Reaction of metalated nitriles with enones

Article abstract of DOI:10.1021/jo0478050

(Chemical Equation Presented) There have been a number of reports of the kinetic conjugate (1,4) addition of metalated arylacetonitriles to enones. Several proposals have been made to explain this behavior based on nucleophile structure or aggregation state or on the HSAB properties of the reactants. A reexamination of these studies showed that in each case the 1,4 adducts resulted from equilibration of the kinetically formed 1,2 adducts to the more stable 1,4 adducts. Thus, no conclusions about the origins of 1,4 selectivity can be drawn from these experiments. The 1,2 addition, retro-1,2 addition, 1,4 addition, and retro-1,4 addition of lithiophenylacetonitrile to benzylideneacetone were examined, and a free energy level diagram was constructed for the reaction.

Full text of DOI:10.1021/jo0478050

Reaction of Metalated Nitriles with Enones
Hans J. Reich,* Margaret M. Biddle, and Robert J. Edmonston
Department of Chemistry, University of Wisconsin, Madison, Wisconsin 53706
reich@chem.wisc.edu
Received December 15, 2004
There have been a number of reports of the kinetic conjugate (1,4) addition of metalated
arylacetonitriles to enones. Several proposals have been made to explain this behavior based on
nucleophile structure or aggregation state or on the HSAB properties of the reactants. A
reexamination of these studies showed that in each case the 1,4 adducts resulted from equilibration
of the kinetically formed 1,2 adducts to the more stable 1,4 adducts. Thus, no conclusions about
the origins of 1,4 selectivity can be drawn from these experiments. The 1,2 addition, retro-1,2
addition, 1,4 addition, and retro-1,4 addition of lithiophenylacetonitrile to benzylideneacetone were
examined, and a free energy level diagram was constructed for the reaction.
The addition of organolithium reagents to R,â-unsat-
urated carbonyl compounds can have synthetic utility
only if regioselectivity (1,2 vs 1,4 addition) can be
controlled. We have been interested in trying to under-
stand the mechanistic features that control selectivity1
and report here our studies of the regioselectivity of
addition of metalated phenylacetonitriles to enones.
were evoked to account for this uncommon regioselectiv-
ity. Stork and Maldonado proposed that the linear N-Li
structure of lithionitriles permitted an extended transi-
tion state (1) leading to 1,4 addition, whereas C-lithiated
reagents favored a more compact four-membered cyclic
transition state (2) which leads to 1,2 addition.
Aggregation Control of 1,2 vs 1,4 Addition. Re-
cently, 1,2 vs 1,4 regioselectivity of lithiophenylaceto-
nitrile has been correlated to the aggregation state of the
anion solution structures5a,b using integrated IR bands
to determine dimer and monomer concentrations in
various solvents. The monomer concentration showed
linear correlation to the 1,2 adduct with benzylidene-
acetone, and the dimer concentration correlated to the
1,4 adduct in THF and THF-toluene solutions (Figure
1). The authors proposed transition states analogous to
those proposed by Stork, with the additional refinement
that the linear N-Li four-center bonded dimer6a,7a led to
1,4 addition, whereas the monomer, suggested by ab
initio calculations (3-21+G* basis set) to have a C-
Stork and Maldonado2 made the synthetically valuable
observation in 1974 that metalated nitriles (protected
cyanohydrins) tend to give a larger proportion of 1,4
adduct than other acyl anion equivalents such as
lithiodithianes or metalated vinyl ethers.3a,4 The factors
controlling this regioselectivity have been a matter of
debate ever since. Several distinct mechanistic rationales
(3) (a) Seebach, D. Synthesis 1969, 17-36. (b) Juaristi, E.; Beck, A.
K.; Hansen, J.; Matt, T.; Mukhopadhyay, T.; Simson, M.; Seebach, D.
Synthesis 1993, 1271-90.
(4) Baldwin, J. E.; Ho¨fle, G. A.; Lever, O. W., Jr. J. Am. Chem. Soc.
1974, 96, 7125-7127.
(1) Reich, H. J.; Sikorski, W. H. J. Org. Chem. 1999, 64, 14-15.
Sikorski, W. H.; Reich, H. J. J. Am. Chem. Soc. 2001, 123, 6527-6535.
(2) Stork, G.; Maldonado, L. J. Am. Chem. Soc. 1974, 96, 5272-
5274.
10.1021/jo0478050 CCC: $30.25 © 2005 American Chemical Society
Published on Web 03/26/2005
J. Org. Chem. 2005, 70, 3375-3382
3375

Reich et al.
FIGURE 2. Effect of para substituents and solvent on the
ratio of 1,2 to 1,4 addition of benzaldehyde O-trimethylsilyl-
cyanohydrins in ether and DME.9a,b
FIGURE 1. Correlation between aggregation state and 1,4
addition for 3 and benzylideneacetone.5a Proposal for aggrega-
tion control of regioselectivity.
base8) properties of electrophiles and nucleophiles. Ahn,
Deschamps, and Seyden-Penne suggested that 1,2 addi-
tions should be more charge controlled (harder electro-
phile) and 1,4 addition more orbital controlled (softer
electrophile).5d Solvent effects on regioselectivity have
been interpreted in terms of ion-pairing effects.9a The
metalated benzaldehyde cyanohydrin 6-H is reported to
form the 1,4 adduct 7 in ether, whereas the silyl-shifted
1,2 adduct 8 is formed in THF or more polar solvent
combinations such as DME or HMPA (Figure 2). It was
reasoned that in ether the “softer” contact ion pair or an
aggregate of it is the reactive species, yielding the 1,4
product, whereas in more polar solvents the “harder”
solvent separated ion pair was responsible for 1,2 addi-
tion.
metalated structure,5c,6b underwent 1,2 addition. Higher
level calculations, including solvent molecules, predicted
that an N-metalated structure would be more favorable
for the monomer as well.14d
HSAB Control of 1,2 vs 1,4 Addition. Several
arguments rationalizing 1,2 vs 1,4 addition propensi-
ties have been made on the basis of frontier molecular
orbital arguments and the HSAB (hard and soft acid-
(5) (a) Strzalko, T.; Seyden-Penne, J.; Wartski, L.; Corset, J.;
Castella-Ventura, M.; Froment, F. J. Org. Chem. 1998, 63, 3295-3301.
