Spirosilanes Activate Gold(I)‐Catalysts in Cycloisomerization and Intermolecular Reactions

Abstract

Gold‐catalyzed reactions have established as a powerful tool in organic synthesis, offering efficient pathways to construct diverse molecular structures and notably scaffolds with high complexity. Most processes rely on the use of LAuCl precatalysts, which are generally activated, i.e., cationized, by silver salts. In this work, we explore the role of silicon‐based Lewis acids based, such as spirosilane, derivatives as alternative activators in lieu of silver salts. It was found that Martin spirosilanes mediate the activation of gold precatalysts, facilitating diverse cyclization and cycloisomerization reactions as well as intermolecular reactions. NMR studies and computational investigations suggest that no real cationization, i.e., formation of an ionic pair, takes place, but rather activation through weak interaction between silicon and the chlorine atom of LAuCl. This preserves the Au─Cl bond and ultimately enables the LAuCl complex to be recovered. On the same line, the LAuCl–silane interaction has also proven to be beneficial for asymmetric catalysis. This work significantly contributes to the expansion of gold‐catalyzed transformations by opening up the prospect of a more sustainable gold catalysis. It also opens perspectives on the use of silicon‐based Lewis acids as versatile cooperative agents in organometallic chemistry.

Spirosilanes activate LAuCl complexes to promote catalysis. This silver‐free activation through weak interactions is nondestructive so that low loadings of LAuCl are possible as well as their recycling. It can also be beneficial for asymmetric catalysis.

Introduction

It is indisputable that π‐activation of C─C double and triple bonds by noble metal catalysis^{[ 1 ]} and cationic gold(I) catalysis in particular,^{[ 2} , ³ , ⁴ , ⁵ , ⁶ , ^{7 ]} has brought a myriad of useful transformations whose synthetic applications range from the synthesis of complex natural products^{[ 8 ]} to asymmetric catalysis^{[ 9 ]} and materials science.^{[ 10} , ^{11 ]} More recently, gold(I) catalysis has extended its range of action by enabling different types of cross‐coupling reactions through the use of hemilabile ligands that allow Au(I)/Au(III) catalytic cycles.^{[ 12} , ^{13 ]} A significant portion of all these reactions rely on the use of gold(I) halide precatalysts and more particularly gold(I) chloride complexes whose strong Au─Cl bond needs to be activated in order to generate in situ a catalytically active species.^{[ 14} , ¹⁵ , ^{16 ]} While polar protic solvents^{[ 15} , ^{16 ]} and notably HFIP^{[ 17} , ^{18 ]} could be used for that purpose, the activation of the Au─Cl bond has been mainly accomplished via chloride scavenging (cationization) to yield [LAu⁺][Y⁻] active species, Y⁻ being a weakly coordinating anion (WCA).^{[ 19 ]} These species are generally generated by anion metathesis between the LAuCl precatalyst and a silver salt AgY,^{[ 20 ]} producing concomitantly AgCl as by‐product. In addition to the hassle of using hygroscopic, light‐sensitive silver salts, this activation step is under equilibrium,^{[ 16} , ^{21 ]} and therefore rarely complete resulting in various metallic species featuring π‐Lewis acidity with potentially competitive catalytic activities that can alter reaction outcomes.^{[ 21} , ²² , ^{23 ]} This “silver effect” is well‐documented as is the influence of the counterion Y⁻ on reaction outcomes.^{[ 24} , ²⁵ , ^{26 ]} So there has been an increasingly clear trend to use silver‐free Au(I) catalysis.^{[ 15} , ^{16 ]} Other salts (Ga(III), Cu(II), among others) have been used,^{[ 27 ]} however the issue remains as to understanding the role of each metallic component. A priori catalytically inactive, alkali metals such as the expensive NaBArF₄ have been used, but they have not shown generality. Preformed cationic complexes are also an alternative but their cost also as well as their non‐trivial preparation in pure‐form (silver free) are often prohibitive. Another option is to use self‐activating Au(I) chloride complexes bearing finely tailored ligands.^{[ 28} , ²⁹ , ^{30 ]} The latter can incorporate WCA,^{[ 31} , ³² , ^{33 ]} hydrogen bond donors based on amide,^{[ 34} , ^{35 ]} urea,^{[ 36 ]} or phosphoric acid,^{[ 37} , ³⁸ , ^{39 ]} and Lewis acid moieties^{[ 40} , ⁴¹ , ⁴² , ⁴³ , ⁴⁴ , ⁴⁵ , ⁴⁶ , ^{47 ]} although they require several synthetic steps and might lack of modularity. The use of redox ligands on gold(I) that can readily undergo one‐electron oxidation has also been envisaged to promote gold catalysis.^{[ 48} , ⁴⁹ , ⁵⁰ , ⁵¹ , ^{52 ]}

