Light Activation and Photophysics of a Structurally Constrained Nickel(II)–Bipyridine Aryl Halide Complex

Abstract

Transition-metal photoredox catalysis has transformed organic synthesis by harnessing light to construct complex molecules. Nickel(II)–bipyridine (bpy) aryl halide complexes are a significant class of cross-coupling catalysts that can be activated via direct light excitation. This study investigates the effects of molecular structure on the photophysics of these catalysts by considering an underexplored, structurally constrained Ni(II)–bpy aryl halide complex in which the aryl and bpy ligands are covalently tethered, alongside traditional unconstrained complexes. Intriguingly, the tethered complex is photochemically stable, but features a reversible Ni(II)–C(aryl) ⇄ [Ni(I)…C(aryl)^●] equilibrium upon direct photoexcitation. When electrophile is introduced during photoirradiation, we demonstrate a preference for photodissociation over recombination, rendering the parent Ni(II) complex a stable source of a reactive Ni(I) intermediate. Here we characterize the reversible photochemical behavior of the tethered complex with kinetic analyses, quantum chemical calculations, and ultrafast transient absorption spectroscopy. Comparison to the previously characterized Ni(II)–bpy aryl halide complex indicates the structural constraints considered here dramatically influence the excited state relaxation pathway and provide insight into the characteristics of excited-state Ni(II)–C bond homolysis and aryl radical reassociation dynamics. This study enriches the understanding of molecular structure effects in photoredox catalysis and offers new possibilities for designing customized photoactive catalysts for precise organic synthesis.

Graphical Abstract

Synopsis: Nickel(II)–bipyridine (bpy) aryl halide complexes can be activated via direct light excitation for cross-coupling reactions. Here, we investigate the effects of molecular structure on the photophysics and photochemical properties of these catalysts via a structurally constrained Ni(II)–bpy complex with covalently tethered aryl and bpy ligands. The tethered complex exhibits prolonged stability under photoexcitation, but can be activated upon addition of electrophiles. It also exhibits distinct excited-state relaxation due to steric constraints, which prohibits the formation of a pseudo-T_d ³LF excited state.

1. Introduction

Photoredox catalysis has motivated modern synthetic organic chemistry through the development of selective bond transformations using light.^1–8 At its core, metallaphotoredox catalysis utilizes photoactive transition metal complexes to facilitate intramolecular charge transfer and/or outer-sphere electron transfer processes to generate catalytically reactive intermediates.^9–16 Nickel(II)–bipyridine (bpy) aryl halide complexes have emerged as a prominent class of light activated cross-coupling catalysts,^17–22 wherein direct photoexcitation leads to population of singlet metal-to-ligand charge transfer (¹MLCT) excited-state potential energy surfaces and, ultimately, Ni(II)–C bond homolysis to yield reactive Ni(I) species.¹⁹ Two bond homolysis mechanisms are outlined in Figure 1a and feature either (1) relaxation to a triplet Ni(II) ligand-field (³LF) state followed by thermal Ni(II)–C bond homolysis,¹⁹ or (2) intersystem crossing to repulsive ligand-to-metal charge transfer (LMCT) states (referred to herein as ISC/LMCT).^21,23 Assessing the interplay between these two (or other) possibilities remains challenging, however, as the quantum yields for Ni(II)–C bond homolysis are small, preventing direct observation using transient spectroscopies. Thus, ongoing efforts seek to understand the photophysical processes involved in photoredox catalysis and to expand the catalytic repertoire and reactivity of nickel catalysts for specific applications.

Figure 1.

a) Energy diagrams showing the proposed direct excitation mechanistic pathways for excited-state Ni(II)–C bond homolysis.^19,23 b) Irradiation of untethered Ni(II)–bpy aryl halide yields the catalytically relevant Ni(I)–bpy halide intermediate and a free aryl radical. c) Irradiation of a tethered Ni(II)–bpy aryl halide complex investigated in this work gives rise to a reversible Ni(II)–C bond homolysis equilibrium. The transient Ni(I)–bpy halide intermediate can be reacted with electrophiles in situ to promote the forward reaction.

Previous catalyst optimization efforts explored novel ligand architectures, bpy substitutions, and coordination environments to adjust the electronic properties of organonickel(II) complexes.^{19–21,24–29} Ni(II)–C bond homolysis serves as a universal photoactive pathway and occurs even upon replacing the imine backbone with aliphatic amines or phosphines.²⁰ However, ligand modifications significantly influence the kinetics of bond homolysis. Upon variation of the bpy and aryl substituents in Ni(II)–bpy aryl halide complexes, rate constants span ~two orders of magnitude, trend linearly with ligand Hammett parameters, and exhibit significant dependence on excitation energy.^21,22 These observations highlight the ability to tune key aspects of excited-state potential energy surfaces through structural modifications. Additionally, ligand substituents provide essential electronic contributions to the reactivity of photogenerated Ni(I) intermediates^24,26,30 and can be finely adjusted to suit different substrates in oxidative addition, to vary the overall solution-phase stability of Ni(I) species to improve selectivity, or to avoid off-cycle reactions.

