Mechanism and scope of the cyanide-catalyzed cross silyl benzoin reaction

Article abstract of DOI:10.1021/ja044086y

In this work, cross silyl benzoin addition reactions between acylsilanes (1) and aldehydes (2) catalyzed by metal cyanides are described. Unsymmetrical aryl-, heteroaryl-, and alkyl-substituted benzoin adducts can be generated in moderate to excellent yie

Full text of DOI:10.1021/ja044086y

Published on Web 01/19/2005
Mechanism and Scope of the Cyanide-Catalyzed Cross Silyl
Benzoin Reaction
Xin Linghu, Cory C. Bausch, and Jeffrey S. Johnson*
Contribution from the Department of Chemistry, UniVersity of North Carolina,
Chapel Hill, North Carolina 27599-3290
Received September 28, 2004; E-mail: jsj@unc.edu
Abstract: In this work, cross silyl benzoin addition reactions between acylsilanes (1) and aldehydes (2)
catalyzed by metal cyanides are described. Unsymmetrical aryl-, heteroaryl-, and alkyl-substituted benzoin
adducts can be generated in moderate to excellent yields with complete regiocontrol using potassium cyanide
and a phase transfer catalyst. From a screen of transition metal cyanide complexes, lanthanum tricyanide
was identified as an improved second-generation catalyst for the cross silyl benzoin reaction. A study of
the influence of water on the KCN-catalyzed cross silyl benzoin addition revealed more practical reaction
conditions using unpurified solvent under ambient conditions. A sequential silyl benzoin addition/cyanation/
O-acylation reaction that resulted in two new C-C bonds was achieved in excellent yield. The mechanism
of cross silyl benzoin addition is proposed in detail and is supported by crossover studies and a number
of unambiguous experiments designed to ascertain the reversibility of key steps. No productive chemistry
arises from cyanation of the more electrophilic aldehyde component. Formation of the carbon-carbon
bond is shown to be the last irreversible step in the reaction.
Scheme 1
Introduction
The R-hydroxycarbonyl group is an important synthon for
the synthesis of natural products, chiral auxiliaries, and ligands.1
Among numerous synthetic strategies for introducing this
moiety, the benzoin condensation and related additions2-10
remain perhaps the most direct. In traditional metal cyanide-
catalyzed benzoin condensations, both aromatic and heteroaro-
matic aldehydes can be transformed into R-hydroxy ketones.
The substrate scope of the classic benzoin condensation was
extended to enolizable aliphatic aldehydes by employing thia-
zolium carbenes as catalysts.4,5 The mechanism of the benzoin
condensation has been extensively studied by many groups;
definitive contributions in the study of the cyanide system came
from Lapworth11 and Schowen,12 and from Breslow13,14 and
Leeper15 for thiazolium carbene catalysis.
Since all of the steps in the benzoin condensation are typically
reversible, its principal limitation arises in the coupling of two
different aldehydes; the product distribution of the cross benzoin
condensation is generally thought to be determined by the
relative thermodynamic stability of the four possible products
(Scheme 1). In some cases, regioisomeric mixtures can be
avoided if one of the two aldehydes is employed in excess.7c
To obtain the contrathermodynamic regioisomer, stoichiometric
carbonyl umpolung reagents, such as dithianes16,17 and protected
cyanohydrin derivatives,18,19 may be applied; however, extra
deprotection steps are required to obtain the benzoin product.
(1) (a) Coppola, G. M.; Schuster, H. F. R-Hydroxy Acids in EnantioselectiVe
Synthesis; Wiley-VCH: Weinheim, Germany, 1997. (b) Hanessian, S. Total
Synthesis of Natural Products: The Chiron Approach; Pergamon: New
York, 1983; Chapter 2.
(2) Ide, W. S.; Buck, J. S. Org. React. 1948, 4, 269-304.
(3) Hassner, A.; Rai, K. M. L. ComprehensiVe Organic Synthesis; Trost, B.
M., Fleming, I., Eds.; Pergamon Press: Oxford, 1991; pp 541-577.
(4) Stetter, H.; Raemsch, R. Y.; Kuhlmann, H. Synthesis 1976, 733-735.
(5) Stetter, H.; Daembkes, G. Synthesis 1977, 403-404.
(6) Johnson, J. S. Angew. Chem., Int. Ed. 2004, 43, 1326-1328.
(7) For some congeners of the classic benzoin reaction, see: (a) Rozwadowska,
M. D. Tetrahedron 1985, 41, 3135-3140. (b) Ricci, A.; Degl’Innocenti,
A.; Chimichi, S.; Fiorenza, M.; Rossini, G. J. Org. Chem. 1985, 50, 130-
133. (c) Heck, R.; Henderson, A. P.; Kohler, B.; Retey, J.; Golding, B. T.
Eur. J. Org. Chem. 2001, 2623-2627. (d) Pohl, M.; Lingen, B.; Mu¨ller,
M. Chem.sEur. J. 2002, 8, 5288-5295. (e) Cunico, R. F. Tetrahedron
Lett. 2002, 43, 355-358. (f) Demir, A. S.; Reis, O. Tetrahedron 2004, 60,
3803-3811. (g) Hachisu, Y.; Bode, J. W.; Suzuki, K. AdV. Synth. Catal.
