Printer Friendly

Base-catalyzed depolymerization of lignin: separation of monomers.

INTRODUCTION

Biofuels production from residual lignocellulosic biomass feedstocks has the potential to provide a spectrum of fuels ranging from ethanol to biodiesel. In the pretreatment of biomass to produce fermentable sugars, lignin is a by-product destined to combustion for heat generation. We have been exploring methods to recover the by-product lignin and explore its possible use as raw material for the production of alternate biofuels and green chemicals. The structure of lignin and its oxyaromatic nature supports such an endeavour.

Lignin, a major constitutive macromolecular component of biomass with an energy content higher than that of carbohydrates, is present in softwoods, hardwoods and grasses. In wood, lignin constitutes 15 to 30% of the weight, depending on tree species. This macromolecule is present primarily between the fibre cells and acts, besides other functions, as a cementing material thus imparting structural rigidity to the wood. Lignin is a three-dimensional amorphous polymer composed of cross-linked phenylpropanoid units. In native or unprocessed lignin, generally two thirds or more of the total linkages are ether bonds, while the remaining are carbon-to-carbon bonds (Kadangode, 2000; Miller et al., 2002).

The chemical structure of lignin suggests that it could be a source of renewable high value fuel additives and green chemicals if it could be depolymerized into low molecular weight compounds which retain the oxyaromatic nature (Johnson, 1997; Miller et al., 2002). Shabtai et al. (1999; 2001) have proven the feasibility of such an approach for converting lignin into a high octane reformulated gasoline. Shabtai and coworkers work consisted of a two-stage catalytic process. In the first stage, a lignin-rich feedstock was subjected to a base catalyzed depolymerization reaction (BCD). The second stage involved catalytic hydroprocessing where the oxygen was partially or totally removed. Products consisted of a mixture of phenyl and cycloalkyl methyl ethers, alkylbenzenes, branched paraffins and alkylated cycloalkanes. When oxygen was completely removed by hydroprocessing in the second stage, a product blend of C-7 to C-10 alkylbenzenes was obtained with yields up to 45 wt% of lignin. The blends were essentially sulphur-free and had high Research Octane Numbers (RON), ranging from 120 to 150.

One way to improve the overall BCD approach is to optimize the amount of monomers generated during the BCD reaction and separate them from the lower value oligomeric products. The latter are the intermediates for fuels production whereas the former will be isolated as green chemicals. Monomer optimization on BCD using Steam Exploded Aspen Lignin (SEAL), a prototype lignin easily extracted from the fibrous wet solid resulting from steam treatment of aspen wood chips, was investigated in our previous work (Vigneault et al., 2006). The concentration of lignin dissolved in the aqueous basic (alkaline) solution was found to be a critical factor since monomer yields were found to be inversely proportional to the lignin content of the solution. Optimum monomer yields could reach close to 10 wt% of the lignin. The monomer distribution was a function of reaction severity.

The economic rationale for transforming biomass and lignin into chemicals and fuel additives was outlined by Vigneault et al. (2006). On a 100 kg lignin feed basis, based on 2004 market prices, a yield of 7 wt% of monomers could provide revenues comprised between 14.0 and US$ 21.0 in chemicals, while dimers, trimers and oligomers could be converted to gasoline blending components, at about 45 wt% yield based on lignin, and sold for an additional income of between 13.0 and US$ 21.6. Total income would then range between 27.0 and US$ 42.6, compared to about US$ 10 if the 100 kg lignin were burned to provide energy. However, no strategy for the separation and purification of the monomers products has been proposed so far in the literature, and such is the goal of this paper. The task is complex: in the BCD monomers rich fraction, the number of identified monomers was greater than 12 and were mixed with higher molecular weight compounds, such as dimers and trimers (Vigneault et al., 2006).

The background on phenolic compounds separation processes comes principally from the vanillin production industry. Goheen in Sarkanen and Ludwig (1971) reviewed different methods for the production and isolation of vanillin. Processes using sulphite waste liquors as the feedstock produced lignin via the following steps: lignin extraction by acidification of the liquor, alkaline oxidation of the lignin, acidification converting the vanillin salt to vanillin, liquid-liquid extraction to remove vanillin from the aqueous solution and distillation to separate vanillin from the solvent. Forss et al. (1986) noted that existing approaches require large amounts of acids for neutralization prior to the extraction of vanillin and that the alternative of extracting the sodium salt of vanillin was ineffective. They proposed a cation-exchange resin in sodium form to solve both problems. Vanillin and other oxidation products were separated from lignin in the column. Li and Yin (2001) reported a crystallization technique to purify vanillin with water (up to 99.8%) in an amount 6 times greater than the vanillin content.

Liquid chromatography is often used as an analytical method to separate phenolic compounds. Reverse phase high performance liquid chromatography (RP HPLC) was the analytical method used throughout the work reported here and separated the mixtures of monomers with a good efficiency. Nomura et al. (1992) did the separation of 9 structural isomers of alkyl phenols with reverse phase liquid chromatography. The eluent was a mixture of methanol and water with a small amount of triethylamine. Peng et al. (1998) isolated lignin dimers with a series of Gel Permeation Chromatography (GPC), Reversed Phase Thin Layer Chromatography (RP TLC) and RP HPLC. Berger and Deye (1991) showed that phenols can be rapidly and efficiently separated by packed column supercritical fluid chromatography. The mobile phase was a mixture of methanol and C[O.sub.2]. Temperatures were on the order of 40[degrees]C to 60[degrees]C and pressures ranged from 12 MPa to 20 MPa. Industrially, the literature is abundant with pharmaceutical and biotechnology processes that use chromatography (Subramanian, 1994; Sofer, 1997). Nevertheless, production is limited to tonnes per year, much smaller than the tonnes per day anticipated with a lignin process. A company in Finland, Cultor, separates xylose and arabinose via liquid chromatography, but no technical data is available on this.

