Additions of Organomagnesium Halides to α-Alkoxy Ketones: Revision of the Chelation-Control Model

Article abstract of DOI:10.1021/acs.orglett.7b01161

The chelation-control model explains the high diastereoselectivity obtained in additions of organometallic nucleophiles to α-alkoxy ketones but fails for reactions of allylmagnesium halides. Low diastereoselectivity in ethereal solvents results from no chelation-induced rate acceleration. Additions of allylmagnesium bromide to carbonyl compounds are diastereoselective using CH₂Cl₂ as the solvent even though rate acceleration is still absent. Stereoselectivity likely arises from the predominance of the chelated form in solution. Therefore, a revised chelation-control model is proposed.

Full text of DOI:10.1021/acs.orglett.7b01161

Letter
pubs.acs.org/OrgLett
Additions of Organomagnesium Halides to α‑Alkoxy Ketones:
Revision of the Chelation-Control Model
Jacquelyne A. Read, Yingying Yang, and K. A. Woerpel*
Department of Chemistry, New York University, New York, New York 10003, United States
S
* Supporting Information
ABSTRACT: The chelation-control model explains the high
diastereoselectivity obtained in additions of organometallic
nucleophiles to α-alkoxy ketones but fails for reactions of
allylmagnesium halides. Low diastereoselectivity in ethereal
solvents results from no chelation-induced rate acceleration.
Additions of allylmagnesium bromide to carbonyl compounds are diastereoselective using CH2Cl2 as the solvent even though
rate acceleration is still absent. Stereoselectivity likely arises from the predominance of the chelated form in solution. Therefore, a
revised chelation-control model is proposed.
dditions of alkylmagnesium and alkyllithium reagents to α-
not more reactive than the nonchelated form. Instead, the
chelated form is more populated in CH2Cl2, so, in accordance
with the Winstein−Holness kinetic model,11 most of the product
is formed by reaction of the chelated intermediate. These
experiments justify a revision of the chelation-control model to
include this scenario.
The low selectivity exhibited in reactions of allylmagnesium
halides with chiral, α-alkoxy ketones is evident when direct
comparisonstootherorganomagnesiumreagentsaremade.12 For
example, we found that whereas addition of n-propylmagnesium
chloride to ketone 2 was highly selective for the chelation-control
product (eq 1), addition of allylmagnesium chloride proceeded
with low stereoselectivity, favoring the opposite stereoisomer (eq
2).13
A
alkoxy carbonyl compounds are powerful reactions for
organic synthesis,1 particularly because these reactions are often
highly diastereoselective.2,3 The stereochemical courses of these
reactions are generally explained using the chelation-control
model,2 which proposes that the α-chelated intermediate (1,
Figure 1) has sterically differentiated diastereotopic faces that
Figure 1. Chelated intermediate formed in reactions of RMgX with α-
alkoxy carbonyl compounds.
react with the nucleophile at different rates. Studies using Me2Mg
established that reactions are diastereoselective because the
chelated intermediate, which is a minor component of the
reaction mixture in ethereal solvents, reacts more rapidly with the
nucleophile than nonchelated intermediates do.4−7 While the
chelation-control model can predict and rationalize the outcomes
of many reactions, it exhibits notable failures.8 For example,
additions of allylmagnesium halides to α-alkoxy carbonyl
compounds generally do not occur with high diastereoselectivity,
which is problematic considering how frequently these reactions
are used in synthesis.9,10
In this paper, we report experiments that demonstrate why the
chelation-control model cannot generally predict or explain
reactions of allylmagnesium halides with α-alkoxy carbonyl
compounds in ethereal solvents. The chelated intermediate is not
theonlyspeciesthatreactsinthiscase. Productsarealsoproduced
from nonchelated intermediates, which react with little
diastereoselectivity. High diastereoselectivity for the allylation
product expected from chelation control can be achieved using
CH2Cl2 as the solvent. The chelation-control model also cannot
explain this phenomenon, however, because the chelated form is
This dichotomy was general (Figure 2). Reactions of ketones
with alkyl- and alkenylmagnesium halides in THF or Et2O gave
the products predicted by the chelation-control model with high
diastereoselectivity.14 By contrast, reactions with allylmagnesium
halides exhibited low selectivity.
