Diastereoselectivity in gas-phase hydride reduction reactions of ketones

Article abstract of DOI:10.1021/ja9910053

The intrinsic diastereoselectivity of the reduction of a series of cyclic ketones by pentacoordinate silicon hydride ions was investigated in the gas phase with the use of the flowing afterglow-triple quadrupole technique. The percent axial reduction of 4-tert-butylcyclohexanone (1), 2-methylcyclohexanone (2), 3,3,5-trimethylcyclohexanone (3), norcamphor (4), 2-tert-butyl-1,3-dioxan-5-one (5), and 2-tert-butyl-1,3-dithian-5-one (6) was determined by collision-induced dissociation experiments. The results show that axial (exo) reduction dominates for 1, 2, and 4, whereas equatorial reduction is dominant for 3, 5, and 6. The trend observed for the reduction diastereoselectivity of compounds 1-4 and 6 matches their condensed-phase behavior; i.e., the percent axial reduction is reduced when steric hindrance of the ketones is increased (1, 99 ± 3%; 2, 68 ± 5; 3, 9 ± 3%). The remarkable consistency of the results obtained in the gas phase and in solution suggests that environmental effects are either unimportant or cancel out and that the reduction diastereoselectivity is a property that can be attributed to the intrinsic nature of the isolated reactants. Qualitatively, the predictions made by Houk et al. regarding the diastereoselectivity of the reduction of 5 and 6 in the gas phase were confirmed, i.e., a preferred equatorial approach of the hydride reducing agent. The preference of compound 5 to undergo equatorial reduction in the gas phase (33 ± 4% axial reduction) contrasts with the almost exclusive. axial reduction reported in solution (93%). This deviation is likely caused by the strong electrostatic repulsion between the nucleophilic hydride reagent and the ring heteroatoms in 5. Compound 6 exhibits an even stronger preference for equatorial reduction (16 ± 4% axial reduction), in agreement with experimental results obtained by others in the condensed phase. Earlier calculations predict an even stronger preference for equatorial reduction. These results are readily rationalized in terms of competition among steric, torsional, and electrostatic effects.

Full text of DOI:10.1021/ja9910053

7130
J. Am. Chem. Soc. 1999, 121, 7130-7137
Diastereoselectivity in Gas-Phase Hydride Reduction Reactions of
Ketones
†
Alexander Artau,* Yeunghaw Ho, Hilkka Kentt a1 maa, and Robert R. Squires
Contribution from the The Department of Chemistry, Purdue UniVersity, West Lafayette, Indiana 47907
ReceiVed March 29, 1999
Abstract: The intrinsic diastereoselectivity of the reduction of a series of cyclic ketones by pentacoordinate
silicon hydride ions was investigated in the gas phase with the use of the flowing afterglow-triple quadrupole
technique. The percent axial reduction of 4-tert-butylcyclohexanone (1), 2-methylcyclohexanone (2), 3,3,5-
trimethylcyclohexanone (3), norcamphor (4), 2-tert-butyl-1,3-dioxan-5-one (5), and 2-tert-butyl-1,3-dithian-
5
-one (6) was determined by collision-induced dissociation experiments. The results show that axial (exo)
reduction dominates for 1, 2, and 4, whereas equatorial reduction is dominant for 3, 5, and 6. The trend observed
for the reduction diastereoselectivity of compounds 1-4 and 6 matches their condensed-phase behavior; i.e.,
the percent axial reduction is reduced when steric hindrance of the ketones is increased (1, 99 ( 3%; 2, 68 (
5
; 3, 9 ( 3%). The remarkable consistency of the results obtained in the gas phase and in solution suggests
that environmental effects are either unimportant or cancel out and that the reduction diastereoselectivity is a
property that can be attributed to the intrinsic nature of the isolated reactants. Qualitatively, the predictions
made by Houk et al. regarding the diastereoselectivity of the reduction of 5 and 6 in the gas phase were
confirmed, i.e., a preferred equatorial approach of the hydride reducing agent. The preference of compound 5
to undergo equatorial reduction in the gas phase (33 ( 4% axial reduction) contrasts with the almost exclusive
axial reduction reported in solution (93%). This deviation is likely caused by the strong electrostatic repulsion
between the nucleophilic hydride reagent and the ring heteroatoms in 5. Compound 6 exhibits an even stronger
preference for equatorial reduction (16 ( 4% axial reduction), in agreement with experimental results obtained
by others in the condensed phase. Earlier calculations predict an even stronger preference for equatorial reduction.
These results are readily rationalized in terms of competition among steric, torsional, and electrostatic effects.
3a
Introduction
approach from the more hindered axial face). Cieplak rational-
izes these results by stereoelectronic control that involves
electron donation from the anti-periplanar σ C-H orbitals of
the cyclohexane ring into the vacant antibonding σ* orbital of
the incipient bond.
Conversion of a ketone to an alcohol by hydride addition is
a key reaction in organic synthesis. Prediction and control of
the stereochemical outcome of these reactions are necessary for
the synthesis of the desired diastereomer or enantiomer of the
product alcohol. This need has inspired numerous qualitative
and quantitative models for the prediction of diastereoselectivity
1
in nucleophilic addition reactions. The Felkin-Anh model is
the one most commonly used to understand the stereoselectivity
of nucleophilic addition to carbonyl compounds. This model
invokes contributions from torsional, steric, and electronic
effects by stereogenic centers adjacent to the reactive carbonyl
group, and it has been supported by several computational
investigations. However, the Felkin-Anh model fails to explain
the diastereoselectivity observed in nucleophilic additions to
substituted cyclohexanones (the reducing agent prefers to
This model successfully explains the observed stereoselectivity
in several cyclohexanone-based systems but fails for other
carbonyl compounds. Furthermore, the model has been heavily
criticized because of the assumption that C-H bonds have
greater electron-donating abilities than C-C bonds and the fact
that electron donation into the antibonding σ* orbital, which is
a bond-weakening interaction, is used to describe the incipient
2
†
Present address: ScinoPharm Taiwan, Ltd., One Ta-Hsueh Rd.,
Incubator of National Chen Kung University, Tainan, Taiwan 701.
(
1) (a) Kamernitzky, A. V.; Akhrem, A. A. Tetrahedron 1962, 18, 705.
b) Morrison, J. D. Asymmetric Synthesis; Academic Press: New York,
984. (c) Deslongchamps, P. Stereoelectronic Effects in Organic Chemistry;
(
2
bond-forming process.
1
Several studies have been carried out in order to probe the
validity of the assumption made by Cieplak that a C-H bond
is a better donor than a C-C bond in hyperconjugative electron
Pergamon: New York, 1983; Chapter 6. (d) N o´ rgr a´ di, M. StereoselectiVe
Synthesis; VCH: Weinheim, Germany, 1987. (e) Wigfield, D. C. Tetra-
hedron 1979, 35, 449. (f) Anh, N. G. Top. Curr. Chem. 1980, 88, 145.
(2) (a) Wu, Y.-D.; Houk, K. N. J. Am. Chem. Soc. 1987, 109, 908. (b)
4
release. Rozeboom and Houk first questioned this idea on the
Houk, K. N.; Paddon-Row, M. N.; Rondan, N. G.; Wu, Y.-D.; Brown, F.
K.; Spellmeyer, D. C.; Metz, J. T.; Li, Y.; Loncharich, R. J. Science 1986,
basis of data obtained by photoelectron spectroscopy and by
ab initio STO-3G calculations. Results obtained on methylpi-
2
31, 1108. (c) Wong, S. S.; Paddon-Row, M. N. J. Chem. Soc., Chem.
