Catalytic, Highly Regio- And Chemoselective Generation of Radicals from Epoxides: Titanocene Dichloride as an Electron Transfer Catalyst in Transition Metal Catalyzed Radical Reactions

Full text of DOI:10.1002/(sici)1521-3773(19980202)37:1/2<101::aid-anie101>3.0.co;2-w

COMMUNICATIONS
with the 6-31G* basis set for the reaction coordinate and frequency
computations. This reduced CAS does not include the s and s*
orbitals of the ªshortº C N bond and the lone-pair orbital on the
alternative N atom. CAS(8,7) produces optimized geometries of MIN,
TS, and LP which differ only slightly from that of CAS(12,10) (see
Figure 1, top). The reaction coordinate was determined with the IRC
method available in Gaussian 94. The conical intersection is easily
located by computing the reaction coordinate until degeneracy of S1
and S0 is achieved. This manifests itself in failure of the MCSCF
calculation to converge owing to S1 ± S0 near-degeneracy. The last
reaction coordinate point before degeneracy (LP) is taken as
representative of the conical intersection.
nistically viewed as a thermally activated process, as previ-
ously established for many triplet ketones.[4,21]
In conclusion, the combined experimental and theoretical
data for the solvent-induced quenching of n,p* excited singlet
azoalkanes support a hydrogen abstraction in which a
transition state is followed by a conical intersection as the
reaction coordinate. This photochemical reaction mechanism
should be general for n,p* singlet excited chromophores and
provides an important rationale for their reactivity and
efficiency in hydrogen abstractions.
[19] a) B. O. Roos, Adv. Chem. Phys. 1987, 69, 399 ± 446; b) the MCSCF
program used is implemented in Gaussian 94, Revision B.2: M. J.
Frisch, G. W. Trucks, H. B. Schlegel, P. M. W. Gill, B. G. Johnson,
M. A. Robb, J. R. Cheeseman, T. Keith, G. A. Petersson, J. A.
Montgomery, K. Raghavachari, M. A. Al-Laham, V. G. Zakrzewski,
J. V. Ortiz, J. B. Foresman, C. Y. Peng, P. Y. Ayala, W. Chen, M. W.
Wong, J. L. Andres, E. S. Replogle, R. Gomperts, R. L. Martin, D. J.
Fox, J. S. Binkley, D. J. Defrees, J. Baker, J. P. Stewart, M. Head-
Gordon, C. Gonzalez, J. A. Pople, Gaussian, Inc., Pittsburgh, PA,
USA, 1995.
[20] a) K. Andersson, P.-A. Malmqvist, B. O. Ross, J. Chem. Phys. 1992, 96,
1218 ± 1226; b) MOLCAS, Version 3: K. Andersson, M. R. A. Blom-
berg, M. Fülscher, V. Kellö, R. Lindh, P.-A. Malmqvist, J. Noga, J.
Olsen, B. O. Roos, A. J. Sadlej, P. E. M. Siegbahn, M. Urban, P. O.
Widmark, University of Lund, Sweden, 1994.
Received: June 11, 1997 [Z10537IE]
German version: Angew. Chem. 1998, 110, 103 ± 107
Keywords: ab initio calculations ´ azo compounds ´ hydro-
gen transfer ´ isotope effects ´ photochemistry
[1] L. Salem, C. Leforestier, G. Segal, R. Wetmore, J. Am. Chem. Soc.
1975, 97, 479 ± 487.
[2] N. J. Turro, J. McVey, V. Ramamurthy, P. Lechtken, Angew. Chem.
1979, 91, 597 ± 612; Angew. Chem. Int. Ed. Engl. 1979, 18, 572 ± 586.
[3] W. M. Nau, F. L. Cozens, J. C. Scaiano, J. Am. Chem. Soc. 1996, 118,
2275 ± 2282, and references therein.
[4] M. A. Garcia-Garibay, A. Gamarnik, R. Bise, L. Pang, W. S. Jenks, J.
Am. Chem. Soc. 1995, 117, 10264 ± 10275.
[21] P. J. Wagner, Q. Cao, R. Pabon, J. Am. Chem. Soc. 1992, 114, 346 ± 348.
[22] P. S. Engel, L. R. Soltero, S. A. Baughman, C. J. Nalepa, P. A. Cahill,
R. B. Weisman, J. Am. Chem. Soc. 1982, 104, 1698 ± 1700.
[5] D. Severance, H. Morrison, Chem. Phys. Lett. 1989, 163, 545 ± 548.
