Bifunctional Chiral ImPy‐Carbene Ligands. H‐Bonding Controlled Reactivity and Enantioselectivity in Au(I)‐Catalysis

Abstract

Bifunctional chiral N‐heterocyclic carbene ligands have been devised for enantioselective gold(I) catalysis. Based on a single C‐stereogenic center derived from chiral pool α‐aminoacids and connecting an imidazopyridine core to an arylurea motif, enantioselective gold(I)‐catalyzed cycloisomerization reactions could be achieved. High enantioselectivities were notably observed for substrates presenting a pendant propargyl alcohol on 2‐naphthol and 1,6‐enyne scaffolds. In the latter case, the catalyst shows a high degree of selectivity for the unprecedented 6‐endo‐dig biscyclization of these substrates to give cyclopropyl‐fused 6‐ring heterocycles with a free OH functionality instead of the previously reported furan‐fused products. This unusual selectivity was investigated by DFT studies, which suggested the dual role of the carbonyl group of the urea moiety: first as an H‐bond acceptor in the catalytic cycle to direct the enantioselectivity and second as a cooperative group in the hydrogen shift leading to deauration.

A new class of catalytically highly efficient bifunctional imidazopyridine (ImPy)‐based gold(I) complexes was synthesized. The ligands are based on tetrahedral C‐stereogenic centers derived from straightforward chiral pool synthesis. Unprecedented cycloisomerization products of 1,6‐enynols were observed in excellent enantioselectivity up to 99:1 er. DFT calculations were utilized for intricate mechanistic investigations.

1. Introduction

Over the last two decades, homogeneous gold(I) catalysis has occupied a prominent role in the selective π‐activation of various unsaturated hydrocarbons such as alkynes, allenes, and alkenes, allowing to reach otherwise difficultly accessible molecular complexity under mild conditions.^{[ 1 ]} An ongoing challenge is enantioselective gold(I) catalysis due to its linear coordination mode, which positions the chiral ligand opposite to the substrate and thus promotes the outer‐sphere attack of a nucleophile, in addition to a low rotational barrier around the gold(I)‐unsaturated substrate bond.^{[ 2 ]} The role of the chiral ligand is therefore crucial since it should create a chiral environment around the metal center and limit the degrees of freedom of the system to induce enantioselectivity. Nevertheless, gold(I) complexes formed by coordination of some mono‐^{[ 3 ]} or bis‐aryl phosphines^{[ 4 ]} with axial chirality have made it possible to achieve high enantioselectivities.^{[ 2 ]} However, many phosphine ligands are difficult to prepare and prone to oxidation.

In this context, C‐based ligands such as N‐heterocyclic carbenes (NHCs)^{[ 5 ]} have become valuable alternatives, notably due to their strong σ‐donor capabilities, allowing the formation of very stable and conformationally rigid complexes. Those have been key factors in their development for a number of applications in the field of catalysis, particularly with gold(I).^{[ 6 ]} Compared to phosphorus‐based ligands, it is relatively easy to modify the NHC structure and consequently the steric environment around the metal center, giving access to a wide variety of chiral gold(I) complexes.^{[ 7 ]} Initially, stereogenic centers were placed on the imidazolium skeleton or on the substituents attached to the nitrogen atom. This approach, however, has proved problematic, as the source of chirality is too far from the gold center in these systems to create a catalytically efficient steric environment. The imidazo[1,5‐a]pyridin‐3‐ylidene (ImPy) carbene, introduced almost simultaneously by the Lassaletta^{[ 8 ]} and Glorius groups,^{[ 9 ]} circumvented this problem by placing both steric and electronic levers in close proximity to the metal center by functionalizing the C5‐position of the ImPy core (Figure 1a). In particular, the introduction of a bulky moiety at this position can create a “lateral wall” for facial selectivity.^{[ 6a ]} The key ImPy structure has since served as a platform for valuable gold complexes.^{[ 10 ]}

Figure 1.

a) C5‐aryl fused ImPy ligands. b) Bifunctional gold(I) catalysts based on ImPy NHC ligands for enantioselective catalysis. c) Chiral indolizy carbene ligand leading to enantioselectivity based on anion and phosphine oxide cooperativity. d) Intermolecular self‐activation of gold(I) complexes. Bottom: Bifunctional ImPy‐based Au(I) chiral catalyst interacting with substrate.

