Secondary deuterium kinetic isotope effects in irreversible additions of hydride and carbon nucleophiles to aldehydes: A spectrum of transition states from complete bond formation to single electron transfer

Article abstract of DOI:10.1021/ja982504r

The competitive kinetics of hydride and organometallic additions to benzaldehyde-H and -D were determined at -78 °C using LiAlH₄, LiBEt₃H, NaBH₄, LiBH₄, LiAl(O-tert-butoxy)₃H, NaB(OMe)₃H, NaB(OAc)₃H (at 20°C). methyl, phenyl, and allyl Grignard, and methyl-, phenyl-, n-butyl-, tert- butyl-, and allyllithium. The additions of hydride were found to have an inverse secondary deuterium kinetic isotope effects in all cases, but the magnitude of the effect varied inversely with the apparent reactivity of the hydride. In the additions of methyl Grignard reagent and of methyllithium and phenyllithium, inverse secondary deuterium isotope effects were observed; little if any isotope effect was observed with phenyl Grignard or n-butyl- and tert-butyllithium. With allyl Grignard and allyllithium, a normal secondary deuterium kinetic isotope effect was observed. The results indicate that rate-determining single-electron transfer occurs with allyl reagents, but direct nucleophilic reaction occurs with all of the other reagents, with the extent of bond formation dependent on the reactivity of the reagent. In the addition of methyllithium to cyclohexanecarboxaldehyde, a less inverse secondary deuterium kinetic isotope effect was observed than that observed in the addition of methyllithium to benzaldehyde, and allyllithium addition to cyclohexanecarboxaldehyde had a kinetic isotope effect near unity. The data with organometallic additions, which are not incompatible with observations of carbonyl carbon isotope effects, suggest that electrochemically determined redox potentials which indicate endoergonic electron transfer with energies less than ca. 13 kcal/mol allow electron-transfer mechanisms to compete well with direct polar additions to aldehydes, provided that the reagent is highly stabilized, like allyl species. Methyl-and phenyllithium and methyl and phenyl Grignard reagents are estimated to undergo electron transfer with endoergonicities greater than 30 kcal/mol with benzaldehyde, so these react by direct polar additions. A working hypothesis is that butyllithium reagents undergo polar additions, despite redox potentials which indicate less than 13 kcal/mol endoergonic electron transfer, because of the great exoergonicity associated with the two-electron addition, which is responsible for a low barrier for polar reactions.

Full text of DOI:10.1021/ja982504r

326
J. Am. Chem. Soc. 1999, 121, 326-334
Secondary Deuterium Kinetic Isotope Effects in Irreversible Additions
of Hydride and Carbon Nucleophiles to Aldehydes: A Spectrum of
Transition States from Complete Bond Formation to Single Electron
Transfer
Joseph J. Gajewski,* Wojciech Bocian, Nathan J. Harris, Leif P. Olson, and
John P. Gajewski
Contribution from the Department of Chemistry, Indiana UniVersity, Bloomington, Indiana 47405
ReceiVed July 15, 1998
Abstract: The competitive kinetics of hydride and organometallic additions to benzaldehyde-H and -D were
determined at -78 °C using LiAlH4, LiBEt3H, NaBH4, LiBH4, LiAl(O-tert-butoxy)3H, NaB(OMe)3H, NaB-
(OAc)3H (at 20° C). methyl, phenyl, and allyl Grignard, and methyl-, phenyl-, n-butyl-, tert-butyl-, and
allyllithium. The additions of hydride were found to have an inverse secondary deuterium kinetic isotope
effects in all cases, but the magnitude of the effect varied inversely with the apparent reactivity of the hydride.
In the additions of methyl Grignard reagent and of methyllithium and phenyllithium, inverse secondary deuterium
isotope effects were observed; little if any isotope effect was observed with phenyl Grignard or n-butyl- and
tert-butyllithium. With allyl Grignard and allyllithium, a normal secondary deuterium kinetic isotope effect
was observed. The results indicate that rate-determining single-electron transfer occurs with allyl reagents, but
direct nucleophilic reaction occurs with all of the other reagents, with the extent of bond formation dependent
on the reactivity of the reagent. In the addition of methyllithium to cyclohexanecarboxaldehyde, a less inverse
secondary deuterium kinetic isotope effect was observed than that observed in the addition of methyllithium
to benzaldehyde, and allyllithium addition to cyclohexanecarboxaldehyde had a kinetic isotope effect near
unity. The data with organometallic additions, which are not incompatible with observations of carbonyl carbon
isotope effects, suggest that electrochemically determined redox potentials which indicate endoergonic electron
transfer with energies less than ca. 13 kcal/mol allow electron-transfer mechanisms to compete well with
direct polar additions to aldehydes, provided that the reagent is highly stabilized, like allyl species. Methyl-
and phenyllithium and methyl and phenyl Grignard reagents are estimated to undergo electron transfer with
endoergonicities greater than 30 kcal/mol with benzaldehyde, so these react by direct polar additions. A working
hypothesis is that butyllithium reagents undergo polar additions, despite redox potentials which indicate less
than 13 kcal/mol endoergonic electron transfer, because of the great exoergonicity associated with the two-
electron addition, which is responsible for a low barrier for polar reactions.
Irreversible addition of hydride and carbon nucleophiles to
Scheme 1
the carbonyl group is of major importance in organic chemistry.1
The mechanism of this process is therefore of concern,
particularly in connection with questions of face selectivity.2
Two extreme mechanisms can be envisioned: a direct attack
of the base with its two electrons on the carbocation-like
carbonyl carbon (polar mechanism) or a single-electron transfer
(SET) from the base to generate a carbonyl radical anion-base
radical cation pair which collapses to the tetrahedral adduct
(Scheme 1).3
In the latter event, rapid radical-radical anion combination
compared with back-electron transfer requires rate-determining
SET, but reversible SET followed by slow combination is
indistinguishable from the direct (two-electron, polar) attack
unless some spectroscopic or trapping technique provides
evidence for SET prior to the rate-determining step. Within this
mechanistic range is a related concern as to the extent of
complexation of the carbonyl oxygen in the transition state with
whatever Lewis acid is present in the system.
While there have been concerns that SET pathways were
involved in lithium aluminum hydride reductions of ketones
based on suggestive but irreproducible experiments involving
alkyl halide reductions,4 large carbon-14 kinetic isotope effects
(1) ComprehensiVe Organic Synthesis; Trost, B. M., Fleming, I., Eds.;
Pergamon: Oxford, 1991; Vols. 1 and 2.
(2) Eliel, E. L.; Wilen, S. H.; Mander, L. N. Stereochemistry of Organic
Compounds; John Wiley & Sons: New York, 1994; pp 875-888. Coxon,
J. M.; Houk, K. N.; Luibrand, R. T. J. Org. Chem. 1995, 60, 418 (for
calculations with AlH3). Wu, Y.-D.; Houk, K. N.; Paddon-Row, M. N.
Angew. Chem., Int. Ed. Engl. 1992, 31, 1019 (for caculations with LiH).
(3) For a review with a penetrating analysis, see: Walling, C. J. Am.
Chem. Soc. 1988, 110, 6846.
(4) For a summary, see: Ashby, E. C.; Welder, C. O. Tetrahedron Lett.
1995, 7171.