2004, 346, 1097-1100.
(8) For highly enantioselective homobenzoin reactions, see: Enders, D.;
Kallfass, U. Angew. Chem., Int. Ed. 2002, 41, 1743-1745.
(9) For an example of highly enantioselective enzyme-catalyzed aryl-aryl cross
benzoin reactions, see: Du¨nkelmann, P.; Kolter-Jung, D.; Nitsche, A.;
Demir, A. S.; Siegert, P.; Lingen, B.; Baumann, M.; Pohl, M.; Mu¨ller, M.
J. Am. Chem. Soc. 2002, 124, 12084-12085.
(10) Enantioselective cross silyl benzoin reactions have been reported: Linghu,
X.; Potnick, J. R.; Johnson, J. S. J. Am. Chem. Soc. 2004, 126, 3070-
3071.
(11) Lapworth, A. J. Chem. Soc. 1903, 83, 995.
(12) Kuebrich, J. P.; Schowen, R. L.; Wang, M.-S.; Lupes, M. E. J. Am. Chem.
Soc. 1971, 93, 1214-1220.
(13) Breslow, R. J. Am. Chem. Soc. 1958, 80, 3719-3726.
(14) Breslow, R.; Kim, R. Tetrahedron Lett. 1994, 35, 699-702.
(15) White, M. J.; Leeper, F. J. J. Org. Chem. 2001, 66, 5124-5131.
(16) Seebach, D.; Corey, E. J. J. Org. Chem. 1975, 40, 231-237.
(17) Groebel, B. T.; Seebach, D. Synthesis 1977, 357-402.
(18) Hu¨nig, S.; Wehner, G. Synthesis 1975, 391-392.
(19) Hu¨nig, S.; Wehner, G. Chem. Ber. 1979, 112, 2062-2067.
9
10.1021/ja044086y CCC: $30.25 © 2005 American Chemical Society
J. AM. CHEM. SOC. 2005, 127, 1833-1840
1833

A R T I C L E S
Linghu et al.
strates33 and an effective catalyst for initiation of the Brook
rearrangement.25,26,28 Preliminary experiments indicated that
KCN/18-crown-6 was also an effective catalyst system for silyl
benzoin addition between benzoyl triethylsilane34 and benzal-
dehyde. With 10 mol % 18-crown-6 and 0.3 equiv of KCN,
silyl-protected benzoin adduct 3a-TES (Table 1) can be obtained
in 90% yield in an operationally convenient reaction time (2
h). Initial attempts to decrease the catalyst loading to 0.1 equiv
of KCN led to a significant increase in reaction time (>12 h).
Scheme 2. Cross Silyl Benzoin Additions Based on [1,2]-Brook
Rearrangement
To evaluate the generality and regiospecificity of the silyl
benzoin reaction, the substrate scope was studied using a variety
of acylsilanes and aldehydes (Table 1). Reactions between aryl
acylsilanes and aryl or heteroaromatic aldehydes gave good to
excellent yields of R-silyloxy ketone products at ambient
temperature (entries 1-11). The reaction times for all entries
were similar (typically 1-5 h). Electron-poor and electron-rich
substrates displayed little difference in reactivity. Significant
steric demand was tolerated without a decrease in yield or an
increase in reaction time (entry 10). The more challenging aryl-
alkyl′ and alkyl-aryl′ benzoin adducts were obtained in
moderate to good yields (entries 12-17). Easily prepared TMS
(trimethylsilyl) and DMPS (dimethylphenylsilyl) acylsilanes35,36
function analogously in the cross silyl benzoin addition;
however, compared with the triethylsilyl group, the more labile
groups were prone to desilylation during purification, and
R-silyloxy ketone products were isolated in somewhat lower
yields (entries 18 and 19). Either regioisomeric benzoin adduct
can be prepared simply through judicious selection of the
acylsilane and the aldehyde (cf. 3b-TES/3c-TES (entries 2 and
3), 3d-TES/3e-TES (entries 4 and 5), 3f-TES/3g-TES (entries
6 and 7), 3h-TES/3i-TES (entries 8 and 9), and 3p-TES/3q-
TES (entries 16 and 17). By way of comparison, thiazolium
carbene-catalyzed benzoin reaction between benzaldehyde and
isobutyraldehyde provides a 2:1 mixture of regioisomeric cross
acyloins,5 while the analogous cross silyl benzoin reaction gives
only one isomer, 3l-TES (entry 12). Benzoin reaction of
p-anisaldehyde with benzaldehyde via cyanide catalysis allows
access to only the thermodynamic isomer (hydroxyl derivative
of 3e-TES).2 In contrast, entries 4 and 5 demonstrate that the
kinetic control inherent in the acylsilane reaction circumvents
this issue and provides access to both regioisomers.
