Photogenerated Ni(I)–Bipyridine Halide Complexes: Structure-
Function Relationships for Competitive C(sp
2
)–Cl Oxidative
Addition and Dimerization Reactivity Pathways
David A. Cagan
†
,
Division of Chemistry and Chemical Engineering, Arthur Amos Noyes Laboratory of Chemical
Physics, California Institute of Technology, Pasadena, California 91125, United States
Daniel Bím
†
,
Division of Chemistry and Chemical Engineering, Arthur Amos Noyes Laboratory of Chemical
Physics, California Institute of Technology, Pasadena, California 91125, United States
Brendon J. McNicholas
,
Division of Chemistry and Chemical Engineering, Arthur Amos Noyes Laboratory of Chemical
Physics, California Institute of Technology, Pasadena, California 91125, United States
Nathanael P. Kazmierczak
,
Division of Chemistry and Chemical Engineering, Arthur Amos Noyes Laboratory of Chemical
Physics, California Institute of Technology, Pasadena, California 91125, United States
Paul H. Oyala
,
Division of Chemistry and Chemical Engineering, Arthur Amos Noyes Laboratory of Chemical
Physics, California Institute of Technology, Pasadena, California 91125, United States
Ryan G. Hadt
Division of Chemistry and Chemical Engineering, Arthur Amos Noyes Laboratory of Chemical
Physics, California Institute of Technology, Pasadena, California 91125, United States
Abstract
We report the facile photochemical generation of a library of Ni(I)–bpy halide complexes
(Ni(I)(
R
bpy)X (R = t-Bu, H, MeOOC; X = Cl, Br, I) and benchmark their relative reactivity
toward competitive oxidative addition and off-cycle dimerization pathways. Structure-function
relationships between ligand set and reactivity are developed, with a particular emphasis
on rationalizing previously uncharacterized ligand-controlled reactivity toward high energy
and challenging C(sp
2
)–Cl bonds. Through a dual Hammett and computational analysis, the
mechanism of the formal oxidative addition is found to proceed through an SNAr-type pathway,
Corresponding Author:
Ryan G. Hadt
Division of Chemistry and Chemical Engineering, Arthur Amos Noyes Laboratory of
Chemical Physics, California Institute of Technology, Pasadena, California 91125, United States; rghadt@caltech.edu.
†
Author Contributions
Contributed equally
Supporting Information
.
The Supporting Information is available free of charge on the ACS Publications website at DOI:
Experimental and computational methods, UV–vis/photochemical data, NMR and EPR spectra, calculated spectra/properties, and
additional comments.
HHS Public Access
Author manuscript
Inorg Chem
. Author manuscript; available in PMC 2023 July 10.
Published in final edited form as:
Inorg Chem
. 2023 June 19; 62(24): 9538–9551. doi:10.1021/acs.inorgchem.3c00917.
Author Manuscript
Author Manuscript
Author Manuscript
Author Manuscript
consisting of a nucleophilic two-electron transfer between the Ni(I) 3
d
(z
2
) orbital and the C
aryl
–Cl
σ
* orbital, which contrasts the mechanism previously observed for activation of weaker C(sp
2
)–
Br/I bonds. The bpy substituent provides a strong influence on reactivity, ultimately determining
whether oxidative addition or dimerization even occur. Here we elucidate the origin of this
substituent influence as arising from perturbations to the effective nuclear charge (Z
eff
) of the Ni(I)
center. Electron donation to the metal decreases Z
eff
, which leads to a significant destabilization
of the entire 3
d
orbital manifold. Decreasing the 3
d
(z
2
) electron binding energies leads to a
powerful two-electron donor to activate strong C(sp
2
)–Cl bonds. These changes also prove to have
an analogous effect on dimerization, with decreases in Z
eff
leading to more rapid dimerization.
Ligand-induced modulation of Z
eff
and the 3
d
(z
2
) orbital energy is thus a tunable target by which
the reactivity of Ni(I) complexes can be altered, providing a direct route to stimulate reactivity
with even stronger C–X bonds and potentially unveiling new ways to accomplish Ni-mediated
photocatalytic cycles.
Graphical Abstract
1. Introduction
Photoredox catalysis has captivated the fields of organic and inorganic chemistry, with
nickel(II)–bipyridine (bpy) aryl halide complexes retaining a prominent place as the metal-
ligand scaffold of choice for numerous cross-coupling reactivity pathways.
1
–
11
Among
these, C–C and C–heteroatom couplings have been facilitated by Ni(II)–bpy complexes
through either the use of an external photosensitizer (e.g., Ir(ppy)
3
) or via direct excitation
of the Ni(II)–bpy aryl halide complex.
12
–
18
Due to the diverse reactivity and intriguing
photophysics of these complexes, much interest has been placed on understanding the
underlying photophysical and thermal processes involved in photoredox mediated cross-
coupling reactivity.
19
–
28
While originally thought to proceed through a Ni(0)/Ni(II) cycle,
6
recent work has instead
supported a Ni(I)/Ni(III) cycle in the direct excitation pathway (Figure 1A).
23
,
29
,
30
Through
analysis of a library of Ni(II)–bpy aryl halide complexes, we have revealed that excited-state
Ni(II)–C
aryl
bond homolysis from the
S
= 0 square planar Ni(II) ground state features a key
ligand-to-metal charge transfer (LMCT) process. This LMCT results in electron excitation
between the Ni–aryl
σ
and
σ
* orbitals, which lowers the bond order from one to zero,
resulting in repulsive homolytic bond cleavage and the generation of an aryl radical and a
three-coordinate Ni(I)–bpy halide species.
27
Cagan et al.
Page 2
Inorg Chem
. Author manuscript; available in PMC 2023 July 10.
