El hidrodeoxigenación (HDO) de aceites obtenidos por pirólisis es una tecnología para producir combustible diésel, gasolina, combustibles de aviación (JP) a partir de madera. En este trabajo, se han analizado los productos y se describen las reacciones químicas que se producen durante el proceso de HDO de biomasa lignocelulósica.
The hydrodeoxygenation (HDO) of bio-oil derived from white oak wood using non-sulfided catalysts was studied in a two zone continuous flow trickle bed reactor system. The major organic components of the pyrolysis oil were pyrolytic lignin (large phenolic polymers), xylose, levoglucosan, organic acids (primarily acetic acid), and hydroxyacetaldehyde. The first zone was a low temperature zone (130 °C) that contained a Ru/C catalyst. In this zone, carbonyl groups were hydrogenated, producing propylene glycol (from hydroxyacetone), ethylene glycol (from hydroxyacetaldehyde), and sorbitol (from levoglucosan). A more severe hydrotreatment was performed in a second zone containing a bifunctional Pt/ZrP catalyst at a temperature between 300 and 400 °C. In the two-stage HDO, an organic phase was produced that consisted of a distribution of hydrocarbons that were primarily cyclic alkanes (naphthenes) ranging from C7 to C24. The organic phase carbon yield decreased with increasing reaction temperature in the second zone. Catalyst deactivation and reactor plugging by coking occurred under all reaction conditions after 55–72 h time on stream (TOS). After ≈55 h TOS, more than 25 % of the carbon in the original bio-oil was accumulated as coke, with increasing amounts for higher temperatures in the second zone. Hydrotreatment gave rise to >C5 hydrocarbon (gasoline and distillate-range fuel) overall yields between ≈30 and 47 carbon % for all experiments compared to the 79.5 % theoretical yield calculated for the bio-oil feedstock. Coke formation and undesired cracking to C1–C4 hydrocarbon gases were the main causes of lower fuel carbon yields.
Introduction
As non-renewable petroleum resources have been deemed a major cause of global climate change, attention has shifted towards renewable and sustainable sources of liquid transportation fuels. Lignocellulosic biomass has received significant attention as a feedstock to make renewable transportation fuels because it is both cheap and abundant. Several approaches are currently being studied to produce renewable transportation fuels from lignocellulosic biomass, including fast pyrolysis, gasification, and biochemical approaches.[1-4] The pyrolysis route for biomass conversion has several technical and economic advantages over gasification and biological approaches, including:
- Lower total capital investment and operating costs for liquid fuel production.[4]
- The ability to easily process mixed feedstocks including agricultural wastes, grasses, and woody biomass.[1, 2, 5]
- Shorter residence time and simpler feedstock pretreatment (drying and grinding) compared to biological approaches.[2, 6]
Several studies discussing the production,[7, 8] properties,[9] applications,[10] and HDO[11-15] of pyrolysis oil and the techno-economic analysis of the pyrolysis process[3, 4, 16] have been performed. Much of this prior work centers on the improvement of bio-oil (pyrolysis oil) properties in relation to traditional petroleum feedstocks. Bio-oils typically contain ≈50 wt % carbon, ≈7 wt % hydrogen, and ≈43 wt % oxygen.[17] The high oxygen and water contents results in bio-oil having a lower energy density (15–19 MJ kg−1) compared to petroleum oil (40 MJ kg−1). Additionally, the bio-oil is unstable, phase separates with time, and is not miscible with petroleum-derived fuels. Due to these limitations, pyrolysis oil cannot be directly used as a liquid transportation fuel in conventional engines and must be upgraded into fungible fuels.
Several approaches have been utilized to upgrade bio-oil into liquid fuels. Vispute and Huber separated the bio-oil into two fractions: a water-soluble and a water-insoluble fraction mainly consisting of pyrolytic lignin.[18] The aqueous fraction made up 50 % of the total C present in bio-oil, with the remaining carbon in the lignin phase. The low-temperature hydrogenation of the aqueous fraction of bio-oil using Ru/C as catalyst yielded polyols (propylene glycol, ethylene glycol) and sorbitol.[18] The hydrotreatment of lignin in the presence of precious-metal catalysts produces naphthenes.[19, 20] Bio-oils can also be converted into renewable aromatics or olefins by combining hydrotreating with a zeolite catalyst.[21]
The entire bio-oil has also been converted into liquid fuels by hydrotreatment or hydrocracking.[12, 14, 22, 23] The hydrotreatment of bio-oil has been carried out in a fixed-bed reactor at high temperature (310–375 °C) using Pd/C as catalyst to produce ketones, acids, esters, phenols, and alkylated phenols. These products were subsequently hydrocracked in the presence of a conventional hydrotreating catalyst at 400 °C and 2000 psi (1 psi=6.89 kPa).[23] Elliott et al. observed that the rate of deactivation of the hydrotreatment catalyst depended on feed flowrate and reaction temperature.[12] Catalyst deactivation during the hydrotreatment reaction is a serious issue as it leads to deactivation of the active catalyst phase.[21] The phenolic compounds, or pyrolytic lignin, adsorb strongly on acidic sites on alumina forming phenate species that were proposed to be precursors for coking.[13] Ab initio studies revealed that phenolic groups preferentially form keto groups in the aqueous phase.[24] Formation of network polymers derived from these keto groups is thought to be a main mechanism for coking.[25] Pyrolytic lignin even forms coke on the catalyst surface at temperatures as low as 25 °C.[20]
Few studies are available in the literature that explore the fundamental chemistry occurring or in which a detailed analysis of the reaction products obtained in this process was performed. One of the objectives of this paper is to study the HDO of a well-characterized bio-oil and analyze the full range of products formed during the two-step catalytic HDO process. By fully characterizing the products after low and high temperature treatments, the fundamental reaction chemistry involved can be delineated. We also analyze the fuel-grade yields from the dual-stage HDO process studied in this work and compare it to other reported yields in academia and industry. This paper thus provides insight into how improved catalysts could be designed for conversion of bio-oil and which biofuel yields can be realistically achieved through the successful implementation of these catalysts
Results and Discussion
Characterization of pyrolysis oil
The composition of pyrolysis oil as characterized by Karl–Fisher, GC, HPLC, and elemental analysis is presented in Table 1. An elemental analysis showed that bio-oil contained 43.5 % C, 7.1 % H, and 49.4 % O. Karl–Fisher analysis showed that the pyrolysis oil contained 21.5 wt % water. Previous characterization of this pyrolysis oil revealed viscosities of 0.1771 Pa s at 25 °C and 0.0638 Pa s at 40 °C.[26] Microfiltration with a 0.8 μm membrane similar to that previously performed in literature resulted in approximately 3 wt % of the bio-oil being removed as char particles.[26] Out of the total carbon content, ≈30 % was present as acids, aldehydes, ketones, and furan-based compounds whereas ≈20 % of the bio-oil was detected in the form of levoglucosan, sorbitol, and xylose. Sugars and their derivatives comprised a large portion of the bio-oil: mainly C5 sugars (11.65 wt %) and levoglucosan (7.17 wt %). Major components derived from hemicellulose include acetic acid (9.73 wt %), hydroxyacetaldehyde (6.17 wt %), dimethyl tetrahydrofuran (2.49 wt %), and hydroxyacetone (1.68 wt %). Solubility tests indicated that 51 % of the carbon (C %) making up the bio-oil was pyrolytic lignin. Carbon derived from the pyrolytic lignin was undetectable in GC and HPLC due to their high boiling point, which is consistent with other studies.[20] The summation of the carbon detected by HPLC, GC, and solubility tests was 102 C %, indicating that our characterization methods were able to detect all the carbon present in the bio-oil with slight calibration errors.
