HYDROLYSIS OF CELLULOSE AND HEMICELLULOSE

YDROLYSIS OF CELLULOSE AND EMICELLULOSE Charles E. Wyman 1, Stephen R. Decker 2, John W. Brady 3, Liisa Viikari 4, and Michael E. immel 2 1. Thayer School of Engineering, Dartmouth College, anover, New ampshire 03755 2. National Renewable Energy Laboratory, Golden, Colorado, 80401 3. Department of Food Science, Cornell University, Ithaca, New York, 14853 4. VTT Technical Research Centre of Finland, FIN-02044, VTT 1

1 INTRODUCTION 5 2 OVERVIEW OF TE RELEVANT STRUCTURAL FEATURES OF CELLULOSIC MATERIALS 7 2.1 Cellulose 7 2.1.1 Chemical Composition 8 2.1.2 Structure and Morphology of Cellulose Fibers 8 2.2 emicelluloses 14 2.2.1 Structural Functionality of emicelluloses 15 2.2.2 Chemical Composition and Sources of emicelluloses Subtypes 17 3 EMICELLULOSE AND CELLULOSE YDROLYSIS FOR BIOMASS CONVERSION 23 3.1 Overview Of Conversion Chemistry 23 3.2 Typical Process Steps 24 3.3 Cellulose and hemicellulose hydrolysis reactions 26 4 ACID YDROLYSIS OF CELLULOSE 28 4.1 System Description and Performance 28 4.2 Kinetic Models 30 5 ACID YDROLYSIS OF EMICELLULOSE 35 5.1 System Description and Performance 35 5.2 Kinetic Models 36 6 CELLULASE SYSTEMS FOR CELLULASE YDROLYSIS 41 6.1 Sources of Cellulase Enzymes 41 6.1.1 Fungal Enzyme Systems 41 6.1.2 Bacterial Enzyme Systems 42 6.2 General Classification of Cellulase Enzymes 44 2

6.2.1 Endoglucanases 44 6.2.2 Exoglucanases 45 6.2.3 Beta-β-Glucosidase 45 6.3 Cellulase Structure and Function 45 6.3.1 Synergism 45 6.3.2 Glycosyl ydrolase Diversity 47 6.3.3 Enzyme Adsorption and Non Productive Binding 48 6.3.4 Processive Cellulases 49 7 ENZYMATIC YDROLYSIS OF CELLULOSE 50 7.1 Experimental Systems 50 7.2 Structural Features Impacting Cellulase Action 53 7.2.1 Crystallinity and Boundary Water Layer 53 7.2.2 emicellulose Content 54 7.2.3 Lignin Content 54 7.2.4 Acetyl Content 58 7.3 Kinetic Modelling of enzymatic hydrolysis of cellulose 58 8 PRETREATMENT TO IMPROVE CELLULOSE YDROLYSIS BY ENZYMES 60 8.1 Effect of Pretreatment on Enzymatic ydrolysis 60 8.2 Desirable Pretreatment Attributes 62 8.3 Pretreatment Types 63 8.4 Chemical Pretreatment at Low to Neutral p 64 8.5 Chemical Pretreatment at igh p 66 9 ENZYMATIC YDROLYSIS OF EMICELLULOSE 68 9.1 Sources of emicellulases 68 3

9.2 General emicellulose ydrolysis 69 9.2.1 Xylan Backbone ydrolysis 70 9.2.2 Mannan ydrolysis 72 9.2.3 Enzymatic glucan hydrolysis 72 9.2.4 Enzymatic hydrolysis of hemicellulose oligomers 74 9.2.5 Enzymatic debranching of hemicellulose 74 9.3 Action of emicellulases on Model Substrates 75 9.4 ydrolysis of Residual emicellulose in the Solid Residue 77 9.4.1 ydrolysis of Solublized emicellulose 78 10 ECONOMICS OF SUGAR PRODUCTION FROM CELLULOSIC BIOMASS 79 10.1 Cellulosic Biomass as a Feedstock 79 10.2 Economics of ydrolysis Based Technologies 80 10.3 Application to Other Products 84 10.4 Opportunities for Cost Reductions 85 11 BENEFITS AND IMPACTS OF CELLULOSIC BIOMASS CONVERSION 87 11.1 Greenhouse Gas Reductions 87 11.2 Strategic Benefits 89 11.3 Solid Waste Disposal 90 11.4 Economic Benefits 91 11.5 International Fuels Market 91 11.6 Sustainable Production of Organic Fuels and Chemicals 92 12 ACKNOWLEDGMENTS 92 13 REFERENCES 92 4

1 INTRODUCTION Cellulosic biomass includes agricultural (e.g., corn stover and sugarcane bagasse) and forestry (e.g., sawdust, thinnings, and mill wastes) residues, portions of municipal solid waste (e.g., waste paper), and herbaceous (e.g., switchgrass) and woody (e.g., poplar trees) crops. Such materials are abundant and competitive in price with petroleum, and cellulosic biomass can provide a sustainable resource that is truly unique for making organic products. Furthermore, cellulosic biomass can be produced in many regions of the world that do not have much petroleum, opening up a new route to manufacturing organic fuels and chemicals. The structural portion of cellulosic biomass is a composite of cellulose chains joined together by hydrogen bonding as long cellulose fibers that are in turn held together with hemicellulose and lignin, allowing growth to large aerial plants that can withstand the weather and resist attack by organisms and insects. Traditionally, humans have employed the strength offered by some forms of cellulosic biomass for structural purposes such as construction of homes, furniture, and baseball bats. Other types of biomass such as grasses can stabilize the soil and provide protein for feeding of animals. In addition, chemicals can react with biomass at high temperatures to remove a portion of the hemicellulose and/or lignin to enhance the properties of the remaining solids for manufacture of paper, particleboard, and other products. owever, the cellulose and hemicellulose portions of biomass, typically representing about 40 to 50% and 20 to 30% of plants, respectively, are polysaccharides that can be broken down into sugars through hydrolysis by enzymes or acids, and these sugars can be fermented or chemically altered to valuable fuels and chemicals. Over the years, such hydrolysis reactions have been employed to release sugars for making ethanol during periods of war by approaches such as the Scholler process during the 1940s and for making various products in controlled economies such as the former Soviet Union. ydrolysis of the polysaccharides in 5

