Dynamics of catalytic resolution of 2-lithio-N-Boc-piperidine by ligand exchange

Article abstract of DOI:10.1021/ja307796e

The dynamics of the racemization and catalytic and stoichiometric dynamic resolution of 2-lithio-N-Boc-piperidine (7) have been investigated. The kinetic order in tetramethylethylenediamine (TMEDA) for both racemization and resolution of the title compound and the kinetic orders in two resolving ligands have been determined. The catalytic dynamic resolution is second order in TMEDA and 0.5 and 0.265 order in chiral ligands 8 and 10, respectively. The X-ray crystal structure of ligand 10 shows it to be an octamer. Dynamic NMR studies of the resolution process were carried out. Some of the requirements for a successful catalytic dynamic resolution by ligand exchange have been identified.

Full text of DOI:10.1021/ja307796e

Article
pubs.acs.org/JACS
Dynamics of Catalytic Resolution of 2‑Lithio‑N‑Boc-piperidine by
Ligand Exchange
Timothy K. Beng,† William S. Tyree,† Trent Parker,† Chicheung Su,‡ Paul G. Williard,‡
and Robert E. Gawley*,†
†Department of Chemistry and Biochemistry, University of Arkansas, Fayetteville, Arkansas 72701, United States
‡Department of Chemistry, Brown University, Providence, Rhode Island 02912, United States
S
* Supporting Information
ABSTRACT: The dynamics of the racemization and catalytic and
stoichiometric dynamic resolution of 2-lithio-N-Boc-piperidine (7) have
been investigated. The kinetic order in tetramethylethylenediamine
(TMEDA) for both racemization and resolution of the title compound and
the kinetic orders in two resolving ligands have been determined. The catalytic
dynamic resolution is second order in TMEDA and 0.5 and 0.265 order in
chiral ligands 8 and 10, respectively. The X-ray crystal structure of ligand 10
shows it to be an octamer. Dynamic NMR studies of the resolution process were carried out. Some of the requirements for a
successful catalytic dynamic resolution by ligand exchange have been identified.
consumes L*. Thus, a variation of dynamic resolution using a
catalytic amount of L* is desirable. Toru and co-workers
reported a catalytic dynamic resolution (CDR) that is driven by
the well-known phenomenon of crystallization-induced asym-
metric transformation (Scheme 2).6 In the Toru example, the
INTRODUCTION
■
Organolithium compounds are widespread in organic syn-
thesis.1 The use of stoichiometric chiral ligands such as
(−)-sparteine complexed to sec-BuLi to effect enantioselective
deprotonation represents one of the methods to generate
scalemic organolithium compounds for use in asymmetric
synthesis.2 Unfortunately, effecting an asymmetric deprotona-
tion on N-Boc-piperidines through the use of the chiral ligand
(−)-sparteine is not as successful as with pyrrolidines,3 but
partial success [enantiomeric ratio (er) of up to 88:12 in
reasonable yields] has been achieved with O’Brien’s diamine.4
In an important development, several chiral organolithium
compounds that are amenable to dynamic thermodynamic
resolution (DTR) have been discovered (Scheme 1).5 In DTR,
Scheme 2. Toru’s Catalytic Dynamic Resolution (CDR)
Scheme 1. Dynamic Thermodynamic Resolution (DTR)
Using a Stoichiometric Amount of the Chiral Ligand L*
resolution is accomplished through dynamic resolution of
planar-chiral organolithium enantiomers by the bisoxazoline
ligand followed by precipitation of homochiral aggregates (an
asymmetric transformation of the second kind). Addition of
aldehyde caused dissolution of the precipitate, and after the
reaction was quenched with Me3SiCl and acidic workup, the
product was isolated in good yield with excellent diaster-
eoselectivity and enantioselectivity.
In 2009, we reported the transmetalation of 2-tributylstannyl-
N-(trimethylallyl)pyrrolidine (1) in the presence of N,N-
diisopropylbispidine (3) to afford racemic organolithium 2
(Scheme 3).7 Addition of ligand 4 (15 mol %) and stirring at
∼0 °C effected a catalytic resolution. Cooling to −78 °C and
a racemic organolithium is generated, and a chiral ligand, L*,
that can resolve the organolithium dynamically (an asymmetric
transformation of the first kind) is then added. Dynamic
resolutions hold significant potential, since they obviate the
need for an enantioselective deprotonation or an asymmetric
synthesis of an enantioenriched precursor stannane. A
disadvantage of DTR can arise if the electrophile used to
react with the organolithium is also reactive toward, and
Received: August 6, 2012
Published: September 11, 2012
© 2012 American Chemical Society
16845
dx.doi.org/10.1021/ja307796e | J. Am. Chem. Soc. 2012, 134, 16845−16855

Journal of the American Chemical Society
Article
implies that the er in the absence of 3 would be 55:45. The
catalytic resolution of 2 by 4 is accelerated by 3, but it is also
apparent that 3 is necessary to achieve a CDR. If 15% of 2 is
resolved to 93:7 er and the remaining 85% is racemic, the er of
the solution would be 56:44, the same as the y intercept within
experimental error. Therefore, a second requirement is that the
chiral and achiral ligands must both be involved in the catalytic
resolution. This could involve a total ligand exchange or a
reorganization of the coordination of the organolithium to
more than one ligand. Aggregation/deaggregation of the
organolithium could be involved in the catalytic resolution by
ligand exchange. Although the aggregation state of 2 was not
determined, the aggregation state of the N-methyl analogue is a
homochiral dimer, and the N-ethyl analogue is a mixture of
aggregates.14
Scheme 3. CDR of 2-Lithio-N-(trimethylallyl)pyrrolidine
(2)9
The mechanistic details of a catalytic resolution by ligand
exchange are of interest. Unfortunately, a detailed mechanistic
investigation of lithiopyrrolidine 2 is hampered by the fragility
of the trimethylallyl group under the conditions of the reaction
(it has a tendency to rearrange, undergo proton transfer, and be
lost during workup). Also, the presence of the achiral bispidine
ligand 315 complicates product purification and recovery of the
conjugate acid of catalyst 4.
