Highly enantioselective zirconium-catalyzed cyclization of aminoalkenes

Article abstract of DOI:10.1021/ja4000189

Aminoalkenes are catalytically cyclized in the presence of cyclopentadienylbis(oxazolinyl)borato group 4 complexes {PhB(C₅H ₄)(Ox^R)₂}M(NMe₂)₂ (M = Ti, Zr, Hf; Ox^R = 4,4-dimethyl-2-oxazoline, 4S-isopropyl-5,5- dimethyl-2-oxazoline, 4S-tert-butyl-2-oxazoline) at room temperature and below, affording five-, six-, and seven-membered N-heterocyclic amines with enantiomeric excesses of >90% in many cases and up to 99%. Mechanistic investigations of this highly selective system employed synthetic tests, kinetics, and stereochemistry. Secondary aminopentene cyclizations require a primary amine (1-2 equiv vs catalyst). Aminoalkenes are unchanged in the presence of a zirconium monoamido complex {PhB(C₅H ₄)(Ox4S-iPr,Me₂)₂}Zr(NMe₂)Cl or a cyclopentadienylmono(oxazolinyl)borato zirconium diamide {Ph₂B(C ₅H₄)(Ox4S-iPr,Me₂)}Zr(NMe₂) ₂. Plots of initial rate versus [substrate] show a rate dependence that evolves from first-order at low concentration to zero-order at high concentration, and this is consistent with a reversible substrate-catalyst interaction preceding an irreversible step. Primary kinetic isotope effects from substrate conversion measurements (k′_obs^(H)/ k′_obs^(D) = 3.3 ± 0.3) and from initial rate analysis (k₂^(H)/k₂^(D) = 2.3 ± 0.4) indicate that a N-H bond is broken in the turnover-limiting and irreversible step of the catalytic cycle. Asymmetric hydroamination/cyclization of N-deutero-aminoalkenes provides products with higher optical purities than obtained with N-proteo-aminoalkenes. Transition state theory, applied to the rate constant k₂ that characterizes the irreversible step, provides activation parameters consistent with a highly organized transition state (ΔS^? = -43(7) cal·mol^-1 K ^-1) and a remarkably low enthalpic barrier (ΔH ^? = 6.7(2) kcal·mol^-1). A six-centered, concerted transition state for C-N and C-H bond formation and N-H bond cleavage involving two amidoalkene ligands is proposed as most consistent with the current data.

Full text of DOI:10.1021/ja4000189

Article
pubs.acs.org/JACS
Highly Enantioselective Zirconium-Catalyzed Cyclization of
Aminoalkenes
Kuntal Manna, William C. Everett, George Schoendorff, Arkady Ellern, Theresa L. Windus,
and Aaron D. Sadow*
Department of Chemistry and U.S. Department of Energy Ames Laboratory, Iowa State University, Ames, Iowa 50011, United States
S
* Supporting Information
ABSTRACT: Aminoalkenes are catalytically cyclized in the
presence of cyclopentadienylbis(oxazolinyl)borato group 4
complexes {PhB(C5H4)(OxR)2}M(NMe2)2 (M = Ti, Zr, Hf;
OxR = 4,4-dimethyl-2-oxazoline, 4S-isopropyl-5,5-dimethyl-2-
oxazoline, 4S-tert-butyl-2-oxazoline) at room temperature and
below, affording five-, six-, and seven-membered N-hetero-
cyclic amines with enantiomeric excesses of >90% in many
cases and up to 99%. Mechanistic investigations of this highly
selective system employed synthetic tests, kinetics, and
stereochemistry. Secondary aminopentene cyclizations require
a primary amine (1−2 equiv vs catalyst). Aminoalkenes are unchanged in the presence of a zirconium monoamido complex
{PhB(C5H4)(Ox4S‑iPr,Me )2}Zr(NMe2)Cl or a cyclopentadienylmono(oxazolinyl)borato zirconium diamide {Ph2B(C5H4)-
2
(Ox4S‑iPr,Me )}Zr(NMe2)2. Plots of initial rate versus [substrate] show a rate dependence that evolves from first-order at low
2
concentration to zero-order at high concentration, and this is consistent with a reversible substrate−catalyst interaction preceding
(D)
an irreversible step. Primary kinetic isotope effects from substrate conversion measurements (k′obs(H)/k′obs = 3.3 0.3) and
(D)
from initial rate analysis (k2(H)/k2 = 2.3 0.4) indicate that a N−H bond is broken in the turnover-limiting and irreversible
step of the catalytic cycle. Asymmetric hydroamination/cyclization of N-deutero-aminoalkenes provides products with higher
optical purities than obtained with N-proteo-aminoalkenes. Transition state theory, applied to the rate constant k2 that
characterizes the irreversible step, provides activation parameters consistent with a highly organized transition state (ΔS⧧ =
−43(7) cal·mol−1 K−1) and a remarkably low enthalpic barrier (ΔH⧧ = 6.7(2) kcal·mol−1). A six-centered, concerted transition
state for C−N and C−H bond formation and N−H bond cleavage involving two amidoalkene ligands is proposed as most
consistent with the current data.
highly enantioselective bis(amidate)16−18 and bis-
INTRODUCTION
■
(carboxyamide)19 complexes (or similar Ta-based systems)20
The pursuit of optically active amines is motivated by their
importance in commodity and specialty chemicals, medicinal
as catalysts. C1-symmetric mixed Cp*(oxazolinylaryloxo)-
zirconium and constrained geometry-type catalysts uniquely
compounds, and natural products.1 Although enantioselective
depart from the predominant axially chiral ligand design.21 Few
olefin hydroamination is potentially a powerful and efficient
approach for amine synthesis, significant challenges remain.2
highly enantioselective late-transition-metal-catalyzed amino-
pentene cyclizations have been reported.22 Although the
Regio- and stereochemical control for additions of aliphatic
enantioselective systems have been generally limited to
alkenes and amines is currently limited; few catalysts provide
terminal amine products (anti-Markovnikov addition) without
control from directing groups, and intermolecular Markovnikov
addition reactions are typically poorly enantioselective.
Products with high optical purities are limited to a few choice
pyrrolidines even in the well-studied intramolecular hydro-
amination/cyclization; asymmetric olefin hydroaminations that
afford functionalized pyrrolidines, piperidines, and azapanes
pyrrolidines, highly efficient and selective intramolecular
catalysts may provide strategies for addressing some of the
tougher challenges in olefin hydroamination.
In this context, catalyst synthesis, substrate screening, and
mechanistic tests may further advance the field. Detailed studies
that probe hydroamination/cyclization mechanisms have
focused on d0 or fnd0 complexes including rare earth,6d,11,23,24
actinide,25 alkaline earth metal,26−28 titanium,29 and zirco-
(among many valuable chiral N-heterocycles) are not yet
nium.25,30−32 Typically, the catalytic cycles associated with
developed.
these species are proposed to have metal-amido resting states.
Still, great progress is highlighted by advances with
metallocene-based and axially chiral lanthanide catalysts,3−11
alkali metal coordination complexes,12 and alkaline earth metal
catalysts.13 Breakthroughs in zirconium-catalyzed alkene hydro-
From that point, the proposed mechanisms for C−N bond
Received: January 7, 2013
Published: April 30, 2013
amination include axially chiral bis(aryloxy),14 bis(amido),15
a
© 2013 American Chemical Society
7235
dx.doi.org/10.1021/ja4000189 | J. Am. Chem. Soc. 2013, 135, 7235−7250

Journal of the American Chemical Society
Article
formation are remarkably divergent to include (A) olefin
insertion into a [M]−NR2 bond and (B) [2π + 2π]
cycloaddition of olefin and [M]NR involving four-centered
metallacyclic intermediates. Recently, (C) a concerted C−N
and C−H bond-forming mechanism was proposed to involve a
six-centered transition state (Scheme 1).
mechanistically distinct from other addition reactions involving
unsaturated C−C moieties and M−E bonds (e.g., M−H).
Mechanistic assignments are further complicated in
zirconium-catalyzed cyclizations by conflicting observations
involving primary and secondary aminoalkenes as substrates for
transformation by cationic and neutral catalysts. Cationic
catalyst systems, such as Cp2ZrMe2/B(C6F5)3, are only active
for secondary aminoalkenes, and catalyst deactivation with
primary amines is proposed to involve formation of inactive,
neutral zirconium imidoalkene species.14,34 In contrast, several
neutral zirconium complexes cyclize primary, but not
secondary, aminoalkenes; this observation is often invoked as
support for a zirconium imido intermediate prior to [2π + 2π]
cycloaddition. Notably, this reaction pathway for imidozirco-
nium compounds and olefins (B in Scheme 1) is not
independently established (although [Zr]NR and alkyne
cycloadditions are well-known).35 In fact, mixtures of an
isolated zirconium imido and alkenes return starting materi-
als.36 Furthermore, an interesting dipyrrolylmethane zirconium
compound cyclizes both primary and secondary amino-
alkenes.37
Scheme 1. Proposed Pathways for C−N Bond Formation in
d0 and fnd0 Metal-Catalyzed Aminopentene Cyclization
Likewise, isolation of the organometallic metal−alkylamine
product from olefin insertion has not yet been described for a
d0 or fnd0 metal amido compound. However, the zirconium
vinylamine product of alkyne insertion into a zirconium−
nitrogen bond was recently isolated and characterized.38 In
addition, a molybdenum amido and alkyne react to provide a d2
molybdenum vinylamine.39 Interestingly, strong support for
olefin insertion into dn metal amidos (n ≠ 0) has been provided
in palladium systems,40,41 and an early catalytic hydroamination
reaction involves norbornene insertion into an [Ir]−NHPh
bond.42
These unusual features, as well as the potential for amine
synthesis, motivated our initial investigations of zirconium-
catalyzed aminoalkene cyclization using an achiral cyclopenta-
dienylbis(oxazolinyl)borate ligand.32 A second-order rate law
(−d[aminopentene]/dt = k[Zr]1[aminopentene]1) and primary
isotope effect were rationalized through reversible substrate
addition, but compelling evidence ruling out “incorrect”
pathways was not available. As noted above, many early
transition-metal hydroaminations are plagued by similar
mechanistic ambiguity.
Additionally, modifications to the olefin insertion mechanism
have included a proposal that the C−N bond formation is
facilitated by a coordinated amine (proton-assisted insertion)3
and a suggestion that the bond-forming step is reversible based
on a computational analysis.33
Unfortunately, the experimental characteristics of most
catalytic systems are similar, complicating mechanistic
elucidation. For example, rate laws often show zero-order
dependence on substrate concentration, irrespective of the
proposed mechanism. Generally, reactions of N-deuterated
amines are slower (i.e., there are primary isotope effects), and
Eyring plots invariably indicate that transition states are highly
ordered. Because these few characteristics describe most
hydroamination/cyclization reactions, insight into subtle effects
on rate and selectivity is often limited.
A few interesting observations, however, provide some
guidance. First, diastereoselectivity in organolanthanide-cata-
lyzed cyclization of racemic aminoalkenes is enhanced with
high substrate concentration and through the addition of
propylamine, but it is lowered with N-deuterated substrates
(Scheme 2).23
Stereochemistry provides an additional tool for assessing the
mechanism; therefore, we prepared optically active compounds
containing chiral oxazolines. The resulting zirconium com-
pounds are remarkably active and highly enantioselective.43
This system is characterized by a number of interesting
observations, including the nonlinear correlation between
substrate concentration and cyclization rate, the requirement
that the substrate/catalyst ratio should be >2 for catalytic
turnover, the positive effect of N-deuteration on the
enantiomeric excess of the product, and the observation that
cyclization of secondary aminoalkenes requires a primary amine
additive. These observations are inconsistent with [2π + 2π]
cycloaddition and insertion pathways (A and B, Scheme 1).
Instead, we proposed concerted C−N and C−H bond
formation through a two-substrate intermediate involving a
six-centered transition state. A similar pathway was suggested
but ultimately excluded for a related zirconium catalyst.30 In
addition, two-substrate pathways have been proposed for
magnesium and calcium,26,27b,c,28 rare earth,11,24 and zirconium
catalysts.31
Scheme 2. Conditions That Influence the cis/trans Ratio in
Cyclization of Racemic Aminoalkenes
Second, zero-order substrate dependence suggests a
unimolecular turnover-limiting step, arguing strongly for the
insertion mechanism (Scheme 1A), while the primary-in-
magnitude isotope effect suggests that N−H bond cleavage
occurs between the catalyst resting state and the turnover-
limiting transition state.23 Insertion of alkenes into M−N bonds
would not be expected to have a significant isotope effect. This
contrast suggests that C−N bond formations may be
7236
dx.doi.org/10.1021/ja4000189 | J. Am. Chem. Soc. 2013, 135, 7235−7250

