Mechanistic insights into photochemical nickel-catalyzed cross-couplings enabled by energy transfer

Various methods that use a photocatalyst for electron transfer between an organic substrate and a transition metal catalyst have been established. While triplet sensitization of organic substrates via energy transfer from photocatalysts has been demonstrated, the sensitization of transition metal catalysts is still in its infancy. Here, we describe the selective alkylation of C(sp3)–H bonds via triplet sensitization of nickel catalytic intermediates with a thorough elucidation of its reaction mechanism. Exergonic Dexter energy transfer from an iridium photosensitizer promotes the nickel catalyst to the triplet state, thus enabling C–H functionalization via the release of bromine radical. Computational studies and transient absorption experiments support that the reaction proceeds via the formation of triplet states of the organometallic nickel catalyst by energy transfer.

After 48 hours, the reaction was filtered through a small bed of celite and silica and concentrated in vacuo. The residue was purified by column chromatography using silica gel (100-200 mesh size) and DCM/hexane or Et2O/pentane as the eluent. In the case of 1,4-dioxane and toluene coupling partners, two 34 W blue LEDs were used to irradiate for 96 hours.

2-(4-phenylbutyl)isoindoline-1,3-dione (28):
An oven-dried screw cap reaction tube and a 5 mL vial equipped with a PTFE-coated stir bar were brought into the N2-filled glove box. NiCl2 glyme (11 mg, 0.05 mmol, 10 mol%), 4,4′-di-tert-butyl-2,2′-dipyridyl (14.8 mg, 0.055 mmol, 11 mol%) and benzene (3.5 mL) was added in to 5 mL vial and stirred well for 15 min (Mixture 1). The other reaction tube was charged with Ir[dF(CF3)ppy]2(dtbbpy)PF6 (11.3 mg, 0.01 mmol, 2 mol%), NaHCO3 (84 mg, 1 mmol, 2 equiv.), N-(3-bromopropyl)phthalimide (134 mg, 0.5 mmol, 1 equiv.), toluene (530 µL, 5 mmol, 10 equiv.), benzene (0.5 mL), and stirred. Mixture 1 was then added to the reaction tube, capped with Teflon septum and parafilmed. The reaction tube was removed from S19 the glove box, and irradiated using 34 W x 2 blue LEDs while stirring at RT (under fan cooling to keep the reaction at room temperature). After 96 hours, the reaction was filtered through a small bed of celite and silica and concentrated in vacuo. The pure product was obtained as a white solid in 31% (43 mg) yield after the column chromatography of the crude reaction mixture (silica gel, gradient 0 to 5% EtOAc/petether). Care should be taken while doing column chromatography since a small amount of dehydrogenated product is seen which is getting eluted almost in the same polarity. Although an Rf of 0.5 in 15% EtOAc/petether is seen, a long column was packed with the polarity not exceeding 4 -5% EtOAc/petether was chosen to isolate the product in pure form by running the column slowly for a long time.

Preparation of Ni(dtbbpy)alkyl bromide complex mixture 11-12
All reagents were prepared in-stock solutions (volumetrically) inside a nitrogen-filled glove box.
For the absorption measurements, an appropriate amount of analyte dispensed in THF was added to 2 mL quartz cuvettes, equipped with PTFE stoppers, and sealed with parafilm inside nitrogenfilled glove-box, removed from the glovebox, and the spectrum was collected. Initially, a 1:1 mixture of Ni(cod)2 and 4,4′-di-tert-butyl-2,2′-bipyridyl (dtbbpy) in THF was stirred for 30 min under argon to give deep purple color with broad absorption at 565.5 nm confirming the formation of (dtbbpy)Ni(cod). To that, (3-bromopropyl)benzene (1.1 equiv.) was added and the mixture was stirred until the reaction mixture has changed from deep purple to light color. Complete consumption of Ni(dtbbpy)(cod) was confirmed by absorption measurements which showed the disappearance of the broad absorption of Ni(dtbbpy)(cod) (max at 565.5 nm) and the appearance of a broad absorption with max at 470 nm due to metal-to-ligand-charge-transfer ( 1 MLCT) ( Supplementary Fig. 1). Another strong absorption band in the ultraviolet region (max = 283 nm) due to ligand-centered π→π* transition is also seen.