10.1021/ja982504r CCC: $18.00 © 1999 American Chemical Society
Published on Web 12/31/1998

Secondary Deuterium Kinetic Isotope Effects
J. Am. Chem. Soc., Vol. 121, No. 2, 1999 327
in all but DIBAL reductions indicate rate-determining C-H
bond formation.5 However, additions of primary and tertiary
Grignard reagents to aromatic ketones6 at room temperature
appear to favor the SET pathway, as judged by the formation
of 5-hexenyl radical cyclization products, although these could
be side reactions unrelated to the addition reaction. Further, a
carbonyl carbon isotope effect of unity in the addition of allyl
Grignard reagent to benzophenone has been cited as evidence
for rate-determining SET, and substantial carbonyl carbon
isotope effects for additions of alkyl and phenyl Grignard
reagents have been interpreted in terms of preequilibrium SET
followed by rate-determining C-C bond formation.7 In the
additions of methyllithium (and lithium dimethylcuprate), al-
lyllithium, and phenyllithium to benzophenone, the carbonyl
carbon-14 kinetic isotope effect is near unity, and all of these
have been interpreted in terms of rate-determining SET.8
Carbonyl carbon isotope effects in additions of carbon
nucleophiles to aldehydes have not been studied, except for
addition of phenyllithium to benzaldehyde, where the effect is
unity, which was interpreted in terms of rate-determining SET,8b
and addition of lithium pinacolate to benzaldehyde, where the
13C effect is 1.019, which suggests rate-determining C-C bond
formation.9 Secondary deuterium kinetic isotope effects (SD-
KIEs) at the carbonyl carbon of aldehydes can also distinguish
between rate-determining polar (two-electron) attack and rate-
determining SET. The polar (two-electron) addition involves a
transition state with one more bond, which usually results in
an increase in bending frequencies at the reacting carbon, which
results in a greater lowering of the zero point energy upon
deuterium substitution than occurs in the aldehyde starting
material. This leads to an inverse SDKIE at the carbonyl
carbon.10 A relevant example involves the observation of a very
large inverse SKDIE in the addition of a silyl ketene acetal to
benzaldehyde, which seems to indicate “late” C-C bond
formation.11 However, rate-determining formation of a ketyl
radical anion, which necessarily has looser bonds than starting
aldehyde, should result in a normal SDKIE at the carbonyl
carbon.12 Finally, rate-determining Lewis acid complexation
might be expected to result in an SDKIE of less than unity since
the low-frequency aldehydic C-H stretch is converted to a C-H
stretch in a carbocation.13a However, prior protonation or
lithiation of a carbonyl oxygen is a less likely process, given
the low basicity of aldehydes and the rapidity of hydride and
organometallic additions. Even in acid-catalyzed additions of
oxygen nucleophiles to aldehydes, general acid catalysis is
observed; further, the secondary deuterium kinetic isotope effect
at the carbonyl carbon is large and inverse, which indicates that
protonation occurs only after or during nucleophilic attack and
not before.13b
A further utility of the SDKIE is its ability to reflect the extent
of bond formation at the site bearing deuterium, provided that
its value can be compared with the isotope effect expected for
complete bond formation. This latter value may be the equi-
librium isotope effect for the overall process, a value interpolated
from tables of deuterium fractionation factors between various
types of carbon,14 or a value calculated from molecular orbital
theory using calculated Harmonic frequencies and the Biegeleis-
en equation, assuming complete bond formation.15
Another important example of the use of SDKIEs is Hill’s
determination of kH/kD ) 1/1.13 for the sodium borohydride
reduction of benzaldehyde carbonyl H/D in isopropyl alcohol
solvent.16a The inverse SDKIE is consistent with a polar
addition, and comparison of this KIE to what would be expected
for complete bond formation (∼1/1.28) suggests that hydride
transfer is roughly 50% complete in the transition state. Thus,
Hill argued for a later transition state in this reaction, which is
consistent with other observations.16b Since the determination
of SDKIEs using aldehydes provides information on processes
closely analogous to those utilized in many synthetic precedures,
we report here the SDKIEs for a number of hydride reagents,
Grignard and lithium reagent additions to benzaldehyde, and
allyl- and methyllithium additions to cyclohexanecarboxyalde-
hyde.
Results
The KIEs for various additions were determined by reaction
of a mixture of benzaldehyde-H and -D with a deficiency
(approximately 0.1 equiv) of reagent in (mostly) ethereal
solvents at -78 °C. The resulting alcohol products were
converted to the methyl ethers by reaction with dimethyl sulfate
in hexane containing triethylamine or by reaction with sodium
hydride followed by treatment with methyl iodide. The methyl
ethers were then purified by preparative gas chromatography
1
and analyzed by H NMR In the case of the hydride addition
product, benzyl methyl ether, the single deuterium shifts the
methylene proton resonance 0.02 ppm upfield to allow integra-
tion of the singlet from diprotio material and the 1:1:1 triplet
(5) (a) Yamataka, H.; Hanafusa, T. J. Org. Chem. 1988, 53, 772. All
reactions had a small positive F value, except for reaction with 9BBN, which
had a negative F value. (b) Yamataka, H.; Hanafusa, T. J. Am. Chem. Soc.
1986, 108, 6643. (NaBH4 in IPA and LiBH4 in ether, 1.066 and 1.089,
respectively; LAH in ether, 1.024.
(6) Ashby, E. C.; Bowers, J. R., Jr. J. Am. Chem. Soc. 1981, 103, 2242.
See also: Ashby, E. C.; Chao, L.-C.; Neumann, H. M.; Laemmle, J. T.
Acc. Chem. Res. 1974, 7, 272. Ashby, E. C. Pure Appl. Chem. 1980, 52,
545; Acc. Chem. Res. 1988, 21, 414. Holm, T. Acta Chem. Scand. 1983,
B37, 569.
(12) For experimental observations with the benzene radical cation,
see: Stevenson, G. R.; Espe, M. P.; Reiter, R. C. J. Am. Chem. Soc. 1986,
108, 532. The 13C isotope effects in this work are controversial, but the
deuterium effects are reproduced by theory: Hrovat, D. A.; Hammonds, J.
H.; Stevenson, C. D.; Borden, W. T. J. Am. Chem. Soc. 1997, 119, 9523.
Thus, to the extent that an olefinic sp2 carbon becomes a radical-like carbon
in the radical anion, the frequencies at that carbon should be lower, see
also: Pacansky, J.; Koch, W.; Miller, M. D. J. Am. Chem. Soc. 1991, 113,
317.
(13) (a) The calculated value for the equilibrium: acetaldehyde-H +
conjugate acid-D h acetaldehyde-D + conjugate acid-H is 0.83 at 25 °C
at the MP2/6-31G** level with no scaling of frequencies. (b) Jencks, W.
P. Acc. Chem. Res. 1976, 9, 425-32. Funderburk, L. H.; Aldwin, L.; Jencks,
W. P. J. Am. Chem. Soc. 1978, 100, 5444-59. Hill, E. A.; Milosevich, S.
A. Tetrahedron Lett. 1976, 4553-54. The possibility that both protonation
and nucleophilic attack are reversible with the rate-determining step
involving general base catalysis in the removal of a proton, which would
also give rise to general acid catalysis, is not involved.
(7) Yamataka, H.; Matsuyama, T.; Hanafusa, T. J. Am. Chem. Soc. 1989,
111, 4912.
(8) (a) Yamataka, H.; Fujimura, N.; Kawafuji, Y.; Hanafusa, T. J. Am.
Chem. Soc. 1987, 109, 4305. (b) Yamataka, H.; Kawafuji, Y.; Nagareda,
K.; Miyano, N.; Hanafusa, T. J. Org. Chem. 1989, 54, 4706.
(9) Yamataka, H.; Sasaki, D.; Kuwantani, Y.; Mishima, M.; Tsuno, Y.
Chem. Lett. 1997, 271; J. Am. Chem. Soc. 1997, 119, 9975.
(10) do Amaral, L.; Bull, H. G.; Cordes, E. H. J. Am. Chem. Soc. 1972,
94, 7579. do Amaral, L.; Bastos, M. P.; Bull, H. G.; Cordes, E. H. J. Am.
Chem. Soc. 1973, 95, 7369. Archila, J.; Bull, H. G.; Lagenaur, C. J. Org.
Chem. 1971, 36, 1345. Pires, J. R.; Stachissini, A. S.; do Amaral, L. J.
Phys. Org. Chem. 1994, 7, 192. Rossi, M. H.; Stachissini, A. S.; do Amaral,
L. J. Org. Chem. 1990, 55, 1300.
(14) Shiner, V. J., Jr.; Neumann, T. E. Z. Naturforsch. 1989, 44a, 337.
For earlier work on fractionation factors, see: Hartshorn, S. R.; Shiner, V.
J., Jr. J. Am. Chem. Soc. 1972, 91, 9002.
(11) (a) Myers, A. G.; Widdowson, K. L.; Kukkola, P. J. J. Am. Chem.
Soc. 1992, 114, 2765. (b) Others (Abu-Hasanayn, F.; Streitwieser, A. J.
