Photolysis of dibenzyl ketones sorbed on MFI zeolites in the presence of spectator molecules: Cage effects, kinetics, and external surface sites characterization

Article abstract of DOI:10.1021/ja801487v

Over the past two decades, the photolytic reactions of dibenzyl ketones sorbed on zeolites have been investigated. The reported results are consistent with a supramolecular model that takes into account the physical and chemical nature of the structure of

Full text of DOI:10.1021/ja801487v

Published on Web 08/01/2008
Photolysis of Dibenzyl Ketones Sorbed on MFI Zeolites in the
Presence of Spectator Molecules: Cage Effects, Kinetics, and
External Surface Sites Characterization#
Alberto Moscatelli,§ Zhiqiang Liu,§,† Xuegong Lei,§ Joanne Dyer,§,‡ Lloyd Abrams,|
M. Francesca Ottaviani, and Nicholas J. Turro*,§
Department of Chemistry, Columbia UniVersity, New York, New York 10027, Central Research
and DeVelopment Department, DuPont Company, Wilmington, Delaware 19880, and Istituto di
Scienze Chimiche, UniVersita` di Urbino, 61029 Urbino (PU), Italy
Received February 27, 2008; E-mail: njt3@columbia.edu
Abstract: Over the past two decades, the photolytic reactions of dibenzyl ketones sorbed on zeolites have
been investigated. The reported results are consistent with a supramolecular model that takes into account
the physical and chemical nature of the structure of the zeolites and their effect on the reactive radical
intermediates produced by photolysis of adsorbed molecules. The model incorporates various phenomena
such as surface coverage, external and internal sorption, surface diffusion, radical sieving, and the resulting
product distributions. This account reports direct evidence for the validation of the model through FT-IR
spectroscopy and through a new method for “titrating” the binding sites via EPR spectroscopy. It is shown
that it is possible to adjust and modulate the photolytic product distribution by varying the parameters of
the system. The effects of co-adsorbed spectator molecules with different polarities, namely water, pyridine,
and benzene, on the photolysis of o-methyldibenzyl ketone and dibenzyl ketone sorbed on MFI zeolites is
examined. This study provides insights into a displacement mechanism caused by spectator molecules
and further demonstrates how the product distribution of photolysis of sorbed ketones can be controlled.
The kinetics of persistent radicals formed by photolysis of ketones sorbed on zeolites is directly monitored
over time by EPR, providing a measure of the lifetime of these reactive organic intermediates. Finally,
measurement of Langmuir isotherms was employed to provide classical evidence for the model.
Introduction
of supramolecular chemistry. Even if not directly responsible
for the bond-making and bond-breaking process, the chemistry
Besides the variety of applications in industry and agriculture
that have benefited from the use of zeolites,1–4 these alumino-
silicate crystals have also attracted the interest of chemists for
basic science applications.5–8 In particular, their channel systems
and cavities have been useful as nanoreactors for chemical
reactions.9 External molecules sorb on the zeolite framework,
forming intermolecular complexes that are addressed by the field
of a reaction can be controlled by modulating the parameters
of such intermolecular interactions. Normally, in the case of
radical reactions in solution, statistics of radical-radical com-
bination dominate the product distribution. However, using
zeolites, the outcome of a radical-radical reaction can be
controlled by supramolecular factors such as the binding strength
and the location of the sorbed radical, as well as its available
diffusion options. Through an understanding of the supramo-
lecular structure and dynamics of the radical@zeolite10 system,
it is possible to control the product distribution of radical
reactions and to diversify the photochemistry of isomeric
precursors that would otherwise exhibit the same photochemistry
in the solution phase.11,12
# Contribution no. 8853 from DuPont Co.
§ Columbia University.
| DuPont Company.
Universita` di Urbino.
† Present address: Colgate-Palmolive Co., Piscataway, NJ 08855.
‡ Present address: Art Conservation Research Center, Carnegie-Mellon
University, Pittsburgh, PA 15219.
Figure 1 shows the size/shape dimensions of a series of
dibenzyl ketones (DBKs) used as guest molecules (Chart 1) and
the MFI zeolite-type framework used as the host system.13 The
cross section of o-MeDBK and o,o′-diMeDBK is ∼6.6 Å,
(1) Csicsery, S. M. Pure Appl. Chem. 1986, 58, 841.
(2) Mumpton, F. A. Proc. Natl. Acad. Sci. U.S.A. 1999, 96, 3463.
(3) Newsam, J. M. Science 1986, 231, 1093.
(4) Breck, D. W. Zeolite Molecular SieVes: Structure, Chemistry, and Use;
Krieger Publishing Co.: Melbourne, FL, 1984.
(5) Calzaferri, G.; Huber, S.; Maas, H.; Minkowski, C. Angew. Chem.,
Int. Ed. 2003, 42, 3732.
(10) The nomenclature probe@zeolite is adopted to represent the supramo-
lecular complex of a guest probe molecule sorbed on zeolite in analogy
to the common guest@host representation of supramolecular systems.
(11) Turro, N. J.; Cheng, C. C.; Lei, X. G.; Flanigen, E. M. J. Am. Chem.
Soc. 1985, 107, 3739.
(6) Sivaguru, J.; Shailaja, J.; Ramamurthy, V. In Handbook of Zeolite
Science and Technology; Auerbach, S. M., Carrado, K. A.;, Dutta,
P. K., Eds.; Marcel Dekker: New York, 2003.
(7) Tung, C.-H.; Wu, L.-Z.; Zhang, L.-P.; Chen, B. Acc. Chem. Res. 2003,
36, 39.
(12) Garciagaribay, M. A.; Zhang, Z. Y.; Turro, N. J. J. Am. Chem. Soc.
1991, 113, 6212.
(8) Chretien, M. N. Pure Appl. Chem. 2007, 79, 1.
(9) Tretyakov, Y. D.; Lukashin, A. V.; Eliseev, A. A. Russ. Chem. ReV.
2004, 73, 899.
(13) Baerlocher, C.; Meier, W. M.; Olson, D. H. Atlas of Zeolite Framework
Types; Elsevier: Amsterdam, 2001.
9
11344 J. AM. CHEM. SOC. 2008, 130, 11344–11354
10.1021/ja801487v CCC: $40.75
2008 American Chemical Society

Photolysis of Dibenzyl Ketones@MFI Zeolite
A R T I C L E S
Figure 2. Schematic representation of photolysis of DBKs@MFI zeolite.
(a) p-MeDBK or DBK is absorbed inside the zeolite channel system and,
upon photolysis, they form persistent A• and B• benzyl radicals. As discussed
in the text, the product distribution, measured by cage effect, changes
dramatically upon adding spectator molecules that displace p-MeDBK or
DBK toward the external surface. (b) o-MeDBK can only adsorb on the
external surface, either by van der Waals interactions, hydrogen-bonding,
or intercalation inside the pores with the unsubstituted benzene ring.
Photolysis of o-MeDBK@MFI produces persistent benzyl radical B• that
sieves inside the pore system and transient radical A• that readily combines
to form A-A product.
Figure 1. MFI zeolite framework and kinetic diameter of p-MeDBK (and
DBK) and o-MeDBK (and o,o′-diMeDBK). Pore-opening sizes are also
reported, according to ref 13. The external surface is composed of strong
binding sites located at the pore openings and weak binding sites located
in the external framework of the zeolite between the pores.