Journal of the American Chemical Society
Article
Table 1. 11B (128 MHz) and 15N NMR (71 MHz) Chemical
Shifts and νCN Values for Oxazoline-Containing Compounds
Here we describe a detailed investigation of this group-4-
based system for asymmetric hydroamination. We have studied
the precatalyst structures, pursued structure/activity/selectivity
relationships for precatalysts and substrate combinations, and
correlated these observations with kinetic effects to better
understand this highly active and highly enantioselective system
for C−N bond formation.
a
a
b
compound
11B NMR
15N NMR
νCN
1589
H2[PhBCp(OxMe )2] (H2[1]) −15.3, −15.6,
−172.0
2
−16.0
H2[PhBCp(Ox4S‑iPr,Me )2]
−15.7
−15.7
−179.1
−179.1
−190.9
−147.0
1584,
1567
2
(H2[S-2])
H2[PhBCp(Ox4R‑iPr,Me )2]
1582,
1560
2
RESULTS
(H2[R-2])
■
H2[PhBCp(Ox4S‑tBu)2] (H2[S- −14.9, −15.3,
1595,
1588
1. Synthesis and Characterization of Oxazolinyl-
boranes and Oxazolinylborates. Four dianionic cyclopenta-
dienylbis(oxazolinyl)borates are prepared as diprotonated
species (Scheme 3).11,32,43 The compounds H[PhB(C5H5)-
3])
−15.6
−8.1
PhB(OxMe )2 (4)
1621,
1601
2
PhB(OxS‑iPr,Me )2 (S-5)
−7.5
−7.3
−131.6
−124.3
1588
2
PhB(OxS‑tBu)2 (S-6)
1617,
1590
Scheme 3. Synthesis of Achiral and Chiral Oxazolinylboranes
and Cyclopentadienylbis(oxazolinyl)borates
H2[Ph2BCp(Ox4S‑iPr,Me )]
−10.0, −11.5,
−12.4
−155.2
1594
2
(H2[S-7])
2H-Ox4S‑iPr,Me
na
−143.3
−148.0
1632
1635
2
2H-Ox4S‑tBu
na
a
Value given in parts per million, measured in acetonitrile-d3 solvent.
b
KBr, cm−1.
(oxazolinyl)borate ligands 1−3, is prepared by reaction of
2Li-Ox4S‑iPr,Me and Ph2BCl followed by addition of Na[C5H5].
2
Exhaustive purification by silica gel column chromatography
yields H[Ph2B(C5H5)(Ox4S‑iPr,Me )] (H2[S-7]) in poor yield (eq
2
1). Three 11B NMR resonances indicates that this cyclopenta-
dienylmono(oxazolinyl)borate is also a mixture of three
isomers.
(OxR)2] (H2[1]; OxMe = 4,4-dimethyl-2-oxazoline), H2[S-2];
2
Ox4S‑iPr,Me = 4S-isopropyl-5,5-dimethyl-2-oxazoline), H2[R-2];
2
2. Synthesis and Characterization of Cyclopenta-
dienyloxazolinylborate Group 4 Compounds. The
preparation of cyclopentadienyloxazolinylborate group 4
complexes involves reaction of homoleptic dimethylamide
compounds M(NMe2)4 (M = Ti, Zr, Hf) and H2[1], H2[S-
2], H2[R-2], H2[S-3], and H2[7] (eq 2). However, the
conditions for synthesis and the complex stability vary
significantly for the metal−ligand pairs.
Ox4R‑iPr,Me = 4R-isopropyl-5,5-dimethyl-2-oxazoline), H2[S-3];
2
Ox4S‑tBu = 4S-tert-butyl-2-oxazoline) are synthesized in two
steps from the appropriate 2H-oxazolines 2H-OxR via neutral
bis(oxazolinyl)phenylborane PhB(OxR)2 intermediates 4,44 S-5,
R-5, and S-6.
The 11B and 15N NMR chemical shifts and infrared CN
stretching frequencies for 4, S-5, and S-6 are listed in Table 1.
The 11B NMR values are consistent with neutral, four-
coordinate boron centers and suggest that a neutral donor
(i.e., acetonitrile-d3) coordinates to the boron center.45
The bis(oxazolinyl)phenylboranes 4−6 react with Na[C5H5]
in THF to provide the desired mixed cyclopentadienyl-
oxazolinylborates H[PhB(C5H5)(OxR)2] after chromatography.
Isolated H2[1], H2[S-2], H2[R-2], and H2[S-3] are obtained as
The achiral H2[1] and M(NMe2)4 (M = Zr, Hf) react rapidly
1
mixtures of three isomers, as indicated by H and 11B NMR
in benzene at room temperature, affording {PhB(C5H4)-
spectroscopy (Table 1). The 1H NMR spectra of these
mixtures were complicated by overlapping resonances in the
phenyl, cyclopentadienyl, and oxazoline regions and were not
useful for characterization or assessment of purity. 11B NMR
spectroscopy was more informative, and the three resonances in
the 11B NMR spectrum of achiral 1 were upfield compared to
that of borane 4 (as an acetonitrile adduct), as expected.45 The
1-C5H5(BR3), 2-C5H5(BR3), and 5-C5H5(BR3) connectivity
accounts for the three isomers. Borylcyclopentadienyl com-
pounds are known to form mixtures of isomers.46
1
(OxMe )2}M(NMe2)2 ({1}M(NMe2)2; eq 3). H NMR spectra
2
A chiral cyclopentadienylmono(oxazolinyl)diphenylborate,
synthesized for comparison with the cyclopentadienylbis-
of micromolar-scale experiments in benzene-d6 indicated that
the reactions proceeded to completion within 10 min.
7237
dx.doi.org/10.1021/ja4000189 | J. Am. Chem. Soc. 2013, 135, 7235−7250

Journal of the American Chemical Society
Article
Table 2. 15N NMR (71 MHz) Chemical Shifts, νCN and [α]2D0 Values for Cyclopentadienylbis(oxazolinyl)borate Group 4
Complexes
compound
15N NMR
νCN
1595
[α]2D0
na
{PhBCp(OxMe )2}Zr(NMe2)2 ({1}Zr(NMe2)2); amorphous
−135
2
{1}Zr(NMe2)2; crystallized
na
1603, 1505
1610, 1533
1595
na
{1}Zr(NMe2)2THF; crystallized
{1}Hf(NMe2)2
na
na
−132
na
{PhBCp(Ox4S‑iPr,Me )2}Zr(NMe2)2 ({S-2}Zr(NMe2)2)
−153, −155
−152, −155
−149, −151
−154, −156
−145, −148
−153, −155
−155
1565
−124.7°
+122.6°
−86.98°
−94.71°
−139.1°
−111.4°
−82.0°
2
{R-2}Zr(NMe2)2
1559
{S-2}Hf(NMe2)2
1559
{S-2}Ti(NMe2)2
{PhBCp(Ox4S‑tBu)2}Zr(NMe2)2 ({S-3}Zr(NMe2)2)
1560
1608, 1506
1578, 1567
1562
{S-2}Zr(NMe2)Cl
{Ph2BCp(Ox4S‑iPr,Me )}Zr(NMe2)2 ({S-7}Zr(NMe2)2)
2
Analytically pure materials are obtained in 98% isolated yield
for both zirconium and hafnium compounds by evaporation of
benzene and the HNMe2 byproduct.
{1}Zr(NMe2)2THF were identical to those of {1}Zr(NMe2)2
in the absence of THF and those of THF in the absence of
{1}Zr(NMe2)2. Interestingly, the 1H NMR spectrum of
{1}Zr(NMe2)2THF at 190 K contained only two broad single
resonances, one assigned to oxazoline methyl and the other to
oxazoline methylene groups. The two methyl groups on each
oxazoline are expected to remain chemically inequivalent at all
temperatures, and a rigid structure for {1}Zr(NMe2)2(THF)
should contain either two or four methyl signals (depending on
the structure). Thus, the fluxional process is too fast on the 1H
NMR time scale to resolve a rigid structure. The hafnium
analogue {1}Hf(NMe2)2 is also fluxional in the presence of 1
equiv of THF, and the room temperature resonances assigned
to oxazoline methyl substituents and the methylene group
coalesce into two broad singlets at 190 K.
We therefore pursued X-ray-quality crystals of THF-free
{1}Zr(NMe2)2, which were obtained by slow diffusion of
pentane into a toluene solution of {1}Zr(NMe2)2 cooled to 243
K (Figure 1). As in {1}Zr(NMe2)2THF, the oxazolinylborate
ligand bonds to Zr through the C5H4B moiety and one
oxazoline group (Zr1−N1, 2.255(1) Å), while the second
The spectroscopic data for {1}Zr(NMe2)2 and {1}Hf-
(NMe2)2 are similar (Table 2). Upon coordination and
1
amine elimination, the H NMR multiplets associated with
the three C5H5B isomers are replaced with two resonances
assigned to C5H4B in a Cs-symmetric molecule. One set of
oxazoline resonances was observed as two singlets for
inequivalent methyl groups and two doublets for the
inequivalent CH2. This pattern is invariant in 1H NMR spectra
acquired at temperatures from 190 to 300 K in toluene-d8.
1H−15N HMBC experiments contained a correlation between
oxazoline methyl and nitrogen, and the 15N NMR chemical
shifts for the oxazoline nitrogen in zirconium and hafnium
compounds at −135 and −132 ppm are only slightly upfield of
noncoordinated 2H-4,4-dimethyl-2-oxazoline (−127 ppm). For
comparison, the 15N NMR chemical shift of {κ3-PhB(OxMe2)3}-
Zr(NMe2)3 is −140 ppm, in which tridentate oxazoline
coordination to Zr is unambiguous.47 In the current system,
the small change in 15N NMR data of complexes versus 2H-
oxazoline does not distinguish two simultaneously coordinated
oxazolines from one coordinated oxazoline rapidly exchanging
with a pendent oxazoline. Analytically pure, amorphous
{1}Zr(NMe2)2 and {1}Hf(NMe2)2, obtained by evaporation
of frozen benzene solutions, were analyzed by IR spectroscopy.
Only one νCN band was observed for either compound (both at
1595 cm−1; KBr). From these data, either (a) the C5H4B group
and both oxazolines are bonded to the metal center or (b) the
C5H4B group and only one oxazoline are coordinated with the
pendent oxazoline and coordinated oxazoline exchanging
1
rapidly on the H NMR time scale in solution and the IR
time scale in the solid state.
A single-crystal X-ray diffraction study of {1}Zr(NMe2)2THF
showed only one oxazoline coordinated to the zirconium
center.32 The coordination geometry of zirconium is a squashed
four-legged piano stool. The dimethylamide groups are transoid
(N−Zr−N: 120.26°) as are the oxazoline and THF ligands
(N−Zr−O: 159.4°). The site trans to the cyclopentadienyl ring
is unoccupied.
The coordination modes of {1} in this structure and the
structure of amorphous {1}Zr(NMe2)2 are not equivalent. In
particular, νCN bands at 1610 cm−1 (pendant oxazoline) and
1533 cm−1 (Zr-OxMe ) were observed for crystals from THF.
2
Figure 1. ORTEP diagram of {PhB(C5H4)(OxMe )2}Zr(NMe2)2
2
Furthermore, the conformation of {1}Zr(NMe2)2THF in the
({1}Zr(NMe2)2). Ellipsoids are plotted at 50% probability, and
hydrogen atoms are not illustrated for clarity. Atomic distances (Å):
Zr1−N1, 2.255(1); Zr1−N3, 2.031(2); Zr1−N4, 2.035(2).
solid state is not representative of the solution structure. At
1
room temperature, the signals in the H NMR spectrum of
7238
dx.doi.org/10.1021/ja4000189 | J. Am. Chem. Soc. 2013, 135, 7235−7250

