,7,17 The preference for binding to the holes is
assumed to be due to favorable dispersion interactions
of an adsorbed benzene ring with the walls of the
pores on the external surface. After saturation of the
holes, further adsorption of ketones occurs onto the
A Wor k in g P a r a d igm for th e Su p r a m olecu la r
Str u ctu r e of Keton e@MF I Com p lexes. The external
(13) Turro, N. J .; Wan, P. J . Am. Chem. Soc. 1985, 107, 678-685.
(14) Turro, N. J .; Cheng, C. C.; Abrams, L.; Corbin, D. R. J . Am.
Chem. Soc. 1987 109, 2449-2456.
(15) (a) Kelly, G.; Willsher, C. J .; Wilkinson, F.; Netto-Ferreira, J .
C.; Weir, A. O. D.; J ohnston, L. J .; Scaiano, J . C. Can. J . Chem. 1990,
68, 812-819. (b) J ohnston, L. J .; Scaiano, J . C.; Shi, J .-L.; Siebrand,
W.; Zerbetto, F. J . Phys. Chem. 1991, 95, 10018-10024.
(16) J ockusch, S.; Hirano, T.; Liu, Z.; Turro, N. J . J . Phys. Chem. B
2000, 104, 1212-1216.
(17) Turro, N. J .; Lei, X.-G.; Li, W.; Liu, Z.; McDermott, A.;
Ottaviani, M. F.; Abrams, L. J . Am. Chem. Soc. 2000, 122, 11649-
11659.

Photolysis of Dibenzyl Ketones
J . Org. Chem., Vol. 67, No. 8, 2002 2609
Sch em e 2. Sch em a tic Rep r esen ta tion of th e Exter n a l Su r fa ce of ZSM-5 Zeolitesa
a
The model assumes that initial binding is stronger for the holes (left), followed by binding to the framework (right) until a monolayer
is formed. Left refers to “low coverage” (up to saturation of the holes); right refers to “high coverage” (beyond saturation of the holes and
up to saturation of the external surface, i.e., formation of a monolayer).
Sch em e 3. Sch em a tic of P ossible Sitin g of Keton es@ZSM-5a
a
See text for discussion.
framework between the holes until a monolayer is formed
(Scheme 2). This paradigm predicts an important change
in the supramolecular structure of the ketone@external
surface complexes as a function of surface coverage:
initially, ketone@hole complexes will be formed, and after
all the holes are filled, ketone@framework complexes will
form until a monolayer is created. It has been shown6,17
that the behavior of the supramolecular radicals pro-
duced by the photochemistry of the two supramolecular
isomers (ketone@hole and ketone@framework) is very
different. The term “low coverage” will refer to situations
up to saturation of the holes, and the term “high
coverage” will refer to situations for which the holes are
completely filled and the framework is partially or
completely covered.
dynamic pathways available for radical-radical reactions
of the secondary radical pair. These pathways are
strongly controlled by (1) the initial siting of the ketone
(on either the internal surface of the zeolite or at the
holes and framework at the interface of the external and
internal surface) and (2) the dynamics for diffusion and
rotational motions of the radicals on the external and
internal surface.
A Mod el Cor r ela tin g Su p r a m olecu la r Str u ctu r e
w ith Ra d ica l P er sisten ce a n d Tr a n sien ce. A model
for photolysis of ketone@hole or ketone@framework
complexes6 is shown in Scheme 3 for three limiting cases.
Imagine a ketone AexCOBin (the subscripts refer to
external surface and internal surface) adsorbed on a MFI
crystal [the molecular cross section of the moiety Aex is
assumed to be too large to allow it to pass through the
hole on the external surface, and the molecular cross
section of the moiety Bin is assumed to be small enough
to fit into the hole and to diffuse into the internal
surface]. Ketone 1-o,o′DiMe and 1 (1-d10) can be com-
pletely adsorbed into the external surface (Scheme 3b)
or internal surface (Scheme 3c) until surface saturation,
respectively. However, at “low coverage”, the initial
supramolecular structures of all of the other ketones in
Chart 1 will be constrained to the interface of the zeolite
crystal and will have a portion of the ketone (a phenyl
group) adsorbed in the hole on the external surface; that
Th e Su p r a m olecu la r P h otoch em ica l P a r a d igm . It
is assumed that the primary photochemical process in
all cases is R-cleavage of the ketone triplet to produce a
geminate radical pair, followed by rapid decarbonylation
to form a secondary radical pair.18,19 The products of the
photochemistry are determined by the supramolecular
(18) Turro, N. J . Modern Molecular Photochemistry; University
Science: Menlo Park, 1991.
(19) (a) Engel, P. S. J . Am. Chem. Soc. 1970, 92, 6074-6076. (b)
Robbins, W. K.; Eastman, R. H. J . Am. Chem. Soc. 1970, 92, 6076-
6077. (c) Robbins, W. K.; Eastman, R. H. J . Am. Chem. Soc. 1970, 92,
6077-6079.

2610 J . Org. Chem., Vol. 67, No. 8, 2002
Turro et al.
F igu r e 1. Steady-state CW-EPR spectra of radicals produced by the photolysis of ketones@ZSM-5: (a) simulation of B; (b) spectrum
produced from photolysis of 1; (c) spectrum produced from photolysis of 1-oMe-d4; (d) spectrum produced from photolysis of 1-oMe;
(e) simulation of B-d5; (f) spectrum produced from photolysis of 1-d10; (g) spectrum produced from photolysis of 1-oMe-d5.
is, these ketones will be examples of a supramolecular
complex, AexCOBin@hole (Scheme 3a). After all the
holes are filled, at “high coverage” the ketones bind to
the framework to generate a new, isomeric supramolecu-
lar complex, AexCOBex@framework (Scheme 3b).
reactions.6 The coverage dependences will correlate with
the filling of the holes on the external surface.
It will be shown directly by EPR spectroscopy that the
persistent radicals@ZSM-5 are quite reactive toward
small molecules such as oxygen and nitric oxide that can
easily diffuse throughout the zeolite surface. In addition,
product analysis demonstrates that radical-radical reac-
tions do occur and that the products of these reactions
depend strongly on the supramolecular structures and
dynamics of the systems investigated.
