Pretreatment and Hydrolysis of Biomass

Design and Installation of Engineering-Scale Countercurrent Pretreatment Reactor System

Research Funded by: U.S. Department of Energy Office of Fuels Development through the National Renewable Energy Laboratory

Project Manager: Robert Wooley 303.384.6825, Robert_Wooley@nrel.gov

Performing Organization: National Renewable Energy Laboratory, 1617 Cole Boulevard, Golden, CO 80401-3393, http://www.nrel.gov

Principal Investigator: Richard Elander, 303.384.6841, richard_elander@nrel.gov

Contract Number: N/A

Contract Period: 10/97–09/99

Contract Funding:

FY 1998: $273,000

FY 1999: $650,100

Objective: The main objective is to achieve initial confirmation of feasibility performance targets in the operation of an engineering-scale countercurrent pretreatment reactor system, including investigation of lower-temperature and lower-pressure operating conditions in the countercurrent stage.

Approach/Background: A two-stage continual shrinking-bed process has been developed and demonstrated in a bench-scale apparatus. Bench-scale hydrolysis yield, detoxification, and hydrolyzate fermentation performance targets selected using the ASPEN process engineering, and economic models have been met. An engineering-scale countercurrent pretreatment reactor system (200 kg dry biomass/day) has been installed in the Alternative Fuels User Facility Process Development Unit. All ancillary process systems, instrumentation, and control systems have also been designed, installed, and tested. System shakedown activities and initial operations are under way.

Status/Accomplishments: A two-stage reactor system at engineering-scale, based upon the countercurrent hydrolysis concept, has been designed, fabricated, and installed in the Process Development Unit; it is currently being operated to collect key process performance data. These performance targets (sugar yields in liquid hydrolyzate streams, preservation of remaining cellulose in the residual solids, and maintenance of an acceptably low liquid:solid ratio) were selected on the basis of achieved bench-scale process performance data and guidance from process engineering models.

Several ancillary systems are required for the operation of this system, and all such systems have been designed and installed. These systems include the dilute acid and deionized water heating and delivery systems, feedstock delivery system, hydrolyzate liquor collection system and pretreated solids collection system, along with numerous utility systems (steam, condensate, process water, cooling water, instrument air, nitrogen, and exhaust gas). In addition, electrical power supply, instrumen-tation, control, and data collection systems were also designed.

A number of operating conditions are being investigated, including lower temperature and pressure operating conditions in the countercurrent stage, as these conditions may be more practical for commercial operation of this technology.

Publications and Presentations:

1. Elander, R., T. Johnston, R. Torget, A. Mokvist, and C. Kajzer. 1999. Systematic design of a novel two-stage engineering-scale dilute acid hydrolysis reactor system. Presented at the Twenty-first Symposium on Biotechnology for Fuels and Chemicals. Fort Collins, CO. May 2–6 [1999].
2. Torget, R., N. Nagle, E. Jennings, K. Ibsen, and R. Elander. 1999. A novel pilot-scale reactor for the aqueous fractionation of hardwood for the improved production of fuel ethanol. Presented at the Twenty-first Symposium on Biotechnology for Fuels and Chemicals. Fort Collins, CO. May 2–6 [1999].
3. Elander, R., T. Johnston, J. Farmer, and R. Torget. 1998. An engineering-scale countercurrent dilute acid hydrolysis reactor to achieve high yields of soluble sugars from biomass. Presented at the Twentieth Symposium on Biotechnology for Fuels and Chemicals. Gatlinburg, TN. May 3–7 [1998].

Summary Date: March 2000

Identification and Evaluation of Cost-Reduction Opportunities for Countercurrent Pretreatment Process

Research Funded by: U.S. Department of Energy Office of Fuels Development through the National Renewable Energy Laboratory

Project Manager: Robert Wooley 303.384.6825, Robert_Wooley@nrel.gov

Performing Organization: National Renewable Energy Laboratory, 1617 Cole Boulevard, Golden, CO 80401-3393, http://www.nrel.gov

Principal Investigator: Richard Elander, 303.384.6841, richard_elander@nrel.gov

Contract Number: N/A

Contract Period: 10/97–09/99

Contract Funding:

FY 1998: $0

FY 1999: $551,700

Objective: We updated the process engineering model for the countercurrent pretreatment process. Based upon program-wide modeling improvements, we completed updates to the technical design basis (which now reflects the most current understanding of the countercurrent pretreatment process) and to several key unit operations. The revised model projected ethanol production costs that indicate a strong potential for a cost-competitive process based upon countercurrent pretreatment technology, especially in the near- to mid-term.

Approach/Background: We revised and up-dated the process engineering and economic models that have been used for all processes under development in the Biofuels Program; these models estimate ethanol production costs and help researchers prioritize and guide research direction. Several subcontracts placed by the Process Engineering Team have provided new information for many areas of the process, including burner and boiler design, wastewater treatment, and distillation. Some of these changes have significantly altered the process design in key areas such as wastewater treatment. During the updates to the countercurrent pretreatment process model, we evaluated the effects of such changes on the overall countercurrent pretreatment process and determined if such changes are rational for this particular process. Sensitivities in key process parameters also affect how to incorporate these changes.

Status/Accoumplishments: The set of technical assumptions that govern the countercurrent pretreatment process have been revised to reflect the most current understanding of the key process parameters. The basic countercurrent pretreatment total hydrolysis process consists of three stages. The first stage is a cocurrent steam treatment at 183ºC for 9 min at 30% solids loading. The second stage is a countercurrent, flowing dilute sulfuric acid (0.07% w/v) at 225ºC and 4 min of solids residence time. The liquid residence time depends on the flow rate of dilute-acid solution relative to the solids residence time. This process parameter has the greatest effect on process economics. Operational data in the engineering-scale reactor system will determine the actual liquor-volume requirements.