Based on the idea to alleviate the tedious preparation of complex ligands, the use of non‐metallic additives that could promote the activation of the Au─Cl bond via halogen bonding,^{[ 53} , ⁵⁴ , ^{55 ]} chalcogen bonding,^{[ 56 ]} hydrogen bonding,^{[ 57} , ^{58 ]} or very recently carbon bonding,^{[ 59 ]} a special case of bonding by group 14 elements collectively coined tetrel bonding^{[ 30} , ⁶⁰ , ⁶¹ , ^{62 ]} has recently witnessed interest (Figure 1).

Figure 1.

Selected examples of LAuCl activation through weak intermolecular interactions.

Intriguingly, silicon activation via silicon‐based Lewis acids has remained underexplored, while silicon forms strong bonds with chlorine (e.g., 113 kcal mol⁻¹ for TMSCl^{[ 63 ]}) and the halogenophilic character of silicon has already been demonstrated for the cationization of gold chloride complexes with silyliums.^{[ 64} , ^{65 ]} Thus, the possible activation of LAuCl precatalysts by silane derivatives would appear highly desirable, whatever the mechanism, from weak interactions to the formation of an ionic pair through cationization, all the more so as the tetravalence of silicon could modulate the activation mechanism.

Results and Discussion

To assess the feasibility of the Au–Cl bond activation process, we focused on two families of silicon Lewis acids. First, we wished to examine Martin spirosilanes [Si‐1] and [Si‐2], whose Lewis acidities are well‐established,^{[ 66} , ⁶⁷ , ⁶⁸ , ⁶⁹ , ^{70 ]} [Si‐2] having already demonstrated higher Lewis acidity than [Si‐1].^{[ 71} , ^{72 ]} The second family consisted of bis(perhalocatecholato)silanes, initiated by Bergman and Tilley^{[ 73 ]} with the bis(perfluorocatecholato)silane derivative, and that have been further classified as Lewis superacids by Greb.^{[ 74 ]} Notably, bis(perbromocatecholato)silane [Si‐3] (Figure 2) has been shown to abstract chloride from tritylchloride.^{[ 75 ]} Additionnally, for both families, the corresponding hypercoordinated chlorosilicates resulting from the addition of a chloride anion have been described in the literature.^{[ 76} , ^{77 ]}

Figure 2.

Spirosilanes [Si‐1–3] and Au(I) precatalysts [Au‐1–7].

Our investigation also aimed at studying the impact of the ligand on Au(I) and we selected representative Au(I) precatalysts bearing phosphine ligands such as chloro(triphenylphosphine) gold(I) [Au‐1] and its bromo analog [Au‐2], JohnPhos gold(I) complex [Au‐3] and the chiral dinuclear BINAP gold(I) complex [Au‐4], as well as carbene ligands such as the NHC IPr gold(I) complex [Au‐5] and our recently developed indolizy C‐centered carbene featuring strong σ‐donating and π‐accepting properties ([Au‐6] complex).^{[ 78} , ^{79 ]}