Jadhav et al. (1991; 1992) published two papers on the separation of phenols by adductive crystallization. The first one describes the separation of p-cresol from 2,6-xylenol with tert-butyl alcohol as solvent. The boiling point difference between these two compounds is only 0.03[degrees]C, whereas the melting points differ by 16[degrees]C. The authors were able to recover from 45 to 77% of the p-cresol with purity varying from 70 to 90%. The second paper relates the separation of phenol from o-cresol with 2-methyl-2-propanol as solvent. The melting point difference between these two compounds is 10[degrees]C. Recovery of phenol was in the range of 84 to 91 mol% and purity was from 74.9 to 83.6 mol%. For both processes, the authors noted that purity can be enhanced with multi-stage crystallization. Industrially, mxylene is separated from p-xylene by crystallization. p-xylene can be obtained with a purity of 99.5% with a series of melting and crystallizations (Weissermel and Arpe, 2003). The difference in melting points between the two compounds is 61[degrees]C, much larger than for Jadhav's compounds.

The literature suggests that to separate phenolic compounds in a single step with conventional methods is most likely impractical. No industrial process purifying a complex mixture, such as the BCD monomers rich fraction, has been found in the literature. Our work proposed a strategy similar to vanillin purification to obtain pure monomers, but combining more steps after the lignin depolymerization: such as acidification, liquid-liquid extraction, vacuum distillation, liquid chromatography and crystallization. Feasibility experiments were conducted with liquid-liquid extraction, vacuum distillation and liquid chromatography. Acidification combined with the removal of water with an adiabatic flash tank, vacuum distillation and liquid-liquid extraction were simulated using Aspen Plus[TM] 11.1. The general objectives of the simulation were to develop basic data for an industrial size process flow diagram, evaluate monomer losses during the separation and the energy requirements.

MATERIALS AND METHODS

Batch Liquid-Liquid Extraction (LLE)

To recover the monomers present in the acidified aqueous phase resulting from the BCD reaction, the performance of 5 different solvents was investigated: diethyl ether (DEE), dichloromethane (DCM), 4-methyl furan (MeFu), ethyl acetate (EtAc) and toluene. All solvents were analytical grade or A.C.S. reagents and used as received. Aqueous acid soluble phases from BCD experiments were extracted. Procedures to obtain the acid soluble phase and to perform the batch LLE were described in Vigneault et al. (2006). Compositions were determined by RP HPLC.

Batch Vacuum Distillation

A synthetic blend and a BCD Monomers Rich Fraction (MRF) were distilled. Compositions of both solutions are presented in Table 1. The synthetic blend was made of analytical grade chemicals, and they were used as received. MRF from Steam Exploded Aspen Lignin (SEAL) was extracted with DEE according to the procedure described in Vigneault et al. (2006).

Vacuum batch distillation was performed in a 50 ml still flask connected to a 15 cm tall vigreux distillation column. A distillation head linked the column with a water cooled condenser, which was followed by a 3 flasks receiver. Vacuum was provided by an electrical vacuum pump. Two thermometers were used to manually monitor temperatures in the still pot and at the distillation column head. Heating was provided by an 80 W hemispherical mantle and was controlled manually with a voltage controller (0-120V, 10A). A glass capillary tube provided an air bleed during the distillation, allowing the mixture to boil smoothly without bumping. The quality of the bleed tube determined the vacuum reached, which was not controlled. The monitoring and adjustment of the temperature of the water fed to the condenser was an important issue to avoid blocking the condenser because of the large range of melting points of the monomers. Water pumped into the condenser came from a water bath in which the temperature was controlled within the range of 4 to 75[degrees]C. The still pot and vigreux column were insulated.

A typical experiment consisted of a batch distillation at about 13 Pa. About 35 g of the mixture were distilled by the system. A voltage controller was set to 50% at the beginning of the distillation. The temperature rose from room temperature and products were distilled and collected. When products stopped condensing, the receiving flask was changed and the voltage controller was increased by 10%. This procedure was repeated until the temperature in the still pot reached about 160[degrees]C. The cooling water temperature in the condenser was adjusted to about 20[degrees]C lower than the column head temperature. Flask compositions were analyzed by RP HPLC.

Flash Liquid Chromatography

Flash Liquid Chromatography (FLC) was performed with silica gel (EM Science, 230-400 mesh). DEE, n-hexane and methanol were the three solvents used. They were A.C.S reagents or analytical grades and used as received. The MRF was the feedstock for the chromatography. The mixture was the same as the one used for the batch distillation (see Table 1 BCD MRF for composition).

A 250 ml column, about 38 cm tall, was filled with silica gel suspended in n-hexane. A 250 ml flask was connected to the top of the column followed by an air distributor. Air pressure was slightly adjusted (Flash LC) to obtain an n-hexane flow rate of about 6 ml/min. The air distributor was disconnected, and about 1.5 g of the MRF was poured into the column. The 250 ml flask was then filled with solvent and the air distributor was reconnected. The composition of the solvent mixture varied from pure n-hexane to pure DEE. Silica gel was washed by methanol in a last step. Figures 3 and 4 indicate the solvents ratios used and the moment of injection in the column. The mobile phase was collected in fractions of 50 ml and directly analyzed by RP HPLC.

RP HPLC

RP HPLC was described in Vigneault et al. (2006).

