“9534‘ 5..hv-‘N., . SEE-g“. ., Y” . :‘t' . #5922? ”'14 “fink ‘ 1):”.ij ;x Pug-‘33 .‘g - ura- a. mm» {Mi u! w. mac 1:252; aw: 112.47.. 145"; may: 3, - 3-“; .‘Q'Z ”:1: ‘95: 11"" “£2 ... .grf»:-rw “(hf-77” gulf UiCHKiAN STATE UNIVERSITY UBRAH ES LIBRARY l/II/l/Ih’l/H/l/Il/lll'lll/I/lllllllll/l/l/ll/l/U/l/ll Michigan State 3 1293 00759 7218 University L This is to certify that the thesis entitled THE EFFECTS OF PHYSICAL AND CHEMICAL CONSTRAINTS ON THE ENZYMATIC HYDROLYSIS OF LIGNOCELLULOSIC MATERIALS presented by DAVID NEIL THOMPSON has been accepted towards fulfillment of the requirements for M' 5- degree in Ch- E Major professor Date NOV, ’7) ’99? 0-7639 MS U is an Affirmative Action/Equal Opportunity Institution PLACE IN RETURNED)! to remove this checkout from your record. TO AVOID FINES return on or bdore due due. DATE DUE . DATE DUE DATE DUE MSU to An Affirmdlve Action/Equal Opportunity Institution ammo-m THE EFFECTS OF PHYSICAL AND CHEMICAL CONSTRAINTS ON THE ENZYMATIC HYDROLYSIS OF LIGNOCELLULOSIC MATERIALS by David Neil Thompson A THESIS Submitted to Michigan State University in partial fulfillment of the requirements for the degree of MASTER OF SCIENCE Department of Chemical Engineering 1989 I (I? C» 4' 1‘5 ABSTRACT THE EFFECTS or PHYSICAL AND CHEMICAL CONSTRAINTS ON THE ENZYMATIC HYDROLYSIS 0F LIGNOCELLULOSIC MATERIALS by David Neil Thompson Mixed hardwood (90 % birch, 10 % maple) was pretreated by four methods which produce substrates differing substantially in chemical composition and structural configuration. The pretreatments included alkaline peroxide delignification, solvent delignification with cthylenediamine, steam autohydrolysis in a plug flow reactor, and dilute acid hydrolysis in a plug flow reactor. The pretreated substrates were then hydrolyzed with Rutgers C30 cellulase at 40.8 FPU/g wood and their crystallinities, lignin contents, and surface areas available to solutes within the cellulase size range determined. The initial glucose yield was found to depend linearly on the surface area available to enzymes with diameters ranging from 24 to 77 A, regardless of the method of increasing the surface area. Empirical correlations were also derived which also showed that crystallinity and lignin content have little effect on the initial rate of enzymatic hydrolysis. ACKNOWLEDGEMENTS ,I would like to thank Dr. Hans E. Grethlein, Dr. Carlos Dosoretz and Dr. H.C. Chen for the encouragement and expert assistance they offered throughout the course of this project. I would also like to thank Vicki S. Freund for her moral support during the long days of lab work. I am also grateful to Michigan Biotechnology Institute, Michigan State University’s Research Excellence Fund and the Microbial Ecology Center for financial assistance. TABLE OF CONTENTS page List of Tables ...................................... vii List of Figures . .................................... ix Chapter I: Introduction . .............................. 1 Chapter 11: Literature Survey ......................... 6 1. Introduction .................................... 6 2. Pretreatment Rationale ............................. 7 3. Ccllulases ...................................... 8 1. Sources of Ccllulases ............................ 8 2. The Cellulase Complex ........................... 9 3. Ccllulase Dimensions ........................... 10 4. Cellulase Adsorption ........................... 11 4. Properties of Lignocellulosics Which Limit Enzyme Accessibility .................................. 13 1. Available Surface Area .......................... 14 2. Cellulose Crystallinity .......................... 20 3. Lignin Content ................................ 23 5. Pretreatments ................................... 25 1. Physical Pretreatments .......................... 26 2. Chemical Pretreatments ......................... 3O 1. Alkali Pretreatments .......................... 30 2. Acid Pretreatments ........................... 31 3. Delignification .............................. 34 4. Cellulose Solvation Pretreatments ................ 36 5. Biological Pretreatments ....................... 37 Chapter 111: Materials and Methods .................... 38 1. Pretreatment Substrate ........................... 38 iv 2. Acid Hydrolysis and Autohydrolysis Pretreatments ...... 38 3. Ethylenediamine Pretreatment .................. . . . . . 4O 4. Alkaline Peroxide Pretreatment ..................... 43 5. Cellulose Crystallinity ........................... 45 6. Substrate Composition ........................... 45 7. Solute Exclusion Technique ....................... 49 8. Enzymatic Hydrolysis ............................ 57 Chapter IV: Results .................... . . . . ......... 61 l. Pretreatments .................................. 61 2. Substrate Chemical Compositions ................... 65 3. Substrate Crystallinities .......................... 71 4. Pore Size Distributions and Surface Areas ............ 71 5. Enzymatic Hydrolysis ............................ 86 1. Choice of Enzyme Level ........................ 86 2. Enzymatic Hydrolysis of Pretreatcd Wood ........... 89 6. Surface Area Correlations ......................... 96 7. Empirical Correlations .......................... 100 Chapter V: Discussion . . . . ...... . ................... 104 1. Significance .................................. 104 2. Pretreatment Effectiveness ....................... 104 3. Cellulase Size ................................. 106 4. Future Directions .............................. 106 APPENDIX.............. ........................... 109 Appendix A: Derivation of equation (3-8) .............. 109 l. Derivation .................................. 109 2. Limitations .................................. 111 Appendix B: Tabulated Data ........................ 112 BIBLIOGRAPHY ..... . ................................ 132 LIST OF TABLES Table title page 3.1 Pretreatment run conditions for the FRWA and FR pretreatment of Native mixed hardwood ............. 41 3.2 Molecular probes used in the solute exclusion technique ......... 51 4.1 Fractional recoveries (R,) and polymerization factors (F,) for the wood carbohydrates ................... 66 4.2 Composition of Wilncr 060 Native mixed hardwood . . . . 66 4.3 (a) Compositional analyses of alkaline peroxide pretreated wood ............................. 67 (b) Compositional analyses of EDA pretreated wood . . . . 67 (c) Compositional analyses of autohydrolysis (FR) pretreated wood ............................. 68 (d) Compositional analyses of acid hydrolysis pretreated wood (1 % H280.) ................... 68 4.4 Relative amounts of lignin and hemicellulose in each pretreated substrate ............... _ .............. 69 4.5 Potential glucose amounts in each pretreated substrate . . 72 4.6 Measured values of CrI for all substrates ............ 74 4.7 Best fit curve parameters and mean square residual errors for all substrates ......................... 77 4.8 Fiber saturation points (ESP) and total specific surface areas (SSA) for all substrates ............... 84 4.9 Specific surface area (SSA) in pores larger than various enzyme sizes over the range 24 to 77 A ....... 85 4.10 Glucose yields from enzymatic hydrolysis at times ranging from 0.5 to 4 hours ...................... 95 vii 4.11 (a) Slopes of the correlations derived for glucose yields at various times and surface areas available to solutes of various sizes ............. 98 (b) Correlation coefficients for the correlations given in Table 4.11(a) ........................ 98 4.12 Constants a, b, c, and d determined by linear regression for the empirical fits to equation (4-2) . . . . 101 4.13 Slopes and correlation coefficients for the correlations derived for actual glucose yield at 2 hr vs predicted yields from the empirical correlations .................................. 101 viii LIST OF FIGURES Figure title 1.1 The cell wall of wood is composed of three layers, 81, S2, and S3, arranged as shown [7] .............. 2.1 The crystalline core envisioned in the Fringed Micellar Model [36]. Figure is from Cowling and Kirk [l8] ................................. 2.2 Packing arrangement of the cellulose cores into elementary microfibrils in the Fringed Micellar Model. Figure is from Cowling and Kirk [l8] ......... 2.3 Arrangement of the glucosidic bonds on the surface of a microfibril [36] ............................ 2.4 The unit Coll of crystalline cellulose, showing the 002 plane [36] ............................. 2.5 Typical monomer units and inter-monomer bonds found in spruce lignin [18] ....................... 3.1 Schematic of the MBI High Temperature Flow Reactor ........................ . .............. 3.2 Separation apparatus for pulp and liquid in the solute exclusion technique ............................ 3.3 Schematic diagram of the modified HPLC apparatus used to measure concentrations of solute solutions ..... 3.4 Parallel plate geometry, where the plates are micro- fibrils in the lamella of the cell wall ............... 4.1 Comparison of the physical appearances of pretreated mixed hardwood (wet) with Native wood: (a) Alkaline H202, 2.5 %, 25 °C, 200 rpm; (b) EDA, 90 °C, 1 hr, 175 rpm ..................................... 16 18 24 39 4.2 4.3 4.4 4.5 4.6 4.7 4.8 4.9 Comparison of the physical appearances of pretreated mixed hardwood (wet) with Native wood: (a) Auto- hydrolysis; (b) Acid hydrolysis, 1 % H280; .......... X-Ray diffractogram for Native mixed hardwood ...... Pore size distribution for Native mixed hardwood ...... Pore size distributions for the alkaline H202 pretreatments. (a) 0.5 hr, (b) 5 hr, (c) 19 hr .......... Pore size distributions for the EDA pretreatments. (a) 10 %, (b) 50 %, (c) 70 % ..................... Pore size distributions for the autohydrolysis (FR) pretreatments. (a) 240 °C, (b) 260 °C, (c) 280 °C ...... Pore size distributions for the acid hydrolysis pretreatments. (a) 180 °C, (b) 200 °C, (c) 220 °C ...... Determination of maximum CCN 3000 activity for FRWA 220 °C wood ............................ 4.10 Comparison of the time course of enzymatic hydro- lysis of FRWA 220 °C wood with (A) Novo CCN 3000 cellulase and (O) Rutgers C30 cellulase ............. 4.11 Time course of the enzymatic hydrolysis of Native (0), and alkaline H20a pretreated wood; (I) 0.5 hr, (A) 5 hr, and (V) 19 hr ................. 4.12 Time course of the enzymatic hydrolysis of Native (0), and EDA pretreated wood; (n) 10 %. (A) 50 %. and (v) 70 % ................. 4.13 Time course of the enzymatic hydrolysis of Native (0), and FR (autohydrolysis) pretreated wood; (I) 240 °C, (A) 260 °C, and (V) 280 °C ....... 4.14 Time course of the enzymatic hydrolysis of Native (0), and FRWA (acid hydrolysis) pretreated wood; (I) 180 °C, (A) 200 °C, and (V) 220 °C ................................ 4.15 Linear correlation between glucose yield at 2 hr and specific surface area available to a solute of 51 A diameter: Native (0), alkaline H202 ([3), EDA (I), FR (.), and FRWA (A) .......................... 