Boosting CO₂ photoreduction by π–π-induced preassembly between a Cu(I) sensitizer and a pyrene-appended Co(II) catalyst

Significance

The sunlight-driven reduction of CO₂ into products like CO, CH₄, etc. can lower the atmospheric CO₂ concentration and provide carbon-neutral energy simultaneously, attracting scientists to design photocatalytic systems to facilitate this process. The rational construction of noncovalent interactions between photosensitizers and catalysts may serve as a versatile strategy to boost catalytic efficiency by speeding up the electron transfer. Besides performance, it is also desirable to lower the expense by developing photocatalytic systems based on earth-abundant elements instead of precious metals. Herein, we present the proof of concept into a fully earth-abundant system by installing dynamic π–π interaction between a dual emissive, pyrene-decorated Cu(I) photosensitizer and a pyrene-appended Co(II) catalyst, achieving remarkable catalytic performance for visible-light–driven CO₂ reduction.

Abstract

The design of a highly efficient system for CO₂ photoreduction fully based on earth-abundant elements presents a challenge, which may be overcome by installing suitable interactions between photosensitizer and catalyst to expedite the intermolecular electron transfer. Herein, we have designed a pyrene-decorated Cu(I) complex with a rare dual emission behavior, aiming at additional π-interaction with a pyrene-appended Co(II) catalyst for visible light–driven CO₂-to-CO conversion. The results of ¹H NMR titration, time-resolved fluorescence/absorption spectroscopies, quantum chemical simulations, and photocatalytic experiments clearly demonstrate that the dynamic π–π interaction between sensitizer and catalyst is highly advantageous in photocatalysis by accelerating the intermolecular electron transfer rate up to 6.9 × 10⁵ s⁻¹, thus achieving a notable apparent quantum yield of 19% at 425 nm with near-unity selectivity. While comparable to most earth-abundant molecular systems, this value is over three times of the pyrene-free system (6.0%) and far surpassing the benchmarking Ru(II) tris(bipyridine) (0.3%) and Ir(III) tris(2-phenylpyridine) (1.4%) photosensitizers under parallel conditions.

Reductive transformation of CO₂ powered by solar energy to produce renewable fuels is auspicious to relieve energy shortage and achieve carbon neutrality (1, 2). The rational design of catalytic systems is the key to accelerating the light-driven CO₂ reduction since this reaction suffers from sluggish multielectron reaction kinetics and strong competition with proton reduction (3, 4). Metal complexes as molecular photosensitizers (PSs) and/or catalysts can serve as essential components in the photocatalytic system, which can be optimized with mechanistic investigations and synthetic modifications on their well-defined structures (5, 6). Ru, Ir, and Re complexes have been vastly utilized as potent PSs in CO₂ photoreduction (7, 8), whereas the scarcity of these noble metals restricts their practical uses (9). In this context, earth-abundant molecular systems with organic dyes (10) or metal-organic chromophores featuring Zn(II) (11), Cu(I) (12), or Al(III) (13) centers have received considerable interest, among which Cu(I)-based PSs are particularly appealing alternatives with their long-lived metal-to-ligand excited states and highly variable luminescent properties with a vast ligand library (14–17). The first report of a Cu(I) PS for CO₂ photoreduction by Ishitani et al. (18) displays the design of a diphosphine-dangled phenanthroline ligand to form a dimeric Cu(I) PS. The tetradentate ligand inhibits the decomplexation of the diphosphine donors around the Cu cores, enabling prominent stability during visible light–driven CO₂ reduction with Fe (18, 19) or Mn (20) catalysts. Beller et al. also developed the use of Cu(I) PSs for CO₂ reduction in synergy with Fe–cyclopentadienone (21) or Mn–diamine complexes (22) as the catalysts. Another breakthrough is from Sakai et al. (12, 23, 24), in which a water-soluble, sulfonated Cu(I) PS was designed for CO₂ photoreduction to CO with several derivatives of Co porphyrins, giving good selectivity (ca. 90%) in fully aqueous conditions which often induce significant proton reduction.

Despite the above progress, the performances of the Cu(I)-sensitized systems for CO₂ reduction are still unparallel to those noble metal–based systems; in turn, the effective knowledge-driven concepts to further improve their efficiencies are highly desirable. On the one hand, rigorous strategies to tailor ligand functionalities are fundamental to improve the photophysical and photoredox properties of Cu(I) PSs in photocatalysis. The discovery of Cu(I) complexes in terms of coordination structures (25) or emission behaviors (26) may promote the efficiency of CO₂ photoreduction. On the other hand, further improvement can be achieved by the construction of additional interaction between PS and catalyst to expedite the electron transfer and thus curb the self-deactivation of the excited PS (27), eventually promoting photocatalysis. This tactic has been applied in the noble metal–based systems with covalent linking (28–30), H-bonding (31), coordinative connection (32), and π–π interactions (33, 34). Among them, noncovalent π–π interactions are advantageous for providing dynamic interactions rather than stable connections between PS and catalyst, in which the “dynamic interactions” can be realized by weak and/or reversible intermolecular attractions, allowing both connection and dissociation in a dynamic manner during the photocatalysis. While accelerating the intermolecular electron transfer, the dynamic interactions can concurrently circumvent the irreversible dissociation between the PS and the catalyst with their affinity to regain the association, which is hardly applicable with the linkages based on fixed covalent bonds (32).

With the above anticipated merits, we have designed a noble metal–free system with π–π interaction between PS and catalyst by employing a Cu(I) PS with two pyrenyl groups (Fig. 1): [Cu^I(xantphos)(pybcp)]PF₆ (CuPYBCP; xantphos = 9,9-dimethyl-4,5-bis(diphenylphosphino)xanthene, pybcp = 2,9-dimethyl-4,7-bis(4-(pyren-1-yl)phenyl)-1,10-phenanthroline) and a pyrene-tethered Co(II) catalyst, [Co^II(PYN5)](ClO₄)₂, (CoPYN5; PYN5 = (2E,12E)-2,13-dimethyl-14-(pyren-1-yl)-3,6,9,12-tetraaza-1(2, 6)-pyridinacyclotridecaphane-2,12-diene), respectively. The prototypical PS and catalyst were also employed for comparison, i.e., [Cu^I(xantphos)(bcp)]PF₆ (CuBCP; bcp = bathocuproine or 2,9-dimethyl-1,10-phenanthroline) and [Co^II(N5)](ClO₄)₂ (CoN5; N5 = (2E,12E)-2,13-dimethyl-3,6,9,12-tetraaza-1(2, 6)-pyridinacyclotri-decaph-ane-2,12-diene), respectively. Interestingly, the pyrene decoration in CuPYBCP endows an unusual dual emission behavior and a sharply decreased lifetime, which has been studied systematically by means of spectroscopic and quantum chemical methods. More importantly, the π–π interaction between CuPYBCP and CoPYN5 has been confirmed to substantially accelerate the electron transfer and improve the photocatalytic performance in CO₂ reduction. Compared to the pyrene-free CuBCP/CoN5 system (6.0%), the π–π-interacted CuPYBCP–CoPYN5 system accomplishes a remarkable apparent quantum yield (Φ) of 19% at 425 nm for the CO₂-to-CO conversion with a maximum selectivity of 97%, also surpassing the parallel performances of benchmarking noble metal PSs. This is an example of noble metal–free molecular system for photocatalytic CO₂ reduction assisted by noncovalent interactions, also presenting a rare case of dual emissive Cu(I) luminescent complex.