Author Manuscript
Author Manuscript
Author Manuscript
Author Manuscript
Related Ni(I)–bpy halide complexes have been prepared by alternate methods, including
pulse radiolysis, electrolysis, and independent synthesis.
30
–
33
Importantly, these compounds
have demonstrated potency for the activation of aryl halide substrates. Following the
work by Vicic
et al
. on Ni(I)–terpyridine complexes,
34
–
36
Bird and MacMillan
et al
.
reported a nickel(I)(4,4
′
-di-
tert
-butyl bipyridine)bromide complex (Ni(I)(
t-Bu
bpy)Br) that
exhibited rapid reactivity toward aryl iodides with second-order rate constants of ~10
4
M
−1
s
−1
.
31
The activation of C(sp
2
)–Br substrates was demonstrated by Doyle
et al
. using
a nickel(I)(diethyl-2,2
′
-bipyridine-4,4
′
-dicarboxylate)chloride complex, Ni(I)(
EtOOC
bpy)Cl.
Activation of the stronger aryl bromide bond proceeded with slower, but still catalytically
relevant, rate constants of ~10
−2
– 10
1
M
−1
s
−1
.
30
Notably, C(sp
2
)–Cl bonds could not be
activated by this Ni(I) complex.
Careful kinetic analysis has been employed to understand the mechanism of this
reactivity;
30
,
37
the Ni(I) species has been suggested to undergo two-electron oxidative
addition to form a short-lived five-coordinate Ni(III)–bpy aryl dihalide complex. This
Ni(III) intermediate rapidly decays by comproportionation with another equivalent of Ni(I),
forming Ni(II)–bpy aryl halide and Ni(II)–bpy dihalide complexes (Figure 1A).
30
Using
three-coordinate pyridinophane Ni(I) model complexes, Mirica
et al
. have also provided
evidence for this Ni(I)/Ni(III) oxidative addition pathway.
29
Therefore, these and related
studies have argued for the importance of three-coordinate Ni(I)–bpy halide complexes to
facilitate the oxidative addition step, without which the cross-coupling catalytic cycle would
not close.
21
,
23
,
38
,
39
While the reactivity of Ni(I)–bpy complexes has proven desirable for organic synthesis, their
solution-phase stabilities vary widely, limiting their scope. Numerous Ni(I) dimerization
products have been characterized, and the resultant polypyridyl species are no longer
reactive toward oxidative addition with aryl halides.
30
,
32
,
40
,
41
Dimerization thereby acts as
an off-cycle sink with diminished catalytic activity (Figure 1A). Hazari
et al
. independently
synthesized and characterized the dimeric [Ni(I)(
t-Bu
bpy)Cl]
2
complex. Once formed, the
compound was stable in solution and exhibited no oxidative addition reactivity even with
weak C(sp
2
)–I bonds.
32
The related [Ni(I)(
t-Bu
bpy)Br]
2
complex was studied by Bird and
MacMillan
et al
. and similarly showed no oxidative addition reactivity with aryl iodides.
31
Additionally, tetrameric [Ni(I)(
EtOOC
bpy)Cl]
4
and dimeric [Ni(I)(
EtOOC
bpy)Cl]
2
complexes
were studied by Doyle
et al
.
30
In solution, both species can dissociate to form the
monomeric Ni(I)(
EtOOC
bpy)Cl complex mentioned above. This three-coordinate species was
reported to exist in solution-phase equilibria with Ni(II)Cl
2
and Ni(0)(
EtOOC
bpy)
2
. However,
the oxidative addition reactivity was favored from the Ni(I) complex, not the Ni(0) species,
and no reactivity was found for the dimeric or tetrameric species.
Understanding the interplay between the oxidative addition and dimerization/
oligomerization pathways available to three-coordinate Ni(I)–bpy halide complexes
is critical for optimized ground-state and metallaphotoredox catalysis and reaction
development (Figure 1B). There is also a fundamental knowledge gap related to specific
structure-function relationships underlying the elementary oxidative addition step in Ni(I)–
bpy complexes, a key component of the scope of bond activations that can be achieved. To
the best of our knowledge, no systematic study has either compared the oxidative addition
Cagan et al.
Page 3
Inorg Chem
. Author manuscript; available in PMC 2023 July 10.
Author Manuscript
Author Manuscript
Author Manuscript
Author Manuscript
reactivity and solution-phase stability of Ni(I)–bpy complexes or rationalized the relative
rates of oxidative addition upon structural perturbation.
Here we accomplish both of these aims through the facile photogeneration of a series of
Ni(I)(
R
bpy)X species (R = t-Bu, H, MeOOC; X = Cl, Br, I; Figure 1C). We provide direct
evidence for C(sp
2
)–Cl bond activation with Ni(I)–bpy complexes, which thus far has only
been implied as part of catalytic cycles.
33
,
42
Furthermore, we broadly detail the kinetics
and mechanisms of their oxidative addition reactivity toward C(sp
2
)–X (X = Cl, Br, I)
bonds, and their thermal barriers for dimerization. Results reported herein implicate key
ligand effects on the electronic structure of the Ni(I)–bpy halide complexes that result in
tunable reactivity and variable solution-phase lifetimes. Ultimately, the ability of Ni(I)–bpy
complexes to activate challenging C(sp
2
)–Cl bonds stems from bpy-induced modulation of
the effective nuclear charge (Z
eff
) of the Ni(I) center. Related analyses reported here have
afforded a molecular orbital-based picture of oxidative addition reactivity, which provides
specific electronic structure benchmarks that need to be achieved to activate strong C(sp
2
)–
X bonds and opens the door for the targeted synthesis of next-generation cross-coupling
catalysts.
