Ligand-to-Metal Charge Transfer of Ag(II) CF₂X Carboxylates: Quantum Yield and Electrophotocatalytic Arene Fluoroalkylation Tuned by X

Abstract

Incorporation of CF₂X groups beyond CF₃ into arene scaffolds is underdeveloped despite these groups’ utility as halogen-bond donors and as precursors to bioisosteres. Herein, we report the synthesis, characterization, and comparative photochemistry of a suite of [Ag(II)(bpy)₂O₂CCF₂X]⁺ and Ag(II)(bpy)(O₂CCF₂X)₂ (bpy = 2,2′-bipyridine, X = F, CF₃, Cl, Br, H, CH₃) carboxylate complexes. We find a dramatic effect of the X substituent on the efficiency of generating CF₂X radicals by ligand-to-metal charge transfer (LMCT), with Ag(II) photoreduction rates varying by over an order of magnitude and quantum yields spanning over 20%. We provide insight into how electronic and structural perturbations of the Ag(II)–O₂CCF₂X core are manifested in the LMCT quantum efficiency. With this information in hand, Ag(II)-mediated electrophotocatalytic CF₂X functionalization is carried out on a range of (hetero)arenes. This work expands the nascent field of Ag(II)-based photocatalysis by allowing for (hetero)aryl–CF₂X functionalization directly from unactivated fluoroalkyl carboxylate precursors.

Graphical Abstract

Fluoroalkylation methods beyond CF₃ incorporation are less common, but highly desirable. Herein, we leverage Ag(II) ligand-to-metal charge transfer (LMCT) to achieve catalytic installation of CF₂X moieties onto (hetero)arenes directly from abundant fluoroalkyl carboxylates. Furthermore, the isolation of a series of well-defined Ag(II) CF₂X carboxylate compounds allows for the study of how LMCT quantum efficiency is modulated by the X substituent.

Introduction

The established benefits of fluorine incorporation into pharmaceuticals^[1,2] have given rise to a variety of strategies to install fluorine into organic molecules. Fluoroalkyl groups feature prominently in FDA-approved pharmaceuticals, with the trifluoromethyl group (CF₃) being among the most prevalent.^[3] Synthetic efforts to incorporate CF₃ groups benefit from the wide availability and well-described reactivity of specialty CF₃ group transfer reagents (Figure 1A).^[4–6] Conversely, fluoroalkyl groups other than CF₃ are less common despite their beneficial properties. For example, CF₂X moieties (X = Cl, Br) can serve as C(sp³)–X halogen bond donors, providing key binding interactions between small molecules and protein targets.^[7] Beyond their utility as functional groups themselves, CF₂X groups can serve as precursors for further elaboration. They can be reduced to CF₂H groups,^[8,9] which are lipophilic hydrogen-bond donors that can act as alcohol or thiol bioisosteres,^[10] or undergo substitution, cross-coupling, or other C–C bond-forming reactions^[11] to furnish R–CF₂–R′ moieties, which can behave as carbonyl bioisosteres.

Figure 1.

(A) Selected examples of common trifluoromethyl group transfer reagents. (B) Fluoroalkyl carboxylates as CF₂X group transfer reagents.

An impediment to installing fluoroalkyl substituents beyond CF₃ into organic molecules arises from a dearth of accessible CF₂X reagents, which either do not exist or require multi-step syntheses to prepare. Current methods for synthesizing aryl–CF₂X^[12–19] or (hetero)aryl–CF₂X^[20–23] functionalized arenes often require pre-installation of an aryl–CX₃ or acyclic CF₂X-functionalized precursor, respectively. The necessity of prefunctionalization in these methods impedes the determination of structure-activity relationships in medicinal chemistry contexts, as each CX₃- or acyclic CF₂X-functionalized precursor must be synthesized individually, in contrast to a diversity-oriented approach in which a single precursor is poised to form several products of interest. More efficient methods that utilize •CF₂X to directly convert aryl–H to aryl–CF₂X are less common.^[9,24–26]

An ideal source of CF₂X groups is their corresponding carboxylates (Figure 1B, left panel), due to their wide availability, low cost, and bench stability. However, the inertness of fluoroalkyl carboxylates, and the consequently harsh conditions required for their oxidation and subsequent decarboxylation, are impediments for their use in synthesis. For example, direct electrochemical oxidation of trifluoroacetate (TFA) requires high anodic potentials (E_ox > 2 V vs SCE) approaching the redox stability windows of common solvents, as well as that of inert arenes such as benzene.^[27,28] To bring the redox activation of fluoroalkyl carboxylates within a workable potential range, the few reported methods for the generation of •CF₂X from fluoroalkyl carboxylate precursors tend to require stoichiometric prefunctionalization of the carboxylate (Figure 1B, middle panel).^[9,24–26] As an alternative strategy, photoredox methods can provide access to the extreme redox potentials needed to furnish radicals directly from the carboxylates under overall mild conditions.^[29–32,33] One such method is ligand-to-metal charge transfer (LMCT) photochemistry which uses light as an energy input to access a LMCT excited-state of an Mⁿ–L complex, allowing for eventual dissociation of a ligand radical, •L, and concomitant formation of a reduced metal center, M^n–1. LMCT photochemistry has been employed to generate a range of radicals from various mid/high-valent metal centers, such as Cu(II),^[34–36] Fe(III),^[37,38] Ce(IV),^[39] Ni(III/II),^[40,41] and V(V).^[42]

