Photochemical Pinacol Rearrangements of Unsymmetrical Diols

Article abstract of DOI:10.1021/jo035439z

The photochemical pinacol reaction of a series of nonsymmetrical 9-fluorenyl-substituted vic-diols was investigated and compared with their acid-catalyzed thermal reaction. Unlike the thermal reaction, the radiation-induced processes involve only fluorenyl cations, as is reflected in differences of product distribution between the two reactions. From the product studies, substituent migratory aptitudes are reversed in the photochemical process, suggesting that kinetic control takes place under neutral conditions unlike the acid-catalyzed thermal reactions. The presence of fluorenyl cation intermediates and their lifetimes were established by laser flash spectroscopy studies.

Full text of DOI:10.1021/jo035439z

P h otoch em ica l P in a col Rea r r a n gem en ts of Un sym m etr ica l Diols
Gabriela Mladenova,† Gurmit Singh,† Austin Acton,† Lie Chen,‡ Olga Rinco,‡
Linda J . J ohnston,*,‡ and Edward Lee-Ruff*,†
Department of Chemistry, York University, Toronto, Ontario M3J 1P3, Canada, and Steacie Institute for
Molecular Sciences, National Research Council Canada, Ottawa, Ontario K1A 0R6, Canada
leeruff@yorku.ca
Received October 1, 2003
The photochemical pinacol reaction of a series of nonsymmetrical 9-fluorenyl-substituted vic-diols
was investigated and compared with their acid-catalyzed thermal reaction. Unlike the thermal
reaction, the radiation-induced processes involve only fluorenyl cations, as is reflected in differences
of product distribution between the two reactions. From the product studies, substituent migratory
aptitudes are reversed in the photochemical process, suggesting that kinetic control takes place
under neutral conditions unlike the acid-catalyzed thermal reactions. The presence of fluorenyl
cation intermediates and their lifetimes were established by laser flash spectroscopy studies.
In tr od u ction
bution is often thermodynamically controlled and de-
pendent on the conditions used.4 Our interests in desta-
bilized carbocations such as 9-fluorenyl cations generated
under photochemical conditions have extended to inves-
tigations of their kinetic stabilities as well as their
tendency to undergo rearrangements.5,6 Since the inter-
mediate carbocations are produced in neutral solutions,
such rearrangements are not subject to thermodynamic
equilibration normally encountered in acid conditions,
and thus product distributions are kinetically controlled
and more reflective of substituent migratory aptitudes.
Several years ago we reported the first photochemical
pinacol rearrangement of symmetrical 9-bifluoren-9-ol to
the corresponding ketone.6 Competitive ketyl formation
by homolytic cleavage was also observed in acetonitrile
solutions. However, the pinacol rearrangement was the
principal pathway in 2,2,2-trifluoroethanol (TFE), a
solvent known to promote carbocation formation.7 To
assess migratory aptitudes of different substituents in
pinacol rearrangements unaffected by acid equilibration,
we investigated the photochemistry of a series of vic-diols
derived from 9-fluorenol. We chose this particular struc-
ture motif because fluorenols are susceptible to hetero-
lytic dissociation to fluorenyl cations on photoexcitation8
and these transients exhibit characteristic visible absorp-
tion features in time-resolved spectra.9 A series of
unsymmetrical fluorenyl diols 1 were prepared, and their
photochemistry was investigated and compared with
their acid-catalyzed dark reaction products.
The pinacol rearrangement of vic-diols has been known
since 1860.1 This transformation involving a dehydration
and skeletal rearrangement is one of the earliest docu-
mented Wagner-Meerwein rearrangements in carbo-
cation chemistry and has been used as a key step in
organic synthesis.2 The intermediate cations are usually
generated under Lewis or Bronsted acid conditions, which
often result in regioisomeric mixtures of ketones and
aldehydes when unsymmetrical diols are used (eq 1).
Regioselectivity is determined by stability of the primary
formed carbocation as well as migratory aptitudes of the
diol substituents, which in turn are predicated by ste-
reoelectronic factors.3 Furthermore, the aldehydes and
ketones produced, under the acid conditions used, are
subject to interconversions such that the product distri-
(4) Fry, A.; Oka, M. J . Am. Chem. Soc. 1979, 101, 6353.
(5) Mladenova, G.; Chen, L.; Rodriquez, C. F.; Siu, K. W. M.;
J ohnston, L. J .; Hopkinson, A. C.; Lee-Ruff, E. J . Org. Chem. 2001,
66, 1109-1114.
(6) Hoang, M.; Gadosy, T.; Ghazi, H.; Hou, D.-F.; Hopkinson, A. C.;
J ohnston, L. J .; Lee-Ruff, E. J . Org. Chem. 1998, 63, 7168-7171.
(7) Eberson, L..; Hartshorn, M. P.; Radner, F. J . Chem. Soc., Chem.
Commun. 1996, 2105-2112.
(8) Wan, P.; Krogh, E. J . Am. Chem. Soc. 1989, 111, 4887-4895.
