Key Parameters for the Synthesis of Active and Selective Nanostructured 3d Metal Catalysts Starting from Coordination Compounds – Case Study: Nickel Mediated Reductive Amination

Graphical Abstract

Reductive Amination: The Ni/Al₂O₃ catalyst is highly active in the reductive amination of ketones using ammonia as the nitrogen source and hydrogen as reducing agents. The role of Ni-salen precursor complexes in the catalyst synthesis was identified: Molecular dispersion on the support gives rise to Ni nanoparticles upon decomposition of the complex, which are stabilized by a defined N-doped carbon layer.

Abstract

The design of nanostructured catalysts based on earth-abundant metals that mediate important reactions efficiently, selectively and with a broad scope is highly desirable. Unfortunately, the synthesis of such catalysts is poorly understood. We report here on highly active Ni catalysts for the reductive amination of ketones by ammonia employing hydrogen as a reducing agent. The key functions of the Ni-salen precursor complex during catalyst synthesis have been identified: (1) Ni-salen complexes sublime during catalyst synthesis, which allows molecular dispersion of the metal precursor on the support material. (2) The salen ligand forms a nitrogen-doped carbon shell by decomposition, which embeds and stabilizes the Ni nanoparticles on the γ-Al₂O₃ support. (3) Parameters, such as flow rate of the pyrolysis gas, determine the carbon supply for the embedding process of Ni nanoparticles.

Introduction

Reductive amination is a very important reaction because the products, alkyl amines, are used intensively as fine and bulk chemicals, pharmaceuticals, and agrochemicals.¹ More specifically, primary amines are the starting material for the production of other amines or N-heterocyclic compounds.² The synthesis of primary amines via reductive amination is very challenging, because overalkylation and other side reactions must be avoided. The reductive amination for the synthesis of primary amines with hydrogen as a reducing agent was introduced 100 years ago by Mignonac, employing a Ni catalyst.³ The use of hydrogen as a potentially sustainable and cost-effective reductant is particularly attractive, however, a catalyst is required for its activation. Furthermore, inexpensive ammonia is utilized as a nitrogen source. Despite the use of Ni catalysts ever since,^{4, 5} no catalyst system for the synthesis of primary amines with a broad scope and functional group tolerance had been found until 2019. We then presented a nanostructured Ni catalyst for the synthesis of primary amines by reductive amination, using ammonia dissolved in water.⁵ Our catalyst had a broad scope and an exceptional tolerance towards functional groups, operated at low temperature and pressure, was highly active, reusable, and easy to handle. The synthesis from a specific Ni complex, namely, a Ni-salen complex, and γ-Al₂O₃ was straightforward, and the ligand-metal combination of this complex was crucial. Other interesting earth-abundant metal catalysts, synthesized via the pyrolysis of salen complexes, were reported by us⁶ and the Beller group.^{7, 8} The superior performance of catalyst systems based on salen precursor complexes was demonstrated in all these publications. Unfortunately, the role of salen complexes in the pyrolysis-based catalyst synthesis is incompletely understood.

Herein, we report on a highly active Ni catalyst for the reductive amination of ketones by ammonia employing hydrogen as reductant. This Ni/Al₂O₃ catalyst was obtained by varying the nickel salen complex precursors used for the catalyst synthesis. Thereby, the key functions of the Ni-salen precursor complex during catalyst synthesis have been identified: (1) The volatility of Ni-salen complexes allows the molecular dispersion of the metal precursor on the support material, which enables an optimal bottom-up approach for the preparation of catalytically active metal sites. (2) Tailored decomposition of the carbon and nitrogen containing salen ligand forms a nitrogen-doped carbon shell that covers the catalytically active nickel nanoparticles, thereby stabilizing them. (3) A specific set of parameters during the pyrolysis step in catalyst generation determines the carbon supply, which is crucial for the embedding process of nanoparticles. A too high carbon supply particularly favors the undesired formation of carbon nanotubes.

Results and Discussion

We developed a Ni complex library to better address the question about the role and the mandatory nature of Ni-salen complexes in the generation of highly active hydrogenation catalysts. Based on this, alumina-supported catalysts were prepared and investigated for their catalytic activity in the reductive amination of acetophenone as a model reaction. The Ni-salen ligand structures were selected by fine-tuning the steric properties and carbon and nitrogen content by choosing appropriate amine precursors and ring substituents. Starting from the known Ni precursor C1⁵ three different aldehydes and four different diamines, including aliphatic, N-heteroaromatic und polyaromatic compounds, were combined. A total of six complexes, C1 to C6, were synthesized, as outlined in Figure 1. The corresponding Ni/Al₂O₃ composite catalysts Cat-1 to Cat-6 were generated along an established three-step protocol for 3d metal catalysts supported on commercial supports^{5, 6}: The γ-Al₂O₃ support was wet impregnated with C1 to C6 (ideally 4 wt.% nickel) in acetonitrile. After removal of the solvent, the samples were pyrolyzed in a nitrogen flow up to 700 °C, followed by a reduction step in forming gas at 550 °C.