The degree of hydrolysis that can be achieved in the second stage is uncertain. There may be hydrodynamic limitations, as well as limitations on an economically acceptable liquor volume, as total hydrolysis of cellulose is approached in the second stage. For now, a 4-min solids residence time in the second stage is assumed. That time interval will convert 55% of cellulose, based upon kinetic modeling and bench-scale experimental results, and it will recover significant percentages of unconverted cellulose in the pretreated solids. A placeholder third stage, based upon a dilute sulfuric-acid plug-flow slurry reactor, is included in the base case countercurrent pretreatment process model to give greater conversion of cellulose.

Several sensitivity scenarios on a three-stage process, as well as a two-stage process with possibly greater cellulose conversion levels, have been developed. In addition, out-year scenarios are included. The basic conclusion of the economic analysis is that if expected liquor volume and sugar yield targets can be demonstrated in the engineering-scale system, then the countercurrent pretreatment process offers an economically competitive (and potentially advantageous) near-term process alternative. Any further process problems are largely related to mechanical reactor design issues that can, in a relatively short time, either be overcome or shown to take longer to address. The countercurrent pretreatment process can be economically competitive until at least 2010. This analysis does not include the potential economic benefits associated with recovery and production of higher-value coproducts from the solubilized lignin that is a product of the countercurrent pretreatment process. The results of this updated model will be used to select performance targets and aid in the design and interpretation of experiments in the engineering-scale countercurrent pretreat-ment system.

Publications and Presentations:

1. Nagle, N., K. Ibsen, and E. Jenning. 1999. A process economic approach to develop a dilute-acid cellulose hydrolysis process to produce ethanol from biomass. Applied Biochemistry and Biotechnology, 77–79:595–607.

Summary Date: March 2000

Pretreatment and Chemical Hydrolysis Reaction Kinetics Verification

Research Funded by: U.S. Department of Energy Office of Fuels Development through the National Renewable Energy Laboratory

Project Manager: Robert Wooley 303.384.6825, Robert_Wooley@nrel.gov

Performing Organization: Auburn University, Department of Chemical Engineering, Auburn, AL 36749-5127

Principal Investigator:Y.Y. Lee, 334. 844-2019, yylee@eng.auburn.edu

Contract Number: XGC-7-17041-01

Contract Period: 10/97–09/99

Contract Funding:

FY 1998: 124,500

FY 1999: $20,800

Objective Our first objective is to understand the effects of operating variables on yields of three products of NREL's multistage (cocurrent–countercurrent) dilute sulfuric-acid hydrolysis process: xylose and glucose from hemicellulose, and glucose from cellulose. Our second objective is to support NREL research staff, both in-house and subcontracted, by providing chemical analyses of biomass samples.

Approach/Background: When lignocellulosic biomass is subjected to thermochemical treatment with dilute sulfuric acid, hemicellu-losic sugars (primarily xylose) and glucose from cellulose are both released. By using experi-mental and modeling studies, we are deter-mining the relationship between particle size, temperature, residence time, acid concentration, and physical shrinkage of the solid mass as a function of hydrolysis and sugar yield.

This study uses yellow poplar sawdust to establish kinetic models that predict yields of sugars treated with dilute sulfuric acid treatment. These kinetic models are incorporated into the appropriate reactor design equations for a multistage simulated countercurrent percolation process. The models determine the optimum operating conditions that maximize sugar yields at concentrations meaningful from a process economic standpoint. Additionally, intraparticle and interparticle mass and heat transfer phenomena are being investigated.

Status/Accoumplishments: Under previous sub-contracts, countercurrent hydrolysis models were developed by Dr. Y.Y. Lee at Auburn University. These models have been modified to incorporate the shrinking-bed effect of this particular hydrolysis reactor design. They also have been modified to predict yield losses due to flow nonideality. The Peclet number analysis, which was conducted in the prototype vertical countercurrent reactor, has been used in this model to predict the yield-loss effects of such nonidealities. The analysis was adjusted to account for higher rates of diffusion expected at process operating temperatures. When this information is incorporated into the model, it predicts no more than a 10% yield loss of glucose owing to nonideality. The overall glucose yield is expected to be 75%–85% in the engineering-scale countercurrent system oper-ated in a total hydrolysis mode.

This shrinking-bed kinetic model is being packaged in a user-friendly menu-driven software format that will allow users to enter the kinetic parameters of different feedstocks.

Additional work has increased our understanding of the effect of reactor design and reaction configuration on the kinetic parameters of hemicellulose and cellulose dilute-acid hydrolysis. This work has indicated that kinetic parameters determined in a batch reactor configuration do not necessarily apply to other reaction configurations, such as packed-bed percolation or countercurrent configurations.

Large quantities of hydrolysis liquors have been produced for subsequent soluble lignin adsorption studies.

Publications and Presentations:

1. Lee, Y.Y., Z. Wu, and R. Torget. 1999. Modeling of countercurrent shrinking-bed reactor in dilute-acid total hydrolysis of lignocellulosic biomass. Bioresource Technol-ogy, 71:29–39.
2. Lee, Y.Y., P. Iyer, and R. Torget. 1999. Dilute acid hydrolysis of lignocellulosic biomass. Advances in Biochemical Engineering and Biotechnology, 65:93–116.
3. Torget, R., N. Nagle, K. Ibsen, E. Jennings, R. Elander, P. Iyer, and Y.Y. Lee. 1998. Production of monomeric carbohydrates from biomass—kinetics and reactor design. Presented at the Twentieth Symposium on Biotechnology for Fuels and Chemicals. Gatlinburg, TN. May 3–7 [1998].