75 78 81 83 88 90 91 94 97 4.16 Variation of the correlation coefficients from the linear correlations in Table 4.11(a): (O) 0.5 hr yields, ([1) 1 hr yields, (I) 2 hr yields, and (O) 4 hr yields ............................... '. 99 4.17 Linear correlation between actual glucose yield at 2 hr and predicted glucose yield from the empirical correlation using surface area available to a solute of 51 A diameter: (0) Native, (:1) alkaline no, (I) EDA, (0) FR, and (A) FRWA ................. 102 CHAPTER I: INTRODUCTION Low value lignocellulosic materials such as wheat straw, sawdust and other agricultural residues are produced at a rate of about one billion tons (dry weight) annually in the United States‘. It has been estimated that about 20 % of this biomass can be economically collected and fed into the production of chemicals‘. It is evident that this represents a renewable resource that would be largely useful if properly utilized. Many chemicals can be produced from these lignocellulosic materials, including ones useful in producing energy, such as ethanol”, ethylene, propylene and related compounds‘. Other uses include production of high energy animal feeds‘, polymer feedstocks‘ and organic solvents‘. Lignocellulosc is a highly heterogeneous 3-dimensional matrix comprised of cellulose, hemicellulose and lignin. Cellulose, the major component of the matrix, comprising about 40-50 %, is a linear glucose polymer consisting of B-D-l,4-anhydroglucopyranose units and is the source of glucose from which many chemicals could be derived. Hemicellulose, comprising about 20-30 % of the matrix, is a mixed polymer consisting mostly of xylose (in hardwoods) or mannose (in softwoods) and some galactose, glucose, arabinose, and glucuronic and galacturonic acids. The remainder of the matrix is mostly lignin, a polymer of various phenolic momomers‘ and is dispersed throughout the matrix. This lignin barrier serves to protect the cellulose and hemicellulose from microbial degradation. 2 The lignocellulosic matrix makes up the cell walls of plants. Surrounding the cell is the middle lamella (see Figure 1.1), a highly lignified region containing lignin and hemicellulose in a ratio of about 70 %:30 %'. In the cell wall, there are four distinct morphological layers. The outermost layer, or primary wall, contains approximately one-third cellulose mixed with lignin and hemicellulose in a loose and random network'. The secondary wall is composed of three layers: S1, 82, and 83’. In the outer layer, or 81, the cellulose microfibrils are wound in flat helices, while in the middle layer (82) the microfibrils are tightly packed in a parallel helix'. It is not clear how the microfibrils are arranged in the innermost (S3) layer. Typically the secondary wall contains a majority of cellulose'. This type of arrangement presents a problem of accessibility for cellulases, which degrade the cellulose polymers, since these enzymes are generally very large in comparison with the spaces between the fibers into which they must diffuse. In addition, it has long been thought that the arrangement of the cellulose in the secondary wall, or its crystallinity, might have some exclusion effects on the enzymatic action since direct contact between the enzyme and fiber are necessary for cellulolytic activity’”. It is a well accepted notion9 that some sort of pretreatment of the lignocellulosic material is necessary to make the cellulose in the matrix more susceptible to cellulolytic attack. Various types of l nln-Hemlcell I hgtrlx uose Figure 1.1: The cell wall of wood is composed of three distinct layers, 81, $2, and S3, arranged as shown [7]. o-‘Flbro Direction -° 4 pretreatments have been employed, such as treatment with acids“'”, alkalis""‘, microwaves“, and many others. Most have produced an increase in enzymatic glucose yield, although to varying degrees. The unifying question among the pretreatments, however, is not the fact that they all increase glucose yield but how they do it. In all cases, some component of the matrix is removed, completely or in part, to open up the pores of the matrix to the cellulase’. Yet some doubt still exists as to the importance of the various factors which are thought to influence the rate of cellulolytic hydrolysis, such as crystallinity”, lignin content“, and cellulose surface area available to the cellulase enzyme". In addition, it is often difficult to compare the results of pretreatments, since researchers invariably use different cellulases and/or different substrates to ten pretreatment effectiveness. In this work, a common substrate, mixed hardwood (90 % birch, 10 % maple), was pretreated by four methods which modify the structure and composition of the lignocellulosic matrix, each in a different manner. These included dilute acid hydrolysis in a plug flow reactor, steam autohydrolysis in a plug flow reactor, organosolvent delignification with ethylenediamine, and alkaline peroxide pretreatment. The lignin content, crystallinity and surface area available to a T. reesei cellulase for each pretreated substrate were determined and combined with enzymatic hydrolysis yields at various times of hydrolysis. These data were used to show a linear 5 correlation between initial glucose yield from enzymatic hydrolysis and surface area available to the cellulase enzyme. Empirical fits of the initial rate data with crystallinity index, glucan to lignin ratio, and available surface area were also determined. These fits also showed that available surface area is the limiting parameter in the initial stages of enzymatic hydrolysis, regardless of how it was created. This thesis is divided into five chapters, including this first chapter (Introduction). Relevant literature are presented in Chapter II, focusing on cellulases, factors limiting cellulolytic hydrolysis of cellulosics and lignocellulosics, and pretreatments to enhance cellulolytic hydrolysis. The third chapter is devoted to pretreatment and analytical pr0tocols for all techniques used to obtain data for this work. Numerical results of this study are presented in Chapter IV, while their interpretation and significance are discussed in Chapter V. Finally, pertinent equations are derived in Appendix A, and data used in figures are listed in tabular form in Appendix B. CHAPTER II: LITERATURE SURVEY 2.1 Introduction Lignocellulosic biomass occurs in nature in a variety of forms. Agricultural residues such as corn stover, wheat straw and chaff, rice straw and chaff, and bagasse (from sugarcane or other sources) are produced at a rate of nearly one billion tons annually in the United States alone, although only 20 % of this can be economically collected and utilized‘. The lumber and pulp and paper induStries also produce large amounts of sawdust, bark, wood chips, and waste paper pulp. In addition, urban refuse contains large amounts of waste paper products, such as newspaper, cardboard, and so on. These lignocellulosic wastes constitute a significant renewable resource, and thus considerable research efforts have gone into developing economically feasible processes for the conversion of lignocellulosics into useful products”"'. The major barrier to these efforts, however, is that the complex lignocellulosic structure, engineered by nature to protect plants from microbial attack, also presents a barrier for the degradation of these materials to their substituents by man. When broken down, cellulose yields glucose, which can be fermented to ethanol or used to produce other chemicals‘. Hemicellulose gives a number of five- and six- carbon sugars, several of which can be fermented to ethanol“. Lignin, which is a complex polymer madevup of a variety of phenylprOpane units’, gives aromatic molecules which could serve as ocrane 7 enhancing additives for conventional fuels. Some of the more relevant topics in the conversion of lignocellulosics will be reviewed in this literature survey, with particular emphasis given to factors affecting cellulolytic conversion of the cellulose in the lignocellulosic matrix, and pretreatments for this conversion. 2.2 Pretreatment Rationale Most of the research to date on the conversion of lignocellulosics has gone into the conversion of the cellulose to glucose, since glucose can easily be fermented to ethanol using common yeasts“. Dilute acid hydrolysis of the cellulose to glucose is a relatively simple process which can give moderately high yields in relatively short times’. Protons are small, so accessibility of the catalyst to the glucosidic bonds is not a problem. The sugar monomers produced, however, can undergo acid-catalyzed reactions to form degradation products. The two most common degradation products, furfural (from xylose) and hydroxymethylfurfural (from glucose), can condense to form insoluble polymers which precipitate and cause reactor problems, besides reducing yields. Unfortunately, while yields increase with increasing hydrolysis temperature, so does the formation of these degradation products". Cellulase enzymes, on the other hand, are specific for cellulose and form only glucose. The hydrolysis conditions are mild (50 °C, pH 4.80)“ and thus there are no degradation products. Ccllulases, however, are relatively large enzymes, much too large to fit into the spaces in the lignocellulosic 8 matrix". Some researchers have tried using methods such as reducing cellulase size20 to increase accessibility and using surfactants to increase cellulase adsorption on the available surface area“, but it is a generally accepted action that some sort of pretreatment must be employed to increase the accessibility to cellulasesi'9'“. Although opinions differ as to which property of lignocellulosics is most important, a "good" pretreatment will normally remove one or more components of the matrix, in addition to swelling the matrix and the crystalline cellulose microfibrils”. 2.3 Ccllulases 2.3.1 Sources of Ccllulases Extracellular cellulases are produced by a number of wood and plant degrading fungi and bacteria. While many microorganisms are capable ’of growing on cellulose, only a relatively small fraction of these can produce extracellular cellulases”. Fungi that produce extracellular cellulases include several Trichoderma Species, such as T. reesei, T. viride, T. kongii, and T. Iignorum. Other fungal species include Penicillium funiculosum, A. wentii, and several others. Bacterial species which produce extracellular cellulases include Cellulomonas and Clostridium thermocellum. In the early 19503, it was found that T. reesei (thought to be a T. viride at that time) produces a cellulase complex which has excellent activity on crystalline cellulose”. T. reesei cellulases, as well as cellulases produced by mutant strains of T. reesei, are among the most common 9 cellulases used today“. In particular, the T. reesei QM9414 and T. reesei Rut-C30 strains have high cellulase productivities’, and T. reesei Rut-C30 has the added benefit that it is resistant to catabolite repression". 