By analogy to functionalized biaryl‐2‐ylphosphines, which bear a Lewis base moiety at the bottom half of the pendant aryl ring allowing secondary interaction with substrate or nucleophile,^{[ 11 ]} Zhang and coworkers designed the first ImPy‐based axially chiral gold(I) complexes by introducing a tetrahydroisoquinoline moiety at the C5‐position (Figure 1b).^{[ 12 ]} The chirality transfer provided by these ligands was due to both “chiral steric hindrance and asymmetric electrostatic attraction”. These complexes have been shown to be effective in a variety of enantioselective catalysis involving alkyne and allene activation reactions. In the same vein, our group recently reported a new family of chiral bifunctional ligands with a phosphine oxide moiety attached to the stereogenic center located on an indolizy core (Figure 1c).^{[ 13 ]} These new gold(I)‐carbene ligands have been shown to be effective in the asymmetric heterocyclization of γ‐allenols. The cooperative effects of the phosphine oxide moiety and of a tosylate counteranion on enantioselectivity have been demonstrated experimentally and supported by DFT calculations.

Based on the concept of self‐activation evidenced by Helaja and coworkers on NHC‐Au(I)Cl complexes flanked with an amide arm,^{[ 14 ]} the group of Echavarren extended this approach to urea‐ or squaramide‐functionalized phosphine ligands^{[ 15 ]} while Lassaletta and coworkers used monosulfonyl squaramide to activate a broad variety of gold(I) chloride complexes by H‐bond donation (HBD) in an intermolecular fashion.^{[ 16 ]} Additionally, Gorai and Teichert recently reported the synthesis of chiral bifunctional NHC/guanidine ligands bearing two stereogenic centers in the linker unit of chiral copper(I) complexes. The guanidine moiety acts as the HBD unit, giving encouraging results in the copper‐catalyzed asymmetric hydrogenation of acetophenone.^{[ 17 ]}

In this context and in our ongoing interest in the design of new chiral ligands for gold complexes,^{[ 13} , ¹⁴ , ^{18 ]} we aimed for a new strategy of bifunctional carbene ligands. Our design was first based on the idea of blocking one side of a chiral pocket created around the metal center. Therefore, we selected an ImPy scaffold incorporating a hindered mesityl group at the C5 position acting as the previously mentioned “lateral wall”.^{[ 6a ]} Second, the chiral pocket environment would rely on a side arm including a stereogenic center stemming from the chiral pool approach and a urea moiety to introduce rigidity into the catalytic system by HBD. Interestingly, related (thio)urea imidazoliums could be used in the organocatalyzed enantioselective cyclopentannulation of enals and enones.^{[ 19 ]}

