We began our investigation by revisiting our previously developed catalytic system for enantioselective cascade Heck-type reactions of alkene-tethered aryl iodides44-48. With the ortho-iodophenol-derived allyl ether 1a as the model substrate, we conducted a preliminary study in isopropyl ether (120 oC) for 24 h with Pd2(dba)3•CHCl3 as the precatalyst, N-Me-Xu-Phos (N-Me-Xu3) as the chiral ligand and Cs2CO3 as the base (Table 1). Our initial efforts resulted in poor yield of the desired product 4a with moderate enantiomeric ratio (e.r.) (67% conversion, 22% yield with 88:12 e.r.) together with a series of byproducts 2, 3 and 5 (Table 1, entry 1), indicating that the synthesis of the highly strained chiral 5,4-spirocycles is indeed very challenging. Subsequently, further evaluation of the reaction conditions revealed that all of bases, additives, Pd salts, and temperature were the key factors to obtain a robust reaction.
Several points are noteworthy: (1) this reaction is very sensitive to the bases (Table 1, entries 2−6). For example, when PMP was chosen as the base, the byproduct 2 was given as the main product via carboiodination of alkene (Table 1, entry 3). Nevertheless, changing to CsOAc, a 1,4-Pd shift of palladacycle followed by ortho-C-H activation led to the fused polycycle 6 (Table 1, entry 4). (2) In Lauten’s work, 1a quickly converts to the target compound 4a in good yield by using the racemic ligand (PtBu3) at 100 oC for 1 hour (entry 9)39. By a huge margin, the rate of reaction is greatly reduced under current conditions without PtBu3 (Table 1, entry 8). Even extending to 72 hours and warming up to 110 oC, the desired product 4a was obtained in only 48% yield (63% conversion). Indeed, adding the racemic ligand was beneficial to the conversion of the reaction (Table 1, entries 2 vs 3, and 7 vs 8). However, such a strong racemic background reaction poses a great challenge to the high enantioselectivity. (3) The potassium bases are beneficial to the enantioselectivity (Table 1, entries 5, 6 vs 2). (4) Reducing the temperature to 110 oC could inhibit the formation of the ring-opening by-product 5 (Table 1, entries 1−6 vs 7, 8). After an extensive evaluation of the reaction conditions, the current conditions were not satisfactory (Table 1, entry 7). Not only the desired product 4a was provided with 92.5:7.5 e.r., but also the conditions have poor universality (more detail see the supporting information, Table S1−S9).
Then, we focused our attention on screening of chiral ligands (Table S10−S12). Firstly, various commercially available chiral ligands such as biphosphines (L1−L7), monophosphine (L8), phosphoramidite (L9), ferrocene-derived P,N-ligands (L10−L11), bis(oxazoline) (L12) were tested occasionally together with the byproduct 7. The biphosphine ligands (L1−L4, L6, L7), ferrocene-derived P,N-ligands (L10−L11) failed to give the desired product 4a. Both a bulky biarylphosphine (L5) and monophosphine (L8) gave 4a in good yield without any enantioselectivity. Only the phosphoramidite (L9) delivered 4a but with low enantioselectivity. Subsequently, we examined the performance of a Sadphos kit (including Ming-Phos55, Xu-Phos44-48, Xiang-Phos56, PC-Phos57, Xiao-Phos58, Wei-Phos59 and TY-Phos60 etc.). The dicyclohexyl phosphine ligands (Xu-Phos) remained the most effective for enantioselective induction. Inspired by these results, other Xu-Phos such as N-Me-Xu4 (Table 1, entry 11) and N-CD3-Xu4 (Table 1, entry 10) were then examined under previous reaction conditions. Of which, the better results could be obtained by using the newly indentified N-CD3-Xu4 as the chiral ligand. To the best of our knowledge, the introduction of deuterium atom in the chiral ligand to increase the enantioselectivity are rarely reported.
