Carbene-stabilized enantiopure heterometallic clusters featuring EQE of 20.8% in circularly-polarized OLED

Bright and efficient chiral coinage metal clusters show promise for use in emerging circularly polarized light-emitting materials and diodes. To date, highly efficient circularly polarized organic light-emitting diodes (CP-OLEDs) with enantiopure metal clusters have not been reported. Herein, through rational design of a multidentate chiral N-heterocyclic carbene (NHC) ligand and a modular building strategy, we synthesize a series of enantiopure Au(I)-Cu(I) clusters with exceptional stability. Modulation of the ligands stabilize the chiral excited states of clusters to allow thermally activated delayed fluorescence, resulting in the highest orange-red photoluminescence quantum yields over 93.0% in the solid state, which is accompanied by circularly polarized luminescence. Based on the solution process, a prototypical orange-red CP-OLED with a considerably high external quantum efficiency of 20.8% is prepared. These results demonstrate the extensive designability of chiral NHC ligands to stabilize polymetallic clusters for high performance in chiroptical applications.

For R-py-Br, due to the weak diffraction and serious disorder of PF6and CH2Cl2 186 molecules in the lattice, we used SADI restraints C-Cl bond of CH2Cl2 molecular; used 187 ISOR restraints F and C atoms of PF6and CH2Cl2 molecular. The twin law was tested 188 but refined to BASF of zero, and hence removed. Two outlier reflections were omitted 189 from the refinements. 190 For S-py-Br, due to the weak diffraction and disorder of PF6and CH2Cl2 molecules 191 in the lattice, we used ISOR restraints F and C atoms of PF6and CH2Cl2 molecular. 192 molecules in the lattice, we used SADI and DFIX restraints P-F bond of PF6molecular; 194 we used DFIX restraints C-C and C-O bonds of Et2O molecular; we used ISOR 195 restraints F atoms of PF6 -. Two outlier reflections were omitted from the refinements. 196 For S-py-I, due to the weak diffraction and serious disorder of Et2O molecules in the 197 lattice, we used DFIX restraints C-C and C-O bonds of Et2O molecular; we used SIMU 198 restraints O and C atoms of Et2O molecular. One outlier reflection was omitted from 199 the refinements. 200 For R-ql-Cl, due to the weak diffraction and serious disorder of Et2O molecules in 201 the lattice, we used DFIX restraints C-C and C-O bond of Et2O molecular; we used 202 SIMU and ISOR restraints F and C atoms of PF6and Et2O. The twin law was tested 203 but refined to BASF of zero, and hence removed.

204
For S-ql-Cl, due to the weak diffraction and serious disorder of PF6and Et2O 205 molecules in the lattice, we used DFIX restraints C-C and C-O bond of Et2O molecular; 206 we used SIMU and ISOR restraints F and C atoms of PF6and Et2O; we used ISOR 207 restraints disorder C atoms of NHC ql ligand. The twin law was tested but refined to 208 BASF of zero, and hence removed.

209
For R-ql-Br, due to the weak diffraction and serious disorder of Et2O molecules in 210 the lattice, we used DFIX restraints C-C and C-O bonds of Et2O molecular; used ISOR 211 and SIMU restraints O and C atoms of Et2O molecular; used ISOR restraints disorder 212 C atoms of NHC ql ligand. The twin law was tested but refined to BASF of zero, and 213 hence removed. Four outlier reflections were omitted from the refinements. atoms of NHC ql ligand. The twin law was tested but refined to BASF of zero, and hence 239 removed. Two outlier reflections were omitted from the refinements.

240
For R-ql-I-250 K, due to the weak diffraction and serious disorder of Et2O molecules 241 in the lattice, we used DFIX restraints C-C and C-O bonds of Et2O molecular; we used 242 SIMU restraints O and C atoms of Et2O molecular. We used ISOR restraints disorder C 243 atoms of NHC ql ligand. Three outlier reflections were omitted from the refinements.