(9) Lew, C. S. Q.; Wagner, B. D.; Angelini, M. P.; Lee-Ruff, E.;
Lusztyk, J .; J ohnston, L. J . J . Am. Chem. Soc. 1996, 118, 12066-12073.
† York University.
‡ Steacie Institute for Molecular Sciences.
(1) For a historical perspective, see: Berson, J . A. Angew. Chem.,
Int. Ed. 2002, 41, 4655-4660.
(2) (a) Aszodi, J .; Fauvau, P.; Melon-Manguer, D.; Eberhard, E.;
Schio, L. Tetrahedron Lett. 2002, 43, 2953-2956. (b) Pettit, G. R.;
Lippert, J . W., III; Herald, D. L. J . Org. Chem. 2000, 65, 7438-7444.
(c) Gasparski, P. M.; Herrinton, P. M.; Overman, L. E.; Wolfe, J . P.
Tetrahedron Lett. 2000, 41, 9431-9435.
(3) Ramart-Lucas, S.-L. C. R. Acad. Sci. 1928, 188, 1301.
10.1021/jo035439z CCC: $27.50 © 2004 American Chemical Society
Published on Web 02/17/2004
J . Org. Chem. 2004, 69, 2017-2023
2017

Mladenova et al.
Resu lts a n d Discu ssion
P r od u ct Stu d ies. Diol 1a was prepared by treating
methyl 9-hydroxy-9H-fluorenecarboxylate with phenyl-
magnesium bromide according to a literature procedure.10
Diol 1b was obtained by condensation of 9,9-dichloro-
fluorene with benzaldehyde via the 9-chlorofluorenyl
anion obtained by metal halogen exchange. The inter-
mediate epoxide underwent rapid ring opening in the
acidic aqueous workup, giving diol 1b in 75% yield. Diol
1c was synthesized by LAH reduction of 9-hydroxy-9-
fluorenecarboxylic acid. The acid-catalyzed thermal reac-
tions were carried out in CDCl3 solutions with catalytic
quantities of toluenesulfonic acid. Diol 1a under these
conditions gave two principal ketones, 2a and 3 (R1 ) R2
) Ph) in a ratio of 3:1 (eq 2) similar to observations
ratio, respectively. Under these conditions no product
from biphenyl migration (i.e., 3) was produced, suggesting
that the primary intermediate involved is the more stable
fluorenyl rather than benzyl cation. By contrast, irradia-
tion of 1b resulted in the formation of 9-phenyl fluorene
(8) as the major product (33%) with minor amounts of
ketone 2b1 (8%) and aldehyde 2b2 (trace amount) (eq 4).
previously reported for the reaction of the same diol.11
This distribution is dependent on the acid concentration,
with the thermodynamic product 3 predominating in
more acidic solutions. Equilibration between the two
ketones takes place in strong acid. Irradiation of 1a in
methanol/acetonitrile (4:1 v/v) solution produced 9-ben-
zoyl-9-phenyl fluorene (2a ) as the major product (31%)
along with epoxide 5 (R1 ) R2 ) Ph) (18%) representing
dehydration of diol 1a . Formation of epoxide intermedi-
ates in acid-catalyzed pinacol rearrangements has been
previously observed.12 The product from methanol quench-
ing (4%) represents a minor component of the reaction
mixture, indicating the relatively high efficiency of the
unimolecular process as compared to solvent quenching.
Some 9-phenylfluorene (8) was observed on prolonged
radiation and was established to be a secondary photo-
product derived from 2a by independent experiments. It
is interesting to note the absence of ketone 3 in the
photochemical process, suggesting that fluorenyl cation
is the principal ionic species involved in the photochemi-
cal reaction. The remaining products 4 and 6a appear to
be derived from radical pathways involving homolytic
C-O or C-C dissociations. Alkene 4 is likely derived
from deoxygenation of epoxide 5 by a carbene such as
tetraphenylmethylene or fluorenylidene, a process which
has been previously documented.13 The same alkene 4 is
obtained from the photolysis of epoxide 5 (vide infra).
The acid-catalyzed thermal reaction of 1b gave two
regioisomeric carbonyl derivatives, 2b1 and 2b2 in a 2:1
Hydrocarbon 8 is likely a secondary photoproduct arising
from photodecarbonylation of aldehyde 2b2. Irradiation
of a solution of aldehyde 2b2 gave clean conversion to
9-phenylfluorene (8) under the same conditions. On the
other hand, photodecarbonylation of ketone 2b1 is not a
favorable process, as was verified independently showing
2b1 to be photostable under these conditions. Thus, the
major pathway involves a phenyl migration in the
photochemical process, in contrast to the acid-catalyzed
dark reaction. A small amount of the methanol quenched
product 7b (2%) from the corresponding 9-fluorenyl cation
accompanied the photoproducts. A small amount of the
radical derived product 6b was also observed in this
mixture.
The photolysis of diols 1a and 1b in TFE, a solvent
used to detect the intermediate carbocations (vide infra),
did not show significant differences in product distribu-
tion from methanol/acetonitrile solutions except for slightly
larger yields of cation-derived products and the absence
of TFE-quenched products.
