Article Cite This: J. Org. Chem. XXXX, XXX, XXX−XXX
pubs.acs.org/joc
Mechanistic Study on Decarbonylative Phosphorylation of Aryl Amides by Nickel Catalysis Yuantai Xu,† Bing Wang,† Julong Jiang,† Haizhu Yu,*,‡ and Yao Fu*,† †
Downloaded via IDAHO STATE UNIV on July 22, 2019 at 09:40:37 (UTC). See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles.
Hefei National Laboratory for Physical Sciences at the Microscale, CAS Key Laboratory of Urban Pollutant Conversion, Anhui Province Key Laboratory of Biomass Clean Energy, iChEM, University of Science and Technology of China, Hefei, Anhui 230026, People’s Republic of China ‡ Department of Chemistry, Center for Atomic Engineering of Advanced Materials, Anhui Provence Key Laboratory of Chemistry for Inorganic/Organic Hybrid Functionalized Materials, Anhui University, Hefei, Anhui 230601, People’s Republic of China S Supporting Information *
ABSTRACT: The phosphorylation of amide represents an unprecedented environmentally friendly and easily achievable method to constitute C−P bonds in organic synthesis. In this study, the mechanisms for the nickel-catalyzed direct decarbonylative phosphorylation of amides recently reported by Szostak’s team were systematically studied with density functional theory calculations. The reaction mainly undergoes four steps: oxidative addition (rate-determining step), phosphorylation, decarbonylation, and reductive elimination. The structures of the substrate and Na2CO3 were found to be critical for the reaction efficiency. Substrates bearing electronwithdrawing groups like carbonyl groups near the amide bond facilitate the reaction by weakening the C−N bond, and Na2CO3 can not only neutralize the H atom in the phosphate ligand as an alkali but also activate the Ni−N bond through the coordination bond with the adjacent carbonyl of the amide group.
■
INTRODUCTION
Scheme 1. (a) Michaelis-Arbuzov Reaction; (b) Hirao Cross Coupling; (c) Michal Szostak’s Work: Amide Hirao Reaction
Organophosphorus compounds are ubiquitous in nature and are frequently used in biochemistry applications, such as cyclin-dependent-kinase/gluconeogenesis inhibitor1,2 and phosphonomycin.3 The traditional ways to create a C−P bond in organic synthesis are the Michaelis−Arbuzov reaction4 (Scheme 1a), the use of conventional Grignard or organolithium reagents, and the Hirao cross-coupling reaction.5−10 These reactions generate organophosphorus from halohydrocarbon. Meanwhile, the transition-metal-catalyzed C−P bond formation via the reactions starting from boronic acids,11 silanes,12 sulfides,13 diazonium salts,14 sulfonates,15−17 esters,18,19 and olefins20 have also been reported in recent years (Scheme 1b). Despite the great progress, these protocols remain suffer from the drawbacks such as toxic/expensive reagents, limited substituent tolerance, or unsatisfactory efficiency.4,21 Lately, Szostak’s group recently reported a catalytic phosphorylation via the palladium/nickel-catalyzed decarbonylative phosphorylation with amides22 (Scheme 1c). This reaction is the first example of the transition-metalcatalyzed direct phosphorylation of the commercially available and cheap amides,23 providing aryl phosphonates in high yields (within 15 h), and is tolerated for different functional groups. These advantages demonstrate the high potential of this reaction to the academic and industrial applications.24−30 © XXXX American Chemical Society
Received: April 8, 2019 Published: June 3, 2019 A
DOI: 10.1021/acs.joc.9b00962 J. Org. Chem. XXXX, XXX, XXX−XXX
Article
The Journal of Organic Chemistry
1 imaginary frequency) and to obtain the thermodynamic corrections of Gibbs free energy. For each transition state, an intrinsic reaction coordinate (IRC) analysis was conducted to verify the connection between the right reactant and product.42,43 On the basis of the optimized structures, the M06 functional44,45 (with the same basis set as the geometry optimization) and IEFPCM solvation model (solvent = 1,4-dioxane) were used for single-point energy calculations. All energies in this study are reported in kcal/mol. In Szostak’s study, different palladium and nickel catalysts were tested, and Ni(dppp)Cl 2 (dppp is short for 1,3-bis(diphenylphosphino)propane) shows higher functional group tolerance and higher yield within 15 h. Therefore, we used the Ni(dppp)Cl2-catalyzed reaction of 1-benzoylpiperidine-2,6-dione with diethyl phosphite in the presence Na2CO3 as the modeling reaction (Scheme 3). According to Szostak’s experiment,31 the yield of this reaction can reach as high as 95% under 160 °C.