Fig. 1.

Structures and photophysics. (A) Chemical structures of the used Cu(I) PSs and Co(II) catalysts. (B) UV–vis absorption spectra of 12 μM CuPYBCP (violet, solid line) or CuBCP (orange, solid line) in CH₃CN. The TDDFT-calculated results of CuPYBCP (violet, dashed and dropped lines) are also shown for comparison. (C) Emission spectra of 50 μM CuPYBCP upon excitation at 380 to 440 nm in deaerated CH₃CN. The emission intensity was normalized to the highest Raman peak of CH₃CN. (D) Excitation spectra and (E) excited-state decay traces of 50 μM CuPYBCP for different emission bands in deaerated CH₃CN.

Photophysics

The heteroleptic Cu(I) complex, CuPYBCP, in the type [Cu(P^P)(N^N)]⁺ was prepared with the pyrene-appended N^N ligand, pybcp, which was synthesized via Suzuki coupling (see Experimental Details in SI Appendix). The chemical compositions and structures of pybcp and CuPYBCP were examined by multiple NMR spectroscopies and high-resolution mass spectroscopy (SI Appendix, Figs. S1–S7).

The UV–vis absorption and steady-state fluorescent spectra of CuBCP and CuPYBCP in CH₃CN show significant differences (Fig. 1B). The UV–vis spectrum of CuBCP displays the characteristic ligand-centered and MLCT bands at 285 and 388 nm (35), respectively. In sharp contrast, several bands appear on the absorption spectra of CuPYBCP with much higher absorption coefficient values, manifesting the significant hyperchromic effect of pyrenyl groups for a stronger visible light absorption ability. Quantum chemical simulations performed at the time-dependent density functional level of theory (TDDFT; SI Appendix, Tables S1 and S2) were carried out to unravel the nature of the electronic transitions underlying the absorption bands of CuPYBCP; fully relaxed structures for all calculated intermediates are collected (36). TDDFT predicts two dipole-allowed MLCT_phen excitations, i.e., into the S₁ and S₂ singlet excited states at 429 and 408 nm from Cu to the lowest two ${\pi }_{\mathrm{p}\mathrm{h}\mathrm{e}\mathrm{n}}^{\ast }$ orbitals of the phenanthroline moiety—associated with the red-sided shoulder at approximately 400 nm (Fig. 1B). According to the simulations, the prominent absorption feature—arising upon incorporation of the pyrene motifs in PS—measured at 344 nm originates from two strongly dipole-allowed intraligand charge transfer (ILCT; S₄ and S₅) transitions at 377 and 372 nm. These transitions involve shifts in electronic density from the ${\pi }_{\mathrm{p}\mathrm{y}\mathrm{r}}$ orbitals of each pyrene group into the lowest-energy ${\pi }_{\mathrm{p}\mathrm{h}\mathrm{e}\mathrm{n}}^{\ast }$ orbital. Furthermore, S₁₂ and S₁₃ (340 and 338 nm) featuring a mixed electronic character involving local excitations of the pyrenes as well as charge transfer from the pyrenes and the metal center toward the phenanthroline acceptor contribute to the 344-nm absorption band.

Next, the emission spectra of CuBCP under excitation at 380 to 450 nm share a similar structureless emission band centered at 580 nm (SI Appendix, Fig. S8), consistent with its emissive MLCT state. However, the emission bands of CuPYBCP varied significantly along with altered excitation wavelengths from 380 to 440 nm (Fig. 1C). In detail, a strong emission band peaking at 460 nm can be excited in the range of 380 to 420 nm, and its intensity was decreased along with the increased excitation wavelength up to 425 nm. Additionally, new emission bands at 550, 653, and 720 nm emerged at 395- to 440-nm excitation, and the strongest emission of these bands was obtained at 425-nm excitation. It is noteworthy that the latter two emission bands are reminiscent of intraligand (IL) excited states of some reported pyrene-appended PSs (33, 37–39). Interestingly, the emission intensity at 460 nm sharply decreased with increasing concentration of CuPYBCP, while the ones at 550, 653, and 720 nm displayed overall enhancement (SI Appendix, Fig. S9). The former trend suggests the aggregation-induced quenching with a higher concentration of CuPYBCP, which should be correlated with the pyrene-involved excited-state species. However, the latter, different tendency infers the presence of another excited state. Moreover, the emission at 460 nm was hardly quenched by air (9%), consistent with singlet emission, whereas the additional emission bands were significantly quenched (over 70%; SI Appendix, Fig. S10) and should be correlated with triplet species.

We further collected the excitation spectra (Fig. 1D) of CuPYBCP for the four emission bands. The excitation spectra are identical for emission between 460 and 550 nm or those between 653 and 720 nm. Notably, an additional band at 420 nm appeared in the excitation spectra responsible for the emission at either 653 or 720 nm, indicating that they should originate from the MLCT_phen states in accordance with the TDDFT calculations (Fig. 1B). On the other hand, the emission at 460 and 550 nm is attributable to the ligand-based excited states, although with different relationship vs. [CuPYBCP] (SI Appendix, Fig. S9) and spin states (SI Appendix, Fig. S10).

More interesting emission behavior was revealed by the measurements of the excited-state lifetimes of CuPYBCP (Fig. 1E). For the ligand-based emission bands, the monoexponential decay of the emission at 460 nm (excited at 365 nm) is different from the biexponential decay of the one at 550 nm (excited at 425 nm) despite their similar short lifetimes (3.4 vs. 2.5 ns), further suggesting their different nature. For the MLCT_phen emission, the lifetimes of the excited-state species at 653 and 720 nm (both excited at 425 nm) are rather similar (9.4 vs. 9.1 ns) with overlapped decay traces, indicating their same origin. Consequently, three kinds of excited-state species and a dual emission behavior (550 and 653/720 nm) at 425-nm excitation can be revealed from the above observations. Also, it is noteworthy that these lifetime values are dramatically shorter than that of CuBCP (289.0 ns), displaying over one-thirtieth shrinkages by appending the pyrene groups. This observation is very unusual since it is contradictory to those pyrene-decorated PSs based on Re (40, 41), Ir (33, 42), and Ru (37, 39) which generally exhibit elongated lifetimes upon the participation of long-lived IL states, highlighting the delicacy of Cu(I) PSs in terms of structural optimizations.