This paradigm suggests a number of consequences for
the photolysis of AexCOBin@ZSM-5 which are experimen-
tally testable: (1) photolysis of AexCOBin@hole (Scheme
3a, e.g., 1-oMe) will produce, after loss of CO, two
radicals, one Aex, which is “directed” toward the external
surface, and a second radical, Bin, which is “directed”
toward the internal surface; (2) photolysis of Aex-
COBex@framework (Scheme 3b, e.g., 1-o,o′DiMe, where
A ) B) will produce, after loss of CO, two radicals, Aex
and Bex, both of which are “directed” toward the external
surface; (3) photolysis of AinCOBin@intersection (e.g., 1,
where A ) B); (4) the radicals that are on the external
surface will be able to diffuse freely and undergo rapid
radical-radical reactions;15,16 (5) the radicals that enter
the internal surface will diffuse slowly and undergo
radical-radical reactions that will be inhibited to some
extent by supramolecular steric effects within the super-
cages of the internal surface; (6) the radicals undergoing
rapid radical-radical reactions on the external surface
will tend to be transient, and those that migrate into the
internal surface and undergo slower radical-radical
reactions on the internal surface will tend to be persis-
tent;7 (7) there will be a coverage dependence of the yield
of persistent radicals and the products of radical-radical
Resu lts
St ea d y-St a t e E P R P r od u ced b y P h ot olysis of
Keton es@ZSM-5. The ketones 1, 1-d10, 1-oMe, 1-oMe-
d4, and 1-oMe-d5 (Chart 1) were adsorbed (ca. 1 wt %/wt)
on ZSM-5 to form supramolecular complexes termed
1@ZSM-5, 1-d10@ZSM-5, 1-oMe@ZSM-5, 1-oMe-d4@ZSM-
5, and 1-oMe-d5@ZSM-5, respectively. Photolysis of sealed
vacuum-degassed samples of 1@ZSM-5, 1-oMe@ZSM-5,
and 1-oMe-d4@ZSM-5 produced experimentally indistin-
guishable EPR spectra (Figure 1, left). In a typical
experiment, samples were photolyzed for ca. 7 min under
comparable conditions. The observed spectra produced
from photolysis of 1, 1-oMe, and 1-oMe-d4 are assigned
to a single species, the benzyl radical (B, Chart 2), based
on spectral simulation (Figure 1, left top) and alternate
methods of preparation (Figure 1).
Photolysis of sealed vacuum-degassed samples of
1-d10@ZSM-5 and 1-oMe-d5@ZSM-5 produced experimen-

Photolysis of Dibenzyl Ketones
J . Org. Chem., Vol. 67, No. 8, 2002 2611
F igu r e 2. Steady-state CW-EPR spectra of radicals produced by the photolysis of ketones@ZSM-5: (a) simulation of R-MeB; (b)
spectrum produced from photolysis of 2; (c) spectrum produced from photolysis of 2-oMe; (d) simulation of R-Me-B-d5; (e) spectrum
produced from photolysis of 2-d10
.
Ch a r t 2. Str u ctu r es of Ra d ica ls Discu ssed in
Th is Rep or t
of ZSM-5 at room temperature), whereas the transient
radical (oMeB) has a kinetic cross section similar to
o-xylene (which cannot be adsorbed into the internal
surface of ZSM-5 zeolites at room temperature).11d,e
Photolysis of 1-o,o′DiMe (Chart 1) serves as a control for
persistent radicals. Photolysis of 1-o,o′DiMe@ZSM-5 com-
plexes, as predicted by the model, leads to no persistent
EPR signals of oMeB within the experimental uncer-
tainty (results not shown), demonstrating that the oMB
radicals are transient (eq 1c).
tally indistinguishable steady-state EPR spectra (Figure
1, right). These spectra are assigned to a single species,
the ring deuterated benzyl radical (Bd5, Chart 2), based
on spectral simulation (Figure 1, right top) and alternate
methods of preparation (Figure 1). All of the EPR spectra
shown in Figure 1 were reproducible at room tempera-
ture and possessed “half-lives” on the order of 10 min.
The results convincingly demonstrate that oMeB and
oMeB-d4 (Chart 2), produced by photolysis of 1-oMe and
1-oMe-d4, are not persistent on the time scale of the
measurements, whereas the benzyl radicals produced by
the photolysis of 1 (1-d5) or 1-oMe (1-oMe-d4) are persis-
tent (eqs 1a and 1b). It is important to note that the
persistent radical (B) has a kinetic cross section similar
to p-xylene (which can be adsorbed in the internal surface
The supramolecular complexes 2@ZSM-5, 2-oMe@ZSM-
5, and 2-d10@ZSM-5 were photolyzed under the conditions
described above and analyzed by steady-state EPR.
Photolysis of sealed vacuum-degassed samples of 2@ZSM-5
and 2-oMe@ZSM-5 produced an experimentally indistin-
guishable EPR spectrum (Figure 2, left), assigned to
a single species, the R-methyl benzyl radical (R-MeB,
Chart 2), based on spectral simulation (Figure 2, top left)
and an alternate method of preparation (Figure 2).

2612 J . Org. Chem., Vol. 67, No. 8, 2002
Turro et al.
F igu r e 3. Steady-state CW-EPR spectra of radicals produced by the photolysis of ketones@ZSM-5: (a) spectrum produced by
the photolysis of 1; (b) spectrum produced from photolysis of 2; (c) spectrum produced from photolysis of 3; (d) spectrum produced
from photolysis of 4.
Photolysis of sealed vacuum-degassed samples of 2-
10@ZSM-5 produced the EPR spectrum shown in Figure
R-alkyl benzyl radical. However, as the alkyl increases
in size, a significant broadening of the EPR signals occurs
compared to those observed upon photolysis of 2@ZSM-5
(and 1@ZSM-5). In addition, the intensity and persistence
of the R-alkyl B radical increases (results not shown) with
the increasing size of the alkyl group, i.e., R-PeB > R-EtB
> R-MeB > B.
Sca ven gin g of Ben zyl a n d r-Meth yl Ben zyl Ra d i-
ca ls by Molecu la r Oxygen . The EPR spectra of the
benzyl radical produced from photolysis of 1@ZSM-5 or
1-oMe@ZSM-5 are transformed into the EPR spectra of
different persistent radicals by addition of air (Figure 4,
left). Similarly, the EPR spectra of the R-methyl benzyl
radical produced from photolysis of 2@ZSM-5 and
2-oMe@ZSM-5 are transformed to the EPR spectra of
different persistent radicals (Figure 4, right) that are
distinct from those produced from photolysis of 1@ZSM-5
or 1-oMe@ZSM-5.
d
2, right. This spectrum is assigned to a single species,
the ring-deuterated R-methyl benzyl radical, based on
spectral simulation (Figure 2, right top). The EPR spectra
shown in Figure 2 were persistent at room temperature
with “half-lives” on the order of 60 min. The results
demonstrate that the oMe-substituted radicals (oMeB,
Chart 2) produced from photolysis of 2-oMe@ZSM-5
are not persistent on the time scale of the measure-
ments, whereas the R-MeB radicals are persistent (eqs
2a and 2b).