The disparate selectivities observed between alkylmagnesium
and allylmagnesium reagents are largely independent of the
precise structure of the reagent. Selectivities of additions to
Received: April 17, 2017
© XXXX American Chemical Society
A
DOI: 10.1021/acs.orglett.7b01161
Org. Lett. XXXX, XXX, XXX−XXX

Organic Letters
Letter
Table 2. Relative Rates of Additions to Chelating Ketone 13
Versus Propiophenone (14)
a
entry
RMgX
MeMgCl
product
15:16
1
2
3
4
5
6
a
a
b
c
104:1
17:1
b
MeMgCl
n-PrMgCl
338:1
27:1
H2CCHMgBr
H2CCHCH2MgCl
H2CCHCH2MgBr
c
c
d
d
49:51
45:55
a
b
Figure 2. Diastereoselectivities of additions of RMgX to α-alkoxy
ketones in THF at −78 °C.
Ratios determined by GC analysis of the reaction mixture. Reaction
run in 2:1 Et2O/THF. Product ratio was corrected for FID response
factors.
c
Table 1. Testing the Effect of Solvent, Halide, and Reactive
Magnesium Species on Diastereoselectivity
that these outcomes parallel observations with Me2Mg,5 the
results with organomagnesium halides provide additional
evidence that the Schlenk equilibrium and aggregation are not
the most important predictors of stereoselectivity. In accordance
with the lack of diastereoselectivity observed for additions of
allylmagnesium halides to chiral, α-alkoxy ketones (Figure 2),
thesereagents reacted withbothketones 13and14atsimilarrates
(Table 2, entries 5 and 6).
The results shown in Table 2 reveal why reactions of highly
reactive21,23 allylmagnesium halides deviate from the chelation-
control model. These reagents do not preferentially react through
chelated intermediates, which are likely to be minor species in
solution.5 Because the majority of the carbonyl compound in
solution is not in a chelated form, low selectivity could result
because these electrophiles would have little steric differentiation
between their diastereotopic faces.
a
entry
RMgX
solvent
dr
b
1
2
3
4
5
6
7
8
MeMgBr
Et2O
Et2O
THF
THF
Et2O
Et2O
THF
THF
>97:3
>97:3
>97:3
>97:3
79:21
76:24
59:41
58:42
MeMgBr
MeMgCl
(Me)2Mg
b
H2CCHCH2MgBr
H2CCHCH2MgBr
H2CCHCH2MgCl
(H2CCHCH2)2Mg
a
Ratios determined by GC or NMR spectroscopic analysis of the
b
reaction mixture. Reactions run at 1.0 M and warmed to 0 °C.
Competition experiments also suggest why additions of
organomagnesium halides to α-alkoxy aldehydes generally
proceed with low stereoselectivity.3,24 The addition of either
MeMgCl or allylmagnesium chloride to a mixture of α-alkoxy
aldehyde 17 and alkyl aldehyde 18 resulted in the formation of
both addition products (eq 4).25 The low selectivity observed in
ketone 12 were not significantly affected by concentration
(entries 1 and 5 compared to entries 2 and 6), which would be
expected to change the aggregation state of the reagent.15 The
position of the Schlenk equilibrium16 (eq 3) also had little
influence on selectivity. As shown in Table 1 (entries 2, 3, 6, and
7), contrasting selectivities between methyl- and allylmagnesium
reagents were observed in Et2O, where the RMgX form
predominates (Keq ≈ 400),17 and in THF, where dialkylmagne-
sium reagents are also present (Keq ≈ 4).18,19 Even in reactions
with prepared R2Mg, where there is no halide ion to establish the
Schlenkequilibrium, selectivitytrendsremainedconstant(entries
4 and 8). In all cases, allylmagnesium reagents exhibited low
diastereoselectivity, whereas methylmagnesium reagents reacted
with high stereoselectivity.20
the reaction of MeMgCl with aldehydes 17 and 18 contrasts
significantly with the high selectivities observed in competition
experiments with ketones 13 and 14 (Table 2, entries 1 and 2).
