Catalytic deoxygenation of wheat straw fast pyrolysis vapors for production of biofuels with improved properties

Abstract

Chemicals and fuels are currently produced from fossil resources, thus contributing to global warming. The demand for renewable alternatives is increasing with tightening legislations on greenhouse gas emissions as well as increasing public pressure. Biomass is the main renewable carbonaceous feedstock that can be converted into chemicals and fuels. This thesis is dedicated to the catalytic treating of biomass fast pyrolysis vapors prior to their condensation, which allows improving the adverse properties of untreated bio-oil and offers a promising route to replace fossil-based liquid energy carriers. The research presented in this work aimed to improve the bio-oil quality by catalytic vapor upgrading without severely reducing the total liquid yield and energy recovery of the deoxygenated and stabilized bio-oil. Specifically, the thermochemical conversion of wheat straw, a high ash-containing agricultural residue, was investigated in a bench scale setup with continuous biomass feeding. A range of commercial and in-house prepared catalysts were tested under atmospheric pressure conditions in either inert (N2) or reducing (H2) atmosphere at catalyst temperatures ranging from 400 to 550 °C. The produced bio-oils were analyzed by e.g. moisture and elemental analysis, total acid number (TAN), basic nitrogen content, size exclusion chromatography (SEC), evaporation characteristics, gas chromatography (GC) with mass spectrometry (MS)/flame ionization detector (FID), GC×GC-ToF/MS or−FID, 1H and 13C nuclear magnetic resonance (NMR) spectroscopy, and 2D HSQC NMR. Besides the catalyst testing at bench scale, catalyst screening was performed in a tandem micro-pyrolyzer system, in which pyrolysis vapors contacted the catalyst in pulses and the non-condensed vapors were directly analyzed via GC-MS/FID/TCD. Characterization of both fresh and spent catalysts included elemental analysis, temperature programmed desorption (TPD) of NH3, CO2, and ethylamine, X-ray diffraction (XRD), electron microscopy (SEM-EDS and TEM), pyridine-FT-IR, 27Al NMR, thermogravimetric analysis (TGA), and pore characterization via argon and N2 physisorption. In both bench and micro-scale systems, the catalyst deactivation was studied as a function of cumulative biomass-to-catalyst ratio (B:C, w/w) and the coke deposits on the catalyst were quantified via oxidative catalyst regeneration. Initially, the effect of catalyst deactivation by coking on the bio-oil properties was studied for a standard microporous HZSM-5 zeolite. Steam-treatment prior to reaction reduced the zeolite’s acidity as such that the subsequent hydrothermal conditions during the catalytic tests did not appreciably further decrease the catalyst acidity. This allowed attributing the catalyst deactivation unambiguously to coking. The stepwise collection of bio-oils during biomass feeding and catalyst coking showed that the yield of deoxygenated hydrocarbons decreased while the yield of oxygenated compounds and the acidity (TAN) of the bio-oil increased at higher B:C ratio. At low B:C ratio, the content of aromatics was clearly enhanced and the content of sugar and aldehydes was reduced, albeit the zeolite’s initial high activity in forming coke and light gases reduced the overall oil yield. Next, mesoporosity was introduced to HZSM-5 with Si/Al = 16, 28, and 39 via desilication, followed by a mild acid wash in an attempt to allow operation to higher B:C due to the improved access of the micropores. Vapor upgrading was performed at 500 °C over the hierarchical HZSM-5 catalysts and their parent versions at the continuous bench-scale pyrolysis unit. The hierarchical samples showed an increased coking propensity compared to their microporous versions. The mesoporous HZSM-5 derived from desilication of HZSM-5 with molar Si/Al ratios of 28 and 39 showed a prolonged activity in deoxygenation and an improved carbon recovery of bio-oil compared to the parent counterparts. The