Mechanisms and applications of cyclometalated Pt(ii) complexes in photoredox catalytic trifluoromethylation

Pt(ii) complexes catalyse the visible light-driven trifluoromethylation of alkenes and heteroarenes with improved quantum yields, due to strict adherence to an oxidative quenching pathway.

Here, we report the rst demonstration of photoredox catalytic triuoromethylation using cyclometalated Pt(II) complexes. Key intermediates in the photoredox catalytic cycle were monitored directly using spectroscopic techniques. The spectroscopic identication of catalytic intermediates in the visible light-driven triuoromethylation reaction is unprecedented. Mechanistic studies revealed that the photoredox catalysis involves an oxidative quenching pathway. The kinetic parameters associated with each step of the catalytic cycle were determined, providing novel insights into enhancing the catalysis performance.

Results and discussion
Synthesis and properties of the Pt(II) complex catalysts A series of prototypical Pt(II) complexes having C^N ligands with different electron densities, 2-(2,4-diuorophenyl)pyridinate, 2-phenylpyridinate, and 2-(3-methoxyphenyl)pyridinate (Ptdfppy, Ptppy, and PtOMe, respectively, in Scheme 1), were tested here. 85 The ligand environment around the Pt(II) center was completed by non-photoactive acetylacetonate ancillary ligands. The complexes were prepared through a modied procedure based on the protocol reported by Zhao and coworkers. 86 All compounds were characterized using standard spectroscopic methods, including 1 H, 13 C, and 19 F NMR spectroscopy, mass spectrometry, and elemental analysis. The spectroscopic identication data agreed fully with the proposed structures (see ESI †).
The UV-vis absorption spectra of the Pt(II) complexes (10 mM in CH 3 CN) featured characteristic MLCT transition bands at 380 nm (log 3 ¼ 3.76; Ptdfppy), 386 nm (log 3 ¼ 3.91; Ptppy), and 420 nm (log 3 ¼ 3.78; PtOMe), with tails that approached the visible regions (Fig. 1a). 85,87,88 Strong photoluminescence emission was observed upon photoexcitation of the MLCT bands at room temperature (see ESI, Fig. S1 † for the photoluminescence spectra). The photoluminescence lifetimes (s obs ) of the emission were on the order of submicroseconds or microseconds, indicating phosphorescence (Fig. 1b). It should be noted that the s obs value increased by two orders of magnitude as the electron richness of the C^N ligand was increased (i.e., in PtOMe). The slow decay was indicative of weak MLCT contributions to the spin-forbidden electronic transition. Indeed, the MLCT character estimated by time-dependent density functional theory (CPCM(CH 3 CN)-TD-B3LYP/LANL2DZ:6-311+G(d,p)//B3LYP/LANL2DZ:6-311+G(d,p)) was signicantly smaller in the PtOMe complex (15%) than in the Ptdfppy complex (28%) ( Table 1). Therefore, the triplet state responsible for phosphorescence emission can be best described as an admixture of the MLCT, ligand-centered (LC) p-p*, and intraligand charge-transfer (ILCT) transitions. The large variations in s obs were useful for examining the notion of whether a longlifetime catalyst could exhibit better photoredox catalytic performance (vide infra). The photophysical data, including the photoluminescence quantum yields (PLQY) and the triplet-state energies (DE T ) of the Pt(II) complexes are summarized in Table 1.