2. Results and Analysis
2.1. Photochemical Synthesis and Spectroscopic Characterization of Ni(I)–bpy halide
complexes
Parent four-coordinate Ni(II)(
R
bpy)(
o
-tolyl)X (R =
tert
-butyl, H, MeOOC; X = Cl, Br,
I) complexes were synthesized according to a previous report.
27
Three-coordinate Ni(I)
(
R
bpy)X species, which are important reaction intermediates in photoredox cross-coupling
catalysis, can be accessed directly from these precursors by air- and moisture-free irradiation
at 370 nm using purple LEDs. Light excitation drives a key LMCT process that results in
Ni(II)–C
aryl
excited-state bond homolysis; this shorter wavelength enhances the quantum
yield of Ni(I)(
R
bpy)X generation relative to more commonly used, longer wavelength blue
light sources (Table S2). Typical irradiation times were ~30 – 60 minutes; irradiation time
variations among complexes depended on the rate of photo-driven decay of the starting
Ni(II) complex and the rate of Ni(I) decomposition. Global kinetic analysis provides strong
support for a first-order process in which the Ni(II)–bpy aryl halide parent complex is
cleanly photolyzed to the Ni(I)–bpy halide (for more details, see Supporting Information
Section S1.5, including Figures S6–S10).
To systematically probe the reactivity and stability of the photochemically generated
Ni(I) complexes, the bpy ligand was varied with electron donating
tert
-butyl groups
(Ni(I)(
t-Bu
bpy)Cl,
1-Cl
), electronically neutral H-atoms (Ni(I)(
H
bpy)Cl,
2-Cl
), and electron
withdrawing methyl ester groups (Ni(I)(
MeOOC
bpy)Cl,
3-Cl
). The halide was also
varied by first synthesizing Ni(II)(
t-Bu
bpy)(
o
-tolyl)Br and Ni(II)(
t-Bu
bpy)(
o
-tolyl)I parent
complexes, followed by irradiation to form Ni(I)(
t-Bu
bpy)Br (
1-Br
) and Ni(I)(
t-Bu
bpy)I (
1-I
),
respectively. Note that
1-Br
is the same as that examined by Bird and MacMillan
et al
. by
pulse radiolysis,
31
while
3-Cl
is similar to that studied by Doyle
et al
.,
30
the only difference
being the appended methyl ester vs ethyl ester substituents, following our previous study.
27
Cagan et al.
Page 4
Inorg Chem
. Author manuscript; available in PMC 2023 July 10.
Author Manuscript
Author Manuscript
Author Manuscript
Author Manuscript
Having direct access to the Ni(I)–bpy halide complexes via photolysis allows for their
characterization using UV-vis electronic absorption and EPR spectroscopies (Figure 2A and
2C, respectively). Irradiation of Ni(II) parent compounds and spectral analysis were done
in tetrahydrofuran (THF). The primary UV-vis absorption band in the ~550 nm – 900 nm
region of the three-coordinate Ni(I)(
R
bpy)X species shifts strongly with changes to the bpy
ligand from 660 nm (
1-Cl
) to 805 nm (
3-Cl
). Moderate shifts are also observed when
changing the halide (640 nm for
1-I
to 660 nm for
1-Cl
, Figure 2A). To further corroborate
the spectral assignment of UV-vis peak to the Ni(I)–bpy halides, we conducted reductive
spectroelectrochemistry on Ni(
t-Bu
bpy)Cl
2
to form Ni(
t-Bu
bpy)Cl; the resultant spectrum is
identical to that of photogenerated
1-Cl
(Figure 2B).
From TDDFT calculated spectra (bottom of Figure 2A; see Computational Details in
Supporting Information Section S2.1), the intensity of the primary absorption band is
attributed to Ni(I)-to-bpy metal-to-ligand charge transfer (MLCT) transitions (3
d
(xz/yz)
→
π
*(1); note these orbitals are labeled according to their parallel (||) or perpendicular
(
⊥
) orientation to Ni
‒
halide bond – see details in Supporting Information Section S2.4).
Consistent with the experimental spectra, smaller shifts are observed for variations in the
halide; the MLCT energies are more sensitive to bpy variation, with an energy trend of
1-Cl
>
2-Cl
>
3-Cl
. Through analyses of the molecular orbital energy diagrams for these
complexes (Figures S40–S44 and Table S7), the MLCT energy shift arises mainly from
the stabilization of the bpy
π
* acceptor orbitals as a function of electron-with-drawing
substituents.
While room temperature EPR analysis provided no resolvable signals, likely due to rapid
spin relaxation times (Figure S31), spectra taken at 5 K in 2-methyl tetrahydrofuran (2-
MeTHF) after irradiation provided signals characteristic of S = 1/2 Ni(I) species (Figure
2C), with
g
values in the range of
g
z
= ~2.22 – 2.37,
g
x
= ~2.04 – 2.17,
g
y
= ~2.07
– 2.20, and
g
iso
= ~2.12 – 2.19 (Figure 2C; Table 1). The axial
g
tensor values are
overall consistent with a single unpaired electron in the Ni(I) 3
d
(x
2
-y
2
) orbital. The EPR
spectra and
g
values of compounds
1-Cl/Br
and
3-Cl
are congruent with previous reports
for Ni(I) halide species with aromatic ligand backbones (e.g., phenanthroline, bipyridine,
and bis(pyrazolyl)pyridines).
30
,
39
,
42
–
45
The
14
N hyperfine values employed in the
3-Cl
simulation are of comparable magnitude to those previously reported for Ni(I)-neocuproine
complexes (0, 50, and 170 MHz).
