Non‐Pincer‐Type Manganese Complexes as Efficient Catalysts for the Hydrogenation of Esters

Abstract

Catalytic hydrogenation of carboxylic acid esters is essential for the green production of pharmaceuticals, fragrances, and fine chemicals. Herein, we report the efficient hydrogenation of esters with manganese catalysts based on simple bidentate aminophosphine ligands. Monoligated Mn PN complexes are particularly active for the conversion of esters into the corresponding alcohols at Mn concentrations as low as 0.2 mol % in the presence of sub‐stoichiometric amounts of KO^tBu base.

The reduction of polar carbonyl moieties is a fundamental organic transformation important for the production of a wide variety of bulk‐ and fine chemicals, such as biofuels, fragrances, and pharmaceuticals. Catalytic processes employing H₂ as the reductant represent an atom‐efficient and sustainable alternative to conventional stoichiometric approaches.1 To date a wide range of versatile and highly active homogeneous ester hydrogenation catalysts based on Ru,2 Os,3 and Ir4 have been described. Driven by economic and environmental considerations, recent efforts have focused on the replacement of the noble‐metal component in such catalysts by cheaper, more abundant, and non‐toxic metals.5 Among these, manganese can be regarded as one of the most desirable candidates in view of its low price, rich chemistry, and exceptional biocompatibility.6 Yet, most examples of non‐noble metal homogeneous hydrogenation catalysts are based on Fe7 and Co,8 while the respective catalytic chemistry of Mn was not known until very recently. In early 2016 Milstein and co‐workers described the first Mn^I‐based catalyst A for the dehydrogenative coupling of alcohols and amines (Scheme 1).9a Later, Kirchner and co‐workers showed that this reaction can also be catalyzed by a related Mn^I PNP pincer complex.9b Shortly afterwards, the groups of Beller10 and Kempe11 independently reported the hydrogenation of ketones with pincer catalysts B and C. Complex B is also active in the reduction of nitriles and aldehydes. Reduction of less‐reactive ester substrates remains a challenge for Mn catalysts with only two examples reported to date. Beller and co‐workers described aliphatic Mn^I PNP‐pincer catalyst D that converts esters into alcohols under basic conditions at 2 mol % catalyst loading (110 °C/30 bar H₂/24 h).12 Milstein and co‐workers reported that lutidine‐derived Mn^I PNN‐pincer catalyst E is active at 1 mol %, but requires addition of KH as the base (100 °C/20 bar H₂/50 h).13 Despite the impressive progress witnessed in recent years in catalytic hydrogenations with non‐noble‐metal complexes, even the most active examples are efficient only at relatively high catalyst loading of 1–3 mol %, significantly limiting their utility as practical alternatives to the more active Ru‐based systems.14 Herein we report the catalytic hydrogenation of esters with three novel non‐pincer‐type Mn PN complexes, based on simple and easily accessible bidentate aminophosphine ligands. They show good performance at an unprecedented loading of only 0.2 mol %, bringing Mn‐catalyzed hydrogenation a step closer to practical implementation.

Scheme 1.

Mn‐based (de)hydrogenation catalysts.

The use of P,N ligands for Ru‐catalyzed ester hydrogenation was first reported by Saudan et al.15 We prepared complexes 1 to 3 by reaction of Mn(CO)₅Br with 1 or 2 equivalents of the corresponding P,N ligand in toluene at 100 °C for 24 h. The isolated complexes were fully characterized by ¹H/³¹P‐NMR, ESI‐MS, FTIR, elemental analysis, and single‐crystal X‐ray structure analysis (see Supporting Information). Single‐crystal X‐ray structure determination revealed the cis‐coordination of the N‐donor groups of the P,N ligands and CO ligands in 1, with the two phosphine moieties bound trans to each other (Figure 1). Their chemical equivalence was also detected in solution by ³¹P NMR, revealing a single resonance for 1 at δ=79.3 ppm. Complex 2 contains a single P,N ligand. The amine and Br⁻ in 2 are bound in a cis fashion, providing a favorable environment for heterolytic H₂ activation across the Mn–N moiety.16

Figure 1.

ORTEP diagrams of 1 (left) and 2 (right). Thermal ellipsoids are set at 30 % probability. Hydrogen atoms have been omitted for clarity.