The feasability of diverse benchmark reactions was examined and the reaction initially studied consisted in the cycloisomerization of enyne malonate 1 for which the activation of [Au‐1] with a silver salt was proven to be necessary in dichloromethane (DCM) at room temperature.^{[ 80 ]} We confirmed this finding and found that in similar conditions, no reaction was taking place with [Au‐1] alone even after overnight stirring (Table 1, entry 1). Pleasingly, using an equimolar mixture (5 mol% loading) of the Au(I) precatalyst [Au‐1] and of the three silane derivatives [Si‐1], [Si‐2], and [Si‐3] proved to be rewarding, since the cycloisomerization process was observed and gave the expected cyclic products 2 in satisfactory (> 95%) NMR yields (Table 1, entries 2, 4, and 9). As previously reported in the literature, 5‐membered products 2a and 2b prevailed over the 6‐membered regioisomer 2c. To test the stability of the catalytic system involved, we ran reactions in open flask mode. Since bis(perbromocatecholato)silane [Si‐3] led to almost no conversion of 1 in these conditions compared to [Si‐1] and [Si‐2] (entry 3 versus entries 5 and 10), we decided to focus for the rest of the study on the more robust catalytic system based on Martin silanes [Si‐1] and [Si‐2]. Note also that silanes alone have no catalytic activity as shown in entry 8 and that the use of [Si‐2] can result in much shorter reaction time. Only 1 h was needed to get full conversion of 1 (entry 9). A marked solvent effect was observed with [Si‐1] since almost no cycloisomerization of 1 was observed in toluene (entry 6). This is reminiscent of Echavarren's findings about the inadequacy of toluene for this reaction using a silver salt activator.^{[ 80 ]} In contrast, toluene could be used with [Si‐2] (entry 11), which is a quite valuable alternative to the use of chlorinated solvents. Other solvents were tested and showed a similar trend (Table S1). We then tested the efficiency of the activation process with a low loading (0.5 mol%) of the catalytic system. In these trials, [Si‐2] proved superior to [Si‐1] by ensuring high conversion and a satisfactory yield of cycloisomerization products 2 (entry 12 versus 7). Silane [Si‐2] also performed better than commonly used silver salts (Table S9 and Figure S13).

Table 1.

Cycloisomerization of enyne 1.

Entry	Silane	[Au] cat.	Yield (%) a)	Ratio 2a:2b:2c:2d a)	1 (%)
1	none	[Au‐1]	n.d.	n.d.	96
2	[Si‐3]	[Au‐1]	95	85:0:10:0	n.d.
3 b)	[Si‐3]	[Au‐1]	13	13:0:0:0	87
4	[Si‐1]	[Au‐1]	95	84:0:9:2	n.d.
5 b)	[Si‐1]	[Au‐1]	89	76:0:6:7	<5
6 c)	[Si‐1]	[Au‐1]	3	3:0:0:0	93
7 d)	[Si‐1]	[Au‐1]	40	40:0:0:0	55
8	[Si‐1]	none	n.d.	n.d.	>95
9 e)	[Si‐2]	[Au‐1]	95	86:0:9:0	n.d.
10 b)	[Si‐2]	[Au‐1]	85	59:15:9:2	n.d.
11 c)	[Si‐2]	[Au‐1]	80	50:9:18:3	n.d.
12 d)	[Si‐2]	[Au‐1]	88	63:0:14:11	n.d.
13	[Si‐2]	[Au‐2]	84	72:5:7:0	n.d.
14 d)	[Si‐2]	[Au‐2]	91	68:0:19:4	n.d.
15	[Si‐2]	[Au‐3]	82	48:25:1:8	n.d.
16	[Si‐2]	[Au‐4]	88	62:9:10:7	n.d.
17	[Si‐2]	[Au‐5]	86	71:7:0:8	n.d.
18 b) , d)	[Si‐2]	[Au‐5]	91	91:0:0:0	n.d.
19	none	[Au‐5]	4	4:0:0:0	90
20 e)	[Si‐2]	[Au‐6]	95	78:22:0:0	n.d.
21 d)	[Si‐2]	[Au‐6]	95	95:5:0:0	n.d.
22 c)	[Si‐2]	[Au‐6]	90	48:35:4:3	n.d.
23 f)	BCF	[Au‐1]	95	78:0:17:0	n.d.
24 b)	BCF	[Au‐1]	98	68:0:9:21	n.d.

