Hydrogenation of unsaturated bonds is dominated by transition metal catalysis. Compared with transition metals, the use of other metals is less explored, especially so for the s-block elements despite their ready availability and low cost. Here, we show that group 2 metal amides (M[N(SiMe3)2]2, M = Mg, Ca, Sr, Ba) unexpectedly catalyse the hydrogenation of aldimines with H2 at 80 °C and a remarkably low H2 pressure of 1–6 bar. Conversion rates increase with metal size: Mg < Ca < Sr < Ba (for Ba, quantitative conversion is reached within 15 min). The key to this catalysis is the unanticipated formation of metal hydride species by deprotonation of H2 (pK a ≈ 49) with a weak base M[N(SiMe3)2]2 (HN(SiMe3)2: pK a ≈ 25.8). Density functional theory calculations suggest that the most favourable pathway indeed involves metal hydride intermediates. The efficient alkaline earth metal-catalysed hydrogenation of imines with molecular hydrogen at remarkably low pressure provides an attractive alternative to transition metal catalysis.
With molecular hydrogen being one of the cleanest reducing agents, catalytic hydrogenation using the more noble transition metals is among the most studied of all chemical processes1. Increasing social pressure towards a sustainable society, however, dictates replacement of costly, and often harmful, precious metals by more abundant first-row transition metals or even biocompatible redox inactive main group metals2,3,4,5,6. The alkaline earth metal calcium does not possess partially filled d orbitals for substrate activation, but has recently shown catalytic activities in the hydrogenation of C=C double bonds with molecular H2 (ref. 7). Although restricted to conjugated C=C bonds, this example strikingly broke the dogma that transition metals are needed for alkene hydrogenation. This was followed by the development of metal-free frustrated Lewis pair (FLP) catalysts8,9,10,11 and, most recently, cationic calcium hydride catalysts that are also able to hydrogenate unactivated alkenes12.
Figure 1a shows a working hypothesis for styrene hydrogenation with a dibenzylcalcium catalyst (CaBn2)7. The first step is the generation of a calcium hydride species, for which ample precedence exists13,14,15,16,17. Further reaction with H2 may cause precipitation of insoluble (CaH2) n , but catalyst loss is partly prevented by aggregation to soluble but undefined Ca x Bn y H z species. Despite a lack of d orbitals, alkene activation proceeds through a weak electrostatic calcium–alkene interaction, recently shown to be of importance in calcium catalysis18. The benzylic calcium intermediate formed after insertion may, after successive styrene insertions, form polystyrene19, but high H2 pressure (20–100 bar) can prevent this side reaction by promoting σ-bond metathesis. The latter step in the cycle is, like the initiation reaction, formally a deprotonation of H2 by a resonance-stabilized benzylic carbanion. Considering the high pK a of H2 (≈ 49)20, this reaction seemed questionable. Stoichiometric conversions of model systems, however, underscored the feasibility of this pathway7. Independent theoretical calculations illustrate that the final σ-bond metathesis step is indeed highly endergonic: Gibbs free energy of activation ΔG ‡ (60 °C, 20 bar) = 25.7 kcal mol−1 (ref. 21).
As the highly atom-efficient catalytic reduction of imines by H2 received much less attention than alkene or ketone hydrogenation22,23,24, it remained an important question whether calcium-catalysed hydrogenation can be extended to imine reduction. Current state-of-the-art imine hydrogenation catalysts can be divided into four categories that vary in terms of substrate activation and nucleophilic power (Fig. 2a–d). Figure 2a shows organometallic metal hydrides that rely on hydride nucleophilicity. Apart from few early transition metal catalysts (Ti25, lanthanides26), these are generally based on late transition metals (Rh, Ir)22. The aluminium hydride compound (iso-butyl)2AlH is an odd example of a main group metal catalyst that can be used, but needs harsh conditions (100 bar H2, 100 °C, 24 h)27. Recently, a zinc hydride catalyst was introduced but similarly high pressures need to be applied (70–100 bar)28,29. Figure 2b shows metal-free organocatalysts. These depend on strong imine activation by protonation, which allows use of weaker nucleophiles such as the Hantzsch ester30,31,32,33. Recently, strong Brønsted-acid activation and organometallic hydride catalysis have been combined34,35. Shvo- or Noyori-type bifunctional catalysts are shown in Fig. 2c, these simultaneously deliver protic Hδ+ and hydridic Hδˉ building blocks and are considered to be intermediate to the catalysts shown in Fig. 2a and b 36. And finally, highly Lewis-acidic borane catalysts, shown in Fig. 2d, can, in interplay with imine Lewis bases, break the H–H bond by FLP activation, but substrates are restricted to bulky imines that do not form B–N adducts37,38,39,40.