Commun. 1990, 456. (d) Wong, S. S.; Paddon-Row, M. N. Aust. J. Chem.
991, 44, 765. (e) Wu, Y.-D.; Tucker, J. A.; Houk, K. N. J. Am. Chem.
1
(3) (a) Cieplak, A. S. J. Am. Chem. Soc. 1981, 103, 4540. (b) Cieplak,
A. S.; Tait, B.; Johnson, C. R. J. Am. Chem. Soc. 1989, 111, 8447.
(4) Rozeboom, M. D.; Houk, K. N. J. Am. Chem. Soc. 1982, 104, 1189.
Soc. 1991, 113, 5018. (f) Frenking, G.; K o¨ hler, K. F.; Reetz, M. T.
Tetrahedron 1991, 43, 9005.
1
0.1021/ja9910053 CCC: $18.00 © 1999 American Chemical Society
Published on Web 07/21/1999

Reduction of Ketones
J. Am. Chem. Soc., Vol. 121, No. 30, 1999 7131
peridines indicate that axial 2-methyl substituents lower the
amine lone pair ionization potential (IP) by ∼0.26 eV, while
equatorial 2-methyl substituents lower the lone pair IP by less
than 0.1 eV. This observation suggests that C-C bonds are
better donors than C-H, assuming that the vicinal antiperiplanar
bonds stabilize the amine radical cation by hyperconjugative
electron release. These results were interpreted by Cieplak as
the consequence of ring distortion. This distortion is evident in
Both the Felkin-Anh and Cieplak models, as well as other
10
6,11
models (classical and modern molecular orbital based ) that
have been proposed to describe stereoselectivity in nucleophilic
additions of carbonyl compounds, can be classified as “gas-
phase” models. In other words, they refer to isolated reactants
and focus entirely on the intrinsic structural and electronic
features of the substrate, despite the fact that the stereochemical
outcome of the reactions of interest often exhibits a marked
sensitivity to the solvent and the type of counterion employed
2,2,6,6-tetramethylpiperidine, wherein the syn-diaxial interac-
tion of the methyl groups effectively increases the size of the
six-membered ring.3b The lowering of the ionization potential
of cyclic amines and ethers by an increase in ring size is a well-
12
with ionic and polar reducing agents.
The intrinsic effects that determine the stereochemistry of
ketone reduction reactions can be examined experimentally in
the gas phase, where extrinsic solvent and counterion effects
are absent. We introduced in a recent communication an
experimental method for distinguishing the diastereomeric
products of gas-phase hydride reduction reactions and presented
preliminary results on the quantitative determination of the
intrinsic diastereoselectivity of reductions of a few alkyl-
5
known fact. Moreover, recent studies on the hydrochlorination
6
and fluorination of 5-substituted adamantan-2-ols and confor-
mational studies of Lewis acid complexed R,â-unsaturated
7
esters seem to confirm the greater electron-donating capability
of C-H bonds over C-C bonds.
The work carried out by le Noble on hydride reductions of
5
-substituted 2-adamantanones further supports the Cieplak
1
3
8
substituted cyclohexanones. The stereochemical outcome of
gas-phase hydride reduction reactions of 4-tert-butylcyclohex-
anone, 2-methylcyclohexanone, and 3,3,5-trimethylcyclohex-
anone by pentacoordinate silicon hydride ions was found to be
remarkably similar to that reported for these substrates toward
model. These molecules show stereoselectivity opposite to what
would be predicted by the widely accepted Felkin-Anh model.
Moreover, studies of nucleophilic additions to 3-substituted
cyclohexanones (A, Y ) O) and electrophilic additions to
3
b
3
-substituted methylenecyclohexanes (A, Y ) CH2) also
3
,12
common reducing agents in solution
and also in agreement
with the diastereoselectivities predicted by MO calculations for
(
10) (a) Cram, D. J.; Abd Elhafez, F. A. J. Am. Chem. Soc. 1952, 74,
5
828. (b) Cram, D. J.; Green, F. D. J. Am. Chem. Soc. 1953, 75, 6005. (c)
Barton, D. H. R. J. Chem. Soc. 1953, 1927. (d) Dauben, W. G.; Fonken, G.
J.; Noyce, D. S. J. Am. Chem. Soc. 1956, 78, 2579. (e) Richer, J. C. J. Org.
Chem. 1964, 30, 324. (f) Marshal, J. A.; Caroll, R. D. J. Org. Chem. 1965,
3
0, 2748. (g) Karabatsos, G. J. J. Am. Chem. Soc. 1967, 89, 1367. (h)
provide results that are in full accord with the Cieplak model.
The presence of an equatorial electron-donating group at C3
was found to decrease the percentage of axial attack, while an
increase occurs for electron-withdrawing substituents. This is
consistent with the Cieplak description of the nucleophile’s
approach in a direction that is anti-periplanar to the best electron-
Ch e´ rest, M.; Felkin, H. Tetrahedron Lett. 1968, 2205. (i) Klein, J.;
Dunkelblum, E.; Eliel, E. L.; Senda, Y. Tetrahedron Lett. 1968, 6127. (j)
Klein, J. Tetrahedron Lett. 1973, 4307. (k) Klein, J. Tetrahedron 1974, 30,
3
349. (l) Wipke, W. T.; Gund, P. J. Am. Chem. Soc. 1976, 98, 8107. (m)
Wigfield, D. C.; Gowland, F. W. J. Org. Chem. 1977, 42, 1108.
11) (a) Anh, N.-T.; Eisenstein, O.; Lefour, J. M.; Tran Huu Dau, M. R.
(
J. Am. Chem. Soc 1973, 95, 6146. (b) Ahn, N.-T.; Eisenstein, O. Tetrahedron
Lett. 1976, 155. (c) Ahn, N.-T.; Eisenstein, O. NouV. J. Chim. 1977, 1, 61.
(d) Liotta, C. Tetrahedron Lett. 1975, 519-523. (e) Ashby, E. C.; Boone,
J. R. J. Org. Chem. 1976, 41, 2890. (f) Royer, J. Tetrahedron Lett. 1978,
donating bond in the ring. The interpretation of these results,
however, has been challenged by Houk,2e who has offered an
alternative explanation in terms of electrostatic effects. An
electron-withdrawing group at C3 would induce a partial positive
charge at that atom, thus making the axial approach of a
negatively charged nucleophile more favorable. The contribution
of electrostatic control to the diastereoselectivity of hydride
additions to carbonyl compounds is also evident in the π-facial
selectivity observed in 1,2-addition reactions of organometallic
1
343. (g) Cieplak, A. S. J. Am. Chem. Soc. 1981, 103, 4540. (h) Giddings,
M. R.; Hudee, J. Can. J. Chem. 1981, 59, 459. (i) Houk, K. N.; Paddon-
Row, M. N.; Rondan, N.; Wu, Y.-D.; Brown, F. K.; Spellmeyer, D. C.;
Metz, J. T.; Li, Y.; Loncharich, R. J. Science 1986, 231, 1108. (j) Wu,
Y.-D.; Houk, K. N. J. Am. Chem. Soc. 1987, 109, 908. (k) Mukherjee, D.;
Wu, Y.-D.; Fronczek, F. R.; Houk, K. N. J. Am. Chem. Soc. 1988, 110,
3328. (l) Wu, Y.-D.; Houk, K. N.; Trost, B. M. J. Am. Chem. Soc. 1987,
1
09, 5560. (m) Cieplak, A. S.; Tait, B. D.; Johnson, C. R. J. Am. Chem.
