Investigation of the C–N Bond-Forming Step in
a Photoinduced, Copper-Catalyzed Enantioconvergent N–
Alkylation: Characterization and Application of a Stabilized
Organic Radical as a Mechanistic Probe
Heejun Lee
,
Jun Myun Ahn
,
Paul H. Oyala
,
Cooper Citek
,
Haolin Yin
,
Gregory C. Fu
*
,
Jonas C. Peters
*
Division of Chemistry and Chemical Engineering, California Institute of Technology, Pasadena,
California 91125, United States
Abstract
Whereas photoinduced, copper-catalyzed couplings of nitrogen nucleophiles with alkyl
electrophiles have recently been shown to provide an attractive approach to achieving a variety
of enantioselective C-N bond constructions, mechanistic studies of these transformations have
lagged the advances in reaction development. Herein we provide mechanistic insight into
a previously reported photoinduced, copper-catalyzed enantioconvergent C–N coupling of a
carbazole nucleophile with a racemic tertiary
α
-haloamide electrophile. Building on the isolation
of a copper(II) model complex whose EPR parameters serve as a guide, we independently
synthesize two key intermediates in the proposed catalytic cycle, a copper(II) metalloradical,
(L*Cu
II
(carb’)
2
) (L* = a monodentate chiral phosphine ligand; carb’ = a carbazolide ligand),
as well as a tertiary
α
-amide organic radical (R·); the generation and characterization of R·
was guided by DFT calculations, which suggested that it would be stable to homocoupling.
Continuous-wave (CW) and pulse EPR studies, along with corresponding DFT calculations, are
among the techniques used to characterize these reactive radicals. We establish that these two
radicals do indeed combine to furnish the C–N coupling product in good yield and with significant
enantiomeric excess (77% yield, 55% ee), thereby supporting the chemical competence of these
proposed intermediates. DFT calculations are consistent with R· initially binding to copper(II) via
a dative interaction from the closed-shell carbonyl oxygen atom of the radical, which positions
*
Corresponding Authors:
gcfu@caltech.edu, jpeters@caltech.edu.
Supporting Information
The Supporting Information is available free of charge on the ACS Publications website. Synthetic procedures for all compounds and
experiments, additional experimental details, spectroscopic characterization data including
1
H NMR and EPR data, details of DFT
calculations, and molecular coordinates.
The authors declare no competing financial interest.
HHS Public Access
Author manuscript
J Am Chem Soc
. Author manuscript; available in PMC 2023 March 09.
Published in final edited form as:
J Am Chem Soc
. 2022 March 09; 144(9): 4114–4123. doi:10.1021/jacs.1c13151.
Author Manuscript
Author Manuscript
Author Manuscript
Author Manuscript
the
α
carbon for direct reaction with the copper(II)-bound carbazole N-atom, to generate the C–N
bond with enantioselectivity, without the formation of an alkylcopper(III) intermediate.
Graphical Abstract
INTRODUCTION
During the past decade, the use of photoredox catalysis to furnish reactive radical
intermediates that enable new enantioselective bond constructions for organic synthesis
has expanded rapidly.
1
,
2
,
3
,
4
Photoredox catalysis offers particular promise for the efficient
asymmetric synthesis of chiral amines, which are ubiquitous in chemistry and biology
while being challenging to synthesize through classical bond-forming processes such as S
N
2
reactions.
Several years ago, we reported a photoinduced, copper-catalyzed method for coupling
certain racemic tertiary alkyl chloride electrophiles with carbazoles and indoles to generate
fully substituted stereocenters with high enantioselectivity (Figure 1, left).
5
These N-
alkylations proceed in the presence of a copper source (CuCl), a chiral monodentate
phosphine (L*), a base (LiO
t
-Bu), and visible-light irradiation (blue LED). This study
combined progress in asymmetric synthesis, base-metal catalysis, and photoredox catalysis
to achieve an uncommon coupling reaction. A number of more recent studies have further
expanded asymmetric C–N couplings via photoinduced, copper-catalyzed couplings,
6
,
7
including a recent report from our labs demonstrating enantioconvergent substitutions of
unactivated racemic electrophiles by amides.
8
There has also been recent progress in
thermally driven, Cu-catalyzed enantioconvergent substitutions using alkyl electrophiles and
nitrogen nucleophiles.
9
,
10
,
11
The ability to efficiently generate a fully substituted stereocenter in high enantiomeric
excess from a racemic tertiary alkyl halide raises interesting mechanistic questions,
especially with regard to the C–N bond-forming step. Whereas we and others have explored
the mechanisms of photoinduced, copper-catalyzed C–X couplings,
12
,
13
,
14
,
15
,
16
,
17
,
18
which
can vary depending on the nucleophile, the electrophile, and the reaction conditions,
there is a paucity of detailed mechanistic investigations of
asymmetric
photoinduced C–
Lee et al.
Page 2
J Am Chem Soc
. Author manuscript; available in PMC 2023 March 09.
Author Manuscript
Author Manuscript
Author Manuscript
Author Manuscript
N couplings. The transformation shown on the left side of Figure 1 hence provided an
attractive opportunity to undertake such a study.
Recapping prior work germane to the present study, we have suggested mechanistic
pathways for photoinduced, copper-catalyzed C–N couplings that involve a single primary
cycle, including a copper (or other) photoreductant that leads to liberation of a persistent
copper(II) radical and an organic radical; these radicals then combine in an out-of-cage step
to construct the C–N bond (e.g., Figure 1, right).
14
We have also presented scenarios that
involve two independent catalytic cycles, one cycle for generation of the organic radical
via photoinduced reduction of the electrophile by an excited-state copper (or other) species,
and a second cycle for C–N bond formation by a different copper complex;
8
,
19
two-cycle
mechanisms are common in photoredox catalysis.
