Ya-Qian
Zhang‡
,
Xue-Yu
Han‡
,
Yue
Wu
,
Peng-Jia
Qi
,
Qing
Zhang
and
Qing-Wei
Zhang
*
Department of Chemistry, University of Science and Technology of China, Hefei 230026, China. E-mail: qingweiz@ustc.edu.cn
First published on 10th March 2022
The catalytic asymmetric synthesis of P-stereogenic phosphines is an efficient strategy to access structurally diverse chiral phosphines that could serve as organocatalysts and ligands to transition metals and motifs of antiviral drugs. Herein, we describe a Ni catalyzed highly regio and enantioselective hydrophosphinylation reaction of secondary phosphine oxides and enynes. This method afforded a plethora of alkenyl phosphine oxides which could serve as valuable precursors to bidentate ligands. A new type of mechanism was discovered by combined kinetic studies and density functional theory (DFT) calculations, which was opposed to the widely accepted Chalk–Harrod type mechanism. Notably, the alkene moiety which could serve as a directing group by coordinating with the Ni catalyst in the transition state, plays a vital role in determining the reactivity, regio and enantioselectivity.
![]() | ||
Fig. 1 P-Stereogenic phosphines and Ni-catalyzed H–[P] reactions. (a) Chalk–Harrod type mechanism, (b) initial result, (c) this work. |
However, the synthesis of P-chiral compounds is challenging.4 The absence of P-stereogenic phosphine precursors in nature has impeded direct access to chiral phosphines by chiral pool syntheses. As a result, catalytic asymmetric syntheses of chiral phosphines have emerged in the last two decades. Among them, the hydrophosphination/hydrophosphinylation reactions (H–P reactions) of secondary phosphines or their derivatives with alkenes or alkynes would be the most straightforward and atom economic strategy. However, the enantioselective reactions were achieved with limited success.5 In addition, the control of P-chirality was much more challenging than carbon chiralities as exemplified by Leung’s,6 Dong’s,7 Wang’s8 and Yin's9 studies, which achieved excellent enantioselectivities of carbon stereocenters but suffered from low d.r. when racemic secondary phosphine oxides were used.
On the other hand, the control of regioselectivity of the H–P reactions of unactivated alkenes or alkynes was elusive.10 Typically, transition metals were introduced as catalysts to achieve better regioselectivities compared with uncatalyzed ones. And the mechanism of transition metal catalyzed H–P reactions was thoroughly investigated to proceed via the Chalk–Harrod type mechanism, i.e. metal catalyzed H–P oxidative insertion, hydrometallation and reductive elimination.11 Ni was also a decent catalyst for H–P reactions which were recognized to comply with the Chalk–Harrod type mechanism.12 However, during our research, phenomena that contradicted the reported mechanism were observed and herein we disclose a new finding in a Ni-catalyzed regio and enantioselective hydrophosphinylation reaction, which was explicitly supported by detailed mechanistic studies combining experimental results and DFT calculations (Fig. 1c).
The scope examination of SPOs (Fig. 2) revealed that the reactivity and enantioselectivity of the reaction were significantly affected by the steric hindrance of the alkyl substituent attached (3ba–3ga). Generally, the bulkier the alkyl substituent, the less reactive the reaction and the higher is the ee. SPOs with phenyl primary alkyl and allyl substituents could all afford corresponding tertiary phosphine oxides 3aa–3ga with both high yields and ee (53–93% yield and 85–93% ee). The aryl or heteroaryl substituents on SPOs tested in Fig. 2 were well tolerated in this reaction, giving products 3ha–3na in moderate to high yields (53–90%) and excellent enantioselectivities (88–95% ee). Among them, the electron rich substrates were less reactive, affording 3ha and 3ia in 68% and 53% yield respectively with excellent enantioselectivities (92% and 95% ee). The reactivities of these SPOs roughly followed such a trend that the more electron donating the aryl substituent was, the less reactive the reaction was, and vice versa. This phenomenon is pivotal in elucidating the reaction mechanism and will be discussed further in the following part. In addition, the absolute configuration of product 3ka was identified by X-ray crystallography analysis (PR).