Summary Date: March 2000

Preparation and Characterization of Novel Titania Designed for Efficient Biomass Fractionation and Upgrading

Research Funded by: U.S. Department of Energy Office of Fuels Development through the National Renewable Energy Laboratory

Project Manager: Robert Wooley 303.384.6825, Robert_Wooley@nrel.gov

Performing Organization: University of Missouri-Columbia, Department of Chemical Engineering, Columbia, MO 65212

Principal Investigator:William Jacoby 573. 822.5037, jacoby@missouri.edu

Contract Number: XCI-9-29056-01

Contract Period: 10/97–09/99

Contract Funding:

FY 1998: $39,000

FY 1999: $30,100

Objective In this project we are preparing and characterizing ultra-high-surface regenerable adsorbents. The adsorbents are to be used to selectively recover solubilized biomass fractions from process streams produced by dilute-acid hydrolysis.

Approach/Background: Process configurations based upon the shrinking-bed–countercurrent dilute-acid hydrolysis concept solubilize as much as 60% of the lignin in hardwood yellow poplar and put it into the hydrolysis liquor stream. The majority of this solubilized lignin must be recovered, either for use as boiler fuel or to serve as an intermediate in the production of value-added fuel additives or chemicals. A process based on adsorbing various fractions of the solubilized lignin upon a regenerable solid adsorbent has been developed for this purpose.

The project’s goal is to improve the adsorption efficiency of this process. We also investigated the possibility of upgrading the adsorbed lignin in situ by using an active-metal-doped adsorbent. Three main tasks are asso-ciated with this project. The first is to prepare and characterize more efficient (i.e., higher surface area) adsorbents. The second is to prepare active-metal-doped adsorbents, and the third is to evaluate these adsorbents in upgrading reactions.

Status/Accoumplishments: A variety of tech-niques are being investigated to improve the efficiency the solid adsorbents. Ultra-high-surface area sol-gel formulations of selected adsorbents have been prepared using aerogel and xerogel techniques. These formulations are being evaluated for surface area, crush strength, adsorption capacity, and regeneration characteristics.

Several promising formulations have been identified. These formulations are now being doped with various amounts of active metals such as nickel, cobalt, and platinum. The doped adsorbents are being characterized by methods such as infrared, thermogravimetric, X-ray diffraction, and Brunauer-Emett-Teller analysis.

Both doped and undoped adsorbents with acceptable properties are being evaluated; evaluation conditions are appropriate for upgrading the adsorbed lignin compounds to liquid fuel additives or other chemicals. These reactions may require high temperatures and pressures similar to those used in conventional industrial hydrogenation reactions.

Publications and Presentations: None

Summary Date: March 2000

Liquid Hot Water Pretreatment of Yellow Poplar Sawdust

Research Funded by: U.S. Department of Energy Office of Fuels Development through the National Renewable Energy Laboratory

Project Manager: Robert Wooley 303.384.6825, Robert_Wooley@nrel.gov

Performing Organization: Hawaii Natural Energy Institute, University of Hawaii-Manoa, Holmes Hall, 2540 Dole St., Honolulu, HI 96822

Principal Investigator: Michael Antal, 808.956.7267, antal@wiliki.hawaii.edu

Contract Number: XXE-8-17099-0

Contract Period: 10/97–09/99

Contract Funding:

FY 1998: $55,000

FY 1999: $60,100

Objective An immersed percolation reactor at the University of Hawaii was used to evaluate the liquid hot water pretreatment of yellow poplar sawdust. The pretreated solids were evaluated at NREL for their performance in simultaneous saccharification and fermentation (SSF) to ethanol.

Approach/Background: A liquid hot water pretreatment approach was developed at the University of Hawaii. An immersed percolation reactor and boiler produces liquid hot water at high temperatures and pressures (up to 230ºC and 400 psig). This water then contacts biomass in a packed percolation mode; the water is in either a static or a flowing mode.

This approach has been tested on several biomass feedstocks over the years and shows potential as a nonacid pretreatment method. Our study is comparing this approach with the existing dilute-acid prehydrolysis approach; the comparison uses a common feedstock (yellow poplar sawdust), common yield calculations, and common solids digestibility and fermentability evaluations.

Status/Accoumplishments: Several liquid hot water pretreatment conditions were tested on yellow poplar sawdust. Temperatures ranged from 188º to 223ºC and residence times ranged from 2 to 7.5 min. The pretreatments took place in an immersed percolation apparatus at the University of Hawaii and used a collected liquor:bone dry solids ratio ranging from 14:1 to 20:1. The most effective condition (using the highest level of xylan solubilization coupled with high levels of xylose recovery as the selection criteria) was at 215ºC at a solids residence time of 2 min and a liquid:solid ratio of 19:1. This condition solubilized 92% of xylan and recovered 85% of the solubilized xylan as xylose (after posttreatment—much of the solubilized xylan is oligomeric).

Pretreated solids were sent to NREL for a set of SSF runs on the washed and pretreated solids using Saccharomyces cerevisiae D₅A. Four solids pretreated with liquid hot water were evaluated, along with a Solka-Floc® control and a control using single-stage dilute-acid prehydrolyzed washed solids from NREL’s Process Development Unit Sunds hydrolyzer. The most effective liquid hot water pretreatment (as described above) outperformed the single-stage dilute-acid prehydrolyzed solids in ethanol yield and rate: an ethanol yield of 97% was achieved at 96 h. This result compares with the dilute-acid pretreated solids ethanol yield of 86% at 96 h (92% at 168 h). Other liquid hot water-pretreated solids that were tested, all of which had considerably lower levels of xylan solubilization during pretreatment, performed much more poorly in SSF.