Journal of the American Chemical Society
Article
oxazoline is not coordinated to Zr (Zr1−N2, 4.31 Å). The
geometry at zirconium is that of a three-legged piano stool; the
Cpcentroid−Zr−N1 angle (103.1°) is similar to that of the THF
adduct, while the Cpcentroid−Zr−N3 and Cpcentroid−Zr−N4
angles (110.47 and 122.35°) are smaller than in {1}Zr-
(NMe2)2THF. The Zr−NMe2 distances (Zr1−N3, Zr1−N4)
are slightly shorter in {1}Zr(NMe2)2 than in the THF adduct
(just outside of 3σ), and the Zr−Noxazoline distance is 0.066 Å
shorter in the THF-free species. We wished to identify whether
this compound would provide the same IR spectrum as the
material obtained from evaporation of benzene. In fact, the two
spectra are inequivalent. A spectrum of the crystalline sample
(KBr) contained two IR bands at 1603 and 1505 cm−1 assigned
to νCN stretching frequencies of noncoordinated and Zr-
coordinated oxazolines. These IR data contrast that of the
material obtained by evaporation, where only one νCN band at
1595 cm−1 was observed.
copy suggests that both oxazolines coordinate to the metal
centers in {S-2}M(NMe2)2. Two 15N NMR signals were
observed at −152.6 and −155.0 in comparison to −143 for 2H-
Ox4S‑iPr,Me . Additionally, only one CN stretching frequency was
2
observed in the IR at 1565 cm−1 (KBr). A νCN associated with a
noncoordinated oxazoline, based on the higher energy νCN of
2H-Ox4S‑iPr,Me (1632 cm−1), was not detected.
2
The solution structure of {S-2}Zr(NMe2)2 is further probed
with a NOESY experiment to identify which groups on the
ancillary ligand interact with the presumed reactive sites that
are occupied by NMe2 groups. The correlations and assign-
ments are illustrated in Figure 2. First, the {S-2} intraligand
In contrast to readily available Zr and Hf species, the reaction
of Ti(NMe2)4 and H2[1] in benzene at room temperature
1
provides a mixture of unidentified products. The H NMR
spectrum does not contain the same spectral signatures as
1
noted above for {1}Zr(NMe2)2. Instead, H NMR spectra of
reaction mixtures in benzene-d6 or toluene-d8 contained vinylic
resonances from isomers of C5H5B; the conversion of these
signals to those associated with a single Cs-symmetric
compound was not observed even upon heating at 120 °C.
HNMe2 was observed in spectra of reaction mixtures within 10
min, which indicated that some ligand substitution occurred.
Four 11B resonances were detected ranging from 33.6 to −14.7
ppm, and the upfield 11B NMR signal suggested that a borane
was among the reaction products.
Interestingly, the {S-2} ligand supports a titanium complex as
well as zirconium and hafnium compounds. The chiral, 4S-
isopropyl-5,5-dimethyl-2-oxazoline-based H2[S-2] and M-
(NMe2)4 (M = Ti, Zr, Hf) react in benzene at room
temperature, providing {S-2}M(NMe2)2 in excellent yield (eq
4; Ti: 25 h, 86%; Zr: 7 h, 96%; Hf: 7 h, 92%).
Figure 2. Correlations detected by NOESY experiments. The
bidentate coordination mode is illustrated.
correlations establish the metal−ligand conformations that are
populated sufficiently to provide NOEs (Figure 2A). The two
sets of correlations discerned are oxazoline−C5H4 cross-peaks
and oxazoline−oxazoline interactions.
Interestingly, the former correlations only involve the lateral
2- and 5-hydrogen on the C5H4 group (Figure 2A) to methyl
and isopropyl groups of both oxazolines. No correlations
between the oxazoline and the 3- and 4-hydrogen were
observed in the NOESY. The lateral correlations suggest
significant population of configurations in which the zirconium-
ligand interaction contains one dissociated oxazoline {η5-κ1-
Ph(Ox4S‑iPr,Me2)B(C5H4)(Ox4S‑iPr,Me2)}Zr(NMe2)2. This confor-
mation is further supported by the X-ray crystal structure of
{1}Zr(NMe2)2. However, there are also through-space
correlations between the two oxazolines that suggest both
oxazolines coordinate to zirconium in an isomer that exists on
an NOE-significant time scale. Thus, we suggest that two
conformations dominate the structure of {S-2}Zr(NMe2)2 and
exchange rapidly on the NMR time scale (eq 5).
The opposite enantiomer {R-2}Zr(NMe2)2 is prepared from
H2[R-2] and Zr(NMe2)4. As expected, the NMR and IR
spectroscopic features of {S-2}Zr(NMe2)2 and {R-2}Zr-
(NMe2)2 are identical. Two sets of oxazoline resonances (e.g.,
two septets and four doublets assigned to the isopropyl groups
and four singlets for the 5-methyl groups) and four downfield
multiplets ranging from 6.7 to 6.06 ppm for the cyclo-
pentadienyl group suggested C1-symmetric species for {S-
2}Zr(NMe2)2 and {R-2}Zr(NMe2)2. Two singlets are observed
for the two inequivalent dimethylamide ligands. Additionally,
the broad 11B NMR resonance at −15.7 ppm resulting from
overlapping signals from isomers of H2[S-2] is replaced with a
sharper signal at −14.5 ppm (122 Hz at half-height). The
compounds {S-2}Ti(NMe2)2 and {S-2}Hf(NMe2)2 have
1
similar H NMR spectra. The C stereocenters in the molecule
necessarily render the two oxazolines inequivalent, so 1H NMR
spectroscopy does not distinguish between zero-, one-, or two-
coordinated oxazolines. However, 15N NMR and IR spectros-
7239
dx.doi.org/10.1021/ja4000189 | J. Am. Chem. Soc. 2013, 135, 7235−7250

Journal of the American Chemical Society
Article
Furthermore, the intra- and interoxazoline correlations
establish the relative orientation of the two oxazolines (Figure
2B). On the basis of the assignments and NOEs between
oxazoline and NMe2 ligands, we assign the signals for the
inequivalent NMe2 groups and the overall configuration of the
zirconium center (Figure 2C). Interestingly, the “back” NMe2
in Figure 2 at 2.83 ppm correlations show through-space
interactions with only the back oxazoline, whereas the “front”
NMe2 at 2.70 ppm contained correlations with resonances from
both oxazoline rings. Thus, the open space between the
cyclopentadienyl and the back oxazoline is occupied by a
dimethylamide group, and the other dimethylamide group is
situated between the two oxazoline rings in the dicoordinated
conformer.
ppm indicated that a single product formed. However, the
complex {S-3}Zr(NMe2)2 decomposes in benzene at room
temperature (t1/2 = 2 h), leading to HNMe2 and unidentified
oxazolinylborate products. In contrast, {S-2}M(NMe2)2 com-
plexes (M = Ti, Zr, Hf) are unchanged in benzene even after 36
h at room temperature.
Additionally, several compounds were required to test the
reactive valences required for hydroamination catalysis and the
nature of ancillary ligand−zirconium interaction and its impact
on cyclization and enantioselectivity. The compound {S-
2}Zr(NMe2)Cl was prepared to address the former point.
Reaction of H2[S-2] and 2 equiv of KCH2C6H5 affords K2[S-2].
Reaction of K2[S-2] and ZrCl4 in THF at room temperature
provides {S-2}ZrCl2 within 30 min.
1
Four doublets and four singlets in the H NMR spectrum
To enhance the clarity of the presentation of through-space
interactions, a rendered image is given in Figure 3. The
zirconium−ligand interaction shows the C1-symmetric environ-
ment that distinguishes the two NMe2 groups.
(THF-d8) assigned to oxazoline isopropyl and methyl groups
were consistent with inequivalent oxazolines expected for a C1-
symmetric complex. Four multiplets in the downfield region
from 6.88 to 6.23 ppm were attributed to the cyclopentadienyl
group in {S-2}ZrCl2. The complex is unchanged in THF-d8 for
7 days at room temperature; however, broad signals were
1
observed in H NMR spectra of {S-2}ZrCl2 after the solvent
was evaporated and the residue was dissolved in THF-d8.
Therefore, the complex {S-2}ZrCl2 is best prepared in situ for
the synthesis of {S-2}Zr(NMe2)Cl.
The desired compound {S-2}Zr(NMe2)Cl is prepared by
addition of 1 equiv of LiNMe2 to {S-2}ZrCl2 (generated in situ,
eq 7). The complex is isolated in 79% yield by evaporation of
the volatile materials from the reaction mixture, followed by
benzene extraction. A mixture of two diastereomers is formed
1
(in 4:1 ratio as detected by H NMR spectroscopy of THF-d8
solutions).
Figure 3. DFT-optimized {S-2}Zr(NR2)2 structure (B3LYP/3-21G).
Hydrogen atoms and the amide substituents are not included for
clarity. Nitrogens are depicted in blue, and the back NMe2 group is
significantly less hindered than the front group, according to NOESY
experiments.
The 1H NMR resonances associated with the ancillary ligand
in each of the two diastereomers indicated the (expected) C1-
symmetry, although many of the resonances are overlapping.
The 4:1 diastereomeric ratio remains unchanged in THF at
room temperature after 12 h.
A mono(oxazolinyl)borato zirconium {S-7}Zr(NMe2)2 com-
A second optically active zirconium complex {PhB(C5H4)-
(Ox4S‑tBu)2}Zr(NMe2)2 ({S-3}Zr(NMe2)2) is prepared by
reaction of H2[S-3] and Zr(NMe2)4 in benzene at room
temperature for 20 min (eq 6).11 In its 1H NMR spectrum, two
pound is prepared by reaction of H[Ph2B(C5H5)(Ox4S‑iPrMe )]
2
and Zr(NMe2)4 in benzene at room temperature. {S-7}Zr-
(NMe2)2 is formed after 20 min and is isolated in 92% yield (eq
8).
singlet resonances at 0.85 and 0.84 ppm were assigned to tert-
butyl substituents on two inequivalent oxazoline groups, and
four downfield multiplets ranging from 6.74 to 6.09 ppm were
assigned to the cyclopentadienyl group. Likewise and as in S-2,
two singlets for two inequivalent dimethylamido groups were
observed, and the overall pattern was consistent with C1
symmetry. In the 11B spectrum, a sharp resonance at −14.4
The 1H NMR spectrum of {S-7}Zr(NMe2)2 in benzene-d6 is
constant at room temperature over at least 12 h. One set of
oxazoline resonances (i.e., one septet and two doublets assigned
to the isopropyl groups and two singlets for the 5-methyl
groups), two singlets for two inequivalent dimethylamido
7240
dx.doi.org/10.1021/ja4000189 | J. Am. Chem. Soc. 2013, 135, 7235−7250