20
,
21
,
22
While our prior report of photoinduced, copper-catalyzed asymmetric C–N coupling was
focused on reaction development, a number of mechanistic observations and analogy with
prior studies led us to postulate the cycle shown in Figure 1 (right) as an initial working
hypothesis for the overall transformation. Here,
A
is a copper–carbazolide adduct, an
example of which, L*
2
Cu(carb), was synthesized and crystallographically characterized,
as well as shown to be competent as a catalyst (eq 1; no C–N bond formation in the absence
of light).
5
Photoexcitation of
A
generates
B
, which reduces the tertiary
α
-chlorocarbonyl
electrophile
23
to furnish radicals R
·
and
C
. These radicals combine, either directly or via
[L
n
Cu
III
(carb)R]X, to form the enantioenriched C–N coupling product and to generate Cu
I
complex
D
, which can undergo ligand substitution by the nucleophile to restart the cycle.
(1)
In the present investigation, we focus on the key C–N bond-forming step. This step, as
well as related transformations that proceed via the combination of an organic (R·) and a
metalloradical species (M·), are challenging to study due to the highly reactive nature of
many radicals. In this report, we describe the independent generation of two radicals, R·
and M·, that are directly relevant to the enantioselective N-alkylation of interest, and we
establish that these radicals undergo C–N coupling in good yield and substantial ee. On the
basis of these studies and of DFT calculations, we propose a refined mechanism (Figure 2)
that provides additional detail about the catalytic cycle, including the enantioselective C–N
bond-forming step.
Lee et al.
Page 3
J Am Chem Soc
. Author manuscript; available in PMC 2023 March 09.
Author Manuscript
Author Manuscript
Author Manuscript
Author Manuscript
RESULTS AND DISCUSSION
Model study: Synthesis and structural characterization of a Cu
II
metalloradical bearing one
phosphine and two carbazolides.
For an initial model study, we targeted a copper(II) complex that features a chelating
phosphine-carbazolide (NP) and a terminal carbazolide ligand (Scheme 1), reasoning that its
characterization might aid our in situ characterization of a catalytically relevant copper(II)
intermediate (vide infra).
Synthesis of this copper(II) model complex proceeds via the addition of the neutral
phosphine-carbazole ligand (HNP) to CuCl in the presence of LiO
t
-Bu, which affords a
1:1 NP:Cu dimer complex (
A
) wherein each bidentate ligand bridges two copper centers.
X-ray analysis of this species shows a Cu–Cu distance of 4.9 Å (see the Supporting
Information). Subsequent addition of the potassium salt of a substituted carbazole (carb’),
followed by the addition of FcPF
6
as an oxidant, affords the target copper(II) complex
(NP)Cu
II
(carb’) (
B
).
24
,
25
Although thermally sensitive, X-ray quality crystals of purple
(NP)Cu
II
(carb’) could be grown from a concentrated solution at −78 °C. X-ray analysis
confirms its monomeric structure bearing an NP chelate and a terminal carbazolide ligand
(Scheme 1).
26
Model Study: Demonstration of C–N bond formation between a Cu
II
metalloradical
((NP)Cu
II
(carb’)) and an organic radical (trityl derivative).
Whereas Gomberg’s dimer (a dimer of trityl radicals)
27
,
28
has been used previously as a
reagent to examine C–N bond formation from a well-defined copper(II) amide precursor,
29
we could not employ this reagent in the present case due to the thermal instability of
(NP)Cu
II
(carb’), which decays to the dicopper(I) dimer
A
and carb’H in toluene. In contrast,
a variant of the trityl radical, tris(
p
-
t
-butylphenyl)methyl radical, can be generated that exists
exclusively as a free radical, rather than as a dimer, in solution.
30
,
31
,
32
Upon treatment of
(NP)Cu
II
(carb’) (
B
) with this radical in toluene at −40 °C, the N-alkylated carbazole was
produced in excellent yield (91%) with concomitant formation of dicopper(I) dimer
A
(eq
2).
(2)
Lee et al.
Page 4
J Am Chem Soc
. Author manuscript; available in PMC 2023 March 09.
Author Manuscript
Author Manuscript
Author Manuscript
Author Manuscript
This transformation provides evidence that a persistent Cu
II
–carb radical can undergo N-
alkylation in the presence of an alkyl radical under our coupling conditions (toluene, −40
°C). In a related observation, the addition of
N
-
t
-butyl-
α
-phenylnitrone, a spin trap, to
(NP)Cu
II
(carb’) (
B
) in toluene at −78 °C also liberates an N-functionalized carbazole (eq 3).
These C–N couplings are consistent with N-centered radical character in (NP)Cu
II
(carb’), in
accord with EPR spectroscopy and DFT calculations (vide infra).
(3)
Evidence for a Cu
II
metalloradical during catalysis: X-band continuous-wave (CW) EPR
spectroscopy.
Focusing on the catalytic asymmetric transformation of interest, X-band (9.4 GHz) CW
EPR spectroscopy was used to probe for the accumulation of a copper(II) species during
the course of catalysis; to provide satisfactory signal-to-noise, the conditions depicted in
Figure 3 were used. In particular, 10 mol% CuCl and 12 mol% L* (note: all mechanistic
experiments were conducted with (
R
)-L*) were present at the outset, and the reaction was
carried out at −78 °C rather than at −40 °C. A representative EPR spectrum, collected
after 2 hours of reaction, is shown in Figure 3 (red trace). A species exhibiting a relatively
broad EPR spectrum is observed, with hyperfine coupling to a single copper nucleus (copper
has two
I
= 3/2 magnetic isotopes,
63
Cu (59.17% abundant) and
65
Cu (30.83% abundant))
suggested by the quartet observed at the low-field edge of the spectrum. Additionally, a
sharp feature centered at
g
~2.0 is observed, which is suggestive of the presence of an organic
radical (vide infra).