Fig. S2 † for the voltammograms), corresponding to the Pt III/II redox couple. 88 These oxidation potentials (E ox ) were more positive than the E ox value (0.47 V vs. SCE) of N,N,N 0 ,N 0 -tetramethylethylenediamine (TMEDA), a sacricial electron donor. Because reduction was not observed until À1.8 V vs. SCE, the reduction potentials (E red ) for the Pt(II) complexes were calculated according to the relationship, E red ¼ E ox À DE g , where DE g is the optical band gap energy. As summarized in Table 1, the E red value shied cathodically as the electron density of the C^N ligand decreased. The same trend was observed in the excitedstate oxidation potential (E * ox ¼ E ox À DE T ), whereas the excitedstate reduction potential (E * red ¼ E red + DE T ) remained relatively unchanged across the series of the Pt(II) complexes ( Table 1). The changes in the E * ox values were dominated by DE T rather than E ox , as inferred from DDE T (0.28 eV) > eDE ox (0.10 eV; e is the elementary charge). The E * ox values (À2.11 to À1.93 V vs. SCE) were more negative than the E red value of CF 3 I (À0.91 V vs. SCE, 89 determined by DPV (ESI, Fig. S3 †)), indicating that the photoexcited Pt complexes could donate an electron to CF 3 I under the driving force for the photoinduced electron transfer ) of 1.02-1.20 eV. The photoexcited complexes could not be reduced by TMEDA, as indicated by the negative driving force (ÀDG PeT

Triuoromethylation of non-prefunctionalized alkenes and heteroarenes
The Pt(II) complexes were evaluated for their activity in the photoredox catalytic triuoromethylation of non-prefunctionalized sp 2 carbons. We employed CF 3 I as a cCF 3 source. Tri-uoromethylation of 0.50 mmol terminal alkenes was carried out in a deaerated acetonitrile solution (2.0 mL) containing 1.0 mol% Pt catalyst, 1.0 mmol DBU and 1.5 mmol CF 3 I. The reaction mixture was photoirradiated using blue LEDs (450 nm, 7 W) at room temperature, and the progress of the reaction was monitored using gas chromatography or thin-layer chromatography. The same method was applied to the tri-uoromethylation of heteroarenes, except that 1.0 mmol TMEDA was used in place of DBU. The reaction conditions were optimized by testing several bases, DIPEA, TEA, K 3 PO 4 , K 2 HPO 4 , and KO t Bu, and by testing several solvents, DMF, CH 3 OH, and CH 2 Cl 2 . The optimization results revealed that the protocols described above worked best (ESI, Tables S1 and S2 †). As demonstrated in Fig. 2, the triuoromethylation of 1-dodecene The scope of the Pt(II) complex-mediated triuoromethylation was investigated over a range of alkenes and heteroarenes (Tables 2 and 3). The isolated yields of the triuoromethylated alkenes exceeded 82%, whereas moderate yields were obtained from the triuoromethylation of heteroarenes. The triuoromethylation reaction was highly tolerant of the presence of a variety of functional groups, including hydroxyl (2c), silylether (2d), ester (2e, 2f and 4b), amide (2g and 2h), carbonate (2i), and sulfonate (2j) groups. Spectroscopic data for the triuoromethylated alkenes and heteroarenes listed in Tables 2 and 3 are summarized in the ESI. † These results successfully demonstrate that the cyclometalated Pt(II) complexes are promising alternatives to conventional photoredox catalysts based on Ir(III) and Ru(II) complexes for triuoromethylation. Scheme 2 illustrates the proposed mechanism underlying the photoredox catalytic triuoromethylation reaction in the presence of the cyclometalated Pt(II) complex (Pt II (C^N) here-aer) catalysts. Photoexcitation of Pt II (C^N) promoted an electronic transition to a singlet MLCT ( 1 MLCT) state, which underwent ultrafast intersystem crossing to a 3 MLCT state (i.e., 3  . An encounter complex formed between the photoexcited catalyst and an electron donor, such as TMEDA, could also have been generated (path A in Scheme 2); however, the negative driving force abrogated the formation of a geminate radical ion pair (vide supra). Two pathways were available to the [[Pt III (C^N)] + CF 3 Ic À ] species: the rst pathway involved quenching to the original species (i.e., Pt II (C^N) and CF 3 I) by back electron transfer (BeT), and the other pathway involved dissociation into [Pt III (C^N)] + and CF 3 Ic À . In the latter path, prompt cleavage of the C-I bond in CF 3 Ic À resulted in the formation of cCF 3 . Radical addition of cCF 3 to substrates formed triuoromethylated radical species. The photocatalytic cycle was completed by the reductive regeneration of Pt II (C^N) with sacricial electron donors, such as TMEDA and DBU. Alternatively, Pt II (C^N) was recovered through the radical-polar mechanism, which involved oxidation of radical species of the triuoromethylated substrate to the cation (path C in Scheme 2). In this case, the resulting cationic species of the triuoromethylated substrate was trapped by the strong nucleophile, I À . It should be noted that the compound bearing both iodide and triuoromethyl groups could also be produced by radical propagation between the CF 3 I and radical species of the triuoromethylated substrate (path D in Scheme 2). In both cases (i.e., paths C and D), E2 elimination assisted by TMEDA or DBU furnished the desired product, with the stoichiometric generation of ammonium iodide salts. The photoredox catalytic cycle described above involved three key electron-transfer steps: forward photoinduced electron transfer (PeT), BeT, and reductive regeneration of the catalyst. The electron-transfer processes and their intermediate species generated during the photoredox catalytic tri-uoromethylation reaction had not been directly observed to date. No kinetic information about electron transfer was available in the literature.