cellulosic biomass for production of fuels and chemicals offers important strategic, environmental, and economic advantages, and although the cost has been too historically too high compared to fossil alternatives, research over the last 2 decades has advanced the technology to the point that it is becoming economically viable. The challenge is to overcome the risk of commercializing first-of-a-kind applications and to continue to advance hydrolysis technologies to become even lower in cost so that fuels and chemicals from cellulosic resources are competitive without subsidies. This chapter presents a comprehensive overview of cellulose and hemicellulose hydrolysis. It begins with a short description of important structural features of cellulosic materials to provide a context on the demands of hydrolyzing these abundant but recalcitrant polysaccharides. The next section focuses on applications, process steps, and stoichiometry for hydrolysis reactions. A discussion of acid hydrolysis approaches and kinetics is then provided followed by a similar treatment for hemicellulose hydrolysis to give insight into opportunities and limitations for acid based processes. At this point, the chapter shifts to enzymatic hydrolysis starting with a short description of different sources of cellulase enzymes for breaking down cellulose to glucose. This section also includes information on the different types of enzyme activities involved in converting cellulose to glucose and features of these components that influence cellulase action. The chapter then examines the structural characteristics of biomass that influence cellulose hydrolysis by enzymes, types of cellulose hydrolysis processes, and experimental results for systems for enzymatic conversion of cellulose, and summarizes some of the factors influencing hydrolysis kinetics. Because cellulosic biomass must be pretreated to achieve high yields of glucose from cellulose by enzymes, an overview is provided of leading options for preparing biomass for enzymatic processing. 6

Many promising pretreatment approaches do not fully hydrolyze hemicellulose to sugar monomers, and use of enzymes provides an attractive route to making them fermentable. emicellulose hydrolysis is also important in many other applications. Therefore, sources of such enzymes, how they act, features impacting their action, experimental results with hemicellulases, and modeling approaches to predict their performance are described. The economics of hydrolysis-based processes are then summarized including a description of advances that have reduced costs to be competitive now, identification of opportunities to reduce costs further, and synergies for production of multiple products from cellulosic materials. The chapter concludes with a brief description of important benefits of applying hydrolysis of biomass to generate sugars for conversion to ethanol and other products. Because covering such a wide spectrum of hydrolysis aspects in a single chapter necessitates being succinct, numerous references are included to help the reader pursue topics in more depth, if desired. 2 OVERVIEW OF TE RELEVANT STRUCTURAL FEATURES OF CELLULOSIC MATERIALS 2.1 Cellulose As a regular, linear homopolymer, cellulose might reasonably be thought to be structurally uncomplicated. This perception would not be entirely correct however, since cellulose chains are organized together into progressively more complex assemblies at increasing size scales. As a result of this emergent structural complexity and heterogeneity, the structure of cellulose has been remarkably difficult to unravel for such a nominally simple substance, and it is only relatively recently that a consensus picture of its organization has begun to emerged. Several 7

recent reviews have surveyed cellulose structures, but this question is still an active topic of study (1-3). 2.1.1 Chemical Composition The chemical structure of cellulose, which is a linear polymer of β-(1 4)-linked D- glucopyranose monomer units, is in fact quite simple. Typically cellulose chains in primary plant cell walls have DPs (degree of polymerization) in the range from 5,000-7,500 glucose monomer units, with the DP of cellulose from wood being around 10,000 and around 15,000 for cellulose from cotton (3). The basic repeating unit of cellulose is cellobiose (Figure 1), the β-(1 4)-linked disaccharide of D-glucopyranose. At ambient temperatures, the relatively rigid glucose rings are all found in their lowest-energy, 4 C 1 puckered chair conformation (see Figure 1), and do not easily make transitions to the other chair conformer or to the various possible twist-boat forms. With the rings in this conformation, all of the hydrogen-bonding hydroxyl and hydroxymethyl substituents of the pyranose rings are equatorial, directed around the periphery of the ring, while all of the hydrophobic aliphatic protons are in axial positions, pointing either up or down relative to the average plane of the rings. This topological feature of cellobiose can be seen readily in Figure 1. 2.1.2 Structure and Morphology of Cellulose Fibers Bulk cellulose has a substantial degree of crystallinity, and its structure has long been the subject of intense study. A crystalline structure for cellulose was first described by Mark and Meyer in 1928 (4). Unfortunately, however, disorder and polydispersity in chain lengths prevent the formation of single crystals, and x-ray diffraction studies of the crystal structure of cellulose have been limited to fiber diffraction experiments. For this reason, the details of the crystal structure of native cellulose are still a matter of debate more than 70 years later. 8

owever, combined with modeling calculations, such experiments can be used to deduce plausible conformations consistent with the limited data. All available evidence indicates that crystalline cellulose chains are in an extended, flat 2-fold helical conformation (described in detail below), but small variations in this conformation or in the packing of the cellulose chains within the crystal give rise to a number of crystalline polymorphs, many of which can be interconverted by various processing treatments (3, 5). Under normal conditions, cellulose is extremely insoluble in water, which is of course necessary for it to function properly as the structural framework in plant cell walls. Seven crystal polymorphs have been identified for cellulose, which are designated as Iα, Iβ, II, III I, III II, IV I, and IV II, as indicated in the following diagram 1.1 (3). (1.1) These polymorphs differ in physical and chemical characteristics such as solubility, density, melting point, crystal shape, and optical and electrical properties (2, 4). In nature, cellulose Iα and Iβ are the most abundant crystal forms and hence are referred to as native cellulose. Initially, X-ray fiber diffraction experiments of different native cellulose sources led to two models of native cellulose differing in the number and orientation of glucose units in the unit cell (6). Later experiments by Atalla and Vanderart in 1984 using 1 -NMR indicated that native cellulose contained two allomorphs which were designated Iα and Iβ (7). This picture has been confirmed by Wada and coworkers using electron diffraction experiments (8). 9