We have recently demonstrated the ability of Ag(II) to catalytically oxidize TFA via LMCT photochemistry.^[29] In contrast to Fe(III)-based methods,^{[30,31,33,43]} Ag(II) LMCT decarboxylation proceeds with visible light (λ ≥ 440 nm) at high quantum efficiencies from isolable, well-defined, carboxylate-bound chromophores. The ability to push LMCT photochemistry into the visible spectral region at high quantum efficiencies is important to the sustainability footprint of a photoredox method,^[44,45] as well as the facility for the method’s practical scale-up.^[46] For LMCT decarboxylations at a Ag(II) center, the highly oxidizing Ag(II) ion possesses extremely low-lying metal-centered acceptor orbitals (Figure 2A), allowing for facile transfer of electron density from carboxylate ligands to the metal using lower energy visible light rather than the higher energy UV or near-UV excitation typically necessary for LMCT reactivity. Given our ability to prepare well-defined Ag(II) TFA complexes that are active catalysts for (hetero)arene trifluoromethylation reactions, we sought to generate a wider array of CF₂X radicals from an expanded palette of Ag(II) CF₂X carboxylate complexes (Figure 1B, right panel). This would not only afford access to fluoroalkylations reactions beyond CF₃ incorporation, but would also allow for LMCT structure-function relationships to be examined. Specifically, we explore the dependence of ligand/metal redox potential on the efficiency of LMCT photochemistry involving metal–ligand bond homolysis, which to date has remains ill-defined.

Figure 2.

(A) Schematic of LMCT photochemistry: a simplified molecular orbital diagram (metal s and p orbitals not shown, σ-bonding only) for a generic D_4h metal complex is shown with an LMCT transition indicated by excitation of an electron in a ligand-centered orbital to a metal-centered orbital. The energies of these orbitals are correlated to the oxidation potential of the ligand and the reduction potential of the metal. (B) The Ag(II) CF₂X carboxylate complexes responsible for fluoroalkylation reactivity exhibit an inverse correlation between LMCT quantum yield and ligand oxidation potential. (C) This work: electrophotocatalytic fluoroalkylation using a Ag catalyst with inexpensive and abundant fluoroalkyl carboxylates.

We now report a suite of well-defined Ag(II) CF₂X (X = F, CF₃, Cl, Br, H, CH₃) carboxylates and determine how the LMCT quantum efficiency of carboxylate bond homolysis is altered as the electronic structure of the Ag(II)-O₂CCF₂X core is perturbed by X (Figure 2B). Building on our understanding of the photochemistry of Ag(II) fluoroalkyl carboxylates, we further develop a general electrophotocatalytic method to induce the chlorodifluoromethylation of a wide array of arenes, including both five- and six-membered heteroarenes, as well as the bromodifluoromethylation and difluoroethylation of a narrower scope of simple arenes or electron-deficient heteroarenes, respectively (Figure 2C). This method simplifies the installation of these underutilized but useful functional groups by directly converting aryl–H to aryl–CF₂X using widely available carboxylates as the CF₂X source, without the need for pre-activation of the arene or the carboxylate. In particular, this method provides convenient access to •CF₂Br, enabling the direct synthesis of aryl–CF₂Br from aryl–H, a transformation for which general methods did not previously exist. Ultimately, the results shown here add to a growing body of electrophotocatalytic decarboxylative radical transformations,^[47–49] and furthermore demonstrate the synthetic utility of Ag(II) LMCT photochemistry as a general platform for radical generation from redox-reticent substrates.

Results and Discussion

Synthesis and characterization of Ag(II) CF₂X carboxylates

Although Ag(II) complexes are rare,^[50] we recently developed a high-yielding procedure to prepare highly oxidizing, bench-stable, and organic-solvent-soluble [Ag(bpy)₂]²⁺ (bpy = 2,2′-bipyridine) complexes.^[29] The [Ag(bpy)₂][OTf]₂ (1[OTf]₂) complex serves as a starting reagent for synthesis of Ag(II)-bound carboxylate compounds. Binding of each carboxylate to [1]²⁺ was assessed through titration experiments monitored by UV-visible absorption spectroscopy. The titration of bis(triphenylphosphine)iminium (PPN) fluoroalkyl carboxylate salts to solutions containing 1[OTf]₂ led to significant UV-vis spectral evolutions consistent with carboxylate binding to the Ag(II) center (Figures S21–S23). While evidence for the formation of multiple Ag(II)–O₂CCF₂X complexes can be seen in the titration data for X = Cl, Br, and CH₃, the carboxylate binding appeared to be weaker (and cation dependent) for X = H (Figure S24). On a preparative scale, complexes [1–O₂CCF₂X][OTf] and 2–(O₂CCF₂X)₂ were obtained by the two general routes shown in Scheme 1. Treatment of 1[OTf]₂ in dichloromethane (CH₂Cl₂) with the appropriate [PPN][O₂CCF₂X] salt (1 equiv) in the presence of bpy (1 equiv) led to the formation of [1–O₂CCF₂X][OTf] (X = Cl, Br, H, CH₃), which could be crystallized as black needles upon vapor diffusion of pentane at −35 °C. Alternatively, treatment of 1[OTf]₂ in CH₂Cl₂ with excess [PPN][O₂CCF₂X] (3–5 equiv) followed by addition of pentane led to the precipitation of 2–(O₂CCF₂X)₂ (X = Cl, Br, CH₃) as microcrystalline orange powders. Conversely, for X = H, even in the presence of excess carboxylate, only [1–O₂CCF₂H][OTf] could be isolated, consistent with the observed titration data. With the exception of 2–(O₂CCF₂CH₃)₂, which is unstable in solution and crystallized as a mixture containing decomposition products, each complex could be isolated as an analytically pure solid. We previously reported the syntheses of the [1–O₂CCF₃][OTf] and 2–(O₂CCF₃)₂ complexes, which are employed in this work as points of comparison against the new Ag(II) CF₂X (X = Cl, Br, H, and CH₃) carboxylate complexes.²⁹ While attempts to synthesize [1–O₂CCF₂CF₃][OTf] were unsuccessful, its synthesis as the tetrafluoroborate salt ([1–O₂CCF₂CF₃][BF₄] is accomplished here starting from 1[BF₄]₂ via the route shown in Scheme 1. Attempts to synthesize [1–O₂CCF₂I][OTf] were unsuccessful, as treatment of [1][OTf]₂ with [Na][O₂CCF₂I] or [PPN][O₂CCF₂I] resulted in decomposition of [1][OTf]₂ and formation of precipitates. Among the decomposition products, a Ag(I) iodide complex was identified (Figure S11).

