(605c) Dynamics and Mechanism of Carbon Filament Formation during Methane Reforming on Supported Nickel Catalysts

Inactive carbon deposits can form during catalytic CH₄ reforming reactions on Ni-based catalysts, leading to commercial processes that must use CO₂/CH₄ or H₂O/CH₄ ratios above those required by stoichiometry in order to suppress coking. These carbon deposits hinder CH₄ reforming reactions by blocking pores and can lead to catalyst disintegration and elution as fines from reactors [1]. Carbon formation is generally considered to proceed via the following sequence: (a) dissociative chemisorption of CH₄ on the metal surface, (b) dissolution of carbon atoms into the metal, (c) bulk diffusion of carbon atoms through the metal, and (d) precipitation of carbon atoms at the metal-carbon interface to form a filament at the supra-equilibrium chemical potentials required for nucleation of the solid phase. Bulk diffusion of carbon through the metal is considered the rate limiting step in this scheme and is driven by a gradient in the thermodynamic activity of surface carbon between the side exposed to reactants (a_C*)_gas and that in contact with the filament (a_C*)_fil [2]. Carbon deposition rates and the morphology of the carbon deposits thus depend on (a_C*)_gas, which is set by the gas composition for equilibrated mixtures, but which depends on the kinetics of formation and removal of C-atoms during steady-state catalysis and thus on the structure and composition of the active surfaces. Most previous studies have investigated carbon formation using equilibrated binary gas mixtures, such as CO-CO₂ or CH₄-H₂, and developed relationships for the carbon activity under equilibrium conditions [3-4]; few studies have systematically examined the kinetics of carbon deposition for CH₄-CO₂-H₂O-CO-H₂ gas mixtures during steady-state catalysis [5]. The conditions that lead to carbon formation during CH₄ reforming before equilibration is achieved and the nature of the requisite driving forces have remained speculative and uncertain. This study describes rate data and their mechanistic interpretation for the formation of surface carbon during CH₄ reforming and develops the appropriate expression for (a_C*)_gas over at a wide range of conditions.

Experiments were carried out on 7% wt. and 15% wt. Ni/MgO samples prepared by incipient wetness impregnation of MgO powders with aqueous Ni(NO₃)₂ · 6H₂O (Alfa, 99.9%), as reported previously [6]. Samples were treated in H₂ (Airgas, UHP, 50 cm³ g^-1 s^-1) at 1123 K (0.167 K s^-1) for 3 h before their use in reactions. The dynamics of carbon formation on these catalysts were measured during reforming reactions using a tapered element oscillating microbalance under controlled reactive environments at 793-973 K; changes in the oscillation frequency of the tapered quartz element are correlated to changes in the mass of the catalyst sample due to deposited carbon. The thermodynamic carbon activity (a_C*)_gas was varied through systematic changes in the inlet CH₄, CO₂, CO, H₂, and H₂O concentrations and in the residence time and chemical conversion for a given inlet gas composition. The effluent was continuously monitored by on-line mass spectrometry to determine CH₄ reforming rates. High-resolution transmission electron microscopy (TEM) was used to characterize Ni particle sizes and the morphology of any carbon deposits formed.

