Synthesis and characterisation of Cu(II)-dye coordination polymers. The installation of a chiral oxazolidinone moiety24 at ortho-position of carboxylic linker of the typical TPA-based ligand tris(4’-carboxybiphenyl)amine (named as H3L-Planar) afforded a new ligand tris[4-(4-carboxy-3-((R)-4-isopropyl-2-oxooxazolidin-3-yl)-phenyl)phenyl]amine (named as H3L-Twisted). The coordination polymer Cu–Twisted was obtained in a 70% yield by the solvothermal reaction between Cu(NO3)2∙ 3H2O and H3L-Twisted at 80 oC for 3 days. Cu–Twisted consisted of truncated cubic cages (internal diameter ca. 2.9 nm, Fig. S2), each delimited by eight L-Twist3− linkers and twelve Cu2(O2C)4 paddle-wheels, forming dumbbell-shaped 1D channels with a cross-sectional area of 15 × 27 Å2 along the a-direction (Fig. 2b).25 The free volume of desolvated Cu–Twisted was estimated as ca. 62% of the porous polymer, and a Brilliant Blue R-250 dye uptake experiment with Cu–Twisted yielded a 54% absorption amount of the coordination polymer weight as determined by UV-Vis spectroscopy (Fig. S9), implying the possibility of accommodating substrates and reagents within the pores.26 The steric hindrance between the bulky oxazolidinone and the Cu2(O2C)4 paddle-wheel forged a twisted dihedral angle between the conjugated carboxylic coordination group and the adjacent phenyl moiety (Fig. 2a),27 which induced the symmetry of coordination polymer into a pto lattice (Figs. 2b and S2).28 This distortion together with the polar carboxylate–copper node provided the possibility to finely modify the electronic communication for the kinetic alleviation of fluorescence quenching of Cu(II) towards the highly reductive excited-state TPA moiety.18
In the absence of bulky auxiliary, a control catalyst was prepared by solvothermal reaction between Cu(NO3)2∙3H2O and H3L-Planar (45% yield). The (3,4)-connected three dimensional (3D) networks (internal diameter ca. 3.8 nm, Fig. S4) were assembled between the tritopic ligands and 4-connected Cu2(O2C)4 paddle-wheels in a high-symmetry tbo topology (Table S1), similar to the well-known HKUST–1.29 Notably, the absence of axial rotational restriction from the bulky auxiliary allowed the conformation of coplanar conjugation between the polar Cu(II)–carboxylate node and the adjacent phenyl moiety of ligand (Fig. 2c). Thus, this control coordination polymer was named as Cu–Planar. It should be noted that the existence or not of distortion along the axial direction of ligand did not affect the identical Cu–NTPA distances (ca. 12.6 Å) in both Cu–Twisted and Cu–Planar (Figs. 2b and 2d), and the non-interpenetrated nature of two coordination polymers precluded the intermolecular luminescence quenching of Cu(II) nodes to the excited state of indirectly connected TPA-moieties. Moreover, the vast internal cavities and open windows of both Cu(II)-TPA coordination polymers were believed to facilitate the rapid mass transfer during their catalytic applications.
Electrochemical analyses of both Cu–Twisted and Cu–Planar exhibited the peaks at ca. 1.0 V corresponding to the redox potentials of the TPA-based ligands (Fig. S11). The similar reductive potentials of the excited-state frameworks were determined as -1.49 V and − 1.59 V for Cu–Twisted and Cu–Planar, respectively (Table S2), based on the free energy change (E0 − 0) between the ground and vibrationally relaxed excited states (Fig. S12),30 demonstrating that both the excited-state coordination polymers can theoretically reduce iodomesitylene dicyclohexanecarboxylate (abbreviated as MesI(OCOCy)2, E1/2red = -1.14 V vs. SCE), one of the prominent iodonium carboxylate-type radical precursors, for the generation of alkyl radicals.7
Electrochemical impedance spectroscopy (EIS) of the two coordination polymers exhibited different arc radii in the high-frequency regions, demonstrating that the charge-transfer resistance (Rct) of Cu–Twisted (ca. 4.8 kΩ) was roughly three times larger than that of Cu–Planar (ca. 1.6 kΩ) (Fig. 3c). The photocurrent responses of two coordination polymers suggested their well reproducible photocurrents upon on/off cycles of the light irradiation (Fig. 3d). Compared with Cu–Twisted, a significant enhancement in the photocurrent response was observed for the case of Cu–Planar, indicating more efficient intramolecular PET process. Clearly, the twisted conjunction mode did not affect the thermodynamic photoreducing abilities of coordination polymers, but remarkably altered the kinetic features of intramolecular electronic communications.
