Mechanism of an Organoboron-Catalyzed Domino Reaction: Kinetic and Computational Studies of Borinic Acid-Catalyzed Regioselective Chloroacylation of 2,3-Epoxy Alcohols

Article abstract of DOI:10.1021/acs.joc.6b02702

A mechanistic study of the borinic acid-catalyzed chloroacylation of 2,3-epoxy alcohols is presented. In this unusual mode of catalysis, the borinic acid activates the substrate toward sequential reactions with a nucleophile (epoxide ring-opening by chloride) and an electrophile (O-acylation of the resulting alkoxide). Reaction progress kinetic analysis of data obtained through in situ FTIR spectroscopy is consistent with a mechanism involving turnover-limiting acylation of a chlorohydrin-derived borinic ester. This proposal is further supported by investigations of the effects of aroyl chloride substitution on reaction rate. The kinetics experiments also shed light on the effects of chloride concentration on reaction rate and indicate that the catalyst is subject to inhibition by the product of the chloroacylation reaction. Computational modeling is employed to gain insight into the effects of the organoboron catalyst on the regioselectivities of the epoxide ring-opening and acylation steps. The density functional theory calculations provide a plausible pathway for selective chlorinolysis at C-3 and benzoylation at O-1, as is observed experimentally.

Full text of DOI:10.1021/acs.joc.6b02702

Article
pubs.acs.org/joc
Mechanism of an Organoboron-Catalyzed Domino Reaction: Kinetic
and Computational Studies of Borinic Acid-Catalyzed Regioselective
Chloroacylation of 2,3-Epoxy Alcohols
Graham E. Garrett, Kashif Tanveer, and Mark S. Taylor*
Department of Chemistry, University of Toronto, Toronto, Ontario M5S 3H6 Canada
S
* Supporting Information
ABSTRACT: A mechanistic study of the borinic acid-
catalyzed chloroacylation of 2,3-epoxy alcohols is presented.
In this unusual mode of catalysis, the borinic acid activates the
substrate toward sequential reactions with a nucleophile
(epoxide ring-opening by chloride) and an electrophile (O-
acylation of the resulting alkoxide). Reaction progress kinetic
analysis of data obtained through in situ FTIR spectroscopy is
consistent with a mechanism involving turnover-limiting acylation of a chlorohydrin-derived borinic ester. This proposal is
further supported by investigations of the effects of aroyl chloride substitution on reaction rate. The kinetics experiments also
shed light on the effects of chloride concentration on reaction rate and indicate that the catalyst is subject to inhibition by the
product of the chloroacylation reaction. Computational modeling is employed to gain insight into the effects of the organoboron
catalyst on the regioselectivities of the epoxide ring-opening and acylation steps. The density functional theory calculations
provide a plausible pathway for selective chlorinolysis at C-3 and benzoylation at O-1, as is observed experimentally.
group and was likely important for facilitating catalyst turnover
(the previously reported examples of boronic acid-promoted
ring-openings of epoxy alcohols have employed stoichiometric
quantities of the boron reagent).7 This transformation also
represented a novel mode of catalytic reactivity for borinic
acids, a class of boron compounds that have not been applied
extensively in catalysis.8,9
INTRODUCTION
■
Methods for selective chemical transformations of 2,3-epoxy
alcohols are needed to make efficient use of the building blocks
generated by diastereo- or enantioselective epoxidations of
allylic alcohols.1 Regioselective ring-opening of 2,3-epoxy
alcohols has been a particularly important goal in this regard.2
More than two decades ago, Sharpless’ group reported that
stoichiometric quantities of Ti(OiPr)4 could be employed to
achieve C3-selective opening with diverse nucleophiles.3
Catalytic methods for regioselective ring-opening of epoxy
alcohols have emerged more recently, with pioneering results
having been achieved by the groups of Yamamoto4 and
Iwabuchi.5 Recently, we reported that 2,3-epoxy alcohols
substrates undergo chloroacylation or chlorosulfonylation in
the presence of borinic ester 2a (Scheme 1).6 The ability to
achieve chloride ring-opening and O-functionalization (acyla-
tion or sulfonylation) in a single operation was a unique aspect
of this catalytic method. It provided a way to differentiate the
three contiguous oxygenated positions of the 2,3-epoxy alcohol
We were motivated to study the mechanism of this
transformation for a number of reasons. This type of domino
reactivity, wherein a boron catalyst apparently activates the
substrate sequentially as an electrophile and a nucleophile, had
not been described previously. The kinetics of this process and
the origins of regioselectivity in the ring-opening (C3-selective)
and O-functionalization (C1-selective) steps were of interest.
Furthermore, we have found it challenging to develop
analogous processes such as fluoroacylation, azidoacylation, or
cyanoacylation of epoxy alcohols. Mechanistic understanding
could point toward ways to overcome these limitations. Herein,
we describe a study of the mechanism of borinic acid-catalyzed
chloroacylation of 2,3-epoxy alcohols using reaction progress
kinetic analysis (RPKA)10 of data obtained using in situ infrared
(IR) spectroscopy. Computational modeling with density
functional theory (DFT) is used to gain insight into the
origins of rate acceleration and regiocontrol for the key borinic
acid-catalyzed steps.
Scheme 1. Chloroacylation and Chlorosulfonylation of 2,3-
Epoxy Alcohols Catalyzed by 2a
Received: November 9, 2016
© XXXX American Chemical Society
A
DOI: 10.1021/acs.joc.6b02702
J. Org. Chem. XXXX, XXX, XXX−XXX

The Journal of Organic Chemistry
Article
RESULTS AND DISCUSSION
Scheme 3. Conditions for Chlorobenzoylation of Epoxy
Alcohol 1 (a) as Initially Reported in Ref 6 and (b) as
Adapted Here for Acquisition of Reaction Rate Data
■
Proposed Catalytic Cycle. The proposed catalytic cycle
for chlorobenzoylation of epoxy alcohol 1 is shown in Scheme
2. Activation of the diphenylborinate precatalyst 2a via
Scheme 2. Proposed Catalytic Cycle for Chloroacylation of
Epoxy Alcohol 1 Using Catalyst 2a
acylation of the ethanolamine ligand releases 2 equiv of
chloride relative to the catalyst. The borinic acid then binds
reversibly to the 2,3-epoxy alcohol, generating complex I.