mesoporous HZSM-5 derived from desilication of HZSM-5 with Si/Al = 16, showing no improved performance, had the lowest microporous volume and crystallinity, and the highest mesoporosity amongst the catalysts tested. This suggests that the addition of a mild level of mesoporosity while largely preserving crystallinity and microporous characteristics benefits the accessibility of active domains during catalyst coking. In continuation, a micro-pyrolyzer study tested a mesoporous HZSM-5 zeolite (Si/Al ~50) that was coated with a thin layer (<20 nm) of silicalite-1, resulting in an average Si/Al ratio of ~70 in order to selectively passivate external acid sites. Even though the coking propensity was highest for the mesoporous HZSM-5 (5.1 wt.% carbon recovery of fed biomass at B:C ~1), it showed improved conversion of oxygenates (15 wt.% oxygen in the non-condensed vapors) and a higher tolerance towards deactivation by coking compared to the conventional HZSM-5 (20 wt.% oxygen in the non-condensed vapors). Coating of mesoporous HZSM-5 with silicalite-1 reduced the number of acid sites per mass of catalyst by ∼20%, which is attributed to the addition of the inert silicalite-1 layer and passivation of the external acid sites of the mesoporous HZSM-5. Compared to mesoporous HZSM-5, the added silicalite-1 shell reduced the coke yield by 43% (from 5.1 to 2.9 wt.% of fed biomass carbon) for the same catalyst mass and by 8% (4.7 wt.% of fed biomass carbon) at an increased catalyst loading corresponding to the same total number of acid sites. The improved tolerance towards deactivation observed for the mesoporous HZSM-5 core was largely preserved for small oxygenates like acids, alcohols and aldehydes. A further study was conducted at the bench scale unit to compare the deoxygenation performance of ϒ-Al2O3 to HZSM-5 and extrudates comprised of both components. In addition, mesoporosity was added to the zeolite component of HZSM-5/Al2O3 via desilication. Compared to ϒ-Al2O3, catalysts containing HZSM-5 promoted aromatization and limited the coke formation due to its shape selective micropores. Nevertheless, ϒ-Al2O3 was effective in deoxygenation and as such offers certain benefits such as low cost and good hydrothermal stability with respect to acidity. Impregnation of ϒ-Al2O3 with Na2CO3 (20 wt.%) added basicity, while the catalyst acidity was reduced by 80%. Micro-pyrolyzer tests showed that Na-Al2O3 particularly decreased the yield of acids and showed a high yield of ketones and CO2, indicating ketonization. Na catalyzed the coke combustion and decreased the combustion temperature by ~100 °C, while decreasing the coke yield. Subsequent bench scale testing investigated catalyst stability during six reaction/regeneration cycles. Na-Al2O3 was highly effective in reducing the acidity of the bio-oils and low TAN (<4 mg KOH/mg) could be maintained up to high B:C ratios of ~13. For a given TAN, this allowed operating to higher B:C ratios and provided higher oil yields compared to using acidic catalysts such as ϒ-Al2O3 and HZSM-5 zeolite for vapor treatment. At bio-oil energy recoveries of ~60-70% relative to raw bio-oil, the deoxygenation was comparable to the acidic catalysts. Operation to higher B:C ratios while still obtaining a good deoxygenation performance of ~60% allowed increasing the energy recovery to ~85% relative to the non-treated bio-oil. The catalyst activity was stable and the Na remained well dispersed on the support despite hydrothermal conditions. In an attempt to further increase the energy recovery and maintain a high level of deoxygenation, atmospheric hydrodeoxygenation (HDO) was investigated over TiO2 supported Pt (0.5 wt.%) and MoO3 (10 wt.%) catalysts, and an industrial molybdenum-based reducible metal oxide catalyst (MoO3/Al2O3). At 50 vol.% H2, all three HDO catalysts effectively reduced the oxygen content of the bio-oils to ~7-12 wt.% (dry basis) compared to a non-catalytic reference (23 wt.