Bis(oxazoline)titanium complexes as chiral catalysts for enantioselective hydrosilylation of ketones - A combined experimental and theoretical investigation

Article abstract of DOI:10.1002/ejoc.200200675

A combined experimental and theoretical investigation has been carried out on a new catalytic system, based on bis-(oxazoline) (BOX) complexes of titanium. These catalytic species are able to reduce aromatic ketones with good enantiomeric excesses and satisfactory yields. The experimental and the computational (DFT) evidence has provided useful information on the nature of the active catalytic species and on the mechanism of the reaction. The most likely reaction path involves a Ti^IV catalytic species. This result agrees with experimentally obtained evidence that seems to rule out the presence of Ti^III species. The analysis of the structure of the transition state corresponding to the reduction process (the addition of the hydride to the carbonyl system), provides an interesting insight on the enantioselectivity that characterizes this reaction. Wiley-VCH Verlag GmbH & Co. KGaA, 69451 Weinheim, Germany, 2003.

Full text of DOI:10.1002/ejoc.200200675

FULL PAPER
Bis(oxazoline)titanium Complexes as Chiral Catalysts for Enantioselective
Hydrosilylation of Ketones ؊ A Combined Experimental and Theoretical
Investigation
[
a]
[a]
[a]
[a]
Marco Bandini, Fernando Bernardi, Andrea Bottoni,* Pier Giorgio Cozzi,*
[
a]
[a]
Gian Pietro Miscione, and Achille Umani-Ronchi
Keywords: Asymmetric catalysis / Density functional calculations / Ketones / Reduction / Transition states
A combined experimental and theoretical investigation has
been carried out on a new catalytic system, based on bis-
agrees with experimentally obtained evidence that seems to
rule out the presence of Ti species. The analysis of the
III
(
oxazoline) (BOX) complexes of titanium. These catalytic
structure of the transition state corresponding to the reduc-
tion process (the addition of the hydride to the carbonyl sys-
tem), provides an interesting insight on the enantioselectivity
that characterizes this reaction.
species are able to reduce aromatic ketones with good en-
antiomeric excesses and satisfactory yields. The experi-
mental and the computational (DFT) evidence has provided
useful information on the nature of the active catalytic spe-
cies and on the mechanism of the reaction. The most likely
reaction path involves a Ti catalytic species. This result Germany, 2003)
( Wiley-VCH Verlag GmbH & Co. KGaA, 69451 Weinheim,
IV
Introduction
the coordination of early transition metals.[8Ϫ15] Among
these, one interesting example is represented by the bis(ox-
[
16]
The development of chiral ansa-metallocene[1] catalysts
azoline) (BOX) ligands (see Scheme 1).
Pioneering stud-
[
17]
[18]
[19]
ies carried out by Pfaltz,
Evans,
Jørgensen,
and
for the stereospecific polymerization of propylene has en-
couraged chemists in industrial and academic laboratories
to extend the use of the species to stereoselective organic
synthesis. For instance, the chiral C2 ansa-metallocenes
[
20]
ʝ
others have shown that these chelating N N ligands are
capable of coordinating many different metal atoms and are
particularly suitable for asymmetric catalysis.
[
(EBTHI)MX ] (M ϭ Ti, Zr; X ϭ OR, Cl, F; EBTHI ϭ
2
ethylenebis(tetrahydroindenyl)] introduced by Brintzinger
can be used to mediate interesting asymmetric
[
2Ϫ3]
transformations.
In particular, chiral titanocene com-
plexes [(EBTHI)TiX ] are able to promote effectively the
2
hydrosilylation of ketones and imines in high enantiomeric
[
4]
excess. This synthetic approach has been included in in-
[
5]
dustrial methods and has been used to prepare biologi-
cally active products.[ However, even though optically ac-
6]
tive [(EBTHI)TiF ] complexes are now commercially avail-
2
able, their use remains quite expensive. The preparation of
optically active chiral titanocenes is not in fact a trivial pro-
cedure, and great skill in organometallic synthesis and un-
[
7]
usual facilities are required.
Recently, the search for more effective and practical poly-
merization catalysts based on non-cyclopentadienyl frame-
works has provided new and interesting ligands suitable for
Scheme 1
[
a]
Dipartimento di Chimica ‘‘G.Ciamician’’, Universita’ di
Bologna,
The objective of the current study is to show that chiral
EBTHI ligands can be replaced by commercially available
or easily prepared chiral bis(oxazolines), avoiding the reso-
Via Selmi 2, 40126 Bologna, Italy
Fax: (internat.) ϩ 39-051/2099456
E-mail: andrea@ciamserv.ciam.unibo.it
Supporting information for this article is available on the
WWW under http://www.eurjoc.org or from the author.
lution process needed to obtain optically pure C metalloc-
2
enes. To this end the various BOX ligands shown in
2972
 2003 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
DOI: 10.1002/ejoc.200200675
Eur. J. Org. Chem. 2003, 2972Ϫ2984

