Screening of the optimal catalysts. In Grubbs’ previous studies on hydration of nitriles and cyanohydrins, more electron-rich groups in the “donor” moiety was discovered to increase the activity of the catalysts49. Therefore, we take the electron-rich 1.1’-[bis(5-methyl-2-furanyl)phosphine]ferrocene “donor” ligand and modulate the electronic nature of the monodentate “acceptor” ligands. As shown in Figure 2, five Pt-catalysts A-E were synthesized via the reaction of 1.1’-[bis(5-methyl-2-furanyl)phosphine]ferrocene platinum dichloride (dmfpfPtCl2) with the corresponding monodentate phosphine ligands (in blue) in the presence of silver triflate, respectively (See Supporting Information). Then the activity of the newly synthesized platinum catalysts was evaluated by performing hydroalkoxylation of 1 and hydroamination of 2 (Figure 2). The results show that catalyst A, with triphenylphosphine as the monodentate ligand, has the worst reactivity for both hydroalkoxylation of 1 and hydroamination of 2. Catalyst B that bears electron-deficient monodentate tris(4-(trifluoromethyl)phenyl)phosphine slightly improved the reactivity for both hydroalkoxylation and hydroamination. Catalyst C harbouring the more electron-deficient monodentate tris(3,5-bis(trifluoromethyl)phenyl)phosphine displays noticeably increased catalytic activity, providing hydroalkoxylation product 1a and hydroamination product 2a in 46% and 44% yields, respectively. The observation that more electron-deficient tri-aryl phosphine ligand in the “acceptor” moiety resulted in higher catalytic activity as demonstrated by the improved performance of catalysts A, B and C supports Grubbs’ “donor–acceptor” catalyst design49. Catalysts D and E bearing other two different electron-deficient monodentate ligands, tris(3,4,5-trifluorophenyl)phosphine and tris(3,5-difluoro-4-(trifluoromethyl)phenyl)phosphine, also show good reactivity for both hydroalkoxylation and hydroamination, and have their own advantages for different substrates. The trend that more elelctron-rich group in the “donor” part and/or the more electron-deficient group in the “acceptor” part increased the activity of the Pt-catalysts could be explained: the cationic Pt(II) was stabilized by the strong σ donating from the electron-rich bidentate ligand, while the electron-deficient monodentate ligand which acts as weak σ donor but strong π acceptor makes the Pt(II) center more electrophilic towards multiple bonds47-48, 51.
Scope for hydroalkoxylation. we next investigated the scope of intramolecular hydroalkoxylation with catalyst C, which contains the commercially available and inexpensive tris(3,5-bis(trifluoromethyl)phenyl)phosphine as the monodentate ligand (Table 1). All the reactions were conducted under mild conditions (either at 23 or 50 °C). The catalytic protocol displays excellent generality with Markovnikov regioselectivity, and is notably applicable to the synthesis of sterically hindered ethers with fused-, bridged- and spiro-ring systems (Table 1, entries 5; 15; 2, 3, 6, 9, 10, 13 and 20). Various hydroxyl groups such as primary, secondary, tertiary alcohols and phenols serve as good nucleophiles. Different olefins with various substitution patterns including trisubstituted (Table 1, entries 1 to 6), 1,1-disubstituted (Table 1, entries 7 to 15), 1,2-disubstituted (both cis and trans) (Table 1, entries 16 to 18), and even mono-substituted (Table 1, entries 19 and 20) double bonds work well in this reaction. Good to excellent yields were obtained in all cases reported.
Table 1. Substrate scope for intramolecular hydroalkoxylationa.