Soc. 1989, 111, 8447. (n) Kurita, Y.; Takayama, C. Tetrahedron 1990, 46,
789. (o) Wong, S. S.; Paddon-Row, M. N. J. Chem. Soc., Chem. Commun.
9
reagents to 4,4-disubstituted dienones. The results, according
3
to Wipf and Kim, can be readily explained by the Felkin-Anh
model and not by Cieplak. Interestingly, a linear correlation
was observed between the calculated (AM1) dipole moment of
some dienones and their characteristic R-face selectivity. This
finding indicates a significant contribution from electrostatic
control to the stereochemical outcome of polar additions to
sterically unhindered carbonyl groups. However, certain dif-
ferences in the π-facial selectivity between monocyclic and
spirocyclic systems cannot be fully explained on the sole basis
of electrostatic control.
1991, 327. (p) Wu, Y.-D.; Houk, K. N.; Florez, J.; Trost, B. M. J. Org.
Chem. 1991, 56, 3656. (q) Wu, Y.-D.; Tucker, J. A.; Houk, K. N. J. Am.
Chem. Soc. 1991, 113, 5018. (r) Frenking, G.; K o¨ hler, K. F.; Reetz, M. T.
Tetrahedron 1991, 47, 8991-9005. (s) Frenking, G.; K o¨ hler, K. F.; Reetz,
M. T. Angew. Chem., Int. Ed. Engl. 1991, 30, 1146. (t) Wong, S. S.; Paddon-
Row, M. N. Aust. J. Chem 1991, 44, 765. (u) Wu, Y.-D.; Houk, K. N.;
Paddon-Row, M. N. Angew. Chem., Int. Ed. Engl. 1992, 31, 1019. (v)
Coxon, J. M.; Luibrand, R. T. Tetrahedron Lett. 1993, 34, 7093-7097.
(
w) Li, H.; le Noble, W. J. Recl. TraV. Chim. Pays-Bas. 1992, 111, 199.
(x) Huang, X. L.; Dannenberg, J. J.; Duran, M.; Bertr a´ n, J. J. Am. Chem.
Soc. 1993, 115, 4024. (y) Huang, X. L.; Dannenberg, J. J. J. Am. Chem.
Soc. 1993, 115, 6017. (z) Shi, Z.; Boyd, R. J. J. Am. Chem. Soc. 1993,
115, 9614. (aa) Williams, L.; Paddon-Row, M. N. J. Chem. Soc., Chem.
Commun. 1994, 353. (bb) Gung, B. W.; Wolf, M. A.; Mareska, D. A.;
Karipides, A. J. Org. Chem. 1994, 59, 4899. (cc) Coxon, J. M.; Houk, K.
N.; Luibrand, R. T. J. Org. Chem. 1995, 60, 418. (dd) Gung, B. W.; Yanik,
M. M. J. Org. Chem. 1996, 61, 947.
(12) (a) Boone, J. R.; Ashby, E. C. In Topics in Stereochemistry; Allinger,
N. L., Eliel, E. L., Eds.; Interscience: New York, 1979. (b) Wigfield, D.
C. Tetrahedron 1979, 35, 4449. (c) Haubenstock, H.; Eliel, E. L. J. Am.
Chem. Soc. 1962, 84, 2363. (d) Fort, Y.; Feghouli, A.; Vanderesse, R.;
Caub e` re, P. J. Org. Chem. 1990, 55, 5911.
(
5) (a) Yoshikawa, K.; Hashimoto, M.; Morishima, L. J. J. Am. Chem.
Soc. 1974, 96, 288. (b) DeLoth, P. C. R. Acad. Sci. Ser. C 1974, 279C,
31.
3
(
(
(
6) Adcock, W.; Cotton, J.; Trout, N. A. J. Org. Chem. 1994, 59, 1867.
7) Gung, B. W.; Yanik, M. M. J. Org. Chem. 1996, 61, 947.
8) (a) Cheung, C.-K.; Tseng, L.-T.; Lin, M.-h.; Srivastava, S.; le Noble,
W. J. J. Am. Chem. Soc. 1986, 108, 1598. (b) Li, H.; le Noble, W. J.
Tetrahedron Lett. 1990, 31, 4391.
(9) Wipf, P.; Kim, Y. J. Am. Chem. Soc. 1994, 116, 11678.
(13) Ho, Y.; Squires, R. R. J. Am. Chem. Soc. 1992, 114, 10961.

7132 J. Am. Chem. Soc., Vol. 121, No. 30, 1999
Artau et al.
reduction by LiH in the gas phase.2a In this paper, we present
a full description of the method as applied to the study of gas-
phase hydride reduction reactions of several cyclohexanone
derivatives. These results are contrasted to those obtained for
cyclic ketones with ring heteroatoms, whose gas-phase hydride
reduction diastereoselectivity has been predicted to be opposite
to that observed in solution.27
silanes has been reported previously (eq 1).13,18 These species
are reactive reducing agents, capable of transferring hydride to
transition metal carbonyls, CO2, boranes, and SiH4.5
,19,20
Al-
dehydes and ketones are reduced by net addition of Si-H bond
across the carbonyl group, producing an alkoxysiliconate ion.
An example of this process is given in eq 2. The structural
assignment of the product of eq 2 was made by direct
comparison of the collision-induced dissociation (CID) spectrum
Experimental Section
-
of the adduct obtained from the reaction of BuSiH4 with
-
All of the gas-phase experiments described in this paper were
performed at room temperature in a flowing afterglow-triple quadru-
pole apparatus that has been described in detail previously.14 The total
pressure and flow rate of the helium buffer gas in the 1 m × 7.3 cm
3-pentanone to that of an authentic BuSiH3(OCH(CH2CH3)2)
ion produced by direct addition of (CH CH ) CHO to BuSiH .
3 2 2 3
-
13
-
-
4
H + RSiH f RSiH
(1)
3
3
flow reactor were 0.4 Torr and 200 STP cm /s, respectively, with a
bulk flow velocity of 9700 cm/s. Hydroxide ions were generated in
the ion source region of the flow reactor by electron ionization of a
BuSiH4- + (CH CH ) CdO f
3
2 2
N
2
O/CH
4
mixture. Fluoride ions were generated by electron ionization
-
-
BuSiH (OCH(CH CH ) ) r (CH CH ) CHO + BuSiH
3
2
3 2
3
2 2
3
of NF . Reagent ions formed in the source region were transported by
3
(
2)
the flowing helium through the reactor, where they were allowed to
react with the gaseous neutral reagents introduced via leak valves. The
alkoxide ions were formed by deprotonation of the corresponding
RO-
R CdO
2
-
-
RSiH3
8 RSiH (OR)
3
8 BuSiH (OR)(OCH(Et) )
2 2
-
alcohols by HO . Those arising from cis- and trans-2-tert-butyl-1,3-
-
(3)
dithian-5-ol were generated by deprotonation with F . The alkoxide
ions were used in the formation of monoalkoxysiliconate ions by
addition to a neutral silane in the flow reactor. Subsequent addition of
a ketone formed the dialkoxysiliconates. Proton-bound clusters were
formed by addition of the alkoxides to alcohols that were introduced
Monoalkoxysiliconate ions, formed either by addition of an
alkoxide to a neutral silane or by reduction of an aldehyde or
ketone by a pentacoordinate silicon hydride, can further reduce
aldehydes and ketones, producing a pentacoordinate siliconate
ion bearing two alkoxy groups (dialkoxysiliconates, e.g., eq 3).