Bis(oxazoline)titanium Complexes as Chiral Catalysts
FULL PAPER
Scheme 1 have been examined. In the new synthetic route Table 1. Enantioselective hydrosilylation of acetophenone by use of
that we propose, the ansa-bis(indenyl) anion is replaced by various BOX ligands
two bis(oxazolinate) anions obtained by deprotonation of
the corresponding neutral ligands. To compare the perform-
ances of our (BOX)titanium complexes to those of ansa-
titanocene complexes, we chose as a reference reaction the
hydrosilylation of aromatic ketones in the presence of
Yield (%)[a]
ee (%)[b]
Box
4
TiX (mol %)
1
TiCl
TiCl
TiCl
TiCl
TiCl
TiCl
4
4
4
2
2
2
(THF)
(THF)
(THF)
2
2
2
(4)
(3)
(4)
(15)
(4)
(3)
60
60
20
87
85
13
86
28
30 (R)
50 (R)
23 (R)
18 (S)
51 (S)
64 (S)
61 (S)
72 (S)
[
c]
1
3
1
4
5
4
4
(
EtO) SiH, since this reaction, mediated by chiral (ETBHI)-
3
(OiPr)
(OiPr)
(OiPr)
(5)
2
titanium complexes, has been widely investigated over the
2
[
21]
In addition to the experimental study,[22] in
last decade.
2
TiF
TiF
4
order to obtain more detailed information on the mechan-
istic aspects of this reaction, we have also carried out a
theoretical investigation at the DFT level.
[
d]
4
(5)
[a]
Isolated yields after chromatographic purification. [b] Percent en-
antiomeric excess was determined by chiral GC analysis (Megadex
). [ The reaction was carried out in the presence of molecular
c]
5
[
d]
2
sieves (1 g). The reaction was run in Et O.
Results and Discussion
ported in Table 1. From these data it is evident that the
best results (yield and enantiomeric excess) were obtained
Experimental Results
[26]
through the use of TiF4
and BOX ligand 4.
'
It is noteworthy that the (BOX)Ti complexes can be ob-
The BOX ligands 1Ϫ5 shown in Scheme 1 were used to
prepare chiral organometallic complexes capable of replac-
tained by a simpler route in which Et N is used as a base
(
3
see Scheme 2, route B); the results obtained with different
ing the [(S,S)-(ETBHI)TiF ] species. While BOX ligands
2
BOX ligands are collected in Table 2. We found that the
titanium complexes prepared in this way were able to re-
duce acetophenone in satisfactory yields. However, in spite
of the easier catalyst preparation, the enantiomeric excesses
were significantly lower than those obtained from the reac-
tion in which the catalyst was prepared with nBuLi. It is
noteworthy that no activation steps are necessary to per-
form the reduction of the acetophenone in either of our
procedures.
2
Ϫ5 are commercially available, BOX ligand 1 was obtained
[23]
by the general procedure described by Evans. The metall-
ation of the bis(oxazoline) species was performed under de-
[
24]
scribed conditions (nBuLi, THF, Ϫ78 °C as summarized
in Scheme 2, route A). Different titanium salts Ϫ TiCl4,
TiCl (THF) , Ti(OiPr) , Ti(OiPr) Cl , and TiF Ϫ were
4
2
4
2
2
4
[
27]
employed with the BOX ligands 1Ϫ5 in the stoichiometric
ratio 1:2. The addition of the titanium salt TiX was carried
4
out at 0 °C, giving a pale orange solution. This procedure
provided various (BOX)titanium complexes that we used in
Table 2. Enantioselective hydrosilylation of acetophenone by use of
the catalytic hydrosilylation of acetophenone. The reaction the (BOX)Ti complexes prepared with Et N
3
was found to be very slow with many silyl hydrides such
as Et SiH, Cl SiH, Ph SiH, PMSH, and (Me Si) SiH. In
3
3
3
3
3
Yield (%)[a]
ee (%)[b]
Box
TiF
4
(mol %)
contrast, though, we obtained much more interesting re-
[25]
sults from the more reactive (EtO) SiH
species, as re-
1
1
1
3
4
4
TiF (10)
92
88
84
84
75
88
25 (R)
24 (R)
24 (R)
18 (R)
51 (S)
49 (S)
3
4
TiF
TiF
TiF
TiF
TiF
4
4
4
4
4
(5)
(5)
(5)
(5)
(5)
[c]
[d]
[a]
Isolated yields after chromatographic purification. [b] Percent en-
antiomeric excess was determined by chiral GC analysis (Megadex
). [ Hünig base was used instead of Et
c]
N.
[d]
The reaction was
5
3
carried out at Ϫ20 °C.
To monitor the formation of the titanium complex we
examined the NMR proton region in the δ ϭ 6.0Ϫ5.4 ppm
range for the chiral BOX ligand 4. The free bis(oxazoline)
has two protons, which appear as two doublets centered at
δ ϭ 5.98 and 5.65 ppm (AB system), respectively, with a
coupling constant of J ϭ 9.0 Hz. The addition of 1 equiv.
of TiF to 2 equiv. of BOX ligand 4 in CD Cl caused the
4
2
2
appearance of two broad signals at δ ϭ 6.0Ϫ5.8 ppm and
Scheme 2
5.8Ϫ5.65 ppm. This evidence suggested a fast exchange be-
Eur. J. Org. Chem. 2003, 2972Ϫ2984
www.eurjoc.org
 2003 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
2973

M. Bandini, F. Bernardi, A. Bottoni, P. G. Cozzi, G. P. Miscione, A. Umani-Ronchi
FULL PAPER
tween a titanium complex and the free BOX ligand 4. Ad- 4.52 ppm, respectively) and those obtained by Singh for the
[
29]
dition of 2 equiv. of Et N caused the appearance of two [Zn(BOX)Cl] complex.
This experimental evidence thus
3
new doublets (δ ϭ 5.89 and 5.41 ppm) and the vanishing of suggests that the (BOX)Ti complex formed in our case has
the two broad signals. The ratio between the two doublets similar structural features. The NMR spectra also indicate
of the free bis(oxazoline) ring and the two doublets of the the formation of highly symmetric (BOX)titanium com-
1
newly formed species is 5.6:1. The H NMR therefore sug- plexes.
gested that the metallation results in the formation of only
Two important points should be stressed. Firstly, the ad-
dition of 1 equiv. of acetophenone into the NMR tube con-
one species. Similar information was obtained from the 13
C
NMR spectra (see Exp. Sect.). Furthermore, after metall- taining the complex does not affect the signals of the com-
ation, the signal of the methine proton of the oxazoline plex. This finding is in agreement with a coordinatively
bridge is shifted to low field: the new titanium species saturated (BOX)Ti complex. Secondly, the addition of 3
formed shows a singlet at δ ϭ 4.69 ppm, while in the free equiv. of triethoxysilane into the NMR tube again leaves
BOX ligand the signal of the protons of the CH bridge the spectra unchanged and no evidence for the formation
2
appear at δ ϭ 3.88 ppm. These results are in agreement with of paramagnetic species could be detected. This result
those reported by Anwander[28] for the methine proton of makes the hypothesis of a mechanism involving Ti spec-
III
the [Ln(BOX) ] complexes (Ln ϭ Y, La; δ ϭ 4.61 and ies unlikely.
2
Table 3. Enantioselective hydrosilylation of ketones[a]
[
a]
Reactions were run with the use of the (BOX)titanium catalyst, prepared in situ from 4 and TiF
4
in THF. [b] Percent enantiomeric
[c]
[d]
excess was determined by chiral GC analysis (Megadex 5). Isolated yield after purification. The enantiomeric excess was evaluated
on the corresponding silyl ether. [ The enantiomeric excess was evaluated by HPLC (Chiralcel OD).
e]
2974
 2003 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.eurjoc.org
Eur. J. Org. Chem. 2003, 2972Ϫ2984

Bis(oxazoline)titanium Complexes as Chiral Catalysts
FULL PAPER
We now discuss the results obtained when the depro- case, each of the two rings of the chelating BOX ligand,
tonation of the BOX species was performed with nBuLi in which is deprotonated in the CH bridge position, was re-
2
[
D ]THF. Under these conditions, the recorded spectra for placed by an OCH group bonded to a carbon atom and a
8 3
the titanium complex were similar to those previously dis- CH group bonded to a nitrogen atom, while the ketone
3
cussed, while the ratio between the proton signals of the substrate was emulated by a formaldehyde molecule. The
III
IV
free BOX ligand and those of the new formed titanium doublet (Ti ) and singlet (Ti ) potential energy surfaces
species changed, becoming 1.5:1 in this case. The NMR associated with these two model reactions are shown in Fig-
analysis therefore indicates that the (BOX)titanium com- ure 1 and the critical points located along them are sche-
plexes obtained by the two different procedures correspond matically represented in Figures 2 and 3 for reaction A, and
to the same species. All the various experimental results that Figures 4 and Figure 5 for reaction B. The values of the
we have discussed in detail, and also the data obtained in most relevant geometrical parameters and the energies rela-
previous crystallographic studies[ on stable titanium com- tive to reactants are collected in the Figures. To check the
plexes, suggest that a plausible formula for the (BOX)Ti reliability of this model system, we replaced the formal-
30]
complex is [Ti(BOX) F ].
dehyde molecule with an acetone molecule and recomputed
2
2
The results obtained from the reduction of several aro- the structures of the critical points associated with the first
matic ketones and substituted α-halo ketones with the best steps of the two model reactions A and B. The results (geo-
catalytic system represented in Scheme 2 (BOX ligand 4, metrical parameters and energy values) obtained for this
nBuLi in THF, and TiF ) are collected in Table 3. These larger model system are reported in Figure 6. Plausible re-
4
IV
III
data show that satisfactory yields and good enantiomeric actions that may produce Ti or Ti complexes are illus-
excesses were obtained.
trated in Scheme 4. For these processes we have assumed
As demonstrated by Buchwald,[ aromatic ketones are for L the simplified form of the BOX ligand chosen in our
4e]
necessary in order to achieve high levels of enantioselectiv- model system and we have replaced (EtO) SiH with (MeO) -
3
3
ity. To explain this observation, a model transition state in SiH. The computed free energy values and the correspond-
which a chiral metallocene titanium() hydride coordinates ing equilibrium constants are reported in Scheme 4.
2
the ketone in a four-center (η ) structure was proposed.
Such a transition state does not seem suitable to explain
the mechanism of the present reaction, based on (BOX)Ti
III
complexes, where the presence of Ti species seems to be
ruled out by the experimentally obtained evidence. Thus, to
obtain more accurate information on the true nature of the
III
IV
active catalytic species present in solution (Ti or Ti
species) and to shed light on the mechanistic details of this
reaction (the most probable reaction path, activation ener-
gies, reaction enthalpies, and free energies), we carried out
a theoretical investigation at the DFT level. The results are
discussed in the following section.
Computational Modeling
In our theoretical study we considered either a ti-
tanium() or a titanium()[ hydride complex as possible
active species capable, in principle, of reducing the carbonyl
substrate. These two species are schematically represented
in the two model reactions A and B in Scheme 3. In each
31]
IV
Figure 1. Energy profiles for model reaction A (Ti catalytic spec-
III
ies) and model reaction B (Ti catalytic species); free energy values
are in parentheses; the MP2 values are in italics; the activation
barriers obtained in the presence of solvent effects are reported in
square brackets
Scheme 3
Eur. J. Org. Chem. 2003, 2972Ϫ2984
www.eurjoc.org
 2003 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
2975

