Synthesis of [TBA]LCu^IICF₃ and LCu^IIICF₃

Treatment of previously reported [TBA]LCu^IIF complex⁴³ with 1.5 equivalents of trimethyl(trifluoromethyl)silane (TMSCF₃) in tetrahydrofuran at room temperature for 30 minutes resulted in the formation of a red copper (II) trifluoromethyl complex [TBA]LCu^IICF₃ complex in 90 % yield (λ=515 nm, ϵ=615 M⁻¹ cm⁻¹ in acetonitrile, Scheme 1). X-ray diffraction (XRD) analysis of [TBA]LCu^IICF₃ single crystals grown from a pentane/dichloromethane mixture shows a square planar Cu center with an average Cu−N bond length of 2.01(4) Å and a Cu−CF₃ bond length of 1.971(3) Å (Figure 2A). Electron paramagnetic resonance (EPR) analysis of [TBA]LCu^IICF₃ in 3 : 1 toluene:acetone mixture at 30 K shows a rhombic S=1/2 signal with g_x = 2.024, g_y=2.034, g_z = 2.153, and a coupling constant A_z = 480 Hz (Figure S57). The cyclic voltammogram of [TBA]LCu^IICF₃ in acetonitrile with 0.1 M TBAClO₄ electrolyte exhibits a quasi-reversible redox couple at E_1/2=−0.15 V vs. Fc/Fc⁺ in 1,2-dichloroethane (DCE) (Figure S2).

Treatment of [TBA]LCu^IICF₃ with oxidant [TBA]₂Ce^IV(NO₃)₆ (E_1/2=0.472 V vs. Fc/Fc⁺ in DCE) in dichloromethane affords a dark purple copper(III) trifluoromethyl complex LCu^IIICF₃ in quantitative yield (λ=540 nm, ϵ=10900 M⁻¹cm⁻¹ in acetonitrile, Scheme 1). The isolated LCu^IIICF₃ is stable at room temperature in a wide range of solvents, e.g., benzene, pentane, acetonitrile, DCE, and dichloromethane. XRD analysis of the single crystals of LCu^IIICF₃ grown from pentane shows a square planar Cu environment with an average Cu−N bond length of 1.90(4) Å and a Cu−CF₃ bond length of 1.9524(17) Å (Figure 2B). Upon oxidation, the Cu−N bonds contract significantly from 2.01(4) to 1.90(4) Å, while the Cu−CF₃ bond length remains roughly the same (Cu^II: 1.971(3) to Cu^III: 1.9524(17) Å). ¹H NMR in CD₂Cl₂ shows six sharp resonances within the 0–9 ppm range, accounting for ligand protons and suggesting a diamagnetic ground state for LCu^IIICF₃ (Figure S35A). ¹⁹F NMR analysis in CD₂Cl₂ shows a single resonance at −41.42 ppm (Figure S35B). Previous studies on formal Cu(III)−CF₃ complexes suggest that the Cu−CF₃ bond might be quite weak.²⁷ Therefore, we evaluated the Cu−CF₃ bond strength in LCu^IIICF₃ and compared it to (Phen)Cu^III(CF₃)₃ reported by Zhang et al..²⁹ Both complexes exhibit Cu−CF₃ bond strengths around 34 kcal/mol (see Supporting Information). The relatively weak Cu−CF₃ bond suggests LCu^IIICF₃ could be utilized for CF₃ group transfer.

C(sp²)−H Trifluoromethylation Reactivity of LCu^IIICF₃

The trifluoromethylation reactivity of complexes [TBA]LCu^IICF₃ and LCu^IIICF₃ was then explored. While [TBA]LCu^IICF₃ is unreactive towards (hetero)arenes, treatment of LCu^IIICF₃ with 50 equivalents of mesitylene and one equivalent of oxidant ([TBA]₂Ce^IV(NO₃)₆) in acetonitrile affords the trifluoromethylated product in 72 % yield after 5 minutes at 85 °C (Table 1, entry 1). Without molecular sieves, the yield of the reaction drops to 62 % (Table 1, entry 2), perhaps due to the residual water introduced by [TBA]₂Ce^IV(NO₃)₆.

Table 1. Optimization of C(sp²)−H trifluoromethylation by LCu^IIICF₃. All yields are determined by ¹⁹F NMR with a hexafluorobenzene internal standard.

It has been shown previously that the inclusion of a co-oxidant in copper-catalyzed C−H trifluoromethylation can improve the reaction yield significantly.^9-15 Zhang et al. attributed these improvements to the S_EAr mechanism of the reaction,²⁹ noting that the rearomatization of the arene CF₃ radical intermediate requires one equivalent of oxidant (Scheme 2). Similarly, we found that the inclusion of one equivalent of oxidant increases the yield from 52 % (without oxidant, Table 1, entry 3) to 72 % (with oxidant). Interestingly, while the reaction with mesitylene requires 30 minutes to complete without oxidant, it reaches completion in only 5 minutes when oxidant is included, as indicated by the disappearance of the dark purple color of LCu^IIICF₃. This observation implicates the involvement of the oxidant in the rate-determining step.