Although the pretreatment yield and SSF results of the best liquid hot water pretreatment appear promising, several difficul-ties are associated with this process. Very large amounts of water are used, resulting in highly dilute liquid streams and high demand for steam to heat the liquid. The high pressures (300 psig at 215ºC) will cause biomass feeding difficulties and other operability concerns in a continuous, large-scale process. A thorough process engineering and economic analysis is necessary to determine if this approach is economically feasible.

Publications and Presentations: None

Summary Date: March 2000

Bench-Scale Development of Countercurrent Pretreatment Process

Research Funded by: U.S. Department of Energy Office of Fuels Development through the National Renewable Energy Laboratory

Project Manager: Robert Wooley 303.384.6825, Robert_Wooley@nrel.gov

Performing Organization: National Renewable Energy Laboratory, 1617 Cole Boulevard, Golden, CO 80401-3393, http://www.nrel.gov

Principal Investigator: Robert Torget, 303.384.6178, robert_torget@nrel.gov

Contract Number: N/A

Contract Period: 10/97–09/98

Contract Funding:

FY 1998: $752,200

FY 1999: $0

Objective Since 1995, we have been developing an advanced-pretreatment– thermochemical-hydrolysis approach based upon the concepts of a shrinking-bed reactor and countercurrent contacting of biomass solids and dilute-acid–liquor streams. Process engineering models developed for this approach serve as a tool for determining specific bench-scale process performance targets that will result in an economically competitive ethanol production process.

Approach/Background: The process engineer-ing and economic model for the countercurrent hydrolysis process identified several key performance expectations that must be met if ethanol production is to meet its competitive-cost target. These expectations include (1) yields of hemicellulosic sugars of 95% or greater; (2) yields of glucose greater than 50% in a partial hydrolysis mode, at least 35% of the glucose remaining as solid cellulose, and concentrations of total sugars greater than 3.5 wt% without process flashing; (3) yields of glucose greater than 75% in a total hydrolysis mode, and concentrations of total sugars greater than 4 wt% without process flashing; (4) yields of ethanol at the 95% level from glucose and at the 60% level from xylose using the cofermenting recombinant Zymomonas mobilis in 48 h, and (5) minimizing reactor capital equipment costs by keeping residence times short (thus minimizing reactor size) and minimizing or eliminating the use of sulfuric acid (thus allowing at least some components of the reactor system to be fabricated from less- expensive stainless steel).

Status/Accoumplishments: All of the above performance expectations were essentially met. A hot-water static first stage was used to hydrolyze approximately 60% of the xylan, followed by a temperature-ramped dilute-acid percolation second stage; this process recovered a mean of 94.6% xylose (standard deviation 1.68%).

Expectation (2) above was exceeded. In a partial hydrolysis mode, glucose recoveries of 55.0% (standard deviation 5.4%) were obtained, 38% of the glucose remained in the solid, and total sugar concentrations were 3.87 wt% without process flashing.

Expectation (3) above was exceeded in one respect. In a total hydrolysis mode, glucose recoveries of 86.5% were obtained. However, owing to a slight increase in the volume of hydrolysis liquor used as compared with the partial hydrolysis mode, the total sugar concentration was only 3.7 wt% (instead of 4.0 wt%) without process flashing. Only limited runs were done in total hydrolysis mode; adding a degree of freedom of varying flow velocities as a function of hydrolysis residence times, sugar concentrations above 4.0 wt% should be achievable while still retaining greater than 86% yields of glucose.

Expectation (4) was essentially met. More than 92% of the fermentation performance expectation was achieved for ethanol production in just 24 h using an adapted recombinant Zymomonas mobilis. We showed that when the flashed hydrolyzate that had a temperature hold step is just overlimed, the adapted strain used 97.2% of the glucose (84.2% converted to ethanol) and 82% of the xylose (70.4% converted to ethanol) in 24 h. In addition, both the glucose and xylose were used simulta-neously in the adapted strain.

The use of hot water only (no added sulfuric acid) in the first stage of the hydrolysis process meets objective (5), lowering reactor capital cost requirements, by allowing for less-expensive stainless steel in at least the first stage of this process.

Achievement of these process perform-ance expectations indicates that an economically competitive process based upon the shrinking-bed–countercurrent hydrolysis approach is possible and that it justifies the follow-on work associated with designing, installing, and operating an engineering-scale reactor system based on this process concept.

Publications and Presentations:

1. Nagle, N., K. Ibsen, and E. Jennings. 1998. A process economic approach to develop a dilute acid cellulose hydrolysis process to produce ethanol from biomass. Presented at the Twentieth Symposium on Biotechnology for Fuels and Chemicals. Gatlinburg, TN. May 3–7 [1998].
2. Torget, R. 1998. A novel pilot scale reactor design for the aqueous fractionation of hardwood. Presented at the AIChE National Meeting. Miami Beach, FL. November [1998].

Summary Date: March 2000

Acid Recovery and Recycle by Solvent Extraction

Research Funded by: U.S. Department of Energy Office of Fuels Development through the National Renewable Energy Laboratory

Project Manager: Robert Wooley 303.384.6825, Robert_Wooley@nrel.gov

Performing Organization: Bioengineering Resources, Inc., 1650 Emmaus Rd. Fayetteville, AR 72701; National Renewable Energy Laboratory, 1617 Cole Boulevard, Golden, CO 80401-3393, http://www.nrel.gov

Principal Investigator:Carl Wikstrom, 501.521.2745, CarlWiks@aol.com; Richard Elander, 303.384.6841, richard_elander@nrel.gov

Contract Number: CRADA 95-042

Contract Period: 10/97–09/99

Contract Funding:

FY 1998: $3,300 (BRI), $4,200 (NREL)

FY 1999: $56,000 (BRI) $47,000 (NREL)

Objective Our objective is to determine the applicability of a proprietary solvent extraction system (developed by Bioengineering Resources, Inc. [BRI]) that removes sulfuric acid, acetic acid, and sugar degradation products from dilute-acid prehydrolyzates.