Scheme 1.

Synthesis of Ag(II) carboxylate complexes.

Single-crystal X-ray diffraction analysis^[51] of the [1–O₂CCF₂X]⁺ series confirms their five-coordinate structure (Figures 3A–3D and S1–S5, Tables S1 and S2). Inspection of the primary coordination sphere of each complex reveals a coordination geometry best described as intermediate between trigonal bipyramidal and square pyramidal (0.45 < τ₅ < 0.71).^[52] The Ag–O1 bond distances are all characteristically long (2.25–2.43 Å), similar to the previously reported [1–O₂CCF₃][OTf] (2.4814(15) Å). The carboxylate ligands assume a κ₁ binding mode, with secondary Ag–O1 interactions that exceed 2.8 Å. As the X group becomes more electronegative, the carboxylate becomes a weaker donor ligand, and accordingly d(Ag–O1) in the [1–O₂CCF₂X][OTf] complexes generally increases along the series X = CH₃ < Br < Cl < CF₃ < F (Figure S6); however, [1–O₂CCF₂H][OTf] displays a longer than expected d(Ag–O1) of 2.363(2) Å. Inspection of the crystallographic environment around the difluoromethyl carboxylate ligand reveals hydrogen bonding interactions between the CF₂–H unit and nearby oxygen atoms of the triflate counterions (Figure S7), which plausibly contribute to the increased distance of the carboxylate from the Ag center. In general, the long Ag(II)–carboxylate bonds across the series of complexes suggest weak interactions that are likely substantially influenced by crystal packing effects. This is evidenced by the observation of two crystallo-graphically distinct molecules in the asymmetric unit of the crystal structure of [1–O₂CCF₂Br][OTf] with d(Ag–O1) values that differ by 0.101(3) Å (Figure S3).^[53] Single-crystal X-ray diffraction analysis of 2–(O₂CCF₂X) (X = Cl, Br, CH₃) (Figures 3E–3F and S8–S10) confirms their four-coordinate square planar nature (0.09 < τ₄ < 0.12). ^[54] The asymmetric units of 2–(O₂CCF₂X) feature tight Ag–O bonding interactions d(Ag–O1) and d(Ag–O3) that range between 2.14–2.16 Å. The 2– (O₂CCF₂X)₂ complexes pack as inversion-symmetric dimers in the solid-state, similar to the previously reported 2–(O₂CCF₃)₂.

Figure 3.

Solid-state structures of (A) [1–O₂CCF₂Cl][OTf], (B) [1–O₂CCF₂Br][OTf], (C) [1–O₂CCF₂H][OTf], (D) [1–O₂CCF₂CH₃][OTf], (E) 2–(O₂CCF₂Cl)₂, and (F) 2–(O₂CCF₂Br)₂ as ascertained by single-crystal X-ray diffraction analysis. Atomic displacement parameters drawn at 50% probability level. Ag, light gray; Br, maroon; Cl, dark green; F, light green; O, red; N, blue. Hydrogen atoms shown as white spheres. Counterions, solvent molecules, and hydrogen atoms on bpy ligands are omitted for clarity.

The stabilities of complexes [1–O₂CCF₂X]⁺ and 2–(O₂CCF₂X)₂ in solution are dramatically affected by the identity of the bound carboxylate. While [1–O2CCF₂X]⁺ (X = F, CF₃, Cl) are relatively stable in acetonitrile (CH₃CN) solutions at room temperature over 12 h, as observed by UV-vis spectroscopy, [1–O₂CCF₂X]⁺ (X = H, CH₃, Br) decay much more rapidly with observed rate constants increasing from Br < H < CH₃ (Figure S25A). The stability of these complexes improves significantly at 10 °C (Figure S25B). Complexes 2–(O₂CCF₂X)₂ are less stable than their five-coordinate counterparts. The UV-vis-NIR features of 2–(O₂CCF₃)₂ degrade by ca. 30% over 12 h, whereas 2–(O₂CCF₂Cl)₂ and 2–(O₂CCF₂Br)₂ completely decompose over the same time frame (Figure S26). This lack of stability for the Ag(II) CF₂X carboxylate complexes for X = Br, H, or CH₃ could be a potential factor in the diminished electrophotocatalytic reactivity observed for these carboxylates (vide infra).

