Mechanism, Reaction Kinetics and Multi-Scale Modelling of CO2 Hydrogenation to Methanol over Trimetallic Heterogeneous Catalysts

Carbon dioxide is a source of potential carbon-containing raw material and also one of the major greenhouse gases [1]. In the past decades, the potential use of CO₂ as an alternative feedstock replacing CO in methanol production has received attention as an effective way of CO₂ utilization [1,2].

A highly efficient catalyst is the key for methanol synthesis via CO₂ hydrogenation. Experiments with the aid of the isotope-labelled CO and spectroscopy have demonstrated that methanol was produced by the hydrogenation of CO₂ rather than CO, while the commonly-used ternary Cu–Zn–Al oxide catalyst for CO hydrogenation was not particularly active for the conversion of pure CO₂ at 5.0–10.0 MPa and 200–250 °C [3]. Mostly used catalyst type for CO₂ hydrogenation was the modified methanol synthesis analogue for CO hydrogenation and the existing studies addressed its chemical composition, supports, additives, different preparation methods/conditions and morphology. Notably, zinc oxide can improve the dispersion and stabilization of copper. ZnO possesses lattice oxygen vacancies, consisting of an electron pair in the lattice, which is active for methanol synthesis. The basicity of ZnO influences the catalyst activity through affecting the dispersion of copper via the precipitate phase of copper [4]. Studies on the precipitate of Cu-containing hydroxocarbonates revealed that some of its phases played important roles in detrmining catalytic performance. The mixed metal oxides, obtained by the controlled thermal decomposition of hydrotalcite-like compounds (HTlcs), with the general formula of [M²⁺_1–xM³⁺_x(OH)₂]^x+(A^y-)_x/y × zH₂O, also exhibited promising results. M²⁺ and M³⁺, divalent and trivalent cation, respectively, possess homogeneous microstructure, a good dispersion of M²⁺ and M³⁺ at atomic level, an enhanced metal–oxide interaction after reduction, the stability against sintering and a high specific surface area, as well as strong basic properties [5]. Therefore, HTlcs are among the most investigated catalyst precursors considering the remarkable properties of the final catalysts. In the present study, the influence of the basicity of alkaline earth metal (i.e. Ba, Ca, Mg and Sr) catalysts, which were prepared via the hydrotalcite route on the selectivity of the methanol production in CO₂ hydrogenation, is reported. The catalysts with the molar ratio of Cu:M:Al = 6:3:1 (M = Ba, Ca, Mg, Sr) were prepared and characterized by various characterization techniques such as scanning (SEM) and transmission (TEM) electron microscopy, X-ray diffraction (XRD), X-ray photoelectron spectroscopy (XPS), CO₂ and NH₃ temperature-programmed desorption (TPD), and N₂O chemisorption.

Table 1: Properties of Cu/M/Al catalysts

Alkaline metal (M)	Surface area (m2 g–1)	Cu dispersion (%)	Crystallite size (nm)	Total basicity (µmol CO2 g–1)
Mg	85	58.2	14	2.8
Ca	69	41.1	11	3.1
Sr	63	38.5	16	3.4
Ba	48	30.7	18	4.1

Powder XRD patterns suggested that the isolated Cu²⁺ species interacted strongly with alumina, to an extent that it may have even formed the surface copper aluminate phase. However, the presence of a highly-dispersed surface CuO was present as well, distributed as smaller crystallites. The SEM and TEM images of the catalysts also supported the findings, obtained by XRD; specifically, no extensive agglomeration of CuO was observed in the respective images. Weak, strong and some moderate basic sites were attributed to the three peaks present in the CO₂ TPD profile of the prepared catalysts, while basicity was in the following order: Ba > Ca > Sr > Mg. The basic character of the catalysts was strongly dominated by the presence of the alkali metal when compared to copper and alumina.

The catalysts were consequently tested in parallel reactor system, which comprised five fixed-bed vessels (Figure 1). The catalytic performance of the alkaline earth metal-containing catalysts was compared with the commercial methanol synthesis catalysts (by Alfa Aesar and Lurgi). Methanol selectivity and CO₂ conversion were examined under the continuous flow in the temperature (T) range between 200 °C and 400 °C. The catalysts were exposed to different space velocities (GHSV) (from 2000 to 6000 h^–1). Prior to catalytic measurements, fresh catalysts were reduced in the stream of H₂ at 300 °C for 3 h under the atmospheric pressure (P; 1 bar). Reactors were then cooled to the ambient temperature (25 °C) introducing reactant gas (N₂:CO₂:H₂ = 1:2:6 molar ratio) flow and raising the pressure to 2.0 MPa.

Figure 1: Parallel reactor system.