Journal of the American Chemical Society
Article
Table 3. Catalytic Cyclization of Aminopentenes with Achiral and Optically Active Group 4 Compounds Coordinated by Mixed
a
Cyclopentadienylbis(oxazolinyl)borato Ligands
a
b
c
1
Reaction conditions: 10 mol % precatalyst. Isolated yield is given in parentheses. Percent enantiomeric excess ( 0.5%) was determined by H
and/or 19F NMR spectroscopy of the Mosher amide derivatives. Absolute configuration assignments are based on literature reports.6d,16a Percent
d
enantiomeric excess measured by 19F NMR of Mosher amide derivatives and also by HPLC of naphthoyl or benzylamide derivatives.
groups, and four downfield multiplets for the cyclopentadienyl
group suggest a C1-symmetric species for {S-7}Zr(NMe2)2. The
1H NMR pattern was constant from 190 to 320 K in toluene-d8.
In the IR spectrum, one νCN at 1562 cm−1 for {S-7}Zr(NMe2)2
indicates that the oxazoline is coordinated to the zirconium
center.
3. Catalytic Hydroamination/Cyclization of Amino-
alkenes. Cyclopentadienylbis(oxazolinyl)borato bis(amido)
group 4 compounds catalyze the cyclization of aminoalkenes
to give racemic and optically enriched pyrrolidine, piperidine,
azepane, and indoline products at room temperature (Tables
3−5). These results are organized into tables by substrate type.
Table 3 gives cyclization reactions of aminopentenes 8a−13a
that have aliphatic or aromatic 3,3-disubstitution; Table 4
contains six- and seven-membered rings from substrates 14a−
16a; Table 5 reports cyclizations of substrates 18a−22a that
give indolines as well as halogenated, acetal, and 2-ethyl-
containing pyrrolidine products. In contrast, cyclization of
aminoalkenes is not observed in the presence of monoamido
{S-2}Zr(NMe2)Cl or mono-oxazoline {S-7}Zr(NMe2)2 under
the conditions tested.
The rates of cyclization depend on the metal center, ancillary
ligand, and substrate. Some general trends of catalytic rate (for
which exceptions are identified below) are summarized as
follows: first, the metal center affects the reaction rate following
the trend Zr > Hf ≫ Ti. Second, the ancillary ligand effect on
cyclization rates of disubstituted aminoalkenes follows the trend
{S-2}Zr(NMe2)2 > {1}Zr(NMe2)2 > {S-3}Zr(NMe2)2.
{S-2}Zr(NMe2)2 provides a more active catalyst than {S-
2}Hf(NMe2)2. In many cases, only starting materials were
detected in reactions of {S-2}Ti(NMe2)2 and aminopentenes
(allowed to stand at room temperature for extended times). In
fact, only three of the most highly reactive substrates, 2,2-
diphenylpent-5-ene amine (8a), 2,2-bis(2-propenyl)-4-
pentenylamine (11a), and C-(1-allylcyclohexyl)methylamine
(12a), are cyclized by {S-2}Ti(NMe2)2 at room temperature
with appreciable conversion.
For zirconium catalysts, the oxazoline substituents affect
hydroamination rates, generally following the trend {S-
7241
dx.doi.org/10.1021/ja4000189 | J. Am. Chem. Soc. 2013, 135, 7235−7250

Journal of the American Chemical Society
Article
2}Zr(NMe2)2 > {1}Zr(NMe2)2 > {S-3}Zr(NMe2)2. For
example, the diphenyl substrate 8a is cyclized to 4,4-
diphenyl-2-methylpyrrolidine 8b in 1.25 h by {S-2}Zr(NMe2)2,
11 h by {1}Zr(NMe2)2, and 18 h by {S-3}Zr(NMe2)2 (Table 3,
entries 2, 1, and 9). This trend applies to the formation of most
pyrrolidine, azepane, and indoline products found in Tables
3−5. However, achiral {1}Zr(NMe2)2 provides a more active
catalyst than {S-2}Zr(NMe2)2 for the unsubstituted amino-
pentene 13a (Table 3, entries 29 and 30) and aminohexenes
14a and 15a (Table 4, entries 1 and 2). Substrate 13a is readily
cyclized by {1}Zr(NMe2)2 even at room temperature, while {S-
2}Zr(NMe2)2 requires 110 °C for only a few turnovers. Only
13a and HNMe2 are detected with {S-2}Hf(NMe2)2, {S-
2}Ti(NMe2)2, or {S-3}Zr(NMe2)2 as precatalysts.
Table 4. Catalytic Hydroamination/Cyclization of
Aminohexenes and Aminoheptenes
a
Several generalizations about the influence of ancillary ligand,
metal center, solvent, and temperature on enantioselectivity are
apparent. Overall, this group 4 system provides pyrrolidines
with excellent optical purities. For example, 8a is cyclized to R-
8b with 98% ee in the presence of {S-2}Zr(NMe2)2 at −30 °C
and in 97% ee with {S-2}Hf(NMe2)2 at 0 °C (Table 3, entries 5
and 7). Impressively, the enantioselectivities of reactions
catalyzed by {S-2}Zr(NMe2)2 are high for a range of
aminoalkenes. In the series of pyrrolidines, spirocyclohexyl
12b (90%), spirocyclopentyl 9b (92%), trimethyl 10b (89%),
and diallyl 11b (93%) are obtained with high enantiopurity at
room temperature using {S-2}Zr(NMe2)2 (Table 3). The
enantioselectivity is not sensitive to catalyst loading. For
example, 2−15 mol % of {S-2}Zr(NMe2)2 provides 8b with
93% ee at all catalyst loadings.
a
Reaction conditions: 10 mol % precatalyst, benzene, room temper-
b
c
ature. Isolated yield is given in parentheses. Percent enantiomeric
excess was determined by 19F NMR spectroscopy of the (R)-Mosher
d
e
amide derivatives. At 85 °C. Percent enantiomeric excess measured
by HPLC.
Table 5. Catalytic Hydroamination/Cyclization of
Functionalized Aminoalkenes
a
In general, {S-2}Zr(NMe2)2, {S-2}Hf(NMe2)2, and {S-
3}Zr(NMe2)2 provide products with better optical purities
than the titanium-based precatalyst {S-2}Ti(NMe2)2. The
isopropyl-5,5-dimethyl-2-oxazoline ligands often give pyrroli-
dines with higher % enantiomeric excesses than the 4S-tert-
butyl-2-oxazoline ligand. In contrast, the trivalent yttrium
catalyst system containing the 4S-tert-butyl-2-oxazoline-based
ligand provides pyrrolidines with higher % ee than {S-
2}YCH2SiMe3.11 In the present zirconium system, tri(allyl)-
methylamine 11a is cyclized with {S-2}Zr(NMe2)2 to give 4,4-
diallyl-2-methylpyrrolidine 11b with 93% ee, whereas the
enantioselectivity provided by {S-3}Zr(NMe2)2 is 88%. Similar
effects are observed with the cyclopentyl substrate (92% vs
88%) and cyclohexyl substrate (90% vs 87%), while the
diphenyl pyrrolidine forms with equivalent % ee using {S-
2}Zr(NMe2)2 and {S-3}Zr(NMe2)2. As expected, the mirror-
image precatalyst {R-2}Zr(NMe2)2 gives equivalent reactivity
and enantioselectivity as {S-2}Zr(NMe2)2 but provides the
products with the opposite absolute configuration (eq 9). Thus,
an additional advantage of the valine-derived ligands versus tert-
leucine is that both enantiomers are readily available for the
former.
a
b
Reaction conditions: 10 mol % catalyst, room temperature. Isolated
c
yield is given in parentheses. Ratio of diastereomers defined as cis/
d
trans. Percent enantiomeric excess ( 0.5%) was determined by 19F
e
f
NMR spectra of Mosher amide derivatives. At −30 °C. Percent
enantiomeric excess verified by HPLC.
The hafnium precatalyst {S-2}Hf(NMe2)2 is also highly
enantioselective, and the examples here are the first in which a
hafnium complex affords hydroamination products with optical
7242
dx.doi.org/10.1021/ja4000189 | J. Am. Chem. Soc. 2013, 135, 7235−7250

Journal of the American Chemical Society
Article
purities >90%. In fact, {S-2}Hf(NMe2)2 provides pyrrolidines
with higher enantioselectivity than {S-2}Zr(NMe2)2 (under
equivalent conditions) for about half of the aminoalkenes that
were tested. In particular, {S-2}Hf(NMe2)2 gives higher % ee in
benzene-d6 at room temperature for cyclopentylpyrrolidine (9b,
95% ee) and cyclohexylpyrrolidine (12b, 93% ee). However,
under these conditions, 4,4-diphenyl (8b), 4,4-dimethyl (10b),
and 4,4-diallyl-2-methylpyrrolidine (11b) are formed with
greater optical purities in reactions catalyzed by {S-2}Zr-
(NMe2)2 than {S-2}Hf(NMe2)2. The enantiomeric excesses
were always lowest with {S-2}Ti(NMe2)2 as the precatalyst,
regardless of the substrate. Thus, the effect of metal center on
enantioselectivity follows the trend Zr ≈ Hf > Ti for this mixed
cyclopentadienylbis(oxazolinyl)borate group 4 system.
High optical purities are obtained from reactions performed
in benzene, toluene, methylene chloride, and THF. However,
the apparent turnover rate is diminished in THF and methylene
chloride compared to benzene and toluene. With the {S-
2}Zr(NMe2)2-catalyzed cyclization of 8a as a representative
example, benzene-d6 and methylene chloride-d2 solvents
provide the same enantioselectivity (93% ee), although the
reaction is faster in benzene-d6. Higher enantiomeric excess is
obtained in THF at room temperature (95%, Table 3, entry 3),
and conditions of lower temperature (−30 °C) in toluene-d8
solvent provide the product with 98% ee.
In fact, the asymmetric induction is significantly enhanced at
lower temperatures, although long reaction times are needed in
several cases. For example, {S-2}Zr(NMe2)2 catalyzes the
formation of 2,4,4-trimethylpyrrolidine 10b at −30 °C with
93% enantiomeric excess, whereas the product is formed with
89% ee at room temperature (Table 3, entries 16 and 17). A
similar trend is also observed for {S-2}Hf(NMe2)2, which
mediates the formation of 4,4-diallyl-2-methylpyrrolidine 11b
with 96% ee at 0 °C (Table 3, entry 21) compared to 90% ee at
room temperature.
Thus, {S-2}Zr(NMe2)2 provides a highly effective catalyst for
the asymmetric cyclization of aminoalkenes to 2-methylpyrro-
lidines. Typically, hydroaminations of aminohexenes and
aminoheptenes (Table 4) are more difficult than the amino-
pentenes of Table 3. In fact, the rates of conversion of
aminohexenes 14a and 15a, as catalyzed by {S-2}Zr(NMe2)2,
are significantly decreased with respect to the aminopentenes.
Additionally, the enantiomeric excesses of the 2-methyl-
piperidine products are significantly lower than the 2-
methylpyrrolidines. The catalytic rate of {S-2}Hf(NMe2)2 is
superior to the rate of the zirconium congener for piperidine
formation, although elevated temperature is required.
including eletriptan and indapamide.48 Previously described
syntheses of highly enantioenriched (>90%) 2-stereogenic
indolines rely on either catalytic asymmetric hydrogenation of
indoles or kinetic resolution of racemic indolines.49,50
Previously reported hydroamination catalysts provide 2-
methylindoline from ortho-allylaniline with moderate enantio-
selectivity.15,22 Therefore, complexes {S-2}Zr(NMe2)2 and {S-
2}Hf(NMe2)2 are promising for preparation of optically active
indoline building blocks from aniline derivatives.
Hydroaminations in methylene chloride and THF, described
above, hint at functional group tolerance. Therefore, we
investigated cyclization reactions of aminopentenes that contain
halogen or ether functionality. Quite promisingly, high
enantiomeric excess is obtained in the conversion of 2,2-
diethoxyaminopentene 18a catalyzed by {S-2}Zr(NMe2)2 (97%
ee) and {S-2}Hf(NMe2)2 (96% ee; Table 5, entries 5 and 6).
The achiral {1}Zr(NMe2)2 mediates the cyclization at
approximately the same rate as {S-2}Zr(NMe2)2, but neither
the 4S-tert-butyl-oxazolinylborate-based {S-3}Zr(NMe2)2 nor
titanium-based compounds are effective.
2-Allyl-2-(4-bromophenyl)pent-4-enylamine 19a is cyclized
by {1}Zr(NMe2)2 to a 2:1 diastereomeric mixture of cis- and
trans-pyrrolidines at room temperature (Table 5, entry 7). The
assignment of the major isomer as cis is supported by NOE
experiments in benzene-d6 in which irradiation of the 2-methyl
signal of the major isomer (1.00 ppm) gives decreased intensity
of the ortho- and meta-phenyl resonances at 7.27 and 6.69 ppm
for that isomer. Interestingly, the optically active {S-2}Zr-
(NMe2)2 is more efficient (full conversion in 20 min) than the
achiral catalyst, gives better diastereoselectivity (cis/trans =
4:1), and provides the products with exceedingly high
enantioselectivity for both diastereomers (cis, 97% ee; trans,
95% ee). The diastereomeric ratio (cis/trans = 2:1) and the
optical purity of the major enantiomer (at 93% ee) are lower
with {S-2}Hf(NMe2)2. High optical purity is obtained with {S-
3}Zr(NMe2)2 (Table 5, entry 11), although the diastereomeric
ratio is 1:1.2 (but now favoring the trans isomer) and the
reaction time is 2 days. Finally and impressively, both
diastereomers are obtained in 99% ee after cyclization at −30
°C in toluene with {S-2}Zr(NMe2)2. However, at that
temperature, the diastereoselectivity decreased to 1:1.3 (with
the trans isomer slightly favored).
In the cyclization of 4-aminomethyl-4-methylhepta-1,6-diene
(20a), the fastest cyclization rate is obtained with {S-
2}Zr(NMe2)2 and the greatest diastereoselectivity is obtained
with {1}Zr(NMe2)2. However, the diastereomeric ratio
obtained with substrate 20b is generally lower than with 19b.
Cyclization of the amino 1,2-disubstituted alkene 21a to 21b
is mediated by {1}Zr(NMe2)2 and {S-2}Zr(NMe2)2 (89% ee).
However, {S-2}Hf(NMe2), {S-2}Ti(NMe2)2, and {S-3}Zr-
(NMe2)2 are not effective at room or elevated temperatures.
Throughout, we have seen that the turnover rate diminishes
with decreasing steric bulk in the 3,3-position of the
aminopentenes, following the often-invoked gem-disubstitution
effect.3,51 As noted above, the parent aminopentene is not
efficiently cyclized by any of these chiral oxazolinylborate
catalysts. Accordingly, the rate of cyclization of 2-phenylpent-4-
en-1-amine 22a is decreased significantly compare to the
diphenylaminopentene 8a. Furthermore, the optical purity
decreases substantially for both diastereomers.
The cyclization of an aminoheptene substrate is catalyzed by
{1}Zr(NMe2)2 to give a racemic azapane and by {S-
2}Zr(NMe2)2 to provide the product in highly optically
enriched form (Table 4, entries 8 and 9). However, the
hafnium and titanium complexes {S-2}Hf(NMe2)2 and {S-
2}Ti(NMe2)2, as well as the tert-butyl-oxazolinylborate
zirconium complex {S-3}Zr(NMe2)2, are not effective even at
elevated temperatures.
The high enantioselectivity in aminopentene cyclizations
motivated further study of five-membered nitrogen-containing
ring formations. The catalysts obtained from {S-2}Zr(NMe2)2
and {S-2}Hf(NMe2)2 cyclize ortho-allylaniline (17a) at room
temperature to generate optically active 2-methylindoline
(17b) with 95 and 94% ee, respectively (Table 5, entries 2
and 3). 2-Substituted indolines are important subunits in
cannabinoid receptor agonists, as well as in drug candidates
4. Spectroscopic and Kinetics Features of Catalytic
Reactions. The olefin hydroamination reactions catalyzed by
{S-2}Zr(NMe2)2 are unique among group 4 systems in terms of
7243
dx.doi.org/10.1021/ja4000189 | J. Am. Chem. Soc. 2013, 135, 7235−7250