The EPR spectra produced by addition of air to pho-
tolyzed samples are assigned to peroxy radicals (eqs 3
and 4) based on computer simulation of a powder
In flu en ce of r-Alk yl Su bstitu en ts on th e EP R
P r od u ced by P h otolysis of r,r′-Dia lk yl Diben zyl
Keton es@ZSM-5. The effect of the length of the R-alkyl
chain on the EPR spectra produced on photolysis of R,R′-
dialkyl ketones was investigated for the R,R′-dialkyl
dibenzyl ketones 3 and 4 adsorbed on ZSM-5. Figure 3
shows the EPR spectra produced upon photolysis of
samples of 3@ZSM-5 and 4@ZSM-5, with the spectra
resulting from photolysis of 1@ZSM-5 and 2@ZSM-5
included for comparison. In each case the dominant
species that carries the EPR signal is assigned to an
spectrum of a peroxy radical and on comparison with the
EPR spectrum obtained from photolysis of cumene peroxy
radical in a solid matrix.20,21 The relative amounts of

Photolysis of Dibenzyl Ketones
J . Org. Chem., Vol. 67, No. 8, 2002 2613
F igu r e 4. Effect of addition of air to the steady-state CW-EPR spectra of radicals produced by the photolysis of ketones@ZSM-5:
(a, d, e, and h) spectra produced by photolysis of 1, 1-oMe, 2-oMe, and 2, respectively; (b, c, f, and g) spectra produced upon
addition of air to the samples corresponding to a, d, e, and h, respectively.
radicals (determined by double integration of the spectra)
were essentially identical before and after aeration of
the samples, indicating that the persistent benzyl and
R-methyl benzyl radicals are converted to persistent
peroxy radicals in high yield. The fact that the spectra
of the peroxy radicals correspond to those expected for a
rigidly held radical indicates that the peroxy radicals are
very restricted in motion.Vacuum removal of the oxygen
in the samples did not affect the peroxy radical spectrum,
indicating that the reaction of the benzyl and R-methyl
benzyl radicals with oxygen is irreversible.22
sealed sample containing nitric oxide resulted in the
appearance of a new EPR spectrum which could be
assigned by computer simulation to that of a nitroxide
shown in eq 5 (Figure 5, top).
Sca ven gin g of r-Meth yl Ben zyl Ra d ica ls by Nitr ic
Oxide. Exposure of a photolyzed vacuum-degassed sample
of 1@ZSM-5 or 2@ZSM-5 to nitric oxide resulted in the
complete disappearance23 of the benzyl and R-methyl
benzyl radical EPR spectrum, respectively. A representa-
tive example is shown in Figure 5 for the addition of nitric
oxide to irradiated samples of 2@ZSM-5. Addition of nitric
oxide (ca. 2 Torr, molar ratio of ketone to NO ) 2) caused
the entire EPR spectrum of the R-methyl benzyl radical
to disappear (Figure 5, middle). Further irradiation of a
P r od u ct An a lysis of th e P h otolysis a s a F u n ction
of Loa d in g. The major product produced by photolysis
of 1@ZSM-5 was found to be 1,2-diphenylethane (eq 6)
and is independent of loading in the range of 0.1-5% (wt/
wt). However, the products produced from photolysis of
2@ZSM-5 are strongly dependent on loading in the range
of 0.1-5%. For loadings in the range from ca. 0.1 to 1%
the ratio of disproportionation (eq 7) to combination (eq
8) is ca. 3-5 (values decreasing with increasing coverage),
and for loadings in the range from ca. 1-5% the ratio of
disproportionation to combination is ca. 0.2-0.8 (decreas-
ing with increasing coverage). In solution, the ratio of
disproportionation to combination is ca. 0.1 for photolysis
(20) Schlick, S.; Kevan, L. J . Phys. Chem. 1979, 83, 3424-3429.
(21) Kevan, L.; Schlick, S. J . Phys. Chem. 1986, 90, 1998-2007.
(22) Hirano, T.; Li, W.; Abrams, L.; Krusic, P. J .; Ottaviani, M. F.;
Turro, N. J . J . Am. Chem. Soc. 1999, 121, 7170-7171.
(23) For examples of trapping of radicals with NO in solution see:
(a) Maruthanmuthu, P.; Scaiano, J . C. J . Phys. Chem. 1978, 82, 1588-
1591. (b) Gyor, M.; Rochenbauer, A.; Tudos, F. Tetrahedron Lett. 1986,
32, 3759-3762.

2614 J . Org. Chem., Vol. 67, No. 8, 2002
Turro et al.
R-cleavage, followed by rapid (k-CO ca. 107 s-1) secondary
thermal decarbonylation24 to form free benzyl radicals.
For both the benzyl radicals (produced from photolysis
of 1) and the R-methyl benzyl radicals (produced from
photolysis of 2) radical-radical recombination proceeds
at close to the rate of diffusion and leads to diphenyl
ethanes (eqs 1 and 6) as the dominant products in ca.
90% yield.25
The primary supramolecular photochemistry of 1 and
2 adsorbed on zeolites is assumed to be identical to the
molecular photochemistry (R-cleavage from the triplet
followed by rapid decarbonylation). However, the reac-
tions of the secondary geminate benzyl and R-methyl
benzyl radical produced are strongly influenced by the
supramolecular structure and dynamics of the radicals@
zeolite complexes. We now analyze the initial supramo-
lecular structure of the ketone@ZSM-5 complexes and
then interpret the experimental result in terms of the
supramolecular model presented above (Scheme 3).
Tr a n sien t a n d P er sisten t Ra d ica ls P r od u ced by
P h otolysis of Keton e@ZSM-5 Com p lexes. Of the
ketones investigated (Chart 1), only 1 and 1-d10 (mini-
mum molecular cross sections <5.5 Å) possess a kinetic
molecular cross section that allows ready access to the
internal surface at room temperature, since the holes
(pores) on the external surface have a “void” cross section
of ca. 5.5 Å (Scheme 1). The ketones 1-oMe, 2, and 2-oMe
(minimum molecular cross sections >5.5 Å) are too large
to enter the internal surface by sieving through the pores
on the external surface. However, one of the phenyl
groups of the latter ketones can be adsorbed into a pore
on the external surface. If such supramolecular struc-
tures are formed, then the part of the ketone molecule
that is adsorbed in the pore will be in good position to be
adsorbed into the internal surface after photochemically
induced R-cleavage; the other larger moiety of the
molecule will be in good position to be adsorbed on the
external surface after photochemically induced R-cleav-
age. Finally, 1-o,o′DiMe possesses a structure that pre-
vents the insertion of either ring into the pores on the
external surface. It is important to note that the internal
surface of ZSM-5 zeolites possesses ca. 500 m2/g of
internal surface and ca. 10 m2/g of external surface.