2. Results and Discussion

2.1. Synthesis of Gold(I) NHC Complexes Au1‐6

To probe this strategy, we selected two families of chiral gold(I) complexes based on NHC and ImPy ligands (see Scheme 1 and details in Supporting Information). Their synthesis relied on the use of β‐amino alcohols from the chiral pool, such as L‐leucinol and L‐phenylalanilol (A) in order to append a chiral side arm linked to the NHC‐nitrogen atom. We started by the synthesis of imidazolium B following the procedure of Vo‐Thanh and Bournaud from L‐leucinol.^{[ 19 ]} A urea moiety was then attached by reaction of the amine group of B with phenylisocyanate, followed by counterion exchange to chloride to yield imidazolium D. Finally, gold(I) complex Au1 was obtained in a one‐pot fashion by treatment of D with Ag₂O followed by transmetalation of the resulting silver complex with Me₂SAuCl in 13.5% over 7 steps. The desired ImPy‐based gold(I) complexes Au2‐6 were obtained through a sequence of nine high‐yielding linear steps, two of which could be performed in a one‐pot fashion without the need for product purification (F1‐2 → G1‐2 and J1‐5 → Au2‐6 complexes). Starting from L‐leucinol or L‐phenylalanilol, aldehyde E,^{[ 20 ]} and formaldehyde in the presence of hydrochloric acid, the corresponding imidazopyridinium scaffolds were obtained and further isolated as hexafluorophosphate salts F1‐2 (84% and 70% yields, respectively), to increase their solubility. The imidazopyridinium site of these compounds showed exceptional stability during the side arm functional layout, opening a path toward a large library of ligands. Thus, alcohols F1‐2 were first mesylated and then underwent nucleophilic substitution with sodium azide to give G1‐2 in 75% and 81% isolated yields, respectively. Catalytic hydrogenation of the azido moiety over Pd/C afforded the corresponding amine derivatives H1‐2. As before, the latter were reacted with arylisocyanates, resulting in the five new enantiopure imidazolium salts I1‐5 bearing urea groups, which were obtained in 71–99% isolated yields. The hexafluorophosphate salts of imidazopyridinium I1‐5 were converted to their corresponding chloride salts, prior to metalation, in excellent yields for all derivatives J1‐5. This strategy was followed for two main reasons: on the one hand, purification by chromatography of the chloride salts was easier; on the other hand, complete removal of PF₆ ⁻ would prevent mixed counterion systems. As shown in the literature, counterions can have substantial effects on gold‐catalyzed reactions.^{[ 21 ]} Concluding the complex synthesis, Au2‐6 were prepared in good yields of 71–81% via sequential metalation/transmetalation, resulting in bench‐stable solids after purification by flash chromatography.

Scheme 1.

Preparations of complexes Au1‐6.

2.2. Screening of Benchmark Reactions

With the six gold complexes Au1‐6 in hand (Scheme 1), we investigated the performance of complexes Au2, Au5, and Au6 in the intermolecular [2 + 4] cycloaddition reaction between allene 1 and diene 2 (Scheme 2a).^{[ 22 ]} We observed good activity for all precatalysts in the presence of AgSbF₆ and obtained the expected cyclohexene adduct 3 in good yields averaging 80%, but with low enantioselectivities. Interestingly, a reversal of enantioselectivity was observed upon switching from the isobutyl group located at the stereocenter (Au2) to benzyl (Au5/6). This phenomenon could be suggestive of π‐stacking interactions taking place between the substrates and Au5/6.

Scheme 2.

a) Intermolecular [4 + 2] cycloaddition of allene 1 with diene 2.^{[ 22 ]} b) Intramolecular cyclization of allenol 4.

The cyclization of allenols 4 to tetrahydrofuran derivatives 5 was assessed (Scheme 2b). A slow product formation was observed at RT without the addition of a silver salt in the case of the reaction of 4a with Au1. Upon addition of a silver salt, the reaction was accelerated significantly, giving yields of 85% with Au1 and 89% with Au2. In each case, however, a racemic mixture of product 5a was observed. The cycloisomerization of 5b, on the other hand, showed an er of 64:36 at RT with Au2. This modest selectivity could be slightly increased to 67:33 upon cooling to −20 °C, with a yield of 92% after 2.5 hours of reaction time. These findings nevertheless showed a slight improvement of the selectivity compared to previous reactions. We theorized that the presence of a free alcohol group on the substrate might provide anchoring to the catalytic system and favor enantiodiscrimination.

For this reason, we then turned our attention to the catalytic dearomatization of 2‐naphthol derivatives 6, which has been studied recently by Silva Lopez, Bandini, and coworkers.^{[ 23 ]} The substrates 6 exhibit an alkyne‐tethered terminal hydroxy group, the structural feature of our interest. In addition, the dearomatization products 7 consist of naphthalenone cores, which are present in various natural compounds (Table 1).^{[ 24 ]}

Table 1.

Dearomatization of 2‐naphthol 6a.