To investigate potential causes of the decreased stability of the complexes when moving from X = F to X = CH₃, we undertook a series of electrochemical measurements. The presence of CF₂X carboxylates in the primary coordination sphere of 1[OTf]₂ substantially alters the electrochemical properties of the complex (Figure 4A). Complex 1[OTf]₂ displays a single, reversible Ag²⁺/Ag⁺ couple at ca. 900 mV vs Fc⁺/Fc (Fc = ferrocene, all potentials hereafter are vs Fc⁺/Fc), showcasing the powerful oxidizing nature of Ag(II). The perturbation of the redox couple by the fluoroalkyl carboxylate was assessed by recording the cyclic voltammogram of in situ generated [Ag(bpy)₂]⁺ (from a mixture of AgOTf and bpy) with titration of the fluoroalkyl carboxylate. A gradual cathodic shift of the Ag²⁺/Ag⁺ couple is observed with increasing carboxylate equivalents (Figures S28–S33) until the invariant cyclic voltammogram shown in Figure 4A is attained. Furthermore, the cyclic voltammograms of the isolated [1–O₂CCF₃]⁺ and 2–(O₂CCF₃)₂ complexes exhibit redox waves that fall intermediate between those of the parent [1]²⁺ complex and those observed in the presence of excess carboxylate (Figure S34). These observations are consistent with weak and fluxional binding of each carboxylate. The Randles-Ševčík plots establish an electrochemically reversible couple (Figures S35–S39). The largest cathodic shift is observed for the more electron-rich carboxylates, ⁻O₂CCF₂H and ⁻O₂CCF₂CH₃, indicating stronger donation into the Ag(II) center by these carboxylates, thereby quenching the metal ion’s electrophilicity. To explore ground-state oxidation as a potential reason for instability, linear sweep voltammograms (LSVs) for the sodium salts of each carboxylate were measured Figure 4B. The peak oxidation potentials of the carboxylates occur at significantly more anodic potentials than that of 1[OTf]₂, thus opening a redox window for the design of an electrophotocatalytic decarboxylation cycle (vide infra).

Figure 4.

(A) Cyclic voltammograms for in situ generated [Ag(bpy)₂]⁺ ([Ag] = 1 mM) and Ag(bpy)(O₂CCF₂X) complexes. (B) Linear sweep voltammograms for 2 mM [Na][O₂CCF₂X] (X = F, CF₃, Cl, Br, H, CH₃). Working electrode, glassy carbon; counter electrode, Pt wire; reference electrode, Ag/AgCl; scan rate: 100 mV s⁻¹; supporting electrolyte, ⁿBu₄NPF₆ (0.15 M).

The [1–O₂CCF₂X]⁺ complexes exhibit nearly identical absorption features (Figure S27), namely, a broad shoulder at 450 nm (ε₄₅₀ = 1210–1560 M⁻¹ cm⁻¹) and a weakly absorbing d-d transition at 700 nm (ε₇₀₀ = 100–200 M⁻¹ cm⁻¹). The corresponding 2–(O₂CCF₂X)₂ complexes are weaker absorbers at 450 nm (ε₄₅₀ = 990–1050 M⁻¹ cm⁻¹) and lack the 700 nm absorption band that is characteristic of the five-coordinate [1–O₂CCF₂X]⁺ complexes. LMCT absorption features are expected to correlate with the difference in redox potential of the ligand and metal.^[55] However, we do not observe this correlation, despite the carboxylate redox potentials spanning a 500 mV range. The nearly identical absorption energies for [1–O₂CCF₂X]⁺ suggest with a weaker, ionic association of the carboxylate with Ag(II), consistent with the longer carboxylate Ag(II) distances observed in the solid state. While little variation is seen in the electronic absorption spectra across the series of compounds, dramatic changes are observed in their LMCT photochemistry (vide infra).

