A reaction mode of carbene-catalysed aryl aldehyde activation and induced phenol OH functionalization

The research in the field of asymmetric carbene organic catalysis has primarily focused on the activation of carbon atoms in non-aromatic scaffolds. Here we report a reaction mode of carbene catalysis that allows for aromatic aldehyde activation and remote oxygen atom functionalization. The addition of a carbene catalyst to the aldehyde moiety of 2-hydroxyl aryl aldehyde eventually enables dearomatization and remote OH activation. The catalytic process generates a type of carbene-derived intermediate with an oxygen atom as the reactive centre. Inexpensive achiral urea co-catalyst works cooperatively with the carbene catalyst, leading to consistent enhancement of the reaction enantioselectivity. Given the wide presence of aromatic moieties and heteroatoms in natural products and synthetic functional molecules, we expect our reaction mode to significantly expand the power of carbene catalysis in asymmetric chemical synthesis.


General Information
Commercially available materials purchased from Alfa Aesar or Sigma-Aldrich were used as received. Proton nuclear magnetic resonance ( 1 H NMR) spectra were recorded on a Bruker BBFO (400 MHz) spectrometer. Chemical shifts were recorded in parts per million (ppm, δ) relative to tetramethylsilane (δ 0.00) or chloroform ( = 7.26, singlet). 1 H NMR splitting patterns are designated as singlet (s), doublet (d), triplet (t), quartet (q), dd (doublet of doublets); m (multiplets), and etc. All first-order splitting patterns were assigned on the basis of the appearance of the multiplet. Splitting patterns that could not be easily interpreted are designated as multiplet (m) or broad (br). Carbon nuclear magnetic resonance ( 13 C NMR) spectra were recorded on a Bruker BBFO (100 MHz) spectrometer. Fluorine ( 19 F) nuclear magnetic resonance ( 19 F NMR) spectra were recorded on a Bruker BBFO (376 MHz) spectrometer. IR spectra were recorded on a Shimadzu IR Prestige-21FT-IR spectrometer as neat thin films between NaCl plates. High resolution mass spectral analysis (HRMS) was performed on Finnigan MAT 95 XP mass spectrometer (Thermo Electron Corporation). The determination of enantiomeric excess was performed via chiral HPLC analysis using Shimadzu LC-20AD HPLC workstation. X-ray crystallography analysis was performed on Bruker X8 APEX X-ray diffractionmeter. Optical rotations were measured using a 1 mL cell with a 1 dm path length on a Jasco P-1030 polarimeter and are reported as follows: [α] rt D (c in g per 100 mL solvent). Analytical thin-layer chromatography (TLC) was carried out on Merck 60 F254 pre-coated silica gel plate (0.2 mm thickness). Visualization was performed using a UV lamp.

S60
General procedure for the catalytic reactions: General procedure for the catalytic synthesis of products 3: To a 10 mL flame-dry Schlenk reaction tube equipped with a magnetic stir bar, was added chiral NHC pre-catalyst C2 (0.005 mmol, 5 mol%, 2.4 mg), urea A3 (0.02 mmol, 20 mol%, 8.76 mg), DABCO (0.1 mmol, 100 mol%, 11.2 mg), oxidant DQ (0.11 mmol, 110 mol%, 45 mg), aldehyde (0.11 mmol) and 4 Å molecular sieves. The Schlenk tube was sealed with a septum, evacuated and refilled with nitrogen (3 cycles). Solvent (CH 2 Cl 2 /Hexane =1:1, 2.0 mL) and trifluoromethyl ketone 2 (0.1 mmol) were then added via syringe. The reaction mixture was allowed to stir for 24 hours at -10℃. After completion of the reaction, monitored by TLC plate, the reaction mixture was concentrated under reduced pressure and the residue was subjected to column chromatography or TLC plate directly using hexane/EtOAc as eluent to afford the desired product 3.
Note: Racemic samples for the chiral phase HPLC analysis were prepared using C as the NHC pre-catalyst.