Entry	AgX	Cata	t [h][ a ]	T [°C]	Yield [%]	Er[ b ]
1	AgSbF6	Au1	21h	−20	62	50:50
2	AgSbF6	Au2	16	−40	67	86:14
3	AgOTs	Au2	19 + 20	−40	74	76:24
4	AgNTf2	Au2	19 + 1	−40	84	87:13
5[ c ]	AgSbF6	Au2	19 + 20	−40	75	72:28
6[ d ]	AgSbF6	Au2	19 + 1	−40	76	83:17
7	AgSbF6	Au2	20 + 1	−20	84	76:24
8	AgSbF6	Au3	20 + 1	−20	79	60:40
9	AgSbF6	Au4	20 + 1	−20	85	68:32
10	AgSbF6	Au5	20 + 1	−20	86	52:48
11	AgSbF6	Au6	20 + 1	−20	81	68:32

^[a] x hours at T + y hours at RT.

^[b] Determined via HPLC with a chiral stationary phase.

^[c] Reaction in THF.

^[d] Reaction in DCM.

With Au1 we obtained only a racemic mixture of 7a (Table 1, entry 1), so we shifted to ImPy complexes Au2‐6. Preliminary reaction with Au2 at −40 °C over 16 hours gave 7a in 67% yield but with an encouraging er of 86:14 (Table 1, entry 2), which is higher than the best value found in the literature (er = 82:18 with (R)‐xylyl‐BINAP(AuCl)₂ 2.5 mol%, AgNTf₂ 5 mol%).^{[ 23 ]} The reaction was reported to transit through two distinct steps, which could be monitored by TLC. The first and stereodetermining step consists of the dearomative rearrangement to an allenol, creating a stereogenic center at C2. The second step is the 5‐endo‐trig cyclization of the allenol to the ring, which was shown to require higher temperature; the reactions were therefore warmed to RT after completion of the first step. It was observed that 1 hour of reaction time at RT was generally sufficient to finalize the reaction, while extended times (entries 3 and 5) did not result in noticeable differences in yields. Regarding enantioselectivity, toluene as a solvent (entry 2, er = 86:14) surpassed both THF (entry 6, er = 72:28) and CH₂Cl₂ (entry 6, er = 83:17). Both silver salts AgSbF₆ (entry 1, er = 86:14) and AgNTf₂ (entry 4, er = 87:13) showed similar enantioselectivities, while AgOTs resulted in a lower 76:24 er (entry 3).

Raising the temperature to −20 °C for the first step had a substantial effect on the reaction (84% yield, entry 7) and resulted in a decrease of the er to 76:24. Finally, through comparing ImPy complexes Au2‐6 (entries 6–10), Au2 was shown to be the optimal catalyst, as modification of the steric demand on the side arm from isobutyl to phenyl (Au4) decreased the er (entry 9). Interestingly, the introduction of the strongly electron‐withdrawing 3,5‐bis‐(trifluoromethyl)phenyl group on the Ar‐urea moiety (as in Au3 and Au5) proved to be highly detrimental. Notably, Au5 (entry 10) resulted in a nearly racemic mixture of 7a. Interestingly, the presence of a para‐methoxy group (Au6) did not prove to be rewarding (entry 11).

We then installed different R substituents at the C6 position of substrates 6 to probe their electronic effects on the reaction with Au2 (Scheme 3). We observed that, while products 7a and electron‐poorer 7d could be isolated both at −20 °C and −40 °C, the formation of electron‐richer products 7b and 7c was only observed at −20 °C. Product 7b was isolated in 92% yield and with an er of 77:23, which is comparable to 7a, while methoxy derivative 7c showed a decreased enantioselectivity (62:38 er). The selectivity shown by 7d, on the other hand, was similar to 7a at both temperatures (79:21 at −20 °C, 87:13 at −40 °C). With those encouraging results that established alkynols as attractive targets for these catalytic systems, we aimed at a larger scope of related compounds.

Scheme 3.

Substrate scope of 2‐naphthol cycloisomerization.