Chart 1. Dibenzyl Ketones Investigated in the Present Work
evidence for the sites of adsorbed molecules.17 It was possible,
for instance, to characterize quantitatively the strong binding
sites18–20 as well as to elucidate the high mobility of adsorbate
molecules on the zeolite surface.20–22 In this paper, these studies
are extended to provide direct evidence for validation of the
model for the external surface of zeolite by FT-IR spectroscopy.
In addition, a new method for “titrating” the strong binding sites
by EPR spectroscopy is developed. The binding energies
between selected adsorbate molecules and the external surface
of zeolites are evaluated by the Langmuir isotherm method.
whereas for DBK and p-MeDBK it is ∼5.3 Å.14 Since the
diameter of the pore opening for MFI zeolites is ∼5.5 Å, DBK
and p-MeDBK can be absorbed inside the zeolite pore openings
and diffuse into the internal surface. By contrast, o-MeDBK
can be adsorbed in the pore opening but cannot diffuse into the
internal surface because of the ortho-methyl substitution. Finally,
o,o′-diMeDBK cannot absorb in the pore openings and therefore
must be entirely adsorbed on the external surface.
The adsorption sites of the initially adsorbed ketones and the
radicals produced by photolysis are shown schematically in
Figure 2. The geminate radical pair (A• + B•) is produced by
photodecomposition followed by rapid decarbonylation (Sch-
eme 1). There are several possibilities for product formation,
depending on the initial location of the ketone on the zeolite
(Figure 2). According to the model, three possible cases are
assumed: (i) the ketone is entirely inside the internal surface of
the zeolite (absorption); (ii) the ketone is interacting entirely
on the external surface of the zeolite (adsorption); and (iii) the
ketone is partially on the external surface and partially in the
pores leading to the internal surface (sorption). For case (i), the
geminate radical pair (A• + B•) is not expected to be able to
diffuse apart readily because both partners are constrained by
The paradigm or model employed to rationalize the results
of photolysis of these ketones, along with the scientific facts
that led to its development and support it, is described in detail
in a recent review.15 The model takes into account the sieving
ability of the zeolites and considers the size/shape/diffusion
characteristics of the photolytic fragments in terms of two
different initial surface-binding sites of the external surface
(Figure 1): strong binding sites, located at the pore openings,
at which adsorbed molecules interact by hydrogen-bonding and/
or partial intercalation, and weak binding sites, located at the
surface “non-openings” of the zeolite framework, at which
hydrogen-bonding and/or van der Waals interactions take place.
Strong binding sites are occupied first, whereas weak binding
sites are occupied when the former become saturated.16 Solid-
state NMR and EPR spectroscopies have provided direct
(17) Turro, N. J.; Lei, X. G.; Li, W.; Liu, Z. Q.; Ottaviani, M. F. J. Am.
Chem. Soc. 2000, 122, 12571.
(18) Liu, Z. Q.; Ottaviani, M. F.; Abrams, L.; Lei, X. G.; Turro, N. J. J.
Phys. Chem. A 2004, 108, 8040.
(19) Ottaviani, M. F.; Lei, X. G.; Liu, Z. Q.; Turro, N. J. J. Phys. Chem.
B 2001, 105, 7954.
(20) Turro, N. J.; Lei, X. G.; Li, W.; Liu, Z. Q.; McDermott, A.; Ottaviani,
M. F.; Abrams, L. J. Am. Chem. Soc. 2000, 122, 11649.
(21) Turro, N. J.; Lei, X. G.; Li, W.; McDermott, A.; Abrams, L.; Ottaviani,
M. F.; Beard, H. S. Chem. Commun. 1998, 695.
(14) Li, W. Ph.D. Dissertation, Columbia University, New York, NY, 1999.
(15) Turro, N. J. Acc. Chem. Res. 2000, 33, 637.
(22) Turro, N. J.; McDermott, A.; Lei, X. G.; Abrams, L.; Ottaviani, M. F.;
Beard, H. S.; Houk, K. N.; Beno, B. R.; Lee, P. S.; Li, W. Chem.
Commun. 1998, 697.
(16) Hirano, T.; Li, W.; Abrams, L.; Krusic, P. J.; Ottaviani, M. F.; Turro,
N. J. J. Org. Chem. 2000, 65, 1319.
9
J. AM. CHEM. SOC. VOL. 130, NO. 34, 2008 11345

A R T I C L E S
Moscatelli et al.
Scheme 1. Reaction Scheme for the Photolysis of Dibenzyl Ketones and Recombination Cases in Supramolecular Systems
the walls of the host structure. Thus, the main product of the
photolysis is the geminate coupling product A-B, corresponding
to a limiting cage effect of +100%. For case (ii), the geminate
radical pair (A• + B•) may be able to diffuse rapidly apart,
considering the expected weak binding of the radicals by the
external surface. In this case, a statistical set of recombination
products A-A, A-B, and B-B is expected in the limit of
complete and irreversible diffusional separation of the geminate
pair, corresponding to a limiting cage effect of 0%. For case
(iii), the partners of the geminate radical pair (A• + B•) are
sorbed in two distinct locations of the zeolite structure: one part-
ner is weakly bound to the external surface and is too large to
diffuse into the pores of the internal surface, and the other
partner is adsorbed in a pore strategically poised to be absorbed
into the internal surface. In this case, in the limit where diffusion
into the internal surface is fast and irreversible compared to
geminate combination, the major products are expected to be
A-A and B-B, corresponding to a limiting cage effect of
-100%.
onto the external surface, followed by a size/shape-selective
diffusion inside the internal pores.24,25 The impact of this first
step on the performance of zeolites depends on the particle size
of the zeolite crystal, because the external surface area (Sext
)
increases faster than the internal surface area as particle size
decreases. Simple geometrical reasoning for a model of a cubic
particle of side l yields eq 2,
Sext
Stot
1
l
)
(2)
where Stot is the total surface area of the zeolite.
Results and Discussion
Photolysis of DBK@MFI: Cage Effect and Kinetics. At low
loadings, water acting as a co-adsorbed “spectator” molecule
displaces DBK from the hydrophilic, Al-containing sites located
in the internal channels, as demonstrated by the decrease of the
cage effect with increasing water molecule loading.26 However,
this is strictly true only for zeolites with high Al content (Si/Al
< 20). The cage effect versus water loading plot displays a
minimum for zeolites with higher Si-Al ratio.27 This was
rationalized by supposing that the displacement is effective up
to a certain amount of water, corresponding to the saturation of
the Al sites. Additional water probably forms globules within
the zeolite channels, thereby restricting migration of radicals
and fostering recombination (resulting in a high cage effect).
However, ketones adsorbed on hydrophobic sites inside the
channels system are not displaced by water. In fact, DBK
remains trapped in the hydrophobic portions of the channels as
the cage effect increases after reaching a minimum. Figure 3
shows plots of the cage effect versus water loading for MFI
zeolite type with different Si-Al ratios. Only for very hydro-
phobic sites with high aluminum contents, such as Si/Al ) 13,
does the cage effect decrease monotonically with water loading.
For Si/Al ratios >13, a minimum is found; its position shifts
toward lower water loadings with decreasing Al content.
Quantitatively, there exists a good correlation (Table 1) between
the water content calculated assuming a 6:1 ratio of H2O to Al
and the percentage of water at the minimum of the cage effect
curve of Figure 3; this explains the cage effect behavior in terms
of the Al content of the zeolite.