244
For R-ql-I-300 K, due to the weak diffraction and serious disorder of Et2O molecules 245 in the lattice, we used DFIX restraints C-C and C-O bonds of Et2O molecular; used 246 SIMU restraints O and C atoms of Et2O molecular. We used ISOR restraints disorder C 247 atoms of NHC ql ligand. The twin law was tested but refined to BASF of zero, and hence 248 removed. Three outlier reflections were omitted from the refinements. 249 6. Quantum chemical calculations. 250 In this study, the PL origin of a series of ligand-protected Au(I)-Cu(I) alloy clusters 251 (R-py-X (X = Br and I) and R-ql-X (X = Cl, Br, and I)) has been studied by the density 10 functional theory (DFT) and time-dependent density functional theory (TD-DFT) 253 calculations. The DFT and TD-DFT calculations were carried out using the Gaussian Hartree/Å, respectively. The theoretical UV-Vis spectra of R-py-X (X = Br and I) and 259 R-ql-X (X = Cl, Br, and I) were calculated at the optimized ground-state (S0) geometries 260 using TD-DFT under PBE0/def2SVP level. SMD solvent model (dichloromethane) was 261 applied for TD-DFT calculations. The calculated absorption spectra were obtained from 262 Multiwfn 3.8 6 . The excited state gradients were calculated to optimize the excited-state 263 geometry. For R-ql-X (X = Cl and Br) the first singlet excited state (S1) and the low-264 lying triplet excited states (Tn, n = 1, 2) were optimized. For R-ql-I and R-py-I, during 265 the structure optimization of T2, we found that the energies of T2 and T1 are almost 266 degenerate, and the energy gap between S1 and T1 geometries is very small, the states 267 primarily relevant for the ISC are the S1 and T1 states.  The radiative rate of fluorescence (kf) and phosphorescence (kp) were evaluated by . For the S1→Tn ISC process, the SOCME 297 is calculated based on the optimized S1 geometry. For the Tn→S1 RISC process, the 298 SOCME is calculated based on the optimized Tn geometry. The contributions of the 299 three degenerate triplet states (Tn,x, Tn,y, and Tn,z) were taken into account by calculating 300 the root sum square of the real and imaginary parts (Re and Im) of the matrix elements, 301 as expressed by the following equation: It is noted that the reorganization energy can be represented as a sum of the 304 contributions from the surroundings and those from intramolecular vibrations.

305
Generally, the contribution of reorganization energy from the surroundings to the 306 system is small, and the calculation error is large, which can be ignored. Therefore, it 307 is generally believed that the reorganization energy of the system mainly comes from 12 the contribution of the intramolecular vibrations, and only the internal reorganization 309 energy is calculated. At present, the most common and simple and effective method for 310 calculating the reorganization energy is to use the classical method. For the S1→Tn ISC 311 process, the reorganization energy is the energy gap between Tn energy level at the 312 optimized S1 geometries and Tn energy level at the optimized Tn geometries. For the 313 Tn→S1 process, the reorganization energy is the energy gap between S1 energy level at 314 the optimized Tn geometries and S1 energy level at the optimized S1 geometries.

315
The interna conversion rate constant (kIC) between the Tn and Tn-1 are obtained by the 316 semiclassical Marcus theory expression (the equation (4)): [13][14][15] 317 where the Δ −1 is the adiabatic energy difference between the Tn and Tn-1 states.

319
The kB is the Boltzmann constant, and the temperature (T) is taken to be 298.15K. VSOC 320 is the spin-orbit coupling matrix elements (SOCME). For the Tn→Tn-1 IC process, the 321 SOCME is calculated based on the optimized Tn geometry. The contributions of the where kB is Boltzmann constant, τ is the experimental average lifetime, ΔE(S1-T1) is 347 the energy gap between S1 and T1 states, k(T1) = 1/τ(T1) and k(S1) = 1/τ(S1) are the 348 decay rates with decay times τ(T1) and τ(S1) of the triplet and singlet excited states, 349 respectively, and T is the temperature.

351
The TA spectra were recorded on a commercial pump-probe system (Helios, Ultrafast

352
Systems LLC) in combination with a femtosecond laser system (Astrella, Coherent).

355
The laser pulses were split to generate pump and probe beams. The pump pulses at 360

634
For R-py-I, the NHC py ligands rendered the resulting R-py-I complexes had CT excited 635 states T1, and more local excitation ( 1 LE) character in S1. At the same time, the atomic 636 orbitals contribution of S1 and T1 had big differences (Supplementary Table 25).
of S1 and T2 for the MECP between S1 and T2 geometries of R-ql-Cl. of S1 and T1 for the S1, T1 geometries of R-py-I. rates of R-ql-X (X = Cl, Br, and I) and R-py-I from S1 to S0 or T1 to S0. 1 and 2 represent high-energy and low-energy emissions, respectively.