2.3. 6‐endo‐dig Biscyclization of Enynols

In this context, we turned our attention to the Au(I)‐catalyzed 6‐endo‐dig biscyclization of heteroatom‐tethered (O or N) 1,6‐enynols 8 with a pendant hydroxy group for the selective formation of hexahydrofuro[3,4‐c]pyridines 9, as reported by the Maestri group.^{[ 25 ]} This reaction was initially worked out with a phosphite‐AuCl/AgSbF₆ catalytic system and features a 6‐endo cycloisomerization‐cyclopropyl ring opening to give polyheterocyclic product 9a in good yields and high diastereoselectivities (see Scheme 4 for an example).

Scheme 4.

Diastereoselective cycloisomerization of 1,6‐enynol 8a²⁵ .

As no enantioselective version of this reaction has, to the best of our knowledge, been reported, we investigated whether the developed gold(I) complexes were able to deliver products 9 with good enantioselectivity. N‐tethered enynol 8a was used as a model substrate. Capitalizing on previous findings, the 2 mol% Au2/AgSbF₆ catalytic system in toluene at RT was selected (Table 2). In sharp contrast to Maestri's results, we obtained the interrupted 6‐endo‐cycloisomerization product 10a, corresponding to an azabicyclo[4.1.0]heptane skeleton, in 89% yield, evidencing that no intramolecular O‐nucleophilic attack leading to 9a occurred.

Table 2.

6‐endo‐dig cyclization 1,6‐enynols 8a with catalysts Au2.

Entry	AgX	Solv[ b ]	t [h]	T [°C]	Yield [%][ c ]	er[ d ]
1	AgSbF6	Tol	48	RT	89	97:3
2	AgSbF6	Tol	22	0	82	99:1
3	AgSbF6	Tol	48	−20	92	99:1
4	AgSbF6	Tol	4	40	93	96:4
5	AgSbF6	CHCl3	22	0	97	95:5
6	AgSbF6	THF	22	0	73	83:17
7		HFIP[ e ]	22	rt	39[ f ]	54:46[ g ]
8		Tol	22	0	94	85:15
9		Tol	22	0	96	98:2
10		Tol	22	40	0	−
11		Tol	22	0	3g	−

^[a] CCDC number 2373846.

^[b] Degassed.

^[c] Isolated.

^[d] Determined via HPLC with a chiral stationary phase.

^[e] HFIP = hexafluoroisopropanol.

^[f] NMR yield.

^[g] 44 % of 8a in the mixture

To our delight, (1S,6R,7R)‐10a was obtained with an er of 97:3 (Table 2, entry 1), whose absolute configuration could be determined via single‐crystal X‐ray diffraction analysis. The reaction showed complete diastereoselectivity according to NMR as well as HPLC analysis. Lowering the temperature to 0 °C or −20 °C further increased the er to 99:1 (entries 2 and 3). Increasing the temperature to 40 °C, on the other hand, sped up the reaction rate while the er remained comparable to the one obtained at RT (entry 4). At this stage, the temperature was set at 0 °C for an optimal combination of reaction kinetics and selectivity. Notably, the enantioselectivity was reduced in polar solvents like THF while toluene was found to be optimal for this reaction, both in terms of yield and selectivity (entry 5 vs. 6). We, therefore, assumed hydrogen bonding between the substrate and the catalyst's side arm to be crucial for the observed excellent enantioselectivity, which would be disrupted in polar solvents. In the same logic, the protic HFIP solvent system in the absence of silver salt led to a dramatic decrease in yield (39%) and enantioselectivity (54:46, entry 6). HFIP is known to activate gold chloride precatalysts via strong hydrogen bonding between the HFIP hydroxyl group and metal‐bound Cl.^{[ 26 ]} In this reaction medium, saturation of the urea's coordination sites by the solvent would no longer allow efficient substrate coordination, leading to a loss of enantioselectivity.

With AgOTs as chloride scavenger, we observed a substantial drop of er (85:15) which, similar to HFIP, may be due to an interaction between the alcohol function of the substrate and the counterion (entry 8). This would be in accordance with previous studies by our group, which showed that the tosylate counterion is able to alter the conformation of transition states.^{[ 13 ]} Good enantioinduction was also observed with AgNTf₂ (98:2) while self‐activation in the absence of silver salt was not observed (entry 10). The effect of water on the reaction was investigated by following its progress via NMR in dry (8 ppm) and wet (100 ppm) toluene (see Supporting Information, Figure S2 for details). While beneficial effects of water were reported in previous studies,^{[ 27 ]} the reactions were very similar in our case; drying or increasing the water content was not necessary with our catalyst system.