We wondered whether catalytic resolution involving ligand
exchange is unique to the N-(trimethylallyl)pyrrolidine system
in Scheme 3 or if other systems that can be resolved
catalytically could be identified. As noted above, we believe
that the free energy barrier for resolution must be lower than
the free energy barrier for racemization. We therefore sought
another system to investigate, and we chose 2-lithio-N-Boc-
piperidine (7), which was shown by the Coldham group5f to
undergo stoichiometric resolution in the presence of
tetramethylethylenediamine (TMEDA) (Scheme 4) and
whose inversion dynamics were investigated jointly by our
two groups.16
quenching with Me3SiCl gave pyrrolidine 5 with an er of 93:7
(R:S).8,9
In this catalytic resolution, organolithium 2 is in excess
relative to the chiral resolving ligand 4. The remaining 2 is
presumably complexed to the achiral ligand 3.10 To achieve a
catalytic resolution, at a minimum it is necessary to ensure that
the racemization of the organolithium is significantly slower
than the dynamic resolution. Therefore, we studied the
enantiomerization of 2 in the presence of the achiral bispidine
ligand 3,9 and the thermodynamic parameters were determined
(ΔH⧧ = 25.1 1.4 kcal/mol; ΔS⧧ = 8.1 5.0 cal mol−1 K−1).
Similarly, the thermodynamic parameters for the resolution of 2
by stoichiometric quantities of 4 in the presence of 3 were
measured (ΔH⧧ = 25.1 1.7 kcal/mol; ΔS⧧ = 14.2 6.1 cal
mol−1 K−1).11 From these data, plots of ΔG⧧ versus
temperature12 revealed that the resolution of 2 by 4 has a
lower barrier than enantiomerization at all temperatures
between −80 and +20 °C.9
It should be noted that the catalytic resolution in Scheme 3
should not be called a dynamic thermodynamic resolution
because it does not employ stoichiometric amounts of the chiral
ligand.5g,13 Neither can it be called a dynamic kinetic resolution,
as that term applies to systems where the organolithium inverts
rapidly and the resolution occurs by asymmetric substitution
under Curtin−Hammett conditions. Thus, we refer to the
process simply as a catalytic dynamic resolution.
Scheme 4. DTR of 75f
We wanted to know whether a CDR was possible in the
absence of the achiral ligand, but since conversion of stannane 1
to organolithium 2 by tin−lithium exchange is not possible in
the absence of the achiral ligand, we tested the possibility by
extrapolation using varying amounts of bispidine 3. Figure 1
shows the fraction of the R enantiomer of 5 as a function of the
number of equivalents of bispidine 3 after treatment of rac-2
with 15 mol % 4 for 1 h at 0 °C.9 The y intercept of 0.55
In lithiopiperidine 7, we have an organolithium that is
synthetically important and readily available. Moreover, the
achiral ligand, TMEDA, is inexpensive. As it turned out (see
Figure 14, below), the moderate final er of the Coldham
resolution (77:23) permitted a crucial experiment that was not
possible in investigating the more highly enantioselective
catalytic resolution of 2. An interesting fact uncovered
previously was that both the stoichiometric resolution and
the racemization of lithiopiperidine 7 are catalytic and first
order in the achiral ligand TMEDA (0−2 equiv of TMEDA for
the DTR and 0.1−1.0 equiv for the racemization). Also, several
other diamine ligands were tested, and all of them failed to fill
this catalytic role in the resolutioneven tetraethylethylenedi-
amine (TEEDA)!16
Figure 1. Resolution of 2 by 4 with varying amounts of bispidine 3 (0
°C, 1 h).9
16846
dx.doi.org/10.1021/ja307796e | J. Am. Chem. Soc. 2012, 134, 16845−16855

Journal of the American Chemical Society
Article
Herein we report the results of our detailed investigation into
the dynamics of the catalytic resolution of 7 by 8. During the
course of this mechanistic investigation, we discovered a new
pair of ligands, 10 and 11, that effect the resolution of 7 and
afford much higher enantioselectivity than 8 (Scheme 5).17
While many of these mechanistic studies were conducted with
ligand 8, we report some experiments with ligand 10 that
provide additional insight into the processes involved.
Scheme 5. DTR of 7 with Ligands 10 and 1117a
Figure 2. CDR of rac-7 by 8.
was found to be very highly enantioselective, and synthetic
applications have been reported.17
Further investigations of the structure of lithiopiperidine 7
and the dynamics of enantiomerization, DTR, and CDR of 7 by
8 and 10 were undertaken at carefully controlled temperatures.
These experiments, which are described in detail below,
revealed the following:
• Conclusive evidence for the solution structure of 7 could
not be obtained because the spectra are consistent with
several possibilities, but evidence of dynamic exchange
phenomena was found.
• There is variable configurational stability of 7 in the
presence of different ligands at low temperature when 7
is generated by tin−lithium exchange.
• Although up to 1 equiv of TMEDA catalyzes the
racemization and DTR of 7,16,17 TMEDA retards
racemization when present in excess; thus, excess
TMEDA increases the configurational stability of 7.
• There is a second-order dependence of the CDR on
[TMEDA].
RESULTS AND DISCUSSION
■
We previously reported the thermodynamic parameters for the
racemization and resolution of 7.16 The stoichiometric DTR of
rac-7 by 8 in the presence of 2 equiv of TMEDA has a large
negative entropy of activation (ΔS⧧ = −55.9 1.8 cal mol−1
K−1) and a small enthalpy of activation (ΔH⧧ = 4.3 0.5 kcal/
mol). The enantiomerization of 7 in the presence of 1 equiv of
TMEDA has ΔH⧧ = 18.1 0.7 kcal/mol and ΔS⧧ = 5.0 3.2
cal mol−1 K−1. At −46 °C, the free energy barriers for
enantiomerization (in the presence of 2 equiv of TMEDA) and
DTR (in the presence of 1 equiv of TMEDA) are equal. Below
this temperature, the free energy barrier for DTR is lower.
Given this information, we tested the possibility of a CDR of 7
at −55 °C in the presence of TMEDA. The time evolution of
the CDR of rac-7 was evaluated as illustrated in Figure 2. In
oven-dried vials, stannane rac-12 was treated with BuLi and 1.2
equiv of TMEDA at −78 °C for 1 h to effect tin−lithium
exchange, affording rac-7·TMEDA. Chiral ligand 8 (15 mol %)
was then added, and the reaction vessel was transferred to a
thermostatted bath at an internal temperature of −55 °C. At
various time intervals over 24 h, a vial was cooled to −78 °C
and quenched with Me3SiCl, and the er was then measured by
chiral-stationary-phase gas chromatography. The final er of the
product (77:23) was the same as that obtained by Coldham et
al.5f using ligand 8 in a stoichiometric DTR; their experiment
was duplicated in our hands as a control experiment. During
the course of this experiment, no precipitate was observed in
the reaction flask (compare with Scheme 2) .