Nucleophilic attack by Cl¯ opens the oxirane ring, forming
tetracoordinate borinic ester intermediate II. Acylation of II by
benzoyl chloride (3a) generates complex III, which after
displacement by the substrate results in catalyst turnover and
release of the product (4). This proposal was consistent with
the following observations from our previous work on borinic
acid catalysis and from our initial studies of epoxy alcohol
chloroacylation: (1) the epoxy ester derived from 1 did not
undergo ring-opening under the reaction conditions, suggesting
that ring-opening precedes acylation; (2) both 2a and related
diarylborinic acids activate diols toward regioselective reactions
with a variety of electrophiles, including acyl and sulfonyl
chlorides;11 (3) N,O-bis-functionalized ethanolamine deriva-
tives have been isolated from diol activation reactions using
ethanolamine ester 2a as precatalyst;12 and (4) in the absence
of catalyst, benzoylation of chlorohydrin diol 5 took place
slowly and with low regioselectivity.
Selection of Model System and Reaction Conditions.
The chlorobenzoylation of 1 was selected for mechanistic study
due to the ease of preparation of the epoxy alcohol substrate
and the high yield and regioselectivity obtained using this
combination of reagents under the optimized conditions (90%
yield, >19:1 C3:C2, 9:1 O1:O2, Scheme 3a). The reaction
conditions were modified to facilitate the acquisition and
analysis of rate data. Acetonitrile was employed as the reaction
solvent, in place of dichloromethane, both to reduce changes in
concentration due to solvent evaporation and to permit analysis
of the C−Cl stretching region of the IR spectrum. Reactions
were kept at a constant temperature of 30 °C in an oil bath
rather than being conducted at ambient temperature. Instead of
ethanolamine ester precatalyst 2a, diphenylborinic anhydride
(2b) was used in combination with tetrabutylammonium
chloride (Bu4N+Cl−). Using 2b and Bu4N+Cl− avoided
complications that could arise from slow entry of precatalyst
2a into the catalytic cycle and permitted independent variation
of the concentrations of diphenylborinic acid and chloride.
Although 2b is a dimer in the solid state, it dissociates rapidly to
monomers in the presence of alcohols or water. When the 2b/
Bu4N+Cl− catalyst system was used, the amount of iPr2NEt
could be decreased from 1.5 equiv to 0.6 equiv relative to epoxy
alcohol (see below). Product 4 was obtained in high yield and
regioselectivity under these modified reaction conditions
(Scheme 3b).
Reaction Monitoring and Rate Determinations. The
course of a chloroacylation reaction conducted under the
1
conditions shown in Scheme 3b was monitored by H nuclear
magnetic resonance (NMR) spectroscopy (Figure 1a). Aliquots
were withdrawn at intervals ranging from 3 to 6 min, dried,
diluted into CDCl3, and analyzed. Clean and complete
conversion of the epoxy alcohol to product 4 (along with a
small amount of its 2-O-acylated regioisomer) took place over
roughly 1 h. Signals corresponding to chlorohydrin diol 5 built
up to a maximum within the first 5 min of the reaction and then
gradually decreased over the course of the experiment,
consistent with the hypothesis that 5 is an intermediate in
the pathway leading to product 4.
To facilitate reaction progress kinetic analysis, the chlor-
oacylation was monitored by in situ FTIR spectroscopy. The
appearance of signals at 717 cm−1 (the C−Cl bond stretching
frequency for product 4 and chlorohydrin diol 5) and 1278
cm−1 (the ester C−O bond stretching frequency for product 4)
was evident as the reaction progressed. The latter signal
enabled quantification of product concentration over time,
while the former was used to infer the concentration of epoxy
alcohol at each time point, based on the assumption that the
conversion of 1 is equal to the sum of the concentrations of 4
and 5. This assumption appears to be justified, given the
absence of other intermediates or byproducts as assessed by 1H
NMR spectroscopy, as discussed above. Indeed, the concen-
trations of 1 and 4 determined by FTIR spectroscopy were in
good agreement with the NMR data (Figure 1b). The
concentration versus time data from in situ FTIR spectroscopy
were fitted to an ordinary differential equation to obtain rate
versus time data (see the Experimental Section for details, and
the Supporting Information for further discussion). Graphical
rate equations of rate versus the concentration of epoxy alcohol
1 for triplicate experiments carried out under the standard
conditions showed good overlay, indicating that this protocol
provided reproducible kinetic data (Figure 1c).
B
DOI: 10.1021/acs.joc.6b02702
J. Org. Chem. XXXX, XXX, XXX−XXX

The Journal of Organic Chemistry
Article
observed upon increasing the concentrations of 3a and 2b.
Normalizing the reaction rates for the total borinic acid
concentration and the concentration of BzCl at each point
resulted in overlay of the graphical rate eqs (Figure 2a),
indicating first-order kinetics in each of these two species.
“Different excess” experiments were conducted, and the
reaction rate plotted against the concentration of BzCl (Figure
2b). The overlay of graphical rate equations is evident at the
initial stages of the reaction points toward pseudo zero-order
(saturation) kinetics in 1. Variation of the concentration of
iPr2NEt showed apparent zero-order kinetics in base (see the
Supporting Information).
Considering that the identity of the base was one of the
important parameters in the original optimization of the
chloroacylation protocol, the observation of zero-order kinetics
in base was somewhat surprising. This apparent inconsistency
was resolved when we explored the effects of base
concentration on reactions conducted under the originally
reported protocol, using precatalyst 2a rather than 2b/
Bu4N+Cl−. Under these conditions, no product formation was
observed in the absence of base, and a positive dependence of
rate on iPr2NEt concentration was observed (see the
Supporting Information). It appears that the base is needed
to neutralize the 2 equiv of HCl generated upon entry of
precatalyst 2a into the cycle, as borinic acid−alcohol binding is
favored under basic conditions.12 We also note that reactions
conducted under the standard conditions for kinetic study did
not proceed to completion in the absence of base, appearing to
“stall” with residual chlorohydrin diol remaining. The origin of
this effect is not clear at present.
As mentioned above, the use of the 2b/Bu4N+Cl−
combination enabled us to vary the amount of chloride while
maintaining a constant concentration of the organoboron
catalyst (Figure 3a). To analyze the data from experiments of
this type, reaction rates at a given epoxide concentration were
plotted against Bu4N+Cl− concentration (Figure 3b). At high
epoxide concentrations, this is equivalent to comparing initial
rates, while at epoxide concentrations ranging from 0.04 to 0.16
M, it is a comparison of steady-state rates. These plots indicate
that the rate increases with Bu4N+Cl− concentration up to
roughly 60 mM and that saturation kinetics are in effect at
higher chloride concentrations. This concentration corresponds
to a 1.4:1 ratio of Cl− to organoboron catalyst, which is lower
than the 2:1 ratio that would result from bis-acylation of the
ethanolamine ligand when using precatalyst 2a.