Aspen Plus[TM] 11.1 Simulation

Aspen Plus[TM] 11.1 simulations were performed. The thermodynamic packages used were Electrolyte NRTL and UNIQUAC, depending on the simulation. Acidification combined with the removal of water in an adiabatic flash tank, liquid-liquid extraction and vacuum distillation were simulated. Specific methodology details and block diagrams on each simulation are given in the "Simulation on Aspen Plus[TM] 11.1" section. The simulations were performed assuming a monomers yield of 7.5 wt% from a lignin feed of 50 tonne/day (monomers yield based on our previous work in Vigneault et al., 2006). To evaluate some of the energy costs for the process, the following data were used: energy for heating was assumed to be provided by steam at a cost of US$ 8.0/GJ. Cooling was assumed to be

done with water and the cost was fixed at US$ 1.9/GJ.

EXPERIMENTAL RESULTS AND DISCUSSION

Batch Liquid-Liquid Extraction

Table 2 shows the performances of 5 solvents for the extraction of monomers. Only DEE and EtAc totally extracted the monomers. As catechol is the major BCD product, toluene was the least efficient solvent, followed by MeFu and DCM.

The goal of the LLE was to extract all monomers from the acidified BCD aqueous solution obtained as one of the two phases of the BCD reaction. The other phase is the oligomeric lignin precipitate. The extracted phase was named the Monomers Rich Fraction (MRF). Previous work done on BCD used DEE to perform the extractions. DEE is well known to form unstable peroxides over time, making its use risky and costly for an industrial process. Toluene would have been an interesting choice, due to its low water solubility, its low cost and its good selectivity toward monomers. Methyl furan was also attractive in the general concept of a biorefinery. The latter could produce MeFu from unfermented pentoses and generate an in-house solvent. Ethyl acetate has three drawbacks. First, it is soluble in water up to 8.08 wt% (Riddick et al., 1986). Second, it is relatively expensive, US$ 1.71/kg (Chemical Market Reporter, 2006), making its recovery necessary. Finally, in aqueous acid solution, it degrades to acetic acid and ethanol when heated, complicating its recovery. For this reason, it was also not possible to analyze for the remaining EtAc in the MRF by ion exclusion chromatography because the conditions for the analysis involved aqueous sulphuric acid and a column temperature of 65[degrees]C. The calculation of selectivity for the EtAc extraction was probably affected by the presence of residual EtAc in the MRF. The solvent was removed using a single stage Rotovap under vacuum (Vigneault et al., 2006). Exhaustive evaporation to completely remove the solvent would likely have resulted in the loss of important monomers.

For BCD, between one half and two thirds of the fraction could not be identified as monomers. They were probably residual dimers and trimers (Vigneault et al., 2006). According to our results, monomers and di- and tri-mers will not be separated by LLE. Hence, to minimize losses, a separation strategy should use a solvent that will extract from water a maximum of monomers and di- and tri-mers. Since EtAc demonstrated very good performance for the extraction, future tests should be performed with vinyl acetate and n-butyl acetate as they have lower solubility in water (2% and 0.7%, respectively) (Riddick et al., 1986), reasonable boiling points (72.5[degrees]C and 126.1[degrees]C, respectively) and similar costs (US$ 1.23/kg and US$ 1.93/kg, respectively, Chemical Market Reporter, 2006). Methyl isobutyl ketone could also be tried as it is known to be a good extractant for low molecular weight lignin and has relatively low solubility in water ( < 2.0%) (Riddick et al., 1986).

Vacuum Batch Distillation

Figures 1 and 2 show the monomers distribution in collected fractions for both synthetic and BCD mixtures. Analysis of the synthetic mixture (Figure 1) reveals that monomers were essentially separated into 4 fractions. The first fraction, which includes phenol and guaiacol, came out between 25[degrees]C and 86[degrees]C. Then catechol and derivatives appeared mainly between 104[degrees]C and 109[degrees]C. The last fraction being distilled, from 110[degrees]C to 164[degrees]C, contained vanillin, methyl-syringol, ethyl-catechol and acetovanillone. Pyrogallol, syringaldehyde and acetosyringone stayed in the still but were mainly lost. Analysis of the BCD mixture (Figure 2) revealed a similar tendency as for the synthetic mixture, despite a poorer mass balance. Three fractions could be seen, instead of the four fractions obtained previously. Phenol and guaiacol needed higher temperatures to be fully distilled. For the distillation of the synthetic mixture, 83.6 wt% of monomers were collected in the different fractions and 13.8 wt% remained in the system (including still) to give a mass closure of 97.4 wt%. With the BCD mixture, 31.5 wt% were collected and 31.0 wt% remained for a total of 62.5 wt%. The lack of closure was likely due to DEE, because this solvent was present in the BCD mixture feed to reduce the viscosity.

Greater losses in monomers for the BCD mixture experiments than for the synthetic mixture were probably due to the limited amount of material used. Since monomers account for only 28 wt% of the BCD mixture, the amount of higher molecular weight compounds was most likely too small in quantity to go through the column and be collected and were mainly lost on the column walls. Higher temperatures were not an option because at 160[degrees]C dimers and trimers were clearly degrading, coking the bottom of the still. Higher molecular weight compounds seemed to shift the range of boiling temperatures higher: it required higher temperatures to distill phenol and guaiacol from the BCD mixture than from the synthetic mixture.

[FIGURE 1 OMITTED]

[FIGURE 2 OMITTED]

[FIGURE 3 OMITTED]

Pure products could not be obtained with this distillation system, which does not include external reflux. Use of a taller vigreux packing (i.e., a higher number of "plates") would involve losing more components to wet the column. If more attempts were made with a similar system, a minimum of 30 g of monomers should be distilled. Hence for the BCD mixture, about 100 g should be distilled (28 wt% monomers case). Overall, the experiments suggested that vacuum distillation is feasible for a primary separation of monomers into at least 3 to 4 fractions, but the dimers and trimers should be minimized before distillation to lower the operating range of the distillation.