The scope of enynes was then examined with SPO 1j (Fig. 2). Aryl enynes with both electron-donating and electron-withdrawing groups, as well as alkyl enynes were all tolerated, providing products 3jb–3jv in moderate to high yields (24–96%) and excellent enantioselectivities (87–99% ee). Generally, the electron rich enynes reacted faster than the electron poor ones which was opposite to that of SPO. It is worth noting that arylbromide that could facilely undergo oxidative addition in the presence of Ni(0) survived the reaction conditions and gave product 3jl in both high yield (90%) and ee (95%). meta-Hydroxyl phenyl enyne with an acidic proton could also give the desired product 3jo during the reaction in 96% ee albeit in moderate yield (45%). Enynes with different alkyl groups could also undergo the hydrophosphinylation reaction, furnishing tertiary alkenyl phosphine oxides 3js–3jv with excellent ee (92–99%) and moderate to high yields (59–92%).
Additionally, this reaction could be accomplished at the 1 mmol-scale (Scheme 1) with comparable yield and enantioselectivity (3aa, 82% yield, and 89% ee). Further transformations of these P-chiral products were investigated (Scheme 1b and 1c). For example, reduction of 3aa was realized to afford the P-chiral phosphine–BH3 adduct 4 in 60% yield with 88% ee.13 The intramolecular olefin metathesis reaction of 3ga under the catalysis of the Hoveyda–Grubbs (II) catalyst proceeded smoothly at room temperature to afford the corresponding cyclization product 5 within 5 min without erosion of the P-chirality (92% yield and 88% ee).
![]() | ||
Scheme 1 Transformations of the TPO product. (a) 1 mmol scale reaction, (b) reduction of the phosphine oxide, (c) Grubbs reaction. |
To identify the structure of the catalyst resting state corresponding to the new peaks, three possible combinations that could lead to the formation of the complex were proposed, and the control experiments in benzene-d6 were carried out and monitored by 31P NMR. As shown in Fig. 3II, when the Ni(cod)2 catalyst was mixed with the (S,S)-BDPP ligand in both 1:
2 and 1
:
1 ratios in benzene-d6, NiL2 (B, δ 28.0 ppm) and Ni(cod)L (C, δ 34.5 ppm) were observed (Fig. 3IIa and b). Delightfully, the two doublets at δ 35.6 ppm and δ 34.6 ppm corresponding to the resting state of the catalyst in the reaction mixture were observed when Ni(cod)2, (S,S)-BDPP and 2a were introduced in a 1
:
1
:
1 ratio (Fig. 3IIc). The structure of the complex was assigned as π-complex A by both 31P and 1H NMR (see the ESI†).
Since secondary phosphine oxides were reported to participate in oxidative insertion with transition metals to form a metal hydride species, we also carried out the experiment of complex C and SPO 1a to verify the validity of the mechanism in our catalytic system. Thus, Ni(cod)2, (S,S)-BDPP and 1a in a 1:
1
:
1 ratio were stirred in benzene-d6 overnight. 31P NMR analysis revealed the presence of both B and C in the mixture (Fig. 3IId). However, no new peak that may be attributed to the coordination of 1a with Ni, or the possible oxidative insertion product P–Ni–H was observed. When enyne 2a was added to the mixture, π-complex A could be observed again accompanying the disappearance of both B and C. These experiments indicated that the formation of species B and C in the presence of cyclooctadiene (cod) was reversible and both species could be transformed to A in the presence of enyne 2a.
The electronic effects of the substituent on both enynes and SPOs were then examined (Fig. 4iv and v). It is interesting to find that although the reaction rate was irrelevant to the concentration of substrate 2a, there was a rate dependence on the electronic properties of enynes (iv). The more electron rich the aryl enyne, the higher the reaction rate, corresponding to a negative slope of the Hammett plot (ρ < 0). These seemingly contradictory results of both the experiments might be explained by the hypothesis that substrate 2a was transformed to an intermediate (A) which participated in the rate limiting step. The reaction also exhibited a rate dependence on the electronic properties of SPOs (v). Conversely, the more electron rich the SPO, the lower the acidity of the SPO as well as the reaction rates, corresponding to a positive slop of the Hammett plot (ρ > 0). This illustrated that both the catalyst resting state A and SPO were involved in the rate limiting step which was consistent with the above results (Fig. 4ii and iii). The inverse correlation of the electronic properties of the two substrates with the reactivity of the reaction i.e. the rate limiting protonation, suggested an acid–base neutralization type mechanism. A kinetic isotope effect experiment was then performed and the KIE from parallel reactions was determined to be 9.1 (vi, please see the ESI† for details). The large KIE was again supportive of such a mechanism, contradicting a rate limiting oxidative insertion or reduction elimination process which usually exhibits a smaller KIE.