With the best reaction conditions in hand, we evaluated Au1‐6 gold(I) catalysts. With Au1, we obtained the expected derivative with an er of 15:85. Comparison of the gold(I) catalysts Au2‐6 was once again insightful, and the same major trends as in the previous study could be observed (Table 3). Exchanging the isobutyl stereocenter present in Au2 to benzyl (Au4), both featuring a phenylurea group, resulted in a decrease in enantioselectivity from 99:1 to 93:7 (Table 3, entries 2 and 4). The second, more spectacular trend was due to the electronic effect of the aryl substituents on urea. With a 3,5‐bis‐(trifluoromethyl)phenyl group (Au3), the er decreased to 70:30 (entry 3) and even became close to 50:50 with Au5 (entry 5). In addition, in the case of Au3, the yield of the reaction decreased significantly to 12%. Complex Au6 bearing an electron‐donating 4‐MeO phenyl substituent (entry 6) also delivered a low yield of 46% and reduced selectivity (er = 83:17). With these optimization conditions in hand (Au2, AgSbF₆, toluene), we examined the scope and limitations of this new cycloisomerization process.

Table 3.

6‐endo‐dig cyclization of 1,6‐enynes 8a with catalysts Au2‐6.

Entry	Catalyst	Yield [%][ b ]	Er[ c ]
1	Au1	56	15:85
2	Au2	82	1:99
3	Au3	12	30:70
4	Au4	97	7:93
5	Au5	94	44:56
6	Au6	46	17:83

^[a] Degassed.

^[b] Isolated.

^[c] Determined via HPLC with a chiral stationary phase.

The substrate scope of the biscyclization is displayed in Scheme 5. We therein varied the substitution patterns on the alkene and the alkyne as well as the nature of the tethering group (X = p‐TsN, MsN, and O).

Scheme 5.

Substrate scope of 1,6‐enynols for biscyclization with various substitution patterns.

With substrate 8b, bearing an electron‐donating 4‐OMe‐phenyl group on the alkene in combination with a p‐NTs‐linker, no reactivity was observed at 0 °C. While at RT, surprisingly only 9b was obtained in a yield of 59%, with a very good er of 94:6 and complete diastereoselectivity. It should be noted that the formation of 9a was not observed in the previous optimizations with 8a and that the 4‐CF₃‐Ph substrate was not reactive under otherwise identical reaction conditions (see Supporting Information).

Changing the terminal aryl group to an alkene on 8c led to a mixture of both derivatives: 10c was obtained in 60% yield and an excellent er of 98:2, while the minor product 9c (30% yield) displayed a lower er = 86:14. The introduction of a methyl group in α‐position to the alkyne moiety in 8d led to only one diastereomer of 10d in severely reduced yield (46%) and a nearly racemic mixture of enantiomers. In contrast, the MsN‐tethered derivative 8e gave the same single product 10e as 8a, however in a lower yield of 64% and with a notable decrease in the er to 80:20. The same reactivity of 8a was observed for the ether derivative 8f. In that case, only product 10f was obtained with a slightly decreased er of 94:6 compared to 10a. In the case of the O‐tethered substrate 8 g, the introduction of a 4‐CF₃‐Ph group on the alkene moiety resulted exclusively in product 10 g, that being in a very good yield of 87% and er of 89:11, in sharp contrast to its unreactive p‐TsN‐tethered analogue. The observation with 8b, in which the presence of an electron‐donating group allowed the formation of a type 9 derivative, was confirmed with 8 h; a comparable mixture of 9 hours (33%, er = 80:20) and 10 hours (57%, 69:31) was obtained. However, as observed for all the O‐tethered derivatives, 9 hours was obtained with a decreased enantioselectivity compared to the p‐TsN‐analogue 9b.