[6] A. E. Dorigo, M. A. McCarrick, R. J. Loncharich, K. N. Houk, J. Am.
Chem. Soc. 1990, 112, 7508 ± 7514.
[23] M. F. Mirbach, M. J. Mirbach, K.±C. Liu, N. J. Turro, J. Photochem.
1978, 8, 299 ± 306.
[7] P. S. Engel, D. W. Horsey, J. N. Scholz, T. Karatsu, A. Kitamura, J.
Phys. Chem. 1992, 96, 7524 ± 7535.
[8] P. S. Engel, D. W. Horsey, D. E. Keys, C. J. Nalepa, L. R. Soltero, J.
Am. Chem. Soc. 1983, 105, 7108 ± 7114.
[9] W. M. Nau, W. Adam, J. C. Scaiano, Chem. Phys. Lett. 1996, 253, 92 ±
96.
[10] M. J. Mirbach, M. F. Mirbach, W. R. Cherry, N. J. Turro, P. Engel,
Chem. Phys. Lett. 1978, 53, 266 ± 269.
[11] P. S. Engel, C. J. Nalepa, Pure Appl. Chem. 1980, 52, 2621 ± 2632.
[12] M. Klessinger, Angew. Chem. 1995, 107, 597 ± 599; Angew. Chem. Int.
Ed. Engl. 1995, 34, 549 ± 551, and references therein.
[13] F. Bernardi, M. Olivucci, M. A. Robb, Chem. Soc. Rev. 1996, 321 ± 328.
[14] B. S. Solomon, T. F. Thomas, C. Steel, J. Am. Chem. Soc. 1968, 90,
2249 ± 2258.
Catalytic, Highly Regio- and Chemoselective
Generation of Radicals from Epoxides:
Titanocene Dichloride as an Electron Transfer
Catalyst in Transition Metal Catalyzed Radical
Reactions**
Andreas Gansäuer,* Marianna Pierobon, and
Harald Bluhm
[15] H. Rau, Angew. Chem. 1973, 85, 248 ± 258; Angew. Chem. Int. Ed.
Engl. 1973, 12, 224 ± 239.
[16] R. P. Bell, The Tunnel Effect in Chemistry, Chapman und Hall,
London, 1980.
During the last two decades the development of efficient
chain reactions has led to an explosive growth in free radical
chemistry.[1] Although the understanding of how substrate
control determines the stereo- and chemoselectivity of these
reactions has reached a high level,[2] to the best of our
knowledge little is known about reagent-controlled, catalytic
transformations of radicals not proceeding as chain reac-
tions.[3] The advantage of this type of reaction is broader
application because the influence of the substrate on the
chemo- and stereoselectivity can ideally be overruled and the
course of the reaction determined solely by the reagent in
catalytic amounts. Therefore a reagent-controlled catalytic
reaction would extend the synthetic utility of free radicals
even further.
[17] For the calculation of kq the ambient-temperature value of t0 ((930 Æ
30) ns, gas phase) was employed for all temperatures. Inclusion of the
much smaller temperature dependence of t0, which implicitly includes
any temperature dependence of kf and kd and the possible contribu-
tion due to photodecomposition, does not affect the activation
parameters within the limits of statistical regression error. For
comparison, the gas-phase lifetimes are 950 and 885 ns at 20 and
658C, respectively; that is, they are within the limits of error of or very
close to the ambient value.
[18] All MCSCF geometry optimizations were carried out with a complete
active space (CAS) including 12 electrons in 10 orbitals (CAS(12,10)).
ˆ
The CAS orbitals comprise the p and p* N N orbitals, the four s and
s* C N orbitals, the two nitrogen lone pair orbitals of the pyrazoline
fragment, and the s and s* orbitals of the reactive C H bond of
CH2Cl2. To improve the description of the H transfer the standard 6-
31G* basis set (double-z ‡ d-type polarization function for atoms of
the first and seconds rows of the periodic system) was augmented with
p-type polarization and s-type diffuse functions on the CH2Cl2
hydrogen atoms and with sp-type diffuse functions (included in
Gaussian 94)[19b] on the nitrogen centers. To improve the energetics by
including the effect of dynamic electron correlation, the S1 energies of
MIN, TS, and LP were recomputed using multireference Mùller±
Plesset perturbation theory with the PT2F method included in
MOLCAS.[20] To reduce the computational cost we used a CAS(8,7)
[*] Dr. A. Gansäuer, M. Pierobon, H. Bluhm
Institut für Organische Chemie der Universität
Tammannstrasse 2, D-37077 Göttingen (Germany)
Fax: Int. code ‡ (49)551-392944
e-mail:agansae@gwdg.de
[**] This work was supported by the Fonds der Chemischen Industrie and
the Socrates Program. We thank Prof. R. Brückner for constant
support and encouragement.