Stereochemistry determination via X-ray crystallographic analysis:
Good quality crystal of 3h (colorless needle crystal) was obtained by vaporization of a hexane/ethyl acetate solution of compound 3c (~500mg). CCDC 1486140 contains the supplementary crystallographic data for this paper. These data can be obtained free of charge from The Cambridge Crystallographic Data Centre via www.ccdc.cam.ac.uk/data_request/cif.

and 1 H spectra for urea and ketone with different solvent
Based on 1 H NMR spectra, we can found that thiourea co-catalyst dose not react with ketone substrate. Based on 19 F NMR spectra, we can see that in non-polar solvents such as CDCl 3 and d8-toluene, in the presence of thiourea, the chemical shift of 19 F was changed compared to no thiourea one (although very small difference, it is also reasonable for weak hydrogen bonding effect). However, in polar solvent such as CD 3 CN, the chemical shift of 19 F totally same no matter with or without thiourea. Therefore, thiourea co-catalyst can serve as H-bonding donor, which is consistent with our proposal.

S64
Computational method for mechanism studies: We performed density functional theory (DFT) calculations using Gaussian 09 software 1 . For geometry optimization calculations, the B3LYP functional [2][3][4] was used in combination with the 6-311G(d) basis set. 5,6 Single-point calculations on the optimized structures were performed using the dispersion-corrected B3LYP (B3LYP-D3) 2-4,7-11 functional and the 6-311+G(d,p) basis set, 5,6,12 with Grimme's D3 correction 7,8 combined with Becke-Johnson (BJ) damping. [9][10][11] The solvent effect of DCM was taken into account using the IEF-PCM method 13 in single-point calculations. Although a mixed solvent (CH 2 Cl 2 /hexane = 1:1) was used in the optimized experimental condition, for the sake of simplicity, we examined the effect of CH 2 Cl 2 because similar e.r. was observed experimentally with this solvent. Vibrational frequency calculations were performed to confirm that each optimized structure resides at a stationary point on the potential energy surface. Free energy corrections were also obtained from frequency calculations. All relative energies are free energy at -10 °C (263.15 K) and 1 atm. Mulliken atomic charges were calculated at the B3LYP/6-311G(d) level.

Conformation of triazolium salt precursor C2
Triazolium salt precursor C2 has two conformers (A and B in Fig. S58) that differ in the geometry of the oxygen-containing ring. Conformer A is more stable than conformer B by 1.3 kcal/mol. Therefore, in the following calculations, conformer A was used for C2. Six conformers of II were obtained from DFT calculations, and II-1 was the most stable one. A weak CH· · · O hydrogen bond between the phenolate oxygen and hydrogen atom of the indane moiety from the NHC catalyst was formed in II-1 but not in II-2, which makes the intermediate II-1 more stable by 0.9 kcal/mol. II-3 and II-4 are even less stable, probably because the phenolate oxygen points away from the NHC group, leading to weaker electrostatic stabilization. II-5 and II-6 ( Fig. S59), in which a carbon atom of the azolium ring forms a covalent bond with the phenolate oxygen atom, are less stable than II-1. The covalent bond in II-5 and II-6 prevents the azolium ring from gaining resonance stabilization that II-1 can have, and this explains the difference in stability of these intermediates. Key Mulliken atomic charges for II-2 are given in Supplementary Table 5. The charge distribution in II-2 is similar to that in II-1.

Mechanisms of [4+2] annulation without urea
The calculations excluding urea suggest that NHC-bound intermediate II can react with ketone substrate 2a in a [4+2] mechanism (Fig. 1c) to afford a ketal-like product, without having too high barriers. Our calculations also suggest that the reaction should consist of two steps: (1) a ring-annulation step and (2) an NHC dissociation step (see Fig. S60). The four conformers (II-1, II-2, II-3, and II-4) in Fig S59 can react with 2a, giving rise to several possible reaction pathways (16 pathways in total). Nevertheless, we here focus on low-energy pathways. When the resultant product does not have the predominant stereochemical pattern observed in the experiment, "minor" is attached to the labels. The o-QM moiety has two distinct faces on which 2a can possibly attack, and thus reaction pathways may be classified into Paths A and B; Path A features the attack of 2a on II from the "indane side" depicted in Fig. S61, whereas Path B involves the attack from the "PhCl 3 side" in Fig. S61. For example, Path 1-A corresponds to the attack of 2a on II-1 from the indane side. Each of Paths A and B can also have two pathways that differ in the orientation of substrate 2a.