With the optimal reaction conditions established, we next examined the generality of this enantioselective tandem Heck reaction/C-H alkylation. First, the effect of allyl substituents was examined, and the results are shown in Table 2. A broad series of allyl substituted aryl iodides 1 including monosubstituted phenyl rings with electron-donating or -withdrawing groups at different positions (ortho, meta or para-), a disubstituted phenyl ring, worked smoothly to afford 4a–4l in 65–96% yields with 91:9–96.5:3.5 ers. The absolute configuration of the product was confirmed by the X-ray diffraction analysis of 4b. In addition, highly strained 5,4-spiro-fused 2,3-dihydrobenzofurans 4m–4n that contain medicinally relevant heterocycles, such as 9,9-dimethyl-9H-fluorene and 9-phenyl-9H-carbazol-1-yl were isolated in 70–71% yields with 95:5–96:4 ers. The effect of substituents on the benzene ring of the 2- iodophenoxy moiety was then investigated under the optimal reaction conditions. The desired products 4o–4ad were also obtained in 44–86% yields with 92.5:7.5–97:3 ers for the reactions with substrates bearing monosubstituted phenyl rings with either electron-rich or electron-deficient group at C4 or C5, a disubstituted phenyl ring and a naphthalene ring. When using N-Me-Xu4 as a chiral ligand, except in one particular case (4e), no better enantioselectivities were given. It is general phenomenon that the introduction of deuterium atom is beneficial to increase the enantioselectivity.
To gain deep insight into the reaction mechanism, several control experiments were carried out (Fig. 2). Although a series of byproducts had been inhibited under the standard conditions, a key mechanistic question is: how does the relative rate of σ-alkyl palladium compare reductive elimination with Sonogashira coupling, Suzuki coupling, reductive Heck and alkyne insertion? To probe this question, several competing reactions were performed (Fig. 2a). Note that no desired product 4a was detected with using PhB(OH)2, cyclohexenylboronic acid or phenylacetylene as the second substrates. These results supported that the C(sp3)-C(sp)and C(sp3)-C(sp2) cross-coupling of the domino Heck/Suzuki or Sonogashira reaction is more favored than this present domino Heck/remote C-H alkylation reaction. However, the C(sp3)-C(sp3) cross-coupling is obviously unfavored. Subsequently, adding HCO2Na or ethyl 3-phenylpropiolate to the reaction mixture, roughly the same amount of the reductive Heck product 3 or alkyne insertion61 product 9 with the spirocycle 4a were furnished, indicating that the rates of these reactions are similar. All these results further confirmed that the realization of domino Heck/remote C-H alkylation reaction is extremely challenging31, 39-40, especially in an enantioselective manner. It's worth noting that er of 3, 8, 9, 12 and 13 are different, which might indicate that the transmetalation step taking place prior to the alkene insertion. Experiments with D2O instead of H2O were conducted under the optimal conditions by using 1a and 1g as the substrates (Fig. 2b), respectively. 75% D-labeled 4a was obtained in 85% yield with 94:6 er. Nevertheless, ortho-methyl aryl group derived 1g could not afford D-labeled 4g. Deuterium (D)-labeling experiments confirmed our assumption that the step of C-H bond activation is reversible. Moreover, the five-membered palladacycle of C(sp2)-H bond activation was favored comparing with the less-strained six-membered palladacycle of C(sp3)-H bond activation under the reaction conditions35. Despite overcoming the high barriers was required to construct the highly strained 5,4-spirocycle, the C-C bond of 4a could be cleaved to give the ring-opening byproduct 5a at 120 oC (Fig. 2c). Nonlinear effect studies on the enantiomeric composition of the chiral ligand N-CD3-Xu4 and product 4a indicated there is a clear first-order dependence was observed for the catalyst (Fig. 2d). These results are consistent with an active catalyst/ligand being of a monomeric nature and the reaction possessing a first-order dependence on catalyst. When carboiodination compound 2ae (5-exo) was subjected to the catalysis of Pd(dba)2/QPhos or Pd2(dba)3•CHCl3/N-Me-Xu3, 6-endo-products 16a and 16a' could be obtained, which indicates that the alkene insertion and carboiodination steps are reversible. Subsequently, we examined the progress of the reaction as a function of time and temperature with ortho-iodophenol-derived allyl ether 1ae as the model substrate (Fig. 2f). At 100 oC, the reaction proceeded smoothly and gave the 5-exo-product 2ae with 93% ee and 95:5 rr. With the increase of reaction time, the ratio of 5-exo-product and 6-endo-product, and ee value hardly changes. However, the proportion of 6-endo-product increased significantly with increasing of the temperature (120 oC and 150 oC). Meanwhile, the ee value of 5-exo-product decreased obviously.