Diol 1c was stable to the acid-catalyzed thermal
reaction conditions used for the pinacol rearrangement
of diols 1a and 1b. On the other hand, photolysis of diol
1c resulted in the formation of fluorenone (22%) and
parent fluorene (10%) as the major products, with a small
yield of the methanol quenched product 7c (eq 5). The
remaining material consists of oligomeric products that
could not be identified. The formation of fluorene can be
rationalized in terms of an ionic process involving hydride
shift of the corresponding fluorenyl cation to give
9-formylfluorene followed by photodecarbonylation. Al-
ternatively, a radical process involving C-C homolysis
(10) Meerwein, H. J ustus Liebigs Ann. Chem. 1913, 396, 200.
(11) (a) Bachmann, W. E.; Sternberger, H. R. J . Am. Chem. Soc.
1933, 55, 3819-3821. (b) Perz, M. R.; Boyer, J . C. R. Acad. Sci. Paris
1968, 267, 1169-1172.
(12) Pocker, Y.; Ronald, B. P. J . Am. Chem. Soc. 1970, 92, 3385-
3392.
(13) (a) Su, M.-D.; Chu, S. Y. Chem. Phys. Lett. 2000, 320, 4775-
480. (b) Pezacki, J . P.; Wood, P. D.; Gadosy, T. A.; Lusztyk, J .;
Warkentin, J . J . Am. Chem. Soc. 1998, 120, 8681. (c) Turro, N. J .;
Aikawa, M.; Butcher, J . A., J r.; Griffin, G. W. J . Am. Chem. Soc. 1980,
102, 5127.
2018 J . Org. Chem., Vol. 69, No. 6, 2004

Pinacol Rearrangements of Unsymmetrical Diols
SCHEME 1
of the corresponding fluorenyl radical would result in
9-fluorenyl radical, which on hydrogen abstraction gives
fluorene (Scheme 1). In a labeling experiment, the
photolysis of 9-(R,R-dideutero-hydroxymethyl)-9-fluorenol
resulted in exclusive formation of 9,9-dideutero-fluorene
with no monodeuterated fluorene detected as evident
in nitrogen-saturated and aerated hexafluoro-2-propanol
(HFIP) are shown in Figure 1. The spectra show a strong
band with λmax at 510 nm and a shoulder at 470 nm that
decays with a rate constant of 2.0 × 105 s-1 in both air-
and nitrogen-purged solutions. There is a second absorp-
tion with λmax at 370 nm that decays with a rate constant
of 3 × 105 s-1 in nitrogen-saturated solution. Part of the
370 nm absorption decays rapidly in the presence of air
or oxygen; the transient spectra shown in Figure 1B are
measured after this rapid oxygen-induced decay. There
is still a weak absorption below 370 nm that decays with
kinetics similar to those of the 510 nm species. This
indicates that there are two species at 370 nm: a weak
band that has kinetics similar to those at 510 nm and a
second species that reacts rapidly with oxygen and thus
is not visible in the spectrum for the aerated sample. The
1
from H NMR spectroscopy and mass spectrometry. The
latter observation is consistent with the ionic pathway
leading to 9-deuteroformyl-9-deuteriofluorene as an in-
termediate followed by a photodecarbonylation via a
radical cage complex. Photodecarbonylation of â,γ-
unsaturated aldehydes occurs with high quantum ef-
ficiencies, with incorporation of the aldehydic hydrogen
as was reported from deuterium labeling.14
The formation of epoxides as intermediates in pinacol
rearrangements has been attributed to internal capture
of the carbocation by the â-hydroxyl group.12 The pos-
sibility of the epoxide-derived zwitterion (from fluorene-
C(9)-O bond cleavage) formed on irradiation of 1a in
neutral solution was explored in view of the significant
amounts of epoxide 5 (R1 ) R2 ) Ph) produced. Such a
species might be expected from heterolytic dissociation
of hydroxide followed by proton transfer from the result-
ing fluorenyl cation. A similar zwitterion was observed
in the photolysis of 9-hydroxyfluorene-9-carboxylic acid
in neutral solution.9 Although the zwitterion is expected
to be the thermodynamic product from the more acidic
carboxylic acid group, similar proton transfer in the case
of the diols is not as apparent. Epoxide 5 (R1 ) R2 ) Ph)
was prepared independently by a literature procedure15
and subjected to photolysis under the same conditions
as 1a . A similar distribution of photoproducts was
observed as in the case of diol 1a with the exception that
the methanol-quenched product 7a was not detected in
the epoxide photolysis. Although the presence of the
zwitterion intermediate in the diol photolysis could not
be unequivocally established from the epoxide photo-
product distribution, the absence of 7a in the epoxide
reaction suggests that such zwitterions if formed are
shorter-lived relative to their conjugate acid (the corre-
sponding fluorenyl cation) so as to preclude solvent
capture. Such behavior for the zwitterion would be
expected as a result of the “push-pull” effect on the
mechanism for rearrangement.
