Liens transverses ENS de Lyon

Agenda de l'ENS de Lyon

A Theoretical Perspective on Hydrogenation and Oligomerization of Acetylene over Pd Based Catalysts

Soutenance de thèse

Vendredi 29 sep 2017
Soutenance de M. Emanuele VIGNOLA de l'IFP Energies Nouvelles sous la direction de M. Philippe SAUTET.


Soutenance de M. Emanuele VIGNOLA de l'IFP Energies Nouvelles sous la direction de M. Philippe SAUTET.

Description générale
Selective hydrogenation of acetylene in ethylene-rich flows is a fundamental process in the petrochemical industry since it allows the purification of ethylene for polymer applications. The reaction is catalyzed by Pd, which features acceptable selectivity towards ethylene compared to the total hydrogenation product, ethane. Pure Pd is, however, deactivated by oligomeric byproducts, known as ”green oil” in the literature.
Therefore, most industrial catalysts are Pd-Ag alloys, where Ag helps to suppress the secondary reactions. This work addresses the formation of initial oligomers on Pd and Ag-Pd catalysts. A mean field based theoretical model was built to efficiently screen the topology of the topper most layer of the alloy catalyst under relevant conditions. This model gave evidence for strongly favored Pd island formation.
To confirm this result, the system was then re-investigated by means of Monte Carlo simulations including the effect of segregation. Emergence of large domains of Pd were confirmed over large ratios of Ag to Pd. Green oil is expected to form on these catalytically active islands. To obtain a detailed view on the oligomerization process, activation energies were computed both for hydrogenation and oligomerization steps by periodic density functional theory on Pd(111). Oligomerization was found to be competitive with hydrogenation, with the hydrogenation of the oligomers being the among the fastest processes. The role of Pd domains to green oil formation is still to be clarified under realistic conditions, where the surface is covered by many different species. A step forward to this goal was taken by developing a machine-learning tool which automatically interpolates model Hamiltonians on graphical lattices based on DFT computations, accounting for lateral interactions and distorted adsorption modes on crowded surfaces.

Salle  de Conférence CBP - Site Monod - ENS de Lyon