32
Sufficiently high concentration samples of
2-Cl
were
precluded due to a precipitation (dimerization) pathway (
vide infra
, Section 2.4); as such,
no EPR signal could be detected. The intermediate peak at ~310 mT in
1-Cl
/
Br
/
I
is likely
attributable to THF solvent coordination (see Supporting Information Section 2.7), as it does
not appear in previous spectra recorded in toluene but does appear in toluene:THF solvent
mixtures and in neat 2-MeTHF.
42
,
43
These published data share excellent agreement with
the spectra collected and presented in Figure 2C. Importantly, the relative intensities of
the
g
z
and intermediate peaks remain approximately constant with variations in photolysis
time, indicating there is not a second Ni(I) species with a distinct kinetic profile (Figures
S33–S35). The DFT calculated
g
values are in general agreement with the experimental
data, with the exception of the g
z
value for 1-I and significantly more predicted rhombicity
throughout (computed
g
x
and
g
y
in Table 1).
Cagan et al.
Page 5
Inorg Chem
. Author manuscript; available in PMC 2023 July 10.
Author Manuscript
Author Manuscript
Author Manuscript
Author Manuscript
We also note the presence of additional complex signals in the half-field region of the
concentrated samples of
1-Cl
in the low-temperature (5 K) EPR (Figure S33–S35). Variable
temperature (VT) UV-vis experiments indicate that these are attributable to a reversible
concentration- and temperature-dependent speciation (see Supporting Information Section
S1.9). That is, these additional species form rapidly upon freezing the EPR samples. Future
studies should consider speciation changes that can occur upon freezing samples for low
temperature spectroscopic characterization. That said, under standard catalysis conditions
(~0.2 mM, room temperature) studied herein, the Ni(I)–bpy halide complexes corresponding
to the S = 1/2 signals are the dominant species after irradiation is terminated (~95% Ni(I)
by VT UV-vis analysis), with the additional species (~5%) being the starting Ni(II)–bpy aryl
halide complex.
2.2. Oxidative Addition Kinetics with 2-Chloro-toluene
Having demonstrated near quantitative conversion of parent Ni(II) complexes to the three-
coordinate Ni(I) complexes, we sought to gauge their relative reactivity toward oxidative
addition. As expected, room temperature addition of excess 2-
bromo-
toluene or 2-
iodo
-
toluene (0.2 mL) to all photogenerated Ni(I) complexes studied herein resulted in immediate
color changes, with loss of the Ni(I)-bpy MLCT features observed in the UV-vis spectra
(Figures S14–S15). The diagnostic S = 1/2 Ni(I) EPR signal is also quenched upon
addition of aryl halide (Figure S36). Again, reactivity with C(sp
2
)–I and C(sp
2
)–Br bonds is
consistent with previous reports.
30
,
31
However, interestingly, many of the Ni(I) intermediates
studied herein also react with 2-
chloro
-toluene – the exception being
3-Cl
. As a control,
addition of 0.2 mL of toluene did not result in UV-vis spectral changes, further implicating
the reactivity of the Ni(I) species with the C(sp
2
)–Cl bond. The lack of reactivity of
3-Cl
is
also significant and is directly related to the bpy ligand, as discussed further in Section 2.3.
Taking
1-Cl
as a representative compound, we sought to confirm the C(sp
2
)–Cl reactivity
by stoichiometric
1
H NMR studies in
d
8
-THF. However, to facilitate reaction turnover at a
rate greater than the decay of the Ni(I) intermediate, a large excess of 2-
chloro
-toluene is
necessary, which over-whelms the
1
H NMR analysis. Using fewer equivalents of aryl halide,
the reaction between 2-
chloro
-toluene and the photogenerated Ni(I) species is too slow and
precludes definitive product speciation assignments (Supporting information Section S1.7).
This can be circumvented using a fluorinated aryl halide and
19
F NMR analysis. As depicted
in Scheme 1, addition of 20 μL of 2-
chloro
-
α
,
α
,
α
-trifluorotoluene (~200-fold excess,
19
F
NMR peak at −63 ppm) to a solution of
1-Cl
afforded a new peak in the
19
F NMR
spectrum at −58 ppm. Independent synthesis
27
confirms this peak originates from the four-
coordinate complex, Ni(II)(
t-Bu
bpy)(CF3Ph)Cl (Figure S18). Thus, this experiment provides
direct evidence of a Ni(I)–bpy halide complex activating an aryl chloride substrate. We
now describe the relative reactivity of the Ni(I)–bpy complexes toward C(sp
2
)–Cl oxidative
addition.
Pseudo
-first order kinetics investigations with quantitative addition of excess 2-
chloro
-
toluene to the photogenerated Ni(I) complexes were carried out by monitoring the fast decay
of the primary MLCT absorption feature via UV-vis spectroscopy (Figure 3A and S20A).
The natural logarithm of the normalized peak absorbance varies linearly with time for ~3–5
Cagan et al.
Page 6
Inorg Chem
. Author manuscript; available in PMC 2023 July 10.
Author Manuscript
Author Manuscript
Author Manuscript
Author Manuscript
half-lives, depending on the magnitude of starting Ni(I) absorption (Figure 3B and S20B).
The slope of this correlation provides
k
obs
values that vary linearly with 2-
chloro
-toluene
concentration (Figure 3C and S20C), yielding second-order rate constants for oxidative
addition,
k
OA
(M
−1
s
−1
) (Table 2). The bimolecular rate constants are on the order of 10
−2
M
−1
s
−1
, six orders of magnitude slower than the reaction of
1-Br
with aryl iodides and two
orders of magnitude slower than the reaction of Ni(I)(
EtOOC
bpy)Cl with aryl bromides.
30
,
31
This reduction in rate constant for the activation of 2-
chloro
-toluene is attributable to the
increased bond dissociation free energy (BDFE) of the carbon–halogen bond (BDFE of
C(sp
2
)–Cl > C(sp
2
)–Br > C(sp
2
)–I, Table S15) and a steric effect of the
ortho
-methyl group
(Figure 4).
