Weakening C-O bonds: Ti(III), a new reagent for alcohol deoxygenation and carbonyl coupling olefination

Article abstract of DOI:10.1021/ja906083c

Investigations detailed herein, including density functional theory (DFT) calculations, demonstrate that the formation of either alkoxy- or hydroxy-Ti(III) complexes considerably decreases the energy of activation for C-O bond homolysis. As a consequence of this observation, we described two new synthetic applications of Nugent's reagent in organic chemistry. The first of these applications is an one-step methodology for deoxygenation-reduction of alcohols, including benzylic and allylic alcohols and 1,2-dihydroxy compounds. Additionally, we have also proved that Ti(III) is capable of mediating carbonyl coupling-olefination. In this sense, and despite the fact that for over 35 years it has been widely accepted that either Ti(II) or Ti(0) was the active species in the reductive process of the McMurry reaction, the mechanistic evidence presented proves the involvement of Ti(III) pinacolates in the deoxygenation step of the herein described Nugent's reagent-mediated McMurry olefination. This observation sheds some light on probably one of the mechanistically more complex transformations in organic chemistry. Finally, we have also proved that both of these processes can be performed catalytically in Cp ₂TiCl₂ by using trimethylsilyl chloride (TMSCl) as the final oxygen trap.

Full text of DOI:10.1021/ja906083c

Published on Web 12/11/2009
Weakening C-O Bonds: Ti(III), a New Reagent for Alcohol
Deoxygenation and Carbonyl Coupling Olefination
Horacio R. Die´guez,† Armando Lo´pez,† Victoriano Domingo,† Jesu´s F. Arteaga,‡
Jose´ A. Dobado,† M. Mar Herrador,† Jose´ F. Qu´ılez del Moral,† and
Alejandro F. Barrero*,†
Department of Organic Chemistry, Institute of Biotechnology, UniVersity of Granada, AVda.
FuentenueVa, 18071 Granada, Spain, and Department of Chemical Engineering, Physical
Chemistry and Organic Chemistry, Faculty of Experimental Sciences, UniVersity of HuelVa,
Campus el Carmen, 21071 HuelVa, Spain
Received August 4, 2009; E-mail: afbarre@ugr.es
Abstract: Investigations detailed herein, including density functional theory (DFT) calculations, demonstrate
that the formation of either alkoxy- or hydroxy-Ti(III) complexes considerably decreases the energy of
activation for C-O bond homolysis. As a consequence of this observation, we described two new synthetic
applications of Nugent’s reagent in organic chemistry. The first of these applications is an one-step
methodology for deoxygenation-reduction of alcohols, including benzylic and allylic alcohols and 1,2-
dihydroxy compounds. Additionally, we have also proved that Ti(III) is capable of mediating carbonyl
coupling-olefination. In this sense, and despite the fact that for over 35 years it has been widely accepted
that either Ti(II) or Ti(0) was the active species in the reductive process of the McMurry reaction, the
mechanistic evidence presented proves the involvement of Ti(III) pinacolates in the deoxygenation step of
the herein described Nugent’s reagent-mediated McMurry olefination. This observation sheds some light
on probably one of the mechanistically more complex transformations in organic chemistry. Finally, we
have also proved that both of these processes can be performed catalytically in Cp2TiCl2 by using
trimethylsilyl chloride (TMSCl) as the final oxygen trap.
Scheme 1. Proposed Mechanism for Ti(III)-Promoted C-O Bond
Homolysis
Introduction
It is well-known that Cp2TiCl (Nugent’s reagent) is a mild
reductant, useful through monoelectronic transfer and widely
employed in organic synthesis.1 Its applications arise from the
interaction of titanocene dichloride, acting as a Lewis acid, with
the heteroatoms of different functional groups, which behave
as Lewis bases. This interaction leads to a balanced complex
that may evolve into different products. Recent research
revealing mechanistic details about this interaction has been
reported.2 In this sense, evidence showing that H2O and
methanol can act, via the corresponding Ti(III) complexes, as
an efficient hydrogen-atom donor has been published.2c,d
With these considerations in mind, it would be reasonable to
postulate that the interaction of an alcohol and Cp2TiCl would
produce the alcohol complex I. The formation of this species,
or of the titanocene alkoxide derivative II, could weaken the
corresponding C-O bond allowing under suitable conditions
and by means of monoelectronic transfer their homolytic
cleavage to give rise to the C-centered radical III, which would
finally evolve to the corresponding products (Scheme 1).