When rate was plotted versus the concentration of 1 for two
reactions conducted under “same excess” conditions, overlay of
the graphical rate equations was not observed (Figure 4a). This
result is consistent with product inhibition or catalyst
deactivation. Borinic acids bind relatively tightly to diols in
solution,13 but their binding to compounds having a single OH
group is less favorable. In this case, both the starting 2,3-epoxy
alcohol and the product contain only one free hydroxy group,
and so they might be expected to show similar affinities for the
organoboron catalyst. Indeed, doping the reaction with 4 under
the “same excess” conditions produced an overlay in the
graphical rate equations and provided better overlap in the
offset concentration versus time plots, indicating that the
chlorohydrin monoester product inhibits this transformation
(Figure 4b).
Figure 1. (a) Progress of the chloroacylation reaction of 1 as assessed
1
by H NMR spectroscopy. (b) Concentration versus time data for
substrate 1 (as assessed by in situ FTIR (blue ) and 1H NMR
○
●
spectroscopy (black )), product 4 (as assessed by in situ FTIR (red
1
■
○
) and H NMR spectroscopy (black )) and chlorohydrin 5 (as
1
◆
assessed by H NMR spectroscopy (black )). (c) Graphical rate
equations (rate versus concentration of 1) for three experiments
conducted under the conditions shown in Scheme 3b.
Reaction Progress Kinetic Analysis of Rate Data. The
kinetic orders in epoxy alcohol 1, (PhB)2O (2b), Bu4N+Cl−,
BzCl (3a), and iPr2NEt were determined by examining
graphical rate equations for reactions conducted at different
concentrations of these species. Increases in reaction rate were
The kinetic orders determined by the analysis described
above are summarized in Figure 5. Taken together, the data
suggest that epoxide ring-opening is rapid relative to ester bond
C
DOI: 10.1021/acs.joc.6b02702
J. Org. Chem. XXXX, XXX, XXX−XXX

The Journal of Organic Chemistry
Article
Figure 2. (a) Graphical rate equations (rate versus concentration of 1) for experiments conducted at different values of the excess (ε = [BzCl] −
[1]) and at different catalyst loadings: ε = 0.38 M, 20 mol % catalyst (black ); ε = 0.61 M, 20 mol % catalyst (red ); ε = 0.38 M, 31 mol %
catalyst (blue ). The graphs are normalized for BzCl concentration and catalyst concentration. (b) Graphical rate equations (rate versus
concentration of BzCl, 3a) for “different excess” experiments: [1]0 = 0.22 M, ε = 0.38 M (black ); [1]0 = 0.28 M, ε = 0.32 M (red ); [1]0 = 0.38
M, ε = 0.21 M (blue ).
Figure 3. (a) Graphical rate equations for reactions conducted with different loadings of Bu4N+Cl−. The Bu4N+Cl− concentrations were 24 mM
(purple ), 44 mM (light-purple ), 68 mM (blue ), 82 mM (black ) and 98 mM (red ). (b) Plots of rate versus Bu4N+Cl− loading for
■
●
epoxy alcohol concentrations of 0.02 M (black ), 0.13 M (purple triangle) and 0.20 M (teal ).
formation and that borinic ester II is the catalyst resting state.
Consistent with this idea, inspection of the IR data from the
early stages of the reaction showed an initial period of roughly
30 s in which the absorbance corresponding to C−Cl bond
formation increased without evident ester C−O bond
formation (see the Supporting Information). After this initial
“burst” of C−Cl bond formation, the C−Cl and C−O FTIR
absorption bands were observed to increase at equal rates,
presumably indicating that the reaction had reached a steady
state. Further support for the assignment of II as the catalyst
resting state was obtained by in situ monitoring of the
reaction mixtures were dried and diluted into CDCl3. For
samples treated this way, chlorohydrin 5 was observed rather
than complex II.
Temperature Dependence of Reaction Rates and
Determination of Activation Parameters. Reaction rates
were determined at temperatures ranging from 0 to 70 °C. For
this set of experiments, the time required for the reaction to
reach complete (>95%) conversion ranged from 16 h at 0 °C to
11 min at 70 °C. Plots of ln(k/T) (where k is the rate constant
for the turnover-limiting step, and T is the reaction temper-
ature) versus 1/T were generated for epoxy alcohol
concentrations within the steady-state regime. A representative
plot at [1] = 0.13 M is shown in Figure 6. The lines of best fit
were used to determine ΔH⧧ and ΔS⧧, the enthalpy and
entropy of activation, using the Eyring−Polanyi equation.
Values of ΔH⧧ = 13.2 0.5 kcal/mol and ΔS⧧ = −26 1 cal/
mol·K were obtained using the average values of the slope and
intercept of all linear fits within the steady state regime, with
1
chloroacylation reaction by H NMR spectroscopy in CD3CN
(see the Supporting Information). A new set of signals
consistent with structure II was observed to build up during
the initial stages of the reaction. These signals were distinct
from those of “free” chlorohydrin 5, which was not observed
under these conditions. It should be noted that this experiment
differs from that described in Figure 1b, in which aliquots of
D
DOI: 10.1021/acs.joc.6b02702
J. Org. Chem. XXXX, XXX, XXX−XXX

The Journal of Organic Chemistry
Article
○
Figure 4. (a) Plots of epoxy alcohol concentration versus time for “same excess” experiments: [1]0 = 0.31 M, ε = 0.40 M (black ), [1]0 = 0.22 M, ε
○
○
= 0.40 M (blue ), [1]0 = 0.22 M, ε = 0.40 M, [4]0 = 0.08 M (red ). (b) Graphical rate equations corresponding to the “same excess” experiments
shown in (a).
Figure 6. Plot of ln(k/T) vs 1/T (where k is the second-order rate
constant for the turnover-limiting step, calculated at [1] = 0.13 M, and
T is the reaction temperature) for temperatures ranging from 0 to 70
°C.
was complete in roughly 7 min (Figure 7a). Contrary to the
results discussed above for the reaction of benzoyl chloride,
Figure 5. Summary of kinetic orders determined by reaction progress
kinetic analysis.
changing the “excess” of 3c had no effect on the rate of the
reaction (Figure 7b). The reaction of 4-nitrobenzoyl chloride
thus shows zero-order kinetics in acylating agent and first-order
kinetics in epoxy alcohol (as inferred from the linear
dependence of rate on [1] in the graphical rate eq (Figure
7b, inset)). This rate law is consistent with a change in the
turnover-limiting step from acylation to epoxide ring-opening
when the more reactive electrophile is employed.
uncertainties based on the standard deviations of the results.
