Article pubs.acs.org/IECR
Modeling the Lignin Degradation Kinetics in a Ethanol/Formic Acid Solvolysis Approach. Part 2. Validation and Transfer to Variable Conditions Daniel Forchheim,†,¶ James R. Gasson,‡,¶ Ursel Hornung,*,† Andrea Kruse,†,§ and Tanja Barth‡ †
Institute of Catalysis Research and Technology, Karlsruhe Institute of Technology, Hermann-von-Helmholtz-Platz 1, DE-76344 Eggenstein-Leopoldshafen, Germany ‡ Department of Chemistry, University of Bergen, Allégaten 41, NO-5007 Bergen, Norway § Conversion Technology and Life Cycle Assessment of Renewable Resources, University of Hohenheim, Grabenstr. 9, DE-70599 Stuttgart, Germany ABSTRACT: A formal kinetic model treating the depolymerization and hydrodeoxygenation of wheat straw lignin in ethanol with formic acid as hydrogen source developed by Gasson et al. [Gasson et al. Ind. Eng. Chem. Res. 2012, 51 (32), 10595−10606] is validated and its applicability in a continuous stirred tank reactor (CSTR) at varying temperatures between 633 K and 673 K is tested. The fitted formal kinetic rate coefficients are compared and sensitivity and flux analyses are performed. Activation energies are estimated for the lumped reactions. The depolymerization to primary products is considerably accelerated when transferring the reaction from batch to a continuous operation. Higher heating rates and continuous feeding of the hydrogen donor formic acid aid in suppressing both gas and char production. Repolymerization of intermediate phenolic compounds is suppressed, which is suggested to be due to the interaction of intermediate and final products. The evaluation shows that a continuous system can aid in avoiding further gasification and charring of the intermediate products in this one-step lignin depolymerization deoxygenation reaction for the production of a phenol-rich bio-oil.
■
and give higher yields of monophenolics.16 Further positive influence can be gained by the use of polar cosolvents17 (e.g. ethanol and iso-propanol).3,18 However, the degradation of the latter to acetone as well as other reactions to form a multitude of ketones and esters at high temperature conditions,19 make it a nonrecyclable solvent with a cost factor that is too high for consideration for this type of reaction process. Lignin-to-Liquid Approach. The one-step system investigated in this paper uses ethanol as a dispersant and solvent medium. Formic acid is used as an in situ hydrogen donor, by thermally decomposing to a mixture of H2 and CO2 or H2O and CO, without necessitating the presence of a catalyst. Reaction temperatures and durations first reported were between 623 K and 673 K and 4−16 h, respectively. Under these conditions, a high yield of alkylated monomeric phenolics with a very low level of oxygenation can be achieved.7 Newer works have concentrated on trying to reduce experimental durations and temperatures, while retaining the originally found product spectrum.19 Both blending in conventional fuels as well as extraction and separation of phenolic compounds from the recovered hydrophobic liquid fraction for the production of platform chemicals are considered viable options to create a value-added product from lignin-rich residual material from lignocellulosic ethanol production in a biorefinery.3
INTRODUCTION Phenols from Lignin. From a chemical perspective, residual lignin from lignocellulosic biomass is most likely the major renewable source for aromatics and, indeed, offers the only potential direct production pathway for phenolics.2 This can be exploited both for the production of aromatic-rich biofuel blends,3,4 as well as the select production of lignin-based platform chemicals.5−7 Both these utilization pathways for lignins are considered within various biorefinery approaches.8 Thermochemical methods have widely been investigated to produce monomeric units from this phenolic-rich polymer and have been well-documented in reviews by Mohan,9 and by Brebu and Vasile.10 The temperature conditions necessary to cleave the largely dominant ether and further existent carbon bonds in lignin unfortunately also promote typical charring and gasification processes. Hydrogenolysis in a supercritical solvent environment is largely able to suppress these two side reactions. A comparison of different approaches, already documented by Meier and Schweers in the beginning of the 1980s, shows the already then present interest to utilize lignin for the production of monomeric phenolics.11 In a more-recent review, Pandey and Kim again found hydrogenolysis to be the most promising approach for the economic production of monomeric phenols from lignin, because the method enables a high conversion rate and a subsequent high yield of monophenolics.12 The hydrogen-rich environment has been found to aid both in depolymerization of the lignin and hydrodeoxygenation of its degradation products.3,13,14 Next to molecular hydrogen in various combinations with catalysts,15 hydrogen donor solvents have also been seen to accelerate the depolymerization process © 2012 American Chemical Society
Received: September 27, 2012 Accepted: October 25, 2012 Published: October 25, 2012 15053
dx.doi.org/10.1021/ie3026407 | Ind. Eng. Chem. Res. 2012, 51, 15053−15063
Industrial & Engineering Chemistry Research
Article
Table 1. Experimental Loadings for Batch and Continuous Reactors Loading (g or g/h) EXP
reactor
formic acida
ethanola
lignina
temperature (K)
(mean) residence time (min)
MA1 MA2 CS1 CS2 CS3 CS4 CS5
batch batch CSTR CSTR CSTR CSTR CSTR
0.390 0.390 18.7 7.72 7.78 7.89 7.96
3.16 3.16 146.3 60.0 60.4 61.3 61.8
0.351 0.351 15.4 6.33 6.37 6.46 6.52
653 653 653 653 633 653 673
60 140 60 136 138 141 137
pressure (MPa) b b
25 30 25 20 25
a
Batch reactors: g, CSTR: g/h. bThe design of the batch reactors used did not enable measurement of resulting reaction pressures. Comparable experiments performed in other batch reactor systems have shown reaction pressures between 29 and 33 MPa.4 The pressure in presented experiments is expected to be reduced by 3−6 MPa, because of a decreased loading ratio of formic acid to ethanol.