| Component | Molecular formula | Concentration [mmol L−1] | TOC [%] |
|---|---|---|---|
|
|||
| methanol | CH3OH | 120 | 0.34 |
| hydroxyacetaldehyde | C2H4O2 | 2160 | 6.17 |
| acetic acid | C2H4O2 | 3406 | 9.73 |
| ethylene glycol | C2H6O2 | 38 | 0.11 |
| methyl acetate | C3H6O2 | 50 | 0.14 |
| hydroxyacetone | C3H6O2 | 589 | 1.68 |
| propanoic acid | C3H6O2 | 177 | 0.51 |
| propylene glycol | C3H8O2 | 43 | 0.12 |
| 2 (5 H)-furanone | C4H4O2 | 182 | 0.52 |
| γ-butyrolactone | C4H6O2 | 68 | 0.19 |
| tetrahydrofuran | C4H8O | 31 | 0.09 |
| 1-hydroxy-2-butanone | C4H8O2 | 325 | 0.93 |
| butanol | C4H10O | 17 | 0.05 |
| 2,3-butanediol | C4H10O2 | 90 | 0.26 |
| furfural | C5H4O2 | 188 | 0.54 |
| R-γ-hydroxymethyl-γ-butyrolactone | C5H8O3 | 8 | 0.25 |
| 2-methyl tetrahydrofuran | C5H10O | 600 | 1.71 |
| cyclopentanol | C5H10O | 23 | 0.06 |
| C5 sugars (HPLC)[a] | C5HxOy | 4079 | 11.65 |
| phenol | C6H6O | 4 | 0.01 |
| catechol | C6H6O2 | 376 | 1.07 |
| 5-hydroxymethylfurfural | C6H6O3 | 198 | 0.57 |
| 3-methyl-1,2-cyclopentanedione | C6H8O2 | 134 | 0.38 |
| levoglucosan (HPLC) | C6H10O5 | 2508 | 7.17 |
| 2,5-dimethyl tetrahydrofuran | C6H12O | 872 | 2.49 |
| cyclohexanol | C6H12O | 282 | 0.80 |
| 1,2-cyclohexanediol | C6H12O2 | 86 | 0.25 |
| hexanoic acid | C6H12O2 | 356 | 1.02 |
| sorbitol (HPLC) | C6H14O6 | 691 | 1.97 |
| guaiacol | C7H8O2 | 63 | 0.18 |
| total C identified by GC and HPLC | 17 843 | 51.0 | |
| pyrolytic lignin | 51 | ||
| total C present in bio-oil | 35 000 | 102 | |
The carbon distribution is presented in Figure 1. Out of the total carbon content ≈50 % of compounds are C2, C5, and C6 carbons. C5 carbons mostly include xylose, with appreciable amounts of 2-methyl tetrahydrofuran and furfural also present. C6 carbons include mostly levoglucosan and smaller amounts of phenols and cyclohexanols. The C7 carbons are mainly guaiacol. The remaining ≈50 % of carbon in the bio-oil is detected as high molecular weight (MW) pyrolytic lignin with a carbon number greater than 7. Pyrolytic lignin consists of phenolic oligomers that can range anywhere from 50 to 3000 Da but that have an average MW between 400 and 1300 Da.[20, 27]
Distribution of compounds present in bio-oil based on GC and HPLC analysis.
HDO of pyrolysis oil
Single-stage hydrogenation of bio-oil over Ru/C
We first studied the low-temperature hydrogenation of pyrolysis oil over Ru/C at 130 °C (Tables 2, 3, 4). Overall mass and carbon balances for the process are shown in Table 2. The carbon selectivity of the liquid phase products [Eq. (1)] and the total amount of carbon identified in the liquid phase [Eq. (2)] were determined using gas chromatography (GC) and total organic carbon (TOC) analysis. In the first 16 h time on stream (TOS), mostly liquid products were obtained at ≈9 C % (≈16 wt %) yield. This suggests that the process has either not reached steady state or that there are large amounts of carbon components being formed on the catalyst surface. After 24 h TOS, more than 55 C % (66 wt %) of the bio-oil is converted to either liquid or gaseous products. The overall material balance indicates that ≈47 C % of the bio-oil is converted to liquid- or gas-phase products, with the remaining ≈53 C % of the products forming an unidentified phase, which is most likely coke on the catalyst surface, over the course of the reaction. After 64 h TOS, the pressure of the reactor increased to over 2200 psi and the reactor bed was plugged due to the accumulation of solid coke on the catalyst surface.
(1)
(2)| TOS | System pressure | Carbon content [wt %] | |||
|---|---|---|---|---|---|
| [h] | [psi] | liquid phase | gas phase | total liquid+gas | unidentified |
|
|||||
| 16 | 2020 | 9.3 (15.5) | 0.7 (0.5) | 10.0 (16.0) | 90.0 (84.0) |
| 24 | 2046 | 54.0 (65.1) | 1.4 (1.1) | 55.4 (66.2) | 44.6 (33.8) |
| 62 | 2154 | 56.5 (70.7) | 1.8 (1.0) | 58.3 (71.6) | 41.7 (28.4) |
| 64 | 2200 | 72.8 (71.6) | 1.4 (1.0) | 74.2 (72.5) | 25.8 (27.5) |
| overall | – | 45.7 (56.2) | 1.5 (0.9) | 47.2 (57.1) | 52.8 (42.9) |
As shown in Table 3, very little carbon (less than 2 C %) was converted to gaseous products in the low-temperature hydrogenation. The gas stream mainly contained excess hydrogen and small amounts of C1–C6 hydrocarbons. The gas-phase selectivity stabilized after 24 h TOS, with the selectivity at 62 h TOS as follows: butane (34 %)>propane (25 %)≈pentane (24 %)>hexane (12 %)>methane (5 %)>ethane (1 %). The goal of the low-temperature hydrogenation step was to reduce some of the more reactive alkene, aromatic, phenolic, and carbonyl groups that are likely responsible for coke formation. A variety of functionalities in the bio-oil feed were hydrogenated during this process (Table 4), including carboxylic acids (formic, acetic, propanoic), aldehydes (acetaldehyde, hydroxyacetaldehyde), ketones (hydroxyacetone, hydroxybutanone, acetone), phenols (guaiacols, syringols, catechols), furfurals (furfural, 5-hydroxymethyl furfural), and lignin-based components. Sugars (xylose), sugar alcohols (sorbitol), and anhydrosugars (levoglucosan) were not reactive in this hydrogenation step. However, the concentration of hydroxyacetaldehyde, acetic acid, 2 (5 H)-furanone, and furfural decreased with TOS. Although hydrogenation of these compounds may have occurred to some extent (suggested by slight increase in ethanol, propylene glycol, and tetrahydrofuran selectivities), the majority of the converted carbonyl groups remained unaccounted for. It is possible that these reactive carbonyl groups were consumed by aldol condensation reactions that result in coke formation. Additionally, the water content of the liquid product increased from 21 to 30 wt % after reaction. This increase in water content has three different sources: i) HDO, ii) carbon loss from coking reactions, or iii) dehydration or aldol condensation reactions. As a large degree of deoxygenation did not occur in this reaction, the water increase is most likely due to coke formation.