Since the formation of an acyl anion equivalent via the
reaction of an aldehyde with cyanide ion or a thiazolium carbene
is the key step in the benzoin and related20 condensations, we
speculated that regiocontrolled direct cross benzoin reactions
may be achieved if alternative methods of acyl anion formation
could place the reaction under kinetic control. In this context,
in situ generation of (silyloxy)nitrile anions from acylsilanes
by cyanide-promoted [1,2]-Brook rearrangement21,22 appeared
to have significant potential. The application of this particular
silicon migration has been reported by several groups.
Degl’Innocenti demonstrated that enones are acylated by
acylsilanes under the influence of cyanide catalysis,23 while
Reich reported that cyanide triggers an addition/rearrangement/
elimination sequence with an R-thiophenyl acylsilane.24 Take-
da’s group has shown that (silyloxy)nitrile anions generated by
the Brook rearrangement in the reaction of (â-(trimethylsilyl)-
acryloyl)silane can undergo methylation at the γ-position.25 Most
recently, cyanide-catalyzed cyanation/1,2-Brook/acylation reac-
tions of acylsilanes have been disclosed.26-28 We anticipated
that with cyanide as the catalyst, unsymmetrical R-silyloxy
ketone products could be prepared by trapping (silyloxy)nitrile
anions with aldehydes (Scheme 2). This projected application
differs from the conceptually related fluoride-promoted acyl-
silane/aldehyde coupling reactions described by Degl’Innocenti
and Heathcock: attack of F- at silicon is thought to be crucial
to the success of those additions.29-31
This article provides a full account of the development of a
new cross silyl benzoin reaction, including improved reaction
conditions and an evaluation of the mechanism.32
Results and Discussion
Influence of Water on the Phase Transfer Process and
Development of More Practical Reaction Conditions. We
presumed that the (silyloxy)nitrile anion intermediates (1-MCN,
Scheme 2) of the cross silyl benzoin reaction would be sensitive
to water. This moisture sensitivity problem has been addressed
in our previous studies of the cyanation/1,2-Brook/acylation of
acylsilanes;26,27 therefore, the reactions in Table 1 were con-
ducted in dry solvent under an inert atmosphere. While yields
were always the same for a given acylsilane/aldehyde pair, the
reaction times were not reproducible (normally, from 1 to 5 h
with 0.3 equiv of KCN/18-crown-6). We also noticed that
Development of a KCN/18-Crown-6 Catalytic System.
KCN in combination with the phase transfer catalyst, 18-crown-
6, is an effective reagent for cyanation of electrophilic sub-
(20) Murry, J. A.; Frantz, D. E.; Soheili, A.; Tillyer, R.; Grabowski, E. J. J.;
Reider, P. J. J. Am. Chem. Soc. 2001, 123, 9696-9697.
(21) Brook, A. G. Acc. Chem. Res. 1974, 7, 77-84.
(22) Moser, W. H. Tetrahedron 2001, 57, 2065-2084.
(23) Degl’Innocenti, A.; Ricci, A.; Mordini, A.; Reginato, G.; Colotta, V. Gazz.
Chim. Ital. 1987, 117, 645-648.
(24) Reich, H. J.; Holtan, R. C.; Bolm, C. J. Am. Chem. Soc. 1990, 112, 5609-
5617.
(25) Takeda, K.; Ohnishi, Y. Tetrahedron Lett. 2000, 41, 4169-4172.
(26) Linghu, X.; Nicewicz, D. A.; Johnson, J. S. Org. Lett. 2002, 4, 2957-
2960.
(27) Nicewicz, D. A.; Yates, C. M.; Johnson, J. S. J. Org. Chem. 2004, 69,
6548-6555.
(33) Evans, D. A.; Truesdale, L. K. Tetrahedron Lett. 1973, 4929-4932.
(34) Acyl triethylsilanes were prepared in three steps from the corresponding
aldehydes. For detailed information, see the Supporting Information of ref
26.
(35) Benzoyl trimethylsilane was prepared by reductive silylation of methyl
benzoate. For detailed information, see: Tongco, E. C.; Wang, Q.; Prakash,
G. K. S. Synth. Commun. 1997, 27, 2117-2123.
(36) Acyl dimethylphenylsilanes were prepared by reductive lithiation of
chlorodimethylphenyl silane, transmetalation, and addition to acid chlorides.
For detailed information, see: Bonini, B. F.; Comesfranchini, M.; Mazzanti,
G.; Passamonti, U.; Ricci, A.; Zani, P. Synthesis 1995, 92-96.
(28) Tanaka, K.; Takeda, K. Tetrahedron Lett. 2004, 45, 7859-7861.
(29) Schinzer, D.; Heathcock, C. H. Tetrahedron Lett. 1981, 22, 1881-1884.
(30) Ricci, A.; Degl’Innocenti, A.; Chimichi, S.; Fiorenza, M.; Rossini, G. J.
Org. Chem. 1985, 50, 130-133.
(31) Degl’Innocenti, A.; Pike, S.; Walton, D. R. M.; Seconi, G.; Ricci, A.;
Fiorenza, M. J. Chem. Soc., Chem. Commun. 1980, 1201-1202.
