Mechanism of a Luminescent Dicopper System That Facilitates Electrophotochemical Coupling of Benzyl Chlorides via a Strongly Reducing Excited State

Abstract

Photochemical radical generation has become a modern staple in chemical synthesis and methodology. Herein, we detail the photochemistry of a highly reducing, highly luminescent dicopper system [Cu₂] (E_ox* ≈ −2.7 V vs SCE; τ₀ ≈ 10 μs) within the context of a model reaction: single-electron reduction of benzyl chlorides. The dicopper system is mechanistically well defined. As we show, it is the [Cu₂]* excited state that serves as the outer-sphere photoreductant of benzyl chloride substrates; the ground-state oxidized byproduct, [Cu₂]⁺, is electrochemically recycled, demonstrating a catalytic electrophotochemical C−C coupling process.

Graphical Abstract

Photochemistry, often in conjunction with transition-metal catalysis, is growing in prominence in modern synthetic methodology.^1–4 Photochemical activation of widely available electrophiles can afford versatile reactive intermediates, such as organic radicals,^5,6 which can be leveraged in a variety of transformations.^7–10 For instance, a recent focus of a number of laboratories, including our own, has been to partner photochemically generated radical intermediates (R^•) with copper-(II)-bound N-nucleophiles in catalytic, photoinduced N-alkylations (Scheme 1, eqs 1 and 2; N_nuc denotes an amidenucleophile).^9,11–18

Scheme 1.

Electrophotochemical Organohalide Reduction

Production of R^• from alkyl halides is integral to many modern organic transformations,^19–21 and hence there is considerable interest in expanding the types of alkyl halides compatible with R^• generation under synthetically useful conditions.^22–24 Alkyl chlorides, with potentials below −2 V vs SCE, are desirable electrophiles but are challenging to reduce;^22,25 the limited examples of their outer-sphere photochemical activation typically feature harsh conditions.^26–28 Phosphine-supported copper-amide excited states^29–33 can be more reducing than those of typical ruthenium or iridium systems,³⁴ providing a sufficient driving force for alkyl chloride reduction. To promote photoinduced R^• generation via a copper species in a generalized fashion (e.g., avoiding the subsequent C−N coupling step as in Scheme 1, eq 2), the copper byproduct of oxidative quenching must be recycled by a suitable reductant.

In 1987, Sauvage demonstrated an elegant solution to photocatalyst regeneration via the electrophotochemical reduction of 4-nitrobenzyl bromide with [Cu(dap)₂]⁺ (E_ox* ≈ −1.4 V; τ₀ = 0.27 μs; dap = 2,9-dianisyl-1,10-phenanthroline).³⁵ Organic photosensitizers have more recently been used to reduce (pseudo)halides under extremely reducing electrophotoredox conditions (E_ox* < −3 V).^36–39 The suggested lifetimes (τ₀ ≈ 1 ns) and nature of the photoreductant intermediates of these processes are still under investigation.⁴⁰

In this study, we explore a dicopper diamond core system (hereafter [Cu₂]), previously developed by our lab³³ and featuring a combination of terminal phosphine and bridging amide ligands, as an attractive electrophotoredox catalyst (Scheme 1, bottom). [Cu₂] is an especially strong excited-state reductant (E_ox* ≈ −2.7 V), with a long-lived excited state in solution at RT (τ₀ ≈ 10 μs). Charge delocalization by the Cu₂(μ-N)₂ diamond core, as well as steric protection from ligand isobutyl and tert-butyl groups, is expected to render the one-electron-oxidized state [Cu₂]⁺ non-nucleophilic. Furthermore, [Cu₂]⁺ can be electrochemically interconverted with [Cu₂]; [Cu₂]⁺ has been isolated and characterized in the solid state.⁴¹

As a representative study of the excited-state intermolecular photochemistry of Cu^I−amide systems, with an eye toward photoreductions using alkyl chlorides as R^• precursors, we explore herein photochemically driven, electrochemically cycled radical couplings using [Cu₂] and benzyl chloride substates (E_p up to −2.5 V vs SCE). The dicopper system described here is mechanistically well defined, and as we show, it is the [Cu₂]* excited state that serves as the outer-sphere photoreductant of benzyl chloride substrates; the ground-state oxidized byproduct, [Cu₂]⁺, is electrochemically recycled to afford a catalytic, electrophotochemical C−C coupling process.

We began by investigating the reactivity of 4-methylbenzyl chloride (1) (E_p = −2.5 V vs SCE) as a model substrate. Benzyl chlorides are important substrates in modern synthesis and methodology^42–44 and also provide a convenient radical termination pathway via diffusion-limited dimerization, simplifying our mechanistic studies.⁴⁵ On exposure of 1 to blue-light irradiation (440 nm) in 1,2-dimethoxyethane (DME), no reaction is observed. However, when [Cu₂] is added, bibenzyl product 1-D is formed quantitatively (Figure 1A).

Figure 1.

Photoreduction of benzyl chlorides: (A) reaction performed for 2 h with yield analyzed by ¹H NMR versus CH₂Br₂ internal standard; (B) Stern−Volmer quenching; (C) Marcus theory analysis in the presence of various benzyl chloride quenchers.