Equation 1 defines a “cage effect” (%CE) for the three cases
above, based on measurement of the product distribution of
A-A, A-B, and B-B. Thus, working backward from the
product distribution, the initial site of the ketone probe can be
computed from simple measurement of the eventual product
distribution produced by the initially geminate radical pair (A•
+ B•).
AB - (AA + BB)
AA + AB + BB
%CE )
(1)
It is possible to adjust and modulate the photolytic product
distribution by changing various parameters of the system, such
as Al content of zeolite, charge-compensating counterion,
loading of adsorbed molecule, and amount of co-adsorbed
spectator molecules. In this paper, the effect of co-adsorbed
spectator molecules with different polarity, namely water,
pyridine, and benzene, on the photolysis of DBKs@MFI is
investigated. This study permits the elucidation of the displace-
ment mechanism by spectator molecules and further demon-
strates the ability to modify product distribution and the cage
effect. In addition, the kinetics of persistent radicals formed by
photolysis is followed over time by continuous wave EPR (cw-
EPR) to measure their lifetimes.23
The importance of investigating the external surface of
zeolites resides also in the fact that the first step for the
interaction of any guest molecule with zeolites is the adsorption
(24) Gilson, J. P.; Derouane, E. G. J. Catal. 1984, 88, 538.
(25) Kennedy, C. R.; Lapierre, R. B.; Pereira, C. J.; Mikovsky, R. J. Ind.
Eng. Chem. Res. 1991, 30, 12.
(26) Turro, N. J.; Wan, P. J. Am. Chem. Soc. 1985, 107, 678.
(27) Turro, N. J.; Cheng, C. C.; Abrams, L.; Corbin, D. R. J. Am. Chem.
Soc. 1987, 109, 2449.
(23) Turro, N. J.; Jockusch, S.; Lei, X. G. J. Org. Chem. 2002, 67, 5779.
9
11346 J. AM. CHEM. SOC. VOL. 130, NO. 34, 2008

Photolysis of Dibenzyl Ketones@MFI Zeolite
A R T I C L E S
Figure 3. Plot of % cage effect versus % water loading for ZSM-5 zeolite
with different Si-Al ratios. The polynomial fitting curve is only for visual
purposes and has no physical meaning.
Table 1. Percentage of Water Absorbed in the MFI Zeolites,
Evaluated by Cage Effect (CE) Measurementsa
Si/Al
x
wt % H2O (6:1)
CE minimum (wt % H2O)
13
20
50
6.9
4.5
1.9
13
8.3
3.5
0.40
8.5 ( 0.5
3 ( 0.5
1 ( 0.5
450
0.21
13
a Unit-cell formula: NaxSi(96-x)AlxO192
.
The minima of Figure 3
agree quite well with the calculated value for a H2O-Al ratio of
6:1.27,56
Figure 5. EPR spectra after photolysis of (a) 1% loading DBK@MFI (Si/
Al ) 20) and (b) 1% loading of o-MeDBK@MFI (Si/Al ) 20) with the
respective signal evolution over time. Line width and hyperfine coupling
constant with the two methylene hydrogens (aH) are also reported. Time
evolution is recorded at the top of the central line, and its value is the result
of three independent measurements.
Figure 4. Plot of % cage effect versus % pyridine loading for ZSM-5
zeolite with different Si-Al ratios. The polynomial fitting curve is only for
visual purposes and has no physical meaning.
of DBK@MFI (Si/Al ) 20) and o-MeDBK@MFI (Si/Al ) 20)
are reported: both show a three-line spectrum as expected from
hyperfine interaction of the unpaired electron with the two
methylene hydrogens.31 Under these conditions, the benzyl
radical half-lifetime is 42 ( 14 s. As a benchmark, diffusion-
controlled radical-radical reactions in nonviscous solvents
results in radical lifetimes of the order of 10-6 s. Such short-
lived radicals cannot be easily detected by cw-EPR and are
termed “transient”. Since the benzyl radicals that carry the signal
of Figure 5 are many orders of magnitude longer lived, they
are termed “persistent”.
The same qualitative results as found for water are also
observed for pyridine as a spectator molecule (Figure 4).
Pyridine exhibits strong binding to the Al-containing sites28 that
enables the displacement of DBK from the hydrophilic sites,
thereby decreasing the cage effect. At high loadings, pyridine
displaces DBK into the hydrophobic portion of the zeolite
channels, inhibits radical diffusion, and raises the cage effect.
Hindered diffusion inside the zeolite channels causes the
benzyl radical pair produced by the photodecomposition of
DBK, followed by fast decarbonylation, to be persistent and
detectable by cw-EPR spectroscopy at room temperature.29,30
In Figure 5, the cw-EPR spectra upon photolysis of 1% loading
The presence of spectator molecules in the EPR spectra
resulting upon photolysis of DBK@MFI was investigated
(Figure 6). Addition of benzene or pyridine above 2% loading
(30) Pushkara Rao, V.; Zimmt, M. B.; Turro, N. J. J. Photochem. Photobiol.
A 1991, 60, 355.
(28) Scaiano, J. C.; Kaila, M.; Corrent, S. J. Phys. Chem. B 1997, 101,
8564.
(31) For these measurements, deuterated benzene rings were employed in
order to avoid complicated EPR spectra due to the hyperfine splitting
with benzene protons.
(29) Turro, N. J.; Lei, X. G.; Jockusch, S.; Li, W.; Liu, Z. Q.; Abrams, L.;
Ottaviani, M. F. J. Org. Chem. 2002, 67, 2606.
9
J. AM. CHEM. SOC. VOL. 130, NO. 34, 2008 11347

A R T I C L E S
Moscatelli et al.
concentration of spectator molecules is not obvious. It is
proposed that, at high concentrations of spectator molecules,
the latter can tightly pack within the zeolite channels and prevent
diffusion of the CO from the immediate vicinity of the benzyl
radical. An equilibrium is thus set up between the phenylacetyl
radical and the benzyl radical and CO. In this way, the
concentration of phenylacetyl radical at equilibrium is sufficient
to allow its detection. By comparison, the spectator molecule
diphenylethane (Figure 6a), which because of its large size packs
more loosely within the zeolite channels, produces only a
reduction of the benzyl radical lifetime with increasing loading,
as a result of DBK displacement from the internal channels.
Photolysis of o-MeDBK@MFI: Kinetics. During the photo-
decomposition of o-MeDBK@MFI, the smaller benzyl radical
B• is initially located at or near the pore opening at the external
surface (since o-MeDBK is initially located at the strong binding
sites at the pores). Since the kinetic diameter of B• is smaller
that the size of the pore opening, it can sieve into the channel
system and becomes persistent. On the other hand, the large
o-methylbenzyl radical (A•) possesses a kinetic cross section
larger than the pore openings and thus is unable to percolate
into the channels. As a result, A• radicals diffuse readily on the
external surface and recombine with other A• radicals, thus being
transient (Figure 2). Consequently, only the persistent B• radicals
that have sieved into the internal surface are seen in the cw-
EPR spectra after the photolysis of o-MeDBK@MFI.29 Figure
5b shows the time-dependent EPR spectrum after photolysis of
1% loading o-MeDBK@MFI (Si/Al ) 20). The three-line
spectrum validates the attribution to the benzyl radical. More-
over, o-methylbenzyl radical (A•) would be easily recognizable
by EPR due to a different hyperfine pattern. For the spectra in
Figure 5, the EPR parameters, such as the line width and the
hyperfine coupling constant (aH), are very similar. However,
the slightly smaller value of aH (a signature of a lower polarity
environment)37 for the spectrum in Figure 5a is consistent with
a more hydrophobic environment sensed by the benzyl radical
inside the internal channels, resulting from a difference in the
initial location of the benzyl radical precursors, DBK and
o-MeDBK.