Analysis of ligand effects reveals that changes to the halide manifest in minor changes
in
k
OA
(Figure S20). These results agree well with the electronic effects of the halide as
predicted by the Hammett parameters (
σ
p
= 0.23 for Cl and Br, and 0.18 for I).
46
More
pronounced effects are observed upon variation of the bpy substituent. Substitution of the
electron donating
tert
-butyl groups (
σ
p
= –0.20) for hydrogens (
σ
p
= 0.0) in going from
1-Cl
to
2-Cl
results in a two-fold decrease in
k
OA
from 7.2 ± 0.2 × 10
−2
M
−1
s
−1
to 3.2 ± 0.2 ×
10
−2
M
−1
s
−1
. Furthermore, introduction of the electron with-drawing methyl ester groups
(
σ
p
= 0.45) eliminates C(sp
2
)–Cl oxidative addition reactivity altogether. As described
further below in Section 2.3, these differences in oxidative addition reactivity can be traced
directly to ligand-induced differences in Z
eff
on the metal, as changes in Z
eff
tune the energy
of the redox active molecular orbital (RAMO) involved in the oxidative addition reaction.
Follow-up work is currently underway to push the limits of Ni(I)-facilitated oxidative
addition reactivity by substituent-based modulation of Z
eff
.
2.3. Oxidative Addition Mechanistic Investigations
Having examined the effect of the substituents on the Ni(I)–bpy halides toward the C(sp
2
)–
Cl oxidative addition reactivity, we next turned to the kinetic dependence of the substituents
on the aryl chloride substrates. We performed a Hammett analysis on the reactivity of series
of para-substituted aryl chlorides with
1-Cl
. The rate constants (
k
OA
) of the formal Ni(I)/
Ni(III) oxidative addition process increased by two orders of magnitude when switching
from electron donating to electron withdrawing substituents (
σ
p
= − 0.27 to 0.50, Figure
4). We further found the slope (
ρ
) of the plot of ln(
k
x
/
k
H
) vs
σ
p
was ~5, indicative of an
S
N
Ar type activation
47
of the C(sp
2
)– Cl bond, wherein the Ni(I) metal center acts as a
nucleophile, attacking the polarized C(sp
2
)– Cl bond at the electron-deficient carbon. While
an activation pathway that proceeds by single electron transfer (SET) also has a fairly large
ρ
value (~4),
48
the reduction potentials of the aryl chlorides are too negative (experimentally
found to be less than –3.2 V vs Fc
+
/Fc, calculated as –3.5 V vs Fc
+
/Fc; for more details
see Supporting Information Section S1.6 and S2.6) to be accessed by the Ni(I)–bpy halides
(calculated Ni(II)/Ni(I) potential of 0.2 V vs Fc
+
/Fc).
49
Interestingly, we note that
ρ
~5 represents a change in the specific mechanism of C(sp
2
)– X
bond activation by Ni(I) halides. Aryl bromides and aryl iodides exhibit Hammett slope
values of
ρ
~2–3, typical of a concerted oxidative addition pathway,
30
,
31
,
50
,
51
but the
Cagan et al.
Page 7
Inorg Chem
. Author manuscript; available in PMC 2023 July 10.
Author Manuscript
Author Manuscript
Author Manuscript
Author Manuscript
activation of the stronger, more polarized C(sp
2
)–Cl bond is better described as nucleophilic
aromatic substitution.
DFT calculations were used to further detail the mechanism of oxidative addition (i.e., step-
wise vs concerted two-electron nucleophilic attack by Ni) and to identify the key frontier
molecular orbitals involved in this elementary reaction step. Oxidative addition of 2-
chloro
-
toluene to a three-coordinate Ni(I)(
R
bpy)X species yields a five-coordinate Ni(III)(
R
bpy)(
o-
tolyl)(Cl)(X) intermediate. Following the initial oxidative addition step, comproportionation
of the Ni(III) complex with an additional Ni(I)(
R
bpy)X species can take place with a
calculated Gibbs free energy of ~–23 kcal mol
−1
, which produces a considerable driving
force for the overall reaction. Comproportionation can occur via transfer of a halide from
the five-coordinate intermediate, which is readily accessible through a barrierless transition
state (Figure 5A). This step therefore results in the regeneration of a parent Ni(II)
‒
bpy
aryl halide complex and the formation of a Ni(II) dihalide (i.e., the speciation determined
experimentally by Diao
et al
. and Doyle
et al
. and
19
F NMR and UV-vis studies presented
herein (
vide supra
, Section 2.2)).
30
,
39
The rate-determining step of the overall reaction is found to be oxidative addition, with
calculated Gibbs free-energy barriers of ~21 – 23 kcal mol
−1
, consistent with the slow
reaction rates at room temperature and the requirement of a large excess of 2-
chloro
-toluene
(Figure 5A). The computed free energy barriers vary by only ~1.3 kcal mol
−1
for
1-Cl
,
1-Br
,
1-I
, and
2-Cl
(i.e., within the general accuracy obtainable with DFT calculations).
55
,
56
However, this observation is still in accord with the differences in
k
OA
for the compounds
observed experimentally (Table 2). Furthermore, the highest computed activation free energy
is observed for
3-Cl
(~23 kcal mol
−1
, red in Figure 5A), consistent with its lack of reactivity
with 2-
chloro
-toluene.