Cellulose I consists of chains arranged in a parallel fashion such that the (1 4) glycosidic bonds point in the same direction along the microfibril. The nature of cellulose biosynthesis requires this parallel packing. Although cellulose I is the enzymatically-synthesized crystal form for this polymer, it is not the lowest-energy form, and the individual fibers are kinetically trapped in this arrangement by the crystal lattice after synthesis. Cellulose Iα is the major allomorph produced by bacterial and fungal sources and is a triclinic P 1 crystal with one cellobiose residue per unit cell (9). The cellulose chains are oriented in a parallel manner as would be expected for a unit cell with one chain. Cellulose Iα is converted to the more stable Iβ form through an annealing process at 270 C in various media (3). Cellulose Iβ is the major crystal form in higher plant species and is monoclinic in nature with two cellobiose moieties per unit cell. Cellulose Iα and cellulose Iβ are found within the same microfibril and hence parallel packing of the cellulose chains in cellulose Iβ is the logical conclusion and is consistent with X-ray diffraction data, analysis with silver staining and cellobiohydrolase digestion (10-12). There are two possible ways to pack the chains in a twoparallel-chain unit cell arrangement: parallel up and parallel down. The two chains are referred to as the corner chain and the origin chain. Differences in the definitions of the unit cell dimensions have lead to confusion as to the precise packing for cellulose Iβ. Current and proper convention, as proposed by Sarko and Muggli (6), designates the chain which lies parallel to the c axis direction as the origin chain and the chain which passes through the center of the a/b plane, and translated in the c axes direction by a/4 with respect to the origin chain, as the center chain (11). The result of this designation denotes the chains of a cellulose Iβ crystal as packing in a parallel up fashion. Cellulose II is the major polymorph in industrially processed cellulose. Cellulose II can be formed upon regeneration or mercerization of cellulose I and is also the most 10

thermodynamically stable allomorph. The nature of the cellulose II structure remains to be definitely agreed upon. For example, arguments continue as to whether cellulose II consists of parallel or antiparallel chains. Nishimura and Sarko proposed that cellulose II is formed after crystalline cellulose I is solvated and the amorphous regions re-crystallize into antiparallel microfibrils (13). Simon and coworkers proposed that the increase in distance between chains in the crystal due to swelling is sufficient for the chain to fold at both ends and then glide along one another to produce the antiparallel arrangement (14). Others claim cellulose II chains align antiparallel after solvation due to chain exchange between microfibrils. Kroon-Batenburg and coworkers have suggested that the difference between cellulose I and cellulose II is simply the rotation of the primary alcohol from the TG conformer 1 to the GT conformer, and that cellulose II is in fact parallel (5). These authors argue that the unit cell dimensions of GT parallel cellulose II are the same as those for antiparallel TG cellulose II, and that it thus is hard to distinguish the packing mode of cellulose II. (It is interesting to note that neither of these primary alcohol conformers is the lowest-energy form for free glucose; both NMR and modeling studies find that the GG conformer has the lowest free energy for the isolated monomer (16-20). Using molecular dynamics techniques, Marhofer and coworkers calculated Young s modulus for cellulose II with chains in both the parallel and antiparallel arrangements (21). Only the parallel arrangement gave good agreement with experimental results, and they too concluded that cellulose II chains have a parallel arrangement. Most recently Langan and coworkers produced a cellulose II structure from neutron fiber diffraction analysis (22). The neutron diffraction data extended to a 1.2 Å 1 tg conformation refers to the rotamer of the primary alcohol group, where the first t (trans) or g (gauche) refers to the orientation of the C4-C5-C6-O6 torsion with respect to O5 and the second letter designates the orientation with respect to C4. 15. R.. Marchessault, and S. Peréz, Conformations of the ydroxymethyl Group in Crystalline Aldohexopyranoses, Biopolymers, 18:2369 (1979). 11

resolution from two highly crystalline fiber samples of mercerized flax, and these workers constructed two models of cellulose II in an anti-parallel arrangement with the primary alcohols in the GT conformation for the origin chain and in the TG conformation for center chain. It is important to note that fiber diffraction studies cannot directly determine the primary alcohol rotamer conformation, which must be deduced from models consistent with the limited diffraction data. Apparently no workers have attempted to fit cellulose II fiber diffraction data using a GG model for either parallel or antiparallel chains, and the question of the structure of cellulose II must be regarded as still unresolved. The problem of describing the overall conformation of polysaccharide chains can be reduced to the problem of specifying the rotational torsion angles φ and ψ around each successive glycosidic linkage, in much the way that the conformation of the backbone of a protein can be specified by listing the peptide Ramachandran torsional angles (23). For single crystals of cellobiose the values of these angles are (42, -18 ) (24). Because the glycosidic linkage is of the equatorial-equatorial type (since the O4 group of glucose is equatorial, as is the O1 group in the β anomer), a broad range of values of φ and ψ are possible which involve no clashes between two linked rings. For cellulose chains, the linkage torsion angles are approximately (28, -30 ), with small variations in different polymorphs but with all cellulose structures having essentially the same extended flat ribbon-like conformation in crystals. Figure 2 illustrates the conformation of the cellobiose repeat unit in cellulose, and Figure 3 shows the conformation of a short hexaose oligomer in this conformation. There are several very important qualitative features of the crystal conformation of the individual cellulose chains that should be noted. As can be seen from Figure 3, in the very extended crystalline conformation each glucose unit is flipped by 180 with respect to the previous and subsequent rings, so that the exocyclic primary alcohol groups alternately point to the right and left of the chain direction. This feature is important for the binding grooves 12