In fact, we have observed up to three consecutive reductions
downstream. The ions were thermalized to ambient temperature by ca.
5
1
0 collisions with the helium buffer gas. Negative ions were extracted
from the flow reactor through a 0.5 mm orifice in a nose cone and
then focused into an EXTREL triple quadrupole analyzer for mass
spectrometric analysis.
-
-
by RSiH4 to produce a trialkoxysiliconate, (RO)3SiHR .
Dialkoxysiliconate ions formed by the process shown in eq 3
provide the key to determine the diastereoselectivity of the
reduction of ketones in the gas phase. Upon collisional activa-
tion, dialkoxysiliconates undergo competitive dissociation reac-
tions involving loss of either of the alkoxide ligands. The relative
yield of the alkoxide fragments appears to be a sensitive function
of their relative basicities. This general behavior is analogous
Collision-induced dissociation experiments were perfomed in the
triple quadrupole mass analyzer using argon as the target gas. The target
pressure in the central, gas tight quadrupole collision chamber (Q2)
was maintained in the 0.08-0.25 mTorr range, which corresponds to
1
5
multiple collision conditions. Collision energies were in the 15-20
eV range (laboratory frame). Mass-analyzed ions were detected with
an electron multiplier. The voltage bias of the third quadrupole (Q3)
was adjusted to optimize product ion collection, and the Q3 mass
resolution and other tuning conditions of the triple quadrupole analyzer
were adjusted to achieve maximum reproducibility ((10% absolute)
in the quantitative measurements of the CID product yield ratios. All
intensities were recorded using a digital counter operating at a 10 s
gate time. The ion intensity measurements, which follow a Poisson
distribution, represent at least 10 replicate measurements. Therefore,
the uncertainty in each counter measurement was calculated by taking
the square root of the intensity. Uncertainties in the yield ratios were
determined by using standard error propagation procedures.16
-
-
+
to that of proton-bound alkoxide clusters, (RO )(R′O )H ,
which can also undergo competitive alkoxide cleavages with
yields reflecting the relative proton affinities of the alkoxide
21
fragments. However, the decomposition of dialkoxysiliconates
also appears to reflect their structures and not just their relative
basicities. A good example of this sensitivity is the dialkoxy-
-
siliconate ion BuSiH2(OCHMe2)(OBu) , formed by the direct
-
addition of Me2CHO to BuSiH3, followed by addition to
butanal. CID of this ion with an argon target at a collision energy
-
of 12 eV (lab frame) yields the alkoxide fragments Me2CHO /
All dioxanone and dithianone derivatives were synthesized using
-
1
7
BuO in the ratio 1.51 ( 0.15. Under similar conditions, CID
literature methods. The compounds were purified according to
standard procedures. All other reagents were obtained from commercial
sources and used as supplied except for degassing of liquid samples
-
-
+
of the corresponding proton-bound dimer (Me2CHO )(BuO )H
gives the alkoxide fragments in essentially identical yields (Me2-
-
-
prior to use. Gas purities were as follows: He (99.995%), SiH
99.995%), N O (99%), CH (99%), NF (99%).
4
CHO /BuO ) 1.00 ( 0.01), a result that reflects their identical
2
2
(
2
4
3
Brønsted basicities. Thus, the secondary alkoxide is prefer-
entially cleaved from the siliconate ion, presumably because of
steric repulsion effects that weaken the Si-OCHMe2 bond. This
behavior is what enabled us to distinguish isomeric alkoxide
ions with similar basicities.
Results
The generation of pentacoordinate silicon hydride ions in a
flowing afterglow instrument by addition of hydride to alkyl-
Dialkoxysiliconate ions are presumed to exhibit a trigonal-
bipyramidal geometry. The ligands undergo facile positional
exchange either in the long-lived ions or during collisional
(
14) (a) Graul, S. T.; Squires, R. R. Mass Spectrom. ReV. 1988, 7, 1. (b)
2
3
Marinelli, P. J.; Paulino, J. A.; Sunderlin, L. S.; Wenthold, P. G.; Poutsma,
J. C.; Squires, R. R. Int. J. Mass Spectrom. Ion Processes 1994, 130, 89.
(
(
15) Graul, S. T.; Squires, R. R. J. Am. Chem. Soc. 1990, 112, 2506.
16) Bevington, P. R. In Data Reduction and Error Analysis for the
(18) Hajdasz, D. J.; Squires, R. R. J. Am. Chem. Soc. 1986, 108, 3139.
(19) Lane, K. R.; Sallans, L.; Squires, R. R. Organometallics 1985, 3,
408.
Physical Sciences; McGraw-Hill: New York, 1969.
17) cis-2-tert-Butyl-1,3-dioxan-5-ol and trans-2-tert-butyl-1,3-dioxan-
(
5
-ol: Weclas, L.; Sokolowski, A.; Burczyk, B. Pol. J. Chem. 1982, 56,
85. 2-tert-Butyl-1,3-dioxan-5-one: Majewski, M.; Gleave, D. M.; Nowak,
(20) Workman, D. B.; Squires, R. R. Inorg. Chem. 1988, 27, 1846.
(21) Cooks, R. G.; Patrick, J. S.; Kotiaho, T.; McLuckey, S. A. Mass
Spectrom. ReV. 1994, 13, 287.
(22) Lias, S. G.; Bartmess, J. E.; Liebmann, J. F.; Holmes, J. L.; Levin,
R. D.; Mallard, W. G. J. Phys. Chem. Ref. Data 1988, 17 (Suppl. 1).
4
P. Can. J. Chem. 1995, 73, 1616. 2-tert-Butyl-1,3-dithian-5-one and cis-
and trans-2-tert-butyl-1,3-dithian-5-ol: Eliel, E. L.; Juaristi, E. J. Am. Chem.
Soc. 1978, 100, 6114.

Reduction of Ketones
J. Am. Chem. Soc., Vol. 121, No. 30, 1999 7133
Scheme 1
Table 1. Diastereomer Yield Ratios in Gas-Phase Hydride Reduction Reactions of Cyclic Ketones (Obtained by CID of the
Dialkoxysiliconates)
-
- a
CID product ratio, RO /R′O
% axial (exo) reduction
ketone reduction
product
this work
(gas phase)
lit.
ketone
cis (exo)
trans (endo)
(solution)b
theoryc
1
2
3
4
5
4-tert-butylcyclohexanone
2-methylcyclohexanone
3,3,5-trimethylcyclohexanone
norcamphor
7.0 ( 0.3
1.43 ( 0.02
1.44 ( 0.05
11.1 ( 0.8
0.26 ( 0.01
0.35 ( 0.03
1.45 ( 0.02
1.82 ( 0.03
7.8 ( 0.4
0.27 ( 0.01
0.87 ( 0.04
99 ( 3
68 ( 5
9 ( 3
93 ( 7
46 ( 3
92, 85
76, 70
21, 42
91, 86
93, 94
88
82
30
3.21 ( 0.02
1.39 ( 0.02
0.42 ( 0.03
1.76 ( 0.05
2-tert-butyl-1,3-dioxan-5-one
3
a
6 2
Measured yield ratio of alkoxide ion fragments from CID of C H13SiH
(OR)(OR′)- ions; R′O- ) 3-pentoxide for ketones 1-3,
-
-
4
-fluorophenethoxide for ketone 5. For cis: RO ) pure cis isomer of reduced ketone (1-5). For trans: RO ) pure trans isomer of reduced
- - b
ketone (1-5). Ketone reduction product: RO formed by reduction of ketone (1-5) by C
6
H13SiH
2
(OR′) . Reported yield of axial reduction
in ether, and NaBH in 2-propanol (5, ref 29).