Initially positing that the copper-containing intermediate might correspond to the one-
electron oxidation product of L*
2
Cu(carb), we studied via X-band EPR analysis (77 K,
toluene) the product of oxidation of L*
2
Cu(carb) by ferrocenium hexafluorophosphate
(FcPF
6
; 1 equiv) at −78 °C. As shown in Figure 3 (black trace), an EPR spectrum with
the same dominant features as the red trace is obtained. This result points to the oxidation
product of L*
2
Cu(carb) as a possible intermediate during catalysis.
To our surprise, an X-band EPR spectrum of the copper(II) model complex (NP)Cu
II
(carb’)
(
B
) (blue spectrum) features an EPR envelope that is qualitatively similar to the red and
black spectra. This result led us to wonder whether the intermediate observed during
catalysis contains
one L* phosphine ligand (rather than two), and
two carbazolide ligands
(rather than one). In accord with this idea, toluene, a non-polar solvent, is optimal for the
enantioconvergent C–N coupling process that we have developed (Figure 1, left). Under
Lee et al.
Page 5
J Am Chem Soc
. Author manuscript; available in PMC 2023 March 09.
Author Manuscript
Author Manuscript
Author Manuscript
Author Manuscript
these conditions, ionic species should be relatively disfavored, and it is hence plausible that
intermediate
C
in our initially proposed catalytic cycle (Figure 1, right) might be converted
to L*Cu
II
carb
2
under the reaction conditions (
C
1
to
C
2
in Figure 2).
Synthesis of the proposed Cu
II
metalloradical observed during catalysis: L*Cu
II
(carb’)
2
.
To further explore the assignment of the paramagnetic copper(II) species observed during
a catalyzed reaction, we targeted the independent synthesis of L*Cu
II
(carb’)
2
; as for
(NP)Cu(carb’) (
B
), the
t
-butyl groups were employed to stabilize the copper(II) species
against undesired reactivity at the 3 and 6 positions of the carbazolide ring,
25
,
33
as well
as to increase its solubility in organic solvents. To this end, one equiv L* was added to
[Cu
I
(carb’)
2
][Na(THF)
3
] to generate [L*Cu
I
(carb’)
2
][Na(THF)
x
] in situ (see the Supporting
Information). Subsequent addition of a slight excess of FcPF
6
(1.1 equiv) at −78 °C
furnished an EPR trace analogous to that obtained from the catalytic reaction using carb’
rather than carb (Figure 4, left column). This experiment provides strong support that the
species that accumulates during catalysis is L*Cu
II
(carb’)
2
.
Support for the structural assignment of L*Cu
II
(carb’)
2
: Pulse EPR and DFT data.
Pulse EPR studies and corresponding DFT calculations were carried out on the model
copper(II) complex (NP)Cu
II
(carb’) (
B
) and the species observed during catalysis to further
establish their structural similarity (see Figure 4 and the Supporting Information). Q-band
(34 GHz) ESE-EPR spectra provided enhanced resolution of the
g
-anisotropy and allowed
for more rigorous estimation of the large hyperfine coupling to the
31
P nucleus of the
phosphine. Furthermore, application of Q-band hyperfine sublevel correlation spectroscopy
(HYSCORE) allowed for detection of nitrogen hyperfine couplings, which are considerably
smaller than copper and phosphorus couplings (and thus not resolved in the field-swept
EPR spectra), to the
S
= 1/2 copper center in both complexes. The assigned
14
N couplings
for both complexes exhibit highly anisotropic tensors of axial symmetry, with similar
magnitudes: (|
A
(
14
N)|= [9, 32, 9] MHz,
a
iso
= 16.7 MHz for (NP)Cu
II
(carb’); |
A
(
14
N)|= [4,
30, 5] MHz,
a
iso
= 13 MHz for L*Cu
II
(carb’)
2
). The significant anisotropic components (
T
= [−7.7, 15.4, −7.7] MHz for (NP)Cu
II
(carb’);
T
= [−9.0, 17, −8.0] MHz for L*Cu
II
(carb)
2
))
of these
14
N hyperfine tensors are almost completely axial, and they are at least an order of
magnitude larger than what could be imparted from purely through-space dipolar coupling
between spin density on Cu and the
14
N nucleus, pointing to significant localization of spin
in a single p-orbital at the carbazole nitrogen in both complexes (estimated p-orbital spin
densities of 0.14 e
–
and 0.15 e
–
for (NP)Cu
II
(carb’) and L*Cu
II
(carb)
2
, respectively).
34
These spectroscopic data are in good agreement with the DFT-predicted spin-density
distributions for (NP)Cu
II
(carb’) and L*Cu
II
(carb)
2
; the spin-density plots (Figure 5) show,
in qualitative terms, similar spin densities at nitrogen for both complexes, as well as
significant spin densities at phosphorus. The N–Cu–N and N–Cu–P angles are distinct
in the two complexes as a result of the chelation in (NP)Cu
II
(carb’); this leads to some
differences in spin distribution, especially in the larger amount of spin at phosphorus
predicted for (NP)Cu
II
(carb), which correlates well with the EPR data (
a
iso
= 562 MHz
for (NP)Cu
II
(carb’);
a
iso
= 422 MHz for L*Cu
II
(carb)
2
).
Lee et al.
Page 6
J Am Chem Soc
. Author manuscript; available in PMC 2023 March 09.
Author Manuscript
Author Manuscript
Author Manuscript
Author Manuscript
Synthesis and characterization of a tertiary alkyl radical relevant to catalytic asymmetric
C–N coupling.
Having achieved one of our key goals, a method for the generation of a persistent copper
metalloradical of possible direct relevance to catalysis (L*Cu
II
(carb’)
2
;
C
2
in Figure 2), we
next turned our attention to accessing the other radical partner in the proposed coupling step
of the catalytic cycle, R· (Figure 2). Relatively few alkyl radicals (other than those stabilized
by an adjacent lone pair) are sufficiently stable to be generated independently and used as
a reagent (such as the substituted trityl radical discussed above), but a DFT study indicated
that the tertiary alkyl radicals resulting from removal of a halide atom from some of our
electrophiles might be sufficiently stable to radical–radical homocoupling to enable their
generation and characterization in solution.