^a) Unless otherwise noted, the reaction was performed by using 1 (0.2 mmol), [Au] precatalyst (5 mol%, 0.01 mmol), and a silane [Si] (5 mol%) in CH₂Cl₂ (2.0 mL) under an argon atmosphere at room temperature for 18 h. ¹H NMR yield was calculated by using 1,3,5‐trimetoxybenzene (TMB) as internal standard.

^b) Open‐flask conditions.

^c) Toluene was used as solvent.

^d) 0.5 mol% of catalyst loading.

^e) Finished in 1 h.

^f) BCF = B(C₆F₅)₃.

Due to its higher reactivity, we therefore selected [Si‐2] as activator for the study of the other Au(I) precatalysts. Interestingly, the less used bromide salt PPh₃AuBr ([Au‐2]) provided comparable results (entries 13 and 14). Other phosphine‐based precatalysts [Au‐3] and [Au‐4] resulted in satisfactory conversion (entries 15 and 16). Carbene‐based Au(I) complexes proved also to be competent. Notably, the IPr‐based [Au‐5]/[Si‐2] catalytic system exhibited very high reactivity and selectivity and the low loading conditions of entry 18, though also open‐flask, yielded the best result. Importantly, we also checked that an activator was needed with [Au‐5] (entry 19). Similar excellent results were obtained with indolizy‐based complex [Au‐6] (entries 20–22). Additionally, the use of tris(pentafluorophenyl)borane (BCF), which had been previously used to cationize gold(III)(OCOCF₃) salts,^{[ 81} , ^{82 ]} led to complete conversion of the starting material (entry 23) with 5 mol% of [Au‐1]. Unlike silanes [Si‐1] and [Si‐2], BCF must be stored and handled in a glovebox since it hydrates rapidly.^{[ 83} , ⁸⁴ , ^{85 ]} Not surprisingly, in open‐flask conditions it led to substantial amount of hydroxycyclization product (21% of 2d, entry 24).

It should also be noted that this activation process is particularly robust as illustrated by the sequence of Scheme 1. A mixture of 1 and 5 mol% [Au‐1] in CH₂Cl₂ was stirred for 2 h at rt. After having checked by ¹H NMR that no cycloisomerization took place, [Si‐2] (5 mol%) was added. After 30 min of stirring, ¹H NMR showed full conversion of 1 to give a mixture of 2a–d, from which 2a was highly major (see page S8 and Figure S2).

Scheme 1.

Silane activation in the course of a reaction.

Adventitious H₂O trapping in some of the previous cycloisomerization reactions of 1 was suggestive that we could run alkoxycyclization reactions.^{[ 86} , ^{87 ]} While the presence of an activator was proved to be necessary (entry 5), [Si‐2] was again the best activator for the methoxycyclization of 1 in terms of conversion and selective formation of 2e (entries 2 versus 1 and 3 of Table 2) though full conversion could be reached only after prolonged reaction time (entry 4).

Table 2.

Alkoxycyclization reaction of 1 in the presence of MeOH. ^a)

Entry	Silane	Yield(%) a)	2e:2a:2c ratio	1 (%)
1	[Si‐1]	23	14:9:0	72
2	[Si‐2]	68	63:5:0	30
3	[Si‐3]	48	41:7:0	51
4 b)	[Si‐2]	93	78:11:4	n.d.
5	–	n.d.	n.d.	96

^a) Unless otherwise noted, the reaction was performed using 1 (0.1 mmol), [Au‐1] cat. (5 mol%), and a silane (5 mol%) in a 3:1 CH₂Cl₂/MeOH mixture (2.0 mL) under argon at rt for 18 h. ¹H NMR yield was calculated by using TMB as internal standard.

^b) 48 h.