To understand the mechanism of the reaction in Scheme 1c, Szostak et al. suggested a four-step pathway (I in Scheme 2) Scheme 2. Two Possible Reaction Mechanisms
Scheme 3. Modeling Reaction Used in This Study
■
RESULTS AND DISCUSSION In this study, the energetics of pathway I, in which phosphorylation occurs before the decarbonylation step, was first examined. The results are given in Figure 1. We started with examining the predominant form of the nickel catalyst. According to the recent studies,46−48 Ni(II) could be easily reduced to Ni(0) by phosphine ligands, and therefore, Ni(0) complex Cat was chosen as the starting point in our calculations. Either solvent (1,4-dioxane) or substrate (1-benzoylpiperidine-2,6-dione) could coordinate with the Ni center in Cat. The coordination of one or two dioxane molecules can generate Cat-D and Cat-2D, respectively. Meanwhile, the substrate could coordinate with the metal center via either the end-on carbonyl coordination49,50 (CatS1) or η6-benzene coordination51 (Cat-S2). According to the calculation results, all of these coordination processes are exergonic. The relative energy of Cat-D is slightly lower than that of Cat-2D, predominantly due to the higher steric hindrance in the latter case. Compared to the solventcoordinated Cat-D/Cat-2D, the relative energy of Cat-S1/ Cat-S2 is significantly lower. The lower energy of Cat-S2 than Cat-S1 is mainly caused by the stronger π-backdonation from the Ni(3d) orbital to the benzene group.52,53 The formation of Cat-S2 from Cat is exergonic by 21.1 kcal/mol. From Cat-S2, oxidative addition occurs via the triangular transition state TS1 to generate the four-coordinated Ni(II) intermediate I-2. The oxidative addition needs to overcome an energy barrier of 26.0 kcal/mol and is exergonic by 13.0 kcal/mol. We anticipated that the two carbonyl groups around the amide N atom weaken the C−N bond and facilitate the oxidative addition. This assumption is supported by the relatively higher energy barrier when the two intracyclic carbonyl groups were replaced with two methylenes (28.3 vs 26.0 kcal/mol, see the SI). Of note, we also considered the oxidative addition with the coordination of the carbonyl of the glutarimide, but this
consisting of oxidative addition (A + 1 → B), hydrogen transfer (B + 2 → C + 3), decarbonylation (C → D + CO), and a final reductive elimination (D → A + 4) step. The key step was suggested to be the oxidative addition step (also known as the metal insertion into the C−N bond step). This mechanism provides a general mechanism for the concerned reaction, but the details, especially regarding the H transfer and decarbonylation steps, remain ambiguous. For example, the details for the H-transfer from the phosphine group to the amine group are still unclear, and direct H-transfer31 or Na2CO3-mediated H-transfer32 are both possible pathways. Meanwhile, according to Shi’s study on the decarbonylative borylation reaction of amides using similar nickel catalysis,33 the direct decarbonylation on Ni(II) might also occur. In this context, an alternative pathway (pathway II in Scheme 2) in which the decarbonylation might occur prior to the H-transfer step could not be excluded. Herein, we conducted density functional theory (DFT) calculations on Szostak’s reaction to elucidate the overall reaction mechanism, the rate-determining step, and the key parameters in determining the reaction efficiency.
■
COMPUTATIONAL DETAILS
All calculations were conducted with the Gaussian 09 package34 using the B3LYP density functional35,36 for structure optimization. A mixed basis set, pseudopotential of SDD37 for Ni, and the total electron basis set of 6-311G(d,p) for the other atoms were used in the unrestricted geometry optimization. The IEFPCM solvation model38−41 (with 1,4dioxane solvent, corresponding to Szostak’s experiments14) was used for all of the calculations to incorporate the solvent effect. The frequency analysis was calculated at the same level of theory with geometry optimization to confirm the nature of the saddle points (local minimum with 0 imaginary frequency and transition state with B
DOI: 10.1021/acs.joc.9b00962 J. Org. Chem. XXXX, XXX, XXX−XXX
Article
The Journal of Organic Chemistry
Figure 1. Energy profiles for pathway I (the Gibbs free energies are given in kcal/mol).