Although homocoupling is predicted to be approximately thermoneutral in the case of a
less-hindered, ethyl-substituted radical at 298 K (eq 4), homocoupling is ~31 kcal/mol
uphill at 298 K in the case of the corresponding isopropyl-substituted radical (eq 5; the
meso isomer is ~35 kcal/mol uphill at 298 K).
35
,
36
,
37
,
38
According to DFT calculations,
the majority of spin in these radicals resides at C
α
, as shown in the spin-density plot for
the isopropyl-substituted radical (Figure 6, left; Loewdin spin population at C
α
is 0.56
e
−
). This prediction, and the fact that its
α
-bromoamide precursor is a competent coupling
partner under catalytic conditions (vide infra), motivated us to pursue the generation and
characterization of the isopropyl-substituted radical from an
α
-haloamide precursor.
(4)
(5)
After canvassing a variety of common strong one-electron reductants, such as sodium-
mercury amalgam and potassium graphite, we settled on the use of a titanium(III)
reagent originally reported by Cummins and coworkers.
39
,
40
The high reactivity of the
titanium reagent enables bromine-atom abstraction from the
α
-bromoamide substrate at low
temperature (−78 °C in toluene; eq 6), with the reaction mixture turning from the green color
Lee et al.
Page 7
J Am Chem Soc
. Author manuscript; available in PMC 2023 March 09.
Author Manuscript
Author Manuscript
Author Manuscript
Author Manuscript
indicative of Ti
III
(N(R)Ar)
3
to the orange color of Ti
IV
Br(N(
t
-Bu)Ar)
3
, which was isolated
and structurally characterized (see the Supporting Information). The desired
α
-amide radical
is persistent in solution at −78 °C for more than 12 hours and can be characterized by
X-band EPR spectroscopy even at room temperature, where a strong signal was observed
after ~30 min.
41
(6)
(7)
To probe whether substantial spin density resides at the tertiary C
α
position, as suggested
by DFT calculations (Figure 6, left), we prepared an isotopically labeled congener with
deuterium substitution at the methyl and methine positions of the isopropyl substituent (eq
7). As shown in Figure 6 (right), the room-temperature X-band EPR spectrum (298 K) of
the radical is highly sensitive to deuterium substitution, with large hyperfine coupling to the
methine proton of the isopropyl substituent. Coupling to the ortho hydrogen atoms of the
arene substituent is also evident in the spectrum. These data indicate that there is substantial
spin density at the
α
carbon atom.
Enantioselective C–N bond formation between two radicals relevant to catalytic
asymmetric C–N coupling: metalloradical L*Cu
II
(carb’)
2
and the
α
-amide organic radical.
Having developed methods for the independent generation and characterization of two
key proposed radical intermediates, the copper(II) metalloradical and the
α
-amide organic
radical, in the catalytic asymmetric coupling process (Figure 2), we sought to determine
whether these two radicals would engage in C–N bond formation, and, if so, whether it
would proceed with significant ee and the same sense of enantioselectivity as the catalyzed
process.
Lee et al.
Page 8
J Am Chem Soc
. Author manuscript; available in PMC 2023 March 09.
Author Manuscript
Author Manuscript
Author Manuscript
Author Manuscript
We generated the organic radical in toluene at −78 °C, next added the independently
prepared copper(II) metalloradical, L*Cu
II
(carb’)
2
, and then allowed them to react at −78 °C
over two hours, which resulted in C–N coupling in 77% yield and 55% ee (Scheme 2, top).
42
This yield and ee can be compared to that obtained from a coupling of the
α
-bromoamide
electrophile with the 2,6-
t
-Bu
2
-carbazole nucleophile catalyzed by L*
2
Cu(carb’) (71% yield,
80% ee; Scheme 2, bottom).
43
Notably, the two processes proceed with the same sense (and
a substantial level) of enantioselectivity; the difference in the specific ee values between
the stoichiometric and the catalyzed reactions may reflect differences in the conditions,
such as the absence of LiO
t
-Bu (and therefore LiBr and HO
t
-Bu as the catalyzed coupling
progresses) and the presence of the titanium byproduct in the stoichiometric reaction; LiO
t
-
Bu is not compatible with the titanium(III) reagent used to generate the organic radical.
The observation of C–N bond formation in good yield as well as substantial ee in the
stoichiometric radical–radical coupling (Scheme 2, top), along with the aforementioned EPR
data, are consistent with the involvement of a persistent metalloradical, copper(II) complex
(L*)Cu
II
(nucleophile)
2
, and an
α
-amide organic radical in the enantioselectivity-determining
C–N bond-forming step during catalysis.
Model for the C–N bond-forming pathway: DFT calculations.
Using DFT calculations, we have explored several C–N coupling pathways to gain insight
into the elementary steps by which this transformation occurs.
44
Given the complexity of
the system (including the number of possible pathways and the number of atoms), we
acknowledge at the outset that our model is a working hypothesis. Nevertheless, we posit a
scenario for the observed transformation that can provide a basis for further investigation.
A central question is whether the organic radical binds to L*Cu
II
(carb)
2
prior to C–N bond
formation (versus direct intermolecular radical–radical coupling at a coordinated carbazole
ligand, which bears considerable spin density on nitrogen (see Figure 5, right)), and, if
so, how it binds. In our recent study of photoinduced, copper-catalyzed enantioconvergent
alkylations of amides by unactivated secondary electrophiles, initial coordination of a
Lewis-basic functional group (e.g., a phosphonyl group) to a chiral copper(II) intermediate,
followed by asymmetric C–N bond formation, provided a plausible rationale for the good
enantioselectivity that was observed. Correspondingly, we anticipated that binding of the
Lewis-basic amide carbonyl to copper(II) might set the stage for C–N bond formation in the
present case (eq 8).
(8)
Lee et al.