Approach/Background: NREL and BRI are collaborating in Phase 2 of a Small Business Technology Transfer Research Program (STTR) project to determine whether BRI's solvent extraction process can be applied to dilute acid prehydrolyzates. This process was originally designed to recover and recycle acid used in concentrated sulfuric- and hydrochloric-acid processes. However, work in Phase 1 of this project demonstrated that this approach might also work as a detoxification step for dilute acid prehydrolyzates.

Although it may not be necessary to recover and recycle sulfuric acid in a dilute-acid process, it is necessary to remove acetic acid and sugar degradation products from dilute-acid prehydrolyzates to allow efficient fermentation to produce ethanol. In Phase 2, NREL and BRI are working jointly on several problems: the toxicity of the solvent to fermenting microorganisms, potential acetic acid and furfural coproduct recovery, overall process flow diagrams, and preliminary process economic evaluations.

Status/Accoumplishments: BRI has further refined the solvent extraction process to remove high percentages of acetic acid, sulfuric acid, and furfural from dilute acid prehydrolyzate liquors at NREL. Early results indicated significant levels of sugar loss (glucose and xylose) from the raffinate stream. BRI has modified the solvent extraction process to reduce sugar losses to 3% or less.

NREL used its cofermenting recombinant Zymomonas mobilis strain in fermentation experiments on BRI-extracted prehydrolyzates to determine if the solvent treatment reduces the toxicity of these streams. Although chemical analysis of the treated prehydrolyzates has indicated near-complete removal of furfural, 5-hydroxymethyl furfural, acetic acid, and sulfuric acid, fermentative ethanol production was negligible. Residual levels of solvent may present in the treated prehydrolyzates that inhibit the ability of the recombinant Zymomonas mobilis to produce ethanol.

Earlier work on solvent-contacted pure sugar solutions showed that it is possible to achieve high yields of ethanol, but only with a significant lag phase that has been attributed to the presence of residual solvent. We believe that the extraction process could be further refined to significantly improve fermentation performance by leaving behind small amounts of the target removal products (acids and aldehydes) but reducing the carryover of solvent into the sugar-rich raffinate. Additional fermentation work by BRI using yeast (Saaccharomyces cerevisiae) has produced good fermentation of the glucose in solvent-treated prehydrolyzates, indicating a greater sensitivity of the recombinant Zymomonas mobilis to residual solvent levels.

The BRI solvent extraction process is currently undergoing a process economic evaluation to determine if it is competitive with other approaches used to condition or detoxify the prehydrolyzate.

Publications and Presentations: None

Summary Date: March 2000

FTIR Rapid Analysis Methods for Pretreatment Slurries

Research Funded by: U.S. Department of Energy Office of Fuels Development through the National Renewable Energy Laboratory

Project Manager: Robert Wooley 303.384.6825, Robert_Wooley@nrel.gov

Performing Organization: National Renewable Energy Laboratory, 1617 Cole Boulevard, Golden, CO 80401-3393, http://www.nrel.gov

Principal Investigator: Melvin Tucker, 303.384.6264, melvin_tucker@nrel.gov

Contract Number: N/A

Contract Period: 10/97–09/99

Contract Funding:

FY 1998: $0

FY 1999: $105,037

Objective Objectives of this project are to (1) develop a wet chemical method for measuring the solids composition of harshly pretreated first- and second-stage pretreated slurries; (2) develop a Fourier transform infrared (FTIR) method of measuring the compositions of solids and liquors from pretreated biomass slurries; and (3) develop a FTIR method for measuring sulfate concentrations in pretreated biomass slurries and liquors.

Approach/Background: Rapid and accurate analysis of chemical composition of solids and hydrolysate liquor is essential in real-time process control of dilute-acid pretreatment–hydrolysis processes. Dilute-acid pretreatment processes require precise control of residence time to obtain maximum sugar recovery. Traditional high-performance liquid chroma-tography (HPLC) methods, although very precise, are time-consuming and therefore not suitable for on-line process control. Rapid analytical methods can also be very cost effective for routine chemical analyses in research projects.

The use of mid-infrared (mid-IR) spectroscopy, in particular FTIR spectroscopy, for the qualitative and quantitative analysis of compounds has increased in importance during the past two decades as more commercial instruments have become available. FTIR analysis is a rapid and nondestructive technique for the qualitative and quantitative identification of compounds in solids and liquids (and some gases) in the mid-IR region. The application of FTIR for analyzing chemical composition of lignocellulosic feedstock and hydrolysates requires accurate and reproducible wet chemical data to calibrate the instrument.

In general, our current wet chemical and HPLC methods for determining chemical composition of lignocellulosic feedstock and acid hydrolysates are very accurate and reproducible. However, the methods are less reliable for determining the composition of residues obtained from harshly pretreated softwood forest thinnings. It was postulated that the high extractive content of the solid residues interferes with the lignin determination assay and thus results in erroneously high lignin values. If this is true, removal of the extractives is necessary before performing the conventional Klason lignin determination. A clean fractionation solvent method developed at NREL was evaluated for removal of extractives.

Insoluble solids and liquor samples from various biomass hydrolysates (softwood forest thinnings, mixed softwood sawdust, yellow poplar sawdust, and mixed solid waste) were analyzed using the improved wet chemical methods and HPLC. Mid-IR spectra were obtained on whole slurry, washed pretreated solids, and hydrolysate liquors by averaging 512 scans from 4000 cm^-1 to 400 cm^-1, at 2-cm^-1 resolution using a Nicolet Avatar 360^® spec-trometer and a six reflection diamond-composite ASI DurasamplIR^TM cell.