Unlike with α-alkoxy ketones, the presence of an α-alkoxy group
in an aldehyde does not accelerate the addition of allyl- or
alkylmagnesium halides.26 As a result, a central premise of the
chelation control model is not met, so stereoselectivity need not
be observed.
Additions of allylmagnesium bromide to α-alkoxy carbonyl
compounds can be highly diastereoselective, however, in non-
Lewis basic solvents.27 When the ethereal solvent of the reagent
was exchanged for CH2Cl2 by evacuating it to dryness (0.2 mbar)
then suspending the resulting solids in CH2Cl2, reactions were
highly diastereoselective (Figure 3). This significant increase in
selectivity compared to selectivities in ethereal solvents was
general for the substrates that had reacted with low selectivity in
Keq
R2Mg + MgX2 HooI 2RMgX
(3)
Competition experiments5,21 provided insight into why
additions of alkyl- and alkenylmagnesium halides were diaster-
eoselective but additions of allylmagnesium halides were not.
Considering that the chelation-control model requires the
chelated intermediate to be the most reactive species in solution,
substratesthatcanchelateshouldbemorereactivethansubstrates
that cannot.4,5 This important condition is met for alkyl- and
alkenylmagnesium halides, where additions occurred almost
exclusively to an α-alkoxy ketone (13) instead of to a ketone
without a chelating group (Table 2, entries 1−4).22 Considering
B
DOI: 10.1021/acs.orglett.7b01161
Org. Lett. XXXX, XXX, XXX−XXX

Organic Letters
Letter
additiontoonediastereotopicfaceofthechelatedintermediatebe
faster than addition to the other face.
1H and 13C NMR spectroscopic studies5,30 of ketone 12 in the
presence of MgBr2 provided evidence that the chelated form of an
α-alkoxy carbonyl compound is more favored in CH2Cl2 than it is
in THF.31 The chemical shifts of ketone 12 in THF-d8 and
CD2Cl2 responded differently to the presence of MgBr2. The
1
downfield shifts (Δδ) for selected H and 13C resonances of
ketone 12 upon binding to MgBr2 are given in Figure 4. In THF-
Figure 3. Diastereoselectivities for additions of RMgBr as suspensions in
CH2Cl2 to α-alkoxy carbonyl compounds.
THF (eq 2 and Figure 2). The allylation of an α-alkoxy aldehyde
to form secondary alcohol 21 was also diastereoselective in
CH2Cl2, and selectivity with an alkylmagnesium reagent
remained high in this solvent.
Figure 4. Downfield 1H and 13C NMR resonance shifts (ppm) observed
upon mixing ketone 12 (0.1 M) with MgBr2 (2 equiv).
Reactions of organomagnesium reagents with carbonyl
compounds in CH2Cl2 were noticeably slower than reactions in
ethereal solvents, however. The reagents in CH2Cl2 are
heterogeneous, suggesting that their aggregation states are quite
different than they are in THF and Et2O. Reactions in CH2Cl2
could be slower because the additional steps of diffusion and mass
transfer between solid and liquid phases could be rate-
limiting.28,29 To observe full conversion, more equivalents of
the reagent (5 equiv compared to 1.2 equiv) and longer reaction
times were necessary (≥3 h compared to ≤15 min).
Even though the nature of the reagents in CH2Cl2 must be
much different than in ethereal solvents, the results of
intermolecular competition experiments were similar. Although
addition of n-propylmagnesium chloride in CH2Cl2 occurred
faster to the α-chelating ketone 13 than to ketone 14,
allylmagnesium bromide showed no preference (eq 5). These
data correlate with those obtained in THF (Table 2).
d8, the chemical shifts associated with the carbonyl group and the
α-methine proton underwent only moderate changes,32 whereas
they were more pronounced in CD2Cl2. Similarly, the downfield
shift oftheresonance fortheα-methoxygroupinCD2Cl2 isnearly
four times its shift in THF-d8. The magnitude of the downfield
shift observed in CD2Cl2 is consistent with earlier observations of
two-point binding involving an α-methoxy group.5 Assuming that
the interactions with MgBr2 are similar to an organomagnesium
reagent, the spectroscopic data suggest that more of the chelated
intermediate resembling 1 is present in CD2Cl2 than it is in THF-
d8.