The present study establishes the photochemical and photophysical properties of an underexplored, tethered Ni(II)–bpy aryl halide complex that has yet to be considered in the context of photoredox catalysis. Covalently tethering the aryl and bpy results in a fixed pseudo-square planar coordination geometry.³¹ Such a structural constraint contrasts traditional untethered analogues and provides a platform to explore how limited ligand flexibility influences excited-state relaxation processes in photoredox catalysis, as we hypothesized that geometric restrictions may modify ground- and excited-state potential energy surfaces and, thus, the ability to produce reactive intermediates upon photoexcitation. Herein, we have examined tethered Ni(II)–bpy aryl halide complex (2) using photochemical methods, ultrafast transient absorption (TA) spectroscopy, and quantum chemical calculations. Direct comparisons are made to the well-characterized, untethered complex (1).

Notably, 2 exhibits remarkable stability upon photoexcitation over a broad wavelength range. However, upon introducing an electrophile during photoirradiation, 2 rapidly converts to a new species. Photochemical behavior studied herein supports the presence of a reversible Ni(II)–C(aryl) ⇄ [Ni(I) + C(aryl)^●] photochemical reaction and demonstrates that a photogenerated Ni(I) species can be productively captured by reaction with electrophile. In addition to distinct photochemical behavior, 1 and 2 exhibit distinct photophysics. While relaxation to a low-energy ³LF state has been established for 1, structural constraints in 2 prohibit ³LF state formation and restrict excited-state relaxation, as probed by ultrafast TA spectroscopy, to MLCT potential energy surfaces. The photoactivation of 2 in the presence of electrophile, despite the lack of ³LF state formation, further suggests an ISC/LMCT pathway (Figure 1a, right) may facilitate excited-state Ni(II)–C bond homolysis. In addition to inhibiting ³LF state formation, structural constraints in 2 prolong ¹MLCT and relaxed-^1/3MLCT lifetimes by factors of ~ten and ~two, respectively. As such, this study holds promise for advancing our understanding of ligand electronic and steric effects in photoredox catalysis and guides approaches to unlocking new potential for the design of efficient and selective photoactive catalysts for organic synthesis.

2. Results and Analysis

2.1. Steady-state UV-vis absorption spectroscopy

Untethered (1) and covalently tethered (2) Ni(II)–bpy aryl halide complexes were synthesized and characterized following previous literature reports (Supporting Information Section S1.2). UV-vis spectra of 1 and 2 in THF revealed two broad bands in the region of ~350 – 600 nm, which were previously assigned to low- (band i) and high-energy (band ii) ¹MLCT transitions (Figure 2).^18,21,31 For 2, band i is red-shifted by ~900 cm⁻¹ (λ_i,max = 485 nm and 507 nm for 1 and 2, respectively) and exhibits a lower extinction coefficient (ε = 4070 M⁻¹ cm⁻¹ and 2800 M⁻¹ cm⁻¹ for 1 and 2, respectively). For 11 corresponding untethered complexes, we have observed a loose trend between increasing ε and longer λ_i,max, which is the opposite trend observed here for 1 and 2.²¹ Therefore, we attribute the lower ε in 2 to the different aryl orientation, which influences the relative overlap between donor and acceptor molecular orbitals involved in the MLCT transitions. Furthermore, peak maxima red-shift and decrease in intensity in toluene relative to THF (Figure S2), corroborating the assignment of bands i and ii as mainly ¹MLCT electronic transitions with Ni(II)-to-^Phbpy π* character.

Figure 2.

a) UV-vis absorption spectra of 1 and 2 in THF. Spectra recorded in toluene are provided in Figures S2 and S7. b) Schematic representation of electronic transitions comprising bands i and ii.

CASSCF/QD-NEVPT2 calculations predict multiconfigurational ground states for 1 and 2, with closed-shell singlet weights in the CAS configuration interaction vector of only ~56 – 57%. Remaining configurations consist of ¹MLCT states (~20% and ~22% for 1 and 2, respectively) and ¹LF states (~17% and ~16% for 1 and 2, respectively) (Tables S4–S5). The lowest vertical excited triplet states of mainly ³LF character are at ~9 300 cm⁻¹ (~26.6 kcal mol⁻¹) and ~10 800 cm⁻¹ (~30.9 kcal mol⁻¹) for 1 and 2, respectively (Figure S28). However, free energy differences between the optimized (adiabatic) triplet and ground states are significantly different, with ΔG = ~8 kcal mol⁻¹ (for 1) and ΔG = ~27 kcal mol⁻¹ (for 2). This difference demonstrates a significant stabilization of the lowest energy ³LF state in 1 due to sterically unencumbered geometric rearrangement, which allows for the formation of a relaxed, pseudo-T_d coordination. The same structure is inaccessible in 2 due to steric constraints provided by the covalently tethered aryl group (Figure S32).

For 1, calculations predict bands i and ii in Figure 2b are mainly ¹MLCT in nature, consisting of excitations between specific Ni 3d donor orbitals and bpy π* acceptor orbitals.^19,21,32 The most intense excitations are found for Ni 3d(xz)/3d(yz) donor orbitals, consistent with a previous report.²¹ However, tethering the aryl group in 2 changes the composition of electronic excitations. While excited states with the largest oscillator strengths in 1 are ¹MLCT states, the ¹LF states in 2 have comparable intensities to the most intense low-energy ¹MLCT band (Figures S29–S30). We attribute this behavior to a better electronic conjugation and overlap between the tethered aryl and bpy ligand, which increases the bpy character (thus mixing some allowed MLCT intensity) in the unoccupied Ni 3d(x²-y²)/C(sp²)* acceptor orbital (Figure S27). The calculated intensity of the low-energy band in 2 also decreases by ~2.5 times relative to 1 (Tables S6–S7), in agreement with the smaller experimental extinction coefficient.