The magnitude of the entropy of activation is consistent with a
bimolecular mechanism and appears to be roughly in line with
existing data for reactions of acyl chlorides with alcohols (− 28
to −34 cal/mol·K).14
Substituent Effects on Epoxy Alcohol Chloroacyla-
tion. Chloroacylations of 1 were conducted using 4-
methoxybenzoyl chloride (3b) and 4-nitrobenzoyl chloride
(3c) to assess the effects of aroyl chloride substitution on
reaction rate. Consistent with acylation being the turnover-
limiting step of the reaction, the chloroacylation rates increased
in the order 3b < 3a < 3c: the reaction of the electron-rich
methoxy-substituted derivative required 5 h to reach
completion, whereas that of the nitro-substituted derivative
Chlorohydrin Diol Complex As a Reactive Intermedi-
ate. The NMR reaction monitoring experiments discussed
above (Figure S7) suggested that chlorohydrin diol complex II
was a reactive intermediate in the chloroacylation of 1.
Compound 5, a precursor to this complex, was synthesized
by the ring-opening of 1 with lithium chloride in the presence
of acetic acid and trimethyl borate7b and subjected to benzoyl
chloride in the presence of catalyst 2b. Unlike the
chloroacylation of 1, this reaction generates HCl as a
E
DOI: 10.1021/acs.joc.6b02702
J. Org. Chem. XXXX, XXX, XXX−XXX

The Journal of Organic Chemistry
Article
○
○
○
Figure 7. (a) Plots of epoxy alcohol concentration versus time for experiments using substituted aroyl chlorides 3a (black ), 3b (blue ) and 3c
○
(red ). (b) Plots of epoxy alcohol versus time for “different excess” experiments using 4-nitrobenzoyl chloride (3c): ε = 0.17 M (black ) and ε =
○
0.40 M (red ); inset: graphical rate equation for the reaction of 1 with 3c (ε = 0.17 M).
byproduct, and so 1.7 equiv of iPr2NEt was added. Reaction
progress was assessed using the appearance of the ester C−O
stretch in the FTIR spectrum to determine the concentration of
product. The rate of this reaction was roughly similar to that of
BzCl with epoxy alcohol 1 (Figure 8), although differences in
interest, since this type of information could not be inferred
from our studies of the reaction rates. The B3LYP DFT
functional and 6-31+G(d,p) basis set were employed for these
calculations. This level of theory was selected due to its low
computational overhead and ease of handling large molecules,
and to enable comparisons with previous computational studies
of 2,3-epoxy alcohol ring-opening reactions.7a,15
We began by modeling the borinic acid-catalyzed ring-
opening step (Figure 9). The lowest-energy calculated
structures of the epoxy alcohol-derived diphenylborinic ester
(of which the most stable places the −OBPh2 substituent in the
pseudoequatorial position of the half-chair) showed only single-
point binding to the boron center. In contrast, previous
computational studies pointed toward chelation of the hydroxy
and epoxide oxygens in analogous complexes of epoxy alcohols
with phenylboronic acid7 or lithium cation.15 However, an
interaction between boron and the developing alkoxide was
evident in the calculated transition state for the ring-opening
step. Attack of chloride at C-3 occurs through a boatlike
transition state, accessed via a higher-energy borinic ester
conformer in which the −OBPh2 substituent is pseudoaxial
(Figure 10). A pathway for chloride attack at C-2 was also
found, proceeding through a twist-chairlike transition-state
structure. This latter transition state was 4.0 kcal/mol higher in
energy than that of the boatlike transition state for attack at C-
3, consistent with the high level of C3-selectivity obtained in
the borinic acid-catalyzed reaction.
Figure 8. Plots of substrate concentration versus time for reactions of
○
BzCl (ε = 0.40 M) with epoxy alcohol 1 (blue ) and chlorodiol 5
○
(red ).
An analogous set of calculations was carried out for the
uncatalyzed epoxide ring-opening (Figures 9 and 10). The
lowest-energy uncatalyzed pathway involves attack at C-2
through a chairlike transition state, leading to a product
conformer having chloro and alkoxide groups in axial positions
the concentration versus time profiles were evident. Exact
overlay of the concentration versus time curves would not be
expected, given that the reactions are not identical and occur
under different conditions. For example, product inhibition was
implicated for the chloroacylation of 1, but is unlikely to be
significant for the benzoylation of diol 5. Nonetheless, the fact
that the reactions of 1 and 5 follow roughly similar time courses
is in keeping with acylation being the rate-determining step of
the epoxy alcohol chloroacylation process.
Computational Modeling. The proposed reaction path-
way was modeled using DFT calculations. The effects of the
borinic acid catalyst on the regiochemical outcome of the
epoxide opening and O-acylation steps were of particular
in accordance with the Furst−Plattner rule.16 Another chairlike
̈
transition state leading to attack at C-3 was found to be only
marginally higher in energy (+1.1 kcal/mol). The lower-energy
of the two transition states arises from the half-chair conformer
in which the OH group occupies a pseudoaxial position and is
thus able to stabilize the developing negative charge through
hydrogen bonding. While experimental data that can be
compared directly to these calculations do not exist, the results
are consistent with previous studies showing that mixtures of
products are formed upon ring-opening of epoxy alcohols in
F
DOI: 10.1021/acs.joc.6b02702
J. Org. Chem. XXXX, XXX, XXX−XXX

The Journal of Organic Chemistry
Article
Figure 9. Potential energy surfaces for borinic acid-catalyzed (black ) and uncatalyzed (blue ) chlorinolysis of epoxy alcohol 1, calculated at the
B3LYP/6-31+G(d,p) level of theory. Pathways leading to ring-opening at C-2 (arrow pointing to the left) and C-3 (arrow pointing to the right) are
shown.
borinic acid-catalyzed reactions of pyranoside derivatives.11a
The transition state leading to the 2-O-benzoylated product
was 3.1 kcal/mol higher in energy, consistent with the high
regioselectivity that was observed experimentally. Benzoylation
was calculated to take place through an SN2-type mechanism, in
which C−O bond formation and C−Cl bond cleavage occur in
a concerted fashion through a tetrahedral transition state
(Figure 12). Numerous alternative computational approaches
were explored, including ab initio HF and MP2 methods as well
as the M06-2X, B97D3, and B2PLYP DFT functionals, but in
no case could a tetrahedral intermediate of the type expected
from an addition−elimination pathway be identified as an
energy minimum. The question of concerted versus addition−
elimination mechanisms for the benzoylation step is not the
primary focus of this work, but we note that several prior
computational studies have identified the SN2-type mechanism
as a viable pathway for nucleophilic substitutions of acyl
chlorides18,19 and that experimental evidence consistent with a
concerted, bimolecular mechanism has also been obtained20 for
solvolysis21,22 and gas-phase acyl transfer reactions.23 This
observation does not appear to be a unique feature of the
Figure 10. Transition-state structures for: borinic acid-catalyzed ring-
opening of 1 at C-3; borinic acid-catalyzed ring-opening of 1 at C-2;
borinic acid-catalyzed acylation, as we were also unable to
locate tetrahedral intermediates as energy minima for the
reactions of the corresponding free alkoxides with BzCl (data
not shown).
uncatalyzed ring-opening of 1 at C-3; and uncatalyzed ring-opening of
1 at C-2, calculated at the B3LYP/6-31+G(d,p) level of theory.
the absence of Lewis acid additives.17 They indicate that the
borinic acid catalyst has a significant effect on the regiochemical
outcome of the epoxide chlorinolysis step.