Preparation of Ni(dOMebpy)alkyl bromide complex mixture 11-12
Initially, a 1:1 mixture of Ni(cod)2 and 4,4′-dimethoxy-2,2′-bipyridyl (dOMebpy) in THF was stirred for 30 min under argon to give deep purple color with broad absorption (max at 567.5 nm) conforming to the formation of (dOMebpy)Ni(cod). To that, (bromomethyl)cyclobutane (1.1 equiv.) was added and the mixture was stirred until the reaction mixture has changed from deep purple to light green color. Complete consumption of Ni(dOMebpy)(cod) was confirmed by absorption measurements which showed the disappearance of the broad absorption of Ni(dOMebpy)(cod) (max at 567.5 nm) and the appearance of a broad absorption with max at 452 nm which we expect due to metal-to-ligand-charge-transfer ( 1 MLCT) ( Supplementary Fig. 2).    The photoluminescence signals were obtained using an automated motorized monochromator. Time-resolved emission data were fit to a single exponential decay to extract the observed rate constant (kobs). Phosphorescence emission spectra and Stern-Volmer plots for each component are given in below.

Stoichiometric reaction of (dtbbpy)Ni-alkyl bromide complex mixture with THF
Reaction in presence of visible light: An oven-dried screw cap reaction tube equipped with a PTFE-coated stir bar was brought into an N2-filled glove box. Ni(cod)2 (27.5 mg, 0.1 mmol, 1 equiv.), 4,4′-di-tert-butyl-2,2′-dipyridyl (26.8 mg, 0.1 mmol, 1 equiv.) and THF (5 mL, 0.02 M) was added and stirred well for 30 min to give a deep purple color solution. To that, (3bromopropyl)benzene (0.1 mmol) was added and the mixture was stirred until the reaction mixture has changed from deep purple to light color with the disappearance of the broad absorption of Ni(dtbbpy)(cod) (max at 565.5 nm) and the appearance of a new broad absorption peak with max at 470 nm due to metal-to-ligand-charge-transfer ( 1 MLCT) (Supplementary Fig. 1). To that, K2CO3 (27.6 mg, 0.2 mmol, 2 equiv.) was added, capped with Teflon septum, and parafilmed. The reaction tube was removed from the glove box, and irradiated using 34 W blue LEDs while stirring at RT.
After 48 hours, the reaction mixture was diluted with ethyl acetate and subjected to Gas Chromatography which does not show product formation.

Kinetic isotope effect by intermolecular competition experiment between a mixture of THF and d8-THF
Kinetic isotope experiments were carried out to understand the nature of the C-H functionalization. An oven-dried screw cap reaction tube and a 5 mL vial equipped with a PTFEcoated stir bar were brought into the N2-filled glove box. Ni(cod)2 (1.37 mg, 5 µmol, 5 mol%), After 48 hours, the reaction was filtered through a small bed of celite and subjected to GC.

Dependence of the reaction rate on light intensity
Dependence of the reaction rate on light intensity is calculated by running the reactions under full and half intensity of the light source at different time intervals and the corresponding yield of 1 is measured by GCMS using dodecane as internal standard. The values suggest that more than one photon is involved in the reaction mechanism.
Supplementary Table 4

Picosecond and nanosecond-TA spectroscopy using Ni(dtbbpy)(o-tolyl)Cl
Ni(dtbbpy)(o-tolyl)Cl. This compound is prepared by the procedure followed by Doyle and coworkers. [11] An oven-dried screw cap reaction tube equipped with a PTFE-coated stir bar was brought into the N2-filled glove box. Ni(cod)2 (275 mg, 1 mmol), 4,4′-di-tert-butyl-2,2′-bipyridine (268.4 mg, 1 mmol), and THF (2.5 mL) was added and stirred for 1 hour at ambient temperature which gave the deep purple color solution. To that, 2-chlorotoulene (6 mL) was added and left to stir for 20 min. The resulting dark red solution was triturated with pentane to give the precipitate which was filtered on a frit, rinsed with pentane, and dried under high vacuum to give the title compound as a light red powder (367 mg, 81%

Computational methods
All the geometries were optimized with the hybrid meta-generalized gradient approximation (meta-GGA) method with Gaussian 09 program packages, 13 using M06 functional. 14