The wet chemical and HPLC results were used to develop statistical models to predict liquor and solids compositions from FTIR spectra of unknown samples. In particular, the FTIR models were used to rapidly predict the concentrations of sugars, acetic acid, furfural, and hydroxyl-methyl-furfural in liquors, and chemical composition of washed pretreated solids. Calibration methods for chemical compositions were determined using the partial least squares regression analysis option in TQ Analyst^TM (ver sion 1.2).

Status/Accoumplishments: The clean fractionation solvent method improved the mass balance closures for glucan and lignin to 100% in the milder pretreatments of softwoods, which are generally found in the single- and first-stage pretreatments. For more severe second-stage pretreatments of softwoods, the clean fraction-ation solvent method improved the lignin mass balance closure from about 125% to about 110%. The additional mass was shown to be insoluble sugar degradation products that coprecipitate or cross-link to the insoluble lignin. Extensive method development would be needed to differentiate insoluble lignin from an intractable insoluble or covalently linked sugar- degradation product.

Spectra from 35 wet-washed pretreated softwood samples were analyzed and gave the following values for correlation coefficients r and r², and for the standard error of estimates:

• 0.9768, 0.9541, and ± 1.4 wt% for glucose;
• 0.9721, 0.9450, and ± 0.1 wt% for mannose;
• 0.7506, 0.5634, and ± 0.1 wt% for galactose;
• 0.7562, 0.5718, and ± 0.4 wt% for xylose;
• 0.9176, 0.8420, and ± 2.3 wt% for lignin.

The moisture content of the washed solid residues in this calibration set varied from 50% to 80%. The large amount of moisture would have made obtaining mid-IR spectra very difficult because of the high background absorbance caused by water; however, the FTIR-attenuated total reflection (ATR) technique was able to overcome this difficulty and gave excellent correlation with the standard wet chemical and HPLC results. The capability of the method in analyzing wet solids is critically important for on-line process control because it eliminates the need for drying the samples, which could cause substantial delay in real-time analysis.

Liquor samples from 23 different pretreated softwood experiments were initially examined using the FTIR-ATR technique. The partial least squares regression analysis of the infrared and HPLC data results in correlation coefficients greater than 0.9 for all the dissolved compounds. The standard error of the estimate was found to be

• ± 5 g/L for glucose,
• ± 2.8 g/L for mannose,
• ± 0.6 g/L for galactose,
• ± 1.5 g/L for xylose,
• ± 0.4 g/L for acetic acid,
• ± 0.3 g/L for hydroxyl-methyl-furfural.

Increasing the number of calibration standards from 23 to 43 by including additional pretreated Quincy Library Group forest thinnings samples and pretreated Sealaska mixed sawdust samples somewhat decreased the accuracy of the partial least squares regression analysis. All of the correlation coefficients for each component in this larger set of calibration standards were greater than 0.9. However, the partial least squares method gave standard error of estimates of

• ± 7.8 g/L for glucose,
• ± 3.0 g/L for mannose,
• ± 0.9 g/L for galactose,
• ± 1.1 g/L for xylose,
• ± 0.7 g/L for arabinose,
• ± 0.6 g/L for acetic acid,
• ± 0.7 g/L for hydroxyl-methyl-furfural,
• ± 0.3 g/L for furfural.

It was possible to use the FTIR-ATR technique to distinguish between liquors from pretreated Quincy Library Group samples and liquors from pretreated Sealaska sawdust.

The FTIR-ATR technique was also successfully applied to monitor the neutralization and overliming of whole slurry from Sunds pretreated yellow poplar sawdust. The technique was able to follow the neutralization of acetic acid by lime in slurry of 16% insoluble solids and 19% total solids. Changes in phenolic infrared absorption bands, sugar absorption and formation of gypsum were also observed. This technique can be used to monitor for any potential loss of sugar during the overliming.

The accuracy of predicting the composition of pretreated liquors should improve as larger databases of FTIR spectra and compositional values are accumulated. However, methods developed using partial least squares for one feedstock must be redeveloped when feedstocks change or reactor conditions are markedly altered. Once that new method has been validated, then the FTIR-ATR technique can be used to monitor pretreatment. It may be possible to use an FTIR-ATR method to control a reactor because options are available for some of the current FTIR spectrometers that allow for 4-20 mA output control with decisions based on values from the partial least squares method. Diamond-composite probes are available with 9, 12, and 23 reflections to allow lower levels of detection of dissolved species.

In summary, several accurate and rapid FTIR-ATR methods were successfully demon-strated in analyzing the chemical composition of wet biomass solids, hydrolysate liquors, and monitoring overliming reactions. This was an important step towards demonstrating on-line pretreatment process monitor and control.

Publications and Presentations:

1. Tucker, M. July 21, 1999. FTIR Method Development. Biofuels Program P-Milestone Report. National Renewable Energy Laboratory. Golden, CO.
2. Tucker, M. September 7, 1999. FTIR method development: Develop validated FTIR method for predicting glucose, xylose, acetic acid, furfural and HMF in pretreated biomass liquors. Biofuels Program P-Milestone Report, National Renewable Energy Laboratory. Golden, CO.
3. Tucker, M. October 19, 1999. FTIR Method Development: Develop validated ftir method for predicting glucan, mannan, galactan, xylan, and lignin compositions in pretreated biomass solids. Biofuels Program P- Milestone Report. National Renewable Energy Laboratory. Golden, CO.
4. Tucker, M., R. Mitri, Q. Nguyen, and J. Webb. 2000 (in press). FTIR quantification of sugars in pretreated biomass liquors. Applied Biochemistry and Biotechnology, 84-86.