This series of findings augured well for further explorations and we wished to delineate the scope of the process with regard to different types of substrates (Scheme 2). Thus, the prototypical reactivity of propargyl esters in inter‐ or intramolecular setting was examined.^{[ 4 ]} The [Si‐1]/[Au‐1] mixture demonstrated efficiency to promote the 1,2‐OAc migration on propargyl ester 3 followed by the trapping of the resulting carbene by styrene to deliver the expected cyclopropyl adduct 4 in 80% isolated yield as Z isomer^{[ 88 ]} and a 9:1 mixture of cis:trans isomers. Enynyl ester 5 could also be converted to the previously observed mixture of tricyclic product 6 originating from a 1,2‐OAc migration accompanied by the 1,3‐OAc migration allenylester product 7a and its hydrolyzed derivative 7b. Note that a reversed 2:1 ratio of cyclopropyl (6)/allenyl products (7) was obtained with [Au1] and AgSbF₆ as activator.^{[ 89 ]}

Scheme 2.

Substrates scope.

The evaluation of substrates possessing polar functions appeared of high interest. Gratifyingly, good conditions were found for the cycloisomerization of N‐2‐propyn‐1‐ylbenzamide 8, which gave 5‐methylene‐4,5‐dihydrooxazole 9a, with only traces of 9b (<5%), consistent with the major intervention of Au(I) catalysis.^{[ 90} , ^{91 ]} For this precursor, [Si‐2] also proved to be the best silane, and even allowed a 0.5 mol% loading of [Au‐1]/[Si‐2] mixture. Good yields of heterocyclizations from allenol 10 and allenyl carbamate 12 precursors were observed upon using the regular conditions consisting of 5 mol% [Au‐5]/[Si‐2]. The silane protocol could even be applied to allenyl carboxylic acid 14 to give lactone 15.

We wished to gain more insight into the activation process and notably monitor the evolution of the species involved. An equimolar mixture of [Au‐1] and of [Si‐2] was analyzed by NMR spectroscopy. Thus, by ¹H NMR (300 MHz), no new species appeared. A slight upfield shift from 8.02 to 8.00 ppm of the proton signals of the aromatic rings of [Si‐2] was observed (Figure 3a). ³¹P NMR also showed an upfield shift from 33.20 to 32.96 ppm for the ³¹P atom of the phosphine ligand (Figure 3b) as observed by Lu and Wang^{[ 59 ]} with the carbon‐bonding activator of Figure 1. These authors attributed the observed upfield shift to the delocalization of electron density from the neighboring atoms to phosphorus to compensate the withdrawing effect of the carbon bonding. A similar phenomenon would be at play with silicon. It was also accompanied by a broadening of the peak, as observed with some cationic Au(I) species.^{[ 92 ]} Similarly, ¹⁹F resonances of the diastereotopic CF₃ groups of [Si‐2] lost resolution in the presence of [Au‐1]. Further investigation in this interaction was pursued by increasing the amount of [Si‐1] relative to [Au‐1] (Figures 4 and S9). From a CD₂Cl₂ solution of [Au‐1] alone to a [Si‐1]/[Au‐1] mixture in a 25:1 ratio, an upfield shift of the ³¹P resonance from 33.78 to 33.55 ppm was observed. Finally, the formation of a hypercoordinate chlorosilicate species from a 1:1 mixture of [Au‐1] and [Si‐2] was discarded by ²⁹Si NMR, which showed only Si resonance at 5.08 ppm, corresponding to the chemical shift of silane [Si‐2] (see ¹H/²⁹Si HMQC of Figure S11).

Figure 3.

NMR study of an equimolar mixture of [Au‐1] and [Si‐2] in CDCl₃ a) ¹H, b) ³¹P, and c) ¹⁹F NMR spectra.

Figure 4.

³¹P NMR monitoring of a [Si‐1]:[Au‐1] mixture in CD₂Cl₂ through increasing the quantity of [Si‐1].