A Reaction Pathway Lump Model. As numerously mentioned in the reviews referenced in the section above, the complexity of lignin and its degradation pathways still challenge the establishment of an economic large-scale production of monomeric phenols in high yield. Although a multitude of reaction parameters have been analyzed in respect to their effect on global bulks, Amen-Chen et al. point out in their review on thermochemical conversion on this matter that only few studies have been performed relating parameters such as temperature, heating rate, biomass origin, and reactor pressure to the produced bio-oil composition.20 However, understanding the chemical reactions in detail and therefore being able to relate the chemical composition to process parameters can largely benefit production optimization and be a crucial step toward developing an economic process. While investigating the properties of hydrogen-enriched lignin solvolysis in ethanol, we identified the main reaction pathways and their importance in the production of deoxygenated phenolics with the help of a formal kinetic model in Part 1 of this work.1 Our analysis shows that the lignin depolymerizes quickly in ethanol to form primary phenolic products, which, in due course, undergo further deoxygenation reactions. The influence of ethanol is largely seen in an improved solubility of both the polymer and its reaction products. In addition, ethanol is seen to act as a mild alkylation reactant, which could be one possible way of producing hindered phenols for fuel applications. Although operating under nonoptimized conditions, formal kinetic modeling results showed that hydro-deoxygenation reactions are slow in comparison to competing gasification and repolymerization processes. This shows the necessity to increase the speed of the consecutive demethoxylation and dehydroxylation reactions, also to counteract repolymerization reactions which were shown to be mainly occurring from oxygenated species, such as catechols. The hydrogen produced by the thermal degradation of formic acid, although rapidly consumed in batch operation, is believed to largely counteract these undesired reactions. Further alternate pathways of significance may be required to explain the experimental observations, depending on the biomass used. Our experiments with wheat straw lignin showed that 4-ethylphenol is produced as a primary product from depolymerization rather than by a degradation pathway including the consecutive reactions from syringol to guaiacol and catechol in this hydrogen-enriched ethanol solvolysis approach. Work Approach. Based on a selected number of experiments in a batch and a continuous stirred tank reactor (CSTR), using the developed formal kinetic model, we aim to explore
and validate our suggested reaction pathway lump model applicability to different reactor systems and reaction temperatures. The applicability of the model to these alternate reaction conditions will be tested and the fit quality will be evaluated. Using sensitivity and flux analysis, we aim to further monitor the influence of temperature and the characteristics of a continuous reactor system on both a molecular level as well as a global bulk level. This will aid in the understanding of the chemical reaction mechanisms and set a basis for future work to create a sustainable production platform for phenolics from lignin.