Dexter triplet-triplet energy transfer (TTET)
In principle, two types of intermolecular energy transfer (EnT) mechanisms between *PS(T1) and D(S0), the Förster 19 and the Dexter 20 mechanisms, are possible. In the Förster mechanism, PS(T1) returns to its singlet ground state with the simultaneous excitation of D(S0) to an excited singlet state. Unfortunately, no excited singlet of Ni(II) is suitable to perform the HAT step because none of the allowed excitations populates the LUMO+3 that corresponds to the * Ni-Br bond, see Supplementary Table 5 and Supplementary Fig. 16. Hence the Förster energy transfer is ruled out.
Simply, the EnT from a photocatalyst excited triplet state to the Ni(II) singlet ground state will lead to singlet ground state Ir and excited triplet state Ni(II) that represents two spatially separated spin reversal processes (violating Wigner's spin conversion rules) 21 which cannot be described by Förster Resonance EnT mechanism. However, the Dexter EnT mechanism, which implies a reversal electron transfer to exchange energy between *PS(T1) and D(S0) following the eq 1, offers a pathway to the triplet excited state of the N(II) intermediate involving the LUMO+3. As mentioned above, the LUMO+3 corresponds to the * Ni-Br bond, which can undergo HAT step to perform the desired reaction.
According to equation 1, D(S0) can be excited to any of the excited triplet states having a smaller singlet-triplet gap than the lowest singlet-triplet gap of the photosensitizer. From Supplementary  Fig. 18 it is obvious that D can be excited up to T5 excited states by PS1. Therefore, the HAT transition state is ~ 9 kcal/mol above from the T5 excited state (Fig. 4 in the manuscript). Interestingly, other photocatalysts, PS2, PS3, and PS4, having smaller singlet-triplet energy gaps, are not capable to excite D to an excited state higher than the T4 state, which corresponds to the electron promoted to the π* of bpy ligand, not to the * Ni-Br bond, which is not preferred for active bromine atom generation. Moreover, the HAT step from an excited state lower than T5 would also be kinetically slow, because a very high free energy barrier (~25 kcal/mol) would then be required to reach the transition state for the HAT step. Therefore, following the experimental findings, PS2, PS3, and PS4 photocatalysts are not effective for the C(sp 3 )-H alkylation reaction.

Energy profile for the electron transfer pathway of reductive elimination step
For energy and other conventions refer to Fig. 4 in the manuscript. a better chance to undergo reductive elimination to form the alkyl-alkyl cross-coupling product. But this product is not observed in the experiment. Therefore, these results indirectly predict that the single electron reduction of D S is unfavored, and hence the Ni(I/III) mechanism can be excluded.  In an alternative pathway, the outer-sphere mechanism of Cα−H activation by active Br atom has been considered. In the PES (potential energy surface) energy is gradually increasing along the coordinate of H and Br bond formation (Supplementary Fig. 20). We have selected one structure for single-point energy calculation from the PES where the H−Br bond is already formed. That structure is unstable by 2.3 kcal/mol than the inner sphere transition state [E-F] T . Finally, the outer sphere mechanism by the active Br atom is less favored than the inner-sphere mechanism.

Discussion regarding the triplet states of nickel
The triplet states of Ni(0) and Ni(II) reported in this study are 3 d-d types, which is in agreement with the recent report. 22 Although a very low singlet-triplet energy gap of 4-5 kcal/mol is found for both Ni(0) and Ni(II) bipyridine complexes (A S and D S ), the 3 d-d type of triplet is thermally inaccessible from the ground state. Photoexcitation of ground state singlet leads to the 1 MLCT type excited state, which eventually gives 3 MLCT type triplet. The 3 MLCT is a higher energy state, which eventually undergoes IC (internal conversion) to the most stable 3 d-d state. 22 However, in the current alkylation reaction the 3 d-d state is unproductive.
In TDDFT calculations we desired to have electronic information of higher energy triplet states. The low lying triplets (T1-T4) in the TDDFT calculations are 3 MLCT types (vertical excitation), whereas the D T is 3 d-d type (optimized geometry). Therefore the energy of D T is not comparable with the TDDFT calculated triplet state.