Having established the optimized conditions, the scope of C(sp²)−H trifluoromethylation was evaluated (Table 2). Overall, electron-rich (hetero)arenes show higher yields than more electron-poor substrates. The selectivity of C(sp²)−H trifluoromethylation is consistent with literature precedent.⁴⁷ For tert-butylbenzene, the functionalization of ortho and meta sites is reduced due to steric hindrance. Drug-like molecules, such as melatonin, can also be successfully trifluoromethylated.

Table 2. Scope of C(sp²)−H trifluoromethylation by LCu^IIICF₃.

• Major products are shown, and the minor product(s) are indicated by asterisks. The ratio of para:meta:ortho selectivity is given in parentheses. Yields without oxidant are given in parentheses. [a] 250 equiv. substrate was used.

The Role of Single Electron Transfer in CF₃ Transfer

The transfer of CF₃ from LCu^IIICF₃ suggests that the rate-limiting step may involve the cleavage of the Cu^III−CF₃ bond. However, this proposal does not align with the notably accelerated rate upon the addition of oxidant. Additionally, under standard reaction conditions, electron-rich substrates exhibit significantly faster reactions (typically 5–10 minutes), while electron-poor substrates require up to 30 minutes of reaction time.

This correlation between the electronic character of the substrates and the reaction kinetics prompted our exploration into the possibility of a SET mechanism. We hypothesize that the reaction can be initiated by the oxidation of (hetero)arene substrate by [TBA]₂Ce^IV(NO₃)₆ to generate an aryl radical cation intermediate, which can then be functionalized by LCu^IIICF₃ to afford Ar−CF₃ (Scheme 3). If this proposed mechanism is followed, the rate-limiting step should be the electron transfer between the oxidant [TBA]₂Ce^IV(NO₃)₆ and the arene substrate. Accordingly, kinetic analysis of the reaction between LCu^IIICF₃ and thioanisole indicates a first-order dependence on the concentration of the oxidant and substrate (Figure 3).

To further evaluate the possibility of a SET mechanism, we conducted a Marcus study,^{44, 48-50} in which the driving force of electron transfer (−▵G_ET) and the kinetics of trifluoromethylation (k₁) for a series of substituted arenes with varying electronic properties (R=H, NO₂, ^tBu, CH₃, OEt, OCH₃, SCH₃, N(CH₃)₂) are assessed. We first determined the oxidation potential (E_ox) of each arene using cyclic voltammetry (CV, Figures S4–S11). Subsequently, we calculated the driving force for electron transfer with [TBA]₂Ce^IV(NO₃)₆ (expressed as −▵G_ET) for each substrate using the equation −▵G_ET=−(E_ox,substrate−E_1/2,oxidant).^{51, 52}

The rate of trifluoromethylation was determined for each substrate by monitoring the UV/Vis profile of LCu^IIICF₃ (λ_max=570 nm) at 60 °C in DCE with 150 equivalents of substrate and one equivalent of [TBA]₂Ce^IV(NO₃)₆. A lower temperature was chosen to ensure reliable rates were properly measured. We also determined that this set of reaction conditions is suitable for our C−H trifluoromethylation reaction (Table 1, entry 4).

The resulting first-order rate constants (k₁) were used to construct a Marcus plot (lnk₁ vs. −▵G_ET, Figure 4). If the reaction proceeds through a rate-limiting electron transfer step, we anticipate a linear correlation with slope represented by α/RT, where α is the Brønsted α value.^{51, 52} Interestingly, we observed two distinct regions: (1) an electron-transfer (ET) region with a linear correlation between −▵G_ET and lnk₁ (R²=0.9985, Figure 4, red) and (2) a non-electron-transfer region (non-ET, Figure 4, blue) where the reaction rate remains similar to that of LCu^IIICF₃ self-decay (Figure 4, black line). The α value of ET region is modest at 0.1177 compared to the ideal value of 0.5, suggesting transition state asynchronicity caused by a delay in the product stabilization factors behind the transition state. The synchronicity of the reaction could be influenced by the aromaticity or steric effects of the arene during ET, or other molecular processes after the ET, such as the release of CF₃ radical from LCu^IIICF₃.^{52, 53}

The two distinct regions in the Marcus plot suggest a mechanism crossover. For electron-rich substrates, the reaction proceeds through rate-limiting electron transfer between the oxidant and arene substrate (Scheme 3). The first-order dependence on the concentration of oxidant and substrate shown in Figure 3 is also consistent with this hypothesis.