Strzalko, T.; Seyden-Penne, J.; Wartski, L.; Froment, F.; Corset, J.
Tetrahedron Lett. 1994, 35, 3935-3936. (b) Croisat, D.; Seyden-Penne,
J.; Strzalko, T.; Wartski, L.; Corset, J.; Froment, F. J. Org. Chem. 1992,
57, 6435-6447. Corset, J.; Castella-Ventura, M.; Froment, F.; Strzalko,
T.; Wartski, L. J. Org. Chem. 2003, 68, 3902-3911. (c) Strzalko, T.;
Seyden-Penne, J.; Wartski, L.; Corset, J.; Castella-Ventura, M.;
Froment, F. J. Org. Chem. 1998, 63, 3287-3294. (d) Deschamps, B.;
Anh, N. T.; Seyden-Penne, J. Tetrahedron Lett. 1973, 14, 527-530.
Sauvetre, R.; Roux-Schmitt, M.-C.; Seyden-Penne, J. Tetrahedron 1978,
34, 2135-2140. (e) Roux-Schmitt, M.-C.; Wartski, L.; Seyden-Penne,
J. J. Chem. Res (S) 1980, 346; Roux-Schmitt, M.-C.; Wartski, L.;
Seyden-Penne, J. J. Chem. Res (M) 1980, 4141-4153. (f) Wartski, L.;
El Bouz, M.; Seyden-Penne, J.; Dumont, W.; Krief, A. Tetrahedron Lett.
1979, 20, 1543-1546. (g) Roux, M. C.; Wartski, L.; Seyden-Penne, J.
Tetrahedron 1981, 37, 1927-34. (h) Wartski, L.; El-Bouz, M.; Seyden-
Penne, J. J. Organomet. Chem. 1979, 177, 17-26.
(6) (a) Zarges, W.; Marsch, M.; Harms, K.; Boche, G. Angew. Chem.,
Int. Ed. Engl. 1989, 28, 1392-1394. Boche, G.; Marsch, M.; Harms,
K. Angew. Chem., Int. Ed. 1986, 25, 373-374. (b) The crystal structure
of a monomeric crown complex showed an N-metalated structure:
Langlotz, I.; Marsch, M.; Harms, K.; Boche, G. Z. Kristallogr. NCS
1999, 214, 509-510.
(7) (a) Carlier, P. R.; Lucht, B. L.; Collum, D. B. J. Am. Chem. Soc.
1994, 116, 11602-11603. (b) Galiano-Roth, A. S.; Kim, Y.-J.; Gilchrist,
J. H.; Harrison, A. T.; Fuller, D. J.; Collum, D. B. J. Am. Chem. Soc.
1991, 113, 5053-5055. Romesberg, F. E.; Collum, D. B. J. Am. Chem.
Soc. 1994, 116, 9198-9202. (c) Thompson A.; Corley E. G.; Huntington
M. F.; Grabowski E. J. J.; Remenar J. F.; Collum D. B. J. Am. Chem.
Soc. 1998, 120, 2028-2038. Briggs, T. F.; Winemiller, M. D.; Xiang,
B.; Collum D. B. J. Org. Chem. 2001, 66, 6291-6298. (d) Sun, X.;
Collum, D. B. J. Am. Chem. Soc. 2000, 122, 2459-2463. (e) DePue, J.
S.; Collum, D. B. J. Am. Chem. Soc. 1988, 110, 5518-5524.
(8) Ho, T.-L. Chem. Rev. 1975, 75, 1-20.
HSAB theory has also been invoked to account for
substituent effects on the regioselectivity of lithiated
nitriles. A series of para-substituted trimethylsilyl benz-
aldehyde cyanohydrins 6-X gave a linear correlation
between the Hammet σF constants and the percent of 1,4
addition to mesityl oxide (Figure 2).9b It was argued that
the p-cyano anion is extensively delocalized and thus a
softer nucleophile which prefers to add 1,4, whereas the
charge is more localized on the carbanion carbon in the
p-NMe2 compound, which then reacts at the harder
carbonyl position. A related paper on the addition of
lithioarylacetonitriles to mesityl oxide suggested that the
1,4 addition of the parent (3) was probably not kinetically
controlled but that of the p-methoxy analogue was.5e
A
rationale for the different behavior of p-methoxyphenyl-
(12) There has been considerable published work on the structure
of lithiophenylacetonitrile and analogues, including spectroscopic
7a, 13, 14a]
studies by NMR[5b,
and IR spectroscopy.[5b] X-ray crystal
structures have been reported,[6a] and the effect of polar cosolvents such
as TMEDA[7a] and HMPA has been studied.[5 14a] The reactivity has
g,
also been studied in several contexts, including the aldol condensa-
[5, 14b]
tion.
There have been computational studies of metalated nitrile
(9) (a) Hu¨nig, S.; Wehner, G. Chem. Ber. 1980, 113, 302-323. (b)
Hu¨nig, S.; Wehner, G. Chem. Ber. 1980, 113, 324-332 (c) Hu¨nig, S.;
Scha¨fer, M.; Schweeberg, W. Chem. Ber. 1993, 126, 205-210.
(10) Dolak, T. M.; Bryson, T. A. Tetrahedron Lett. 1977, 1961-64.
Brown, C. A.; Yamaichi, A. J. Chem. Soc., Chem. Commun. 1979, 100-
101. Cohen, T.; Abraham, W. D.; Myers, M. J. Am. Chem. Soc. 1987,
109, 7923-24.
structure and reactivity.[5c, 14d, 15, 16]
(13) Abbotto, A.; Bradamante, S.; Pagani, G. A. J. Org. Chem. 1993,
58, 449-455.
(14) (a) Carlier, P. R.; Lo, C. W.-S. J. Am. Chem. Soc. 2000, 122,
12819-12823. (b) Carlier, P. R.; Lo, K.-M. J. Org. Chem. 1994, 59,
4053-4055. Carlier, P. R.; Lo, K.-M.; Lo, M. M.-C.; Williams, I. D. J.
Org. Chem. 1995, 60, 7511-7517. (c) Carlier, P. R.; Lo, C. W.-S.; Lo,
M. M.-C.; Wan, N. C.; Williams, I. D. Org. Lett. 2000, 2, 2443-2445.
(d) Carlier, P. R.; Madura, J. D. J. Org. Chem. 2002, 67, 3832-3840.