The overall calculated reaction pathway for chloroacylation is
depicted in Figure 13. Quantitative agreement between the gas-
phase calculations and the experimental data would not be
expected, but to the extent that the calculations indicate that
the chlorohydrin-derived borinic ester is the resting state of the
catalyst and that benzoylation is turnover-limiting, the results
are qualitatively consistent with the experimentally determined
kinetic orders. The calculated energies of complexation of
Ph2BOH (used as a proxy for the (Ph2B)2O precatalyst that was
employed in the experiments) with substrate 1 and product 4
were +5.5 kcal/mol and +6.4 kcal/mol, respectively. That these
energies are within 1 kcal/mol of each other is compatible with
We then turned our attention to the benzoylation step
(Figure 11). The tetracoordinate borinate formed by attack of
chloride at C-3 can undergo relaxation from the boat
conformation to two lower-energy chairlike conformers, with
the most stable (by 0.7 kcal/mol) having the chlorine
substituent in an equatorial position. However, the lowest-
energy benzoylation pathway was found to proceed from the
chairlike conformer having the axial chloro group. In this
conformation, it is the equatorial oxygen that undergoes
functionalization, in agreement with the selectivity pattern for
G
DOI: 10.1021/acs.joc.6b02702
J. Org. Chem. XXXX, XXX, XXX−XXX

The Journal of Organic Chemistry
Article
Figure 11. Potential energy surface for benzoylation of borinic ester II, calculated at the B3LYP/6-31+G(d,p) level of theory. Pathways leading to
benzoylation at O-2 (arrow pointing to the left) and O-1 (arrow pointing to the right) are shown.
the borinic acid-catalyzed chloroacylation process in several
important ways. The experimental kinetic orders and the effects
of aroyl chloride substitution on reaction rate indicate that the
acylation step is turnover limiting. Thus, even at the relatively
low chloride ion concentrations resulting from entry of the
ethanolamine ester precatalyst into the cycle via reaction of the
ligand with BzCl, epoxide ring-opening is fast in comparison to
the reaction of the formed borinic ester with BzCl. However, if
a sufficiently reactive acylating agent is used (e.g., 4-nitro-
benzoyl chloride), epoxide ring-opening becomes turnover-
limiting. The primary role of the base is to neutralize the acid
generated in the precatalyst activation step, and the reaction
shows overall zero-order dependence on base concentration.
The calculations reveal an important role for the borinic acid in
the ring-opening step, both in providing rate acceleration and in
favoring attack of chloride at C-3 through a boatlike transition
state. Turnover-limiting acylation apparently proceeds from a
borinic ester conformer in which the chloro group occupies an
axial position. This pathway may allow for a sterically favorable
approach of the electrophile to the position that is not
deactivated by the inductive electron-withdrawing effect of the
Figure 12. Transition-state structures for benzoylation of II at O-1
(major product) and O-2 (minor product), calculated at the B3LYP/6-
31+G(d,p) level of theory.
the product inhibition inferred from the “same excess”
experiments (Figure 4).
CONCLUSIONS
■
The kinetics experiments and computational modeling that we
have carried out have solidified the mechanistic proposal
depicted in Scheme 2 and have deepened our understanding of
Figure 13. Potential energy surface for borinic acid-catalyzed chloroacylation of 1, calculated at the B3LYP/6-31G+(d,p) level of theory.
H
DOI: 10.1021/acs.joc.6b02702
J. Org. Chem. XXXX, XXX, XXX−XXX

The Journal of Organic Chemistry
Article
chloro substituent. The “same excess” kinetics experiments
reveal an appreciable level of product inhibition, a phenomenon
we have not encountered in borinic acid-catalyzed reactions of
diols. The computational results also point toward product
inhibition, as they show that both substrate and product
interact with the catalyst through single-point binding and with
similar affinities. The relatively low affinity of the catalyst for
the epoxy alcohol may be one of the issues to be addressed if
this type of reactivity is to be expanded to other substrate
classes. Given the numerous types of functionalized and
stereochemically defined products that can be accessed from
epoxy alcohols and related substrates, such expansions are
worthy of further study.
spectra were consistent with previously reported data.6 1H
NMR (500 MHz, chloroform-d) δ 4.19−4.10 (m, 2H), 3.55
(app dt, J = 9.3, 2.3 Hz, 1H), 2.71 (s, 1H), 2.35 (s, 1H), 2.21
(dddd, J = 11.2, 5.5, 4.4, 2.8 Hz, 1H), 1.98−1.89 (m, 1H),
1.82−1.68 (m, 1H), 1.69−1.61 (m, 1H), 1.60−1.45 ppm (m,
2H). 13C NMR (126 MHz, chloroform-d) δ 105.2, 69.4, 62.9,
34.4, 30.0, 19.7 ppm.
Reaction Monitoring Protocol and Data Analysis.