M. Bandini, F. Bernardi, A. Bottoni, P. G. Cozzi, G. P. Miscione, A. Umani-Ronchi
FULL PAPER
A
A
A
A
1 2 1 3
Figure 2. Schematic representation of the structures of the four critical points M , M , TS , and M ; the energy and the free energy
Ϫ1
A
˚
2
ϩ H CO) and the most relevant geometrical parameters [A] are reported;
the absolute total energy and free energy values of reactants are Ϫ2121.0583 and Ϫ2120.6691 Hartree, respectively
(
in parentheses) values [kcal·mol ] relative to reactants (M
1
A
A
Figure 3. Schematic representations of the structures of the two critical points TS
2
and M
4
; the energy and the free energy
Ϫ1
A
˚
(
in parentheses) values [kcal·mol ] relative to reactants (M
1
ϩ H CO) and the most relevant geometrical parameters [A] are given
2
2976
 2003 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.eurjoc.org
Eur. J. Org. Chem. 2003, 2972Ϫ2984

Bis(oxazoline)titanium Complexes as Chiral Catalysts
FULL PAPER
B
B
B
B
Figure 4. Schematic representations of the structures of the four critical points M
1
, M
2
, TS
1
, and M
3
; the energy and the free energy
Ϫ1
B
˚
(
in parentheses) values [kcal·mol ] relative to reactants (M
1
ϩ H
2
CO) and the most relevant geometrical parameters [A] are given;
the absolute total energy and free energy values of reactants are Ϫ2021.6734 and Ϫ2022.2904 Hartree, respectively
Model Reaction A: The [(L) TiFH] Catalyst
˚
2
OCH group (the H···O distance is 1.921 A). This interac-
3
The energy profile for model reaction A is shown in Fig- tion seems to be responsible for the significant stabilization
ure 1, part a. The reaction is a three-step process, and is observed for M2A, although we should point out that this
Ϫ1
highly exothermic (∆E ϭ Ϫ74.13 kcal·mol ). Six critical stabilization is probably overestimated because of the small
A
A
points were located along the energy profile: M , M2
,
basis set on the oxygen atom. This hydrogen bond has the
1
TS1 , and M3 (Figure 2), together with TS and M4A effect of placing the substrate in the best position for the
Figure 3). In M1 , which represents the active form of the subsequent hydride transfer (transition state TS1 ). The
A
A
A
2
A
A
(
catalyst, the coordination of the metal atom is octahedral role played by this H bond in determining the spatial orien-
˚
A
with two TiϪN bonds in axial positions (2.060 and 2.103 A, tation of the formaldehyde is equally evident in TS1 , in
respectively). The first intermediate M2A forms from M1A which the H···O distance is only slightly longer (2.116 A)
and an isolated formaldehyde molecule without any barrier than in M2A. It is important to point out that the orienting
˚
A
Ϫ1
(
M2 is 24.15 kcal·mol lower in energy than the reac- effect of the hydrogen bond is not a shortcoming of the
tants) and corresponds to a long-range complex in which model system used here, since the involved oxygen atom is
the formaldehyde molecule weakly interacts with the cata- present in the same position in the real system. The re-
lyst. The distance between the hydrogen atom bonded to ported geometrical parameters indicate that the transition
the titanium atom and the formaldehyde carbon atom is state TS1A has a strongly reactant-like character, since the
˚
˚
˚
still 3.281 A, but a hydrogen bond forms between one for- breaking TiϪH bond is still 1.802 A (1.715 and 1.764 A in
maldehyde hydrogen atom and the oxygen atom of one M1 and M2A, respectively) and the newly forming CϪH
A
Eur. J. Org. Chem. 2003, 2972Ϫ2984
www.eurjoc.org
 2003 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
2977

M. Bandini, F. Bernardi, A. Bottoni, P. G. Cozzi, G. P. Miscione, A. Umani-Ronchi
FULL PAPER
B
B
Figure 5. Schematic representations of the structures of the two critical points TS
2 4
and M ; the energy and the free energy (in paren-
Ϫ1
B
˚
theses) values [kcal·mol ] relative to reactants (M
1 2
ϩ H CO) and the most relevant geometrical parameters [A] are given
A
A
B
B
1 1 1 1
Figure 6. Schematic representations of the structures of the four critical points M *, TS *, M *, and TS *; the energy and the free
Ϫ1
˚
energy (in parentheses) values [kcal·mol ] relative to reactants and the most relevant geometrical parameters [A] are given; total and
free energies of reactants: Ϫ2199.7168 (E) and Ϫ2199.2759 (G) Hartree (reaction A); Ϫ2100.3319 (E) and Ϫ2100.8972 (G) Hartree
(reaction B)
2978
 2003 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.eurjoc.org
Eur. J. Org. Chem. 2003, 2972Ϫ2984