A hypothetical mechanism for calcium-catalysed reduction of imines (Fig. 1b) could be similar to that for alkene hydrogenation (Fig. 1a). However, given the fact that the last step in the catalytic cycle (σ-bond metathesis) represents a deprotonation of H2, such catalysis was expected to be unlikely: the calcium amide intermediate would be an even more inferior base for H2 deprotonation than a benzylic carbanion. We were therefore surprised to find that aldimines can be efficiently hydrogenated by Ca[N(SiMe3)2]2, a simple, easily accessible, calcium amide complex41,42. This stimulated us to investigate the scope and working principles of alkaline earth metal-catalysed aldimine hydrogenation. Whereas existing methodology uses either noble metals and/or high pressures, we here introduce simple alkaline earth metal amide catalysts (Mg, Ca, Sr, Ba) that can hydrogenate imines of various substitution patterns at low pressure (down to 1 bar) and relatively mild conditions (80 °C), while showing fair functional group tolerance.
Catalytic imine hydrogenation
The catalysts M[N(SiMe3)2]2 (abbreviated MN″2, M = Mg, Ca, Sr, Ba)41,42 can be conveniently prepared on a large scale from the metal and are the most widely used precursors in group 2 metal chemistry. Using the imine PhC(H)=NtBu as a benchmark substrate, the influence of temperature, pressure and metal on the catalytic conversion was tested. Substrate scope was evaluated by systematic variation of substituents (Table 1).
Using CaN″2, the substrate PhC(H)=NtBu was cleanly hydrogenated at a remarkably low pressure of 6 bar and a temperature of 80 °C (Table 1, entry 1). Doubling the H2 pressure decreased the reaction time (entry 2), but the effect of pressure was small. Full conversion was still observed at pressures as low as 1 bar (entry 3). Increase of temperature to 120 °C had a strong accelerating effect (entry 4). The high catalyst concentration could be halved to 5 mol% (entry 5), but this resulted in strongly increased reaction times (lower catalyst loadings are even more ineffective). The inability to lower the catalyst concentration is not due to catalyst decomposition by hydrolysis. Although organocalcium complexes are substantially more sensitive towards hydrolysis than most transition metal catalysts, often catalyst (or initiator) concentrations down to 0.1 mol% are feasible5,19,43. We attribute the need for higher catalyst concentrations to formation of larger aggregates (vide infra).
Screening of various aldimine substrates shows that catalytic conversion is governed by the steric and electronic effects of the substituents. Presuming a mechanism as shown in Fig. 1b, the substitution pattern can either affect the electrophilicity of the imine (insertion step) or the basicity of the intermediate amide responsible for H2 deprotonation (σ-bond metathesis step). Decreasing steric hindrance at N by replacing tBu for iPr had a pronounced accelerating effect (entry 6), but replacing tBu for Ph significantly increased the reaction time (entry 7). This is likely due to the stabilization of the intermediate anion by resonance (Fig. 3a), thus making it a less efficient base for H2 deprotonation than PhCH2(tBu)N‒. Replacing Ph for a mesityl substituent (Mes) accelerated conversion (entry 8). The ortho Me substituents in the intermediate anion enforce a perpendicular orientation of the Mes substituent (Fig. 3b). This partially shuts off delocalization of negative charge, thus increasing its basicity. Changing the Ph group at the imine C for a ferrocenyl unit (Fc) reduced reaction times (entry 9), whereas a Mes substituent at C increased reaction times (entry 10). The latter may be explained by the perpendicular orientation of the Mes substituent (Fig. 3c), blocking the imine C for hydride attack. Reducing the electrophilicity at C by a bulky, electron-releasing, tBu substituent (Fig. 3d) similarly increased reaction times (entry 11). Slight reduction of electron-releasing and steric effects by replacing the tBu group for iPr gave a significant acceleration (entry 12). Attempts to hydrogenate Ph2C=NtBu were not successful and currently the method is limited to aldimines.