DFT calculations proved to be helpful to account for this activation process by notably following the evolution of the Au─Cl─Si triad during a cycloisomerization reaction (Figure 5). Interacting [Au‐1] with [Si‐1] leads to the formation of intermediate A, which exhibits a weak long‐range interaction (3.74 Å) between the Cl and Si atoms. A transition state could be located starting from A (7.3 kcal mol⁻¹ above A) and resulting in the new intermediate B, wherein the increased proximity of the chlorine atom to the silicon induces a significant change in geometry, from a tetrahedral as in A (see Table S10 for a description of [Si‐1]) to a quasi‐trigonal‐bipyramidal structure (O─Si─Cl, 87.6°‐86.9°; O─Si─O, 173.8°; C─Si─C, 136.3°; C─Si─O, 96.6°, 92.3°).^{[ 67} , ^{68 ]} A Si─Cl distance of 2.33 Å was found, which matches the calculated d(Si─Cl) of a chorosilicate with the PPh₃Au⁺ counter‐cation (See Figure S14). Still, this approach does not result in a substantial separation of the Cl atom from the gold (2.39 Å compared to 2.31 Å in [Au‐1]) and the transition from A to B is endothermic by 4.2 kcal mol⁻¹, indicating that the formation of B is highly reversible. This could explain why a solution of [Au‐1] and [Si‐1] showed no dramatic changes by NMR (see Figures 3 and S9).

Figure 5.

Computed reaction path using DFT at B3PW91‐D3(BJ)/def2‐TZVP(dichlo)//B3PW91‐D3(BJ)/def2‐SVP level Gibbs free energy in kcal mol⁻¹, distance in Å.

When enyne 1 was coordinated to B, the system evolved toward the formation of a species with a stabilised energetic profile, resulting in the generation of complex C. It is noteworthy that subsequent to the coordination of 1 on gold, the Cl atom persisted in its migration toward Si, with a shorter Cl─Si distance of 2.20 Å (2.33 Å in B) and a longer Cl─Au distance of 2.78 Å (2.39 Å in B, 2.31 Å for [Au‐1] alone). This outcome is not unexpected, given that upon coordinating 1 to B, the electronic contribution of the triple bond to gold has reduced its cationic character and consequently its interaction with chlorine atom, which has migrated closer to the Si atom. Interestingly, in complex C, the interaction between gold and chloride remains present. In order to gain insight into the TS associated with the formation of the C─C bond of the cycloisomerization process, a scan of the potential energy surface (PES) corresponding to the shortening of the distance between the two carbon atoms involved in the bond was conducted. This approach was utilized for the formation of the five‐ and six‐membered rings intermediates, respectively D and E, which lead to the experimental cycloisomerization products (the six‐membered ring being the minor product).^{[ 93 ]} An analysis of the variation in the distances d(Si─Cl) and d(Cl─Au) as a function of the distance between the two carbon atoms involved in the formation of the bond at the origin of the cyclisation reveals that the “formation/release” of the [L‐Au]⁺ catalyst occurs progressively at an advanced stage (see Scheme S3).

Thus, it appears that as the carbon atoms move closer together, the transfer of the Cl atom from Au to Si is facilitated, ultimately leading to the formation of a transition state in which the cationic nature of gold effectively induces cyclisation. In other words, the spirosilane does not facilitate the formation of the L‐Au cationic catalyst directly. Rather, it serves to capture the Cl atom when, from a certain distance between the two C atoms involved in the C─C bond formation, the cationic form must be available to carry out cyclisation. The introduction of the spirosilane into this cyclisation process appears to be an effective approach, as evidenced by the barrier energies that must be surmounted. This is exemplified by the case of TS_C5 (giving the major product). The barrier values for the cyclisation are 7.8 kcal mol⁻¹ for [Ph₃PAu]⁺ alone and 9.9 kcal mol⁻¹ in the presence of spirosilane. It should also be noted that a similar reaction pathway was calculated with [Si‐2] and it appeared energetically more favorable (see Figure S17).

