Research report: Improving hydrodeoxygenation over Ni-Mo/Al2O3 by tuning the textural properties of the support via doping with lanthanoid group metals

The standard fix is hydrodeoxygenation, a hydrotreating process that strips oxygen from these molecules under hydrogen, heat and pressure.

The researchers, led by Tove A. Kristensen and Sara Blomberg, started from a known dilemma. Heating alumina hard enough opens up its pores, which helps the chunky phenolic molecules in lignin oil diffuse to active sites instead of getting stuck at the entrance. But the same heat treatment also sinters the material, collapsing the surface area on which the nickel and molybdenum must disperse. Lose too much surface and the catalyst loses activity. Lanthanum and cerium are widely used in other catalytic fields precisely because they stabilize alumina against this kind of thermal damage, yet their role in hydrodeoxygenation had only recently begun to attract attention. Earlier work by the same group had hinted that a catalyst doped with 1 weight percent of each metal stayed active longer in the hydrodeoxygenation of vanillin, a common lignin model compound, and that lanthanum was probably the main driver of pore enlargement.

To turn that hint into a design rule, the team prepared a family of doped alumina supports using three different impregnation recipes. In the first, lanthanum and cerium nitrates were soaked into the alumina pellets together and then calcined at 1100 degrees Celsius. In the second, the two dopants were added in separate steps with calcination at 1100 and 500 degrees respectively, and in the third they were added separately but both calcined at 1100 degrees. The dopant loadings ranged from zero to 5 weight percent for each metal. Rather than testing combinations one by one, the group used response surface methodology with a central composite design, then analyzed the results with analysis of variance to build statistically significant regression models covering the whole designed composition space from a reduced number of experimental points. The central point of each design was repeated three times to capture experimental error.

The modeling delivered a clean division of labor. Across all three impregnation methods, lanthanum emerged as the dominant factor controlling pore diameter, while cerium mainly governed surface area. Only when the two metals were impregnated together did cerium influence pore diameter as well, and in that case the pore diameter grew with increasing lanthanum and could be maximized with roughly 2 to 3 weight percent cerium. For the two separate-impregnation routes, the models pointed to pore enlargement driven by lanthanum alone. When the team asked the models for the composition that maximizes pore diameter while keeping the surface area of the untreated alumina, the answer was 5 weight percent lanthanum for every method, with 0.9 weight percent cerium added only in the simultaneous-impregnation case. Validation experiments with fresh supports confirmed the models at a 90 percent confidence interval.

Characterization filled in the mechanistic picture. Crucially, the drastic surface-area collapse that signals complete conversion to alpha-alumina never appeared. X-ray diffraction backed this up, revealing mixtures of delta, theta and alpha phases in all heat-treated samples rather than a single alpha phase, which the authors read as evidence that the treatment induced only partial transformation. Where 5 weight percent lanthanum was present, the diffractograms showed lanthanum aluminate rather than lanthanum oxide, and cerium oxide reflections grew stronger as the lanthanum loading increased, indicating larger crystalline ceria particles. Ceria calcined at 500 degrees appeared in finer particles than ceria calcined at 1100 degrees.

The dopants also reshaped the acidity of the finished nickel-molybdenum catalysts. Ammonia chemisorption and temperature-programmed desorption showed that catalysts on supports carrying 5 weight percent lanthanum had lower total acidity, lower acid-site density and fewer strong acid sites, the latter identified by an ammonia desorption peak near 460 degrees Celsius. The authors speculate that the dopants block strong acid sites and convert them into sites of intermediate strength. Temperature-programmed reduction added a further twist: within the 400 to 500 degree reduction window, the ratio between the peak assigned to polymeric molybdenum species and the peak assigned to well-dispersed molybdenum shifted toward the polymeric form on the 5 weight percent lanthanum catalysts, a change absent on undoped heat-treated alumina. The team proposes that blocking the strong acid sites weakens the interaction between support and molybdenum, letting the active metal agglomerate and reducing its dispersion.