LMCT photochemistry

With confirmation that the well-defined Ag(II) CF₂X carboxylate complexes are formed across the series of X substituents, we sought to examine their photolytic activity. Changes in the solution-state FT-IR spectra upon the irradiation of [1–O₂CCF₂X][OTf] with visible light (Figure S40) are consistent with the LMCT photoreaction shown in Figure 5A. Asymmetric and symmetric carboxyl stretches at 1685 and 1360 cm⁻¹, respectively, of [1–O₂CCF₂Cl][OTf] in CD₃CN gradually disappear with irradiation, and a new peak prominently appears at 2341 cm⁻¹, which is attributed to CO₂ (Figure 5B).^[56] Taken together, the FT-IR data for photolyzed [1–O₂CCF₂Cl][OTf] solutions support the decarboxylation of ⁻O₂CCF₂Cl to release CO₂ and •CF₂Cl upon LMCT excitation. The FT-IR spectra of the remaining complexes in the [1–O₂CCF₂X]⁺ series show similar changes upon steady-state photolysis (Figures S41–S44). FT-IR difference spectra of the [1–O₂CCF₂X]⁺ complexes are compared in Figure 5C. Whereas relatively small growth of higher frequency signals (1700–1900 cm⁻¹) are observed for [1–O₂CCF₂X]⁺ (X = CF₃, Cl, Br, and CH₃), a large, sharp signal appears at 1756 cm⁻¹ during photolysis of the X = H complex. This stretch is attributed to difluoroacetic acid (HO₂CCF₂H), which suggests that the •O₂CCF₂H radical formed from LMCT undergoes H-atom transfer (HAT) competitively with decarboxylation. Consistently, less CO₂ is observed following photolysis of [1–O₂CCF₂H][OTf] as compared to the other five-coordinate complexes (Figure 5C). The 2–(O₂CCF₂X)₂ complexes exhibit similar FT-IR spectral changes upon photolysis (Figures S45–S47). However, for these complexes, the asymmetric carboxyl stretch does not bleach completely, consistent with the retention of one intact carboxylate following LMCT. The sharpened asymmetric carboxyl stretch coincides with that of the trigonal planar Ag⁺ complexes, Ag(bpy)(O₂CCF₂X) (Figure S47).

Figure 5.

(A) Schematic of LMCT-induced decarboxylation of the [1–O₂CCF₂X][OTf] complexes. IR-active functional groups of interest are indicated by color (green: triflate; blue: carboxyl; purple: CF₂X; red: CO₂). (B) Steady-state FT-IR spectral evolution of a solution of [1–O₂CCF₂Cl][OTf] (5 mM, CD₃CN) under irradiation (λ_exc = 440 nm). Assigned IR stretches are labeled with the color-coded stars shown in (A). (C) Steady-state FT-IR difference spectra for solutions (5 mM, CD₃CN) of [1–O₂CCF₂CF₃][BF₄] (cyan lines), [1–O₂CCF₂Cl][OTf] (green lines), [1–O₂CCF₂Br][OTf] (red lines), [1–O₂CCF₂H][OTf] (blue lines), and [1–O₂CCF₂CH₃][OTf] (yellow lines) under irradiation (λ_exc = 440 nm).

The formation of fluoroalkyl radicals (•CF₂X) and Ag(I) as primary photoproducts (Figure 5A) is reflected in complementary spectroscopies. The observed loss of axial EPR signals upon irradiation of [1–O₂CCF₂X]⁺ and 2–(O₂CCF₂X)₂ is consistent with the generation of diamagnetic Ag(I) (Figures S48–S54), which was further validated by NMR spectroscopy. While solutions of [1–O₂CCF₂X]⁺ or 2–(O₂CCF₂X)₂ are ¹H NMR silent, irradiation produces new aryl resonances attributable to bpy ligands (Figures S55–S59). Furthermore, the appearance of HCF₂X and CF₂X₂ (X = Cl, Br) in ¹⁹F spectra (Figures S60–S61) are consistent with •CF₂X-derived products resulting from radical abstraction reactions.^[57]

Figure 6A shows the UV-vis spectral changes of [1–O₂CCF₂Cl][OTf] upon irradiation with 450 nm light. Similar spectra for the remaining [1–O₂CCF₂X]⁺ and 2–(O₂CCF₂X)₂ complexes are shown in Figures S65–S71. Complete bleaching of the visible absorption bands (inset of Figure 6A) is observed following irradiation, consistent with the generation of the Ag(I) photoproduct. In the UV spectral region, a shoulder at ca. 345 nm (ε₃₄₅ = 5000 M⁻¹ cm⁻¹) bleaches to baseline and a peak at 292 nm (ε₂₉₂ = 31000 M⁻¹ cm⁻¹), attributed primarily to bpy π–π* transitions, blue shifts and decreases in intensity. A similar ligand-centered transition at 244 nm (ε₂₄₄ = 21000 M⁻¹ cm⁻¹) also decreases in intensity with the concomitant growth of a new feature at 237 nm. Overall, four isosbestic points at 281, 260, 237, and 229 nm are observed during photolysis, suggesting clean conversion of [1–O₂CCF₂Cl][OTf] to the Ag(I) photoproduct. In contrast, [1–O₂CCF₂H][OTf] exhibits distinct photochemical reactivity from the rest of the [1–O₂CCF₂X]⁺ series (Figure S68). At early time points, similar bleaching of all visible absorption features is observed while a band at 373 nm persists (Figure S72A). Upon continued irradiation, this feature steadily decreases with the rise of a new broad absorption band extending from ca. 600–850 nm, with an isosbestic point at 411 nm (Figure S72B). Attempts to isolate this photoproduct were unsuccessful. While the 2–(O₂CCF₂X)₂ complexes behave similarly to the five-coordinate complexes, early time points are characterized by the growth of the spectral feature at 700 nm before subsequent decay of this feature at later time points (Figures S70–S71), indicating transient formation of [1–O₂CCF₂X]⁺ during photolysis. Though the above photochemical studies show that both the 5- and 4-coordinate complexes undergo LMCT reactivity with visible light to produce •CF₂X, under catalytic conditions (vide infra), when an excess of carboxylate is present, 2–(O₂CCF₂X)₂ complexes are likely the major Ag(II) photoreagents in solution as showcased by the UV-vis titration data (Figures S21–S24). Further insight into the photochemistry of the Ag(II) complexes was provided by time-resolved femtosecond absorption spectroscopy. The extremely short-lived excited-state lifetime measured for 1[OTf]₂ (τ < 20 ps) indicates that the carboxylate must be within the primary coordination sphere to undergo electron transfer (Figure S73).