[FIGURE 4 OMITTED]

[FIGURE 5 OMITTED]

Flash Liquid Chromatography

Results for Flash Liquid Chromatography (FLC) on silica gel are presented in Figures 3 and 4. Both figures are from the same experiments but show different compounds for the purpose of clarity. FLC separated phenolic compounds in a similar way to vacuum distillation. Phenol and guaiacol appeared first, followed by all catechols and syringol, ending up with higher molecular weight compounds, ranging from vanillin to syringaldehyde. Figure 3 demonstrates that FLC does not give a separation superior to that of distillation for lower molecular weight monomers. Phenol and guaiacol came out of the column almost at the same time; all catechols and syringol were basically mixed together. Nonetheless, separation between these two fractions was clearly obtained. The gain was more obvious in Figure 4, which shows higher molecular weight compounds only. Relatively good separation was obtained for all compounds, except for syringaldehyde and acetosyringone. Nevertheless, the quantity of all higher molecular weight compounds was small compared to catechols. Tests at higher concentrations should be performed to validate the feasibility of the separation.

Straight vertical lines on both figures indicate the eluent volumes when the solvent composition was changed. Since the column volume was about 250 ml, the last solvent change came out at about 250 ml after it was initiated. At 0 ml on the figures, the solvent inside the column was pure n-hexane, and after 250 ml it was 3:1 hexane: DEE solution was added. Hence, only pure hexane was necessary to elute phenol and guaiacol. A ratio of 2:1 hexane: DEE was necessary to extract most of the catechols and syringols. The same ratio eluted vanillin and acetovanillone. A ratio of 1:1 hexane: DEE extracted pyrogallol, acetosyringone and most of the syringaldehyde. Dimers and trimers remained in the column until pure methanol was used to elute these compounds. The RP HPLC chromatogram of the final fraction (at 1775 ml), contained a broad UV absorbance at around 25 min (not shown here). This absorbance was linked in Vigneault et al. (2006) to dimers and trimers. Thus, silica gel could serve as a selective adsorption system to separate monomers from dimers and trimers.

Although not shown on the figures for clarity, methyl guaiacol and ethyl guaiacol came out with guaiacol while p-cresol and 1,2,3-trimethoxy benzene followed the phenol peak. 3-methyl catechol and 4-ethyl catechol appeared slightly before 4-methyl catechol. Some detected compounds had similar UV spectra to other compounds but different residence times on the RP HPLC chromatogram. This was the case for pyrogallol, acetovanillone, acetosyringone and methyl syringol. Since no standard was available for methyl pyrogallol and ethyl syringol, it was possible that those two compounds were also present.

Overall, these experiments suggested that higher molecular weight monomers could be separated with liquid chromatography and dimers and trimers could be selectively adsorbed on silica (and recovered by desorption). New separation techniques, such as supercritical chromatography with carbon dioxide, also merit investigation as success has been demonstrated in the literature with phenols (Berger and Deye, 1991). The capital investment would be considerable, but separation of monomers from solvent would be trivial, by depressurization of the C[O.sub.2].

Separation Strategy Guidelines

Table 3 summarizes the physical data for the separation of all monomers produced from BCD of SEAL.

Analyzing Table 3, it is clear that no single step would completely separate all valuable monomers. A multi-step approach is thus proposed and the following guidelines were followed for the design: (1a) crystallization cannot work with more than three major components in the feed; (1b) a minimum difference of 10[degrees]C between melting points is necessary; (2a) distillation cannot be performed at temperatures higher than 160[degrees]C to avoid thermal degradation; (2b) the condenser temperature must be high enough to avoid solidification; (2c) a minimum difference of 15[degrees]C between boiling points is necessary to achieve separation; and (2d) about 30 theoretical stages represent the higher limit for a distillation column, which should be shorter than 20 m for construction reasons.

Overall Strategy

Figure 5 presents the overall strategy proposed to make the separation of the 12 major monomers produced during the BCD reaction: phenol, guaiacol, catechol, 4-methyl catechol, 3-methoxy catechol, 4-ethyl catechol, syringol, vanillin, acetovanillone, pyrogallol, syringaldehyde and acetosyringone. As seen in Vigneault et al. (2006), the BCD process leads to a complex mixture of C[O.sub.2], small organics, monomers, dimers, trimers, oligomers and an insoluble solid waste fraction. Acidification and filtration leave an aqueous phase containing essentially monomers, dimers and trimers. As the following step involves a LLE, the cost of this operation would be lower if the amount of water could be reduced: the extractor would be smaller and the water will dissolve less solvent. It has been previously proposed at NREL to vaporize part of the water by reducing sequentially the pressure, using an adiabatic flash tank separator. NREL's initial work suggested the performance of the extraction before the filtration, taking all solvent soluble material from the solid and liquid phase, all at the same time and in a single mixer. It is doubtful that a good recovery yield could be achieved with a single stage process, without a large quantity of solvent. Most likely, a counter-current multi-stage extractor would be needed and solid-liquid extraction and LLE will require separate equipment.