(32) A portion of this work has been communicated: (a) Linghu, X.; Johnson,
J. S. Angew. Chem., Int. Ed. 2003, 42, 2534-2536. (b) Bausch, C. C.;
Johnson, J. S. J. Org. Chem. 2004, 69, 4283-4285.
9
1834 J. AM. CHEM. SOC. VOL. 127, NO. 6, 2005

Cross Silyl Benzoin Additions Catalyzed by Metal Cyanides
A R T I C L E S
Table 2. Influence of Water on the Phase Transfer Process
Table 1. Cross Silyl Benzoin Additions of Acylsilanes and
Aldehydes Catalyzed by KCN/18-Crown-6a
added H2O
(equiv)b
yield
(%)c
entry
catalyst
time
1
KCN/18-crown-6
(30 mol %)
KCN/18-crown-6
(10 mol %)
KCN/18-crown-6
(10 mol %)
KCN/18-crown-6
(10 mol %)
0
5 h
79
76
83
75
70
low
2
3
4
5
6
7
0
24 h
6 h
2
10
3 h
KCN/18-crown-6
(10 mol %)
KCN/18-crown-6
20
3 h
excessd
0
24 h
<10 min
(10 mol %)
conversion
74
[18-crown-6‚K]CNe
(10 mol %)
a Et2O (10 mL) was dried by passage through a bed of activated Al2O3
under Ar. b Equiv of H2O relative to KCN. c Isolated yield of analytically
pure material. d A 2:1 Et2O/H2O biphase was employed. e [18-Crown-
6‚K]CN complex was prepared by the literature method (ref 37).
reactions sealed with new septa required longer reaction times
than those sealed with used septa. Those observations led us to
believe that trace amounts of water may not halt catalysis by
quenching 1-MCN; on the contrary, water could increase the
reaction rate by facilitating the phase transfer process. Several
test reactions between benzoyl triethylsilane (1a-TES) and
p-anisaldehyde (2c) were designed to evaluate the effect of H2O;
the results are summarized in Table 2.
All experiments were performed in oven-dried flasks sealed
by new septa, and dry Et2O was obtained from an activated
alumina column.38 Under anhydrous conditions, the reaction
with 30 mol % KCN and 30 mol % 18-crown-6 was complete
in 5 h in 79% yield (entry 1), while the reaction using 10 mol
% catalyst required 24 h and afforded a slightly lower yield
(entry 2). By introducing 20 mol % H2O, the reaction time
dropped to 6 h (10 mol % catalyst), while the yield increased
from 76 to 83% (entry 3). The addition of more H2O (100 mol
%) resulted in further reduction of the reaction time (from 6 to
3 h), but no further change was observed with 200 mol % H2O
(entries 4 and 5). The role of water in solid-liquid phase transfer
catalysis of KCN has been studied by Liotta and co-workers.39
They observed similar results in the reaction of benzyl halides
with KCN in the presence of 18-crown-6: small amounts of
added water resulted in a significant rate enhancement. They
postulated that water added to the solid-liquid system could
coat the surface of the solid particles to form a third phase that
was termed the omega phase. This third phase consists of water,
“dissolved” KCN, organic solvent, and crown ether extracted
from organic phase. The complexation of KCN by 18-crown-6
occurs in the omega phase. It was proposed that in the absence
of added water, the mass transfer of KCN from the solid phase
directly into the organic phase is slow and becomes the limiting
step of the reaction, while in the presence of small amounts of
a R1C(O)SiR3 (1.0 equiv), R2CHO (1.1 equiv), 18-crown-6 (0.1 equiv), and
KCN (0.3 equiv) in Et2O at 25 °C for 1-5 h, unless otherwise noted. See
Supporting Information for details. b Isolated yield of analytically pure material.
c Reaction time ) 10 h. d [18-Crown-6‚K]CN complex (0.2 equiv); the silyloxy
ketone product was subjected to deprotection with 1 M HCl in MeOH.
e R1C(O)SiR3 (1.0 equiv), R2CHO (4.0 equiv), 18-crown-6 (0.3 equiv), and KCN
(0.5 equiv) in Et2O at 25 °C. f [18-Crown-6‚K]CN complex (0.1 equiv).
(37) Livinghouse, T. Org. Synth. 1981, 60, 126-132.
(38) Alaimo, P. J.; Peters, D. W.; Arnold, J.; Bergman, R. G. J. Chem. Educ.
2001, 78, 64.
(39) Liotta, C. L.; Berkner, J.; Wright, J.; Fair, B. ACS Symposium Series 659;
American Chemical Society: Washington, DC, 1997; pp 29-40.
9
J. AM. CHEM. SOC. VOL. 127, NO. 6, 2005 1835

A R T I C L E S
Linghu et al.
Table 3. Evaluation of M(CN)n Complexes as Cross Silyl Benzoin
Catalysts
added water, the omega phase is an integral part of the reaction
system and the mass transfer of the complex into the organic
phase is faster.