Page 9
J Am Chem Soc
. Author manuscript; available in PMC 2023 March 09.
Author Manuscript
Author Manuscript
Author Manuscript
Author Manuscript
A search of plausible structures (see the Supporting Information) suggests that association
of the organic radical to (L*)Cu
II
(carb)
2
through the carbonyl oxygen to form a weak
Cu–O interaction is plausible (ΔE = +7.2 kcal mol
–1
with respect to (L*)Cu
II
(carb)
2
and
the organic radical for the isopropyl-substituted radical; ΔE = +4.8 kcal mol
–1
for the
ethyl-substituted radical; these were minimized as broken-symmetry singlet structures; see
Figure 7 for a structural model of the isopropyl-substituted radical). Calculation of the
Loewdin spin distribution shows a high degree of radical character at C
tert
of the organic
radical (
i
-Pr: −0.57 e
−
) and at copper (Cu: +0.32 e
−
), as well as significant spin on the N
atoms of the two carbazolide ligands and the P atom of the phosphine ligand. We anticipate
that C–N bond formation occurs directly from such an intermediate (Figure 7, rotated view
shows image along the C
tert
–N
carb1
vector; see the Supporting Information for a calculated
transition-state (TS) structure for C–N bond formation
45
), with the entropic cost of bringing
the two radicals together having been offset partly by the favorable O–Cu interaction and
possibly also by antiferromagnetic electronic coupling between the organic radical and the
copper center. We did not identify an energetically plausible alkylcopper(III) intermediate as
an alternative (see SI).
46
,
47
Notably, a ligand-exchange pathway wherein the organic radical substitutes for a carbazolyl
radical at copper(II) was significantly higher in energy. Alternative mechanisms that involve
initial electron transfer between the copper(II) complex and the organic radical yielding, for
example, either copper(I) and a carbocation or copper(III) and a carbanion, were also ruled
out on energetic grounds via DFT calculations (>+40 kcal mol
–1
).
CONCLUSION
We have provided insight into the critical enantioselective C–N bond-forming step of a
metal-catalyzed N-alkylation process. Specifically, we have synthesized and characterized
two proposed key intermediates, a copper(II) metalloradical and an organic radical, in
the enantioconvergent C–N coupling of carbazoles with racemic tertiary
α
-haloamides. In
the case of the copper(II) metalloradical, the synthesis and structural characterization of
a copper(II) model complex was critical to the identification and characterization of the
persistent copper(II) radical that is present during a catalyzed coupling (but too unstable to
isolate). In the case of the organic radical, DFT calculations led to the recognition that the
organic radical derived from a hindered electrophile might be resistant to radical–radical
homocoupling and therefore accessible as a free radical in solution; this represents a rare
example of an organic radical that is sufficiently stable to characterize and then use in
reaction chemistry.
Upon independent generation and then mixing, the copper(II) metalloradical and the
organic radical do indeed couple at −78 °C in good yield and with the same sense of
enantioselectivity as a catalyzed process. Supported by DFT calculations, we suggest that
the organic radical initially binds to the copper(II) complex via the carbonyl group of the
organic radical, which brings two atoms with substantial spin density, the
α
carbon of the
electrophile and the nitrogen of the carbazole, into close proximity for C–N bond formation.
Lee et al.
Page 10
J Am Chem Soc
. Author manuscript; available in PMC 2023 March 09.
Author Manuscript
Author Manuscript
Author Manuscript
Author Manuscript
The current study leads us to refine our earlier, preliminary mechanistic outline (Figure
1, right) into a more detailed pathway (Figure 2). Previously characterized complex
L*
2
Cu(carb) (
A
) serves as a photoreductant (via excitation to
B
) of the tertiary
α
-
halocarbonyl electrophile R–X to generate R·, a comparatively stable organic radical, and
[L*
2
Cu
II
(carb)]X (
C
1
). A ligand substitution step to install a second carbazolide ligand
with loss of one L* affords the newly characterized intermediate L*Cu
II
(carb)
2
(
C
2
). This
intermediate reacts with R·, through initial tethering of R· to copper(II) via the carbonyl
group, to generate (
C
3
); this pre-organization step sets the stage for coupling the
α
carbon with a copper-bound carbazolide ligand to form the N-alkylation product with
good enantioselectivity, as well as to regenerate copper(I) complex
A
upon binding L*.
Additional studies to further elucidate the mechanisms of metal-catalyzed enantioconvergent
N-alkylations are underway.
Supplementary Material
Refer to Web version on PubMed Central for supplementary material.
ACKNOWLEDGMENTS
Support has been provided by the National Institutes of Health (National Institute of General Medical Sciences:
R01-109194). We are grateful to the Dow Innovation Fund for support of our EPR facility and to the Beckman
Institute X-ray Crystallography Facility. The Resnick Sustainability Institute at Caltech is acknowledged for its
support of enabling facilities. J. M. A acknowledges the National Sciences and Engineering Research Council
(NSERC) of Canada for a graduate research fellowship.
REFERENCES
1. Prier CK; Rankic DA; MacMillan DWC Visible Light Photoredox Catalysis with Transition Metal
Complexes: Applications in Organic Synthesis. Chem. Rev 2013, 113, 5322–5363. [PubMed:
23509883]
2. Meggers E Asymmetric catalysis activated by visible light. Chem. Commun 2015, 51, 3290–3301.
3. Brimioulle R; Lenhart D; Maturi MM; Bach T Enantioselective Catalysis of Photochemical
Reactions. Angew. Chem. Int. Ed 2015, 54, 3872–3890.