2.3.2 The Cellulase Complex The term "ccllulasc" refers not to a single enzyme, but to a family of enzymes which act synergistically to degrade the cellulose polymers. The separate components can be grouped into three classes: endoglucanases, exoglucanases and B-glucosidases (or cellobiases). Endoglucanases, or endo-B-1,4-glucanases, attack the glucosidic bonds between molecules in the interior of the chains, forming short oligomers. These enzymes constitute 15-20 % of the total extracellular protein produced when T. reesei is grown on cellulose”. As many as six endoglucanases have been reported“; each showed significant differences in its specificity and in the number of glucose units in its final oligomeric product. Exoglucanases, or exo- B-l,4-glucanases, catalyze the hydrolysis of the glucosidic bonds in an end-wise manner, releasing cellobiose; because of this, exoglucanases are often called cellobiohydrolases. Exoglucanases, of which there are two in T. reesei cellulase, comprise 35-85 % of the extracellular protein when T. reesei is grown on cellulose“. B-glucosidases (cellobiases), as their name implies, cleave cellobiose molecules to give two moles of glucose per mole of cellobiose. These enzymes comprise less than 1 % of the extracellular protein 10 produced when T. reesei is grown on cellulose. For this reason, it is often necessary to supplement T. reesei cellulase with B-glucosidase from another source”, for example Aspergillus niger”. This helps to keep cellobiose levels low. 2.3.3 Cellulase Dimensions Ccllulases from various sources are quite different in size and molecular weight. Since there are so many different enzymes in the cellulase complex, it is difficult to make a "universal" estimate of cellulase dimensions. Depending on the substrate and the fermentation conditions, different amounts of each component will be produced, resulting in different average dimensions. Cowling‘ compiled data on molecular weights and gel filtration for cellulases from various sources and estimated the dimensions of their cellulase complexes. The results showed that if the cellulases are hydrodynamic spheres in solution, their diameters range from 24 to 77 A, while if they are ellipsoids with an axial ratio of about six, their sizes range from about 13 x 79 A to 42 x 252 A in width and length, respectively‘. Both spherical and ellipsoidal dimensions have been used by researchers in the past. Van Dyke“ used ellipsoidal dimensions (35 x 210 A) in a study of cellulase accessibility, while Grethlein" preferred to use the mean of the spherical diameter range (51 A) in a similar study. Stone et al.27 concluded that the cellulase produced by rumen microorganisms was excluded from pores smaller than 40A in swollen cutten linters, using the solute exclusion 11 technique”. All values previously used appear to be within the ranges reported by ,Cowling‘, although it is Still uncertain what steric conformation is maintained in solution by the cellulase complex. A recent study using FPLC and small-angle X-Ray scattering to characterize T. reeset' cellulases suggests that the shape of one of the cellobiohydrolases from the complex is that of a tadpole, with a maximum width at the head of 44 A and a total length of 180 A”. The size of this enzyme is therefore well within the reported ranges for cellulase size, although it is only one of the enzymes involved. One possibility is that the separate enzymes are excluded from pores based on their own dimensions and not on the basis of the cumulative cellulase complex dimensions. Then the size ranges reported by Cowling‘ might be viewed as an average over the separate ccllulase components. 2.3.4 Cellulase Adsorption One of the first steps in the enzymatic hydrolysis of cellulose is the physical adsorption of the cellulase onto the substrate. Reese3° showed that cellulase adsorption is a function of enzyme concentration, available surface area and its nature, physical properties of the enzymes present, and adsorption conditions (for example, pH, salt concentration and temperature). Kim et al.“ reported strong pH effects on the adsorption of T. viride cellulase using microcrystalline cellulose as subStrate. Adsorption was maximum at pH 4.8, falling sharply with pH in either direction. This 12 result is contrary to results reported by other researchers”, in which it was shown that pH had little effect on adsorption in the range 3.5-5.5 although they showed a maximum, also at pH 4.8. The discrepancy may be due to the use of different enzymes, or to differences in the amounts of each component enzyme in the complex, since endoglucanases preferentially adsorb on crystalline substrates”. Temperature also has a large effect“31 on the adsorption of cellulases. Mandels et al.“ showed that that adsorption rate of cellulase on Solka Floc increased with temperature, but the maximum amount adsorbed showed the opposite trend”. Kim et al." also also reported a decrease in protein adsorbed with increasing temperature using microcrystalline cellulose, while Bisaria and Chase“ found that the specific adsorption increased with temperature up to a maximum at 50 °C, using alkali-treated bagasse as substrate. Perhaps the differences in the trends occur because of choice of substrate; both groups that reported decreasing adsorption with increasing temperature used essentially pure cellulose, while the opposite trend was observed using a lignocellulosic substrate. It was observed" that even though adsorbed protein decreased with increasing temperature, enzyme activity increased to a maximum at 50 °C. This led the authors to speculate that the active enzyme was bound more strongly to the cellulose surface at higher temperatures. Whatever the trends, 50 °C seems to be the optimum hydrolysis temperature at the pH optimum of 4.8. 13 Specific surface area was also found to be a major determinant in protein adsorption’"“”. Lee and Fan”, using Solka Floc as substrate, came to the following conclusions: (1) The hydrolysis rate is influenced little by mass transfer limitations of the enzyme; (2) The enzymatic portion of the soluble protein (cellulase preparations are inherently impure) is more readily adsorbed than the nonenzymatic portion; (3) The maximum hydrolysis rate and maximum extent of soluble protein adsorption is attained at the onset of hydrolysis; (4) The initial extent of soluble protein adsorption and the initial hydrolysis rate are linearly related to the initial enzyme concentration; (5) The extent of initial soluble protein adsorption increases approximately linearly as the specific surface area increases; and (6) The hydrolysis rate depends on: (a) the initial extent of protein adsorption, which depends on initial cellulose concentration, enzyme concentration and specific surface area; and (b) the effectiveness of the adsorbed protein to promote the hydrolysis, which is related to the crystallinity of the cellulose. 2.4 Properties of Lignocellulosics Which Limit Enzyme Accessibility Several physical properties of lignocellulosics limit the accessibility of the cellulose to the cellulase complex’. These include specific surface area available to the enzyme, cellulase crystallinity, and lignin content. Researchers studying microcrystalline cellulose tend to place more emphasis on l4 crystallinity, while with lignocellulosics, their observations normally place more importance on surface area, and in a related fashion, distribution of lignin in the matrix. These points will be discussed in the following sections. Since hemicellulose is easily hydrolyzed and is normally removed in most pretreatments“, it will not be discussed here. 2.4.1 Available Surface Area Wood and cotton fibers contain two types of voids: (1) gross capillaries (0.02 to 10 um diameter) including cell lumina and pit aperatures, and (2) cell-wall capillaries, which include spaces between amorphous regions in the cellulose and between microfibrils. The cell wall capillaries are closed off when the material is dry, but when fully saturated with water, they are often as large as 0.02 pm in diameter". Removal of matrix components serves to increase this pore size. Two models have been proposed for the microstructure of cellulose fibers. In the Fringed Micellar Model", the cellulose fibers are envisioned to have a crystalline core 50 x 100 A in cross section, surrounded by an amorphous sheath containing regions of order and disorder along the length of the fiber (see Figure 2.1). It was postulated that these fibers associate into microfibrils which contain 15-40 cellulose chains and are segmented into crystalline and paracrystalline regions (Figure 2.2). The Folding Chain Model” proposes that the basic unit is the platellite, with a cross section of 15 4— 100A ———9 .. I/l /,3fi”,gl<fiug ‘/ IalflfflJ/////”' "/17///////////'/ I r////////////”_ 50A -"'////////////1. I '.'///////////// l /I///f///rr/// / "t.-"'t’.-'.t’-‘ ”:7 "' .'.- I / Figure 2.1: The crystalline core envisioned in the Fringed Micellar Model [36]. Figure is from Cowling and Kirk [18]. 1|in ”Illinlill-Ijni' MA M W'AV MA Av VAN v M M M Crystalline region ———< ~ - 5' ‘$ ‘ Pamcrystalltne region { i l Figure 2.2: Packing arrangement of the cellulose cores into elementary microfibrils in the Fringed Micellar Model. Figure is from Cowling and Kirk [18]. 17 36 x 6 A and a degree of polymerization (DP) .of up to 1000 glucose residues in each platellite. Several of these platellites pack into an elementary fibril, called a crystallite, and sixteen crystallites pack into a fibril, giving a cross section of approximately 120 A square. Four fibrils pack together to give a microfibril". Three or four glucose bonds are involved in each fold, providing weak, susceptible links for cellulase attack. During dilute acid or enzymatic hydrolysis, these microfibrils are partially hydrolyzed, preferentially in the amorphous regions. It has been observed that the final DP for a given substrate (called the "leveling-off" degree of polymerization, or LODP) at the same conditions is constant”'”. This evidence seems to support the folding-chain model with its orderly structure. Regardless of the correct model, the cellulose chains are linearly aligned in microfibrils with the B-1,4-glucosidic bonds between the anhydroglucopyranose units buried one-half molecule deep on the surface of the microfibril“ (see Figure 2.3). Thus, the number of glucosidic bonds accessible to the enzyme will be proportional to the surface area of the microfibrils that is accessible. As with most enzymes, the cellulase must come into direct contact with the glucosidic bond to catalyze its hydrolysis". Thus, the rate of enzymatic hydrolysis should depend directly on the available surface area. Whether or not the dependence is linear will depend on the effects of other factors, such as enzyme concentration, crystallinity, and lignin content. If these effects are small or 18 accessible hydroxyls \ H N inaccessible '3 5" H..;11hydroxyls {3.