(11) Carreira, E. M.; Du Bois, J. J. Am. Chem. Soc. 1995, 117, 8106-
8125.
3376 J. Org. Chem., Vol. 70, No. 9, 2005

Reaction of Metalated Nitriles with Enones
acetonitrile and 6-OMe (which gives mostly 1,2 addition)
was suggested based on the possibly more pyramidal
structure of the siloxy substituted carbanion center in
6-OMe.
We became interested in this question, and particularly
the aggregation hypothesis of Figure 1,5a as a follow up
on our studies of the effect of polar cosolvents such as
HMPA and DMPU on the 1,2/1,4 dichotomy in the
reaction of sulfur-substituted organolithium reagents
with enones and enals.1 In this work we provided support
for the notion5f,10 that ion-pair status (contact vs sepa-
rated ion pairs) and lithium solvation levels were major
determinants of the propensity for 1,4 addition. The
direct correlation between dimer concentration and 1,4
addition for the lithiophenylacetonitrile reactions repre-
sents a rare case where selectivity could be directly
attributed to aggregate structure, a concept frequently
invoked to rationalize selectivity changes in synthetically
useful reactions.11 This observation implies strongly that
the Curtin-Hammett principle does not apply, i.e., that
the rate of reaction is faster than the rate of aggregate
interconversion, an effect possibly subject to direct ex-
perimental test.
Implicit in the aforementioned proposals is the pre-
sumption that products were formed under kinetic con-
trol, since product ratios formed under thermodynamic
control provide no information as to the nature of
transition states. Indeed, all of the workers in the area
recognized the possibility that the 1,4 adduct, which is
usually thermodynamically more stable than the 1,2
adduct, could arise simply by equilibration if the addition
were reversible. The isomerization of 1,2 to 1,4 adducts
in aldol additions to enones has been detected many
times, including during the additions of lithioaceto-
nitriles.5d,h,9c We report here that in all of the above cited
cases, each of which implied or explicitly claimed kinetic
formation of 1,4 -addition products, these arise not from
any special transition state effects, but from equilibration
of the kinetically formed 1,2 adducts to the thermody-
namically favored 1,4 adducts.
FIGURE 3. Isomerization of 9 to 10 in (a) THF and (b) 30:70
THF/toluene at -78 °C. Each point is a separate experiment:
9, 1,2 adduct 9 formed by reaction of 3 with benzylidene-
acetone (the solid lines are best fit kinetic simulations of these
points using the bimodal scheme and rate constants shown);
4, 1,2 adduct formed by deprotonation of 4. f 1,2 to 1,4
isomerization in THF in the presence of 1 equiv of LiOPh (the
dashed line (k′2) corresponds to first-order conversion).
SCHEME 1
quench with a mixture of propionic acid and benzalde-
hyde, the latter to establish that there was some free
lithiophenylacetonitrile left. Such an experiment at -98
°C with an 8 s reaction time gave 94% conversion to 4,
corresponding to k1,2 ) 80 M-1 s-1, ∆Gq ) 8.5 kcal/mol. A
high dilution experiment (0.001 M) at -78 °C gave 83%
Results and Discussion
Aggregation Control. Our examination of the reac-
tion of lithiophenylacetonitrile (3)12 with benzylidene-
acetone gave quite different results from those reported
in Figure 1.5a We observed a time dependence on the
product ratios, with high ratios of 1,2 to 1,4 adduct if
reaction times were short (<30 s) and an effective quench
was used (Figure 3). At longer times, 1,4 products
increasingly predominated, with eventual complete con-
version to the 1,4 adduct.
This behavior can be explained with the hypothesis
shown in Scheme 1. The lithiophenylacetonitrile adds
quickly and reversibly at the carbonyl carbon to afford
the 1,2 adduct. Upon retro 1,2 addition, a slower conju-
gate addition affords the 1,4 adduct. We set out to
estimate values for the rate constants with the goal of
constructing an approximate free energy diagram for this
scheme.
conversion in 3 s, corresponding to k1,2 ) 400 M-1 s-1
,
∆Gq ) 8.9 kcal/mol. The diastereomer ratio of the 1,2
addition products 4A and 4B (relative stereochemistry
was not assigned) in these experiments varied slightly
depending on conditions. At the shortest reaction times
(<10 s), where the diastereomer ratio is close to kineti-
cally controlled, ratios as high as 67/33 were observed.
Later in the reaction (10-120 min), when substantial
product had built up, the ratio became close to 1:1. Since
the 1,2-adducts must be at equilibrium on any time scale
greater than 1 min or so, the small drift in ratios observed
as the isomerization of 9 to 10 proceeded can be at-
tributed to changes in the equilibrium ratio as various
mixed aggregates between 9 and 10 are formed.
The Retro-1,2 Addition (Retro-Aldol, k-1,2). The
alkoxide 9 formed by deprotonation of the 1,2 adduct 4
with LDA was converted to the 1,4 adduct 10 at -78 °C
in THF at a rate comparable to reaction of 3 with
benzylideneacetone (Figure 3). This alone does not es-
tablish that the retro-aldol process was occurring, since
The 1,2 Addition (Aldol, k1,2). The rate of 1,2
addition was too fast to measure at -78 °C at the
concentrations usually used (0.08 M). Typical experi-
ments involve an excess of the benzylideneacetone and
J. Org. Chem, Vol. 70, No. 9, 2005 3377

Reich et al.
benzaldehyde.17 With these limitations in mind, the
trapping occurs on a comparable time scale to that
observed for the diastereomer interconversion. We esti-
mate ∆Gq
for the retro-aldol (k-1,2) as 12.4 kcal/mol.
-78
The 1,2 to 1,4 Isomerization. Initial 1,2 to 1,4 ratios
as high as 98/2 were achieved in THF at -78 °C. In 30:
70 THF/toluene the highest ratio observed at -78 °C was
91/9, but ratios as high as 97/3 were observed at -98 °C.
We presume that our quenching procedure (addition of
a 3 M solution of propionic acid in ether) does not work
quite as well in the less polar solvent mixture as in pure
THF. This may also be an explanation for the reported
increase in the fraction of 1,4 adduct as solvent polarity
was decreased (Figure 1). The quenching procedure used
(addition of 2 M aqueous HCl) might become increasingly
ineffective as the solvent became less polar, resulting in
more isomerization before the reaction was stopped.