4. Jiang C; Chen W; Zheng W-H; Lu H Advances in asymmetric visible-light photocatalysis, 2015–
2019. Org. Biomol. Chem 2019, 17, 8673–8689. [PubMed: 31536083]
5. Kainz QM; Matier CD; Bartoszewicz A; Zultanski SL; Peters JC; Fu GC Asymmetric copper-
catalyzed C–N cross-couplings induced by visible light. Science 2016, 351, 681–684. [PubMed:
26912852]
6. Zhou Z; Li Y; Han B; Gong L; Meggers E Enantioselective catalytic
β
-amination through proton-
coupled electron transfer followed by stereocontrolled radical–radical coupling. Chem. Sci 2017, 8,
5757–5763. [PubMed: 28989615]
7. Nakafuku KM; Zhang Z; Wappes EA; Stateman LM; Chen AD; Nagib DA Enantioselective radical
C–H amination for the synthesis of
β
-amino alcohols. Nat. Chem 2020, 12, 697–704. [PubMed:
32572164]
8. Chen C; Peters JC; Fu GC Photoinduced copper-catalysed asymmetric amidation via ligand
cooperativity. Nature 2021, 596, 250–256. [PubMed: 34182570]
9. Bartoszewicz A; Matier CD; Fu GC Enantioconvergent Alkylations of Amines by Alkyl
Electrophiles: Copper-Catalyzed Nucleophilic Substitutions of Racemic
α
-Halolactams by Indoles.
J. Am. Chem. Soc 2019, 141, 14864–14869. [PubMed: 31496239]
10. Zhang Y-F; Dong X-Y; Cheng J-T; Yang N-Y; Wang L-L; Wang F-L; Luan C; Liu J; Li Z-L;
Gu Q-S; Liu X-Y Enantioconvergent Cu-Catalyzed Radical C–N Coupling of Racemic Secondary
Lee et al.
Page 11
J Am Chem Soc
. Author manuscript; available in PMC 2023 March 09.
Author Manuscript
Author Manuscript
Author Manuscript
Author Manuscript
Alkyl Halides to Access
α
-Chiral Primary Amines. J. Am. Chem. Soc 2021, 143, 15413–15419.
[PubMed: 34505516]
11. Guo W; Zuo L; Cui M; Yan B; Ni S Propargylic Amination Enabled the Access to Enantioenriched
Acyclic
α
-Quaternary
α
Amino Ketones. J. Am. Chem. Soc 2021, 143, 7629–7634. [PubMed:
33988363]
12. Creutz SE; Lotito KJ; Fu GC; Peters JC Photoinduced Ullmann C–N Coupling: Demonstrating the
Viability of a Radical Pathway. Science 2012, 338, 647–651. [PubMed: 23118186]
13. Johnson MW; Hannoun KI; Tan Y; Fu GC; Peters JC A mechanistic investigation of the
photoinduced, copper-mediated cross-coupling of an aryl thiol with an aryl halide. Chem. Sci
2016, 7, 4091–4100. [PubMed: 28044096]
14. Ahn JM; Ratani TS; Hannoun KI; Fu GC; Peters JC Photoinduced, Copper-Catalyzed Alkylation
of Amines: A Mechanistic Study of the Cross-Coupling of Carbazole with Alkyl Bromides. J. Am.
Chem. Soc 2017, 139, 12716–12723. [PubMed: 28817770]
15. Mitani M; Kato I; Koyama K Photoaddition of alkyl halides to olefins catalyzed by copper(I)
complexes. J. Am. Chem. Soc 1983, 105, 6719–6721.
16. Kern J-M; Sauvage J-P Photoassisted C–C coupling via electron transfer to benzylic halides by a
bis(di-imine) copper(I) complex. J. Chem. Soc., Chem. Commun 1987, 546–548.
17. Bagal DB; Kachkovskyi G; Knorn M; Rawner T; Bhanage BM; Reiser O
Trifluoromethylchlorosulfonylation of Alkenes: Evidence for an Inner-Sphere Mechanism by a
Copper Phenanthroline Photoredox Catalyst. Angew. Chem. Int. Ed 2015, 54, 6999–7002.
18. Michelet B; Deldaele C; Kajouj S; Moucheron C; Evano G A General Copper Catalyst for
Photoredox Transformations of Organic Halides. Org. Lett 2017, 19, 3576–3579. [PubMed:
28598630]
19. Ahn JM; Peters JC; Fu GC Design of a Photoredox Catalyst that Enables the Direct Synthesis
of Carbamate-Protected Primary Amines via Photoinduced, Copper-Catalyzed N-Alkylation
Reactions of Unactivated Secondary Halides. J. Am. Chem. Soc 2017, 139, 18101–18106.
[PubMed: 29200268]
20. Tellis JC; Primer DN; Molander GA Single-electron transmetalation in organoboron cross-
coupling by photoredox/nickel dual catalysis. Science 2014, 345, 433–436. [PubMed: 24903560]
21. Zuo Z; Ahneman DT; Chu L; Terrett JA; Doyle AG; MacMillan DWC Merging photoredox with
nickel catalysis: Coupling of
α
-carboxyl sp
3
-carbons with aryl halides. Science 2014, 345, 437–
440. [PubMed: 24903563]
22. Skubi KL; Blum TR; Yoon TP Dual Catalysis Strategies in Photochemical Synthesis. Chem. Rev
2016, 116, 10035–10074. [PubMed: 27109441]
23. While Li(carb) can also serve as a photoreductant of alkyl electrophiles (R–X) to furnish radicals
carb· and R·, as established in a recent study (see ref
14
), Li(carb) is poorly soluble in the solvent
and at the temperature used for the present coupling (toluene, −40 °C).
24. Herein, copper(II) is used as a formal oxidation-state assignment. We believe that there
is significant spin delocalization (see Figure 5) and that a copper(I)-aminyl descriptor also
contributes to the electronic structure, as has been discussed for related phosphine-supported
copper(II)–amido complexes. For example, see reference
25
.
25. Mankad NP; Antholine WE; Szilagyi RK; Peters JC Three-Coordinate Copper(I) Amido and
Aminyl Radical Complexes. J. Am. Chem. Soc 2009, 131, 3878–3880. [PubMed: 19253942]
26. Ferrocene, which is a co-product of the oxidation reaction, co-crystallizes with (NP)Cu
II
(carb’)
(
B
). The thermal instability of (NP)Cu
II
(carb’), along with solubility properties similar to that of
ferrocene, precluded its isolation in analytically pure form. See the Supporting Information.