We have also computationally reproduced the Hammett analysis for the reaction of
1-Cl
with 4-substituted aryl chlorides and have found excellent agreement with experiment
(calculated
ρ
~5.8; see Supporting Information Section S2.6). Furthermore, the calculated
bond lengths in the transition state show that the new Ni–Cl bond is ~0.3 Å longer than
the new Ni–C bond (potentially implying the Ni–C bond is formed prior to the Ni–Cl bond
and corroborating an SNAr-type mechanism, Figure 5B). Additionally, we observe that the
experimental rate constants trend linearly and positively with the Löwdin atomic charges,
the Mulliken atomic charges, and the natural population analysis (NPA) charges at the
carbon in the Ar–Cl bond, but they do not trend well with the energy of the LUMO of the
aryl chlorides (Figure S66). This observation is also indicative of an S
N
Ar mechanism as the
electrophilicity of this carbon is related to the rate and not the reduction potential of the aryl
chlorides (as would be expected for SET).
57
To further elucidate the origin of the different reactivity profiles for the different Ni(I)
species, we analyzed the initial oxidative addition step in the context of intrinsic bond orbital
(IBO) progression.
58
,
59
IBO analysis allowed for the identification of orbital changes (
δ
-orb)
along the intrinsic reaction coordinate (IRC), wherein the separated Ni(I) and aryl halide
reactants come together to form the five-coordinate Ni(III) adduct. Although the unpaired
electron is located in the Ni 3
d
(x
2
-y
2
) orbital (
cf
. EPR analysis in Section 2.1), the largest
Cagan et al.
Page 8
Inorg Chem
. Author manuscript; available in PMC 2023 July 10.
Author Manuscript
Author Manuscript
Author Manuscript
Author Manuscript
δ
-orb effects are observed in the doubly-occupied Ni 3
d
(z
2
) orbital and C
aryl
–Cl
σ
orbital.
These are transformed into new Ni
‒
C
aryl
and Ni
‒
Cl
σ
bonds, respectively (Figure S39).
This change in bonding is the outcome of two-electron transfer from the occupied Ni 3
d
(z
2
)
orbital into the virtual C
aryl
– Cl
σ
* orbital (Figure 5C). Direct observation of Ni-to-(C
aryl
–
Cl)
σ
* backbonding at the transition state is evidenced by significant mixing of filled 3
d
orbital character into the unoccupied C
aryl
–Cl
σ
* orbital along the IRC (Figure 5C, middle).
For example, from
1-Cl
to
2-Cl
to
3-Cl
, the Löwdin Ni 3
d
character varies from 24.0 to 29.4
to 39.4. Previous analysis on Pd catalyzed aryl chloride activation found that the increase
in Pd backbonding to the C
aryl
– Cl
σ
* orbital was related to an increase in the barrier for
the reaction.
60
This same trend is seen in our Ni-based C(sp
2
)–Cl bond activation. Thus,
backbonding in the transition state may play an important role in activating stronger C(sp
2
)–
Cl bonds by Ni(I) species. As discussed further below, this backbonding will increase with
less negative electron binding energies on the metal.
At the transition state, the
α
- and
β
-type orbitals are transformed in conjunction, suggesting
a two-electron nucleophilic attack by the Ni(I) 3
d
(z
2
) orbital on the C
aryl
atom, yielding a
Ni(III) intermediate. This concerted two-electron transfer is supported by a small change
of Löwdin spin density on the Ni center throughout the reaction (~1 unpaired electron
throughout; see inset table in Figure 5C); one-electron transfer or stepwise two-electron
transfer would reveal more substantial metal-based spin density changes along the IRC. The
small increase in the spin density at the transition state arises from the slightly misaligned
attack initiated by
β
Ni 3d(z
2
) orbital, which is ~0.8 eV higher in energy than the
α
Ni 3d(z
2
) orbital due to spin polarization arising from the presence of five vs four 3
d
electrons in
α
and
β
manifolds, respectively (Table S7). That C
aryl
–Cl bond-breaking is
rate-determining is consistent with the more facile oxidative addition activation of 2-
bromo
-
toluene and 2-
iodo
-toluene substrates (e.g., their reactivity is observed even with
3-Cl
due to
their weaker C
aryl
–halogen
σ
bonds and lower energy C
aryl
–Cl
σ
* orbitals, which facilitates
backbonding from the metal).
From the preceding analysis, the reactivity of individual Ni(I) complexes should correlate
with the energies of the Ni 3
d
(z
2
) orbitals. Indeed, the calculated Ni(I)–bpy 3
d
(z
2
) orbital
energies trend with the
k
OA
rate constants (Figure 5C and Table 2). For example, the most
and least reactive complexes,
1-Cl
and
3-Cl
, have
β
3
d
(z
2
) orbital energies of −4.92 eV
and −5.44 eV, respectively (Δ = −0.52 eV). There is also a linear trend between the energy
of the 3
d
(z
2
) orbital and the bpy
σ
p
Hammett parameters for
1-Cl
,
2-Cl
, and
3-Cl
(Figure
S45). The less-negative electron binding energies of
1-Cl
will increase nucleophilicity and
propensity for a two-electron reduction. As discussed below, this effect on the 3
d
orbital
energies is not necessarily specific to 3
d
(z
2
) alone, specifically because the bpy-based
σ
interaction only involves the torus of the 3
d
(z
2
) orbital. Rather, the bpy ligand ultimately
tunes the entire 3
d
orbital manifold via changes in Z
eff
of the metal, making the ligand
substitution effects a key predictor of reactivity. In further support of these changes, the
calculated Ni(I) 1
s
and 2
p
(x,y,z) orbital energies of
1-Cl
,
2-Cl
, and
3-Cl
all trend linearly
with bpy-based Hammett parameters and oxidative addition rate constants, with slopes
similar to the 3
d
orbitals (Figure S45–S46). Moving forward, the calculated 3
d
orbital
energies of three-coordinate Ni(I) intermediates will be a useful predictor for oxidative
Cagan et al.
