Mechanism of titanocene-mediated epoxide opening through homolytic substitution

Article abstract of DOI:10.1021/ja067054e

The mechanism of titanocene-mediated epoxide opening was studied by a combination of voltammetric, kinetic, computational, and synthetic methods. With the aid of electrochemical investigations the nature of a number of Ti(III) complexes in solution was established. In particular, the distribution of monomeric and dimeric Ti(III) species was found to be strongly affected by the exact steric conditions. The overall rate constants of the reductive epoxide opening were determined for the first time. These data were employed as the basis for computational studies of the structure and energies of the epoxide-titanocene complexes, the transition states of epoxide opening, and the β-titanoxy radicals formed. The results obtained provide a structural basis for the understanding of the factors determining the regioselectivity of ring opening and match the experimentally determined values. By employing substituted titanocenes even more selective epoxide openings could be realized, Moreover, by properly adjusting the steric demands of the catalysts and the substrates the first examples of reversible epoxide openings were designed.

Full text of DOI:10.1021/ja067054e

Published on Web 01/17/2007
Mechanism of Titanocene-Mediated Epoxide Opening through
Homolytic Substitution
Andreas Gansa¨uer,*,† Andriy Barchuk,† Florian Keller,† Martin Schmitt,†
Stefan Grimme,*,‡ Mareike Gerenkamp,‡ Christian Mu¨ck-Lichtenfeld,‡
Kim Daasbjerg,*,§ and Heidi Svith§
Contribution from the Kekule´-Institut fu¨r Organische Chemie und Biochemie der UniVersita¨t
Bonn, Gerhard Domagk Strasse 1, 53121 Bonn, Germany, Organisch-Chemisches Institut der
Westfa¨lischen Wilhelms-UniVersita¨t, Correnstrasse 40, 49149 Mu¨nster, Germany, and
Department of Chemistry, UniVersity of Aarhus, Langelandsgade140, 8000 Aarhus C, Denmark
Received October 2, 2006; E-mail: andreas.gansaeuer@uni-bonn.de; grimmes@uni-muenster.de; kdaa@chem.au.dk
Abstract: The mechanism of titanocene-mediated epoxide opening was studied by a combination of
voltammetric, kinetic, computational, and synthetic methods. With the aid of electrochemical investigations
the nature of a number of Ti(III) complexes in solution was established. In particular, the distribution of
monomeric and dimeric Ti(III) species was found to be strongly affected by the exact steric conditions. The
overall rate constants of the reductive epoxide opening were determined for the first time. These data
were employed as the basis for computational studies of the structure and energies of the epoxide-
titanocene complexes, the transition states of epoxide opening, and the â-titanoxy radicals formed. The
results obtained provide a structural basis for the understanding of the factors determining the regioselectivity
of ring opening and match the experimentally determined values. By employing substituted titanocenes
even more selective epoxide openings could be realized. Moreover, by properly adjusting the steric demands
of the catalysts and the substrates the first examples of reversible epoxide openings were designed.
have recently attracted considerable interest.4 Important ex-
amples of rare reactions such as enantioselective radical
generation3d,i or 3-exo- and 4-exo-cyclizations3n,p controlled by
polarity matching of radical reduction have been developed. A
number of pertinent applications in the synthesis of complex
molecules have emerged, e.g., titanocene-catalyzed epoxypoly-
Introduction
Over the last decades radical chemistry has been increasingly
used in the synthesis of complex molecules. This is mainly due
to the mildness of radical generation, the excellent tolerance of
these reactive intermediates toward numerous functional groups,
and the high predictability of a number of key radical reactions.1
In the field of metal-induced radical chemistry titanocene-
mediated2 and titanocene-catalyzed3 epoxide opening reactions
(3) For pertinent catalytic reactions see: (a) Gansa¨uer, A.; Pierobon, M.; Bluhm,
H. Angew. Chem., Int. Ed. 1998, 37, 101-103. (b) Gansa¨uer, A.; Bluhm,
H.; Pierobon, M. J. Am. Chem. Soc. 1998, 120, 12849-12859. (c) Gansa¨uer,
A.; Bluhm, H. Chem. Commun. 1998, 2143-2144. (d) Gansa¨uer, A.;
Lauterbach, T.; Bluhm, H.; Noltemeyer, M. Angew. Chem., Int. Ed. 1999,
38, 2909-2910. (e) Gansa¨uer, A.; Pierobon, M. Synlett 2000, 1357-1359.
(f) Gansa¨uer, A.; Pierobon, M.; Bluhm, H. Synthesis 2001, 2500-2520.
(g) Gansa¨uer, A.; Pierobon, M.; Bluhm, H. Angew. Chem., Int. Ed. 2002,
41, 3206-3208. (h) Gansa¨uer, A.; Rinker, B. Tetrahedron 2002, 58, 7017-
7026. (i) Gansa¨uer, A.; Bluhm, H.; Rinker, B.; Narayan, S.; Schick, M.;
Lauterbach, T.; Pierobon, M. Chem. Eur. J. 2003, 9, 531-542. (j) Barrero,
A. F.; Rosales, A.; Cuerva, J. M.; Oltra, J. E. Org. Lett. 2003, 5, 1935-
1938. (k) Gansa¨uer, A.; Rinker, B.; Pierobon, M.; Grimme, S.; Gerenkamp,
M.; Mu¨ck-Lichtenfeld, C. Angew. Chem., Int. Ed. 2003, 42, 3687-3690.
(l) Justicia, J.; Rosales, A.; Bun˜uel, E.; Oller-Lo´pez, J. L.; Valdivia, M.;
Ha¨ıdour, A.; Oltra, J. E.; Barrero, A. F.; Ca´rdenas, D. J.; Cuerva, J. M.
Chem. Eur. J. 2004, 10, 1778-1788. (m) Gansa¨uer, A.; Rinker, B.; Ndene-
Schiffer, N.; Pierobon, M.; Grimme, S.; Gerenkamp, M.; Mu¨ck-Lichtenfeld,
C. Eur. J. Org. Chem. 2004, 2337-2351. (n) Gansa¨uer, A.; Lauterbach,
T.; Geich-Gimbel, D. Chem. Eur. J. 2004, 10, 4983-4990. (o) Justicia, J.;
Oltra, J. E.; Cuerva, J. M. J. Org. Chem. 2004, 69, 5803-5806. (p)
Friedrich, J.; Dolg, M.; Gansa¨uer, A.; Geich-Gimbel, D.; Lauterbach, T.
J. Am. Chem. Soc. 2005, 127, 7071-7077. (q) Justicia, J.; Oller-Lopez,
J. L.; Campan˜a, A. G.; Oltra, J. E.; Cuerva, J. M.; Bun˜uel, E.; Ca´rdenas,
D. J. J. Am. Chem. Soc. 2005, 127, 14911-14921. (r) Barrero, A. F.; Qu´ılez,
del Moral, J. F.; Sa´nchez, E. M.; Arteaga, J. F. Org. Lett. 2006, 8, 669-
672.
† Universita¨t Bonn.
‡ Westfa¨lische Wilhelms-Universita¨t.
§ University of Aarhus.
(1) For leading references see: (a) Curran, D. P.; Porter, N. A.; Giese, B.
Stereochemistry of Radical Reactions; VCH: Weinheim, Germany, 1996.
(b) Radicals in Organic Synthesis; Renaud, P., Sibi, M. P., Eds.; Wiley-
VCH: Weinheim, Germany, 2001. (c) McCarroll, A. J.; Walton, J. C.
Angew. Chem. 2001, 113, 2282-2307; Angew. Chem., Int. Ed. 2001, 40,
2225-2250. (d) Zard, S. Z. Radical Reactions in Organic Synthesis; Oxford
University: Oxford, 2003.
(2) For pertinent stoichiometric reactions see: (a) Nugent, W. A.; RajanBabu,
T. V. J. Am. Chem. Soc. 1988, 110, 8561-8562. (b) RajanBabu, T. V.;
Nugent, W. A. J. Am. Chem. Soc. 1989, 111, 4525-4527. (c) RajanBabu,
T. V.; Nugent, W. A.; Beattie, M. S. J. Am. Chem. Soc. 1990, 112, 6408-
6409. (d) RajanBabu, T. V.; Nugent, W. A. J. Am. Chem. Soc. 1994, 116,
986-997. (e) Ferna´ndez-Mateos, A.; de la Nava, E. M.; Coca, G. P.; Silva,
A. R.; Gonza´lez, R. R. Org. Lett. 1999, 1, 607-609. (f) Hardouin, C.;
Doris, E.; Rousseau, B.; Mioskowski, C. Org. Lett. 2002, 4, 1151-1153.
(g) Barrero, A. F.; Oltra, J. E.; Cuerva, J. M.; Rosales, A. J. Org. Chem.
2002, 67, 2566-2571. (h) Anaya, J.; Ferna´ndez-Mateos, A.; Grande, M.;
Martia´n˜ez, J.; Ruano, G.; Rubio Gonza´lez, R. Tetrahedron 2003, 59, 241-
248. (i) Ferna´ndez-Mateos, A.; Mateos Buro´n, L.; Rabanedo Clemente,
R.; Ramos Silva, A. I.; Rubio Gonza´lez, R. Synlett 2004, 1011-1014. (j)
Ferna´ndez-Mateos, A.; Mateos Buro´n, L.; Mart´ın de la Nava, E. M.;
Rabanedo Clemente, R.; Rubio Gonza´lez, R.; Sanz Gonza´lez, F. Synlett
2004, 2553-2557. (k) Leca, D.; Fensterbank, L.; Lacoˆte, E.; Malacria, M.
Angew. Chem., Int. Ed. 2004, 43, 4220-4222. (l) Banerjee, B.; Roy, S. C.
Eur. J. Org. Chem. 2006, 489-497.
(4) (a) Gansa¨uer, A.; Bluhm, H. Chem. ReV. 2000, 100, 2771-2788. (b)
Gansa¨uer, A.; Narayan, S. AdV. Synth. Catal. 2002, 344, 465-475. (c)
Gansa¨uer, A.; Lauterbach, T.; Narayan, S. Angew. Chem. 2003, 42, 5556-
5573. (d) Cuerva, J. M.; Justicia, J.; Oller-Lo´pez, J. L.; Oltra, J. E. Top.
Curr. Chem. 2006, 264, 63-92.
9
10.1021/ja067054e CCC: $37.00 © 2007 American Chemical Society
J. AM. CHEM. SOC. 2007, 129, 1359-1371
1359