[4+2] annulation pathways without urea for intermediates II-1 or II-2
Our DFT calculations on the reactions of II-1 and II-2 with 2a without urea showed that this reaction consists of two steps: (1) a ring-annulation step and (2) an NHC dissociation step. It should be noted that Path 1-A proceeds in a concerted fashion and the optimized structure of TS2 1-A is similar to those of several other TS2s for the NHC dissociation step. After the formation of the product (the minor enantiomeric product), the NHC catalyst is recovered and may abstract a proton from H + -DABCO (protonated base) to regenerate the triazolium salt precursor. The regeneration of the triazolium salt precursor is exothermic by 8.2 kcal/mol (Fig. S62). *Geometry optimization was performed at the B3LYP/6-311G(d) level.

Supplementary
Supplementary Figure 62. Regeneration of the triazolium salt precursor from the NHC catalyst.
[4+2] annulation pathways without urea for intermediates II-3 or II- 4 We also examined the reactions of conformers II-3 and II-4 with 2a, and in these conformers, the phenolate oxygen points away from the NHC group. The reaction on Path 3-A can take place via a 6-membered ring intermediate (Int1 3-A ), which should be structurally equivalent to Int1 2-B , as illustrated in Fig. S63. In a similar fashion, Paths 3-B, 4-A, and 4-B also involve 6-membered ring intermediates, which should be structurally equivalent to Int1 2-A , Int1 1-B , and Int1 1-A , respectively. However, as shown in Tables S9 and S10, Int1's on Paths 3-A, 3-B, 4-A, and 4-B are not as stable as Int1's on Paths 2-B, 2-A, 1-B, and 1-A, respectively. It should be noted that these energy differences are small, which therefore suggests that Int1's on Paths 3-A, 3-B, 4-A, and 4-B may be readily converted into the corresponding Int1's (Int1's in Tables S9 and S10) before the NHC dissociation process. Thus, we assumed that TS2's and PC's on

Reaction mechanisms with urea co-catalyst
Our DFT calculations without urea showed that step 2 is more important than step 1. We calculated the reaction mechanism in Path 1-A with thiourea A1 and obtained three transition states, namely A1-TS1 1-A , A1-TS1' 1-A , and A1-TS2 1-A , as shown in Fig. S64. As the reaction without urea, the NHC dissociation step has the highest energy. Overall, the reaction looks endothermic, but after the formation of PC, NHC can be further stabilized by forming a triazolium salt precursor (Fig. S62). Note that A1 is attached on the label when A1 is added in the system. Thiourea A1 interacts with ketone through a hydrogen bond that should make ketone more reactive, and therefore the activation energy for the ring-annulation step should be lowered. In the calculations with A1, we examined several binding modes of A1 to find the most stable TS1 1-A and S69 TS2 1-A . We obtained one binding mode for A1-TS1 1-A (Fig. S64) and two binding mode for A1-TS2 1-A (Fig. S65). A1-TS2 1-A , in which A1 forms hydrogen bonds with the carbonyl oxygen of o-QM, is more stable than A1-TS2 1-A' , in which A1 forms hydrogen bonds with the CF 3 group of ketone. It should be noted that A3-TS2's, which are involved in the reaction with urea A3, should be the rate-determining step because the urea moiety of both A1 and A3 should interact with the same atom(s) through hydrogen bond(s) in each key species, and therefore A1-TS1's and A3-TS1's should be stabilized by the same hydrogen bond(s).

Energy differences of A3-TS2's
We further investigated each A3-TS2's to identify the key factors that control the enantioselectivity (Table S12). A3-TS2 1-A , which is the lowest-energy in A3-TS2, was found to have attractive π-π stacking between the pentafluorophenyl group of urea A3 and the indane moiety of the catalyst (Fig. 2c). Such π-π stacking was also observed in A3-TS2 1-A-minor , but was not found in other A3-TS2's. Therefore, A3-TS2 1-A and A3-TS2 1-A-minor are more stable than the other A3-TS2's. Note that we examined several binding modes between TS2 and urea A3 in each pathway to obtain the lowest-energy A3-TS2 (Table S13).