Attempts to carry out the reaction under liquid-liquid biphase
conditions failed to give appreciable product (entry 6), probably
due to an unfavorable partition coefficient. Since KCN and 18-
crown-6 eventually form [18-crown-6‚K]CN, we speculated that
the catalytic reaction may be more efficient with a preformed
complex. Since [18-crown-6‚K]CN37 has appreciable solubility
in Et2O, the solid-liquid biphasic reaction was converted to a
homogeneous catalytic process. The reaction time with 10 mol
% of this complex decreased dramatically to less than 10 min
in the absence of water (entry 7). This suggested that in the
presence of water and without the preformed complex, steps
involving the mass transfer of KCN(s) from the solid phase to
the omega phase and complexation of KCN(ω) by 18-crown-6
were still rate limiting.
Considering the effect of water on phase transfer catalysis,
we expected to develop a more practical and efficient cross silyl
benzoin reaction. Taking advantage of a trace amount of water
in unpurified Et2O, R-silyloxy ketone can be obtained in 88%
yield with 10 mol % catalyst loading (eq 1). The reaction is
complete in an operationally convenient time (6 h) and can be
performed without the need for an inert atmosphere. Tetrabu-
tylammonium cyanide, which is commercially available or easily
prepared from NaCN and a cheaper phase transfer catalyst,
tetrabutylammonium bisulfate,40 is an alternative catalyst for
the cross benzoin reaction. Within 30 min, the reaction with 10
mol % catalyst loading afforded 3d-TES in 88% yield (eq 2).
entry
catalyst
yielda(%)
1
2
3
4
5
6
7
8
9
(salen)Al-CN
Er(iOPr)3/Me3SiCN
Yb(iOPr)3/Me3SiCN
0
42-66
30-55
40
50
75
51
5
68
b
Ce(CN)3
Er(CN)3
La(CN)3
b
b
b
Sm(CN)3
b
Y(CN)3
b
Yb(CN)3
a Isolated yield of 4b. b Catalyst prepared via treatment of 0.1 equiv of
(n-Bu)3M with 0.3 equiv of Me3SiCN (see text).
SiCN failed to provide reproducible results (entries 2 and 3);
however, Ln(CN)3 generated from (n-Bu)3Ln/Me3SiCN was
found to be an effective catalyst and provided consistent
yields.41,42 According to the Utimoto protocol,41 LaCl3 (0.1
equiv) was treated with n-butyllithium (0.3 equiv) to give a
tributyllanthanum species, which upon subsequent addition of
Me3SiCN, provided the active catalyst. Six different (cyanide)-
lanthanum catalysts were evaluated in the reaction between
benzoyl dimethylphenylsilane (1a-DMPS) and 4-chlorobenzal-
dehyde (2b); the results are summarized in Table 3. All catalysts
were generated in situ, and the starting materials 1a-DMPS and
2b were added to the catalyst suspension. While all of the Ln-
(CN)3 complexes delivered the desired R-hydroxy ketone 4b
(after acidic workup), La(CN)3 afforded the highest yield (entry
6).
With the optimized catalyst in hand, the scope of the cross
benzoin reaction was further studied (Table 4). From our earlier
experience that the more labile TMS (trimethylsilyl) and DMPS
(dimethylphenylsilyl) groups underwent some desilylation dur-
ing purification, all substrates that contained those two silyl
groups were subjected to silyl ether deprotection with aqueous
HCl upon completion of the reaction. Compared with the KCN/
18-crown-6-catalyzed cross benzoin reaction, La(CN)3 typically
catalyzed the reactions in less than 5 min (cf. 1-5 h for reactions
with KCN/18-crown-6). The La(CN)3 catalyst gave yields
comparable to those of the KCN/18-crown-6 catalyst previously
reported for aryl-aryl′ benzoin adducts (entries 1-3, 6, and
7). The La(CN)3 system gave excellent yields in coupling
PhCOSiEt3 and heteroaromatic aldehydes (entries 4 and 5).
PhCOSiMe3 underwent selective catalyzed 1,2-addition to an
R,â-unsaturated aldehyde (entry 8). Aliphatic aldehydes, includ-
ing those that are R-branched (cyclohexane carboxaldehyde,
entry 10) and tert-alkyl (pivaldehyde, entry 11), are also
effective acceptor substrates. It is noteworthy that the latter
aldehyde is completely unreactive in the KCN/18-crown-6
system.
Development of a La(CN)3 Catalytic System. Although the
substrate scope of KCN/18-crown-6-catalyzed cross silyl ben-
zoin additions was reasonable for aryl-aryl′ combinations, the
reaction produced lower yields for alkyl-aryl′ or aryl-alkyl′
adducts and afforded <20% alkyl-alkyl′ products. We won-
dered if a transition metal counterion could provide a more
tolerant catalyst. To enhance the reactivity of both aromatic and
aliphatic substrates and to further provide a platform for the
development of enantioselective variants, transition metal
cyanides were examined as new catalysts for the cross silyl
benzoin reaction.
The recent observation that (salen)Al-CN complexes react
with acylsilanes to give (silyloxy)nitrile anions (1-MCN, M )
Al(salen)) led us to consider these complexes as potential
catalysts,27 but our efforts in this area were not fruitful (entry
1). A possible reason for the absence of catalysis will be
discussed in the context of our mechanistic studies (vide infra).