La ser F la sh P h otolysis. Time-resolved absorption
spectroscopy was used in order to probe for the presence
of fluorenyl cations in these photoreactions. Transient
spectra obtained following 308 nm excitation of diol 1b
F IGURE 1. Transient absorption spectra obtained after 308
nm excitation of diol 1b in (A) nitrogen- and (B) air-saturated
HFIP. The delays for the spectra are (0) 240 ns, (2) 1.16 µs,
and (O) 6.0 µs after the laser pulse for A and (0) 480 ns, (2)
2.5 µs, (O) 5.7 µs, and ([) 11.3 µs after the laser pulse for B.
(14) (a) Wolf, H.; Gonzenbach, H.-U.; Mu¨ller, K.; Schaffner, K. Helv.
Chim. Acta 1972, 55, 2919. (b) Baggiolini, H.; Hamlow, H. P.; Schaffner,
K. J . Am. Chem. Soc. 1970, 92, 4906.
(15) Luisa, M.; Franco, T. M. B.; Herold, B. J .; Maercker, A. J . Chem.
Soc., Perkin Trans. 2 1991, 119.
J . Org. Chem, Vol. 69, No. 6, 2004 2019

Mladenova et al.
F IGURE 3. Transient absorption spectra measured at delays
of (0) 320 ns, (2) 1.5 µs, and (O) 4.7 µs after 308 nm excitation
of diol 1c in aerated HFIP.
F IGURE 2. Transient absorption spectra measured at delays
of (0) 120 ns, (2) 640 ns, and (O) 2.2 µs after 308 nm excitation
of diol 1b in aerated TFE. The inset shows the quenching of
the signal centered at 370 nm with the triplet quencher, 1,3-
CHD.
Laser flash photolysis of diol 1c in aerated HFIP gave
the spectra shown in Figure 3, which show a strong band
at 500 nm with a small shoulder at 470 nm and a much
weaker band at 350 nm. The short and long wavelength
absorptions decay with similar kinetics (k ) 1.8 × 105
s-1) and are insensitive to oxygen, suggesting that both
belong to cation 9c. This is confirmed by the fact that
both are quenched by methanol with a second-order rate
constant of 9.4 × 106 M-1 s-1. Nitrogen-purged solutions
give a slightly larger amount of signal in the 350-370
nm region, indicating a small amount of triplet formation.
By contrast, a strong 370 nm triplet signal is observed
in aerated TFE. Similarly to results for diol 1b, the cation
can still be detected in 1:1 TFE/HFIP but decays rapidly
(k ∼7.4 × 106 s-1).
510 nm species reacts with methanol; a plot of the
observed rate constant for decay at 510 nm vs methanol
concentration gives a second-order rate constant of 5.9
× 106 M-1 s-1. The 510 nm transient is assigned to the
fluorenyl cation 9b on the basis of its absorption proper-
ties, which are similar to those of many other fluorenyl
cations, its insensitivity to oxygen, and its methanol
reactivity.
Transient spectra obtained following 308 nm excitation
of diol 1a in oxygen-saturated HFIP are shown in Figure
4A. The main feature is a species with λmax at 370 nm
that is oxygen-sensitive and that we assign to a triplet
as for diols 1b and 1c. There is also a weak band at 500
nm that decays with a rate constant of 1.8 × 107 s-1; the
kinetics were measured in the presence of sufficient 1,3-
cyclohexadiene to reduce the triplet lifetime to <10 ns
in order to ensure that the decay kinetics at 500 nm were
not complicated by contributions from the triplet. On the
basis of these characteristics and its similarity to other
fluorenyl cations,5 the 500 nm species is assigned to
cation 9a . Transient spectra in TFE show only a triplet
at 360 nm (which is quenched by 1,3-cyclohexadiene with
a rate constant of 2.6 × 109 M-1 s-1) with no evidence for
a short-lived cation at longer wavelength. Spectra ob-
tained after complete decay of the triplet absorption show
a weak residual absorption with λmax at 340 and 500 nm
that is consistent with ketyl radical.17
Excitation of diol 1b in aerated TFE gave a spectrum
with λmax at 370 nm and only very weak absorption at
510 nm (Figure 2). The 370 nm transient decays with
rate constants of 1 × 106 s-1 and 3.9 × 106 s-1 in nitrogen-
and air-saturated solutions, respectively. Because the
oxygen sensitivity is consistent with either radical or
triplet species, an additional experiment with 1,3-cyclo-
hexadiene was carried out. The diene quenched the 370
nm transient with a rate constant of 2.8 × 109 M-1 s-1
,
indicating that it can reasonably be assigned to a triplet.
This assignment is also consistent with the reported
triplet-triplet absorption of fluorene at 376 nm in
ethanol.16 After decay of the 370 nm transient in oxygen-
or air-purged samples there is a weak and longer lived
signal with λmax at 390 and 500 nm. There is no evidence
for a cation with a lifetime in excess of 20-30 ns in HFIP.