(5) based on ab initio (MP2/6-31G*//3-21G) molecular orbital
product formed with LiAlH in tetrahydrofuran and NaBH
4
4
in 2-propanol (1-4, ref 12), LiAlH
4
4
c
-
5
Predicted yield of axial reduction product formed with LiH (1-4) and SiH
calculations (refs 2 and 27).
activation. This is demonstrated by the equivalence of the CID
spectra obtained for dialkoxysiliconate ions produced by the
two complementary alkoxide addition/ketone reduction se-
quences (eq 2). For instance, attachment of the cyclohexyl-
alkoxide ion to RSiH3 (R ) n-hexyl) followed by reduction of
alkoxide group for this particular set of experiments after
extensive screening of alkoxide/ketone pairs for a system that
gave measurable and reliable fragment ion yield ratios. Genera-
tion of 3-pentoxide by proton abstraction, followed by attach-
ment to RSiH and subsequent reduction of 2-methylcyclohex-
3
3
-pentanone gives a dialkoxysiliconate ion that yields a CID
anone with this adduct, produces the dialkoxysiliconate whose
spectrum identical to that obtained for the ion formed by
addition of the 3-pentoxide ion to RSiH3 followed by reduction
of cyclohexanone. Thus, the two reaction sequences yield
dialkoxysiliconate ions with a common structure or mixture of
structures, presumably via rapid Berry pseudorotation of the
fluxional pentacoordinate silicon ion,24 regardless of the order
in which the alkoxy ligands become attached.
stereochemical structure was to be determined (2M). CID of
2M by using an argon target (0.10 mTorr) and 20 eV collision
-
energy (lab frame) yields the two alkoxides in a ratio (C7H13O /
-
C5H11O ) of 1.82 ( 0.03. Subsequently, calibration experiments
were carried out in which diastereomerically pure cis- and trans-
2
-methylcyclohexyl alkoxide ions were generated in the flow
reactor by proton abstraction from the pure alcohols. Addition
of RSiH3 (R ) n-hexyl) to form the monoalkoxysiliconate,
followed by reduction of 3-pentanone with this ion, yields the
dialkoxysiliconate ions 2C and 2T. CID of mass-selected 2C
and 2T under the same conditions used for the experiments with
Given these general characteristics of the bimolecular and
unimolecular reactions of pentacoordinate silicon hydride ions,
we recently13 reported the following strategy for determining
the stereochemical outcome of the gas-phase reduction of 4-tert-
butylcyclohexanone (1), 2-methylcyclohexanone (2), and 3,3,5-
trimethylcyclohexanone (3), substrates commonly examined in
condensed-phase studies of carbonyl reduction stereo-
2
M gives the two alkoxide ion fragments in yield ratios
-
-
(C7H13O /C5H11O ) of 3.21 ( 0.02 for 2C and 1.44 ( 0.05
25
for 2T (Table 1). Deconvolution of the diastereomeric mixture
ratio from these data was then carried out using the algebraic
expression shown in Scheme 1. The final percentages are 68 (
1
0-12
chemistry.
An illustration of the experimental protocol is
given in Scheme 1. 3-Pentoxyl was selected as the auxiliary
5
% trans and 32 ( 5% cis ((1 standard deviation). Thus,
(
23) Tandura, S. N.; Voronkov, M. G.; Alekseev, N. V. Top. Curr. Chem.
986, 131, 99.
24) Windus, T. L.; Gordon, M. S.; Burggraf, L. W.; Davis, L. P. J. Am.
Chem. Soc. 1991, 113, 4356 and references therein.
1
(
(25) The absolute total cross sections for the decomposition of 2C, 2T,
and 2M were found to be essentially identical at 12 ( 3 Å.

7134 J. Am. Chem. Soc., Vol. 121, No. 30, 1999
Artau et al.
Table 2. Diastereomer Yield Ratios in Gas-Phase Hydride Reduction Reactions of Cyclic Ketones (Obtained by CID of the Proton-Bound
Clusters)
-
O- a
CID product ratio, R
trans
1
O /R
2
% axial reduction
ketone reduction
product
this work
(gas phase)
lit.
(solution)
ketone
cis
theoryb
5
6
2-tert-butyl-1,3-dioxan-5-one
2-tert-butyl-1,3-dithian-5-one
0.43 ( 0.06
0.28 ( 0.04
8.3 ( 0.2
2.6 ( 0.1
0.98 ( 0.04
0.43 ( 0.03
33 ( 4
16 ( 4
93, 94c
15
3
0
d
a
-
+
-
-
Measured yield ratio of alkoxide ion fragments from CID of [R
1
O ‚‚‚H ‚‚‚ OR
2
] ions. R
1
O ) 2,2-difluoroethanol for 5, 2,2,2-trifluoroethanol
-
-
for 6. For cis: R
O
2
O
) pure cis isomer of reduced ketone. For trans: R
2
O
) pure trans isomer of reduced ketone. Ketone reduction product:
-
-
b
-
R
2
formed by reduction of ketone by C
6
H13SiH
4
.
Predicted yield of axial reduction product formed with SiH
5
based on ab initio molecular
in 2-propanol
c
orbital calculations (MP2/6-31G*//3-21G) (ref 27). Reported yield of axial reduction product formed by LiAlH
ref 29). Reported yield of axial reduction of 1,3-dithian-2-phenyl-5-one by LiAlH in ether (ref 29).
4
4
in ether and NaBH
4
d
(
reduction of 2-methylcyclohexanone by a pentacoordinate
silicon hydride ion in the gas phase occurs mainly from the
axial direction to give the more stable trans product.
1,3-dioxan-5-one (5) and 2-tert-butyl-1,3-dithian-5-one (6). The
tert-butyl group at C-2 makes the ring structures of 5 and 6
rigid, a property that is essential for distinguishing the products
resulting from axial reduction from those arising from equatorial
approach of the reducing agent. However, the results of these
experiments can be applied to 1,3-dioxan-5-one and 1,3-dithian-
5-one because the tert-butyl group is too removed to exert any
steric influence at the reaction center. In addition, its inductive
and field effects should be minimal.
Use of analogous procedures with the same 3-pentoxide/3-
pentanone auxiliaries for 1, 3, and 4 leads to the results shown
in Table 1. Reduction occurs almost entirely from the axial
direction for compound 1, a rigid substrate representing a
relatively unhindered chair cyclohexanone. In contrast, nearly
exclusive equatorial reduction occurs for ketone 3, wherein the
axial face of the carbonyl group is effectively blocked by the
axial methyl at C-3. The complete inversion in the preferred
mode of attack between 1 and 3 illustrates the intrinsic
preference for axial addition to unhindered cyclohexanones, even
by a relatively bulky reducing agent such as an alkoxysiliconate
ion.26 The relatively greater yield of cis product for 2 compared
to 1 is believed to be due, in part, to axial reduction of the
higher energy conformer with the methyl group in the axial
position. Ketone 4, containing a rigid bicyclic structure with
well-defined hindered (endo) and unhindered (exo) faces,
undergoes a hydride reduction primarily from the exo direction,
producing the less stable endo isomer of the product. The results
obtained for ketones 1-3 (Table 1) are in good agreement with
In contrast to all of the experiments discussed above,
3-pentoxyl cannot be used as the auxiliary group (R′O) for the
dioxanone system because the gas-phase basicity of 3-pentoxide
is greater than that of the cis and trans alkoxide ions corre-
sponding to 5. Survey experiments with several different systems
-
eventually pointed to CH CH(Ar)O (Ar ) 4-fluorophenyl) as
3
a suitable partner alkoxide. The corresponding ketone, 4-fluo-
roacetophenone, has enantiotopic faces. This means that a
racemic mixture of monoalkoxysiliconates is formed upon
reduction of the dioxanone. However, since the carbonyl carbon
in the substrate is not prochiral, only one pair of diastereomeric
products is generated. Hence, the values obtained for R and Rt
c
reflect only the differences in the structures of the alkoxides
2
theory. No theoretical predictions have been published on the
that correspond to the ketone of interest and are not dependent
-
diastereoselectivity of the hydride reduction of 4.