Journal of the American Chemical Society
Article
the mild conditions of conversion (room temperature to −30
°C), the broadly high enantioselectivity, and the relative
insensitivity of the enantioselectivity to solvent choice. We have
collected additional data to characterize some of the kinetic and
stereochemical features of our system to rule out possible
catalytic pathways and possibly identify the features that
provide high rates and high enantioselectivity. We have also
investigated effects of ancillary ligand modifications on
cyclization rates that, along with the kinetic data, provide a
mechanistic hypothesis that best explains the currently available
data.
4.1. Reaction Time Course and Its Kinetic Features.
1
HNMe2 is observed by H NMR spectroscopy upon addition
of the aminoalkenes to the precatalyst. The resting state of the
catalyst is not readily assigned from 1H NMR spectra of
micromolar reactions performed in deuterated solvents.
However, the active species is formed at room temperature
1
within 1 min, based on the appearance of product in the H
NMR spectrum of the reaction mixture.
A 1H NMR spectrum of the reaction of 1 equiv of C-(1-allyl-
cyclohexyl)methylamine (12a) and {S-2}Zr(NMe2)2 at 230 K
in toluene-d8 contained resonances assigned to HNMe2 (2.20
ppm) and new olefin moieties (5.81 and 5.10 ppm) associated
with transformation of 12a but not its cyclization. In addition,
resonances assigned to the ancillary ligand and ZrNMe2 were
detected. Peaks from 12a and the {S-2}Zr(NMe2)2 starting
material are not visible in this mixture. For comparison, the 1H
NMR spectrum of 12a at 230 K contained olefinic resonances
at 5.69 and 4.99 ppm. Aminopentene 12a cyclizes upon
warming to room temperature. The final spectrum after 12 h
contained signals from HNMe2 and a mixture of resonances
tentatively assigned as the two diastereomers of {S-2}Zr(12b-
H)(NMe2).
Figure 4. Plots of ln[12a] versus time for the {S-2}Zr(NMe2)2-
catalyzed conversion of 12a into pyrrolidine 12b. Three experiments
are illustrated, in which [{S-2}Zr(NMe2)2] = 9.7 mM and 0, 4, and 6
equiv of the product 12b is added prior to conversion. Linear least-
squares best fits for ∼2.5 half-lives corresponds to conversion to
substrate/Zr molar ratio = 2.5:1 for the three data sets.
Schafer pointed out that the curvature might result from a
change in rate law rather than product blocking catalytic sites.31
We therefore pursued this point by measuring rates and
determining the rate laws in the presence of product.
Cyclization rates of 12a catalyzed by {S-2}Zr(NMe2)2 were
measured with 4 and 6 equiv of product 12b (using 90% ee
material) relative to catalyst, and pseudo-first-order rate
constants (0 equiv of 12b, kobs = 1.08 × 10−3 s−1; 4 equiv of
12b, kobs = 0.32 × 10−3 s−1; 6 equiv 12b, kobs = 0.22 × 10−3 s−1)
were smaller in the presence of 12b (measured with equivalent
catalyst concentration {S-2}Zr(NMe2)2). Thus, kinetic experi-
ments indicate that addition of product results in slower
conversion (without a change in rate law) consistent with
product inhibition.
Interestingly, even in the presence of 6 equiv of 12b, plots of
ln[12a] versus time are linear only up to ∼75% conversion. In
general, the ln[12a] versus time plots deviate from first-order
substrate dependence at a ratio of [substrate]/[catalyst] ∼ 2.5.
This apparent effect may be related to the requirement that two
molecules of 12a are needed for catalytic turnover (see the
initial rates experiments below). Thus, these experiments
suggest that the curvature in first-order plots results from a
change in mechanism at low substrate concentration in addition
to the effects of high product concentration. These effects
should be minimized under conditions of high [12a].
A 1H NMR spectrum of the reaction of {S-2}Zr(NMe2)2 and
2 equiv of 12a, mixed, maintained, and acquired at 230 K,
1
contained signals assigned to HNMe2. The H NMR signals
previously assigned to ZrNMe2 and 12a were absent from the
1
spectrum. However, the H NMR spectrum of the reaction
mixture was complicated by broad overlapping resonances in
the phenyl and cyclopentadienyl regions, and structural
assignment was not possible. Similarly, broad resonances for
{S-2}Zr-species were observed in catalytic reaction mixtures at
low temperature; therefore, assignment of the structure of a
catalytic resting state is not possible at this point.
Instead, the reaction pathway was characterized using kinetic
studies. The concentrations of the “cyclohexyl” substrate 12a
1
and the cyclized spiro product 12b were monitored by H
NMR spectroscopy over the course of the catalytic conversion.
Plots of ln[12a] versus time are linear (up to 75−82%
conversion, Figure 4), and this is consistent with first-order
dependence on substrate concentration. Using the first 75% of
the reaction, a series of pseudo-first-order rate constants (kobs
)
An additional observable that can provide insight into
possible mechanism change is the enantioselectivity, which will
be sensitive to catalyst structure. Therefore, the % ee was
monitored as a function of conversion (Figure 5). Notably, this
plot shows that the % ee is constant from 35 to 70%
conversion, and the enantioselectivity begins to decrease above
75% conversion. Furthermore, the lower enantioselectivity
occurs at the same % conversion at which the rate appears to
deviate from first-order. However, the observed % ee is
invariant over experiments in which the absolute concentration
of the reaction mixture is varied. Furthermore, experiments in
which 12b product is added prior to catalysis provide the total
are obtained for a range of initial concentrations of {S-
2}Zr(NMe2)2. A linear relationship between kobs and [{S-
2}Zr(NMe2)2] provides the empirical rate law: −d[12a]/dt =
k′obs[12a]1[{S-2}Zr(NMe2)2]1 (k′obs = 0.128 0.007 M−1 s−1;
21 °C).52
However, after two half-lives, the reaction rate decreases to a
greater degree than expected for a first-order reaction. Similar
observations have been reported in other early metal-catalyzed
amino-olefin cyclizations (although typically these showed
zero-order substrate concentration dependence), and the
nonlinearity was attributed to product inhibition.53,54 Recently,
7244
dx.doi.org/10.1021/ja4000189 | J. Am. Chem. Soc. 2013, 135, 7235−7250

Journal of the American Chemical Society
Article
Figure 5. Plot showing the correlation of enantiomeric excess of
pyrrolidine product 12b with catalytic conversion of substrate 12a.
product with identical optical purity (i.e., the product is not
influencing the enantioselectivity).
Figure 6. Plots of initial rates of cyclization −d[12a]/dt versus
[12a]ave for cyclization of 12a at 273, 294, 314, and 322 K. The curves
represent nonlinear least-squares regression analysis of the data to the
equation −d[12a]/dt = k2[Zr][12a]/{K′ + [12a] + KSI[12a]2}.
Mechanistic studies of early transition-metal, lanthanide-, and
actinide-catalyzed hydroamination/cyclization have probed the
kinetic features (isotope effects, temperature dependence)
based on some percentage of catalytic conversion due to
apparent product inhibition. Based on those precedents, the
kinetic analysis here is performed using ∼2.5 half-lives of data.
Conversion of C-(1-allylcyclohexyl)methylamine-ND2 (12a-
d2), catalyzed by {S-2}Zr(NMe2)2, is much slower than
ln[substrate] versus time plots, the reaction rate drops
precipitously at low substrate concentrations.
The nonlinear rate dependence on concentration indicates
that the rates of the elementary reaction steps are inequivalently
influenced by substrate concentration; the result is that
changing substrate concentration modifies the contribution of
the elementary steps to the rate of catalysis. Nonlinear least-
squares regression analysis of the data in Figure 6 provides
good correlation with eq 10, which includes the rate constant k2
for the irreversible step, the catalyst formation constant K′
({k−1 + k2}/k1), and the substrate inhibition constant KSI. The
form of the rate equation resembles the Michaelis−Menten
equation, with the addition of a reversible substrate inhibition
step (eq 11).55 Additionally, the substrate concentration in eq
11 is modified by the approximated requirement that 1 equiv of
substrate 12a is needed to convert {S-2}Zr(NMe2)2 precatalyst
into an active form of the catalyst, giving the terms {[12a]-[{S-
2}Zr(NMe2)2]} in eq 10.
(D)
conversion 12a itself. The second-order rate constant k′obs
(0.039 0.002 M−1 s−1) and the large k′obs(H)/k′obs value
(D)
(3.3 0.2) is consistent with a primary isotope effect (i.e., a
N−H or N−D bond is broken in the transition state of the
turnover-limiting step). In addition, second-order rate con-
stants (k′obs) for cyclization of 12a were measured at
temperatures from 266 to 314 K. The plot of ln(k′obs/T)
versus 1/T from these data provides the values of activation
parameters: ΔH⧧ = 11.0(6) kcal·mol−1 and ΔS⧧ = −24(2)
cal·mol−1 K−1.
4.2. Initial Rates. The potential simplification of kinetic
studies at high [12a] and low [12b] motivated investigations of
the instantaneous rate dependence on substrate concentration
for the first 8−20% of the catalytic conversion (corresponding
to ∼2 catalytic turnovers). At low concentrations (0.0098 to
0.07 M), the average initial substrate conversion rates increase
linearly as the substrate concentration increases; however, at
higher substrate concentrations (70 mM), the initial rates reach
a maximum. A slight decrease of the initial rate is observed at
[12a] > 70 mM. Plots of initial rates versus substrate
concentration thus show saturation in substrate concentration
(Figure 6), and experiments that probe initial rate dependence
on substrate concentration performed at 273, 314, and 322 K,
as well as using N-deuterated substrate 12a-d2, provide similar
curves. As can been seen in Figure 6, the initial rate plots also
contains a nonzero x-intercept that coincides with catalyst
concentration. This nonzero intercept suggests that 1 equiv of
substrate is required before one catalytic cycle can be
completed. That statement is an approximation, in fact, because
the low-temperature experiments with 1:1 substrate to catalyst
ratio indicate that cyclization can occur at that low ratio
(presumably resulting from formation of a mixture that
contains {S-2}Zr(12a-H)2). However, as suggested by the
−d[12a]
=
dt
(k2[{S‐2}Zr(NMe2)2]{[12a] − [{S‐2}Zr(NMe2)2]})
(K′ + {[12a] − [{S‐2}Zr(NMe2)2]}
+ KSI{[12a] − [{S‐2}Zr(NMe2)2]}2)
(10)
−d[substrate]
dt
(k2[catalyst][substrate])
+ [substrate] + KSI[substrate]2
=
k−1 + k2
k1
(
)
(11)
The three parameters obtained from the curve fit are k2 (1.7
0.3 × 10−2 s−1), K′ (2.8 0.7 × 10−2 M), and KSI (5 2 M−1).
The rate law in eq 10 corresponds to the reaction mechanism
shown in Scheme 4, which shows reversible substrate and
catalyst association followed by an irreversible step. This rate
7245
dx.doi.org/10.1021/ja4000189 | J. Am. Chem. Soc. 2013, 135, 7235−7250