| Component | Selectivity [%] | |||
|---|---|---|---|---|
| 16 h | 24 h | 62 h | 64 h | |
|
||||
| CH4 | 0.00 | 0.00 | 4.64 | 4.13 |
| C2H6 | 0.00 | 0.00 | 1.26 | 1.33 |
| C3H8 | 0.00 | 23.51 | 24.75 | 23.73 |
| C4H10 | 0.00 | 31.37 | 34.05 | 33.39 |
| C5H12 | 67.91 | 30.41 | 23.57 | 25.58 |
| C6H14 | 32.09 | 14.71 | 11.73 | 11.83 |
| Component | Number of | Selectivity [%] | ||||
|---|---|---|---|---|---|---|
| C atoms | feed | 16 h | 24 h | 62 h | 64 h | |
|
||||||
| methanol | 1 | 0.34 | 0.00 | 0.00 | 0.00 | 0.00 |
| ethanol | 2 | – | 0.22 | 0.04 | 0.00 | 0.00 |
| acetic acid | 2 | 9.73 | 3.95 | 1.25 | 1.19 | 7.22 |
| hydroxyacetaldehyde | 2 | 6.0 | 0.08 | 0.11 | 0.00 | 0.00 |
| methyl acetate | 3 | 0.14 | 29.35 | 14.98 | 1.34 | 6.46 |
| 1-propanol | 3 | – | 0.44 | 0.03 | 0.06 | 0.09 |
| propanoic acid | 3 | 0.51 | 0.40 | 0.40 | 0.40 | 2.38 |
| propylene glycol | 3 | 0.12 | 0.39 | 0.08 | 0.07 | 0.25 |
| tetrahydrofuran | 4 | 0.09 | 3.52 | 0.22 | 0.00 | 0.00 |
| 1-hydroxy-2-butanone | 4 | 0.93 | 0.00 | 0.12 | 0.00 | 0.00 |
| γ-butyrolactone | 4 | 0.19 | 0.00 | 0.04 | 0.09 | 0.62 |
| 2 (5 H)-furanone | 4 | 0.52 | 0.00 | 0.15 | 0.00 | 0.00 |
| 1-pentanol | 5 | – | 0.00 | 0.06 | 0.07 | 0.52 |
| furfural | 5 | 0.54 | 0.00 | 0.07 | 0.00 | 0.00 |
| C5 sugars (HPLC) | 5 | 11.65 | 13.76 | 14.06 | 20.12 | 18.25 |
| hexanoic acid | 6 | 1.02 | 1.02 | 0.54 | 0.22 | 1.44 |
| 3-methyl cyclopentanol | 6 | – | 0.00 | 0.11 | 0.00 | 0.00 |
| phenol | 6 | 0.03 | 0.00 | 0.00 | 0.00 | 0.03 |
| levoglucosan (HPLC) | 6 | 7.17 | 8.27 | 12.85 | 13.97 | 23.35 |
| sorbitol (HPLC) | 6 | 1.97 | 2.36 | 3.23 | 4.39 | 2.76 |
| C % identified | 51 | 63.8 | 48.3 | 41.9 | 63.4 | |
| C % unidentified | 49 | 36.2 | 51.7 | 58.1 | 36.6 | |
The concentrations of sugar and levoglucosan increased with time, and very few hydrolysis products, such as ethylene glycol, propylene glycol, and 1,4-butanediols, were observed probably because of the absence of acid sites on the catalyst. This increase is probably attributable to cleavage of oligomeric sugar species. Unidentified products in the aqueous phase were likely similar highly oxygenated species our GC could not identify. This is consistent with the work of Vispute and Huber, who reported that the sugar and levoglucosan present in the aqueous phase of bio-oil do not undergo hydrogenation at 130 °C in the presence of a Ru/C catalyst.[18]
The bio-oil and low-temperature hydrogenated products were further characterized using gel permeation chromatography (GPC; Figure 2). The bio-oil has high MW species ranging from 100–10 000 Da. The broad bands observed at 1000, 300, and 120 Da most likely represent distinct lignin polymeric species.[28] The hydrogenated bio-oil products have similar features as the bio-oil feed. The hydrogenated bio-oil obtained at 24 h TOS is characterized by a broad peak at ≈2000 Da. With an increase in TOS from 24 to 44 h, the broad peak shifted towards lower MWs, suggesting that either the HDO resulted in the decomposition of the large-MW species to smaller molecules or the larger MW species were polymerized inside the reactor forming coke deposits. It is possible that this shift occurred due to the cleavage of ether linkages in the pyrolytic lignin. Only small concentrations of light components in the GPC are observed after reaction, indicating that the smaller-MW bio-oil components were converted to gaseous products or products that were not detectable with our LC detector.
GPC of hydrogenated bio-oil samples collected from low-temperature hydrogenation (130 °C) over Ru/C.
Two-stage HDO of bio-oil
The bio-oil was then treated in a two-stage trickle bed reactor in which the first reactor contained the same Ru/C catalyst as in the one-stage 130 °C hydrogenation and the second stage contained a Pt/ZrP catalyst at three different temperatures (300, 350, and 400 °C). Organic-liquid-phase, aqueous-liquid-phase, and gas-phase products were observed for all three reaction conditions. The reaction was continued until the reactor beds were plugged, which was indicated by an increase in system pressure. A similar two-stage setup for HDO of aqueous carbohydrate streams derived from maple wood and of the water-soluble portion of bio-oil was studied in the past.[29, 30] We also studied the HDO of sorbitol and developed a reaction pathway for the conversion of sorbitol.[29, 31]
Ru/C–Pt/ZrP (300 °C)
Product analysis of the two-stage HDO with Ru/C at 130 °C and Pt/ZrP at 300 °C is presented in Tables 5, 6, 7. The overall mass and carbon balances are presented in Table 5. The overall yields of the different phases were 40.5 C % (≈21 wt %) organic-phase, 3.3 C % (≈22 wt %) aqueous-phase, 27.8 C % (≈14 wt %) gas-phase products, with 28.3 C % unidentified. A hydrogen consumption of 7.30 g H2 per 100 g dry bio-oil and a degree of deoxygenation of at least 95.6 % was calculated from the observed products. While coking plays a major role at early TOS, the majority of the aqueous phase at longer TOS is likely water formed as byproduct. The increase in water production for the two-stage HDO versus the low temperature hydrogenation (Table 2) is due to the higher degree of deoxygenation. After 45 h TOS, approximately 80 wt % and 80 C % in the original bio-oil was detected. According to Table 6, the major aqueous-phase products in the first 25 h TOS are methanol and ethanol, which account for ≈75 % of total carbon present in the aqueous phase. At 45 h TOS, the selectivity toward acetic acid, methyl acetate, 1-propanol, and levoglucosan increased. The selectivity toward methanol and ethanol decreased with TOS. Over 14 C % was converted into gas-phase molecules after 45 h TOS. This is more than achieved for the single-zone reaction, which only converted 2 % of the carbon into gas-phase products. Gas-phase selectivity at 66 h TOS decreased in the order methane (≈29 %)>ethane (≈21 %), >hexane (≈13 %)>propane (≈15 %)>butane (≈12 %)>pentane (≈9 %) (Table 7). After 66 h TOS, the pressure increased to a level at which the gas flow controllers shut off (above 2200 psi) and the experiment was shut down.