Addition of 4% loading of pyridine or 6% loading of p-xylene
“quenches” the EPR spectrum completely (Figure 7). This result
is consistent with the spectator molecules readily absorbing into
the channel system, displacing the adsorbed ketones to the
external surface, and also inhibiting the sieving of the photolysis
products from the external surface to the internal surface. As a
result, the B• radicals produced by photolysis are restricted to
the external surface, transforming B• radicals from persistent
to transient.38
Titration of the Strong Binding Sites on the External
Surface of MFI Zeolite by EPR Spectroscopy. There are two
important reasons why nitroxides have been used to probe and
characterize zeolite surfaces: (i) the high sensitivity of the EPR
technique, which allows very low loadings of adsorbed probe
Figure 6. EPR spectrum after irradiation of 1% DBK@MFI (Si/Al ) 20)
in the presence of (a) diphenylethane, (b) benzene, and (c) pyridine at
different loadings.
totally “quenches” the benzyl EPR signal (Figure 6b,c). This is
explained by the model as follows: Before photolysis, as the
system is in equilibrium, the added spectator molecules dis-
place the initially sorbed DBK located in the internal channels
to the external surface. Photolysis produces benzyl radicals on
the external surface, where radical mobility is greater and
recombination occurs rapidly. As a result, the benzyl radical
that is “persistent” (long-lived) in the absence of benzene or
pyridine become “transient” (short-lived) in the presence of these
spectator molecules. This observation is in agreement with the
decrease of the cage effect with increased loading of spectator
molecules (Figures 3 and 4). Interestingly, at loadings greater
than 2% a new persistent two-line EPR spectrum appears (Figure
6b,c). The shape of the spectrum is characteristic of an acetyl
radical adsorbed on a surface.32,33 Considering the photochem-
istry of the system, the signal is assigned to the phenylacetyl
radical formed after Norrish type I photoreaction (R-cleavage)
of DBK as primary photochemical product, or from the
secondary photolysis of the rearrangement product meth-
yldeoxybenzoin. A control experiment shows that the photolysis
of 4-methylbenzyl phenyl ketone on ZSM-5 produces a super-
position of EPR spectra of benzyl and benzoyl radicals, a feature
not observed in our study. Therefore, it is concluded that the
acetyl signal in Figure 6b,c derives from the phenylacetyl
radical. Decarbonylation of phenylacetyl radicals in solution
occurs in microseconds or less.34,35 In general, also in the zeolite
media, decarbonylation occurs too fast for the cw-EPR to be
able to follow it.36 However, in the case of the Figure 6b,c,
decarbonylation of phenylacetyl radical, a unimolecular process,
is inhibited and causes the radical to be persistent. The
mechanism for the inhibition of decarbonylation by the high
(32) Fischer, H. In Magnetic Properties of Free Radicals; Hellwege, K.-
H.; Hellwege, A. M., Eds.; Landolt-Bo¨rnstein: Group II Molecules
and Radicals; Springer-Verlag: Berlin, 1965.
(37) Ottaviani, M. F.; Martini, G.; Nuti, L. Magn. Reson. Chem. 1987, 25,
897.
(38) Because the samples in Figures 6 and 7 are loaded with the same
amount of ketone and irradiated for the same amount of time, and the
spectra are acquired under the same physical and instrumental
conditions, they also provide an integrated snapshot of the kinetic
profile from 0 to 24 s from the end of the irradiation, where 24 s is
the time taken to acquire the cw-EPR spectrum. Therefore, the fact
that, at high loading of spectator molecule, the spectrum of benzyl
radical disappears indicates a shorter lifetime under such conditions,
in accordance with the model of Figure 2.
(33) Turro, N. J.; Lei, X. G.; Niu, S. F.; Liu, Z. Q.; Jockusch, S.; Ottaviani,
M. F. Org. Lett. 2000, 2, 3991.
(34) Lunazzi, L.; Ingold, K. U.; Scaiano, J. C. J. Phys. Chem. 1983, 87,
529.
(35) Turro, N. J.; Gould, I. R.; Baretz, B. H. J. Phys. Chem. 1983, 87,
531.
(36) Lei, X.; Jockusch, S.; Ottaviani, M. F.; Turro, N. J. Photochem.
Photobiol. Sci. 2003, 2, 1095.
9
11348 J. AM. CHEM. SOC. VOL. 130, NO. 34, 2008

Photolysis of Dibenzyl Ketones@MFI Zeolite
A R T I C L E S
Figure 8. 4-DPA-TEMPO EPR spectra at different loadings (black line),
with relative simulation (blue line), and schematic representation of the
loading of the external surface of zeolite in the light of the model described
in the text. The intrinsic line width (∆H) and the spin-spin exchange
frequency (ωex) used to obtain the simulated spectra are reported. They are
averages of the all probes adsorbed onto the different binding sites. These
spectra refer to the silicalite S1.
Figure 7. EPR spectrum after irradiation of 1% o-MeDBK@MFI (Si/Al
) 20) in the presence of (a) p-xylene and (b) pyridine.
Chart 2. Structures of the Nitroxide Probes Used in the Present
Work
In Figure 8, 4-DPA-TEMPO EPR spectra, at different surface
loadings, are shown (simulated spectra shown on upper traces):
line shapes are very dependent on the loading, and in particular,
the three hyperfine lines broaden to complete coalescence around
1.8% loading. A schematic interpretation of the EPR spectra in
terms of the location of the probes on the external surface is
presented to the right of each spectrum. The line shapes of the
spectra in Figure 8 for the 1.8% loading (and lower loadings)
are consistent with the partial resolution of the magnetic
dipole-dipole interaction anisotropies. This interaction depends
on the distance and on the relative orientation of the magnetic
dipoles and is modulated by the rotational diffusion motion of
the probe. The probe is expected to first bind to the strong
binding sites on the external surface. Because of this interaction,
the probe’s motion is inhibited. As a result, the dipole-dipole
interaction between electron spins does not average to zero (as
it does when there is fast isotropic motion of the probe). Under
the assumption that the broadening is dominated by dipolar
interactions between adjacent inhibited probes in strong binding
sites, the following simple relationship between the average
probe distance (d, expressed in Å) and spectral line width (∆Hd,
in G) has been developed:42–44
(on the order of 10-3% by weight)39 to be studied, and (ii) the
ability to measure the mobility and the environmental polarity
of the probes, because of the dramatic change of line shape of
their EPR spectra under hindered conditions and the sensitivity
of aH to the polarity of the environment.40 As a result, EPR
provides a very elegant method to discriminate probes adsorbed
on relatively strong binding sites from probes adsorbed on
20
relatively weak binding sites. In this paper, a direct method
to “titrate” the strong binding sites and to distinguish them from
the weak binding sites on the external surface of zeolites is
developed. In order to restrict variables that might influence
the pertinent parameters, monodisperse MFI silicalite crystals
(synthesized as per the Experimental Section) are used. These
crystals possess virtually no Al (Si/Al > 1000). Both 4-DPA-
TEMPO and 4-oxo-TEMPO (Chart 2) are employed as EPR
probes. These two probes were chosen because they have a low
and a high interacting strength toward the surface sites,
respectively.18 Furthermore, the TEMPO moiety, with its kinetic
diameter of 11 Å, adsorbs exclusively on the external surface
of MFI zeolite crystals.41
1
∆Hd ) 3 × 104
(3)
3
(d )
Equation 3 provides a means of determining the distance
between adjacent spin probes on the external surface of silicalite.