The results showed that Cu/Mg/Al catalyst was at least as good as the commercial catalysts (Figure 2), while the main products detected were methanol, water and carbon monoxide. Selectivity for methanol was between 10 and 40%, depending on reaction conditions and the catalyst used. The experimental data, paralleled by model predictions showed that CO₂ conversion increased with elevated temperature, while on the contrary, methanol selectivity dropped with increasing temperature.

Figure 2: Methanol selectivity and CO₂ conversions over different catalysts at GHSV = 2000 h^–1, T = 200 and 250 °C, and P = 20 bar.

It is well-known that copper metallic surface area is an important parameter for the methanol synthesis by CO₂ hydrogenation. The relationship between copper surface area and the catalytic activity for the reactions over the copper-based catalysts has been studied extensively, though some controversies still remain. In the present study, the effect of the surface area of the metallic copper on the activity of the methanol synthesis through CO₂ hydrogenation was established and the results are presented in Figure 3. It is observed that the attained catalytic activity increases analogously with surface area versus the amount of basic sites. This indicates that among the preparation techniques applied, ultra-sonic method improves the dispersion of copper particles without changing the intrinsic activity of CuO–ZnO catalyst. This observation was also supported by the results, obtained by N₂O dissociation reaction and XPS.

Figure 3: Catalyst activity versus metallic copper surface area and basic sites for catalytic CO₂ hydrogenation at T = 250 °C and P = 20 bar.

In parallel to extensive experimental trials (varying catalysts and process operating conditions), density functional theory (DFT) calculations were used to determine elementary reaction steps and their activation energies, as well as pre-exponential factors (Figure 4), yielding the rate constants for different operating conditions. These parameters were used in the kinetic Monte Carlo (KMC) and micro-kinetic modelling to obtain the model for a packed bed reactor (PBR). The two critical variables for the design of PBR, the packing void fraction and the pressure drop across PBR, are usually predicted using empirical correlations that were and still are a matter of an ongoing discussion among researchers. The generalization of typical parameters brings about a high error risk and leads to oversized reactors due to the capacity safety factor. In contrast to the correlations and experimental investigations of adaptive parameters, several numerical approaches are available today to investigate the flow within particle packings, mostly based on computational fluid dynamics (CFD), which were also applied in this study. The common work-flow for the CFD simulations of a realistic packed bed thus consisted of reactor bed packing, domain meshing, boundary conditions setting, simulation running, and finally, collecting the results (Figure 5), coupling CFD with reaction kinetics upon operating outside a fully-developed turbulent regime, in which the solution could have been approximated by plug flow.

Figure 4: Reaction mechanisms and pathways employed.

Figure 5: CFD simulations of packed bed reactors.

In overall, the study comprised the preparation of various non-noble metal heterogeneous catalysts with differing in respective surface composition. While a typical Cu–Zn–Al process performance was modelled in an inherently multi-scale manner; specifically, obtaining a full reaction kinetic parameters set from DFT (in turn coupling the latter with a realistic reactor operation using transport phenomena and fluid mechanics), for other heterogeneous catalytic formulations, the micro-kinetic reaction model was maintained in its original form, adopting a regression of the pinpointed rate determining step(s) (RDS(s); also by DFT), whereas each reaction kinetic term was de facto correlated with the corresponding sites available (meaning Cu, ZnO, Al₂O₃, etc.). As a result of this it is therefore possible to, to some extent, reverse engineer heterogeneous catalytic material, or adopt the kinetic scheme proposed, should a new catalytic formulation be applied, respectively. The developed modeling framework also allows for a detailed pathway analysis also upon shifting the desired product selectivity, e.g. to syngas via the reverse water–gas shift reaction (RWGS), or varying process conditions applied.

Acknowledgement

The presented work was partially funded by the EU Framework Programme for Research and Innovation Horizon 2020 under the grant agreement No 637016 (MefCO2). The authors also gratefully acknowledge the financial support of the Slovenian Research Agency (ARRS) through the Programme P2–0152.

References

[1] K.-D. Jung, O.-S. Joo, Catal. Lett., 84 (2002) 21-25.

[2] Slamet, H.W. Nasution, E. Purnama, S. Kosela, J. Gunlazuardi, Catal. Commun., 6 (2005) 313-319.

[3] J. Gao, Q. Liu, F. Gu, B. Liu, Z. Zhong, F. Su, RSC Adv., 5 (2015) 22759-22776.

[4] S.G. Jadhav, P.D. Vaidya, B.M. Bhanage, J.B. Joshi, Chem. Eng. Res. Des., 92 (2014) 2557-2567.

[5] Á. Mastalir, Á. Patzkó, B. Frank, R.Schomäcker, T.Ressler, R.Schlögl, Catal.Commun., 8 (2007) 1684-1690.