Summary Date: March 2000

Dilute-Acid Cellulose Hydrolysis Process Development and Conversion of Softwoods to Ethanol

Research Funded by: U.S. Department of Energy Office of Fuels Development through the National Renewable Energy Laboratory

Project Manager: Robert Wooley 303.384.6825, Robert_Wooley@nrel.gov

Performing Organization: National Renewable Energy Laboratory, 1617 Cole Boulevard, Golden, CO 80401-3393, http://www.nrel.gov

Principal Investigator: Quang Nguyen, 303.384.6868, quang_nguyen@nrel.gov

Contract Number: N/A

Contract Period: 10/97–09/99

Contract Funding:

FY 1998: $0

FY 1999: $1,091,598

Objective In FY 1998, our objective was to develop a conceptual design of a dilute acid hydrolysis process for conversion of softwood forest thinnings to ethanol. The target total hemicellulosic sugar and glucose from cellulose conversion yields are 75% and 50%, respectively.

In FY 1999, our objective was to develop a ±30% capital and operating cost estimate for a softwood-to-ethanol facility to be located at selected sites in California and Alaska. These design packages were to be made available to Sealaska and Wheelabrators for evaluation of options for converting wood residues to value-added products.

We also provided research and process development support to industrial and Cooperative Research and Development Agreement partners.

Approach/Background: In FY 1998, our strategy to meet the showstopper performance goals focused on the following: (1) We employed a two-stage dilute sulfuric acid hydrolysis process to maximize sugar recovery yield. The first-stage processing conditions would be at low temperature to minimize degradation of heat-labile hemicellulosic sugars. The second-stage processing conditions would be more severe to hydrolyze the cellulose. (2) We adapted a Saccharomyces cerevisiae strain to the hydrolysate to obtain high ethanol yield.

In FY 1999, after surpassing the show-stopper performance targets in FY 1998, we entered the process development stage focused on improving sugar yields and process efficiencies, and on developing the experimental data required for material and energy calculations. After reviewing the literature and our original conceptual design of a two-stage dilute acid hydrolysis process, we identified several areas that need clarification or improvement. These include

• Design and cost estimates of the hydrolysis reactors. The previous cost estimates for the hydrolyzers from Sunds Defribator, Inc. were very high.
• Solid-liquid separation equipment for both hydrolysis stages. The second-stage hydrolysate is particularly difficult to handle because it contains fine particles. Whether current commercial solid-liquid separation equipment can process the second-stage hydrolysate efficiently is uncertain.
• There were suggestions that perhaps a single-stage dilute acid hydrolysis would be more cost effective than a two-stage process.

Status/Accoumplishments: In FY 1998, we surpassed our show-stopper performance targets in FY 1998 by achieving total sugar recovery yield of 89% for mannose, 82% galactose, and 50% glucose, and ethanol yield as high as 95% of theoretical. Additionally, a Saccharomyces cerevisiae yeast strain was successfully adapted to the inhibitors produced during the acid hydrolysis of mixed softwood forest thinnings.

In FY 1999, to provide solutions to the technical issues described in the background section, we completed the following activities:

• Performed a set of preliminary optimization experiments for single- and two-stage dilute sulfuric acid hydrolysis to provide data for comparing sugar yields of the two processes, and to provide data for designing the two-stage process.
• Developed design specifications for vertical hydrolyzers to be used in both stages, and identified vendors and manufacturers who could fabricate the equipment and provide cost information. We passed this information to Merrick & Company (Aurora, CO) for cost evaluation. These vertical hydrolyzers were used in both the Merrick design for the Martell, CA, plant and the NREL ASPEN model.
• Developed a new process block-flow diagram, which eliminates the second-stage solid-liquid separation step. This simplified process eliminates not only the uncertainty of this unit operation but also reduces the total capital cost. Furthermore, it reduces the amount of water added to fermentors; thus, it increases the ethanol concentration in the beer. This concept splits the fermentation train into two stages. The first-stage hydrolysate liquor is fermented in the first-stage fermentors, and the second-stage hydrolysate slurry is combined with the partially fermented first-stage liquid. This arrangement also allows for recycling the yeast and thus adaptation to inhibitors.
• Performed a set of fermentation experiments and confirmed that this process is feasible.

After the new conceptual process was confirmed in the laboratory, process flow diagrams were developed. Both the NREL and the Merrick models were based on the same process flow diagrams for the stand-alone plant design.

An ASPEN Plus simulation model was developed based on the process flow diagrams. The results from this model were then used in an economic spreadsheet to calculate material and energy balances, and equipment sizes and costs. Once the equipment was costed, a discounted cash-flow analysis was performed, based on a set of assumptions, to produce an overall cost of producing ethanol by the process. This cost per gallon figure can then be used to compare the process against other biomass-to-ethanol technologies or against other fuel technology costs. Other important calculations are annual ethanol production rate, total capital investment, ethanol production yield, and equipment capital costs.

With limited optimization work to date, we have achieved hemicellulosic sugar yield as high as 89% and glucose yield as high as 59% from California mixed softwood forest thinnings (White fir and Ponderosa pine). A Sacchar-omyces cerevisiae yeast strain was successfully adapted to first-stage hydrolysate liquor at total solids concentrations as high as 24% (solubles plus insoluble solids). Using the adapted strain, ethanol yields of 85%–90% theoretical were obtained from hexose sugars in the combined first- and second-stage hydrolysates without overliming requirements.

Based on the NREL process design and ASPEN Plus simulation model for softwood-to-ethanol plants using two-stage dilute sulfuric acid hydrolysis technology, Merrick & Com-pany developed a preliminary design for an 800 BDT/d plant for the Quincy Library Group area. Two scenarios were investigated: a stand-alone plant and plant colocated with a biomass power plant in Martell, CA.

