Metal‐Free Aryl Cross‐Coupling Directed by Traceless Linkers†

Abstract The metal‐free, highly selective synthesis of biaryls poses a major challenge in organic synthesis. The scope and mechanism of a promising new approach to (hetero)biaryls by the photochemical fusion of aryl substituents tethered to a traceless sulfonamide linker (photosplicing) are reported. Interrogating photosplicing with varying reaction conditions and comparison of diverse synthetic probes (40 examples, including a suite of heterocycles) showed that the reaction has a surprisingly broad scope and involves neither metals nor radicals. Quantum chemical calculations revealed that the C−C bond is formed by an intramolecular photochemical process that involves an excited singlet state and traversal of a five‐membered transition state, and thus consistent ipso–ipso coupling results. These results demonstrate that photosplicing is a unique aryl cross‐coupling method in the excited state that can be applied to synthesize a broad range of biaryls.


Experimental Procedures General synthesis procedures
All reagents were obtained from commercial suppliers (Sigma Aldrich, TCI, Alfa Aesar, etc.) and used without further purification unless otherwise explained. Reactions were carried out under inert gas (argon) by using the Schlenk technique in dried solvents. Dichloromethane (DCM), acetonitrile (MeCN), methanol (MeOH) and chloroform were used from a solvent purification system (Innovative Technologies). Open column chromatographic separations were executed on silica gel (Kieselgel 60, 15-40 µm, Merck KGaA). Reaction progresses were monitored by thin layer chromatography (TLC) (silica gel on aluminium sheets 20 × 20 cm with fluorescent indicator 254 nm, Merck KGaA), GC-MS or HPLC-(HR)MS. Photoreactions were performed on a self-made photo reactor made from a milled aluminum block (16.9 × 31.9 × 0.03 cm, reaction volume ~15 mL) covered with a quartz glass slide. A SIMDOS 02 dosing pump from KNF was attached to the photo reactor with FEP tubes. A Herolab UVT-40 S equipped with 6 tubes Philips TUV 15 W/G15 T8 (total power 90 W; 44 W/m 2 at 245-260 nm) was used as the UV light source to illuminate the samples from above. A cryostat attached to the aluminum block allowed the control of the temperature and a small container stopped any air bubbles from the pump from reaching the photo reactor. Flow rates (0.5-10 mL/min) were selected in the course of reaction optimization.

Excess of toluene during the photosplicing of sulfonamide 2s
A solution of sulfonamide 2s (18.9 mg, 61.9 μmol) in a mixture of MeCN and toluene (v/v = 1:1) (4 mL) was prepared. The solution was loaded on the photo reactor with a flow rate of 1 mL/min (MeCN/toluene (v/v = 1:1)) and irradiated with UV light at room temperature. The solvent fraction containing the photoproducts was collected and the solvent was removed under reduced pressure. The crude product was analyzed by HPLC-HRMS in comparison with 1s, 2s and 2a. The results are shown in Figure S4.

Computational details
The ground state potential energy surface of 1a, 1p and 1aa were sampled by means of relaxed scans and conformer analysis. All ground state density functional theory (DFT) calculations were carried out using the range-separated exchange correlation functional CAM-B3LYP 4 and the all-electron def2-TZVP triple-ζ basis set 5 . Dispersion interactions were taken into account by Grimme's D3model with Becke-Johnson damping [6][7] .
Relaxed scans were performed with Gaussian 16, Revision B.01 8 at the DFT level (CAM-B3LYP/def2-TZVP) by varying the central (C-N-S-C)-dihedral angle in the sulfonamide-linker from -180° to 180° in 37 steps while equilibrating the remaining degrees of freedom at each step. Possible conformers of the biaryl substrates were generated by the simulated annealing procedure as implemented in Grimme's extended tight binding code GFN-xTB 5.8 using the GFN2-parametrization. Effects of solvation (acetonitrile) on the conformer geometries were taken into account by the generalized Born solvent area (GBSA) continuum solvation model 9 . Improved energies 10 for the conformers (generated by xTB) were calculated with the domain-based local pair natural orbital coupled cluster approach with triples corrections (DLPNO-CCSD(T)) as implemented in ORCA 4.0.1.2 [11][12] . The def2-QZVPP and the corresponding auxiliary basis sets were utilized [13][14][15] . Solvent effects (acetonitrile, ε = 36.6, n = 1.344) on the coupled cluster single point energies were taken into account by the conductor-like polarizable continuum model (CPCM) [16][17] . Tight criteria were used for the self-consistent-field convergence and the truncation threshold in the DLPNO procedure (TightPNO, TCutPairs = 10 -5 , TCutPNO = 10 -7 , TcutMKN = 10 -4 ) 18 . To visualize the correlation between the conformer bonding parameters and their coupled cluster energies, a principal component analysis was conducted using a set of eight internal coordinates as features. Transition state (TS) optimizations, subsequent vibrational analysis and reaction path calculations were carried out at the DFT level, while the nature of the first-order saddle points was confirmed by vibrational analysis. Minimum energy paths were obtained by calculating the intrinsic reaction coordinate (IRC) using the Hessian predictor-corrector method as implemented in Gaussian [19][20] to verify that the optimized TS connects the presumed educt and product of the biaryl coupling reaction. The exact Hessian was recalculated every seventh IRC step. Equally spaced geometries were sampled from both sides of the IRC every sixth step, yielding reaction paths for the subsequent excited state calculations. All excited-state calculations were carried out in Gaussian 16 using time-dependent DFT (TDDFT) and along the sampled IRCimages. The same basis set and dispersion correction model as for the preliminary ground state DFT calculations was applied. This computational setup allows a balanced description of local as well as of charge transfer excitations among the π-systems of educt and product states. 21 Vertical excitation energies and oscillator strengths for the six lowest singlet excited states were calculated for the sampled geometries along the IRC. Solvent effects (acetonitrile) on the vertical excitation energies and oscillator strengths were taken into account by CPCM. Excited state characters were interpreted in terms of natural transition orbitals (NTOs) 22 as calculated by Multiwfn 3.5 23 .

Conformational analysis
Definition of internal coordinates for the PCA Figure S6: Atom labels used in defining the internal coordinates for the principal component analysis (PCA) of 1a, 1p and 1aa.    S38 Nomenclature Figure S18: π-orbitals located on the S-linked phenyl residue (green) are denoted by a S-subscript (πS), whereas π-orbitals located on the CH2-NH-linked phenyl residue (blue) are given a N-subscript (πN).

Comparison of excited states along the IRC for 1a and 1p
TDDFT spectrum of 1aa, horseshoe

H and 13 C NMR Spectra
Figure S104: 13 C NMR spectrum of 1ad. Figure S105: 1 H NMR spectrum of 1ae.