Hammett analysis of photodecarbonylation in crystalline 1,3-diarylacetones

Article abstract of DOI:10.1021/ol0480527

(Chemical Equation Presented) The relative quantum yields and chemical efficiencies of crystalline p,p′-disubstituted 1,3-diphenyl-2-propanones with 4-MeO, 4-Me, 4-F, 4-CF₃, and 3,4-diMeO groups were determined by parallel irradiation of polycrystalline samples. Variations in quantum yields that span a factor of 4 are analyzed in terms of the effects of substituents on the stability of the benzylic radical. All solid-state reactions proceeded with 100% chemoselectivity and in >95% chemical yield.

Full text of DOI:10.1021/ol0480527

ORGANIC
LETTERS
2005
Vol. 7, No. 3
371-374
Hammett Analysis of
Photodecarbonylation in Crystalline
1,3-Diarylacetones
Marino J. E. Resendiz and Miguel A. Garcia-Garibay*
Department of Chemistry and Biochemistry, UniVersity of California,
Los Angeles, California 90095-1569
mgg@chem.ucla.edu
Received September 24, 2004
ABSTRACT
The relative quantum yields and chemical efficiencies of crystalline p,p′-disubstituted 1,3-diphenyl-2-propanones with 4-MeO, 4-Me, 4-F, 4-CF3,
and 3,4-diMeO groups were determined by parallel irradiation of polycrystalline samples. Variations in quantum yields that span a factor of
4 are analyzed in terms of the effects of substituents on the stability of the benzylic radical. All solid-state reactions proceeded with 100%
chemoselectivity and in >95% chemical yield.
It has been suggested that the intrinsic reactivity of organic
molecules may be less important than packing effects when
reactions in crystals are concerned.1 However, recent work
has shown that some reactions may be reliably predicted from
well-known trends in reactive intermediates.2 In particular,
rapidly accumulating evidence shows that the photodecar-
bonylation of crystalline ketones depends on the energy
content of the triplet state and the bond dissociation energies
of the R-bonds,3 as long as they have no intramolecular
quenching interactions.4 Remarkably, whereas reaction fea-
sibility relies on the intrinsic reactivity of the excited ketone,
the selectivity and specificity of the solid-state reaction are
determined by the rigidity and homogeneity of the crystalline
medium. In fact, we have shown the synthetic potential of
the reaction in the preparation of optically pure compounds
with adjacent quaternary stereogenic centers5 and in the total
synthesis of the natural product herbertenolide (Scheme 1).6
However, for the reaction to be considered a robust synthetic
method, it is important to show that is amenable to structure-
Scheme 1
(1) (a) Ramamurthy, V.; Venkatesan, K. Chem. ReV. 1987, 87, 433-
481. (b) Cohen, M.; Schmidt, G. M. J. J. Chem. Soc. 1964, 1996-2000.
(2) (a) Garcia-Garibay, M. A. Acc. Chem. Res. 2003, 36, 491-498. (b)
Garcia-Garibay, M. A.; Shin, S. H.; Sanrame, C. Tetrahedron 2000, 56,
6729-6737.
(3) Campos, L. M.; Ng, D.; Yang, Z.; Dang, H.; Martinez, H. L.; Garcia-
Garibay, M. A. J. Org. Chem. 2002, 67, 3749-3754.
(4) Ng, D.; Yang, Z.; Garcia-Garibay, M. A. Tetrahedron Lett. 2001,
42, 9113-9116.
10.1021/ol0480527 CCC: $30.25
© 2005 American Chemical Society
Published on Web 01/12/2005