A R T I C L E S
Gansa¨uer et al.
ene cyclizations.3l,o,q Moreover, the titanocene-mediated epoxide
opening has been used with excellent success for the initiation
of the polymerization of styrene.5 Titanocene-mediated or
titanocene-catalyzed pinacol couplings have attracted less at-
tention due to their restricted substrate scope, even though the
first example of an enantioselective coupling has been reported.6
Scheme 1. Reactions Featuring Homolytic Substitution Reactions
with Metal Oxygen Bonds (porph ) Porphyrin)
An attractive feature of titanocene catalysis is that the
regioselectivity of ring cleavage is opposite to SN2 reactions,7
such as Jacobsen’s outstanding catalytic epoxide openings.7d-g
Therefore, the preparative usefulness of epoxides,8 in organic
synthesis is increased even further. Despite this conceptual and
preparative usefulness of the titanocene-mediated or titanocene-
catalyzed transformations, to the best of our knowledge nothing
is known about the exact mechanism of the titanocene-mediated
epoxide opening. Here, we present a full account of the first
study elucidating this mechanism by a combination of electro-
chemical, kinetic, computational, and synthetic methods.9 The
results are of fundamental interest due to the insights in the
molecular details of the reaction and lead the way for the
development of even more selective catalysts or processes.
Results and Discussion
We selected four alkyl-substituted titanocene dichlorides,
(RCp)2TiCl2, with the aim of identifying the principal species
formed in Zn-reduced solutions and their reactivity toward
epoxides. The substituent at the cyclopentadienyl ring ranges
from the relatively small Me over tert-butyl to the bulky
menthyl13 group along with an ethylene-bridged tetrahydroin-
denyl moiety (Chart 1).14 These substituents are interesting not
only with respect to their steric and electronic effects, in general,
but also because of their potential synthetic applications in
stereoselective epoxide openings.
Moreover, our investigation provides a unique link between
the mechanism of reductive opening of epoxides, where a C-O
bond is cleaved through a homolytic substitution reaction, and
reactions where C-O bonds are formed through homolytic
substitutions featuring metal oxygen bonds as radical traps.
Examples for the latter transformations are alkoxylation reac-
tions such as epoxidations via â-metaloxy radicals,10 hydroxy-
lations of hydrocarbons via C-H bond activation by organo-
metallic complexes,11 and our titanocene-catalyzed THF
synthesis.3k,m Even the oxygen rebound step, that is intensely
debated for the P450 enzymes, may be regarded as another case
of this type of reaction.12 These reactions are highlighted in
Scheme 1.
Chart 1. Structures of the Substituted Titanocenes Used in This
Study
(5) (a) Asandei, A. D.; Moran, I. W. J. Am. Chem. Soc. 2004, 126, 15932-
15933. (b) Asandei, A. D.; Moran, I. W. J. Polym. Sci., Part A: Polym.
Chem. 2006, 44, 6028-6038. (c) Asandei, A. D.; Moran, I. W. J. Polym.
Sci., Part A: Polym. Chem. 2005, 43, 6039-6047. (d) Asandei, A. D.;
Moran, I. W. J. Polym. Sci., Part A: Polym. Chem. 2006, 44, 1060-1070.
(6) (a) Handa, Y.; Inanaga, J. Tetrahedron Lett. 1987, 28, 5717-5718. (b)
Gansa¨uer, A. Chem. Commun. 1997, 457-458. (c) Gansa¨uer, A.; Bauer,
D. J. Org. Chem. 1998, 63, 2070-2071. (d) Gansa¨uer, A.; Bauer, D. Eur.
J. Org. Chem. 1998, 2673-2676. (e) Dunlap, M. S.; Nicholas, K. M. Synth.
Commun. 1999, 27, 1097-1106. (f) Halterman, R. L.; Zhu, C.; Chen, Z.;
Dunlap, M. S.; Khan, M. A.; Nicholas, K. M. Organometallics 2000, 19,
3824-3829.
(7) (a) Nugent, W. A. J. Am. Chem. Soc. 1992, 114, 2768-2769. (b) Martinez,
L. E.; Leighton, J. L.; Carsten, D. H.; Jacobsen, E. N. J. Am. Chem. Soc.
1995, 117, 5897-5898. (c) Jacobsen, E. N.; Kakiuchi, F.; Konsler, R. G.;
Larrow, J. F.; Tokugana, M. Tetrahedron Lett. 1997, 38, 773-776. (d)
Larrow, J. F.; Schaus, S. E.; Jacobsen, E. N. J. Am. Chem. Soc. 1996, 118,
7420-7421. (e) Tokugana, M.; Larrow, J. F.; Kakiuchi, F.; Jacobsen,
E. N. Science 1997, 277, 936-938. (f) Jacobsen, E. N. Acc. Chem. Res.
2000, 33, 421-431. (g) Nielsen, L. P. C.; Jacobsen, E. N. In Epoxides and
Aziridines in Organic Synthesis; Yudin, A. K., Ed.; Wiley-VCH: Weinheim,
Germany, 2006; pp 229-269.
(8) (a) Johnson, R. A.; Sharpless, K. B. In Catalytic Asymmetric Synthesis;
Ojima, I., Ed.; Wiley-VCH: New York, 2000; pp 231-280. (b) Lane,
B. S.; Burgess, K. Chem. ReV. 2003, 103, 2457-2474. (c) Shi, Y. In Modern
Oxidation Methods; Ba¨ckvall, J.-E., Ed.; Wiley-VCH: Weinheim, Germany,
2004; pp 51-78. (d) Yang, D. Acc. Chem. Res. 2004, 37, 497-505. (e)
Shi, Y. Acc. Chem. Res. 2004, 37, 488-496. (f) McGarrigle, E. M.;
Gilheany, D. G. Chem. ReV. 2005, 105, 1564-1602.
(9) Daasbjerg, K.; Svith, H.; Grimme, S.; Gerenkamp, M.; Mu¨ck-Lichtenfeld,
C.; Gansa¨uer, A.; Barchuk, A.; Keller, F. Angew. Chem., Int. Ed. 2006,
45, 2041-2044.
For comparative purposes unsubstituted titanocene dichloride,
Cp2TiCl2, was included in the study. In previous work we have
succeeded in elucidating the composition of the electrochemi-
cally and metal-induced reductions of Cp2TiCl2 in THF by
means of cyclic voltammetric and kinetic measurements.15 We
were able to demonstrate that the principal constituents of such
solutions are monomeric and dimeric Ti(III) species. Trinuclear
(11) (a) Groves, J. T.; Viski, T. J. Org. Chem. 1990, 55, 3628-3634. (b) Larrow,
J. F.; Jacobsen, E. N. J. Am. Chem. Soc. 1994, 116, 12129-12130. (c)
Hamachi, K.; Irie, R.; Katsuki, T. Tetrahedron Lett. 1996, 37, 4979-4982.
(d) Hamada, T.; Irie, R.; Mihara, J.; Hamachi, K.; Katsuki, T. Tetrahedron
1998, 54, 10017-10028. (e) Katsuki, T. In ComprehensiVe Asymmetric
Catalysis II; Jacobsen, E. N., Pfaltz, A., Yamamoto, H., Eds.; Springer:
Berlin, 1999; pp 791-799. (f) Katsuki, T. Synlett 2003, 281-297.
(12) For some recent reviews see: (a) Newcomb, M.; Toy, P. H. Acc. Chem.
Res. 2000, 33, 449-455. (b) Shaik, S.; Cohen, S.; de Visser, S. P.; Sharma,
P. K.; Kumar, D.; Kozuch, S.; Ogliaro, F.; Danovich, D. Eur. J. Inorg.
Chem. 2004, 207-226. (c) Meunier, B.; de Visser, S. P.; Shaik, S. Chem.
ReV. 2004, 104, 3947-3980. (d) Shaik, S.; Kumar, D.; de Visser, S. P.;
Thiel, W. Chem. ReV. 2005, 105, 2279-2328. (e) Denisov, I. G.; Makris,
T. M.; Sligar, S. G.; Schlichting, I. Chem. ReV. 2005, 105, 2253-2277.
(13) (a) Cesarotti, E.; Kagan, H. B.; Goddard, R.; Kru¨ger, C. J. Organomet.
Chem. 1978, 162, 297-309. Practical synthesis: (b) Gansa¨uer, A.; Narayan,
S.; Schiffer-Ndene, N.; Bluhm, H.; Oltra, J. E.; Cuerva, J. M.; Rosales, A;
Nieger, M. J. Organomet. Chem. 2006, 691, 2327-2331.
(10) (a) Samsel, E. G.; Srinivasan, K.; Kochi, J. K. J. Am. Chem. Soc. 1985,
107, 7606-7617. (b) Srinivasan, K.; Michaud, P.; Kochi, J. K. J. Am. Chem.
Soc. 1986, 108, 2309-2320. (c) Zhang, W.; Loebach, J. L.; Wilson, S. R.;
Jacobsen, E. N. J. Am. Chem. Soc. 1990, 112, 2801-2803. (d) Irie, R.;
Noda, K.; Ito, Y.; Katsuki, T. Tetrahedron Lett. 1990, 31, 7345-7348. (e)
Daly, A. M.; Renehan, M. F.; Gilheany, D. G. Org. Lett. 2001, 3, 663-
666.
(14) (a) Wild, F. R. W. P.; Zsolnai, L.; Huttner, G.; Brintzinger, H. H.
J. Organomet. Chem. 1982, 232, 233-247. (b) Collins, S.; Kuntz, B. A.;
Taylor, N. J.; Ward, D. G. J. Organomet. Chem. 1988, 342, 21-29.
9
1360 J. AM. CHEM. SOC. VOL. 129, NO. 5, 2007