However, a cation is readily detectable at 510 nm in 1:1
TFE/HFIP and decays with a rate constant of ∼7 × 106
s-1, over an order of magnitude larger than that mea-
sured in HFIP.
Excitation at 308 nm of epoxide 5 in HFIP gave results
that were very similar to those for diol 1a (Figure 4B). A
strongly absorbing oxygen-sensitive transient at 370 nm
and a weak signal at 500 nm that decayed with a rate
constant of 1.3 × 107 s-1 were both observed and were
assigned to triplet and cation, respectively. The cation
(16) Carmichael, I.; Hug, G. L. J . Phys. Chem. Ref. Data 1986, 15,
1-250.
(17) Biczok, L.; Berces, T.; Linschitz, H. J . Am. Chem. Soc. 1997,
119, 11071-11077.
2020 J . Org. Chem., Vol. 69, No. 6, 2004

Pinacol Rearrangements of Unsymmetrical Diols
TABLE 2. Activa tion En er gies a n d En er gies for
Con ver sion of F lu or en yl Ca tion s
∆Eq (kcal mol-1
(level of calcn)
)
∆E (kcal mol-1
)
cation lifetime
(s-1) (HFIP)
cation
(level of calcn)
10
26.1 (AM 1)
20.0 (HF-3-21G*) -14.6 (HF-3-21G*)
21.3 (AM 1)
17.0 (HF-3-21G*)
27.1 (AM 1)
20.8 (HF-3-21G*)
25.4 (AM 1)
-19.0 (AM 1)
2.3 × 10-6
5.6 × 10-8
5.0 × 10-6
5.6 × 10-6
9a
-7.7 (AM 1)
-4.8 (HF-3-21G*)
+10.7 (AM 1)
+8.7 (HF-3-21G*)
+5.9 (AM 1)
+1.0 (HF-3-21G*)
9ba
9c
30.6 (HF-3-21G*)
a
Energies represent phenyl migration. ∆Εq for hydride migra-
tion is +20.5 (AM1) and +27.2 (HF/3-21G*) kcal mol-1
.
only weak signals that were consistent with a small yield
of a fluorene-like triplet, with additional weaker residual
absorptions that may be due to the fluorenyl ketyl
radical.
Cation 9a is significantly shorter-lived relative to 10,
the cation derived from photoheterolysis of 9,9′-bifluo-
rene-9,9′-diol (Table 1).8 This is attributed to the relative
rates of formation of the phenonium ion intermediates
11 involved in these aryl migrations.12 The corresponding
phenonium ion 12 associated with the biphenylene re-
arrangement of ion 10 has a strained bicylo[3.1.0]heptane
moiety imbedded in its structure leading to an expected
larger activation barrier relative to that of 1a . The longer-
F IGURE 4. Transient absorption spectra measured following
308 nm excitation of (A) diol 1a in HFIP and (B) epoxide 4 in
acetonitrile. The delays for the spectra are (0) 56 ns, (2) 208
ns, and (O) 928 ns after the laser pulse for A and (0) 28 ns,
(2) 96 ns, ([) 244 ns, and (O) 424 ns after the laser pulse for
B.
TABLE 1. Lifetim es a n d Meth a n ol Qu en ch in g Ra te
Con sta n ts of F lu or en yl Ca tion s
fluorenyl
cation
lifetime (s)
(solvent)
KMeOH (M-1 s-1
)
(solvent)
lived cation 9b as compared to 9a is likely the result of
the absence of a gem-phenyl group in the former, which
would stabilize the phenonium ion by possible π-inter-
action. MO calculations for the corresponding phenonium
ions show a shallow energy surface near or at the
transition state. The calculated activation energies (at
AM 1 and HF/3-21G* levels) for the rearrangement of
the fluorenyl cations 9a -c and 10 to the corresponding
oxonium ions exhibit a qualitative trend with their
lifetimes (Table 2).
9a
9b
5.6 × 10-8 (HFIP)
1.4 × 10-7 (1;1 TFE/HFIP)
5.0 × 10-6 (HFIP)
5.9 × 106 (HFIP)
9.4 × 106 (HFIP)
9c
10
1.4 × 10-7 (1:1 TFE/HFIP)
5.6 × 10-6 (HFIP)
6.7 × 10-7 (1:1 TFE/HFIP)a
2.3 × 10-6 (HFIP)a
1.2 × 107 (1:1TFE:HFIP)
2.1 × 106 (HFIP)a
8.3 × 10-8 (TFE)a
a
From ref 8.