( )-7-Oxabicyclo[4.1.0]heptan-2-ol (1) (17.7 mg, 0.16
mmol) was weighed into a two-neck conical flask equipped
with a magnetic stir bar. A 6.35 mm attenuated total reflection
(ATR) probe with diamond sensor, connected by fiber optic
cable to a mid-infrared FTIR spectrometer, was used for
reaction monitoring. The probe was inserted into the flask, and
the second neck sealed with a fresh septum. The flask was then
purged with argon for 15 min and then 235 μL of acetonitrile
was added to the flask. The vessel was submerged into an oil
bath set to 30 °C with the stir rate set to 250 rpm. N,N-
Diisopropylethylamine (iPr2NEt) (15 μL, 11.1 mg, 0.09 mmol)
was added to the flask and allowed to stir. Weighed into a 1/2-
dram vial in the glovebox, 1,1,3,3-tetraphenyldiboroxane (2b)
(6.6 mg, 0.02 mmol) and tetra-n-butylammonium chloride
(Bu4N+Cl¯) (14.4 mg, 0.06 mmol) were dissolved in 500 μL of
acetonitrile with sonication and light heating. Benzoyl chloride
(3a) (50 μL, 57.9 mg, 0.41 mmol) was then injected into the
flask, followed immediately by 400 μL of the acetonitrile
solution of 2b and Bu4N+Cl¯. The total volume for the reaction
was 700 μL. Spectra were acquired at roughly 15-s intervals,
and the reaction conducted under these conditions required
roughly 60 min to reach completion. The appearance of the
signal at 1278 cm−1 (ester C−O stretch) was used to determine
the concentration of product. The concentration of starting
material was determined indirectly, by taking the difference
between the initial epoxy alcohol concentration and the total
concentration of chlorine-containing products inferred from the
absorbance at 717 cm−1 (C−Cl stretch). The concentration vs
time data were then fitted to the ordinary differential equation
in eq 1, where a, b, c, d, e, and f are fitted parameters and [1] is
calculated from the difference between the initial epoxide
concentration and the amount of C−Cl bond formation. A full
discussion of the derivation and validation of the fitting
methodology is provided as Supporting Information.
EXPERIMENTAL SECTION
■
General. Commercially available reagents were used as
received with the following exception: acetonitrile was purified
by being passed through columns of activated alumina under
argon and then degassed for 25 min by sparging with argon.
Stainless steel needles and gastight syringes were used to
transfer all liquids. Deionized water was obtained from an in-
house supply. NMR spectra were obtained on a 500 MHz
instrument equipped with a 5 mm cryogenically cooled probe
or a 5 mm H/F{X} probe. Chemical shifts δ are reported in
ppm relative to tetramethylsilane, referenced to residual
protium in the solvent. Spectral features are reported in the
following order: chemical shift (δ, ppm); multiplicity (s, singlet;
d, doublet; t, triplet; q, quartet; m, complex multiplet; br,
broad); number of protons; coupling constants (J, Hz). High-
resolution mass spectra (HMRS) were obtained using a time-
of-flight mass spectrometer equipped with a direct analysis in
real time (DART) ion source.
Substrate and Catalyst Preparation. ( )-7-Oxabicyclo-
[4.1.0]heptan-2-ol (1) was prepared according to a previously
reported protocol.6 NMR spectra were consistent with
previously reported data.24 1H NMR (500 MHz, chloroform-
d) δ 3.99 (dddd, J = 9.0, 7.9, 4.9, 2.9 Hz, 1H), 3.33 (m, 1H),
3.30 (m, 1H), 2.13 (d, J = 8.9 Hz, 1H), 1.90−1.81 (m, 1H),
1.81−1.72 (m, 1H), 1.61−1.49 (m, 2H), 1.49−1.38 (m, 1H),
1.31−1.19 ppm (m, 1H). 13C NMR (126 MHz, chloroform-d)
δ 67.2, 55.6, 55.5, 29.1, 23.2, 18.3 ppm.
1,1,3,3-Tetraphenyldiboroxane (2b) was prepared by hydrol-
ysis of 2-aminoethyl diphenylborinate (2a). Compound 2a
(732.5 mg, 3.25 mmol) was suspended in 4.2 mL of a 50/50
methanol/acetone solution and placed in an ice bath to cool to
0 °C. To this was added 4.2 mL of 1 M HCl(aq), and the
mixture stirred at 0 °C. After 40 min, the reaction was diluted
with 4 mL of water and then extracted three times with 4 mL of
diethyl ether. The organics were then pooled and washed twice
with 4 mL of water and once with 4 mL of brine. The organics
were then dried over magnesium sulfate, filtered, and
concentrated in vacuo without heating. The material was
dried further under high vacuum until a white solid resulted.
1,1,3,3-Tetraphenyldiboroxane (2b) was obtained as a white
solid (419.2 mg, 1.21 mmol, 75% yield). The 1H NMR
spectrum (DMSO-d6 with a drop of D2O) was consistent with
the reported data for Ph2BOH.25 The solid was stored at 0 °C
in a glovebox under nitrogen. 1H NMR (500 MHz, DMSO-d6)
δ 7.64−7.59 (m, 4H), 7.45−7.40 (m, 2H), 7.39−7.34 ppm (m,
4H). 13C NMR (126 MHz, DMSO-d6) δ 139.0, 135.6, 131.7,
129.0 ppm.
d[1]
dt
a[1] + b[1]2 + c[1]3
=
1 + d[1] + e[1]2 + f[1]3
(1)
After fitting, inputting [1] into eq 1 produced the graphical rate
equations used in this study.
Reaction Products. (1R*,2R*,3R*)-( )-3-Chloro-2-hy-
droxycyclohexyl benzoate (4) was isolated from a reaction
conducted according to the monitoring protocol described
above (0.16 mmol scale). Analysis of the unpurified reaction
1
mixture by H NMR spectroscopy showed a 16:1 ratio (1-
OBz:2-OBz) of benzoylated regioisomers. Solvent was removed
in vacuo, and the residue was purified by flash chromatography
on silica gel, eluting with hexanes/ethyl acetate (gradient
elution, 100/0 → 80/20). The product was obtained as a white
solid (28.1 mg, 0.11 mmol, 71% yield). The 1H NMR spectrum
(DMSO-d6) was consistent with the reported data.6
(1R*,2R*,3R*)-( )-3-Chloro-2-hydroxycyclohexyl 4-nitro-
benzoate was isolated from a reaction conducted according to
the monitoring protocol described above (0.16 mmol scale),
using 4-nitrobenzoyl chloride as the electrophile. Analysis of
(1R*,2R*,3R*)-( )-3-Chlorocyclohexane-1,2-diol (5). Pre-
pared according to a previously reported protocol.7b,6 NMR
I
DOI: 10.1021/acs.joc.6b02702
J. Org. Chem. XXXX, XXX, XXX−XXX

The Journal of Organic Chemistry
Article
the unpurified reaction mixture by 1H NMR spectroscopy
showed a 9.3:1 ratio (1-OCOAr:2-OCOAr) of acylated
regioisomers. Solvent was removed in vacuo, and the residue
was purified by flash chromatography on silica gel, eluting with
hexanes/ethyl acetate (gradient elution, 100/0 → 80/20). The
ASSOCIATED CONTENT
* Supporting Information
The Supporting Information is available free of charge on the
ACS Publications website at DOI: 10.1021/acs.joc.6b02702.