The catalyst was also tested for functional group tolerance. Whereas the presence of a nitro substituent in the para position of the Ph group fully inhibited catalysis, (p-Cl-C6H4)C(H)=NtBu and (p-MeO-C6H4)C(H)=NtBu were efficiently hydrogenated (entries 13 and 14).
All test reactions were run in the moderately polar solvent benzene (relative polarities for the solvents pentane (0.009), benzene (0.111) and tetrahydrofuran (THF, 0.207) were taken from ref. 44). Use of CaN″2∙(THF)2 instead of the THF-free catalyst CaN″2, elongated the reaction time from 3.0 to 8.5 h (entry 15). In agreement with this decelerating effect of polar co-solvents, reactions in pure THF showed after 24 h no conversion, whereas a solvent change from benzene to the less polar hexane accelerated the conversion (entry 16). Since catalytic runs in pure imine as a polar solvent gave no conversion, the decelerating effects of polar solvents cannot entirely be explained by blocking of imine–calcium coordination.
The free amine HN(SiMe3)2 that is formed during catalyst initiation also had a decelerating effect. Addition of HN(SiMe3)2 lowered the conversion rate dramatically (entry 17). This is likely due to the fact that the catalyst initiation in Fig. 1b is reversible: CaN″2 + H2 ⇄ N″CaH + N″H. This led us to probe the strongly basic dibenzylcalcium catalyst Ca(DMAT)2∙(THF)2 (1, Fig. 4)45, which after initiation produces DMAT–H, which is inert to deprotonation by the calcium hydride intermediate (DMAT: 2-dimethylamino-α-trimethylsilyl-benzyl). Despite presence of two polar THF ligands, extremely fast conversion was observed (<15 min; entry 18), even after reducing catalyst loading (entry 19).
Finally, we evaluated the influence of the alkaline earth metal. Complex MgN″2 is also catalytically active but needed a longer reaction time (entry 20), which was reduced significantly at higher temperature (entry 21). The heavier group 2 metal Sr, however, gave a noticeable acceleration (entries 22 and 23). The Ba amide catalyst BaN″2∙(THF)2 was, despite the presence of THF co-solvent, the most active (entries 24 and 25). Increased activity along the series Mg < Ca < Sr < Ba suggests that bond ionicity in the catalyst is more important than the Lewis acidity of its metal. Replacing the alkaline earth metal for the alkali metal Li (entry 26), however, gave only 6% conversion under standard conditions after 24 h, showing that the basicity of the N″ anion is not the sole criterion. Increase of metal size, Li+ < Na+ < K+, slightly accelerated conversion (entries 26–28), but the higher Lewis acidity of the 2+ alkaline earth cations is essential for efficient conversion.