To validate this approach, we chose two synthetic transforma-
tions including C-O cleavages as key processes, namely,
deoxygenation-reduction of alcohols and carbonyl coupling-
olefinations.
† University of Granada.
‡ University of Huelva.
(1) The single electron transfer bis(cyclopentadienyl)-titanium(III)chloride,
Nugent’s reagent, can be generated in situ by stirring commercial
Cp2TiCl2 with dust Mn or Zn. For pioneering reports on the use of
this reagent, see: (a) RajanBabu, T. V.; Nugent, W. A. J. Am. Chem.
Soc. 1994, 116, 986-997 and references cited therein. For pertinent
reviews on the use of this reagent, see: (b) Gansa¨uer, A.; Bluhm, H.
Chem. ReV. 2000, 100, 2771–2788. (c) Gansa¨uer, A.; Pierobon, M. In
Radicals in Organic Synthesis; Renaud, P., Sibi, M. P., Eds.; Wiley-
VCH: Weinheim, Germany, 2001; Vol. 2, pp 207-220. (d) Gansa¨uer,
A.; Lauterbach, T.; Narayan, S. Angew. Chem., Int. Ed. 2003, 42,
5556–5573. (e) Barrero, A. F.; Quilez del Moral, J. F.; Sanchez, E. M.;
Arteaga, J. F. Eur. J. Org. Chem. 2006, 1627–1641. (f) Cuerva, J. M.;
Justicia, J.; Oller-Lo´pez, J. L.; Bazdi, B.; Oltra, J. E. Mini-ReV. Org.
Chem. 2006, 3, 23–35.
(2) (a) Enemrke, R. J.; Larsen, J.; Skrydstrup, T.; Daasbjerg, K. J. Am.
Chem. Soc. 2004, 126, 7853–7864. (b) Gansa¨uer, A.; Barchuk, A.;
Keller, F.; Schmitt, M.; Grimme, S.; Gerenkamp, M.; Mu¨ck-Licht-
enfeld, C.; Daasbjerg, K.; Svith, H. J. Am. Chem. Soc. 2007, 129,
1359–1371. (c) Cuerva, J. M.; Campan˜a, A. G.; Justicia, J.; Rosales,
A.; Oller-Lo´pez, J. L.; Robles, R.; Cardenas, D. J.; Bun˜uel, E.; Oltra,
J. E. Angew. Chem., Int. Ed. 2006, 45, 5522–5526. (d) Jin, J.;
Newcomb, M. J. Org. Chem. 2008, 73, 7901-7905 and references
cited therein.
9
254 J. AM. CHEM. SOC. 2010, 132, 254–259
10.1021/ja906083c  2010 American Chemical Society

Ti(III)-Mediated Weakening of C-O Bonds
A R T I C L E S
Scheme 2. Proposed Mechanism for Ti(III)-Mediated Alcohol
Alcohol deoxygenation constitutes a powerful synthetic tool
especially used in complex natural product synthesis.3 Most of
the known synthetic procedures take place via a number of steps,
the Barton-McCombie methodology being the most commonly
used, mainly for secondary alcohols, due to its compatibility
with different functional groups.4 Few procedures involving one-
step deoxygenations have been described,5 which in our opinion
makes necessary further efforts in this subject.
Deoxygenation-Reduction
Table 1. Calculateda ∆Gq and ∆Grxn Values for C-O Bond
Homolytic Cleavage of Different Alcohols
On the other hand, the reductive coupling of carbonyls into
olefins by use of low-valent titanium species (LVT), known as
the McMurry reaction, has been extensively used in organic
synthesis, and both inter- and intramolecular couplings have
been described to proceed with remarkable efficiency.6 This kind
of reaction is usually carried out in two consecutive steps,
namely, reduction of TiCl4 or TiCl3 followed by addition of
the carbonyl compound. In this regard, a number of reducing
agents such as Li, Na, K, Mg, Zn, KC8, Zn(Cu), LiAlH4, and
others were used, in an attempt to overcome the reproducibility
problems usually associated with this reaction. To this end,
different improved protocols have been reported.7 From a
mechanistic point of view, there are three main features to
consider: formation of the LVT species, coupling reaction, and
finally, deoxygenation of the intermediates leading to the olefin.