• The CDR has a fractional-order dependence on the
concentration of the chiral ligand (8 or 10).
NMR Studies. Before describing the dynamics of
lithiopiperidine 7, we begin with the results of our study of
its solution structure. It is known from in situ IR studies that
the lithium ion of 7 is coordinated to the carbonyl oxygen, and
the monomer structure in Figure 3 was proposed to be the
structure in solution.4a However, there are dimeric and
polymeric structures that are consistent with the IR data. We
used 6Li and 13C NMR spectroscopy to investigate the possible
structures illustrated in Figure 3.
Transmetalation of stannanes rac-12 and S-12 with n-Bu6Li
6
in ether/TMEDA provided {6Li}-7. The Li and 13C spectra
were recorded at temperatures ranging from −80 to −30 °C.
6
The Li spectra taken at several temperatures are shown in
Figure 4. The racemate showed a single peak at 2.80 ppm
assigned to the Li+ on 7 as well as small amounts of solvated
lithium ion18 at 0.5−1.0 ppm at all temperatures between −30
and −80 °C. A single lithium signal is consistent with the
monomer, dimer-1, dimer-2, and polymer structures shown in
Figure 3. In contrast, the enantioenriched compound showed
These data are noteworthy for two reasons. First, a CDR is
operative, and second, the S organolithium is configurationally
stable for at least 12 h under the reaction conditions, implying
the achievement of an equilibrium state. During the course of
this mechanistic investigation, the CDR using ligands 10 and 11
16847
dx.doi.org/10.1021/ja307796e | J. Am. Chem. Soc. 2012, 134, 16845−16855

Journal of the American Chemical Society
Article
Figure 5. Partial 13C spectra of (left) {6Li}-rac-7 and (right) {6Li}-S-7
(93:7 er) in TMEDA/Et2O at −80 °C, showing the metal-bearing
carbon at 56.2 ppm, referenced to external CD3OD at 49.0 ppm.
and C-2 of a contact ion pair; at −30 °C, this coupling was no
longer evident, consistent with a solvent-separated ion pair.
Being indicative of a single lithium in contact with the
carbanionic carbon, this triplet rules out dimer-1 and dimer-3.
The resonance of the carbonyl carbon appeared at 160.1 ppm
for both rac-7 and S-7. Thus, no difference between the
chemical shifts of this peak for racemic and enriched 7 in
TMEDA/Et2O could be detected, even though we took care to
reference the spectra to the chemical shift of methanol-d4 in a
concentric capillary. These similarities suggest that the structure
of the organolithium species being observed is the same
whether the species is racemic or enantioenriched. It should be
noted, however, that these observations do not preclude the
presence of other organolithium species in the solution.
Figure 3. Possible solution structures for 7 (neglecting the absolute
and relative configurations of the metal-bearing carbons).
an approximately 1:1 ratio of 7 and solvated Li+ at −80 °C that
changed as the temperature increased. The two lithium signals
observed at −80 °C are consistent with the triple ion and
dimer-3 structures. The growth of a shoulder on the upfield
signal as the temperature was raised indicates a second type of
solvated lithium ion, and the changes in line shape suggest a
dynamic exchange process among solution species.
6
To summarize: for rac-7, the Li and 13C NMR data are
consistent with the monomer, dimer-2, or polymer structure
Partial 13C spectra of rac-7 and S-7 (93:7 er) at −81 °C
(Figure 5) showed a peak at 56.2 ppm that was split into a
1:1:1 triplet (J = 11.9 Hz) as a result of 1J coupling between 6Li
but not the triple ion, dimer-1, or dimer-3 structure. For S-7,
6
the Li spectrum is consistent with the triple ion or dimer-3
structure at −80 °C, with other structures evident at
6
Figure 4. Li NMR spectra of 7 obtained by transmetalation of 12 in Et2O in the presence of 4 equiv of TMEDA: (left) {6Li}-rac-7; (right)
6
enantioenriched {6Li}-S-7. The peaks at 0 ppm are external 1.0 M LiCl in CD3OD in a concentric capillary.
16848
dx.doi.org/10.1021/ja307796e | J. Am. Chem. Soc. 2012, 134, 16845−16855

Journal of the American Chemical Society
Article
temperatures greater than or equal to −60 °C. The coupling
seen in the partial 13C spectra are consistent with the monomer,
triple ion, dimer-2, or polymer structures.
X-ray Crystal Structure of Ligand 10. The dilithio ligand
10 formed beautiful crystals when an ether solution of the
ligand was cooled in the absence of TMEDA. The ligand
crystallized as an octamer formed by stacking of two cyclic
tetramers on top of each other. Each tetramer features a 16-
membered ring consisting of four −O−Li−NMe−Li− units.
Figure 7a shows a ball-and-stick structure, including two ethers
of crystallization. Figure 7b is a ChemDraw rendering of one of
the 16-membered tetramer rings, with one monomer high-
lighted in blue. It should be noted that for each monomer, one
of the lithium atoms is chelated by two nitrogens and the
oxygen, while the second lithium bridges the methylated
nitrogen to the oxygen of the next monomer. Figure 7c shows
the octamer, with the second tetramer rendered in red. There
are two types of bridging between the two 16-membered rings:
the methylated nitrogens in one tetramer are coordinated to the
bridging lithiums of the other tetramer, while the chelated
lithiums in one tetramer are coordinated to the oxygens in the
other tetramer. This results in eight four-coordinate lithium
atoms and eight 3-coordinate lithium atoms in the crystal. It is
noteworthy that the eight N-methyl substituents of these chiral
ligands completely fill the interior cavity of the octomer.
Configurational Stability of 7. In these studies, we chose
to generate organolithium 7 by tin−lithium exchange from
stannane 12 because we wanted to avoid the possibility of
incomplete deprotonation of 6, which could complicate the
analysis of the product mixtures, and because we wished to
generate 7 efficiently while also having evidence of its
formation.