Attempts employing lanthanum i-propoxides activated by Me3-
A significant improvement has been observed for alkyl-aryl′
and alkyl-alkyl′ benzoin adducts. In our previous report,
catalyst loading and electrophile concentration had to be
(41) Matsubara, S.; Onishi, H.; Utimoto, K. Tetrahedron Lett. 1990, 31, 6209-
6212.
(42) Schaus, S. E.; Jacobsen, E. N. Org. Lett. 2000, 2, 1001-1004.
(40) Dehmlow, E. V.; Kunesch, E. Liebigs Ann. Chem. 1985, 1904-1909.
9
1836 J. AM. CHEM. SOC. VOL. 127, NO. 6, 2005

Cross Silyl Benzoin Additions Catalyzed by Metal Cyanides
A R T I C L E S
Table 4. Catalyzed Silyl Benzoin Addition Reactions of
step of the classic benzoin reaction is aldehyde cyanation, while
the cross silyl benzoin reaction relies on productive cyanation
of the acylsilane at the onset of the catalytic cycle. Because
aldehydes are more electrophilic than acylsilanes, successful
catalysis of the cross silyl benzoin reaction is dependent on
reVersible and nonproductiVe aldehyde cyanation: the derived
alkoxide products, 6-K and 6-La, cannot be stable. If aldehyde
cyanation were irreversible, those alkoxides would become a
nonproductive shunt of metal cyanide catalysts and acylsilane
cyanation would be impossible. The absence of benzoin catalysis
with (salen)Al-CN complexes (Table 3, entry 1) is an example
where a stable tetrahedral intermediate sequesters a metal
cyanide that has been previously demonstrated to react with
acylsilanes.
Acylsilanes and Aldehydesa
(i) Aldehyde Cyanation. To assess the reversibility of
aldehyde cyanation in both KCN/18-crown-6- and La(CN)3-
catalyzed cross silyl benzoin reactions, mandelonitrile (5) was
employed as a catalyst precursor in conjunction with bases that
would furnish the appropriate counterion, KH or (n-Bu)3La,
respectively. In the KCN/18-crown-6 system, the reaction
between KH and the cyanohydrin should be rapid and irrevers-
ible to generate alkoxide 6-K. If aldehyde cyanation is revers-
ible, benzaldehyde extrusion should yield the competent catalyst
KCN. In the test experiment, after KH and 18-crown-6 were
introduced to an Et2O solution of PhCH(OH)CN, (p-MeOPh)-
COSiEt3, and PhCHO, the cross benzoin reaction was complete
within 5 min. The desired R-silyloxy ketone product, (p-
MeOPh)COCH(OSiEt3)Ph (3e-TES), was obtained in 81% yield
(Scheme 3). Analogously, La(CN)3 can be generated by treating
Scheme 3. Test for Reversibility of Aldehyde Cyanation (KCN/
18-Crown-6 System)
Scheme 4. Test for Reversibility of Aldehyde Cyanation (La(CN)3
System)
a R1C(O)SiR3 (1.0 equiv), R2CHO (1.1 equiv), and La(CN)3 (0.1 equiv)
in THF at 25 °C for <5 min, unless otherwise noted. See Supporting
Information for details. b Yield of isolated analytically pure material. c The
silyloxy ketone product was subjected to deprotection with 1 M HCl in
MeOH. d R1C(O)SiMe3 (1.0 equiv), R2CHO (1.5 equiv), and La(CN)3 (0.1
equiv) in THF at 25 °C for 15 min.
significantly increased to achieve satisfactory yields. La(CN)3
catalyzes the silyl benzoin reaction of acetyl trimethylsilane and
4-chlorobenzaldehyde in reasonable yield (entry 12), and for
the first time that we are aware, cyanide catalysis has proven
effective (albeit in a modest yield) for the synthesis of an alkyl-
alkyl′ acyloin (entry 13).
a THF solution of PhCH(OH)CN with a solution of freshly
prepared (n-Bu)3La. After introducing a THF solution of (p-
MeOPh)COSiEt3 and PhCHO, the cross silyl benzoin product,
3e-TES, was isolated in 82% yield (Scheme 4). Both results
were consistent with our working hypothesis.
Mechanism of the Cross Silyl Benzoin Addition. (A)
Evaluation of Potential Competing Reactions. The initiating
9
J. AM. CHEM. SOC. VOL. 127, NO. 6, 2005 1837

A R T I C L E S
Linghu et al.
These results further suggested a practical improvement in
executing the catalyzed reactions. Acetone cyanohydrin (8) is
a latent source of -CN that is cheap, stable, and easily handled.