These calculations showed that, throughout the catalytic cycle, the silane is in close proximity to the gold center, which naturally could have important consequences for catalysis. This led us first to test an asymmetric transformation, and the enantioselective cyclization of allenol 10 to generate the enantioenriched tetrahydrofuran product 11 was envisaged to probe this hypothesis (see Table 3). For that purpose, we prepared optically pure (R)‐[Au‐4] from (R)‐BINAP. When the reaction was run with [Si‐2] as activator in DCE at 25 °C (entry 1), degradation of the reaction product was observed so the temperature was decreased to −20 °C in DCE (entry 2) and the desired product 11 was obtained in 78% yield and 53% ee, which is the best ee observed on this substrate with the [Au‐4] catalyst.^{[ 94 ]} Running this reaction at −50 °C in CH₂Cl₂ (entry 4) led to a decrease in both yield (15%) and ee (23%), evidencing diminished reactivity and selectivity at lower temperature. This was in fact consistent with variable temperature (VT) NMR studies, which suggested diminished coordination of (R)‐[Au‐4] to [Si‐2] when cooling the mixture (Figure S12). Performing the reaction in toluene (entry 3) resulted in slightly reduced yield (62%) and ee (35%) compared to DCE. Comparing AgSbF₆ with [Si‐2] (entry 5) or AgOTf (entry 6) showed diminished but consistent with literature^{[ 94 ]} enantioselectivity (respectively 14 and 17.5%) and lower yields, confirming the highly beneficial role of the silane additive in inducing enantioselectivity (Table 3).

Table 3.

Enantioselective cyclization of allenol 10. ^a)

Entry	Activator	T (°C)	Yield of 11 (%) a)	ee of 11 (%)
1	[Si‐2]	25	n.d.	–
2	[Si‐2]	−20	78	53
3 b)	[Si‐2]	−20	62	35
4 c)	[Si‐2]	−20	60	52
5 c)	[Si‐2]	−50	15	23
6	AgSbF6	−20	55	14
7	AgOTf	−20	14	17.5

^a) Unless otherwise noted, the reaction was performed using 10 (0.2 mmol), [Au] cat. (2.5 mol%), and a silane or Ag salt (5 mol%) in DCE (1.0 mL, 0.2 M) under argon at rt for 18 h by isolated yield.

^b) Solvent was toluene.

^c) Solvent was CH₂Cl₂.

We also studied the asymmetric cycloisomerization of N‐tethered enynol 16.^{[ 95} , ^{96 ]} Using the ( R)‐[Au‐4]/ [Si‐2] combination provided product the expected bicyclic product 17 in 72% yield and 25% ee (Table 4, entry 1). However, using the (R)‐[Au‐7] complex, flanked with the DTB‐MeO‐BIPHEP ligand resulted in a substantial enhancement in enantioselectivity 64% ee. When the activator AgSbF₆ was employed (entry 3), once again enantioselectivity decreased to 49%, confirming the cooperative role of [Si‐2].

Table 4.

Enantioselective cyclization of N‐tethered enynol 16.

Entry	[Au]	Activator	Yield of 17 (%) a)	ee of 17 (%)
1	(R)‐[Au‐4]	[Si‐2]	72	25
2 b)	(R)‐[Au‐7]	[Si‐2]	61	64
3	(R)‐[Au‐7]	AgSbF6	57	49

^a) Unless otherwise noted, the reaction was performed using 14 (0.2 mmol), [Au] cat. (2.5 mol%), and a silane or Ag salt (5 mol%) in DCE (1.0 mL, 0.2 M) under argon at rt for 18 h by isolated yield.

^b) Catalysts’ loading were ( R‐[Au‐7] = 5 mol% and [Si‐2] = 10 mol%.