To avoid the structural distortion during the direct Htransfer process, we then examined the energy demand for an Na2CO3-mediated H-transfer process. As shown in Figure 1, Na2CO3 can neutralize the acidic H atom in the phosphate ligand of I-3 as an alkali, generating the sodium salt I-16. The overall coordination environment of the Ni center in I-16 is similar to that in I-3, avoiding the strong structural distortion as shown in TS-2. After the removal of H+, the two O(Et) atoms of the phosphate ligand locate away from the benzene of dppp ligand (Figure 2), reducing the steric hindrance among ligands. As a result, this process (I-3 → I-16) is exergonic by 30.6 kcal/mol. As shown in Figure 2, the P−O(Me) and the carbonyl group of the glutarimide ligand constitute complex Ni−O interactions with the Na+ in I-16. Specifically, the CO → Na+ interaction remarkably weakens the conjugation of the amide bond and leads to the lengthened Ni−N bond (2.051 Å in I-16 vs 1.947 Å in I-3). This proposal is also supported by the natural bond orbital (NBO) charge analysis on I-16 and the formed I-21 (see Figure S2), as both the Ni and Na centers become less electron-deficient after the Ni−N bond dissociation step. Such structural characteristics facilitate the Na+mediated dissociation of the glutarimide group, which results in a vacant coordination site on I-21. The subsequent dissociation of the glutarimide sodium salt (I-17) on I-21 generates I-15 to prepare for the subsequent carboxylation step. Besides, additional calculations indicate that the other alkali metal cations, i.e., K+ and Cs+, might also show similar beneficial effect to the phosphorylation step (see Figure S6). From I-15, the decarbonylation occurs via the triangular transition state TS-3 to generate the four-coordinated Ni(II) compound I-5. This process is exergonic by 7.6 kcal/mol, with an energy barrier of only 2.4 kcal/mol. After that, the carbonyl ligand leaves in the form of carbon monoxide, generating I-7 as the product. This process is exergonic by 13.7 kcal/mol. The final reductive elimination on I-7 occurs via the triangular transition state TS-5, generating the η6-benzene-coordinated I8. The energy barrier of TS-5 is 17.7 kcal/mol, and this process is endergonic by 0.3 kcal/mol. Finally, the ligand exchange of organophosphorus 4 with the substrate 1 occurs to regenerate CAT-S2 and release the product. According to the calculation results in Figure 1, pathway I goes through four main steps: oxidative addition, Na2CO3mediated phosphorylation, decarbonylation, and reductive
pathway was found to be highly energy-demanding (see Figure S3). From I-2, the direct attack of phosphate to the Ni center is unlikely due to the high steric hindrance. Alternatively, the partial-dissociation of one arm of the diphosphine ligand might occur first54,55 to provide a vacant coordination site (I-9 in Figure 1). This Ni−P bond dissociation process is endergonic by 18.8 kcal/mol, and formed I-9 could easily undergo the coordination of the phosphate 2 to generate the intermediate I-3 (exergonic by 9.4 kcal/mol). From I-3, the possibility for the formation of phosphate via an 1,2-H transfer of the phosphate ligand was also examined but was excluded in view of the high-energy barrier (over 60 kcal/mol, see Figure S5). For the subsequent phosphorylation step, the direct Htransfer was first examined. This step occurs via the concerted transition state TS-2 through a five-membered ring (Figure 2).
Figure 2. Optimized structure of selected species in Figure 1. The bond distances and angles are given in angstroms and degrees, respectively.
However, the energy barrier of TS-2 is 60.3 kcal/mol, which is too high for this reaction to occur under the experimental conditions (200 °C). The high energy demand for the direct H-transfer might originate from the high steric hindrance among the ligands in TS-2. The Ni−P distance changes from 3.214 Å in I-3 to 2.712 Å in TS-2. In addition, the Ni−O−P and N−H−P angles in TS-2 are 53.3° and 176.5°, indicating a highly twisted pentagon transition state of direct H-transfer. According to these results, the direct H-transfer via TS-2 was excluded. In addition, we also calculated a stepwise direct Htransfer using the carbonyl as a hydrogen acceptor with a following 1,3-H transfer, but this mechanism needs to overcome the energy barrier of over 60 kcal/mol (see Figure S4). C
DOI: 10.1021/acs.joc.9b00962 J. Org. Chem. XXXX, XXX, XXX−XXX
Article
The Journal of Organic Chemistry
Figure 3. Energy profiles for the transformation of I-9 to I-7 via pathway II (the Gibbs free energies are given in kcal/mol).
elimination. The phosphorylation, decarbonylation, and reductive elimination are all facile (with energy barrier
