Introduction

Transition-metal-catalyzed asymmetric hydroformylation (AHF) represents an 100% atom-economic and clean approach for the preparation of value-added chiral aldehydes with high efficiency1,2,3,4,5,6,7,8,9,10. Since Takaya and Nozaki reported the milestone chiral ligand Binaphos in 199311, many diphosphine ligands, such as Chiraphite12, Bis-diazaphos13, Ph-BPE14, Yanphos15,16, Bobphos17, Bettiphos18 and others19,20,21,22,23,24 are successfully and frequently applied for AHF of monosubstituted, 1,2- and 1,1-disubstituted olefins, providing the corresponding aldehyde with high regioselectivity and excellent enantioselectivity (Fig. 1a). However, ligands for AHF of sterically deactivated trisubstituted alkenes to afford advanced chiral aldehydes are seldom disclosed, due to the challenges in obtaining useful activity and selectivities25. Recently, our group reported a rhodium-catalyzed AHF of cyclopropyl-functionalized-trisubstituted alkenes utilizing (R,S)-DTBM-Yanphos ligand (Fig. 1b)26. However, overriding the Keulemans’ rule27, AHF of trisubstituted alkenes to afford chiral quaternary aldehydes is yet to be explored16. Such advanced chiral aldehydes and their derived alcohols, carboxylic acids and alkenes are of great importance for pharmaceuticals, organic synthesis advanced materials, and others.

Fig. 1: Rh-catalyzed asymmetric hydroformylation of different alkenes.
figure 1

a The state-of-the-art (asymmetric hydroformylation) AHF of alkenes; b Previous work: AHF of cyclopropyl-functionalized trisubstituted alkenes; c This work: AHF of trisubstituted cyclopropenes affording chiral quaternary cyclopropanes.

Chiral cyclopropanes are important structural motifs widely occurring in pharmaceuticals and natural products and as well widely used as versatile building blocks in organic synthesis and materials science28,29,30,31,32,33,34,35. In the past few decades, Simmons-Smith cyclopropanation36,37, transition-metal-catalyzed decomposition of diazoalkanes28,38, Michael-initiated ring closure28, enzymatic methods28,39, hydrofunctionalization of achiral cyclopropenes40,41,42,43,44,45,46,47,48,49,50,51,52,53,54,55,56,57,58,59,60,61, among others are documented for the preparation of chiral 1,2-disubstituted, 1,2,3-trisubstituted and 1,1,2,3-tetrasubstituted cyclopropanes (Fig. 2). However, to the best of our knowledge, there are limited methods reported for efficient asymmetric synthesis of chiral quaternary 1,1,2,2-tetrasubstituted cyclopropanes that are highly strained cyclopropanes with potential applications in medicinal research38,62. Some representative bioactive molecules are shown in Fig. 2, including isovelleral63 and as well inhibitors of aggrecanase64, FASN65, and LRRK266. Therefore, a generally useful catalytic method to access chiral quaternary cyclopropanes with high efficiency remains a challenge.

Fig. 2: The substituted cyclopropane motifs and representative bioactive molecules.
figure 2

Upper part: different substituted cyclopropane skeletons; Lower part: representative bioactive molecules.

In this paper, we disclose a rhodium-L1-catalyzed highly efficient chemo-, regio-, and enantioselective AHF of trisubstituted cyclopropenes for construction of chiral quaternary cyclopropanes (Fig. 1c). Compared to the well-established diphosphine ligands, the simple L1 ligand exhibited high chemoselectivity and regioselectivity, furnishing the target aldehydes with high yields and excellent enantioselectivities. The synthetic utilities of this methodology are highlighted by scale-up reactions as well as follow-up transformations to the corresponding alcohol, acid, ester, lactone, and lactam.