and pockets of cellulase enzymes that must accommodate cellulose chains. Secondly, the chain is stabilized by strong hydrogen bonds along the direction of the chain, from the exocyclic O6 primary alcohol hydroxyl group to the O2 secondary alcohol hydroxyl group of the subsequent residue, and from the O3 hydroxyl group to the O5 ring oxygen of the next sugar residue. These hydrogen bonds help to maintain and reinforce the flat, linear conformation of the chain beyond the rigidity that might be expected for this type of linkage. Importantly, in most models for crystalline cellulose, there are no hydrogen bonds between chains in different crystal layers. Finally, with all of the aliphatic hydrogen atoms in axial positions and all of the polar hydroxyl groups in equatorial positions, the top and bottoms of the cellulose chains are essentially completely hydrophobic, while the sides of the chains are hydrophilic and capable of hydrogen bonding. This topology is extremely important for the packing of chains into crystals. In all of the proposed crystal packing schemes, the chains are stacked with a pairing of hydrophobic faces, and these hydrophobic regions must make some contribution to the insolubility of cellulose under normal conditions. Furthermore, binding sites for cellulose segments in proteins must contain hydrophobic surfaces such as tryptophan or phenylalanine side chains to pair up with these non-polar faces. In native cellulose found in plant sources, the cellulose chains have extended segments with the regular, repeating crystal conformation, and a number of independent chains are packed together into bundles or microfibrils which typically consist of from around 30 to 200 independent chains. In these microfibrils, long sequences of the individual chains are found in the extended conformation and similar segments from adjacent chains pack together into highly regular microcrystalline regions. The microfibrils are not entirely crystalline, however, even in so-called bacterial microcrystalline cellulose samples, as the regular segments of the individual chain conformations alternate with segments which are irregular in conformation. This alternation of regular microcrystalline regions with more flexible disordered regions is 13

almost certainly the result of evolutionary selection, since in order to serve effectively as the principal structural component of plant cell walls, both great strength and flexibility are needed. 2.2 emicelluloses emicelluloses are generally classified according to the main sugar residue in the backbone, e.g. xylans, mannans, and glucans, with xylans and mannans being the most prevalent. Depending on the plant species, developmental stage, and tissue type, various subclasses of hemicellulose may be found, including glucuronoxylans, arabinoxylans, linear mannans, glucomannans, galactomannans, galactoglucomannans, β-glucans, and xyloglucans. These different subtypes can be grouped into two general categories based on the hydration of the fibers. Low hydration polysaccharides include the arabinoxylans, glucuronoxylans, xyloglucans, and linear mannans. With the exception of the linear mannans, which serve mainly as a seed storage compound, this class of hemicellulose functions primarily to stabilize the cell wall through hydrogen-bonding interactions with cellulose and covalent interaction with lignin. Again, with the exception of the linear mannans, these compounds are water soluble in their native state, generally due to their branched construction. Standard extraction protocols using alkali de-esterifies them of some of their sidechains, disrupting the entropy of the water of solublization enough to cause the partially debranched chains to aggregate through interchain hydrogen bonding and fall out of solution. Other hemicelluloses, such as galactoglucomannans, glucomannans, galactomannans, and β-glucans comprise a subset of highly hydrated hemicelluloses. This class, comprised mainly of hydrocolloids, is used primarily as an extracellular energy and raw materials storage system and as a water retention mechanism in seeds. As they have fewer, if any, ester-linked sidechains, alkaline extraction does not usually render them insoluble and they tend to be heavily hydrated. 14

2.2.1 Structural Functionality of emicelluloses The majority of the polysaccharides found in plant cell walls belong to the cellulose, hemicellulose, or pectin classes. Cellulose, the dominant structural polysaccharide of plant cell walls, is a linear β-(1 4)-D-glucopyranoside polymer. Although cellulose functions as the rigid, load-bearing component of the cell wall, the rigidity of the cellulose microfibril is strengthened within a matrix of hemicelluloses. Also referred to as cross-linking glycans, the hemicelluloses are believed to be involved in regulation of wall elongation and modification Unlike lignin, the hemicelluloses are thought to have a strong effect on the interactions between cellulose microfibrils and to heavily influence the strength and rigidity of the crosslinking matrix. This matrix is further complexed and reinforced with the pectins. Pectins are non-cellulosic acidic cell wall polysaccharides. The pectins vary widely and are divided into three classes, homogalacturonan, rhamnogalacturonan I, and rhamnogalacturonan II. Pectins function as a sol-like matrix, providing water and ion retention, support and facilitation of cell wall modifying enzymes, cell wall porosity, cell-to-cell adhesion, cell expansion, cell signalling, developmental regulation, and defense. emicelluloses are loosely defined as non-cellulose, non-pectin cell wall heteropolysaccharides consisting of several polymers, varying in composition of monosaccharides and glycosidic linkages, substitution pattern and degree of polymerization. Their chemical composition and structural features vary widely across species, sub-cellular location, and developmental stages. They have often been reported as chemically associated or cross-linked to other polysaccharides, proteins or lignin. Studies of bacterial cellulose produced by Acetobacter xylinum in the presence of various hemicellulose fibers have shown that these hemicelluloses readily become complexed into the interior and along the surface of the cellulose microfibril (25-27). The results of this interaction are dependent upon the type 15