Starting from the hypothesis of using two distinct pieces of equipment, the extract from the LLE will contain monomers, dimers and trimers only. A silica gel bed is then used to retain most of the dimers and trimers by adsorption, as was demonstrated during the FLC experiments. The adsorption column could then be washed with methanol, supposing that the latter will dissolve negligibly the silica for regeneration. The purpose of this operation is to simplify the downstream distillation step by removing the highly viscous and thermally unstable material (i.e., dimers and trimers). Furthermore, Aspen simulation demonstrated (results not presented here) that temperatures higher than 200[degrees]C in the reboiler would be necessary to separate dimers and trimers from the monomers by distillation, increasing the risk of thermal degradation of the monomers as well. Following the LLE unit and di-trimers silica-adsorption unit, a series of 4 distillation columns, operating under vacuum, produce 5 streams: phenol, guaiacol, a catechol rich stream, a methoxy catechol rich stream and a stream with higher molecular compounds starting with ethyl catechol. Crystallizations are necessary to purify components in both catechol and methoxy catechol rich streams. The higher molecular weight stream would be submitted to liquid chromatography on silica gel. As observed during the FLC experiments (Figure 4), syringaldehyde and acetosyringone require an additional crystallization step to be separated.

[FIGURE 6 OMITTED]

Simulation on Aspen Plus[TM] 11.1

Water Flash Removal and Acidification

Figure 6 presents the block diagram used to simulate the water flash removal and acidification unit. The system, simulated with the Electrolyte NRTL thermodynamic package, included NaOH, [H.sub.2]S[O.sub.4] (and their respective ions and salts), water, phenol, guaiacol, catechol and bisphenol. Since Electrolyte NRTL did not support molecules estimated by UNIFAC, catechol was present in a quantity equal to all monomers identified, less phenol and guaiacol. Catechol being the third major volatile product, the simulation was conservative (we were looking at the amount of monomers lost in the vapour phase). Bisphenol, a dimer present in the Aspen compound databank, was the compound used to represent all dimers and trimers. Oligomers were assumed to be non-volatile and their effect negligible on the removal of water. Oligomers were therefore not included in the simulation.

The first flash tank reduced the pressure from 12.7 to 6.6 MPa and the second from 6.6 to 0.1 MPa. In the third flask tank, the mixture was acidified. The heat generated during acidification vaporized a fraction of the mixture, and with the help of vacuum (3 kPa) additional water could be withdrawn.

From the component flow rate data in Figure 6, it can be calculated that almost 60 wt% of the water can be removed. Phenol losses were considerable at 13 wt% of the inlet. However it was not an issue, since the production of the latter was too small to be economically interesting. Losses of guaiacol and catechol were reasonable, below 7 wt% and 3 wt% of the "BCD products" input, respectively. Losses of dimers and trimers, represented by bisphenol, were negligible at 0.1 wt%. Hence, water removal by adiabatic flash tanks was realistic according to the simulation data.

LLE and Removal of Dimers and Trimers

Figure 7 shows the block diagram used to simulate the LLE unit, the dimers and trimers removal adsorption unit and solvent recycling (excluding pumping). Since the NRTL Electrolyte package did not lead to reasonable results, the UNIQUAC thermodynamic package was chosen for this simulation. Hence, ionic species could not be included. UNIFAC estimated molecules could be included, but they were always shown to extract with 100% yield. Ethyl acetate was chosen as solvent for LLE. Experimentally, it was demonstrated that EtAc could easily extract monomers but also higher molecular weight compounds soluble in water. The extract, containing monomers, dimers and trimers, was directed to the di-trimers removal adsorption unit. The adsorption column must be regenerated with solvent. In the laboratory, methanol proved to efficiently extract dimers and trimers from silica gel while DEE and hexane could not. Methanol was then chosen for the simulation, but more laboratory experiments are needed to find optimal operating conditions and make this part of the simulation relevant. The stream leaving the adsorption column, containing only monomers, was then submitted to a flash separator followed by a vacuum distillation. The flash tank goal was to evaporate most of the solvent and hence reduce the size and energy duty of the distillation column.

[FIGURE 7 OMITTED]

As pointed out earlier, one of the drawbacks with EtAc is its relatively high solubility in water, which is about 8 wt% (Riddick et al., 1986). The raffinate coming out of the extractor will contain EtAc that will have to be recovered. It was proposed to use an activated carbon bed to adsorb the organic solvent. Recovery could be performed with vacuum and heat or washing the bed with methanol, which will add cost to the process. The practical use and efficiency of the activated carbon bed will have to be validated in the laboratory. If vinyl acetate or n-butyl acetate proved to efficiently extract monomers and di-trimers, the recovery would be easier, since their solubility in water is much lower, respectively, 2% and 0.7% (Riddick et al., 1986).

The LL extractor, designed with 10 theoretical stages, with a EtAc flow rate set at about 27 times larger than the amount of monomers (100 tonne/day EtAc vs 3.6 tonne/day monomers), could recover more than 99.7% of all monomers, according to the simulation. The UNIQUAC method predicted easy extraction of catechol with EtAc at standard conditions. However, simulations with toluene and methyl furan showed similar results, even though catechol was poorly extracted with these two solvents. Hence, the degree of confidence was limited for this part of the simulation. Calculated from the component flow rate data, losses of economically interesting monomers were all below 0.5 wt% of their respective amount entering the unit.

The LLE unit showed promising results, and a solvent recycling strategy was thus proposed (shown in Figure 7). Solvent recycling will have to be validated experimentally to confirm its feasibility. The di-trimers removal unit was shown as a front end but also needs more experimental results with reliable operating conditions to make a relevant simulation and test its economic feasibility.