However, adsorption on the huge internal surface can
only occur for molecules or reactive intermediates whose
size/shape characteristics allow diffusion into the internal
surface. Scheme 3 shows the supramolecular structural
paradigm used to interpret the results. At loadings on
the order of ca. 0.5% or less, all of the ketones except 1,
1-d10, and 1-o,o′diMe are adsorbed in or near the holes
of the external surface, with a benzyl (or R-methyl benzyl)
moiety associated with the hole on the external surface
(Scheme 3a). On the other hand, 1 will be primarily
adsorbed on the internal surface (Scheme 3c), plausibly
at the spherical intersections which possess diameters
of ca. 9 Å, and 1-o,o′diMe will be adsorbed exclusively on
the external surface (Scheme 3b).26 Photolysis of 1@ZSM-5
produces, after decarbonylation, secondary geminate
radical pairs which either recombine to form 1,2-diphenyl-
F igu r e 5. Effect of the addition of nitric oxide to the steady-
state CW-EPR spectra of radicals produced by the photolysis
of 2@ZSM-5: (a) spectrum produced from photolysis of 2; (b)
spectrum produced upon addition of nitric acid; (c) spectrum
produced upon further photolysis of sample after addition of
nitric oxide; (d) simulated spectrum of the nitroxide R-MeBNit.
of 2, i.e., a value comparable to that for the higher
loadings of 2@ZSM-5.
Discu ssion
The molecular photochemistry of the dibenzyl ketone
family of ketones has been well established.18,19 In
common organic solvents after photochemical excitation,
the initially produced singlet excited state of 1 is trans-
formed nearly quantitatively to the triplet state and
undergoes rapid (kR ca. 109 s-1) primary photochemical
(24) (a) Lunazzi, L.; Ingold, K. U.; Scaiano, J . C. J . Phys. Chem.
1983, 87, 529-530. (b) Turro, N. J .; Gould, I. R.; Baretz, B. H. J . Phys.
Chem. 1983, 87, 531-532. (c) Gould, I. R.; Baretz, B. H.; Turro, N. J .
J . Phys. Chem. 1987, 91, 925-929.
(25) (a) Baretz, B. H.; Turro, N. J . J . Am. Chem. Soc. 1983, 105,
1309-1316. (b) Ghatlia, N. D.; Turro, N. J . J . Photochem. Photobiol.
A: Chem. 1991, 57, 7-19.

Photolysis of Dibenzyl Ketones
J . Org. Chem., Vol. 67, No. 8, 2002 2615
ethane (eq 6) or diffuse apart to form free benzyl radicals.
The geminate coupling of benzyl radicals occurs on a very
short time scale, and the radicals that combine gemi-
nately are transient and are not detected by steady-state
EPR. The benzyl radicals that diffuse apart become free
radicals, some of which become persistent and are
detected by steady-state EPR. This feature of the model
is consistent with the observation that the cage effect for
secondary radical recombination is ca. 60-70% for
1@ZSM-5.27
tions.15,16 These radicals will be transient, and their EPR
spectra will not be observed on the time scale of seconds.
It should be noted in passing that in favorable cases
(diphenyl methyl) transient radicals adsorbed on the
external surface can be observed by laser flash photolysis
methods and that the radicals decay on a microsecond
time scale.15,16 The radicals formed in the pores will tend
to diffuse into the internal surface and become persistent
by the definition employed in this study.
With this supramolecular structural and dynamic
model in mind, we now consider the implications on the
EPR spectra produced by photolysis of 1@ZSM-5 and
1-oMe@ZSM-5 (Figure 1). The observed spectra are
assigned to persistent benzyl radicals based on the good
fit of the simulated spectra to the experimental spectra
and on the fact that the same spectrum is produced from
two precursors. The absence of a contribution from the
ortho-methyl benzyl radical (which has a distinctly
different EPR spectrum from the benzyl radical)17 is
mechanistically significant, since it confirms the supra-
molecular structural assumption that the ortho-methyl
benzyl moiety (oMeB, Chart 2) cannot enter the internal
surface and become persistent. Although ortho-methyl
benzyl radicals are produced (as demonstrated by product
analysis),14 they rapidly diffuse on the external surface,
undergo recombination reactions, and are not persistent
on the time scale of seconds or longer, and their steady-
state EPR spectrum is not observed.
The lifetime of the B radicals produced by photolysis
of 1@ZSM-5 or 1-oMe@ZSM-5 is qualitatively the shortest
of the radicals investigated. This is attributed to the
faster diffusional dynamics of the B radical compared to
the other radicals investigated and to the smaller cross
section of the coupling product formed by combination
of B radicals at an intersection (eq 6). We speculate that
of all the persistent radicals investigated, only B can
undergo radical-radical combination at close to the rate
of diffusion on the internal surface; that is, there is a
minimal supramolecular steric effect on the combination
of two B radicals in a ZSM-5 intersection.
The EPR produced by photolysis of 2@ZSM-5 and
2-oMe@ZSM-5 (Figure 2) are assigned to the R-methyl
benzyl radical based on the good fit of the simulated
spectra to the experimental spectra and on the fact that
the same spectrum is produced from two precursors. As
in the case of photolysis of 1-oMe@ZSM-5, the absence
of a contribution of the oMeR-MeB radical from the
photolysis of 2-oMe@ZSM-5 is significant and demon-
strates that, although the R-MeB radical can diffuse into
the internal surface and become persistent, oMeR-MeB
cannot. The latter radicals are produced on the external
surface where they diffuse and react rapidly and are
transient, as is the case for the o-MeB radicals produced
from photolysis of 1-oMe and 1-o,o′DiMe.
In summary, the EPR results of Figures 1 and 2 are
completely consistent with the supramolecular model
(Scheme 3), which predicts that benzyl and R-methyl
benzene radicals that are able to be adsorbed in the
internal surface can become persistent and are respon-
sible for the observed EPR spectra, but the oMe-
substituted radicals (oMeB and oMeR-MeB) are adsorbed
only on the external surface and therefore are transient.
In flu en ce of r-Alk yl Su bstitu en t on th e EP R
Sp ectr u m . The variation of the R-alkyl substituent has
three effects on the EPR spectra (in addition to the
specific changes of hyperfine couplings due to changes
Thus, the interpretation of the EPR spectra produced
from photolysis of 1@ZSM-5 (Figure 1) is that for each
system the spectra are due to persistent benzyl radicals
that have escaped from the cage in the zeolite internal
surface in which they were produced. The persistence
(lifetime) of these radicals is determined by the rate of
radical-radical combination reactions. The rate-limiting
feature of radical-radical reactions may be the time scale
for a pair of diffusing free radicals to find each other at
an intersection as they percolate through the porous
internal surface of the zeolite (supramolecular diffusion-
controlled combination). Since geminate recombination
of benzyl radicals has been shown to occur relatively
efficiently in the supercages at the intersections,14 it is
likely that the rate-limiting feature is related to diffu-
sionally controlled limited recombination on the internal
surface. Other mechanistic possibilities can be envisioned
(e.g., diffusion to and recombination on the external
surface) but are considered less likely. Thus, we conclude
that the lifetime of the persistent benzyl radicals is
limited by the rate at which radicals diffuse together and
encounter at an intersection of the internal surface and
that the reaction mechanism is analogous to diffusion-
controlled reaction of radicals in fluid solution.