of hemicellulose involved. These results have been supported by FT-IR work on intact wood fibre (28). Probably no chemical bonds exist between cellulose and hemicellulose, but mutual adhesion is provided by hydrogen bonds and van der Waals forces. Evidence has been reported for linkages between hemicellulose and lignin in wood and other plant materials. The side groups of xylan may have an important role in the bonding of lignin to hemicellulose. Both ester linkages between lignin and methylglucuronic acid residues and ether bonds from lignin to arabinosyl groups have been reported (29). Low molecular weight phenolic components, such as ferulic acid and p-coumaric acid, are covalently bound to plant cell wall polysaccharides of annual plants (30). Side groups, especially the acetyl substituents, affect the physicochemical properties and biodegradability of soluble or matrix-bound hemicelluloses. Acetylation has been shown to increase the solubility of polysaccharides in water, probably by hindering the aggregation of molecules, and thus making the substrate more accessible to enzymes (31, 32). On the other hand, the presence of ester-linked non-carbohydrate residues impedes the ability of individual glycanases to liberate uniform degradation products and therefore decreases the enzymatic degradability of polysaccharides. Chemical de-acetylation of aspen wood and wheat straw has been shown to increase their enzymatic degradation. Xylan became 5 7 fold more digestible which in turn improved significantly the accessibility of cellulose to enzymatic hydrolysis (33). The effect of acetyl side groups on the enzymatic hydrolysis of hemicelluloses was first demonstrated by Biely and co-workers (34). 16

2.2.2 Chemical Composition and Sources of emicelluloses Subtypes 2.2.2.1 Xylans The term xylan is a catchall for polysaccharides that have a β-(1 4)-D-xylopyranose backbone with a variety of sidechains. Xylan is the predominant hemicellulose in most plant cell walls, generally comprising about 1/3 of the total plant biomass (35). Xylan is found primarily in the secondary cell wall and seems to accumulate on the surface of the cellulose microfibril as the cellulose synthase complex extrudes the microfibril (36). This deposition continues as the additional cell wall material is laid down, resulting in heavier xylan deposition on the cellulose microfibrils in the outermost layers of the secondary cell wall. Xylans function primarily by forming cross-links between the other cell wall components, such as cellulose, lignin, other hemicelluloses, and pectin. This interaction is carried out by hydrogen bonding to the other polysaccharides and by covalent linkages through the arabinofuranosyl sidechains to the ferulic and coumaric acids found in lignin. The composition and linkages of the sidechains determines the specific variety of xylan. The sidechains, usually attached at the O2 and/or O3 positions, include glucuronic acid, 4-Omethylglucuronic acid, L-arabinofuranose, xylose, and acetyl groups. Removal of these side chains generally enhances the rate of degradation by endoxylanase enzymes (37). The types and levels of sidechains are dependent on the particular plant. Due to these differences in sidechains, xylans from grasses and annuals having an elevated level of α-larabinofuranoside substituents are referred to as arabinoxylans, while hardwood xylans, highly substituted with 4-O-methyl glucuronic acid, are termed glucuronoxylans. ardwood glucuronoxylan, typically comprising 15-30% of the biomass dry weight, has high acetyl and glucuronic acid moieties. 4-O-methylglucuronic acid is linked to the xylan backbone by α- (1 2) glycosidic bonds and the acetic acid is esterified at the carbon 2 and/or 3 hydroxyl 17

group. The molar ratio of xylose:glucuronic acid:acetyl residues is about 10:1:7 (38). Side chains on grasses and other annual plant xylans, also typically 15-30% of the plant cell dry weight, are mainly arabinofuranose and acetyl groups and are even more heterogeneous than those from woody tissues. The L-arabinofuranoside branches are linked α-(1 2,3) to the xylopyranose backbone. The ester-linked acetyl groups are also attached to the C2 or C3 hydroxyl. In contrast with hardwoods and annual plants, softwood hemicellulose is dominated by galactoglucomannan (15-20% dry weight), with xylans comprising only 7-10% of the biomass dry weight. The galactoglucomannans are divided into two subtypes of low and high galactose content with galactose:glucose:mannose ratios of 0.1:1:4 and 1:1:3 respectively (38). Softwood xylans, arabino-4-o-methylglucuronoxylans, are not acetylated, but the xylan backbone is substituted at carbon 2 and 3 with 4-O-methyl-α-D-glucuronic acid and α-larabinofuranosyl residues, respectively. Softwood xylan has a xylose:glucuronic acid:arabinose ratio of approximately 8:1.6:1 (39) to 10:2:1 (38). Softwood xylan also has a lower DP of about 100 compared to hardwood xylans with DP around 200 (40) 2.2.2.2 Mannans and mannan derivatives The term mannan indicates a linear polymer of β-(1 4)-linked mannopyranosyl residues. The structure, and hence the degradation, of mannan is very analogous to cellulose, both being linear β-(1 4) linked monosaccharide polymers (41). Mannan, however, is found in only a few particular plants, notably in the endosperm of the ivory-nut from the Tagua Palm (Phytelephas macrocarpa) and a few other plants (41). The mannan polymer can be branched however, with various combination of mainly glucose and galactose residues, giving rise to much more common glucomannans, galactomannans, and galactoglucomannans. 18