} tiocn2 Figure 2.3: Arrangement of the glucosidic bonds on the surface of a microfibril [36]. 19 negligible, the dependence will be essentially linear. Stone et al.” reported a linear relationship between the initial rate of enzymatic hydrolysis and available surface area in acid-swollen cotton. Grethlein” reported the same linearity in a study of dilute acid hydrolysis of various lignocellulosic materials. Grous er al.‘° also reported a linear relationship between initial rate of enzymatic hydrolysis and surface area in steam exploded poplar. In this study the same enzyme and activity used by Grethlein" were used; remarkably, the slopes of both correlations were nearly identical (0.375 and 0.326 %/m’, respectively). These results suggest that surface area available to the enzyme is the initial limiting factor, although the correlations were not tested with delignified substrates. Fan et al.“"’ developed an empirical correlation for microcrystalline cellulose relating the relative extent of hydrolysis in eight hours (REH) to specific surface area (SSA) measured by nitrogen adsorption, and crystallinity index (CrI): REH =- 0.380 (SSA)""‘ (100 - CrI)"°‘ (2-1) which indicates that SSA has only a small effect, while crystallinity is much more important, at least in the case of microcrystalline cellulose. For wheat straw, however, Gharpuray et a1.“ developed a similar correlation which showed a much greater dependence on surface area, while exhibiting a much smaller effect of crystallinity and a moderate lignin effect: REH = 2.044 (SSA)°'”' (100 - CrI)°'”7 (Lignin)'°'3“ (2-2) 20 These results clearly show that the enzymatic hydrolysis of microcrystalline cellulose and lignocellulosic materials are fundamentally different. 2.4.2 Cellulose Crystallinity. The close association of cellulose chains and the nature of its monomers, i.e. hydroxyl groups, creates an ideal situation for interchain hydrogen bonding in a highly oriented manner”. Crystallinity may be measured by a number of methods, which can be placed into two categories, those giving c0tton a crystallinity of about 70 %, and those giving..a cotton crystallinity of about 90 %". Methods in the first category include X—Ray diffraction in its older interpretation, density, deuterium exchange, infrared absorption, by dielectric constant, water sorption, and heat of wetting". Those in the second category include thallation, formylation, acid hydrolysis, and periodate oxidation". In the past there has been considerable disagreement on which method is best, although the X-Ray diffraction method of Segal et al.“ is widely accepted today. The structure of the unit cell of crystalline cellulose is depicted in Figure 2.4, with the lattice planes labeled. The 002 plane diffracts near 20 = 22.8°, while completely amorphous cellulose gives a diffracted minimum near 20 :- 18°. The crystallinity index (CrI) of Segal et al.“ is then defined as: CrI = {(I, - I,)/I,} x 100 (2-3) where I, is the maximum diffracted X-Ray intensity (near 29 = 22.8°) 21 tat I l / 002 plane: I ¢O.¢. d -. u i 4,1147%? l i .' 11 ones a a a: a ° ° fi I i I Figure 2.4: The unit cell of crystalline cellulose, showing the 002 plane. Figure is from Scallan [36]. 22 and I, is the minimum diffracted X-Ray intensity (near 20 = 18°). Ccllulases act very differently on crystalline and amorphous celluloses. Ryu et al." found that the affinity constant of the enzyme for amorphous cellulose was one-tenth of that for crystalline cellulose, while the maximum rate of hydrolysis was 40 times that for crystalline cellulose. Fan et al.“ showed an inverse linear relationship between Cd and glucose production from Solka Floc using T. reesei cellulase. Their later correlation", equation (2-1), has an exponent of 1.04 for the crystallinity term, indicating that crystallinity has a large effect on the enzymatic hydrolysis of a cellulosic substrate. Klesov and Sinitsyn“ hydrolyzed cotton linters with T. react and A. niger cellulases and found that the initial rate of hydrolysis increased as CrI decreased and as surface area increased. Lee and Fan” reported that surface area affected protein adsorption, while the effect of crystallinity was to decrease the effectiveness of the adsorbed enzyme. For lignocellulosic substrates, the dependence of the hydrolysis on CrI seems to be less pronounced. The small exponent of 0.257 in equation (2-2), reported by Gharpuray et a1." underlines the differences between microcrystalline cellulose hydrolysis and hydrolysis of lignocellulosics: surface area is the more important factor for lignocellulosics, while CrI dominates for microcrystalline cellulose. Fan and Lee", however, noted that although the initial surface area may be important in the initial stages of the hydrolysis, 23 as amorphous cellulose is removed, the crystallinity may become the overriding factor. Several groups""’ have reported that in the enzymatic hydrolysis of lignocellulosics, sometimes pretreated substrates with higher crystallinities than the untreated material gave higher rates and extents of hydrolysis. It was concluded" that an increase in crystallinity in the pretreated substrate was due to the removal of amorphous cellulose and that the CrI was probably not as important as previously thought. 2.4.3 Lignin Content Lignin, a complex aromatic polymer based on phenylpropane units, contains few hydrolyzable bonds and thus is not very susceptible to hydrolytic enzymes. The various types of monomer units and inter-monomer bonds in spruce lignin are shown in Figure 2.5. Together, lignin, hemicellulose, and cellulose form a very resistant material, both through association in the matrix and through chemical linkages”. The resulting material is resistant to both chemical and biological attack". The presence of hemicellulose also limits accessibility to the cellulose, although it is easily hydrolyzed and does not present as much of a barrier as does lignin”. Little is presently known about the role of hemicellulases in fungal wood degradationm’. Blanchette et at.“ has recently shown that hemicellulose is removed along with lignin during fungal degradation, suggesting that a similar accessibility problem to that for cellulases may exist for both hemicellulases and ligninases. Lignin and the 24 mean it, on HC— 0 HCOH CH1 "at“ c "a. HngH 0 H00 OM. Riga” :3 o 0 e _9H blitCOH CH; o———cu ‘— HC— —0 CH '6 '6 "C d) HC om “go ":90” £3) OH: OH H H -— Ohio OH MeO H,cou q oc/°\C.H, q Ht} CH H HCOH IO... OH o— mean o—QH H390}! OMe neon H 0 $ “M mo —¢ H 20H OMe "£0" ca 0 cute ' __ .. 3.7. ‘3’? i" .. . g 0 Me0¢— ‘0“. $0!“ —0 OH -0 Figure 2.5: Typical monomer units and inter-monomer bonds found in spruce lignin [18]. 25 compounds associated with it have been found to inhibit cellulase production, although no inhibition of in vitro cellulase aetivity was found”. Delignification enhances the enzymatic hydrolysis of softwoods more than for hardwoods"; this is possibly because softwoods are more heavily lignified than are hardwoods". Sudc et al.“ reported that the increase in the rate of cellulolytic hydrolysis in softwoods was small with initial removal of lignin, but increased greatly with increased delignification. Millet et a1." and Fan et al.”, using alkali- treated hardwood and wheat straw pretreated by several methods, respectively, reported that cellulose hydrolysis increased sharply with initial delignification and leveled off after about half the lignin had been removed. Stranks‘° suggested that the physical relationship between lignin and the carbohydrates is responsible for the effect of lignin, and not the actual amount of lignin. Saddler et at.“ agreed, stating that the effect of lignin on accessibility of cellulases is due to its distribution in the matrix rather than the lignin content itself. The complex association of lignin with the cellulose and hemicellulose polymers reported by Ruel et a1.“ would seem to support these suggestions. 2.5 Pretreatments Pretreatments to enhance enzymatic hydrolysis can generally be placed into two major categories: physical and chemical. Physical pretreatments, which include various types of grinding, milling, and 26 irradiation pretreatments, aim to physically modify the crystallinity, specific surface area, or both. Chemical pretreatments can be classed into three major types: (1) those which sclubilize the cellulose; (2) those which remove one or more matrix components to increase cellulase accessibility; and (3) those which swell the matrix. Some chemical pretreatments may fit into one or more of these classes, since often swelling of the matrix occurs when a matrix component is removed, as is the case with alkali treatments". Physical pretreatments are generally more effective with cellulosics, while chemical pretreatments are generally better with lignocellulosics. This is expected since cellulosics are very nearly pure cellulose and have only small amounts of other components that can be removed by chemical pretreatments; a possible exception is with the swelling action of alkali treatments. 2.5.1 Physical Pretreatments Ball milling, the most common form of physical pretreatment, decreases crystallinity and particle size, causing an increase in specific surface area and increasing water solubility". Mandels et al.“ found ball milling to be the most effective pretreatment for newsprint, although Wilke et al.“ found it to be too expensive in a process design study. In a study of ball milled hardwoods and softwoods, Matsumura et at.“ observed that the rate of hydrolysis increased as crystallinity and particle size decreased, although they concluded that it was the physical and chemical state of the lignin 27 that was the major factor inhibiting the hydrolysis. Millet et at." ball milled octton linters, red oak, and a variety of cther lignocellulosics and noted that the CrI decreased to zero as the milling time was increased. Complete enzymatic conversion was attained in ten days with cotton linters milled for one hour, and 93 % conversion was attained with red oak milled for four hours. Rivers and Emert", in a comparative study of wet and dry ball attrition of lignocellulosics, found that the terminal CrI (after enzymatic hydrolysis) decreased after wet ball attrition, but increased after dry attrition; even though the particle size was smaller for the dry milled substrates than for the wet, there was not a significant difference in their enzymatic conversion. This is possibly because they were lignocellulosics, since lignocellulosics exhibit less dependence on particle size and Cd than do cellulosics. Surface area in cellulosics is mainly external since the lignin and hemicellulose matrix has already been removed, and the remaining cellulose collapses to form a crystalline material. Lignocellulosics, on the other hand, contain most of their surface area in the interior of the matrix, in pores. Since ball milling increases external surface area only, only a small effect is expected for lignocellulosics. Another type of milling which is potentially useful in pretreating lignocellulosic is compression or roll milling. In one report, Tassinari and Macy" found that the extent of hydrolysis of wood chips increased 7 to 17 times after compression milling, while 28 the hydrolysis of newspaper was increased only modestly (2 times). The susceptibility of hardwoods and agricultural residues are increased more by compression milling than for softwoodsml, possibly because of plasticization of the softwoods due to their chemical content. Gharpuray et al.