Dual emission behavior is not frequently reported for common PSs, while it could be noticed when inefficient electronic conduction between two distinct emissive excited states in a unified molecule (39). However, it is essential to exclude the factor of a minor impurity inducing the second emissive state. To rule out this possibility, multiple batches of CuPYBCP were prepared by three researchers on different scales, all of which afforded identical results within experimental error. Moreover, besides elemental analysis, all batches of CuPYBCP were proven to be >99.9% pure by high-performance liquid chromatography. Consequently, we have eliminated the impurity factor for the observation of dual emission.

In order to address the origins of the dual emission and the rather short excited-state lifetime of CuPYBCP, we fully relaxed the low-lying triplet excited states of interest (SI Appendix, Tables S3–S5). Equilibration of the lowest triplet state from the Franck–Condon geometry by means of DFT (ΔSCF approach) yields a ³MLCT_phen state and an emission wavelength of 818 nm (1.52 eV). Accordingly, TDDFT predicts that this ³MLCT_phen emission is slightly higher in energy—at 732 nm (1.69 eV). Therefore, the quantum chemical simulations suggest the long-wavelength emission observed at 720 nm to be of ³MLCT character. The rather short lifetime of this ³MLCT_phen state of merely ~9 ns possibly results from the pronounced relativistic effects. On the one hand, the sizable spin–orbit couplings (SOCs) allow the efficient population transfer to the ³MLCT_phen states from the energetically close ¹MLCT_phen states, i.e., S₁ ¹MLCT_phen to T₅ and T₉ ³MLCT_phen states with SOCs of 119 and 136 cm⁻¹, respectively (SI Appendix, Table S3). On the other hand, these relativistic effects also foster a rapid deactivation of the lowest ³MLCT_phen to the singlet ground state ( $〈T_1\left|H_{\mathrm{SOC}}\right|S_0〉=32$ cm⁻¹). The second emission band measured at 653 nm is of similar lifetime (9.4 ns) and stems very likely from a higher-lying ³MLCT_phen state, such as Franck–Condon state T₅ at 419 nm (2.96 eV) which is initially accessible upon intersystem crossing (ISC) from the dipole-allowed S₁ excitation or from the lower-lying T₄ (471 nm, 2.63 eV).

In addition, excited-state relaxation channels yielding the population of the two low-lying ³IL_pyr states—the locally excited ππ* states—of each pyrene moiety were investigated by means of scalar relativistic (SR)TDDFT. Based on the predicted SOCs, these ³IL_pyr states are inaccessible from the singlet excited states as the metal center is not involved in the triplet acceptor states (SOCs of 0 to 1 cm⁻¹, SI Appendix, Table S3). This finding is consistent with previous studies on transition metal complexes combining an inorganic (^1/3MLCT) and an organic chromophore (^1/3IL), revealing that an efficient ISC is exclusively observed in scenarios where the metal center is involved in the singlet donor and in the triplet acceptor state (43–45). However, the two ³IL_pyr states might be accessible upon internal conversion from higher-lying ³MLCT_phen states. A potential ³IL_pyr deactivation is predicted by DFT and by TDDFT at merely 1,069 and 1,072 nm (1.16 eV; ΔSCF) for the two ³IL_pyr states, respectively. However, their low energy and their inaccessibility by direct ISC suggest that the locally excited states of the pyrene are not involved in the dual emission of CuPYBCP. As mentioned above, this finding is at first counterintuitive as Ir (33, 42) and Ru (37, 39) PSs bearing pyrene groups typically feature extended excited-state lifetimes due to the population of long-lived ³IL states which are due to their very small SOCs decoupled to the singlet ground state. However, as we could show in a recent joint synthetic–spectroscopic–theoretical study, such long-lived state is accessible for a related iridium complex due to a pronounced mixing of ^1/3IL states with ^1/3MLCT states (33). This mixed electronic character results in sizable SOCs up to almost 1,000 cm⁻¹ and thus fosters the population of the IL state.

Finally, based on the above computational investigations, the emission at 550 nm is unlikely to originate from the monomer form of CuPYBCP. Considering its broad, unstructured photoluminescence, as well as its positive concentration dependency (SI Appendix, Fig. S9) which is opposite to that of the pyrene-based emission band at 460 nm, the emission at 550 nm could be tentatively attributed to the excimer emission (46, 47) of CuPYBCP.

Redox Properties

Furthermore, cyclic voltammetry with the above spectroscopic results was employed for the estimation of the redox potentials (48, 49) of the Cu(I) PSs (SI Appendix, Fig. S11), which are also included in Table 1. Potentials are footnoted versus normal hydrogen electrode (vs. NHE) unless otherwise noted. The results first indicate that the decoration of pyrenyl groups makes the reversible reduction wave (E_red = −1.37 vs. −1.43 V) of CuPYBCP less negative to CuBCP, suggesting the apparent electron-withdrawing effects of the pyrenyl groups (6, 50). In turn, the reductive quenching potential (E_q,red = 1.73 vs. 1.39 V) of CuPYBCP is much more positive than that of CuBCP. Quantum chemical simulations on the singly reduced doublet species of CuPYBCP reveal that the first reduction event is mainly associated with the population of the lowest-energy ${\pi }_{\mathrm{p}\mathrm{h}\mathrm{e}\mathrm{n}}^{\ast }$ orbital (LUMO). However, upon single reduction, this orbital partially extends onto one of the decorating pyrene groups (D₀ in SI Appendix, Table S6), which explains the slightly less negative E_red with respect to CuBCP. Photoexcitation of this doublet species yields several low-lying dipole-allowed doublet transitions of ²ILCT nature—mainly from the ${\pi }_{\mathrm{p}\mathrm{h}\mathrm{e}\mathrm{n}}^{\ast }$ orbital toward the ${\pi }_{\mathrm{p}\mathrm{y}\mathrm{r}}^{\ast }$ orbitals (SI Appendix, Tables S6 and S7 and Fig. S12). The above comparison demonstrates a trade-off with pyrene modifications between the two values, both of which serve as the direct or indirect driving forces of reductive quenching pathways for most Cu(I) PSs (22, 27), in which their E_red values are already sufficient to drive the catalysis of CoN5 or CoPYN5 (ca. −1.1 V) (33).

Table 1.