Galactomannan is composed of a polymeric β-(1 4) mannopyranosyl backbone highly substituted with β-(1 6) linked galactopyranose residues (42, 43). The degree of substitution varies with source. Areas of galactose-free mannosyl residues can form junction zones through inter-chain hydrogen bonding where there are six or more unsubstituted mannose residues. The number of free mannose regions and degree of overall substitution determines properties such as viscosity and solubility. More highly substituted galactomannans tend to have high degrees of solubility and form more viscous solutions. Galactomannans are generally believed to be an extracellular non-amylose form of carbohydrate storage found predominantly, if not exclusively, in seeds (44). Found mainly in the legume family, there are two major types of galactomannan, differentiated by the source and degrees of galactose substitution. Locust bean gum (LBG), derived from the carob tree (Ceratonia siliqua), contains an average of 2000 sugar residues, with a galactose about every 3.5-mannosyl residues. The other major commercial source is guar gum, from the seed of the leguminous shrub Cyamopsis tetragonoloba. Guar gum contains more galactose residues than LBG, having a galactose every 1.5 to 2 mannose units, and is longer than LBG, with residue counts of around 10,000. The galactomannans form a set of polysaccharides called gums that include other non-hemicellulosic polysaccharides such as carrageenan (alternating β-d- and α-d-galactoses from some seaweeds) and xanthan gum (produced by the bacterium Xanthomonas campestris). Carrageenans are subdivided into λ-, ι-, and κ-carrageenans based on the sulfonation and 3,6 anhydro bridges present on the individual monomers. The linkages alter between α-(1 3) and β-(1 4). Glucomannan is a storage polysaccharide found mainly in the root of the Konjac plant (Amorphophallous konjac). It consists of a β-(1 4) linked mannopyranose and glucopyranose backbone in a ration of 1.6:1 (45). The backbone residues are substituted in a 19

β-(1 3) linkage with several sugars and short oligosaccharides, as well as with O-linked acetyl groups about every 15 residues (46, 47). ardwoods contain low levels of glucomannan, with a glucose to mannan ratio of 1:1 to 1:2 (38). Although most grasses do not contain glucomannan in the mature tissue, ramie has been shown to contain approximately 7.5% glucomannan in the whole plant (48). Other plants found to contain glucomannan include aspen, lily, dates, western hemlock, lodge pole pine, and spruce (47, 49-52). Galactoglucomannan is a hemicellulose found mainly in softwoods, although they have also been described from fern, with fern galactoglucomannan having a lower DP than softwood galactoglucomannan (53). Structurally, galactoglucomannan is comprised of β-(1 4)-linked β-d-glucopyranosyl and β-d-mannopyranosyl units which are partially substituted by α-dgalactopyranosyl and acetyl groups (40). Two types of galactoglucomannans can be separated; water and alkali soluble fractions, with ratios of mannose:glucose:galactose:acetyl residues 3:1:1:0.24 for the water soluble fraction and 3:1:0.1:0.24 for the alkali soluble fraction (54). 2.2.2.3 Glucans In addition to cellulose, the dominant polysaccharide in plant biomass, two other β-glucans play important roles in plant cell wall structure and function. β-glucan and xyloglucan are structurally similar to cellulose, being based on a β-linked glucose backbone. β-glucan consists of mixtures of β-(1 3) and β-(1 4)-linked glucose residues while xyloglucan is a straight β-(1 4) glucopyranose polymer with varying degrees of α-(1 6)-linked xylose residues. Both structures are involved in support and cross-linking of the cellulose matrix through hydrogen bonding interactions with cellulose, other hemicelluloses, and pectins. Xyloglucan is believed to be extensively involved in support of the cell wall through hydrogen bonding interactions with the cellulose (55, 56). These hydrogen bonds are 20

extended to multiple cellulose microfibrils, giving strong support between these cellulose bundles and forming what is referred to as the cellulose/xyloglucan network (57). Rapid freezing, deep-etching electron microscopy evidence indicates that the cellulose-xyloglucan interactions extend beyond surface interactions to include embedding of the xyloglucan into the cellulose microfibril (56). This is supported by the hydrolysis and extraction work of Pauly and co-workers (58). Inclusion of xyloglucan chains in the media during cellulose formation by Acetobacter aceti spp. xylinum has also shown that the interaction of xyloglucan with cellulose extends into the microfibril(59). This interaction is believed to play a part in regulating cell wall expansion, mainly through weakening of the rigid structure of the cellulose microfibrils, allowing flexibility, and modification of these cross-links by specific enzymes (55, 58). Xyloglucans are polysaccharide polymers composed of a linear backbone of β-(1 4) linked glucopyranose moieties with some monomers substituted with xylopyranose in an α-(1 6) linkage. In this aspect, they are very similar to cellulose. The difference arises due to the substitution that occurs on some of the glucose residues in the backbone. In xyloglucan, some of these glucosyl residues are substituted with xylopyranose in an α-(1 6) linkage. The two predominant patterns that occur are based in a repeated tetramer, either three sequential substituted glucoses followed by a single unsubstituted glucose, or a pair of substituted glucoses followed by a pair of unsubstituted glucoses. These patterns are referred to as XXXG and XXGG respectively (60). The xylose side chains can in turn be substituted with one or more of the following disaccharides: α-(1 2)-L-fucosylpyranose-β-(1 2)-Dgalactopyranose or α-(1 2)-L-galactopyranose-β-(1 2)-D-galactopyranose, with the fucose residues being found mainly in primary cell wall (57, 61-65). As the linkage goes through a galactose α-linked to the glucose, these are designated by an L in the pattern designation, 21