Org. Chem. 1998, 63, 2954) found a large inverse equilibrium SDIE in the
addition of a lithium enolate to benzaldehyde.
(15) For a systematic theoretical study of small-molecule deuterium
fractionation factors, see: Harris, N. J. J. Phys. Chem. 1995, 99, 14689.
(16) (a) Hill, E. A.; Milosevich, S. A. Tetrahedron Lett. 1976, 3013. (b)
For an excellent review, see: Wigfield, D. C. Tetrahedron 1979, 35, 449.

328 J. Am. Chem. Soc., Vol. 121, No. 2, 1999
Gajewski et al.
Table 1. KIEs for Irreversible Additions to PhCHO/PhCDO
reagenta
solvent
THF
THF
EtOH
EtOH
THF
Et2O/THF
HOAc
Et2O
Et2O
Et2O
Et2O
Et2O
temp (°C)
KIEb
KIE (25 °C)c
EIE (25 °C)d
i (25 °C)
LiAlH4
LiBEt3H
-78
-78
-78
-78
-78
-78
20
-78
-78
25
-78
-78
-78
-78
-78
-78
-78
-78
-78
-78
0.975 ((0.01)
0.965 ((0.01)
0.91
0.984
0.977
0.94
0.88
0.92
0.92
0.75
0.97
0.90
0.93
1.00
1.03
1.03
1.06
0.90
0.91
0.92
0.97
1.0
0.85
0.85
0.78
0.78
0.85
0.85
0.78
0.78
0.78
0.78
0.78
0.78
0.78
0.78
0.78
0.78
0.78
0.78
0.78
0.78
0.12
0.14
0.25
0.51
0.51
0.51
1.16
0.13
0.44
0.31
0
SET
SET
SET
0.44
0.40
0.37
0.13
0
NaB(OMe)3H
NaBH4
LiBH4
0.82
0.885 ((0.01)
LiAl(OtBu)3H
NaB(OAc)3H
0.88
0.75
0.95
0.87
CH3MgBr (0.1 M)
CH3MgBr (0.01 M)
CH3MgBr (0.01 M)
phenylMgBr (0.1 M)
allylMgBr (0.1 M)
allylMgBr (0.01 M)
allyllithium
1.00
1.04
1.04
1.095
0.87
0.88
0.89
0.95
0.99
0.99
Et2O
Et2O
Et2O
Et2O
CH3Li (0.1 M)
CH3Li (0.01 M)
CH3Li-0.5 equiv of MeOH
Phenyl-Li (0.1 M)
n-butyllithium
Et2O
Et2O
Et2Oe
pentane
tert-butyllithum
1.0
0
a Commercially available reagents were used where available. b Average deviation is (0.03 except as noted. c KIE values measured at -78 °C
were corrected to 25 °C assuming the KIEs depend only on ∆H. d Theoretical EIE values from RCHO + RCD(R′)OM ) RCDO + RCH(R′)-OM,
where R′ ) H or alkyl and M ) H or Li (or Mg) from MP2/6-311+G** on acrolein-1-d, allyl alcoholate derivatives, acrolein, and allyl-1-d-
alcoholate derivatives; thus, complete C-C bond formation is the standard here (see text). This value applies only to polar, two-electron mechanisms.
e Butyllithium in pentane was added to an ether solution of the aldehyde.
Table 2. KIEs for Irreversible Additions to Cyclohexanecarboxaldehyde H/D
reagenta
solvent
temp (°C)
KIEb
KIE (25 °C)c
EIE (25 °C)d
i (25 °C)
MeLi
allyl-Li
Et2O
Et2O
-78
-78
0.92 ((0.03)
0.98 ((0.03)
0.95
0.99
0.78
0.78
0.20
0.05
a Commercially available reagents were used where possible. b Average deviation is (0.03. c See footnote c, Table 1. d See footnote d, Table 1.
of the monodeuterio material. Capillary GC on an optically
active nickel compound was also utilized since the deuterium
compound has a shorter retention time than the protio com-
pound. The composition of the mixture of aldehyde H and D
was determined by reaction with an excess of reagent and
analysis of the product methyl ether by the same technique as
used for the competition experiments in an effort to reduce
systematic errors. Under these circumstances, kH/kD ) H/D
product(reagent limiting)/H/D product(aldehyde limiting). The
results from various hydride, Grignard, and lithium reagent
additions to benzaldehyde-D are given in Table 1.
less accessible with aliphatic aldehydes due to their higher
reduction potentials. These data are given in Table 2.
To provide a standard for comparison of the KIEs, the
equilibrium constant for the equilibration (the deuterium
fractionation factor, eq 1) of acrolein-1-d and allyl alcoholate
CH2dCHCHO + CH2dCHCHDO-M h
CH2dCHCDO + CH2dCHCH2O-M (1)
M ) H+, Li+, nothing
In the case of methyl and allyl Grignard reagents, the
concentrations of the reactants were varied over a factor of 10,
leading to diminished inverse isotope effects at higher concen-
trations only in the case of methyl Grignard. In the case of
methyl Grignard, the reaction was also performed at room
temperature at the highest dilution studied at low temperatures,
and this led to a KIE that is less inverse than the KIE from low
temperatures extrapolated to room temperature. In the case of
methyllithium additions, inverse SDKIEs were also observed,
but their magnitudes did not depend on concentration. Further,
prior reaction of methyllithium with 0.5 equiv of methanol did
not dramatically affect the SDKIE for addition of methyllithium
to benzaldehyde. The SDKIE determinations with allylmagne-
sium bromide and allyllithium additions to a mixture of
benzaldehyde and benzaldehyde-d were conducted at 0.1 M
concentrations in diethyl ether solvent, and allyl Grignard
determination was also conducted at 0.01 M, all at -78 °C.
All allyl reagent additions revealed a normal SDKIE.
with acrolein-1-H and allyl-1-d alcoholate in the gas phase was
determined from the harmonic frequencies obtained from MP2/
6-311+G** calculations.15 The counterion of allyl alcoholate
was either H+, Li+, or nothing. The values for these equilibria
are 1/1.28, 1/1.17, and 1.11, respectively, at 25 °C. While these
values provide appropriate comparisons for the hydride addition,
they were corrected downward by 11% for additions of carbon
nucleophiles, recognizing that replacement of hydrogen by
carbon results in increased stabilization of deuterium relative
to hydrogen on the carbonyl carbon when the substitution
occurs.14 It was further assumed that this fractionation factor
equilibrium (FF) with magnesium counterion is the same as that
for the lithium counterion to provide a standard for the Grignard
additions.
To make the comparisons, the kinetic isotope effects were
extrapolated to room temperature, assuming that there are no
entropic contributions so that the exponent, i, of the linear free
energy relationship, SDKIE ) FFi, between the SDKIE and the
FF data could be obtained. Such a relationship assumes that
there are no factors affecting the transition state that are not
reflected in the relative stabilities of H- and D-substituted
reactants and products.
Since the SDKIEs with benzaldehyde varied over a wide
range, a limited SET of experiments with methyl- and allyl-
lithium reaction with cyclohexanecarboxaldehyde-d were con-
ducted, recognizing that rate-determining SET mechanisms are

Secondary Deuterium Kinetic Isotope Effects
J. Am. Chem. Soc., Vol. 121, No. 2, 1999 329
While all of the SDKIEs reported in the tables were the result
of analysis of the product alcohol as its methyl ether, attempts
were made to examine the hydrogen/deuterium fractionation in
the starting benzaldehyde in the reactions with methyl Grignard
and methyllithium as well as with allyl-, n-butyl-, and tert-
butyllithium.17 Generally, the SDKIEs determined in this manner
at short reaction times were consistent with those determined
from the product alcohol (converted to its methyl ether).
However, at greater extents of conversion, the SDKIEs deter-
mined from benzaldehyde with n-butyl- and allyllithium became
larger than unity and proportional to the extent of conversion.
Further small amounts of benzyl alcohol were formed in each
reaction (typically 1% of the product). The use of a deficiency
of reagent and long reaction times (overnight) before quenching
led to substantial amounts of benzyl alcohol. These results
suggest incursion of a large, normal primary hydrogen isotope
effect due to hydride transfer which dramatically fractionates
starting material, particularly at long reaction times.