Over the years, computational methods have gained increasing prominence as a tool for investigating mechanisms of catalytic asymmetric transformations62,63,64. Notably, detailed mechanistic studies on catalytic AHF reactions employing DFT computations have been recently reported63,65,66. Lack of molecular insights on the AHF of chiral trisubstituted cyclopropenes leading to chiral quaternary aldehydes motivated us to carry out DFT calculations to elucidate the molecular factors responsible for the regio- and enantio-selectivities of the title reactions. Improved understanding on the mechanism would help exploit the scope of asymmetric variants of AHF reactions.

Results

We initially investigated a variety of well-known chiral diphosphine ligands (Fig. 3) in AHF of the trisubstituted cyclopropene 1a (Table 1). The classical axial ligand Binap showed poor chemoselectivity, resulting in a complex mixture containing mainly decomposed reactants. (S,R)-Yanphos67,68, (S,S)-Yanphos and (S,S)-DTBM-Yanphos69,70,71,72 which are the representative ligands in AHF of monosubstituted and 1,1-disubstituted alkenes, unfortunately (entry 2-4), also gave mainly substrate-decomposition reactions and only afforded the desired product 2a with negligible yields. Diphosphine ligands like BenzP, iPr-Duphos, Duanphos, and Ph-BPE, which have five-member chelating ability, were also not satisfactory (entry 5-8). Although the desired product 2a could be generated with moderate yield and good enantiomeric excess (ee) with Ph-BPE, the side reactions could not be suppressed effectively. To our surprise, our previously designed easy made variant from Bisdiazaphos, L1 that contains a sterically hindered well-defined chiral environment38, exhibited high chemo-, regio- and enantioselectivity, delivering the desired aldehyde 2a with high yield and excellent enantioselectivity (rr > 20:1, 99% ee, entry 9-11).

Fig. 3: Diagram of ligand structure.
figure 3

Bidentate phosphine ligands evaluated for AHF of 1a.

Table 1 Results of AHF of 1a with chiral ligandsaView full size image

With the optimized reaction conditions in hand, we set out to investigate the functional group tolerance and generality of this asymmetric catalysis methodology. As depicted in Fig. 4, for most cases, the tetrasubstituted cyclopropanes bearing a quaternary carbon stereocenter could be obtained with high yields and excellent ee’s, revealing the unique property of the L1 ligand in controlling chemo-, regio- and enantioselectivity. Various functional groups, such as halides (2d, 2e, 2f), methoxy (2g, 2h, 2j), methyl (2k), tertiary butyl (2l), trifluoromethyl (2m), nitro (2n), phenyl (2o) and ester (2p), were well tolerated. Although partially hydrogenation by-products were detected for electron-donating substituents (2j, 2l) bearing substrates, the desired aldehydes were produced with good yields and excellent ee’s. Interestingly, the reaction did not occur even at 90 °C when the methoxy group (1i) occupied the para-position of the phenyl ring. Aromatic functions including naphthyl (2q) and benzofuryl (2r) were also compatible with this asymmetric transformation, affording the target products with high yields and excellent ee’s.

Fig. 4: Scope of substrates.
figure 4

All reactions were performed on 0.2 mmol in 1 mL toluene with 1 mol % Rh(acac)(CO)2 and 2 mol % ligand under 5/5 bar (CO/H2) at 60 °C for 24 h. Rr was Determined by 1H NMR spectroscopy. Isolated yield. Ee was determined by HPLC analysis using a chiral stationary phase after a Wittig reaction with methyl (triphenylphosphoranylidene)acetate. The yields of 2t’-2x’ belong to the products after the Wittig reaction.

Importantly, besides aromatic substituents, alkyl substituents bearing substrates that do not have inherent regioselectivity control via pi-stabilization effects17,73,74,75 also provided the corresponding chiral quaternary products with good to excellent ee’s and useful regioselectivities (2s−2x’) instead of generally electronically and sterically favored linear aldehydes. Unfortunately, a cyclohexyl-bearing substrate 1y and the tetrasubstituted reactant 1z could not proceed the AHF reaction probably because of the significantly increased steric hindrance on the C=C double bond.