Titanocene-Mediated Epoxide Opening
A R T I C L E S
Chart 2. Proposed Structures of Solvated Cp2TiCl (5a) and
(Cp2TiCl)2 (5b)
this reaction is of much less importance for substituted ti-
tanocene dichlorides as the analysis of the chemically reduced
solutions will reveal.
Cyclic voltammograms recorded of the various solutions of
(RCp)2TiCl2 are collected in Figure 1. In consistency with the
detailed analysis provided elsewhere15b for Cp2TiCl2 the most
profound feature is the appearance of the redox wave around
-1.4 V vs Fc+/Fc (abbreviation for ferrocenium/ferrocene)
complexes or ionic clusters were not present, despite their
isolation and characterization in solid state.16
-
pertaining to the (RCp)2TiCl2/(RCp)2TiCl2 couple. The peak
In Chart 2 the proposed structures of the two principal Ti-
(III) species Cp2TiCl (5a) and (Cp2TiCl)2 (5b) are depicted
assuming that any free coordination site would be occupied by
solvent molecules, i.e., THF. An analysis of the solid-state
structure of the dimeric species pointed to a symmetric arrange-
ment.17 However, since kinetic measurements involving electron
acceptors such as benzyl chloride, benzaldehyde, and epoxides
revealed that the dimer had a surprisingly high reactivity in
solution, we suggested that the half-open arrangement shown
in Chart 2 with an accessible coordination site actually might
be the active reagent in solution.15c,d This interpretation was
recently supported by computational studies.9 A particular
interesting aspect to consider herein is therefore the extent by
which the presence of various substituents at the cyclopenta-
dienyl ring may influence the monomer/dimer distribution. To
obtain a picture as comprehensive as possible the present study
was carried out employing a combination of cyclic voltammetric,
kinetic, computational, and synthetic methods.
separation increases nonlinearly as a function of sweep rate as
expected for a quasi-reversible charge-transfer process. At the
same time a small wave positioned at less negative potentials
originating from the oxidation process of (RCp)2TiCl tends to
appear on the oxidative sweep, i.e., at high sweep rates the lower
part of the square scheme comes into play. Upon addition of
tetrabutylammonium chloride as a chloride source to the
solution, enhancement of the rate of the reaction between
(RCp)2TiCl and Cl- induces a further shift of the equilibrium
-
reaction toward (RCp)2TiCl2 to favor once again the upper
part of the square scheme.
In our previous work the dissociation equilibrium constant
-
for Cp2TiCl2 was estimated on the basis of such experiments
and extensive simulations to a value of 10-4 M.15b Because
essentially the same or a slightly larger amount of tetrabutyl-
ammonium chloride is required (>400 mol %) to deplete the
small (RCp)2TiCl wave in the case of 1-4, this study reveals
that the equilibrium constant is essentially of the same magnitude
for the substituted titanocene dichlorides. Thus, the overall effect
of the bulkiness of the cyclopentadienyl ligands on this particular
reaction is relatively small. In contrast, the substituent effect
on the redox properties of the (RCp)2TiCl2/(RCp)2TiCl2- system
is clearly detectable. While the standard potential was deter-
mined to be -1.27 V vs Fc+/Fc for Cp2TiCl2,15b we find values
of -1.40, -1.34, -1.37, and -1.48 V vs Fc+/Fc for 1, 2, 3,
and 4, respectively. A discussion of these data along with others
obtained will be given below.
Electrochemical Reduction of (RCp)2TiCl2. First our focus
will be on the electrochemical reduction of (RCp)2TiCl2 with
the expectation that the mechanistic scheme can be described
in terms of the square scheme already elucidated in the case of
Cp2TiCl2 (Scheme 2).15b
Scheme 2. Square Scheme Describing the Electrochemical
Reduction of (RCp)2TiCl2
Chemical Reduction of (RCp)2TiCl2. The second part of
the voltammetric analysis is concerned with the study of Zn-
reduced solutions of (RCp)2TiCl2, denoted Zn-(RCp)2TiCl2.
Such an investigation of Ti(III)-containing solutions is not only
important because it complements the above study focusing on
Ti(IV) as starting material but also because it may provide
additional information pertinent to the many important synthetic
applications using Zn-(RCp)2TiCl2. The present analysis is
based on the reasonable assumption that the mechanistic scheme
already resolved for Zn-Cp2TiCl2 can be applied for the four
substituted cases 1-4 as well.15c
The essential feature of this scheme is that the electrochemical
reduction of (RCp)2TiCl2 to (RCp)2TiCl2 on the reductive
-
sweep in cyclic voltammetry is followed by a reversible
fragmentation reaction, in which the electroactive (RCp)2TiCl
and Cl- are generated. On the oxidative sweep, oxidation of
(RCp)2TiCl2- and/or (RCp)2TiCl results in the direct or indirect
regeneration of the parent (RCp)2TiCl2 compound. In the case
of (RCp)2TiCl the conversion goes through the (RCp)2TiCl+
intermediate that is involved in a fast equilibrium reaction with
Cl-. Eventually, (RCp)2TiCl may dimerize to afford the
electroactive dimer, [(RCp)2TiCl]2, although the inclusion of
The general conclusion from our previous work15c-e is that
the principal species formed in THF solutions of Zn-Cp2TiX2
(X ) Cl, Br, and I) are a mixture of Cp2TiX, (Cp2TiX)2, and
Cp2Ti+, the distribution of which is dependent on the halogen
considered.15d For example, Cp2Ti+ is present in quite substantial
amounts only for X ) Br and I, whereas it is completely absent
for X ) Cl. In all three solutions the amounts of Cp2TiX and
(Cp2TiX)2 are appreciable being formed in a mixture having a
dimerization equilibrium constant of (1-3) × 103 M-1. The
(15) (a) Enemærke, R. J.; Hjøllund, G. H.; Daasbjerg, K.; Skrydstrup, T. C. R.
Acad. Sci. 2001, 4, 435-438. (b) Enemærke, R. J.; Larsen, J.; Skrydstrup,
T.; Daasbjerg, K. Organometallics 2004, 23, 1866-1874. (c) Enemærke,
R. J.; Larsen, J.; Skrydstrup, T.; Daasbjerg, K. J. Am. Chem. Soc. 2004,
126, 7853-7864. (d) Enemærke, R. J.; Larsen, J.; Hjøllund, G. H.;
Skrydstrup, T.; Daasbjerg, K. Organometallics 2005, 24, 1252-1262. (e)
Larsen, J.; Enemærke, R. J.; Skrydstrup, T.; Daasbjerg, K. Organometallics
2006, 25, 2031-2036.
(16) (a) Sekutowski, D. G.; Stucky, G. D. Inorg. Chem. 1975, 14, 2192-2199.
(b) Sekutowski, D.; Jungst, R.; Stucky, G. D. Inorg. Chem. 1978, 17, 1848-
1855.
(17) Coutts, R. S. P.; Wailes, P. C.; Martin, R. L. J. Organomet. Chem. 1973,
47, 375-382.
-
anion Cp2TiCl2 is of no importance, unless the reduction of
Cp2TiCl2 is carried out electrochemically. This is due to the
fact that the zinc chlorides present in the Zn-reduced solutions
will be able to take up additional chlorides.
9
J. AM. CHEM. SOC. VOL. 129, NO. 5, 2007 1361