The UV-Vis absorption spectra of solid-state Cu–Twisted and Cu–Planar exhibited the bands covering a broad visible range (400 ~ 550 nm, Fig. S10) that attributed to the absorption of TPA-based ligands, and the typical d-d transition peaks of dicopper paddle-wheel units were observed above 600 nm. As exhibited in Fig. 3a, upon photoirradiating the Cu–Twisted suspension in 1,4-dioxane, the obvious 536 nm-centred luminescence band indicated the weak intramolecular quenching of Cu(II) nodes towards the highly reductive excited-state TPA moieties. In comparison, the remarkably lower fluorescence intensity and shorter fluorescence lifetime (2.1 ns vs. 2.9 ns) of Cu–Planar suspension in 1,4-dioxane than those of Cu–Twisted under the identical conditions (Figs. 3a and 3b) suggested the more efficient intramolecular fluorescence quenching in the coordination polymer Cu–Planar.
Electron paramagnetic resonance (EPR) studies revealed that both coordination polymers exhibited the characteristic signals of Cu(II) ions with g value of ca. 2.09 (Fig. 5a).31 Photoirradiation with compact fluorescent lamps for 15 min lead to significant depression of Cu(II) signal in EPR spectra of Cu–Planar. In comparison, the identical photoirradiation did not cause obvious Cu(II) signal variation in EPR spectra of Cu–Twisted, confirming that the twisted conjugation between TPA-moiety and polar carboxylate–copper node effectively hampered the intramolecular PET process between them (Figs. 2a and 2c). Moreover, the much greater amount of Cu(II) sites that retained in the photoirradiation of diode-like Cu–Twisted was believed to benefit the Cu(II) catalytic steps in its combination with photocatalysis (Fig. 1c).
Heterogeneous synergistic photoredox and copper catalysed decarboxylative C(sp 3 )-heteroatom coupling. Noted that the carboxylic acids are more abundant, stable, and less toxic chemical feedstocks compared with alkyl halides, the classical alkyl radical precursor, thus the redox-activated carboxylic acids like iodonium carboxylates are considered as the practical and sustainable radical sources in this case. We next examined the synergistic photoredox and copper catalytic performances of the obtained Cu(II)–TPA coordination polymers using decarboxylative C(sp3)-heteroatom coupling of iodonium carboxylates and heteroatom nucleophiles as the benchmark.7 Upon addition of MesI(OCOCy)2 into the suspensions of Cu–Twisted or Cu–Planar in 1,4-dioxane, the luminescences of coordination polymers were markedly quenched (Fig. S15). This result suggested a productive intermolecular PET from the excited state of TPA moiety to MesI(OCOCy)2 was allowed for the generation of alkyl radicals. In a typical procedure, a mixture of MesI(OCOCy)2, nitrogen-centred nucleophile 3-chloroindazole 2a, basic additive BTMG (BTMG = 2-tert-butyl-1,1,3,3-tetramethylguanidine), and 2.5 mol% Cu–Twisted in 1,4-dioxane was subjected to visible-light irradiation from a 405-nm LED under a N2 atmosphere, and the desired C(sp3)–N coupling product was obtained in 84% isolated yield after 5 hours (Fig. 4, 3a). Subsequently, a series of iodonium carboxylates derived from the primary linear, secondary acyclic and cyclic, and tertiary substituted alkyl carboxylic acids were found applicable in this protocol to deliver the corresponding N-alkyl heteroaryl products 3 in medium to good efficiencies (Fig. 4, 3a–3f). Especially, the use of tertiary carboxylic acid 3e successfully introduced a sterically demanding adamantine moiety into the product in 86% yield; it should be noted that N-alkylation employing tertiary alkyl halides is elusive using traditional nucleophilic substitution. Interestingly, an ibuprofen derivative valuable to medicinal chemistry was accessible by this methodology (3f). In addition to indazole, other nitrogen heterocycles like indole, pyrrole, and phthalimide were also amenable to this N-alkylation protocol (3 g–3i). This catalytic system well tolerated functional groups such as terminal (3c) or internal (3d) olefins, which were inherently susceptible to radical addition, implying that Cu–Twisted might harness the behavior of radical species in an inner-spheric manner.