The kinetic stabilities of these fluorenyl cations as
measured by methanol quenching rates (Table 1) show
a trend that correlates with steric accessibility of the
cationic center.
decay kinetics were measured in the presence of sufficient
diene to completely quench the triplet. Both spectra and
kinetics were similar for the 500 nm transients derived
from diol 1a and epoxide 5, although the short lifetime
and weak signals make these results less reliable than
those for diols 1b and 1c. A zwitterion formed by
photolysis of epoxide 5 would be expected to have a
spectrum and/or decay kinetics different than those of
the initial cation derived from diol 1a , based on earlier
observations for 9-hydroxyfluorene-9-carboxylic acid.9
There is no evidence for this, although the weak and
short-lived signals obtained would make it difficult to
pick up small differences. These observations do not
necessarily preclude the presence of the zwitterion in the
photochemistry of 1a and 5. The zwitterion once formed
may undergo diffusion-controlled protonation with the
protic solvents used for the LFP experiments. Laser flash
photolysis experiments for epoxide 5 in acetonitrile gave
Su m m a r y
The above results demonstrate that there are signifi-
cant differences between the acid-catalyzed thermal and
photochemical reactions of nonsymmetrical pinacols 1a -
c. Whereas the thermal reaction products are predicated
by the formation of the more stable cation, as well as
other factors, the photochemical processes of these diols
result in the initial formation of the fluorenyl cations as
evident from the LFP observations. These differences are
evident in the product distributions obtained from diol
1a . Whereas biphenylene rearrangement products are
observed in the acid-catalyzed dark reactions, the prod-
ucts from the ionic pathway in the photochemical process
J . Org. Chem, Vol. 69, No. 6, 2004 2021

Mladenova et al.
originate from the fluorenyl cation. The acid-catalyzed
thermal reaction of diol 1b gives products derived only
from the fluorenyl cation. The corresponding benzyl
cation from 1b would be expected to be less stable than
the fluorenyl cation. Hydride migration appears to be
preferred over phenyl shift by a ratio of 2:1 in the dark
reaction. The photolysis of 1b leads to secondary photo-
products that originate from the pinacol rearrangement.
The product distribution indicates a preference for phenyl
over hydride migration. This difference may be explained
in terms of the presence of acid in the dark reaction,
which could result in thermodynamic equilibration be-
tween 2b1 and 2b2 favoring the former, although this
speculation could not be independently confirmed under
the conditions of the dark reaction. Alternatively, a later
transition state in the acid reaction favoring the ther-
modynamic product may be influencing the course of the
dark reaction. Similar observations have been reported
by other groups where pinacol rearrangement of 1,2-
diphenylethane-1,2-diol under neutral heterogeneous
conditions results in products favoring phenyl over hy-
dride migration by a factor of 3:1.19 These observations
are in sharp contrast to classical studies of the same
reaction carried out in acidic conditions.20 Thus, our
results show that in neutral solutions phenyl migration
is preferred over hydride even though the phenyl group
is better than hydrogen in stabilizing the secondary
oxonium ion. The pinacol rearrangement of diol 1c still
takes place on photoexcitation unlike the dark acid
conditions. The primary product, 9-formyfluorene, un-
dergoes rapid decarbonylation under these conditions as
evident from the labeling studies.
Exp er im en ta l Section
Photolyses were performed using a 450 W medium-pressure
mercury arc lamp in a water-cooled quartz immersion well.
Pyrex tubes containing samples dissolved in the solvent system
specified were strapped around the immersion well, and the
whole assembly was immersed in a water bath. Deaeration of
the solutions was accomplished by purging with nitrogen or
argon. Diol 1a was prepared according to a literature proce-
dure.10 NMR spectra were obtained from a 400 MHz spec-
trometer.
Laser flash photolysis experiments were carried out at room
temperature using 6 ns 308 nm pulses, from an excimer laser
(XeCl; <40 mJ /pulse). The laser system has been described in
detail.21 Solutions were prepared by adding a small amount
of a stock solution (2 × 10-4 M) of the samples in TFE or HFIP.
Solutions were contained in 7 × 7 mm2 Suprasil quartz cells
and were either aerated or purged with nitrogen or oxygen
before laser excitation.
9-(Hydr oxy-ph en yl-m eth yl)-9H-flu or en -9-ol (1b). To 9,9-
dichlorofluorene22 (0.5 g, 2.14 mmol) in THF (25 mL) cooled
to -78 °C was added butyllithium (0.85 mL, 2.5 M in hexane).
After 10 min at -78 °C, the mixture was treated with a
solution of benzaldehyde (0.22 g, 2.14 mmol) in THF (5 mL).
The mixture was left to warm to ambient and added to an
aqueous solution (30 mL) acidified with 2 M HCl. The organic
layer was extracted with dichloromethane (2 × 10 mL), and
the extract was evaporated to dryness. The mixture was
separated by chromatography (silica gel; hexane/ethyl acetate
(8:1)) and yielded 0.46 g (75%) of the diol as a white solid, mp
127 °C (lit.23 mp 126 °C): 1H NMR δ 2.82 (s, 1H, OH), 3.17 (d,
1H, OH, J ) 1.96 Hz), 5.27 (d, 1H, J ) 3.76 Hz), 6.75 (d, 2H,
J ) 7.6 Hz), 6.89 (t, 2H, J ) 7.56 Hz), 6.97 (d, 1H, J ) 8.36
Hz), 7.28-7.23 (m, 4H), 7.35 (t, 2H, J ) 3.32 Hz), 7.56(d, 1H,
J ) 3.36 Hz), 7.64 (d, 1H, J ) 7.16 Hz).