Complexes 1–3 are active catalysts for ester hydrogenation. Table 1 summarizes the results of the initial catalytic tests using methyl benzoate as a model substrate. Monoligated complex 2 was found to be considerably more active than 1 and 3 (Table 1, entries 1–3). This is remarkable as the related Ru‐PN catalyst is biligated.15 Reaction at 80–100 °C gave similar benzyl alcohol (BnOH) yields, while the yield decreased substantially at higher temperatures owing to formation of methyl benzyl ether (Table 1, entries 2,4,5). Increasing the amount of KO^tBu led to improved yields (Table 1, entries 6–8). Ultimately, quantitative BnOH yield was obtained with 0.75 equivalents of KO^tBu relative to the substrate (Table 1, entry 8).

Table 1.

Hydrogenation of methyl benzoate with 1–3.^[a]

Entry	Catalyst	KOtBu [mol %]	T [°C]	Conv. [%]	YBnOH [%]17
1	1	10	100	43	24
2	2	10	100	75	66
3	3	10	100	13	3
4	2	10	80	74	65
5	2	10	120	57	43
6	2	25	100	86	80
7	2	50	100	96	91
8	2	75	100	99	98

[a] Conditions: 1 mmol methyl benzoate, 10–75 mol % KO^tBu, 1.0 mol % Mn, 2 mL THF, 80–120 °C, 50 bar H₂, 20 h. Yield determined by GC.

After full conversion was achieved, we sought to optimize crucial process parameters such as solvents, bases, reaction temperature and H₂ pressure to enable use of 2 at reduced catalyst loading. With 0.5 mol % of 2 in THF a BnOH yield of 87 % could be achieved in just 3 h. Importantly, 2 could also be formed in situ without significant loss of activity, thus eliminating the need for catalyst isolation (Table S1 in the Supporting Information). Mercury poisoning did not evidence inhibition, suggesting the homogeneous nature of catalysis with 2 (Table S1).18 Replacement of THF for 1,4‐dioxane resulted in a higher product yield, while the use of 2‐methyl‐THF and MTBE led to inferior performance (Table S2). KO^tBu was found to be the superior base for the current catalytic system (Table S3). An increase in temperature and reduction in H₂ pressure resulted in lower BnOH yields (Table S4).

Next, we expanded the scope of the substrates and further decreased the catalyst loading to 0.2 mol %. Under the optimized conditions, 2 was able to convert aromatic and aliphatic esters into their corresponding alcohols in good to excellent yields (Scheme 2). Reduction of hexanoate esters A1–A3 led to good yields of 1‐hexanol with hexyl hexanoate as the only by‐product. Interestingly, more sterically hindered esters (A4–A6) were almost quantitatively hydrogenated, whereas these are typically more difficult to reduce than their methyl and ethyl analogues.1 Aromatic benzoate esters with varied steric bulk or electronic properties were all hydrogenated to benzyl alcohol in high yield (B1–B4). Similar to aliphatic esters, the reduction of bulky tert‐butyl benzoate was more efficient than the less‐sterically hindered substrates. Hydrogenation of functionalized esters B5 and B6 gave high yields of the corresponding alcohols with the functional group being preserved and only trace amount of the methyl ether side products detected by GC‐MS. Hydrogenation of unsaturated esters with 2 was fully chemoselective for substrates with the C=C bond distant from the ester moiety, such as fatty acid methyl esters C1 and C2. Methyl cinnamate (C3), however, was fully converted into hydrocinnamyl alcohol. No products associated with the Claisen condensation were observed for the enolizable substrates.

Scheme 2.

Hydrogenation of various esters with 2. Conditions: 1 mmol substrate, 75 mol % KO^tBu, 0.2 mol % 2, 2 mL 1,4‐dioxane, 100 °C, 50 bar H₂, 16 h. [a] 0.5 mol % 2, 6 h.

To get better insight into the effect of the base in catalysis with 2 we carried out additional catalytic tests using four different benzoate substrates at varied base concentration (Figure 2). For all substrates, the elevated base loading resulted in a higher product yield. The hydrogenation of methyl‐ and ethyl benzoates was more sensitive to changes in the base concentration than for the tert‐butyl‐ and benzyl benzoate substrates. We attribute this to catalyst inhibition by the short‐chain alcohols produced in the reaction. This effect is in line with the lower activity achieved with KOMe and KOEt bases (Table S3). Product inhibition via metal‐alkoxide formation is well‐known for P,N‐type complex catalysts and is consistent with both the lower observed rates for methyl‐ and ethyl esters as well as the increased TON at reduced catalyst loading.19

Figure 2.

Effect of ester alkoxy group and KO^tBu amount on the degree of hydrogenation (equal to sum of benzyl alcohol, methyl benzyl ether, and $1/2$ benzyl benzoate yields).