In the coupling step, the involvement of acyl radical and/or
carbenoid intermediates in the metallopinacolate formation is
accepted, although their structure is claimed to be influenced
by the nature of the carbonyl group, the titanium species, and
the reducing agent.6g In this sense, even more uncertainties exist
about the actual mechanism of the deoxygenation of these
metallopinacolate intermediates, although it has been widely
a Calculated via UM05/Ahlrichs-VDZ.
accepted that species with low valence states of Ti, either Ti(II)
or Ti(0), were required in this reductive process.
(3) (a) Zard, S. Z. Xanthates and Related Derivatives as Radical Precursors.
In Radicals in Organic Synthesis, Vol. 1; Renaud, P., Sibi, M. P.,
Eds.; Wiley-VCH: Weinheim, Germany, 2001; pp 90-108. (b)
McCombie, W. S. In ComprehensiVe Organic Synthesis, Vol. 8; Trost,
B. M., Fleming, I., Eds.; Pergamon Press: Oxford, U.K., 1991; pp
811-833.
Results and Discussion
A. Deoxygenation-Reduction of Alcohols. i. Mechanistic
Proposal and DFT Calculations. The mechanistic proposal for
deoxygenation-reduction of alcohols is depicted in Scheme 2.
Thus, the C-centered radical III, originating from homolysis
of the corresponding C-O bond, could evolve to the corre-
sponding hydrocarbon after being trapped by another molecule
of Cp2TiCl, generating alkyltitanium V, which would then be
protonated easily to afford the alkane. In this regard, the direct
reduction of the radical III by species such as I, IV, or the
solvent should not be ruled out.
To verify this hypothesis, density functional theory (DFT;
UM05/Ahlrichs-VDZ)8 calculations were carried out. Thus, the
energy barriers (∆Grxn and ∆Gq) for homolytic cleavage of the
C-O bond in the alcohol complex I were calculated for different
type of alcohols (Table 1 and Figure 1).
(4) (a) Zard, S. Z. Angew. Chem., Int. Ed. 1997, 36, 672–685. (b) Barton,
D. H. R.; Jang, D. O.; Jaszberenyi, J. C.; Joseph, C. J. Org. Chem.
1993, 58, 6838–6842. (c) Barton, D. H. R.; Jang, D. O.; Jaszberenyi,
J. C.; Joseph, C. Synlett 1991, 435–438. (d) Barton, D. H. R.;
Motherwell, W. B.; Stange, A. Synthesis 1981, 743–745.
(5) (a) Corey, E. J.; Achiwa, K. J. Org. Chem. 1969, 34, 3667–3668. (b)
McMurry, J. E.; Silvestri, M. G.; Fleming, M. P.; Hoz, T.; Grayston,
M. W. J. Org. Chem. 1978, 43, 3249–3254. (c) Ledon, H.; Tkatchenko,
I.; Young, D. Tetrahedron Lett. 1979, 20, 173–176. (d) Sato, F.;
Tomuro, Y.; Ishikawa, H.; Oikawa, T.; Sato, M. Chem. Lett. 1980,
103–106. (e) Lee, J. T.; Alper, H. Tetrahedron Lett. 1990, 31, 4101–
4104. (f) Crevier, T. J.; Mayer, J. M. J. Am. Chem. Soc. 1997, 119,
8485–8491. (g) Zhang, L.; Koreeda, M. J. Am. Chem. Soc. 2004, 126,
13190–13191. (h) Spiegel, D. A.; Wiberg, K. B.; Schacherer, L. N.;
Medeiros, M. R.; Wood, J. L. J. Am. Chem. Soc. 2005, 127, 12513–
12515.
(6) (a) McMurry, J. E. Chem. ReV. 1989, 89, 1513–1524. (b) Robert-
sonG. M. In ComprehensiVe Organic Synthesis; Trost, B. M.,
Fleming, I., Pattenden, G., Eds.; Pergamon: Oxford, U.K., 1991; Vol.
3, p 563. (c) DushinR. In ComprehensiVe Organometallic Chemistry
II; Hegedus, L. S., Ed., Pergamon: Oxford, U.K., 1995; Vol. 3, p 1071.
(d) Lectka, T. In ActiVe Metals: Preparation, Characterization,
Applications; Fu¨rstner, A., Ed.; Wiley-VCH: Weinheim, Germany,
1995; p 85. (e) Fu¨rstner, A.; Bogdanovic, B. Angew. Chem., Int.
