of 8
Molecular
Dynamics
Investigations
of
Dienolate
[4
+ 2]
Reactions
Peng-Jui
Chen,
§
Alexander
Q.
Cusumano,
§
Kaylin
N.
Flesch,
Christian
Santiago
Strong,
William
A. Goddard,
III,
*
and
Brian
M.
Stoltz
*
Cite
This:
J. Am. Chem.
Soc.
2024,
146, 12758−12765
Read
Online
ACCESS
Metrics
& More
Article
Recommendations
*
Supporting
Information
ABSTRACT:
We report
quantum
mechanics
calculations
and
quasiclassical
trajectory
simulations
of [4 + 2] reactions
using
three
common
dienolate
substrates:
siloxy
dienes,
Li dienolates,
and
conjugated
Pd enolates.
Asynchronous
transition
structures
and
unequal
bond
formation
were
invariably
found,
with average
time
gaps of developing
bonds
ranging
from 26.5 to >251.0
fs. The results
display
a spectrum
of dynamically
concerted
and stepwise
[4 + 2]
reactions,
offering
insights
into the origin
of the stereochemical
outcomes
of such reactions.
INTRODUCTION
[4 + 2] cycloadditions
are among
the most
widely
utilized
reactions
in organic
synthesis
because
of their ability
to readily
construct
complex
motifs
through
the concomitant
formation
of two C
C
bonds
and up to four
stereocenters.
1
An
important
feature
of canonical
Diels
Alder
reactions
is its
stereospecificity,
which
guarantees
control
over the relative
stereochemistry
at two centers
of point
chirality.
Additionally,
when
the diene
is suitably
substituted,
there
is an additional
element
of stereoselectivity
that must
be considered
regarding
the relative
facial
bias of the two reactive
partners.
Fortunately,
many
[4 + 2] cycloadditions
proceed
with
high
levels
of
stereoselectivity
and
often
favor
the so-called
“endo”
product.
1,2
[4 + 2] reactions
involving
nonpolarized
or symmetrically
polarized
substrates
generally
follow
a concerted
and stereo-
specific
mechanism
(
TS1
).
3
In contrast,
polarized
dienes
and
dienophiles
can electronically
support
charge-separated
inter-
mediates
(
8
); thus,
reactions
with polarized
substrates
may
proceed
via two separate
C
C
bond-forming
events,
and the
stereochemical
outcome
in such processes
may be determined
by the stereoselectivity
of the system
(Figure
1A).
4
Cyclohexenone-derived
dienes
are common
polarized
substrates
utilized
in [4 + 2] reactions
for rapid
access
to
ketone-containing
[2.2.2]
bicycles
(e.g.,
10
,
13
, and
16
, Figure
1B
D),
which
are ubiquitous
motifs
in various
complex
natural
products.
5
Representative
substrates
include
siloxy
dienes
(
9
),
6
Li dienolates
(
12
),
7
and more
recently,
chiral
conjugated
Pd enolates
reported
by our lab (
15
, Figure
1B
D).
8
Siloxy
diene
Diels
Alder
cycloadditions
are believed
to
afford
[2.2.2]
bicyclic
products
in a stereospecific
manner
with
respect
to the stereochemistry
of the dienophile,
9
which
is
consistent
with
a concerted
pathway
that lacks
long-lived
intermediates.
Conversely,
despite
nearly
identical
reaction
and
stereochemical
outcomes
from
[4 + 2] reactions
with
Li
dienolates,
Li dienolate
reactions
have
typically
been
considered
to involve
stepwise
double
Michael
additions
because
of their mild reaction
conditions
(e.g.,
78
°
C)
10
and
the observation
of single
Michael
adducts
(
19
)
in certain
reactions
(Figure
1E), consistent
with sequential
conjugate
addition
pathways.
11
Recently,
our lab reported
the catalytic
asymmetric
synthesis
of [2.2.2]
bicycles
that proceed
via chiral
conjugated
Pd enolates.