Summary of the photophysical and redox properties of the Cu(I) PSs^*

Cu PS	λab (nm)λab	λem (nm)	τ (ns)	Isosbestic point (nm)	E0-0 (V)	Ered§ (V)	Eox¶ (V)	Eq,red (V)	Eq,ox (V)
CuBCP	320 and 388	580†	289.0†	445	2.79	−1.40	1.52	1.39	−1.27
CuPYBCP	240, 279, and 344	460†, 550‡, 653‡, and 720‡	3.5 (460 nm)†, 2.5 (550 nm)‡, 9.4 (653 nm)‡, and 9.1 (720 nm)‡	400	3.10	−1.37	1.59	1.73	−1.51

^*Measured in deaerated CH₃CN solution at 293 K. Potentials are footnoted vs. NHE.

^†Excitation wavelength is 365 nm.

^‡Excitation wavelength is 425 nm.

^§Half-wave potentials (E_1/2) of reduction waves.

^¶Peak potentials of oxidation waves.

NMR Titration with DFT Modeling

¹H NMR titration was then applied for evaluations on the binding affinity between CuPYBCP and CoPYN5, as well as their own self-interaction tendencies. It can be observed that the addition of CoPYN5 into CuPYBCP could induce small shifts of the pyrenyl proton signals (SI Appendix, Fig. S13), which were recorded and fitted into several models (SI Appendix, Table S8), in which the noncooperative 1:2 model was found as the optimal model to afford a 1:1 binding constant (K₁₁) of 207 ± 10 M⁻¹. This value is at the same order of magnitude of the π–π binding constant between a pyrene-dangled Ir(III) PS and CoPYN5 (199 ± 7 M⁻¹), consistent with the reversible, dynamic nature of the π–π interaction. Moreover, the self-titration experiments of CoPYN5 (SI Appendix, Fig. S14) and CuPYBCP (SI Appendix, Fig. S15) reveal dimerization constants of 132 ± 5 and 50 ± 6 M⁻¹, respectively. These above binding behaviors indicate that the heteromolecular binding can be in a competitive relationship with the self-interactions, but it is still more eminent for its higher binding constant.

With experimental indications, DFT simulations were performed to assess the binding modes and the binding energy between CuPYBCP and CoPYN5 in more detail (36). To this aim, π-stacked structures were generated based on the previously sampled conformer space for a related Ir-based PS interacting with CoPYN5 (33). The DFT-predicted binding energy between the nonreduced CuPYBCP (singlet) and the Co(II)-based catalyst (doublet) was calculated to be 0.86 eV (83 kJ mol⁻¹); upon reductive quenching of the excited CuPYBCP, the interaction gets stronger with a higher binding energy of 0.92 eV (89 kJ mol⁻¹). These calculations suggested a pronounced intermolecular interaction and thus a preassembly between CuPYBCP and CoPYN5, which is in good agreement with previous studies on related π-stacked systems (33, 51, 52). On the other hand, sandwich-like structures, where the pyrene moiety of CoPYN5 is localized between both pyrene moieties of CuPYBCP, were investigated computationally. In both cases of the nonreduced and singly reduced CuPYBCP interacting with CoPYN5, their binding energies are almost doubled (nonreduced: 1.37 eV and 132 kJ mol⁻¹; singly reduced: 1.82 eV and 176 kJ mol⁻¹). Furthermore, these sandwich-like structures are also thermodynamically more stable than those stemming from the interactions among only two pyrene moieties (SI Appendix, Table S9). However, these sandwiched structures spontaneously feature a singly reduced CoPYN5 after structural optimization, presumably due to the electron delocalization from the two closely stacked pyrenes (detailed discussion in SI Appendix). Therefore, such sandwiched structures should not be involved in the initial preassembly between the (reduced) CuPYBCP and CoPYN5 but may be of importance for subsequent catalytic cycles and in the underlying electron transfer processes of these later cycles. More details with respect to the binding modes and relative energies are presented in SI Appendix [SI Appendix, Table S9 with uploaded data (36)].

Photocatalytic CO₂ Reduction

Noble-metal–free systems by pairwise combinations of the two Cu(I) PSs and the two Co(II) catalysts were evaluated for visible-light–driven CO₂ reduction in a CH₃CN solution containing BIH (1,3-dimethyl-2-phenyl-2,3-dihydro-1H-benzo[d]imidazole) as the sacrificial electron donor, triethylamine (TEA) mainly as the deprotonation agent for the oxidized BIH (53), 2,2,2-trifluoroethanol as the proton source, and the xantphos ligand to suppress the dissociation of Cu(I) PSs (54, 55) during photocatalysis. We note that our previous studies (33, 50) indicate that 2,2,2-trifluoroethanol is the optimal proton source for the Co(II) macrocyclic catalysts over TEA, phenol, or water. CO and H₂ were detected as the main products, and all the above components are needed to achieve the maximum CO formation (SI Appendix, Table S10). The ¹³CO₂-labeled reaction generated significant ¹³CO (m/z = 29; SI Appendix, Fig. S16), proving that the source of CO is CO₂ rather than the decomposed organic components.

As summarized in Table 2 (SI Appendix, Fig. S17), the first pairwise combinations with 0.10 mM catalyst and 0.10 mM PS exhibit the highest TOF_1h, TON, selectivity, and Φ of the CuPYBCP–CoPYN5 system, suggesting the promotive effect of π–π interaction (Fig. 2A). The same activity trend was observed when the catalyst concentration was reduced to 0.01 mM, in which a higher TON of 630±88 was obtained for the most efficient CuPYBCP–CoPYN5 pair (SI Appendix, Fig. S18 and Table S11). It is interesting to note that the Φ value with CuBCP is over three times higher than that with CuPYBCP when the pyrene-free CoN5 acted as the catalyst (6.0% vs. 1.8%), while an inverse order of Φ (12% vs. 19%) was observed with CoPYN5 instead. These comparisons indicate the relatively unfavorable photoredox properties of CuPYBCP to drive the catalyst with no additional interaction, which, however, have been significantly circumvented by dynamic π–π interaction with CoPYN5. The maximum Φ value of 19% at 425 nm during 1 h is higher than our previous π-interacted system with a pyrene-appended Ir(III) PS and CoPYN5 (14.3 ± 0.8% at 425 nm) (33), which is also higher than most state-of-the-art noble metal–free systems, including purpurin/[Fe(qpy)(OH₂)₂](ClO₄) (56) (Φ_CO = 1.1% at 450 nm; qpy = 2,2′:6′,2″:6″,2‴-quaterpyridine), Cu(II) purpurin/FeTDHPP (57) (Φ_CO = 6% at 450 nm; FeTDHPP = chloroiron(III) 5,10,15,20-tetrakis (2′,6′-dihydroxyphenyl)-porphyrin), [Cu₂(P₂bph)₂]²⁺/Fe(dmp)₂(NCS)₂ (18) (Φ_CO = 6.7% at 436 nm; P₂bph = 4,7-diphenyl-2,9-di(diphenylphosphinotetramethylene)-1,10-phenanthroline; dmp = 2,9-dimethyl-1,10-phenanthroline), CuBCP with a Mn(I) catalyst (22) (Φ_CO = 9.1% at 415 nm), etc. However, it is still lower than the system incorporating [Cu₂(P₂bph)₂]²⁺ and another Mn catalyst (20) (Φ_HCOOH+CO = 57% at 436 nm), which can be mainly attributed to the short-lived excited state of CuPYBCP. The main CO production was contributed by the first 2 h of reaction as indicated by a long-term test (Fig. 2B). We checked the stability of CuPYBCP and CoPYN5 after the reaction by adding fresh component into the deactivated reaction mixture (SI Appendix, Fig. S19), where the CO evolution could be largely resumed by the addition of fresh CoPYN5 instead of CuPYBCP, indicating the main deactivated component is the Co(II) catalyst (33). Further delicate screening of PS and catalyst with additional interactions is warranted in our laboratory for further improvement in catalytic performances.