on the chirality of the partner alkoxide CH CH(Ar)O . The final
3
The success of the method depends critically on two factors:
1) the auxiliary alkoxide RO fragment and the other alkoxide
percentage obtained for the reduction of 5 (46 ( 3% axial
reduction, Table 1) is in disagreement with the theoretical
prediction (97% equatorial reduction). However, the results
reflect a trend predicted by theory, i.e., less axial reduction than
that observed for reduction of 4-tert-butylcyclohexanone.
A problem was encountered when examining ketone 6: the
CID spectra obtained in the calibration experiments (Scheme
1) did not yield significantly different values for R and R . In
other words, the relative yields of the two alkoxide fragments
are not sensitive to the differences in the structure of the two
diastereomeric pentacoordinate silicon ions. To enhance ste-
reodifferentiation in the CID process, a series of different
alkylsilanes (n-hexylsilane, diethylsilane, tert-butylsilane) were
examined. An increase in the steric bulk around the pentacor-
dinate silicon bridge was expected to promote the dissociation
of the alkoxy ligands to different extents for the two diastere-
omers. However, these values remained essentially the same
for all alkylsilanes used. Therefore, the dialkoxysiliconate
approach, while successful for the dioxanone system, did not
seem to work for the analysis of the reduction of 6 and had to
be abandoned.
-
(
fragment must be observable upon CID so that the ratios can
be reliably measured; (2) the CID fragment ion ratios measured
for the two pure diastereomers in the calibration experiments
must be as different as possible in order to minimize quantitative
uncertainty in the mixture analysis; (3) the absolute cross
sections for the decomposition of the diastereomeric dialkoxy-
siliconates, e.g., 2C and 2T, have to be similar. These
requirements actually became an obstacle in the studies involv-
ing 2-tert-butyl-1,3-dithian-5-one (6), as discussed below.
Recently, Houk et al. made some provocative predictions
c
t
-
regarding the preferred mode of hydride delivery by SiH5 to
1
,3-dioxan-5-one and 1,3-dithian-5-one in the gas phase. An
almost exclusive equatorial approach of the reducing agent was
calculated for these substrates.27 This result contrasts with the
behavior observed in solution, where 94% of the reduction by
12a,28,29
LiAlH4 in ether occurs from the axial direction (Table 2).
To probe the intrinsic reactivity, we determined the diastereo-
selectivity of the gas-phase hydride reduction of 2-tert-butyl-
(26) Extremely bulky reducing agents in solution, such as the trialkyl-
In the course of these studies, however, it became evident
that 6 has a sufficiently high hydride affinity to react with
monoalkylsilicon hydride ions through an overall (bimolecular)
hydride transfer to produce the free alkoxide (eq 4). This
observation, which was also noted in earlier studies,30 im-
borohydrides, display a preference for equatorial attack with 1: Smith, K.;
Pelter, A.; Norbury, A. Tetrahedron Lett. 1991, 32, 6243.
(
27) Wu, Y.; Houk, K. N. J. Am. Chem. Soc. 1993, 115, 10993.
(28) Senda, Y.; Teresawa, T.; Ishiyama, J.; Kamiyama, S.; Imaizumi, S.
Bull. Chem. Soc. Jpn. 1989, 62, 2948.
29) (a) Kobayashi, Y. M.; Lambrecht, J.; Jochims, J. C.; Burkert, U.
(
Chem. Ber. 1978, 111, 3442. (b) Jochims, J. C.; Kobayashi, Y.; Skrzelewski,
E. Tetrahedron Lett. 1974, 571, 575.
(30) Ho, Yeunghaw, Ph.D. Thesis, Purdue University, 1995.

Reduction of Ketones
J. Am. Chem. Soc., Vol. 121, No. 30, 1999 7135
mediately suggested a solution to the lack of stereodifferentiation
of CID of dialkoxysiliconates: the use of proton-bound clusters.
Proton-bound clusters21 (R1O ‚‚‚H ‚‚‚ OR2) are readily gener-
-
+
-
ated in the flow reactor by the (termolecular) addition of an
alkoxide R1O to an alcohol R2OH. Examination of CID of
-
these clusters allows the stereochemical analysis to be carried
out as described above. For these experiments, 2,2,2-trifluoro-
ethanol was chosen as the auxiliary alcohol, since its proton-
bound clusters synthesized in the flow reactor exhibited excellent
stereodifferentiation upon CID (Rc ) 0.28 ( 0.04, Rt ) 2.6 (
-
-
-
+
Figure 1. Partial CID spectra for (C
2
H
3
F
2
2
O )(cis-C
8
H
H
15
O
O
3
)H ,
-
-
+
-
+
(C H
2 3
F
2
O )(trans-C
H O
8 15 3
)H , and (C
2
H
3
F
O )(mix-C
8
15
3
)H .
0
.1) [the gas-phase acidity (∆Hacid) of trans-2-tert-butyl-1,3-
dithian-5-ol is about 6 kcal/mol lower than that of cis-2-tert-
butyl-1,3-dithian-5-ol ].
(
∼0.2% natural abundance) generated from the protolytic
31
cleavage of R-protons of 5 to the total intensity of the parent
ion selected for CID, the correction made for the analysis of
the reduction of 6 was also employed for this system. The final
percentage obtained was 33 ( 4% for axial reduction of 5 (Table
This new approach seemed to effectively solve the obstacles
encountered in the calibration experiments described above for
pure cis- and trans-2-tert-butyl-1,3-dithian-5-ol. However, it
introduced a new problem. In addition to overall hydride
transfer, protolytic cleavage of an R-proton occurs with the
dithianone (6) upon reaction of the pentacoordinate silicon
hydride ions to produce an enolate, 7 (eq 5), instead of the
2
), a value that is well above the theoretical prediction. However,
this result demonstrates a dramatic reversal of the behavior
observed in solution, in agreement with theory.
Discussion
Among the alkyl-substituted cyclohexanones studied, 1 is of
particular interest, since it has the least flexible ring structure
and the least hindered axial face for hydride addition to the
carbonyl group. The tert-butyl group is locked in an equatorial
position and is spatially removed from the reaction center.