The 1,2 to 1,4 adduct isomerization showed a distinct
bimodal dependence on time. Isomerization was fast up
to about 50% conversion, with a half-life of ca. 4 min at
-78 °C. The reaction then slowed drastically, with the
last half proceeding about 1/20 as fast (Figure 3).18 It is
this two-phase process that can lead to a false impression
that the product ratio formed in the first few minutes is
the kinetic one. In fact, Strzalko and co-workers used
reaction times of 5 and 30 min in their study,5a just the
interval in which the fast phase is almost completed, and
the slow phase has progressed only partially. This
problem is exacerbated by the inherent temperature
increases caused by addition of the reactant and quench-
ing solutions to the reaction mixture.
One possible explanation for the bimodal kinetics is
that the aggregate of the 1,2 adduct 9 isomerizes rapidly,
whereas mixed aggregates between 9 and 10, formed as
the reaction proceeds, do the retro-aldol much more
slowly. Although the actual aggregation states have not
been determined, and our data is not accurate enough
to individually define the role of multiple aggregates, the
kinetics can be simulated nicely (solid line in Figure 3)
by the following simple kinetic model. We assume that
initially the 1,2 adduct exists as a dimer (9)2, which
isomerizes rapidly to the mixed aggregate 9‚10. This
species isomerizes to (10)2 at about 1/20 the rate of the
homodimer.19 To observe this product inhibition the
mixed aggregate 9‚10 must be thermodynamically fa-
vored over the homodimers, although it could also be
FIGURE 4. Interconversion of the diastereomers of 4 in THF
at -78 °C (relative stereochemistry was not assigned). The
lines are calculated for first-order conversion of each isomer
to the equilibrium 60/40 mixture with k ) 0.06 s-1
.
a 1,3-sigmatropic rearrangement mechanism is conceiv-
able. We performed two types of experiments to estimate
the actual rate of the retro-aldol process: interconversion
of the 1,2 adduct diastereomers and external trapping
of lithionitrile 3.
The alkoxide diastereomers 9A and 9B are present in
a 60:40 ratio at equilibrium.14b,c Chromatographically
purified alcohol 4 enriched in each of its diastereomers
4A and 4B was deprotonated with LDA at -78° C in THF
and then quenched with a propionic acid-Et2O or pro-
pionic acid-methanol solution after reaction times of
5-100 s. Less than 8% of 1,4 adduct was formed in these
experiments, but 1-11% of benzylideneacetone was
formed, suggesting that the retro-aldol was competitive
with mixing of the reagents, leading to protonation of the
lithiophenylacetonitrile by the alcohol. There was too
much scatter in the points to obtain reliable rate con-
stants, presumably because mixing and temperature
control is difficult in these very short experiments, but
we estimate a half-life of 11 s for the isomerization
(Figure 4). The very fast interconversion of diastereomers
(about 50 times as fast as regioisomer interconversion)
means that kinetic and thermodynamic diastereoselec-
tivity can be misidentified even more easily than
regioselectivity.5g,14b
To provide a second estimate of the retro-aldol rate we
performed a series of experiments which involved trap-
ping of the product 3 with an electrophile. Benzaldehyde
was shown to be an effective trap for lithionitrile 3 by
adding a 1:1 mixture of benzaldehyde and benzylidene-
acetone to the lithiophenylacetonitrile and observing a
99/1 ratio of aldols favoring the benzaldehyde adduct 11
(a 80:20 ratio of anti/syn isomers was formed14b). When
a mixture of 4 and excess benzaldehyde were treated with
LiN(SiMe3)2, 11 was formed rapidly, although these
experiments also showed considerable scatter (half-life
of 10-30 s). In addition to the short reaction times and
mixing problems present in the diastereomer intercon-
version, there are other potential problems. Addition of
LiN(SiMe3)2 to the benzaldehyde in competition with
deprotonation of 4 would lower the concentration of both
benzaldehyde and the alkoxide 9. The concentration of
benzaldehyde could also be lowered by addition of 9 to
the benzaldehyde to form a lithiohemiacetal, similar to
those detected by McGarrity and co-workers in their
rapid injection NMR study of the reaction of BuLi with
(15) Wu, Y. D.; Houk, K. N.; Florez, J.; Trost, B. M. J. Org. Chem.
1991, 56, 3656-3664. Wong, S. S.; Padden-Row, M. N.; Li, Y.; Houk,
K. N. J. Am. Chem. Soc. 1990, 112, 8679-8686.
(16) Kaneti, J.; Schleyer, P. v. R.; Clark, T.; Kos, A. J.; Spitznagel,
G. W.; Andrade, J. G.; Moffat, J. B. J. Am. Chem. Soc. 1986, 108, 1481-
1492. Dorigo, A. E.; Morokuma, K. J. Am. Chem. Soc. 1989, 111, 4635-
4643.
(17) McGarrity, J. F.; Ogle, C. A.; Brich, Z.; Loosli, H.-R. J. Am.
Chem. Soc. 1985, 107, 1810-1815.
(18) The rates in both phases were independent of concentration
within experimental error over a factor of 12 in concentration.
(19) (a) Abu-Hasanayn, F.; Streitwieser, A. J. Am. Chem. Soc. 1996,
118, 8136-8137. (b) Streitwieser, A.; Juaristi, E.; Kim, Y.-J.; Pugh, J.
K. Org. Lett. 2000, 2, 3739-3741.
3378 J. Org. Chem., Vol. 70, No. 9, 2005

Reaction of Metalated Nitriles with Enones
SCHEME 2
SCHEME 3
by a factor of 4. There was no significant substituent
effect on the equilibration, a 1:1 ratio of either 10 and
12 or 13 and 3 each went to a 1:1 ratio of 10 and 13 at
comparable rates. Rates were measured at eight tem-
peratures between -20 and 20 °C. An Eyring plot gave
∆Hq ) 15.1 ( 1.0 kcal/mol, ∆Sq ) -19 ( 4 eu. Extrapola-
kinetically trapped if interconversion among the ag-
gregates (9)2, (10)2, and 9‚10 was slower than isomer-
ization. The rates measured for conversion of 1,2 to 1,4
adduct in THF at -78 °C correspond to ∆Gq
) 13.5
-78
tion of the rates to -78 °C gave ∆Gq
) 18.7 kcal/mol.
kcal/mol for the initial fast phase and ∆Gq-78 ) 14.6 kcal/
mol for the slower second phase. We emphasize that this
scheme is a functional kinetic model only. The actual
process could involve a series of mixed tetramers with
the rate decreasing sharply as alkoxides 9 are replaced
byproduct enolates 10 in the aggregates.20 An equivalent
kinetic scheme to the one above would replace dimers
(9)2, 9‚10, and (10)2 by the tetramers (9)4, (9)2‚(10)2, and
(10)2.