Control experiments demonstrated that coordination polymer catalyst and light irradiation were indispensable for the coupling reaction (Table S3, entries 2 and 3), neither the dark condition nor the absence of Cu–Twisted gave the formation of 3a. There was no noticeable further conversion after hot filtration of Cu–Twisted, suggesting the heterogeneous nature of reaction (Table S3, entry 14). After photocatalysis, the coordination polymer was easily isolated from the reaction mixture by centrifugation and could be reused at least three times without a marked decrease in reactivity (83 ~ 79%; Fig. S25). The PXRD pattern of the recovered catalyst indicated that it maintained its structural integrity (Fig. S21). The nearly similar catalytic performances of the pristine crystals of coordination polymers compared to the finely ground samples ruled out the vital influence of particle sizes of heterogeneous catalysts towards catalytic efficacy (Fig. S22), implying that the reaction mainly occurred in the pores of coordination polymer (Table S3, entries 1, 4, 11 and 12).
When a radical scavenger tetramethylpiperidine-N-oxyl (TEMPO) was added to the reaction mixture, the conversion was inhibited immediately (Table S3, entry 13), suggesting a radical mechanism. The only use of ester form of ligand Et3L-Twisted or Cu(II) salt as catalyst gave no reactions (Table S3, entries 7 and 9), reflecting the necessity of two kinds of catalytic sites. The simple combinations of Cu(II) salt either with Et3L-Twisted or Me3L-Planar afforded the low conversions, probably owing to the futile intermolecular fluorescence quenching of Cu(II) salt (Table S3, entries 5 and 6). The reference coordination polymer Cu–Planar gave a 41% yield of 3a. Although this conversion was only half the level of Cu–Twisted (Table S3, entries 1 and 4), at least the incorporation of TPA moieties together with polar Cu(II)–carboxylate nodes into coordination polymer partially hampered the intermolecular fluorescence quenching that limited the direct combination of photoreductive dye and Cu(II) catalyst in the solution (Figs. 3a and 5a). Furthermore, if the similar redox potentials (Table S2), substrate encapsulating abilities, and intra-pore substrate diffusion kinetics (Fig. S24) of Cu–Twisted and Cu–Planar were also taken into account, the distinct catalytic efficiencies of two coordination polymers (Fig. S23) might be attributed to their remarkably different intramolecular electronic communications.
The circular dichroism (CD) spectra of Cu–Twisted exhibited a negative signal at approximately 346 nm and positive dichroic band centered at 275 nm similar to that of H3L-Twisted (Fig. S8), and the Cu(II) ions here were enveloped in the semi-open chiral pores (Fig. S2), which possibly provided stereoinduction in the radical capture of Cu(II) sites. When employing the iodonium carboxylate prepared from chiral natural derivative (1R,3R,4S)-p-menthane-3-carboxylic acid, the net resulting stereochemistry of the carboxylic α-carbon was fortunately maintained during C(sp3)-N coupling (dr. > 20:1, Fig. 4, 3j) in the presence of proximal chiral centres on the menthane scaffold. After photoreductive decarboxylation, the frequent flipping of the formed C(sp3)-centred radical erases the original stereoinformation,32 thus it should be practical to use the radical precursors derived from the racemic α-substituted carboxylic acids. For 3 k and 3 l, chiralities were successfully introduced into the achiral α-positions of the carboxylic moieties by employing cheap racemic camphanic acid (1 k) and norbornene carboxylic acid (1 l) instead of their expensive enantioenriched homologs, demonstrating the inner-spheric redox process between the in situ generated radical species and the well-retained Cu(II)33 within asymmetric local environment for controlling this diastereoselectivity.