After PhCOSiEt3 1a-TES and p-anisaldehyde 2c were mixed
with acetone cyanohydrin 8 (0.1 equiv) and Bu4NBr (0.1 equiv)
in the presence of powdered KOH (0.1 equiv), R-silyloxy ketone
product 3d-TES can be obtained in 77% yield within 5 min
(eq 3). This represents an operationally simple and economically
attractive method of catalysis.
hyde participate in a KCN/18-crown-6-catalyzed silyl benzoin
reaction. Upon consumption of the acylsilane (TLC analysis),
ethyl cyanoformate (10) was added. The R-silyloxy ketone
intermediate (3) underwent a subsequent catalyzed cyanation,
1,4-silyl migration, and O-acylation with ethyl cyanoformate
to afford silyl cyanohydrin carbonate 11-TBS in 91% yield as
a single regioisomer (d.r. ) 3.3:1, relative stereochemistry
unassigned). The analogous reaction between benzoyl triethyl-
silane and p-chlorobenzaldehyde affords TES-cyanohydrin
carbonate 11-TES in 74% yield as a single regioisomer (d.r. )
2.3:1, relative stereochemistry unassigned). Conveniently, or-
thogonal protection of the product’s two carbinol groups is an
inherent characteristic of the reaction.
Due to the migration of the silyl group between two adjacent
oxygen atoms of the intermediates 9-a and 9-b (Scheme 6),
both intermediates can, in principle, react with cyanoformate,
and two possible regioisomers could be obtained; however, only
the regioisomer derived from acylation of 9-b was observed.43
The deactivating geminal nitrile group and steric congestion
associated with alkoxide 9-a apparently renders the tertiary site
unreactive. Equilibration via 1,4-silyl migration is clearly faster
than acylation of 9-a, and the more nucleophilic secondary
alkoxide 9-b reacts with cyanoformate to give the observed
regioisomer.
(ii) Aldehyde Dimerization. The classic benzoin condensa-
tion is conducted in warm alcohol solvent.2 A control experiment
showed that under typical cross silyl benzoin reaction conditions
(Et2O, ambient temperature), benzoin (4a) was not observed
over extended periods of time (eq 4 in Scheme 5). To verify
Scheme 5. Test for Competing Homobenzoin Reaction under
Cross Silyl Benzoin Addition Conditions
The new tandem reaction provided us with a useful piece of
mechanistic information: the last two steps of cross silyl benzoin
addition are reversible.44 At this stage, there were two questions
left to obtain the integrated catalytic cycle information: (1) Does
the intermediate 9-b undergo C-C bond cleavage during the
reaction process (9-b f 9-c)? (2) Are acylsilane cyanation and
[1,2]-Brook rearrangement reversible under the reaction condi-
tions, that is, is 9-c formed reversibly?
(C) Evaluating Reversibility of the C-C Bond Forming
Step. Because of our interest in the development of enantiose-
lective silyl benzoin reactions,10 we were concerned about the
reversibility of C-C bond formation. We designed two cross-
that benzoin is not formed reversibly, 4a was subjected to
normal cross silyl benzoin conditions (eq 5 in Scheme 5). The
absence of benzaldehyde further indicates that no productive
chemistry arises from aldehyde cyanation under the mild
reaction conditions employed.
(B) Tandem Silyl Benzoin Addition/Cyanation/O-Acyla-
tion and Underlying Mechanistic Information. The cross silyl
benzoin addition has been coupled to a second C-C bond
construction that occurs in the same reaction pot (Scheme 6).
1
over experiments to investigate this point. On the basis of H
NMR analysis, the reaction between triethylsilyloxy phenyl
acetophenone (3a-TES) and p-chlorobenzaldehyde (2b) under
standard conditions afforded none of the crossover product 3b-
TES over 36 h (eq 6 in Scheme 7). If C-C bond cleavage had
occurred, (silyloxy)nitrile anion 9-c would have been generated
and aldehyde 2b would have been incorporated to give the
crossover product 3b-TES, a compound that we have fully
characterized. The absence of 3b-TES confirms that 9-b and
9-c are not in equilibrium. This conclusion was verified by a
second experiment in which enantiomerically enriched silyloxy
benzoin 3e-TES was treated with 10 mol % [18-crown-6‚K]-
CN for an extended period of time (eq 7 in Scheme 7). The
fact that the reaction only gave slightly diminished enantiomeric
excess over 24 h indicated that on the time scale of the reaction,
racemization is negligible and C-C bond formation is irrevers-
ible. Base-catalyzed racemization via enolization is also dis-
counted by this experiment.
Scheme 6. Sequential One-Pot Silyl Benzoin Addition/Cyanation/
O-Acylation Reaction
(43) Details of the regiochemical proof can be found in the Supporting
Information.
(44) A reviewer correctly pointed out that this experiment suggests, but does
not definitively verify, reversibility of the retrocyanation and silyl migration
under the silyl benzoin conditions since we have introduced a new variable
(i.e., presence of ethyl cyanoformate). We agree that the probability that
these equilibria are “turned on” only in the presence of ethyl cyanoformate
is small, but not zero.