-78
A long-term experiment at -78 °C (120.5 h) gave a barely
detected amount of crossover product (0.4%) which cor-
responds to a k-1,4 of 9.2 × 10-9 s-1, (∆Gq ) 18.4 kcal/
mol), in good agreement with the results extrapolated
from higher temperatures.
We observed a downward drift of the infinity point
(from the expected value of 67% conversion) in some of
the kinetic runs of Scheme 2. A potential explanation for
this observation is that the 1,4 adduct enolates (10 and
13) are probably more basic than the metalated nitrile
14 and its m-methoxy analogue, judging from the pKa
(DMSO) of diethyl ketone (27.127a) and R-phenylpropio-
nitrile (23.127b), so that rearrangement of 10 to 14 could
occur. We performed a control experiment by treating
benzylideneacetone with 3 equiv of 3 to mimic the
presence of 2 equiv of 12 in the experiment of Scheme 2.
After reaction time long enough for >80% conversion (2
h at 0 °C) the reaction was quenched with DOAc.
We have performed a test for this behavior by prepar-
ing 9 in the presence of 1 equiv of lithium phenoxide, an
unreactive surrogate for the enolate 10. We hypothesized
that formation of mixed aggregates between 9 and LiOPh
might inhibit the isomerization of 9 to 10 similarly to
that provided by mixed aggregates between 9 and 10.21a
In fact, with lithium phenoxide present there is little or
no “fast phase” for the isomerization (Figure 3); the entire
isomerization now proceeds at a rate comparable to that
of the slow phase in the absence of phenoxide.22
The rate of 1,2 addition must be 50 times as fast as
that of 1,4 addition (k1,2/k1,4 Scheme 1), since we have
observed kinetic 1,2 to 1,4 ratios as high as 50:1. Thus,
1
2
Analysis by H and H NMR spectroscopy showed 84%
deuteration R to the ketone and only 2% deuteration at
the benzylic nitrile position (Scheme 3). Thus, enolate
equilibration was insignificant during this particular
experiment.
the rate of isomer interconversion (essentially k-1,2
)
should be 50× as fast as the fast initial phase of the 1,2
to 1,4 isomerization. The observed value is a factor of 20.
These numbers are probably within experimental error.
The Retro 1,4 Addition (Retro-Michael, k-1,4). A
series of crossover experiments were performed to de-
termine the rate of retro-1,4 addition of 10 (Scheme 2).
A solution of 10 was prepared in situ by the reaction of
lithiophenylacetonitrile with benzylideneacetone and
warming to 0 °C to complete the isomerization. This
solution was then treated with a 2-fold excess of lithio-
m-methoxyphenylacetonitrile 12, and the crossover prod-
uct 13, formed by reaction of 12 with the benzylidene-
acetone released in the retro 1,4 addition, was monitored.
The formation of 13 was readily detected at -25 °C or
higher. The rate of conversion followed reasonable first-
order kinetics, and the rate was unaffected after dilution
From these data, a rough energy diagram of the reac-
tion between R-lithiophenylacetonitrile and benzylidene-
acetone in THF can be constructed (Figure 5). The
highest selectivity for 1,2 vs 1,4 addition that we have
measured is about 50/1, corresponding to ∆∆Gq ) 1.5
kcal/mol, whereas the same selectivity measured from
the kinetic data was 1.1 kcal/mol (factor of 20), which
we consider good agreement. From the energy diagram
we can determine that the free energy difference favoring
the 1,4 adduct is about 4.9 kcal/mol. The exothermicity
of the 1,2 addition is >3.5 kcal/mol, and the exothermicity
of the 1,4 addition is >8 kcal/mol.
The HSAB Hypothesis. Interpretation of the linear
correlation between σpara and the ratio of 1,2 to 1,4
(23) (a) Hilmersson, G.; Malmros, B. Chem. Europ. J. 2001, 7, 337-
341. (b) Sott, R.; Granander, J.; Hilmersson, G. Chem. Europ. J. 2002,
8, 2081-2087.
(24) Corruble, A.; Davoust, D.; Desjardins, S.; Fressigne, C.; Giess-
ner-Prettre, C.; Harrison-Marchand, A.; Houte, H.; Lasne, M.-C.;
Maddaluno, J.; Oulyadi, H.; Valnot, J.-Y. J. Am. Chem. Soc. 2002, 124,
15267-15279.
(25) Williard, P. G.; Hintze, M. J. J. Am. Chem. Soc. 1990, 112,
8602-8604.
(26) Sugasawa, K.; Shindo, M.; Noguchi, H.; Koga, K. Tetrahedron
Lett. 1996, 37, 7377-7380.
(27) (a) Matthews, W. S.; Bares, J. E.; Bartmess, J. E.; Bordwell, F.
G.; Cornforth, F. J.; Drucker, G. E.; Margolin, Z.; McCallum, R. J.;
McCollum, G. J.; Vanier, N. R. J. Am. Chem. Soc. 1975, 97, 7006-
7014. (b) Bordwell, F. G.; Cheng, J.-P.; Bausch, M. J.; Bares, J. E. J.
Phys. Org. Chem. 1988, 1, 209-223.
(20) Lithium tert-butoxide, lithium phenoxide and the lithium
enolate of acetophenone are all approximately tetrameric in THF:
Arnett, E. M.; Moe, K. D. J. Am. Chem. Soc. 1991, 113, 7288-7293.
(21) (a) For studies of Lithium phenoxides as enolate surrogates
see: Jackman, L. M.; Smith, B. D. J. Am. Chem. Soc. 1988, 110, 3829-
3835. Jackman, L. M.; Chen, X. J. Am. Chem. Soc. 1992, 114, 403-
411. Jackman, L. M.; Chen, X. J. Am. Chem. Soc. 1997, 119, 8681-
8684. (b) Jackman, L. M.; Dunne, T. S. J. Am. Chem. Soc. 1985, 107,
2805-2806. Jackman, L. M.; Lange, B. C. J. Am. Chem. Soc. 1981,
103, 4494-4499.