■
S
Kinetic data and analysis methodology, copies of 1H and
13C NMR spectra, coordinates and energies for calculated
structures (PDF)
1
product was obtained as a white solid. H NMR (500 MHz,
chloroform-d) δ 8.31−8.27 (m, 2H), 8.22−8.18 (m, 2H),
5.64−5.60 (m, 1H), 4.26−4.19 (m, 1H), 3.84 (app dt, J = 9.1,
3.0 Hz, 1H), 2.58 (d, J = 3.2 Hz, 1H), 2.39−2.29 (m, 1H),
2.15−2.04 (m, 1H), 1.87−1.64 ppm (m, 4H). 13C NMR (126
MHz, chloroform-d) δ 164.1, 150.8, 135.7, 130.9, 123.7, 75.0,
73.6, 62.0, 33.7, 28.3, 20.2 ppm.
AUTHOR INFORMATION
Corresponding Author
*E-mail: mtaylor@chem.utoronto.ca.
ORCID
Graham E. Garrett: 0000-0001-8558-2027
Mark S. Taylor: 0000-0003-3424-4380
Notes
■
The minor regioisomer ((1R*,2R*,6R*)-( )-2-chloro-6-
hydroxycyclohexyl 4-nitrobenzoate) was was obtained as a
1
white solid. H NMR (500 MHz, chloroform-d) δ 8.35−8.29
(m, 2H), 8.29−8.24 (m, 2H), 5.17 (dd, J = 9.2, 2.8 Hz, 1H),
4.43 (app td, J = 9.8, 4.5 Hz, 1H), 4.36 (dd, J = 5.3, 2.7 Hz,
1H), 2.35−2.26 (m, 1H), 2.00−1.91 (m, 1H), 1.87−1.77 (m,
3H), 1.77−1.66 ppm (m, 2H). 13C NMR (126 MHz,
chloroform-d) δ 164.0, 150.9, 135.2, 131.1, 123.8, 79.4, 68.5,
57.2, 30.5, 19.2 ppm.
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
■
This work was supported by NSERC (Discovery Grants and
Canada Research Chairs programs), the Canada Foundation for
Innovation (Project nos. 17545 and 19119), the Province of
Ontario, and by fellowships to G.E.G. (NSERC and the
Province of Ontario). The DFT calculations were carried out at
the Centre for Advanced Computing (http://cac.queensu.ca).
(1R*,2R*,3R*)-( )-3-Chloro-2-hydroxycyclohexyl 4-me-
thoxybenzoate was isolated from a reaction conducted
according to the monitoring protocol described above (0.16
mmol scale), using 4-methoxybenzoyl chloride as the electro-
1
phile. Analysis of the unpurified reaction mixture by H NMR
REFERENCES
■
spectroscopy showed a 11:1 ratio (1−OCOAr:2−OCOAr) of
acylated regioisomers. Solvent was removed in vacuo, and the
residue was purified by flash chromatography on silica gel,
eluting with hexanes/ethyl acetate (gradient elution, 100/0 →
80/20). The product was obtained as a white solid (6.5 mg,
0.02 mmol, 15% yield). Difficulties in separating the two
regioisomers resulted in a relatively low isolated yield. 1H NMR
(500 MHz, chloroform-d) δ 7.98 (app d, J = 9.0 Hz, 2H), 6.92
(app d, J = 9.0 Hz, 2H), 5.54 (app dt, J = 5.0, 2.6 Hz, 1H), 4.23
(ddd, J = 9.9, 8.9, 4.4 Hz, 1H), 3.86 (s, 3H), 3.80 (dd, J = 9.0,
3.0 Hz, 1H), 2.35−2.26 (m, 1H), 2.11−2.01 (m, 1H), 1.80−
1.61 ppm (m, 4H). 13C NMR (126 MHz, chloroform-d) δ
165.8, 163.7, 131.8, 122.6, 113.8, 75.2, 72.4, 62.1, 55.6, 33.9,
28.4, 20.2 ppm. HRMS (DART, m/z): calculated for
C14H17ClO4 ([M + H]+): 285.08936, found: 285.08954.
The minor regioisomer ((1R*,2R*,6R*)-( )-2-chloro-6-
hydroxycyclohexyl 4-methoxybenzoate) was obtained as an oil
(0.8 mg, 0.003 mmol, 2% yield). 1H NMR (500 MHz,
chloroform-d) δ 8.04 (app d, J = 9.0 Hz, 2H), 6.95 (app d, J =
9.0 Hz, 2H), 5.13 (dd, J = 8.9, 2.8 Hz, 1H), 4.43−4.36 (m, 1H),
4.31 (app dt, J = 5.6, 2.8 Hz, 1H), 2.27 (dt, J = 12.0, 4.4 Hz,
1H), 1.98−1.86 (m, 1H), 1.86−1.62 ppm (m, 4H). 13C NMR
(chloroform-d) δ 165.5, 163.9, 132.0, 122.1, 113.9, 77.9, 68.6,
57.5, 55.6, 30.3, 29.9, 19.3 ppm.
Computational Methods. DFT calculations were carried
out using the Gaussian 09 suite of programs26 at the B3LYP/6-
31+G(d,p) level of theory.27 Vibrational frequency calculations
were carried out for each stationary point to ensure they were
either an energy minimum (no imaginary frequencies) or a
transition state (one imaginary frequency) on the potential
energy surface. All transition-state trajectories were confirmed
using the intrinsic reaction coordinate method. All frequency
calculations were carried out at 1 atm and 298.15 K. Structures
were visualized using GaussView 5.0.
(1) (a) Katsuki, T.; Martin, V. Asymmetric Epoxidation of Allylic
Alcohols: the Katsuki−Sharpless Epoxidation Reaction. In Organic
Reactions; John Wiley & Sons, Inc.: Hoboken, NJ, 2004. (b) Katsuki,
T.; Sharpless, K. B. J. Am. Chem. Soc. 1980, 102, 5974−5976.
(2) (a) Hanson, R. M. Chem. Rev. 1991, 91, 437−475. (b) Pena, P. C.
A.; Roberts, S. M. Curr. Org. Chem. 2003, 7, 555−571. (c) Ko, S. Y.;
Sharpless, K. B. J. Org. Chem. 1986, 51, 5413−5415.
(3) Caron, M.; Sharpless, K. B. J. Org. Chem. 1985, 50, 1557−1560.
(4) (a) Wang, C.; Yamamoto, H. J. Am. Chem. Soc. 2014, 136, 6888−
6891. (b) Wang, C.; Yamamoto, H. Org. Lett. 2014, 16, 5937−5939.