Water Distillation vs LLE Unit

A water distillation unit was simulated using NRTL as the thermodynamic package. The NRTL package was used without base or acid, which tends to give erroneous results in distillation. To remove the 186 tonne/day of water, remaining after the flash removal, simulation indicated that the energy requirement of the reboiler was very high, 6525 kW, even though the reflux ratio was small at 0.2 with a design of 7 theoretical plates (condenser pressure set to 110 kPa). By comparison, the heating requirement for the LLE unit was 466 kW for the flash tank and 71 kW at the reboiler. Water enthalpy of vaporization, which is 2257 kJ/kg at standard conditions (Sandler, 1999), makes distillation about 12 times more energy demanding than for the LLE unit (excluding energy requirement to obtain "Recycle 1" on Figure 7 and all vacuum and pumping energy). The energy costs for the flash tank and the distillation column of the LLE unit were calculated as US$ 396/day or US$ 0.79/100 kg lignin fed.

Vacuum Distillation Unit

The vacuum distillation unit was simulated using UNIQUAC as the thermodynamic package. Figure 8 presents 4 distillation columns operating under vacuum and dedicated to monomers separation. The first column separated phenol (Stream "101-182"[degrees]C) and the second made a stream rich in guaiacol (Stream "183+"[degrees]C). The third separated higher molecular weight compounds having a normal boiling point higher than 262[degrees]C. The lighter fraction (Stream "236-261"[degrees]C) went to the last distillation column to make a catechol and methoxy catechol rich streams. Table 4 shows inlet and outlet stream conditions and compositions. As indicated in Table 4, 7 compounds were created with the group contribution method UNIFAC. To make the simulation more relevant, available boiling points from the literature and vapour pressure data were provided to the simulator. The mass fractions shown in bold type indicate the compounds with economic potential. Distillation provided a first purified compound, guaiacol, with small losses, 0.6 wt%. All other streams would require further separation steps to be purified. 3-methoxy catechol and syringol had the biggest losses, but these were still small at 1.7 wt% each. Both compounds could not be completely isolated in one stream.

[FIGURE 8 OMITTED]

Low pressure drop was a key issue for the design of the vacuum columns. The pressure must be as low as possible to reduce component boiling points. Since flow rates are relatively small, column diameters would be equally small (< 1 m diameter). For small diameter and low pressure applications, randomly packed columns are usually preferred (Kister, 1992; Peters et al., 2003). Pressure drop estimates were based on the minimum pressure drop curve presented in Eckert's generalized correlation chart (Kister, 1992). The design was based on a 25 mm plastic Ball random packing. For this type of packing, a height equivalent to a theoretical plate (HETP) was fixed at 0.5 m (Peters et al., 2003). Calculated column diameters were all below 1 m and most of them were below 0.5 m. The largest one had 31 stages and should approach 20 m tall, reflecting the difficulty of making further separation via distillation. Energy requirements were relatively small for the distillation: combining and converting the 4 reboiler and condenser duties into energy cost, US$ 81/day or US$ 0.16/100 kg lignin fed would be spent for this unit (excluding pumping cost).

CONCLUSION

This paper proposed a strategy for separating 12 major monomers generated during the base catalyzed depolymerization (BCD) of lignin. The approach was to combine liquid-liquid extraction (LLE), vacuum distillation, chromatography, and crystallization operations. The feasibility of extraction, distillation and liquid chromatography was tested in the laboratory. EtAc and DEE were the best solvents for the extraction of monomers. Since DEE can generate unstable peroxides, EtAc was preferred, even though it is soluble in water, up to 8 wt%. Batch distillation under vacuum with a 15 cm vigreux column showed that up to 4 distinct fractions of monomers can be achieved. An alternate strategy via flash liquid chromatography (FLC) on silica gel made the separation of higher molecular weight compounds, from methyl syringol to syringaldehyde, conceivable. However, lower molecular weight compounds could not be separated by FLC. Simulation showed that about 60 wt% of water in the aqueous liquor (i.e., the liquor that contains the monomers rich fraction, MRF) coming out the BCD reactor could be removed by adiabatic flash separators. Dimers and trimers present in the MRF should be removed prior to the monomers separation processes to avoid degradation during high temperature distillation. Removal of dimers and trimers by a silica gel adsorber was proposed as a front end, but its economic feasibility has yet to be proven. Simulation reveals that, with a 10-stage counter-current extractor, a 99.7 wt% recovery for each monomer could be envisaged with EtAc. Solvent recycling is crucial, and its technical feasibility needs to be proven. Water distillation, to replace the LLE unit, would require more than 12 times more heat than LLE. Simulation demonstrates that, after the LLE unit, a series of 4 vacuum distillation columns could make 5 streams of monomers, where losses would be small (< 2 wt%). Three streams require further chromatography and crystallization operations to purify them.
NOMENCLATURE

BCD Base Catalyzed Depolymerization
DCM Dichloromethane
DEE Diethyl Ether
EtAc Ethyl Acetate
FLC Flash Liquid Chromatography
GPC Gel Permeation Chromatography
HETP Height Equivalent of a Theoretical Plate
LC Liquid Chromatography
LL Liquid-Liquid
LLE Liquid-Liquid Extraction
MeFu Methyl Furan
MRF Monomers Rich Fraction
NREL National Renewable Energy Laboratory
RON Research Octane Number
RP HPLC Reverse Phase High Performance Liquid
 Chromatography
SEAL Steam-Exploded Aspen Lignin


Manuscript received September 29, 2006; revised manuscript received April 3, 2007; accepted for publication June 15, 2007.

REFERENCES

Berger, T. A. and J. F. Deye, "Separation of Phenols by Packed Column Supercritical Fluid Chromatography," J. Chromatogr. Sci. 29, 54-59 (1991).

"Chemical Market Reporter," 27 February-5 March 2006, Retrieved February 2006, from chemicalmarketreporter.com (2006).