Figure 1

A library concept of 6 Ni-salen complex precursors: Ni-salen complexes C1 to C6 synthesized by fine-tuning the steric properties and carbon and nitrogen content (left). The catalytic activity of Ni/Al₂O₃ catalysts Cat-1 to Cat-6 studied in the reductive amination of acetophenone depending on the Ni source (right). In order to ensure comparability, the reaction conditions from Ref. [5] were adjusted so that no complete conversion of the reactant occurred: 1.2 mol% Ni/Al₂O₃ catalyst (0.006 mmol Ni, 0.35 mg Ni), 0.5 mmol acetophenone, 0.5 mL aq. NH₃ (25 %, 6.7 mmol), 2 mL H₂O, 80 °C, 0.5 MPa H₂, 20 h. Yields were determined by gas chromatography using n-dodecane as an internal standard. Ni content was analysed by ICP-OES.

The catalytic activity of the Ni/Al₂O₃ catalysts Cat-1 to Cat-6 was compared in the reductive amination of acetophenone as a model reaction. The table in Figure 1 gives an overview of the catalytic activity of the catalysts and the respective loading of nickel nanoparticles. We discovered an obvious dependence of the performance of Ni/Al₂O₃ catalysts on the Ni precursor complex used. Cat-4 showed the highest activity with a yield of 60 % of 1-phenylethylamine, followed by Cat-3 (55 %). By contrast, Cat-5 (2 %) and Cat-6 (9 %) were nearly inactive in the reductive amination of acetophenone under these reaction conditions. It was shown that the catalytic activity of the Ni/Al₂O₃ catalyst published,⁵ Cat-1 (35 %), was surpassed by fine-tuning the salen ligand, which, similar to Cat-2 (34 %), led to only a moderate yield. Salen complexes are known to sublime without decomposition.⁹ Consequently, the volatility of the Ni-salen complexes was investigated by thermogravimetric analysis (TGA) analogous to the catalyst preparation parameters (Supporting Information, Figure S1). All Ni salen complexes C1 to C6 volatilize into the gas phase above 400 °C (Supporting Information, Table S1). Intriguingly, the compounds C3 and C4 show significantly higher volatility than the other Ni-salen complexes. Possibly, the attached di-tert-butyl substituted aromatic rings and aliphatic backbones are beneficial. Inductively coupled plasma optical emission spectroscopy (ICP-OES) was used to analyze the amount of Ni in the Ni/Al₂O₃ catalysts. Depending on the volatility of the Ni-salen complex used, the nickel loading of the Al₂O₃ support decreased due to removal in the gas flow during catalyst preparation. Cat-3 and Cat-4 showed the lowest Ni loadings of 2.8 wt.% and 2.6 wt.%, respectively, while less volatile Ni precursors resulted in higher Ni contents (Figure 1). In addition, the carbon and nitrogen contents were determined by elemental analysis, reflecting the atomic composition of the salen precursors (Supporting Information, Table S2, Figure S2). Transmission electron microscopy (TEM) analysis of Ni/Al₂O₃ materials (Supporting Information, Figure S3) showed a subordinate dependence of the Ni particle size on the precursor complex. The average Ni particle size of the catalysts varied from 8.0 to 11.0 nm with narrow size distribution. Only Ni salen complex C2 was decomposed to larger Ni particles of 30 nm in diameter. The Ni particle size did not alter due to catalysis, as exemplified for Cat-4 (Supporting Information, Figure S4). Initial results did not give a clear indication that the Ni particle size and/or the amount of nickel, carbon and nitrogen in the Ni/Al₂O₃ catalyst is the determining factor for a high catalytic activity. However, we observed a qualitative effect of the precursor complex itself. Since there might be a correlation between the high hydrogenation activity of Cat-4 and the volatility of the Ni-salen complex C4, this led to the hypothesis that volatility might play a crucial role in the generation of catalytically active metal sites. Therefore, we investigated the role of the Ni-salen complex during active catalyst formation exemplified by complex C4. The following three key properties were identified:

(1) Molecular Dispersion of the Metal Precursor on the Support Material. Firstly, we focused on the interaction of the Ni-salen complex C4 with the Al₂O₃ support material during the impregnation and pyrolysis steps in catalyst preparation. We found by TEM analysis of wet impregnated C4/Al₂O₃ (Supporting Information, Figure S6) that C4 crystallizes in needles several micrometers long, starting from Al₂O₃ agglomerates as nucleation centers once the solvent is removed. This method of impregnation does not yield molecular dispersion of C4 on Al₂O₃ to produce small Ni nanoparticles directly during pyrolysis. Comparative TGA analysis (Figure 2a) of salen ligand L4, Ni salen complex C4 and impregnated C4/Al₂O₃ showed that the salen ligand is already volatile, being thermally stabilized by the chelating coordination of nickel. C4 itself sublimes at a temperature of 469 °C without decomposing, confirmed by mass spectrometry of the residue (Supporting Information, Figure S5). As briefly discussed above, a Ni loading of 2.6 wt.% implies that about 65 % of C4 was decomposed on the Al₂O₃ support during pyrolysis. Considering the almost quantitative sublimation of the pure C4 (92 %), this is a clear indication of an attractive interaction of the Lewis-acidic Al₂O₃ surface and the Ni-salen complex. In addition, we investigated the interaction of the Ni-salen complex C4 with the Al₂O₃ surface using diffuse reflectance infrared Fourier transform spectroscopy (DRIFTS). As measured in pure Al₂O₃, the vibrational band of acidic Al₃−OH is centered at 3696 cm⁻¹ and the broad signal at 1250–900 cm⁻¹ originates from Al−O (Figure 2b).¹⁰ The DRIFTS spectrum shows a very broad band in the hydroxyl spectral region (4000–2500 cm⁻¹) since the untreated Al₂O₃ sample is surface hydrated by physically absorbed water molecules. The DRIFTS spectrum of Ni-salen complex C4 shows signals in the fingerprint region for wavenumbers lower than 1680 cm⁻¹, with the characteristic signal at 1625 cm⁻¹ originating from the C=N imine stretching. The bands between 2951 and 2843 cm⁻¹ were assigned to C−H stretching vibrations. When impregnated C4/Al₂O₃ was heated to 469 °C as during catalyst formation and then cooled to room temperature, the band at 3696 cm⁻¹ was no longer visible in the spectrum. The lack of a characteristic Al₃−OH band suggests a surface-absorbed complex C4 interacting with acidic centers of Al₂O₃. The TEM analysis of this material showed no crystals of C4 (Supporting Information, Figure S6). As exemplified for C4, Ni-salen complexes sublime during Ni/A_l2O₃ catalyst generation with negligible decomposition. This property allows the molecular dispersion of single complex molecules on the Al₂O₃ surface from the gas phase. Since this dispersion cannot be achieved by wet impregnation with C4, the volatility of the complexes plays a key role in this bottom-up approach to generate catalytically active metal sites.

Figure 2

(a) TGA analysis (heating ramp 10 K/min, in N₂ flow) of L4 (dashed), C4 (black) and impregnated C4/Al₂O₃ (grey) demonstrating the volatility of salen compounds. (b) DRIFTS analysis of Al₂O₃ (black), C4 (red) and C4/Al₂O₃ (blue, heated to T_subl 469 °C of C4) confirming the interaction of the Ni-salen complex C4 with the Al₂O₃ support. The Al₃-OH band is shown in the inset.

(2) Formation of a Nitrogen-doped Carbon Shell for the Stabilization of Nickel Nanoparticles. We recorded X-ray photoelectron spectroscopy (XPS) survey spectra between 0–1100 eV, given in Figure S7. All expected lines for the Al₂O₃ support were identified, and in addition, carbon and minor traces of nitrogen were found. Small Ni signals were detected in the Ni 2p region (Figure 3b), but further Ni lines were not identified due to their small photoemission cross section and their partial overlap with Al lines. The Ni/Al₂O₃ catalyst shows a combination of metallic Ni⁰ signals located at 852.6 eV and a broad signal located at around 854.6 eV which we assigned to oxidized Ni²⁺. Contributions of Ni³⁺ are also possible due to the width of the signal. The 6 eV satellite of Ni is found as a broad feature between 858.6 and 865 eV.¹¹ The intensity ratio of Ni⁰:Ni²⁺ is approximately 1.5 : 2. The C 1s signal observed at 284.7 eV is close to what is typically observed for graphitic carbon.¹² The binding energy of the N 1s signal observed is at around 399.2 eV, possibly a remnant of the N-containing precursor molecule and its decomposition products (Supporting Information, Figure S7). High-angle annular dark-field scanning TEM (HAADF-STEM) analysis of Ni/Al₂O₃ confirmed the homogeneous distribution of Ni nanoparticles with an average size of 8.0 nm on the Al₂O₃ support material (Figure 3a, c). Energy dispersive X-ray (EDX) elemental maps for nickel and carbon recorded in the same image section illustrate the embedding of Ni nanoparticles in a carbon matrix covering the entire support material (Figure 3d, e). Electron energy loss spectroscopy (EELS) was used to study the near environment of a single nickel nanoparticle. The carbon and the nitrogen component cover the Ni particle and the surrounding support material (Figure 3f). We conclude, from XPS and HAADF-STEM investigations that the decomposition of the salen ligand on γ-Al₂O₃ during catalyst generation provides a defined N-doped carbon shell that stabilizes the Ni nanoparticles on the support material.