Thus, benzoyl tert-butyldimethylsilane and p-chlorobenzalde-
9
1838 J. AM. CHEM. SOC. VOL. 127, NO. 6, 2005

Cross Silyl Benzoin Additions Catalyzed by Metal Cyanides
A R T I C L E S
Scheme 7. Test for Reversibility of C-C Bond Formation
(D) Interrogating the Reversibility of Acylsilane Cyanation
and [1,2]-Brook Rearrangement. Hu¨nig and co-workers
reported lithiation of phenyl trimethylsilyloxy cyanohydrin 12a
followed by addition to aldehydes to form unsymmetrical
benzoins.18,19 Intermediate 9-c (M ) Li) is inferred as the key
nucleophilic species. We wished to synthesize the proposed
intermediate 9-c (M ) [18-crown-6‚K]) via an analogous
method. Phenyl silyloxy cyanohydrin 12a was treated by KH
in the presence of 18-crown-6 (Scheme 8) in an effort to
aprotic solvent. To provide experimental support for the
proposed mechanism, it was deemed critical to establish the
nature of the silyl group transfer. A simple crossover experiment
was designed (Scheme 9); two acylsilanes, benzoyl tert-
butyldimethylsilane (1a-TBS) and p-methoxybenzoyl triethyl-
silane (1c-TES), which are differentiated in the identity of both
the aryl and silyl groups, underwent reaction with a single
aldehyde (2a, 2 equiv) under normal cross silyl benzoin
conditions. The product distribution was evaluated by 1H NMR
spectroscopy. Only 3a-TBS and 3e-TES were observed, con-
firming the intramolecular silyl group transfer pathway.
Scheme 8. Test for Reversibility of Cyanation/Brook
Rearrangement Steps
Proposed Catalytic Cycle for Cross Silyl Benzoin Addi-
tions. The collected mechanistic information allows us to de-
rive a reasonable catalytic cycle for cross silyl benzoin addi-
tions (Scheme 10). Aldehyde cyanation occurs, but is neither
productive nor irreversible; therefore, cyanide is not sequestered
and is available to undergo nucleophilic addition to acylsilane
1, followed by [1,2]-Brook rearrangement to yield (silyloxy)-
nitrile anion 9-c. There was no direct evidence for the revers-
ibility of those two steps. The cross silyl benzoin reaction is
regiospecific by virtue of an irreversible carbon-carbon bond
forming step. The silyl group shuttles between two adjacent
oxygen atoms in intermediates 9-b and 9-a, the latter of which
reversibly releases the metal cyanide catalyst and leads to the
desired product.
generate (silyloxy)nitrile anion 9-c in the same environment as
it is formed in the cross silyl benzoin reaction. If reverse Brook
(silyl-Wittig) rearrangement45,46 and retrocyanation occur, -CN
and acylsilane 1a-TES would be formed. We attempted to verify
the generation of acylsilane 1a-TES by 1H NMR spectroscopy,
but in no case were we able to observe its NMR signatures.
We sought indirect evidence by trapping the generated -CN
with other electrophiles; however, neither the reaction with
p-methoxybenzoyl triethylsilane (aqueous workup) nor the
reaction with benzyl bromide afforded the expected products.
Both reactions showed only undefined decomposition of the
starting material 12a; again, acylsilane 1a-TES was not
observed. At this time, we cannot conclude whether 9-c is
formed reversibly under the reaction conditions.
Conclusions
A new cross silyl benzoin addition reaction with complete
regiochemical control has been developed to generate R-silyloxy
ketone products. The KCN/18-crown-6 catalyst system can
afford good to excellent yields for aryl and heteroaryl substrates.
The scope was extended to alkyl and R,â-unsaturated substrates
with an improved La(CN)3 catalyst. Operationally simple
reaction conditions were discovered that allow the alkali
cyanide-catalyzed reactions to be conducted under ambient
(E) Test for Silyl Group Transfer Pathway. Since tradi-
tional benzoin reactions are run in ethanol, the proton-transfer
steps could proceed through an intramolecular and/or intermo-
lecular pathway. In the cross silyl benzoin addition, silyl transfer
takes the place of proton transfer, and the reaction proceeds in
Scheme 9. Test for Silyl Group Transfer Pathway
9
J. AM. CHEM. SOC. VOL. 127, NO. 6, 2005 1839

A R T I C L E S
Linghu et al.
Scheme 10. Proposed Mechanism for Cross Silyl Benzoin
Addition
silyl benzoin additions was proposed based on detailed mecha-
nistic studies.
Acknowledgment. We are grateful to Research Corporation,
the donors of the American Chemical Society Petroleum
Research Fund, and the National Institutes of Health (National
Institute of General Medical Sciences, GM068443) for partial
support of this research. X.L. acknowledges a Burroughs
Wellcome Fellowship. J.S.J. is an Eli Lilly Grantee and a 3M
Nontenured Faculty Awardee.
Supporting Information Available: Detailed experimental
procedures and compound characterization data. This material
is available free of charge via the Internet at http://pubs.acs.org.
JA044086Y
conditions (rt, air). A tandem silyl benzoin addition/cyanation/
O-acylation reaction that resulted in two new C-C bonds was
achieved in excellent yields. Finally, the catalytic cycle of cross
(45) Wright, A.; Ling, D.; Boudjouk, P.; West, R. J. Am. Chem. Soc. 1972, 94,
4784-4785.
(46) Linderman, R. J.; Ghannam, A. J. Am. Chem. Soc. 1990, 112, 2392-2398.
9
1840 J. AM. CHEM. SOC. VOL. 127, NO. 6, 2005