(22) The role of alkoxide and amide mixed aggregates has been
explored in a variety of contexts: lithium enolate/LiX,[7b, 19a, 21b] BuLi/
BuOLi,[17] RLi/LiNR2,[23a, 24] lithium acetylide/ROLi,[7c] lithium enolate/
LiNR ,[3b, 7d, 25] R2NLi/LiX,[7e, 26] lithionitrile/LiNR2.
[6a, 7a, 23b]
2
J. Org. Chem, Vol. 70, No. 9, 2005 3379

Reich et al.
FIGURE 6. Time course of the fraction of 1,4 adduct in the
reaction of 17-X with mesityl oxide in THF at -78 °C to form
20-X and 21-X.
FIGURE 5. Energy diagram for the 3/9/10 system at -78 °C
in THF.
CF3, CN)28 which should be electronically similar to the
O-trimethylsilyl analogues 6-X (Scheme 5).
SCHEME 4
The results obtained for the p-NMe2 (17-NMe2) com-
pound in THF parallel those reported for the silyl
analogue 6-NMe2 in DME, giving only 1,2 adduct in high
yield at -78 °C even after reaction times of several hours
(Figure 6). Only at much longer reaction times or when
the temperature was raised well above -78 °C or when
the polar cosolvent HMPA was added was isomerization
to the 1,4 adduct observed. HMPA is known to facilitate
retroaldol reactions of lithionitriles with aldehydes.14c
With compounds 17-Cl and 17-H our results were quite
different than Hu¨nig’s results for 6-Cl and 6-H. For each
we were able to observe >95% 1,2 addition at short
reaction times (5-20 s). Thus, here, as for all the systems
studied except 17-CN and 17-CF3, the kinetic product
is almost completely the 1,2 adduct. Our experiments
were done in THF, whereas Hu¨nig’s were done in DME,
so we have performed an experiment with 17-Cl in DME,
and found that 1,2 adduct was also the kinetically formed
product at short reaction times.
SCHEME 5
Compound 17-CN, on the other hand, gave only the
1,4 adduct under all conditions tried, like that reported
for 6-CN. With short reaction times (40 s at -78 °C) only
30% conversion to products was observed, showing that
the 1,4 adduct was a true kinetic product. A reasonable
explanation for this observation is that the 1,2 addition
has become endothermic (k-1,2 > k1,2 in Scheme 4) for the
strongly stabilized p-CN lithionitrile. We estimated above
that the benzylideneacetone-lithiophenylacetonitrile 1,2
addition was around 4 kcal/mol exothermic (Figure 5).
In DMSO the additional anion stabilization provided by
the p-CN group in phenylacetonitrile is about 8 kcal/mol
(the pKa of phenylacetonitrile in DMSO is 21.9, for the
p-cyano it is 16.027b). Although this energy difference
would be considerably smaller under the ion-paired and
aggregated conditions of our experiments in THF, it could
easily be enough to invert the stability of 1,2 adduct vs
lithionitrile and enone. It is not unusual for aldols with
ketone acceptors to be thermodynamically unfavorable.29
The p-CF3 compound showed behavior intermediate
between that of 17-CN and 17-Cl. The addition was slow
addition reported by Hu¨nig and Wehner (Figure 2)9b is
complicated by the O to O silyl migration of the initial
1,2 adduct 15-X, which irreversibly leads to formation
of the ketone 8-X. An alternative explanation of their
results is that silyl shift (kSi) proceeds at a relatively
constant rate for the various p-substituents, whereas the
retro-aldol rate(k-1,2), and hence, the rate of the isomer-
ization of the 1,2 to the 1,4 adduct increases in the series
from p-NMe2 to p-CN as the anion becomes increasingly
stabilized (Scheme 4). The correlation between 1,4 ad-
dition and σpara in Figure 2 is then the result of competi-
tion between the silyl shift and the retro-1,2 addition (kSi
and k-1,2) and not due to any inherent HSAB properties
of the lithium reagents.
(28) Utimoto, K.; Wakabayashi, Y.; Shishiyama, Y.; Inoue, M.;
Nozaki, H. Tetrahedron Lett. 1981, 22, 4279-4280.
(29) Conant, J. N.; Tuttle, N. Org. Synth. Coll. Vol I, 1932, 199-
202.
To avoid this problem, we have investigated a series
of O-methyl cyanohydrins (17-X, X ) NMe2, H, Cl, Br,
3380 J. Org. Chem., Vol. 70, No. 9, 2005

Reaction of Metalated Nitriles with Enones
5 mL, long-necked (5 cm, to minimize warming by conduction
of heat) round-bottom flask. To this was added 3 mL of dry
THF, the system was cooled to -78 °C under positive nitrogen
pressure, and 120 µL (0.25 mmol) of 2.08 M n-BuLi was added.
The nitrogen inlet needle was removed, the puncture septum
was greased to prevent quenching, lithiation was continued
for 20 min, and 100 µL of 2.5 M benzylideneacetone solution
was injected into the flask. The solution was stirred for the
appropriate time and quenched with 200 µL of 3 M propionic
acid in diethyl ether. The solution was taken up in 1:1 hexanes/
diethyl ether, washed once with saturated NaHCO3 solution,
twice with water, and once with brine, dried with anhydrous
MgSO4, and rotary evaporated. To this was added pentachloro-
ethane, (15 µL, 0.125 mmol) as an internal NMR standard,
and the solution was mixed thoroughly with deuterated
chloroform with 0.03% tetramethylsilane. Spectra were taken
on a 300 MHz NMR spectrometer. Data are displayed in
Figure 3 and in Table S-1.31
High-Dilution Experiment. The high-dilution experiment
to measure k1,2 was performed as above with the following
changes: phenylacetonitrile (22 µL, 0.19 mmol) was added to
a nitrogen-purged 250 mL RBF containing 90 mL of THF
metalated with 75 µL (0.19 mmol) of 2.6 M n-BuLi, and 100
µL of 0.95 M benzylideneacetone solution was injected into the
flask. The solution was stirred for 3 s and quenched with a
mixture of 50 µL of benzaldehyde and 100 µL of 3 M propionic
acid in diethyl ether.
enough at -78 °C that the progress of the reaction could
be observed (half-life ca. 20 s), but the 1,2 adduct never
built up to more than 10% of the product. Here the rate
of retro-aldol (k-1,2 of Scheme 4) has become comparable
to the addition rate k1,2 (i.e., the free energy of 1,2 adduct
18-CF3 is similar to that of 17-CF3 and mesityl oxide).