In summary, covalently tethering the aryl and bpy groups modulates the electronic properties of 2 relative to 1, as manifested in the experimental and computational changes in the excited-state compositions and intensities of bands i and ii, with band ii being responsible for light-activation in direct excitation nickel photoredox catalysis. In terms of other well-studied Ni(II)–bpy aryl halide complexes, the tethered complex 2 provides an important reference for a comprehensive investigation of the influences of electronic and steric perturbations on its photophysical and photochemical characteristics.

2.2. Photochemical activity

Irradiating 2 in THF using a 390 nm LED (exciting high-energy ¹MLCT band ii) resulted in no significant photochemical activity after 15 hours (Figure 3a, right). Slow decay of characteristic absorption features occurs after prolonged irradiation, likely due to unproductive photodegradation. 2 is also photo-inactive over a broad range of excitation wavelengths (370 nm, 427 nm, 456 nm, 525 nm; Figure S6) and in toluene (Figure S7). This behavior is in stark contrast to 1, which affords almost full conversion to a previously characterized Ni(I)–bpy chloride species after one hour of photoirradiation (Figure 3a, left) using 390 nm LED with the same power (k_obs = 3.8 × 10⁻² min⁻¹ and Φ = 0.5 × 10⁻³; LED power of 170 mW cm⁻²).

Figure 3.

Photoirradation of 1 and 2 in THF using a 390 nm LED. a) Time-dependent photolysis profiles of 1 and 2. b) Time-dependent photolysis profiles of 1 and 2 in the presence of 1000 eq. of mesityl bromide.

The distinct photochemical behavior of 1 and 2 (Figure 3a) indicates that 2 either a) does not follow the same photoexcitation/relaxation pathways as 1, or b) excited-state Ni(II)–C bond homolysis, although potentially viable, is reversible due to the covalent tethering of the bpy and aryl groups. We note that the absence of any photoproduct formation following irradiation of 2 excludes alternate pathways for its irreversible decomposition, such as through Ni(II)–Cl bond homolysis. To distinguish between the two cases above, we have repeated photochemical experiments in the presence of a large excess (1000 eq.) of mesityl bromide, which should react if a Ni(I) intermediate is formed in situ. The steady-state UV-vis spectra do not change appreciably upon addition of mesityl bromide (Figure S3); no reactivity is observed between Ni(II) and mesityl bromide without irradiation. However, bands i and ii of parent Ni(II) complexes 1 and 2 decay more rapidly upon 390 nm LED irradiation in the presence of mesityl bromide (Figure 3b).

Compounds 1 and 2 exhibit distinct photochemical differences when irradiated in the presence of a large excess of mesityl bromide. Untethered 1 exhibits precedented reactivity; the accumulation of the Ni(I)–bpy chloride intermediate complex is fully suppressed by mesityl bromide (Figure 3b, left). The proposed pathway for this reactivity consists of oxidative addition of mesityl bromide to Ni(I) to yield a high-energy Ni(III) intermediate,^30,33–35 which can then undergo comproportionation with another Ni(I) in each reaction cycle to produce a new photoactive Ni(II)–bpy aryl halide and a photoinactive Ni(II)–bpy dihalide in a 1:1 ratio.^16,33 With extended irradiation, all of the starting Ni(II)–bpy aryl halide complex is expected to ultimately convert to a triplet d⁸ Ni(II)–bpy dihalide (Scheme S8), which does not possess distinctive UV-vis absorption features in the visible region (Figure S4). Note that the sloping UV-vis spectra in Figure 3b, left is indicative of either precipitation of Ni(II)–bpy dihalide or formation of an off-pathway [Ni(I)–bpy halide]₂ binuclear species from two Ni(I)–bpy halide units.³⁰

Conversely, irradiation of 2 with 1000 equivalents of mesityl bromide produces a new, stable species with a red-shifted absorption band at 555 nm (Figure 3b, right). Interestingly, a species with a similar absorption profile (absorption bands centered at 547, 399, and 352 nm; see Figure S11 for comparison) has been observed by Klein and co-workers via one-electron reductive dissociation of the bromide anion from a related tethered Ni(II)–bpy phenyl bromide complex.³⁶ Based on spectroelectrochemistry, that species was assigned as the tethered, tridentate Ni(I)–bpy phenyl complex. Similar to our photochemical process, no isosbestic points were detected in the spectroelectrochemical conversion of the parent species to the proposed Ni(I)–bpy phenyl intermediate, suggesting more complex, multistep reactivity.