Journal of the American Chemical Society
Article
Scheme 4. Hydroamination Mechanism Based on
Interpretation of [12a] versus Initial Rate Plots
Table 6. Effect of N-d2 Substitution on % ee in
Enantioselective Hydroamination
a
a
b
Reaction conditions: 23 °C, C6D6, >95% yield. At 0 °C in toluene or
c
THF. At −30 °C in THF.
with higher optical purities compared to the corresponding
proteo-heterocycles. Thus, this stereochemical isotope effect is
systemic, at least within this cyclopentadienylbis(oxazolinyl)-
borato group 4 family of catalysts.
DISCUSSION
■
Catalyst Structure. Cyclopentadienylbis(oxazolinyl)-
borato-supported group 4 diamide compounds are exception-
ally active for cyclization of aminopentenes. Within the ligand
class, variation of 4- and 5-substitution on the oxazoline affects
complex stability, reaction rate in catalytic hydroaminations,
and enantioselectivity. Achiral {1}Zr(NMe2)2 is more active for
cyclization of the aminohexenes and the unsubstituted
aminopentene, while {S-2}Zr(NMe2)2 is much more active
for 3,3-disubstituted aminopentenes.
law and the observations given above provide a rudimentary
reaction pathway for which some rate constants are known and
the speciation of some intermediates can be surmised. The
(D)
isotope effect for conversion k′obs(H)/k′obs could thus result
from an equilibrium isotope effect K′(H)/K′(D), a kinetic isotope
effect k2(H)/k2(D), or a combination of both.
Room temperature kinetic investigations provide an
equilibrium isotope effect (for the formation constant K′) of
1.9 0.5. The kinetic isotope effect that employs the turnover-
limiting rate constant k2 is 2.3 0.4.
These effects warrant discussion of the coordination
properties of the oxazoline groups. X-ray crystal structures of
{1}Zr(NMe2)2 and {1}Zr(NMe2)2THF show that a single
In order to evaluate the transition state features of the
irreversible step, experiments that probe the effect of
temperature on initial rate versus [12a] were pursued. From
the nonlinear least-squares best fits, k2 values were determined
for catalysis at temperatures of 273, 294, 314, and 322 K. Using
the Eyring equation, the plot of ln(k2/T) versus 1/T provides
ΔH⧧ = 6.7(2) kcal·mol−1 and ΔS⧧ = −43(7) cal·mol−1 K−1.
Notably, the “saturation kinetics” show a highly organized
transition state and low activation enthalpy. The values from
the initial rate analysis are not equal to the values obtained from
conversion kinetics described above, and this reflects k′obs as a
combination of more than one elementary step.
Cyclization reactions of N-deutero-aminoalkenes give 2-
methylpyrrolidines and a 2-methylpiperidine with higher
enantiomeric excesses than the corresponding proteo-amino-
alkene. As shown in Table 6, the % ee values for deutero-
heterocycles are higher compared to the corresponding proteo-
heterocycles with the {S-2}Zr(NMe2)2. Using this method, the
% ee can be enhanced from 90 to 97% under otherwise
identical reaction conditions! Interestingly, even the piperidine
formations that proceed with lower enantioselectivity show the
same trend. Similarly, the {S-2}Hf(NMe2)2 and {S-3}Zr-
(NMe2)2 compounds provide deutero-heterocyclic products
1
oxazoline is coordinated to zirconium. Low-temperature H
NMR spectroscopy indicated fluxionality in {1}Zr-
(NMe2)2THF, and IR studies showed two oxazoline bands.
In contrast, the 1H NMR spectrum of THF-free {1}Zr(NMe2)2
is invariant over a large temperature range, and the IR spectrum
of amorphous material contained one oxazoline stretch. The IR
spectrum of crystalline material, however, was consistent with
the X-ray analysis. The coordination of {1}, {S-2}, and {S-3}
1
under catalytic conditions could not be discerned from H
NMR spectra of reaction mixtures, even at low temperature.
Therefore, a mono(oxazolinyl)diphenylborate ligand H[S-7]
and its zirconium complex {S-7}Zr(NMe2)2 were prepared (eq
8). The zirconium diamide is isolable and fully characterized.
Mixtures of {S-7}Zr(NMe2)2 and aminopentenes 8a or 12a
contained resonances associated with HNMe2, the {S-7} ligand
bonded to zirconium, and aminopentene. The expected
pyrrolidine product is not detected. Even after 4 days at
room temperature or heating at 140 °C for 2 days, the 2-
1
methylpyrrolidine is not detected in the H NMR spectra. We
conclude that two oxazoline groups are required for catalysis to
occur, although the mechanistic role of the two oxazolines is
not defined by this experiment. However, given the observed
7246
dx.doi.org/10.1021/ja4000189 | J. Am. Chem. Soc. 2013, 135, 7235−7250

Journal of the American Chemical Society
Article
NMe2 substitution by 12a without formation of 12b product, it
is likely that both oxazolines are involved in the C−N/C−H
bond formation step.
the enantioselectivity might be expected to be affected by the
concentration of the substrate. While the % ee for the
pyrrolidine product decreases at high conversion, the overall
enantioselectivity is identical in experiments that systematically
vary the overall reaction mixture concentration (as long as the
comparison involves equivalent conversions). All of our kinetic
evidence and chemical observations point to a conversion in
which a zirconium center, an oxazolinylborate ancillary ligand, a
“cyclizable” amidoalkene ligand, and a primary amide ligand are
present in the turnover-limiting step. The pathway that best fits
the data discussed above is described by path C of Scheme 1.
The nature of the proposed transition state in the turnover-
limiting step of path C, then, is probed by activation parameters
associated with the initial rate measurements and isotope effects
on rate and stereochemistry. In particular, the ΔS⧧ for the
turnover-limiting step is large and negative, indicating a highly
ordered transition state; the activation enthalpy contributes
only a small component of the overall barrier. Interestingly, the
activation parameter values from conversion kinetics show
more significant enthalpic contribution and a less significant
entropic contribution to the overall barrier. The distinction
between initial rate kinetics and conversion kinetics is
important because the reversible step in the catalytic cycle is
included in the k′obs rate constant from the empirical rate law
for conversion. Temperature-dependent equilibria thus impact
and disguise the activation parameters that describe the
irreversible step in the cycle. As a result, activation parameters
associated with k′obs provide an incorrect description of the
transition state. Because the initial rate measurements provide
the rate constant k2 for the irreversible step (or at least the
combination of elementary steps that summarize to k2), the
association of substrate and catalyst is separated from the
irreversible step. That is, the highly negative ΔS⧧ in the initial
rate kinetics describes a highly organized unimolecular
transformation.
The cyclopentadienylborate ligand occupies two valent sites
of tetravalent zirconium, leaving two active amides. The
reactivity of {S-2}Zr(NMe2)Cl in aminopentene cyclizations
was therefore explored to identify the importance of two active
sites. In fact, aminopentenes are unchanged in the presence of
catalytic amounts of {S-2}Zr(NMe2)Cl. Although {S-2}Zr-
(NMe2)Cl and 8a react to give HNMe2, pyrrolidine is not
observed in the reaction mixture. The {S-2}-related resonances
are broad, and the zirconium-containing product is not
characterized. The lack of activity for this zirconium
monoamido chloride complex is remarkable in comparison to
the related constrained geometry (CG = Me2Si(C5Me4)Nt-Bu)
system,25 in which the mixed {CG}Zr(NMe2)Cl is more
reactive than the corresponding diamide complex for
cyclizations.
Characterization of the Reaction Pathway. Mixtures of
secondary N-methylaminoalkene 8a-Me and {S-2}Zr(NMe2)2
results in substitution of the dimethylamide groups. Thus,
HNMe2 and a new species assigned as diamido {S-2}Zr(8a-
Me)2 were observed by in situ 1H NMR spectroscopic
experiments. However, conversion to the N-Me-pyrrolidine
was not observed at room temperature or upon heating to
elevated temperatures. Interestingly, 25% conversion of NMe-
8a to NMe-8b (ca. 2.5 catalyst turnovers) occurs upon addition
of a catalytic amount of amylamine or propylamine (1−2 equiv
vs catalyst). Additional quantities of amylamine and/or elevated
temperatures provide equivalent catalytic conversion (25%).
These results suggest that, under these conditions, the
zirconium center is bonded to an amidoalkene ligand and a
second amido that contains an NH group; both ligands are
needed for C−N bond formation to occur.
Kinetic experiments and isotopic effects on stereochemistry
further support this proposal. Specifically, the plots of initial
rate versus substrate concentration, extrapolated to the zero
value of initial rate, suggest that the concentration of
aminoalkene substrate must be greater than catalyst concen-
tration (although this observation is limited by the quality of
low substrate concentration/rate data). Thus, the slower
cyclization rates observed at high conversion are in part
attributed to low concentrations of aminoalkene substrate and
in part to inhibition by the secondary amine product.
The increased enantioselectivity obtained with N-deuterated
substrates demonstrates that NH functionality is intimately
involved in the stereochemistry-determining step. This stereo-
chemical effect, along with the kinetic isotope effect on the
turnover-limiting step obtained from initial rate kinetics,
suggests that the elementary step described by k2 is also the
step in which the absolute configuration is established. These
observations disfavor mechanisms involving olefin insertion
into a zirconium−nitrogen bond (path A, Scheme 1) and [2π +
2π] cycloaddition of a zirconium imido (path B, Scheme 1).
Here, we should note that reversible insertion (or 2π + 2π
addition) followed by irreversible protonation, suggested for
magnesium, yttrium, and zirconium catalyses from computa-
tional investigations,56 is unlikely given the observations with
{S-2}Zr(NMe2)Cl as a precatalyst precursor and the need for a
NH-containing amido ligand to be present in the turnover-
limiting step. For the observed saturation kinetics to involve
reversible insertion, reversible proton transfer prior to
irreversible protonation is also required. If this were the case,
At the same time, the primary kinetic isotope effect shows
N−H bond cleavage in the transition state. Additionally, the
(H)
initial rate data (k2 and k2(D)) and the enantiomeric excess
values for 12b and 12b-d2 provide isotope effects for the
diastereomeric pathways leading to the major and minor
(D)
enantiomers. The kinetic isotope effect (k2(H)/k2 = 2.2(5))
that provides the major enantiomer is relatively small; however,
the corresponding isotope effect for the minor pathway is
significant (7.7(1)). Thus, the increase in enantioselectivity
with 12a-d2 versus 12a is a result of the minor pathway slowing
to a greater extent than the major pathway. The larger isotope
effect may suggest a more symmetrical proton transfer for the
less favored pathway. Although specific interactions between
the amidoalkene and the ancillary ligand {S-2} in the transition
state cannot be probed directly by experiments, the relative
positions of the two reactive amide sites with respect to {S-2}
are suggested by NOESY experiments. The two sites are
inequivalent by symmetry, and NOESY experiments also
indicate that there are significantly dissimilar interactions
between the amide groups and the oxazoline ligands. One
amide is situated between two coordinated oxazolines, whereas
the second amide group is located in the open space between
one oxazoline and the cyclopentadienyl group. We speculate
that cyclization of one of the amides is favored and leads to the
R product, while cyclization from the other amide site provides
the minor 2S-2-methylpyrrolidine (Scheme 5).
This idea is further supported by the effect of deuteration on
the enantioselectivity. In particular, the isotope effect indicates
7247
dx.doi.org/10.1021/ja4000189 | J. Am. Chem. Soc. 2013, 135, 7235−7250