The exact composition of the catalyst(s) for imine hydrogenation is unclear. The precatalyst CaN″2 forms a complex with (E)-PhC(H)=NtBu (crystal structures are shown in Supplementary Figs. 7–9), but no further reaction is observed on heating to 80 °C. Pressurizing a benzene solution of CaN″2 with H2 (80 °C, 6 bar), however, led to formation of N″H and undefined hydride species. This unexpected deprotonation of H2 (pK a ≈ 49)20 by a weak amide base (N″H: pK a = 25.8)46 is at first sight thermodynamically unfavourable, but the driving force for formation of hydrides could originate from exothermic aggregation to larger clusters and finally (CaH2)∞. Possible existence of a variety of larger aggregated species, Ca x N″ y H z , with molecular weights up to 7,500 is supported by DOSY NMR (Supplementary Fig. 6). Very recently, we reacted MN″2 (M = Ca or Sr) with PhSiH3 in the presence of MeN(CH2CH2NMe2)2 (PMDTA) and isolated the first well-defined larger Ca and Sr hydride clusters with molecular weights up to 1,500: M6H9N″3∙(PMDTA)3 (2, Fig. 4)47. Heating these complexes to 100 °C gave decomposition to larger undefined M x N″ y H z clusters. Similar formation of large metal hydride aggregates during imine hydrogenation would drastically lower the catalyst concentration, thus explaining the need for higher catalyst loadings.
The undefined nature of the catalyst precludes an accurate mechanistic study. Therefore, we performed an extensive density functional theory (DFT) study on the model catalyst N″CaH using the M06 hybrid functional, which has been shown to be reasonably accurate for main group thermochemistry, kinetics and noncovalent interactions48. Gibbs free energies at 80 °C and 6 bar (M06/6-311++G(d,p)//M06/6-31++G(d,p)) were corrected for benzene solvent effects using the polarizable continuum model with radii and nonelectrostatic terms from Truhlar and co-worker’s solvation model based on density (SMD)49.
Three conceptually different mechanisms have been investigated. (1) Calcium hydride catalysis (mechanism A, Fig. 5), which involves precoordination and activation of the imine followed by nucleophilic hydride attack. (2) Bifunctional catalysis (mechanism B, Fig. 5), in which a transient calcium hydride complex with a coordinated N″H ligand reduces the imine by concerted protic and hydridic attack (cf. Noyori- or Shvo-type catalysts). (3) Metal hydride-free catalysis (mechanism C, Fig. 5) through a six-membered-ring transition state similar to that proposed by Berkessel for ketone hydrogenation by KOtBu under harsh conditions (20 mol% cat, 200 °C, >100 bar H2)50. This mechanism would circumvent generation of a high-energy metal hydride species.
The calcium hydride cycle (route A in Fig. 5) starts with the thermodynamically unfavourable deprotonation of H2 by the weak CaN″2 base. The first amide–hydride exchange through transition state A1* (+18.7 kcal mol−1) is endergonic by 11.9 kcal mol−1, while exchange of the second amide ligand is even less favourable (A10, +27.3 kcal mol−1). Full cycles for both hydride catalysts, N″CaH and CaH2, have been calculated, but the lower energy route for N″CaH is more likely. The highest point along the pathway is the transition state for hydride to imine addition (A5*) with an energy of 8.1 kcal mol−1 relative to N″CaH (A3). It should be noted that the anion PhCH2(tBu)N− is a stronger base than N″−, which makes the deprotonation of H2 to form the final product (A6 + H2 → A7* → A3 + product) slightly easier (+18.1 kcal mol−1) and less endergonic (+9.1 kcal mol−1) than the initiation CaN″2 + H2 → A1* → A3 + NH″. The last step cycles back to A3, thus giving an overall reaction energy of ‒11.4 kcal mol−1.