Ed. Engl. 1996, 35, 2442–2449. (f) Wirth, T. Angew. Chem., Int.
Ed. Engl. 1996, 35, 61–63. (g) Ephritikhine, M. Chem. Commun.
1998, 2549–2554.
Although the calculated values were endothermic for all
compounds studied (1-6), the allylic and benzylic alcohols
showed, as expected, lower activation and reaction energies
(8) (a) Frisch, M. J. Gaussian 03, ReVision E.01; Gaussian, Inc.,
Wallingford, CT, 2004. For complete reference, see Supporting
Information. (b) Zhao, Y.; Schultz, N. E.; Truhlar, D. G. J. Chem.
Phys. 2005, 123, 161103. (c) Schaefer, A.; Horn, H.; Ahlrichs, R.
J. Chem. Phys. 1992, 97, 2571–2577. (d) Feller, D. J. Comput. Chem.
1996, 17, 1571–1586. (e) Schuchardt, K. L.; Didier, B. T.; Elsethagen,
T.; Sun, L.; Gurumoorthi, V.; Chase, J.; Li, J.; Windus, T. L. J. Chem.
Inf. Model. 2007, 47, 1045–1052.
(7) (a) McMurry, J. E.; Letcka, T.; Rico, J. G. J. Org. Chem. 1989, 54,
3748–3749. (b) Fu¨rtsner, A.; Hupperts, A.; Ptock, A.; Janssen, E. J.
Org. Chem. 1994, 59, 5215–5229.
9
J. AM. CHEM. SOC. VOL. 132, NO. 1, 2010 255

A R T I C L E S
Die´guez et al.
Table 2. Ti(III)-Mediated Deoxygenation-Reduction of Alcohols,,
Figure 1. Energy diagram for C-O bond homolytic cleavage in allylic
alcohol 1 (activation energy ∆Gq and reaction energy ∆Grxn).
(Table 1). In any case, when these data were compared with
those calculated theoretical and experimentally for the uncom-
plexed homolytic C-O bond dissociation,9 significant higher
values were found for the uncomplexed species, which clearly
show that this homolytic cleavage is strongly affected by
coordination to Ti(III).
In light of these observations, we surmised that the necessary
energy for this homolysis could be obtained by simply heating
organic solutions of the corresponding alcohol-Cp2TiCl com-
plexes. Thus, a solution of benzyl alcohol 4 in THF (0.2 M)
was treated with 2.1 equiv of TiCp2Cl2 and 8.0 equiv of Mn in
THF. After the mixture was refluxed for 45 min, the reaction
was proven to be completed, and gratifyingly, only toluene was
obtained in an excellent yield of 93%.
Variations on the quantities of Nugent’s reagent were carried
out in order to gain further insight into the course of the reaction.
Thus, when only 1.1 equiv of Cp2TiCl2 was used, the reaction
stopped when approximately half the starting material 4
remained unaltered, toluene being obtained in a 47% yield. This
fact suggests that the C-centered radical (III) is efficiently
trapped by a second equivalent of Ti(III) leading to the formation
of a benzyl-titanium intermediate (V). Consequently, 2.0 equiv
of Ti(III) is required to drive the reaction to completion. With
respect to the reducing agent, its quantity can be lowered up to
1.5 equiv, and the transformation still remains efficient (92%).
ii. Scope of the Reaction. We devoted our efforts to inves-
tigate the scope of this transformation (Table 2). Ti(III) (2.0
equiv), 1.5 equiv of reducing agent, and tetrahydrofuran (THF)
as solvent were used, except for entry 19, where toluene was
employed. Thus, different benzylic alcohols were transformed
efficiently into the corresponding deoxygenated-reduced com-
pounds (Table 2, entries 1-5). Allylic alcohols reacted in a
similar way producing the expected deoxygenated-reduced
products (Table 2, entries 6-12). Thus, farnesol 19 gave rise
mainly to hydrocarbon 20 (Table 2, entry 7). In this case, the
absence of cyclization products resulting from an intermediate
allylic radical also supports the fast reduction of this species
(Scheme 2). This result is in good accordance with previous
results by some of us, which reported the high tendency of allylic
radicals to evolve to allyltitanium species in presence of excess
a 2.0 equiv of Ti(III) and 1.5 equiv of reducing agent per mole of
starting material were used; THF was used as solvent. b 0.3 equiv of
Ti(III), 4.0 equiv of TMSCl equiv, and 8 equiv of reducing agent per
mole of starting material were used; THF was used as solvent. c Toluene
was used as solvent.