The situation for 1-o,o′DiMe@ZSM-5 is quite different,
as there is no significant EPR signal observed upon
photolysis. This is the expected result for the situation
in Scheme 3b for which the radicals are produced on the
external surface and for which the size of the radicals is
too large to allow kinetic entry to the internal surface.
In this case, rapid diffusion and radical-radical reactions
occur on the external surface. In contrast to the situa-
tion for 1@ZSM-5 and 1-o,o′DiMe, the ketones in the
1-oMe@ZSM-5, 2@ZSM-5, and 2-oMe@ZSM-5 complexes
are not adsorbed on the internal surface, but are mainly
adsorbed on the framework or pore system on the
external surface. Although photolysis of 1-oMe@ZSM-5,
2@ZSM-5, and 2-oMe@ZSM-5 results in R-cleavage and
decarbonylation, the resulting structure of the resulting
radical pair is supramolecularly asymmetric, since one
radical is formed adsorbed in a hole and the other radical
is formed on the external surface (Scheme 3a). The
radicals formed on the external surface are able to diffuse
rapidly on the external surface, mainly encounter other
radicals of the same type, and undergo rapid, probably
diffusion-controlled radical-radical combination reac-
(26) We note at this point that if the exact supramolecular structure
does not have the phenyl group protruding in the hole, it must have
characteristics that are equivalent to this structure. In particular, the
actual structure must be stoichiometrically related to the number of
holes, such that the smaller moiety of the molecule tends to be adsorbed
into the internal surface. One could imagine, for example, an alternate
structure that is bonded near a hole through SiOH bonds and for which
the proximity to the hole enhances the probability of adsorption into
the internal surface, which is only possible for the smaller fragments
produced by photolysis.
(27) Lei, X.-G. Unpublished results.

2616 J . Org. Chem., Vol. 67, No. 8, 2002
Turro et al.
in the coupled protons on going from a CH3 substituent
to a CH2R substituent), as shown in Figure 3: the line
widths, the half-lives (not shown), and the intensity of
the spectra (not shown). The trend is that, as the size of
the alkyl substituent increases, the line width, the half-
life, and the intensity of the spectra increase. These
observations are consistent with a dominating steric
effect on the motion and diffusion of the radicals. As the
alkyl group gets larger, the steric effects increase, leading
to an inhibition of the rotational and diffusional motion
of the persistent radicals. The inhibition of rotation
results in broader lines; the inhibition of diffusion results
in slower radical-radical reaction, which, in turn, leads
to a higher persistent radical concentration and a stron-
ger signal and half-lives. These conclusions are consistent
with measurements of the rates of diffusion of molecules
such as toluene, ethylbenzene, and n-propylbenzene in
ZSM-5.12b
Oxygen Sca ven gin g of P er sisten t Ra d ica ls. The
addition of air to samples containing persistent benzyl
radicals or persistent R-methyl benzyl radicals (Figure
4, eqs 3 and 4) shows that both types of radicals react
with oxygen to produce persistent peroxy radicals; that
is, the EPR spectrum of the carbon-centered radical is
transformed essentially quantitatively into the EPR of
the peroxy radical. These results are consistent with
earlier investigations of the diphenyl methyl radical.22
However, it was found in the case of the diphenyl methyl
radical that the addition of oxygen was reversible at room
temperature. In the case of the benzyl and R-methyl
benzyl radicals, the reaction with oxygen is irreversible
at room temperature. These results demonstrate the
ability to perform reactions with the persistent radicals
and to track the reactions in situ by steady-state EPR
spectroscopy.
Nitr ic Oxid e Sca ven gin g of P er sisten t Ra d ica ls.
The reaction of persistent radicals adsorbed on zeolites
with NO is quite remarkable (Figure 5, eq 5). Initial
addition of NO results in complete disappearance of the
EPR signal. Photolysis of the sample after addition of NO
(ca. 2 Torr) causes the creation of a strong EPR signal,
comparable in intensity to that of the original persistent
radical signal. From simulations the new EPR is assigned
to a nitroxide radical (eq 5), formed by addition of an
R-methyl benzyl radical to the nitroso compound formed
by scavenging.23 This is an important result, because it
confirms the conclusion that diffusion of R-methyl benzyl
radicals occurs mainly within the internal surface and
is an example of scavenging of persistent radicals by a
trap that is situated on the internal rather than on the
external surface. This conclusion is supported by product
distribution studies discussed in the next section.
P r od u cts fr om P h otolysis of 1 a n d 2. Photolysis of
1 produces 1,2-diphenylethane (eq 6) as the exclusive
product. The formation of 1,2-diphenylethane as the
exclusive product from radical-radical combination reac-
tions of B radicals (eq 6) is not surprising, because the
size (ca. 9 Å, Scheme 1) of the intersection of the ZSM-5
internal surface allows the product to fit comfortably.
This feature and the experimental EPR have led to the
conclusion that recombination of B radicals is probably
at a rate close to diffusion controlled in the internal
surface. In this case, persistence is limited by the rate of
diffusion.
radicals (eq 8) is so large that it will possess a high
“strain” energy when adsorbed in the intersections of the
internal surface of MFI zeolites. As a result, radical-
radical combination, essentially the exclusive reaction of
R-methyl benzyl radicals in solution,25 is expected to be
strongly inhibited by this supramolecular steric effect.
However, some radical-radical reactions must occur,
since the adsorbed R-methyl benzyl radicals do disappear
over a period of hours. Thus, the R-methyl benzyl radicals
are persistent because they are adsorbed into the internal
surface and because combination reactions are strongly
inhibited due to supramolecular steric effects to combina-
tion. In contrast with the case of B radicals, which
undergo diffusion-controlled combination, R-MeB radicals
undergo reaction (or conformation)-limited dispropor-
tionation. The high energy barrier for radical-radical
combination reaction of R-MeB radicals compared to the
comparable reaction of B radicals is confirmed by com-
puter simulation.
Examination of the product distribution from the
photolysis of 2@ZSM-5 as a function of loading provides
insight into the reaction responsible for removal of the
persistent radicals. At lower loadings, the main products
are styrene and ethylbenzene (eq 7), structures expected
from radical-radical disproportionation, whereas at high
loadings 2,3-diphenylbutane (eq 8), the result of radical-
radical combination, is the major product. At low loading
of 0.3%, the ratio of disproportionation to combination is
3, whereas at a high loading of 5% the ratio of dispro-
portionation to combination (0.3) is comparable to the
value for photolysis of 2 in benzene.
These results on product distributions are interpreted
as follows. At low loadings, molecules of 2 are mainly
adsorbed in the holes on the external surface and after
photolysis some of the R-MeB radicals generated are
sieved into the internal surface. These radicals are more
persistent than are B radicals, because radical-radical
reactions of R-MeB are strongly inhibited in the internal
surface. However, the inhibition to radical-radical dis-
proportionation of R-MeB radicals is evidently less than
that for radical-radical combination. This is plausible
since the disproportionation reaction produces two mol-
ecules and does not produce a product that is too large
to fit in the zeolite intersection. Thus, the steric require-
ments for disproportionation are less than those for
combination.