Page 9
Inorg Chem
. Author manuscript; available in PMC 2023 July 10.
Author Manuscript
Author Manuscript
Author Manuscript
Author Manuscript
addition reactivity and will bracket relative reactivity for specific C(sp
3
)–X and C(sp
2
)–X
bonds.
2.4. Thermodynamics of the Dimerization of Ni(I)–bpy halides
Having established the reactivity of the Ni(I) complexes toward oxidative addition, we
sought to better understand their general stability in solution. Their stability is of particular
interest, as unreactive halide-bridged dimers have been proposed to form in bpy and related
systems.
30
–
32
,
40
Photogeneration of the Ni(I) compounds was again achieved using 370 nm
LEDs in THF. Even in the absence of aryl halide, the characteristic MLCT band of the
Ni(I)–bpy halide complexes decayed over time; this decay was accelerated by increased
temperature. In all cases, this decomposition results in the formation of a precipitate.
However, we found that the precipitate of complexes
1-Cl
/
Br
exhibited slight solubility
in THF, likely owing to the bulky, non-polar
tert
-butyl substituents on the bipyridine.
Taking
1-Cl
as a representative complex, we thus collected the precipitation product and
analyzed it by UV-vis,
1
H NMR, and EPR (in THF,
d
8
-THF, and 2-MeTHF, respectively;
Figures S28–S30). We found that its spectra matched that of independently synthesized
[Ni(I)(
t-Bu
bpy)Cl]
2
, suggesting that the primary thermal decomposition product of the
Ni(I)–bpy halides is their halide-bridged dimers (see Supporting Information Section S1.8).
Furthermore, this thermal decay pathway of the Ni(I)–bpy halides can be monitored by the
decrease of their characteristic UV-vis spectra over time (Figure 6, top). Linear fits were
obtained when plotting the reciprocal of the absorbance change vs time (Figure 6, middle),
consistent with a decay process that is second-order in nickel concentration. Together with
the comparison to independent synthesis, this pathway is therefore assigned to Ni(I)–bpy
halide dimerization.
The slope of the linear fit yielded second-order dimerization rate constants,
k
D
, which were
observed to be temperature-dependent (Figure 6,
bottom
and Table 3). Eyring analysis of
the temperature-dependent rate constants afforded enthalpic and entropic thermodynamic
parameters, ΔH
‡
and ΔS
‡
(Table 3).
61
–
63
Rather large enthalpic barriers are observed for
all compounds, ranging from ~11 kcal mol
−1
to 19 kcal mol
−1
. The most significant trend
follows ΔH
‡
of
3-Cl
>
2-Cl
>
1-Cl
. ΔS
‡
values ranged from ~ − 15 cal mol
−1
K
−1
to −38 cal
mol
−1
K
−1
.
The fairly large, negative values of ΔS
‡
are consistent with an associative (i.e., bimolecular)
transition state. ΔG
‡
(298 K) values yielded small overall differences, however, falling in the
range of ~22 kcal mol
−1
to 25 kcal mol
−1
. The largest ΔG
‡
(298 K) is observed for
3-Cl
; at
~25 kcal mol
−1
, this barrier is consistent with very slow decomposition at room temperature
and is in agreement with the overall stability of the related Ni(I)(
EtOOC
bpy)Cl complex.
30
We note, however, that dimer precipitation can offset the chemical equilibrium, acting as a
thermodynamic sink and driving the Ni(I) monomer to dimer conversion.
We again turn to substituent-based electronic effects to rationalize the changes in
dimerization rate constants and thermodynamics. The ΔH
‡
for dimerization is lowest for
1-Cl
/
Br
/
I
and higher for
2-Cl
/
3-Cl
, a trend similar to that observed for oxidative addition
(
vide supra
, Section 2.2). Electron withdrawing effects weaken the nucleophilicity of the
Ni(I) center, lowering the propensity for dimerization. Conversely, electron donation from
Cagan et al.
Page 10
Inorg Chem
. Author manuscript; available in PMC 2023 July 10.
Author Manuscript
Author Manuscript
Author Manuscript
Author Manuscript
the ligands enhances the reactivity of the Ni(I) center for both oxidative addition and
dimerization. The halide appears to have a lesser effect on the dimerization thermodynamics
than the bpy, with
1-Br
exhibiting the lowest enthalpic barrier of the three halide variants
(Figure S27).
2.5. Dimerization Mechanistic Investigations
DFT calculations also support the decay mechanism of Ni(I)
‒
bpy halides through
dimerization. The experimental second-order decay in Ni(I) allows for a direct 2 Ni(I)
(
R
bpy)X
→
[Ni(I)(
R
bpy)X]
2
pathway (Figure 7A).
We note that the experimental kinetic analysis does not rule out the inclusion of an
intermediate step, 2 Ni(I)(
R
bpy)X
→
Ni(0)(
R
bpy) + Ni(II)(
R
bpy)X
2
→
[Ni(I)(
R
bpy)X]
2
.
In either case, the starting reactants are two Ni(I)(
R
bpy)X species, and the ultimate
decomposition product is the [Ni(I)(
R
bpy)X]
2
dimer, making the overall reaction free energy
(ΔG) the same for both pathways. We have elected to computationally examine the simpler
case which does not invoke a Ni(0) species.
Like the oxidative addition mechanism discussed in Section 2.3, DFT finds that dimer
formation is initiated by nucleophilic attack by the doubly-occupied Ni 3
d
(z
2
) orbital. Rather
than interacting with the C(sp
2
)–X
σ
* orbital of an aryl halide, the acceptor orbital for
dimerization is an unoccupied halide
p-
orbital of another three-coordinate Ni(I) complex.