![]() | ||
Fig. 5 Free energy profile for the new reaction pathway. Computed at the SMD(mesitylene)/M06/6-311++G(d,p)/SDD//B3LYP-D3(BJ)/6-31G(d)/LANL2DZ level. |
The catalyst resting state generated from Ni(cod)2 and methyl enyne which has been identified by NMR studies was used as the starting point. To our surprise, two isomers A1 and A2 with alkene and alkyne functional groups coordinating with nickel respectively were located with almost the same Gibbs free energy (0 vs. 0.05 kcal mol−1). This result contradicted the NMR studies above in which only one conformer was observed. Further NMR studies to elucidate the new catalyst resting state were carried out using alkyl enyne 2u with Ni(cod)2 as a comparison with the result above. Gratifyingly, two pairs of doublets in a roughly 1:
1 ratio which could be attributed to A1 and A2 were detected by 31P NMR (please see the ESI†), which was in perfect match with the DFT calculations.
The key transition state of the reaction of A1 or A2 with 1b was then located with the phosphorus atom of the secondary phosphine oxide coordinated with the Ni centre. Two transition states that could lead to the obtained product were found with alkene (TS-1a) and the alkyne moiety (TS-1b) respectively coordinated with Ni. The calculated energy barrier of the key intramolecular protonation step viaTS-1a (18.8 kcal mol−1) vs.TS-1b (25.7 kcal mol−1) indicated that the reaction probably proceeded through the former pathway which was favored by 6.9 kcal mol−1. This result was consistent with the experimental study that enynes were much more reactive than alkynes and could react at a very low temperature (−20 °C). As a comparison, the regioisomer could only be generated by transition state TS-1-regio in which the alkyne moiety was coordinated with Ni with an activation barrier of 25.8 kcal mol−1. This result strongly supported the excellent regioselectivity of the reaction.
In transition state TS-1a, the Ni centre and the hydroxyl group resided superficially with the enyne. As a result, simultaneous cis protonation and metalation occurred to afford Ni(II) intermediate INT-2 specifically. Reductive elimination of INT-2via transition state TS-2 (13.6 kcal mol−1) followed by ligand exchange could afford the final product and regenerate the catalyst resting state. Overall, intramolecular protonation served as the rate limiting step, which was also consistent with the kinetic studies. In addition, the oxidative insertion of SPO to Ni was also calculated to exhibit a high energy barrier (37.1 kcal mol−1), again consistent with the above experimental and calculation results (please see the ESI† for details).
The enantioselectivity could be also rationalized by the DFT calculations (please see the ESI† for details).14 The transition state of both secondary phosphine oxides (R)-1b and (S)-1b was located. The conformers with the lowest energy of TS-1a and TS-1a-S are shown in Fig. 6 with the former being more favored by 3.0 kcal mol−1 after correction. The ee was predicted to be 99% which matched with the experimental results of alkyl enynes. In summary, the cis, regio and enantioselectivity were well rationalized by DFT calculations which is fully in accordance with the experimental results. Obviously, the enyne coordinated with Ni(0) through a double bond was the key to both the high reactivity and enantioselectivity of the reaction.
![]() | ||
Fig. 6 Geometries and corrected relative free energies at −20 °C of enantioselective protonation transition states TS-1a and TS-1a-S. |
Footnotes |
† Electronic supplementary information (ESI) available. CCDC 2056434. For ESI and crystallographic data in CIF or other electronic format see DOI: 10.1039/d2sc00091a |
‡ These authors contributed equally to this work. |
This journal is © The Royal Society of Chemistry 2022 |