To gain insights in the origin of the regioselectivity and enantioselectivity, we conducted control experiments using substrates bearing dialkyl functions or aryl-ester functions instead of diester groups. Due to the challenges of preparing such substrates via diazo method, we made substrate 1aa bearing dialkyl functions upon the reduction of substrate 1a. Under the standard condition, 1aa proceeded smoothly to afford the corresponding chiral quaternary aldehyde 2aa with good regioselectivity and enantioselectivity (Figs. 5a, 5.2:1 rr and 80% ee), implying that the diester groups are not necessary for the excellent selectivity control. In a sharp contrast, only dominated linear products were observed using Rh-PPh3 under the identical conditions (Fig. 5b). More importantly, the racemic substrate 1ab that contains an aryl group and an ester group proceeded smoothly to yield chiral quaternary aldehyde with excellent enantioselectivity, regioselectivity and diastereoselectivity (96% ee, 7:1 dr, and >20:1 rr) via kinetic resolution process (Fig. 5c), implying that the remote substituents not on the C=C bonds also play an important role in the selectivity determination process.

Fig. 5: Control experiments and kinetic resolution reaction.
figure 5

a AHF of 1aa under standard condition; b AHF of 1aa under standard condition with PPh3; c Kinetic resolution AHF of (±) 1ab under standard condition.

Taking into account the decomposition of the reactants and simultaneously aiming for good conversion rates, the kinetic resolution reactions were conducted with shorter reaction times and lower syngas pressure. As shown in Fig. 6, the desired chiral product 2ab and 2ae were obtained with high yields and excellent enantioselectivities, and the enantio-enriched reactants (R)−1ab and (R)−1ae were recovered with good yields. Substituents on the Ar1 or Ar2 ring were also amenable, providing the chiral aldehydes with high yields and satisfactory enantioselectivities (2ac, 2ad, 2af-2ai). Meanwhile, the recovered chiral cyclopropenes were obtained with high yields and good to excellent enantioselectivities, demonstrating the robust capability of this catalytic system in asymmetric kinetic resolution reactions.

Fig. 6: Scope of kinetic resolution reactions.
figure 6

All reactions were performed on 0.2 mmol in 1 mL toluene with 1 mol % Rh(acac)(CO)2 and 2 mol % ligand under 2.5/2.5 bar (CO/H2) at 60 °C for 10 h. Isolated yield. Ee was determined by HPLC analysis using a chiral stationary phase. aPerformed at CO/H2 = 1/1 bar, 60 °C for 12 h. s = ln[(1 − conv)(1 − ee1)]/ln[(1 − conv)(1 + ee1)], conv = ee1/(ee1 + ee2).

To demonstrate the practical synthetic utility, a scale-up reaction of 1o was conducted in the presence of 0.2 mol% catalyst, successfully providing the chiral cyclopropane 2o with high yield (Fig. 7). The enantioselectivity was detected after a Wittig reaction and the corresponding α,β-unsaturated ester 5 was obtained in high yield76,77. When the chiral aldehyde 2o was treated with LiAlH4, all of the ester groups and the formyl group were reduced, affording the chiral triols 6 with moderate yield. The formyl group could also be transformed into chiral acid 7 with high yield and excellent ee. Upon using the reduction reagent NaBH3CN, cascade reduction/lactonization and reductive amination/lactamization reactions proceeded smoothly, delivering the corresponding lactone 8 and lactam 9, respectively, with maintained enantioselectivity.

Fig. 7: Synthetic utility.
figure 7

Left part: scale-up reaction of 1o; Right part: transformations of the versatile chiral aldehydes.