Bis(oxazoline)titanium Complexes as Chiral Catalysts
FULL PAPER
reaction, which is exothermic by Ϫ70.02 kcal·molϪ1, is
again a three-step process characterized by six critical
B
B
B
B
points along the energy profile: M , M , TS , and M
1
2
1
3
B
B
(
Figure 4), together with TS2 and M4 (Figure 5). The ac-
tive complex M1 corresponds to a distorted trigonal bi-
pyramid in which two TiϪN bonds (2.076 A) are situated in
B
˚
axial positions. As previously observed along the reaction
B
Ϫ1
profile A, a preliminary complex M , 26.35 kcal·mol
2
B
lower in energy than the reactants, forms from M1 and an
isolated formaldehyde molecule. This transformation again
B
occurs without any barrier. In the M2 species the formal-
dehyde molecule forms two hydrogen bonds with the oxy-
Scheme 4
gen atom of one OCH group. These two hydrogen bonds
3
˚
bond is only 1.841 A. A slight lengthening of the carbonyl are fairly strong, as suggested by the two H···O distances
˚
˚
bond (1.236 A), which is losing its double-bond character, (2.025 and 2.290 A, respectively) and the energy lowering
is also observed. TS1A leads to a second intermediate M3
A
,
B
of the system. From M it is possible to reach a new inter-
2
˚
B
B
B
in which the new CϪH bond is almost complete (1.189 A) mediate M
through the transition state TS . In TS ,
1 1
3
and the negative oxygen atom interacts slightly with the which has a product-like character, the formaldehyde mol-
positive titanium atom, the TiϪO distance now being 3.774 ecule moves away from the H
CO group to coordinate the
CO molecule interacts signifi-
2
3
A. The transformation M Ǟ M3A requires the overcom- metal atom. Here, the H
˚
A
2
ing of an energy barrier of 4.45 kcal·mol , and M3A is cantly with the titanium atom (the TiϪO distance is 2.960
Ϫ1
Ϫ1
A
˚
4
.75 kcal·mol lower in energy than M . An additional A), while the distance between the titanium-bonded hydro-
2
transition state TS leads from M3A to the final product gen atom and the carbonyl carbon atom is still large (2.803
A
2
A
A
˚
B
M4 . In TS2 the H CO group rotates in such a way as to A). In the intermediate M , the TiϪO and C···H distances
position the negative oxygen atom closer to the metal atom. become 1.987 and 2.843 A, respectively. Thus, in this com-
3
3
˚
˚
Here the TiϪO distance becomes 3.073 A and the hydrogen plex, which corresponds to a square bipyramid with the for-
atom still weakly interacts with the metal atom, the TiϪH maldehyde molecule in equatorial position, the TiϪO bond
˚
Ϫ1
B
B
distance being 2.172 A. A barrier of 6.03 kcal·mol
characterizes this transformation. In the final product, 12.60 kcal·mol and the barrier required for the trans-
is almost complete. M
3
is higher in energy than M
2
by
Ϫ1
A
B
B
M4 , only the oxygen atom of the H CO fragment is coor- formation M
Ǟ M
(the rate-determining step of the
3
2
3
Ϫ1
[32]
dinated to the titanium atom. The TiϪO bond is now 1.874 process) is 20.82 kcal·mol . A computation of the intrinsic
A and the carbonyl group has definitely lost its double- reaction coordinate (IRC) was carried out, starting from
bond character (the CϪO bond is 1.394 A).
in both reactant and product directions, to demon-
In addition to the total energy values, we also computed strate how the three critical points M
˚
˚
B
TS
1
B
B
B
, TS , and M are
2
1 3
the free energy values, which are reported for the various connected on the reaction surface. It is interesting to note
critical points in parentheses in the diagram of Figure 1, the different natures of the corresponding critical points lo-
A
A
A
part a. The free energy profile is similar to that of the total cated for model reaction A (i.e., M
, TS
, and M
, in
). In the former
thermic (55.47 kcal·mol ) and the two free energy barriers case we showed that the hydrogen bond, which immediately
2
1
3
B
B, and M
B
energy previously discussed. The reaction is still highly exo- comparison to those of M , TS
2 1 3
Ϫ1
A
that must be overcome to obtain the product remain quite forms in M , has the effect of positioning the substrate
small (7.09 and 7.59 kcal·mol , respectively). Thus, both molecule in the right position for the hydride attack. In the
2
Ϫ1
sets of values (total energies and free energies) indicate that latter case, in contrast, two hydrogen bonds must be broken
IV
carbonyl reduction promoted by the Ti catalytic species to move the formaldehyde molecule close enough to the
is a quite easy reaction. However, it is important to stress TiϪH bond and to make the reduction process possible.
that, while the activation energies and the activation free This could explain the larger activation barrier found for
B
B
energies are very similar, more significant variations are ob- TS
. A second transition state (TS ), with an activation
1
2
Ϫ1
B
served when we compare the relative energies and relative barrier of 7.72 kcal·mol , affords the final product M
.
4
free energies of M2 , M3 , and M4A. For instance the en- In TS
A
A
B
the formaldehyde molecule rotates around the
2
ergy and free energy (relative to the asymptotic limit) of CϪO bond to position the molecular plane orthogonal to
A
Ϫ1
M2 are 24.15 and 13.31 kcal·mol , respectively. This re- the approaching direction of the hydride ion. This tran-
markable change is certainly due to the anomalous contri- sition state has a strongly reactant-like character and the
butions of low frequencies appearing in the presence of distance between the hydrogen atom and the formaldehyde
A
˚
˚
weakly interacting fragments such as in M ϩ H CϭO carbonyl atom is still large (2.808 A). In addition, no weak-
1
2
and M2A
.
ening of the TiϪH bond (1.792 A) is observed. The final
B
product M4 is a trigonal bipyramid in which the H CO
fragment, obtained after reduction, is situated in an equa-
3
B. Model Reaction B: The [(L) TiH] Catalyst
2
The energy profile for model reaction B (see Figure 1, b) torial position and two nitrogen atoms are in axial posi-
˚
is very similar to that discussed in the previous section. The tions. The final OϪTi distance is 1.918 A, while the CϪO
Eur. J. Org. Chem. 2003, 2972Ϫ2984
www.eurjoc.org
 2003 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
2979