(c) Wang, C.; Yamamoto, H. Angew. Chem., Int. Ed. 2014, 53, 13920−
13923. (d) Wang, C.; Luo, L.; Yamamoto, H. Acc. Chem. Res. 2016, 49,
193−204.
(5) Uesugi, S.-i.; Watanabe, T.; Imaizumi, T.; Shibuya, M.; Kanoh,
N.; Iwabuchi, Y. Org. Lett. 2014, 16, 4408−4411.
(6) Tanveer, K.; Jarrah, K.; Taylor, M. S. Org. Lett. 2015, 17, 3482−
3485.
(7) (a) Sasaki, M.; Tanino, K.; Hirai, A.; Miyashita, M. Org. Lett.
2003, 5, 1789−1791. (b) Tomata, Y.; Sasaki, M.; Tanino, K.;
Miyashita, M. Tetrahedron Lett. 2003, 44, 8975−8977.
(8) (a) Ishihara, K.; Kurihara, H.; Yamamoto, H. Synlett 1997, 1997,
597−599. (b) Ishihara, K.; Kurihara, H.; Yamamoto, H. J. Org. Chem.
1997, 62, 5664−5665. (c) Mori, Y.; Manabe, K.; Kobayashi, S. Angew.
Chem., Int. Ed. 2001, 40, 2815−2818. (d) Tanaka, Y.; Hasui, T.;
Suginome, M. Synlett 2008, 2008, 1239−1242. (e) Oishi, S.; Saito, S.
Angew. Chem., Int. Ed. 2012, 51, 5395−5399. (f) El Dine, T. M.;
Rouden, J.; Blanchet, J. Chem. Commun. 2015, 51, 16084−16087.
́
(9) Dimitrijevic, E.; Taylor, M. S. ACS Catal. 2013, 3, 945−962.
(10) (a) Blackmond, D. G. Angew. Chem., Int. Ed. 2005, 44, 4302−
4320. (b) Blackmond, D. G. J. Am. Chem. Soc. 2015, 137, 10852−
10866.
(11) (a) Lee, D.; Taylor, M. S. J. Am. Chem. Soc. 2011, 133, 3724−
3727. (b) Taylor, M. S. Acc. Chem. Res. 2015, 48, 295−305.
(12) Lee, D.; Williamson, C. L.; Chan, L.; Taylor, M. S. J. Am. Chem.
Soc. 2012, 134, 8260−8267.
(13) Chudzinski, M. G.; Chi, Y.; Taylor, M. S. Aust. J. Chem. 2011,
64, 1466−1469.
(14) Kevill, D. N.; Daum, P. H.; Sapre, R. J. Chem. Soc., Perkin Trans.
2 1975, 963−965.
J
DOI: 10.1021/acs.joc.6b02702
J. Org. Chem. XXXX, XXX, XXX−XXX

The Journal of Organic Chemistry
Article
(15) Infante, I.; Bonini, C.; Lelj, F.; Righi, G. J. Org. Chem. 2003, 68,
3773−3780.
(16) Kirby, A. J. Stereoelectronic Effects; Oxford Science Publications:
New York, 1996.
(17) Behrens, C. H.; Sharpless, K. B. J. Org. Chem. 1985, 50, 5696−
5704.
(18) Fox, J. M.; Dmitrenko, O.; Liao, L.-a.; Bach, R. D. J. Org. Chem.
2004, 69, 7317−7328.
̈
(19) Ruff, F.; Farkas, O. J. Phys. Org. Chem. 2011, 24, 480−491.
(20) Williams, A. Acc. Chem. Res. 1989, 22, 387−392.
(21) (a) Bentley, T. W.; Carter, G. E.; Harris, H. C. J. Chem. Soc.,
Chem. Commun. 1984, 387−389. (b) Bentley, T. W.; Carter, G. E.;
Harris, H. C. J. Chem. Soc., Perkin Trans. 2 1985, 983−990.
(22) Kevill, D. N.; Kim, C.-B. J. Chem. Soc., Perkin Trans. 2 1988,
1353−1358.
(23) (a) Asubiojo, O.; Brauman, J. I. J. Am. Chem. Soc. 1979, 101,
3715−3724. (b) Han, C.-C.; Brauman, J. I. J. Am. Chem. Soc. 1987,
109, 589−590.
́ ́
́
(24) Mello, R.; Alcalde-Aragones, A.; Olmos, A.; Gonzalez-Nunez, M.
̃
E.; Asensio, G. J. Org. Chem. 2012, 77, 4706−4710.
(25) Lee, D.; Newman, S. G.; Taylor, M. S. Org. Lett. 2009, 11,
5486−5489.
(26) Frisch, M. J.; Trucks, G. W.; Schlegel, H. B.; Scuseria, G. E.;
Robb, M. A.; Cheeseman, J. R.; Scalmani, G.; Barone, V.; Mennucci,
B.; Petersson, G. A.; Nakatsuji, H.; Caricato, M.; Li, X.; Hratchian, H.
P.; Izmaylov, A. F.; Bloino, J.; Zheng, G.; Sonnenberg, J. L.; Hada, M.;
Ehara, M.; Toyota, K.; Fukuda, R.; Hasegawa, J.; Ishida, M.; Nakajima,
T.; Honda, Y.; Kitao, O.; Nakai, H.; Vreven, T.; Montgomery, J. A., Jr.;
Peralta, J. E.; Ogliaro, F.; Bearpark, M. J.; Heyd, J.; Brothers, E. N.;
Kudin, K. N.; Staroverov, V. N.; Kobayashi, R.; Normand, J.;
Raghavachari, K.; Rendell, A. P.; Burant, J. C.; Iyengar, S. S.;
Tomasi, J.; Cossi, M.; Rega, N.; Millam, N. J.; Klene, M.; Knox, J. E.;
Cross, J. B.; Bakken, V.; Adamo, C.; Jaramillo, J.; Gomperts, R.;
Stratmann, R. E.; Yazyev, O.; Austin, A. J.; Cammi, R.; Pomelli, C.;
Ochterski, J. W.; Martin, R. L.; Morokuma, K.; Zakrzewski, V. G.;
Voth, G. A.; Salvador, P.; Dannenberg, J. J.; Dapprich, S.; Daniels, A.
̈
D.; Farkas, O.; Foresman, J. B.; Ortiz, J. V.; Cioslowski, J.; Fox, D. J.
Gaussian 09; Gaussian, Inc.: Wallingford, CT, 2009.
(27) Becke, A. D. J. Chem. Phys. 1993, 98, 5648−5652.
K
DOI: 10.1021/acs.joc.6b02702
J. Org. Chem. XXXX, XXX, XXX−XXX