Attempts were made to monitor changes in the solution
structure during a catalytic resolution. To that end, a mixture of
racemic stannane 12, TMEDA (4 equiv), and the conjugate
acid of ligand 10 (10 mol %) was placed in an NMR tube and
cooled to −78 °C. A solution of n-Bu6Li was added, and the
sample was transferred to an NMR probe cooled to −80 °C.
The temperature was maintained at −80 °C for 2 h to complete
the tin−lithium exchange, then raised to −45 °C for 3 h to
effect the dynamic resolution, and finally cooled back to −80
°C (we previously determined that enantioenriched 7 is
configurationally stable at −80 °C16). Four spectra were
obtained, with an accumulation time of ∼1 h in each case
(Figure 6). Most noticeably, the 1:1:1 triplet indicating a
Scheme 6 summarizes the experiments used to evaluate the
configurational stability of 7 formed by tin−lithium exchange in
Figure 6. Partial 13C spectra of a CDR of 7 conducted in an NMR
tube.
Scheme 6. Transmetalation and Configurational Stability
Studies
contact ion pair comprising one carbon and one lithium atom is
evident in all four spectra. Changes were also seen in the region
between 50 and 53 ppm: a series of unassigned peaks are barely
evident in the top spectrum, disappear in the spectra at −45 °C,
and become much more prominent at the end of the resolution.
Quenching the solution in the NMR tube with phenyl
isocyanate and subsequent workup afforded an 89:11 ratio of
pipecolic acid anilide enantiomers [see the Supporting
Information (SI) for details].17a
Figure 7. X-ray crystal structure of dilithio ligand 10: (a) ball-and-stick rendering of the octameric crystal structure comprising two stacked 16-
membered rings and two molecules of diethyl ether; (b) one 16-membered-ring tetramer, with one monomer colored blue; (c) stacked tetramers,
with the rear tetramer colored red.
16849
dx.doi.org/10.1021/ja307796e | J. Am. Chem. Soc. 2012, 134, 16845−16855

Journal of the American Chemical Society
Article
the absence or presence of (−)-sparteine (13), bispidine 3, or
TMEDA. We previously found that tin−lithium exchange of S-
12 in the absence of any ligand at temperatures of −60 °C or
above result in racemic product.16 Below −60 °C, the lithiation
is too slow to be synthetically useful, but it was attempted at
−78 °C to investigate the configurational stability in the
absence of ligands. At this temperature, the half-life for tin−
lithium exchange is >5 h (see the SI for kinetic details). After
treatment of 12 with n-BuLi for 5 h at −78 °C, quenching with
Me3SiCl produced a mixture of 12 and 9, but the latter was
racemic. In the presence of either 13 or 3, TLC monitoring
showed that transmetalation of S-12 (73:27 er) to give 7 was
complete after 5 h; however, quenching with Me3SiCl afforded
racemic 9. This result is in contrast to the 87:13 er of 7
obtained by deprotonation using sec-BuLi in the presence of
13,3a perhaps reflecting different mechanisms for the formation
of 7 involving different types of lithium ligation in different
solvents or in the presence of different amines. In
tetrahydrofuran in the absence of diamines, for example, tin−
lithium exchange occurs with complete retention of config-
uration.4a As noted previously,16 TMEDA plays a unique role in
the racemization and DTR of lithiopiperidine 7: both processes
are first order in [TMEDA] when 0−1 molar equiv of TMEDA
is used. After treatment of stannane 12 (73:27 er) with n-BuLi
for only 30 min at −78 °C in the presence of 1 equiv of
TMEDA, quenching with Me3SiCl revealed complete trans-
metalation, and 9 was obtained in 73:27 er. Of the ligands
tested, only TMEDA is suitable for the CDR of 7 following
tin−lithium exchange, consistent with its previously-noted
uniqueness.16
where [S]t is the fraction of the S enantiomer at time t and [S]0
and [S]eq are the fractions of the S enantiomer at t = 0 and t =
∞, respectively, at the temperature of interest (here, [S]0 = 0.5
and [S]eq = 0.77). The data revealed a half-life of ∼3 h at this
temperature.
The Effect of [8] on the Dynamic Resolution of 7. To
evaluate how varying the amount of chiral ligand 8 affects the
dynamic resolution, experiments were conducted using catalysts
loadings of 15, 20, 50, 80, and 100 mol %. After 5 h at −55 °C,
er values of 63:37, 66:34, 71:29, 75:25, and 78:22 were
observed for 9 (Figure 9, diamonds). The data in Figure 9
Figure 9. Effect of varying [8] on the dynamic resolution of rac-7 with
TMEDA (1.2 equiv) at −55 °C in ether for 5 h, after quenching with
Me3SiCl. The circles and the straight line indicate the fractions
calculated if the system were to contain rac-7·TMEDA and 7·8 with an
er of 77:23. The diamonds indicate the experimental values observed
after 5 h.
DTR Kinetics at −55 °C. Previously, the barriers for
inversion in the DTR of rac-7 by stoichiometric quantities of 8
were determined at four temperatures ranging from −10 to +20
°C.16 Because we planned to investigate catalytic resolution at
considerably lower temperatures, we decided to measure the
kinetics of stoichiometric DTR at −55 °C. The time evolution
of the stoichiometric DTR of rac-7 by 8 in the presence of 1.2
equiv of TMEDA was measured, as illustrated in Figure 8. The
evolution of the major enantiomer (in this case, the S
enantiomer) is given by the equation
clearly show that catalysis occurred, since the observed er is
always higher than what would have been predicted if 7 not
bound to 8 was racemic (Figure 9, circles), as was realized at
higher temperatures (see Figure 15, below). This is all the more
surprising since the values should be lower than the predicted
value if equilibrium were not yet reached. The SI contains a
similar plot for ligand 10. Thus, a quantitative evaluation of the
rates of resolution with differing amounts of 8 and 10 was
undertaken.
obst
[S]t = [S]eq + ([S]0 − [S]eq ) e−k
(1)
The dependence of the rate of resolution of lithiopiperidine
7 on chiral ligands 8 and 10 was investigated at −20 °C in the
presence of TMEDA (1.0 equiv) in Et2O (see the SI for
details). Figure 10 shows plots of the observed first-order rate
constant versus the amount of 8 or 10 over the range 0−1
equiv. A nonlinear fit of the data to the equation kobs = k[L*]n
afforded n = 0.5 and k = (1.59 0.06) × 10−3 M−0.5 s−1 for 8. A
similar fit of the observed rate constants plotted against [10]
revealed n = 0.265
0.022 and k = (9.49
0.12) × 10−4
M−0.265 s−1 (see the SI for details). Thus, the reaction is half
order in [8] and approximately one-fourth order in [10]. A
fractional order is indicative of a mechanism involving
deaggregation of the chiral ligand.