8
While
the stereochemistry
of most
products
is consistent
with that expected
from
Diels
Alder
reactions
(i.e., stereospecific
with respect
to the dienophile
olefin
geometry
in addition
to high endo/exo
diastereoselec-
tivity),
it was ambiguous
whether
the reaction
is, indeed,
concerted
and stereospecific
or completely
stereoselective.
One possible
delineation
between
concerted
and stepwise
pathways
depends
on whether
an intermediate
and a second
transition
state
is featured.
On the potential
energy
surface
(PES),
a stepwise
pathway
would
be characterized
by two local
maxima
(transition
structures)
connected
by a local minimum
(intermediate)
along
the reaction
coordinate
(Figure
2, top
PES).
However,
analysis
solely
based
on electronic
energy
(i.e.,
the PES)
can be misleading.
12
For various
pericyclic
reactions,
Received:
February
22, 2024
Revised:
April
9, 2024
Accepted:
April
10, 2024
Published:
April
29,
2024
Article
pubs.acs.org/JACS
© 2024
The Authors.
Published
by
American
Chemical
Society
12758
https://doi.org/10.1021/jacs.4c02681
J. Am. Chem.
Soc.
2024,
146, 12758
12765
This article is licensed under CC-BY-NC-ND 4.0
flat regions
may exist
along
the reaction
coordinate
PES
(Figure
2, middle
PES).
Molecules
may
spend
extended
amounts
of time in these
regions
as long-lived
intermediates
despite
the absence
of a local potential
energy
minimum.
13,14
Such
scenarios
can arise from the entropic
penalty
associated
with
the progression
along
the reaction
coordinate,
and
consequently,
these
intermediates
are often
described
as
entropic
intermediates,
which
exhibit
shallow
free energy
minima
but no potential
energy
minima.
While
these
reactions
appear
concerted
because
of the lack of an intermediate
on the
PES,
they can be dynamically
stepwise
if long-lived
entropic
intermediates
are present.
As entropic
intermediates
are not potential
energy
minima,
they are challenging
to identify
on the basis
of transition
structure
and PES studies.
This also leads
to the difficulty
in
characterizing
dynamically
stepwise
reactions.
Complementary
to transition
state
analysis,
molecular
dynamics
simulations
have been
shown
effective
in describing
reactions
in a time-
resolved
manner,
which
is key to discerning
dynamically
concerted
and stepwise
processes.
14,15
In the context
of the Diels
Alder
reaction,
Houk
and co-
workers
demonstrated
that highly
asynchronous
transition
structures
(i.e., those
with large differences
in bond
distances
in the two forming
bonds)
may undergo
two temporally
distinct
bond-forming
steps,
even
in the absence
of an
intermediate
on the PES.
However,
because
the time
gap
between
these
steps is generally
less than or equal
to the period
of a typical
C
C
bond
stretch
(
30
60
fs), the authors
defined
these
processes
to be “dynamically
concerted.”
16
In this report,
we present
the results
of molecular
dynamics
simulations
of [4 + 2] reactions
involving
cyclohexenone-
derived
diene
substrates
to describe
a spectrum
of mechanisms
from synchronous
and dynamically
concerted
to asynchronous
and dynamically
stepwise.
Our findings
are consistent
with the
usual
classifications
of [4 + 2] reactions
with siloxy
dienes
and
Li dienolates
and demonstrate
the influence
of substrate
choice
on possible
stereochemical
outcomes.
EXPERIMENTAL:
COMPUTATIONAL
METHODS
All quantum
mechanics
calculations
were carried
out with the ORCA
program.
17
Geometry
optimizations,
harmonic
frequency
calculations,
and single-point
energy
evaluations
were
carried
out with density
functional
theory
(DFT).