Improved understanding of the mechanism of AHF reactions is crucial for broadening the applicability of such methods. Previous experimental studies have extensively proposed the alkene insertion as the most likely selectivity-determining event of these reactions4,78,79,80. However, trisubstituted alkenes as a novel candidate in AHF reaction demands additional attention. We have undertaken a detailed computational study to shed light on the molecular factors responsible for the observed regio- and enantio-selectivities. In particular, we focus on the transition states of the alkene insertion to Rh−H active catalyst. For ease of comprehension, a schematic diagram of the key transition states leading to 2a and 3a as the final products is provided in Fig. 8. Transition state (TS) for the alkene insertion, initiated by the hydride addition to unsubstituted carbon, leading to R and S enantiomers of 2a, are respectively denoted as TSHAU(re) and TSHAU(si) indicating the corresponding prochiral face as involved. Similarly, TS for the hydride addition to the substituted carbon leading to product 3a is designated as TSHAS. Here, subscripts HAU and HAS respectively imply the hydride addition to the unsubstituted and substituted end of the alkene.

Fig. 8: Explanation of regio- and enantioselectivity based on DFT calculation.
figure 8

a A simplified representation of the enantiocontrolling (TSHAU(re) and TSHAU(si)) and regio-controlling (TSHAS) transition states for the alkene insertion that respectively lead the formation of 2a (R and S enantiomers) and 3a in the AHF reaction of 1a. b Composite depictions rendered using a combination of space-filling and stick models derived from the enantiocontrolling transition states. Distances are in Å. All hydrogen atoms, except for the Rh-H, are omitted for improved clarity. A more detailed representation of these transition states, see Fig. 9.

The initial step of the catalytic cycle involves the coordination of alkene to the Rh−H active catalyst [RhH(CO)(L1)] and the subsequent insertion into the Rh−H bond triggered by the hydride addition to alkene. This event is irreversible and is likely to be the enantio-controlling step as evident from the previous computational studies on this front63,65,81,82. The alkene insertion step can also be considered as the regio-controlling step since the corresponding TS pertains to the Rh-hydride addition to the substrate (1a), which could exhibit a preference toward one of the two alkene carbon atoms. For instance, the hydride addition to the unsubstituted carbon of the alkene leads to the major products 2a (R/S), followed by the formylation at the phenyl-bearing end of the alkene (Fig. 8a). Conversely, the hydride addition to the phenyl-substituted carbon leads to the formation of the minor product 3a. Since the phenyl-bearing end of alkene is prochiral, two stereochemically distinct modes of alkene insertion en route to product 2a are considered. For example, the alkene insertion to the re prochiral face of alkene via TSHAU(re) can give access to the R enantiomer of 2a, while the S enantiomer arises through the si face insertion (TSHAU(si)). Therefore, the origin of enantioselectivity should stem from the stereoelectronic features in the alkene insertion transition states.

The important geometric details of the enantio- and regio-controlling transition states and the corresponding relative Gibbs free energies are provided in Fig. 8. The Gibbs free energy of the TS for the alkene insertion through the re prochiral face (TSHAU(re)), responsible for the formation of the R enantiomer of the product, can be noted as 2.3 kcal/mol lower than TSHAU(si) that corresponds to the S product. The predicted energy difference suggests an ee of 95%, which is in good agreement with the experimentally reported value of 99% in favor of the R enantiomer. On the other hand, the energy of TSHAS that leads to product 3a is found to be 4.3 kcal/mol higher than the most preferred TSHAU(re), again in good agreement with the experimentally observed regiochemical preference for product 2a over 3a.