Cp2TiCl.10 The conversion of 4-hydroxy-3-methylbut-2-enyl
benzoate 22 to isopentenyl benzoate 23 (Table 2, entry 10)
deserved to be emphasized, since it mimicks the function of
(10) (a) Seemann, M.; Bui, B. T. S.; Wolff, M.; Tritsch, D.; Campos, N.;
Boronat, A.; Marquet, A.; Rohmer, M. Angew. Chem., Int. Ed. 2002,
41, 4337–4339. (b) Rohmer, M. Nat. Prod. Rep. 1999, 16, 565–574.
(c) Eisenreich, W.; Bacher, A.; Arigoni, D.; Rohdich, F. Cell. Mol.
Life Sci. 2004, 61, 1401–1426. (d) Xiao, J.; Chu, L.; Sanakis, Y.; Liu,
P. J. Am. Chem. Soc. 2009, 131, 9931–9933.
(9) Complementary calculations on nonmediated homolytic C-O bond
cleavages gave values larger that 56 kcal mol-1 in any case. The
reported experimental value for MeOH is 102 kcal mol-1: Kerr, J. A.
Chem. ReV. 1966, 66, 465-500. For supplementary data, see Sup-
porting Information.
(11) (a) Barrero, A. F.; Herrador, M. M.; Qu´ılez del Moral, J. F.; Arteaga,
P.; Arteaga, J. F.; Die´guez, H. R.; Sa´nchez, E. M. J. Org. Chem. 2007,
72, 2988–2995. (b) Barrero, A. F.; Herrador, M. M.; Qu´ılez del Moral,
J. F.; Arteaga, P.; Arteaga, J. F.; Piedra, M.; Sa´nchez, E. M. Org.
Lett. 2005, 7, 2301–2304.
9
256 J. AM. CHEM. SOC. VOL. 132, NO. 1, 2010

Ti(III)-Mediated Weakening of C-O Bonds
A R T I C L E S
Scheme 3. Proposed Mechanism for Ti(III)-Promoted
Deoxygenation of 1,2-Diols
Scheme 4. Proposed Mechanism for Ti(III)/Mn-Promoted Carbonyl
Olefination
the IspH enzyme in the natural biosynthesis of C5 units of the
terpenoids through the deoxyxylulose phosphate pathway.11
With respect to 1,2-diols, treatment of 1,2-diphenylethane-1,2-
diol 28 with 2.1 equiv of TiCp2Cl2 and 1.5 equiv of Mn led
almost quantitatively to trans-stilbene 29 (Table 2, entry 14).
Other 1,2-diols such as tetradecane-1,2-diol 32 and pinanediol
34 were easily deoxygenated to 33 and ꢀ-pinene 35, respec-
tively, in high yield (Table 2, entries 17 and 18). The
stoichiometry of the reagents and good yield seem to indicate
that the double bond would be formed via an initial deoxygen-
ation from the diol complex VI leading to the C-centered radical
VII (Scheme 3). The presence of this first radical would facilitate
the second C-O cleavage to lead to the corresponding olefin.
When nonactivated alcohols such as octadecanol 36 were used
as starting material, no reaction was observed (see ∆Grxn in
Table 1). To overcome this unreactivity, other solvents with
higher boiling point were tested. Thus, when a mixture of
Nugent’s reagent and 36 were heated in toluene at reflux, 37
and 38 (1:1) were obtained as result of deoxygenation-reduction
and dehydration processes, respectively (Table 2, entry 19).
When we focused our attention on the presence of different
functional groups in the alcohols to be deoxygenated, good
yields of the corresponding deoxygenation products were also
obtained in the presence of hydroxyls protected as methyl ethers
(Table 2, entries 3 and 4), although the acid-labile silyl ethers
did not tolerate the reaction conditions. Other hydroxyl protect-
ing groups such as acetate also decompose under the reaction
conditions. However, most robust esters such as the benzoate
or, even more efficiently, the pivalate proved to be appropriate
protecting groups in our deoxygenation protocol (Table 2, entries
10 and 11). Furthermore, alcohols containing carboxylic acids
protected as methyl esters, as well as chloro derivatives, also
deoxygenated in acceptable yields (Table 2, entries 5 and 15).