At high loadings, molecules of 2 are adsorbed in all of
the holes and also on the framework between the holes
and in multilayers. Thus, photolysis of 2@ZSM-5 at high
loadings produces a significant number of radicals that
cannot enter the internal surface, because the holes are
plugged by ketone molecules. These radicals, however,
can diffuse on the external surface, encounter, and
undergo combination reactions without steric inhibition.
Finally, product analysis from the photolysis of
1-o,o′DiMe@ZSM-5 results in a product distribution for
which the coupling of oMeR-MeB radicals dominates.
This result demonstrates that these radicals do not enter
the internal surface and react dominantly on the external
surface.
Con clu sion s
The photolysis of the supramolecular complexes formed
by adsorption of the ketones shown in Chart 1 onto
ZSM-5 produces persistent radicals (Figures 1-3). The
Computer simulations suggest that the molecular cross
section of the coupling product of R-methyl benzyl

Photolysis of Dibenzyl Ketones
J . Org. Chem., Vol. 67, No. 8, 2002 2617
1,3-Dip h en yl-2-p r op a n on e (DBK) (1) was purified by
recrystallization from 5% (v/v) ether in hexane.
detailed features of the EPR spectra depend on the
loading and the molecular structure of the guest ketone.
These features are predictable on the basis of the initial
supramolecular structure, ketone@ZSM-5, which under-
goes photolysis (Scheme 3), and on the dynamics and
reactions of radicals@ZSM-5 that result from photolysis.
These initial supermolecular structures determine the
supramolecular structures of the carbon-centered radicals
produced by photolysis. Radicals that are constrained to
the external surface (e.g., oMeB, oMe, R-MeB, Chart 2)
diffuse on the external surface, undergo rapid radical-
radical reactions, are transient, and are not observed by
steady-state EPR. Radicals that are generated in the
internal surface or that diffuse into the internal surface
undergo slow radical-radical reactions and are persistent
and observed by steady-state EPR.
The persistent radicals react with oxygen to form
persistent peroxy radicals. Persistent radicals react with
nitric oxide to form diamagnetic nitroso compounds that
can undergo further reaction with radicals to form
nitroxides.
The products produced from persistent radicals on the
internal surface are determined by steric constraints or
radical-radical reactions in the supercages of the inter-
nal surface. Benzyl radicals are small enough to undergo
radical-radical recombination in the supercages within
a time frame of minutes. On the other hand, R-methyl
benzyl radicals are too large to undergo radical-radical
combination in a supercage and instead undergo the
sterically less demanding radical-radical disproportion-
ation on the time scale of hours.
1-(2-Meth ylp h en yl)-3-p h en yl-2-p r op a n on e (o-MeDBK)
(1-oMe) was synthesized following a method designed for 1-(4-
methylphenyl)-3-phenyl-2-propanone using 2-(methylphenyl)-
acetic acid instead of 4-(methylphenyl) acetic acid.29 1H NMR:
δ 7.35-7.26 (m, 3H), 7.21-7.11 (m, 5H), 7.11-7.02 (m, 1H),
3.74 (s, 2H), 3.71 (s, 2H), 2.13 (s, 3H). 13C NMR: δ 205.9, 137.2,
134.2, 133.0, 130.6, 128.9, 127.6, 127.2, 126.4, 49.3, 47.5, 19.7.
MS m/z (relative intensity): 224.05 (M+, 2.6), 133 (4.0), 132
(6.6), 119 (2.1), 118 (4.2), 105 (100), 91 (69), 92 (24), 77 (15.8).
1,3-Di(p h en yl-d 5)-2-p r op a n on e (DBK-d 10) (1-d 10) was
synthesized as follows: Benzene-d6 was treated with oxalyl
chloride to yield benzoyl-d5 chloride,30 which was reduced with
LiAlH4 to yield benzyl-d5 chloride.31 The latter was treated
with KCN and refluxed with a concentrated KOH solution.32
The resulting acid was treated with thionyl chloride to yield
phenyl-d5 acetic chloride. Then the procedure of o-MeDBK
1
synthesis was employed to produce DBK-d10. H NMR: δ 7.32
(s, 0.03H), 7.15 (s, 0.06H), 3.72 (s, 4H). 13C NMR: δ 205.9,
133.9, 129.5, 129.2, 128.9, 128.7, 128.4, 128.0, 127.0, 126.7,
126.4, 49.1. MS m/z (rel int): 220.2 (M+, 7.6), 123 (7.4), 96
(100).
1-[2-Meth yl(p h en yl-d 4)]-3-p h en yl-2-p r op a n on e (o-MeD-
BK-d 4) (1-oMe-d 4) was synthesized as follows: To a 120 mL
of LiAlH4 solution in THF (1.0 M, 0.12 mol) was added 10 g of
phthalic-d4 acid (0.06 mol) to yield 8 g of (2-hydroxymethyl-
phenyl-d4)methanol (0.056 mol, 94% yield). The latter was
stirred with 25 mL of hydrochloric acid (37 wt % aqueous
solution) and extracted with ether. After drying over MgSO4
the solvent was removed under vacuum and the product was
purified by flash column chromatography to yield 6 g of (2-
chloromethylphenyl-d4)methanol (0.037 mol, 67% yield). The
latter was treated with 40 mL of LiAlH4 solution in THF (1.0
M, 0.04 mol) to yield 5 g of (2-methylphenyl-d4)methanol
(0.0396 mol, 107%). The latter was stirred with 20 mL of
hydrochloric acid (37 wt % aqueous solution) to yield 4.1 g of
(2-methylphenyl-d4)methyl chloride (0.028 mol, 72% yield).
Then the procedure for o-MeDBK synthesis was followed to
From time-resolved laser flash experiments it has been
shown that radical-radical recombination on the exter-
nal surface of ZSM-5 occurs in microseconds,15,16 a time
scale typical of radical-radical reactions in solution.
Thus, the rates of radical-radical reactions on the
external surface (microseconds) can be orders of magni-
tude faster than reaction of the same radicals on the
internal surface (minutes to hours), demonstrating the
strong control of supramolecular steric effects on the
dynamics of radical-radical reactivity.
1
yield the final product 1-oMe-d4. H NMR (ppm): 7.35-7.25
(m, 3H), 7. 17-7.14 (m, 2H), 3.74 (s, 2H), 3.70 (s, 2H), 2.13 (s,
3H). 13C NMR: δ 205.8, 137.0, 134.2, 132.9, 129.6, 128.8, 127.2,
49.2, 47.4, 19.6. MS m/z (rel int): 228.15 (M+, 3.0), 137 (4.0),
136 (6.7), 119 (2.4), 118 (4.6), 110 (8.8), 109 (100), 108 (7.2),
92 (24), 91 (66), 82 (7.2).