Given the high significance of the energy difference between the enantiocontrolling TSHAU(re) and TSHAU(si), we have first analysed the stereoelectronic factors that contribute to this vital energy difference by using the activation strain analysis (refer to Supplementary Fig. 1 and Supplementary Table 1 for more details)83. The activation energy is considered as the sum of destabilizing distortion (ΔEd) and stabilizing interaction (ΔEi) energies in these transition states. Here, we employ relative ΔEd and ΔEi, calculated with respect to the lower energy TS, for ease of discussion. The higher energy TSHAU(si) is found to experience 2.7 kcal/mol higher than that in the most preferred TSHAU(re). The higher ΔEd in TSHAU(si) primarily arises due to an increased distortion in the catalyst as compared to the distortion in the alkene moiety. This distortion in TSHAU(si) is discernible by inspecting the encircled region as shown in Fig. 8b, where the geometric disposition of the lower half of the chiral ligand is notably lifted upwards toward the alkene, causing a large deviation from its native trigonal bipyramidal geometry in the pre-catalyst. The extent of distortion also manifests in the Rh-P2-C-N dihedral angle, which gets narrower in TSHAU(si) (86°) as compared to the undistorted pre-catalyst (131°). On the other hand, the distortion of the ligand backbone in the lower energy TSHAU(re) is much lower, as evident from the Rh-P2-C-N dihedral angle of 120°, which is closer to that in the native geometry of the pre-catalyst. Interestingly, TSHAU(re) enjoys a slightly improved stabilizing interaction (ΔEi of 0.9 kcal/mol) than that in TSHAU(si) . The net effect of ΔEd and ΔEi results in 3.7 kcal/mol lower activation strain in TSHAU(re) than that in TSHAU(si) rendering it the most preferred stereochemical mode of alkene insertion.

Additional analyses of the enantiocontrolling transition states are carried out by examining the noncovalent interactions (NCIs) as deciphered using the atoms in molecule (AIM) topological analysis63,64,84,85,86,87,88,89 In the present case, two distinct sets of NCIs are most relevant; (i) catalyst-alkene interactions and (ii) interactions within the catalyst as shown in Fig. 9. We have identified the presence of seven distinct NCIs (C−H···Cl, C−H···π, C−H···O, l.p···π, π···π, C−H···N, and C−H···H−C) in TSHAU(re) and TSHAU(si). Given that these transition states exhibit several NCIs belonging to sets (i) and (ii), a comprehensive mapping all in a single figure could affect the clarity and understanding of the associated discussion. Therefore, in Fig. 9, we place the full geometry of the TS at the center while the crucial regions labeled as A, B,…, D, carrying significant NCIs within the TSs are shown magnified along the periphery. Magnified views provide finer details of the NCIs in the relevant regions and help bring out the differences in the NCIs between TSHAU(re) and TSHAU(si). For improved clarity, the orientation within a magnified region had to be realigned as compared to that in the full geometry shown at the center.

Fig. 9: Optimized geometries for the enantiocontrolling TSs (TSHAU(re) and TSHAU(si)) and regioicontrolling TSHAS highlighting the important NCIs.
figure 9

Key interatomic distances (in Å) and the corresponding electron densities at the bond critical points (ρbcp × 10–2) for various NCIs are given in parentheses. Hydrogen atoms not involved in significant interactions are omitted for improved clarity. a NCIs in enantiocontrolling TSHAU(re); b NCIs in enantiocontrolling TSHAU(si); c NCIs in regioicontrolling TSHAS.

As discussed before, the fundamental differences between TSHAU(re) and TSHAU(si) are in the prochiral faces of the alkene and the associated geometric disposition of the chiral ligand, as shown using the space-filling representation in Fig. 8b. This disparity seems to have implications to both sets of NCIs, i.e., Cat-Rh···alkene as well as those within the catalyst. To begin, we focus on the key differences in the NCIs as displayed in Fig. 9a, b, and their relative efficiencies. One of the major differential NCIs is noted within the chiral catalyst. For instance, eight C–H···Cl contacts denoted as b3, b4, b5, b6, b7, b8, b9, and b10 are found in regions B and C within the chiral catalyst in the case of the lower energy TSHAU(re), whereas only two such interactions (b'2 and b'3 in regions B’ and C’) are present in the higher energy TSHAU(si) (refer to Supplementary Table 3 for details). The l.p(Cl)···π interactions involving the amido region within the chiral catalyst (d2, d3, d4, and d5 in regions B and C and d'2, d'3, d'4, d'5, and d'6 in regions B’ and C’) are slightly more efficient in TSHAU(re), according to the ρ values at the respective bond critical points found in our AIM analysis.