| TOS | System pressure | Carbon content [wt %] | ||||
|---|---|---|---|---|---|---|
| [h] | [psi] | organic phase | aqueous phase | gas phase | total liquid+gas | unidentified |
|
||||||
| 25 | 2030 | 0.0 (0.0) | 0.7 (13.0) | 46.5 (23.7) | 47.2 (36.7) | 52.8 (63.3) |
| 45 | 2099 | 56.4 (31.3) | 5.7 (39.2) | 18.5 (9.4) | 80.6 (79.9) | 19.4 (20.1) |
| 66 | 2158 | 73.6 (37.5) | 4.2 (16.7) | 14.5 (7.4) | 92.2 (61.5) | 7.8 (38.5) |
| overall | – | 40.5 (21.4) | 3.3 (22.1) | 27.8 (14.2) | 71.7 (57.7) | 28.3 (42.3) |
| Components | Number of | Selectivity [%] | |||
|---|---|---|---|---|---|
| C atoms | feed | 25 h | 45 h | 66 h | |
|
|||||
| methanol | 1 | 0.34 | 63.24 | 13.95 | 5.89 |
| ethanol | 2 | – | 11.59 | 9.37 | 3.88 |
| acetic acid | 2 | 9.73 | 0.00 | 7.45 | 0.18 |
| methyl acetate | 3 | 0.14 | 0.00 | 2.78 | 1.16 |
| 1-propanol | 3 | – | 0.00 | 2.36 | 1.72 |
| tetrahydrofuran | 3 | 0.09 | 0.00 | 0.67 | 0.23 |
| hexanoic acid | 3 | 1.02 | 0.00 | 0.75 | 0.00 |
| 3-methyl cyclopentanol | 3 | – | 0.00 | 0.65 | 0.33 |
| 3-methyl-1,2-cyclopentanedione | 3 | 0.38 | 0.00 | 0.03 | 0.21 |
| ethyl acetate | 4 | – | 0.00 | 0.26 | 0.09 |
| 1-butanol | 4 | 0.05 | 0.00 | 0.25 | 0.45 |
| 2-butanol | 4 | 0.05 | 0.00 | 0.05 | 0.18 |
| Tetrahydro-2-methyl furan | 4 | 1.71 | 0.18 | 0.00 | 0.00 |
| propanoic acid | 4 | 0.51 | 0.00 | 0.47 | 0.79 |
| guaiacol | 4 | 0.18 | 0.00 | 0.05 | 0.05 |
| tetrahydrofuran-2,5-dimethyl | 5 | 2.49 | 0.00 | 1.10 | 4.82 |
| ethylene glycol | 5 | 0.11 | 0.00 | 0.03 | 0.16 |
| furfural | 5 | 0.54 | 0.00 | 0.00 | 0.12 |
| γ-butyrolactone | 5 | 0.19 | 0.00 | 0.07 | 0.70 |
| C5 sugars (HPLC) | 5 | 11.65 | 0.00 | 0.00 | 0.00 |
| 2-pentanol | 6 | – | 0.00 | 0.09 | 0.13 |
| phenol | 6 | 0.01 | 0.00 | 0.01 | 0.01 |
| γ-valerolactone | 6 | – | 0.00 | 0.03 | 0.20 |
| R-γ-hydroxymethyl-γ-butyrolactone | 6 | 0.25 | 0.18 | 0.11 | 0.00 |
| levoglucosan (HPLC) | 6 | 7.17 | 0.00 | 4.87 | 5.13 |
| sorbitol (HPLC) | 6 | 1.97 | 0.00 | 0.00 | 0.00 |
| C% identified | 75.19 | 45.49 | 26.34 | ||
| C% unidentified | 24.81 | 54.51 | 73.46 | ||
| Component | Selectivity [%] | ||
|---|---|---|---|
| 25 h | 45 h | 66 h | |
|
|||
| CH4 | 37.31 | 33.23 | 29.22 |
| C2H6 | 22.61 | 16.05 | 20.99 |
| C3H8 | 11.91 | 12.70 | 15.49 |
| C4H10 | 10.22 | 10.80 | 12.33 |
| C5H12 | 8.87 | 10.36 | 8.83 |
| C6H14 | 9.09 | 16.85 | 13.15 |
Nitric oxide ionization spectroscopy evaluation (NOISE) was carried out for the organic-phase products. NOISE analysis quantifies the relative amounts of various functionalities (isoparaffins, n-paraffins, naphthenes, aromatics, and naphthalenes) in complex mixtures by combining NO+ ionization and GC–MS spectroscopy. This is possible through the unique interactions between NO+ ions and different hydrocarbon functional groups. This is very useful for the analysis of bio-oil and bio-oil HDO products and can provide insight into the change in the product distribution under various conditions, such as reaction temperature discussed herein. The selectivities of the organic phase products at 66 h TOS are presented in Figure 3. The plot shows that the organic products mostly consist of cyclic alkanes (naphthenes), aromatics, and naphthalenes. Most of these compounds have carbon numbers between 9 and 17. Very small amounts of iso- and normal paraffins were obtained from this process.
Carbon distribution for organic products obtained from 130–300°C two-stage HDO of bio-oil after 66 h TOS.
The GPC of aqueous and organic products are presented in Figure 4 a and b, respectively. The bio-oil has sharp peaks at 100, 200, and 400 Da and broad bands from 1000–5000 Da. The aqueous-phase products show sharp peaks at ≈12, ≈60, ≈100, and ≈150 Da and broader peaks at ≈400 and ≈900 Da. The intensity of the products slightly increased with increasing TOS. The organic products obtained at 45 and 66 h TOS show peaks with less intensity at 200 Da (Figure 4 b). Moreover, the intensity of the broad band between 1000 and 5000 Da decreased substantially. This suggests that most of the large molecules were hydrogenated to produce smaller molecules or converted to coke.
GPC of a) aqueous-phase and b) organic-phase products from 130–300°C two-stage HDO of bio-oil.
Ru/C–Pt/ZrP (350 °C)
Product analysis of the two-stage HDO over Ru/C at 130 °C and Pt/ZrP at 350 °C is presented in Tables 8, 9, 10. Based on the mass and carbon balances in Table 8, the organic phase first appeared after 48 h TOS and increased from ≈13 to 28 C % (≈5 to 11 wt %) when increasing TOS from 48 to 72 h. The amount of unidentified products decreased with reaction time, but overall 27 % of the carbon was unidentified. The aqueous phase of the product contained less than 7 C %. The majority of the carbon in the aqueous phase was not detectable by the GC and HPLC methods used in this paper. The major aqueous-phase product in the first 24 h is methanol, which accounts for ≈15 % of total carbon present in the aqueous phase (Table 9). The majority of the carbon compounds was detected in the gas phase (≈43–54 C %). This is significantly more than that detected for the single zone process and slightly more than that obtained for the two-stage hydrogenation at 300 °C. The higher temperature in the second bed is responsible for the increase in the gas-phase carbon yield. A decrease in the cracking activity of the catalyst to methane and ethane was observed with increasing TOS (Table 10). The reaction was stopped after 72 h TOS due to reactor plugging.
| TOS | System pressure | Carbon content [wt %] | ||||
|---|---|---|---|---|---|---|
| [h] | [psi] | organic phase | aqueous phase | gas phase | total liquid+gas | unidentified |
|
||||||
| 25 | 2030 | 0.0 (0.0) | 6.7 (18.2) | 42.9 (21.8) | 49.6 (40.0) | 50.4 (60.0) |
| 48 | 2090 | 12.7 (5.1) | 3.5 (39.5) | 54.0 (27.5) | 70.2 (72.1) | 29.8 (27.9) |
| 72 | 2174 | 27.7 (11.3) | 1.6 (26.5) | 52.3 (26.6) | 81.6 (64.5) | 18.4 (35.5) |
| overall | – | 14.5 (5.4) | 4.4 (27.8) | 54.1 (25.2) | 73.0 (58.4) | 27.0 (41.6) |
| Component | Number of | Selectivity [%] | ||
|---|---|---|---|---|
| C atoms | 25 h | 48 h | 72 h | |
|
||||
| methanol | 1 | 14.47 | 5.01 | 8.33 |
| ethanol | 2 | 0.23 | 0.00 | 0.00 |
| 2-propanol | 3 | 1.06 | 0.45 | 2.12 |
| methyl acetate | 3 | 0.21 | 0.00 | 0.00 |
| tetrahydrofuran | 3 | 0.11 | 0.00 | 0.00 |
| 2-butanol | 4 | 0.11 | 0.00 | 0.00 |
| 2-pentanol | 5 | 0.14 | 0.00 | 0.00 |
| C5 sugars (HPLC) | 5 | 0.16 | 0.16 | 0.00 |
| levoglucosan (HPLC) | 6 | 0.05 | 0.03 | 0.00 |
| sorbitol (HPLC) | 6 | 0.00 | 0.00 | 0.00 |
| C % identified | 16.54 | 5.65 | 10.45 | |
| C % unidentified | 83.48 | 94.35 | 89.55 | |
| Component | Selectivity [%] | ||
|---|---|---|---|
| 25 h | 48 h | 72 h | |
|
|||
| CH4 | 92.39 | 43.46 | 11.99 |
| C2H6 | 4.79 | 21.46 | 7.63 |
| C3H8 | 1.09 | 9.64 | 46.81 |
| C4H10 | 1.73 | 8.08 | 3.49 |
| C5H12 | 0.00 | 8.14 | 19.58 |
| C6H14 | 0.00 | 9.23 | 10.50 |
The selectivities of the organic products obtained from NOISE analysis are presented in Figure 5. The plot shows that the organic products mostly consist of cyclic compounds with carbon number 11 to 19. Very small amount of paraffins and aromatics are obtained during the HDO at 350 °C compared to 300 °C. The organic phase mostly contained alkyl cycloalkane components. These molecules were mostly in the C13–C17 range.