Benzyl chloride photoreduction was mechanistically interrogated via Stern−Volmer (SV) studies to establish outer-sphere electron transfer (ET) and to probe rates of ET. Time-resolved photoluminescence spectroscopy confirmed that electronically distinct benzyl chlorides 1–8 quench [Cu₂] in a dynamic (i.e., diffusional) process. The rates of quenching, determined from linear SV plots (Figure 1B), were in the range of ~10⁸–10¹⁰ M⁻¹ s⁻¹ for K_SV/τ₀. These values indicate rapid quenching, reaching diffusion-limited values with electron-poor 2. Using benzyl chloride peak potentials obtained from cyclic voltammetry (E_p = −1.7 to −2.5 V; see the Supporting Information), the quenching rates could be analyzed as a function of driving force, using E_ox* ≈ −2.7 V. Notably, a quadratic relationship between log K_SV and driving force was observed, consistent with the behavior predicted by Marcus theory for outer-sphere electron transfer (Figure 1C).⁴⁶ Although such outer-sphere dynamic quenching is commonly assumed in photoredox mechanisms, this contrasts with the behavior of some organic electrophotoredox catalysts hypothesized to involve preassembly of the photocatalyst and substrate to compensate for short lifetimes.³⁸ These photophysical measurements thus indicate a rapid dynamic oxidative quenching step in which [Cu₂] undergoes outer-sphere electron transfer to benzyl chloride electrophiles.

We expected oxidative quenching to produce the stable, red-brown, mono-oxidized species [Cu₂]⁺ (Figure 2A).⁴¹ A 440 nm irradiation of [Cu₂] and 1 in DME produces a pale yellow solution, the UV−vis spectrum of which is mostly featureless (Figure 2B). Thus, the expected UV−vis features for [Cu₂]⁺ at 520, 600, and 800 nm were not observed. Surprisingly, this suggests that the oxidative quenching reaction may involve either degradation following quenching or chemical steps at copper.

Figure 2.

Influence of chloride on oxidized copper products. UV−vis spectra in DME of (A) [Cu₂]⁺ and (B) a mixture of [Cu₂] and 4-methylbenzyl chloride irradiated (440 nm) for 5 min. (C) ³¹P NMR spectra of chloride-bound copper products and (D) their structures.

We hypothesized that the stability of [Cu₂]⁺ could be compromised by chloride, a byproduct of benzyl chloride reductive C−Cl bond cleavage. Accordingly, addition of lithium chloride to a solution of [Cu₂]⁺ in DME resulted in a loss of red-brown color over several hours, producing a yellow solution. Off-white crystals isolated from the reaction mixture were characterized by two ³¹P NMR peaks (Figure 2C), and single-crystal XRD revealed the presence of two independent dimers, each comprised of two CuCl (chloro-cubane) or one CuCl (chloro-diamond) per H-PNP^tBu ligand equivalent: i.e., [(H-PNP^tBu)Cu₂ (μ-Cl)₂]₂ or [(H-PNP^tBu)Cu(μ-Cl)]₂, respectively. An independent synthesis of chloro-cubane and chloro-diamond (see the Supporting Information), produced white solids whose ³¹P NMR resonances reproduced those of the cocrystalline material (Figure 2C), and the characterization of chloro-diamond enabled its identification as a reaction product in the stoichiometric reaction described in Figure 1A (SI).

We sought to detect and track the fate of [Cu₂]⁺ in the presence of chloride via a UV−vis time course analysis, photolyzing [Cu₂] and 1 under 440 nm irradiation (Figure 3A). Bands characteristic of [Cu₂]⁺ grow in throughout 15–30 s, after which the 520 nm absorbance rapidly decreases. This accounts for our failure to observe the presence of [Cu₂]⁺ in Figure 2B. Knowing that chloride in the form of lithium chloride slowly degrades [Cu₂]⁺ over a period of several hours, we investigated whether lithium salts could sequester chloride via tight ion pairing to mitigate degradation of [Cu₂]⁺.⁴⁷

Figure 3.

Stability and regeneration of [Cu₂]⁺. Time course studies for a mixture of [Cu₂] and 1 under 440 nm irradiation. (A) UV−vis spectra and (B) 520 nm absorbance vs time in the presence and absence of 0.2 M LiNTf₂. (C) 77 K EPR spectrum recorded after 15 s of irradiation in the presence of LiNTf₂. (D) UV−vis spectra pre- and postirradiation, as well as after 5 min of −0.15 V applied potential in the dark.

When [Cu₂] and 1 were irradiated in the presence of 0.2 M LiNTf₂, bands for [Cu₂]⁺ became persistent, decreasing in intensity by only ~20% after 20 min (Figure 3B). This is consistent with kinetic measurements which indicate a rate of ~3 × 10⁻² M⁻¹ s⁻¹ for the reaction between [Cu₂]⁺ and tetrabutylammonium chloride in the presence of 0.2 M LiNTf₂; without LiNTf₂, the reaction is almost instantaneous (Supporting Information). An analysis of [Cu₂] photolyzed in the presence of 1 and 0.2 M LiNTf₂ by EPR provided orthogonal support for assigning the product as [Cu₂]⁺ (Figure 3C).⁴⁸ Thus, these analyses indicate [Cu₂]⁺ to be the oxidative quenching product and corroborate its degradation by chloride.