The study includes process design, heat and material balance, process flow diagrams, equipment selection, capital and operating cost estimates, and ethanol market assessments. The process design is essentially identical to the NREL simulation model except that the throughput is matched to feedstock available in the area. The co-located plant study at Martell identifies specific modifications required to fit the infrastructure and operation. Based on this evaluation, a co-located softwood-to-ethanol plant in the Martell area is an economically attractive concept. Assuming an ethanol selling price of $1.20/gal, the estimated internal rate of return on investment is about 36% on 25% equity.

A set of preliminary dilute-acid hydrolysis experiments was completed using Sealaska feedstock. Hemicellulose yield as high as 94% was obtained. However, glucose yield was only 48%. Further optimization of second-stage hydrolysis is required to improve the glucose yield.

Based on NREL experimental data and process design, Merrick engineers also com-pleted a preliminary engineering design of an ethanol plant for Sealaska. The plant is to be located at a site of a shutdown pulp mill in Ketchikan, AK.

Publications and Presentations:

1. Merrick & Company. June 14, 1999. Softwood biomass to ethanol feasibility study, final report. NREL subcontract No. AXE-8-18020-01. National Renewable Energy Laboratory, Golden, CO.
2. Nguyen, Q. and A. Aden. Nov 8, 1999. Preliminary design of an ethanol plant based on two-stage dilute acid hydrolysis technology. Biofuels Program P-milestone report. National Renewable Energy Laboratory, Golden, CO.
3. Nguyen, Q., M. Tucker, F. Keller, and F.P. Eddy. 2000 (in press). Two-stage dilute acid pretreatment of softwoods. Applied Biochem-istry and Biotechnology, 84–86.
4. Tucker, M., R. Mitri, Q. Nguyen, and J. Webb. 2000 (in press). FTIR quantification of sugars in pretreated biomass liquors. Applied Biochem-istry and Biotechnology, 84–86.
5. Nguyen, Q., M. Tucker, F. Keller, D. Beaty, K. Connors, and F.P. Eddy. 1999. Dilute acid hydrolysis of softwoods. Applied Biochemistry and Biotechnology, 77–79:133–142.
6. Nguyen, Q., M. Tucker, B. Boynton, F. Keller, and D. Schell. 1998. Dilute acid pretreatment of softwoods. Applied Biochemistry and Biotechnology, 70–72:77–87.
7. Keller, F., D. Bates, R. Ruiz, and Q. Nguyen, 1998. Yeast adaptation on softwood prehydrol-ysate. Applied Biochemistry and Biotechnology, 70–72:137–148.

Summary Date: March 2000

Softwood Extractives Characterization

Research Funded by: U.S. Department of Energy Office of Fuels Development through the National Renewable Energy Laboratory

Project Manager: Robert Wooley 303.384.6825, Robert_Wooley@nrel.gov

Performing Organization: Kemestrie Inc., 4245, rue Garlock Sherbrooke, Quebec J1L 2C8, Canada , http://www.enerkem.com

Principal Investigator: Jean Michel Garro, 819.569.4888, kem@interlinx.qc.ca

Contract Number: ACO-8-18037-01

Contract Period: 05/98–10/98

Contract Funding:

FY 1998: $62,047

Objective The overall objectives of the project were to investigate methods for recovering the extractives from softwood forest thinnings before pretreatment, to characterize the recovered extractives, and to determine the effect of extractive removal on steam pretreatment.

Approach/Background: Preliminary results of dilute acid pretreatment and yeast fermentation of mixed softwood forest thinnings at NREL indicate that the high extractive content in the feedstock has several negative effects on the process. The extractives tend to cause deposits on pretreatment and fermentation systems and are toxic to fermentative organisms. One way to reduce these effects is to remove the extractives before pretreatment. An added advantage to this strategy is the potential recovery of valuable products from the extractives.

Using previous knowledge in extractive removal method and equipment already available at its laboratory, the Subcontractor evaluated two methods for recovering the extractives from softwood forest thinnings. These methods are as follows: Method 1: Presteaming followed by hot dilute acid soaking (variables are 50^o–80^oC temperature and 1–3 h soaking time); Method 2: Presteaming followed by hot dilute acid and ethanol soaking (variables are 50^o–80^oC temperature and 5–20 wt% ethanol concentration). The set of extraction conditions that give the highest extractives removal and minimal degradation of carbohydrates will be referred to as standard conditions. Samples produced under these conditions are designated as standard samples: the liquid samples designated as standard extracts and the solid samples as standard extracted chips.

The recovered extractives were characterized and evaluated for their potential applications.

The standard extracted chips were pretreated using a steam explosion reactor. The pretreated samples were analyzed for solubilized sugar yields and total solids recovery.

Status/Accoumplishments: Highest removal of extractives and minimal degradation of carbohydrates present in softwood forest thinnings were obtained by impregnation followed by extraction with 50% v/v ethanol at 80^oC.

Bench-scale extraction with 50% v/v ethanol gave 6.1% extractives yield based on the difference between initial and final dry weight of ground biomass. Pilot-scale extractives yield was 3.5% based on dry extractive powder obtained after concentration and lyophilization of the extract.

The major component of the extractive powder was identified as lignin (85%) but its value seems overestimated by the analytical method used (Klason Lignin). Further work is necessary to determine the purity of the lignin obtained.

It is interesting to note that the extractives contain 7.8 wt% proanthocyanidins Assuming a conservative yield of 30% and a market price of $333/kg for purified proanthocyanidins, its potential value is $270/ dry metric ton of forest thinnings.

Soluble sugar recovery yield following dilute acid steam pretreatment of the ethanol-extracted material is similar to that obtained from unextracted wood.

Publications and Presentations:

1. Kemestrie Inc. Dec 15, 1998. Softwood extractives: recovery and characterization, final report. NREL subcontract no. ACO-8-18037-01. National Renewable Energy Laboratory, Golden, CO.

Summary Date: March 2000