The hydroprocessing reactions of kraft lignin and four of its model compounds, o-hydroxydiphenylmethane (OHD), diphenylmethane (DPM), 4-phenoxyphenol (PP), and 4-methylguaiacol (MG) were studied over a sulfided CoMo/γ-Al2O3 catalyst. The reactions of OHD were also studied over sulfided Ni-Mo/γ-Al2O3, Ni-W/γ-Al2O3, and Ni-Mo/zeolite catalysts. Reaction pathways, kinetics, and catalyst deactivation were resolved. Experiments with the actual kraft lignin substrate showed that catalytic hydroprocessing led to higher yields of single-ring products and lower yields of light gas compared to hydropyrolysis alone. The CoMo/γ-Al2O3 catalyst increased the selectivity to non-coking monophenolics and hydrocarbon products. The model compound results permitted interpretation. The hydroxyl substituents on OHD and PP directed bond cleavage strongly. The atom linking the two aromatic rings remained almost exclusively with the unsubstituted ring after cleavage. The substituent increased reactivity by at least an order of magnitude. The catalytic reactions of OHD, DPM, and PP were 103-104 times faster than their thermal reactions. Deoxygenation was appreciable in these reactions and in the catalytic hydroprocessing of MG. Collectively, these results suggest that the improved activity and selectivity in catalytic lignin liquefaction is due to enhancement of link cleavage and the transformation of coking dioxygen-substituted phenolic coke pecursors (e.g., catechols and guaiacols) to non-coking phenols. Hydrotreating catalysts with alumina supports appear suitable for lignin conversion. The highly active and rapidly deactivating zeolite-supported hydrotreating catalyst afforded a complex mixture of products. Regeneration of the alumina-supported catalysts would be required. Notes ∗Correspondence Additional informationNotes on contributorsPeter M. Train Present Address: Amoco Oil Company, Research and Development, P. O. Box 3011, Naperville, IL 60566.
The aging anticipated of a typical lignin hydrotreating catalyst was probed via reactions of the model compound o-hydroxydiphenylmethane (OHD). Reactions over Co-Mo/y-AI2O3 at 250°C, 2250 PSIG, and fixed-bed space velocities of about 0.5 hr' led to toluene and phenol as major products with minor amounts of benzene and/or cyclonexane and o-cresol. Deoxygenation was not significant. The observed catalyst decay of 76% was fully recoverable upon resulfidation. Reaction at higher temperature led to some deoxygenation. Reaction over the same catalyst with feed-added methanol led to the formation of a methyl-substituted OHD (MOHD) as well as toluene, phenol, and deoxygenation products. For this case, catalyst decay was only partially recoverable upon resulfidanon. At MeOH/OHD = 7.9, only 25% of the cracking function was recoverable whereas fully 100% of the deoxygenation activity was recoverable. Finally, reaction over NiMo/y-AljOj led to toluene, phenol, and deoxygenated products. Catalyst decay was only partially recoverable. These results led to the categorization of three types of oxygen species: "phenolic" oxygen, methanol, and "product" oxygen all affected the catalyst differently. Phenolic oxygen led to reversible catalyst decay. Methanol exhibited reversible inhibition and promoted deoxygenation to product oxygen, which led to irreversible decay of, largely, the catalyst's cracking function
An a priori Monte Carlo simulation of the product spectrum resulting from the thermal and catalytic depolymerization of lignin has been developed. The simulation combines model compound reaction pathways and kinetics, deactivation parameters, and stochastic models of polymer diffusion into a Markov-chain based simulation of the reaction of lignin polymers. Predicted product class ratios of single-ring phenolics were in reasonable agreement with experimental data pertaining to lignin liquefaction, especially as regards neat pyrolysis. The advantage of catalytic liquefaction was suggested by the increase in single-ring product yields, especially those of phenols and hydrocarbons, relative to pyrolysis.