M. Bandini, F. Bernardi, A. Bottoni, P. G. Cozzi, G. P. Miscione, A. Umani-Ronchi
FULL PAPER
˚
carbonyl bond has become 1.401 A, as one can expect for bond between a methyl hydrogen atom of the ketone and
a single CϪO σ bond. the oxygen atom of one methoxy group. The H···O distance
˚
˚
A
The free energy values (reported in parentheses in Fig- is 2.047 A (it was 1.921 A in M ). Similarly, the distance
2
ure 1, b) provide information similar to that obtained from between the titanium-bonded hydrogen atom and the ace-
˚
the total energy curve. The reaction is still highly exother- tone carbon atom (3.058 A) does not differ significantly
Ϫ1
B
˚
from the value found in the case of formaldehyde (3.281 A).
mic (Ϫ51.21 kcal·mol ) and, even if the intermediate M3
is significant higher in energy (it becomes almost degenerate TS1 * also has a structure similar to that of TS1 . As
A
A
B
to reactants), the two energy barriers associated with TS1
and TS2 do not change significantly, being 21.32 and 7.32 characterized here by an H···O distance of 2.122 A, is im-
kcal·mol respectively.
pointed out in the case of formaldehyde, the H bond,
B
˚
Ϫ1
portant for maintaining the spatial orientation of the sub-
A
strate. The structural parameters computed for TS1 * indi-
A
The Solvent Effect
cate a more strongly product-like character than TS1 , since
the breaking TiϪH bond and the newly forming CϪH
bond are now 1.822 and 1.663 A, respectively. However,
The activation energies of the first reaction step for both
˚
model systems (M2 Ǟ TS2A and M2 Ǟ TS2 transform-
ations) were recomputed with allowance for solvent effects.
The emulated solvent in the COSMO computations is tetra-
hydrofuran (dielectric constant ε ϭ 7.58), the solvent used
in the real experiment. In the presence of the solvent the
activation barriers for the two model reactions A and B
A
B
B
this stronger product-like character is associated with an
increase in the activation energy, which becomes 6.34
Ϫ1
Ϫ1
kcal·mol
(4.45 kcal·mol
with formaldehyde). It is
reasonable to ascribe this larger barrier to an increase in the
steric hindrance associated with the acetone methyl groups.
B
As observed for reaction A, the results obtained for M *
2
(
2
values in square brackets in Figure 1) become 0.54 and
B
B
3.29 kcal·molϪ1, respectively. Thus, even when the environ-
and TS * are also very close to those described for M
1
2
B
B
and TS , respectively. The preliminary complex M * is
1
2
ment solvent effect is taken into account, the reaction chan-
Ϫ1
IV
30.82 kcal·mol lower in energy than that of the reactants
M1 and isolated acetone) and its formation does not re-
nel involving Ti species is still highly favored.
B
(
quire any energy barrier. The reduction in energy with re-
spect to the reactants is again caused by the formation of
two hydrogen bonds between the methyl hydrogen atoms
MP2 Computations
The total energies of the first three critical points along
A
A
A
the two reaction profiles (M ϩ H CO, M , TS1 and
1
2
2
and the oxygen atom of one H CO group. The H···O dis-
3
B
B
B
M1 ϩ H CO, M , and TS ) were recomputed at the
˚
B
2
2
1
tances in this case are 2.737 and 2.563 A. In TS * the
1
MP2 level. The results are reported in italics in Figure 1.
For both reaction pathways the formation of the first inter-
acetone molecule behaves similarly to that of the formal-
B
dehyde molecule in TS1 and thus moves closer to the ti-
A
B
mediate (M2 and M ) again provides a strong stabiliz-
˚
2
tanium atom. The TiϪO distance becomes 1.850 A and that
Ϫ1
ation of the system: 32.16 and 33.76 kcal·mol , respec-
tively. Furthermore, the activation barriers for the first step
between the titanium-bonded hydrogen atom and the car-
˚
bonyl carbon atom 2.379 A. To approach the metal atom,
A
A
B
B
(
M2 Ǟ TS
and M2 Ǟ TS ) are 3.65 and 35.73
1
1
the acetone molecule must break one hydrogen bond as ob-
Ϫ1
kcal·mol . The MP2 method thus provides the same mech-
anistic scenario as obtained at the DFT level: the reaction
pathway corresponding to model reaction A (TiIV species)
is again much more likely than that found for model reac-
B
served in TS . The breaking of this bond determines the
1
B
entity of the activation barrier for the transformation M *
2
B
Ǟ TS *. Even if this barrier decreases with respect to the
1
simpler model based on formaldehyde (it now becomes
III
tion B (Ti species). These results also indicate that the
Ϫ1
1
2.20 kcal·mol ), it remains much larger than that found
large stabilization found for the first intermediate complex
is not a shortcoming of the computational method.
IV
for reaction A. The pathway involving Ti species is thus
also favored over that based on Ti complexes in the pres-
III
ence of acetone as substrate. A similar conclusion can be
reached from the free energy values reported in parentheses
The Model System Involving Acetone
A
and TS A in Figure 6; the free activation barriers for the two model
The structures of the critical points M
2
1
B
B
Ϫ1
(
model reaction A) and M2 and TS1 (model reaction B) reactions A and B are, in fact, 7.04 and 15.28 kcal·mol ,
were recomputed after the formaldehyde molecule had been respectively. These results suggest that the simpler model
replaced with an acetone molecule. The four new computed system based on formaldehyde emulates the reaction satis-
structures are denoted as M2A*, TS1A *, M *, and TS * factorily.
B
B
2
1
(
see Figure 6).
The intermediate M2A* is 27.30 kcal·molϪ1 lower in en-
A
ergy than the reactants (isolated acetone ϩ M ). This Discussion
1
A
Ϫ1
value is fairly close to that for M (24.15 kcal·mol ). The
2
structures of M2 and M2A* are quite similar. M2 * again
A
A
Comparison of the two reaction profiles (A and B) sug-
IV
corresponds to a long-range complex in which the acetone gests that the reaction involving the Ti species should be
molecule interacts weakly with the titanium catalyst. The faster than that promoted by the Ti complex. Although
III
key interaction, which moves the substrate to a suitable po- the two reactions are characterized by similar exothermicity
sition for the subsequent hydride transfer, is the hydrogen values, the activation barrier of the rate-determining step in
2980
 2003 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.eurjoc.org
Eur. J. Org. Chem. 2003, 2972Ϫ2984

Bis(oxazoline)titanium Complexes as Chiral Catalysts
FULL PAPER
the TiIII case (20.82 kcal·molϪ1), is much higher than the
IV
two barriers found along the Ti reaction pathway (4.45
Ϫ1
and 6.03 kcal·mol , respectively). This evidence, and the
data obtained for the reactions reported in Scheme 4, which
are assumed to produce the two possible active species, may
help in the proposal of a reasonable mechanistic scenario
for the whole catalytic process.
The free energy values and the corresponding equilibrium
constants reported in Scheme 4 indicate that the equilib-
rium corresponding to reaction (1) is moved to the left, so
IV
only a few [L] Ti (F)H molecules are available to catalyze
2
the reaction. This result is in agreement with the experimen-
tally obtained NMR spectroscopic data: as previously
pointed out, we were not able to identify any new species
generated by the reaction between the silyl hydride and the
(
BOX)titanium complex from the NMR spectra.
Similar computational evidence was found for the Ti -
Scheme 5
III
based catalyst. The formation of this catalytic species re-
quires two consecutive reactions: reaction (2), which gives
III
the [L] TI F complex, and reaction (3), in which the fluor-
2
to accommodate the more cumbersome phenyl group in the
less crowded molecular region, as indicated in Figure 7.
This particular arrangement should reduce the steric hin-
drance between the titanium catalyst and the benzene ring.
At the same time it allows the formation of a stabilizing
interaction (hydrogen bond) between the oxygen atom of
ine atom is replaced by the hydrogen atom to give the active
complex [L] Ti H. Even if reaction (2) is thermo-
dynamically favored, however, reaction (3) is highly endo-
thermic with a very small equilibrium constant, so the for-
mation of the [L] Ti H catalyst is highly disfavored. The
low concentrations of both catalytic species in the reaction
medium should favor the reduction pathway characterized
by the lowest activation barriers: that involving the
L] Ti (F)H complex. Since this is an easy process (low
III
2
III
2
one BOX ligand and one hydrogen atom of the carbonyl
[33]
substrate.
In our model system (in particular that based
on acetone) the formation of this hydrogen bond is evident.
It involves one methyl hydrogen atom of acetone and one
oxygen atom of the BOX ligand (as indicated in Figures 6
and 7). It is worth pointing out that, even in the case of the
simple model system used here to emulate the (BOX)titan-
ium complex, the expected absolute configuration of the al-
cohol product obtained from acetophenone is (S) in agree-
ment with the experimentally obtained observation.
IV
[
2
activation barriers and high exothermicity), it is reasonable
to believe that the TiIV active species is immediately in-
volved in the reduction as soon as it is formed and that
reaction (1) is forced to produce additional [L] Ti (F)H
complexes to restore the equilibrium constant. This hypoth-
esis points to the Ti complex as the catalytically active
species truly involved in the reduction, in agreement with
the experimentally obtained data, from which we had no
evidence for the existence of Ti species.
IV
2
IV
III
The mechanism that we propose on the basis of our com-
bined experimentally determined and computational results
corresponds to a catalytic cycle such as that schematically
represented in Scheme 5. In this cycle the catalytic species
IV
[
[
L] Ti (F)H (L ϭ deprotonated BOX ligand) forms from
2
IV
L] Ti F and (EtO) SiH (reaction 1 in Scheme 4). The re-
2
2
3
duction of the carbonyl compound (R R CϭO) affords the
1
2
IV
[
L] Ti OCHR R complex, as found in the investigation of
2 1 2
the reaction surface. This complex reacts with the (EtO)3-
SiH species to provide the silyl alcohol (EtO) SiOCHR R
2
3
1
IV
(
final product) and the [L] Ti (F)H complex, which re-
2
stores the catalytic species.
As pointed out in the previous discussion, the transition
state TS1A describes the reduction process (i.e., the addition
of the hydride ion to the carbonyl system). It is interesting
A
to analyze the structure of TS1 , since it can explain the
experimentally observed stereoinduction in this reaction.
When, instead of the formaldehyde or acetone model sub-
strate, the acetophenone molecule approaches the active
complex M2 , it is much easier for the substrate molecule tophenone substrate in the TS
Figure 7. Schematic representation of the arrangement of the ace-
A
A
1
transition state (model reaction A)
Eur. J. Org. Chem. 2003, 2972Ϫ2984
www.eurjoc.org
 2003 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
2981