Among the other NCIs within the chiral catalyst, a l.p(O)···π (c2 in region C) and a C–H···O (e3 in region C) is unique to the lower energy TSHAU(re). C–H···N interaction is also found to be better in TSHAU(re) than in TSHAU(si). Both TSs also feature interactions such as π···π and C–HHC within the catalyst backbone, interestingly with modestly higher efficiency in the higher energy TSHAU(si). In the cat-Rh···alkene interactions (regions A’, D’, and A), one C–H···O, three C–H···π, a C–H···Cl, a l.p(O)···π, and a π···π interaction are common to both the TSs. However, the lower energy TS has a unique l.p(Cl)···π as well as an additional C–H···Cl interaction. In contrast, the higher energy TS exhibits two unique C–HHC, two additional C–H···π, and an additional l.p(O)···π interactions. However, the total number of NCIs is found to be more in the case of the lower energy TSHAU(re). In addition, the cumulative strength of such NCIs in TSHAU(re) is about 9.2 kcal/mol more than that in TSHAU(si) (refer to Supplementary Table 3 for details). The predicted energy difference between TSHAU(re) and TSHAU(si) (2.3 kcal/mol) can therefore be considered to gain from the lower net distortion in TSHAU(re) (by 2.7 kcal/mol as obtained from the activation strain analysis) than that in TSHAU(si). These factors together render TSHAU(re) energetically more favored.

To probe the molecular basis of our computed energy difference between the regiocontrolling transition states TSHAU(re) and TSHAS, we have performed activation strain analysis and have examined the NCIs. The higher energy TSHAS is found to bear a greater degree of distortion as compared to its lower energy regioisomeric alternative for the alkene migration via TSHAU(re). The relative distortion energy (ΔEd) in the reacting partners is 15.7 kcal/mol higher in TSHAS as compared to that in TSHAU(re), wherein the catalyst experiences a significantly more distortion than the alkene moiety. Although the stabilizing ΔEi in the higher energy TSHAS is more effective by 8.6 kcal/mol than in the lower energy TSHAU(re), it is found to be insufficient to offset the destabilization caused by the distortion in the catalyst. The net effect of these opposing factors renders TSHAU(re) the most favored transition state for the alkene migration leading to the observed regiochemical preference.

Additional analyses of the alternative regiochemical mode of alkene insertion to the phenyl substituted C2 carbon of cyclopropene via a higher energy TSHAU were performed by examining the NCIs present therein. As shown in Fig. 9a, c, both TSHAU(re) and TSHAS have plenty of NCIs such as C–HCl, C–HO, C–HHC, C–Hπ, l.pπ, and ππ interactions. Here, we focus on the major differences in the NCIs and their relative efficiencies. Interestingly, in the higher energy TSHAS, two unique C–HH–C interactions are found between the ortho hydrogen of the phenyl group of the alkene and the hydrogen atoms of the aryl group of the catalyst (g2′′ and g3′′ in region A′′). Two unique ππ interactions between the phenyl group of the alkene and the aryl group of the catalyst (f1′′ and f2′′ in region A′′) are noted in TSHAS. These interactions are absent in the lower energy counterpart. However, the total number of NCIs between the catalyst and the alkene is more in the lower energy TSHAU(re) than in TSHAS. In considering the interactions within the chiral catalyst, two C–HN interactions (h1 and h2 in region B) and one ππ (f2 in region C) interaction are exclusive to the lower energy TSHAU(re), while a C–Hπ (a3′′ in region C′′) interaction is unique to the higher energy TSHAS. The difference in the number of NCIs between the catalyst and the substrate as well as those within the catalyst itself, is found to influence the regioselectivity. The cumulative strength of all the important NCIs is −126.1 kcal/mol in TSHAU(re) while that in TSHAS is −101.4 kcal/mol (refer to Supplementary Table 3 for details). In other words, more effective NCIs contribute to the additional stabilization of the lower energy TSHAU(re), as compared to TSHAS, thereby rendering TSHAU(re), more preferred over the regiochemical alternative TSHAS.