“ found a more than three-fold increase in the extent of hydrolysis with compression milled wheat straw, which accounted for about 75 % of that attainable with ball milling. Hammer milling does not disrupt the fiber structure, but chops the fibers into smaller pieces, creating only new ends. Thus, this type of milling has little or no effect on enzymatic hydrolysis“'”. Irradiation pretreatments are another class of physical pretreatment which shows potential utility. Kumakura and Kaetsu” pretreated rice straw, rice husks and cedar sawdust with electron beam irradiation at 2 x 10‘ rad/min and found that with a 5 x 10‘ rad exposure, 80-90 % yield of reducing sugars was attainable in all cases after 35 hours of enzymatic hydrolysis. Khan et al." found that with spruce chips, electron beam irradiation produced a substrate approximately equivalent to untreated filter paper cellulose; 90 % conversion was observed in 24 hours using T. r'eesei Rut-C30 cellulase. Pigden et at." pretreated forage cr0ps with y-ray irradiation and found that with 2.7 x 10' rad exposure, digestibility increased but that the dose was too small to solubilize the cellulose, while Han et al." y-irradiated various lignocellulosics in swelling 29 agents such as NaOH and ethylenediamine and found that mostly oligosaccharides and glucose degradation products were formed. Microwave irradiation has also been studied extensively“°"'". When the pretreatment is done in water, autohydrolysis occurs as acetyl groups are cleaved from the hemicellulose (so microwave treatments might also be termed acid treatments)". Ooshima et at." irradiated rice straw and bagasse of 84 to 94 % moisture content with microwaves and observed 1.6 and 3.2- fold increases, in enzymatic conversion, respectively. Azuma et al."'" pretreated various lignocellulosics with microwaves and found that the maximum enhancement of enzymatic susceptibility occurred in the temperature range 223-229 °C, regardless of the lignocellulosic source. Yields of 77 to 84 % were obtained using bagasse, rice straw, and rice hulls, while yields ranged from 65.9 to 93 % for a softwood, a hardwOod, and bamboo. A possible reason why they found no difference in the optimal enhancement temperature could be that the xylose was degraded to furfural, which then polymerized and precipitated, blocking the pores of the matrix. This is plausible, since they reported that b0th acidity and furfural production increased with increasing temperature". Other types of physical pretreatments used with lignocellulosic materials include freeze-explosion79'“ and cryomilling'm. 30 2.5.2 Chemical Pretreatments Chemical pretreatments are designed to increase accessibility to cellulolytic enzymes by removing a matrix component, swelling the matrix, dissolving the cellulose, or by a combination of these. Many different chemicals have been used to pretreat lignocellulosics. Five general classes of chemicals may be conveniently grouped: (1) alkali pretreatments; (2) acid pretreatments, including dilute mineral acids and autohydrolysis; (3) delignification; (4) cellulose solvation treatments; and (5) pretreatment with living organisms or in vitro enzymatic treatments. 2.5.2.1 Alkali Pretreatments Alkali pretreatments disrupt the structure of the lignocellulosic matrix by removing part of the lignin and hemicellulose, swelling the fibers and often decreasing the cellulose crystallinity". NaOH is the most common alkali used”'", although aqueous or liquid NH, is often dSI.33.84.“ 13.90 use Ammonia explosion" and supercritical ammonia pretreatments have also been used. Other systems include Ca(OH)2 and Na,CO,“, and alkaline peroxidem'”. With NaOH treatments, concentrations will commonly vary between 0.5 and 5 %, and up to 1 g NaOH/g substrate. Temperatures vary from room temperature up to 160 °C, with reaction times as short as 30 minutes or as long as 2 to 3 days. Higher temperatures allow for lower NaOH concentrations and shorter reaction times while increasing hemicellulose removal, often more so than lignin removal“. 31 The NaOH is normally removed by washing. NH, or NH,OH treatments provide an economical alternative to NaOH pretreatment, since NH, can be recovered by a simple evaporation process; however, they are not nearly as effective pretreatments as NaOH". Alkaline peroxide often gives increased enzymatic susceptibility over NaOH alone; in addition, room temperature is adequate and liquor recycle is possible". Mild alkali (less than 10 % NaOH) cause intercrystalline swelling, while stronger alkalis (greater than 10 % NaOH) tend to cause intracrystalline swelling”. Hemicellulose is removed by solubilization, rather than by hydrolysis of its glycosidic bonds’. Lignin removal appears to increase with alkali concentration up to a maximum removal and then level off". Alkalis have little effect on cellulose, with little or no removal at 5 % NaOH or less“. The major effect of alkali treatments is the hydrolysis of ester- type lignin-carbohydrate bonds, solubilizing the hemicellulose and swelling the matrix”. As an example, alkali treatment of a hardwood increased accessible protein size from a molecular weight of around 30,000 to 40,000-50,000”. The fiber saturation point, as measured using the nonsolvent water content technique”, increased from 40 % to 70-80 %. 2.5.2.2 Acid Pretreatments In a dilute acid pretreatment, hemicellulose is preferentially attacked, followed by amorphous cellulose". In contrast to alkali treatments, during acid hydrolysis the hemicellulose is removed by 32 hydrolysis of its glycosidic bonds; often, nearly complete removal of the hemicellulose can be attained”. Critical parameters in acid pretreatments include temperature (and pressure), time, and acid to solids ratio if the solids concentration is above 10 %. The rate of hemicellulose degradation increases with temperature; however, monomer degradation is highly time dependent, so reactor designs tend to use higher temperatures and shorter times"'°’. Actual conditions will vary with equipment design and the aim of the process“""‘°°“°’. Acid-treated cellulose is also susceptible to recrystallization“. During pretreatment, lignin is often depolymerized (from carbohydrate-lignin bond breakage) and then repolymerized’°"’°‘; some of this lignin can then be extracted with organic solvents“. Autohydrolysis, which is similar to dilute acid hydrolysis, occurs when the dry substrate is “cooked" under high temperature (high or low pressure) steam“"°"°"‘°’. The catalyst in this case is acetic acid, which is formed when acetyl groups are cleaved from the hemicellulose at high temperature’"°‘. Steam explosion is carried out on dry substrates at high temperature and pressure, but unlike autohydrolysis, the mixture is explosively decompressed to atmospheric pressure, causing the fibers to explode. According to Delong‘“, the beneficial effects of steam explosion include: (1) the reaction is rapidly quenched; (2) liquid is evaporated; (3) volatile by-products are flashed; (4) expansion of the 33 matrix and rendering of lignin separable by organic extraction. In the Iotech Process‘”, low moisture substrates are cooked at 185- 240 °C under steam, which plasticizes the hemicellulose and lignin; the cellulose is separated on explosive decompression. Reaction times are short, so little degradation of products occurs and high temperatures may be used‘“. Enhancement of enzymatic hydrolysis of Substrates pretreated by the Iotech Process is reportedly high. Steam explosion pretreatments have been used for a variety of substrates, including wheat straw‘°°'“°, sallow‘”, sugarcane bagassem'm, and aspen“"“‘. Morjanoff and Gray"2 found that by adding 1 % H,SO, to the reaction, they could get 83 % yield of glucose and 84 % xylose yield. Brownell et al."’, using a similar procedure, reported nearly quantitative recovery of all components of aspen wood. They concluded that preimpregnation with dilute acid not only improves the solubility of the pentosans, but it also allows for the use of lower temperatures, which destroys less of the pentosans and gives a better substrate for enzymatic hydrolysis. In addition, they found that the explosive decompression had no effect on the accessibility of the pretreated aspen to enzymatic hydrolysis and was therefore unimportant to the process’”. Ccllulases are inhibited by impurities in steam exploded wood, but washing of the pretreated wood removes the impurities and thus alleviates the problem”. 34 2.5.2.3 Delignification Delignification pretreatments aim to remove part or all of the lignin, and sometimes alter the distribution of the remaining lignin in the matrix. Kraft and sulfite pulping are commonly used in the pulp and paper industry to remove lignin; however, the purpose of these treatments is to recover cellulose fibers to use in paper, not to disrupt these fibers. Two general classes may be defined for delignifying chemicals used for pretreatments to enhance enzymatic susceptibility: (1) oxidizing agents, which oxidize the lignin and often destroy the crystalline structure of the cellulose"“"; and (2) organic solvents, such as amines or alcohols, with a catalyst to degrade lignin and hemicellulose (so-called "organosolv" processes)”. Chemicals used as oxidizing agents include NaClO,"”°, NaOCP'm, H,O, with a heavy metal ion such as Mn(IV), Mn(II). or Fe(II)"‘““”, ozonez'm'm, SO,"", and many others’. Very high extents of delignification can be achieved in addition to high removal of hemicellulose and little loss of cellulose using NaClO2 on rice straw‘”. One oxidation pretreatment which shows considerable promise is pretreatment with ozone. Ozone is attractive since‘“: (1) no toxic, acidic, or basic residues remain after pretreatment; (2) residual ozone is easily removed by conversion to 0,; (3) ozone can be cheaply generated as necessary on-site; and (4) ambient temperature and pressure are possible in most cases. In addition, particle size need act be smaller than 0.5 mm, and reaction times are 35 short (10-30 minutes with excess ozone). Using ozone pretreatments on wheat straw resulted in large amounts of lignin and hemicellulose removal""” and very high yields of glucose after enzymatic hydrolysis, on the order of 75 96”“. Binder et al.” found that they could increase enzymatic glucose yields after ozone pretreatment of wheat straw to 99 % by adding a step using concentrated H,PO, to dissolve the cellulose. Organosolv reagents consist of a catalyst such as an acid or base to degrade hemicellulose and lignin, and an organic solvent to extract the degraded fragments‘“. Acids such as H,SO.,“"”"m and acetic acid"" are commonly used, but often NaOH""’“ or other bases such as NH, are used‘“. Aliphatic and aromatic alcohols are often used as solvents“"”‘°”‘, although organic amines such as ethylenediamine, ethanolamine, and n-butylamine are often used since they act as both catalyst and solventm'm'm. Other catalytic solvents include NH,OH, EDTA (ethylenediaminetetraacetic acid), and DMSO‘”. Lee et al.