These experiments lead to a deeper understanding of the
chelation-control model in reactions with organomagnesium
halides. Experiments originally conducted with Me2Mg inTHF4,5
stipulated that, in accordance with the Curtin−Hammett
principle,11 the chelated form of the carbonyl compound must
be more reactive than the nonchelated form. The experiments
reported here show that this situation also holds for reactions of
ketones in THF and Et2O with the more commonly used RMgX
formofthereagentinsteadofR2Mg. Inthecaseofallylmagnesium
halides, however, this requirement is not met, so chelation-
controlled selectivity was not observed.
These studies also illustrate that the chelation-control model
requires revision. The chelated form of the carbonyl compound
need not react faster than the nonchelated form if the chelated
intermediate predominates in solution; it is only necessary that
the two diastereotopic faces of the carbonyl compound react at
different rates. This revision broadens the model to incorporate
not only a Curtin−Hammett kinetic scenario but also the
Winstein−Holness kinetic scenario in which the major product is
formed from the most abundant reactive intermediate.11
Consequently, caution should be used when interpreting results
involving additions to α-alkoxy carbonyl compounds using the
chelation-control model. Should a reaction be stereoselective, it
should be established whether chelation is rate-accelerating or
whether the chelated form predominates in solution, which is
possible in CH2Cl2.
The competition experiments shown in eq 5 reveal that the
chelation-control model is not consistent with the stereo-
selectivities in Figure 3. Because the chelated form was not
more reactive than other species present, application of the
chelation-control model would lead to a prediction that reactions
should not be diastereoselective, as has been observed for β-
alkoxy carbonyl compounds.3−5 Instead, these reactions were
highly diastereoselective (Figure 3).
An alternative hypothesis can be proposed for why additions of
allylmagnesium bromide in CH2Cl2 are diastereoselective. The
requirementthatthechelatedform(1)ofthecarbonylcompound
be more reactive than the nonchelated form5 is necessary only if
thechelatedform isaminorcomponent ofthe reactionmixture, asitis
in ethereal solvents. In THF or Et2O, the chelated reaction pathway
mustbesignificantlyfasterthancompetingnonchelatedpathways
for diastereoselectivity to be observed. If most of the carbonyl
compound were in the chelated form resembling 1, however, it
could react at the same rate as nonchelated forms would, as
observed for allylmagnesium halides, and the reactions could be
diastereoselective. All that is required for selectivity is that
ASSOCIATED CONTENT
* Supporting Information
■
S
TheSupportingInformationisavailablefreeofchargeontheACS
Publications website at DOI: 10.1021/acs.orglett.7b01161.
C
DOI: 10.1021/acs.orglett.7b01161
Org. Lett. XXXX, XXX, XXX−XXX

Organic Letters
Letter
1
K.; Kaburagi, Y.; Fukuyama, T. J. Am. Chem. Soc. 2003, 125, 4048.
(b) Cardona, F.; D'Orazio, G.;Silva, A. M. S.;Nicotra, F.; La Ferla, B. Eur.
J. Org. Chem. 2014, 2549. (c) Zhang, J.; Zheng, S.; Peng, W.; Shen, Z.
Tetrahedron Lett. 2014, 55, 1339.
(11) Seeman, J. I. Chem. Rev. 1983, 83, 83.
(12) For examples where allylmagnesium halides exhibit low selectivity
Experimental procedures, characterization data, H and
13C NMR spectroscopic data, GC traces, stereochemical
correlations, and X-ray structures (PDF)
X-ray data for compound S5 (CIF)
X-ray data for compound S11′ (CIF)
but other reagents are selective, see: (a) Marco, J. A.; Carda, M.;
́
Gonzalez, F.; Rodríguez, S.; Castillo, E.; Murga, J. J. Org. Chem. 1998, 63,
AUTHOR INFORMATION
■
698. (b)Qian, X.;Sujino, K.;Otter, A.;Palcic, M. M.;Hindsgaul, O. J. Am.