Finally, to determine the magnitude of the SDKIE at
aldehydic carbon expected in rate-determining formation of an
organolithium complex, MP2/6-31G** calculations of harmonic
frequencies were performed on acrolein-H and -D as well as
on the carbonyl oxygen complex with methyllithium. After
geometry optimization to a structure with the oxygen, the
lithium, and the methyl nearly collinear, the lithium is syn to
the aldehydic H (or D). The equilibrium constant calculated for
eq 2 at 25 °C is 0.94 with unscaled frequencies (0.95 with
appropriately scaled ones).
coincidence of mechanisms for the two substrates. In the case
of lithium aluminum hydride reductions, the nature of the
reagent is of some concern, but the SDKIE determinations
reported here provide little information about this. The SDKIEs
reflect only the extent of change in bonding at the carbonyl
carbon. Nonetheless, it should be noted that lithium aluminum
hydride is ionized at low concentrations in tetrahydrofuran but
forms solvent-separated ion pairs above 0.001 M.19a Further-
more, lithium cation is essential for the reaction, since lithium
cation sequestering agents inhibit the reaction.19b In addition,
lithium aluminum hydride reductions in tetrahydrofuran occur
with first-order kinetics,19c but in diethyl ether solvent, the
reduction is second order in lithium aluminum hydride.19d
Nonetheless, the primary deuterium kinetic isotope effect for
LAH/LAD reductions is similar in both solvents, suggesting
similar transition-state structure in the vicinity of the carbonyl
group.
Still another concern in the hydride reductions is the changing
nature of the reagent being studied during the reaction. The
SDKIE determinations require a deficiency of reagent, which
therefore means that all of the reagent is consumed, and, in the
case of lithium aluminum hydride and tetrahydridoboron
reagents, it must change in response to changes in the ligands
due to replacement of hydride by alkoxides, which may also
result in changes in aggregation. The SDKIE determinations of
Tables 1 and 2, therefore, represent an average over all the
changes in the reagent during the reaction. One clear case of
this is the sodium borohydride reduction. The SDKIE deter-
mined for sodium trimethoxyborohydride, the reagent present
in the last 25% of the borohydride reduction in methanol, is
less inverse than that observed for the overall reaction. This
suggests that the SDKIE from reaction of tetrahydridoboron
anion itself must be more inverse than that observed for the
overall reduction. Nonetheless, in the discussions below, the
focus will be on the average nature of the reagent, as revealed
by the SDKIE.
The results with hydride reagents reveal dramatic changes
in transition-state structure. The very reactive hydrides, namely
lithium aluminum hydride, lithium triethylborohydride, and
sodium trimethoxyborohydride, give SDKIEs which approach
unity, but the least reactive, namely sodium triacetoxyborohy-
dride, reacts with a very large inverse SDKIE that is comparable
to the equilibrium fractionation factor for the reaction. There
are also examples of hydride reagents whose inverse SDKIEs
are intermediate in value, namely sodium and lithium borohy-
dride and lithium aluminum tri-tert-butoxy hydride. Further, as
indicated above, it is likely that the i value for reaction of
borohydride ion is larger than the average i value for all reducing
agents in the sodium borohydride reaction, so even here the
transition state appears to be “late” with respect to reactants
and products. This is surprising, considering the exothermicity
of the reaction; however, it has been previously suggested in
work that is not contradicted by the current results that the
exothermicity in the borohydride reduction results from carbonyl
oxygen-boron bond formation after rate-determining generation
of the tetrahedral addition product.20 In hydroxylic media, this
is not unreasonable, since protonation of the alkoxide in the
tetrahedral species (or perhaps before its formation in acetic
CH2CHCHO + CH2CHCHDO-Li-CH3 h
CH2CHCDO + CH2CHCH2O-Li-CH3 (2)
Discussion
Hydride Reductions. With the hydride reagents studied, all
of which are anionic, the SDKIEs are inverse, suggesting rate-
determining C-H bond formation. Further, the extent of hydride
transfer as measured by the i values varies from small (0.12 (
0.05) with lithium aluminum hydride to apparently more than
complete (1.16 ( 0.2) with sodium triacetoxyborohydride. In
the latter case, the excessively large i value is still within
experimental error of 1.0. While there may be some concern
about the maximum expected SDKIE in all cases, it is unlikely
that there are large discrepancies unless cation bonding to the
oxygen anion is less than complete. If this were the case, then
the extent of bonding would be greater than that suggested by
the i values of Table 1. This is because deuterium lowers the
zero point energy of the aldehyde to a greater extent than it
lowers the zero point energy of an unsolvated alkoxide ion
product. This is probably the result of low frequencies associated
with groups attached to carbon bearing the oxyanion due to
negative ion hyperconjugation.18
It is interesting to note that the extent of hydride transfer to
benzaldehyde suggested by the SDKIEs (the i values) at -78
°C are not very different than the extent of hydride transfer to
benzophenone near room temperature as deduced from carbon-
14 effects for the same reagent (lithium aluminum hydride,
sodium and lithium borohydride).5 Thus, there may be a
(17) See, for example: Delmonte, A. J.; Haller, J.; Houk, K. N.;
Sharpless, K. B.; Singleton, D. A.; Strassner, T.; Thomas, A. A. J. Am.
Chem. Soc. 1997, 119, 9907.
(18) DeFrees, D. J.; Bartmess, J. E.; Kim, J. K.; McIver, R. T., Jr.; Hehre,
W. J. J. Am. Chem. Soc. 1977, 99, 6451. These authors showed that the
equilibration of methanol and trideuteriomethoxide with trideuteriomethanol
and methoxide favored the latter pair by 0.5 kcal/mol at room temperature
in an ICR cell.
(19) (a) Hogen-Esch, T. E.; Smid, J. J. Am. Chem. Soc. 1966, 88, 318.
Ashby, E. C.; Dobbs, F. R.; Hopkins, H. P., Jr. J. Am. Chem. Soc. 1973,
95, 2823. (b) Pierre, J. L.; Handel, M.; Perrand, R. J. Tetrahedron 1975,
31, 2795. (c) Eliel, E. L.; Senda, Y. Tetrahedron 1970, 26, 2411. Ashby,
E. C.; Boone, J. R. J. Am. Chem. Soc. 1976, 98, 5524. (d) Wiegers, K. E.;
Smith, S. G. J. Am. Chem. Soc. 1977, 99, 1480.
(20) Adams, C.; Gold, V.; Reuben, D. J. Chem. Soc., Perkin Trans. 2
1977, 1472.

330 J. Am. Chem. Soc., Vol. 121, No. 2, 1999
Gajewski et al.
acid) can easily occur. Whether the same is true, i.e., extent of
metalation of the oxygen, of the reductions in ethereal solvents
is of concern. However, the SDKIE and the i value for lithium
aluminum tri-tert-butoxy hydride are consistent with substantial
interaction of the carbonyl oxygen with lithium cation, or the
SDKIE would be much less inverse, as suggested from the
calculation of fractionation factor above with no counterion for
the oxygen.
addition, the SDKIE in the addition of allyl Grignard is
independent of concentration. These results indicate that these
reactions occur by rate-determining SET. It is interesting that
there is only a small difference (0.16 V) in the oxidation
potentials of allyl Grignard and allyllithium determined by
Holm22 and Breslow,23 respectively, once corrected for the
reference electrode, but neither determination could be con-
ducted under conditions of reversible electron transfer, and the
oxidation of the lithium reagents was conducted in THF, so an
exact comparison cannot be obtained. However, it is important
to note that the ease of oxidation of the allyl reagents is ca.
0.5-0.9 V greater than those of methyl Grignard and methyl-
lithium determined in the same way. Therefore, single-electron
transfer is more favorable with the allyl reagents than with the
methyl reagents. Finally, if electron transfer were rate-determin-
ing with allyllithium reacting with benzaldehyde, then this
pathway should be less favored with a nonconjugated aldehyde
since the reduction potentials of these are more negative by at
least 0.3 V.24 Indeed, this appears to be the case from the data
of Table 2, where the SDKIE is near unity.