M. Bandini, F. Bernardi, A. Bottoni, P. G. Cozzi, G. P. Miscione, A. Umani-Ronchi
FULL PAPER
Conclusions
Ϫ78 °C for 10 min, and then warmed to 0 °C. TiF
4
(9.3 mg,
0.075 mmol), stored under nitrogen in sealed vials, was then added,
We have carried out a combined experimental and theor- and the resulting mixture was rapidly stirred at room temperature
until the formation of an orange solution. The resulting solution
was transferred by syringe into a flame-dried NMR tube, stored in
a dried Schlenk flask. The tube was closed with a stopcock and
etical investigation of a new catalytic system based on bis-
(oxazoline) (BOX) complexes of titanium, which is able to
reduce aromatic ketones with good enantiomeric excesses
and satisfactory yields. We have demonstrated that these
BOX complexes can effectively replace the more expensive
stored under nitrogen until the spectra had been recorded. Diag-
1
nostic signals for the titanium complex: H NMR ([D
8
]THF,
2
00 MHz): δ ϭ 5.87 (d, J ϭ 8.4 Hz, 2 H), 5.41 (d, J ϭ 8.4 Hz, 2
chiral C metallocene species.
13
2
H), 4.59 (s, 1 H) ppm. C NMR ([D
8
]THF, 50 MHz): δ ϭ 30.48,
The experimentally acquired evidence and the compu-
tational results, obtained at the DFT level, have provided
useful information on the nature of the active catalytic spec-
ies and on the mechanism of the reaction. The DFT com-
putations demonstrated the existence of two possible reac-
5
8.61, 69.43, 85.34, 137.48, 139.93, 168.40 ppm.
Computational Details: All the DFT computations were performed
with the Gaussian 98[ series of program, with use of the non-
local hybrid Becke’s three-parameter exchange functional denoted
as B3LYP. [ It has been shown elsewhere that this functional is
able to provide reliable results for organotitanium complexes. These
results agree very well both with experimentally determined data
34]
35]
IV
tion paths: one originating from a Ti catalytic species
III
(
(
model reaction A) and the other from a Ti complex
model reaction B). Although both reaction channels are
and with the results obtained at higher levels of theory, such as the
thermodynamically favored (they are highly exothermic), MP2 and CASPT2 approaches.[ In all cases the geometries of the
36]
the former is characterized by much lower activation ener- various critical points on the reaction surface were fully optimized
IV
gies than the latter. This result points to the (BOX)Ti
complex as the true active catalyst involved in the reaction
and agrees with the experimentally acquired evidence
by the gradient method available in Gaussian 98. A heterogeneous
basis set was used. The DZVP basis set,[ which is a Local Spin
Density (LSD)-optimized basis set of double-zeta quality in the
valence shell plus polarization functions, was used for the titanium
atom, the hydrogen atom involved in the reduction process, and the
37]
(
NMR spectra), which seems to rule out the presence of
III
Ti species. The analysis of the structure of the transition
state corresponding to the reduction process [i.e., the ad-
dition of the hydride ion to the carbonyl system (TS1A)],
also provides an interesting insight into the enantioselectiv-
ity that characterizes this reaction.
[
38]
atoms of the formaldehyde molecule. The 3-21G basis set
was
employed for the atoms directly bonded to the metal atom, while
[39]
the minimal STO-3G basis was used to describe all the remain-
ing atoms. A computation of the harmonic vibration frequencies
was carried out to determine the natures of the various critical
The computation of the solvent effects for the two trans- points. To check the reliability of the computational level used here
formations M2 Ǟ TS1A and M2 Ǟ TS1 does not change (basis set and B3LYP functional), we compared the computed geo-
A
B
B
metrical parameters to the experimentally derived X-ray param-
the mechanistic picture, and the pathway corresponding to
model reaction A remains highly favored. Similar con-
clusions were obtained from MP2 computations on the
DFT-optimized structures.
eters. Since the X-ray structures for the titanium complexes investi-
gated in this work have not been determined, we chose as a refer-
III
ence compound a similar Ti complex (represented in Figure 8),
the structure of which is available in the literature.[ Inspection of
40]
the results reported in Figure 8 shows that the computed structure
agrees fairly well with the experimentally determined one. This sug-
gests that our computational approach is acceptable for investi-
Experimental Section
gation of this class of metal complexes. In addition to total energy
General Procedure for the Asymmetric Hydrosilylation of Ketones
by Use of [Ti(BOX)
sired BOX ligand (1Ϫ5, 0.16 mmol) was dissolved in degassed THF
2 mL) in a flame-dried flask, which was cooled to Ϫ78 °C. A solu-
2 2
F ] Complexes Prepared with nBuLi: The de-
(
tion of nBuLi in hexane (1.6 , 0.08 mmol) was added, the reaction
mixture was stirred for 10 min, and the pale yellow solution was
then warmed to 0 °C. Solid TiF (9.8 mg, 0.08 mmol) was added,
4
and the reaction mixture was vigorously stirred. The solution
turned orange and was stirred at room temperature for 1 h. The
ketones (1 mmol) and triethoxysilane (WARNING: The vapors can
cause blindness!) (2 mmol) were added, and the reaction mixture
was stirred for 48Ϫ80 h. The reaction mixture was diluted with
Et
stirred at room temperature for 30 min, and the organic phase was
separated. The aqueous phase was extracted with Et O. The com-
2
O, and NaOH (2 ⁾ was then carefully added. The mixture was
2
bined organic phases were dried with sodium sulfate and then con-
centrated under reduced pressure to give an oil, which was purified
by flash chromatography (cyclohexane/Et
2
O, 8:2 to 9:1).
Figure 8. Schematic representation of the titanium complex re-
ported in ref.[ , with the values of the most important geometrical
parameters obtained at the B3LYP level; the corresponding exper-
imentally determined values are reported in parentheses (bond
40]
NMR Studies: [D ]THF was vacuum-transferred into a flame-dried
8
flask. Bis(oxazoline) 4 (70 mg, 0.15 mmol) was added, and the re-
sulting solution was cooled to Ϫ78 °C. nBuLi (2.5 , 60 µL,
˚
0.15 mmol) was added, the pale yellow solution was stirred at
lengths are in A and angles in °)
2982
 2003 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.eurjoc.org
Eur. J. Org. Chem. 2003, 2972Ϫ2984