Forss, K. G., E. T. Talka and K. E. Fremer, "Isolation of Vanillin from Alkaline Oxidized Spent Sulfite Liquor," Ind. Eng. Chem. Process Des. Dev. 25, 103-108 (1986).

Jadhav, V. K., M. R. Chivate and N. S. Tavare, "Separation of p-Cresol from its Mixture with 2,6-Xylenol by Adductive Crystallization," J. Chem. Eng. Data 36, 249-251 (1991).

Jadhav, V. K., M. R. Chivate and N. S. Tavare, "Separation of Phenol from its Mixture with o-Cresol by Adductive Crystallization," J. Chem. Eng. Data 37, 232-235 (1992).

Johnson, D. K., "Characterization Methodology and Its Application to Lignin and BCD Tar Product: Rationale, Method Development, Results, Discussion and Conclusions," report, National Renewable Energy Laboratory, Golden, CO, (1997).

Kadangode, S. M., "Lignin Conversion into Reformulated Hydrocarbon and Partially Oxygenated Gasoline Composition," Thesis, University of Utah, Salt Lake City, UT, (2000).

Kister, H. Z., "Distillation Design," McGraw-Hill, New York, (1992).

Li, H. and D.-L. Yin, "The Technical Advance in Synthesis of Vanillin," Jinshu Xuebao 22(1), 90-93 (2001).

Miller, J. E., L. R. Evans, Jason E. Mudd and Kara A. Brown, "Batch Microreactor Studies of Lignin Depolymerization by Bases. 2. Aqueous Solvents," report, Sandia National Laboratories, Albuquerque, NM (2002).

Nomura, N., R. Kaneko and M. Hara, "Reversed-Phase Liquid Chromatography of Alkylphenols Depending on their Dissociation Constants," J. Liq. Chromatogr. 15(5), 885-896 (1992).

Peng, J., F. Lu and John Ralph, "The DFRC Method for Lignin Analysis. 4. Lignin Dimers Isolated from DFRC-Degraded Loblolly Pine Wood," J. Agric. Food Chem. 46, 553-560 (1998).

Peters, M. S., K. D. Timmerhaus and R. E. West, "Plant Design and Economics for Chemical Engineers," McGraw-Hill, New York (2003).

Riddick, J. A., W. B. Bunger and T. K. Sakano, "Organic Solvents Physical Properties and Methods of Purification," Wiley, New York (1986).

Sandler, S. I., "Chemical and Engineering Thermodynamics," Wiley & Sons, New York (1999).

Sarkanen, K. V. and C. H. Ludwig, Eds., "Lignins Occurrence, Formation, Structure and Reactions," New York, Wiley & Sons (1971).

Shabtai, J. S., W. W. Zmierczak and E. Chornet, "Process for Conversion of Lignin to Reformulated Hydrocarbon Gasoline," U.S. Patent 5,959,167 (1999).

Shabtai, J. S., W. W. Zmierczak and E. Chornet, "Process for Conversion of Lignin to Reformulated, Partially Oxygenated Gasoline," U.S. Patent 6,172,272 (2001).

Sofer, G. K., "Handbook of Process Chromatography," Academic Press, Millborne, CA (1997).

Subramanian, G., Ed., "Process Scale Liquid Chromatography," VCH, New York (1994).

Vigneault, A., D. K. Johnson and E. Chornet, "Advance in the Thermal Depolymerization of Lignin via Base-Catalysis," in "Science in Thermal and Chemical Biomass Conversion Conference," CPL Press, Victoria, BC, pp. 1401-1419, (2006).

Weissermel, K. and H.-J. Arpe, "Industrial Organic Chemistry," Wiley-VCH, Weinheim, Germany (2003).

Alexandre Vigneault (1), David K. Johnson (2) and Esteban Chornet (1,2)

(1.) Departement de genie chimique, Universite de Sherbrooke, Sherbrooke, QC, Canada J1K 2R1

(2.) National Renewable Energy Laboratory, 1617 Cole Blvd, Golden, CO, U.S.A. 80401

* Author to whom correspondence may be addressed.

E-mail address: mchornet@interlinx.qc.ca
Table 1. Synthetic mixture and BCD MRF compositions for vacuum
batch distillation

Compounds Synthetic BCD MRF
 wt% wt% (a,b)

Phenol 7.8% 8.4%
Guaiacol 19.9% 10.0%
1,2,3-Trimethoxy benzene 0.3% 1.7%
Catechol 27.4% 29.6%
3-Methoxy catechol 14.5% 14.2%
4-Methyl catechol 14.0% 12.5%
Syringol 7.8% 8.8%
Vanillin 2.5% 1.8%
4-Methyl syringol 0.1% 1.5%
4-Ethyl catechol 0.3% 4.1%
Acetovanillone 1.4% 1.2%
Pyrogallol 0.9% 0.3%
Syringaldehyde 1.5% 1.2%
Acetosyringone 1.7% 3.4%

(a.) wt% over total identified monomers

(b.) identified monomers represented 28 wt% of the total fraction

Table 2. Solvents performances (a) in extracting monomers and
selectivity (b, c)

 DEE DCM Toluene Ethyl 4-methyl
 acetate furan
Monomers
Catechol 100% 32% 8.70% 100% 27%
3-Methoxy catechol 100% 92% 42% 100% 74%
Phenol 100% 100% 100% 100% 77%
4-Methyl catechol 100% 59% 26% 100% 63%
Vanillin 100% 100% 100% 100% 100%
Guaiacol 100% 100% 100% 100% 100%
Syringaldehyde 100% 100% 100% 100% 100%
Acetovanillone 100% 100% 100% 100% 100%
Syringol 100% 100% 100% 100% 100%
4-Ethyl catechol 100% 100% 100% 91% 100%
Selectivity (b, c) 0.42 0.51 0.72 0.35 0.54