Carbon distribution for organic products obtained from 130–350°C two-stage HDO of bio-oil after 72 h TOS.
Ru/C–Pt/ZrP (400 °C)
Product analysis of the two-stage hydrogenation over Ru/C at 130 °C and Pt/ZrP at 400 °C is presented in Tables 11, 12, 13. Table 11 shows that no separate organic phase formed at 24 h while only 35 wt % and 12.7 C % was detected at this time. However, an organic phase was observed at 55 h TOS and the mass balance and the carbon balance improved to 100 wt % and 87 C %, respectively. In the first 24 h TOS, the products ethylene glycol, hydroxyl butanone, cyclohexanediol, and butanediol showed carbon selectivities of roughly 24 %, 11 %, 24 %, and 20 %, respectively (Table 12). However, after 55 h TOS, these products disappeared and only small amounts of organic products (ethyl cyclohexane, propyl cyclohexane, and trans-methyl cyclohexane) are observed in the aqueous phase. At 55 h TOS, only 1.8 % of the carbon is present in the aqueous phase. The selectivities of various components of organic products obtained after hydrogenation at 400 °C are presented in Figure 6. The plot shows that the organic products mostly consist of naphthene compounds with carbon numbers of 8 to 16. Very small amounts of isoparaffins are obtained during the hydrogenation at 400 °C. Overall, ≈21 C % are transformed to gas phase products, much less than for the 350 °C second-zone experiment. The selectivity toward gas-phase products at 55 h decreased in the order propane (≈22 %)>methane (≈20 %)>pentane (≈19 %)≈ethane (≈16 %)>hexane (13 %)≈butane (≈9 %) (Table 13). After 55 h TOS, the reactor was plugged.
| TOS [h] | System pressure | Carbon content [wt %] | ||||
|---|---|---|---|---|---|---|
| [psi] | organic phase | aqueous phase | gas phase | total liquid+gas | unidentified | |
|
||||||
| 24 | 2030 | 0.0 (0.0) | 0.7 (29.0) | 12.0 (6.1) | 12.7 (35.2) | 87.3 (64.8) |
| 55 | 2130 | 57.1 (20.8) | 1.8 (64.9) | 28.3 (14.4) | 87.3 (100.1) | 12.7 (0.0) |
| overall | – | 32.2 (11.7) | 1.3 (49.3) | 21.2 (10.8) | 54.7 (71.8) | 45.3 (28.2) |
| Component | Number of | Selectivity [%] | |
|---|---|---|---|
| C atoms | 24 h | 55 h | |
|
|||
| 2-pentanol | 5 | 0.81 | – |
| ethylene glycol | 2 | 24.29 | – |
| propylene glycol | 3 | 3.47 | – |
| 1-hydroxy-2-butanone | 4 | 11.02 | – |
| 1,2-butanediol | 4 | 8.21 | – |
| 1,4-pentanediol | 5 | 2.28 | – |
| 1,4-butanediol | 4 | 20.17 | – |
| 1,2-cyclohexanediol | 6 | 24.29 | – |
| cyclohexane | 6 | – | – |
| methyl cyclohexane | 7 | – | – |
| trans-dimethyl cyclohexane | 8 | – | 1.79 |
| ethyl cyclohexane | 8 | – | 47.10 |
| propyl cyclohexane | 9 | – | 14.37 |
| trans-ethyl-methyl cyclohexane | 9 | – | 20.20 |
| C % identified | 94.54 | 83.46 | |
| C % unidentified | 5.46 | 16.54 | |
Carbon distribution for organic products obtained from 130–400°C two-stage HDO of bio-oil after 55 h TOS.
| Component | Selectivity [%] | |
|---|---|---|
| 24 h | 55 h | |
|
||
| CH4 | 37.26 | 20.12 |
| C2H6 | 16.29 | 16.29 |
| C3H8 | 8.82 | 22.43 |
| C4H10 | 5.20 | 8.50 |
| C5H12 | 16.17 | 19.17 |
| C6H14 | 16.26 | 13.26 |
Table 14 summarizes the NOISE analysis of the organic fraction from HDO as a function of temperature of the second-reactor zone. The analysis shows that the organic layer contains linear and branched paraffins, aromatics, naphthenes, and naphthalenes. The isoparaffin selectivity increases with temperature with a maximum of ≈9 % at 400 °C. Naphthenes are the major products present in the organic phase product with 90 % selectivity at 350 and 400 °C. With an increase in temperature, the amount of monoaromatics and naphthalenes decreases to <4 %. The average MW of the product also decreases with temperature from 215 Da at 350 °C to 155 Da at 400 °C, suggesting the cracking of larger molecules into smaller ones.
| Product | Amount [wt %] | ||
|---|---|---|---|
| 300 °C | 350 °C | 400 °C | |
| isoparaffins | 3.11 | 2.58 | 8.94 |
| n-paraffins | 1.49 | 0.74 | 1.57 |
| naphthenes (methyl, ethyl, propyl cyclohexanes) | 50.67 | 93.18 | 89.29 |
| monocyclic aromatics | 30.30 | 3.49 | 0.19 |
| naphthalenes | 12.33 | 0.00 | 0.00 |
| average molecular weight [g mol−1] | 214.40 | 214.7 | 155.1 |
Reaction pathways and coke formation
Low-temperature hydrotreatment over Ru/C was expected to result in the preferential hydrogenation of reactive carbonyl and furanic groups. Ru has been shown to be the most active monometallic metal for hydrogenation of aqueous-phase acetic acid to ethanol.[32] It was demonstrated in this work that some of the cellulosic and hemicellulosic fractions of bio-oil were successfully hydrogenated by Ru at low temperature to yield value-added chemicals whereas the remainder underwent polymerization reactions to form coke. It has been proposed that aldol condensation is one of the major polymerization pathways at low temperatures.[24] One reason for this is the relative stability of keto versus enol tautomers (as demonstrated for phenol hydrogenation products) due to stabilization of the former in the aqueous phase.[24] This is likely the result of straight-chain condensation of carbonyl-containing compounds such as acetic acid, hydroxyacetaldehyde, and furfural. Furthermore, the relatively high selectivities toward C5 sugars, sorbitol, and levoglucosan throughout the reaction suggest that these multifunctional compounds did not play a role in coke formation. The pyrolytic lignin (PL) fraction of the bio-oil, which has been shown to form coke as low as 25 °C over Ru catalysts, [20] likely polymerized in the low-temperature bed. In contrast, hydrogenation of the aqueous-phase products formed through biomass hydrolysis over a Ru/C at 130 °C (with no PL present) resulted in a stable catalyst for over 200 h TOS.[29] Clearly, more research is needed to focus on the low-temperature stabilization of the pyrolysis oils.