Journal of the American Chemical Society
Article
catalyst−product interaction gives poorer enantioselectivity.
There is, however, a product inhibition effect because addition
of excess pyrrolidine (90% ee) and substrate to the catalytic
reaction mixture gives a slower catalytic reaction.
Scheme 5. Proposed Stereochemical Model
Therefore, our current hypothesis is that cyclization at
aminoalkene/catalyst ratios >2:1 follows the proposed highly
selective pathway, whereas under conditions with less substrate,
the pathway is less enantioselective. In principle, these
alternative pathways could involve other elementary steps,
including but not limited to those suggested in Scheme 1. An
alternative, however, might involve a coordinated amine acting
as the proton transfer agent through a similar six-centered
transition state. Further support for competing pathways in
hydroaminations will require insight into the evolution of the
zirconium speciation during catalytic conversions. Currently,
experimental NMR spectroscopic studies on this catalytic
system have been limited by fluxionality under reaction
conditions, and we are currently exploring new systems to
obtain additional data. However, this idea could explain the
substantial effects of conditions, substrate identity, and catalyst
structure on enantioselectivity in d0 and fnd0 metal-mediated
hydroaminations that have caused difficulties in the hunt for
broadly applicable enantioselective catalysts.
that the two diastereomeric transition states are dissimilar. Two
sterically inequivalent sites provided by the C1-symmetric
catalyst provide a reasonable mechanistic rationalization for
generating sufficiently inequivalent shapes of the transition
state.
A second consideration is the coordination geometry of the
{S-2} ligand during the catalytic cycle. The fluxionality of the
ligands is enhanced under conditions of excess THF;
furthermore, the {S-2}Zr system is highly fluxional under
catalytic conditions. An achiral model species, in which one
oxazoline is dissociated and the THF is coordinated, was
characterized by an X-ray crystallographic analysis. Despite
these data, a few points favor a cyclization requirement of both
oxazolines interacting with Zr. First, the mono-oxazoline
compound {S-7}Zr(NMe2)2 reacts with substrates to give
HNMe2 and zirconium amido compounds, but pyrrolidine is
not formed even under forcing conditions (>140 °C, 2 days).
Because C−N and C−H bond formation (cyclization) is
turnover-limiting and occurs from the diamide species, a
reasonable supposition is that the second oxazoline in the {S-2}
Zr species is required for cyclization. Second, the enantiose-
lectivity of the catalytic species is unaffected by changes in
concentration of the substrate (until the final stages of the
reaction, as discussed in the next paragraph and illustrated in
Figure 5). The rate, however, is diminished at high substrate
concentrations because of a substrate inhibition, suggesting that
the zirconium species containing three substrate molecules
(and presumably a dissociated oxazoline) is not active or
considerably less active.
A third important consideration is the apparent change in
mechanism at the late stages of the catalytic conversion. The
observations of (slightly) lower % ee and slower than expected
−d[12a]/dt occur concurrently at greater than 75% conversion.
These two observations may be correlated; the implication of
correlated rate law and stereoselectivity deviations is that
another mechanism, possibly involving an alternative speciation
of the {S-2}Zr-based catalyst, becomes more important at the
substrate/catalyst ratio of ∼2:1.
This effect is independent of “product inhibition” because
the same rate law for the first ∼75% conversion was measured
with and without excess pyrrolidine. Furthermore, the final %
ee values are identical in catalytic experiments with and in the
absence of excess 2-methylpyrrolidine. Note that the catalytic
conditions (that give 90% ee) and 2-methylpyrrolidine additive
(90% ee) were matched so that the optical purity of the final
product (which is a mixture of pyrrolidine produced in the
catalysis and present as the additive) would be 90% unless the
CONCLUSION
■
Cyclopentadienylbis(oxazolinyl)borate-supported zirconium
and hafnium diamide compounds provide highly active and
effective, enantioselective catalysts for hydroamination/cycliza-
tion of amino-olefins. These catalysts provide optically enriched
pyrrolidines at synthetically useful levels, even in the presence
of functional groups; furthermore, highly enantioenriched
indoline and azepanes are accessible with these catalysts. Our
survey of several aminoalkenes highlights some of the
possibilities, as well as some of the limitations, associated
with {S-2}Zr(NMe2)2 in hydroamination catalysis. However,
minor modifications to the ancillary ligand significantly affect
precatalyst stability, catalytic conversion rates, and stereo-
selectivity. These effects suggest that ligand modifications,
centered around the modular oxazolinylborate platform, can
further advance catalytic activity and selectivity.
ASSOCIATED CONTENT
* Supporting Information
■
S
Synthetic, catalytic, and kinetics procedures, error analysis,
tables with detailed yield and % ee data for all substrates, NMR
and HPLC conditions and data for evaluating optical purity,
configuration assignments. This material is available free of
charge via the Internet at http://pubs.acs.org.
AUTHOR INFORMATION
Corresponding Author
sadow@iastate.edu
■
Notes
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
■
Prof. M. Jeffries-EL is thanked for providing access to her
HPLC, and Prof. A. Bakac is gratefully thanked for valuable
suggestions regarding kinetics experiments. This research was
supported by the U.S. Department of Energy, Office of Basic
Energy Sciences, Division of Chemical Sciences, Geosciences,
and Biosciences through the Ames Laboratory (Contract No.
7248
dx.doi.org/10.1021/ja4000189 | J. Am. Chem. Soc. 2013, 135, 7235−7250