Corresponding Author
*E-mail: kwoerpel@nyu.edu.
ORCID
̌
Chem. Soc. 1999, 121, 12063. (c) Smaltz, D. J.; Svenda, J.; Myers, A. G.
Org. Lett. 2012, 14, 1812.
(13) Selectivities were determined by 1H and 13C NMR spectroscopy.
The first number in each product ratio corresponds to the indicated
stereoisomer. Details of stereochemical proofs are provided as
Supporting Information.
K. A. Woerpel: 0000-0002-8515-0301
Notes
(14) The 1-propenyl Grignard reagent was used as an E/Z mixture.
(15) Walker, F. W.; Ashby, E. C. J. Am. Chem. Soc. 1969, 91, 3845.
(16) Schlenk, W.; Schlenk, W., Jr. Ber. Dtsch. Chem. Ges. 1929, 62, 920.
(17) Ashby, E. C.; Laemmle, J.; Neumann, H. M. Acc. Chem. Res. 1974,
7, 272.
(18) Parris, G. E.; Ashby, E. C. J. Am. Chem. Soc. 1971, 93, 1206.
(19) Schnegelsberg, C.; Bachmann, S.; Kolter, M.; Auth, T.; John, M.;
Stalke, D.; Koszinowski, K. Chem. - Eur. J. 2016, 22, 7752.
(20) A variety of Lewis acid additives such as MgBr2·Et2O, SnCl4, and
TiCl4 were screened, and no increase in diastereoselectivity was
observed. Attempts at influencing the selectivity by modifying the
ligands on magnesium also did not give satisfactory results.
(21) Read, J. A.; Woerpel, K. A. J. Org. Chem. 2017, 82, 2300.
(22) Rate acceleration due to the presence of α-alkoxy substituents has
also been observed in Mukaiyama aldol and hetero-Diels−Alder
reactions: Reetz, M. T. Acc. Chem. Res. 1993, 26, 462.
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
■
Acknowledgment is made to the Donors of the American
Chemical Society Petroleum Research Fund for partial support of
this research (57206-ND1). Additional support was provided by
the National Institutes of Health, National Institute of General
Medical Sciences (GM-61066). J.A.R. was supported by a
Margaret Strauss Kramer Fellowship from the NYU Department
of Chemistry. K.A.W. thanks the Global Research Initiatives,
NYU and NYU Florence, for a fellowship. We thank Dr. Elizabeth
́
M. Valentın (NYU) for valuable discussions. We thank Dr. Chin
Lin (NYU) for assistance with NMR spectroscopy and mass
spectrometry and Dr. Chunhua Hu (NYU) for assistance with
crystallographic studies.
(23) Benkeser, R. A. Synthesis 1971, 347.
(24) For examples of α-alkoxy aldehydes reacting with low
diastereoselectivity, see: (a) Yang, W.-Q.; Kitahara, T. Tetrahedron
2000, 56, 1451. (b) Denmark, S. E.; Yang, S.-M. J. Am. Chem. Soc. 2004,
126, 12432. (c) Suzuki, A.; Sasaki, M.; Nakagishi, T.; Ueda, T.; Hoshiya,
N.; Uenishi, J. Org. Lett. 2016, 18, 2248.
(25) Obtaining precise relative rates of addition in this case is
complicated, presumably due to competitive enolization. HRMS
confirmed the presence of multiple aldol addition and condensation
products in the reaction mixture.
(26) The absence of chelation-induced rate acceleration in these
reactions with α-alkoxy aldehydes may be due to the markedly higher
reactivity of aldehydes compared to ketones: Brown, H. C.; Wheeler, O.
H.; Ichikawa, K. Tetrahedron 1957, 1, 214.
(27) Charette, A. B.; Benslimane, A. F.; Mellon, C. Tetrahedron Lett.
1995, 36, 8557.
DEDICATION
■
WededicatethispapertothememoryofthelateProfessorRobert
A. Benkeser of Purdue University (1920−2017), who made
important contributions to the study of allylic Grignard reagents
and their reactions.