A R T I C L E S
Gansa¨uer et al.
Figure 1. Cyclic voltammograms of 2 mM solutions of 1 (A), 2 (B), 3 (C), and 4 (D) recorded at sweep rates of 0.1 (black), 0.5 (red), 1 (green), 5 (blue),
and 10 V s-1 (magenta) in 0.2 M Bu4NPF6/THF. Currents are normalized with respect to sweep rate.
Figure 3. Cyclic voltammograms of solutions containing Zn-4 in
concentrations of 0.7 (black) and 1.5 mM (red) recorded at a sweep rate of
1 V s-1 in 0.2 M Bu4NPF6/THF. Currents are normalized with respect to
concentration.
Figure 2. Cyclic voltammograms of solutions containing Zn-1 in
concentrations of 0.7 (black), 1.1 (red), and 1.8 mM (green) recorded at a
sweep rate of 0.1 V s-1 in 0.2 M Bu4NPF6/THF. Currents are normalized
with respect to concentration.
The earlier cyclic voltammetric studies also gave immediate
access to a characterization of the oxidation processes pertaining
to the electroactive species Cp2TiCl2-, (Cp2TiCl)2, Cp2TiCl, and
Cp2Ti+. Although all of these four species are not present
initially in the solution, they may be generated in followup
reactions taking place at the electrode surface during the sweep
in cyclic voltammetry. Such followup reactions include a
father-son reaction between Cp2TiCl and Cp2TiCl+, a direct
fragmentation of the dimer cation (Cp2TiCl)2+, and chloride
transfer reactions between various titanium-based species and
the zinc chlorides.
One of the essential questions raised by the present study is
how the substituent on the cyclopentadienyl ring will affect the
parameters associated with the reaction scheme. While the first
9
1362 J. AM. CHEM. SOC. VOL. 129, NO. 5, 2007

Titanocene-Mediated Epoxide Opening
A R T I C L E S
Scheme 3. Proposed Mechanistic Scheme for the Electrochemical Oxidation of Zn-(RCp)2TiCl2 (R ) H, Me)
part of our study has revealed already a substituent effect on
the redox properties of the (RCp)2TiCl2/(RCp)2TiCl2 couple,
At first sight the overall appearance with the presence of two
characteristic oxidation waves is the same as for Zn-Cp2TiCl2
and Zn-1, but there is a distinct difference in the sense that
there is no splitting of the first wave. The same lack of a splitting
is also seen for the Zn-reduced solutions of 2 and 3 (Supporting
Information). This lends support to the hypothesis that the
dimerization reaction for the more highly substituted titanocene
chlorides is less favorable. In these cases the mechanism thus
becomes somewhat simplified since the blue part involving the
dimer in Scheme 3 can be omitted.
Discussion of Cyclic Voltammetric Data. In the further
analysis of the cyclic voltammograms the goal was to obtain
quantitative information about the composition of the solutions.
In general, all relevant potentials, heterogeneous and homoge-
neous rate constants as well as equilibrium constants were
extracted by digital simulations (using the program DigiSim
3.03)19 of the experimentally obtained voltammograms recorded
at different sweep rates and concentrations. A detailed descrip-
tion of the exact model parameters used as well as relevant fits
can be found in the Supporting Information. Below the focus
will be on the most important parameters extracted, i.e., the
standard potentials and the dimerization equilibrium constant.
The standard potentials are summarized in Table 1.
-
it is yet to be elucidated if this will apply also to the
[(RCp)2TiCl]2+/[(RCp)2TiCl]2, Cp2TiCl+/Cp2TiCl, and Cp2Ti2+
/
Cp2Ti+ redox couples, not to mention the possible effect the
substituent may exert on the chemical reactions.
In Figure 2 we have collected cyclic voltammograms recorded
at three different concentrations of Zn-1. The oxidation wave
appearing at -0.8 V vs Fc+/Fc actually consists of two
processes, with the first one becoming more dominant as the
concentration increases. The two-component nature of the wave
is particularly pronounced for the intermediate concentration.
Thus, the first process is attributed to the oxidation of the
[(MeCp)2TiCl]2 dimer and the second one to the oxidation of
the (MeCp)2TiCl monomer. The oxidation peak at -0.2 vs Fc+/
Fc originates from the oxidation of (MeCp)2Ti+ generated in
the chemical reactions occurring during the cyclic voltammetric
sweep. Surprisingly, the peak potential of the (MeCp)2Ti+ wave
varies appreciably as a function of concentration, but this may
be related to the strong adsorption tendency of this species as
revealed by the rather broad and almost bell-shaped feature of
the wave. In any case this will introduce some additional
uncertainty in the determination of the standard potential of the
(MeCp)2Ti2+/(MeCp)2Ti+ redox couple. As seen in Scheme 3
the generation of (MeCp)2Ti+ is accompanied by the formation
of (MeCp)2TiCl2, the presence of which is substantiated by the
reduction wave that is about to appear at ca. -1.5 V vs Fc+/Fc
on the reverse sweep (compare with Figure 1).18 At this point
we may therefore state that introducing a methyl group at the
cyclopentadienyl ring does not change the overall mechanistic
scheme, although the exact values of the associated parameters
of course may differ (see below).
In general, for a given substituent the standard potentials
-
determined for the redox couples (RCp)2TiCl2/(RCp)2TiCl2
,
[(RCp)2TiCl]2+/[(RCp)2TiCl]2, (RCp)2TiCl+/(RCp)2TiCl, and
(RCp)2Ti2+/(RCp)2Ti+ increase (i.e., becomes less negative) in
the order listed. This is in line with the results already obtained
and discussed in the case of Cp2TiCl2.15b-d From a thermody-
namic point of view the electron-donating ability of the Ti(III)
-
species is thus RCp2TiCl2 . (RCp2TiCl)2 > RCp2TiCl .
RCp2Ti+.However,asalreadyshownbyourearlierinvestigations15b-d
this order does not hold from a kinetic point of view (see below).
A particularly interesting issue to discuss in this context is
the influence of the substituent on the potentials. Considering
only the inductive effect of electron-donating alkyl groups one
would predict that the potentials should become more negative
going from Cp2TiCl2 to 1, 2, or 3, and further to 4. Indeed this
With respect to the solution of Zn-4, voltammograms
recorded at two different concentrations are shown in Figure 3.
(18) In general, application of potentials below -1.4 V vs Fc+/Fc had to be
avoided because reduction of the zinc ions present could cause deleterious
adsorption of metal on the electrode surface.
(19) Rudolph, M.; Feldberg, S. W. DigiSim Version 3.03; Bioanalytical Systems,
Inc.: West Lafayette, IN.
9
J. AM. CHEM. SOC. VOL. 129, NO. 5, 2007 1363