Based on the results of the above experiment and previous works, a reasonable pathway for this palladium-catalyzed Heck/intramolecular C-H alkylation reaction is shown in Fig. 3. First, the arylpalladium species I was generated by oxidative addition of 1 or 14 with Pd(0) complex. Then, followed by an intramolecular 5-exo-Heck-type or 6-endo-Heck-type reactions form chiral σ-alkylpalladium species II or II’. Next, aryl (Ar2) C-H bond activation via the σ-alkylpalladium intermediate II deprotonation gave a spiropalladacycle III, which produces the corresponding product 4 or 15 and regenerates the Pd(0) catalyst to the next catalyzed cycle from reductive elimination. Nevertheless, σ-alkylpalladium species II through C-I elimination also afforded the iodide 2. The β-H elimination of 6-endo-cyclization σ-alkylpalladium species II’ could delivered the endo-products 16a and 16a'. Here, it is worth noting that the C(sp2)-H activation, the carbopalladation and the C−I reductive elimination steps are reversible.
Encouraged by these excellent results, we applied this method to the construction of the optically active 5,5-spirocycles. Both enantiomers of a chiral molecule are often required in organic synthesis, biological chemistry, and the medicinal and pharmaceutical industries. Generally, the synthesis of the enantiomer of a chiral molecule could be obtained by using the enantiomer of chiral ligands. Herein, we envisaged that the synthesis of a pair of enantiomers might be achieved by changing the position of the aromatic or heteratomic aromatic rings (Table 3). Using a variety of aryl iodides 14 bearing electron-donating/withdrawing groups at the para- and meta-positions of phenyl rings, both (R)- and (S)-enantiomers of the spiro-indane-dihydrobenzofuran products 15a–15f were obtained in high yields (80–99% for the (R)-enantiomers and 82–98% for the (S)-enantiomers) with excellent enantioselectivity (95:5–97:3 ers for the (R)-enantiomers and 95:5–97.5:2.5 ers for the (S)-enantiomers) under the standard condition, respectively. When the linker at C2 or C3 of heteratomic aromatic rings, such as thiophene, benzothiophene and furan, the corresponding 5,5-spirocycle products 15g–15n were also afforded with good results (93–99% yields with 95:5–98.5:1.5 ers for the (R)-enantiomers and 94–>99% yields with 95:5–98.5:1.5 ers for the (S)-enantiomers).
Then, we further extended the scope of alkene-tethered aryl iodides 14 (Table 4). In more than 40 examples, the desired products 15o–15bc were synthesized exclusively in good yields with excellent enantioselectivity (up to 98.5:1.5 er). Substrates bearing monosubstituted phenyl rings with electron-donating or -withdrawing groups at different positions (ortho or para), a disubstituted phenyl ring, a trisubstituted phenyl ring, a naphthalene, a pyrene and medicinally relevant heterocycles (dibenzo[b,d]thiophene, benzofuran, thiophene and indole) could be well tolerated. Notably, the present asymmetric catalytic system is applicable to BocN as a tether, delivering the indolines 15al–15am in good yields and with good ee values.