(a.) Performance for each monomers = Extracted Mass / Mass before
extraction

(b.) Selectivity = Total mass of extracted and identified
monomers / MRF total mass

(c.) Selectivity for DEE, DCM and Ethyl acetate may be lowered by
remaining solvent not quantified

Table 3. Summary of physical data for separation of monomers

Separation Method Crystallization (a) Distillation (a)

Properties Melting points Boiling points
 ([degrees]C) ([degrees]C)

Phenol 41 182
Guaiacol 30 205
4-Methyl guaiacol 2 222
4-Ethyl guaiacol 15 235
1,2,3-Trimethoxy benzene 45 241
Catechol 104 245
3-Methyl catechol 67 250 (c)
4-Methyl catechol 68 251
3-Methoxy catechol 42 250 (c)
Syringol 55 261
Vanillin 82 285
4-Methyl syringol 40 294 (c)
4-Ethyl catechol 39 295 (c)
Acetovanillone 115 298
Pyrogallol 134 309
Syringaldehyde 112 324 (b)
Acetosyringone 126 331 (b)
Di-trimers

Separation Method Chromatography

 Flash (d) RP HPLC (e)

Properties Max peak Peak width Peak
 (ml) (ml) (min)

Phenol 75 30-180 10.9
Guaiacol 15 0-120 14.1
4-Methyl guaiacol 15 0-75 21.6
4-Ethyl guaiacol 15 0-90 25.9
1,2,3-Trimethoxy benzene 135 120-185 23.7
Catechol 375 375-575 6.6
3-Methyl catechol 315 270-375 13.5
4-Methyl catechol 425 330-575 12.1
3-Methoxy catechol 525 425-825 8.9
Syringol 425 425-575 16.3
Vanillin 675 625-725 13.1
4-Methyl syringol 525 475-575 22.7
4-Ethyl catechol 360 255-475 20.0
Acetovanillone 825 725-875 15.1
Pyrogallol 1000 925-1075 3.6
Syringaldehyde 1175 1175-1425 15.1
Acetosyringone 1175 1125-1225 16.6
Di-trimers 1775 25.5

(a.) Unless otherwise specified, data are taken from manufacturer
web sites, such as Sigma-Aldrich and Lancaster Synthesis

(b.) Bold = From UNIFAC property estimation

(c.) Italic = Estimation according to vacuum distillation results

(d.) Data extracted from the flash chromatography experiments

(e.) Data extracted from the RP HPLC analytical calibration table

Table 4. Simulation of vacuum distillation unit: main inlet and
outlet stream data

 Inlet Outlet

Stream Names Monomers 101- 183- Catechol
 182 235 Rich

Operating Parameters
Temperature ([degrees]C) 152.5 78.1 109.5 105.2
Pressure (kPa) 13.3 5.6 4 0.7
Total Mass Flow (kg/day) 3546 306 443 1650
Mass Fraction
Water 0 0.005 0 0
Ethyl Acetate 0 0.001 0 0
Phenol 0.086 0.993 0.005 0
Guaiacol 0.125 0.001 0.995 (d) 0.001
Catechol 0.32 0 0.001 0.685
4-Methyl Catechol (a) 0.157 0 0 0.309
3-Methoxy Catechol (a) 0.105 0 0 0.004
Syringol (a) 0.048 0 0 0.001
Vanillin 0.014 0 0 0
4-Methyl Syringol (a) 0.009 0 0 0
4-Ethyl Catechol (a) 0.07 0 0 0
Acetovanillone 0.012 0 0 0
Pyrogallol 0.01 0 0 0
Syringaldehyde (a) 0.007 0 0 0
Acetosyringone (a) 0.018 0 0 0
Bisphenol A 0.018 0 0 0

 Outlet % Lost

Stream Names MeO 262+ During
 Catechol Distillation
 Rich

Operating Parameters
Temperature ([degrees]C) 137.9 159.5
Pressure (kPa) 2 0.8
Total Mass Flow (kg/day) 615 532
Mass Fraction
Water 0 0
Ethyl Acetate 0 0
Phenol 0 0
Guaiacol 0 0 0.60%
Catechol 0.005 0 0.40%
4-Methyl Catechol (a) 0.076 0 0.00%
3-Methoxy Catechol (a) 0.595 0 1.70%
Syringol (a) 0.272 0 1.70%
Vanillin 0 0.094 0.00%
4-Methyl Syringol (a) 0.052 0
4-Ethyl Catechol (a) 0.001 0.465 0.00%
Acetovanillone 0 0.081 0.00%
Pyrogallol 0 0.067 0.00%
Syringaldehyde (a) 0 0.047 0.00%
Acetosyringone (a) 0 0.122 0.00%
Bisphenol A 0 0.122

(a.) Compound simulated by UNIFAC

(b.) 100%--Component mass flow (c) econonic potential outlet
stream(s) / Component mass flow "Monomers" stream

(c.) Component mass flow = Total mass flow * Component mass fraction

(d.) Bold = corresponds to compound with economic potential
COPYRIGHT 2007 Chemical Institute of Canada
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 2007 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:Vigneault, Alexandre; Johnson, David K.; Chornet, Esteban
Publication:Canadian Journal of Chemical Engineering
Date:Dec 1, 2007
Words:7414
Previous Article:Heavy metal removal and neutralization of acid mine waste water--kinetic study.
Next Article:Characterization of copper foam as catalytic material in ethanol dehydrogenation.

Terms of use | Copyright © 2017 Farlex, Inc. | Feedback | For webmasters