Hydrogen Assisted Catalytic Biomass Pyrolysis for Green Fuels

Magnus Zingler Stummann*

*Corresponding author for this work

Research output: Book/ReportPh.D. thesis

137 Downloads (Pure)

Abstract

This thesis is dedicated to the investigation and optimization of catalytic hydropyrolysis of biomass for the production of renewable hydrocarbon fuels. This has been achieved through studying the effect of process conditions and the type of catalyst in the fluid bed hydropyrolysis reactor.
The experiments were conducted in a bench scale setup up, where up to 1 kg of biomass, mainly beech wood, was continuously fed into a fluid bed reactor in a high pressure hydrogen atmosphere containing a catalytic material, which produces char, light gasses (COx, C1-C3), water, and oil. The char was removed with a filter and the vapors were sent to a deep hydrodeoxygenation (HDO) reactor, which could be by-passed, followed by a 3 step condensation system. Support materials, commercial catalysts, and catalysts prepared in-house were tested in the experimental setup and the produced organic phases were extensively analyzed e.g. gas chromatography (GC) with atomic emission detector (AED) and GC×GC-mass spectrome-try (MS)/flame ionization detector (FID). The prepared fresh and spent catalysts were characterized with NH3 temperature programmed desorption (TPD), X-ray diffraction (XRD), Raman spectroscopy, electron microscopy (SEM and HAADF-STEM) and the surface area and elemental composition were determined.
An oxygen free (<0.01 wt.%) oil with a condensable organic (condensed organics and C4+ in gas) yield of 16.6-22.5 wt.% dry ash free (daf) was obtained using a commercial CoMoS/MgAl2O4 catalyst supplied by Haldor Topsøe A/S in the fluid bed reactor (365-511°C) and a NiMoS/Al2O3 catalyst in the HDO reactor (350-400°C) and operating at a hydrogen pressure of 15.9 bar or higher. GC×GC-MS/FID of the condensed organic phase showed that it consisted of 42-75 % FID-area aromatics. and that the concentration was kinet-ically controlled at fluid bed temperatures below 430°C and controlled by the thermodynamics at higher temperatures. Decreasing the hydrogen partial pressure to 8.0 and 3.0 bar increased the oxygen content in the organic phase to 3.3 and 7.8 wt.% dry basis (db), respectively, where the remaining oxygen was mainly in the form of phenols. The maximum concentration of monoaromatics (57 % FID-area) was detected at 15.9 bar and further increasing the hydrogen pressure decreased the monoaromatics concentration forming naph-thenes. This indicated that the concentration of monoaromatics is controlled by the thermodynamics at hy-drogen partial pressures of 15.9 bar and higher, while the concentration is controlled by the kinetics for con-version of phenols at lower hydrogen partial pressures.
Using pure support materials (MgAl2O4 and H-ZSM-5-Al2O3) in the fluid bed reactor increased the char and coke yield up to 21.1 wt.% daf, while it was between 11.4 and 13.1 wt.% daf when using different commer-cial catalysts. Having a catalyst in the fluid bed reactor decreases the degree of coking due to stabilization of the reactive oxygenates by hydrogenation. Furthermore, an energy recovery of up to 58 % in the condensable organics was obtained when using bog iron, a cheap and non-toxic natural mineral, in the fluid bed reactor, while the highest energy recovery obtained with a commercial catalyst was 54 % (NiMoS/H-ZSM-5-Al2O3). This indicates that bog iron can replace the more expensive and toxic Co(/Ni)Mo catalysts in catalytic hy-dropyrolysis.
To investigate the difference in deoxygenation activity, selectivity and product composition for sulfided CoMo, NiMo, and MoS, these catalysts were prepared, using MgAl2O4 as support, and tested in the fluid bed reactor without the HDO reactor. This showed that the NiMo catalyst had the highest hydrogenation, crack-ing, decarboxylation and/or decarbonylation activity, while the Mo catalyst had the lowest. The carbon re-covery in the condensable organics for the NiMo catalyst was 37 %, while it was 39 % for both the CoMo and Mo catalyst. The CoMo catalyst had the highest hydrodeoxygenation activity and the Mo the lowest, thus the CoMo catalyst is considered to be the most suitable catalyst for the fluid bed reactor.