A R T I C L E S
Gansa¨uer et al.
Table 1. Relevant Data Extracted from Cyclic Voltammograms Recorded on Zn-Reduced Solutions of 1-4
Zn−Cp2TiCl2
Zn−1
Zn
−2
Zn
−3
Zn−4
E1°/V vs Fc+/Fca
E2°/V vs Fc+/Fca
E3°/V vs Fc+/Fca
E4°/V vs Fc+/Fca
-1.27 ( 0.04b
-0.81 ( 0.03d
-0.75 ( 0.03d
-0.43 ( 0.03d
-1.40 ( 0.03c
-0.91 ( 0.03
-0.81 ( 0.03
-0.35 ( 0.08
-1.34 ( 0.03c
-1.37 ( 0.03c
-1.48 ( 0.03c
-0.81 ( 0.03
-0.55 ( 0.03
-0.84 ( 0.03
-0.54 ( 0.03
-0.94 ( 0.03
-0.62 ( 0.03
a Potentials can be converted to SCE by adding 0.52 V. b From ref 15b. c Obtained from cyclic voltammograms recorded on solutions of 1-4 (see above).
d From ref 15c.
Scheme 4. Mechanistic Scheme for the Reaction between Zn-Reduced Titanocene Dichlorides and 6
expectation is met for the three series involving [(RCp)2TiCl]2,
(RCp)2TiCl, and (RCp)2Ti+. The only outlier pertains to the
cation system for R ) Me, but presumably this should be
attributed to the adsorption problems encountered for this
particular system. In contrast, unexpected results are obtained
for the (RCp)2TiCl2 series, in that the tert-butyl- (2) and
menthyl-substituted titanocene dichlorides (3) have redox
potentials being intermediate to those of Cp2TiCl2 and (MeCp)2-
TiCl2. The bridged ansa-compound (4) with its fixed geometry
gives rise to the most negative value as also seen for the other
series.
The unusual order of potentials emphasizes the importance
of effects other than purely inductive. The influence of substit-
uents on the electrochemical reduction of titanocene dichlorides
has been addressed already in other studies.20 In general, a
correlation is expected between the reduction potentials and the
Hammett substituent coefficients, i.e., systems having electron-
donating substituents are more difficult to reduce.20b However,
in the series of methyl-substituted titanocene dichlorides excep-
tions pertaining to the sterically encumbered tetra- and pen-
tamethylcyclopentadienyl ligands exist. In the latter case a
positive shift in the reduction potential is even encountered.20c
It has been suggested that such a behavior can be brought about
by steric strain between the cyclopentadienyl ligands which will
not only induce additional hindrance to rotation but also lower
the dihedral angle between the cyclopentadienyl ring planes.20c
Our results corroborate this interpretation in the sense that the
potentials for the sterically encumbered 2 and 3 are less negative
than that of 1. The steric effects on the potentials are smaller
for the series involving [(RCp)2TiCl]2, (RCp)2TiCl, and
(RCp)2Ti+, in which one or both chloride ligands are absent.
The issue of solvent (i.e., THF) coordination and the possible
lack of such a coordination for the sterically most hindered
compounds seem to have only a minor effect on the standard
potentials.20c,21
Ti+ are present in any appreciable amount (although they may
be generated at the electrode surface during the cyclic volta-
mmetric sweep), independent of the substituent, the exact
distribution of the monomer (RCp)2TiCl and the dimer
[(RCp)2TiCl]2 is strongly affected. As previously determined
for the solution derived from Zn-Cp2TiCl2 the two species are
in rapid equilibrium with an equilibrium constant of 3 × 103
M-1 15b-d
For 1 the dimer still remains an important constituent
.
having an equilibrium constant of 103 M-1, but already to a
lesser extent.9 Furthermore, bulkier ligands as in 2-4 make the
monomer the only detectable species in cyclic voltammetry; a
dimerization equilibrium constant of 100 M-1 can be set as a
limiting value as deduced from digital simulations. For a 2 mM
Zn-reduced solution of either Cp2TiCl2, 1, or 2-4, this will
imply that the dimer/monomer ratio will be 1.5, 0.8, and <0.15,
respectively.
These results are reasonably in line with EPR studies on the
series of (C5H5-nMen)2TiX (n ) 0-5), in which it was found
that compounds with n ) 3-5 were monomeric, whereas those
with n ) 0-2 were dimers, at least in toluene. In 2-methyltet-
rahydrofuran, on the other hand, the latter compounds dissoci-
ated due to a solvent molecule coordination.21 At this point we
do not know to which extent the less favorable dissociation
process we seem to observe for the Zn-reduced solutions of
Cp2TiCl2 and 1 in THF should be attributed to the concentration
ranges considered, the presence of supporting electrolyte in
cyclic voltammetry, or other factors.
Kinetic Studies. Kinetic analysis of the epoxide opening with
Zn-reduced solutions of 2-4 was carried out with the aim of
characterizing in further details the reactive Ti(III) species. As
model compound for the epoxide we selected 2-methyl-2-
phenethyl-oxirane (6), because under the reaction conditions
chosen, 1,1-disubstituted epoxides are converted into titanocene
alkoxides, such as 6a, in high yield. The kinetic analysis is
therefore not complicated by interfering pathways. Epoxide 6
was also used in our previous kinetic analysis of Zn-reduced
solutions of Cp2TiCl2 and 1.9 The relevant rate constant, k, was
extracted by monitoring the disappearance of the Ti(III) species
using absorption spectroscopy (Scheme 4).
The substituents on the cyclopentadienyl ring also exert a
strong effect on the equilibrium distribution of the Ti(III) species
in the Zn-reduced solutions. While the simulations of the cyclic
-
voltammograms reveal that neither (RCp)2TiCl2 nor (RCp)2-
While the kinetic analysis of the solutions involving Zn-
Cp2TiCl2 and Zn-1 was complicated by the fact that an
equilibrium mixture of both the reactive monomer and dimer
is present,9,15c the situation in the case of Zn-(2-4) is much
simpler because of the absence of the dimer. Employment of 6
in an excess will confine the kinetics to simple pseudo-first-
(20) (a) Mugnier, Y.; Fakhr, A.; Fauconet, M.; Moise, C.; Laviron, E. Acta
Chem. Scand. 1983, B37, 423-427. (b) Johnston, R. F.; Borjas, R. E.;
Furilla, J. L. Electrochim. Acta 1995, 40, 473-477. (c) Langmaier, J.;
Samec, Z.; Varga, V.; Hora´cˇek, M.; Mach, K. J. Organomet. Chem. 1999,
579, 348-355.
(21) (a) Mach, K.; Raynor, J. B. J. Chem. Soc., Dalton Trans. 1992, 683-688.
(b) Mach, K.; Varga, V.; Schmid, G.; Hiller, J.; Thewalt, U. Collect. Czech.
Chem. Commun. 1996, 61, 1285-1294.
9
1364 J. AM. CHEM. SOC. VOL. 129, NO. 5, 2007

Titanocene-Mediated Epoxide Opening
A R T I C L E S
Table 2. Rate Data Obtained in the Epoxide Opening of 6 with Zn-Reduced Solutions of Cp2TiCl2 and 1-4
Zn
−Cp2TiCl2
Zn
−Cp2TiCl2
Zn
−1
Zn
−1
Zn
−2
Zn
−3
Zn−4
Ti(III) reagent
monomer
dimer
monomer
dimer
monomer
monomer
monomer
k/M-1 s-1
0.5a
1.4a
1.3a
3.9a
0.10
0.74
4.8
a From ref 9.
are interesting features. In the former case this may be attributed
to the presence of the sterically demanding tert-butyl groups
which apparentlysand even to a larger extent than the menthyl
groups in Zn-3sare able to prevent a close pairing up of the
Ti(III) reagent with the epoxide 6. The high reactivity of the
Zn-Cp2TiCl2 and Zn-1 dimers is surprising in the sense that
an analysis of the solid-state structure has pointed to a symmetric
structural arrangement with no vacant coordination site.17 We
therefore suggest that the actual solution structure is of the half-
open type 5b depicted in Chart 2.9,15c,d In fact, the availability
of a coordination site would be a prerequisite for having a fast
inner-sphere electron-transfer reaction. In this sense the dimer
can be considered as a highly Lewis acidic intermediate formed
from the monomer 5a (Chart 2) according to the principle of
activation of electrophiles through dimeric association.23
-
It is also noteworthy that previous investigations on Cp2TiCl2
Figure 4. Logarithm of the observed rate constant, k, for the Ti(III)-
mediated opening of 6 vs the standard potential E° for the relevant Ti(IV)/
Ti(III) system.
have suggested15b,c that its lack of reactivity as an electron donor
can be attributed to the fact that the coordination around the Ti
core is that of a distorted tetrahedron.24 Because a pentavalent
arrangement in the transition state would appear highly unlikely,
this would leave the outer-sphere electron-transfer reaction as
the only viable pathway for this species. In keeping with this
interpretation is the finding that the cation Cp2Ti+ with its easily
accessible coordination site is a rather potent electron donor in
spite of its relatively high oxidation potential (see Table 1).15c,d
order conditions. As in the synthetic work, 1,4-cyclohexadiene
(CHD) was added to chemically reduce the tertiary radical
intermediate formed upon the stereoselective electron-transfer-
mediated epoxide opening,22 thus precluding efficiently side
reactions involving the radical and the Ti(III) species. This is
in keeping with the 1:1 reaction stoichiometry observed for the
Ti(III) and epoxide reagents. Note that the alternative formation
of the primary radical intermediate is less favorable. The kinetics
followed a simple exponential decay that was unaffected by
the exact concentration of CHD employed. Thus, electron
transfer constitutes the rate-controlling step of the overall
reduction. This renders any mechanism with a quick and
reversible epoxide opening before radical trapping unlikely.
We also note that in the previous synthetic work22 the yield
of the product 2-methyl-4-phenylbutan-1-ol was determined to
be larger than or equal to 55% for all the systems studied herein,
the exception being Zn-4 with a disappointing 21% yield.
Those reactions were carried out under catalytic conditions using
collidine hydrochloride as proton source to liberate the Ti(IV)
species from the alkoxides formed. Thus, it seems that for Zn-4
another step in the catalytic cycle is too inefficient for a satis-
fying performance of the reaction. Most likely, this is the pro-
tonation with collidine hydrochloride for catalyst regeneration.
In Table 2 we have collected all relevant rate data obtained
in the present and previous work.9 In Figure 4 the data are
plotted against the relevant standard potentials. The general
expectation on thermodynamic grounds that the rate should
diminish as the driving force of the reaction (expressed as -FE°)
is decreased is by and large fulfilled. For instance, the fact that
Zn-1 constitutes a better electron-transfer agent than Zn-Cp2-
TiCl2 can be attributed to its higher reducing ability and that
the steric demands of the C5H4CH3 and C5H5 ligands are similar.
The low reactivity of the Zn-2 monomer along with the high
reactivity of the two dimeric systems of Zn-Cp2TiCl2 and Zn-1
Computational Studies. On the basis of the above experi-
mental results, we turned our attention toward insights into the
reaction mechanism on a molecular level. We studied the
complexation of a number of epoxides by Cp2TiCl (5a) and its
dimer (5b), the activation and reaction energies of ring opening,
and the structures of all pertinent intermediates, transition states,
and products by density functional theory (DFT) calculations
with the BP functional and a TZVP basis set.25 This pure density
functional is suited for the description of reactions of transition
metal compounds and has been successfully used in the literature
for chemically similar systems.26 One must be aware, however,
that the absolute barriers of radical reactions predicted by this
method are usually too low.27 Since we are mainly interested
in the comparison of reactivities of various epoxides, this
systematic error seems acceptable and does not affect the
conclusions of our work.
(23) Negishi, E.-i. Chem. Eur. J. 1999, 5, 411-420.
(24) Clearfield, A.; Warner, D. K.; Saldarriaga-Molina, C. H.; Ropal, R. Can.
J. Chem. 1975, 53, 1622-1629.
(25) Technical details of computation: All DFT calculations were performed
with the TURBOMOLE program package (see 25a). The geometries and
energies were obtained with the BP functional (see 25b,c) and a polarized
valence triple-ú basis set (TZVP) (see 25d) using unrestricted wavefunctions.
Spin densities for the TS were calculated using the Mulliken population
analysis scheme. (a) TURBOMOLE 5.6; Ahlrichs, R.; et al. Universita¨t
Karlsruhe, 2003. For the complete citation, see the Supporting Information.
(b) Becke, A. D. Phys. ReV. B 1988, 38, 3098-3100. (c) Perdew, J. P.
Phys. ReV. B 1986, 33, 8822-8824. (d) Scha¨fer, A.; Huber, C.; Ahlrichs,
R. J. Chem. Phys. 1994, 100, 5829-5835.
(26) (a) Abashkin, Y. G.; Collins, J. R.; Burt, S. K. Inorg. Chem. 2001, 40,
4040-4048. (b) Jacobsen, H.; Cavallo, L. Organometallics 2006, 25, 177-
183.
(27) For an example see: Hemelsoet, K.; Moran, D.; Van Speybroeck, V.;
Waroquier, M.; Radom, L. J. Phys. Chem. A 2006, 110, 8942-8951.
(22) Gansa¨uer, A.; Barchuk, A.; Fielenbach, D. Synthesis 2004, 15, 2567-2573.
9
J. AM. CHEM. SOC. VOL. 129, NO. 5, 2007 1365