Phenyl- and Butyllithium Additions. It is interesting that
phenyl-, n-butyl- and tert-butyllithium additions to benzaldehyde
have SDKIEs near unity. Whether this reflects the extreme
reactivity of these reagents leading to an early transition state
or a competition between rate-determining SET and polar
addition is unclear, but we would suggest that these reactions
are very exothermic polar additions with very early transition
states, especially in the case of the phenyl reagents, which are
not easy to oxidize by one electron, in contrast to the
butyllithium reagents (see below).
Redox Potentials and Electron Transfer and Comparison
to Polar Additions. Efforts to assess the involvement of single-
electron transfer in organometallic additions to carbonyl com-
pounds often focus on redox potentials of the reagents.25 Arnett’s
study of the reduction potentials of benzaldehydes and ben-
zophenone and the reversible oxidation potential of lithium
pinacolate indicates that rate-determining electron transfer is
unfavorable by ca. 40 kcal/mol.25 Though this would appear to
rule out rate-determining SET, Arnett was cautious in not
dismissing the possibility, given the differences in the media
between the actual reaction and the electrochemical determina-
tions. However, in the case of addition of lithium pinacolate to
benzaldehyde, carbonyl carbon KIEs are substantial,9 and
unpublished results in our laboratory indicate an inverse SDKIE
(0.94 at -78 °C). Both of these results suggest rate-determining
C-C bond formation. It is important to note that the reduction
potentials of benzaldehyde and benzophenone are similar and
equal to roughly -1.8 V relative to SCE,24 so the evidence for
SET mechanisms in additions to benzophenone is not simply a
function of the ease of reduction of it relative to benzaldehyde.
The apparent oxidation potential of methyl Grignard in diethyl
ether under irreversible conditions is -0.25 V relative to a
standard hydrogen electrode.22 By comparison, the oxidation
potential of methyllithium in THF under irreversible conditions
is -0.75 V relative to SCE.23 Since the difference in the
reference electrodes is 0.242 V, methyl Grignard is more
Additions of Carbon Nucleophiles. Methyl Grignard
Reagents and Methyllithium. Additions of methyl Grignard
were conducted at different concentrations and different tem-
peratures. The SDKIE at low concentrations (0.01 M) was more
inverse than the SDKIE at a 10-fold higher concentration at
-78 °C. The data indicate that the reaction does not involve
rate-determining SET. Further, with the assumption that the
fractionation factor for the alkoxide with a magnesium coun-
terion is the same as that with the lithium counterion calculated
above, the extent of CC bond formation in the transition state
for methyl Grignard addition to benzaldehyde is roughly 30%
in the more dilute reactions. At higher concentrations of
Grignard reagent, it is reasonable that a higher concentration
of dimeric species is present, but why this results in a less
inverse SDKIE is unclear. It may be significant that the SDKIE
for reaction of methyl Grignard in dilute solution at room
temperature is roughly intermediate between that extrapolated
from the -78 °C experiments at both 0.1 and 0.01 M, but more
careful determinations are necessary to make distinctions.
Consistent with the current results are the carbonyl carbon
KIEs in additions of methyl and phenyl Grignard reagents to
benzophenone, which are substantial.7 These have been inter-
preted in terms of a polar mechanism or one involving
preequilibrium SET with rate-determining C-C bond formation.
It is also important to note that Holm found inverse secondary
deuterium kinetic isotope effects (SDKIEs) at the R carbon of
methyl and ethyl Grignard in their additions to benzophenone,
and these were interpreted in terms of a polar mechanism.21
Interestingly, earlier Holm found that the logarithms of the
relative rates of addition of many Grignard reagents to ben-
zophenone (with benzophenone in excess) correlated with the
ease of oxidation of the reagent, and these reactions were
interpreted in terms of an SET mechanism.22
The reactions of methyllithium with benzaldehyde and with
cyclohexanecarboxyaldehyde in diethyl ether at -78 °C reported
in Tables 1 and 2, respectively, reveal inverse SDKIEs,
indicative of rate-determining C-C bond formation in each case,
with perhaps less bond formation with the aliphatic aldehyde.
These results are consistent with a polar reaction that has an
earlier transition state with the less stabilized aldehyde than with
benzaldehyde. By contrast, the addition of methyllithium to
benzophenone in diethyl ether at 0.0 °C occurs with no carbonyl
carbon-14 KIE, which was interpreted in terms of rate-
determining SET.8a Finally, it was important to find that the
SDKIE determined for a methyllithium reagent which had
reacted partially with methanol was nearly the same as that for
methyllithium itself. This reveals that the reagent, though
changing during the reaction, has similar reactivity in its various
forms.
(23) Jaun, B.; Schwanz, J.; Breslow, R. J. Am. Chem. Soc. 1980, 102,
5741. Approximate Eo (relative to SCE; Eo relative to SHE, 0.242 V): allylLi
) -1.24 V; BuLi ) -1.41 V; MeLi ) -0.72 V; PhLi ) -0.34 V, all in
THF at -60 °C, except allyl at -25 °C.
Allyllithium and Allyl Grignard SET Reactions. The
additions of allyllithium and allyl Grignard reagent to benzal-
dehyde at -78 °C are remarkable in that they have SDKIEs
which are greater than unity. Further, unlike the methyl Grignard
(24) Evans, D. H. Carbonyl Compounds. In Encyclopedia of Electro-
chemistry of the Elements, Vol XII; Bard, A. J., Lund, H., Eds.; Marcel
Dekker: New York, 1978; pp 1-259.
(21) Holm, T. Acta Chem. Scand. 1994, 48, 362.
(22) Holm, T. Acta Chem. Scand. 1983, B37, 569. Eo(relative to SHE):
allylMgBr ) -1.16 V; MeMgBr ) -0.25 V, in diethyl ether at 25 °C. For
phenylMgBr, Eo ) 0. Eberson, L. Acta Chem. Scand. 1984, B38, 439.
(25) Palmer, C. A.; Ogle, C. A.; Arnett, E. M. J. Am. Chem. Soc. 1992,
114, 5619. This paper provides an excellent review of the efforts to
distinguish between rate-determining C-C bond formation and SET.

Secondary Deuterium Kinetic Isotope Effects
J. Am. Chem. Soc., Vol. 121, No. 2, 1999 331
difficult to oxidize than methyllithium by ca. 0.25 V. However,
the values suggest that the free energy for electron transfer to
benzaldehyde from these methyl reagents is 30-35 kcal/mol,
which appears to be very large relative to the low barriers that
exist in these reactions. The apparent SET reaction of methyl-
lithium with benzophenone relative to the polar reaction of
methyl Grignard as suggested from carbonyl carbon KIEs might
arise because of the 5 kcal/mol greater ease of oxidation of
methyllithium and the steric difficulty encountered in polar
additions to benzophenone relative to benzaldehyde. On the
other hand, the free energy change involved in the electron
transfer from the allyl reagents to benzaldehyde is roughly +13
kcal/mol based on electrochemical measurements, which suffer
from irreversibility in the oxidation of the organometallic and
counterion and solvent effects in the reduction of benzalde-
hyde.25 Nonethess, this relatively low endoergonicity is sufficient
to allow the SET mechanisms to dominate over the polar
addition pathway, even with benzaldehyde. The fact that allyl
Grignard reacts much faster than expected on the basis of its
oxidation potential relative to other Grignard reagents is also
suggestive of a different mechanism, and this would appear to
be the SET pathway and not the cyclic mechanism. The latter
mechanism was proposed26 to occur with allyl Grignard to
distinguish it from the other Grignard reagents which, at the
time, were assumed to proceed via SET but now should be
recognized as polar, two-electron additions.