Beside the iodonium carboxylates, the redox-active N-hydroxyphthalimide (NHPI) esters34 derived from alkyl carboxylic acids were also the prominent alkyl radical precursors in decarboxylative cross-coupling reactions, as the surrogates for alkyl halides. Like that of iodonium carboxylate MesI(OCOCy)2, the addition of NHPI ester 4c into the Cu–Twisted suspension strongly quenched the luminescence (Figs. 5b and S16), whereas NHPI ester 4c hardly quenched the fluorescence of Et3L–Twisted (with a quite low quenching constant Ksv < 0.01 µmol− 1·L). As the excited-state reductive potential of Et3L-Twisted was more negative than that of Cu–Twisted (Table S2), the Cu(II) sites of coordination polymer Cu–Twisted were speculated to improve the fluorescence quenching process of NHPI ester. IR spectra of Cu–Twisted crystals that soaked in the solution of NHPI ester 4c exhibited the red-shifted carbonyl stretching peak of 4c compared with that of the free molecule (Figs. 5c and S18), this potential interaction between the carbonyl of radical precursor 4c and the Cu(II) site of coordination polymer might draw the radical precursor closer to the nearby photocatalytic centre,35 which might be correlated to the above-mentioned enhancement of fluorescence quenching effect. Similarly, a bathochromic shift in the N–H stretching vibration of ortho-anisidine 5a (3460 cm− 1 to 3444 cm− 1) was also observed after substrate incubation within Cu–Twisted, verifying that the nitrogen atom of N–H was the nucleophilic binding site and could be activated during substrate adsorption within the coordination polymer (Figs. 5d and S19).
With minor changes of reaction conditions, this heterogeneous synergistic catalytic system was also applicable to the coupling of diverse primary, secondary, and tertiary alkyl redox-active NHPI esters with anilines in good yields (Fig. 6, 6a–6e).36,37 Then, the substrate scope of this heterogeneous approach could be easily expanded from anilines to chalcogen-centred nucleophiles.38 Phenols were successfully employed in C(sp3)–O coupling (6f–6 h), and the ortho-phenylphenol fungicide derivative 6 h was obtained in 55% yield. Further extension of chalcogen-type nucleophiles to thiophenols facilitated rapid access to thioether-derived Naproxen 6 l, showcasing the efficient late-stage functionalisation for drugs and the perfect resistance to the potential sulphur-poisoning effect.
As shown in Fig. 7, our heterogeneous synergistic catalytic strategy using a twisted-conjugated Cu(II)–dye coordination polymer supplies a unified mechanistic paradigm to meet the diversified needs of reaction kinetics between photoredox and copper catalytic cycles3 when employing the different types of substrates and radical precursors. Upon light irradiation, the diode-like coordination polymer Cu–Twisted hampers the intramolecular fluorescence quenching to switch on the intermolecular PET from the excited-state TPA moiety of ligand to the alkyl radical precursor that possibly activated by the Cu(II) site. The in situ generated alkyl radical is subjected to an inner-spheric redox process with the Cu(II) site that coordinates with the heteroatom-centred nucleophile, delivering Cu(I) species and the desired coupling product, as depicted by the pioneering homogeneous protocols.39 Then, the radical cation of the oxidized TPA moiety retrieves one electron from the Cu(I) site through the twisted connection at the ground state, which completes the productive closed-circuit electron transfer route for the synergy of photocatalytic and copper catalytic cycles and simultaneously regenerates the resting-state Cu–Twisted for the next round of reaction.14 Within the confined environments of Cu–Twisted, the well-retained Cu(II) sites enriches substrates and reagents to improve the local concentration of activated nucleophiles and in situ generated alkyl radicals. This key kinetic modulation is believed to furnish the heterogeneous synergistic photoredox and copper catalysis in a concise and easy handling manner, which alleviates the reliance on the varied and elaborative reaction conditions obtained by the massive screening efforts that typically needed in the homogeneous protocols when using different types of nucleophiles and radical precursors. In a comparison, the low concentration of the transient Cu(II) species that in situ generated from the homogeneous Cu(I)-photocatalyst system might hamper them from exerting the readily accessible modulation effects.3,40