Bis(oxazoline)titanium Complexes as Chiral Catalysts
FULL PAPER
A. Caselli, L. Giannini, E. Solari, C. Floriani, N. Re, A. Chiesi-
Villa, C. Rizzoli, Organometallics 1997, 16, 5457Ϫ5469.
D. P. Steinhuebel, S. J Lippard, Organometallics 1999, 18,
[
[
[
12]
13]
values, free energy values were also considered for the various criti-
cal points. Free energies were obtained from frequency calculations
within the RRHO approximation. This approximation can provide
only a rough estimate of free energy trends. This is especially true
in situations such as those examined here, in which, in the presence
of weakly interacting fragments, low frequencies and thus anomal-
ous contributions to free energy can be expected. For rough evalu-
ation of the effect of the solvent environment, we carried out single-
point computations on the gas-phase optimized structures for
some critical points, by use of the continuous model approach
1
09Ϫ111.
14] [14a]
G. D. Whitener, J. Hagadorn, J. Arnold, J. Chem. Soc.,
[
14b]
Dalton Trans. 1999, 1249Ϫ1255.
J. R. Hagadorn, J. Ar-
nold, Organometallics 1998, 17, 1355Ϫ1368. [
Shmulinson, C. Averbuj, A. Lisovskii, F. T. Edelmann, M.
14c]
V. Volkis, M.
Eisen, Organometallics 1998, 17, 3155Ϫ3157.
[15] [15a]
G. Rodriguez, G. C. Bazan, J. Am. Chem. Soc. 1995, 117,
[15b]
1
0155Ϫ10156.
R. Kempe, P. Arndt, Inorg. Chem. 1996, 35,
[
41]
[42]
[15c]
COSMO as implemented in the Turbomole package.
2644Ϫ2649.
B. Qian, W. J. Scanlon IV, M. R. Smith III,
[
15d]
D. H. Motry, Organometallics 1999, 18, 1693Ϫ1698.
A. J.
Ashe, III, X. Fang, J. W. Kampf, Organometallics 1999, 18,
Acknowledgments
[15e]
1
363Ϫ1365.
S. J. Brown, X. Gao, D. G. Harrison, L. Koch,
R. E. v. H. Spence, G. P. A. Yap, Organometallics 1998, 17,
We would like to thank the C.N.R. and M.U.R.S.T. (Progetto
Nazionale ‘‘Stereoselezione in Sintesi Organica: Metodologie ed
Applicazioni’’) and Bologna University (funds for selected research
topics) for the financial support of these researches.
445Ϫ5447. [
Int. Ed. Engl. 1997, 36, 2450Ϫ2452.
Tedenborg, D. A. Alonso, P. G. Anderson, Organometallics
999, 18, 1281Ϫ1286.
15f]
G. G. Lavoie, R. G. Bergman, Angew. Chem.
5
[15g]
S. K. Bertilsson, L.
1
[
16]
For a recent review see: P. Braunstein, F. Naud, Angew. Chem.
Int. Engl. 2001, 40, 680Ϫ699, and ref. therein.
A. Pfaltz, Acc. Chem. Res. 1993, 26, 239Ϫ345.
[
1] [1a]
H. H. Brintzinger, D. Fisher, R. Mülaupt, B. Rieger, R. M.
Waymouth, Angew. Chem. Int. Ed. Engl. 1995, 34, 1143Ϫ1170.
[
[
17]
18]
[1b]
D. A. Evans, J. A. Murry, P. V. von Matt, R. D. Norcross, S.
A. H. Hoveyda, J. P. Morken, in Metallocene (Eds.: A.
[
18a]
J. Miller, Angew. Chem. Int. Ed. Engl. 1995, 34, 798Ϫ800.
Togni, R. L. Halterman), Wiley-VCH, Weinheim, 1998, p. 625,
D. A. Evans, S. J. Miller, T. Lectka, P. von Matt, J. Am. Chem.
Soc. 1999, 121, 7559Ϫ7573.
and references therein.
A. H. Hoveyda, J. Morken, Angew. Chem. Int. Ed. Engl. 1996,
[
[
2]
[
19] [19a]
3
5, 1262Ϫ1284.
J. Thorhauge, M. Johannsen, K. A. Jørgensen, Angew.
3] [3a]
[19b]
S. Collins, Y. Hong, B. A. Kuntz, J. Org. Chem. 1989, 54,
Chem. Int. Ed. 1998, 37, 2404Ϫ2406.
S. Yao, M.
[
3b]
4
154Ϫ4158.
ganometallics 1994, 13, 3411Ϫ3412.
Jensen, D. Y. Kodankov, S. M. Wang, J. Am. Chem. Soc. 1994,
16, 8404Ϫ8405.
T. Takahashi, D. Y. Kodankov, N. Suzuki, Or-
Johannsen, H. Audran, R. G. Hazell, K. A. Jørgensen, J. Am.
Chem. Soc. 1998, 120, 8599Ϫ8605.
A. K. Ghosh, P. Mathivanan, J. Cappiello, Tetrahedron: Asym-
metry 1998, 9, 1Ϫ45, and references therein.
[3c]
E. Negishi, M. D.
[
20]
1
[
4] [4a]
[21] [21a]
[21b]
M. B. Carter, B. Schiøtt, A. Guti e` rrez, S. L. Buchwald, J.
H. Imma, M. Mori, T. Nakai, Synlett 1996, 1229Ϫ1230.
R. L. Halterman, T. M. Ramsey, Z. Chen, J. Org. Chem.
Am. Chem. Soc. 1994, 116, 11667Ϫ11670. [ X. Verdaguer, U.
E. W. Lange, M. T. Reding, S. L. Buchwald, J. Am. Chem. Soc.
4b]
1994, 59, 2642Ϫ2644.
4c]
1
996, 118, 6784Ϫ6785. [ X. Verdaguer, U. E. Lange, S. L.
[22]
Part of these results has been already presented: M. Bandini,
P. G. Cozzi, L. Negro, A. Umani-Ronchi, Chem. Commun.
[4d]
Buchwald, Angew. Chem. Int. Ed. 1998, 37, 1103Ϫ1107.
M.
C. Hansen, S. L. Buchwald, Org. Lett. 2000, 2, 713Ϫ715. [ J.
Yun, S. L. Buchwald, J. Am. Chem. Soc. 1999, 121, 5640Ϫ5644.
The titanocene methodology was patented by: S. L. Buchwald,
R. B. Grossman, A. Guti e` rrez, US 5227538 assigned to M. I.
T., 1993.
4e]
1999, 39Ϫ40.
[23]
[24]
D. A. Evans, G. S. Peterson, J. S. Johnson, D. M. Barnes, K.
R. Campos, K. A. Woerpel, J. Org. Chem. 1999, 64, 678Ϫ678.
M. Nakamura, A. Hirai, M. Sogi, E. Nakamura, J. Am. Chem.
Soc. 1998, 120, 5846Ϫ5847.
The reactivity of silanes has been shown to decrease in the
3 2 2 2
order (EtO) SiH Ͼ (EtO) Si(Me)H Ͼ PMHS Ͼ Ph SiH .
However, the reagent has potential hazards, see: K. Barr, S. C.
Berk, S. L. Buchwald, J. Org. Chem. 1994, 59, 4323Ϫ4326. As
the (EtO) SiH is reported to be a toxic reagent, other hydride
3
[
5]
[
[
6]
7]
J. Yun, S. L. Buchwald, J. Org. Chem. 2000, 65, 767Ϫ774.
See for example:
[
25]
[7a]
G. M. Diamond, R. F. Jordan, J. L. Pet-
[
7b]
ersen, J. Am. Chem. Soc. 1996, 118, 8024Ϫ8033.
Q. Yang,
M. D. Jensen, Synlett 1996, 147Ϫ148.
[
8] [8a]
R. Uhrhamer, D. G. Black, T. G. Gardner, J. D. Olsen, R.
F. Jordan, J. Am. Chem. Soc. 1993, 115, 8493Ϫ8494. [
8b]
E. B.
source were examined. We discovered that catecholborane can
be used as hydride source with BOX ligands, but that lower
enantiomeric excess was determined in all cases.
Tjaden, D. C. Swenson, R. F. Jordan, J. L. Petersen, Organome-
tallics 1995, 14, 371Ϫ386. [
8c]
A. J. Blake, J. M. McInnes, P.
Mounford, G. I. Nikonov, D. Swallow, D. J. Watkin, J. Chem.
Soc., Dalton Trans. 1999, 379Ϫ392.
[
[
26]
27]
The reaction between the deprotonated BOX ligand and TiF
can be performed by direct addition of TiF , or the TiF can
be added dissolved in CH CN.
The active catalyst, starting from the pre-catalyst [(S,S)-(EB-
4
[
9] [9a]
4
4
R. R. Schrock, R. Bauman, S. M. Reid, J. T. Goodman, R.
3
Stumpf, W. M. Davis, Organometallics 1999, 18, 3649Ϫ3670.
[
9b]
T. I. Goutntchev, T. Don Tilley, Organometallics 1999, 18,
[
9c]
THI)TiF ], was prepared by Buchwald by addition of phe-
nylsilane, pyrrolidine and MeOH sequentially, and heating at
0 °C until the color of the solution changed from yellow to
5
1
661Ϫ5667.
A. D. Horton, J. de With, Organometallics
2
997, 16, 5424Ϫ5436. [
9d]
N. A. H. Male, M. Thortnon-Pett,
6
M. Bochman, J. Chem. Soc., Dalton Trans. 1997, 2487Ϫ2494.
[9e]
green. No explanation for this procedure and no investigations
on the species prepared in such a way has been presented. The
previous procedure suggests that the activation step depends
on the way nBuLi is added to the reaction mixture (see the
experimental procedure described in ref.[
R. Baumann, W. M. Davis, R. R. Schrock, J. Am. Chem.
Soc. 1997, 119, 3830Ϫ3831. [ J. D. Scollard, D. H. McCon-
9f]
[9g]
ville, S. J. Rettig, Organometallics 1997, 16, 1810Ϫ1812.
J.
D. Scollard, D. H. McConville, J. Am. Chem. Soc. 1996, 118,
4a]
[
9h]
1
008Ϫ1009.
M. Schubart, L. O’Dwyer, L. H. Gade, W. S.
[
28]
29]
Li, M. McPartlin, Inorg. Chem. 1994, 33, 3893Ϫ3898.
H. W. Görlitzer, M. Spiegler, R. Anwander, J. Chem. Soc., Dal-
ton Trans. 1999, 4287Ϫ4288.
[
10]
11]
H. Brand, J. Arnold, J. Am. Chem. Soc. 1992, 114, 2266Ϫ2267.
H. Brand, J. Arnold, Organometallics 1993, 12, 3655Ϫ3665.
L. Giannini, C. Caselli, E. Solari, C. Floriani, A. Chiesi-Villa,
C. Rizzoli, M. Re, A. Sgamellotti, J. Am. Chem. Soc. 1997,
[
In a series of [Zn(BOX)X] complexes the signal of the CH
2
[
protons, which appears at δ ഠ 3.3 ppm in the free bis(oxazol-
ines), were found to be shifted to δ ഠ 4.7 ppm (CH form). See:
R. P. Singh, Bull. Soc. Chim. Fr. 1997, 134, 765Ϫ768.
119, 9709Ϫ9719.
Eur. J. Org. Chem. 2003, 2972Ϫ2984
www.eurjoc.org
 2003 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
2983