Hydrogenation according to a bifunctional mechanism (route B in Fig. 5) starts with the first step of the former cycle, that is, endergonic formation of intermediate A2 (+10.8 kcal mol–1). A2 features hydridic and protic H atoms for concerted attack at the imine, but all attempts to optimize a six-membered-ring transition state led to B1* in which only the hydridic H attacks the imine (Fig. 5d; B1*: the N″H∙∙∙N(imine) distance of 3.350 Å is long). This is due to a strong affinity of Ca2+ for the imine N (Ca–N = 2.610 Å) that effectively competes with the N″H∙∙∙N(imine) interaction. Rather than a concerted bifunctional mechanism, a second step is necessary (protonation of the amide via B4*). It should be noted that the transition state for bifunctional ketone reduction according to Noyori shows very advanced hydridic attack, while the proton transfer is at an early stage51. Most recent calculations even conclude a true two-step mechanism in which, analogous to our findings, hydridic attack is followed by protonation52. Note that route B is connected to A: addition of amine N″H to transition state A5* gives B1*. As this process is endergonic by 5.7 kcal mol−1, pathway B is unlikely. The additional N″H also plays a role in final product formation by protonation of the anion PhCH2(tBu)N− along B3 → B4* → B5, a pathway that is easier than protonation of this amide by H2 (A6 → A7* → A8). The transfer A6 + N″H → B3 could therefore be an attractive low-energy alternative for route A; however, it cycles back to CaN″2 instead of to A3.
The metal hydride-free route (route C in Fig. 5) starts with the coordination of imine to CaN″2 to form the prereactive complex C1 (Fig. 5c). This complex, and similar Mg and Sr imine complexes, indeed could be experimentally observed. The DFT-optimized complex C1 shows a close fit with its crystal structure (Supplementary Fig. 8). However, all attempts to find the proposed six-ring transition state for H2 activation by a pincer-like interaction with the N″ anion and the positively charged imine C were unsuccessful. Instead, H2 deprotonation by the polarized Caδ+–N″δ‒ bond was found (C2*). This transition state shows a very long distance of 3.057 Å between the imine C and the hydridic Hδ‒ (Fig. 5d), which means there is essentially no interaction. The activation energy C1 → C2* → C3 is very high (25.6 kcal mol−1). Alternatively, C3 can be formed from intermediate A2 and react further through C4*, the transition state for hydride addition and the highest point along the route. Route C is also related to A: the transition states C2* and C4* are connected to A1* and A5* by bonding of an additional imine or N″H, respectively. Coordination of these ligands is, in both cases, endergonic, which makes pathway C unlikely. Additional ligands create steric stress and are generally endergonic: for example, A2 + imine → C3 (+9.0 kcal mol−1). The crowded coordination sphere in C3 causes an extremely short H∙∙∙H distance of 1.803 Å between the protic N″Hδ+ and hydridic CaHδ‒. This makes loss of H2 via C3 → C2* → C1 with an activation barrier of only 1.8 kcal mol−1 more likely than imine reduction via C4*, which needs an activation energy of 6.4 kcal mol−1. Note that transition state C4* is very similar to B1*, differing only in its three-dimensional arrangement. The pathways B and C are therefore also connected to each other. The metal hydride-free six-membered-ring pathway C is therefore non-existent. It is of interest to note that recent modelling of Berkessel’s K-catalysed ketone hydrogenation led to a similar multistep route52.
The computational study can be summarized by claims (1)–(5). (1) As expected, conversion of CaN″2 with H2 to give the reactive hydride species A3 is highly endergonic (+11.9 kcal mol−1) and the step with the highest activation energy (18.7 kcal mol−1). It is, however, likely that aggregation into larger hydride bridged clusters, Ca x N″ y H z , is exergonic and allows for this transformation. (2) Bifunctional species containing hydridic (Ca)Hδ‒ and protic (N″)Hδ+ (for example, A2) can be formed, but a Noyori-type mechanism (B) with concerted hydridic/protic attack is non-existent. The high affinity of Ca2+ for the imine N enforces a two-step mechanism. (3) Attempts to model a six-membered-ring transition state for the metal hydride-free route C were fruitless. Owing to the high affinity of Ca2+ for the hydridic Hδ‒, a transition state with a four-membered ring (C2*) is preferred. Proposed concerted pathway C is therefore split in a two-step pathway in which calcium hydride formation (C2*) is followed by hydride-imine addition (C4*). (4) The proposed pathways A, B and C interconnect to form a complex web of possibilities. Pathways B and C are related to A by additionally coordinated N″H or imine ligands. Coordination of an additional Lewis base generally leads to steric crowding and higher energies. All three mechanisms can therefore be reduced to pathway A. (5) Taking CaN″2 as the starting point, the highest single step needs an activation energy of 18.7 kcal mol−1 (A1*) and the highest point along the pathway is 20.0 kcal mol−1 (A5*). Under catalytic conditions, A3 is formed after each cycle. Taking this complex as the starting point, an overall activation free energy of +8.2 kcal mol−1 is estimated, however, the side reaction A3 + N″H → A2 → A1* → CaN″2 with a lower activation free energy of +6.8 kcal mol−1 should be taken into account. This escape route explains the decelerating effect of N″H on catalysis and demonstrates why the dibenzylcalcium catalyst Ca(DMAT)2∙(THF)2 (1) is more effective. Analysing the energy profiles with the energetic span model53 gives an overall activation free energy of 22.3 kcal mol−1 for a cycle with a CaN″2 catalyst or 17.2 kcal mol−1 for N″H-free catalysis starting with A3 as the catalyst (Supplementary Fig. 11).