such as XXLG, where glucoses 1 and 2 are substituted with xylose, glucose 3 is linked to a galactose, and glucose 4 is unsubstituted(66). α-(1 2)-L-arabinofuranose has also been shown to be substituted onto either the main glucose chain or onto the galactose or xylose side groups (57, 66, 67). In solanaceous plants, such as tobacco, tomato, and potato, the xyloglucan has a high degree of arabinosyl residues attached to up to 60% of the xylose sidechains in an O-2 linkage (57). These polymers are referred to as arabinoxyloglucans. It has been shown that xyloglucans are acetylated through O-linkages to the arabinose or galactosyl sidechains (57, 64, 67, 68). Despite this side chain substitution, a specific acetylxyloglucan esterase has not been discovered. Xyloglucan is found predominantly in dicotyledons, although non-graminaceous monocotyledons such as onion also contain these polysaccharides (58). In mature wood from hybrid poplar, xyloglucan comprises approximately 3% of the cell wall dry weight, reduced from the 6% found in the early primary cell wall (69). In cocoa, xyloglucan makes up 8% of the dry weight of cell wall material (70), while pea stems contain 21% xyloglucan(58). Corn coleoptiles (3 days) contain 2.3 and 6.7% xyloglucan in epidermal and mesophyll cells respectively(71). β-glucan is a glucopyranose polymer containing either β-(1 3) or mixed β-(1 3), β- (1 4) linkages. This polymer is found in plant cell walls under a variety of conditions and in several specific developmental stages. The ratio of (1 4) to (1 3) linkages varies by species and gives specific properties to individual β-glucan polymers. In maize coleoptiles, the mixed β-(1 3), β (1 4) glucan is deposited early in cell development, comprising greater than 70% of the cell wall material in the endosperm, and decreases rapidly after day 5 (72). The second type of plant β-glucan consists of callose, a β-(1 3) linked glucose polymer found primarily in rapidly growing structures such as pollen tubes and developing 22

seeds (73-77). Additional non-plant β-(1 3) forms include curdlan, produced by the bacterium Agrobacterium sp. ATCC 31750 (formerly Alcaligenes faecalis subsp. myxogenes)(78), and krestin, from the fungus Coriola versicolor (79). A subset of the β- (1 3) glucans include β-(1 6) branch points, and while there are multiple forms of this subset produced by fungi, the primary plant produced form is laminarin, produced by a brown algae Laminaria digitata (80). 3 EMICELLULOSE AND CELLULOSE YDROLYSIS FOR BIOMASS CONVERSION 3.1 Overview Of Conversion Chemistry The cellulose and hemicellulose in lignocellulosic biomass can be hydrolyzed to sugars that can be microbially fermented into various products such as ethanol or chemically converted into other products (81-90). In such a case, cellulose is broken down to glucose, the same sugar released by hydrolysis of starch. Thus, the same fermentation and chemical conversion processes as now being used to manufacture products such as ethanol, lactic acid, and lysine from starch glucose can be applied to manufacturing products from the glucose released by hydrolysis of cellulose. The primary challenge is that the glucose in cellulose is joined by beta bonds in a crystalline structure that is far more difficult to depolymerize than the alpha bonds in amorphous starch (91). On the other hand, hemicellulose is an amorphous polymer that is more easily hydrolyzed into its component sugars than is cellulose. owever, hemicellulose is typically made up of five different sugars - arabinose, galactose, glucose, mannose, and xylose as well as other components such as acetic, glucuronic, and ferulic acids. Native organisms do not efficiently ferment this range of sugars to products, and the heterogeneity of the mixture can impact product purity for chemical conversion routes. As 23

presented previously, the remaining 20 to 30% or so of cellulosic biomass is often mostly lignin, a phenyl-propene aromatic compound that cannot be fermented but can be used as a high energy content boiler fuel. In addition, cellulosics contain protein, minerals, oils, and other components with the amount varying with species. Ideally, each of these fractions can be utilized to make fuels, chemicals, food, and feed (92). 3.2 Typical Process Steps There are many possible routes to hydrolysis of cellulose and hemicellulose to sugars, but we will focus on a coupled acid and enzymatic hydrolysis approach considered by many to be most ready for near term applications (81-85, 87-89, 92). Figure 5 presents a simplified process flow diagram for making ethanol from hemicellulose and cellulose sugars released during the pretreatment and enzymatic hydrolysis steps, respectively, based on this approach. The process starts with feedstock transportation to the process facility for possible storage at the site or immediate feeding to the process. In either event, feedstock is then moved into the process using suitable conveying equipment coupled with cleaning operations to remove foreign objects such as rocks and dirt that could jam or otherwise interfere with process operations. The storage, conveying, and cleaning operations are important steps, and experience with these operations is vital to insure successful operations. Next, the material is pressurized for feeding to pretreatment reactors capable of achieving the high solids concentrations that are vital to minimizing energy (steam) use and maximizing sugar concentrations, and acid and steam are added to promote the hemicellulose hydrolysis to its component sugars. It is generally desired to limit acid use to less than about 1% in the hemicellulose hydrolysis reactors to minimize acid use as well as subsequent neutralization and disposal costs, with actual conditions determined by tradeoffs among pretreatment time, temperature, and acid use with some variability in the best conditions among feedstocks. Up 24

to about 80 to 90% of the hemicellulose sugars can be recovered in a an optimized pretreatment operation (93-95). Following pretreatment, the reactor contents are usually rapidly cooled by flashing to a lower pressure to stop sugar degradation reactions and maximize sugar yields. The solids containing most of the cellulose and lignin in the feed are washed in special equipment to recover nearly all the sugars released from the hemicellulose in a liquid stream that is then conditioned with lime to neutralize the p and remove fermentation inhibitors, and ion exchange is applied, if necessary, to remove acetic acid and other components that can retard downstream operations. Depending on enzyme loading requirements and enzyme production yields, about 4 to 9% of the washed solids and possibly some liquid from the hemicellulose sugar recovery and conditioning steps are used to grow an organism such as Trichoderma reesei that can make enzymes known as cellulase that can be added back to the bulk of the pretreated solids to hydrolyze cellulose to glucose. Many operations are currently based on adding the cellulase to the pretreated solids along with the conditioned hemicellulose sugar stream and fermenting the resultant sugar stream with an organism such as genetically engineered E. coli that ferments all of the five sugars in hemicellulose and cellulose to ethanol with the high yields vital to commercial success. In this so-called simultaneous saccharification and cofermentation or SSCF process configuration, the glucose released by enzymatic hydrolysis of cellulose is converted into ethanol as rapidly as it is released. Ethanol is recovered at appropriate purity from the fermentation broth by a combination of distillation and dehydration operations. The bottoms from the distillation column contain unreacted cellulose, lignin, cellulase, fermentative organisms, and other residual ingredients as well as water, and the solids are recovered and burned to generate all the heat and electricity needed 25