Angew. Chem. Int. Ed. 1998, 37, No. 1/2
ꢀ WILEY-VCH Verlag GmbH, D-69451 Weinheim, 1998
1433-7851/98/3701-0101 $ 17.50+.50/0
101

COMMUNICATIONS
Table 1. Optimization of the reductive opening of epoxide 1 with 5 mol%
[Cp2TiCl2] and 1,4-cyclohexadiene.
An interesting alternative to radical chain reactions is the
reductive opening of epoxides by single electron transfer
(SET) from titanocene chloride dimers.[4] The b-titanoxy
radical thus formed can undergo the typical radical reactions.
Epoxides can be prepared conveniently from a number of
functional groups, for example olefins[5] and carbonyl com-
pounds,[6] and can thus serve as radical precursors. An
attractive feature of this reaction is that an alcohol is formed
after hydrolysis rather than a hydrocarbon, as in the reactions
of halides with stannanes.[7] However, both the electron
transfer reagent precursor, titanocene dichloride, and the
reductant, usually zinc powder, are employed in at least
stoichiometric quantities. This is clearly disadvantageous
when more complex titanocenes are used.[8] If the reaction
could be conducted with a catalytic amount of titanocene, this
problem would be solved, and the way would be paved for
transition metal catalyzed radical reactions with the potential
for efficient reagent control. To achieve this goal the
titanocene alkoxide formed must be converted into a deriv-
ative of the alcohol or the alcohol itself and titanocene
dichloride, the catalyst precursor. In situ reduction by the
stoichiometric reductant would yield the redox-active titan-
ium(iii) complex[9] and complete the catalytic cycle.
Recently Fürstner et al. have devised efficient catalytic
McMurry[10] and Nozaki ± Hiyama couplings[11] by adding
chlorotrimethylsilane to the reaction mixture. The metal
alkoxides and oxides formed in the couplings are thus
silylated and the catalytically active metal species are
regenerated. This method is, however, not applicable to
epoxides because silylated chlorohydrins would form by
Lewis acid initiated ring opening. We reasoned that proto-
nation of the titanium-bound oxygen atom should also allow
efficient regeneration of the catalyst. For this purpose the acid
must be strong enough for the protonation, but neither may
the metal powder be oxidized nor the catalyst deactivated by
complexation of the resulting base. Pyridinium hydrochlor-
ides (Py-HCl) seemed to be appropriate based on their pKa
values[12] and were therefore tested as acids in the reaction
Hydrochloride
M
t[h]
1:2:3[a]
Py-HCl
Zn
Zn
Zn
Mn
35
36
36
16
28:32:26
30:43:22
15:71[b]:12
0:88[b]:0.5
2,4-Me2Py-HCl
2,4,6-Me3Py-HCl
2,4,6-Me3Py-HCl
[a] Determined from the 1H NMR spectrum of the crude mixture.
[b] Yield of the isolated product.
having protonation as the key step in the catalytic cycle:
1. Protonations are amongst the fastest reactions. 2. 2,4,6-
Collidine can be recovered simply by acid ± base extraction or
distillation; treatment with HCl then provides the hydro-
chloride. Hydrolysis of the silylated alcohols yields hexame-
thyldisiloxane, which cannot be reconverted to chlorotrime-
thylsilane in a straightforward manner. These advantages
should become especially important in large-scale applica-
tions.
In hydrogen atom abstraction reactions with 1,4-cyclo-
hexadiene, manganese is superior to zinc as the stoichiometric
reductant in two respects. When manganese is used, no
chlorohydrin 3 is formed by Lewis acid initiated ring opening
of the epoxides, and the reaction proceeds significantly faster.
Apparently the zinc chloride formed in the reaction com-
plexes the epoxide, prevents complete conversion into
product, and allows the slow chloride-mediated ring opening
to compete. Gratifyingly the much less Lewis-acidic manga-
nese dichloride[11] does not interfer with the catalytic reaction
and thus the products are obtained cleanly in a reasonable
time. It should be noted that other metals such as aluminum[13]
do not lead to fast reduction of titanocene dichloride and are
therefore not suitable as stoichiometric reductants.