The use of redox potentials to characterize the nature of the
addition reactions is reinforced by the current observations, even
to the point of rationalizing the SDKIE results of the additions
to cyclohexanecarboxaldehyde. Since the reduction potential of
aliphatic aldehydes is roughly 5 kcal/mol higher than that of
benzaldehyde,24 the SET mechanism is less dominant. The only
remaining concern given the data of Table 1 is the observation
of SDKIEs of unity in additions of n-butyl- and tert-butyllithium
to benzaldehyde, despite the apparent fact that the ease of
oxidation of these reagents is higher than that of allyllithium.23
We would suggest that these reagents can also add via a two-
electron polar mechanism in an exceedingly exothermic fashion
so as to effectively compete with the SET pathway. The
exothermicity of the polar, two-electron pathway suggests an
early transition state leading to an SDKIE of near unity, which
is consistent with the observations. Evidence for the great
exothermicity derives from Arnett’s determination of the heat
of reaction of various lithium reagents with tert-butyl alcohol
(-50 kcal/mol for reaction of n-butyllithium in 9:1 hexane/
ether).27 Simple bond dissociation considerations would suggest
that additions of lithium reagents to aldehydes and ketones
would be even more exothermic. For example, the summation
of reaction 4 (∆H ) -5.4 kcal/mol28) and reaction 3, which is
the reaction of methyllithium with tert-butyl alcohol to give
lithium tert-butoxide and methane with a heat of reaction of
-41.6 kcal/mol, provides the exothermicity of the addition of
methyllithium to acetone (reaction 5), which is -47 kcal/mol.
Figure 1. Semiquantitative relationship between the rates of two-
electron (polar) and SET additions of lithium reagents to benzaldehyde
based on the energetics of each reaction. Solid vertical arrows represent
the favorable pathway.
for the addition of phenyllithium to carbonyl compounds to be
similar to that for addition of methyllithium. However, substitu-
tion of methyllithium by other butyllithium reagents should
provide even more exothermic addition reactions. On the other
hand, allyllithium is stabilized relative to simple aliphatic lithium
reagents due to electron delocalization, so it should not provide
as exothermic a reaction in either proton removal or a two-
electron, polar addition. Allyllithium is, of course, not stabilized
with respect to loss of an electron since a delocalized allyl
radical is the result, which is not the case with loss of an electron
from n-butyl- or tert-butyllithium.
The analysis above is, however, not quantitatively correct in
the case of lithium pinacolate addition to benzaldehyde in THF
solvent at 25 °C, where Arnett measured the reaction enthalpy
to be -16 kcal/mol,27 a value roughly 15 kcal/mol more
exothermic than might be suggested by the analysis using
reactions similar to 3-5. Differential solvent effects may be
important here. However, the relative exothermicities for
addition, as suggested by the heat of reaction of the lithium
reagents with tert-butyl alcohol, are probably not out of line.
An important, perhaps obvious, conclusion from these results
is that the combination of an easily oxidized organometallic
reagent that is not highly basic and a good electron acceptor
that is also sterically hindered provides the best opportunity for
rate-determining SET pathway to compete with a two-electron,
polar addition. It is also reasonable that less easily oxidized
reagents are unlikely to undergo single-electron transfer prior
to the rate-determining step. An example of the latter case is
the lithium aldolate addition to benzaldehyde, where the
electrochemically determined endoergonicity is roughly 45 kcal/
mol.25 Here the exoergonicity in the polar, two-electron addition
is only -16 kcal/mol,27 yet this pathway dominates, as judged
by isotope effects, because of the very unfavorable SET reaction.
Other examples having the SET pathway with very unfavorable
energetics are the additions of phenyllithium and phenyl
Grignard to benzaldehyde. Finally, in the additions to carbonyl
compounds where the electron-transfer pathway is less endo-
ergonic than roughly 13 kcal/mol, a highly exoergonic two-
electron, polar addition can compete effectively with the SET
pathway. Figure 1 is an attempt to provide a semiquantitative
relationship between the two reaction pathways and the energet-
ics of each pathway in the reaction of lithium reagents with
benzaldehyde. It is necessarily incomplete because steric effects
should retard the polar addition relative to the SET pathway,
so highly hindered organometallic reagents, e.g., tertiary Grig-
nard reagents, reacting with hindered carbonyl compounds, e.g.,
MeLi + (CH3)3-OH f
Me-H + (CH3)3-OLi + 41.6 kcal/mol (3)
Me-H + (CH3)2CdO f (CH3)3-OH + 5.4 kcal/mol (4)
MeLi + (CH3)2CdO f (CH3)3-OLi + 47.0 kcal/mol (5)
(27) Arnett, E. M.; Moe, K. J. Am. Chem. Soc. 1991, 113, 7068. See
also: Arnett, E. M.; Palmer, C. A. J. Am. Chem. Soc. 1990, 112, 7354.
(28) From Benson’s Group Additivity Tables: Benson, S. W. Thermo-
chemical Kinetics, 2nd ed.; John Wiley & Sons: New York, 1976; pp 272-
275.
Since the heat of protonation of phenyllithium is the same
as that of methyllithium, one might expect the exothermicity
(26) Holm, T. J. Am. Chem. Soc. 1993, 115, 916.

332 J. Am. Chem. Soc., Vol. 121, No. 2, 1999
Gajewski et al.
gas chromatograph with thermal conductivity detection and either a
benzophenone, may prefer the SET pathway over the polar one,
as has been suggested by Ashby.
1
1
DB-5 (6 ft × /4 in.) column or a Carbowax (12 ft × /4 in.) column.
A Varian 3700 gas chromatograph with flame ionization detection was
used for analytical gas chromatography. The capillary columns used
were DB-5 (50 m × 0.25 mm), Supelcowax-10 (60 m × 0.35 mm),
SPB-5 (100 m × 0.25 mm) in series with SP-2330 (60 m × 0.25 mm),
and nickel bis(3-heptafluorobutyryl-(1R)-camphorate/OV-1 (25 m ×
0.25 mm), a Schurig column. The latter two columns were used for
isotopic separations. Integration of relative peak heights was ac-
complished with either a Varian 4240 or a Hewlett-Packard 3390A
integrator. Dry tetrahydrofuran and diethyl ether were obtained by
distillation under N2 from sodium benzophenone ketyl. All glassware
for addition reactions was flame-dried under vacuum with an N2 purge.
Precise temperature measurements were performed with a Newport
series 2600 digital thermocouple indicator with a type E thermocouple.
Benzaldehyde-d (Adapted from Ref 30). A mixture of anhydrous
diethyl ether (1.5 dmol3), 379.2 g (2 mol) of anhydrous stannous
chloride, and 628.0 g (8 mol) of acetyl chloride was added to a 5-L
three-neck flask equipped with a stirrer, a dropping funner, and a
mercury bubbler. The flask was cooled to -10 °C, and 150.2 g (7.5
mol) of deuterium oxide was added in small portions such that the
temperature did not rise above 10 °C. To the homogeneous solution
was added 103.1 g (1 m) of benzonitrile, and stirring was continued
for 12 h at room temperature. The precipitated crystalline stannic
chloride complex was filtered away from the solution, washed with
ether, and dried in the dark. Hydrolysis of the complex to benzalde-
hyde-D was accomplished upon addition to hot water. Extraction of
the benzaldehyde-D from the aqueous solution by diethyl ether followed
by drying the ethereal solution over anhydrous magnesium sulfate,
removal of the solvent at reduced pressure, and distillation at reduced
pressure, provided quantitative yield of benzaldehyde-D with 98.0%
deuterium at the aldehydic carbon.
General Procedures for Methyl Ether Formation from Benzylic
Alcohols: Procedure A. A dispersion of sodium hydride in mineral
oil was washed thoroughly with pentane, and the residue was dried
and weighed (0.133 g, 5.5 mmol) into a flask with with methyl iodide
(0.68 g, 4.8 mmol) with 5 mL of dry THF. The dried alcohol derived
from aldehyde and reagent addition (typically 0.35 mmol) was added
to the mixture and allowed to stir at room temperature for 8 h. Then,
3 mL of a 1 M aqueous sodium hydroxide solution was added dropwise.
The aqueous layer was extracted with ether (3 × 3 mL), and the
combined organic layers were dried over anhydrous sodium sulfate and
then filtered and concentrated by rotory evaporation. The solution was
either analyzed directly on capillary or subjected to purification by
preparative GC and analyzed by capillary GC.