A R T I C L E S
Gansa¨uer et al.
Table 3. Activation and Reaction Energies of the Epoxide
Binding of the Epoxides by Titanocene(III) Complexes.
Binding of the epoxides 7-11, that serve as simplified sur-
rogates for the preparatively employed substrates, to 5a was
studied first. These complexes are relatively easy to describe
computationally, they are involved in epoxide opening, and they
provide a valuable model for higher substituted titanocenes. The
binding enthalpies of 7-11 relative to those of the 5a*THF
complex are summarized in Scheme 5.
Complexes 5a*7-5a*11 in kcal mol-1
Scheme 5. Epoxides Investigated Computationally and Their
Complexation Energies by Cp2TiCl (5a) Relative to Those of THF
Interestingly, however, the Ti-O bond is slightly longer in 5a*8
than in 5a*10.
These differences in the geometry and energy of substrate
binding are general. As for 5a, binding of 10 by Zn-3 is
energetically disfavored compared to that of (R)-8 or (S)-8. At
the same time 10 is rotated by 90° compared to both enantiomers
of 8. These complexation studies provide important quantitative
and qualitative insights into the initiation of epoxide opening
mediated by Cp2TiCl or substituted titanocenes. The properties
of the other predominant species in solution (Cp2TiCl)2 are even
more relevant to most synthetic applications.
As expected, the calculations with the closed structure of (Cp2-
TiCl)2, that possesses no vacant coordination site, did not
provide a stable complex with 8. However, the already
mentioned half-open structure 5b was found to readily form a
complex with 8. In this complex, the second Cp2TiCl unit is
pointing away from the other titanocene and the chlorine atom
of the second Cp2TiCl unit has no contact with the first Ti center.
Thus, the complexes of 5a and 5b with 8 are structurally
surprisingly similar in the vicinity of 8. This is also the case
for the complex of 10 with 5b. The epoxide is rotated by 90°
exactly as in the complex with 5a.
Replacement of THF from 5a by 7-9 is thermodynamically
favorable. This is in line with the experimental observation that
mono and cis-1,2-disubstituted epoxides react swiftly with
titanocene complexes in THF. However, this situation changes
markedly for 10 and 11. In both cases THF replacement is
thermodynamically disfavored. These epoxides are nevertheless
rapidly opened by 5a, and hence both epoxide and THF binding
must be considered reversible and fast. The 1,1-disubstitution
pattern of 10 and 11 constitutes the reason for this difference
as can be inferred from the structures of their Cp2TiCl complexes
that are shown in Figure 5.
The hydrogen atom at the substituted carbon of the epoxide
in 8 cannot be replaced with a methyl group without causing
geometrical changes in the binding of the epoxide. A rotation
by 90° is necessary to enable the less favorable binding.
Discussion of the Energies of Activation and the Transition
Structures. We turned our attention to studying the key step
of the overall reaction, the epoxide opening through the
homolytic substitution reaction, next. For the same reasons as
above, the reactions of the Cp2TiCl complexes of 7-11 were
studied first. The activation energies of ring opening, ∆Eq, are
summarized in Table 3. They are in the range of 7.0-10.3 kcal
mol-1 and indicate that radical generation should be facile at
room temperature, in agreement with the experimental results.
Formation of the higher substituted radicals is kinetically
favored.
In the transition structures the spin density on the evolving
radical center is typically lower (approximately 0.3) than on
titanium (approximately 0.7). Thus, radical stability is most
likely not the only contribution to the control of regioselectivity
of epoxide opening. As the other important factor, we suggest
steric interactions between the epoxide and the cyclopentadienyl
ligands. This will be especially relevant when two radicals with
the same substitution pattern or radicals with relatively low
Figure 5. Structure (BP86/TZVP) of the 5a complexes of 8 (top) and 10
(bottom). Distances in angstrom.
9
1366 J. AM. CHEM. SOC. VOL. 129, NO. 5, 2007

Titanocene-Mediated Epoxide Opening
A R T I C L E S
Figure 6. Transition structures of the opening of 8 and 11.
Scheme 6. Transition Structures, Activation, and Reaction
Energies of the Opening of 8 by 5b
stability differences, such as primary and secondary radicals or
secondary and tertiary radicals, are formed during ring opening.
Inspection of the transition structures of the opening of 5a*8,
TS8a and TS8b, that are shown in Figure 6 as representative
examples, and of the opening of 5a*11, TS11a and TS11b
provides substantiation for this hypothesis. In TS8a relatively
close contacts between the cyclopentadienyl ligands and 8 were
observed. Both Hcis and Htrans are placed at a distance of 2.39
Å (and the CH3 group at more than 2.9 Å) from the cyclopen-
tadiene. In TS8b Hcis is placed at 2.40 Å from the ligand and
the CH3 group at only 2.30 Å. Thus, in addition to the stability
difference of the forming radicals, the larger CH3 group gets
into closer contact with the ligand during the less favored
opening to the primary radical.
The activation energies for the opening of 5b*8 were
calculated to be 6.3 kcal mol-1 for the formation of 12a and
7.5 kcal mol-1 for the formation of 12b. Both values are about
2 kcal mol-1 lower than for the corresponding complex 5a*8.
Even though the calculations do not include solvent effects, they
predict a substantially higher reactivity of the 5b-derived epoxide
complexes exactly as observed experimentally. This can be
explained by a larger stabilization of the strongly Lewis acidic
half-open dimer through the developing charge in the transition
structure. In TS12a and TS12b the Cl of the second Cp2TiCl
group is pointing away from the epoxide that is being opened
as shown in Figure 7. As a consequence the geometries of the
transition structures’ half-open epoxide complexes are very
similar to TS8a and TS8b.
This is also reflected in identical stability differences between
TS12a and TS12b and TS8a and TS8b, respectively. In the
case of an irreversible epoxide opening identical regioselec-
tivities are expected for 5a and 5b. Therefore, the Cp2TiCl-
derived transition structures, that are much easier to calculate,
can be used as excellent models for the computationally more
demanding (Cp2TiCl)2-derived structures.
A similar situation arises in the opening of 11. In TS11a only
the hydrogen atom and the methyl group attached to the
secondary carbon are positioned at a close distance to the ligands
(2.23 and 2.35 Å), whereas in TS11b all three methyl groups
are in close contact to the cyclopentadienyl ligands. As in the
case of TS8a, TS11a is therefore electronically and sterically
favored. The stability difference of the transition structures due
to the steric differentiation should be increased even further,
when the cyclopentadienyl ligands contain properly positioned
bulky alkyl substituents. Support for this hypothesis and its
applications in stereoselective catalysis is presented in the
context of the experimental results.
These findings are highly pertinent for the discussion of the
reactions of substituted titanocenes. However, for 1 and the Cp2-
TiCl2-derived catalysts, that are most frequently used in synthetic
applications reactions of epoxide, complexes of the highly Lewis
acidic dimers may constitute the prevalent (depending on the
exact concentrations employed) and more reactive species in
solution. Our calculations for the complex 5b*8 are summarized
in Scheme 6.
The reliability of the computational results can be qualitatively
estimated by comparing predicted and experimentally deter-
9
J. AM. CHEM. SOC. VOL. 129, NO. 5, 2007 1367