Our attempts to isolate and characterize this soluble photogenerated species have been unsuccessful due to complications associated with irradiation of larger quantities of starting material (e.g., low solubility and low light penetration) and the presence of a large excess of mesityl bromide, Ni(II)–bpy dihalide (vide infra), and unreacted Ni(II)–bpy aryl halide in the reaction mixture after irradiation. However, according to UV-vis and paramagnetic ¹H-NMR spectroscopic characterization of the crude reaction mixture, we note that the photogenerated species is paramagnetic (¹H-NMR peaks at ~18, ~23, and ~69 – 76 ppm), stable under high temperature and concentration conditions, and reacts rapidly with air (see Section S1.4 of the Supporting Information). These findings corroborate the results of Klein and co-workers associating the UV-vis features to a tridentate Ni(I)–bpy phenyl species.³⁶ We also note that calculated (CASSCF/QD-NEVPT2) absorption spectra are consistent with this assignment, with the tethered, tridentate Ni(I)–bpy phenyl species exhibiting a slight red shift of the main absorption feature from the parent Ni(II) complex (Figure S31). The calculated oscillator strengths further suggest that ~50% of the parent Ni(II) species is converted to this Ni(I)–bpy phenyl species.

Additionally, from the same reaction mixture, we were able to isolate an appreciable amount of sparingly soluble Ni(II)–bpy dihalide; its spectroscopic features resemble an independently synthesized Ni(II)–^Phbpy dichloride (Figure 4 and Section S1.4 of the Supporting Information). We note that possible presence of mixed chloride and bromide species cannot be ruled out, since the peaks in the paramagnetic region in Figure 4 are fairly broad. Moreover, the ¹H-NMR spectrum of the sample was collected from the crude reaction mixture as a precipitate, washed with THF, and redissolved in D₂O for the NMR analysis. Therefore, different solubilities of the chloride/bromide complex might result in altered speciation. Nonetheless, this comparison supports photodissociation of the tethered aryl ligand is indeed feasible in 2 as in 1, as irradiation of 2 also yields an intermediate that is captured by mesityl bromide and ultimately results in the formation of a Ni(II)–bpy dihalide complex.

Figure 4.

Spectroscopic characterization and comparison between the independently synthesized Ni(II)–^Phbpy dichloride (top) and the isolated product of irradiation of 2 with mesityl bromide (bottom). a) Paramagnetic ¹H-NMR spectra measured in D₂O. b) Steady-state UV-vis absorption in H₂O.

Photochemical reactions of 1 and 2 exhibit distinct dependence on mesityl bromide concentrations, with increasing rates observed with increasing mesityl bromide concentrations (Figure S19). In both 1 and 2, the decay of Ni(II) is assisted by the presence of mesityl bromide in solution. As discussed above, this observation is attributed to a reversible equilibrium for Ni(II)–C bond homolysis in 2. For 1, this behavior can likely be attributed to a solvent-caging effect, which prevents aryl radical diffusion from Ni(I), allowing sufficient time for reformation of the Ni(II)–C bond.³⁷ This effect is less pronounced when mesityl bromide is present in sufficiently high concentrations (Figure S19), likely due to limited formation of caged radicals due to more rapid Ni(I) reaction with electrophile in solution. Although complex kinetics (especially at higher mesityl bromide concentrations) preclude obtaining well-defined rate constants for the Ni(II)–C bond homolysis step (see further discussion in Supporting Information Section S1.5), we note that the irradiations of 1 with and without the addition of quantitative excess mesityl bromide reveals a nearly 2–3 fold increase in Ni(II) conversion. Such estimation offers valuable insights into the quantum yields of Ni(II)–C photochemical bond homolysis and aids in deconvolution between ‘productive’ ligand dissociation and ‘unproductive’ recombination kinetics. It is worth noting that the addition of mesityl bromide does not influence the ultrafast photophysical relaxation pathways as discussed below in Section 2.3 and instead facilitates the dissociation of aryl radicals from the [Ni(I)…C(aryl)^●] cage. Ongoing work involves detailed quantum chemical molecular dynamics simulations to better understand the behaviors of 1 and 2 after photoexcitation and the impact of mesityl bromide addition on photolysis rate constants.

To better understand the nature of the photoinitiated reactivity of 2 with mesityl bromide, we have performed irradiations of 2 with a variety of substituted aryl bromides (Figures S14–S18). We find that the soluble products seen after prolonged irradiations exhibit comparable absorption features, consistent with the same tethered Ni(I)–bpy phenyl product seen upon irradiation of 2 with mesityl bromide. Interestingly, we have observed an electronic influence on the apparent rate of conversion of 2; electron-donating substituents on the aryl bromide contribute to an enhancement of this rate. Steric effects appear to play a lesser role, but with a poorly discernible trend. These preliminary results suggest that a concerted oxidative addition mechanism for the activation of aryl bromides via conversion of 2 may not be operative, as it would predict enhanced reactivity with aryl bromides with electron withdrawing substituents and reduced sterics.^26,33 Current efforts are underway to thoroughly detail the precise mechanism of reactivity of aryl halides with light-activated 2.

2.3. Transient absorption spectroscopy

To further understand the photochemical differences between 1 and 2, we examined their excited-state relaxation pathways using ultrafast TA spectroscopy. The photophysics of 1 and other untethered Ni(II)–bpy aryl halide complexes have been described;^18,19 in this study, we include 1 for direct comparison with 2. A consistent relaxation pathway for Ni(II)–bpy aryl halides has been outlined, regardless of whether low-energy ¹MLCT (λ_exc. = ~530 nm) or high-energy ¹MLCT states (λ_exc. = ~400 nm) are accessed (bands i and ii in Figure 2b).^18,19 However, only the high-energy ¹MLCT band ii was proposed to be relevant for catalysis, demonstrating excited-state Ni(II)–C bond homolysis and formation of a reactive Ni(I) intermediate.^21,22 This assignment is exemplified by the photochemical activity of 1, which is considerably wavelength-dependent and requires a minimum energy threshold (λ_exc. < 525 nm) for photolysis (Figure S5).