9-Hyd r oxym eth yl-9H-flu or en -9-ol (1c). To a solution of
LiAlH4 (0.03 g, 0.78 mmol) in 10 mL THF was added 9-hy-
droxy-9-fluorenecarboxylic acid (0.2 g, 0.88 mmol) in THF (15
mL), and the mixture was left to stir for 30 min. The mixture
was quenched with methanol (5 mL) and then water (10 mL).
The mixture was extracted with dichloromethane (3 × 5 mL)
and run through a silica gel plug, and the extract was
evaporated to dryness to yield 0.18 g (97%) of the diol as a
white solid, mp 94-96 °C: 1H NMR δ 7.64 (dd, 4H, J ) 7.52,
7.48 Hz), 7.40 (t, 2H, J ) 7.42 Hz), 7.31 (t, 2H, J ) 7.44 Hz),
3.85 (d, 2H, J ) 6 Hz), 2.51 (s, 1H, OH), 2.33-2.30 (t, 1H,
OH), 1J ) 6.46 Hz) ppm. MS data of this compound were
previously reported.24
The intermediacy of a zwitterion, resulting from pho-
toheterolysis of diol 1a followed by proton transfer, was
examined by comparison of photoproducts obtained from
epoxide 5, as well as by comparison of transients in the
LFP experiments. Product distributions were similar
except for the absence of the methanol-quenched product
7b in the photolysis of epoxide 5. Although this suggests
that the zwitterion is shorter-lived than its conjugate acid
(9a ), the LFP results indicate that transients with similar
spectra and lifetimes are obtained from both diol and
epoxide. On the basis of the combined transient and
product studies, it is difficult to conclude whether the
transient generated from epoxide 5 is an initial zwitterion
or whether rapid protonation leads to the conjugate acid,
cation 9a , which is also generated by photolysis of diol
1a . An alternate possibility is that the zwitterion and
conjugate acid 9a are in equilibrium in HFIP, although
the differences in product distribution suggest that this
is not the case in methanol/acetonitrile. Because nucleo-
philic trapping gives little or no product in all of the
solvents examined, rearrangement is likely to be the
primary reaction in each case. Further studies associated
with related diols and epoxides substituted with electron-
withdrawing groups so as to stabilize such zwitterions
are in progress.
Did eu ter a ted 9-Hyd r oxym eth yl-9H-flu or en -9-ol (1c-
d 2). The same procedure as for the preparation of diol 1c was
followed, but LiAlD4 was used instead of LiAlH4. The yield
was the same, and the 1H NMR spectrum was identical to that
of diol 1c, except that the peak at δ 3.8 ppm was absent.
Acid -Ca ta lyzed Da r k Rea ction s. Gen er a l P r oced u r e.
Diols 1a -c (10 mg) were dissolved in 1 mL of CDCl3 containing
a crystal of toluenesulfonic acid and monitored by NMR
spectroscopy with comparison to authentic samples prepared
according to the literature.
P r od u ct Stu d ies. Gen er a l P r oced u r e. Nitrogen-purged
solutions of diols 1a -c (2 × 10-3M) in 50 mL of methanol/
acetonitrile (4:1) were irradiated in a Pyrex tube for 1 h. After
evaporation of the solvent the residue was separated by
preparative TLC (silica gel, ethyl acetate/hexane (1:8)).
P h otop r od u cts fr om Diol 1a . 9-Ben zoyl-9-p h en ylflu o-
r en e (2a ): 31%; mp 168-170 °C (lit.25 172 °C); 1H NMR δ 7.80
(18) Matsumoto, Kiyoshi; Goto, Ryozo; Sera, Akira; Asano, Tsutomu.
Nippon Kagaku Zasshi 1966, 87, 1076-82.
(21) Kazanis, S.; Azarani, A.; J ohnston, L. J . J . Phys. Chem. 1991,
95, 4430.
(22) Ray, F. E.; Albertson, C. E. J . Chem. Soc. 1905, 87, 1249.
(23) Corey, E. J .; Noe, M. C. J . Am. Chem. Soc. 1996, 118, 11038.
(24) Verardo, G.; Strazzolini, P.; Giumanini, A. G.; Poiana, M. J .
Photochem. Photobiol., A 1989, 48, 115.
(19) Hsien, M.; Sheu, H.-T.; Lee, T.; Cheng, S.; Lee, J .-F. J . Mol.
Catal. A: Chem. 2002, 181, 189-200.
(20) Daniloff, S.; Venus-Danilova, E. Ber. Dtsch. Chem. Ges. 1926,
59B 1032-43.
2022 J . Org. Chem., Vol. 69, No. 6, 2004

Pinacol Rearrangements of Unsymmetrical Diols
(d, 2H, J ) 7.52 Hz), 7.45-7.38 (m, 4H), 7.32-7.23 (m, 10H),
7.12 (t, 2H, J ) 7.64 Hz) ppm.
P h ot op r od u ct s fr om Diol 1b . 9-P h en ylflu or en e (8):
33%; comparison with a commercial sample.