Stabilizing [Cu₂]⁺ enables the prospect of electrochemically regenerating [Cu₂]. [Cu₂]⁺ was photochemically generated from [Cu₂] and 1 in DME, with LiNTf₂ serving as both a chloride sequestrant and the electrolyte, and then transferred into a two-compartment electrochemical cell. Applying E_app = −0.15 V for 5 min using a carbon cloth working electrode, cathodic of E_ox = 0 V for [Cu₂]^0/+, 0.76 e⁻ equivalents of current was passed (Figure 3D). One electron is required to fully reduce [Cu₂]⁺ to its photoactive neutral state; thus, up to 76% could be reduced. The UV−vis spectrum of this solution showed recovery of the 440 nm peak characteristic for [Cu₂], albeit with ~60% of its original intensity, indicating successful, albeit incomplete, regeneration.⁴⁹ An electrochemical analysis of chloro-cubane and chloro-diamond indicated no electron transfer pathway for recovering [Cu₂] at our operating potential, highlighting the importance of stabilizing [Cu²]⁺ and rationalizing the incomplete regeneration of [Cu₂].

The described reactivity of the [Cu₂] system constitutes the requirements for an electrophotoredox cycle (Scheme 1); thus, we turned to catalytic investigations under controlled-potential conditions (Table 1). Indeed, [Cu₂] is a competent electrophotoredox catalyst, generating 1-D from 1 in 89% yield using 3 mol % [Cu₂] (entry 1). Reactions of additional substrates 2–8 proceeded in 68–91% yield (entries 12−18). No reaction was observed in the absence of either [Cu₂] or light (entries 2 and 3). In the absence of an applied potential, only the expected stoichiometric amount of 1-D relative to [Cu₂] was produced (entry 4).

Table 1.

Electrophotocatalytic Benzyl Chloride Reduction^a

Entry	Variation	Yieldb
1	none	90c
2	no [Cu2]	0
3	no light	0
4	no applied potential	2
5	2 equiv. H2O	62
6	5 mL air	4 {7}d
7	LiCIO4 instead of LiNTf2	30
8	TBAPF6 instead of LiNTf2	10
9	TBANTf2 instead of LiNTf2	11
10	chloro-cubane instead of [Cu2]	0
11	chloro-diamond instead of [Cu2]	0
12	2 → 2-D	77c
13	3 → 3-D	77c
14	4 → 4-D	68c
15	5 → 5-D	81c
16	6 → 6-D	77c
17	7 → 7-D	91c
18	8 → 8-D	75c

^aPerformed for 1.5−3 h with 0.15 mmol of benzyl chloride.

^bYields of known products determined by ¹H NMR versus CH₂Br₂ internal standard.

^cAverage of two runs.

^dValue for 4-methylbenzaldehyde in braces.

The intermediacy of benzyl radicals during catalysis is supported by several pieces of circumstantial evidence. The production of 1-D in the presence of added water (Table 1, entry 5) and the dimerization of tertiary and ester-substituted benzyl chlorides are inconsistent with the intermediacy of benzyl anions. Although the reaction is highly sensitive to air due to quenching of [Cu₂]* (entry 6), 4-methylbenzaldehyde becomes the major product (7% yield). Benzaldehydes are known products of the reaction between benzyl radicals and oxygen.^35,50 Attempts to trap benzyl radicals with the radical trap TEMPO were unsuccessful, as TEMPO quenches [Cu₂].⁵¹

The catalytic reaction is very sensitive to factors that alter chloride binding to [Cu₂]⁺. Li⁺ from LiNTf₂ likely interacts with chloride through ion pairing as a Lewis acid; electrolytes expected to exhibit weaker ion pairing with chloride, such as tetrabutylammonium salts, performed notably worse (Table 1, entries 7–9). The poorer performance of LiClO₄ (entry 7) is attributed to the fact that in DME ClO₄⁻ is more tightly associated with Li⁺ than is NTf₂⁻,⁵² possibly limiting the sequestration of Cl⁻. Isolated chloro-cubane and chloro-diamond (Figure 2D) were catalytically inactive under the conditions (entries 10 and 11). Therefore, the detection of chloro-diamond by ³¹P NMR at the end of the standard reaction (entry 1) suggests one pathway by which catalysis ceases.

To close, we have described the electrophotochemical reactivity of [Cu₂] in the presence of benzyl chloride substrates. Our mechanistic studies enable the assignment of facile electron transfer from the excited state [Cu₂]* with substrate to liberate [Cu₂]⁺, Cl⁻, and a benzyl radical that undergoes homocoupling to produce bibenzyl. By tracking down off-path copper-cubane and -diamond chloride sinks, and devising a means of sequestering the chloride produced, we were able to demonstrate the electrophotocatalytic chemistry of interest. Our study complements other recent reports employing organo-photocatalysts for R(Ar)−X electrophotochemical couplings where the nature of the photoreductants are still being studied.