Finally, although in a low-yield, nonoptimized process, a
primary allylic alcohol was deoxygenated in the presence of a
carbonyl group (Table 2, entry 12), which in our opinion widens
promisingly the possibilities of this transformation.
with Cp2TiCl in refluxing THF.13 Gratifyingly, after this mixture
was heated for 3 h, gas chromatographic (GC) analysis of the
crude product proved the generation of toluene, which seemed
to confirm the first premise of our hypothesis.2d Encouraged
by this result, we made benzaldehyde 39 react with 2 equiv of
titanocene(III) in THF at reflux. Stilbene 29 was thus obtained
with high efficiency (93% yield) after 1 h (Table 3, entry 1).
Considering that the system Cp2TiCl2/Mn was previously
reported to mediate pinacol coupling of aldehydes and ketones,12
we focused our efforts to gain an insight into the real species
participating in the deoxygenation step.
To this end, we subjected benzaldehyde 39 to the same
experimental conditions described above but now maintained
the temperature at 25 °C. After 10 min of stirring at this
temperature, analysis of an aliquot taken from the green reaction
mixture revealed complete consumption of the starting material,
benzopinacol being the only product formed. Excess Mn was
then removed from the remaining reaction mixture, and the
filtrate was heated at reflux for 1 h to afford again stilbene 29.
It is worth noting the progressive color change, from green to
blue, during the heating. At this point, it was reasonable to
assume the involvement of a Ti(III) pinacolate in this olefination
process, according to the mechanistic proposal depicted in
Scheme 4. The excess Mn present in the reaction medium would
reduce the initially generated Ti(IV) pinacolate species12 to the
corresponding Ti(III) species. The fact that, to the best of our
knowledge, Mn is not able to reduce Ti to a valence lower than
3 at room temperature, together with the green color of the
solution, constitutes a strong hint to confirm the presence of
this Ti(III) metallopinacol.14
A final piece of evidence to denote the involvement of Ti(III)
pinacolates in the deoxygenation step was derived from the
following experience. Thus, addition of only 0.5 equiv of Zn
or Mn to a solution of 1 equiv of Cp2TiCl2 in THF at room
temperature, after 5 min of stirring, caused the expected color
change in the solution, which turned from red to green, thus
confirming the reduction of the Ti(IV) species to Ti(III).
Addition at this point of 1 equiv of benzaldehyde 39 caused
the solution to show an orange hue within a few minutes, most
likely due to the presence of Ti(IV) pinacolates. NMR analysis
of an aliquot confirmed the disappearance of benzaldehyde and
B. Carbonyl Coupling-Olefination. i. Mechanistic Proposal.
On the basis of the above findings, namely, the dramatic
decrease in the calculated homolytic C-O bond dissociation
energy for RH2C-OH compared to that calculated for the
complex Cp2TiIII(Cl)HO-CH2R and the observed double-bond
formation upon treatment of diols 28, 30, 32, and 34 with
titanocene(III), we surmised that the Nugent’s reagent (Cp2TiCl)
would be able to promote carbonyl olefinations via homolysis
of the carbon-oxygen bonds present in the Ti(III) pinacolates
(IX, Scheme 4). Species IX would originate after reduction of
the corresponding Ti(IV) pinacolates (VIII, Scheme 4), which
are described as intermediates in Ti(III)-mediated pinacol
coupling of carbonyls.12
(12) (a) Barden, M. C.; Schwarzt, J. J. Am. Chem. Soc. 1996, 118, 5484–
5485. Enemærke, R. J.; Larsen, J.; Hjøllund, G. H.; Troels Skrydstrup,
T.; Daasbjerg, K. Organometallics 2005, 24, 1252–126. (b) Paradas,
´
M.; Campan˜a, A. G.; Este´vez, R. E.; Alvarez de Cienfuegos, L.;
Jime´nez, T.; Robles, R.; Cuerva, J. M.; Oltra, J. E. J. Org. Chem.
2009, 74, 3616-3619 and references cited therein.
(13) For the preparation of biscyclopentadienyl titanium alkoxides, see:
Huffman, J. C.; Moloy, K. G.; Marsella, J. A.; Caulton, K. G. J. Am.
Chem. Soc. 1980, 102, 3009-3014.