Expanding our investigation to intermolecular hydroalkoxylation, we discovered that catalyst D was the optimal catalyst whereas C gave slightly reduced yields over a range of different substrates (Table 2). Higher catalyst loadings (5 mol%) and slightly elevated temperatures are required for optimal conversion of the intermolecular reactions reported here. Alkenes with ring strain such as norbornene (23), comphene (24), and four-membered carbocycles (25) all productively undergo intermolecular hydroalkoxylation reactions with alcohols. Acyclic alkenes such as 26 and 27 that are less reactive substrates than their cyclic counterparts were coupled with alcohols in moderate yields. These reactions generally stalled at 30 to 70% conversions depending on the nature of the nucleophile used; longer reaction times or higher catalyst loadings do not help to convert the reactions further. Intermolecular hydroalkoxylation of norbornene (23) with para-substituted phenols were also investigated. Phenols have a potential to engage in hydroarylation reactivity57; however, we only observed hydroalkoxylation products (ethers) when para-substituted phenols (CF3, F, Cl, Br) and norbornene (23) were treated with D/AgOTf at either room temperature or 50 °C. In contrast, bis-ortho-hydroarylation products were exclusively generated when triflic acid (TfOH) alone was used as the catalyst (See the Supporting Information for further details). These results suggest a distinct mechanistic pathway for our catalytic system, different from that of a Brønsted acid-catalyzed hydroalkoxylation.
Table 2 Substrate scope for intermolecular hydroalkoxylationa.
Substrate scope for hydroamination. The “donor–acceptor“-type Pt-catalysts not only are capable of catalyzing hydroalkoxylation, but can also effect hydroamination of unactivated alkenes in generally good yields. As complied in Table 3, the catalyst E, even used at lower loading of 1 mol%, was shown to be capable of effecting intramolecular alkene hydroamination with sulfonamides at ambient temperatures, providing various nitrogen-heterocycles that are important structraul motifs in naturally occuring and pharmaseuticals. A wide range of alkene substitution patterns are tolerated, including 1,1-disubstitutied (Table 3, entries 1 to 8), 1,2-disubstituted (Table 3, entry 9), mono-substituted (Table 3, entries 10 and 11) and trisubstituted alkenes (Table 3, entries 12 to 14). The relative stereochemistry of 41a was unambiguously confirmed by X-ray crystallography (Table 3, entry 14). Next, we examined the viability of more difficult intermolecular alkene aminations with p-toluenesulfonamide, (p-tolylsulfonyl)methylamine, N-tosyl-4-methoxyaniline and methanesulfonamide (Table 4). A number of alkene substitution patterns were accommodated, including trisubstituted (Table 4, entries 1 to 4), terminal (Table 4, entries 5 to 7) and 1,2-disubstituted alkenes (Table 4, entries 8 and 9). Both cyclic and acyclic alkene partners can be aminated successfully. The structure of 23f was verified by X-ray crystallogrphy.
Table 3. Substrate scope for intramolecular hydroaminationa.
Table 4. Substrate scope for intermolecular hydroaminationa.
Mechanistic Studies. Recently, several groups have independently demonstrated that triflic acid can catalyze the additions of oxygen- and nitrogen-based nucleophiles to simple alkenes with comparable efficiency/selectivity as some metal triflates53-56. The aforementioned reports raised the question of a competitive acid-catalyzed pathway when metal triflates are employed. Therefore, we elected to perform a detailed analysis of the possible mechanistic pathways for the “donor–acceptor” Pt-catalyzed hydroalkoxylation of alkenes.
The mechanistic study commenced with monitoring the reaction of norbornene and 4-trifluromethylphenol catalyzed by D and AgOTf in deuterated dichloromethane at 23 oC by 31P, 19F and 1H NMRs (Figure 3 and please also see Supporting Information). When equal molar of AgOTf was added into catalyst D in deuterated dichloromethane, 31P NMR revealed the appearance of a new peak at d -22.5 (s, 1JPt-P = 4022 Hz) ppm (Figure 3c). This peak was confirmed to be complex G that can be generated independently by mixing 1 equivalent of 1,1’-[bis(5-methyl-2-furanyl)phosphine]ferrocene platinum dichloride (dmfpfPtCl2) with 2 equivalents of AgOTf (Figure 3b). It is worth to note that when the chloride ion in the catalyst D was abstracted by AgOTf, a more electron-deficient Pt(II) center was formed; as a result, the monodentate phosphine ligand was dissociated from the Pt center and coordinated with Ag(I) ion to form new complexes such as [(Ar3P)xAgOTf]n (H) with 19F signals at d -131.2 (d, J = 21.3 Hz) and -154.2 (t, J = 21.0 Hz) ppm (please see 19F NMR in Supporting Information)58.