However, since the dipole-dipole interaction only accounts
for the line broadening and does not explain the observed
coalescence of the hyperfine lines, it is necessary to take into
consideration the spin-spin exchange mechanism between
probes adsorbed on the weak binding sites of the external surface
of zeolite. This interaction occurs when the probes undergo free
diffusion and collide at a certain frequency (as in the case of
very concentrated solutions), thus undergoing overlapping
(39) For such low loaded zeolites, in order to avoid exponential numbers
in the percentage, it is usually preferable to use millimolality (mm,
millimoles of adsorbate per gram of zeolite). However, in this paper,
we choose to keep percentage for consistency and comparison
(40) Spin Labeling: Theory and Applications; Berliner, L. J., Ed.; Academic
Press: New York, 1976.
(41) Khodakov, A. Y.; Kustov, L. M.; Bondarenko, T. N.; Dergachev, A. A.;
Kazansky, V. B.; Minachev, K. M.; Borbely, G.; Beyer, H. K. Zeolites
1990, 10, 603.
(42) Sackmann, E.; Trauble, H. J. Am. Chem. Soc. 1972, 94, 4482.
(43) Sackmann, E.; Trauble, H. J. Am. Chem. Soc. 1972, 94, 4492.
(44) Trauble, H.; Sackmann, E. J. Am. Chem. Soc. 1972, 94, 4499.
9
J. AM. CHEM. SOC. VOL. 130, NO. 34, 2008 11349

A R T I C L E S
Moscatelli et al.
Table 2. Critical Loading Values (c*) and Corresponding Average
Probe Distance for Different MFI Zeolites
c* (wt %)
distance at c*
(Å)
symbola Si/Alb ESAc (m2 ·g-1
)
EPR probe
measd calcd
S1
S2
S3
S4
S5
S6
S7
S7
C1
C2
C3
>1000
>1000
>1000
>1000
>1000
>1000
>1000
>1000
1000
55.1
20.3
9.27
5.71
29.8
3.75
2.75
2.75
4-DPA-TEMPO 1.83 1.89
4-DPA-TEMPO 0.66 0.70
4-DPA-TEMPO 0.35 0.32
4-DPA-TEMPO 0.17 0.20
4-DPA-TEMPO 0.92 1.02
4-DPA-TEMPO 0.04 0.13
4-DPA-TEMPO 0.02 0.05
4-oxo-TEMPO 0.02 0.05
4-oxo-TEMPO 0.017
13.4
13.3
13.3
13.0
12.7
13.2
13.1
12.7
13.2
13.0
12.9
400
90
4-oxo-TEMPO 0.013
4-oxo-TEMPO 0.051
a Home-made silicalites have symbol Sn, and commercial zeolites
have symbol Cn. b Si/Al > 1000 corresponds to silicalite crystals.
c Measured by mercury porosimetry.
orbitals. In order to distinguish the portion of line width that
originates from the dipole-dipole mechanism and the spin
exchange, a computer-aided simulation of the adsorbed nitroxide
probes spectra is performed. The simulation procedure is
described in the Experimental Section. Some representative
simulation output spectra are shown in Figure 8 as the upper
traces, along with the corresponding values of line broadening
(∆H) and spin exchange frequency (ωex).
From the simulations, at low surface coverage (0.037%), the
EPR spectrum of 4-DPA-TEMPO displays a line shape that is
characteristic of nitroxide probes undergoing slow rotational
motion. Furthermore, the spectrum can be simulated without
inclusion of any spin-exchange interaction and is only slightly
broadened by the dipole-dipole interaction. For such a spec-
trum, ∆H ) 1.5 G, which converts to a probe distance of 27 Å
using eq 3.
By tracking the EPR parameters as a function of loading,
the spectra are found to exhibit a qualitative change in
parameters at a loading of ∼1.8%. This change is considered a
signature of the saturation of the strong binding sites and the
onset of the transition to occupation of the weak binding sites.
This loading is termed a “critical” loading (c*). At c*, the
computation indicates the maximum value for the line width,
∆H ) 12.5 G, which corresponds to a probe separation of 13.4
Å. At this distance, a small spin-exchange contribution is needed
to properly simulate the spectrum (ωex ) 3.5 × 107 s-1). It is
noteworthy that the distance between the centers of the pore
openings on the silicalite external [010] surface is 12 Å and
that on the [001] surface is 14 Å, as measured by XRD.45 This
excellent agreement between the probe separation from eq 3
and the pore separation distance measured by XRD is consistent
with the saturation of the pore opening at c*.
At a surface coverage of 2.9%, the EPR spectrum of the probe
collapsestoasingleline.Simulationindicatesanelectron-electron
exchange interaction (ωex ) 3.2 × 108 s-1), in addition to a
strong but essentially constant dipolar broadening (∆H ) 12.8
G ≡ 13.3 Å). These results are consistent with the hypothesis
of a constant dipolar contribution when the pores are saturated
and with a loading-dependent contribution from spin-spin
interactions above c*. To test this hypothesis, the EPR spectra
for monodisperse zeolite crystal of different sizes (S1 to S7, Table
2) were measured as a function of loading, and the values of
c* were evaluated by simulation of the spectra. Although the
Figure 9. Behavior of (a) the EPR line width and (b) the spin-spin
exchange frequency versus 4-DPA-TEMPO surface loading. These param-
eters are EPR spectra simulation outputs. When the line width reaches the
plateau, the strong binding sites are saturated. Solid lines are for clarity
only and do not have physical meaning.
distance between the probes remains essentially the same (the
dipolar broadening reaches a plateau, as shown in Figure 9a
for silicates S1, S2, and S3.) the local concentration of radical
adsorbed increases constantly with loading, as demonstrated by
the larger and larger values of ωex. This is explained in the light
of the two-binding-sites paradigm, assuming that the strong
binding sites are saturated at c* and that above c* the probe
mobility and diffusion increase dramatically, producing a high
collision frequency between probe molecules. These dynamic
features are reflected in the EPR spectrum as increasing
spin-spin exchange frequency (Figure 9b).
A more rigorous test of the model is provided by the
investigation of a systematic series of monodisperse crystals of
differing size, since the number of available pore openings
(strong binding sites) is related to the crystal size. From
knowledge of the crystal size, the critical loading and the
external surface should have a direct relationship, namely Stot
) lSext (eq 2). Thus, a plot of Stot versus Sext is expected to
yield a straight line, within the limits of the approximation of
eq 2. Since Stot is proportional to the EPR-determined critical
loading c* and Sext is proportional to the external surface area,
a linear relationship between c* and the external surface area
is expected. Since the external surface area, in turn, is related
(45) Flanigen, E. M.; Bennett, J. M.; Grose, R. W.; Cohen, J. P.; Patton,
R. L.; Kirchner, R. M.; Smith, J. V. Nature 1978, 271, 512.
9
11350 J. AM. CHEM. SOC. VOL. 130, NO. 34, 2008

Photolysis of Dibenzyl Ketones@MFI Zeolite
A R T I C L E S
Figure 10. Critical loading of nitroxide 4-DPA-TEMPO versus external
surface area for different monodisperse silicalites. The two parameters show
a strong linear relationship, providing significance to the EPR analysis for
the titration of the strong binding sites of the external surface of MFI crystals.