To test the validity of this hypothesis, we began by examining
whether homolytic cleavage of the C-O bond in the species
PhCH2OTiIIICp2 could be obtained by treating sodium benzoxide
(14) Villiers and Ephritikhine postulated the involvement of U(III)
analogues in the reductive coupling of ketones: Villiers, C.; Ephri-
tikhine, M. Chem.sEur. J. 2001, 14, 3043-3051.
9
J. AM. CHEM. SOC. VOL. 132, NO. 1, 2010 257

A R T I C L E S
Die´guez et al.
Table 3. Titanocene(III)-Mediated Reductive Carbonyl Coupling,,
deoxygenation process, we devoted our effort to investigate the
scope of this transformation. Thus, excellent to good yields and
reproducibility were also found with other aromatic and
conjugated aldehydes (Table 3, entries 1-10). This protocol
proved also to be applicable, less efficiently, to aromatic and
aliphatic ketones (Table 3, entries 11 and 12), where longer
reaction times were necessary to complete the reductive cou-
pling.15 Different aliphatic and aromatic aldehydes bearing
additional functional groups led efficiently to the coupling
products, which increases substantially the potential of this
transformation. Thus, aldehydes containing hydroxyl groups
protected as ethers or esters, and even unprotected hydroxyl
derivatives, coupled efficiently (Table 3, entries 2, 4, and 10).
Methoxycarbonyl and halogen derivatives also reacted with
acceptable yields (Table 3, entries 3 and 5).
With the aim of widening the scope of this transformation,
we checked the ability of titanocene(III) to promote intermo-
lecular cross-coupling reaction between dissimilar ketones. Thus,
we submitted benzophenone 60 and excess (4 equiv) cyclohex-
anone 59 to the standard reaction conditions to obtain the
intermolecular cross coupling adduct 61 at 65% yield (Table 3,
entry 13). Similar results were obtained with a mixture of citral
49 and benzaldehyde 39 in excess (Table 3, entry 14). It was
thus proven that unsymmetrical olefins can be generated by
reaction of aromatic and aliphatic carbonyls.
C. Catalytic Processes in Titanium. Our final efforts were
devoted to render these processes catalytic. Among the types
of reagents that have been described to regenerate Cp2TiCl2 from
species of oxygen-bonded titanium,16 chlorosilanes were claimed
to promote carbonyl coupling reactions catalytic in titanium by
regenerating TiCl3 from the corresponding titanium oxide,
namely, ClTidO, and accumulating the corresponding hexam-
ethyldisiloxane, which is finally removed in vacuum. Since we
propose the formation of titanocene oxides or hydroxytitanocene
derivatives in the deoxygenation step of these processes, we
decided to make farnesol 19 react with 0.3 equiv of Cp2TiCl2
(2 equiv is required in the noncatalytic route) in the presence
of excess Mn and 4 equiv of trimethylsilyl chloride (TMSCl).
To our delight, the deoxygenation product was obtained after
55 min in an exceptional 92% yield (Table 2, entry 8). The
deoxygenation of diol 30 could also be achieved by employing
substoichoimetric quantities of Ti and TMSCl as regenerator.
The yield of the corresponding alkene (86%) was again higher
than that obtained in the stoichoimetric version (Table 2, entry
16).
Additionally, this catalytic protocol could also be also used
to promote the reductive coupling of benzylic aldehydes, and
thus, by using only 0.3 mmol of Cp2TiCl2 and 4 equiv of
TMSCl, 46 was transformed into 31 in 95% yield (Table 3,
entry 6), which again meant an increase in the efficiency of
transformation when compared to the stoichiometric version.
The catalytic protocol for this McMurry reaction can be
rationalized as depicted in Scheme 5. Thus, the titanium oxo
species resulting from the deoxygenation step would react with
TMSCl in view of the pronounced oxophilicity of silicon.