The addition of a second stage led to the production of naphthenes and alcohols from the HDO of bio-oil. The active Pt metal was responsible for hydrogenation and dehydrogenation reactions whereas the ZrP activated the C−O bond cleavage through hydrogenolysis reactions. The organic phase consisted of a distribution of hydrocarbons that were primarily cyclic alkanes ranging from C7 to C24. The organic-phase carbon yield decreased with increasing reaction temperature in the second zone. The PL formed methyl-, ethyl-, and propyl-cycloalkane products. Small amounts of isoparaffins, n-paraffins, and aromatics were also obtained. The aqueous phase mostly consisted of hydrogenation and hydrogenolysis products of the cellulose and hemicellulose fractions of bio-oil. Ethylene glycol, propylene glycol, and methyl acetate were hydrogenated to form methanol and ethanol whereas the sorbitol and glucose formed propanediols.
The two-stage deoxygenation process could be carried out only for 55–72 h because of catalyst deactivation due to coke formation in the reactor. The amount of unidentified carbon increased with an increase in the second-stage reaction temperature. The oligomeric species present in pyrolysis oil are reactive at temperatures higher than 300 °C and are partly responsible for the coking of the catalyst. The presence of aromatic rings and naphthalenes at lower temperatures (300 °C) and naphthenes at higher temperatures (350 and 400 °C) is consistent with literature on coke formation during pyrolysis-oil HDO.[33-35] According to Zhang et al., coke formed during bio-oil HDO can be broadly characterized into “soft coke” and “hard coke”.[34] Soft coke is formed at lower temperatures and is highly oxygenated and active to further polymerization reactions. Soft coke can be solubilized by dichloromethane and its formation is reversible. Conversely, hard coke is very hard to solubilize and its formation is considered irreversible.[34] Polyaromatics derived from the PL portion of the bio-oil are likely responsible for hard coke formation inside the catalyst pores.[35] Evidence of hard-coke formation is provided by the high yields of aromatics and naphthalenes at 300 °C. At temperatures 350 °C or higher, the naphthalenes are likely further hydrogenated to a more graphite-like form of hard coke, as suggested by the high yields of naphthenes. The PL likely responsible for high temperature coke formation is a complicated mixture of polymeric phenolic compounds.[28] Chen et al. have shown that PL does not contain the β-O-4, phenylcoumaran, and resinol interunit linkages prominent in lignin.[20] Furthermore, it was shown that hydrogenation over Ru/TiO2 saturates the carbonyl and aromatic carbon atoms to C−C and C−O aliphatic carbon atoms.[20] This converts the aromatic structure of the phenolic units into naphthenic structures. The lack of oxygenated aromatics in our product streams suggests that these compounds have likely undergone HDO to monoaromatics or dehydrogenation/hydrogen transfer to polyaromatics.
Extensive coke formation and reactor plugging are common problems for hydroprocessing of bio-oil.[36] More detailed studies are needed to more clearly define the species that form coke and the coking mechanism. The coke can form from both homogeneous and heterogeneous reactions. A wide variety of species in the bio-oil form coke in the aqueous phase. Carbohydrates will decompose to coke (or humins) when heated in water at temperature above 120 °C.[37] In addition, there are many phases possible in the reactor including a gas phase, an aqueous phase, and potentially multiple organic phases. It should be noted that the pyrolysis oil itself is an emulsion that can contain up to three different phases. During the HDO process, the pyrolysis oil undergoes a phase change from being mostly hydrophilic to becoming hydrophobic.
Catalyst stability
To successfully advance pyrolysis-oil HDO technology, the problem of coking needs to be solved. One potential solution to coking lies in the upstream production of more stable pyrolysis oils. This is based on the concept of removing destabilizing carboxylic acids and other carbonyls and reducing overall oxygen content during catalytic fast pyrolysis (CFP). Pyrolizing biomass over basic catalysts has been claimed to drastically reduce the carboxylic acid and oxygen content. Basic catalysts such as MgO, CaO, CsX, or hydrotalcite have been employed in the pyrolysis step.[38] A bio-oil feed with the enhanced stability implied by these properties would greatly improve the catalyst lifetime of bio-oil HDO processes in addition to the low-temperature hydrogenation step proposed in this work.
Another potential source of catalyst deactivation are minerals originating from the biomass feedstock ending up in the pyrolysis oil. Some of the most common minerals making up bio-oil ash include potassium, calcium, magnesium, sodium, and phosphorous.[39] Other impurities in the bio-oil such as chlorine and sulfur can also lead to deactivation. Mortensen et al. demonstrated reversible deactivation of Ni-based HDO catalysts caused by K and Cl but irreversible deactivation through only 0.05 wt % S.[40] Despite improved post-pyrolysis filtration to remove ash particles, the possibility of the negative effects of ash on catalyst activity should be accounted for when performing model compound studies.
Fuel product yields from bio-oil HDO
Figure 7 shows the overall and instantaneous (55 h TOS) carbon yields for HDO of the bio-oil for the 130–400 °C duel-bed reaction and compares it to the theoretical HDO yields possible (assuming no coke is formed). Boiling point cuts for all analyses are as follows: light gases/liquefied petroleum gas (LPG) <30 °C; gasoline 30–160 °C; jet fuel/diesel 160–340 °C; residual oil >340 °C. The theoretical yields were calculated from the bio-oil feedstock in Table 1 assuming complete HDO and no molecular cracking or C−C bond formation. It was also assumed that the pyrolytic lignin fraction could be hydrodeoxygenated into a 50 % jet fuel and 50 % diesel fuel mixture, a valid assumption based on the boiling points of the resulting C6–C24 hydrocarbons. This analysis indicates that the theoretical yields for HDO of the bio-oil are 50 C % yield to distillate-range fuels, ≈30 C % yield to gasoline, and ≈20 C % yield to light gases/LPG. The actual instantaneous yields of distillate-range products for the 400 °C reaction at 55 h TOS are only 6 C % lower than the theoretical value. However, when taking into account the overall yields over the course of the whole reaction, the distillate range yields are reduced, that is, 25 C % lower than the theoretical value. This speaks for the importance to look at overall yields in addition to instantaneous yields for systems such as these where catalyst deactivation is a problem. These results demonstrate that only reporting these instantaneous yields could lead to a drastic underestimation of the production costs for HDO of pyrolysis oils.
Comparison of theoretical carbon yields (light grey), observed carbon yields from 400 °C dual bed reaction at 55 h TOS (dark grey), and observed overall carbon yields from 400 °C dual bed reaction (black).
Sankey diagrams are used to compare carbon balances of our process with recent work by Elliott et al. (Figure 8),[14] who employed a dual-stage, single-reactor system for a low-temperature hydrotreatment over sulfided Ru/C at 170 °C and a high-temperature hydrotreatment over sulfided CoMo or NiMo at 400 °C. Between 94 and 99 % oxygen was removed before reactions were terminated between 90–99 h TOS due to reactor plugging. Liquid product was accumulated for roughly 20 h TOS between sample collections. The first sample of each run was collected at ≈23 h TOS and the last sample collected at >80 h TOS and were labelled accordingly in the analysis. Comparable amounts of product and coke formation were observed in both studies. In general, our work demonstrated slightly higher gasoline and distillate fuel yields despite larger coke formation at early reaction times and quicker plugging of the reactor. Our lower residual oil yields are likely due to the higher cracking activity of the Pt/ZrP catalyst compared to the CoMo and NiMo used by Elliott et al., as both HDO steps were performed at 400 °C. We observed components with boiling points in the residual oil range for the 350 °C HDO, but negligible amounts for the 400 °C HDO. In both studies, higher coke formation was detected at early reaction times. Elliott et al. concluded that a decrease in catalyst activity over time resulted in the shift to higher-boiling residual oil products and that coke buildup at the low temperature–high temperature bed interface was the primary cause of deactivation.