The steady-state absorption profile of 2 is red-shifted relative to 1 (Figure 2a). This ultimately allows enhanced accessibility of the higher-energy ¹MLCT states in 2 (band ii) with lower-energy light. Still, both species have notable absorption intensity associated with a high-energy ¹MLCT state near 400 nm. Therefore, 400 nm excitation has been utilized for TA to probe relaxation pathways from ¹MLCT excited states of 1 and 2, thereby addressing key excited states relevant for catalysis.

Direct photoexcitation of 1 with pulsed 400 nm light in THF results in an excited-state lifetime on the nanosecond timescale (Figure S21). Global analysis revealed a three-component sequential decay model composed of a fast time component (τ₁ = ~400 fs), an intermediate component (τ₂ = ~9 ps), and a longer-lived state (τ₃ = ~3.5 ns). This relaxation pathway agrees well with and is analogous to that characterized by Doyle and co-workers for 1 with λ_exc. = 530 nm (Table S3).¹⁹ As illustrated in Figure 5, a ¹MLCT excited state is first accessed upon photoexcitation. This state decays into either a relaxed ¹MLCT state through vibrational cooling or to a ³MLCT state via intersystem crossing. This process is observable through the blue-shifting of the excited-state absorption (ESA) in the red region of the spectrum (Figure S21B), which occurs on the picosecond timescale. The ^1/3MLCT state then relaxes into an excited ³LF state observed through the decay of the ESA feature between a ~5 – 20 ps time delay (Figure S21C). Finally, this ³LF excited state undergoes a longer, spin-forbidden relaxation back to the singlet ground state in several nanoseconds (Figure S21D), which is reflected by the recovery of the ground state bleach (GSB).

Figure 5.

Proposed mechanisms of relaxation for 1 and 2 after photoexcitation into their excited ¹MLCT states with 400 nm pulsed light.

Photoexcitation of 2 into the high-energy ¹MLCT state resulted in distinct excited-state dynamics relative to 1. While the non-zero TA signal for 1 persists into the nanosecond regime, the entire excited-state manifold of 2 relaxes on the tens of picoseconds timescale, a significant shortening of excited-state lifetime by 2 – 3 orders of magnitude (Figure 6). Global analysis of the excited-state dynamics of 2 revealed a two-component sequential model consisting of a shorter (τ₁ = ~3 ps) and longer time component (τ₂ = ~14 ps). Analogous to 1, the first ~10 ps of the TA difference spectra for 2 are dominated by a blue-shifting ESA alongside recovery of the GSB (Figure 6c). However, the longer time dynamics differ between the complexes, with the ESA decaying monoexponentially and with the same time constant associated with the recovery of the GSB (Figure 6b and Figure 6d). This isosbestic behavior can be interpreted as a recovery of the ground state directly from the relaxed-¹MLCT or ³MLCT excited state. In all, the photophysical behavior of 2 does not support the formation of ³LF excited states with appreciable quantum yield.

Figure 6.

Ultrafast TA spectra of 2 in deoxygenated THF upon 400 nm excitation. a) Evolution-associated spectra obtained from a two-component, sequential global model of the data. b) Decay of the ESA feature at 568 nm (13.7 ± 1.8 ps, orange) and recovery of the GSB feature at 506 nm (13.2 ps ± 1.8 ps, blue), with their fits to monoexponential functions in black. c) TA difference spectra for short time delays between pump and probe. d) TA difference spectra for longer time delays between pump and probe.

We also considered the influence of 100 eq. of mesityl bromide on the excited-state dynamics of 1 and 2. Although UV-vis monitoring provides evidence of reactivity for 1 and 2 with 100 eq. of mesityl bromide during 390 nm LED photoirradiation (Figures S9–S10), the relaxation pathways of 1 and 2 probed by TA spectroscopy remain largely unaffected by the addition of electrophile, as relaxation time constants and spectral profiles are consistent between samples with and without mesityl bromide (see Section S1.6 in the Supporting Information). These observations indicate that TA of 1 and 2 probe the dominant background relaxation processes and are insensitive to the low quantum yield processes related to light-activated reactivity with electrophiles (vide infra).

3. Discussion

Significant efforts have been dedicated to investigating the influence of steric and electronic factors on the photoredox reactivity of organonickel(II) complexes. These studies have unveiled structure-function relationships that elucidate how the ligand scaffold or electron-donating/-accepting substituents impact the formation of reactive Ni intermediates in cross-coupling catalysis. However, to the best of our knowledge, no data have been reported regarding organonickel(II) complexes with light-induced activation and dissociation of a coordinated aryl group tethered to the backbone ligand. While a number of previously synthesized tethered Ni(II)–bpy aryl halide complexes have been examined for electrochemical/reductive catalysis, including detailed spectroelectrochemical investigations, an explicit pathway for productive, direct light activation has not been outlined.^31,36,38 This study adopted a previously described tethered Ni(II)–bpy aryl chloride complex 2 and explored its photochemical and photophysical properties, particularly through a comparison to an untethered Ni(II)–bpy aryl halide photocatalyst 1.