Figure 11. FT-IR spectum of o-MeDBK@MFI at different loadings. The
two peaks correspond to the two sites of interaction with the external surface
of zeolite: 1716 cm-1 (weak binding sites) and 1696 cm-1 (strong binding
sites). (Inset) FT-IR spectrum of 5% loading p-MeDBK@MFI (black line)
and 5% loading o,o′-diMeDBK@MFI (blue line). The three peaks for
p-MeDBK record the three sites of interaction with the zeolite: 1716 and
1696 cm-1 are for the external surface, whereas 1685 cm-1 is for the internal
pores (blue); for o,o′-diMeDBK there is only one peak in this case for the
interaction of the adsorbate molecule with the weak binding sites on the
external surface, as it cannot intercalate inside the pore openings. These
spectra refer to silicalite S7.
to the crystal size, a linear relationship is expected between c*
and the crystal size.
This rather rigorous test of the model requires determination
of the external surface area of a set of monodisperse crystals of
varying size. Accordingly, a set of monodisperse silicalite
crystals of external surface area varying from ∼3 to ∼50 m2
g-1 were synthesized (Table 2).The external surface area was
measured by mercury porosimetry, a widely used technique to
determine the external surface area of porous materials,46,47
while c* was obtained by the EPR analysis described above
for 4-DPA-TEMPO as a probe. Figure 10 reports the plot of c*
versus external surface area for seven silicalites with different
particle sizes: the linear relationship strongly validates the critical
loading, c*, as a quantitative parameter for characterization of
the external surface of silicalites crystals. Furthermore, from
simulation of the spectra, the probe distance corresponding to
c* remains constant throughout all the zeolite samples at ∼13
Å (Table 2). The c* values for 4-DPA-TEMPO were also
calculated from the external surface areas by assuming two holes
per 330 Å2 (average of the two different faces of the MFI
crystals). The calculated versus measured values of c* show
good agreement, within the experimental error (Table 2).
In order to adapt this approach to more commonly available
materials, we tested a second probe, 4-oxo-TEMPO, and the
laboratory-synthesized silicalite S7 and compared the results to
those obtained with a commercial zeolite. The EPR analysis
shows very good agreement for 4-oxo-TEMPO with the results
for 4-DPA-TEMPO probe (Table 2), within the error limit of
the computer-aided spectral simulation. Next, 4-oxo-TEMPO
was adsorbed on commercial MFI zeolites with different Al
contents. The critical loading, which ranges between 0.013 and
0.051%, is reported in Table 2, along with the correspondent
average probe distances. Although these zeolites have the same
nominal particle size of 2 µm, their high polydispersity does
not permit a linear relationship between c* and the external
surface area. However, the average probe distance at c*, which
depends only on the crystal structure, remains constant for all
the commercial zeolites tested and is in very good agreement
with the XRD data for pore opening distance. This demonstrates
the utility of EPR analysis for the determination of critical
loading, hence the method for the “titration” of the surface pore
openings, e.g., the strong binding sites, independent of Al
content.
FT-IR Analysis: Three Peaks for Three Types of Interactions.
Infrared spectroscopy is a powerful tool for the investigation
of the interactions between adsorbed molecules and surfaces,
because binding sites of different strengths cause a shift in a
probe’s vibrational peaks in the FT-IR spectrum. For example,
in a nonpolar solution phase, the carbonyl stretch, ν(CO), of
o-MeDBK and p-MeDBK occurs at the same frequency, ∼1717
cm-1. In polar media, ν(CO) shifts to lower energies. Thus, it
is expected that a ketone probe adsorbed on a MFI zeolite will
exhibit values of ν(CO) that are related to the polarity of its
binding site. Since the more polar binding sites are expected to
be more strongly binding, the shift to lower energy should be
a signature of interactions with the strong binding sites. It is
natural, therefore, to consider how FT-IR spectroscopy can be
employed to provide information on the binding of the ketones
studied in this investigation (Chart 1).
In Figure 11, the FT-IR spectrum of o-MeDBK@MFI is
shown at different loadings. At 1% loading, two peaks at 1716
and 1696 cm-1 are present; this provides direct evidence to
support the two-site binding model of the external surface of
zeolites. The weak binding site interaction is assigned to the
peak at 1716 cm-1, because this value most closely matches
that of o-MeDBK in solution, whereas the interaction with the
strong binding sites produces the peak at 1696 cm-1. This is
even more convincing if compared with the inset of Figure 11,
where the FT-IR spectrum of p-MeDBK@MFI is reported for
an intermediate loading. In accordance with the assumption that
p-MeDBK can diffuse inside the zeolites pores, three different
ν(CO) energies are centered at 1716, 1696, and 1685 cm-1
.
The first two, being also present in the spectrum of o-MeDBK,
are assigned to the interaction with the external surface, whereas
the third one, at the lowest frequency, corresponds to the
interaction with the internal pores of the MFI zeolite.
(46) Gregg, S. J.; Sing, K. S. W. Adsorption, surface area, and porosity;
Academic Press: London, 1982.
(47) Abrams, L.; Keane, M.; Sonnichsen, G. C. J. Catal. 1989, 115, 410.
9
J. AM. CHEM. SOC. VOL. 130, NO. 34, 2008 11351

A R T I C L E S
Moscatelli et al.
respectively, by the adsorbate (in mg), and ks and kw are related
to the equilibrium constants for the adsorption onto the strong
and weak sites, respectively.
The parameters in Table 3 show that 4-oxo-TEMPO binds
about 3 times more strongly with the strong binding sites than
o-MeDBK. This can be due to (i) a reduced steric hindrance
because of the absence of bulky phenyl groups, (ii) a different
approach of the molecule to the binding sites that modifies the
thermodynamics of the interaction, or (iii) the fact that 4-oxo-
TEMPO binds with the zeolite surface also through the N-O•
moiety. For comparison, the Langmuir isotherm for TEMPO
was also constructed. In this case, no carbonyl group is present,
and the experimental data can be fit with the one-site model
(Figure 12). The fitting parameters reported in Table 3 show a
value of ks about 100 times lower for TEMPO than for 4-oxo-
TEMPO, confirming the assumption that the carbonyl group is
largely responsible for the adsorption of the latter molecule, as
the N-O• group itself interacts very poorly with the zeolite
surface. With such a low interaction energy, the weak binding
sites are not effective in binding TEMPO (kw ) 0); this results
in a one-site model fitting of the Langmuir isotherm. Moreover,
comparing the value of Ms for 4-oxo-TEMPO (converted to
coverage percentage), which gives an estimation of the maxi-
mum amount of adsorbate needed to occupy all the strong
binding sites, with the value of c* for the same probe in silicalite
S7, we found them in reasonable agreement, Ms ) 0.0155%
and c* ) 0.0180%, to confirm the accuracy of our EPR
approach.