■
MATERIALS AND METHODS Materials and Reactors. Materials. Formic acid, absolute ethanol, and other consumables were purchased from Sigma− Aldrich, Germany, in analytical quality and were used without further purification. Protobind 1000, a lignin-rich (>90 wt %) material produced from wheat straw, from ALM Indiva Pvt. Ltd., India, was used in all experiments. The lignin was dried for 48 h at 353 K and ground to a maximum diameter of 200 μm. The lignin contains C (63.8 wt %), H (6.0 wt %), and O (27.0 wt %). This results in a H/C atomic ratio of 1.1 and a O/C atomic ratio of 0.3. The residual moisture of the lignin after drying and storage was 4.5 wt %. Batch Reactors. For batch reactor experiments, custom-built 5-mL batch reactors made of 1.4571 stainless steel were used. These have an inner cylindrical shape with a diameter of 11.5 mm and height of 48.1 mm. The reactors have a metal-on-metal seal and can withstand pressures of up to 40 MPa and a maximum temperature of 673 K. The reactors were heated in a HP-5890 GC oven, which can be heated and cooled using a temperature program. The reactors were filled with a solution of ethanol, formic acid, and biomass, according to the experimental loadings given in Table 1. They were then purged with nitrogen and sealed. The reactors were placed in an oven and heated at a rate of 40 K/min until the reaction temperature of 653 K was reached. After completion of the chosen residence time at this temperature, the reactors were removed from the oven and rapidly quenched in icy water. These were thereafter ventilated, and the gas was quantified volumetrically and sampled on GC. The liquids and solids were recovered, the latter being weighed, dried for 16 h at 378 K, and reweighed to account for liquid loss. The filtered and quantified liquids were sampled using gas chromatography−flame ionization detection (GC-FID). Continuous Stirred Tank Reactor. The custom in-house built CSTR has an inner volume of ∼190 mL, is made of the nickel-based alloy inconel 625, and is heated by incorporated 15054
dx.doi.org/10.1021/ie3026407 | Ind. Eng. Chem. Res. 2012, 51, 15053−15063
Industrial & Engineering Chemistry Research
Article
claimed to be reached. Several liquid samples (including solid particles) were taken throughout the entire experiment by emptying the flash separator via a valve at the bottom of the container. Further workup procedures were conducted analogous to the workup described for samples from batch reactor experiments. Experimental Section. Experiments using comparable loading ratios and identical reaction temperature were performed in both batch reactors and the CSTR at varying (mean) residence times. Further experiments were conducted in the CSTR varying the temperature in the range between 633 K and 673 K. The experimental conditions for both sets of experiments are given in Table 1. The ratio of formic acid to lignin was 1.1:1 in the batch reactor and 1.2:1 in the CSTR. The ratio of ethanol to lignin was 9:1 in the batch experiments and 9.5:1 in the experiments conducted in the CSTR. Analytical Methods. GC Analysis of the Gases. Gas-phase GC analysis was performed on a Agilent 7890A with a 2-m Molsieve 5A column in series with a 2-m Porapak Q column equipped with a FID front and TCD back detector. The system was controlled by an Agilent laboratory data system. Injections were carried out by manually injecting 100 μL of the ventilated gas from the reactor. Temperature program: initial temperature 323 K for 22 min, then heating at 20 K/min to 423 K, holding at that temperature for 15 min. Further heating at 50 K/min to 503 K, holding at that temperature for 10 min. The injection port was at a temperature of 523 K, the flame ionization detection (FID) was at a temperature of 503 K, and the pressure was kept constantly at 255 kPa. Components were quantified by varying amounts of standard gas mixtures. GC-FID Analysis of the Liquids. Quantitative GC-FID analysis of liquid products was performed on a HP 5890-II GC system that was equipped with an HP 5890 autosampler, a 30-m Rtx-1MS dimethylpolysiloxan column, and a FID detector. The system was controlled by an HPChem laboratory data system. Temperature program: initial temperature of 313 K for 6 min, then 5 K/min heating to 453 K, then with a rate of 30 K/min continued heating to 533 K, kept for 5 min. Further heating increase at 30 K/min to a final 573 K and kept for 12 min. The injection port had a temperature of 548 K and the FID was at 603 K. The crude liquid samples were diluted in a 1:2 ratio with a prepared standard solution of pentacosane in ethyl acetate (1002 mg/L) as an internal standard and injected by the autosampler system. Monoaromatic components were individually calibrated for quantification by running dilution series with the same internal standard present. Modeling. Base Model. The model was originally developed based on a time-dependent series of lignin solvolysis experiments performed under typical Lignin-to-Liquid (LtL) conditions in batch reactors.1 A schematic of the reaction network and the individual pathways used in the rate coefficient determination is given in Figure 2. The simplified model illustrates the proposed main reaction pathways within the LtL reaction system with a focus on the reactions from lignin to monomeric aromatics and includes additional pathways characteristic to the experimental system, such as the observed alkylation reactions by ethanol and the integration of 4ethylphenol as a primary depolymerization product from wheat straw lignin. Analyzed and quantified key-components were lumped where possible to simplify the model and generalize trends based largely on the structural similarities between the single components.
heating cartridges. It is designed to withstand temperatures up to 923 K and pressures up to 100 MPa. The stirrer is made of 1.4571 stainless steel. The transmission of the torque between the motor and the shaft of the stirrer is realized by means of a magnet clutch. The reactor is heated electrically. The magnet coupling must be maintained below the Curie temperature (1041 K) to guarantee the magnetic properties of the material and the movement of the stirrer throughout the experiment. Feeding of the reaction suspension into the reactor is realized via two screw presses, each with a maximum inner volume of 60 mL. The suspension of ethanol, formic acid, and lignin is sucked into the screw press cylinder through a tube on the front end of the cylinder and pressed out of the screw press cylinder through a tube at the lower part of the cylinder. In order to avoid plugging by means of sedimentation in the feeding tubes, long and curved tubes in the construction were avoided (feeding tubes were