The Effect of [TMEDA] on CDR of 7. As noted above,
TMEDA plays a unique role in the inversion dynamics of
lithiopiperidine 7 (both enantiomerization and DTR).16 It was
therefore of interest to explore the kinetic role of TMEDA in
the CDR of 7. The time evolution of S-7 under CDR
conditions in the presence of varying amounts of TMEDA (0−
4 equiv) was followed, and the rate constants were measured
Figure 8. Evolution of the fraction of the S enantiomer in the
stoichiometric DTR of 7 by 8 (1.0 equiv) in the presence of TMEDA
(1.2 equiv) at −55 °C. A nonlinear least-squares fit of the data to eq 1
using [S]eq = 0.77 and [S]0 = 0.5 yielded kobs = (7.48 0.63) × 10−5
s−1.
16850
dx.doi.org/10.1021/ja307796e | J. Am. Chem. Soc. 2012, 134, 16845−16855

Journal of the American Chemical Society
Article
generating organolithium 7 (er 85:15 or 96:4 S:R) using tin−
lithium exchange in Et2O at −78 °C with BuLi and 2 or 4 equiv
of TMEDA, warming the reaction mixture to the desired
temperature for different time periods, and then cooling to −78
°C and quenching with excess Me3SiCl as previously described
(see the SI for details).16 The rate constants were determined
by nonlinear fits to the zero-order plots. Eyring analysis of the
rate constants at their respective temperatures in the presence
of varying amounts of TMEDA provided the thermodynamic
parameters listed in Table 1. The plots of Gibbs free energy as a
Table 1. Thermodynamic Parameters for Racemization of 7
with Varying Amounts of TMEDA
equiv of
TMEDA
ΔH⧧
ΔS⧧
entry
RLi·L
(kcal/mol)
(cal mol−1 K−1
)
a
1
7·TMEDA
7·TMEDA
7·TMEDA
1.0
2.0
4.0
18.1 0.7
20.1 1.8
22.5 2.0
5.0 3.2
9.9 7.6
16.7 8.8
2
3
a
Data taken from ref 14.
function of temperature11 for various amounts of TMEDA
(Figure 12a) reveal that racemization has a significantly higher
energy barrier with 4 equiv of TMEDA than with 1 or 2 equiv
at all temperatures between −80 and 0 °C. Figure 12b shows
that in the presence of 2 equiv of TMEDA, racemization and
resolution of 7 by stoichiometric amounts of 8 have equal free
energy barriers at −33 °C.11 Below that temperature, resolution
has a lower barrier.
Figure 10. Plots of the observed rate constant kobs vs the
concentrations of (top) 8 and (bottom) 10 for the dynamic resolution
of rac-7 (0.06 M) in the presence of TMEDA (1.0 equiv) at −20 °C in
Et2O. The point at the origin in each graph is not experimental.
(Figure 11). Upward curvature is emblematic of a higher-order
dependence of the rate on [TMEDA]. A nonlinear parametric
In a preliminary communication, we showed that the rate of
enantiomerization of 7 is first order in TMEDA up to 1 equiv.16
However, the above experiments revealed that the barrier for
enantiomerization is raised significantly in the presence of
excess TMEDA. Because our CDR studies were carried out
using excess TMEDA, we decided to study the rate of
enantiomerization with excess TMEDA. The details are given
in the SI, and the results are plotted in Figure 13. There is an
inverse dependence of the racemization rate on [TMEDA]
when TMEDA is in excess relative to 7. Thus, for the purposes
of developing the CDR, 2−4 equiv of TMEDA beneficially
retards racemization. In the preparative experiments described
later and also those using ligands 10 and 11,17 we employed 4
equiv of TMEDA, which served to accelerate resolution and
retard racemization.
Approach to the −55 °C Equilibrium from Either
Direction. The zero-order plots for the CDR starting from
rac-7 approach but never exceed the er value observed in the
stoichiometric DTR (77:23). In our study of the CDR of N-
(trimethylallyl)pyrrolidine 2, we showed that R-2 (93:7 er)
maintained configurational stability under the catalytic
conditions for 90 min.9 For lithiopiperidine 7, we employed
stannylpiperidine S-12 with 96:4 er,17a and it was of interest to
see whether this er would be maintained under the catalytic
conditions or drop to the 77:23 er observed when starting with
racemate. We subjected both enantiomers to catalytic
conditions (15 mol % 8, 2 equiv of TMEDA, −55 °C) and
recorded the evolution of the er, as shown in Figure 14. In both
experiments, the final er matched the er in the stoichiometric
DTR. The rate constant kSR for equilibration of S-7 (96:4 er) in
the presence of 2 equiv of TMEDA and 15 mol % 8 is (1.93
0.11) × 10−5 s−1. Similarly, the rate constant kRS for CDR of R-
7 (72:28 er) is (6.24 0.45) × 10−5 s−1. The kRS/kSR ratio is
Figure 11. Effect of varying [TMEDA] on the CDR of 7 (0.06 M) in
the presence of 8 (0.15 equiv) in ether at −55 °C for 4 h. The solid
curve is a fit to the equation kobs = k[TMEDA]n with k = (4.02 0.11)
× 10−3 M−2 s−1 and n = 1.99.
fit of the equation kobs = k[TMEDA]n revealed a second-order
dependence on [TMEDA], with n = 1.99 and k = (4.02 0.11)
× 10−3 M−2 s−1. This indicates that the transition structure of
the rate-limiting step in the CDR contains two TMEDA
molecules.
Racemization of 7 with Excess TMEDA. Since the
dynamics of racemization of S-7 were previously determined in
the presence of 1 equiv of TMEDA, and since we observed a
second-order dependence of the CDR on [TMEDA], we re-
examined the racemization kinetics, this time in the presence of
excess TMEDA. The enantiomerization was followed by
16851
dx.doi.org/10.1021/ja307796e | J. Am. Chem. Soc. 2012, 134, 16845−16855

Journal of the American Chemical Society
Article
Figure 12. (a) Plots of ΔG⧧ vs temperature for the enantiomerization of 7 with varying amounts of TMEDA. (b) Plots of ΔG⧧ vs temperature for
racemization and stoichiometric DTR of 7 in the presence of 2 equiv of TMEDA.11
3.23, which is similar to the S-7·8/R-7·8 ratio in a DTR (3.35
for 77:23 er).