Varying the CoMo loading between 4.0 and 12.0 wt.% and using MgAl2O4 as support showed that increasing the CoMo loading increased the yield of light gasses (C1-C3), decreased the oxygen content from 9.0 to 4.7 wt.% db, but decreased the carbon recovery of the condensable organics from 39 to 37 %. The effect of vary-ing the catalyst acidity was also investigated by using a mixture of zeolite and alumina (H-ZSM-5-Al2O3) instead of MgAl2O4, while maintaining a CoMo concentration at approximately 4.0-4.3 wt.%. This showed that using a more acidic support increases the hydrocracking activity and decreases the oxygen content in the condensed organic phase from 9.0 wt.% db to between 5.2 and 6.1 wt.% db, depending on the zeolite to alu-mina ratio. However, the carbon recovery was not affected by the acidity, most likely because the zeolite increases the alkylation activity, which increases the aromatic yield through incorporation of light oxygen-ates and olefins into the condensable organics.
In order to test the process stability a 5 day semi-continuous experiment was conducted with a commercial CoMoS/MgAl2O4 catalyst in the fluid bed reactor and a NiMoS/Al2O3 catalyst in the HDO reactor. The total time on stream was 16.2 h and approximately 5 kg of biomass was used with 50 g of catalyst. The condensa-ble organic yield was fairly stable during the experiment and varied between 21.2 and 23.2 wt.% daf, but the oxygen content in the organic phase increased during the experiment from 40 to 2832 wt-ppm, indicating that some deactivation of the catalyst in fluid bed and the HDO reactor may have occurred. It should also be noted that 40 wt.% of the catalyst in fluid bed reactor was lost during the experiment due entrainment, which most likely also decreased the conversion of reactive oxygenates in the fluid bed reactor and thereby acceler-ated the deactivation of the HDO reactor. Furthermore, analysis of the spent catalyst from the fluid bed reac-tor showed that the carbon content increased with time on stream. It was 3.7 wt.% after 3.5 h, but 7.2 wt.% after 16.2 h, indicating that the carbon deposition rate decreased with time on stream. Interestingly the potas-sium and calcium content on the spent catalysts increased proportionally to the time on stream, hence propor-tionally to the amount of biomass used, thus it was 0.14 and 0.075 wt.% after 3.5 hours, respectively, but 0.67 and 0.28 wt.% after 16.2 hours, respectively. Calcium was found as single particles (40-200 nm) and therefore only expected to have a minor effect on the catalytic activity, while potassium was well-distributed on the catalyst and could therefore have a larger impact on the activity. To investigate this a CoMo catalyst was prepared and doped with 1.9 wt.% potassium prior to the sulfidation, tested in the fluid bed reactor and compared to a similar un-doped CoMoS catalyst. This showed that potassium decreases the cracking and hydrogenation while increasing the decarboxylation activity and only led to a small decrease in the total de-oxygenation activity. However, it should be noted that potassium also altered the catalyst morphology by increasing the MoS2 slab length and increasing the degree of stacking. This most likely led to the formation of the more active type II sites, which may have enhanced the deoxygenation activity. Interestingly doping the catalyst with potassium led to encapsulation of the catalyst particles with coke, indicating that potassium can act as a catalyst for polymerization reactions. Furthermore, using wheat straw instead of beech wood as feedstock, which contains 10 times more potassium, led to fast defluidization due to agglomeration and SEM images of the agglomerates showed that that potassium had acted as a catalyst for polymerization reactions.
Based on the experimental results the important chemical reactions in catalytic hydropyrolysis have been identified and a mechanistic model for catalytic hydropyrolysis is proposed. Catalytic hydropyrolysis has also been compared to other pyrolysis technologies, which indicated that it is promising process for the pro-duction of renewable liquid hydrocarbon fuels.
Original languageEnglish
Place of PublicationKgs. Lyngby
PublisherTechnical University of Denmark
Number of pages327
Publication statusPublished - 2018

Cite this

Stummann, M. Z. (2018). Hydrogen Assisted Catalytic Biomass Pyrolysis for Green Fuels. Kgs. Lyngby: Technical University of Denmark.