The mechanistic interpretation of CH₄ reforming rates on Ni-based catalysts, provided by extensive isotopic and theoretical inquiries [6], lead to (a_C*)_gas values that are proportional to either P_COP_CH4/P_CO2 (χ) and P_H2P_CH4/P_H2O (φ) at CH₄ reforming gas mixtures far from equilibrium. The equivalence of these two ratios reflects their proportional values through the equilibrium constant for water-gas shift, a reaction that is at equilibrium during CH₄-CO₂ and CH₄-H₂O reforming reactions. For each Ni/MgO sample and temperature, carbon deposition rates depended only on these concentration ratios at the conditions of our study and increased with (a_C*)_gas. Three carbon growth regimes are evident from TEM analysis: minimal carbon deposition (regime I), filamentous carbon growth (regime II), and metal particle encapsulation by carbon layers (regime III). These regimes also aligned with trends noted in measured carbon deposition rates. TEM detected only surface carbon patches on 5.4 nm Ni crystallites in regime I (χ < 1.1 kPa); these patches minimally covered the Ni particles. CH₄ reforming reaction rates remained constant with time, and carbon formation rates were negligible for these low χ values. Hollow carbon nanotubes with diameters similar to those of the metal crystallites formed in regime II (1.1 < χ < 4.2 kPa). CH₄ reforming and carbon formation rates also remained constant with time throughout this regime, even after the deposition of C/Ni_surface ratios >800 and the extensive formation of filamentous carbon structures evident from TEM. The filaments do not hinder reaction rates in this regime because the rate of dissolution and diffusion through the Ni particle remains equal to the rate of C-addition to the filaments, thus continuously unblocking active Ni surfaces as chemisorbed carbon forms. Encapsulation of the metal particle in onion-like carbon shells was evident in micrographs for regime III (χ > 4.2 kPa), during which reforming and carbon formation rates both decreased with time after an initial induction period. Such induction periods are characteristic of nucleation-growth processes that require supersaturation levels of carbon at metal surfaces to nucleate a new phase. These onion-like layers in regime III appear to reflect the simultaneous nucleation of a carbon phase at several points in Ni particles, thus breaking the symmetry and preferential diffusion direction present on particles with a single nucleation point, which contain a growing filament that continuously deplete the C-pool and maintain the carbon activity below supersaturation levels. Carbon formation rates were consistently higher on 11.1 nm than on 5.4 nm Ni particles (by ~ 0.3 C-atoms surface-Ni^-1 s^-1 ; 873 K ) at each given value of χ (or φ), apparently because of the more facile nucleation of carbon in larger particles and lower carbon activity for filaments with larger diameters, which tend to form on larger particles [7]. Carbon deposition rates decreased with increasing temperature for a given value of χ (or φ), reflecting presumably lower concentrations of surface carbon and surface carbon-forming precursors at higher temperatures. Removal of carbon from the catalyst by the reverse of their formation was detected at χ < 0.07 kPa on samples with carbon deposits formed during deposition experiments.

These results are consistent with the diffusion-limited growth of carbon filaments, driven by a gradient in carbon activity between the gas-side Ni surface and the precipitated carbon filament. Carbon deposition does not occur until the carbon activity in the Ni particle exceeds that required for nucleation at the metal-filament interface, which depends on carbon activity of the filament and thus on its diameter and that of the Ni particle. The carbon activity was described by the partial pressure ratio χ or φ, and carbon formation rates were a single valued function of these parameters on each supported Ni catalyst and at each temperature. Carbon deposition rates increased with Ni crystallite size because of the lower barrier for nucleation and lower carbon activity in the filament in these particles. These results provide a framework for determining CH₄ reforming conditions at which (a_C*)_gas can be kept below the threshold value that leads to carbon formation. It also provides predictive guidance for the design of CH₄ reforming catalysts that can locally control the values of χ and φ by imposing concentration gradients and by spatially preventing the formation of carbon filaments. Such properties can be conferred by metal nanoparticles that reside within void environments that impose such gradients, and which also prevent nanoparticle growth even at the high temperatures of CH₄ reforming catalysis [8]. Such small clusters would impose higher carbon activity thresholds for nucleation and growth as a result of their small diameters and their confined pore environment. Diffusional constraints for reactants and products in these confining materials are also expected to decrease (a_C*)_gas in CH₄-H₂O reforming by increasing their intracrystalline H₂O/CH₄ concentration ratios because of the faster diffusion of H₂O than CH₄ within materials that have pore diameters similar in size to those of these molecules.

[1] Rostrup-Nielsen, J. R., Catalytic Steam Reforming. In Catalysis: Science and Technology, Anderson, J. R.; Boudart, M., Eds. Springer Berlin Heidelberg: 1984; pp 1-117.

[2] Holstein, W. L., J Catal. 152 (1995) 42-51.

[3] Audier, M., Coulon, M., Carbon. 23 (1985) 317-323.

[4] Debokx, P. K., Kock, A. J. H. M., Boellaard, E., Klop, W., Geus, J. W., J Catal. 96 (1985) 454-467.

[5] Jablonski, G. A., Geurts, F. W. A. H., Sacco, A., Carbon. 30 (1992) 99-106.

[6] Wei, J. M., Iglesia, E., J Catal. 224 (2004) 370-383.

[7] Rostrup-Nielsen, J. R., Steam Reforming Catalysts. Teknisk Forlag: 1975.

[8] Otto, T., Zones, S. I., Iglesia, E., J Catal. 339 (2016) 195-208.

Financial support from BP as part of the Methane Conversion Cooperative Research Program at the University of California at Berkeley and from MEXUS grant CN-02-76 are gratefully acknowledged.