Finally, the Langmuir isotherm for the o,o′-diMeDBK is
constructed. Since such ketones cannot intercalate into the
pore openings of the zeolite crystals as o-MeDBK can, a
weaker interaction with the external surface of zeolite is
expected, as observed in its FT-IR spectra, which shows one
slightly shifted peak. The experimental data can again be
fitted with the one-site adsorption model (Table 3). The value
kw ) 10 implies an interaction with the surface more than
an order of magnitude weaker than for o-MeDBK on the same
type of crystals (Si/Al ) 20). In light of consistent FT-IR
measurements, it is concluded that, for dibenzyl ketones, the
strong binding sites on the external site of the zeolite are
due to the intercalation of part of the molecule into the pore
openings; if this is impeded, as in the case of o,o′-diMeDBK,
interaction energies are significantly lower, by up to an order
of magnitude, typical of interaction with weak binding sites.
This simple comparison validates the postulate that the strong
binding sites are the pore openings of the external surface
of the zeolites.
Figure 12. Langmuir isotherm for 4-oxo-TEMPO and TEMPO adsorbed
onto silicalite S7. While the experimental points of 4-oxo-TEMPO can
only be fitted by a two-site model (eq 4), TEMPO is well fitted by a
one-site model, i.e., Kw ) 0. A summary of fitting parameters is reported
in Table 3.
Table 3. Langmuir Isotherm Fitting Parameters
molecule
Ms (mg)
ks
Mw (mg)
kw
o-MeDBKa,b
o,o′-diMeDBKb
4-oxo-TEMPOc
TEMPOc
0.40
0
0.37
1.0
350
0
1090
115
0.59
1.0
0.63
0
2.1
10
2.0
0
a Reference 20. b For MFI with Si/Al ) 20. c For silicalite S7.
In order to fully test the external surface model, we completed
the cases sketched in Figure 2 measuring the FT-IR spectrum
of o,o′-diMeDBK@MFI, a molecule that is unable to intercalate
a benzene ring inside the channels, and thus unable to occupy
the strong binding sites. The spectrum (lower trace in the inset
of Figure 11) shows predominantly only one peak centered at
1713 cm-1, confirming the prediction of the model.
Langmuir Isotherm. The model and the spectroscopic ex-
perimental evidence presented above clearly demonstrate that
there is both a strong binding site and a weak binding site
associated with the external surface of MFI zeolite. Additional
evidence for the presence of two binding sites with different
energies on a surface is obtained by constructing the classical
Langmuir isotherm for the probes investigated in this study.
Moreover, a fit of the isotherm allows a quantitative estimation
of the binding energies of the strong and weak binding sites.48
The Langmuir isotherm for o-MeDBK@MFI (Si/Al ) 20) is
reported in the literature and shows the two binding sites’
behavior.16 We constructed the Langmuir isotherm for 4-oxo-
TEMPO, in order to ensure that our EPR analysis was indeed
consistent with the two-site model. The result is shown in Figure
12. The best one-site fitting curve is clearly not appropriate in
this case; however, the two-site curve offers a good agreement
with the experimental data. The analytical expression of the
Langmuir fitting curve is given by eq 4, while the best fitting
parameters for 4-oxo-TEMPO and o-MeDBK are summarized
in Table 3.
Conclusions
The impact of the external surface on the physical-chemical
properties of zeolites increases with decreasing particle size. In
order to optimize zeolites’ performances, knowledge of the
external surface adsorption properties is essential. A model in
which the external surface of zeolites is composed of strong
binding sites, located on the pore openings, and weak binding
sites, situated between the pores, has been extensively inves-
tigated using a multi-technique approach.
ksC
s(1 + ksC)
kwC
w(1 + kwC)
A ) M
+ M
(4)
FT-IR spectroscopy has been used to show directly the three
types of interactions, characterized by different ν(CO) peak
shifts, that selected dibenzyl ketones (Chart 1) can have with
the zeolite framework. The weak and strong sites on the external
In eq 4, A (in mg) is proportional to the surface coverage, C is
the total loading of the adsorbate (in wt/wt), Ms and Mw give
an estimation for the relative coverage of strong and weak sites,
surface are clearly revealed by ν(CO) at 1716 and 1696 cm-1
,
(48) Os´cik, J. Adsorption; Halsted Press: New York, 1982.
respectively (Figure 11).
9
11352 J. AM. CHEM. SOC. VOL. 130, NO. 34, 2008

Photolysis of Dibenzyl Ketones@MFI Zeolite
A R T I C L E S
By EPR spectroscopy, a direct method to specifically probe
the strong binding sites on the external surface of MFI zeolites
has been developed. A measure of the critical loading, c*, at
which the strong binding sites are saturated and cross-
comparison with mercury porosimetry and XRD data validates
the EPR method for characterizing the external surface of MFI
zeolites. The exact value of the external surface area for
monodisperse silicalites can be determined by the EPR spin
probing method described here, provided that a pre-characterized
and calibrated silicalite sample is available as a standard.
The interaction energies for the dibenzyl ketones and the
nitroxides used as EPR probes have been determined by
measuring the Langmuir isotherm. The results are consistent
with a two-site energy surface in which the intercalation of the
unsubstituted benzene ring of o-MeDBK inside the channels
system provides the most significant contribution to the interac-
tion energy for the adsorption to the external surface of zeolite.
Other mechanisms, such as hydrogen-bonding and van der
Waals interactions, are more than an order of magnitude less
intense, as demonstrated by the o,o′-diMeDBK Langmuir
adsorption isotherm (Table 3); these interactions are typical of
the weak binding sites. Since the latter molecule cannot
intercalate inside the pore openings because of structural
impedances, previous assignment of the strong binding sites to
the pore openings proves to be correct in the light of the present
investigation.
Figure 13. Schematic representation of the home-made apparatus used
for loading zeolites with spectator solvent molecules. It is composed of
two parts: a branched cell for EPR measurements (blue line), where the
zeolite, previously loaded with DBKs, is placed and that will also serve as
EPR sample holder, and the solvent addition system (orange line), where a
controlled vapor pressure can be created and then let adsorb to the loaded
zeolite. All parts are made of Pyrex glass, except for the EPR cell, which
is made of quartz for irradiation.
Finally, the photochemistry of o-MeDBK@MFI and DBK
@MFI (Scheme 1) in the presence of spectator molecules, such
as water, pyridine, and benzene, has been investigated both by
measuring the concentration of final products to calculate the
cage effect and by following the persistent benzyl and acyl
radicals in real time by cw-EPR spectroscopy. The presence of
co-adsorbed spectator molecules adds another level of variability
to this supramolecular system for adjusting and controlling the
product distribution of such radical reactions. In particular, it
has been found that, during the photolysis of DBK@MFI in
the presence of pyridine and benzene, the decarbonylation step
is slowed so greatly that the phenylacetyl radical becomes
persistent (time scale of minutes) and thus detectable by cw-
EPR: a remarkable outcome, if compared to the sub-microsecond
time scale for the decarbonylation step for dibenzyl ketones in
solution.49,50
4-oxo-2,2,6,6-tetramethyl-1-piperidinyl-oxy (4-oxo-TEMPO) and
2,2,6,6-tetramethyl-1-piperidinyl-oxy (TEMPO) were purchased
from Aldrich and used without further purification.
Laboratory-synthesized, monodisperse silicalite crystals (Sn) were
synthesized and fully characterized as described in a previous
publication.18 Commercial zeolited (Cn) were purchased from
Zeocat. It is important to emphasize that initial studies using
amorphous silicalites (defect laden) showed the same photochemi-
cally induced reaction outcome as did those using the highly
crystalline zeolite.27,47 Therefore, within measurement and calcula-
tion error, perfect, monodispense crystals have been assumed.