The calcium hydride pathway A was further confirmed by stoichiometric reactions with a recently introduced well-defined calcium hydride complex16 (3, Fig. 6). Complex 3 reacted smoothly with PhC(H)=NtBu to a mixed hydride-amide species (4, crystal structure in Supplementary Fig. 10), but, even under forced conditions, a second insertion is not observed. Pressurizing complex 4 with H2 gave PhCH2N(H)tBu and the original calcium hydride complex 3. This stoichiometric reactivity suggests catalytic activity, which indeed could be observed (Table 1, entries 29 and 30). The protecting spectator ligand prevents formation of larger calcium hydride clusters, thus enabling a lower catalyst concentration (2.5 mol%).
The reduction of a variety of imines with H2 can be catalysed by simple group 2 metal amides under relatively mild conditions (80 °C, 1–6 bar H2) and shows functional group tolerance to aromatic Cl and MeO substituents. Polar solvents inhibit conversion and the best results were obtained in apolar solvents such as hexane. The amine N″H, formed during catalyst initiation, inhibits imine hydrogenation, likely by reversibly reacting with the metal hydride catalyst. A more Brønsted-basic dibenzylcalcium catalyst gave extremely fast conversion (<15 min). Catalytic activities for MN″2 complexes increase with metal size: Mg < Ca < Sr < Ba; for Ba, conversion times <15 min have been observed. Although this may lead to the conclusion that basicity is more important than the metal’s Lewis acidity, the poor catalytic performance of alkali metal amide catalysts suggests that the higher Lewis acidity of the 2+ alkaline earth metal cation is crucial.
The catalyst for the calcium-mediated imine hydrogenation is an undefined aggregate of Ca2+, H− and N″− ions. Starting from CaN″2 as a model precatalyst, DFT calculations for three conceptually different pathways (A, B and C in Fig. 5) showed that metal hydride-free or Noyori-type mechanisms with a bifunctional Hδ¯/Hδ+ catalyst are not relevant. Most favourable is a metal hydride mechanism (pathway A), which according to the energetic span model has an overall activation free energy of 22.3 kcal mol−1. This mechanism finds support from stoichiometric conversions on a well-defined model system. The bottle-neck is the highly endergonic formation of the catalytically active metal hydride complex N″CaH, which in the presence of acidic N″H may be deactivated by H2 loss.
The herein introduced hydrogenation of aldimines with simple group 2 metal amide catalysts at remarkably low hydrogen pressures illustrates that the fast-growing field of early main group metal catalysis can be an attractive alternative to transition metal catalysis. We are currently targeting the more challenging ketimine hydrogenation and aim to control activities and selectivities by spectator ligands.