reactivity correlations analogous to those commonly carried
out for a wide variety of reactions in solution.
Knowing that aliphatic ketones with aryl substituents at
their two R-positions display an efficient reaction in the solid
state,7 we decided to test the effect of substituents in several
2,2,4,4-tetramethyl-1,3-diphenylacetones (Scheme 2). Com-
from the constraints of the σ bond, the fragments separate
and the free acyl radical loses CO. Free radical encounters
in solution lead to the second radical pair (RP2), which forms
diphenylethane 2 by bond formation and products 3 and 4
by disproportionation. Quantum yields of reactant consump-
tion determined in C6H6 solutions using diphenylacetone (Φ
≈ 0.6 ( 0.05)9 as a chemical actinometer were essentially
identical for all tetramethyl-substituted compounds (Φ ≈ 0.8
( 0.1).
Scheme 2
Reactions in crystals proceeded with ideal chemoselec-
tivity, giving the diphenylethanes 2a-2f in >95% yield. To
determine the effects of the substituents on the quantum
yields of reaction, competition experiments were set up by
mechanically mixing 15 mg of independently powdered
samples of each ketone, irradiating the powder mixture at
298 K with a medium-pressure Hg arc Hanovia lamp using
a λ > 300 nm cutoff filter, and analyzing the extent of
reaction by gas chromatography as a function of conversion
every 5 min for 1-2 h (e.g., Figure 1).
pounds 1a-1f, prepared in two steps as illustrated in Scheme
2, were crystalline solids with melting points shown to be
conveniently high for all solid-state reactions to be carried
out at ambient temperature without melting.8
Photochemical reactions of 1a-1f carried out as control
experiments in ca. 0.1 M benzene with λ > 300 nm produced
the expected diphenylethanes (2a-2f), styrenes (3a-3f), and
isopropyl benzenes (4a-4f). As indicated in Scheme 3, the
Scheme 3
Figure 1. Representative plot of reaction progress with crystals
of 1,3-diphenylacetones 1d (4-Me), 1e (4-CF3), and 1f (4-F).
Given the excellent overlap in the UV spectra (Supporting
Information) and assuming that absorption and scattering of
the various powders are similar, the initial slopes of the plot
in Figure 1 are proportional to the quantum yields (Φ) of
reaction. As shown in Figure 1, conversion values varied
by a factor of 4 between the most (1b, 4-CF3) and least
reactive (1c, 4-F) ketones. Internally consistent results from
repeated experiments suggest that the ca. 5% difference
observed between compounds 1a (H) and 1f (4-Me) are
reliable. Relative quantum yields [Φ(Y)/Φ(H)] are sum-
marized in Table 1 along with the experimentally determined
ratios of the solution rate constants of 1,3-diphenylacetones
(please see below) and the differences in bond dissociation
energies of the benzylic C-H bonds [BDE(Y) - BDE(H)]
reported in the literature.11-13
reaction is known to start by R-cleavage from the triplet
excited state to give an acyl-alkyl radical pair (RP1).9,10 Free
(5) Ellison, M. E.; Ng, D.; Dang, H.; Garcia-Garibay, M. A. Org. Lett.
2003, 5, 2531-2534.
(6) Ng, D.; Yang, Z.; Garcia-Garibay, M. A. Org. Lett. 2004, 6, 645-
647.
(7) Choi, T.; Peterfy, K.; Khan, S. I.; Garcia-Garibay, M. A. J. Am. Chem.
Soc. 1996, 118, 12477-12478.
(8) Keating, A. E.; Garcia-Garibay, M. A. Molecular and Supramolecular
Photochemistry; Ramamurthy, V., Schanze, K., Ed.; Marcel Dekker: New
York, 1998; Vol. 2, pp 195-248.
(9) Engel, P. S. J. Am. Chem. Soc. 1970, 92, 6074-6076.
(10) (a) Weiss, D. Organic Photochemistry; Padwa, A., Ed.; Marcel
Deker: New York, 1981; Vol. 5, pp 347-420. (b) Baretz, B.; Turro, N. J.
J. Am. Chem. Soc. 1983, 105, 1309-1316. (c) Turro, N. J.; Cherry, W. R.
J. Am. Chem. Soc. 1978, 100, 7431-7432.
372
Org. Lett., Vol. 7, No. 3, 2005