Effect of Temperature on the Position of the
Equilibrium. The above discussion emphasizes the importance
of staying below the crossover temperature to ensure that the
free energy barrier for resolution is below the barrier for
racemization. Two control experiments demonstrated the
importance of temperature control. Racemic lithiopiperidine 7
was subjected to catalytic conditions (4 equiv of TMEDA and
15 mol % 8 or 10 at −55 °C), and the er was measured at
several time intervals until it reached 76:24 for 8 and 95:5 for
10. The temperature was then raised to 0 °C, and the er was
monitored again. The dramatic results are illustrated in Figure
15. At 0 °C, the er dropped dramatically to ∼55:45. Clearly,
careful control of the temperature is necessary to achieve a
Figure 13. Effect of 1−4 equiv of TMEDA on the rate of racemization
of 7 (0.25 M) at −40 °C.
CDR.
Optimized Conditions for CDR. Given the above findings,
we decided on the optimum conditions of 5−15 mol % 10 and
4 equiv of TMEDA, which we subsequently used in several
Figure 14. Evolution of CDR starting from enantioenriched R- or S-12.
16852
dx.doi.org/10.1021/ja307796e | J. Am. Chem. Soc. 2012, 134, 16845−16855

Journal of the American Chemical Society
Article
2. The 6Li spectra of S-7 exhibit dynamic phenomena as the
temperature is raised to −60 °C and above.
3. Splitting of the 13C NMR signal for the carbanionic
carbon persists throughout the CDR. Other signals
appear in the same region, suggesting the presence of
other species.
4. Consistent with faster organolithium dynamics at higher
temperatures, rapid racemization is observed when the
temperature of the reaction mixture containing resolved
7 is raised to 0 °C.
5. Chiral ligands 8 and 10 increase the rate of CDR. The
kinetic order for CDR of lithiopiperidine 7 is 0.5 for 8
and 0.265 for 10. Fractional orders implicate deaggrega-
tion of the chiral catalysts. Ligand 10 is an octamer in the
solid state.
6. Excess TMEDA accelerates the CDR. Although the
stoichiometric DTR is first order in [TMEDA],16 the
CDR (0.15 equiv of 8) is second order in [TMEDA].
7. Excess TMEDA retards racemization. The rate constant
for enantiomerization of lithiopiperidine 7 is at a
maximum when the ratio 7:TMEDA = 1. Racemization
is first order in [TMEDA] up to 1 equiv of TMEDA
(however, racemization is very fast in the absence of
TMEDA) and inverse order when excess TMEDA is
present (up to 4 equiv).
8. The final er values of the stoichiometric DTR and the
CDR are the same.
9. Temperature control is critical to the success of the
CDR.
Figure 15. Proof of a low-temperature process that effects a catalytic
resolution at −55 °C but which at 0 °C: CDR of 7 in the presence of
TMEDA (4.0 equiv) at −55 °C followed by warming to 0 °C for 30
min results in rapid racemization using either (a) ligand 8 or (b) ligand
10.
synthetic applications. Both direct electrophilic substitutions
and transmetalations with ZnCl2 followed by Cu(I) or Pd(0)
cross-couplings were demonstrated by numerous examples with
a high degree of enantioselectivity, usually ≥95:5 er.17
Examples using trimethylsilyl chloride and phenyl isocyanate
as electrophiles are given in Scheme 7 for comparison of ligands
8 and 10.
10. For the CDR that converts R-7 to S-7 using L* as the
catalyst, the rate law is
d[S‐7]
dt
= k[R‐7][TMEDA]2[L*]n
where n = 0.5 for 8 and 0.265 for 10.
We have observed CDR for five combinations of the
organolithium species, catalyst, and achiral ligand (Figure 16):
2-lithio-N-(trimethylallyl)pyrrolidine 2 as the organolithium
with either lithium diaminoalkoxide 4 or (−)-sparteine as the
catalyst and N,N′-diisopropylbispidine 3 as the achiral ligand9
and 2-lithio-N-Boc-piperidine 7 as the organolithium with
either lithium diaminoalkoxide 8 or dilithio ligands 10 or 11 as
the catalyst and TMEDA as the achiral ligand.17 What is clear
from these studies is that ligand exchange is an important
process in dynamic resolutions of 7 and 2 both stoichiometric
and catalytic conditions. The effect of a ligand on the rate of
carbanion inversion of a chiral organolithium species can
change with the ligand/organolithium molar ratio. It is also
clear that an achiral ligand and careful control of temperature
are necessary to achieve a catalytic resolution. As a guide for
choosing the temperature, we used the “crossover temper-
ature”, where the free energy barrier for racemization and
dynamic resolution are equal: below this temperature, CDR can
occur, but above this temperature it cannot. Further refinement
of the temperature is based on the free energy barrier for the
CDR such that a reasonable half-life can be achieved.
SUMMARY AND CONCLUSIONS
Several important details have emerged from these studies:
■
1. NMR data are inconclusive as to the solution structure of
7, although differences in the 6Li spectra of rac-7 and S-7
are evident.
Scheme 7. Optimized CDR Conditions
16853
dx.doi.org/10.1021/ja307796e | J. Am. Chem. Soc. 2012, 134, 16845−16855

Journal of the American Chemical Society
Article
EXPERIMENTAL SECTION
■
For general experimental conditions, see the SI.
General Procedure for Determining the Effect of Varying [8]
on Dynamic Resolution of 7 in the Presence of TMEDA. In
oven-dried tubes equipped with stir bars and capped with rubber septa,
rac-12 (0.06 M in ether) was treated with BuLi (1.2 equiv) and
TMEDA (1.2 equiv) at −78 °C for 1 h under argon to effect tin−
lithium exchange, affording rac-7·TMEDA. The concentration of 8 was
varied (0.0 to 100 mol %), and the tubes were quickly transferred to
the −55 °C bath. After 5 h at this temperature, the tubes were
transferred to the −78 °C bath and rapidly quenched with excess
Me3SiCl. The silanes were subsequently analyzed by chiral-stationary-
phase supercritical fluid chromatography (CSP-SFC) with monitoring
at 210 nm. See the SI for details.