4 equivalents of norbornene were then added to the above solution which led to rapid disappearance of complexes G and H, and 1H NMR analysis revealed that diagnostic alkene peaks move downfield (Figure 3d and also see 19F and 1H NMRs in Supporting Information). These results together indicate that the mono-phosphine ligand re-combined to the Pt center when the alkene was coordinated to the Pt. Finally, when 4-trifluoromethylphenol was added, 1H NMR confirmed that the hydroalkoxylation product 23b was formed with the concurrent regeneration of complexes G and H (Figure 3e and also see Supporting Information).
To ascertain whether the nucleophile is reacting or coordinating with the Pt center, we changed the sequence of adding 4-trifluoromethylphenol and norbornene (Figure 4). When 4 equivalents of 4-trifluoromethylphenol were added into the solution of D and AgOTf, there was no change in 1H, 19F and 31P NMRs (Figure 4b and also see Supporting Information). Next, 4 equivalents of norbornene were added and the resulting solution was immediately analyzed: 31P and 19F NMRs revealed that both the 31P signal for complex G and 19F signals for complex H decreased (Figure 4c and see Supporting Information). These results suggest that both the mono-phosphine ligand and norbornene were coordinating to the Pt center. As the reaction proceeds for 1 hour, the hydroalkoxylation product 23b was formed, and both complexes G and H were generated again (Figure 4d, and also see 1H and 19F NMRs in Supporting Information).
Based on the NMR experiments, we propose a plausible mechanism for “donor–acceptor”-type Pt-complex catalyzed hydroalkoxylation of unactivated alkenes (Figure 5). Catalyst D reacts with AgOTf to form Pt(II) species F and precipitates AgCl. Since complex F is more electron-deficient and unstable, the monodentate phosphine ligand dissociates from the Pt center and reacts with an additional equivalent of AgOTf to generate complexes G and H. Then complex G coordinates with the alkene to form complex I. Due to its s-donating nature, the alkene makes the Pt center in complex I more electron-rich than in complexes F and G. Therefore, the electron-deficient mono-phosphine re-associates to form a 16-electron, “donor-acceptor”-type Pt complex J, thus further activating the alkene. Nucleophilic attack on the bound alkene by a free alcohol provides complex K. Proton transfer thereafter provides the hydroalkoxylation product and regenerates complexes G and H.
Finally, we explored the catalytic asymmetric hydroalkoxylation14, 59-63 and hydroamination64-69 with the “donor–acceptor” catalytic system (Figure 6). Various chiral Pt-complexes generated in situ from chiral bisphosphine platinum dichlorides with electron-deficient tris(3,4,5-trifluorophenyl)phosphine were examined in the reaction of 21 and 2 under standard conditions. Disappointingly, none of the chiral Pt-catalysts produced enantioselectivities (See Supporting Information). We envisioned that a bi-functional catalysis where the monodentate ligand has a basic group that can form a hydrogen-bonding with the nucleophile would assist recognition of prochiral faces of the alkenes61, 70. To our delight, catalysts N and M bearing (R)-BINAP and (R)-Difluorphos as “donor” ligands and the monodentate “acceptor” ligand with a hydrogen bonding site induced moderate enantioselectivities in both hydroalkoxylation of 21 and hydroamination of 2. The observed enantioselectivity strongly supports our proposed mechanism that “donor-acceptor” Pt-catalyzed hydroalkoxyations and hydroaminations are true metal-catalyzed processes53-56.