Table 2.

Results of photocatalytic CO₂ reduction experiments^*

Entry	PS	Catalyst	n(CO) (μmol)	n(H2) (μmol)	TON(CO)†	TOF1 h(CO)‡(h−1)	TOF1 h(H2)‡(h−1)	CO%	Φ 1 h
1	CuBCP	CoN5	35.2 ± 4.6	7.1 ± 2.4	84.5 ± 11.4	84.5 ± 11.4	11.8 ± 5.5	88	6.0%
2	CuBCP	CoPYN5	71.2 ± 8.2	7.7 ± 2.3	178.0 ± 20.6	178.0 ± 20.6	19.3 ± 5.8	90	12%
3	CuPYBCP	CoN5	10.2 ± 1.0	3.4 ± 0.8	25.4 ± 2.5	25.4 ± 2.5	8.6 ± 2.0	75	1.8%
4	CuPYBCP	CoPYN5	135.2 ± 17.4	4.0 ± 1.5	338.0 ± 43.5	263.7 ± 16.1	13.6 ± 3.7	97	19%
5	RuBPY	CoPYN5	1.8 ± 0.6	0.6 ± 0.1	4.5 ± 1.5	2.9 ± 1.3	1.3 ± 0.2	76	0.3%
6	IrPPY	CoPYN5	9.0 ± 1.5	0.8 ± 0.2	22.5 ± 3.9	22.5 ± 3.9	2.0 ± 0.5	92	1.4%

^*Standard condition: Cu PS (0.10 mM), Co catalyst (0.10 mM), TFE (1.0 M), TEA (1.0 M), and BIH (20 mM) in 4.0 mL CH₃CN under 425-nm LED irradiation (50 mW cm⁻²). The yields of CO and H₂n(CO) and n(H₂), were recorded for the maximum values within 120-min irradiation (SI Appendix, Fig. S18). The experimental errors represent the SDs of three independent measurements.

^†TON = n(CO)/n_catalyst, in which n_catalyst = 0.1 mM × 4.0 mL = 0.4 μmol.

^‡TOF_1h = n(CO)_1h/(n_catalyst ·1 h).

Fig. 2.

Photocatalytic CO₂ reduction. (A) TOF_1h values of CO (yellow) and H₂ (red) yields in the four pairs of Cu(I) PSs (0.10 mM) and Co(II) catalysts (0.10 mM) under 425-nm irradiation for photocatalytic CO₂ reduction. The comparisons with Ru and Ir benchmarking PSs are shown for comparison. Specific conditions and time profiles are shown in SI Appendix, Figs. S17 and S20. The error bars represent the SDs of three independent measurements. (B) Time profiles of photocatalytic CO (star) and H₂ (circle) formation with CuPYBCP (0.10 mM), CoPYN5 (0.10 mM), TFE (1.0 M), TEA (1.0 M), and BIH (20 mM) in 4.0 mL CH₃CN under 425-nm irradiation (50 mW cm⁻²).

Additionally, higher activity and selectivity for CO₂ reduction over hydrogen evolution were attained with CoPYN5 over CoN5 regardless of the used Cu(I) PS. These differences can be ascribed to our previous finding (33, 50) that the additional pyrene group in CoPYN5 increased both the redox activity of the pyridyldiimine moiety and the hydrophobic nature of the pristine CoN5 catalyst. It is interesting to note that the photocatalytic evaluations with the use of more concentrated Cu(I) PSs (0.5 mM) display the much-suppressed performances from the systems with CuPYBCP (SI Appendix, Fig. S20 and Table S12), which are even lower than those with CuBCP under parallel conditions. These observations should be mainly attributed to the reinforced self-stacking of CuPYBCP at high concentrations, as suggested by the NMR self-titration results (SI Appendix, Fig. S15), which may cause substantial recombination of photoexcited electrons and holes among the π–π-interacted pyrene-tethered PSs.

We further demonstrated the excellence of this additionally interacted, non–noble metal system by using noble metal benchmarking PSs, RuBPY (= [Ru(bpy)₃]Cl₂, bpy = 2,2′-bipyridine) and IrPPY (= fac-Ir(ppy)₃, ppy = 2-phenylpyridine), in which the Φ value of CuPYBCP–CoPYN5 pair is over 60 and 10 times higher than those of systems with RuBPY or IrPPY as the PS under identical conditions (19% vs. 0.3%/1.4%; Table 2, Fig. 2A and SI Appendix, Fig. S21), as well as a higher CO selectivity (97% vs. 76%/88%), respectively. These intriguing performances which are superior to noble metal PSs further suggest the merits of additional interactions between PS and catalyst to facilitate the photocatalytic CO₂ reduction.

Overall, the above photocatalytic results demonstrate the promising application of π–π interaction in improving the noble metal–free systems for photocatalytic CO₂-to-CO conversion.

Photoinduced Electron Transfer

We then deployed a sequence of fluorescent spectroscopies excited at 425 nm to study the photoinduced electron transfer pathways and kinetics. The estimated quenching constants via Stern–Volmer plots (details in SI Appendix) are summarized in Table 3. The results of quenching experiments with time-resolved fluorescent spectroscopies indicate that the photoinduced electron transfer follows the reductive quenching pathway (Eqs. 1–3) in the system with CuBCP or CuPYBCP. This is because the lifetimes of excited states could be effectively quenched by the addition of BIH rather than the two Co catalysts (SI Appendix, Figs. S22–S24). Especially, with BIH addition, much higher dynamic quenching constants were obtained in the case of CuPYBCP at 550/653/720-nm emission than the one of CuBCP (1.7~2.3 × 10¹¹ vs. 8.5 × 10⁹ M⁻¹ s⁻¹), approaching to the diffusion limit (58). This tendency is consistent with the much more positive E_q,red of CuPYBCP than that of CuBCP (1.73 vs. 1.39 V).

$$\mathrm{C}\mathrm{u} \mathrm{P}\mathrm{S}+h\upsilon \to \mathrm{C}\mathrm{u} {\mathrm{PS}}^{*},$$ [1]

$$\mathrm{Cu} {\mathrm{PS}}^{*}+\mathrm{B}\mathrm{I}\mathrm{H}\stackrel{k_{\mathrm{q},\mathrm{red}}}{\to }\mathrm{C}\mathrm{u} \mathrm{P}{\mathrm{S}}^{-}+\mathrm{B}\mathrm{I}{{\mathrm{H}}^{·}}^{+},$$ [2]

Table 3.