All of this mattered only once the catalysts faced real reaction conditions, so the team ran hydrodeoxygenation of vanillin in a pressurized fixed-bed reactor at 314 degrees Celsius and 5 bar gauge pressure, with a weight-hourly space velocity of 35 per hour, using hydrogen sulfide in hydrogen to maintain the molybdenum sulfide active phase. Every catalyst tested pushed conversion above 99 percent, so the interesting question was not whether vanillin disappeared but how it disappeared. The reaction can reach cresol, the main deoxygenated product, through two routes: a direct hydrodeoxygenation pathway that begins by removing the aldehyde oxygen, or a cracking pathway that starts by breaking the molecule apart. By tracking the selectivity to cresol relative to the methane-to-carbon-monoxide ratio, the researchers could tell the routes apart, since a lower value signals a stronger preference for direct deoxygenation.

On that measure, the catalysts supported on alumina doped with just 1 weight percent lanthanum and 1 weight percent cerium showed the highest propensity for the direct hydrodeoxygenation pathway. The catalysts carrying 5 weight percent lanthanum instead leaned toward cracking, which the authors attribute to the poorer metal dispersion implied by the polymeric molybdenum signal: with fewer, larger active sites, the bulky aromatic ring can adsorb more freely, and adsorbed rings are known to favor cracking over deoxygenation of the oxygenated functional groups. Elemental analysis of the spent catalysts added the second half of the story. Sulfur deposition was similar across all samples, so it could not explain the performance differences, but carbon deposition told a clearer tale. Measured per unit of total acidity, the low-dopant catalysts carried the least carbon, while the 5 weight percent lanthanum catalysts, despite having fewer and weaker acid sites, accumulated more. The authors link this to the increased cracking route and to lanthanum-driven basicity promoting carbonate formation, consistent with earlier observations on platinum catalysts.

If the only goal were the widest possible pores, the regression models say to load the alumina with 5 weight percent lanthanum. But that formulation degrades the very properties a hydrodeoxygenation catalyst needs, weakening metal dispersion, shifting the reaction toward cracking and encouraging carbon laydown. The modest 1 weight percent of each dopant, the formulation carried over from the group’s previous work, delivered the better catalytic outcome: a stronger direct deoxygenation pathway and less coke, which the authors interpret as prolonged activity over the eight-hour test.

The study also sketches the limits of the current evidence. The activity tests ran for eight hours on a single model compound rather than on real lignin oil, and the authors themselves note that their heat treatment appears not to have driven the alumina to full phase transformation, meaning the full potential of lanthanum and cerium doping for this purpose may not yet have been exhausted. Earlier literature they cite adds further caveats: cerium’s coke-suppressing effect over a nickel-molybdenum catalyst in guaiacol upgrading did not carry over to real bio-oil, and the order of dopant impregnation has been shown to matter for activity and selectivity in related systems. Even so, the work demonstrates that a statistically designed tuning of support texture, dopant identity and calcination sequence can be turned into concrete catalyst formulations, and that in this system less dopant, not more, is what keeps the deoxygenation chemistry on track.

Subject of Research: Tuning alumina-supported Ni-Mo hydrodeoxygenation catalysts by doping the support with lanthanum and cerium

Article Title: Improving hydrodeoxygenation over Ni-Mo/Al2O3 by tuning the textural properties of the support via doping with lanthanoid group metals

Article References: Kristensen, T. A., Selimi, J., Nordlander, J. E., Abdelaziz, O. Y., Hulteberg, C. P., & Blomberg, S. (2025). Improving hydrodeoxygenation over Ni-Mo/Al2O3 by tuning the textural properties of the support via doping with lanthanoid group metals. Discover Catalysis, 2(1), Article 25. https://doi.org/10.1007/s44344-025-00031-6

Image Credits: AI Generated

DOI: 10.1007/s44344-025-00031-6

Keywords: hydrodeoxygenation, lignin, Ni-Mo catalyst, alumina support, lanthanum doping, cerium doping, pore diameter, surface area, coke formation, vanillin, response surface methodology, catalyst deactivation