Computational Prediction of Ammonia-Borane Dehydrocoupling and Transfer Hydrogenation of Ketones and Imines Catalyzed by SCS Nickel Pincer Complexes

Inspired by the catalytic mechanism and active site structure of lactate racemase, three scorpion-like SCS nickel pincer complexes were proposed as potential catalysts for transfer hydrogenation of ketones and imines with ammonia-borane (AB) as the hydrogen source. Density functional theory calculations reveal a stepwise hydride and proton transfer mechanism for the dehydrocoupling of AB and hydrogenation of N-methylacetonimine, and a concerted proton-coupled hydride transfer process for hydrogenation of acetone, acetophenone, and 3-methyl-2-butanone. Among all proposed Ni complexes, the one with symmetric NH2 group on both arms of the SCS pincer ligand has the lowest free energy barrier of 15.0 kcal/mol for dehydrogenation of AB, as well as total free energy barriers of 17.8, 18.2, 18.0, and 18.6 kcal/mol for hydrogenation of acetone, N-methylacetonimine, acetophenone, and 3-methyl-2-butanone, respectively.


INTRODUCTION
Compared with direct hydrogenation, transfer hydrogenation (TH) avoids the use of hazardous molecular hydrogen and high pressure equipment by adopting low-cost and safe hydrogen provider compounds (Gladiali and Alberico, 2006;Ikariya and Blacker, 2007). Transition-metal-catalyzed TH has attracted increasing attentions in pharmaceutical, agrochemical, fragrance and other fine chemical industries as a powerful and practical way for the production of valuable chiral alcohols and amines (Blaser et al., 2003(Blaser et al., , 2007Klingler, 2007;Saudan, 2007;Hansen et al., 2009).
In contrast to the encouraging progress archived in iron catalysts, ATH of ketones catalyzed by cobalt complexes have rather low enantioselectivities (Morris, 2015). Only a few Ni catalysts have been reported so far (Hamada et al., 2008;Hibino et al., 2009;Dong et al., 2012;Yang et al., 2014;Xu et al., 2015). In 2008, Hamada and co-workers (Hamada et al., 2008) applied nickel-bisphosphine complexes to catalyze asymmetric hydrogenation of α-amino-β-keto ester hydrochlorides and achieved high diastereo-and enantioselectivities (88-93% ee) for the production of anti-β-hydroxy-α-amino esters. They also used the same catalyst for asymmetric hydrogenation of substituted aromatic α-aminoketone hydrochlorides to FIGURE 2 | Free energy profile for the dehydrocoupling of AB and TH of acetone catalyzed by 1 A .
Frontiers in Chemistry | www.frontiersin.org produce β-aminoalcohols and found excellent diastereo-and enantioselectivities (Hibino et al., 2009). In 2012, Dong et al. (2012) reported Ni(II) complexes chelated by PNO ligands for catalytic ATH of aromatic ketones with 2-propanol as the hydrogen source and obtained optical alcohols up to 84% ee under mild conditions. In 2014, Yang et al. (2014) reported a SCHEME 2 | 1 A catalyzed TH of N-methylisopropylamine with stepwise proton and hydride transfers. Frontiers in Chemistry | www.frontiersin.org highly active Ni(OAc) 2 /Binapine catalyst for the synthesis of αand β-amino acid trough ATH of olefins with formic acid. Later on, they reported a highly active NiCl 2 (dme)/Binapine catalysts for ATH of hydrazones and other ketimines (Xu et al., 2015).
In a typical TH process, low-cost compounds such as 2propanol (Haack et al., 1997) and formic acid (Fujii et al., 1996), are used as hydrogen donors. However, reversibility is a major drawback while 2-propanol is used as the hydrogen source. Excessive 2-propanol are usually required for the formation of alcohol. Formic acid releases stable CO 2 after its dehydrogenation and does not suffer from equilibrium problems but can only be used for a limited range of stable complexes. Compared to alcohols, amine-boranes are easy to handle, irreversible, and potentially recyclable hydrogen donors (Nixon et al., 2011). Therefore, amine-boranes, especially ammonia-borane (AB), are promising to serve as solid hydrogen surrogates in TH reactions (Nixon et al., 2011). In recent years, Pagano et al. (2015) reported cobalt catalyzed dehydrocoupling of AB and TH of alkenes and alkynes. Fu et al. (2016), Shao et al. (2016), and Ai et al. (2019) reported the first cobaltcatalyzed selective TH of alkynes, and the only example of cobalt-catalyzed TH of nitriles with dehydrocoupling of AB under mild conditions. Yang et al. (2010Yang et al. ( , 2011a recently reported direct TH of polarized C=N, C=C, and C=O bonds in imines, olefins, ketones, and aldehydes from AB. Li et al. (2016Li et al. ( , 2017, Meng et al. (2018) reported frustrated Lewis pair catalyzed ATH of imines with AB and ATH of 2,3-disubstituted quinoxalines with AB. Ding et al. (2017) reported TH of Nheterocycles from AB prompted by B(C 6 F 5 ) 3 . Korytiakova et al. (2017) reported Cu(I) catalyzed transfer semihydrogenation of alkyne and conjugate transfer hydrogenation of enoates with the dehydrocoupling of AB. Barrios-Francisco and García (2010) reported semihydrogenation of alkynes from AB catalyzed by Ni(0) complexes. Chong et al. (2014) reported 1,3,2diazaphospholenes-catalyzed metal-free TH of N=N double bond using AB. Although some iron and nickel catalysts for acceptorless dehydrogenation of AB have also been developed (Keaton et al., 2007;Yang and Hall, 2008;Zimmerman et al., 2009a,b;Vogt et al., 2011;Bhunya et al., 2016;Lunsford et al., 2016;Rossin and Peruzzini, 2016;Chakraborty et al., 2017;Coles et al., 2017), effective base metal catalyst for TH of ketones using AB under mild conditions was rarely reported.
Inspired by the catalytic mechanism and the active site structure of lactate racemase (Desguin et al., 2015), as well as previously reported SCS nickel pincer complexes (Meguro et al., 2008;Xu et al., 2017), we recently proposed and computationally predicted several promising scorpion-like SCS nickel pincer complexes for catalytic lactate racemization and ATH of 1acetonaphthone based on density functional theory (DFT) calculations (Qiu and Yang, 2017;Qiu et al., 2019). In this paper, we further examined three SCS nickel pincer complexes as potential catalysts for fast dehydrocoupling of AB and TH of ketones and imines. Figure 1 shows our previously proposed scorpion-like SCS nickel pincer complexes 1 A , 1 B , and 1 C with imidazole tails. 1 A and 1 B have symmetric arms but different sizes of amino groups in the SCS ligand, which are similar to the Ni complexes with pincer SCS ligands synthesized by Hu and co-workers (Xu et al., 2017). 1 C is a mimic of the active site of lactate racemase with a carbonyl arm in the SCS pincer ligand.