Sankey diagrams for a) 130–400 °C HDO experiment from this paper at 55 h TOS, b) HDO of bio-oil from Elliott et al. at ≈23 h TOS,[14] and c) HDO of bio-oil from Elliott et al at >80 h TOS)[14].
A recent collaborative effort between national laboratories focused on the field-to-fuels pyrolysis and HDO of several biomass feedstocks.[15] The resulting pyrolysis oils underwent HDO in a dual-zone reactor over Ru/C at 220 °C in the first zone and CoMo/Al2O3 at 400 °C in the second zone. In addition, there was a Ru/C catalyst bed (termed a “guard bed”) added at the reactor inlet to prevent polymerization in the non-isothermal zone. A simulated distillation curve of the oil product resulted in average gasoline, diesel fuel, jet fuel, and residual oil mass yields of 44 %, 32 %, 11 %, and 13 %, respectively, at a supposed “steady-state”.[15] This corresponds to carbon yields of 32, 23, 8, and 10 C %, with an average light-gas yield of 25 C %. In the study by Howe et al.,[15] the light-gas (25–19 C %) and gasoline (32–24 C %) yields were higher and the distillate range fuels yields were lower (31–44 C %) than in our system at 55 h TOS (400 °C second stage). The higher-distillate-range fuels obtained in our work are likely due to increased cracking of residual oil components as compared to Howe et al. (who used a similar CoMo catalyst as Elliott et al).[14, 15] Importantly, Howe et al. were able to close the carbon balance for most of the bio-oil feedstocks tested, with only the poplar- and switchgrass-derived bio-oils giving >5 C % missing carbon.[15] This speaks for the viability of mild hydrogenation treatment zones similar to that presented in this work in limiting coke formation in bio-oil HDO processes.
Unfortunately, there have been setbacks in the implementation of bio-oil HDO technologies on an industrial scale. One company built a demonstration facility that was designed to produce 13 million gallons of liquid (gasoline, diesel, and jet) fuels per year by pyrolysis followed by HDO of the resulting bio-oil. Unfortunately, this company was only able to operate at 25 % of the designed capacity. It was reported that this facility produced yields from ≈20–35 gallons of fuel per ton of dry biomass (GTB).[41] To our knowledge, the highest yields obtained in the literature is 82 GTB obtained by Howe et al. (clean pine). This corresponds to a 100 % theoretical HDO yield from our bio-oil if we assume the same pyrolysis step yield as Howe et al. (see Figure 7).[15] The yields reported in industry are lower than that of our overall yield (41 GTB) and of our 55 h TOS/400 °C second bed reaction yield (70 GTB) assuming the same bio-oil yield from pyrolysis.[41] Challenges often occur when processes are scaled up to industrial scale.[42] Academic researchers should pay special attention to try and identify the fundamental chemistry that causes coke formation and develop approaches to mitigate this issue.
Although progress has been made in increasing catalyst lifetime and the degree of deoxygenation in hydrotreating processes, there is a clear need to understand the fundamental catalytic and homogeneous chemistry involved in the HDO of bio-oil for its potential as a feedstock in producing renewable fuels and chemicals to be fully realized. Some proposed methods to achieve a more efficiently bio-oil HDO include: 1) improved pre-upgrading filtration or chemical modifications to the bio-oil,[43] 2) separation of the bio-oil into water-soluble and organic fractions and upgrading each portion separately,[18, 30] 3) performing HDO in stabilizing alcohol solvents,[44] and 4) improved bio-oil fractionation and subsequent upgrading of individual components.[8] Another procedure that has been claimed to drastically reduce coke formation is the use of a hydrogenation catalyst guard bed at the isothermal zone of the reactor inlet.[15]
Conclusions
We studied the hydrodeoxygenation (HDO) of a bio-oil into liquid hydrocarbon products. The bio-oil contained 49.4 % oxygen, 43.5 % carbon, and 7.1 % hydrogen. About 30 % of the carbon present in the bio-oil was in the form of acids, aldehydes, ketones, and furans. Roughly 20 % of the total carbon content of the bio-oil was C5 sugars, levoglucosan, sorbitol, and xylose. The remaining ≈50 % of the carbon in the bio-oil was present as pyrolytic lignin. The catalytic HDO of the bio-oil was carried out in a continuous-flow two-stage catalytic reactor system that contained a mild hydrogenation zone at 130 °C over Ru/C as catalyst followed by a more severe HDO zone between 300–400 °C over a Pt/ZrP catalyst. The first stage Ru/C catalyst converted the aqueous phase of bio-oil into functionalized alcohols, such as propylene glycol, ethylene glycol, and sorbitol. Polymerization reactions occurring in this first stage resulted in the plugging of the reactor at less than 70 h time on stream (TOS).
The addition of the high-temperature second zone with a Pt/ZrP catalyst resulted in the production of a hydrocarbon oil similar to petroleum-based fuels. The Pt/ZrP catalyst was successful in the HDO of the oxygenated feed, with monoaromatics, naphthalenes, and naphthenes being produced at 300 °C and primarily naphthenes at 350 and 400 °C. High molecular-weight polyaromatics were likely responsible for coke formation at high reaction temperatures.
Coke formation led to a decrease in overall oil yields, with coke yields of 45 C % for the highest temperature (400 °C) dual-bed reaction. This is in contrast to coke yields of just 13 C % at 55 h TOS, which points to the importance of reporting overall yields versus instantaneous yields. Hydrotreatment yielded a combined gasoline- and distillate-range fuel overall yield of 30 C % for the same reaction, compared to the 79.5 C % theoretical yield calculated for the HDO of the bio-oil feedstock. This represents a large deviation from ideal carbon yields and points to serious coking problems associated with the hydrotreating of bio-oil and the need for improved bio-oil HDO processes or the integration of other methods such as bio-oil fractionation and selective upgrading.
Experimental Section
Bio-oil feed
The bio-oil was synthesized by National Renewable Energy Laboratory (NREL), Golden, Colorado under typical pyrolysis conditions from white oak pellets using thermochemical process development unit (TCPDU). A description of the pyrolysis unit can be found elsewhere.[45] Membralox® TI-70 microfiltration membranes with nominal pore sizes of 0.8 μm were employed to remove char particles prior to HDO of bio-oil as described elsewhere.[26] The oil before and after filtration was stored at ≈5 °C to avoid aging and prevent polymerization.
Catalyst synthesis
Zirconium phosphate: The zirconium phosphate support was synthesized using a co-precipitation method from precursor salts of zirconium and phosphorus metals. Zirconium oxychloride (ZrOCl2⋅8 H2O, Sigma–Aldrich) and ammonium phosphate monobasic (98.5 %, NH4H2PO4, Sigma–Aldrich) were used as precursor salts for zirconia and phosphorous, respectively. A solution of zirconium oxychloride (22.5 g) in water (70 mL) was prepared. The P/Zr molar ratio of 2.0 was maintained by mixing ammonium phosphate (16.1 g) with water (140 mL). The zirconium oxychloride solution was then added dropwise to the phosphate solution to obtain a precipitate. The precipitate was subsequently filtered, washed with water, and left at room temperature overnight. The paste was then dried at 100 °C for 6 h before calcination in air at 450 °C for 4 h.[46] The obtained zirconium phosphate was used as support for the synthesis of 4 % Pt/ZrP catalyst.
Pt/zirconium phosphate: The 4 % Pt/ZrP catalyst was prepared through the incipient wetness impregnation method. A solution of tetraammineplatinum(II) nitrate (99.995 % metals basis, H12N6O6Pt, ACROS Organics) was prepared by using a volume of water equivalent to the incipient volume of zirconium phosphate support. The desired impregnation was performed in two stages to achieve better dispersion over the zirconium phosphate support. The catalyst was then dried at room temperature overnight and at 100 °C for 6 h followed by calcination at 400 °C for 2 h.