A R T I C L E S
Gansa¨uer et al.
Figure 7. Transition structures (BP86/TZVP) of the opening of 8 with 5b. Distances in angstrom. (Ti-Cl distances in angstrom are the following. TS12a:
Ti1-Cl1 2.558, Ti2-Cl1 2.699, Ti2-Cl2 2.454, Ti1-Cl2 4.313. TS12b: Ti1-Cl1 2.553, Ti2-Cl1 2.687, Ti2-Cl2 2.454, Ti1-Cl2 4.317.)
Scheme 7. Computed and Observed Regioselectivity of an Irreversible Opening of Various Epoxides
mined relative values of the regioselectivity of epoxide opening.
For an irreversible ring opening reaction the calculations predict
ratios of 88:12 for 8a and 8b, 97:3 for 10a and 10b, and 95:5
for 11a and 11b at 25 °C as shown in Scheme 7. In the case of
a swift radical trapping the same ratios are expected in the final
products, also.
Discussion of the Reaction Energies of Ring Opening. Even
though the kinetic measurements suggest that epoxide opening
is kinetically controlled, analysis of the reaction energies of ring
opening provides additional and essential insight into important
features of this process. Epoxide opening is exothermic, and
the formation of the higher substituted radicals is thermody-
namically favored in all cases investigated as summarized above
in Table 3.
Quite surprisingly, the differences in the reaction energies
of a number of â-titanoxy radicals do not correlate with the
stability differences of related free alkyl radicals based on bond
dissociation energies (BDE).28 For example, 7a and 8a are
formed with about the same reaction energy (∆E ≈ -4.0 kcal
mol-1) from their Cp2TiCl complexes, even though secondary
radicals are more stable than primary radicals by approximately
3 kcal mol-1 based on their BDE. This must be caused by steric
interactions with the cyclopentadienyl ligand in 8a, which fully
counterbalances the radical’s increased stability. A similar trend
of product stability is observed in the generation of the radicals
10b and 11b.
of the looser structure of 10a releases some of the energy
associated with the less favorable complexation in 5a*10.
Another strong indication of the pertinence of steric effects is
obtained from the energy of ring opening of 9. Even though a
secondary radical is formed as in the case of 8, the energy of
ring opening is higher by about 2 kcal mol-1. This is due to
relatively strong interactions of the additional methyl group with
the cyclopentadienyl ligands.
For the dimer 5b the opening of 8 was investigated compu-
tationally (Scheme 6). The reaction energies of ring opening to
12a and 12b are -6.4 and -4.6 kcal mol-1, respectively, and
thus about 3 kcal mol-1 more favorable than for 5a*8. This
can be explained by a higher stabilization of the more Lewis
acidic dimer through the negatively charged oxygen. The
stability differences between 12a and 12b are only slightly lower
than for 8a and 8b.
An important mechanistic perspective arises from the low
energy of activation and low energy of formation of the primary
radicals 8b and 10b and the secondary radicals 8a, 9a, and 11b.
The ring closure of these â-metaloxy radicals to the initial
titanocene epoxide complexes should become possible in the
(28) See for example: (a) McMillen, D. F.; Golden, D. M. Annu. ReV. Phys.
Chem. 1982, 32, 493-532. (b) Ru¨chardt, C.; Beckhaus, H.-D. Top. Curr.
Chem. 1985, 130, 1-22. (c) Bordwell, F. G.; Zhang, X.-M. Acc. Chem.
Res. 1993, 26, 510-517. (d) Brocks, J. J.; Beckhaus, H.-D.; Beckwith,
A. L. J.; Ru¨chardt, C. J. Org. Chem. 1998, 63, 1935-1943. (e) Blanksby,
S. J.; Ellison, G. B. Acc. Chem. Res. 2003, 36, 255-263. (f) Zipse, H.
Top. Curr. Chem. 2006, 263, 163-189 and references therein. For an
interesting discussion concerning radical stabilization, see: (g) Coote, M.
L.; Pross, A.; Radom, L. Org. Lett. 2003, 5, 4689-4692. (h) Gronert, S.
J. Org. Chem. 2006, 71, 1209-1219. (i) Gronert, S. J. Org. Chem. 2006,
71, 7045-7048.
The formation of 10a is more favorable by 4.5 kcal mol-1
compared to 8a. This is even higher than the typical stability
difference between a tertiary and a secondary radical. Generation
9
1368 J. AM. CHEM. SOC. VOL. 129, NO. 5, 2007

Titanocene-Mediated Epoxide Opening
A R T I C L E S
Scheme 9. Selectivities in the Opening of 13 with the Catalysts
Cp2TiCl2 and 1-2
absence of efficient radical traps. As demonstrated below, we
have indeed been able to obtain the very first experimental
evidence for this hypothesis of a reversible epoxide opening.
In general, the reaction energies of ring opening are surpris-
ingly low when considering the release of the oxirane’s ring
strain and the formation of the strong titanium-oxygen bond.
We suggest that the difference in stability between the titanium-
centered and carbon-centered radical is responsible for these
low values. Apparently, the high stability of the metal-centered
radical almost compensates the loss of ring strain and the
favorable formation of the titanium-oxygen bond.
This suggests that cleavage of aziridines or even oxetanes,
that are less strained than epoxides, with Cp2TiCl will be
difficult, if not impossible. Moreover, it should be feasible to
prepare relatively unstrained rings through a homolytic substitu-
tion reaction of titanium-oxygen bonds by radicals. This has
indeed been observed by us for tetrahydrofurans,3k,m as shown
in Scheme 8, and later by others, also.2k,29 The energy of
formation of the cis-bicyclo[3.3.0] system was calculated to be
exothermic by about 12 kcal mol-1
.
Scheme 8. Tetrahydrofuran Formation via Homolytic Substitution
obtained from the reactions involving tert-butyl acrylate suggests
that even under the catalytic conditions epoxide opening
constitutes the slowest step of the overall reaction. Thus,
reductive trapping of the enoyl radical generated through
addition to the acrylate, that is favored by polar effects, seems
to be substantially faster than the disfavored polymerization of
the acrylate.2b,3n Note that collidine hydrochloride serves as
proton donor to liberate the Ti(IV) species from the ionic
products formed.
The substituted catalyst 1 resulted in a slightly but noticeably
higher selectivity of ring opening as expected if selectivity is
influenced by steric factors in the transition states. However,
in the γ-terpinene case, the isolated yields are too low to be
synthetically useful. Catalyst 2 performs distinctly worse as
already suggested by the kinetic experiments. Clearly, tertiary
alkyl substitution is not suitable for an improved performance
of the catalytic reaction.
This ring closure is mechanistically related to the alkoxylation
of Mn-O and Cr-O bonds via radicals that is discussed as
one of the key steps of the Mn(salen) and Cr(salen)-catalyzed
epoxidations.10 It is even reminiscent of the postulated oxygen
rebound step of the P450 enzymes for the hydroxylation of
radicals.12
The titanocene-catalyzed reductive epoxide opening consti-
tutes the microscopic reversal of the ring-forming step of the
epoxidations. Thus, in homolytic substitution reactions involving
metal-oxygen bonds a subtle interplay between the stability
of the metal- and carbon-centered radicals, the metal-oxygen
bond strength, and the size of the rings formed or broken is
operating.
Synthetic Studies. The electrochemical, kinetic, and com-
putational investigations have provided a wealth of useful
mechanistic information and predictions about the outcome of
epoxide opening. In the following sections we describe our
efforts aimed at verifying these predictions, at understanding
and designing stereoselective reactions of enantiomerically pure
catalysts with enantiomerically pure substrates, and at the
development of the first reversible epoxide openings. To confirm
the predicted regioselectivities, the reactions of 13, a nonvolatile
substitute of 8, were carried out with Cp2TiCl2, 1, and 2, using
the hydrogen atom donor γ-terpinene as reducing agent or tert-
butyl acrylate as radical trap as shown in Scheme 9.
The regioselectivities observed with Cp2TiCl2 are in excellent
agreement with the predicted values and independent of the
reducing/trapping agent. Thus, our mechanistic proposal of an
irreversible ring opening obtained from the kinetic studies is
strongly supported for the Cp2TiCl2-derived catalysts. The
observation that no substantial amounts of polymers were
A more powerful approach to the understanding of the
importance of steric effects in the ring opening of epoxides is
constituted by the reaction of enantiomerically pure substrates
and catalysts. In these cases, diastereomeric epoxide titanocene
complexes are formed that may display differing reactivity
profiles. In this respect, ring opening of (R)-13 with both
enantiomers of Zn-3 in the presence of γ-terpinene or tert-
butyl acrylate leads to instructive results. Matched and mis-
matched cases of regioselectivity were observed as expected
for an irreversible course of the reactions as shown in Scheme
10. Gratifyingly, in the matched cases complete selectivity of
epoxide opening was combined with higher yields of the
products.
An interesting outcome was observed for the mismatched
cases. The ratio of primary to secondary alcohols in the products
depended noticeably and reproducibly on the radical trap
employed. We suggest that a closure of the intermediate primary
radical is faster than its trapping, and thus epoxide opening of
13 to the primary radical becomes reversible. The approach of
trapping agents such as γ-terpinene is hindered as the radical
center is positioned in the sterically blocked environment of
the catalyst’s pocket. Considering that the ring opening of
(29) (a) Trost, B. M.; Shen, H. C.; Surivet, J.-P. Angew. Chem., Int. Ed. 2003,
42, 3943-3947. (b) Trost, B. M.; Shen, H. C.; Surivet, J.-P. J. Am. Chem.
Soc. 2004, 126, 12565-12579. (c) Leca, D.; Song, K.; Albert, M.;
Goanc¸alves, M. G.; Fensterbank, L.; Lacoˆte, E.; Malacria, M. Synthesis
2005, 1405-1420.
9
J. AM. CHEM. SOC. VOL. 129, NO. 5, 2007 1369