27. Gomberg M An Instance of Trivalent Carbon. J. Am. Chem. Soc 1900, 22, 757–771.
28. Lankamp H; Nauta WT; MacLean C A new interpretation of the monomer-dimer equilibrium of
triphenylmethyl- and alkylsubstituted-diphenyl methyl-radicals in solution. Tetrahedron Lett 1968,
9, 249–254.
29. Jang ES; McMullin CL; Käß M; Meyer K; Cundari TR; Warren TH Copper(II) Anilides in sp
3
C–H Amination. J. Am. Chem. Soc2014, 136, 10930–10940. [PubMed: 24940616]
30. Dünnebacke D; Neumann WP; Penenory A; Stewen U Über sterisch gehinderte freie Radikale,
XIX. Stabile 4,4
′
,4
′′
-trisubstituierte Triphenylmethyl-Radikale. Chem. Ber 1989, 122, 533–535.
Lee et al.
Page 12
J Am Chem Soc
. Author manuscript; available in PMC 2023 March 09.
Author Manuscript
Author Manuscript
Author Manuscript
Author Manuscript
31. Colle TH; Lewis ES Hydrogen atom transfers from thiophenols to triarylmethyl radicals. Rates,
substituent effects, and tunnel effects. J. Am. Chem. Soc 1979, 101, 1810–1814.
32. Eisenberg DC; Lawrie CJC; Moody AE; Norton JR Relative rates of hydrogen atom H· transfer
from transition-metal hydrides to trityl radicals. J. Am. Chem. Soc 1991, 113, 4888–4895.
33. Mallick S; Maddala S; Kollimalayan K; Venkatakrishnan P Oxidative Coupling of Carbazoles:
A Substituent-Governed Regioselectivity Profile. J. Org. Chem 2019, 84, 73–93. [PubMed:
30521754]
34. Morton JR; Preston KF Atomic parameters for paramagnetic resonance data. J. Magn. Reson 1969,
30, 577–582.
35. At 195 K (−78 °C), homocoupling of the ethyl-substituted radical is estimated to be downhill
by ~6 kcal mol
−1
, whereas homocoupling of the isopropyl-substituted radical is estimated to be
uphill by ~24 kcal mol
−1
. Hence, for the ethyl-substituted radical, a dimer/radical equilibrium
ratio ([R–R]/[R•]) of approximately 27 is estimated at 1 mM concentration at 195 K, a typical
concentration for an EPR experiment; by contrast, for the isopropyl-substituted radical, the organic
radical vastly dominates, even at low temperature.
−1
−1
36. The calculated bond dissociation enthalpy (BDE) of the R–R dimer is predicted to be ~17
kcal mol
−1
for the dimer of the ethyl-substituted radical, while for the dimer of the isopropyl
substituted radical, it is ~–11 kcal mol
−1
, strongly disfavoring dimerization.
−1
−1
37. For comparison, a previously reported C–C coupled dimer, in equilibrium with free phenoxy
radical, exhibits a BDE of ~6 kcal mol
−1
. See: Wittman JM; Hayoun R; Kaminsky W; Coggins
MK; Mayer JM A C–C Bonded Phenoxyl Radical Dimer with a Zero Bond Dissociation Free
Energy. J. Am. Chem. Soc 2013, 135, 12956–12959. [PubMed: 23952108]
38. For related BDE data pertaining to triarylmethane radicals, see: Neumann WP; Uzick W; Zarkadis
AK Sterically hindered free radicals. 14. Substituent-dependent stabilization of para-substituted
triphenylmethyl radicals. J. Am. Chem. Soc 1986, 108, 3762–3770.
39. Wanandi PW; Davis WM; Cummins CC; Russell MA; Wilcox DE Radical Synthesis of a
Heterobinuclear μ-Oxo Complex: Reaction of V(O)(O-
i
-Pr)
3
with Ti(NRAr)
3
(R = C(CD
3
)
2
CH
3
,
Ar = 3,5-C
6
H
3
Me
2
). J. Am. Chem. Soc 1995, 117, 2110–2111.
40. Peters JC; Johnson AR; Odom AL; Wanandi PW; Davis WM; Cummins CC Assembly of
Molybdenum/Titanium μ-Oxo Complexes via Radical Alkoxide C−O Cleavage. J. Am. Chem.
Soc1996, 118, 10175–10188.
41. Stable Radicals. Fundamentals and Applied Aspects of Odd-Electron Compounds; Hicks RG, Ed.;
Wiley: West Sussex, United Kingdom, 2010.
42. In a control experiment, we established that the titanium(IV) byproduct (eq 6) is unreactive toward
L*Cu
II
(carb’)
2
at low temperature.
43. The corresponding addition of the organic radical to the copper(II) model complex (NP)Cu
II
(carb’)
(
B
), using an analogous protocol, furnished the C–N coupling product in 64% yield (see the
Supporting Information).
44. Electronic energies were obtained by DFT structural optimizations in ORCA, with TPSS
functional, def2-TZVP(Cu/Cl/P/O/N/C
α
)-def2-SVP(C,H) basis set, RIJCOSX approximation, and
CPCM(Toluene) solvation model.
α
45. We have calculated a transition state structure for C–N bond formation for the ethyl-substituted
radical (ΔG
‡
= 23.5 kcal mol
−1
at 298 K; 17 kcal mol
−1
at 195 K). See the Supporting
Information.
46. For the ethyl-substituted radical, an alternative pseudo 5-coordinate intermediate structure that
exhibits
κ
2
-O,C-binding of the organic radical was found at significantly higher energy (ΔE =
+14.4 kcal mol
–1
). This structure still features a comparatively long Cu–C
tert
distance of 2.43 Å,
indicative of a weak Cu–C
tert
interaction (see the Supporting Information). No
κ
2
structure could
be identified for the isopropyl-substituted radical.