Under an atmosphere of pure nitrogen, the catalyst (0.05 mmol, 10 mol% or 0.025 mmol, 5 mol%) was dissolved in C6D6 (1.0 ml) in a miniature steel autoclave (15 ml) and the imine (0.5 mmol) was added to the solution. The autoclave was sealed tightly and the desired hydrogen pressure (1, 6 or 12 bar) was applied. While stirring with a stirring bar, the reaction mixture was quickly heated to the specified temperature (80 or 120 °C) in a heating block. Reaction times were determined by sampling the reaction mixture in 15 min intervals under nitrogen atmosphere and determining the conversion by 1H NMR measurement and integration of significant signals. As the reactions did not show by-products, conversions were estimated by determination of the product/substrate ratios. The Supplementary Methods includes the syntheses of all catalysts and substrates.
Preparative scale catalytic conversion
Under an atmosphere of pure nitrogen, MgN″2 (68.8 mg, 0.2 mmol) was dissolved in benzene (4 ml) in a glass autoclave and N-benzylidene-tert-butyl-amine (332.5 mg, 2.0 mmol) was added to the solution. The autoclave was sealed tightly and hydrogen pressure (6 bar) was applied. While stirring with a stirring bar, the reaction mixture was heated to 120 °C for 3 d. After complete conversion was confirmed by gas chromatography (GC) measurement, the mixture was hydrolysed with water (5 ml) followed by extraction with pentane (2 × 2 ml). The organic phase was separated and dried over MgSO4. Distillation under atmospheric pressure yielded the hydrogenated product, N-benzyl-tert-butyl-amine (isolated yield: 250.2 mg, 1.53 mmol, 77 %).
Stoichiometric reactions with a well-defined calcium hydride complex
The synthesis and structure of complex 3 (Fig. 6) have been reported previously16. Stoichiometric conversions of 3 with imine and subsequently with H2 (6 bar) were performed in a high-pressure NMR tube and were monitored by 1H NMR (see Supplementary Methods). The mixed amide-hydride intermediate 4 has been fully characterized by NMR methods, CHN analysis (see Supplementary Methods) and X-ray diffraction (see Supplementary Methods). The crystal structure of 4 is shown in Supplementary Fig. 10. Details for catalytic conversion with catalyst 3 can also be found in the Supplementary Methods.
All the chemical species along the reaction pathways A, B and C (Fig. 5) were fully optimized and characterized by harmonic vibrational frequency calculations, which also provided the zero-point energy and the thermodynamic contributions to the gas-phase enthalpy (ΔH) and Gibbs free energies (ΔG) for each species. For all optimizations, the M06 hybrid functional and a 6-31++G(d,p) basis set have been used. The M06 suite of density functionals has been shown to be fairly accurate for main group thermochemistry, kinetics and noncovalent interactions48. The Gibbs free energy correction were evaluated at 80 °C and 6 bar pressure. Accurate electronic energies and solvation energies were obtained by single point calculations at the M06/6-311++G(d,p) level using the polarizable continuum model with radii and nonelectrostatic terms from Truhlar and co-worker’s SMD model49. Gibbs free energy profiles (kcal mol−1) for the different pathways of calcium-catalysed imine hydrogenation are shown in Fig. 5. For further details see Supplementary Methods and Supplementary Data 1 (the XYZ coordinate file containing information on the geometries of all intermediates).
Data for the crystal structures reported in this paper have been deposited at the Cambridge Crystallographic Data Centre (CCDC) under the deposition numbers CCDC 1550247–1550250. Copies of these data can be obtained free of charge via www.ccdc.cam.ac.uk/data_request/cif. All other data are available within the paper and its Supplementary Information files, or from the corresponding authors on request.
Publisher’s note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
M.A. thanks the Fund for Scientific Research–Flanders (FWO-12F4416N) for a postdoctoral fellowship and the Free University of Brussels (VUB) for financial support. F.D.P. acknowledges the Research Foundation Flanders (FWO) and Strategic Research Program funding of the VUB. He also acknowledges the Francqui foundation for a position as ‘Francqui research professor’. S.H. acknowledges the Deutsche Forschungsgemeinschaft for financial support (HA 3218/7-1).
About this article
Nature Catalysis (2018)