Related data and calculated rate constants from fluorescent quenching and ns-TA spectroscopy with Cu(I) PSs and Co(II) catalysts^*

PS	Time-resolved fluorescent quenching		ns-TA
	τ0 (ns)	kq(BIH) (M−1 s−1)	τ0′ (ns)	τ1 (μs)	kr(CoN5) (× 109 M−1 s−1)	kr(CoPYN5) (× 109 M−1 s−1)
CuBCP	289.0	8.5 × 109	329.0	42.8	0.87	0.91
CuPYBCP	2.5 (550 nm) 9.4 (653 nm) 9.1 (720 nm)	1.7 × 1011 (550 nm) 2.3 × 1011 (653 nm) 2.0 × 1011 (720 nm)	N.A.	10.9	1.4	6.9

^*τ₀ and τ₀′ are the excited-state lifetimes, k_q(BIH) is the reductive quenching constant, τ₁ are the reduced-state lifetimes, k_r(CoN5) and k_r(CoPYN5) are the second-order reaction constants between the reduced-state PS and the ground-state catalyst.

$$\mathrm{C}\mathrm{u} \mathrm{P}{\mathrm{S}}^{-}+\mathrm{C}{\mathrm{o}}^{\mathrm{I}\mathrm{I}}\mathrm{T}\mathrm{C}\mathrm{P}\mathrm{c} \stackrel{k_{\mathrm{r}}}{\to }\mathrm{C}\mathrm{u} \mathrm{P}\mathrm{S}+\mathrm{C}{\mathrm{o}}^{\mathrm{I}}\mathrm{T}\mathrm{C}\mathrm{P}\mathrm{c}.$$ [3]

After forming the reduced state of the Cu(I) PS, the electron of the reduced species will be transferred to the Co(II) catalyst (Eq. 3). Therefore, nanosecond transient absorption (ns-TA) spectroscopy was utilized to investigate the kinetics of this intermolecular electron transfer process under the influence of π–π interaction. The ns-TA spectra of the reduced Cu(I) PSs could be obtained in the presence of excess BIH and 425-nm excitation. First, for CuBCP, the excitation induced an excited-state absorption (SI Appendix, Fig. S25) at 330 and 530 nm, in which the latter can be attributed to the absorption of the ³MLCT species (SI Appendix, Fig. S8). The excited species has a lifetime (τ₀') of 329 ns, roughly consistent with its fluorescent lifetime τ₀ = 289.0 ns. The addition of an excess of BIH (12 equivalents) induces a long-lived species (τ₁ = 42.6 μs; Fig. 3A) with an altered TA spectrum displaying an absorption band at ca. 350 nm, suggesting the generation of the reduced CuBCP, which further confirms the reductive quenching pathway. Subsequently, increasing concentration of either CoN5 or CoPYN5 in the above reaction mixture could reduce the lifetime of the reduced-state species. The second-order reaction constant (k_r), namely the electron transfer rate from PS to catalyst, can be calculated for each Co(II) catalyst (32) (Fig. 3 B–D, see SI Appendix for details and Table 3 for values). The k_r with CoPYN5 is slightly higher than that with CoN5 (9.1 vs. 8.7 × 10⁸ M⁻¹ s⁻¹; Fig. 3H), consistent with the relatively easy reduction of CoPYN5 to form Co(I) species.

Fig. 3.

TA experiments. (A) TA spectra of 0.05 mM CuBCP with 0.60 mM BIH. Kinetic traces of 0.05 mM CuBCP with 0.60 mM BIH and (B) 0 to 35 μM CoN5 or (C) 0 to 30 μM CoPYN5. (D) Plot of (τ₁/τ−1) versus the concentration of CoPYN5 (red) or CoN5 (blue) with linear fitting for the CuBCP system with BIH. (E) TA spectra of 0.05 mM CuPYBCP with 0.60 mM BIH. The DFT-simulated absorption spectrum (violet, dashed line) and structure (Inset) of CuPYBCP are shown. Kinetic traces of 0.05 mM CuPYBCP with 0.60 mM BIH and (F) 0 to 50 μM CoN5 or (G) 0 to 40 μM CoPYN5. (H) Plot of (τ₁/τ−1) versus the concentration of CoPYN5 (red) or CoN5 (blue) with linear fitting for the CuPYBCP system with BIH. The kinetic data were collected by following the spectra at 350 nm in Ar-saturated CH₃CN upon excitation at 425 nm.

In contrast, the ns-TA spectrum of CuPYBCP did not show detectable species due to its short-lived excited state (< 10 ns). Then, the addition of excess BIH led to the emergence of a long-lived reduced species (τ₁ = 10.9 μs). The TA temporal evolution of the reduced CuPYBCP shows a bleaching band at 350 nm and a positive absorption in 400 to 470 nm range (Fig. 3E). According to TDDFT results, this absorption originates from several dipole and spin-allowed metal-to-ligand and intraligand charge transfer (MLCT and ILCT) transitions of the singly reduced doublet species (D₀), i.e., into D₂₆, D₂₈, and D₃₀ (SI Appendix, Tables S6 and S7 and Fig. S12). Further addition of each Co(II) catalyst also decreased the longevity of the reduced CuPYBCP (Fig. 3 F and G). Notably, the estimated k_r with CoPYN5 is nearly five times of that with CoN5 (6.9 vs. 1.4 × 10⁹ M⁻¹ s⁻¹; Fig. 3H). This is in sharp contrast with the ns-TA results of the CuBCP system (Fig. 3D and Table 3) in which much smaller differences were observed between the two catalysts, which well demonstrates the promotive effect of dynamic π–π interaction in accelerating the intermolecular electron transfer process. A very fast PS-to-catalyst electron transfer rate of 6.9 × 10⁵ s⁻¹ (with 0.1 mM catalyst) (32) was achieved, which is faster than the covalently linked Ru(II)–Re(I) dyads (10⁴~10⁵ s⁻¹) (59) and a coordinatively interacted Ir(III)–Co(II) combination (2.9 × 10⁵ s⁻¹) (32) but slower than our previous Ir(III)–Co(II) system with π–π interaction (2.6 × 10⁶ s⁻¹) (33). We should note that the differences in the utilized transient spectroscopies and the electron transfer mechanisms should lead to deviations in estimating the electron transfer kinetics. Overall, the facilitated electron delivery successfully overcomes the negative impact of the very short-lived excited state of CuPYBCP, finally achieving a remarkable catalytic efficiency with CoPYN5 via dynamic π–π interaction which is superior to those pyrene-free systems.

