Dynamics and Mechanism of Carbon Filament Formation during
Methane Reforming on Supported Nickel Catalysts
Samuel L. Leung1, Junmei Wei1, William L. Holstein1, Miguel Avalos-Borja2, Enrique Iglesia1
1Department of Chemical and Biomolecular Engineering, University of California – Berkeley, Berkeley, California 94720, United States
2Department of Advanced Materials, Instituto Potosino de Investigación Cientifica y Technológia, San Luis Potosí, Mexico
Keywords: Methane reforming, carbon deposition, carbon filaments, nickel catalyst
Carbon filaments tend to form on Ni-based catalysts during the steady-state catalysis of CH4 reforming reactions, thus necessitating H2O/CH4 or CO2/CH4 ratios in excess of those required by stoichiometry in order to suppress coking. Bulk diffusion of carbon from exposed surfaces to those in contact with the growing filament has been proposed to limit the rate of filament growth. This diffusive flux depends, in turn, on the gradient in the thermodynamic activity of carbon between metal surface exposed to gaseous reactants (aC,s) and that in contact with the filament (aC,f) . The rate of filament formation therefore reflects a value of aC,s that is determined by the relative rates of the elementary steps that form and remove C-atoms during steady-state CH4 reforming catalysis and thus also on the structure and composition of the active surfaces. Sperle et al. previously used similar arguments determine a critical H2O/CH4 ratio below which carbon growth would not occur . This study describes a mechanistic basis for interpreting CH4 reforming rates, which accounts also for the activity of surface carbon during steady-state CH4 reactions, over a broad range of conditions relevant to the practice of reforming catalysis.
Experiments were performed on 7% wt. and 15% wt. Ni/MgO samples prepared by incipient wetness impregnation . The dynamics of carbon deposition were measured during CH4 reforming reactions using a tapered element oscillating microbalance under controlled reactive and hydrodynamic environments at 843-973 K. The value of aC,s was varied through systematic changes in the inlet CH4, CO2, CO, H2, and H2O concentrations and the chemical conversion for each inlet gas composition. The effluent was continuously monitored by on-line mass spectrometry to concurrently determine CH4 reforming rates. High-resolution transmission electron microscopy (TEM) was used to determine Ni particle sizes and the morphology of the carbon deposits formed after exposure to reactive environments.
The mechanistic interpretation of CH4 reforming rates on Ni-based catalysts, previously derived from isotopic and theoretical inquiries , shows that aC,s values are proportional to PCOPCH4/PCO2 (χ) (or equivalently to PH2PCH4/PH2O (j), because of the prevalent water-gas shift equilibration) during CH4-H2O and CH4-CO2 reforming at conditions far removed from its thermodynamic equilibrium. For each Ni/MgO sample and temperature, the carbon morphology and carbon deposition rate were solely determined by the magnitude of χ (or j). Three regimes, leading to distinct carbon morphologies, were evident from TEM analysis with increasing χ or j values for 7% wt. Ni/MgO: minimal formation of carbon deposits (regime I, χ < 1.1 kPa), formation of filamentous carbon without detectable changes in reforming turnover rates (regime II, 1.1 < χ < 4.2 kPa), and formation of encapsulating carbon shells with a concomitant decrease in reforming rates (regime III, χ > 4.2 kPa). These regimes also aligned with trends noted in steady-state carbon deposition rates (Figure 1). Carbon deposition rates were consistently higher on 11 nm than on 5.4 nm average diameter Ni particles and decreased with increasing temperature for a given value of χ.
Carbon deposition rates are negligible in regime I because filament growth cannot occur until the carbon activity in the Ni particle exceeds that required for nucleation of a carbon filament phase. The activity of carbon filaments can therefore be estimated from the threshold for filament growth. The linear dependence of carbon deposition rates on aC,s in regime II is consistent with the proposed diffusion model, while the lower than expected rates in regime III from extrapolation of rates in regime II is caused by the encapsulation and subsequent deactivation of the Ni clusters. CH4 turnover rates and carbon deposition rates were not influenced by the formation of carbon filaments in regime II because such growth of filaments continuously depletes the carbon pool, maintaining carbon activity below supersaturation levels, and does not hinder access by gas-phase reactants to active Ni surfaces. Carbon deposition rates increased with increasing Ni crystallite size, reflecting lower thermodynamic activities for the larger diameter filaments that tend to form on larger clusters , leading to steeper carbon activity gradients. Carbon deposition rates decreased with increasing temperature for a given value of χ, reflecting the higher activation energy for the removal of surface carbon than for its deposition [3, 5] and therefore the more rapid removal of surface carbon and thus lower values of aC,s at higher temperatures.
Carbon diffusion and filament growth is driven by a gradient in thermodynamic activity across the metal catalyst particle. The carbon activity at the gas-metal interface is described by the pressure ratio χ or j at conditions before equilibration is achieved, and carbon deposition rates are a single valued function of these parameters on a given supported Ni catalyst and temperature. These results provide a framework for determining CH4 reforming conditions at which aC,s can be kept below threshold values that lead to carbon deposition.
 Holstein, W. L., J Catal. 152 (1995) 42-51.
 Sperle, T., Chen, D., Lodeng, R., Holmen, A., Appl Catal a-Gen. 282 (2005) 195-204.
 Wei, J., Iglesia, E., J Catal. 224 (2004) 370-383.
 Rostrup-Nielsen, J. R., Steam Reforming Catalysts. Teknisk Forlag: 1975.
 Snoeck, J. W., Froment, G. F., Fowles, M., Ind Eng Chem Res. 41 (2002) 4252-4265.
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.