In contrast to untethered analogues, 2 exhibits superior stability upon photoirradiation across a broad range of excitation wavelengths spanning ~370 – 525 nm. Despite this prolonged stability, the covalently tethered Ni(II) complex can still be light-activated for reactivity in the presence of electrophile. This reactivity can be attributed to a shift in reversible Ni(II)–C(aryl) ⇄ [Ni(I)…C(aryl)^●] photochemical equilibrium, in which the resulting Ni(I) reacts with an electrophile in solution. Such improved stability could potentially be leveraged in photoredox catalysis. For instance, it was previously observed that the presence of electron-withdrawing substituents like methyl or ethyl esters at the bpy 4,4’ positions significantly enhances the photochemical reactivity of 1, while simultaneously shifting the steady-state absorption features to lower energies by ~2000 cm⁻¹ (λ_i,max = 532 nm).²¹ A similar effect may be expected for 2, which would result in even more markedly red-shifted MLCTs, making them accessible with lower-energy light, but also with the tethered ligand providing stability to the otherwise highly photochemically reactive Ni(II).

Although we have studied 2 in terms of its competency for photoredox catalysis, such photochemical robustness also offers the potential for regulating reactivity specific to experimental conditions. For example, by tuning concentrations of other reactive species in solution, the unique photochemistry of 2 could be harnessed for a tailored release of Ni(I), thereby potentially advancing reaction selectivity (Figure 7). This possibility could unveil intriguing and unexplored applications leveraging specific roles of Ni catalysts based on experimental conditions. A similar approach was employed to enhance yields and selectivity in cross-electrophile coupling reactions.^39,40 In ref. 39, adjusting the reduction potentials of homogeneous organic reductants allowed for the precise time-release of alkyl radicals from Katritzky salt precursors and, thus, synchronized their production rates with those of the Ni(II)–aryl intermediates proposed for their capture. This strategy minimized the probability of undesired side reactions and Ni catalyst decomposition. Similarly, Reisman and co-workers modulated formation rates of benzyl radicals, which yielded an analogous effect for the cross-coupling of alkenyl and benzyl electrophiles.⁴⁰ Analogously, the released benzyl radical is intercepted by a Ni(II)–alkenyl resting state, establishing a direct connection between radical generation rate and the selectivity of homo- vs. cross-coupling reactivity. Modulation of the rate of Ni(I) formation based on the concentration of electrophile, as demonstrated in this work, presents an alternative route to a similar strategy.

Figure 7.

Cross-coupling selectivity in the Ni ground- and excited-state catalysis can be improved by synchronizing the generation of alkyl or aryl radicals (R₂^●) with the rate of formation of the LNi(II)R₁X intermediate.^39,40 In this work, we propose an alternative strategy by controlling the generation of the LNi(I)X complex from the parent Ni(II) precursor instead, offering an independent handle for the rate of nickel cross-coupling catalysis.

Further, regarding reaction selectivity, 1 and 2 revealed different product formation upon photoirradiation in the presence of mesityl bromide. Prolonged irradiation of 1 culminated in the formation of Ni(II)–bpy dihalide, as monitored by the disappearance of the Ni(II)–bpy aryl halide UV-vis absorption features, which is consistent with the previously proposed oxidative addition and comproportionation reaction mechanism.^16,30,33 Contrastingly, 2 yielded a new Ni species with a unique UV-vis spectrum (Figure 3b, right) in addition to a Ni(II)–^Phbpy dihalide. A species with a similar UV-vis spectrum was accessed via spectroelectrochemical one-electron reduction of the bromine analogue of 2 and was attributed to the tethered, tridentate Ni(I)–bpy phenyl complex.³⁶ Here we propose a photochemical pathway to this reaction intermediate. The formation of Ni(II)–^Phbpy dihalide in the reaction mixture also supports the activation of mesityl bromide (Figure 4). However, this species is not the dominant product, as the higher stability of the tethered Ni(I)–bpy phenyl species and its lack of reactivity with mesityl bromide terminates the catalytic cycle at this stage. The low reactivity of the Ni(I)–bpy phenyl is likely due to steric hindrance of the tridentate ligand coordinating and protecting the Ni(I) center. Under catalytic conditions, this stable Ni(I) intermediate could potentially be reactivated via reactivity with radicals,^41,42 sterically accessible alkyl bromides,⁴³ or through outer-sphere single-electron transfer processes⁴⁴.

An examination of mesityl bromide concentration-dependent photochemistry highlighted controllable Ni(II) species decomposition rates (Figure S19), further supporting the notion of reversibility in the excited-state bond homolysis. Intriguingly, a similar observation was made for both 2 and 1, wherein reversibility in the Ni(II)–C(aryl) ⇄ [Ni(I)…C(aryl)^●] excited-state bond homolysis in the latter was not previously considered. While the concept of solvent caging and retaining the aryl radical in the proximity of Ni(II) has been posited for an oxidative addition of aryl halides reacting through a halogen-atom abstraction mechanism,³³ the impact of this reversibility on photochemical reactivity remains unexplored. Through a comparative assessment of the rate constants of photochemical decay of 1 with varying mesityl bromide concentrations, we observed a five-fold increase in the rate constant. While the experimental data reported here were obtained solely in THF, a thorough investigation of the solvent dependence of this rate enhancement would provide valuable insights for understanding the effect of caged radical recombination in photoredox catalysis.⁴⁵ Unfortunately, the inherently low quantum yields of the Ni(II)–C bond homolysis and very short MLCT excited state lifetimes hinder a direct, comprehensive experimental survey of the excited-state bond homolysis. Traditional computational methods – often static and only probing vertical energies and vertical excited states – are also insufficient.^21,23,46,47 Given the complexities of the excited-state dynamics and bond homolysis behavior, which may also require a multireference treatment, insights derived from the experimental data in this study could provide an important reference for benchmarking more advanced molecular dynamics simulations.