“' found that at 95 °C, 97.4 % delignification of corn stover was possible using ethylenediamine, including significant removal of hemicellulose. Over 90 % delignification was also achieved using ethanolamine (93.2 %) and n-butanol with 0.5 % H,SO, and at 160 °C (93.9 %). Aromatic alcohols are generally more effective lignin solvents than aliphatic alcohols” , perhaps due to their similarities to the lignin monomer structures. High enzymatic conversions have been obtained with alkaline-ethanol pretreatment of 36 corn stover‘", and treatment of wheat straw with 50% ethylenediamine”. Since ethylenediamine is very toxic, however, its commercial use has proven to be limited; in this respect, catalySt- alcohol systems offer the advantages that they are relatively non- toxic and could be removed by evaporation and washing. 2.5.2.4 Cellulose Salvation Pretreatments According to Ladisch et al.‘”, cellulose solvents can be grouped into four classes: (1) strong mineral acids; (2) quaternary ammonium bases; (3) aprotic solvents; and (4) metal complexes. Cadoxen‘”, which consists of 25-30 % ethylenediamine, 70-75 % water, and 4.5- 7 % cadmium oxide, has been found to be very effective in selectively solubilizing the cellulose. Less toxic analogs of cadoxen, such as CMCS, and aqueous solution of 17 % sodium tartrate. 6.6 % FeCl,, 7.8 % NaOH, and 6 % sodium sulfite, show increased ease of recovery and often better enzymatic glucose yields than does cadoxen”. Strong mineral acids are also very effective cellulose solvents""‘”'“‘. Binder et at.“ pretreated wheat straw with concentrated H,PO4 and obtained 80 % conversion of the cellulose to glucose after enzymatic hydrolysis, and Szczodrak et 01)” obtained a high conversion of oak sawdust using 85 % H.,PO4 at 4 °C for four hours. Chen and Gong’“ dissolved avicel in ZnCl2 and 0.5 % HCl at 145 °C and found that complete enzymatic hydrolysis was achieved in 72 hours. 37 Although cellulose solvation pretreatments are very effeCtive, the costs of chemicals and their toxicity has prevented them from being used in practical industrial applications“. 2.5.2.5 Biological Pretreatments Biological pretreatments are plausible alternatives to physical and chemical pretreatments, since fungi and some bacteria possess the enzymes necessary to degrade lignin3 and hemicellulosem‘. An excellent review of the lignin-degrading enzymes of Phanerochaete chrysosporium has been published recently by Kirk’. Pretreatment of bagasse with a mutant strain of P. chrysosporium lacking cellulase was recently found to yield pulps that were highly susceptibly to cellulases‘”. Fungal pretreatments have also been found to increase yields in paper pulping‘“. Recent evidence”""137 suggests that the ligninases and hemicellulases also have accessibility problems; thus, if enzymes are used to pretreat the substrate, it is reasonable to expect that more effective pretreatments will use both types of enzymes. The field of biological pretreatments thus shows great promise for future applications, both for enhancing cellulolytic action and for the pulp and paper industry. CHAPTER III: MATERIALS AND METHODS 3.1 Pretreatment Substrate The substrate for all pretreatments was 60-mesh, air dried mixed hardwood (9O % birch, lO % maple) Type 060 obtained from Wilner Wood Products (Norway, MB). This untreated (hereafter referred to as "Native") mixed hardwood was presumed to have a homogeneous particle size distribution. 3.2 Acid Hydrolysis and Autohydrolysis Pretreatments Dilute acid hydrolysis (FRWA) and autohydrolysis (FR) pretreatment of Native mixed hardwood were performed in a high temperature plug flow reactor system‘” as described by Allen et at”. A schematic diagram of the reactor system used (MBI High Temperature Flow Reactor, Michigan Biotechnology Institute, Lansing, MI) is shown in Figure 3.1. The reactor is fed from a 20 L feed tank by a Moyno pump (moving cavity positive displacement pump, Robbins-Myer, Springfield, OH) with 18 stages, which was designed to handle slurries at up to 1650 psig and flow rates from 1 to 8 L/min. The reactor is constructed of‘0.5” o.d. stainless steel tubing with 0.048" wall thickness, with a total reactor volume of 190.1 mL. High temperatures are attained by the injection of steam through a series of drill holes near the inlet of the reactor. A 1 mm orifice is located near the end of the manor at which the reaction mixture is flashed from reactor pressure and temperature to atmospheric pressure (and 100 °C). The remaining steam is then 38 39 HIGH TEMPERATURE FLOW REACTOR ORHWCE STEAM INJECTOR a REACTOR F i I i } Passe: ,I/fig PRODUCT PRESSURE BOfl£R sax vac/u Figure 3.1: Schematic diagram of the MBI High Temperature Flow Reactor. 4o condensed and the pretreated slurry collected at the reactor (outlet. Temperature is controlled by steam addition, and residence time is controlled by mass flow rate of the slurry in conjunction with steam addition. Operating pressure is typically about 50 psi above the steam saturation pressure at the desired reactor temperature. Both pretreatments were carried out on 5.0 % (w/v) slurries, with 1.0 wt. % H2804 added for the FRWA treatment. Pretreatments were run at a nominal residence time of 7.3 seconds, although actual residence times were larger because higher flow rates of steam were needed to attain the higher temperatures. Run temperatures, slurry flow rates, and actual residence times for each pretreatment are given in Table 3.1. Each pretreatment was run at three temperatures to give varying degrees of treatment severity. Recovery and washing steps included filtration on Whatman No. 41 filter paper and storage at pH 1-2 under distilled water at 4 °C to preserve the pore structure of the wood (drying of the wood collapses its pores, which then are again inaccessible). The pretreated wood in each case was washed with distilled water until the filtrate was clear and colorless. 3.3 Ethylenediamine Pretreatment Organosolv pretreatment with ethylenediamine (EDA) was carried out using the conditions given by Lee et 01.”. Reaction was carried out on 15 gram batches of air dried Native mixed hardwood, which were mixed together after recovery and washing. Three EDA 41 Table 3.1: Pretreatment run conditions for the FRWA and FR pretreatment of Native mixed hardwood. Flow Rate Res. time Pretreatment T (°C) (kg/min) 1 (sec) FRWA 180.0 :1: 0.3 1.38 8.27 FRWA 200.0 :I: 0.1 1.36 8.39 FRWA 220.0 :i: 0. 5 1.34 8.51 FR 240.0 i 0.9 1.29 8.84 FR 260.0 + 2.1 1.21 9.43 - 1.0 FR 280.0 :I: 0. 5 0.90 12.7 42 concentrations were used, including 10 %, 50 %, and 70 % by volume of aqueous EDA, and all batches were run in a shaker bath at 90 i 3 °C and 175 rpm. Each batch was prepared for pretreatment by weighing (to the nearest 0.01 g) 15.00 g of Native wood into a 500 mL Erlenmeyer flask and adding 300 mL. of a solution of the desired concentration of aqueous EDA. The flask was then stoppered, mixed thoroughly, and allowed to presoak for one hour (1.5 hours for the 70 % EDA batches) at room temperature with occasional mixing. After the presoak was completed, the flask was transferred to the shaker bath, which was set at the desired conditions, and the reaction was allowed to take place for one hour. Since the resulting reaction liquor (containing EDA, lignin and hemicellulose) was viscous and thus difficult to filter, the reaction mixture was then centrifuged for 15 minutes at 10,000 rpm and the supernatant was discarded. After centrifugation, filtration was still difficult, so a rather lengthy washing and neutralization procedure was employed, in which the solids were washed cyclically with distilled water and 0.001 M HCl. On a per batch basis, this normally required a t0tal water volume of about 5 L and about 1.5 L of 0.001 M HCl. The criteria for wash termination were that the filtrate be clear and colorless and that the final filtrate pH 'be in the range 6.5-7.5 (each batch separately). When all the batches at a condition met these criteria, they were thoroughly mixed together and stored under distilled water 43 at 4°C (to preserve the pore structure). Later it was found that neutralization of the reaction mixture immediately after centrifugation and removal of the supernatant yielded a more easily filtered solid. Thus, the 10 % and 50 % EDAbatches were first neutralized (initial pH was around 13) with 4 M HCl and then washed as described earlier, but with only 5 L of distilled water and no acid. In all cases, filtration was performed using Whatman GF/C glass fiber filter paper (pore size of about 1.2 pm). 3.4 Alkaline Peroxide Pretreatment Alkaline peroxide pretreatment was performed according to the protocol established by Gould‘"”'122 and Gould and Freer‘”. Five percent (w/v) slurries were pretreated in 15 gram batches using H202 at a concentration of 2.5 % (w/v), or 0.5 g H,O,/g wood), 25 °C, pH 11.50, and 200 rpm in a shaker bath. Reaction times were varied to produce pretreated substrates with varying degrees of delignification and hemicellulose removal; the reaction times chosen were 0.5, 5, and 19 hours. Following pretreatment and washing, all batches at a particular reaction time were mixed together. One batch run consisted of the following steps. Fifteen grams of Native wood were weighed, to the nearest 0.01 g, into a 500 mL Erlenmeyer flask. Next, 300 mL of a solution of aqueous 2.5 % (w/v) H202 was added and the flask was stoppered and thoroughly mixed. The pH of the resulting slurry was then adjusted to 11.50 with the addition of approximately 22 mL of 8 N NaOH. 44 This adjustment caused the mixture to foam extensively; thus, the mixture was shaken and vented to the atmosphere periodically over a 3-5 minute period. The flask was then placed in the shaker bath, which was set at the desired conditions, and the reaction was allowed to take place for the desired reaction time. Upon completion of the run, the mixture was centrifuged for 20 minutes at 10,000 rpm, and the supernatant was discarded. The solids were then resuspended in distilled water and the pH of the slurry adjusted to 3.4-3.5 with the addition of about 15 mL of 4 M HCl. This pH adjustment decreased the water-absorbing capabilities of the pretreated wood’“, which made filtration much easier. The slurry was then filtered on Whatman GF/D glass fiber filter paper (pore size approximately 2.7 pm). Because the fibers were still quite hydroscopic, filtration still proved difficult. Thus, large volumes of distilled water (total of about 4 L per batch) and 0.0002 M HCl (total of about 0.5 L per batch) were used in a cyclical washing and neutralization procedure (water, acid, water, acid, etc.) using 150 mL aliquots of acid and 1 L aliquots of distilled water. The final pH of the filtrate was in all cases between 6.5 and 7.5, and in all cases the filtrate was clear and colorless after washing. All batches for a particular reaction time were mixed together thoroughly and stored under distilled water at 4 °C (to preserve the pore structure). 