Figure 6.

(A) Steady-state irradiation (λ_exc = 450 nm) of [1–O₂CCF₂Cl][OTf] (0.13 mM, CH₃CN) at 23 °C. Spectra were recorded at the following intervals: 0, 1, 2, 3, 4, 5, 7.5, 10, 15, 20, 30 min. Inset shows an expanded visible region of the UV-vis spectrum. (B) Absorbance vs time plots for the decay of solutions (ca. 0.1 mM, CH₃CN) of [1–O₂CCF₃][OTf] (purple), [1–O₂CCF₂CF₃][BF₄] (cyan), [1–O₂CCF₂Cl][OTf] (green), [1–O₂CCF₂Br][OTf] (red), [1–O₂CCF₂H][OTf] (blue), and [1–O₂CCF₂CH₃][OTf] (yellow) under irradiation (λ_exc = 450 nm, power = ca. 50 mJ/s) at 10 °C. (C) Photoreduction quantum yield of [1–O₂CCF₂X]⁺ at λ_exc = 440 nm at 10 °C. Error bars represent the standard deviation of experiments performed in triplicate. (D) Action spectra of [1–O₂CCF₃][OTf] and [1–O₂CCF₂Cl][OTf] at 23 °C; quantum yields for [1–O₂CCF₃][OTf] (purple circles) and [1–O₂CCF₂Cl][OTf] (green circles) are superimposed on the normalized UV-vis spectrum of [1–O₂CCF₃l][OTf] (purple line) and [1–O₂CCF₂Cl][OTf] (green dashed line). (E) Overlay of the transient absorption kinetics of [Ru(bpy)₂]²⁺ in H₂O at 490 nm (gray trace) as a reference, as well as [1–O₂CCF₃][OTf] (purple trace) in CH₃CN at 400 nm. The TA kinetics of [1–O₂CCF₃][OTf] were smoothed by adjacent averaging over 25 points. Per pulse energy = 60 μJ.

The rates of [1–O₂CCF₂X]⁺ photoreduction were measured upon 450-nm laser excitation by tracking the visible absorption decay over time; the photoreaction was performed at 10 °C to ensure thermal stability of the complexes (Figure S25B) during the course of photolysis. The observed lifetimes (τ) varied by an order of magnitude, with [1– O₂CCF₂CH₃][OTf] decaying at the fastest rate (τ = 27.4 s) and [1–O₂CCF₃][OTf] decaying the slowest (τ = 440 s) (Figure 6B). For [1–O₂CCF₂H][OTf], irradiation was constrained to early time points in which the visible absorption contributions from the unknown photoproducts are minimized.

Photoreduction quantum yields (ϕ_red) were calculated from the observed lifetimes following previously reported methods (Supporting Information, Section G).^[58] Due to the thermal instability and complicated speciation of the 2–(O₂CCF₂X)₂ complexes, quantification of their quantum yields was not undertaken. Table 1 tabulates the measured quantum yields at a single wavelength for the [Na][O₂CCF₂X] complexes alongside the peak oxidation potentials (E_p) of their associated carboxylate ligand as well as action spectral data for for [1–O₂CCF₃]⁺ and [1–O₂CCF₂Cl]⁺ complexes. The increase in ϕ_red across the series (F ~ CF₃ < Cl < Br << H ~ CH₃, Figure 6C) is correlated with the oxidation potentials of the carboxylate ligands (Figure 4B), with ϕ_red increasing as the carboxylate ligand becomes easier to oxidize. The jump in ϕ_red between X = Br and X = H is reflected in the large difference in oxidation potential between [Na][O₂CCF₂Br] and [Na][O₂CCF₂H] (ΔE_ox = ca. 200 mV). To investigate how ϕ_red changes with irradiation wavelength, action spectra were collected at room temperature for both [1–O₂CCF₃][OTf] and [1–O₂CCF₂Cl][OTf]. For both complexes, the action spectra show an increase in ϕ_red from 525 nm to 340 nm (Figure 6D); we note a modest temperature dependence on ϕ_red is observed between 10 and 23 °C measurements (Figure S94). For [1–O₂CCF₃][OTf], ϕ_red appears to level off near 340 nm, as we previously observed,^[29] whereas ϕ_red for [1–O₂CCF₂Cl][OTf] shows no signs of leveling. Attempts to measure ϕ_red at λ < 340 were not possible due to competitive absorption of photogenerated Ag(I) products. ϕ_red was also measured for [1–O₂CCF₃][OTf] at λ_exc = 355 nm using nanosecond transient absorption spectroscopy with [Ru(bpy)₃]²⁺ as a reference (Supporting Information, Section G).^[59,60] A persistent bleach allows for a ϕ_red determination of ca. 8% (Figure 6E), in agreement with the steady state data The bleach signal persists for >3 μs without recovery, indicating that parasitic processes on ϕ_red, such as back electron transfer, occur on a faster timescale than ca. 20 ns.