Therefore, the gas-phase diastereoselectivity of 1 should ef-
fectively represent the ratio of axial and equatorial attack on
the chair conformation of cyclohexanone in the absence of
solvent, i.e., the intrinsic diastereoselectivity. The 99 ( 3% value
obtained for the axial reduction of 1, corresponding to formation
of the more stable trans isomer of the product, is in excellent
agreement with the behavior observed for this compound in
solution and consistent with the theoretical prediction of a
preferred axial reduction in the gas phase (Table 1). Torsional
strain has been proposed to be the dominant factor in the strong
preference for axial attack observed for 1 in solution.3
The diastereoselectivity observed for the reduction of 2 in
the gas phase (68 ( 5% axial reduction) is also consistent with
the reported condensed-phase behavior of this compound toward
the common reducing agents LiAlH4 and NaBH4. However,
theory predicts a somewhat greater amount of axial reduction
for LiH in the gas phase (82%, Table 1) than what was measured
in our experiments. We believe that this difference is due, in
part, to the fact that only the conformer with the methyl group
in the equatorial position was considered in the theoretical study
desired free alkoxide. This product has a mass-to-charge ratio
that is two units less than that of the corresponding alkoxide.
The enolate (7) forms a proton-bound cluster with 2,2,2-
trifluoroethanol, just like the desired alkoxide produced by
reduction of 6 (eq 4). Furthermore, the proton-bound cluster
containing the enolate moiety (with a 34S atom, ∼4% natural
abundance) produces fragment ions at the same mass-to-charge
value as the proton-bound cluster containing the alkoxide moiety
(
with two 32S atoms), thus introducing an error in the value
obtained for Rm (Scheme 1). This problem was addressed by
determining the percentage of enolate-containing proton-bound
clusters that undergo CID along with the alkoxide-containing
proton-bound clusters (by making use of the 3 S natural
abundance and monitoring the intensity of enolate-containing
proton-bound clusters in the flow reactor with two 3 S atoms)
and by independently measuring the absolute collision cross
section of the enolate-containing proton-bound cluster. After
making the appropriate subtraction to the fragment ion intensities
in the CID spectra, we obtained a value of 16 ( 4% for axial
2,33
4
2
-
reduction of 6 by n-C6H13SiH4 . This result agrees well with
2
9
theory (Table 2).
To make a direct comparison between the dithianone and
dioxanone systems, it was necessary to reanalyze the dioxanone
system by using the proton-bound cluster approach. This
comparison exposes any possible biasing of the final results
due to the use of different reducing agents [alkylsilicon hydride
(Table 1). At ambient temperature, 2 is reported to exist as a
mixture of conformers composed of 5% of the high-energy
conformer with the methyl group in the axial position and 95%
of the low-energy conformer with the methyl group in the
3
4
equatorial position. The diastereoselectivity observed for the
reaction of cis-4-tert-butyl-2-methylcyclohexanone (in which
the 2-methyl group is fixed in an equatorial position) and LiAlH4
in tetrahydrofuran (83% axial reduction) supports this hypoth-
(eq 5) vs monoalkoxysiliconate (eq 4)] or different types of
parent ions in the diastereomeric mixture analysis (dialkoxy-
siliconates vs proton-bound clusters). 2,2-Difluoroethanol was
chosen as the auxiliary alcohol for the reexamination of the
reduction of 5. CID of the proton-bound clusters of this alcohol
produced measurable ion yield ratios with excellent stereodif-
ferentiation (Figure 1). To eliminate the contribution of the
1
2
esis.
Electronic, torsional, and steric effects have been proposed
to explain the stereoselectivity in the hydride reduction of cyclic
1
8
enolate-containing proton-bound clusters with an O atom
(32) Ch e´ rest, M.; Felkin, H.; Prudent, N. Tetrahedron Lett. 1968, 2199.
(33) Klein, J. Tetrahedron Lett. 1973, 4307.
(31) Artau, A.; Squires, R. R. Manuscript in preparation.
(34) Coxon, J. M.; Luibrand, R. T. Tetrahedron Lett. 1993, 34, 7097.

7136 J. Am. Chem. Soc., Vol. 121, No. 30, 1999
Artau et al.
ketones.11 The torsional component is believed to overwhelm
the steric component as a control factor in systems in which
axial hydride addition dominates, such as in 1 and 2. However,
hydride addition to alkyl-substituted cyclohexanones from the
equatorial direction becomes favorable if a large steric hindrance
exists for axial attack. Ketone 3, in which the axial face of the
carbonyl group is effectively blocked by the axial methyl group
at C-3, is ideal for investigating this effect. Experimental results
obtained in solution as well as theoretical predictions for the
gas phase indicate that hydride addition to 3 occurs from the
equatorial rather than the axial direction. The 9% axial reduction
observed for 3 in the gas phase, which represents a complete
inversion in the preferred mode of attack when compared to 1,
verifies that the measured diastereomer ratios are kinetically
determined in these experiments (cis-3,3,5-trimethylcyclohex-
anol is more stable than the trans isomer by ca. 1.5 kcal/mol,
and an even larger energy difference is expected for the
dialkoxysiliconate ion reduction products of 3).39 The preference
for the axial reduction of 3 observed in this study is, however,
somewhat less than that predicted by theory (Table 1). This
inconsistency can be explained by differences in the steric effects
considered in the theoretical studies and those present in the
real system. LiH is the reducing agent in the theoretical study,
a species that is much smaller than the monoalkoxysiliconate
substrates in solution toward common reducing agents in, for
example, LiAlH4 and NaBH4, and with the diastereoselectivities
predicted by MO calculations for reduction by LiH in the gas
phase (Table 1). The observation of the same diastereoselectivity
in the gas phase and in solution implies that extrinsic factors,
such as specific solvation, ion-pairing, and/or metal ion coor-
dination effects, need not necessarily be invoked; i.e., that
diastereoselectivity of these ketones is a property ascribed to
intrinsic characteristics of the isolated reactants.
Wu and Houk predicted a strong departure in the stereo-
chemical outcome between the gas phase and solution for the
- 27
hydride reduction of 1,3-dioxan-5-one by SiH5 . The basis
for this prediction is the calculated strong electrostatic repulsion
upon attack of the hydride reducing agent from the axial face
of the dioxanone. The intrinsic controlling factor in this
reduction is repulsion between the nucleophilic hydride and the
ring heteroatoms, which overwhelms the torsional effect as-
sociated with equatorial attack (in solution, this repulsion is
attenuated by the presence of solvent). Furthermore, owing to
the short C-O bonds in the ring, the six-membered ring of the
dioxanone is flatter than that of cyclohexanone, increasing the
torsional strain associated with equatorial attack of the hydride
nucleophile relative to cyclohexanone. Hence, the diastereose-
lectivity observed for reduction of 1,3-dioxan-5-one by LiAlH4
in solution is one in which axial attack occurs to a greater extent
-
C6H13SiH3(OCHEt2) .