Heterogeneous synergistic photoredox and copper catalysed trifluoromethylation-chloration difunctionalisation. As mentioned above, the well-retained Cu(II) centres during PET process enabled the inner-spheric capture of single-electronic radical intermediates and the fixation of double-electronic heteroatom-centred nucleophiles and carbonyl-containing radical precursors, which shed light on different-typed value-added catalytic application by integrating nucleophiles, carbonyl-containing substrates, and radical species in a step-economic manner, such as the pharmaceutically important trifluoromethylation-chloration difunctionalisation41 of olefins of α,β-unsaturated compounds. Appending CF3 and Cl moieties to olefins was highly dependent on the copper-mediated inner-spheric processes,42,43 since that the olefinic moieties were prone to either the intermolecular radical oligomerisation/polymerisation or the intramolecular cyclisation when encountered with free radicals. The negative enough reductive potential of excited-state Cu–Twisted allowed the photoreduction of triflyl chloride (TfCl) (E1/2red = -0.18 V vs. SCE),44 the typical CF3 radical precursor. As a consequence, the quenching of 512-nm luminescence of the acetonitrile suspension of Cu–Twisted crystals upon the addition of TfCl was indicative of the PET process from excited-state TPA-based ligand to TfCl for the generation of CF3 radicals under light irradiation (Fig. S17).
Given the propensity of the methacrylate ester to undergo the radical oligomerisation/polymerisation,45 the phenyl methacrylate (Fig. 8a, 7a) was chosen as the model substrate of difunctionalisation by Cu–Twisted. In a typical procedure, a mixture of phenyl methacrylate 7a, TfCl, basic additive (2,4,6-collidine), and 2.5 mol% Cu–Twisted in acetonitrile was subjected to visible-light irradiation from two household compact fluorescent lamps (CFLs), obtaining the trifluoromethylation-chloration product 8a in an isolated yield of 92% (Fig. 8a), and no radical oligomerisation/polymerisation products could be detected. The only use of Et3L- Twisted resulted in the dominative oligomerisation/polymerisation of the substrate (Table S5, entry 7), and the simple combination of Cu(II) salt and Et3L-Twisted gave a hugely diminished conversion (13%), reflecting the necessity of the holo coordination polymer for hampering the undesirable intermolecular fluorescence quenching (Table S5, entry 9). When using the controlled catalyst Cu–Planar, a lower yield of 45% was detected, implying the importance of well-rectified intramolecular electronic communication in the diode-like catalyst for the efficient diffunctionalisation (Table S5, entry 4). Furthermore, various α,β-unsaturated esters (7b to 7e) were transformed to the corresponding α-chloro-β-trifluoromethyl ester products in good to high yields. Besides, the α,β-unsaturated amides (7f to 7 h) also participated well in this difunctionalisation, and the well-known radical cyclisations with intramolecular N-phenyl moieties were effectively depressed (see also Fig. S28).46 Those results indicated that the carbon-centred radical intermediates formed after CF3 radical addition might be intercepted by the Cu(II) nodes to prohibit the side-pathways like oligomerisation/polymerization or cyclisation. Attaching bulky auxiliaries to the neighbouring positions of carbonyl groups of the substrates should benefit the mutual stereo-recognition and induction between substrates and local environments (7i to 7 k). In particular, the reaction of 7 k bearing a bio-relevant dihydrocholesterol fragment, afforded the targeted product nearly as sole diastereomer (Fig. 8a, 8 k).