% O). MoO3/TiO2 was least efficient in conversion of acids (TAN = 28 mg/KOH), while Pt/TiO2 and MoO3/Al2O3 obtained oils with TAN ~13 mg KOH/g (non-catalytic = 66 mg KOH/g). Compared to the TiO2 supported catalysts, the industrial MoO3/Al2O3 catalyst produced higher yields of coke at the expense of condensed bio-oil. At 50 vol.% H2, MoO3/TiO2 performed identical to Pt/TiO2 in terms of deoxygenation and energy recovery of condensed bio-oil, and at increased H2 concentration the energy recovery could be further increased. Pt/TiO2 was more selective to aliphatics while MoO3 based catalysts favored aromatics due to a lower hydrogenation activity. Pt/TiO2 showed the lowest coke yields as the coke yield at B:C ~8 was only 0.6 wt.% of fed biomass while obtaining bio-oil with 11 wt.% O at a carbon and energy recovery of 34 C% and 42%, respectively. The fluidized catalytic cracking of bio-oil obtained from vapor treatment with HZSM-5/ϒ-Al2O3 and bio-oils obtained without treatment from wheat straw and wood was investigated in collaboration with Equinor using a micro-activity testing unit. The bio-oils were co-processed at a 20/80 weight blend (bio-oil/FCC feed) and the change in product distribution was studied. At constant conversion (77.5%) at the MAT, the wood pyrolysis oil showed a product distribution quite similar to the reference oil while the wheat straw pyrolysis oil gave a 1.6 percentage points higher coke yield and a 1.2 percentage point lower liquid petroleum gas yield. For the catalytically treated wheat straw pyrolysis oil, an even higher coke yield (2.6 percentage points) and 1.9 percentage points lower LPG yield resulted. The observations are attributed to the higher content of aromatics, phenolics, and nitrogen containing compounds of the catalytically upgraded straw fast pyrolysis oil, leading to faster catalyst coking. Lastly, the deoxygenation of tar vapors that were generated in a low temperature circulating fluidized bed (LT-CFB) gasifier was investigated as a flexible process to co-produce high quality bio-oil, nutrient rich char/ash, and utilize the producer gas for heat and power production. With decrease in the LT-CFB pyrolysis temperature from 690 to 570 °C, the producer gas contained less gas and more condensable organics. The higher heating value (HHV) of the bio-oil decreased from ~35 to 30 MJ/kg due and the oxygen content, moisture content, and acidity of the bio-oil increased, while the concentration of aromatics and phenols decreased. The yield of the condensable organics obtained with the LT-CFB gasifier operating in the temperature range of 630-660 °C was lower (~12-16 wt.% of fed biomass) compared with fast pyrolysis temperatures of 500-550 °C for maximum liquid production (~36 wt.% of fed biomass). ϒ-Al2O3 and HZSM-5/ϒ-Al2O3 were tested for tar upgrading and both catalysts improved the bio-oil quality in terms of increased HHV and decreased oxygen content, moisture content, TAN, and basic nitrogen content. The vapor treatment with HZSM-5/ϒ-Al2O3 decreased the energy yield of bio-oil from ~22% to ~20%, while the oxygen content and TAN of the bio-oil decreased from 13 wt.% O and 34 mg KOH/g to 11 wt.% O and 3 mg KOH/g, respectively. The catalytically treated bio-oils thus are better suited for further processing in existing oil refineries. This work shows that a variety of heterogeneous catalysts are active in vapor deoxygenation, but rapid deactivation tends to occur for catalysts with strong acidic nature. Modification of the pore-structure can to some extent lower the rate of deactivation and improve the performance of HZSM-5 zeolite. However, alternative catalyst systems such as Na-Al2O3 appear stable and can obtain comparable or even better energy recovery of bio-oil compared to HZSM-5 at reasonable deoxygenation of ~60-70%. The energy recovery of bio-oil improves further when applying TiO2 supported MoO3 or Pt catalyst under hydrogen atmosphere. Our results suggest that mild deoxygenation may be a viable way of pretreating wheat straw-derived pyrolysis vapors before their further processing via FCC or hydrotreating in conventional refineries or direct application as a renewable fuel for ship diesel engines.