Ep oxid e 5 (R1 ) R2 ) C6H5): 18%; 1H NMR δ 7.67 (d, 2H,
J ) 7.52 Hz), 7.31-7.23 (m, 12H), 6.87 (t, 2H, J ) 7.60 Hz),
6.11 (d, 2H, J ) 7.68 Hz) compared with a sample prepared
independently according to a literature method.15
(9H-F lu or en -9-yl)-d ip h en yl-m eth a n ol (6a ): 10.5%; 1H
NMR δ 7.70 (dd, 6H, J ) 7.60, 7.60 Hz), 7.39-7.30 (m, 8H),
6.99 (t, 2H, J ) 7.60 Hz), 6.52 (d, 2H, J ) 7.68 Hz), 5.28 (s,
1H), 1.95 (s, 1H, OH). All spectral features identical with a
sample prepared independently and fully characterized in our
laboratory.5
9-Ben zoylflu or en e (2b1): 8%; mp 133-134 °C (lit.25 135
1
°C); (7.6%), H NMR δ 7.84 (d, 2H, J ) 7.60 Hz), 7.73 (d, 2H,
J ) 5.92 Hz), 7.50 (t, 2H, J ) 7.08 Hz), 7.44-7.33 (m, 5H),
7.26 (t, 2H, J ) 7.60 Hz), 5.60 (s, 1H) ppm; MS m/z 270 (5),
165 (5), 105 (20).
9-P h en yl-9-flu or en eca r boxa ld eh yd e (2b2): 1%; decom-
poses on silica gel; 1H NMR δ 9.56 (s, 1H), 7.82 (d, 2H), 7.48-
7.24 (m, 11H); MS m/z 270 (5), 241 (100).
(9H-F lu or en -9-yl)-p h en yl-m eth a n ol (6b): 3.1%; mp 120-
3
121 °C (lit.26 118.5-119 °C); 1H NMR δ 5.09 (d, 1H, J ) 3.72
Meth a n ol Ad d u ct 7a : this material coelutes with alcohol
Hz), 4.40 (d, 1H, J ) 6.04 Hz).
1
1
6a ; 0.8 mg (4% by H NMR spectroscopy); H NMR δ 7.68 (d,
1H, J ) 7.56 Hz), 7.53 (d, 1H, J ) 4.56 Hz), 7.37-7.25 (m,
12H), 6.83 (t, 2H, J ) 7.60 Hz), 6.65 (d, 2H, J ) 7.80 Hz), 2.94
(s, 3H, OCH3).
(9-Meth oxy-9H-flu or en -9-yl)-ph en yl-m eth an ol (7b): 1.5%;
1
mp 185-187 °C (lit.27 189-190 °C); H NMR δ 7.68 (d, 1H, J
) 7.60 Hz), 7.54 (d, 1H, J ) 4.64 Hz), 7.37-7.27 (m, 6H), 6.96
(t, 1H, J ) 7.28 Hz), 6.83 (t, 2H, J ) 7.64 Hz), 6.65 (d, 2H, J
) 7.68 Hz), 5.28 (s, 1H), 3.57 (s, 1H, OH), 2.94 (s, 3H, OCH3);
MS m/z 195 (100), 107 (20), 77 (40).
9-Ben zh yd r ilyd en eflu or en e (4): 10%; this material was
previously characterized5 and compared with a commercial
sample.
P h otop r od u cts fr om Diol 1c: fluorene (10%), fluorenone
(22%), (9-m eth oxy-9H-flu or en -9-yl)-m eth a n ol (7c) (4%); 1H
NMR δ 7.69 (d, 2H, J ) 5.68 Hz), 7.58 (d, 2H, J ) 7.44 Hz),
7.44 (t, 2H, J ) 7.44 Hz), 7.34 (t, 2H, J ) 7.44 Hz), 3.84 (d,
P h otop r od u cts fr om Ep oxid e 5 (R1 ) R2 ) C6H5).
Products compared with those from diol 1a : 9-benzoyl-9-
phenylfluorene (2a ) (40%), 9-benzhydrilydenefluorene (4) (7%),
and 9-phenylfluorene (8) (10%) (this compound formed as a
secondary photoproduct from photoexcitation of 2a as shown
by an independent experiment).
3
3
1H, J ) 6.40 Hz) 2.89 (s, 3H, OCH3) 2.41 (t, 1H, OH), J )
6.36 Hz).
Ack n ow led gm en t. We thank the Natural Sciences
and Engineering Council of Canada (N.S.E.R.C.) for
financial support.
(25) Rouzaud, J .; Cauquil, G.; Boyer, J . Bull. Soc. Chim. Fr. 1965,
2345-2353.
(26) Kliegl, A.; Weng, F.; Wiest, G. Ber. Dtsch. Chem. Ges. 1930,
63, 1262.
(27) J onczyk, A. J . Org. Chem. 1979, 44, 1192-1194.
J O035439Z
J . Org. Chem, Vol. 69, No. 6, 2004 2023