Structurally complex small molecules containing repeating functional groups,47 like multiple olefinic natural products, have an extraordinary capacity for a wide range of useful functions.48 The similar thermodynamic reactivities of different olefinic sites of natural derivatives made the discrimination of olefinic groups a challenging task in the presence of highly active CF3 radicals, like in the case of trifluoromethylation-chloration difunctionalisation (Table S6).49 Here, the natural flavour derivative 7 l containing a carbonyl-adjacent olefin and a carbonyl-free olefin in the molecular scaffold was chosen as the model substrate.50 DFT calculation revealed the thermodynamic feasibility of docking the carbonyl site of substrate 7 l through Cu–O interaction (Table S7), with a free energy change of ca. 12.44 kcal∙mol− 1 in 7 l over the adsorption upon Cu(II) node. The IR spectra of Cu–Twisted with encapsulated 7 l suggested that the C = O stretching peak of 7 l was red-shifted relative to that of the free molecule (1744 to 1732 cm− 1, Fig. S20), verifying the possible fixation of the carbonyl moiety of substrate on the Cu(II) site. As a consequence, the distances between the multiple olefinic sites of the docked substrate and the catalytic centres should be effectively differentiated within Cu–Twisted, which might provide an ideal model of the Hammond postulate-typed site-selectivity control (Fig. S30).51
Under the typical reaction conditions of Fig. 8a, the trifluoromethylation-chloration difunctionalisation of 7 l solely occurred on the carbonyl-adjacent olefinic site and afforded the formation of product 8 l in an isolated yield of 71% (Fig. 8b), and the carbonyl free olefinic site of allylbenzene terminal of 7 l was well retained. To the contrast, the olefinic site of the free standing control substrate allylbenzene well participated in the difunctionalisation under the identical catalytic condition (Fig. S29).43 It was deduced that the spatial proximity between the TPA moiety and the docked carbonyl of substrate might facilitate the formation of a product-like late transition state52 by restricting the photogenerated CF3 radical near to the fixed α,β-unsaturated olefinic group in the confined space, which might be important to kinetically distinguish the bonded and unbonded olefinic sites within the same molecules (Fig. S30). When the biologically interesting steride scaffolds containing olefinic sites were merged together with the α,β-unsaturated esters, the corresponding steride derivatives 7p and 7q delivered excellent regio- and diastereocontrol simultaneously, and the carbonyl-adjacent olefinic terminals of substrates were converted with perfect diastereoselectivity while retaining single or even multiple carbonyl-free alkenes in the fragments of cholesterol and ergosterol (Fig. 8b, 8p and 8q), illustrating the potential of this heterogeneous synergistic photoredox-copper catalytic system in drug discovery.
Then, DFT calculations were performed to investigate the role of readily accessible Cu(II) sites of Cu–Twisted in discriminating different olefinic sites during the synergistic catalytic difunctionalisation. As shown in Fig. 9, the computed energy profiles of the reaction pathways of the carbonyl-adjacent olefinic site (pathway a, in red) of 7 l and the opposite-side carbonyl-free allyl terminal (pathway b, in blue) were compared. Addition of the CF3 radical to the olefinic site nearby the docked carbonyl 7 l occurs via transition state TS I by an energy barrier of 16.76 kcal∙mol− 1 to form carbon-centred radical intermediate II. Subsequently, the chloride anion coordinates to the copper centre and replaces the carbonyl moiety, and the copper–Cl species mediates Cl-atom transfer to form 8 l.53 The overall transformation via pathway a is exergonic by 12.19 kcal∙mol− 1. In contrast, addition of the CF3 radical to the carbonyl-free allyl terminal of 7 l occurs through transition state TS I’ with a higher energy barrier of 25.6 kcal∙mol− 1, and all subsequent steps also proceed through higher energy barriers despite pathway b being more thermodynamically favourable overall (exergonic by 24.21 kcal∙mol− 1).
Consequently, the two thermodynamically allowed pathways were distinguished within coordination polymer, and the difunctionalisation of olefinic moiety that adjacent to carbonyl binding site in the same molecules was kinetically favoured (Fig. 8b and Table S6). Moreover, it was revealed that the adsorption of substrate 7 l was more favoured than that of product 8 l (the comparison of DFT calculated free energy changes, -12.44 < -11.03 kcal·mol− 1, see Table S7). This result implied that the generated product may be crowded out from Cu(II) centre by the competitive adsorption of a substrate molecule to trigger the new round of reaction, facilitating the recyclability of the heterogeneous catalysts.