These calculations were also suggestive that the Au─Cl bond can be preserved during the cycloisomerization process. Since structures D and E are of the same type as the previously calculated skeletal rearrangement intermediates involved in the formal cycloisomerization of 1,^{[ 93} , ^{97 ]} giving the identical products 2, it was interesting to examine theoretically the fate of the PPh₃Au⁺ fragment that is released at the end of the skeletal rearrangement process. To achieve this, we placed the PPh₃‐Au⁺ fragment at a distance greater than 4 Å (Au─Cl distance) from the chlorosilicate [Si‐1‐Cl]⁻ and optimized this starting point. In a somewhat unanticipated manner (it was initially hypothesised that Cl would move toward Au), the structure obtained [(Si‐1‐Cl);O─AuPPh₃] is one in which Au has coordinated to an oxygen atom (Figure S16). The energy balance of such an association is characterised by an exothermicity of 38.1 kcal mol⁻¹. The transfer of AuPPh₃ to the Cl atom from [(Si‐1‐Cl);O‐AuPPh₃] requires an activation barrier of only 2.7 kcal mol⁻¹, this transition state leading directly to B, which can then undergo a new reaction cycle. So, this drove us to investigate the possibility of recycling LAuCl catalysts^{[ 58 ]} and we tested this with a mixture [Au‐1] and [Si‐1] on precursor 1 (Scheme 3). After a first run, [Au‐1] could be recovered through a very simple protocol. Following evaporation under reduced pressure, addition of hexane to the reaction mixture resulted in the formation of a white precipitate which was filtered off and which consisted of [Au‐1] (90% recovery yield). Purity of [Au‐1] was confirmed by ³¹P and ¹H NMR (see Figure S8). Catalytic activity was tested and provided products 2 in 82% yield. A second recycling run gave similar yields of recycled [Au‐1] and products 2 (Table S7 and Figure 6).

Scheme 3.

Catalyst recycling evidenced with [Au‐1] and [Au‐5].

Figure 6.

Recycling of [Au‐5] over 7 consecutive runs.

Furthermore, recycling of the [Au‐5] catalytist could be accomplished over seven consecutive reaction runs. In this series, the recycled [Au‐5] provided yields of products 2 between 80% and 96% (Figure 6). A progressive decrease in the initial [Au‐5] content was observed from 90% after the first run to 15% after the seventh run (in average 79% yield). The purity of the recovered [Au‐5] catalyst was confirmed by ¹H NMR spectroscopy and high‐resolution mass spectrometry (HRMS) (Figure S8). Thus, despite the significant depletion of [Au‐5], the reaction yield remained consistently high (≥80%) through all cycles, peaking at 96% in the sixth run. These findings confirmed that a minimal quantity of [Au‐5] is sufficient to maintain high catalytic performance, highlighting its robustness and efficiency under recycling conditions. A comparable result was observed for [Au‐5] and [Si‐1], achieving yields of 92% and 85% in the first and second cycles, respectively. The recovery of [Au‐5] was also efficient, with 93% and 84% recovered in the two runs (see Table S7 and Figure S7). This catalyst recycling was also performed on substrates 5 and 10 (see Table S8 and Scheme S2).

Conclusion

To conclude, we have shown that Martin spirosilanes [Si‐1] and [Si‐2] mediate the activation of LAuCl (L = carbene or phosphine ligands) precatalysts for promoting cycloisomerization and intermolecular reactions on diversely functionalized substrates. The use of the more Lewis acidic [Si‐2] silane shows higher performance as only 0.5 mol% of loading of gold(I) and silane can be used. While NMR studies did not evidence cationization, i.e., formation of an ionic pair, computational investigations suggested that Cl shuttles back and forth between gold and silicon along the catalytic cycle. This preserves the Au─Cl bond and ultimately enables the LAuCl precatalyst to be recycled. Another advantage of this activation lies in the proximity of the silane to the catalytic gold(I) center, which not only does not prevent asymmetric catalysis, but allows to amplify in comparison to silver salts the asymmetric induction provided by a chiral ligand. This work significantly contributes to the prospect of a more sustainable gold catalysis. It also opens exciting perspectives about the possibility to confer cooperative properties to the silanes, notably in asymmetric catalysis.

Conflict of Interests

The authors declare no conflict of interest.

Supporting information

Supporting Information

Dedicated to Professor Iwao Ojima at the occasion of his 80 birthday

Mutalliev L., Beudy E., Ballarín‐Marión M., Delattre V., Troufflard C., Mouriès‐Mansuy V., Gimbert Y., Fensterbank L., Angew. Chem. Int. Ed. 2026, 65, e18534. 10.1002/anie.202518534

Data Availability Statement

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