45 3.5 Cellulose Crystallinity The crystallinity of the cellulose in each sample was measured by X-Ray diffraction using the method of Segal et 01.“. Wet samples were first suction dried on filter paper and broken up into small pieces with a spatula. The samples were then filled into a cell mount and scraped with a spatula to give a smooth, flat surface which was flush with the surface of the cell mount, and ideally consisted of randomly oriented fibers. The samples were then exposed to X-Rays and the fraction diffracting from 29 = 12° to 28° were measured and stored in a data acquisition system. Samples were measured using the horizontal wide-angle goniometer section (D2155R6) of the Rigaku D/Max-RBx Horizontal X-Ray Powder Diffractometer (Rigaku/USA, Danvers, MA). Settings were at 45 kV and 16 mA with a 12 kW rotating copper anode. Crystallinities were expressed as Crystallinity Indexes (CrI) as defined by Segal et at.“ and as given in equation (2-3). 3.6 Substrate Composition Extractives determinations were made according to the TAPPI Standard Method T-12 (1959) with ethanol-benzene in a Soxhlet extractor apparatus. Approximately 500.0 mg of ground, oven dried sample was weighed, to the nearest 0.1 mg into a clean, tared coarse porosity fritted glass extraction thimble and extracted with 100 mL of 2:1 benzenezethanol by volume. Extraction was continued until no further change in sample mass occurred with further extraction. The 46 sample was then sucked reasonably dry under vacuum and dried at 105 °C to constant weight. The percent extractives was then calculated as: % Extractives = {(m, - m,)/m,] x 100 (3-1) where m, is the initial sample mass and mr is the final sample mass (both in grams). Carbohydrate and lignin compositions were determined using the Quantitative Saccharification Method of Saeman et 013“. In this procedure, unextracted lignocellulosic samples were subjected first to hydrolysis with 72 % H180, at 30 °C for one hour and subsequently with 4 % H280, at 121°C for one hour in an autoclave. The remaining insoluble portion was defined as Klasson lignin with extractives and ash. The liquor from the hydrolysis was measured for carbohydrates by HPLC, using a BioRad HPX-87P Carbohydrate column. The procedure was as follows. To begin, 500.0 i 50.0 mg of ground, oven-dried sample was weighed, to the nearest 0.1 mg, into an 18x150 mm test tube. Five milliliters of 72 % (w/w) H3SO4 were then added and the mixture was mixed thoroughly with a glass stir-rod (which was left in the tube). The sample was then placed in a 30 °C water bath for one hour and was mixed with the stir rod at 10, 20 and 40 minutes. At the end of one hour, the sample was transferred quantitatively to a 500 mL Erlenmeyer flask using 28 mL of distilled water per 1. mL of 72 % H180, originally used (140 mL of distilled water). The flask was 47 then covered with aluminum foil and autoclaved for one hour at 121 °C. The acid insoluble portion was quantitatively collected by filtration on a tared 30 mL coarse porosity fritted gooch crucible to which a 1.2 pm porosity Whatman GF/C glass fiber filter paper disk had been added to improve the filtration rate and recovery of small particles. Subsequent drying of the insoluble portion to constant weight in a 105 °C oven, cooling in a desiccator and weighing yielded the mass of Klasson lignin with extractives and ash. The liquor from the hydrolysis was quantitatively diluted with distilled water to 200.00 :I: 0.10 mL in a volumetric flask. Twenty milliliters of this solution were then stored in a leak-proof polyethylene bottle at 4 °C for further analysis. Determinations of the carbohydrate concentrations in the diluted hydrolysis liquor were made by HPLC using a BioRad HPX-87P column. Degassed HPLC-grade water (from a Millipore deionizing system) was used as eluent. Initial conditions for the chromatography were 0.400 mL/min and 83 °C, but as column performance (degree of separation) decreased, these were lowered to 0.380 mL/min and 80 °C. For a description of the HPLC apparatus and its components, refer to Figure 3.3 (page 55). Note that in this application, the zero dead-volume fitting shown in Figure 3.3 is replaced with the carbohydrate column and its precolumn system. The pre-column system consisted of carbohydrate deashing system (BioRad, Holder Cat. No. 125-0139, Cartridges Cat. No. 125-0118) 48 and a GC620 Guard Column System (Interaction Chemicals) which contained a lead-based ion-exchange resin similar to that of the chromatography column. Samples were prepared by first neutralizing 3mL with a scoop of PbCO3 (powder form, Fisher Scientific), centrifuging for 3 min at 14,000 rpm and filtering the supernatant through a 0.45 pm syringe filter into an autoinjector vial. Concentrations determined by HPLC were corrected for recovery losses incurred during the high temperature step by performing an additional experiment to determine the recoveries of the various wood sugars when subjected to this step in monomer form. In this recovery experiment, solid samples containing all five wood sugars were subjected to the autoclave step of the procedure and then treated as normal lignocellulosic samples. The fraCtional recoveries were then defined as: R, = m,,,/m,., (3-2) where R, is the fractional recovery of sugar i, m,,, mass of sugar i recovered from the experiment (measured by HPLC), and m,,, is the mass of sugar i initially present in the sample. In addition, in their polymer forms, wood carbohydrates are present in anhydro- forms, so to complete the material balance, the mass of each sugar must be multiplied by a "polymerization factor", as defined below: F, = (MW, - MW.)/MW, (3-3) where F, is the polymerization factor for sugar 1', MW, is the 49 molecular weight of sugar 1', and MW" is the molecular weight of water. For any wood sugar, then, the potential amount in the original sample, expressed as as percent, is: P, = {20 C,}/[W R1} (3-4) where P, is the percent p0tential sugar 1', C, is the diluted liquor concentration (g/L) of sugar 1', and W is the mass of original lignocellulosic sample (g). R, is used as defined in equation (3-2); note that R, must be between 0 and 1. It should also be noted that since the P, values are based on the masses of their free sugars and not their anhydro- forms, the material balance need not sum to 100 % or less. The carbohydrate contents, expressed in their anhydro- forms (for example, glucan instead of glucose), are then: P“, = P, F, (3-5) where PM is the percent potential sugar monomer i, and F, and P, are as defined in equations (3-3) and (3-4), respectively. The values of PW, then, must sum to 100 % or less. When combined with the percent lignin with extractives and ash, calculated by dividing the mass of the acid insoluble fraction by the initial sample mass and multiplying by 100 %, these values complete the material balance (neglecting uronic acid content and recovery errors). 3.7 Solute Exclusion Technique Measurements of the water in pores inaccessible to solutes of various molecular sizes were made according to the solute exclusion 50 technique of Stone and Scallan”""". In this technique, a known mass of aqueous solute solution is added to a wet presscake and allowed to diffuse into accessible pores. The change in concentration is then measured and the inaccessible water is determined by a solute balance before and after contacting the wet pulp to give: 5, p = (w + q) - w (C,/Cg) (3-6) where 8, is the specific mass of inaccessible water for solute of size 1' (g inaccessible water/g wood), w is the mass of solute solution added (g), q is the mass of water in the wet presscake (g), p is the dry weight of presscake (g), C, is the initial solute solution concentration (%). and C, is the final solute (diluted) concentration, corrected for minor water soluble compounds in the presscake (%). In all cases the density of water at ambient conditions was assumed to be 1 g/mL and thus the volume of water inaccessible to a solute of size 1' is: d, = 8, (3-7) Dextrans ranging in molecular weight from 6000 to 2 x10‘ and having equivalent spherical diameters in solution“ from 38 to 560 A were used as solutes. In addition, four low molecular weight carbohydrates (see Table 3.2) ranging in diameter from 8 to 12 A were used to supplement the dextran size range. Solute concentrations were 0.5 % or 2 %, prepared by dissolving 0.5 or 2 g of the anhydrous solute in 100 mL of distilled water (without heating) and storing overnight at 4 °C. Solute solutions were made 51 Table 3.2: Molecular probes used in the solute exclusion technique. Estimated Diameter‘ Solute Vendor M.‘ MJM,‘ (A) Glucose Sigma 180 1.0 8 Fructose Sigma 180 1.0 8 Maltose Fisher 342 1.0 10 Raffinose Sigma 504 . 1.0 12 Dextran 6K Fluka 6,000 N.A.” 38 Dextran T10 Pharmacia 10,500 1.74 51 Dextran 15-20K Polysciences 17,500' N.A. 61 Dextran T40 Pharmacia 38,800 1.60 90 Dextran T70 Pharmacia ' 72,200 1.88 118 Dextran 200-300K Polysciences 2.50x10" N.A. 204 Dextran T500 Pharmacia 5.07x10‘ 2.16 270 Dextran T2000 Pharmacia 2.0x10‘ N.A. 560 a Obtained from vendor’s lot analysis except where indicated by (*), where M, is calculated as the midpoint of the molecular weight range reported. b N.A. = Not Available c Probe diameters estimated using the values given in the original work by Stone and Scallan [28] and from Weimer and Weston [71]. 52 fresh before each experiment. Next, wet pulp samples sufficient to represent one gram dry weight of fibers were weighed, to the nearest 0.0001 g, into 20 mL polyethylene scintillation vials with leak-proof caps (Baxter Scientific, Cat. No. R2555-3). Approximately 20 mL of solute solution was then quantitatively added (by mass) and the bottle was tightly capped. The bottles needed only to be air-dried, since no difference in mass was observed between b0ttles before and after handling. Also, since the solute solutions were at 4 °C at the time of addition to the bottles, the bottles (with samples and solutes) were allowed to come to room temperature before being weighed. Three replicates (bottles) were used per solute per sample. Three bottles containing only distilled water and wet pulp were also prepared as "blanks," controls used to subtract out the minor contributions of water soluble extractives from the wOod to the final concentration measurements. The validity of this method was proven with a preliminary experiment which showed that refractive indexes (when taken relative to water) and optical rotations are additive. The final concentration measurements were made first by refractive index and later by optical rotation. After equilibration overnight with occasional vigorous shaking, the wet presscake from each bottle was collected quantitatively on a tared 30 mL coarse porosity fritted gooch crucible as shown in Figure 3.2. The liquid fraction was collected in a test tube, which was then removed from the vacuum 53 30 mL Coarse Fritted lass Gooch Crucible Rubber Seal \\\\\\\\\\\\\\\\\\\\\\‘ Glass Crucible Holder —""’ to vacuum ._ 500 mL Vacuum Flask A r \r Test Tube \ 1 Figure 3.2: Separation apparatus for pulp and liquid in the solute exclusion technique. 