Spectroscopic observation of the photoredox catalytic cycle
The electron transfer reactions in each step of the photoredox catalytic cycle were investigated using time-resolved photoluminescence and laser ash photolysis measurements aer  Fig. S4 †) did not permit spectral resolution. Fig. 3a shows the photoluminescence decay traces (l obs ¼ 543 nm) for the deaerated acetonitrile solutions of 50 mM PtOMe aer nanosecond pulsed photoexcitation at 377 nm. The decay trace followed a monoexponential decay model. s obs was as long as 11.6 ms, but the incremental addition of CF 3 I (0-20 mM) to the solution signicantly shortened s obs . Other Pt complexes show similar photoluminescence quenching behaviors (ESI, Fig. S5 †). Photoluminescence quenching accompanied the generation of cCF 3 , as evidenced by a weak photoinduced ESR signal with a g value of 2.004, corresponding to a free radical, 90,91 although the broad ESR spectrum hindered resolution of the hyperne coupling due to the uorine nuclei (ESI, Fig. S6 †). These results unambiguously indicate the occurrence of oxidative quenching of 3 [Pt III (C^N)c À ]* by CF 3 I. As inferred from the negative driving force (vide supra), reductive quenching was not observed, even at 20 mM TMEDA (ESI, Fig. S7 †). The oxidative PeT rates were calculated from 1/s obs (CF 3 I) À 1/s obs , where s obs (CF 3 I) and s obs are the photoluminescence lifetimes in the presence and absence of CF 3 I, respectively. The rate constants of PeT (k PeT ), determined by the pseudo-rst order t of the PeT rates vs. the CF 3 I concentrations, were 8.8 Â 10 8 M À1 s À1 , 7.9 Â 10 8 M À1 s À1 , and 4.4 Â 10 8 M À1 s À1 for Ptdfppy, Ptppy, and PtOMe, respectively (Fig. 3b). Apparently, k PeT increased in proportion to ÀDG PeT , indicating electron transfer in the Marcus normal region (vide infra). The fraction (f) of 3 [Pt III (C^N)c À ]* that underwent oxidative quenching was estimated based on the empirical relationship, f ¼ (k PeT Â 1 mM)/(k r + k nr + k PeT Â 1 mM), assuming that 1 mM CF 3 I was present and other conditions, including diffusion, were held constant. In this relationship, k r and k nr are the rate constants for the radiative transition and non-radiative transition, respectively, which compete with PeT (      Table 4, these values are comparable to the k PeT values but are approximately one order of magnitude smaller than k BeT . The classical Marcus theory for adiabatic outer-sphere electron transfer provided a valuable basis for correlating the kinetic parameters (i.e., k PeT , k BeT , and k regen ) to the driving force. The electron transfer steps could be best described using eqn (1) with consideration for the diffusion process: 92,93 In eqn (1), k et , h, DG eT , and l are the rate for adiabatic outersphere electron transfer, the Planck constant, the free energy change, and the reorganization energy for electron transfer, respectively. 94 As shown in Fig. 6, the k PeT values adhere well to the theoretical curve calculated using the Marcus theory of electron transfer, with a l value of 2.7 eV. The large reorganization energy indicates bond scission in CF 3 I aer electron transfer. The positive dependence of k PeT on ÀDG PeT points to the occurrence of PeT in the Marcus normal region (i.e., ÀDG PeT < l). The Marcus normal behavior suggests two possible strategies for accelerating PeT: (1) raising E * ox of the photoredox catalyst by incorporating ligands with wide band gap energies (i.e., high DE T ) and weak p-backbonding abilities (i.e., cathodically shiing E ox ), and (2) using cCF 3 sources with large (more positive) E red values, such as the Shibata reagent. 66 The former approach involves a tradeoff because the effect may be offset by a reduction in the visible absorption cross-section.
The analyses using eqn (1) revealed a positive dependence of k BeT on ÀDG BeT , with a l value of 2.7 eV. This l value was identical to that obtained under PeT, as expected for forward and reverse electron transfer. One potential strategy for retarding the hazardous BeT involves raising the E ox of the photoredox catalysts using ligands with weak p-accepting or strong s-donating properties. Alternatively, the use of cCF 3 sources with large (more positive) E red values may be advantageous.