Norcamphor (4), which contains a rigid bicyclic structure, is
another example of a ketone with well-defined hindered and
unhindered faces. However, reduction of 4 with LiAlH4 in ether
is highly selective in solution. Reduction (91%) occurs from
the exo direction in solution (Table 1), which is less sterically
hindered. No theoretical predictions have been published on the
diastereoselectivity of the hydride reduction of 4. The gas-phase
results (93 ( 7% exo reduction) are consistent with the reduction
of 4 by LiAlH4 in tetrahydrofuran (Table 2). Compound 4
exhibits a behavior similar to that of 3; i.e., the hydride
29
than in cyclohexanone. In the gas phase, however, the
electrostatic repulsion between the nucleophilic hydride and the
ring heteroatoms overwhelms the torsional effect, and the
reduction is predicted to occur almost exclusively (97%) from
the equatorial direction. The 46 ( 3% axial reduction obtained
here for the gas-phase reduction of the tert-butyl derivative (5)
is inconsistent with the theoretical prediction. This inconsistency
may arise from the different reducing agent employed in the
-
-
two studies (SiH5 vs C6H13SiH3OC8H8F ). The monoalkoxy-
siliconate ion has an electron-withdrawing group attached to
the Si atom, which may result in less electron density on the
3
5
approaches 4 from the least hindered direction. Steric effects,
torsional interaction,36 and frontier orbital interactions have
been used to rationalize the outcome of the hydride reduction
of 4. It appears that steric effects overwhelm the torsional effects
as a control factor in the reduction of 4 by a monoalkoxysili-
conate ion in the gas phase (DePuy et al. experimentally
determined that there is no difference in the intrinsic thermo-
dynamic stability of its reduction products, gaseous exo-
norborneol and endo-norborneol ). The steric repulsion results
from the interaction between the reducing agent and the endo
hydrogens at C-5 and C-6. Alternatively, examination of the
LUMO arising from hyperconjugation of the â C-C σ* orbital
with the carbonyl π* orbital (B) reveals a distortion of the vacant
37
-
hydrides compared to SiH5 . Therefore, less electrostatic
repulsion between the monoalkoxysiliconate ion and the ring
-
oxygens may be present than for SiH5 . This expectation is
supported by the results obtained in the gas-phase experiments
involving proton-bound clusters as the substrates for CID (Table
2), which reflect a decrease of 13% in axial reduction compared
to the systems in which monoalkoxysiliconates are used as
reducing agents (Table 1). In the proton-bound clusters, the Si
atom in the reducing agent does not contain an electron-
withdrawing group. Therefore, the electron density on the
hydrides may be greater than in monoalkoxysiliconates, resulting
in an increased electrostatic repulsion between the reducing
agent and the ring oxygen atoms and, therefore, a reduced
percent axial reduction. Finally, Houk’s prediction does not
include any interactions between the Si atom and the carbonyl
group oxygen. In the real system, a net addition of an Si-H
bond occurs across the carbonyl group. The carbonyl oxygen
may slightly stabilize the transition state for axial attack by
interacting with the electropositive Si atom. The change in
selectivity relative to 1, however, is in the predicted direction.
38
π* orbital that favors exo attack of the hydride reducing agent,
in agreement with our results in the gas phase and the observed
diastereoselectivity of hydride reductions in solution (Table 1).
The gas-phase stereochemical results obtained for ketones
The preferred mode for hydride addition to 1,3-dithian-5-
one is predicted to occur from the equatorial direction.27 As in
the case of 1,3-dioxan-5-one, this preference can be explained
by electrostatic repulsion between the incoming negative hydride
nucleophile and the lone pairs of the two sulfur atoms in the
ring. However, the dithianone should exhibit an even larger
preference for equatorial reduction than the dioxanone due to
torsional effects. The long C-S bonds and small C-S-C angles
1-4 are generally consistent with the reported behavior of these
(35) Giddings, M. R.; Hudge, J. Can. J. Chem. 1981, 59, 459.
(36) Xie, M.; le Noble, W. J. J. Org. Chem. 1989, 54, 3836.
(37) Ashby, E. C.; Boone, J. R. J. Org. Chem. 1976, 41, 2890.
(38) DePuy, C. H.; Story, P. R. J. Am. Chem. Soc. 1960, 82, 627.
(39) Eliel, E. L.; Gilbert, E. C. J. Am. Chem. Soc. 1969, 91, 5487.

Reduction of Ketones
J. Am. Chem. Soc., Vol. 121, No. 30, 1999 7137
cause the ring to be significantly puckered relative to cyclo-
hexanone, favoring equatorial addition.27 This trend is indeed
what was observed in the gas phase. Interestingly, the gas-phase
stereochemical results are in excellent agreement with the
observed diastereoselectivity for reduction by LiAlH4 in solution
) 99 ( 3, 68 ( 5, 9 ( 3, and 93 ( 7 for 1, 2, 3, and 4,
respectively] are readily rationalized in terms of competition
between steric and torsional effects. Hence, concepts of hyper-
conjugative electron donation from the anti-periplanar σ C-H
orbitals of the cyclohexane ring into the vacant antibonding σ*
orbital of the incipient bond need not be invoked. The preference
for hydride attack from the axial direction in unhindered
cyclohexanones (1 and 2) arises from a torsional barrier for the
equatorial attack. This is true even for a relatively bulky reducing
(Table 2). Hence, it appears that, in the case of axial attack, the
electrostatic repulsion between the incoming reducing agent and
the ring sulfur lone pairs is so large that it is not effectively
muted by the solvent molecules. This proposition is supported
by theory.27 The energy calculated for the equatorial transition
26
agent such as a monoalkoxysiliconate ion. The extent of axial
-
state in the reaction with SiH5 (MP2/6-31G* SCRF//HF/3-
attack is reduced when steric hindrance is introduced. Equatorial
attack becomes the preferred mode of addition when steric
effects overcome torsional strain (as in the case of 3 and 4).
21G) is 2.9 kcal/mol lower than that of the axial transition state,
which gives a selectivity of 99:1 in favor of equatorial attack.
The equivalent energy difference predicted for the dioxanone
system is 2.8 kcal/mol. This leads to a selectivity of 1:99 in
favor of axial attack, which is in agreement with the reported
selectivity in solution.
Houk’s predicted trend of a preferred equatorial attack of the
reducing agent in the gas-phase hydride reduction of 2-tert-
2
7
butyl-1,3-dioxan-5-one (5), in sharp contrast to the behavior
observed in solution, was experimentally confirmed. This
observation is readily rationalized by electrostatic effects.
Although the gas-phase results obtained for 5 do not exactly
match the theoretical prediction, the lower percent axial reduc-
tion observed in 5 relative to 1 is in agreement with the
Conclusions
An experimental method for distinguishing the diastereomeric
products of gas-phase hydride reduction reactions has been
developed and applied to the determination of the intrinsic
diastereoselectivity of reduction of cyclic ketones commonly
used as substrates in solution studies. The results obtained for
alkyl-substituted cyclohexanones (axial attack of the hydride
reducing agent to sterically unhindered ketones and equatorial
attack to ketones with a sterically hindered axial face) and
norcamphor (exo attack) are in excellent agreement with the
behavior reported for the same substrates in solution toward
2
7
computed trends. Houk’s theoretical prediction of a larger
preference for equatorial attack in the gaseous dithianone
compared to the dioxanone system was also verified. The
diastereoselectivity observed for hydride reduction of 2-tert-
butyl-1,3-dithian-5-one (6), i.e., almost exclusive equatorial
approach of the reducing agent, is also in good accord with the
behavior observed for the 2-phenyl-substituted dithianone system
3
,12
29
common reducing agents, such as NaBH4 and LiAlH4,
and
in solution. Experiments are in progress to distinguish enan-
are consistent with the diastereoselectivities predicted by
tiomeric products of gas-phase hydride reduction reactions of
prochiral ketones by using the same stereochemical protocol.
-
molecular orbital calculations for reduction by LiH and SiH5
2,27
in the gas phase. The similarity between the condensed-phase
and gas-phase behavior implies that environmental effects, such
as solvation, ion-pairing effects, and metal ion coordination,
are either canceling or unimportant and that reduction diaste-
reoselectivities are the result of the intrinsic properties of the
isolated reactants.
Acknowledgment. This article is dedicated to the memory
of Professor Robert R. Squires. We thank Professor K. N. Houk
of the University of California, Los Angeles, for stimulating
discussions. We are also grateful to the National Science
Foundation for financial support of this research.
The differences observed for the reduction diastereoselec-
tivities among the alkyl-substituted ketones [% axial (exo) attack
JA9910053