M. Bandini, F. Bernardi, A. Bottoni, P. G. Cozzi, G. P. Miscione, A. Umani-Ronchi
FULL PAPER
[
30]
P. G. Cozzi, C. Floriani, A. Chiesi-Villa, C. Rizzoli, Inorg.
P. M. W. Gill, B. Johnson, W. Chen, M. W. Wong, J. L. Andres,
M. Head-Gordon, E. S. Replogle, J. A Pople, Gaussian 98, re-
vision A.6, Gaussian, Inc., Pittsburgh, PA, 1998.
Chem. 1995, 34, 2921Ϫ2930.
[
31]
For the structure of a titanium() atom, surrounded by non-
Cp ligands, see: J. B. Love, H. S. C. Clark, F. G. N. Cloke, J.
C. Green, P. B. Hitchcock, J. Am. Chem. Soc. 1999, 121,
[
[
[
35] [35a]
[35b]
A. D. Becke, J. Chem. Phys. 1993, 98, 5648Ϫ5652.
Lee, W. Yang, R. G. Parr, Phys. Rev. 1988, B37, 785.
C.
36]
37]
F. Bernardi, A. Bottoni, G. P. Miscione, Organometallics 1998,
6
843Ϫ6849.
17. 16.
[
[
32]
33]
C. Gonzalez, H. B. Schlegel, J. Chem. Phys. 1989, 90,
154Ϫ2161.
N. Godbout, D. R. Salahub, J. Andzelm, E. Wimmer, Can. J.
Chem. 1992, 70, 560Ϫ571. UniChem. DGauss, Version 2.3.1,
Cray Research, Inc., Seattle, 1994.
2
Recent evidence for this kind of hydrogen bond has recently
been provided in theoretical and experimental investigations:
[38]
[39]
J. S. Binkley, J. A. Pople, W. J. Hehre, J. Am. Chem. Soc. 1980,
[
33a]
N. Bernardi, A. Bottoni, S. Casolari, E. Tagliavini, J. Org.
1
02, 939Ϫ947. W. J. Pietro, M. M. Francl, W. J. Hehre, D. J.
Chem. 2000, 65, 4783. [
33b]
E. J. Corey, T. W. Lee, Chem. Com-
Defrees, J. A. Pople, J. S. Binkey, J. Am. Chem. Soc. 1982,
mun. 2001, 1321Ϫ1329.
104, 5039Ϫ5048.
[
34]
M. J. Frisch, G. W. Trucks, H. B. Schlegel, G. E. Scuseria, M.
A. Robb, J. R. Cheeseman, V. G. Zakrzewski, J. A. Jr.
Montgomery, R. E. Stratmann, J. C. Burant, S. Dapprich, J. M.
Millam, A. D. Daniels, K. N. Kudin, M. C. Strain, O. Farkas, J.
Tomasi, V. Barone, M. Cossi, R. Cammi, B. Mennucci, C. Po-
melli, C. Adamo, S. Clifford, J. Ochterski, G. A. Petersson, P.
Y. Ayala, Q. Cui, K. Morokuma, D. K. Malick, A. D. Rabuck,
K. Raghavachari, J. B. Foresman, J. Cioslowski, J. V. Ortiz, B.
B. Stefanov, G. Liu, A. Liashenko, P. Piskorz, I. Komaromi, R.
Gomperts, R. L. Martin, D. J. Fox, T. Keith, M. A. Al-Laham,
C. Y. Peng, A. Nanayakkara, C. Gonzalez, M. Challacombe,
W. J. Hehre, R. F. Stewart, J. A. Pople, J. Chem. Phys. 1969,
51, 2657. J. B. Collins, P. v. R. Schleyer, J. S. Binkley, J. A Pople,
J. Chem. Phys. 1976. 64, 5142.
[
[
[
40]
41]
42]
P. H. M. Budzelaar, A. B. van Oort, A. G. Orpen, Eur. J. Inorg.
Chem. 1998, 1485Ϫ1494.
A. Klamnt, J. Phys. Chem. 1995, 99, 2224. A. Klamnt, J. Phys.
Chem. 1996, 100, 3349.
Turbomol, version 5.6, Institut für Physikalische Chemie und
Elektrochemie Lehrstuhl für Theoretische Chemie, Universität
Karlsruhe, Kaiserstr. 12, 76128 Karlsruhe, Germany, 2002.
Received December 2, 2002
2984
 2003 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.eurjoc.org
Eur. J. Org. Chem. 2003, 2972Ϫ2984