When considering the reversibility of C(aryl)^● formation, it is noteworthy to emphasize the research by Park and co-workers who observed a similar phenomenon with Ni(II) metallacycle compounds (Figure 8).²⁰ Regardless of the specific backbone ligand, metallacycles featuring cycloneophyl (–C₆H₄–o–C(CH₃)₂CH₂–) and its oxa- (–OC₆H₄–o–C(CH₃)₂CH₂–) and thia- (–SC₆H₄–o–C(CH₃)₂CH₂–) derivatives have been proposed to undergo light-induced carbon radical generation through Ni(II)–C(sp³) bond homolysis. In the case of thia-metallacycles, radical formation can be followed by C–S bond formation through reductive elimination. In cycloneophyl and its oxa- derivative, however, radical formation is followed by recombination to reform the reactant state (Figure 8). While the covalent Ni–X linkage in the metallacycles and their overall chemical behavior resemble that of 2, the proposed observation of reversibility in 1 and the photophysics presented in this work suggest that reversible radical recombination to Ni might represent a more general paradigm to consider in Ni photoredox and single-electron chemistry.

Figure 8.

Irradiation of Ni(II) metallacycle compounds results in the formation of transient bound-C(sp³) radicals that can either facilitate reductive elimination or recombine to recover the Ni(II) ground state.

Using transient spectroscopies, Doyle and co-workers have provided significant insights into the excited-state dynamics of untethered Ni(II)–bpy aryl halide complexes related to 1.^18,19 They observed comparable excited state dynamics using either 400 or 530 nm excitation and further provided valuable insights into the lifetimes of key MLCT and LF excited states. Their findings established that ^1/3MLCT lifetimes are short (< ~15 ps in a series of eight Ni(II)–bpy aryl halide complexes), while ³LF states are significantly longer-lived (~2 – 8 ns). From this analysis, in conjunction with corroborating ¹H-NMR 2D Exchange Spectroscopy (EXSY) and DFT-calculated bond strengths, it was inferred that the excited state responsible for Ni(II)–C(aryl) bond homolysis is the ³LF state. From comparative TA experiments and comparisons to previous literature, we find that ¹MLCT states are formed upon initial 400 nm excitation for both 1 and 2.^18,21,31 The excited-state dynamics of both complexes share a similar subsequent relaxation step, indicated by a recovery of the GSB and a concomitant blue-shifting in the ESA by ~20 nm (Figures S21 and S24). Similar relaxation profiles, including the blue-shifting ESA, have been observed in other 1^st- and 2^nd-row transition metal complexes, including Ni(II)–bpy aryl halides.^19,48 The precise nature of the relaxation process is currently unclear and can be indicative of either vibrational cooling or intersystem crossing between ^1/3MLCT states. Both processes can occur on the timescale consistent with our data, and vibrational cooling can even mediate intersystem crossing to a ³MLCT state. We note that CASSCF/QD-NEVPT2 calculations predict multiple overlapping ^1/3MLCT states that could be accessible (Figure S28); thus, we cannot unambiguously differentiate vibrational cooling vs. intersystem crossing. Notably, the blue-shifting of the ESA has also been observed for Cu(I) bis-phenanthroline complexes and was associated with the flattening of the T_d Franck-Condon geometry in the excited-state Cu(II)* on the picosecond timescale, which is also associated with intersystem crossing.⁴⁹ In the Ni(II) d⁸ systems studied here, an interconversion between square planar and pseudo-T_d geometries near the Franck-Condon point of the excited-state ¹MLCT surface might take place. Such behavior could yield significantly distinct ¹MLCT lifetimes, as was observed in this work for the flexible 1 (τ₁ = ~400 fs) and more constrained 2 (τ₁ = ~3 ps).

The excited-state dynamics of 1 and 2 are also distinct after accessing the relaxed ^1/3MLCT excited state. Due to steric constraints in 2, distortion to a pseudo-T_d geometry is prohibited (Figure 9 and S32), which strongly destabilizes the ³LF state relative to the ground state, likely making it inaccessible from the MLCT excited state manifold. From quantum chemical calculations, the optimized ³LF state of 1 is ~19 kcal mol⁻¹ lower in ΔG than the analogous ³LF state in 2. As a consequence, while 1 relaxes from the MLCT excited-state manifold to a longer-lived ³LF state with a time constant of ~9 ps, 2 exhibits a prolonged ^1/3MLCT excited state that relaxes back to the singlet ground state with a time constant of ~14 ps. A similar structural control over excited state dynamics has been quantified for Cu(I) bis-phenanthroline complexes, in which a combined experimental and computational approach has determined that steric effects can provide up to ~20 kcal mol⁻¹ in tuning excited-state potential energy surfaces, which further tunes the MLCT lifetime of those complexes over four orders of magnitude.⁵⁰

Figure 9.