Low temperature ¹H‐NMR experiments were performed with 8f to gain kinetic insight. It was observed that at −20 °C, a clean turnover from starting material to product takes place without notable side product formation or deactivation of the catalyst (for details, see Supporting Information, Figures S3‐S5). The catalysis was found to be rather sensitive toward substrates with different electronic or geometric properties – no consumption of starting material was observed with substrates bearing a prenyl moiety or amine as nucleophile (see Supporting Information). It is also important to note that all these reactions were fully diastereoselective; that is, only one diastereoisomer of derivatives 9 and 10 was obtained.

Due to the highly specific reactivity of our complexes, all racemic alcohols were prepared using an achiral Au7 complex flanked by a urea arm lacking a stereogenic center or racemic Au4 (see Supporting Information).

2.4. Mechanistic Investigations

To gain theoretical insight into the reaction mechanism of the Au2‐catalyzed 6‐endo‐dig biscyclization of 1,6‐enynes, we performed DFT calculations (TPSS‐D3/def2‐TZVP//TPSS‐D3/def2‐SVP, CPCM = Toluene, see details in Supporting Information) and probed the energetics of the reaction pathway (Scheme 6). N‐Methanesulfonyl compound 8e was chosen as the model substrate in combination with cationic complex [Au2]⁺. Due to the conformational flexibility arising from several freely rotating bonds in catalyst [Au2]⁺ and substrate 8e, their conformational space was first sampled with the CREST tool, part of the AQME software (see Supporting Information section “Computational Details”).^{[ 28 ]}

Scheme 6.

a) DFT energy profile for the biscyclization of 10e catalyzed by cationic Au2. b) Proposed catalytic cycle.

The computation of viable reaction pathways was initiated using energy minimum geometries evidenced through conformational screening. Notably, structure I exhibited the hydrogen donor (O‐H and N‐H) and acceptor sites (C─O and S═O) of bonding interactions (Scheme 6b). The search of the first transition state (TS‐I) revealed that the hydroxyl group remains H‐bonded to the urea carbonyl during the C─C bond formation in the case of Si but not during the corresponding Re face attack. The former attack is favored with a ca. 3 kcal/mol lower free energy barrier than the latter one, which is in good agreement with the observed dominant formation of 1S,6R,7R‐10e (Scheme 6). The energy advantage of TS‐Ia is obviously related to the abundant noncovalent attractive interactions in the structure, visualized in Figure 2a, which include dispersion interactions connected to the tight folding in addition to the H‐bonding interactions.

Figure 2.

Plots of noncovalent interactions (NCI) for TS‐Ia (a) and TSII‐CO‐mediated H‐shift (b) showing strong attractive interactions in blue (e.g., H‐bonding) and weaker attractive interactions in green circles (details in Supporting Information).

To rationalize the formation of a fused cyclopropyl ring instead of the previously reported furan ring,^{[ 25 ]} the benzylic carbocation intermediate (structure II) observed after the first bond‐forming step appears to be a critical intermediate. The hydroxyl group is kept remote from the cationic site by H‐bonding with the urea carbonyl, which prevents it from interfering with the cyclopropyl ring formation. The catalytic cycle is closed by hydride shift from the α‐position to the amino group to the Au(I)‐bonded carbon, which releases the cationic gold species from the cycle (deauration step) in a favorably low energy barrier of 4 kcal/mol.

Interestingly, without coordination of an electron pair of H₂O or the urea carbonyl, this hydrogen shift exhibits the highest energy barrier in the reaction pathway (Scheme 6a: TS‐II, 25 kcal/mol). Yet, when the migrating hydrogen (TS‐II) is interacting with the oxygen atom of a water molecule, the barrier is lowered below the energy required for the Si face attack (TS‐Ia, 15.1 kcal/mol; see Supporting Information, Figure S6). This type of water‐mediated proton shuttling mechanism has been rationalized by the groups of Guo and Xia in a DFT study^{[ 29 ]} for an AuCl₃‐catalyzed bromofuran synthesis that was reported earlier by the Gevorgyan group.^{[ 30 ]} Conversely, we discovered that in our case, the urea carbonyl coordination with the migrating hydrogen atom offers an alternative, effective shuttling mechanism with a low energy barrier (TS‐II_{C = O} , 12 kcal/mol) being lower than the one mediated by H₂O (14 kcal/mol, see Supporting Information). Notably, the TS structure exhibits two additional H‐bonding interactions between substrate and urea: one between the substrate's hydroxyl group and the urea oxygen and another between the substrate's sulfonyl oxygen and the urea NH. The strong H‐bonding interactions are visualized with blue circles in Figure 2. The carbonyl‐assisted proton migration adequately matches the experimental observation that a small amount of water in toluene slows down the rate of catalysis to some extent (see Supporting Information, Figures S2, S6, TS‐II + H₂O).