RESULTS AND DISCUSSION
Scheme 1 is the proposed reaction cycle for the dehydrocoupling of AB and hydrogenation of acetone to 2propanol catalyzed by 1 A . Figure 2 reports the corresponding free energy profile. The optimized structures of key intermediates and transition states in the reaction are displayed in Figure 3. The reaction cycle, free energy profile, and key structures of the TH of N-methylacetonimine catalyzed by 1 A are shown in Scheme 2, Figures 4, 5, respectively. The transfer hydrogenation reaction starts with dehydrocoupling of AB. When an AB molecule approaches 1 A , a hydride on borane could be transferred to the sp 2 carbon coordinated to nickel through transition state TS 2,3−A ( G = 15.0 kcal/mol) and forms an unstable intermediate 3 A . An ammonia proton then transfers to the imidazole nitrogen through transition state TS 3,4−A ( G = 13.9 kcal/mol). The exchange of acetone and H 2 NBH 2 in 4 A forms a 7.2 kcal/mol stable intermediate 6 A1 . Then the proton and hydride obtained from AB transfer from the pincer ligand in 6 A1 to acetone in one-step through TS 6,7−A1 (Figure 3) with a free energy barrier of 17.8 kcal/mol for the formation of 2-propanol.
Different from the hydrogenation of acetone, the proton and hydride are transferred from the pincer ligand in 6 A2 to N-methylacetonimine in a stepwise way with two transition states, TS 6,7−A2 and TS 7,8−A2 . The total free energy barrier   of the hydrogenation of N-methylacetonimine is 18.2 kcal/mol (7 A2 → TS 7,8−A2 ). Table 1 lists calculated free energy barriers of the hydrogenations of acetone and N-methylacetonimine ( G acetone and G N−methylacetonimine ) catalyzed by 1 A , 1 B , and 1 C . The difference of those relative free energies are less than 4 kcal/mol, which indicates a rather weak steric effect with different functional groups in the pincer ligand. Figures 6, 7 report free energy profiles for 1 A catalyzed TH of acetophenone to 1-phenyl-ethanol and TH of 3-methyl-2-butanone to 3-methyl-2-butanol, respectively. Some key structures in those catalytic reactions are displayed in Figure 8.
The AB dehydrocoupling process has the same free energy barrier of 15.0 kcal/mol in Figures 6, 7. The hydrogenation of acetophenone goes through a concerted one-step proton and hydride transfer transition state TS 6,7−A3 (Figure 8). Although TS 6,7−A3 is 17.7 kcal/mol higher than 7 A3, the total free energy barrier of this catalytic TH reaction is 18.0 kcal/mol (7 A3 → TS 2,3−A ) after considering the barrier of AB dehydrogenation.
The THs of acetophenone and 3-methyl-2-butanone catalyzed by 1 A , 1 B , and 1 C have similar mechanisms but slightly different energy barriers, which are listed in Table 2. We can see the free energy barriers of 1 B and 1 C are higher because of the stronger steric effects of dimethyl groups in them.
It is worth to note that the role of the imidazole groups in those scorpion-like SCS nickel pincer complexes are proton reservoirs facilitating proton transfer in the dehydrocoupling of AB and hydrogenation of C=O and C=N bonds. The ethylene group connecting the pyridinium ring and the imidazole group ensures the adjustability of the imidazole group's position for the hydrogenation of different ketones and imines. The substituents on the arms of the SCS ligand can slightly influence the reaction barriers through their steric effects. We believe the stepwise hydride and proton transfers in hydrogenation of 3-methyl-2butanone are caused by the weak polarity of the C=N bond in it.