Unable to separate much beyond the sum of their van der
1
Waals radii, a rapid spin flip (kTS) to form RP1 makes it
Table 1. Relative Quantum Yields in the Solid State with
Relative Rates Constants for R-Cleavage and Decarbonylation
and with Differences in Bond Dissociation Energies for
Substituted Diphenylacetones
possible to form the σ bond and go back to ground-state
starting material (dashed arrow in Scheme 3). However, if
the rate of decarbonylation (k-CO) is substantially greater than
the rate of intersystem crossing (kTS), the yield of decarbon-
ylation will be high (eq 3).
BDE(Y) -
Φ(Y)/Φ(H] kR(Y)/kR(H) k-CO(Y)/k-CO(H) BDE(H)
Y
3,4-MeO
4-MeO
4-Me
4-F
H
4-CF3
1.08
1.65
1.05
0.54
1.0
n.a.
3.18
1.53
1.05
1.0
n.a.
1.55
1.14
n.a.
1.0
-1.1a
-0.7b
-0.5b
0.1c
0
0.3b
k-CO
Φ-CO(Y) )
(3)
k-CO + kTS
Local diffusion of CO should render decarbonylation
2.18
n.a.
1.06
3
irreversible so that formation of RP2 should determine
a
Reference 13. b Reference 11. c Reference 12
product formation. If this assumption is correct, substituent
effect in the following steps (kTS′ and kbond) should not affect
the observed quantum efficiency.
Knowing that the effects of substituents on radical reac-
tions are generally small, the results in Figure 1 and Table
1 are reassuring, as large variations due to undetermined
“packing effects” would undermine our hypothesis. As
discussed below, the relatively poor correlation with sub-
stituent constants is consistent with observations in solution.
The quantum yields of reaction for compounds 1a-1f
[Φ(Y)] depend on several consecutive steps (eq 1), and the
mechanism in Scheme 3 illustrates the intermediates to
consider in going from ketones 1 to diphenyl ethanes 2.
The lack of significant substituent effects on the solution
quantum yields implies that the rate constants for the product-
limiting steps are greater than those of other competing
processes. It is known that fast separation of the acyl-alkyl
radical pair (3RP1) makes R-cleavage product formation
3
limiting. In crystals, the fate of RP1 is determined by a
competition between decarbonylation, which goes on to give
1
products, and intersystem crossing to RP1, which returns
to the starting material and decreases the solid-state quantum
yield.
The effects of light atom substituents on intersystem
crossing rates are expected to be small. Similarly, Johnston
et al. reported that the relative rates of R-cleavage in solution
[kR(Y)/kR(H), Table 1] span a factor of 3, with a reactivity
ranking that includes p-MeO > p-Me > p-F ≈ H.14 Finding
a reasonable fit to the Hammet equation when the substituent
σ+ values were used,15 the authors inferred a transition state
stabilized by electron-donating substituents. The effects of
substituents on the absolute rate constant of decarbonylation
in several symmetric ketones (Y-PhCH2COCH2Ph-Y) were
investigated by Nau using transient absorption methods.16
The decarbonylation rate constants turned out to be remark-
ably insensitive to p-MeO, p-Me, p-Cl, and p-CF3 substit-
uents as indicated by k-CO(Y)/k-CO(H) ≈ 1 values in Table
1.
At a first approximation, one may expect that variations
in solid-state Φ values may correlate with kR and σ+(Y).
Alternatively, because a good correlation is expected between
the height of the barrier and the heat of reaction (as suggested
by the Hammond postulate),10a,17 the results may reflect the
effect of substituents on the bond dissociation energy [BDE]
of the R bonds. In fact, there is a reasonable correlation
between ∆BDE(Y) and σ+(Y). However, although results
with 4-MeO, 4-Me, 4-H, and 4-F compounds would be as
expected, the plot of Φ(Y)/Φ(H) vs ∆BDE(Y) in Figure 2
shows a poor correlation with the 3,4-diMeO- and CF3-
ketones deviating at the two ends of the plot.
Φ(Y) ) ΦT(Y)‚ΦR(Y)‚Φ-CO(Y)
(1)
Assuming that photons are absorbed at the same rate by
all solids and that the loss of CO is irreversible, the Φ(Y)
values will depend on the effects of substituents on the yields
of triplet formation [ΦT(Y)], R-cleavage [ΦR(Y)], and the
loss of CO [Φ-CO(Y)].
Although the R-cleavage reaction may proceed in solution
from singlet and triplet manifolds, singlet radical pairs in
the solid state are likely to retrace their trajectory back to
the original σ bond rather than going on to products.
Fortunately, ΦT(Y) values for ketones are generally high
[e.g., ΦT(Y) ≈ 0.5-1.0].10a After reaching 31*, the quantum
efficiency of product formation depends on the quantum yield
of R-cleavage [ΦR(Y)], which is given by the rate of
R-cleavage (kR) as compared to the rate of return to the
ground state by phosphorescence (kP) and radiationless decay
(kd) as indicated in eq 2.
kR
ΦR(Y) )
(2)
kR + kP + kd
Notably, while efficient R-cleavage reactions in solution
lead to the products thanks to the rapid separation of RP1
to form free radicals, the same is not true in the solid state.
3
(14) Johnston, L. J.; de Mayo, P.; Wong, S. K. J. Am. Chem. Soc. 1982,
104, 307-309.
(11) Pratt, D. A.; Dilabio, G. A.; Mulder, P.; Ingold, K. U. Acc. Chem.
Res. 2004, 37, 334-340.
(15) Brown, H. C.; Okamoto, Y. J. Am. Chem. Soc. 1958, 80, 4979-
(12) Fy, Y.; Liu, L.; Lin, B.-L.; Mou, Y.; Cheng, Y.-H.; Guo, Q.-X. J.
Org. Chem. 2003, 68, 4657-4662.
4987.
(16) Zhang, X.; Nau, W. M. J. Phys. Org. Chem. 2000, 13, 634-639.
(17) Fisher, H.; Paul, H. Acc. Chem. Res. 1987, 20, 200-206.
(13) Fox, T.; Kollman, P. A. J. Phys. Chem. 1996, 100, 2950-2956.
Org. Lett., Vol. 7, No. 3, 2005
373