REFERENCES
■
(1)Forrecentexamplesinnaturalproductsynthesis, see:(a)Crimmins,
M. T.; Dechert, A.-M. R. Org. Lett. 2012, 14, 2366. (b) Matthies, S.;
Stallforth, P.; Seeberger, P. H. J. Am. Chem. Soc. 2015, 137, 2848.
(c) Trost, B. M.; Biannic, B.; Brindle, C. S.; O’Keefe, B. M.; Hunter, T. J.;
Ngai, M.-Y. J. Am. Chem. Soc. 2015, 137, 11594.
(2) Cram, D. J.; Kopecky, K. R. J. Am. Chem. Soc. 1959, 81, 2748.
(3) Reetz, M. T. Angew. Chem., Int. Ed. Engl. 1984, 23, 556.
(4) Frye, S. V.; Eliel, E. L.; Cloux, R. J. Am. Chem. Soc. 1987, 109, 1862.
(5) Chen, X.; Hortelano, E. R.; Eliel, E. L.; Frye, S. V. J. Am. Chem. Soc.
1992, 114, 1778.
(6) Mori, S.; Nakamura, M.; Nakamura, E.; Koga, N.; Morokuma, K. J.
Am. Chem. Soc. 1995, 117, 5055.
(7) This rate acceleration resulting from chelation to magnesium is not
present in reactions of β-alkoxy ketones, but a kinetic advantage is
observed when β-alkoxy ketones chelate with titanium and chromium
(28) Wen, C. Y. Ind. Eng. Chem. 1968, 60, 34.
(29) We cannot definitively assign the rate-determining step, however.
Due to the heterogeneity of the reaction mixture, the reaction could
occur in solution, within the suspended solid, or at the surface of that
solid.
(30) For other NMR studies of complexes between carbonyl
compounds and Lewis acids, see: (a) Keck, G. E.; Castellino, S. J. Am.
Chem. Soc. 1986, 108, 3847. (b) Reetz, M. T.; Hullmann, M.; Seitz, T.
̈
Angew. Chem., Int. Ed. Engl. 1987, 26, 477. (c) Stanton, G. R.; Johnson, C.
N.; Walsh, P. J. J. Am. Chem. Soc. 2010, 132, 4399.
reagents: Kauffmann, T.; Moller, T.; Rennefeld, H.; Welke, S.;
̈
(31) Binding constants could not be determined because the signal
associated with the ketone decreased upon addition of MgBr2. After
addition of 2 equiv, only 25% of ketone 12 was in solution, as determined
by 1H NMR spectroscopy with an internal standard.
(32)Inthechelationexperimentsreportedinref5, aphenylketonewith
an α-OTBS group (not generally capable of chelation) shows larger
downfield shifts in CD2Cl2 than those observed here with ketone 12 in
THF-d8.
Wieschollek, R. Angew. Chem., Int. Ed. Engl. 1985, 24, 348.
(8) Mengel, A.; Reiser, O. Chem. Rev. 1999, 99, 1191.
(9) For recent examples, see: (a) Crimmins, M. T.; Ellis, J. M.; Emmitte,
K. A.; Haile, P. A.; McDougall, P. J.; Parrish, J. D.; Zuccarello, J. L. Chem. -
Eur. J. 2009, 15, 9223. (b) Yamashita, S.; Ishihara, Y.; Morita, H.;
Uchiyama, J.; Takeuchi, K.; Inoue, M.; Hirama, M. J. Nat. Prod. 2011, 74,
́
357. (c) Moume-Pymbock, M.; Furukawa, T.; Mondal, S.; Crich, D. J.
Am. Chem. Soc. 2013, 135, 14249. (d) Kita, M.; Oka, H.; Usui, A.;
Ishitsuka, T.; Mogi, Y.; Watanabe, H.; Tsunoda, M.; Kigoshi, H. Angew.
Chem., Int. Ed. 2015, 54, 14174.
(10) Some reactions of allylmagnesium halides can give the product
expectedbythechelation-controlmodel. Forexamples, see:(a)Shimada,
D
DOI: 10.1021/acs.orglett.7b01161
Org. Lett. XXXX, XXX, XXX−XXX