To investigate the intermolecular electron transfer processes between the PS (CuPYBCP) and the Co(II) catalyst (CoPYN5) in more detail, quantum chemical simulations were performed to assess the associated thermodynamic properties, i.e., driving forces (ΔG), reorganization energies (λ), the underlying reaction coordinate, and the nature of the involved (photo)redox intermediates. We focused our computational studies exclusively on the reductive quenching pathway (Eqs. 1–3), as experimentally suggested by the Stern–Volmer kinetics. To this aim, the previously discussed (nonsandwich-like) dimer structure, combining the singly reduced PS and the nonreduced catalyst (SI Appendix, Table S9), was used as the electron donor state (D in Fig. 4). The respective Co(I) intermediate with a doubly occupied d_z²(Co) orbital was considered as the electron acceptor state. Due to computational reasons, the donor state was investigated within triplet multiplicity (open-shell), while the acceptor state was evaluated as (closed-shell) singlet state at the density and the time-dependent density levels of theory (DFT and TDDFT). A linear-interpolated reaction coordinate (R_ET) was constructed that connects the fully equilibrated structures of the donor and the acceptor states (36). The respective diabatic potential energy curves of the electron donor state and of the electron acceptor state are visualized in Fig. 4A, while the underlying reaction coordinate is shown in Fig. 4B. The most prominent structural rearrangement along R_ET is associated with the formation of the Co(I) species with a stable d⁸ configuration, which leads in consequence to a lowered coordination number and the dissociation of one CH₃CN molecule (solvent), reminiscent of the computational studies on CoN5 for CO₂ reduction (60). Based on the energies of the two diabatic states of interest in their respective equilibria, a driving force of merely 0.03 eV (ΔG) and reorganization energies of 0.80 (λ_D) and 1.29 eV(λ_A) were obtained. Unfortunately, the rate constants for such electron transfer process within the semiclassical Marcus picture, as previously done for light-driven electron transfer processes in photocatalytic dyads (61–63), could not be calculated due to the approximated spin states of the donor (triplet) and the acceptor (singlet) states. However, the weak driving force suggests a rather slow rate and the formation of an equilibrium. Therefore, the initial computational results present a clear contradiction to the rapid electron transfer observed for the CuPYBCP–CoPYN5 couple.

Fig. 4.

Electron transfer model. (A) Simulated potential energy curves obtained along a linear-interpolated internal reaction coordinate (R_ET) connecting (B) the fully optimized ³[CuPYBCP⁺⁰ + CoPYN5²⁺] donor state (in red; charge density difference is shown, charge transfer takes place from blue to red) and the ³[CuPYBCP⁺⁰ + CoPYN5²⁺] acceptor state (in green; doubly occupied d_z²(Co) acceptor orbital is shown). Quadratic polynomials were fitted to the donor (in red) and acceptor (in blue) states. Driving force (ΔG) and reorganization energies (λ_D and λ_A) are indicated. A further intermediate is shown in blue, ³[CuPYBCP⁺ + CoPYN5²⁺], which features a reduced catalyst with the two unpaired electrons localized in a π_pyr* orbital and in the d_z²(Co), see spin density. The reaction coordinate—associated with dissociation of one solvent molecule from the Co center—is visualized.

Nevertheless, a closer look at further available electron transfer states yields an alternative energetically low-lying acceptor state, where the electron is not transferred from the ${\pi }_{\mathrm{p}\mathrm{y}\mathrm{r}}^{\ast }$ of the reduced PS into the d_z(Co)-orbital to form the Co(I) species but into the respective low-lying ${\pi }_{\mathrm{p}\mathrm{y}\mathrm{r}}^{\ast }$ of the pyrene moiety of CoPYN5 (Fig. 4B, in blue). This process also yields a singly reduced catalyst, while it is an open-shell intermediate which is in all investigated structures energetically well below the Co(I) closed-shell species and also thermodynamically more favorable than the donor state. Consequently, the population of such alternative acceptor state with the electron being transferred to the pyrenyl π-system of CoPYN5 shows a stronger driving force of at least −0.49 eV (Fig. 4A) as predicted along R_ET. In contrast, such species is not available for CoN5 due to the absence of the pyrene group. Eventually, the combined experimental and computational results suggest that the population of different electron acceptor states in CoPYN5 and CoN5 accounts for the sharply varied electron transfer rates, which include a rather inefficient and slow population of the d_z(Co) orbital (1.4 × 10⁹ M⁻¹ s⁻¹; Fig. 3H) which is available for both catalysts, and a more efficient and faster (6.9 × 10⁹ M⁻¹ s⁻¹; Fig. 3H) population that generates the open-shell species only in CoPYN5 based on its highly accessible low-lying π* orbital.

Proposed Mechanism

With the confirmed reductive quenching pathway, and our previous investigations on the catalytic mechanism of CoPYN5 (33, 50), an overall photocatalytic cycle can be proposed for the CuPYBCP–CoPYN5 system in CO₂ reduction in Fig. 5. The excited state of CuPYBCP will react with BIH via the reductive quenching pathway to generate its reduced state. Then, the electron can be facilely transferred from the reduced-state species to CoPYN5 through π–π interaction. Notably, the π-system of the pyrene is not only involved in the preassembly of the PS and the catalyst but also acts—according to the performed (TD)DFT simulations—as the potential electron acceptor of CoPYN5. The two electron transfer processes discussed above have been experimentally identified by spectroscopic analyses and theoretical modeling. Next, in the cycle of CoPYN5 catalyst, its binding with CO₂ requires the second one-electron reduction, possibly by another reduced-state species of CuPYBCP (Fig. 5) or by the strongly reducing BI^· radical (53) from the deprotonation of BIH^·+. The C atom of CO₂ will coordinate to the Co center of the doubly reduced CoPYN5, which induces a bending of CO₂ due to its partial reduction by the catalyst and population of one ${\pi }_{\mathrm{C}\mathrm{O}}^{\ast }$ orbital (50, 60). The Co-CO₂ adduct will be protonated by the proton source and undergo C-OH cleavage to afford Co-carbonyl species, which releases CO and recovers the catalyst.

Fig. 5.

Mechanism. Proposed photocatalytic mechanism for the reduction of CO₂ using the CuPYBCP–CoPYN5 system.