1-(2-Meth ylp h en yl)-3-(p h en yl-d 5)-2-p r op a n on e (o-MeD-
BK-d 5) (1-oMe-d 5) was synthesized following the procedures
of 1-oMe-d4 synthesis, using phthalic acid instead of phthalic-
d4 acid, and benzyl-2,3,4,5,6-d5 chloride instead of benzyl
Exp er im en ta l Section
1
chloride. H NMR (ppm): 7.21-7.15 (m, 3H), 7. 11-7.05 (m,
2H), 3.73 (s, 2H), 3.71 (s, 2H), 2.12 (s, 3H). 13C NMR: δ 205.8,
137.1, 134.0, 133.0, 130.6, 127.5, 126.3, 49.1, 47.4, 19.6. MS
m/z (rel int): 229.15 (M+, 1.8), 133 (3.9), 132 (5.8), 124 (1.7),
123 (4.3), 105 (100), 97 (22), 96 (76.8), 77 (16.7).
Gen er a l Meth od s. Gas chromatography was performed on
a HP5890 gas chromatograph equipped with a 29 mHP-1
capillary column and FID detector. 1H NMR (300 MHz and
400 MHz) and 13C NMR (75 and 100 MHz) were recorded on
a Bruker NMR spectrometer using CDCl3 (TMS) as solvent.
1,3-Bis(2-m eth ylp h en yl)-2-p r op a n on e (o,o′-d iMeDBK)
(1-o,o′DiMe) was synthesized following a published procedure
for the synthesis of 1,3-bis(4-methylphenyl)-2-propanone using
2-(methylphenyl)acetic acid instead of 4-(methylphenyl)acetic
acid.29 The product was purified by column chromatography
with hexane/ether (19:1 v/v) and was recrystallized from
GC-MS (EI) spectra were recorded on
a HP5792 mass
selective detector, connected to a HP5890II gas chromatograph
with a 20 mHP-5 capillary column. The MS detector was
calibrated with perfluorotributylamine (PFTBA) prior to data
acquisition. EPR spectra were recorded on a Bruker ESP 300
spectrometer interfaced to a computer with Bruker ESP1600
system software.
1
hexane/ether as white crystals. H NMR: (ppm) δ 7.22-7.13
(3H, m), 7.09-7.06 (2H, m), 3.73 (4H, s), 2.17 (6H, s). 13C
NMR: (ppm) δ 205.9, 137.1, 133.1, 130.7, 130.6, 127.6, 126.4,
47.5, 19.8. MS m/z (rel int): 238.20 (M+, 1.8), 132 (3.4), 133
(2.1), 105 (100), 77 (18).
All chemicals, unless noted otherwise, were obtained from
Aldrich and used as received.
Ma ter ia ls. Zeolites. Sodium-exchanged ZSM-5 (Si/Al ) 20)
was a Chemie Uetikon product obtained as a generous gift
from Dr. V. Ramamurthy, Department of Chemistry, Tulane
University, New Orleans, LA. The mean crystal lateral dimen-
sion of the silicalite was ca. 1 µm and that of ZSM-5 was ca.
0.3 µm as characterized by scanning electron microscopy. The
external surface area of ZSM-5 was determined by the mercury
porosimetry method28 to be 16 m2/g.
2,4-Dip h en ylp en ta n -3-on e (2) was synthesized with iodo-
methane and 1 following a published procedure25 The meso:
(29) Turro, N. J .; Weed, G. C. J . Am. Chem. Soc. 1983, 105, 1861-
1868.
(30) Sokol, P. E. In Organic Syntheses; Baumgarten, H. E., Ed.; J ohn
Wiley: New York, 1973; Coll. Vol. 5, p 706.
(31) Nystrom, R. F.; Brown, W. G. J . Am. Chem. Soc. 1947, 69,
1197-1199.
(32) Ruhoff, J . R. In Organic Syntheses; Blatt, A. H., Ed.; J ohn
Wiley: New York, 1943; Coll. Vol. 2, p 292.
(28) Abrams, L.; Keane, M.; Sonnichsen, G. C. J . Catal. 1989, 115,
410-419.

2618 J . Org. Chem., Vol. 67, No. 8, 2002
Turro et al.
d,l ratio was determined by 1H NMR to be 92:8 comparing the
chemical shifts of the diastereomerically pure meso- and d,l
-2. 1H NMR (ppm): δ 7.37-6.95 (m, 10H) (consisting of 7.37-
7.27 and 7.22-7.13, m, 6.24H and 7.01-6.95, m, 3.67H), 4.0-
3.7 (m, 2H) (consisting of two quartets: 3.88, 1.81H, J ) 7.04
Hz and 3.77, 0.16H, J ) 6.88 Hz), 1.4-1.1 (m, 6H) (consisting
of two doublets: 1.38, 5.68H, J ) 7.04 Hz and 1.25, 0.48H, J
) 6.87 Hz). 13C NMR: δ 211.5, 210.5, 140.2, 128.6, 128.2, 127.0,
51.7, 18.5. MS m/z (rel int): 238.05 (M+, 3.3), 133 (7.3), 105
(100), 77 (16).
2-(2-Meth ylp h en yl)-4-p h en ylp en ta n -3-on e (2-oMe) was
synthesized by 1-oMe with iodomethane following the same
procedure as for 2. 1H NMR (ppm): δ 7.34-6.93 (m, 9H), 4.2-
3.4 (m, 2H) (consisting of 4 quartets: 4.06, 0.56H, J ) 6.97
Hz; 3.91, 0.35H, J ) 6.82 Hz; 3.79, 0.55H, J ) 7.07 Hz; 3.65,
0.37H, J ) 6.86 Hz), 2.3-1.9 (m, 3H) (consisting of two
singlets: 2.20, 0.88H; 2.06, 1.61H), 1.45-1.10 (m, 6H) (consist-
ing of four doublets: 1.38, 1.98H, J ) 7.10 Hz; 1.34, 1.69H, J
) 7.01 Hz; 1.28, 1.11H, J ) 6.90 Hz; 1.21, 0.92H, J ) 6.82
Hz). 13C NMR: δ 211.8, 211.2, 141.2, 140.4, 139.6, 138.5, 135.9,
135.8, 131.0, 130.6, 129.1, 128.4, 128.1, 128.0, 127.8, 127.2,
127.2, 127.0, 126.9, 126.9, 126.3, 51.4, 51.0, 48.0, 47.0, 19.6,
19.5, 19.0, 18.5, 17.9, 17.4. MS m/z (rel int): 252.10 (M+, 1.7),
147 (5.0), 146 (4.3), 133 (1.4), 132 (1.4), 119 (100), 105 (52.5),
91 (16), 77 (16).
2,4-Di(p h en yl-d 5)p en ta n -3-on e (2-d 10) was synthesized by
1-d 10 and iodomethane following the same procedure as for 2.