Table 1.

Steady state photoreduction quantum yields (ϕ_red): (left) at 450 nm and 10 °C, compared to the peak oxidation potential for [Na][O₂CCF₂X] (E_p, V vs Fc⁺/Fc) for the [1–O₂CCF₂X]⁺ series (right) at various excitation wavelengths for [1–O₂CCF₃]⁺ and [1–O₂CCF₂Cl]⁺ at 23 °C (right side). Error represents the standard deviation of experiments performed in triplicate.

Single wavelength			Action spectra
X	EP (V)	ϕred (%), 450 nm, 10 °C	λexc/nm	ϕred (%), 23 °C
F	1.79	1.72 ± .05		X=F	X=CI
cf 3	1.71	1.81 ± 0.07	340	7.67 ± 0.35	13.14 ± 0.28
Cl	1.60	3.72 ± 0.04	370	6.38 ± 0.52	9.54 ± 0.09
Br	1.57	4.90 ± 0.14	405	4.94 ± 0.07	8.02 ± 0.43
H	1.28	21.37 ± 0.21	450	2.22 ± 0.09	4.36 ± 0.09
ch 3	1.20	21.62 ± 0.55	525	0.66 ± 0.02	1.71 ± 0.02

The high ϕ_red values reported here are significant, as quantum yields have been recognized as a determinant factor in the energy costs of photoredox methods.^[46] Notably, appreciable quantum yields are maintained even with lower energy green light. These high photoreduction quantum yields, in both the visible and UV spectral regions, likely contribute to the fast reaction times observed for the Ag(II)–O₂CCF₂X complexes compared to related methods involving LMCT. Notwithstanding, the observed quantum yields only measure the efficiency of charge transfer from bound fluoroalkyl carboxylates to Ag(II). Subsequent steps important for fluoroalkylation reactivity, such as •CF₂X radical addition rates, can vary across the X series with trends differing from that observed for LMCT quantum efficiency. To probe the efficiency of these subsequent chemical steps, we undertook stoichiometric fluoroalkylation experiments using the isolated [1–O₂CCF₂X][OTf] complexes.

In the presence of excess benzene, [1–O₂CCF₂Cl][OTf] is rapidly consumed (20 min) under 440 nm irradiation, yielding (chlorodifluoromethyl)benzene in 45% yield (based on 1 equiv Ag), which is close to the theoretical maximum of 50% yield for this stoichiometric reaction, as 1 equiv of Ag(II) is needed to re-aromatize the ring (Figure S95A). The only other observed product by ¹⁹F NMR spectroscopy is (H)O₂CCF₂Cl. In the absence of light, a 10% yield of (chlorodifluoromethyl)benzene is obtained after 20 h of heating at 60 °C, and degradation of the carboxylate to unidentified side products is observed (Figure S95B). Similarly, efficient production of (bromodifluoromethyl)-benzene is only observed under illumination conditions, in which 37% yield was achieved during 20 min of irradiation (λ_exc = 440 nm) with minimal byproducts (4% CF₂Br₂). If the reaction mixture is instead heated in the dark (60 °C, 18 h), only trace (bromodifluoromethyl)-benzene is produced (Figure S96). In contrast to [1–O₂CCF₂X][OTf] (X = Cl, Br), only trace amounts of benzene functionalization by CF₂X is observed upon irradiation of [1–O₂CCF₂H][OTf] or [1–O₂CCF₂CH₃][OTf] despite the higher ϕ_red values for these complexes (Figures S97–S98). It is known that the nucleophilic •CF₂H radical does not add efficiently to benzene,^[61] as radical polarity is known to play a significant role in dictating the kinetics of radical addition.^[62] Therefore, we attempted CF₂X (X = H, CH₃) functionalization of the more electron-deficient substrate 1-methylquinoxalin-2(1H)-one, given the nucleophilicity of the •CF₂H and •CF₂CH₃ radicals. Small quantities of the CF₂X-functionalized product were observed for X = H (5%, Figure S99) with higher quantities obtained for X = CH₃ (18%, Figure S100). Competing decomposition pathways likely contribute to the low yields of arene difluoromethylation (vide supra). Notably, a significant amount of difluoroacetic acid is observed following the LMCT of [1–O₂CCF₂H][OTf], suggesting that either the relative rate of HAT of the CF₂H carboxyl radical is faster than the other fluoroalkyl carboxyl radicals studied here, or its decarboxylation rate is slower. While [1–O₂CCF₂CH₃][OTf] possess the highest ϕ_red, it is also the most thermally unstable (vide supra), suggesting that low difluoroethylation yields may be observed due to competing thermal degradation pathways intrinsic to the Ag(II)–O₂CCF₂CH₃ complexes. This observation is consistent with the lack of chloro- and bromodifluoromethylation when complexes [1–O₂CCF₂Cl][OTf] and [1–O₂CCF₂Br][OTf] are heated in the presence of benzene (Figures S95–S96).