Another unusual regioselectivity reported was that the
addition of 6-H to mesityl oxide in ether produced only
1,4 adduct, whereas the reaction in DME or THF gave
largely 1,2 adduct (Figure 2).9a This was also rationalized
in HSAB terms, with the softer contact ion pair in ether
undergoing 1,4 addition, whereas the harder solvent
separated ion pair (SIP) in THF and DME gave 1,2
addition. Here the experiment may be complicated by the
fact that 6-H was insoluble in ether, and an aqueous
NH4Cl quench was used, which may not have reacted at
-78 °C. We have performed an experiment with 17-H
and mesityl oxide in ether and find that 98% of 1,2 adduct
was formed (<2% of unreacted starting materials, 80%
yield) at -78 °C after 12 s reaction time. We thus suggest
that none of the product ratios in ether summarized in
Figure 2 represent kinetic product ratios, and these
results permit no conclusions about the nature of interac-
tions at the transition state.
Test for Unreactive Mixed Aggregate with LiOPh. The
same conditions were used as above with the following
changes: phenol (24 mg, 0.25 mmol) was added to the flask
before purging with nitrogen, 200 µL (0.50 mmol) of 2.5 M
n-BuLi was added, and the reactions were quenched with 400
µL of 3 M propionic acid in diethyl ether. Data are displayed
in Figure 3 and in Table S-2.31
We also reinvestigated the report that lithio-p-meth-
oxyphenylacetonitrile underwent kinetically controlled
1,4 addition to mesityl oxide in THF.5e This report is also
in error. We find that >99% of the product was the 1,2
adduct (7 s at -78 °C), which isomerized to 48% 1,4
adduct in 19 min and 93% after 2 h.
Preparation of 4A and 4B. The same reaction conditions
were used as for 3 reacting with benzylideneacetone, except
that the scale was increased to 2 mmol, the temperature
decreased to -100 °C to suppress formation of 5, the reaction
time was 5 s, and products were purified on a silica gel column
eluted with 2:1 hex/Et2O to separate diastereomers. 4A: Rf )
0.19. 4B: Rf ) 0.10. 3-Hydroxy-2,5-diphenyl-3-methylpent-4-
enenitrile (4A): 1H NMR (300 MHz, CDCl3) δ 7.4-7.2 (m,
10H), 6.62 (d, J ) 16.0 Hz, 1H), 6.28 (d, J ) 16.0 Hz, 1 H),
3.95 (s, 1H), 2.27 (s, OH), 1.49 (s, 3H). (4B): δ 7.4-7.2 (m,
10H), 6.58 (d, J ) 16.0 Hz, 1H), 6.27 (d, J ) 16.0 Hz, 1H),
3.92 (s, 1H), 2.25 (s, OH), 1.52 (s, 3H). Full characterization
data can be found in the Supporting Information.5d
Sample Reaction of 1,2 Adduct (4) Rearranging to 1,4
Adduct (5). Chromatographically purified (2:1 hex/Et2O, silica
gel) 4 (33 mg, 0.125 mmol) was added to a 5 mL long-necked
RBF, the flask was purged with nitrogen, 3 mL of dry THF
was added, the system was cooled to -78 °C under positive
nitrogen pressure, and 125 µL of 1.0 M LDA (0.25 mmol) was
added. The resulting solution was stirred for the appropriate
time and quenched with 100 µL of 3 M propionic acid in diethyl
ether. The solution was worked up as in reaction of 3 with
benzylideneacetone. Data are presented in Figure 3 and
tabulated in Table S-3.31
Summary. We have reexamined the experimental
evidence supporting several widely held and appealing
hypotheses to explain the course of addition of metalated
nitriles to R,â-unsaturated ketones: that soft anions
prefer 1,4 additions, whereas hard anions prefer 1,2
additions, and that the N-metalated linear structures of
dimeric lithioacetonitriles favor 1,4 additions whereas
monomers or SIPs form 1,2 adducts. For the specific cases
of the addition of 3 to benzylideneacetone and addition
of 17-H, -Cl, and -NMe2 (and by extension to 6 analogues)
to mesityl oxide, we find that the experimental evidence
does not support an interpretation in terms of these
hypotheses, but that in each case the 1,4 products are
formed under thermodynamic control by isomerization
of the kinetically favored 1,2 adducts. That is not to say
that these effects do not exist, but in these systems they
do not overwhelm the high kinetic preference for 1,2
addition. The special cases of 17-CF3 and 17-CN with
mesityl oxide can be rationalized on the basis that the
1,2 addition has become endothermic, so the 1,2 adduct
18-X cannot build up in the reaction mixture.
Diastereomer Interconversion Studies: Retro-Aldol.
Chromatographically purified (2:1 hex/Et2O, silica gel) 4 (44
mg, 0.16 mmol) enriched in the minor diastereomer (4A/4B )
20:80) was added to a 5 mL long-necked RBF. The flask was
purged with nitrogen, 2 mL of THF was added, and this was
Experimental Section
General Methods. Tetrahydrofuran was freshly distilled
from sodium benzophenone ketyl before use. All glassware was
placed in an oven at 110 °C overnight or flame-dried prior to
use. n-BuLi was titrated with n-propanol in THF at -78 °C
with 1,10-phenanthroline as an indicator. Dry ice/acetone
slurry was used to achieve -78 °C temperature baths and
N2(l)/ethanol to attain -100 °C baths. The phenylmethoxyac-
etonitriles were prepared by reaction of benzaldehyde di-
methylacetals30 with Me3SiCN.28
(30) The dimethyl acetals were prepared using the procedure
reported by: Creary, X.; Aldridge, T. J. Org. Chem. 1991, 56, 4280-
4285. All are known compounds. p-H, p-Cl, p-NMe2 were reported by:
Hassner, A.; Wiederkehr, R.; Kascheres, A. J. J. Org. Chem. 1970, 35,
1962-1964, among others and p-CN and p-CF3 were reported by
Mueller, K.; Guerster, D.; Piwek, S.; Wiegrebe, W. J. Med. Chem. 1993,
36, 4099-4107.