Activation Parameters for Racemization of 7 with 2.0 equiv
of TMEDA. In a typical kinetic run, S-12 (85:15 er, 0.06 M in ether) in
the presence of TMEDA (2.0 equiv) was treated with BuLi (1.2 equiv)
at −78 °C for 30 min under nitrogen in oven-dried septum-capped
tubes equipped with a stir bars to effect tin−lithium exchange. The
tubes were quickly transferred to a second thermostatted bath and
warmed to the desired temperature. At various time intervals a tube
was cooled to −78 °C and rapidly quenched with excess Me3SiCl.
After 4 h, MeOH (2 mL) was added, and the mixture was extracted
into Et2O. The silanes were subsequently analyzed by CSP-SFC with
monitoring at 210 nm. The rate constants were determined by
nonlinear fits to the zero-order plots, and Eyring analysis of the
temperature-dependent rate constants provided the thermodynamic
parameters. See the SI for details.
Details of X-ray Crystal Structure Determination of Ligand
1 0 .
A
c l e a r , c o l o r l e s s c h u n k l i k e s p e c i m e n o f
Figure 16. Organolithium and ligand combinations amenable to CDR.
C96H192Li16N16O8·(C4H10O)2 (approximate dimensions 0.280 mm ×
0.300 mm × 0.340 mm) was used for the X-ray crystallographic
analysis. The X-ray intensity data were measured on a Bruker Apex 1K
diffractometer. Integration of the data using a triclinic unit cell yielded
a total of 21 236 reflections to a maximum θ angle of 26.48° (0.80 Å
resolution), of which 18 549 were independent (average redundancy =
1.145, completeness = 83.5%, Rint = 3.81%, Rsig = 7.20%) and 14 814
(79.86%) were greater than 2σ(F2). The final cell constants were
determined to be a = 11.668(4) Å, b = 17.093(6) Å, c = 18.174(6) Å,
α = 64.950(5)°, β = 76.278(5)°, γ = 88.581(6)°, and V = 3178.2(19)
Å3 on the basis of refinement of the XYZ-centroids of reflections above
20σ(I). Data were corrected for absorption effects using the multiscan
method (SADABS). The ratio of minimum to maximum apparent
transmission was 0.534−0.711.
The structure was solved and refined using the Bruker SHELXTL
software package in the space group P1 with Z = 1 for the formula unit
C104H212Li16N16O10. The final anisotropic full-matrix least-squares
refinement on F2 with 1343 variables converged at R1 = 7.53% for the
observed data and wR2 = 22.77% for all data. The goodness of fit was
1.036. The largest peak in the final difference electron density
synthesis was 0.869 e/Å3, and the largest hole was −0.564 e/Å3, with a
root-mean-square deviation of 0.076 e/Å3. On the basis of the final
model, the calculated density was 1.023 g/cm3 and F(000) was 1076e.
If the organolithium species and the chiral ligand are
monomeric, CDR is not possible because the process would
violate the principle of microscopic reversibility. However, as
shown by Toru and co-workers,6 irreversible population of a
homochiral aggregate from a racemic monomer or a
heterochiral aggregate by CDR is possible. In the case of the
piperidines studied in this work, there is some evidence of
differences between the solution structures of racemic and
enantioenriched 7. In this system, we found no visible evidence
of aggregate precipitation but did see evidence of variable
solution structure that is dependent on temperature.
Furthermore, we also observed a change in the final enantiomer
ratio that is dependent on temperature. In addition, the
fractional order of the CDR with respect to L* provides clear
evidence that deaggregation of the resolving ligand is operative
in the dynamic resolution of 7. We have also demonstrated that
the catalytic resolution requires exchange with an achiral ligand,
in this case TMEDA.19 These observations may be due to a
change in solution structure between racemic and enantioen-
riched 7, such as the formation of a homochiral aggregate at low
temperature. At higher temperatures such as 0 °C, new
pathways for racemization become operative, such that the
organolithium that is presumably complexed to TMEDA
racemizes rapidly while the remainder, complexed to L*,
remains enantioenriched. Additionally, the fractional kinetic
order in ligands 8 and 10 implicates changes in the aggregation
state of the resolving ligand in the rate-determining step. Our
working hypothesis of this phenomenon is that a successful
CDR depends on ligand exchange with an achiral ligand and on
aggregation−deaggregation phenomena of the species in
solution, including the chiral organolithium and/or the
resolving ligand.
ASSOCIATED CONTENT
■
S
* Supporting Information
Full experimental details, including kinetic plots, data analysis,
and the NMR experiments. This material is available free of
charge via the Internet at http://pubs.acs.org.
AUTHOR INFORMATION
■
Corresponding Author
bgawley@uark.edu
Notes
The authors declare no competing financial interest.
16854
dx.doi.org/10.1021/ja307796e | J. Am. Chem. Soc. 2012, 134, 16845−16855

Journal of the American Chemical Society
Article
(12) Plots of ΔG⧧ vs temperature over a large temperature range
assume that ΔH⧧ and ΔS⧧ are constant over the entire temperature
range. Since the data in our Eyring plots were obtained over a small
temperature range of 20−30 °C, the obtained values of ΔH⧧ and ΔS⧧
are valid over the temperature range where the measurements were
made, and we assume that they offer a first approximation over the
larger range.
ACKNOWLEDGMENTS
■
We thank the National Science Foundation (CHE 0616352 and
1011788) and the Arkansas Biosciences Institute for direct
support of this work. T.P. thanks the NSF-REU Program for a
summer fellowship (CHE 0851505). Core facilities were
funded by the Arkansas Biosciences Institute and the National
Institutes of Health (P30 RR031154 and P30 GM103450). We
also thank Professors Donna Blackmond, Iain Coldham, David
Collum, and Jonathan Clayden for helpful discussions.
(13) Lee, W. K.; Park, Y. S.; Beak, P. Acc. Chem. Res. 2009, 42, 224−
234.
(14) (a) Low, E.; Gawley, R. E. J. Am. Chem. Soc. 2000, 122, 9562−
9563. (b) Gawley, R. E.; Klein, R.; Ashweek, N. J.; Coldham, I.
Tetrahedron 2005, 61, 3271−3280.