Electrophotocatalysis

Building upon the observed photochemical behavior and stoichiometric fluoroalkylation reactivity of the [1–O₂CCF₂X][OTf] complexes, the scope of Ag(II)-mediated fluoroalkylation under catalytic conditions was explored by performing the photolyses under an applied potential in a divided electrochemical cell to electrochemically oxidize the Ag(I) photoproduct to regenerate Ag(II). We sought to improve upon the reaction conditions utilized for X = F and X = CF₃ in our previous report.^[29] Starting from simple Ag(I) salts instead of the preformed 1[OTf]₂, as well as utilizing lower ligand loadings and carboxylate equivalents (Table S4), the electrophotocatalytic chlorodifluoromethylation of benzene proceeded in moderate yields with turnover numbers exceeding 20 (Table 2, entry 1). Under this improved condition, the trifluoromethylation (58% = 53% mono + 5% bis) and pentafluoroethylation (83% = 72% mono + 11% bis) of benzene also proceeded in good yield. Constant current electrolyses could also be performed and afforded product in slightly diminished yield (entry 2). Control experiments indicated that light, bpy, and Ag(I) were all required for productive reactivity (Entries 3–5). When AgOTf was substituted with Cu(OTf)₂, no product was detected (Entry 6). This lack of reactivity is not due to an inability to form fluoroalkyl-carboxylate-ligated Cu complexes, as several of these complexes were isolated and crystallographically characterized (Figures S12–S13). Rather, the photoredox chemistry is obviated by the absence of accessible LMCT bands in the visible region (Figure 2A), consistent with the weaker oxidizing power of Cu(II) as compared to Ag(II).

Table 2.

Reactions were performed in a divided, three-electrode electrochemical cell with benzene (4a, 0.5 mmol) as the limiting reagent. Yields were determined by ¹⁹F NMR spectroscopy of the crude reaction mixtures, using PhOCF₃ as the internal standard. Constant potential reactions were halted at 6 h or when the measured current dropped to less than 1 mA. Constant current reactions were halted when a sharp increase in potential was observed.

Entry	Deviation from optimal conditions	Yield (%)
1	None	57 = 49 mono + 8 bis
2	Constant current: 7.5 mA	50 = 45 mono + 5 bis
3	No Ag	0
4	No bpy	0
5	No light	0
6	Cu(OTf)2 instead of AgOTf	0

Chloro- and bromodifluoromethylation, as well as difluoroethylation, may be extended beyond benzene to include other arenes, including pharmaceutically relevant heteroarenes (Table 3). A pyrazole substructure (4f), which is among the most common five-membered heteroarenes in FDA-approved pharmaceuticals,^[63–65] was efficiently chlorodifluoromethylated, predominantly at the 5-position. Medicinally prevalent six-membered heteroarenes, including pyrazine (4g), pyrimidine (4h), pyrimidone (4i), and uracil (4j) and quinoxalone (6) were also functionalized effectively. A variety of functional groups were accommodated, including aryl halides (4d–e, 4g, 5c–d), fluoroalkyl halides (4b, 5b), and N–H heterocycles (4i). Additionally, despite the potential for H-atom abstraction at benzylic positions under radical reaction conditions, benzylic protons (4c) were well-tolerated. The radical reaction mechanism resulted in the formation of multiple regioisomers or difunctionalized products from some substrates, most of which could be chromatographically separated. For substrates of lower yields, we note that there are currently few options to synthesize these products, reflecting the difficulty associated with these reactions. Indeed, for many of these CF₂X functionalized (hetero)arenes, this is the first report of their synthesis and comprehensive characterization (e.g., ortho-7b, meta-7b, bis-7c, 5–7f, ortho-7e, 7g, bis-7g, 7h, 7i, ortho-8b, meta-8b, meta-8c, ortho-8d, ortho-8f, meta-8f).

Conclusion

Fluoroalkyl carboxylates are a widely available and general source of fluoroalkyl radicals, and the results presented here indicate that these ideal CF₂X sources can be activated by Ag(II) complexes. A series of bipyridyl Ag(II)–O₂CCF₂X (X = F, CF₃, Cl, Br, H, CH₃) complexes undergo LMCT photoreactivity under visible light irradiation, leading to decarboxylation of the ligated ⁻O₂CCF₂X anion to liberate CO₂ and •CF₂X radicals, allowing for the straightforward installation of the CF₂X group via C(sp²)–C(sp³) bond formation. The isolable nature of these complexes permits the study of how the quantum efficiency of bond homolysis systematically varies as a function of ligand identity. We find that the LMCT quantum yields scale with ligand oxidation potential. With the ability to tune the electronic properties of well-defined Ag(II) complexes, we anticipate that they will continue to find further use as LMCT photocatalysts for the activation of redox-reticent substrates.

Supplementary Material

Materials and methods, synthetic procedures, crystallographic data, UV-vis, EPR, FT-IR, quantum yield data, and NMR spectra. The authors have cited additional references within the Supporting Information.^[63–81]

Supporting information for this article is given via a link at the end of the document.

Figure 7.

Electrophotocatalytic CF₂X functionalization. Reactions were performed in a divided, three-electrode electrochemical cell with arene (4, 5, or 6, 0.5 mmol) as the limiting reagent. Yields were determined by ¹⁹F NMR spectroscopy of the crude reaction mixtures, using PhOCF₃ or PhCF₃ as the internal standard. Isolated yields for non-volatile products are given in parentheses. Reactions were halted at 6 h or when the measured current dropped to less than 1 mA. Legend: ^a1.3 V vs Fc⁺/Fc. ^bCarboxylate (10 equiv). r.t. = room temperature with fan cooling.