for the conversion process with excess electricity sold for extra revenue. A portion of the water is recycled to the process while the rest passes to waste water treatment for proper disposal. Methane gas produced during anaerobic treatment is also fed to the boiler to generate heat and electricity. The SSCF configuration is favored to reduce equipment costs and the powerful inhibition of cellulase activity by glucose and particularly cellobiose. Adding cellulase, pretreated cellulose, and fermentative organisms in the SSF configuration has been clearly demonstrated to be favorable for reducing ethanol production costs (82). The low concentration of free glucose and presence of ethanol also make it more difficult for invading microorganisms to takeover the fermentation reaction and form undesirable products. Although SSCF has similar potential to improve fermentation of cellulosic sugars to products other than ethanol, most of these other products are less volatile than water, and separation of the product from the more complex fermentation broth at the completion of fermentation is more challenging. In addition, other combinations of pretreatment and hydrolysis reactions can be applied to release sugars from cellulose and hemicellulose fractions, as described in other sections of this chapter. 3.3 Cellulose and hemicellulose hydrolysis reactions Cellulose is a long chain of glucose molecules covalently joined by beta bonds with many of the neighboring chains typically linked together through hydrogen bonding in a crystalline structure (96). Acids or enzymes known as cellulases can catalyze the reaction of water with the glucose molecules in these chains to release single glucose molecules monomers by the following reaction: ( C6 10 O5 ) n + n 2O nc6 12 O6 (3.1) 26

in which 2 O is water, (C 6 10 O 5 ) n is a chain made up of n glucose molecules that is often termed glucan, and C 6 12 O 6 is a glucose monomer. Thus, each glucose unit in the long chain combines with a water molecule, and 180 mass units of glucose are released from 162 mass units of glucan and 18 mass units of water, an 11.1% mass gain. Oligomers made up of several glucose molecules may also be released as intermediates in cellulose hydrolysis and often contain only 2 to perhaps 3 glucose units. Cellulase enzymes are very specific in only catalyzing addition of water to glucan chains, and the optimum temperature needed for reaction 3.1 is only about 50 o C, virtually eliminating degradation reactions. Thus, only glucose is formed via enzymatically driven hydrolysis of cellulose, and yields can approach 100%. On the other hand, use of dilute acids (e.g., 1.0% sulfuric acid) to drive reaction 3.1 requires much higher temperatures of about 220 o C, and the acid also triggers the breakdown of glucose to degradation products such as hydroxymethyl furfural, limiting yields. Concentrated acids (e.g., 75% sulfuric acid) can be used at moderate temperatures to achieve high yields similar to those for cellulase enzymes. emicellulose can also be hydrolyzed by the addition of water to release individual sugar chains contained in the longer hemicellulose molecule. The stoichiometry for the reaction of the hexose sugars galactose, glucose, and mannose that are in hemicellulose is the same as shown in equation 3.1, and an 11.1% mass gain results for these molecules. On the other hand, addition of water to the five-carbon sugar molecules arabinan and xylan in the hemicellulose molecule follows the following stoichiometry: ( C5 8 O4 ) n + 2 O nc5 10 O5 (3.2) with (C 5 8 O 4 ) n being a chain made up of n arabinose or xylose (pentose) molecules that can be termed arabinan or xylan, respectively, and C 5 10 O 5 being one of the corresponding pentose sugars formed by hemicellulose hydrolysis. Based on this reaction, we can see that 27

the molecular weight of the sugar molecule released increases from 132 mass units before hydrolysis to 150 mass units of pentose sugars formed, a gain of over 13.6%. emicellulose is open to attack at intermediate positions along its long backbone with the release of oligomers made up of many sugar molecules that can be successively broken down to shorter chained oligomers before single sugar molecules are formed (97). A suitable cocktail of enzymes known collectively as hemicellulase can catalyze addition of water to hemicellulose with high specificity at modest temperatures, thus avoiding sugar degradation and resulting in high sugar yields. Dilute acids (e.g., sulfuric) can also catalyze hemicellulose hydrolysis to sugars at temperatures of about 100 to 200 o C, but furfural and other degradation products are formed from the sugars at these temperatures if one targets good yields of hemicellulose sugars in solution. Nonetheless, degradation of the sugars released can be modest enough to recover about 80 to 90% of the maximum possible sugars (93-95). On the other hand, operation without adding acid limits recovery of hemicellulose sugars to about 65% or less, with most in oligomeric form (98). 4 ACID YDROLYSIS OF CELLULOSE 4.1 System Description and Performance When heated to high temperatures with dilute sulfuric acid, long cellulose chains break down to shorter groups of glucose molecules that then release glucose that can then degrade to hydroxymethyl furfural (99-109). Generally, the majority of cellulose is in crystalline form, and harsh reaction conditions (high temperatures, high acid concentrations) are needed to liberate glucose from these tightly associated chains. Furthermore, it is found that the rate of cellulose hydrolysis increases faster with increasing temperature than the rate of glucose degradation does, and as a result, yields increase with temperature and acid concentration, 28