Journal of the American Chemical Society
Article
(8) (a) Kim, J. Y.; Livinghouse, T. Org. Lett. 2005, 7, 1737−1739.
(b) Kim, H.; Kim, Y. K.; Shim, J. H.; Kim, M.; Han, M.; Livinghouse,
T.; Lee, P. H. Adv. Synth. Catal. 2006, 348, 2609−2618.
(9) (a) Meyer, N.; Zulys, A.; Roesky, P. W. Organometallics 2006, 25,
4179−4182. (b) Benndorf, P.; Jenter, J.; Zielke, L.; Roesky, P. W.
Chem. Commun. 2011, 47, 2574−2576.
(10) (a) Xiang, L.; Wang, Q.; Song, H.; Zi, G. Organometallics 2007,
26, 5323−5329. (b) Wang, Q.; Xiang, L.; Song, H.; Zi, G. Inorg. Chem.
2008, 47, 4319−4328. (c) Zi, G.; Xiang, L.; Song, H. Organometallics
2008, 27, 1242−1246.
(11) Manna, K.; Kruse, M. L.; Sadow, A. D. ACS Catal. 2011, 1,
1637−1642.
(12) (a) Martinez, P. H.; Hultzsch, K. C.; Hampel, F. Chem.
Commun. 2006, 2221−2223. (b) Ogata, T.; Ujihara, A.; Tsuchida, S.;
Shimizu, T.; Kaneshige, A.; Tomioka, K. Tetrahedron Lett. 2007, 48,
6648−6650. (c) Deschamp, J.; Collin, J.; Hannedouche, J.; Schulz, E.
Eur. J. Org. Chem. 2011, 3329−3338.
(13) (a) Buch, F.; Harder, S. Z. Naturforsch. 2008, 63b, 169−177.
(b) Horrillo-Martínez, P.; Hultzsch, K. C. Tetrahedron Lett. 2009, 50,
2054−2056. (c) Neal, S. R.; Ellern, A.; Sadow, A. D. J. Organomet.
Chem. 2011, 696, 228−234. (d) Wixey, J. S.; Ward, B. D. Chem.
Commun. 2011, 47, 5449−5451. (e) Wixey, J. S.; Ward, B. D. Dalton
Trans. 2011, 40, 7693−7696. (f) Zhang, X.; Emge, T. J.; Hultzsch, K.
C. Angew. Chem., Int. Ed. 2012, 51, 394−398.
DE-AC02-07CH11358). NMR measurements were performed
on instruments supported by NSF (CRIF-0946687 and MRI-
1040098). This research was also supported by an allocation of
advanced computing resources provided by the National
Science Foundation (A.D.S. and T.L.W.). The computations
were performed on Kraken at the National Institute for
Computational Sciences.
REFERENCES
■
(1) (a) Nugent, T. C. Chiral Amine Synthesis: Methods, Developments
and Applications; Wiley-VCH: Weinheim, Germany, 2010. (b) Ricci, A.
Modern Amination Methods; Wiley-VCH: Weinheim, Germany, 2000.
(c) King, A. G.; Meinwald, J. Chem. Rev. 1996, 96, 1105−1122.
(2) (a) Togni, A.; Grutzmacher, H. Catalytic Heterofunctionalization:
̈
From Hydroanimation to Hydrozirconation; Wiley-VCH: Weinheim,
Germany, 2001. (b) Hong, S.; Marks, T. J. Acc. Chem. Res. 2004, 37,
673−686. (c) Hultzsch, K. C. Adv. Synth. Catal. 2005, 347, 367−391.
(d) Muller, T. E.; Hultzsch, K. C.; Yus, M.; Foubelo, F.; Tada, M.
̈
Chem. Rev. 2008, 108, 3795−3892.
(3) (a) Gagne, M. R.; Brard, L.; Conticello, V. P.; Giardello, M. A.;
Stern, C. L.; Marks, T. J. Organometallics 1992, 11, 2003−2005.
(b) Giardello, M. A.; Conticello, V. P.; Brard, L.; Gagne, M. R.; Marks,
T. J. J. Am. Chem. Soc. 1994, 116, 10241−10254. (c) Douglass, M. R.;
Ogasawara, M.; Hong, S.; Metz, M. V.; Marks, T. J. Organometallics
2002, 21, 283−292. (d) Hong, S.; Tian, S.; Metz, M. V.; Marks, T. J. J.
Am. Chem. Soc. 2003, 125, 14768−14783. (e) Ryu, J.-S.; Marks, T. J.;
McDonald, F. E. J. Org. Chem. 2004, 69, 1038−1052. (f) Yu, X.;
Marks, T. J. Organometallics 2007, 26, 365−376.
(4) (a) O’Shaughnessy, P. N.; Scott, P. Tetrahedron: Asymmetry 2003,
14, 1979−1983. (b) O’Shaughnessy, P. N.; Knight, P. D.; Morton, C.;
Gillespie, K. M.; Scott, P. Chem. Commun. 2003, 1770−1771.
(c) O’Shaughnessy, P. N.; Gillespie, K. M.; Knight, P. D.; Munslow,
I. J.; Scott, P. Dalton Trans. 2004, 2251−2256.
(5) (a) Collin, J.; Daran, J.-C.; Schulz, E.; Trifonov, A. Chem.
Commun. 2003, 3048−3049. (b) Collin, J.; Daran, J.-C.; Jacquet, O.;
Schulz, E.; Trifonov, A. Chem.Eur. J. 2005, 11, 3455−3462.
(c) Riegert, D.; Collin, J.; Meddour, A.; Schulz, E.; Trifonov, A. J.
Org. Chem. 2006, 71, 2514−2517. (d) Riegert, D.; Collin, J.; Daran, J.-
C.; Fillebeen, T.; Schulz, E.; Lyubov, D.; Fukin, G.; Trifonov, A. Eur. J.
Inorg. Chem. 2007, 1159−1168. (e) Aillaud, I.; Lyubov, D.; Collin, J.;
Guillot, R.; Hannedouche, J.; Schulz, E.; Trifonov, A. Organometallics
2008, 27, 5929−5936. (f) Hannedouche, J.; Aillaud, I.; Collin, J.;
Schulz, E.; Trifonov, A. Chem. Commun. 2008, 3552−3554. (g) Aillaud,
I.; Wright, K.; Collin, J.; Schulz, E.; Mazaleyrat, J.-P. Tetrahedron:
Asymmetry 2008, 19, 82−92. (h) Aillaud, I.; Collin, J.; Duhayon, C.;
Guillot, R.; Lyubov, D.; Schulz, E.; Trifonov, A. Chem.Eur. J. 2008,
14, 2189−2200. (i) Chapurina, Y.; Hannedouche, J.; Collin, J.; Guillot,
R.; Schulz, E.; Trifonov, A. Chem. Commun. 2010, 46, 6918−6920.
(j) Aillaud, I.; Collin, J.; Hannedouche, J.; Schulz, E.; Trifonov, A.
Tetrahedron Lett. 2010, 51, 4742−4745. (k) Chapurina, Y.; Ibrahim,
H.; Guillot, R.; Kolodziej, E.; Collin, J.; Trifonov, A.; Schulz, E.;
Hannedouche, J. J. Org. Chem. 2011, 76, 10163−10172. (l) Queffelec,
(14) Knight, P. D.; Munslow, I.; O’Shaughnessy, P. N.; Scott, P.
Chem. Commun. 2004, 894−895.
(15) Watson, D. A.; Chiu, M.; Bergman, R. G. Organometallics 2006,
25, 4731−4733.
(16) (a) Wood, M. C.; Leitch, D. C.; Yeung, C. S.; Kozak, J. A.;
Schafer, L. L. Angew. Chem., Int. Ed. 2007, 46, 354−358. (b) Ayinla, R.
O.; Gibson, T.; Schafer, L. L. J. Organomet. Chem. 2011, 696, 50−60.
(17) Gott, A. L.; Clarke, A. J.; Clarkson, G. J.; Scott, P.
Organometallics 2007, 26, 1729−1737.
(18) Zi, G.; Zhang, F.; Xiang, L.; Chen, Y.; Fang, W.; Song, H. Dalton
Trans. 2010, 39, 4048−4061.
(19) Reznichenko, A. L.; Hultzsch, K. C. Organometallics 2010, 29,
24−27.
(20) (a) Reznichenko, A. L.; Emge, T. J.; Audorsch, S.; Klauber, E.
̈
G.; Hultzsch, K. C.; Schmidt, B. Organometallics 2011, 30, 921−924.
(b) Zhang, F.; Song, H.; Zi, G. Dalton Trans. 2011, 40, 1547−1566.
(21) (a) Gott, A. L.; Clarke, A. J.; Clarkson, G. J.; Scott, P. Chem.
Commun. 2008, 1422−1424. (b) Gott, A. L.; Clarkson, G. J.; Deeth, R.
J.; Hammond, M. L.; Morton, C.; Scott, P. Dalton Trans. 2008, 2983−
2990.
(22) (a) Shen, X.; Buchwald, S. L. Angew. Chem., Int. Ed. 2010, 49,
564−567. (b) Turnpenny, B. W.; Hyman, K. L.; Chemler, S. R.
Organometallics 2012, 31, 7819−7822.
(23) (a) Gagne, M. R.; Marks, T. J. J. Am. Chem. Soc. 1989, 111,
4108−4109. (b) Gagne, M. R.; Stern, C. L.; Marks, T. J. J. Am. Chem.
Soc. 1992, 114, 275−294.
(24) Hangaly, N. K.; Petrov, A. R.; Rufanov, K. A.; Harms, K.;
Elfferding, M.; Sundermeyer, J. Organometallics 2011, 30, 4544−4554.
(25) Stubbert, B. D.; Marks, T. J. J. Am. Chem. Soc. 2007, 129, 6149−
6167.
̀
C.; Boeda, F.; Pouilhes, A.; Meddour, A.; Kouklovsky, C.;
Hannedouche, J.; Collin, J.; Schulz, E. ChemCatChem 2011, 3, 122−
126.
(26) Dunne, J. F.; Fulton, D. B.; Ellern, A.; Sadow, A. D. J. Am. Chem.
Soc. 2010, 132, 17680−17683.
(27) (a) Crimmin, M. R.; Arrowsmith, M.; Barrett, A. G. M.; Casely,
I. J.; Hill, M. S.; Procopiou, P. A. J. Am. Chem. Soc. 2009, 131, 9670−
9685. (b) Arrowsmith, M.; Crimmin, M. R.; Barrett, A. G. M.; Hill, M.
(6) (a) Gribkov, D. V.; Hultzsch, K. C.; Hampel, F. Chem.Eur. J.
2003, 9, 4796−4810. (b) Gribkov, D. V.; Hampel, F.; Hultzsch, K. C.
Eur. J. Inorg. Chem. 2004, 4091−4101. (c) Gribkov, D. V.; Hultzsch, K.
C. Chem. Commun. 2004, 730−731. (d) Gribkov, D. V.; Hultzsch, K.
C.; Hampel, F. J. Am. Chem. Soc. 2006, 128, 3748−3759.
(e) Reznichenko, A. L.; Hampel, F.; Hultzsch, K. C. Chem.Eur. J.
2009, 15, 12819−12827. (f) Vitanova, D. V.; Hampel, F.; Hultzsch, K.
C. J. Organomet. Chem. 2007, 692, 4690−4701. (g) Reznichenko, A.
L.; Nguyen, H. N.; Hultzsch, K. C. Angew. Chem., Int. Ed. 2010, 49,
8984−8987.
S.; Kociok-Kohn, G.; Procopiou, P. A. Organometallics 2011, 30,
̈
1493−1506. (c) Brinkmann, C.; Barrett, A. G. M.; Hill, M. S.;
Procopiou, P. A. J. Am. Chem. Soc. 2011, 134, 2193−2207.
(28) Liu, B.; Roisnel, T.; Carpentier, J.-F.; Sarazin, Y. Angew. Chem.,
Int. Ed. 2012, 51, 4943−4946.
(29) (a) Pohlki, F.; Doye, S. Angew. Chem., Int. Ed. 2001, 40, 2305−
2308. (b) Muller, C.; Saak, W.; Doye, S. Eur. J. Org. Chem. 2008,
̈
2731−2739. (c) Janssen, T.; Severin, R.; Diekmann, M.; Friedemann,
M.; Haase, D.; Saak, W.; Doye, S.; Beckhaus, R. Organometallics 2010,
29, 1806−1817.
(7) Heck, R.; Schulz, E.; Collin, J.; Carpentier, J.-F. J. Mol. Catal. A
2007, 268, 163−168.
7249
dx.doi.org/10.1021/ja4000189 | J. Am. Chem. Soc. 2013, 135, 7235−7250

Journal of the American Chemical Society
Article
(30) Allan, L. E. N.; Clarkson, G. J.; Fox, D. J.; Gott, A. L.; Scott, P. J.
Am. Chem. Soc. 2010, 132, 15308−15320.
(31) (a) Leitch, D. C.; Platel, R. H.; Schafer, L. L. J. Am. Chem. Soc.
2011, 133, 15453−15463. (b) Leitch, D. C.; Payne, P. R.; Dunbar, C.
R.; Schafer, L. L. J. Am. Chem. Soc. 2009, 131, 18246−18247.
(32) Manna, K.; Ellern, A.; Sadow, A. D. Chem. Commun. 2010, 46,
339−341.
(33) Tobisch, S. Chem.Eur. J. 2011, 17, 14974−14986.
(34) Gribkov, D. V.; Hultzsch, K. C. Angew. Chem., Int. Ed. 2004, 43,
5542−5546.
(35) (a) Walsh, P. J.; Hollander, F. J.; Bergman, R. G. J. Am. Chem.
Soc. 1988, 110, 8729−8731. (b) Baranger, A. M.; Walsh, P. J.;
Bergman, R. G. J. Am. Chem. Soc. 1993, 115, 2753−2763. (c) Straub,
B. F.; Bergman, R. G. Angew. Chem., Int. Ed. 2001, 40, 4632−4635.
(36) (a) Kissounko, D. A.; Epshteyn, A.; Fettinger, J. C.; Sita, L. R.
Organometallics 2006, 25, 1076−1078. (b) Leitch, D. C.; Schafer, L. L.
Organometallics 2010, 29, 5162−5172.
(37) Majumder, S.; Odom, A. L. Organometallics 2008, 27, 1174−
1177.
(38) Leitch, D. C.; Turner, C. S.; Schafer, L. L. Angew. Chem., Int. Ed.
2010, 49, 6382−6386.
(39) Katayev, E.; Li, Y.; Odom, A. L. Chem. Commun. 2002, 838−
839.
(40) (a) Neukom, J. D.; Perch, N. S.; Wolfe, J. P. J. Am. Chem. Soc.
2010, 132, 6276−6277. (b) Neukom, J. D.; Perch, N. S.; Wolfe, J. P.
Organometallics 2011, 30, 1269−1277.
(41) (a) Tye, J. W.; Hartwig, J. F. J. Am. Chem. Soc. 2009, 131,
14703−14712. (b) Hanley, P. S.; Markovic, D.; Hartwig, J. F. J. Am.
Chem. Soc. 2010, 132, 6302−6303.
(42) Casalnuovo, A. L.; Calabrese, J. C.; Milstein, D. J. Am. Chem.
Soc. 1988, 110, 6738−6744.
(43) Manna, K.; Xu, S.; Sadow, A. D. Angew. Chem., Int. Ed. 2011, 50,
1865−1868.
(44) Dunne, J. F.; Manna, K.; Wiench, J. W.; Ellern, A.; Pruski, M.;
Sadow, A. D. Dalton Trans. 2010, 39, 641−653.
(45) Kidd, R. G. Boron-11; Academic Press: New York, 1983; Vol. 2.
(46) (a) Herberich, G. E.; Fischer, A. Organometallics 1996, 15, 58−
̂
67. (b) Roitershtein, D.; Domingos, A.; Marques, N. Organometallics
2004, 23, 3483−3487.
(47) Dunne, J. F.; Su, J.; Ellern, A.; Sadow, A. D. Organometallics
2008, 27, 2399−2401.
(48) (a) Ralevic, V. Eur. J. Pharmacol. 2003, 472, 1−21. (b) Nicolaou,
K. C.; Roecker, A. J.; Pfefferkorn, J. A.; Cao, G.-Q. J. Am. Chem. Soc.
2000, 122, 2966−2967. (c) Sun, H.; Moore, C.; Dansette, P. M.;
Kumar, S.; Halpert, J. R.; Yost, G. S. Drug Metab. Dispos. 2009, 37,
672−684. (d) Gueritte, F.; Fahy, J. In Anticancer Agents from Natural
Products; Cragg, G. M., Kingston, D. G. I., Newman, D. J., Eds.; CRC
Press: Boca Raton, FL, 2005; pp 123−135.
(49) (a) Kuwano, R.; Kashiwabara, M. Org. Lett. 2006, 8, 2653−
2655. (b) Kuwano, R.; Kashiwabara, M.; Sato, K.; Ito, T.; Kaneda, K.;
Ito, Y. Tetrahedron: Asymmetry 2006, 17, 521−535.
(50) Arp, F. O.; Fu, G. C. J. Am. Chem. Soc. 2006, 128, 14264−
14265.
(51) Jung, M. E.; Piizzi, G. Chem. Rev. 2005, 105, 1735−1766.
(52) The k′ value is calculated from a linear least-squares regression
analysis of kobs vs [{S-2}Zr(NMe2)2].
(53) For group 4 catalysts that show product inhibitions and/or
nonlinear kinetics, see refs 25 and 31.
(54) For rare earth catalysts that show product inhibition and
nonlinear kinetics, see refs 3b, 6d, and (a) Li, Y.; Marks, T. J. J. Am.
Chem. Soc. 1996, 118, 9295. (b) Hong, S.; Kawaoka, A. M.; Marks, T.
J. J. Am. Chem. Soc. 2003, 125, 15878. (c) Ryu, J. S.; Li, G. Y.; Marks,
T. J. J. Am. Chem. Soc. 2003, 125, 12584.
(55) Cornish-Bowden, A. Fundamentals of Enzyme Kinetics, 3rd ed.;
Portland Press: London, 2004.
(56) (a) Tobisch, S. Dalton Trans. 2012, 41, 9182−9191. (b) Tobisch,
S. Inorg. Chem. 2012, 51, 3786−3795.
7250
dx.doi.org/10.1021/ja4000189 | J. Am. Chem. Soc. 2013, 135, 7235−7250