To locate
saddle
points
of the reactions
and
propagate
trajectories,
the Perdew
Burke
Ernzerhof
(PBE)
func-
tional
18
paired
with
Becke
Johnson
damped
D4 dispersion
corrections
19
(PBE-D4)
and implicit
conductor-like
polarizable
continuum
model
(CPCM)
for the respective
solvents
20
were used.
For geometry
optimization
and harmonic
frequency
calculations,
Pd is
described
by the def2-TZVP
basis set
21
and the ECP28MWB
small-
core (18 explicit
valence
electrons)
quasi-relativistic
pseudopoten-
tial,
22
while
C, H, N, and P are assigned
the def2-SV(P)
basis.
Diffuse
functions
are added
to oxygen
(ma-def2-SV(P)).
Quasiclassical
trajectories
were
initialized
by normal
mode
sampling
at the saddle
points
in which
each normal
mode
includes
zero-point
energy,
as well as a Boltzmann
sampling
of geometries
corresponding
to thermal
energy
available
at the respective
reaction
temperatures
with a randomized
phase.
Trajectories
were propagated
in the forward
and reverse
directions
with a velocity
Verlet
algorithm
in 1.0 fs time
steps,
as implemented
in Singleton’s
Progdyn
program,
23
until either
formation
of cycloadduct
(forming
both C
C bond
lengths
< 1.5 Å), formation
of starting
material
(the shorter
C
C
bond
length
> 3 Å and the longer
> 4 Å), or reaching
max time
(500 fs).
Considering
the size of the system,
a hundred
trajectories
were
sampled
to maintain
tractability,
and we limited
the DFT
to the
generalized
gradient
approximation
(GGA)
functional
PBE for the
dynamic
simulations.
To investigate
the influence
of DFT
functional
choice
on simulation
results,
a suite
of control
experiments
was
performed
on a model
reaction
using
hybrid,
metahybrid,
and range-
separated
functionals.
While
simulations
with different
functionals
yielded
slightly
variable
results,
all simulation
results
described
in this
report
were performed
with the same
functionals
for comparison.
For
additional
details
and discussion,
see the Supporting
Information
(SI).
RESULTS
AND
DISCUSSION
Molecular
dynamics
simulations
provide
time-resolved
studies
of pericyclic
reactions,
which
enables
identification
of long-
Figure
1.
(A) Potential
concerted
and stepwise
[4 + 2] reaction
pathways
of nonpolarized
and polarized
dienes.
(B) Siloxy
diene
[4 +
2] reactions.
(C) Lithium
base-promoted
[4 + 2] reactions.
(D) Pd-
catalyzed
asymmetric
[4 + 2] cycloaddition.
(E) Isolated
intermediate
from intended
[4 + 2] reaction
with Li dienolate.
Figure
2.
Example
PES of a stepwise
[4 + 2] reaction
(top),
concerted
[4 + 2] reaction
(bottom),
and a dynamically
stepwise
(one-step,
two-stage)
[4 + 2] reaction
(middle).
Journal
of
the
American
Chemical
Society
pubs.acs.org/JACS
Article
https://doi.org/10.1021/jacs.4c02681
J. Am. Chem.
Soc.
2024,
146, 12758
12765
12759
lived
entropic
intermediates
in reactions
where
bond
formations
are not synchronous.
13,14
Herein,
we define
the
average
time
gap (
Δ
t
)
as the difference
in time
between
formation
of the two developing
C
C
bonds
and adopt
60 fs,
approximately
the period
for a typical
C
C
bond
stretch,
as
the time
criterion
to discern
dynamically
concerted
and
stepwise
reactions,
as proposed
by Houk
and co-workers.
13
Previously,
the Houk
lab reported
an average
time gap of 3.9 fs
in the symmetrical
[4 + 2] reaction
of butadiene
and ethylene
(
TS1
,
Figure
3A).
16
While
more
asynchronous
transition
structures
led to time gaps of up to 56.5 fs (
TS3
,
Figure
3A),
the reported
reactions
are all deemed
dynamically
concerted.