Discussion

In conclusion, we report an efficient and straightforward method for preparing quaternary tetrasubstituted cyclopropanes via rhodium-catalyzed highly chemo-, regio- and enantio-selective hydroformylation of challenging trisubstituted cyclopropenes. By employing an easy L1 ligand, the side reactions can be effectively suppressed, and useful activity and selectivities are achieved. This methodology exhibits good substrate compatibility and excellent stereocontrol, providing desired molecules with high yields and excellent enantioselectivities. Scale-up reactions and follow-up transformations demonstrated the practical utility of this methodology in organic synthetic chemistry. Preliminary mechanistic studies suggest the electron-with-drawing diester groups are not required for control over the challenging regioselectivity and enantioselectivity.

Our analysis of the regio- and enantio-selectivity determining transition states as obtained using the DFT computations suggests that the alkene insertion into the Rh-H bond of the catalyst, initiated by hydride addition to two distinct ends of the alkene, can account for the observed selectivities. The Gibbs free energy of the alkene insertion transition state for the hydride addition to the unsubstituted end of the alkene leading to the major product is found to be 4.3 kcal/mol lower than the TS for the hydride addition to the substituted end of alkene. Furthermore, in this regiochemically preferred pathway, alkene insertion to the re prochiral face of the alkene is found to be 2.3 kcal/mol energetically more preferred over that to the si face. These findings suggest that the predicted enantioselectivity of 95% in favour of the R product is consistent with the experimentally observed enantiomeric excess of 99%. The high energetic advantage for the insertion to the re prochiral face of alkene leading to the R enantiomer of the product could be traced to a lower distortion in the corresponding transition state complemented by relatively more effective noncovalent interactions within the catalyst backbone. The origin of regioselectivity is traced to a series of more effective noncovalent interactions, such as C–H···Cl, C–H···N and l.p(Cl)···π, within the catalyst backbone and between the catalyst and the alkene in the lower energy regioisomeric transition state. Further exploration of the construction of highly valuable chemicals via AHF and as well in-depth operation mechanisms is ongoing in our laboratory.

Methods

General procedure for Rh-catalyzed hydroformylation

In a glovebox filled with argon, to a 5 mL vial equipped with a magnetic bar was added L1 (0.004 mmol) and Rh(acac)(CO)2 (0.002 mmol in 1 mL toluene). After stirring for 10 min, the mixture was charged to substrate (0.2 mmol). The vial was transferred into an autoclave and taken out of the glovebox. The argon gas was replaced with hydrogen gas for three times, and then hydrogen (5 bar) and carbon monoxide (5 bar) were charged in sequence. The reaction mixture was stirred at 60 °C (oil bath) for 24 h. The reaction was cooled to room temperature and the pressure was carefully released in a well-ventilated hood. The solution was concentrated and the product was isolated by column chromatography, and the eluents were petroleum ether (PE) and ethyl acetate (EA), and the ratio of PE/EA was from 20/1 to 10/1 in order to obtain pure products. For HPLC analysis: The reaction solution was transferred into a solution of methyl (triphenylphosphoranylidene)-acetate (0.24 mmol) in 2 mL dichloromethane, and the mixture was stirred at room temperature for 2 h and the corresponding esters were isolated for HPLC analysis. The enantiomeric excesses of 2a-2g, 2j-2x’, 5, 7, 8, 9, 2aa-2ai, and (R)−1ab – (R)−1ai were determined by HPLC analysis using a chiral stationary phase.