CONCLUSIONS
In summary, we computationally examined three scorpion-like SCS nickel pincer complexes, 1 A , 1 B , and 1 C , with different steric effects as potential catalysts for catalytic TH of ketones and imines. Our calculations reveal stepwise hydride and proton transfer processes for the dehydrocoupling of AB and hydrogenation of imine, and a proton coupled hydride transfer process for hydrogenation of ketones. Among three examined Ni complexes, 1 A with symmetric NH 2 groups in the pincer ligand has the lowest free energy barriers of 17.8, 18.2, 18.0, and 18.6 kcal/mol for transfer hydrogenations of acetone, Nmethylacetonimine, acetophenone, and 3-methyl-2-butanone, respectively. 1 B and 1 C have slightly higher barriers for the same reactions because of their stronger steric effects. Such low barriers and exothermicities shown in calculated free energy profiles indicate that those (SCS)Ni pincer complexes are promising catalyst candidates for efficient TH of ketones and amines under mild conditions. The steric effects of the substituents on the arms of the pincer ligand have very weak influence on the energy barriers of the catalytic reactions. Our computational predictions not only provide prototypical base metal catalysts for dehydrocoupling of AB and transfer hydrogenation of ketones and imines, but also shed a light for further development of cost-effective catalysts for the hydrogenation of polarized double bonds. Further design of base metal complexes with SCS pincer ligands for more hydrogenation and dehydrogenation reactions is underway.

COMPUTATIONAL DETAILS
The Gaussian 09 suite of ab initio programs (Frisch et al., 2010) was employed to perform all DFT calculations at the ultrafine (99,590) numerical integration level for the ωB97X-D (Chai and Head-Gordon, 2008) functional with Stuttgart relativistic effective core potential (ECP10MDF) basis set (Martin and Sundermann, 2001) and all-electron 6-31+G(d,p) basis set (Hehre et al., 1972;Hariharan and Pople, 1973;Francl et al., 1982) for Ni and all other atoms, respectively. Without other specific instruction, all structures in this paper were fully optimized in THF by using the integral equation formalism polarizable continuum model (IEFPCM) (Tomasi et al., 2005) with SMD (Marenich et al., 2009) atomic radii solvent corrections. The ground states of all structures were confirmed as singlet through comparison with optimized high-spin analogs. Thermal corrections were considered under 298.15 K and 1 atm pressure through frequency calculations using the same method on optimized structures. The optimized structures were confirmed to have no imaginary vibrational mode for intermediates and only one imaginary vibrational mode for each transition state, which was further confirmed by intrinsic reaction coordinate (IRC) calculations to ensure proper stationary points were connected. The 3D molecular structures displayed in the text were drawn by using the JIMP2 program (Manson et al., 2006). We also evaluated the performance of various density functionals and the influence of solvent effect corrections for this Ni catalytic system. The calculation results and discussions are provided in the Supplementary Materials (Tables S1 and S2).

DATA AVAILABILITY
All datasets generated for this study are included in the manuscript and/or the Supplementary Files.

AUTHOR CONTRIBUTIONS
XY proposed the computational catalyst design project and the catalytic transfer hydrogenation reactions. BQ designed and computed all of the catalysts with the help of XY. BQ and WW wrote the paper with the help of XY.

FUNDING
This work was supported by the National Natural Science Foundation of China (21873107, 21673250). We are also grateful to Institute of Chemistry Chinese Academy of Sciences for funding.