Dedicated kinetic experiments were next carried out to further study the role of the base (Figure 3).21 Near‐complete hydrogenation was achieved with 0.75 equiv. KO^tBu, while in the presence of 0.1 equiv. base the reaction progress was limited to around 20 %. Remarkably, catalytic activity could be instantaneously restored upon addition of 0.65 equiv. KO^tBu. Regardless of the base loading sequence, nearly identical initial rates of about 1100 h⁻¹ were observed (see Figure S14). This is consistent with our hypothesis on Mn‐alkoxide inhibition, which upon reaction with KO^tBu convert into the catalytically active manganese amide. A similar mechanism of in situ catalyst regeneration has been proposed previously for related Ru‐based catalysts.20

Figure 3.

Kinetic traces of methyl benzoate hydrogenation with 2. Conditions: 15 mmol methyl benzoate, 10–75 mol % KO^tBu, 0.5 mol % 2, 28 mL THF, 100 °C, 50 bar H₂.

Next, the reaction mechanism with 2 was studied by density functional theory (DFT) calculations at the PBE0/6‐311G(d)//6‐31G(d) level (Gaussian 09 D.01).22 Methyl acetate (MeOAc) was chosen as the model substrate. The proposed mechanism, along with the reaction and activation Gibbs free energies for elementary steps, ΔG°373K_solv, are summarized in Figure 4. Prior to the catalytic reaction, 2 is activated via a base‐assisted hydrogenolysis to produce hydrido complex I (see Supporting Information). The cycle starts with an exergonic complexation of MeOAc with I to give H‐bonded intermediate II, which then converts into an activated gem‐acetal III via a hydride attack with a free energy barrier of 97 kJ mol⁻¹. The addition of H₂ to III yields σ‐complex IV, which after hydrogenolysis produces CH₃OH and CH₃CHO. Methanol elimination gives VI, from which the final stage of the catalytic cycle, that is, aldehyde hydrogenation, proceeds. This step is significantly more favorable than the initial ester activation. The first hydride transfer is exergonic by −8 kJ mol⁻¹ and shows a free energy barrier of only 29 kJ mol⁻¹ (VI→VII). The resulting alkoxy anion is stabilized by a partial deprotonation of the NH₂‐moiety of the ligand, thereby resulting in a trigonal bipyramidal configuration of Mn in VII. The interaction with the basic ethoxide facilitates complexation with H₂ to form VIII that is followed by a barrierless and highly exergonic heterolytic dissociation to produce I. The overall free energy barrier for this alkoxide‐assisted catalyst regeneration is 59 kJ mol⁻¹, in which the major energy losses originate from the structural distortions upon the formation of σ‐H₂ complex VIII. The alternative path via ethanol elimination from VII followed by the metal–ligand cooperative H₂ activation shows a free energy barrier of about 100 kJ mol⁻¹.

Figure 4.

Proposed catalytic cycle for methyl acetate hydrogenation by H₂ and 2 (ΔG and G ^≠ stand for the reaction and activation Gibbs free energy changes in kJ mol⁻¹ at 373 K).

DFT calculations also reveal a competing side‐path for the decomposition of III, resulting in CH₃CHO elimination and the formation of a stable Mn‐alkoxide complex (see Supporting Information). From this point, the formation of I requires a base‐assisted hydrogenolysis similar to that proposed for the activation of pre‐catalyst 2. This provides additional support for our proposal on catalyst inhibition by stable Mn‐alkoxide resting states. In line with the experimental results, the hydrogenolysis of the bulkier Mn‐O^tBu adduct shows a much lower energy barrier than Mn‐OMe (89 vs. 106 kJ mol⁻¹, respectively).

In summary, we have synthesized and fully characterized three novel Mn P,N ligand complexes, of which monoligated complex 2 is a highly active catalyst for the hydrogenation of aliphatic and aromatic esters. Considering the high catalytic performance and the simple and straightforward preparation, complex 2 holds a great promise as a cheap and practical non‐noble metal‐based ester hydrogenation catalyst. Based on the complementary experimental and computational results, we provide a mechanistic proposal that points to a potential for further improvement of the Mn‐based catalysts under study.

Conflict of interest

The authors declare no conflict of interest.

Supporting information

As a service to our authors and readers, this journal provides supporting information supplied by the authors. Such materials are peer reviewed and may be re‐organized for online delivery, but are not copy‐edited or typeset. Technical support issues arising from supporting information (other than missing files) should be addressed to the authors.

Supplementary

R. van Putten, E. A. Uslamin, M. Garbe, C. Liu, A. Gonzalez-de-Castro, M. Lutz, K. Junge, E. J. M. Hensen, M. Beller, L. Lefort, E. A. Pidko, Angew. Chem. Int. Ed. 2017, 56, 7531.