1H NMR (ppm): δ 7.20-7.16 (m, 0.09H), 6.98 (s, 0.06H), 3.92-
3.73 (m, 2H) (consisting of two overlapping quartets: 3.92-
3.84, 1.82H, J ) 7.04 Hz, 6.93 Hz and the third quartet: 3.77,
0.16H, J ) 6.86 Hz), 1.39-1.24 (m, 6H) (consisting of two
overlapping doublets: 1.39-1.35, 5.58H, J ) 7.06 Hz, 6.85 Hz
and the third doublet: 1.25, 0.49H, J ) 6.88 Hz). 13C NMR: δ
211.5, 210.5, 208.2, 140.9, 140.1, 134.3, 128.3, 128.0, 127.7,
127.4, 126.8, 126.4, 126.1, 51.6, 18.5. MS m/z (rel int): 248.25
(M+, 1.6), 138 (4.8), 110 (100), 83 (11).
3,5-Dip h en ylh ep ta n -4-on e (3) was synthesized by 1 and
iodoethane following the same procedure as for 2. It was
obtained as white crystals. 1H NMR: δ 7.16-7.12 (m, 6H),
6.96-6.93 (m, 4H), 3.60 (t, 2H), 2.10-1.94 (m, 2H), 1.75-1.60
(m, 2H), 0.78 (t, 6H). 13C NMR: δ 210.9, 138.7, 128.7, 128.5,
127.0, 60.7, 26.2, 12.4. MS m/z (rel int): 266.15 (M+, 3.2), 147
(7.6), 119 (76), 91 (100).
photolyzed sample was (within 2 min) placed into the ESR
cavity for ESR measurement. For peroxy radical detection, the
samples were open to the air briefly after photolysis and
degassed to 10-4 Torr (ca. 3 min), and then the ESR spectra
were acquired. For NO scavenging, the samples were im-
mediately connected to the vacuum line after photolysis,
degassed, and equilibrated with the NO gas (previously
deaerated) at 2 Torr. The EPR spectra of the sample were
taken, and then the sample was photolyzed for another 8 min
and the EPR spectra were taken again for comparison.
P r od u ct An a lysis. The products of photolysis of 2 were
extracted from zeolite by stirring the zeolite with benzene (0.5
mL per 100 mg of zeolite) overnight. Then 4,4′-dimethylbenzo-
phenone was added to the slurry as an internal standard. The
supernatant was analyzed by GC and GC-MS analysis. The
identity of products was determined by comparison to com-
mercial samples of 2,3-diphenylethane, styrene, and ethyl-
benzene on GC and on GC-MS. The ratios of each compound
were determined by the integral of corresponding peaks on
the GC trace assuming the same detector response. The mass
balance was calculated by combining the amount of the
starting ketones and all the products from GC traces, with
the starting ketones calibrated against the internal standard.
Sp ectr a l Sim u la tion s. Computer simulations of the EPR
spectra were performed on the Simfonia program (Bruker),
assuming partially hindered rotation of the radicals, which
averages the anisotropies but allows a significant broadening
of the hyperfine lines. The simulation of EPR spectra of B and
R-MeB radicals was based on published proton-electron
hyperfine coupling constants.33 The deuterium-electron hyper-
fine coupling constants for the phenyl ring perdeuterated B
and R-MeB radicals were assumed to be about one-fifth of its
protic counterpart.
The simulations of EPR spectra of R-MeBNit were per-
formed with the following parameters. For g factor, gxx ) 2.008,
gyy ) 2.006, gzz ) 2.002 were assumed. For Aii(N), Axx ) 4 G,
Ayy ) 4 G, and Azz ) 35 G were obtained from calculation.
Coupling constants from spectral simulations were 18 and 27
G for the two R-H’s and 0.5 G for the protons in R-Me groups.
The line width was set to be 3.5 G.
For the simulation of the EPR spectrum of R-MeBO2, two
components in the ratio of 1:1 were independently computed
and added to provide a good fit with the experimental
6,8-Dip h en yltr id eca n -7-on e (4) was synthesized by 1 and
1-iodopentane following the same procedure as for 2. It was
obtained as an off-white liquid at room temperature. 1H NMR:
δ 7.40-7.10 (m, 6H), 6.96-6.93 (m, 4H), 3.68 (t, 2H, J ) 7.41
Hz), 1.99-1.91 (m, 2H), 1.69-1.57 (m, 2H), 1.3-1.0 (m, 12H),
0.82 (t, 6H, J ) 6.6 Hz,). 13C NMR: δ 210.9, 139.0, 128.6, 128.5,
126.9, 58.8, 33.0, 31.8, 27.4, 22.6, 14.2. MS m/z (rel int): 350.25
(M+, 1.3), 189 (2.5), 161 (53), 91 (100).
All of the purified ketones were above 97% pure by GC
analysis after column chromatography and/or recrystallization.
Typical impurities in the 2 series are the corresponding
compounds from mono- and trialkylation.
spectrum. The simulation of one component involves gxx
)
2.0015, gyy ) 2.009, gzz ) 2.0029 as g factor, and a line width
of 4.3G for adsorption along the X and Y axes and 8.5 G along
the Z axis. The simulation of the second component involves
gxx ) 2.001, gyy ) 2.0084, gzz ) 2.023 as g factor and a line
width of 4.7 G for adsorption along the X and Y axes and 10.0
G along the Z axis.
The simulations of EPR spectra of BO2 employed gxx
)
2.0033, gyy ) 2.0087, gzz ) 2.029 as g factor and a line width
of 10.0 G for adsorption along X and Y axes and 12.0 G along
the Z axis.
Sa m p le P r ep a r a tion . ZSM-5 was activated in a furnace
at 500 °C for 2 h and was cooled in a desiccator to room
temperature before use.
A typical sample preparation was as follows. A specific
loading was achieved by stirring a weighed amount of ketone
in a solution of 0.8 mL of isooctane with 300 mg of zeolite for
an hour. After evaporation of the solvent under argon flow,
the samples were dried under vacuum.
GC analysis was employed to determine the photolysis
products with 1-phenyldodecane as an internal standard.
P h otolysis. Typical photolysis conditions employed 100 mg
of dried zeolite samples in a quartz cell equipped with a
sidearm for vacuum degassing. After pumping to 1 × 10-5 Torr,
the samples were irradiated for 8 min with a 450 W mercury
lamp equipped with a chromate filter solution. The resulting
Ack n ow led gm en t. The authors at Columbia thank
the National Science Foundation for its generous sup-
port of this research (Grant No. CHE 98-12676). The
research was supported in part by the Department of
Energy and the National Science Foundation by Grant
No. CHE 98-10367 to the Environmental Molecular
Sciences Institute (EMSI) at Columbia University.
M.F.O. thanks the Italian Ministero dell’Universita e
della Ricerca Scientifica (MURST) for financial support.
J O011047L
(33) Conradi, M. S.; Zeldes, H.; Livingston, R. J . Phys. Chem. 1979,
83, 633-639.