A R T I C L E S
Gansa¨uer et al.
Scheme 10. Matched and Mismatched Selectivity in the Opening
of (R)-13 by 3 and ent-3
Scheme 11. Enantioselective Opening of meso-Epoxide 17 with 3
and Structure of Zn-3*17
group and the ligand during the preferred formation of (S)-19.
Thus, the experimentally observed formation of (S)-19 is readily
explained by our simple model of minimizing steric interactions.
Finally, we investigated the cis-1,2-disubstituted epoxide 18
with two substituents displaying distinctly differing steric bulk.
This was done primarily with the goal of understanding the
interactions of the sterically demanding cyclohexyl group with
the ligands of Zn-3. An additional point of interest, that is also
shown in Scheme 12, arises from the bulky substituent of 18.
Due to this shielding reactions of the radical with sterically
demanding radical traps may become particularly slow. There-
fore epoxide opening could become reversible.
monosubstituted epoxides to primary radicals is only just
thermodynamically favorable and the activation energies of ring
opening are low (Table 3), this ring closure should not constitute
a problem thermodynamically or kinetically.
While these mismatched cases of ring opening provide the
first reliable hints for a reversible epoxide opening, the analysis
of the results is somewhat complicated by the presence of both
electronic and steric effects on the regioselectivity of the overall
transformation. To exclude the influence of electronic contribu-
tions on the regioselectivity and reversibility of ring opening,
we investigated the reactions of two cis-1,2-disubstituted
epoxides, denoted meso-epoxide 17 and 18, featuring a smaller
and a large substituent.
Scheme 12. Issues of Reactivity in the Opening of 18 ([Ti] )
Cp2TiCl or Zn-3)
Meso-epoxides constitute the easiest starting point for a study
of the steric control of ring opening. Both substituents of the
epoxide are identical, and the reactions are not complicated by
kinetic resolutions. An example of our highly enantioselective
opening of meso-epoxide 17 is shown in Scheme 11 together
with the epoxide catalyst complex Zn-3*17.3d,i
The highly enantioselective preparation of 19 amply dem-
onstrates the validity of our concept of steric control in the
absence of an electronic preference for ring opening. A
qualitative explanation for the high selectivity can be derived
by a correlation between the substrate titanocene complex Zn-
3*17 and the transition structures as already carried out above
for TS8a and TS8b. Breaking of the left-hand C-O bond in
Zn-3*17 results in an increased and energetically unfavorable
interaction of the right-hand CH2OCH2CH3 group with the
ligand compared to the complex. In this manner (R)-19 will be
formed after reduction and protonation.
The results of our investigations with Cp2TiCl2 and 3 as
catalysts and CHD and the commercially available γ-terpinene
as hydrogen atom donors are summarized in Scheme 13. With
the combination of Cp2TiCl2 and CHD the observed regiose-
lectivity (4.7:1 in favor of 20) is in line with a minimization of
the interactions of the cyclopentadienyl ligand with the cyclo-
hexyl group. However, matters are more complicated as the
regioselectivity markedly depends on the hydrogen atom donor
Breaking of the right-hand C-O results in a diminishing of
the steric interactions between the right-hand CH2OCH2CH3
9
1370 J. AM. CHEM. SOC. VOL. 129, NO. 5, 2007

Titanocene-Mediated Epoxide Opening
A R T I C L E S
Scheme 13. Opening of 18 by Cp2TiCl and 3
to be strongly affected by the exact steric conditions. The kinetic
measurements strongly suggest an irreversible epoxide opening
in the presence of efficient radical traps. These results were used
as the experimental basis for the calculation of the structure
and energy of the titanocene(III)-epoxide complexes, the
transition states of ring opening, and the product â-titanoxy
radicals. For steric reasons 1,1-di- and higher substituted
epoxides bind differently than mono- and 1,2-disubstituted
epoxides. The transition structures and activation energies
suggest that both the stability differences of the forming radicals
and steric interactions between the catalysts’ ligands and the
substrate are essential in controlling the regioselectivity of ring
opening. The relevance of the computational study is highlighted
by the excellent agreement between the predicted and experi-
mentally observed regioselectivities.
The use of Kagan’s catalyst 313 in the opening of monosub-
stituted and cis-1,2-disubstituted epoxides has led to highly
selective and preparatively useful epoxide openings on several
occasions that can be readily rationalized with our steric model
of controlling the regioselectivity of ring opening. In the
mismatched cases the first examples of reversible epoxide
openings were observed. With cis-1,2-disubstituted epoxides
containing one bulky and one small substituent the course of
the epoxide opening depends on the catalyst employed. Our
results provide a unique link between the mechanism of the
titanocene-mediated epoxide opening and that of epoxidation
reactions featuring â-metal oxy radicals,10 and of hydroxylation
reactions featuring C-H activation steps by organometallic
reagents11 or the P450 enzymes.12
employed. With the sterically more demanding γ-terpinene22
the ratio of 20 to 21 was reduced to 1.6:1. The sterically hindered
radical 18a is reduced slower than 18b by the latter H atom
donor. These findings exclude an irreversible ring opening in
this specific situation and strongly support a reversible epoxide
opening.
However, this situation changes completely with catalyst 3.
Under all conditions 20 was obtained as the sole product of an
irreversible epoxide opening. Alcohol 20 was obtained in almost
racemic form (e.r. < 60:40). Thus, 3 does not constitute an
efficient catalyst for the resolution of cis-1,2-disubstituted
epoxides. Moreover, the steric interactions of either enantiomer
of the bulky catalyst with the cyclohexyl group are by far more
important than any other factor including the slow reduction of
the sterically hindered radical 18a. Epoxide opening appears to
be irreversible. Thus, it is in principle possible to switch between
an irreversible and reversible epoxide opening by properly
adjusting the steric demand of the catalyst, the radical’s
substituents, and the trapping reagent.
Acknowledgment. We are indebted to the Danish Natural
Science Research Council, the Deutsche Forschungsgemein-
schaft, and the Fonds der Chemischen Industrie for continuing
and generous financial support.
Supporting Information Available: Details of the cyclic
voltammetric studies, atomic coordinates of all structures, and
experimental procedures. This material is available free of
charge via the Internet at http://pubs.acs.org.
Conclusion
In summary this study has established the structure of a
number of Ti(III) catalysts in solution and the kinetics of the
titanocene-mediated epoxide reduction. In particular, the dis-
tribution of monomeric and dimeric Ti(III) species was found
JA067054E
9
J. AM. CHEM. SOC. VOL. 129, NO. 5, 2007 1371