Extent of C-C Bond Formation in Lithium Reagent
AdditionssComparison to Calculations. The i values repre-
sent one approach to the question of the extent of bond formation
to the carbonyl carbon. As described above, the i value is equal
to the fraction of the kinetic isotope effect to the expected
equilibrium isotope effect in free energy terms. It therefore
assumes, as indicated above, that there are no factors affecting
the 3N - 7 vibrations of the transition state that cannot be
interpolated between reactant and product state. Some caveats
were described. In particular, concern was expressed for the
extent of cation binding to the developing negative charge on
the carbonyl oxygen. However, there is another approach to
assessing transition-state structure which has been used ef-
fectively in recent years, namely the calculation of kinetic
isotope effects from ab initio calculations of reactant and
transition state structures.11b,16 If the experimental KIEs can be
reproduced by the calculations, then the structures giving rise
to the differences in frequencies are assumed to be the correct
ones. In the case of addition of the lithium enolate of
p-(phenylsulfonyl)isobutyrophenone to benzaldehyde-d, a large
inverse SDEIE (0.74) was determined at room temperature.11b
HF/6-31++g** calculations of the ground state and transition
state for gas-phase addition of lithium enolate to formaldehyde
provided an SDKIE of 0.84, even though the C-C bond being
formed was long, namely 2.221 Å. Since the calculated EIE
was 0.82, the transition-state structure would suggest a much
smaller inverse KIE by the reasoning proposed above. Just why
the calculated KIE is not a bad match to the experimental EIE
despite a transition-state structure that would descredit the
relationship between KIEs and EIE proposed above is unclear.
We note, however, that like all gas-phase ion-molecule
reactions, the transition state is more stable than reactants, -12.2
kcal/mol in this case, so solvents play an important role and
may be responsible for the discrepancies. We shall continue to
employ the linear free energy relationship, KIE ) EIEi, as well
as pursue calculations to establish a meaningful relationship
between calculations on such transition states and the LFER.
Finally, we note that the calculated equilibrium deuterium
isotope effect for formation of the methyllithium-acrolein
complex prior to the transition state is 0.95 using scaled
harmonic frequencies. This value might be interpreted to indicate
that the origin of the inverse SDKIEs in the current study is
rate-determining complex formation. However, the very small
SDKIEs in the additions of butyllithium to benzaldehyde suggest
that complex formation is not rate-determining in these cases
and perhaps in no other organolithium addition. This is not to
argue that Lewis acid-carbonyl complexes29 could not be
formed in the rate-determining step of additions catalyzed by
strong Lewis acids such as boron trifluoride.
Procedure B. A solution of benzyl alcohol (0.5 g, 4 mmol) in hexane
(2 mL) was stirred with dimethyl sulfate (0.7 g, 5.5 mmol) and
triethylamine (0.020 g) over sodium hydroxide pellets (0.5 g, 12.5
mmol) for 8 h. The solution was then heated at reflux for 3 h, cooled,
and diluted with water (1 mL). After the mixture was stirred for 30
min, the organic layer was removed, and the aqueous layer was
extracted with benzene (1 mL). The combined organic layers were
washed with brine, dried over anhydrous magnesium sulfate, and
concentrated to give a yellow oil. The product was eluted through a
short column of silica gel with light petroleum ether/diethyl ether (5:
1). Preparative GC on carbowax 20M was performed prior to analysis
1
by H NMR.
1H NMR (methyl benzyl ether): δ 3.38 (s, 3H), 4.46 (s, 2H), 7.3
(m, 5H). For R-D-methyl benzyl ether, all resonances are in nearly the
same positions, except for that at δ 4.46, which appears at δ 4.44 (three
lines, J ) 1.6 Hz, 1H).
General Procedure for Determination of Kinetic Isotope Effects
from Lithium Aluminum Hydride, Lithium Triethylborohydride,
and Lithium Borohydride Reductions by GC. For the lithium
aluminum hydride, lithium triethylborohydride, and lithium borohydride
reductions in tetrahydrofuran, a typical experiment was conducted in a
50-mL three-neck flask which had been oven-dried and fitted with a
thermometer, magnetic stirrer, nitrogen inlet, and septum. The flask
was charged with 25 mL of dry THF and 2 mL of a 1:1 mixture of
Experimental Section
1
General. H NMR spectra were obtained on Bruker AM-500 (500
MHz), Varian VXR 400 (400 MHz), or Varian XL-300 (300 MHz)
spectrometers. Chemical shifts are reported in δ units relative to TMS
in chloroform-d solution, where chloroform-H is the reference which
is assumed to be at δ 7.26 ppm. Column chromatography were
performed on Merck silica gel (100-200 mesh), and thin-layer
chromatography was performed on Merck silica gel (0.25 mm)
precoated glass sheets. IR spectra were obtained on a Perkin-Elmer
298 or on a Mattson Instruments Galaxy 4020 Fourier transform
spectrophotometer. Preparative GC was performed using a Varian 4700
(29) Denmark, S. E.; Almstead, N. G. J. Am. Chem. Soc. 1993, 115,
3133 and references therein.
(30) Turner, L. J. Chem. Soc. 1956, 1686.

Secondary Deuterium Kinetic Isotope Effects
J. Am. Chem. Soc., Vol. 121, No. 2, 1999 333
benzaldehyde-H and benzaldehyde-D. After the flask was cooled to
-78 °C, 1 mL of a 1.0 M solution of lithium triethylborohydride as
added via syringe at such a rate as to maintain the temperature. After
addition, the reaction mixture was poured slowly into 50 mL of a 5%
aqueous sodium bicarbonate solution. Then, 30 mL of diethyl ether
was added, the organic layer was separated, and the aqueous layer was
extracted with 20 mL of diethyl ether. The combined organic layers
were washed with water, concentrated aqueous sodium bisulfate, water,
and saturated brine and then dried over anhydrous sodium sulfate. After
chromatography on silica gel, eluting with 10% ethyl acetate in hexane,
73 mg of benzyl alcohol was obtained and converted to the methyl
ether by procedure A. Capillary gas chromography on the nickel bis-
(3-heptafluorobutyryl-(1R)-camphorate/OV-1 column at 23 psi and 75
°C resulted in a separation of the D and H compounds by ca. 40 s
peak-to-peak at 19 min. The separation was within 10% of the baseline
relative to the smallest peak.
ether by procedure A above. 1H NMR was used to determine the isotope
ratios by comparing the integration of the methine and methoxy proton
signals after purification of the methyl ether by preparative GC on a
Carbowax column. Long delay times (20 s) were used between pulses
to prevent saturation. In cases where the kinetic isotope effect for loss
of benzaldehyde was determined, the average weighted ratio of the
aldehydic proton to the ortho, meta, and para protons, which are
separated to baseline at 400 MHz, was used for analysis.
1H NMR (methyl methylphenylcarbinyl ether): δ 1.45 (d, J ) 6.4
Hz, 3H), 3.22 (s, 3H), 4.29 (q, J ) 6.4 Hz, 1H), 7.3 (m, 5H).
1H NMR (methyl diphenylcarbinyl ether): δ 3.38 (s, 3H), 5.24 (s,
1H), 7.3 (bm, 10H).
1H NMR (methyl butylphenylcarbinyl ether): δ 0.87 (t, J ) 7.2
Hz, 3H), 1.3 (m, 4H), 1.62 (m, 1H), 1.80 (m, 1H), 3.20 (s, 3H), 4.07
(dd, J ) 6.8, 6.0, 1H), 7.3 (m, 5H).
1H NMR (methyl tert-butylphenylcarbinyl ether): δ 0.88 (s, 9H),
3.18 (s, 3H), 3.77 (s 1H), 7.3 (m, 5H).
The procedure was repeated with excess lithium triethylborohydride
to obtain the ratio of H and D in the starting mixture. All analyses
were performed in triplicate.
1H NMR (methyl allyl phenylcarbinyl ether): δ 2.49 (AB multiplet,
2H), 3.22 (s, 3H), 4.11 (dd, J ) 7.5, 6.0 Hz, 1H), 5.0 (br d, J ) 10.2,
5.04 (br d, J ) 17.0 Hz, 1H), 5.76 (ddt, J ) 17.0, 10.2, 7.0, 1H), 7.3
(m, 5H).