To explore the generality of the optimized reaction
conditions a number of epoxides were tested in inter- and
intramolecular C ± C bond forming reactions and reductions
(Table 2). Terminal, 1,1- and 1,2-disubstituted (cis and trans,
entry 3), and trisubstituted epoxides give the desired products
in good yields. It should be noted that deoxygenation of the
epoxide was not detected for any substrate. The regioselec-
tivity of the epoxide opening is opposite to that of SN2
reactions and seems to be governed only by the stability of the
radical formed. Therefore the radical is produced only on
the more substituted carbon atom of the epoxide. The
observed diastereoselectivities of the cyclizations are typical
for radical reactions and their well-known conformational
effects.
For the intermolecular addition of the radical to methyl
acrylate, zinc was used as the stoichiometric reductant
(entries 12 and 13). The lactones were isolated in 75 ± 80%
yield after 40 h but the corresponding chlorohydrins (roughly
5%) were also detected. The reaction was much slower with
manganese. After 40 h the conversion was estimated at 15%.
This is presumably due to deactivation of the titanium catalyst
by the MeOH generated. Zinc chloride seems to be Lewis-
acidic enough to complex MeOH and thus prevent catalyst
deactivation. When manganese was employed in combination
with zinc chloride, essentially the same results were obtained
of 1 with cyclohexadiene as the hydrogen atom donor, a
stoichiometric reductant, and 5 mol% titanocene dichloride
(Table 1).
All of the pyridine hydrochlorides tested, except the bulky
2,6-di-tert-butylpyridine hydrochloride, led to conversion.
2,4,6-Trimethylpyridine hydrochloride (collidine hydrochlor-
ide) was the most efficient. In addition to the milder reaction
conditions required, there are two more general advantages to
102
ꢀ WILEY-VCH Verlag GmbH, D-69451 Weinheim, 1998
1433-7851/98/3701-0102 $ 17.50+.50/0
Angew. Chem. Int. Ed. 1998, 37, No. 1/2

COMMUNICATIONS
Table 2. Titanocene-catalyzed reductive opening of epoxides under the
optimized conditions.[a]
obtained by hydroboration and oxidation, the method descri-
bed here is superior because distant double bonds and ketones
are not tolerated by the hydroboration protocol.[16] Also
hydroboration of 2-methyl-1,11-dodecadiene takes place
preferentially at the less substituted double bond and is
therefore complementary to our reaction (entry 4).[17]
We have demonstrated that the reductive opening of
epoxides with concommitant hydrogen atom abstraction and
inter- or intramolecular C ± C bond formation can be cata-
lyzed efficiently and in a highly chemoselective manner by
titanocene dichloride if manganese or zinc is used as the
stoichiometric reductant and 2,4,6-collidine hydrochloride is
used as an acid to protonate the titanium oxide and regenerate
titanocene dichloride.
Entry Substrate
t[h]
Yield[%]
Product
1
30
78[b]
2
3
30
30
55[c]
83[d]
4
5
16
18
72
58
Experimental Section
All new compounds were characterized by the usual means. Preparation of
cyclododecanol:
A suspension of collidine hydrochloride (197 mg,
6
7
8
18
12
18
86
67
74
69
1.25 mmol), titanocene dichloride (12.4 mg, 0.05 mmol), manganese
(60.0 mg, 1.1 mmol), 1,4-cyclohexadiene (0.425 mL, 4.5 mmol), and cyclo-
dodecene oxide (182 mg, 1.0 mmol) in THF (10 mL) was stirred at room
temperature for 32 h. After extraction (5 mL 2m HCl, 2 Â 20 mL H2O) and
chromatography on silica gel (diethyl ether/petroleum ether 1/3) cyclo-
dodecanol (153 mg, 0.83 mmol) was obtained in 83% yield.
Received: July 18, 1997 [Z10701IE]
German version: Angew. Chem. 1998, 110, 107 ± 109
9
18
60
10
70
95
Keywords: chemoselectivity
catalysis ´ radical reactions ´ regioselectivity
´ epoxides ´ homogeneous
4-phenyl-2-butanone
4-phenyl-2-butanone
11
30
76[e]
[1] a) B. Giese, Radicals in Organic Synthesis: Formation of Carbon-
Carbon Bonds, Pergamon, Oxford, 1986; b) W. B. Motherwell, D.
Crich, Free Radical Chain Reactions in Organic Synthesis, Academic
Press, London, 1991.
[2] a) A. L. J. Beckwith, C. J. Easton, A. K. Serelis, J. Chem. Soc. Chem.