Sample Reaction of 3 with Benzylideneacetone. Phenyl-
acetonitrile (29 µL, 0.25 mmol) was added to a nitrogen-purged
(31) Figures and Tables with an S prefix are in the Supporting
Information.
J. Org. Chem, Vol. 70, No. 9, 2005 3381

Reich et al.
cooled to -78 °C under positive nitrogen pressure and meta-
lated with 100 µL of 1.59 M LDA (0.16 mmol). The solution
was stirred for the indicated time (Table S-4) and quenched
with 140 µL of 3 M propionic acid in diethyl ether. The solution
was worked up as in reaction of 3 with benzylideneacetone.
Data are reported in Figure 4 and tabulated in Table S-4.31
Benzaldehyde Trapping Studies: Retro-Aldol. To a 5
mL nitrogen-purged long-necked RBF was added 69 mg (0.26
mmol)of 4, 100 µL of benzaldehyde (1.0 mmol), and 3 mL of
THF. The solution was cooled to -78 °C under positive
nitrogen pressure and metalated with 260 µL of 1.0 M
LiN(SiMe3)2 (0.26 mmol). The resulting solution was stirred
for the appropriate time and quenched with 100 µL of 3 M
propionic acid in diethyl ether. The solution was worked up
as in reaction of 3 with benzylideneacetone. Data are tabulated
in Table S-5.31
n-BuLi; 0.25 mmol of benzylideneacetone; 0.50 mmol of 12)
was performed in a flask equipped with a stopcock (Schlenk
flask) at -78 °C for 120.5 h, and quenched with 60 µL of acetic
acid-d. The solution was worked up as in the reaction of 3 with
benzylideneacetone. Analysis by 1H NMR showed a 78:22 ratio
of d-10:h-10 by loss of integration at δ 3.17-2.90, so the
reaction mixture had been quenched to the extent of 22%. The
level of m-methoxy incorporation (13) was determined to be
0.4% by line shape simulation of quantitative spectra and
integration relative to the 13C satellites of 10 as a standard.
Test for Proton Transfer To Form 14. A solution of
phenylacetonitrile (87 µL, 0.75 mmol) in 3 mL of THF in a
nitrogen-purged flask was cooled to -78 °C and metalated with
0.29 mL (0.75 mmol) of 2.6 M n-BuLi. The reaction was allowed
to stir 20 min, 250 µL of 1.0 M (0.25 mmol) of benzylidene-
acetone solution in THF was added, and the flask was warmed
to 0 °C for 2 h to generate 10 in situ and quenched with 50 µL
of acetic acid-d (DOAc). The solution was worked up as in the
reaction of 3 with benzylideneacetone. The 1H NMR spectrum
showed 84% deuteration R to the ketone by integration of the
AB quartet signals at δ 3.15 and 2.98, and integration of the
2H NMR spectrum showed a d-10/d-14 ratio of >98:2 by
integration of peaks at δ 3.15 and 2.98 compared to the R-cyano
protons peaks at δ 4.35 and 4.07.
Sample Reaction of 17-H with Mesityl Oxide. Methoxy-
phenylacetonitrile (37 mg, 0.25 mmol) was added to a 5 mL,
long-necked, round-bottom flask, the flask was purged with
nitrogen, 3 mL of dry THF was added, and the system was
cooled to -78 °C under positive nitrogen pressure. n-BuLi (95
µL of 2.65 M, 0.25 mmol) was added, the mixture was stirred
for 20 min, and 100 µL of 2.5 M mesityl oxide in THF was
injected. The solution was stirred for the indicated time,
quenched with 200 µL of 3 M propionic acid in diethyl ether,
and worked up as in reaction of 3 with benzylideneacetone.
Data are displayed in Figure 6 and in Table S-6.31
Benzaldehyde Trapping Studies: Control Experi-
ment. Phenylacetonitrile (29 µL, 0.25 mmol) was added to a
nitrogen-purged 5 mL, long-necked, round-bottom flask. Three
milliliters of dry THF was added, the system was cooled to
-78 °C under positive nitrogen pressure, and 90 µL (0.25
mmol) of 2.83 M n-BuLi was added. A mixture of benzaldehyde
and benzylideneacetone (0.5 mmol each in 500 µL of THF) was
injected into the flask and quenched after 2 min with 200 µL
of 3 M propionic acid in diethyl ether. The solution was worked
up as in the reaction of 3 with benzylideneacetone. The ratio
1
of 11/4 was >99:1, as determined by H NMR integration of
doublets at δ 4.97-4.93 compared to singlets at δ 3.95 and
3.93.
m-Methoxy Crossover Studies: Retro-Michael Trap-
ping. Phenylacetonitrile (58 µL, 0.50 mmol) and 6 mL of THF
were added to a nitrogen-purged flask, the flask was cooled to
-78 °C under positive nitrogen pressure, and 0.190 mL (0.5
mmol) of 2.62 M n-BuLi was added. The reaction was allowed
to stir 20 min to ensure complete metalation, and 200 µL (0.5
mmol) of 2.5 M benzylideneacetone solution in THF was added.
The solution was allowed to warm to 0 °C for 1 h to generate
10 in situ. The solution was cooled to the appropriate tem-
perature with an EtOH/CO2 or ice bath, and a solution of 12
(1 mmol in 12 mL of THF) was added via cannula. Samples (3
mL) were withdrawn from the stock solution at various times
with a jacketed syringe cooled below the sample temperature
and were quenched with 200 µL of 3 M propionic acid (0.6
mmol) in ether. The solution was worked up as in the reaction
of 3 with benzylideneacetone. Data are displayed in Figure
S-1. The rate constant data and Eyring plot are shown in
Figure S-2.
Acknowledgment. We thank the National Science
Foundation (Grant Nos. CHE-0074657 and CHE-
0349635) for support of this research. This material is
available free of charge via the Internet at http://
pubs.acs.org.
Supporting Information Available: Preparation of start-
ing materials, tables of data for the kinetic experiments,
characterization data and NMR spectra of reaction products
(4, 5, 11, 20-X, 21-X). This material is available free of charge
via the Internet at http://pubs.acs.org.
Long-Term Experiment at -78 °C. A similar experiment
(3 mL of THF; 29 µL, 0.25 mmol of 3; 0.100 mL, 0.25 mmol of
JO0478050
3382 J. Org. Chem., Vol. 70, No. 9, 2005