Conclusion

Heteroleptic Cu(I) complexes have been extensively deployed in the noble metal–free systems for photocatalytic CO₂ reduction, while the further elevation of their catalytic performances presents challenging. The elegant design of ligand scaffold should allow the creation of Cu(I) PSs with photophysical properties for promoting the photocatalysis. After screening out excellent PSs and catalysts upon ligand engineering, a more fundamental strategy can be the installation of additional interactions between them, aiming at a boosted intermolecular electron transfer to facilitate the formation of catalytically active species. In this work, we expanded the library of Cu(I) PSs for CO₂ photoreduction with a rare dual emissive Cu(I) complex derived from pyrene decorations on the diimine ligand. Compared to its pyrene-free prototype, CuPYBCP exhibits unusual photophysical properties besides the dual emission, including the much stronger visible light absorption (Fig. 1B), the excitation-dependent emission behavior (Fig. 1C), and the much more short-lived excited states (ca. 9 vs. 289 ns, Fig. 1D). With experimental and computational proofs, under excitation at 425 nm, the emission at 653 and 720 nm can be rationalized as an ³MLCT excited state and the other one at 550 nm can be tentatively attributed to an excimer emission.

More impressively, as a proof of concept fully based on earth-abundant elements, the cooperation between CuPYBCP and the pyrene-appended CoPYN5 catalyst successfully established a noble metal–free system for visible light–driven CO₂ reduction possessing π–π interactions between PS and catalyst. As demonstrated by NMR titration (SI Appendix, Fig. S13), DFT simulations (SI Appendix, Table S9 and Fig. 4), and ns-TA spectroscopies (Fig. 3), the dynamic π–π interaction plays a key role in expediting the electron transfer from the reduced Cu(I) PS to the pyrene-appended catalyst, circumventing the unfavorable photocatalytic capabilities of CuPYBCP. Consequently, a remarkable Φ of 19% at 425 nm can be accomplished for selective CO₂-to-CO conversion, substantially outperforming the parent CuBCP and two noble metal–based PSs under parallel conditions. Compared to our previous work on a π–π-interacted Ir(III)–Co(II) system (33), the next-generation Cu(I)–Co(II) is highly advantageous in the absence of precious metals, the higher catalytic performance (Φ = 19% vs. 14%), and a clearer mechanism for the accelerated electron transfer demonstrated by quantum chemical calculations. Based on the above interesting results, we believe that our work opens an avenue to the exploitation and investigations of efficient earth-abundant systems for CO₂ photoreduction and offer valuable insights for the delicate synthetic modifications on the photophysics of Cu(I) PSs.

Methods

Materials.

CuBCP (64), BIH (65), CoN5 (66), and CoPYN5 (50) were prepared following the previously reported methods, and other chemicals were commercially available and used without further purification. Caution! Perchlorate salts of metal complexes with organic ligands are potentially explosive and should be handled in small quantities with care.

Synthesis of pybcp Ligand.

The pyrene-decorated ligand was prepared via Suzuki coupling reaction. 1 mmol 4,7-dibromo-2,9-dimethyl-1,10-phenanthroline, 2.5 mmol (4-pyrenylphenyl)boronic acid, 15 mol% P(PPh₃)₄, and 3 mmol K₂CO₃ were dissolved in 50 mL THF/H₂O (v:v = 5:1) and degassed with argon. After refluxing for 1 d, the reaction mixture was cooled to room temperature and subjected to extraction with DCM, followed by washing with brine twice and then drying over MgSO₄. The column chromatography with DCM/CH₃OH (2%) afforded the yellow powder with 52% yield. Elemental analysis (C₅₈H₃₆N₂): calculated: C, 91.55; H, 4.77; and N, 3.68; measured: C, 91.46; H, 4.84; and N, 3.62. ¹H NMR (400 MHz, CDCl₃) δ 8.34−8.18 (m, 12H), 8.17−8.00 (m, 18H), 7.88−7.74 (m, 11H), 7.73−7.62 (m, 5H), 7.60−7.42 (m, 4H), and 3.11 (s, 6H). ESI-MS(+): [pybcp+H]⁺ (m/z = 761.3).

Synthesis of CuBCP/CuPYBCP.

The synthesis follows a similar procedure for preparing heteroleptic Cu(I) PSs (16). In a 50-mL Schlenk tube, [Cu(CH₃CN)₄]PF₆ (186 mg, 0.5 mmol) and xantphos (279 mg, 0.5 mmol) or xantphos (289 mg, 0.5 mmol) are dissolved in 10 mL dry DCM at room temperature under N₂ and darkness. The resulting solution is stirred at 40 °C for 2 h. The reaction mixture was then slowly added with the bcp/pybcp ligand (0.5 mmol) under N₂ flow. The resulting mixture is then heated at 45 °C for 2 h. The reaction mixture was then cooled to room temperature and underwent precipitation by pouring into 200 mL dry hexane, giving CuBCP/CuPYBCP as bright yellow powder (86%/75% yield). For CuPYBCP, elemental analysis (CuC₉₇H₆₈N₂OP₃F₆): calculated: C, 75.26; H, 4.43; and N, 1.81; measured: C, 74.99; H, 4.53; and N, 1.72. ¹H NMR (4.0 mM, 400 MHz, CD₃CN) δ 8.37 – 8.17 (m, 13H), 8.16−8.02 (m, 6H), 7.93 (s, 2H), 7.85−7.76 (m, 6H), 7.71−7.66 (m, 4H), 7.59 (s, 2H), 7.31 (dt, J = 18.1, 7.9 Hz, 7H), 7.24−7.16 (m, 8H), 7.16−7.05 (m, 11H), 2.33 (s, 6H), and 1.75 (s, 6H). ¹³C NMR (126 MHz, CD₃CN) δ 158.11, 154.86, 149.29, 143.56, 141.76, 136.44, 135.59, 133.91, 133.44, 133.02, 132.96, 132.90, 131.65, 131.52, 131.44, 131.40, 130.88, 130.84, 130.21, 130.01, 129.73, 128.65, 128.61, 128.58, 128.21, 127.95, 127.84, 127.71, 127.65, 127.38, 126.47, 125.62, 125.52, 125.39, 125.13, 124.94, 124.70, 124.61, 124.49, 123.29, 121.66, 121.56, 121.46, 35.94, 27.90, and 26.90. ³¹P NMR (202 MHz, CD₃CN) δ −9.90. Q-TOF HR-ESI-MS(+): [Cu(xantphos) (pybcp)]⁺ (measured: 1,401.4065; simulated: 1,401.4097). FT-IR (cm⁻¹): 3,049 (w), 2,925 (w), 1,603 (w), 1,577 (w), 1,481 (w), 1,434 (m), 1,403 (s), 1,224 (m), 1,095 (w), 1,025 (w), 1,003 (w), 834 (s), 742 (m), 722 (m), 693 (m), 556 (m), and 512 (m).

Supplementary Material

Appendix 01 (PDF)

Data, Materials, and Software Availability

Experimental details, supplementary figures and tables, and computational data are available from the authors. Most data generated in this study are provided in SI Appendix/Source Data file. Source Data are provided with this paper.