An overlay of DFT-optimized geometries of 1 and 2 in their ground-state singlet and triplet states, highlighting the prohibited distortion of 2 due to steric constraints. ΔG was obtained from CASSCF/QD-NEVPT2 electronic energies (see Section 2.1 of the Supporting Information); RMSD = root-mean-square deviation of atomic positions between superimposed structures of optimized singlet and triplet states.

Notably, despite the lack of ³LF state formation due to strong steric constraints, 2 still exhibits reactivity with electrophiles after photoirradiation. This reactivity suggests access to the ³LF state may not be necessary for the generation of a reactive Ni(I) intermediate. We further note that the excited state dynamics of 1 and 2 probed by TA are not affected by the addition of mesityl bromide, which indicates that the chemical reactivity with mesityl bromide occurs via a separate process and reactive species that are not observed as part of the excited-state dynamics probed by this experiment. This observation is reminiscent of the nature of the excited-state bond homolysis in 1 and related untethered complexes. Due to small quantum yields for homolysis, transient spectroscopies probe effectively unproductive, background excited-state relaxation processes that do not result in Ni(I) intermediates and organic radicals. Similarly for 2, the TA reports on MLCT relaxation but not excited-state bond homolysis.

The primary distinction between the unique excited-state dynamics of 1 and 2 lies in their geometric properties and excited-state distortions/constraints, which render 2 incapable of stabilizing ³LF excited states. We can also draw comparisons between the photophysics of 2 and other tethered Ni(II) species. A recent report by Wenger and co-workers investigated the excited-state behavior of a tethered Ni(II)–bpy aryl chloride complex with an NĈ^N binding motif rather than ĈN^N in 2.⁵¹ Analogous to 2, upon 400 nm excitation, the NĈ^N tethered complex accesses a ¹MLCT excited state, which was proposed to undergo decay through another undefined-spin MLCT state to a ³LF state. The time constant associated with the first decay varies with solvent, ranging from ~700 – 900 fs. This fast time constant accounts for the relaxation of both singlet and triplet MLCT manifolds. Furthermore, a solvent-dependent time constant associated with ³LF relaxation to the singlet ground state ranged from ~9.2 – 14 ps. While the proposed excited-state decay pathway for the NĈ^N tethered complex is similar to that proposed for untethered analogues,¹⁹ the time constants appear comparable to those observed for 2 in this work. However, the authors hypothesized the NĈ^N isomer does not require distortion from square planar to pseudo-T_d to stabilize a ³LF state. Here we have observed a similar degree of geometric distortion upon optimization of the ³LF state in the NĈ^N isomer and 2 (Figure S33). However, the smaller RMSD between the superimposed singlet and triplet structures of the former suggests that the NĈ^N ligand remains more planar in the ³LF state, with the main distortion featuring larger out-of-plane bending of the Ni–Cl bond. Therefore, it remains unclear whether the variance in accessing a ³LF state between the tethered isomers can be attributed to disparities in ĈN^N vs. NĈ^N coordination or potential solvent effects on the excited state dynamics. In 2, the MLCT-based ESA feature decays monoexponentially with the same time constant as the recovery of the GSB, which allows us to associate the longer-lived time component as MLCT in origin, not ³LF. In both cases, the tethered isomers demonstrate significantly faster relaxation compared to untethered Ni(II)–bpy aryl halide analogues. Ultrafast X-ray spectroscopies may prove useful for obtaining more precise mechanistic information regarding excited-state relaxation dynamics in tethered vs. untethered complexes.

4. Conclusions

We have compared a covalently tethered, structurally constrained Ni(II)-bpy aryl halide complex 2 and its untethered analogue 1, exploring the significant distinctions in their photochemical reactivity and excited-state dynamics. Key findings are a) unlike 1, the tethered complex 2 exhibits prolonged stability under photoexcitation across a wide range of LED wavelengths, which indicates excited state bond homolysis products from 2 do not effectively accumulate over time; b) the introduction of an electrophile during photoirradiation of 2 enables productive capture of a species capable of C(sp²)–Br activation, which we propose to be a Ni(I) intermediate based on comparison to previous literature. This reactivity offers a promising avenue for enhancing nickel photoredox catalysis and refining reaction selectivity; c) from concentration-dependent photochemical experiments in the presence of electrophile, both 1 and 2 exhibit behavior consistent with reversible Ni(II)–C(aryl) ⇄ [Ni(I)…C(aryl)^●] excited-state bond homolysis; and d) distinct excited-state relaxation mechanisms occur in 1 vs. 2, which can be attributed to steric constraints introduced through the covalent linkage between the phenyl and bpy ligands, prohibiting the formation of a pseudo-T_d geometry and access to ³LF excited states.

Future investigations should focus on practical applications of the tethered complex in photoredox cross-coupling catalysis to assess its photochemical behavior under catalytic conditions and whether influences on Ni(I)–organic radical cage escape processes play an important role. Relatedly, exploring controlled Ni(I) generation holds potential for enhancing reaction selectivity in organic synthesis. Further research into the impact of ligand substituents on the tethered bpy backbone, especially electron-withdrawing groups, may enable the use of lower-energy light and improve the stability of highly reactive intermediates, offering opportunities to tune reaction conditions in nickel ground-state and photoredox catalysis.