General Procedure for Determination of Kinetic Isotope Effects
from Lithium Tri-tert-butoxy Hydride, Sodium Borohydride,
Sodium Trimethoxy Borohydride, and Sodium Triacetoxyborohy-
dride Reductions by 1H NMR. For the lithium tri-tert-butoxy hydride
reduction in tetrahydrofuran, the sodium borohydride reduction in
ethanol, the sodium trimethoxy borohydride reduction in methanol, and
the sodium triacetoxyborohydride reduction in acetic acid, after the
reaction was quenched with water and the excess benzaldehyde removed
by washing the ether or ether-THF layer with aqueous sodium bisulfite,
followed by conversion of the resulting alcohol to the methyl ether by
procedure B above, 1H NMR was used to determine the isotope ratios.
The methylene singlet of PhCH2OCH3 was separated by 8.5 Hz from
the methine triplet of PhCHDOCH3 at 400 MHz, and the methoxy group
of the deuterated and nondeuterated compounds were separated by 1.5
Hz, while each had a line width of 0.4 Hz. Thus, the ratio of the
methylene singlet of protio material to the methoxy singlet of the
deuterated material was used to determine the kinetic isotope effects.
In a separate determination of the isotope effect in the sodium
borohydride reduction in methanol at -78 °C using CG on the nickel
complex capillary column, kH/kD) 0.792 was obtained, which compares
favorably with the value determined from the reaction in ethanol by
1H NMR, namely 0.82.
Procedure for Determination of Kinetic Isotope Effects from
Addition of Methyl, Phenyl, and Allyl Grignard to Benzaldehyde.
In a typical procedure, the Grignard reagent in ether was added to a
mixture of benzaldehyde and benzaldehyde-d, which was dissolved in
ether and cooled to -78 °C under a nitrogen atmosphere. To determine
the ratio of H and D benzaldehyde, excess Grignard reagent was used.
To determine the kinetic competition for reaction with H and D
benzaldehyde, 0.1 equiv of Grignard reagent was used. In each case,
the reaction was quenched with water, and in the case of use of a
deficiency of Grignard reagent, the excess benzaldehyde was removed
by washing the ether layer with aqueous sodium bisulfite. In all cases
when the resulting alcohol was converted to the methyl ether by
procedure A above, 1H NMR was used to determine the isotope ratios
by comparing the integration of the methine and the methoxy hydrogen
signals.
1H NMR (methyl methylcyclohexylcarbinyl ether): δ 1.0 (d
superimposed on multiplet, 5H), 1.2 (m, 3H), 1.39 (m, 1H), 1.62 (m,
3H), 1.74 (m, 3H), 3.02 (quint, J ) 6.1 Hz, 1H), 3.30 (s, 3H).
1H NMR (methyl allylcyclohexylcarbinyl ether): δ 1.1 (m, 5H), 1.45
(m, 1H), 1.7 (m, 3H), 2.28 (symm m, 2H), 2.94 (quart, J ) 5.5 Hz,
1H), 3.33 (s, 3H), 5.04 (d, J ) 10.0 Hz, 1H), 5.08 (d, J ) 19.0 Hz,
1H), 5.85 (symm m, 1H).
Determination of Kinetic Isotope Effects in the Recovered
Benzaldehyde. Reactions of benzaldehyde-H/D with methyllithium
were conducted as described above except that more than 0.1 equiv of
methyllithium was used. The NMR integration of the reaction mixture
after workup was used to determine the extent of reaction as well as
the deuterium content of the recovered benzaldehyde by integration of
the aldehyde and ortho hydrogens to allow calculation17 of the SDKIE.
At 17.2% conversion at -78 °C, the KIE was 0.90; at 69.8% conversion
the SDKIE was 1.04. In the addition of n-butyllithium at 17.6%
conversion the SDKIE was 0.99; at 66.8% conversion the SDKIE was
1.18. With tert-butyllithium addition at 5.6% conversion, the SDKIE
was 1.22, and at 69.3% conversion the SDKIE was 1.17. With
allyllithium addition, the SDKIE at 35.1% conversion was 1.12, and
at 64.8% conversion the SDKIE was 1.24. With phenyllithium addition
at 30.0% conversion, the SDKIE was 1.10.
Ab Initio Computations. Acrolein and its Hydride Addition
Product. Computations of the equilibrium isotope effect values were
performed using the Gaussian32 program at the MP2/6-311+G** level.
s-trans-Acrolein-H(D) was used to model benzaldehyde, and prope-
noxide was used to model benzyl oxide. Energy second derivatives
for Cs cisoid propenoxide, Cs methanol, C3V methoxide, C3V lithium
methoxide, and C3V sodium methoxide were computed at the minimum
energy geometries. The force constants were used to compute the
harmonic frequencies for protio and deuterio isotopomers. The frequen-
cies were scaled33 by a factor of 0.983 before being used to compute
the reduced isotopic partition function ratio34 for each molecule. The
reduced isotopic partition function ratios for propenol and the sodium
and lithium propenoxide salts were estimated from (s2/s1)fROM ) [(s2/
-
-
3
-
s1)fRO ][s2/s1)fCH OM/(s2/s1)fCH O ], where (s2/s1)fRO denotes the reduced
3
General Procedure for Determination of Kinetic Isotope Effects
from Methyllithium, Allyllithium, Phenyllithium, n-Butyllithium,
and tert-Butyllithium Additions to Benzaldehyde and Methyl- and
Allyllithium to Cyclohexanecarboxaldehyde by 1H NMR. Additions
of methyllithium in ether, allyllithium in ether,31 phenyllithium in ether,
n-butyllithium in ether, and tert-butyllithium in pentane to the aldehyde-
H/D were conducted by adding 10 mol % (or higher if H/D fractionation
in benzaldehyde was to be examined) of the reagent to the aldehyde at
-78 °C at the concentrations indicated in Tables 1 and 2. Within 10
min of addition, each reaction was quenched with acetic acid, and then
the organic layer was washed with aqueous bicarbonate followed by
aqueous sodium bisulfite to remove the excess aldehyde. After
evaporation of the solvent, the alcohol was converted to the methyl
-
isotopic partition function ratio for propenoxide anion, (s2/s1)fCH O
3
denotes that for methoxide, and (s2/s1)fCH OM denotes that for methanol,
3
lithium methoxide, or sodium methoxide. The value of (s2/s1)fCH OH
3
for methanol differs depending on whether the deuterium substitution
(32) Frisch, M. J.; Trucks, G. W.; Head-Gordon, M.; Gill, P. M. W.;
Wong, M. W.; Foresman, J. B.; Johnson, B. G.; Schlegel, H. B.; Robb, M.
A.; Replogle, E. S.; Gomperts, R.; Andres, J. L.; Raghavachari, K.; Binkley,
J. S.; Gonzalez, C. Martin, R. L. Fox, D. J.; Defrees, D. J.; Backer, J.;
Stewart, J. J. P.; Pople, J. A. Gaussian 92, Revision C; Gaussian, Inc.:
Pittsburgh, PA, 1992.
(33) Pople, J. A.; Scott, A. P.; Wong, M. W.; Radom, L. Isr. J. Chem.
1993, 33, 345.
(34) Bigeleisen, J.; Mayer, M. G. J. Chem. Phys. 1947, 15, 261.
Wolfsberg, M. Acc. Chem. Res. 1972, 5, 225. Hout, R. F., Jr.; Wolfsberg,
M.; Hehre, W. J. J. Am. Chem. Soc. 1980, 102, 3296.
(31) Seyferth, D.; Weiner, M. A. J. Org. Chem. 1961, 26, 4797.

334 J. Am. Chem. Soc., Vol. 121, No. 2, 1999
Gajewski et al.
is anti or gauche to the O-H bond. The value of (s2/s1)fpropenol was
computed using the geometric mean of these two values.
Acknowledgment. We thank the National Science Founda-
tion and the Department of Energy for financial support, the
Lubrizol Co. for a graduate fellowship to N.J.H., and Professor
Dennis Peters for helpful discussions.
Acrolein and Its Methyllithium Complex. Computations of the
equilibrium isotope effect values were performed at the MP2/6-31G**
level. The frequencies were used unscaled, although the calculated CH
stretching frequencies were roughly 8% high. Correction would reduce
the calculated value for the equilibrium in eq 2 to roughly 0.95.
JA982504R