This attack results in a transition state with a
pseudo-
tetrahedral geometry (i.e., a formal
Ni(I)–bpy dihalide complex bridged via a halide to a second Ni(I)–bpy (Figure 7B)). The
predicted barrier for this transformation is ~17 – 22 kcal mol
−1
across the complexes. Two
distinct dimerization product geometries were obtained by following the downhill reaction
coordinate from the transition state, both featuring two bridging halides but with different
orientations (i.e., “peaked” vs “flat” geometry in Figure 7A). At the DFT level, both dimer
geometries feature two nickel sites in the Ni(I) oxidation state (Figure S68). Although
both dimers are rather high in energy – alone not providing enough driving force for the
dimerization – they show distinct similarities to the dimer structures obtained previously in
the solid phase.
30
,
32
,
64
Given the low solubility of the [Ni(I)(
R
bpy)X]
2
dimers in THF, we
argue that precipitation drives the dimerization process in the forward direction and makes
the dissociation of the dimer (i.e., the back-reaction) less favorable.
The free energies of activation are comparable to the experiment for
1-Cl
/
Br
/
I
; however,
the DFT transition states are predicted too low in energy for
2-Cl
and
3-Cl
(orange and
red data in Figure 7A). The discrepancy between experimental vs DFT values may again
be attributed to the better solubility of
1-Cl/Br
in THF, influencing the reversibility of the
speciation. The higher multiconfigurational character for Ni(I)
‒
bpy halides
26
,
27
and likely
for the dimeric structures (not captured by DFT) could also contribute to the differences
between computed and experimental data. These should be present to a greater degree in
complexes with electron-withdrawing substituents on the bpy (i.e., the electron withdrawing
trend of
3-Cl
>
2-Cl
>
1-Cl
). Therefore, it may be the case that the starting three-coordinate
Ni(I)
‒
bpy halides and their corresponding dimers are predicted to be too high in energy
for
2-Cl
and
3-Cl
due to insufficient account of ligand non-innocent radical character on
Cagan et al.
Page 11
Inorg Chem
. Author manuscript; available in PMC 2023 July 10.
Author Manuscript
Author Manuscript
Author Manuscript
Author Manuscript
the bpy ligand, which may result in smaller effective barriers for dimerization. Nevertheless,
computational analysis provides a plausible mechanism for Ni(I)–bpy halide dimerization.
3. Discussion
Photoredox catalysis mediated by electronically excited nickel complexes, coupled to
thermal reactivity pathways, can accomplish bond transformations with broad substrate
versatility. Ni(I) intermediates are thought to be critical for specific bond activation
steps facilitating reaction turnover (Figure 1A). Here we have demonstrated the facile
photochemical generation of a library of Ni(I)(
R
bpy)X complexes (R = t-Bu, H, MeOOC;
X = Cl, Br, I). Through experimental and computational analyses, we have evaluated their
competitive reactivity with challenging C(sp
2
)–Cl bonds and tendency toward dimerization
and have developed structure-function relationships between ligand set and reactivity.
UV-vis electronic absorption and EPR spectroscopies have provided an initial
characterization of the electronic structures of the photochemically generated Ni(I)–bpy
complexes. By UV-vis spectroscopy, all of the Ni(I) species exhibit relatively low energy
MLCT transitions (Figure 2A), the energies of which trend with the electronic effects of
the bpy substituents. More electron-withdrawing functionalities lead to the shift of the most
intense low-energy MLCT bands from ~660 nm to ~805 nm going from
1-Cl
to
3-Cl
.
This shift is attributed to the stabilization of the bpy
π
* acceptor orbitals. In contrast, the
variation of the halides shows a rather small influence on the MLCT energies.
EPR spectra reflect
S
= 1/2 Ni(I) ground state electronic configurations with
g
iso
values of
~2.12 – ~2.19 (Table 1). The
g
iso
observed here deviate somewhat from the free electron
g
value (2.0023) due to the presence of ground state orbital angular momentum. They are
larger than those observed for methyl Ni(II)–terpyridine
•–
(
g
iso
= 2.02) and mesityl Ni(II)–
phenanthroline
•–
(
g
iso
= 2.01) but smaller than observed for Ni(I)–bisoxazoline bromide
(
g
iso
= 2.24).
34
,
39
,
45
These
g
iso
values are consistent with greater ligand character for
the terpyridine and phenanthroline alkyl complexes and greater Ni(I) character for the
bisoxazoline bromide complex. While the
g
iso
of Ni(I)-bipyridine halide compounds would
be consistent with Ni(I) character intermediate to these complexes, a full analysis of the
aggregate
g
iso
values will require the additional assignment of the ligand field transitions
that spin-orbit couple with the ground state, giving rise to orbital angular momentum.
Current efforts are underway to assign the UV-vis transitions across various isolable
Ni(I) complexes to elucidate the amount of ligand vs metal character, as well as
ligand field vs MLCT contributions to the observed
g
values. These may be useful to
evaluate the degree of electron delocalization onto the bpy ligand. Indeed, multireference/
multiconfigurational calculations predict significant multiconfigurational character (i.e.,
Ni(I)(bpy)X vs Ni(II)(bpy
•–
)X).
26
,
27
A full analysis of the Ni(I)
g
values will require
experimental characterization of the Ni-based ligand field transitions. The
g
values report
on the spin-orbit coupling in the ground state of Ni 3
d
(x
2
-y
2
) SOMO parent-age. However,
importantly, the electronic effects of the bpy ligand and its substituents extend beyond the Ni
3
d
(x
2
-y
2
) SOMO and allow for energetic tuning of the lower-energy, occupied Ni 3
d
orbitals
Cagan et al.
Page 12
Inorg Chem
. Author manuscript; available in PMC 2023 July 10.
Author Manuscript
Author Manuscript
Author Manuscript
Author Manuscript