As illustrated in Figure 3, the conformations of 1a and 1f
are quite similar. They have two R-methyl groups eclipsed
Figure 2. Relative quantum yields [log(ΦY/ΦH)] plotted against
the difference in bond dissociation energies [∆BDE(Y) ) BDE-
(Y) - BDE(H)] of the substituent Y for several crystalline 1,3-
diphenylacetones.
Figure 3. Molecular structures of 1a and 1f determined by X-ray.
Only one of two disordered positions is shown for the CF3 groups.
Although the ∆BDE(Y) values differ by only 1.5 kcal/
mol and the variation in Φ(Y)/Φ(H) is only a factor of 4,
we decided to see whether the unexpected behavior of 1f as
compared to 1a may be explained by differences in the
conformation of the aryl groups in their respective crystals.
As aryl rotation in crystals should be limited, the prospective
radicals may not enjoy the same extent of benzylic delocal-
ization. In fact, experiments18 and calculations19 indicate that
the difference in energy between benzyl radical conforma-
tions with parallel and orthogonal p-π-orbital alignment is
ca. 12.5 kcal/mol. Recent results from our group with 1,1,3,3-
tetramethyl-2-indanone 5 (Scheme 4) are particularly relevant
with the CdO bond, as expected by a dipole-induced dipole
interaction,21 and their two R-phenyl groups are anti to each
other. The dihedral angles made by the plane of each phenyl
group and the R-bond in 1a are D1 ) 50.6° and D2 ) 50.3°,
respectively. The corresponding values in 1f are D1 ) D2 )
46.3°. Because the orientation of the π-system is orthogonal
to the aromatic plane, the ideal benzylic stabilization would
occur for D ) 90° (i.e., Θ ) 90° - D). The resonance energy
expected if Θ were to remain constant during the solid-state
reaction would be E(1a) ) 12.5[cos2(39.4°)] ) 7.3 kcal/
mol and E(1f) ) 12.5[cos2(43.7°)] ) 6.5 kcal/mol, respec-
tively. This would predict 1f to react less efficiently than
1a, suggesting that small ground-state conformational dif-
ferences in the crystal may not be playing a determining role.
Other factors to consider include the effect of substituents
on the “polarity” of the medium, which is known to affect
the rates of decarbonylation in solution,16 and differences in
packing interactions leading to different steric barriers for
the bond-cleavage processes.
Scheme 4
In conclusion, although the effect of aromatic substituents
on the R-cleavage and decarbonylation reactions of 1,3-
diarylacetones is relatively modest, it is important that
reactions in crystals are as tractable and not much more
complex than analogous reactions in solution. This gives
support to our suggestion that reactions in crystals can be
engineered from known molecular structure parameters.
in this context.20 Closely related to the 1,3-diphenyl acetones
studied here, tetramethyl indanone 5 reacts efficiently in
solution where bond rotations allow benzylic stabilization
to take place, but it is completely stable in the solid state.
Knowing that interaction energies have an angular depen-
dence E(Θ) ) Emax cos2 Θ, where Θ is the angle between
the two orbital systems and Emax is the maximum stabilization
energy when Θ ) 0, we decided to investigate the X-ray
structures of 1a (Y ) 4-H) and 1f (Y ) CF3).
Acknowledgment. Financial support by the National
Science Foundation is gratefully acknowledged.
Supporting Information Available: Preparation and
characterization of compounds 1a-1f, 2a-2f, and 5a-5f
and crystallographic data for compounds 1a (tables) and 1f
(in CIF format). This material is available free of charge
via the Internet at http://pubs.acs.org.
(18) (a) Nonhebel, D. C.; Walton, J. C. J. Chem. Soc., Chem. Commun.
1984, 731. (b) Conradi, M. S.; Zeides, H.; Livingston, R. J. Phys. Chem.
1979, 83, 2160-2161.
(19) (a) Hrovat, D. A.; Borden, W. T. J. Phys. Chem. 1994, 98, 10460-
10464. (b) Dorigo, A. E.; Li, Y.; Houk, K. N. J. Am. Chem. Soc. 1989,
111, 6942-6948.
OL0480527
(20) Ng, D.; Yang, Z.; Garcia-Garibay, M. A. Tetrahedron Lett. 2002,
43, 7063-7066.
(21) Wiberg, K. B.; Murcko, M. A. J. Comput. Chem. 1988, 9, 488-
494.
374
Org. Lett., Vol. 7, No. 3, 2005