Consequently, hexamethyldisiloxane, which can be easily
removed, would be the final acceptor of oxygen. In this sense,
the fact that TMSCl, a known regenerator of titanium chloride
from titanium oxides, is capable of regenerating the final
a 1.2 equiv of Ti(III) and 2.4 equiv of reducing agent per mole of
starting material were used; THF was used as solvent. b 0.3 equiv of
Ti(III), 4.0 equiv of TMSCl equiv, and 8 equiv of reducing agent per
mole of starting material were used; THF was used as solvent. c For
details, see text.
the sole presence of benzopinacol as reaction product. This
orange solution remained unaltered for hours, in both color and
composition. At this point, addition of a further 0.5 equiv of
Zn or Mn led the reaction mixture to recover its former green
color (15 min), which was rationalized by assuming a new
reduction from the corresponding Ti(IV) to Ti(III) pinacolate
(Scheme 4). At this moment, benzopinacol was again proven
to be the only component of the reaction mixture. As happened
previously, no change in the color or composition was observed
after several hours of stirring. However, gentle heating until
refluxing temperature was reached produced the formation of
stilbene 29, as well as a last color change, in this case, from
green to dark blue, which was attributed to the production of
titanocene oxide as a result of the deoxygenation step.
(15) Although in these cases, other reductive coupling mechanisms possibly
coexist, it should be noted that the corresponding pinacols deriving
from acetophenone 55 and geranylketone 57 could be isolated.
(16) Fu¨rstner, A.; Hupperts, A. J. Am. Chem. Soc. 1995, 117, 4468–4475.
ii. Scope of the Reaction. Once we obtained conclusive
evidence confirming the involvement of Ti(III) species in the
9
258 J. AM. CHEM. SOC. VOL. 132, NO. 1, 2010

Ti(III)-Mediated Weakening of C-O Bonds
A R T I C L E S
Scheme 5. Proposed Catalytic Cycle for Reductive Carbonyl
Olefination
function of the IspH enzyme in the natural biosynthesis of C5
units of the terpenoids through the deoxyxylulose phosphate
pathway.11
We have also described that Nugent’s reagent has the
capability to induce reductive carbonyl coupling in reasonable
short reaction times and with total reproducibility. The mecha-
nistic evidence presented, proving the involvement of Ti(III)
pinacolates in the deoxygenation step of this McMurry olefi-
nation, is particularly noteworthy since, in our opinion, it casts
some light on one of the mechanistically more obscure
transformations in organic chemistry.
Additionally, we have proven how these processes can be
performed catalytically in Cp2TiCl2 by using TMSCl as the final
oxygen trap. In this sense, we should underscore the fact that
keto alcohol 26 was efficiently transformed into the correspond-
ing deoxygenated compound 27 by using the catalytic protocol,
a process where the less reactive aliphatic keto group was
completely discriminated in favor of the allylic alcohol.
Our current interest is focused to gain a comprehensive
understanding of the overall reactions and to widen the scope
of these innovative processes. We are also engaged in testing
the applicability of these processes in the synthesis of bioactive
natural products.
titanium species produced in this transformation strongly
supports the proposed formation of Cp2TidO in the deoxygen-
ation process.
A final piece of evidence of the potential of this protocol
was obtained when the keto alcohol 26 was subjected to catalytic
conditions, giving an excellent 85% yield of the deoxygenated
compound. The yield improved significantly on that obtained
in the stoichiometric version, an increase that was noticed in
all cases where both versions were compared.
Acknowledgment. This research was supported by the Spanish
Ministry of Science and Innovation, Project CTQ2006-15575-C02-
01. We also thank the “Centro de Servicios de Informa´tica y Redes
de Comunicaciones” (CSIRC), University of Granada, for allowing
us to use its supercomputing facilities (UGRGrid). V.D. thanks the
Spanish Ministry of Science and Technology for a predoctoral grant
enabling him to purse these studies.
Conclusion
We describe two new synthetic applications of Nugent’s
reagent in organic chemistry as a direct consequence of the
dramatic decrease in homolytic C-O bond dissociation energy
for RH2C-OH compared to that for the complex
Cp2TiIII(Cl)HO-CH2R. The first of these applications is an easy
and reproducible deoxygenation-reduction procedure for al-
cohols and 1,2-diols. This method complements alternative
previously reported deoxygenations and is of general interest
to organic synthesis. Its application on the allylic terpenoid
alcohol 22 deserved to be emphasized, since it mimicks the
Supporting Information Available: Computational details,
1
experimental procedures, spectroscopic data, and H and 13C
NMR spectra for compounds 8, 10, 12, 14, 16, 18, 20, 23, 25,
27, 29, 31, 33, 41, 43, 45, 48, 50, 52, 54, 56, 58, 61, and 62
and complete ref 8a. This material is available free of charge
via the Internet at http://pubs.acs.org.
JA906083C
9
J. AM. CHEM. SOC. VOL. 132, NO. 1, 2010 259