Ru/C: The Ru/C [Restek, Catlog # 44-4050] catalyst (surface area of 635 m2 g−1) was obtained as a dry-powder. H2 chemisorption yielded a dispersion of 6.7 % of metal over the catalyst. A detailed procedure is provided elsewhere.[29]
Pyrolysis oil HDO
Block diagram of HDO of pyrolysis oil: A block-flow diagram for pyrolysis of biomass followed by HDO is presented in Scheme 1. The solid biomass was dried and ground prior to the pyrolysis process. The organic vapor generated from pyrolysis was rapidly quenched to yield the liquid bio-oil. CO and CO2 gases along with char particles were byproducts of the pyrolysis process. Two trickle-bed reactors were used in series for HDO. Hydrogen was obtained from high-pressure hydrogen cylinders (UHP 99.999 %) although hydrogen for HDO should be supplied through methane steam reforming in an industrial process, as it is currently the cheapest technology available.[47] The hydrogen was then pressurized to reaction pressure and fed over the fixed bed catalyst along with liquid bio-oil in a three-phase system.
Block diagram for conversion of biomass to liquid transportation fuels by pyrolysis and HDO.
After hydrotreatment, the products could be separated through boiling points into light gases/LPG, gasoline, and distillate-range fuels. The hydrogen could be recycled to the HDO reactors. The non-condensable light gases produced in this process had to be separated or purged along with the H2 gas recycle stream. The present work solely focused on the low- and high-temperature hydrotreatment reactor units in this process.
Hydrogenation reactors: The hydrogenation of bio-oil was carried out in continuous flow-trickle bed reactors at high pressure. For single-stage hydrogenation, one reactor with Ru/C catalyst (2.2 g) was used. For the two-stage hydrogenation process, the first reactor with Ru/C (2.2 g) and the second reactor with Pt/ZrP (5.8 g) were connected in series to carry the hydrogenation products from the first to the second reactor. An Inconel Alloy 600 tube capable of resisting corrosive acid and withstanding high temperature and high pressure was selected as the material for the reactor tubing. The typical size of the reactor tube was 4.57 mm ID, 6.35 mm OD, and 12 cm in length. The powder catalyst was supported by quartz wool on both ends. Typically, a catalyst bed height of 10 cm was maintained using 1 cm thick quartz wool on both ends to seal the reactor. The temperature in the first reactor was reached using a heating tape with reactor vertically mounted on a split-open furnace equipped with a temperature controller.
For each run, both reactors were loaded with catalysts and treated with hydrogen to remove water and other species adsorbed on catalyst surface. The Ru/C and Pt/ZrP reactors were maintained at 260 and 450 °C, respectively, for 8 h at atmospheric pressure. Both reactors were then cooled down at a rate of 1 °C min−1 to the desired reaction temperatures to avoid sintering of catalysts.
The system was then slowly pressurized to 2000 psi by introducing hydrogen through a mass-flow controller. The system was allowed 1 h to reach a steady hydrogen flowrate of 200 mL min−1 and reaction temperature and pressure. An ISCO syringe pump (Teledyne Inc, Model: 500 HP) was used to pump bio-oil at a flowrate of 5–8 µL min−1 to the pressurized reactor. These flowrates correspond to a 185–300 g H2/100 g bio-oil feed ratio. Although our H2/bio-oil ratio was much higher than those that would be economically employed in industry (≈8 g H2/100 g bio-oil in a recent bio-oil HDO design case carried out by the Department of Energy), this should not significantly affect the HDO chemistry observed in this work.[48]
The liquid product formed during the reaction was collected in a cylinder. The liquid product contained two phases (a water-soluble phase and an organic phase) which were analyzed separately. Prior to collecting the liquid products formed during the reaction, the gas products were analyzed in an on-line GC. The pump was subsequently stopped and the liquid products were drained. This resulted in a decrease in the system pressure, which was then allowed 30 min to reach the steady state before again feeding bio-oil into the system. Each run was terminated when the system experienced a sudden increase in pressure.
Product analysis
The water content of each sample was determined using a Karl–Fisher titration (Mettler Toledo, V20 Volumetric Titrator) instrument following the standard ASTM E203-96 method. The TOC content for solid and liquid samples were carried out in SSM-5000A and TOC-V CPH Shimadzu TOC analyzers, respectively.
Water was added to the bio-oil in a 4:1 mass ratio for complete dissolution of the water-soluble fraction of bio-oil. The portion of bio-oil insoluble in water was separated and designated pyrolytic lignin. The sugars, sugar alcohols, and anhydrosugars were analyzed using a Shimadzu High Performance Liquid Chromatography (HPLC) (model #: LC 20AT) equipped with a refractive index (RI) detector (RID-10A). A 0.005 m H2SO4 mobile phase flowrate of 1 mL min−1 was maintained through the Aminex column. A Varian Mesopore column was u for gel-permeated column (GPC) characterization using a HPLC instrument for size analysis of liquid products. Stabilized tetrahydrofuran (THF) at a flowrate of 0.5 cm3 min−1 was used as the mobile phase. The liquid samples were diluted with THF prior to GPC analysis in the instrument.
The feed composition and liquid (both water-soluble and organic) products were analyzed in three GCs. The lighter hydrocarbons present in liquid products were analyzed in an Agilent Technologies GC (model# 7890A) equipped with a flame ionization detector (FID). All components except the sugars, sugar alcohols, and levoglucosan were identified and quantified in this GC equipped with a Restek Rtx-VMS column. Heavy hydrocarbons were identified and quantified in a Shimadzu GC (model # GC-2010) through an FID detector. Response factors (mol C per L-peak area) were determined by analyzing a known concentration of a standard solution for each compound. The gas products obtained from this setup were analyzed in a Shimadzu online GC (model # GC-2014) equipped with an FID and thermal conductivity detector (TCD) for analysis of hydrocarbons and hydrogen, respectively.
The simulated distillation analysis was carried out using a Varian Star 3400 GC equipped with a Supelco column. A Finnigan 4500 GC–MS using a CTC A200S automatic sample injection system was used for simulated distillation analysis. The simulated distillation and nitric oxide ionization spectrometry evaluations (NOISE) analysis were performed by Triton Analytics Corp., Houston, Texas. In NOISE analysis, NO+ ions formed by Townsend discharge of nitric oxide (NO) ionize different hydrocarbon functionalities (naphthenes, naphthalenes, aromatics, isoparaffins, and n-paraffins) uniquely based on their bond structure. Combined with GC–MS, the functionality and carbon number distribution of the complete organic sample could be quantified.
Acknowledgements
This work was supported by the Defense Advanced Research Project Agency through the Strategic Technology Office (BAA 08-07) (Approved for Public Release, Distribution Unlimited). The views, opinions, and/or findings contained in this article/presentation are those of the author/presenter and should not be interpreted as representing the official views or policies, either expressed or implied, of the Defense Advanced Research Projects Agency or the Department of Defense. K.R. would like to thank Drs. Ning Li, Tushar P. Vispute, Hakan O. Olcay, and Aniruddha Upadhye of Chemical Engineering for their assistance in several characterization experiments. K.R. would also like to acknowledge the help extended by Triton Analytics Corporation, Houston, Texas in conducting the Simulated Distillation and NOISE analysis of bio-oil and its hydrogenated products. The authors wish to express their gratitude to Professors Geoffrey A. Tompsett and Michael F. Malone of University of Massachusetts and Dr. Max Cai of Logos Technologies for helpful discussions.
Fuente: http://onlinelibrary.wiley.com