Commun. 1980, 482; b) W. Smadja, Synlett 1994, 1; D. P. Curran, N. A.
Porter, B. Giese, Stereochemistry of Radical Reactions, VCH, Wein-
heim, 1996.
12
13
40
40
77[e]
82[e,f]
[3] a) A. Gansäuer, Chem. Commun. 1997, 457; b) Synlett 1997, 363.
[4] a) W. A. Nugent, T. V. RajanBabu, J. Am. Chem. Soc. 1988, 110, 8561;
b) T. V. RajanBabu, W. A. Nugent, ibid. 1989, 111, 4525; c) T. V.
RajanBabu, W. A. Nugent, M. S. Beattie, ibid. 1990, 112, 6408; d) T. V.
RajanBabu, W. A. Nugent, ibid. 1994, 116, 986.
[a] See Experimental Section. [b] cis:trans ˆ 88:12. [c] cis:trans ˆ 85:15.
[d] Substrate was a mixture of cis und trans isomers. [e] Zn was used as
reductant. [f] cis:trans ˆ 14:86.
[5] A. S. Rao in Comprehensive Organic Synthesis, Vol. 7 (Eds.: B. M.
Trost, I. Fleming, G. Pattenden), Pergamon, Oxford, 1991, pp. 358±375.
as with zinc as reductant. This result supports the above-
mentioned hypothesis of catalyst deactivation.
Â
[6] J. Aube in Comprehensive Organic Synthesis, Vol. 1 (Eds.: B. M. Trost,
I. Fleming, G. Pattenden), Pergamon, Oxford, 1991, pp. 819 ± 834.
[7] J. O. Metzger, Methoden Org. Chem. (Houben-Weyl) 4th ed. 1952 ± ,
Vol. E19a/1, pp. 147 ± 168.
Since titanocene-mediated SET is the key step in the
epoxide opening, a number of functional groups sensitive to
electron transfer reagents such as samarium diiodide[14] were
subjected to the optimized conditions. Gratifyingly benzyl
ethers, chlorides, and even ketones (see Table 2, entry 10 for
the competition experiment) are perfectly stable under the
reaction conditions, and the corresponding products were
isolated in good yields. Chlorides are not tolerated in radical
chain reactions using tributylstannane (dehalogenation), and
therefore the reaction described here allows a mild and highly
chemoselective entry to radical chemistry.[7] Protecting groups
that can migrate under basic conditions like pivalate (Piv),
tosylate (Ts), and tert-butyldiphenylsilyl ether (OTBDPS) do
not migrate even when it would be kinetically favorable, as in
the case of the TBDPS ethers.[15] Also elimination of tosylate
was not observed. Although these alcohols can also be
[8] R. L. Halterman, Chem. Rev. 1992, 92, 965.
[9] M. L. H. Green, C. R. Lucas, J. Chem. Soc. Dalton Trans. 1972, 1000.
[10] A. Fürstner, A. Hupperts, J. Am. Chem. Soc. 1995, 117, 4468.
[11] A. Fürstner, N. Shi, J. Am. Chem. Soc. 1996, 118, 2533, 12349.
[12] Handbook of Chemistry and Physics, 66th ed. (Ed.: R. C. Weast),
CRC, Boca Raton, FL, 1985, D159 ± D161.
[13] R. S. P. Coutts, P. C. Wailes, R. L. Martin, J. Organomet. Chem. 1973,
47, 375.
[14] G. A. Molander, Chem. Rev. 1992, 92, 29.
[15] T. W. Greene, P. G. M. Wuts, Protective Groups in Organic Synthesis,
2nd ed., Wiley, New York, 1991, p. 83, p. 98.
[16] a) H. C. Brown, P. Heim, J. Am. Chem. Soc. 1970, 92, 1637; b) E.-I.
Negishi in Comprehensive Organometallic Chemistry, Vol. 7 (Eds.: G.
Wilkinson, F. G. A. Stone, E. W. Abel), Pergamon, Oxford, 1982,
p. 255; c) H. Molin, B. G. Pring, Tetrahedron Lett. 1985, 26, 677.
[17] J. A. Soderquist in Encyclopedia of Reagents for Organic Synthesis,
Vol. 2 (Ed.: L. Paquette), Wiley, New York, 1995, pp. 622 ± 630.
Angew. Chem. Int. Ed. 1998, 37, No. 1/2
ꢀ WILEY-VCH Verlag GmbH, D-69451 Weinheim, 1998
1433-7851/98/3701-0103 $ 17.50+.50/0
103