Conversely,
highly
polarized
systems,
such
as cyclo-
hexenone-derived
dienolates,
may lead to dynamically
stepwise
reactions.
Dynamically
stepwise
[4 + 2] reactions
would
feature
long-lived
intermediates
analogous
to sequential
Michael
additions.
These
reactions
are considered
stereo-
selective,
rather
than stereospecific,
because
bond
rotation
of
the intermediate
about
the dienophile
is potentially
allowed,
which
can lead
to formation
of additional
diastereomers
(Figure
3B).
To investigate
the timing
of the bond
formations
in [4 + 2]
reactions
with different
cyclohexenone-derived
substrates,
we
report
the molecular
dynamics
simulations
of three
different
cyclohexenone
systems:
siloxy
dienes,
Li dienolates,
and
conjugated
Pd enolates.
2.1.
Reactions
with
Siloxy
Dienes.
[4 + 2] reactions
of
siloxy
dienes
and dienophiles
are generally
considered
to be
concerted
processes
and are referred
to as Diels
Alder
reactions
in the literature.
Aiming
to investigate
whether
these
reactions
are concerted
in a dynamical
sense,
we studied
the model
intramolecular
reaction
of siloxy
cyclohexadiene
25
(Figure
4A).
Reported
examples
of [4 + 2] reactions
with siloxy
dienes
are typically
performed
in the presence
of Lewis
acid catalysts
or in high-boiling
solvents
at elevated
temperatures.
To
eliminate
potential
influences
of exogenous
additives,
our
computations
were
performed
for the additive-free
reaction
with the CPCM
implicit
solvation
model
for toluene.
The
vibrational
frequency
calculations
and trajectory
initialization
in the molecular
dynamics
simulation
are in accord
with
thermal
energies
available
at 383.15
K.
Quantum
mechanical
(QM)
calculations
revealed
an
asynchronous
transition
state (
TS4
)
for the modeled
reaction
with an imaginary
frequency
of 323 cm
1
. Bond
distances
of
C1
C2
(
d
1
) and C3
C4
(
d
2
) were found
to be 2.08 and 2.71
Å, respectively.
Comparison
of the free energies
of the
transition
state with the corresponding
reactant
indicates
an
activation
energy
barrier
of 23.2 kcal/mol.
From
the identified
transition
state structure,
molecular
dynamics
simulations
were
employed
to further
investigate
the synchronicity
of the bond
formations.
In practice,
molecules
will be found
with
geometries
distributed
along
the dividing
surface
defined
by the calculated
transition
structure
depending
on the available
thermal
energy.
To simulate
this distribution,
weighted
Boltzmann
sampling
was performed
to yield a set of starting
points
represented
by
the blue
dots
in Figure
4B. From
these
starting
points,
trajectories
are propagated
in both
directions,
which
means
that the atoms
are assigned
initial
velocities
of opposite
signs
for the two directions.
The
trajectories
are subsequently
plotted
on a 2D potential
energy
surface
defined
by
d
1
and
d
2
.
Productive
trajectories
lead to the formation
of the cyclo-
adduct
in one direction
and separation
of reactants
in the
other,
while
unproductive
trajectories
yield
separated
starting
material
(SM)
in both
propagated
directions.
These
unproductive
pathways
are termed
recrossing
to the starting
material
(SM)
and are omitted
from the analysis
of
Δ
t
.
Figure
3.
(A) Examples
of previously
studied
[4 + 2] reactions.
(B)
Rotation
of dienophile
bond
in dynamical
intermediates
can lead to
formation
of diastereomers.
Figure
4.
(A) Computed
thermodynamics
of intramolecular
[4 + 2]
reaction
with siloxy
diene.
(B) Trajectories
derived
from
molecular
dynamics
simulation
originating
from
TS4
.
Journal
of
the
American
Chemical
Society
pubs.acs.org/JACS
Article
https://doi.org/10.1021/jacs.4c02681
J. Am. Chem.
Soc.
2024,
146, 12758
12765
12760