The reductive regeneration by TMEDA was found to be essential because the triuoromethylated products could not be obtained if TMEDA was replaced with a Brønsted base, such as K 3 PO 4 , K 2 HPO 4 or KO t Bu (ESI, Tables S1 and S2 †). As shown in Fig. 6, the regeneration by TMEDA followed a typical Rehm-Weller behavior with a l value of 1.2 eV. The large positive dependence indicated that the rate of regeneration increased using catalysts with large E ox values; however, predicting the inuence of k regen on the overall cycle was not straightforward due to the presence of an additional regeneration path involving radical species generated on the triuoromethylated substrate (path C in Scheme 2). The above analyses based on the Marcus theory of electron transfer established that the photoredox catalysis performance could be improved by implementing the following molecular controls: (1) E ox of the catalyst should be as small as possible to speed up PeT and to slow down BeT; (2) E ox  of the catalyst should be more positive than E ox of the sacricial electron donor to warrant regeneration; (3) the value of DE T for the catalyst may be optimized, since larger values of DE T will accelerate oxidative PeT but very large DE T values eventually initiate competing reductive quenching.
Having established the photoredox catalytic cycle, we sought to understand the inuence of the C^N ligand structures on the overall catalytic performance. The quantum yields of the three Pt(II) complexes for triuoromethylation (QY) of 1-dodecene and N-methylpyrrole were determined using standard ferrioxalate actinometry (6.0 mM K 3 [Fe(C 2 O 4 ) 3 ], quantum yield ¼ 1.1 at 420 nm; see the ESI † for experimental details). As summarized in Table 5, the QY values exceeded 100% in all cases. These results strongly indicated the signicant involvement of radical propagation (path D in Scheme 2). This hypothesis was supported by the observation of a dark-state reaction aer shutting off photoirradiation (ESI, Fig. S10 †). It is noted that the Pt(II) catalyst prepared with an electron-poor C^N ligand produced a larger value of QY for the alkene, whereas the QY values for the heteroarene exhibited an opposite trend. Because the radical propagation and subsequent steps, including iodination and E2 elimination, were independent of the identity of the photoredox catalyst, variations in QY could be explained in terms of the photoredox cycle. Obviously, the QYs for 1-dodecene followed a trend opposite to those of f and k BeT for the series of Pt(II) complexes, suggesting that the effects of f and k BeT would be marginal. One possible explanation for the QY trend is that regeneration is the limiting step in the overall catalytic performance. Of note, in Table 5 are the greater QY values of the Pt(II) complexes over those of the well-established Ir(III) and Ru(II) catalysts. The improved photon economy of the Pt(II) catalysts is likely due to elimination of the hazardous reductive quenching pathway that exists in the Ir(III) and Ru(II) catalyst systems.

Conclusions
We demonstrated the photoredox catalytic properties of a series of cyclometalated Pt(II) complexes for use in the tri-uoromethylation of non-prefunctionalized alkenes and heteroarenes under visible light irradiation. The Pt(II) complexes displayed excellent catalytic performances, as demonstrated by high yields and functional group tolerance. The oxidative quenching pathway was exclusively allowed in the photoredox catalysis cycle due to the high excited-state redox potentials (E * ox and E * red ) of the Pt(II) catalysts. Direct spectroscopic measurements revealed Marcus normal behaviors for the photoinduced electron transfer, back electron transfer, and reductive regeneration processes. These results suggested several molecular strategies that could be used to enhance the catalyst performances. Forward photoinduced electron transfer to generate cCF 3 could be accelerated by incorporating ligands with a high triplet-state energy and weak p-accepting properties, whereas hazardous back electron transfer could be minimized through the use of ligands having weak p-accepting or strong s-donating properties. Alternatively, cCF 3 reagents with low (i.e., more positive) reduction potentials, such as the Shibata reagent, could have identical effects. Correlations between the quantum yields and the rate constants for electron transfer pointed to the notion that regeneration by a sacricial electron donor or radical species of the triuoromethylated substrate could be a limiting process in the overall catalysis cycle. Evidence for radical propagation suggests an oxidation potential smaller than À0.91 V vs. SCE for the radical species of the triuoromethylated substrate. Accordingly, a large driving force exceeding 1.43 eV for catalyst regeneration by the radical species of the triuoromethylated substrate can be estimated, strongly supporting that regeneration by the sacricial electron donor may be the limiting process. We hope that the research described in this work will provide useful insights into the future development of photoredox catalysts for a range of organic transformations. a Quantum yields for triuoromethylation of 1-dodecene. b Quantum yields for triuoromethylation of N-methylpyrrole. The quantum yields were determined using standard ferrioxalate actinometry (6.0 mM K 3 [Fe(C 2 O 4 ) 3 ], QY ¼ 1.1 at 420 nm (light intensity ¼ 6.7 Â 10 À10 Einstein s -1 )).