Conversely, within the non-ET region, the mechanism is likely dominated by the rate of Cu^III−CF₃ bond homolysis. In this case, the reaction rate is not substrate-dependent and closely resembles the self-decay rate of LCu^IIICF₃ (Figure 4, black line), which directly corresponds to the rate of CF₃ radical release from LCu^IIICF₃ when the CF₃ radical release/S_EAr mechanism is followed (Scheme 4). Consistent with this hypothesis, EPR analysis of the products of LCu^IIICF₃ self-decay in MeCN at 85 °C shows the formation of LCu^IIMeCN in >99 % yield (Scheme 4, Figures S58 and S59), suggesting that the decay of LCu^IIICF₃ quantitatively releases CF₃ radical.

Since the redox potential of LCu^IIICF₃ is quite oxidizing (−0.15 V vs. Fc/Fc⁺), the LCu^IIICF₃ complex itself could potentially engage in ET with electron-rich arenes. To examine this hypothesis, we performed an analogous Marcus study in the absence of oxidant. In this case, the driving force of ET is calculated using the redox potential of LCu^IIICF₃ (−0.15 V vs. Fc/Fc⁺). Indeed, the resulting Marcus plot also suggests a mechanism crossover when the ET driving force approaches −1.0 eV (Figure 4B). While the majority of arenes exhibit a rate similar to the self-decay of LCu^IIICF₃, the rates of the -SMe and -NMe₂ derivatives are markedly faster, suggesting electron transfer. Notably, the rate of release of CF₃ radical from the LCu^IIICF₃ is slower in the absence of an oxidant. This is due to the further reaction of LCu^II with [TBA]₂Ce^IV(NO₃)₆ to generate LCu^IIINO₃, which drives the equilibrium of CF₃ radical release to the completion (Scheme 4).

To further investigate the mechanism dichotomy, we designed a substrate, 1,2,3-trimethyl-5-(allyloxy)-benzene (Figure 5A, 1), which has a redox potential between the ET region and the non-ET region (E_ox=0.927 V vs. Fc/Fc⁺ in DCE, −▵G_ET=−0.455 eV). In the case of CF₃ radical release, the S_EAr reaction of arene and CF₃ radical is expected to produce 2, and the CF₃ radical addition to the alkene is expected to produce the cyclized product 3.^{54, 55} However, if the SET pathway is followed, only the cyclized product 3 is expected.⁵⁶ This selectivity of the SET pathway to 3 is because the oxidation of 1 takes place at the arene group rather than the allyl ether, as the oxidation potential of 3 (E_ox, 0.927 V vs. Fc/Fc⁺) closely matches that of 4 (0.928 V vs. Fc/Fc⁺) and is much lower than 5 (1.210 V vs. Fc/Fc⁺, Figure 5B, see Supporting Information). Since the SET pathway exclusively yields compound 3, a higher selectivity for 3 is anticipated when the reaction is carried out in the presence of a strong oxidant. Indeed, when the trifluoromethylation reaction of 1 is conducted in the presence of one equivalent of [TBA]₂Ce^IV(NO₃)₆ oxidant, the ratio of 2 to 3 shifts from 1.2 : 1 (without oxidant) to 1 : 1.3 (with oxidant). The flipped product distribution is consistent with the proposed mechanism dichotomy – the SET driving force is higher in the presence of [TBA]₂Ce^IV(NO₃)₆, therefore increasing the contribution of SET pathway to trifluoromethylation.

The SET mechanism was further tested by trapping the aryl radical cation intermediate. Trifluoromethylation of mesitylene was performed in the presence of excess amount of H₂O₂ (24 equiv. in 56 equiv. H₂O). Under these conditions, the yield of the trifluoromethylated product 6 drops from 72 % to 3 %, while phenol product 7 is observed at a 14 % yield, suggesting the aryl radical cation intermediate is trapped with H₂O₂ or H₂O prior to functionalization by LCu^IIICF₃ (Figure 5C). Finally, to evaluate whether LCu^IIICF₃ can capture aryl radical cations, we investigated its reaction with thioanthrene radical cation PF₆ salt.^{57, 58} The stable radical cation serves as a model for the aryl radical cation intermediate generated via the electron transfer mechanism. Treatment of LCu^IIICF₃ with two equivalents of thioanthrene radical cation afforded Ar−CF₃ coupled products 8 and 9 in a total 11 % yield (Figure 5D). The modest yield is attributed to competitive S−CF₃ coupling, which produced 5-(trifluoromethyl)-5H-thianthren-5-ium (10) in 65 % yield. Despite the low efficiency, the successful Ar−CF₃ bond formation supports the proposed coupling between radical cations and LCu^IIICF₃ complex.