(15) Barker, G.; O’Brien, P.; Campos, K. R. ARKIVOC 2011, No. v,
217−229.
(16) Coldham, I.; Leonori, D.; Beng, T. K.; Gawley, R. E. Chem.
Commun. 2009, 5239−5240; Corrigendum with correct enthalpy and
entropy values: 2010, 9267−9268.
REFERENCES
■
(1) (a) Clayden, J. Organolithiums: Selectivity for Synthesis; Pergamon
Press: London, 2002. (b) Organolithiums in Enantioselective Synthesis;
Hodgson, D. M., Ed.; Springer: Berlin, 2003; Vol. 5. (c) The Chemistry
of Organolithium Compounds; Rappoport, Z., Marek, I., Eds.; Wiley:
Chichester, 2004. (d) Organometallics: Compounds of Group 1 (Li ...
Cs); Majewski, M., Snieckus, V., Eds.; Thieme: Stuttgart, 2006; Vol. 8a.
(e) Stereochemical Aspects of Organolithium Compounds; Gawley, R. E.,
Ed.; Topics in Stereochemistry, Vol. 26; Wiley-VCH: Weinheim,
Germany, 2010.
(17) (a) Beng, T. K.; Gawley, R. E. J. Am. Chem. Soc. 2010, 132,
12216−12217. (b) Beng, T. K.; Gawley, R. E. Org. Lett. 2011, 13,
394−397. (c) Beng, T. K.; Gawley, R. E. Heterocycles 2012, 84, 697−
718.
̈
(18) Reich, H. J.; Sikorski, W. H.; Gudmundsson, B. O.; Dykstra, R.
R. J. Am. Chem. Soc. 1998, 120, 4035−4036.
(19) In the catalytic resolution of 2 by 4, the achiral ligand required is
bispidine 3 (see ref 9).
(2) (a) Hoppe, D.; Hense, T. Angew. Chem., Int. Ed. Engl. 1997, 36,
2283−2316. (b) Hoppe, D.; Christoph, G. In The Chemistry of
Organolithium Compounds; Rappoport, Z., Marek, I., Eds.; Wiley:
Chichester, 2004; pp 1055−1164. (c) Beak, P.; Johnson, T. A.; Kim,
D. D.; Lim, S. H. Top. Organomet. Chem 2003, 5, 139−176.
(d) Kizirian, J.-C. Top. Stereochem. 2010, 26, 189−251. (e) Mitchell,
E. A.; Peschiulli, A.; Lefevre, N.; Meerpoel, L.; Maes, B. U. W.
Chem.Eur. J. 2012, 18, 10092−10142.
(3) (a) Bailey, W. F.; Beak, P.; Kerrick, S. T.; Ma, S.; Wiberg, K. B. J.
Am. Chem. Soc. 2002, 124, 1889−1896. (b) Coldham, I.; O’Brien, P.;
Patel, J. J.; Raimbault, S.; Sanderson, A. J.; Stead, D.; Whittaker, D. T.
E. Tetrahedron: Asymmetry 2007, 18, 2113−2119. (c) Metallinos, C.;
Dudding, T.; Zaifman, J.; Chaytor, J. L.; Taylor, N. J. J. Org. Chem.
2007, 72, 957−963.
(4) (a) Stead, D.; Carbone, G.; O’Brien, P.; Campos, K. R.; Coldham,
I.; Sanderson, A. J. Am. Chem. Soc. 2010, 132, 7260−7261. (b) Dixon,
A. J.; McGrath, M. J.; O’Brien, P. In Organic Syntheses; Wiley: New
York, 2009; Collect. Vol. 11, pp 25−37.
(5) (a) Beak, P.; Anderson, D. R.; Curtis, M. D.; Laumer, J. M.;
Pippel, D. J.; Weisenburger, G. A. Acc. Chem. Res. 2000, 33, 715−727.
(b) Coldham, I.; Dufour, S.; Haxell, T. F. N.; Howard, S.; Vennall, G.
P. Angew. Chem., Int. Ed. 2002, 41, 3887−3889. (c) Coldham, I.;
Dufour, S.; Haxell, T. F. N.; Vennall, G. P. Tetrahedron 2005, 61,
3205−3220. (d) Whisler, M. C.; MacNeil, S.; Snieckus, V.; Beak, P.
Angew. Chem., Int. Ed. 2004, 43, 2206−2225. (e) Coldham, I.; Dufour,
S.; Haxell, T. F. N.; Patel, J. J.; Sanchez-Jimenez, G. J. Am. Chem. Soc.
2006, 128, 10943−10951. (f) Coldham, I.; Raimbault, S.; Chovatia, P.
T.; Patel, J. J.; Leonori, D.; Sheikh, N. S.; Whittaker, D. T. E. Chem.
Commun. 2008, 4174−4176. (g) Coldham, I.; Sheikh, N. S. Top.
Stereochem. 2010, 26, 253−293. (h) Coldham, I.; Raimbault, S.;
Whittaker, D. T. E.; Chovatia, P. T.; Leonori, D.; Patel, J. J.; Sheikh, N.
S. Chem.Eur. J. 2010, 16, 4082−4090.
(6) (a) Nakamura, S.; Hirata, N.; Kita, T.; Yamada, R.; Nakane, D.;
Shibata, N.; Toru, T. Angew. Chem., Int. Ed. 2007, 46, 7648−7650.
(b) Nakamura, S.; Hirata, N.; Yamada, R.; Kita, T.; Shibata, N.; Toru,
T. Chem.Eur. J. 2008, 14, 5519−5527.
(7) Bispidine 3 was first used as an exchangeable achiral ligand in
catalytic deprotonations. See: McGrath, M. J.; O’Brien, P. J. Am. Chem.
Soc. 2005, 127, 16378−16379.
(8) If retentive substitution is assumed, this implies accumulation of
the R organolithium.
(9) Beng, T. K.; Yousaf, T. I.; Coldham, I.; Gawley, R. E. J. Am. Chem.
Soc. 2009, 131, 6908−6909.
(10) Tin−lithium exchange was used to generate 2 from 1, a process
that fails in the absence of a diamine ligand.
(11) In all cases, the best fits of the kinetic data were to reversible
first-order plots.
16855
dx.doi.org/10.1021/ja307796e | J. Am. Chem. Soc. 2012, 134, 16845−16855