EPR Analysis with Nitroxide Spin Probes. In a typical
experiment, 300 mg of calcined zeolites was first activated in an
aerated furnace at 500 °C for 2 h and then placed in a desiccator
to allow cooling to room temperature before use. Next, 3 mL of
solution in 2,2,4-trimethylpentane (isooctane), containing an ap-
propriate amount of adsorbate molecule, was added to the activated
crystal. Isooctane was chosen as solvent because of its inability to
penetrate the pore openings of the MFI zeolite, as its kinetic
diameter is larger than the pore openings.53 The system was allowed
to equilibrate under magnetic stirring for about 10 h. To obtain the
dry loaded sample, the solvent was first removed under gentle Ar
stream, and then the sample was degassed under vacuum (1 × 10-5
Torr) in a branched EPR cell (Figure 13, blue line). Measurements
were performed using an EMX Bruker spectrometer with a
microwave power of 2.01 mW and modulation amplitude of 1.0
G. Spectral simulations were made using the software NLSL for
slow tumbling motion regime.54,55 The simulation strategy approach
is described in detail elsewhere.19
Experimental Section
Materials. Syntheses of 1,3-diphenyl-2-propanone (DBK), 1-(2-
methylphenyl)-3-(phenyl)-2-propanone (o-MeDBK), their deuter-
ated benzene counterparts, and 1,3-bis(2-methylphenyl)-2-pro-
panone (o,o′-diMeDBK) were performed as previously reported,29
as was that of 1-(4-methylphenyl)-3-phenyl-2-propanone (p-
MeDBK).51 Pyridine, benzene, and p-xylene were purchased from
Aldrich and used as received.
Nitroxide probe 2,2,6,6-tetramethyl-4-[(diphenylacetyl)oxy]-1-
piperidinyl-oxy (4-DPA-TEMPO) was synthesized following a
published procedure for esterification at room temperature, using
4-hydroxy-2,2,6,6-tetramethyl-1-piperidinyl-oxy and diphenylacetic
acid, in 66% yield.52 It was recrystalized using ethyl acetate. MS
(FAB+): 366.2 (M - H)+, 368.2 (M + H)+. Nitroxide probes
FT-IR Measurements. Preparation of loaded zeolite followed
closely that for EPR measurements except for the amount of zeolite
used, which in this case was typically 20 mg, and the total volume
(49) Gould, I. R.; Baretz, B. H.; Turro, N. J. J. Phys. Chem. 1987, 91,
925.
(53) Jacobs, P. A.; Martens, J. A.; Weitkamp, J.; Beyer, H. K. Faraday
Discuss. 1981, 72, 353.
(50) Tsentalovich, Y. P.; Kurnysheva, O. A.; Gritsan, N. P. Russ. Chem.
Bull. 2001, 50, 237.
(54) Budil, D. E.; Lee, S.; Saxena, S.; Freed, J. H. J. Magn. Reson., Ser.
A 1996, 120, 155.
(51) Hrovat, D. A.; Liu, J. H.; Turro, N. J.; Weiss, R. G. J. Am. Chem.
Soc. 1984, 106, 5291.
(52) Hassner, A.; Alexanian, V. Tetrahedron Lett. 1978, 4475.
(55) Earle, K. A.; Budil, D. E. In AdVanced ESR Methods in Polymer
Research; Schlick, S., Ed.; Wiley-Interscience: Hoboken, NJ, 2006.
(56) Hill, S. G.; Seddon, D. Zeolites 1985, 5, 173.
9
J. AM. CHEM. SOC. VOL. 130, NO. 34, 2008 11353

A R T I C L E S
Moscatelli et al.
of adsorbate solution, which was 200 µL. After drying in the
vacuum line, 5 mg of loaded zeolite was combined with 500 mg
of commercially prepared KBr crystals (purchased from Thermo).
The diffuse reflectance infrared spectroscopy by Fourier transform
(DRIFT) technique was used for these measurements. The FT-IR
spectra were obtained at room temperature on a Nicolet Nexus 360
spectrometer (equipped with a DTGS detector and DRIFTS Smart
accessory) at 2 cm-1 resolution, 64 scans.
It is important to highlight that the dibenzyl ketone is loaded
before the spectator molecule for the following reasons: (i) to be
consistent with similar studies previously reported in which water
was used as spectator molecule27 and (ii) to make sure that the
photolysis of DBKs is studied starting from the same conditions
of zeolite loading, allowing a more straightforward comparison
among the spectator molecules.
For the cage effect measurement, the loaded, degassed zeolite
was irradiated with a 450 W medium-pressure Hg lamp through a
chromate filter solution with continuous cell tumbling. The conver-
sion of the reaction was controlled to less than 25%, and the
irradiated samples were extracted with ether or benzene. The
product distribution was measured by GC. For the kinetics
measurements with cw-EPR, the branched cell was put inside the
EPR cavity and irradiated for 6 s with a 300 W Xe lamp with an
optical filter to cut the wavelengths below 300 nm. The EPR spectra
or the time decays were recorded immediately after the photolysis.
Langmuir Isotherms. Langmuir isotherms were constructed
using a standard depletion method, in which the amount of adsorbed
molecule was calculated from the difference between the initial
and the final concentrations in the solution. This was followed by
doubly integrated EPR spectra for nitroxides and absorption at 295
nm for o,o′-diMeDBK. Starting solutions were prepared in isooctane
and added to 200 mg of zeolite. The system was allowed to
equilibrate under magnetic stirring for about 10 h before measuring
the concentrations of the supernatant.
Photolysis of DBKs and Cage Effect Measurement. Dibenzyl
ketones and spectator molecules were loaded to the zeolite in two
separate stages. First, the desired DBK was loaded following the
procedure described for the EPR analysis; however, CH2Cl2 was
used in this case as solvent to allow the solution to permeate inside
the pores of the zeolite. The zeolite was placed in a branched cell
for EPR measurements (Figure 13, blue line) and dried under
vacuum at 1 × 10-5 Torr. At this point, spectator molecules were
added by a solvent addition system (Figure 13, orange line). This
system, designed in-house by X.L. and manufactured by Mr. Fred
Krummer (glassblower), is composed of a 250 mL round-bottom
flask connect to a mercury pressure gauge, which was calibrated
before use, a solvent reservoir containing 1 mL of liquid, which
was degassed by pump-freeze-thaw cycles before addition to the
zeolite, the branched cell for EPR measurements, and a vacuum
line. To add a desired amount of solvent to the zeolite, we first
evacuated the whole system (except the solvent reservoir) to 1 ×
10-5 Torr and then, closing the connections to the vacuum line
and to the branched cell, let the solvent vapors occupy the rest of
the volume. The vapor pressure in these conditions was calculated
by the movement of the mercury column. The amount of solvent
to be added to the zeolite was converted to the pressure difference
and mercury level, according to the volume of the total system
and the molecular weight of the solvent. The last step was to open
the connection to the EPR branched cell until the desired mercury
level was reached. The system proved to be very accurate, as the
solvent loading error caused by the mercury level reading is less
than 2%.
Acknowledgment. The authors thank the National Science
Foundation for its generous support of this research through grants
CHE 04-15516 and CHE-07-17518. Moreover, authors are grateful
to Dr. Steffen Jockusch at Columbia University for the help in
setting up the cw-EPR measurements with photolysis and helpful
discussion.
JA801487V
9
11354 J. AM. CHEM. SOC. VOL. 130, NO. 34, 2008