To further evaluate the role of urea but also the nature of the ligand connected to the gold center, NHC vs. phosphite, we examined the reaction of 8a with IPrAuCl. In this case, we observed the formation of both derivatives 9a and 10a in 42% and 50%, respectively, yield (Scheme 7).

Scheme 7.

Catalytic process with IPrAuCl.

This first confirms the importance of the interaction between urea and alcohol on this cycloisomerization outcome since substantial amount of 9a was formed. Second, it gives insight in the role of electronic properties of the Au(I) ligand. Indeed, with NHC ligands, as well documented in the literature, carbene III intermediate is the most favored, leading to 10. With phosphite as ligand, as used the Maestri group, carbocationic intermediate IV predominates, favoring nucleophilic attack of the alcohol and formation of 9.^{[ 31 ]} So the use of IPr gives a result corresponding to a compromise between a relatively free nucleophilic alcohol but a less electrophilic carbene center, while the ImPy ligand displays relatively similar σ‐donicity but enhanced π‐acidity compared to IPr.^{[ 6} , ^{32 ]}

3. Conclusion

We herein describe the synthesis and characterization of a new family of chiral urea‐tethered gold(I) NHC complexes based on imidazole and ImPy core structures. The central chirality is based on a single stereogenic center stemming from the chiral pool. It is easily appended using simple commercially available β‐amino alcohols, giving easy access to a large family of molecules from affordable starting materials.

The corresponding gold(I) complexes were first tested in several benchmark reactions, which drove us to examine the more challenging dearomatization reaction of 2‐naphthol 6a. This proved rewarding since we obtained the best er value found in the literature to date. In addition, these new catalysts showed exceptional activity and selectivity in the biscyclization of 1,6‐enynols, yielding unprecedented products by retaining a free hydroxyl group in the substrates. The reactions delivered excellent yields and enantioselectivities through novel mechanistic pathways which were supported by DFT calculations. Besides high enantioselectivity, the catalysts showed complete diastereoselectivity for a variety of substrates.

We hereby showed the first example of a high‐performing ImPy catalyst featuring a single stereogenic center without additional axial chirality. Building on the remarkable H‐bonding capabilities of a urea group to assist enantioinduction and demetallization steps, the generalization of this type of strong substrate‐catalyst interaction opens the way to new catalytic methodologies.

Supporting Information

Additional refrences are cited in supporting file.^{[ 33} , ³⁴ , ³⁵ , ³⁶ , ³⁷ , ³⁸ , ³⁹ , ⁴⁰ , ⁴¹ , ⁴² , ⁴³ , ⁴⁴ , ⁴⁵ , ⁴⁶ , ⁴⁷ , ⁴⁸ , ⁴⁹ , ⁵⁰ , ⁵¹ , ⁵² , ⁵³ , ⁵⁴ , ⁵⁵ , ⁵⁶ , ⁵⁷ , ⁵⁸ , ⁵⁹ , ⁶⁰ , ⁶¹ , ⁶² , ⁶³ , ⁶⁴ , ⁶⁵ , ⁶⁶ , ⁶⁷ , ⁶⁸ , ⁶⁹ , ⁷⁰ , ^{71 ]}

Conflict of Interests

The authors declare no conflict of interest.

Supporting information

Supporting Information

Data Availability Statement

The data that support the findings of this study are available in the supplementary material of this article.