Figure 3

Characterization of Ni/Al₂O₃: (a) TEM image of the Ni/Al₂O₃ catalyst shows that the Al₂O₃ support is covered with homogeneously distributed Ni nanoparticles. The size distribution of Ni particles is given in the inset. (b) XPS analysis of the Ni 2p_3/2 region revealed minor traces of Ni within the composite material. The Ni⁰ nanoparticles are surface oxidized to Ni²⁺/Ni³⁺ species. (c) HAADF-STEM analysis of the Ni/Al₂O₃ catalyst and (d, e) representative EDX element maps of nickel (red) and carbon (blue). (f) Overlapped EELS element maps of nickel (red), carbon (blue) and nitrogen (green) demonstrating the embedding of a Ni particle in a N-doped carbon layer.

(3) Determination of the Carbon Supply during the Pyrolysis Step. Impregnated C4/Al₂O₃ material was sealed in a quartz glass ampoule in inert atmosphere and treated according to the standard catalyst synthesis process. Carbon nanotubes with a diameter of 30–40 nm grew in this confined gas space, starting from Ni particles of the same size (Supporting Information, Figure S11). The growth of carbon nanotubes initiated by 3d metal particles is well established.¹³ We assume that excess carbon, in the form of the salen ligand and its volatile decomposition products, could not be removed due to the absence of a gas flow, thereby favoring the formation of the carbon nanotubes. Thus, the pyrolysis parameters for heating rate and gas flow were varied during the catalyst synthesis. Changing the heating ramp by ±5 K/min resulted in consistent catalytic activity of the Ni/Al₂O₃ catalysts, as the sublimation properties of C4 were not significantly different within this chosen range (Supporting Information, Figure S12). By contrast, changing the N₂ flow has a major impact on the catalytic activity. Reducing the nitrogen flow by 75 % causes the catalytic activity to collapse (Figure 4a), while increasing the gas flow by 25 % to the maximum gas flow of the device shows no effect. The TEM analysis of the less active catalyst material (Figure 4b) revealed a broader Ni particle size distribution and a slightly larger mean diameter of 12.5 nm than for the standard Ni/Al₂O₃ catalyst. The elemental composition of this material also showed higher mass fractions of nickel, carbon, and nitrogen (3.3 wt.% Ni, 8.1 wt.% C, 0.5 wt.% N) compared to Cat-4 (2.6 wt.% Ni, 4.6 wt.% C, 0.3 wt.% N). This led us to conclude that the volatility of the Ni-salen complex in combination with judiciously chosen pyrolysis parameters regulates the carbon supply during catalyst preparation.

Figure 4

(a) The variation of the heating ramp (standard 10 K/min±5 K/min) and N₂ gas flow (1.25× and 0.25× standard flow) during Ni/Al₂O₃ catalyst synthesis identified the flow as a critical parameter during the pyrolysis step. Reaction conditions: 1.2 mol% Ni/Al₂O₃ catalyst (0.006 mmol Ni, 0.35 mg Ni), 0.5 mmol acetophenone, 0.5 mL aq. NH₃ (25 %, 6.7 mmol), 2 mL H₂O, 80 °C, 0.5 MPa H₂, 20 h. Yields were determined by gas chromatography using n-dodecane as an internal standard. (b) TEM analysis and the size distribution of nickel particles are shown for Ni/Al₂O₃ synthesized under reduced N₂ flow.

Conclusion

In conclusion, key parameters for the synthesis of active and selective nanostructured 3d metal catalysts starting from a coordination compound were found, which may enable a more rational design of such catalysts in the future.

Acknowledgements

This work was supported by the German Research Foundation (DFG SFB 840, B1). The authors also acknowledge the support of the Bavarian Polymer Institute (University of Bayreuth, KeyLab Electron and Optical Microscopy). We thank T. Feller (TGA measurements) and B. Brunner (ICP-OES measurements). The help of H. Kurz, J. Timm and R. Marschall (DRIFTS measurements) is gratefully acknowledged. Furthermore, we thank the Elite Network Bavaria for financial and other support, and E. Arzt for his support through INM. The authors would like to thank the group of Prof. K. Mayrhofer from the Helmholtz Institute Erlangen-Nürnberg for granting access to the XPS setup. Open access funding enabled and organized by Projekt DEAL.

Conflict of interest

The authors declare no conflict of interest.