47. Casitas A; King AE; Parella T; Costas M; Stahl SS; Ribas X Direct Observation of Cu
I
/Cu
III
Redox Steps Relevant to Ullmann-Type Coupling Reactions. Chem. Sci 2010, 1, 326–330.
Lee et al.
Page 13
J Am Chem Soc
. Author manuscript; available in PMC 2023 March 09.
Author Manuscript
Author Manuscript
Author Manuscript
Author Manuscript
Figure 1.
(left) Reaction conditions for our previously reported photoinduced, catalytic C–N coupling
of tertiary
α
-chlorocarbonyl electrophiles with carbazole nucleophiles. (right) Previously
postulated outline of a catalytic cycle for this coupling. Notes: (
R
)-L* is used throughout in
this study; X in
C
may be an inner-sphere or an outer-sphere ligand.
Lee et al.
Page 14
J Am Chem Soc
. Author manuscript; available in PMC 2023 March 09.
Author Manuscript
Author Manuscript
Author Manuscript
Author Manuscript
Figure 2.
Updated outline of a catalytic cycle for the photoinduced, copper-catalyzed
enantioconvergent coupling of a carbazole with a racemic tertiary electrophile.
Lee et al.
Page 15
J Am Chem Soc
. Author manuscript; available in PMC 2023 March 09.
Author Manuscript
Author Manuscript
Author Manuscript
Author Manuscript
Figure 3.
EPR spectra (9.4 GHz, 77 K, toluene). Red trace: Spectrum of a sample taken from a
catalyzed reaction under the conditions shown, after 2 h of irradiation, showing a copper(II)
species. Black trace. Spectrum of the oxidation product of L*
2
Cu(carb), using ferrocenium
hexafluorophosphate (FcPF
6
) as the oxidant. Blue trace: Spectrum of the model complex
(NP)Cu(carb’) (B).
Lee et al.
Page 16
J Am Chem Soc
. Author manuscript; available in PMC 2023 March 09.
Author Manuscript
Author Manuscript
Author Manuscript
Author Manuscript
Figure 4.
Left column: X-band CW EPR spectra (top) and pseudomodulated Q-band ESE-EPR spectra
(bottom) of independently generated L*Cu
II
(carb’)
2
(black trace, see text for details) and
a catalytic mixture using carb’ in place of carb (red trace). Middle: Q-band HYSCORE
spectrum of independently synthesized L*Cu
II
(carb’)
2
at a magnetic field (1192.5 mT)
corresponding to
g
= 2.046. Experimental data are plotted in the top panel, while in
the bottom panel simulations are overlaid in red using parameters in Table S1 over the
experimental data in gray. Right: Q-band HYSCORE spectrum of L*Cu
II
(carb’)
2
generated
by freeze quenching a catalytic mixture at a magnetic field (1192.5 mT) corresponding
to
g
= 2.046. Experimental data are plotted in the top panel, while in the bottom panel
simulations are overlaid in red using parameters in Table S1 over the experimental data
in gray.
14
N simulation parameters: |
A
(
14
N)| = [4, 30, 5] MHz, |
e
2
qQ/h
(
14
N)|
≈
0.5 MHz
(synthesized), 0.3 MHz (catalytic mixture),
η
(
14
N)
≈
1.
Lee et al.
Page 17
J Am Chem Soc
. Author manuscript; available in PMC 2023 March 09.
Author Manuscript
Author Manuscript
Author Manuscript
Author Manuscript
Figure 5.
DFT-predicted Loewdin spin-density plots for: (left) (NP)Cu
II
(carb’). (right) L*Cu
II
(carb’)
2
.
See the Supporting Information.
Lee et al.
Page 18
J Am Chem Soc
. Author manuscript; available in PMC 2023 March 09.
Author Manuscript
Author Manuscript
Author Manuscript
Author Manuscript
Figure 6.
DFT-predicted Loewdin spin-density plot (TPSS, def2-TZVP(N,O,C
tert
)/def2-SVP(C,H),
toluene solvation) and EPR characterization of
α
-amide organic radicals. X-band EPR
data (9.4 GHz, 298 K, toluene) are shown with experimental (black traces) and simulated
(red traces) spectra, showing the response of the spectrum to D-vs-H incorporation in the
isopropyl substituent. Simulation parameters:
g
iso
= 2.0031,
A
H
α
= 25.9 MHz,
A
H
β
1
= 12.2
MHz,
A
H
β
2
= 11.6 MHz,
A
H(ortho1)
=
A
H(ortho2)
= 12.9 MHz,
A
H(para1)
=
A
H(para2)
= 4.9
MHz. See the Supporting Information.
Lee et al.
Page 19
J Am Chem Soc
. Author manuscript; available in PMC 2023 March 09.
Author Manuscript
Author Manuscript
Author Manuscript
Author Manuscript
Figure 7.
DFT-optimized structure (two views) featuring a complex between L*Cu
II
(carb)
2
and the
carbonyl oxygen of the organic radical (alkyl =
i
-Pr; Ar = 4-chlorophenyl), including the
energy of bring the two radicals together (minimized as a broken-symmetry singlet; TPSS,
def2-TZVP(Cu,Cl,P,N,O,C
tert
)/def2-SVP(C,H), toluene solvation), key bond distances, and
Loewdin spin populations.
Lee et al.
Page 20
J Am Chem Soc
. Author manuscript; available in PMC 2023 March 09.
Author Manuscript
Author Manuscript
Author Manuscript
Author Manuscript
Scheme 1.
Synthesis of model copper(II) complex (NP)Cu
II
(carb’). Fc = ferrocenium.
Lee et al.
Page 21
J Am Chem Soc
. Author manuscript; available in PMC 2023 March 09.
Author Manuscript
Author Manuscript
Author Manuscript
Author Manuscript