54 flask. Care was taken to collect liquid samples in 15 seconds or less to prevent evaporation of the sample under vacuum‘“. The presscake in the gooch crucible was then washed free of solutes and dried to constant weight at 105 °C. In cases where filtration was very slow (5 mL/min or less), the solids were not washed, but dried immediately and their dry weights corrected for solutes by the equation: 9 = {p’ - wC.(1 - [a./a.])}/{l + (wCJa.) (a./p).....l (3-8) where p’ is the dry weight of presscake and residual solute (g), on, is the optical rotation of the final (diluted) solute solution (°), a, is the optical rotation of initial solute solution (°), and (on/p)“, is the average of (0t,/p) for the blanks (°/g), and p, w, C, , and C, are as defined earlier. This equation is derived in Appendix A. The solute concentrations before and after the procedure were measured initially by refractive index as described by Lin er al.“"“‘; however, the variances in readings obtained with this method were large. The measurement apparatus is shown in Figure 3.3. Since it is a flow system, the high viscosities of the 2 % dextran solutions necessitated the use of 0.5 % solutes, but even this did not improve reproducibility of readings. Thus, optical rotations and 2 % solutes were used in later experiments. These were measured with a precision polarimeter (Carl Zeiss, West Germany) which was accurate to d: 0.01°. A sodium lamp (589.1 nm) was used. Final concentrations were corrected for water soluble extractives by subtracting out the specific reading per unit mass of 55 li'j effluent . Water Reservoir Model 2350 HPLC Pump (Isco, Inc., Lincoln, NE) . Autoinjector AN-728 (Anspec, Inc., Ann Arbor, MI) . VICI Injection Valve (Valco Instruments, Inc., Houston, TX) Zero dead-volume fitting ERC-7522 Differential Refractive Index Detector (Erma, Inc., Tokyo, Japan) . Waters 745 Data Module (integrator), (Millipore Corp., ‘Waters Chromatography Div., Milford, MA) 0 mpuop> Figure 3.3: Schematic diagram of the modified HPLC apparatus used to measure concentrations of solute solutions. 56 wood from the blanks: R... = R: - P (R./p).i... (3-9) where R,° is the corrected final solution reading (refractive index or optical rotation), (units), R, is the actual final solution reading (units), p is the dry weight of the sample presscake, and (R,/p),,,,,,, is the averaged correction from the bl_ar_1_k_s_, (units/g). The solutes (dextrans) were assumed to be spheres in solution‘" and thus they represented pore diameters when compared with their inaccessible water values. The resulting pore volume distributions were fit, using the program. EUREKA (Borland International, Scotts Valley, CA), to the equation’“: d(t) = K [1 - e'“ [cosh(Bt) + C sinh(Bt)}] (3-10) A = CIT (3-11) B = (C1 - mm ("a-12) C = C/(C2 - -1)°'5 (3-13) t = log,,,D - log,,,4 (3-14) where D is the molecular diameter (pore diameter, A), d is the inaccessible water (mL/g wood), and K, C, and 1: are constants. This equation describes the response of a second order overdamped (C > 1) system to a unit impulse input and has the desired properties; namely, a sigmoidal curve which increases monotonically. Since the pore size distributions represent increases in pore volume with increasing pore diameter, the curves muSt monotonically increase. Water, with a diameter of 4 A is assumed to be completely accessible 57 to all pores; thus, d = 0 at D = 4 A. Assuming that the pore geometry is adequately described by the parallel plate model‘“, which is represented in Figure 3.4, the surface area in the pores can be calculated by the following equation“: AA,,, = (2 AV,_,)/w,., (3-15) where AA,,, is the specific surface area in pores in the size range 1'- j (m’lg), AV,,, is the increase in pore volume over the pore size range i-j (m’lg), and w,., is the average pore width over the pore size range H (m). As defined, w,., is the average of w, and w,, the pore widths at sizes i and j, respectively (equal to i and j). By summing the values of AA over a range of pore widths, the surface area in pores within that size range can be calculated. A molecule of diameter 560 A is assumed to be completely inaccessible”, so the sum of AA over 4 to 560 A represents the total surface area. Likewise, summing from 51 to 560 A gives the surface area in pores larger than 51 A. 3.8 Enzymatic Hydrolysis Enzymatic hydrolysis of all pulp samples was carried out in stoppered 250 mL Erlenmeyer flasks with crude cellulase powder from Trichoderma reesei (Rutgers C30, U.S. Army Natick Research and Development Center, Natick, MA) with a reported cellulase activity of 0.38 FPU/mg enzyme powder (0.408 FPU/mg dry enzyme powder). In addition, the cellulase, which has very low levels of B-glucosidase’, was supplemented with Novozym TN 188, a liquid B-glucosidase from Aspergillus niger (Novo Industries, Denmark) with 58 AA / % Figure 3.4: Parallel plate pore geometry, where the plates are microfibrils in the lamella of the cell wall [142]. 59 a B-glucosidase activity of 588 Units/mL. This was done to minimize cellobiose accumulation, which not only lowers the measured glucose levels, but also inhibits the cellobiohydrolase activity of the T. reesei cellulase complex“. Hydrolysis was performed at the optimum“ conditions of 50.0 °C and pH 4.80. The reaction mixture, preheated to 50 °C by the time of enzyme addition, consisted of citrate buffer (50 mM, pH 4.80, 3.08 mM in sodium azide to inhibit bacterial growth), Rutgers C30 cellulase (40.8 FPU/g substrate), Novozym TN 188 B-glucosidase (40.8 Units/g substrate), and 2 % (w/v) oven-dried equivalent of wet substrate. The high enzyme to substrate ratio ensured that there was no rate limitation due to lack of enzyme, and was determined by a series of experiments with the most accessible substrate (FRWA 220 °C) which were designed to find the maximum enzymatic activity (defined at 1.1.... the maximum initial rate). The value of 40.8 FPU/g substrate used was above the maximum amount of enzyme necessary to achieve maximum enzymatic activity. The cellulase used in these experiments was Novo CCN 3000 (Novo Industries, Denmark), a liquid cellulase from T. reesei with a reported cellulase activity of 52.1 FPU/mL. A comparison between the CCN 3000 cellulase at 49.6 FPU/g wood and the C30 cellulase at 38.8 FPU/g wood gave nearly identical initial rates and thus 1 unit of C30 is equivalent to 1.28 units of CCN 3000. Reaction was carried out at 200 rpm in a shaker bath. Homogeneous 2 mL samples were withdrawn at time intervals of 0.5, 60 l, 2, 4, .12, and 24 hours and quenched by adding 10 1.1L of 72 % H280, (this stops the reaction by precipitating the enzyme). The samples were then neutralized with a scoop of BaCOa (powder, Mallinckrodt) and analyzed for glucose with a YSI enzymatic glucose analyzer (YSI Model 27, Yellow Springs Instrument Co., Yellow Springs, OH). Cellobiose was measured by HPLC with a BioRad HPX-87H column using 0.008 N H280, as eluent. With equivalent activities of CCN 3000 and Novozym TN 188, it was reported’“ that no cellobiose was formed in the hydrolysis of Sigmacell 20; since equivalent activities of cellulase and B-glucosidase were used in this study (in terms of equivalent C30), cellobiose was measured only in selected early samples. No detectable cellobiose was found in any of the samples from this study. Glucose yield from the enzymatic hydrolysis was calculated as the percent of potential glucose in the original sample which was formed as a result of the hydrolysis, or (glu = glucose): % Yield = (g glu formed)/(g potential glu)x100 (3-16) CHAPTER IV: RESULTS 4.1 Pretreatments Pretreatment with alkaline peroxide resulted in substrates which appeared to have markedly reduced particle sizes, as evidenced by the passage of large amounts of small particles through the filter paper (maximum mean pore diameter of 2.7 um). Native wood, when washed, passed no particles through the filter paper (Whatman GF/D). In addition, the water absorbancy of the alkaline H202 pretreated wood increased markedly. This point will be made more clear when the fiber saturation points for all the pretreated substrates are shown. Because of this increase in water absorbancy, filtration of the pretreated wood was extremely difficult; the wood tended to load the filter paper rapidly. Only adjustment of the slurry pH from around 13.5 to a slightly acidic pH helped increase the speed of filtration. Alkaline peroxide pretreated wood was whitish in color due to the bleaching effect of peroxide (a similar process is used in the pulp and paper industry to bleach paper pulp). Degree of bleaching generally increased with pretreatment time (see Figure 4.1(a)). Ethylenediamine (EDA) pretreated wood was reddish-brown in color (Figure 4.1(b)), and reaction liquors were nearly black. This pretreatment produced substrates which were difficult to filter due to the EDA remaining in the pores of the pretreated wood (EDA is relatively viscous), and also because the slurries were at a very high pH (about 14). Removal of the EDA by centrifugation and 61 62 (a) (b) Figure 4.1: Comparison of the physical appearances of pretreated mixed hardwood (wet) with Native wood: (a) Alkaline H103, 2.5 %, 25 °C, 200 rpm, (b) EDA, 90 °C, 1 hr, 175 rpm. 63 subsequent neutralization of the slurry resolved the filtration problems. With 50 % and 70 % EDA concentrations, excess EDA was present, since EDA vapors were observed in the head space of the reaction flasks after reaction was completed. With 10 % EDA, however, no EDA vapors were observed, suggesting that the EDA had completely reacted; since EDA is both the reactant and the solvent, this brings into question whether the lowered molecular weight lignin molecules formed in the latter stages of the reaction were extracted or just redistributed. Autohydrolysis pretreatment in the plug flow reactor (FR) formed dark brown substrates, with darker color as temperature increased (Figure 4.2(a)). Lignin has a softening temperature below 240 °C; since this was the lowest autohydrolysis temperature, lignin was melted and separated from the matrix at all three temperatures. This separate lignin-rich fraction was easily separated by its preferential loading on the surface of the filter paper. Filtration was quite easy with Whatman No. 41 filter paper once this fraction, which appeared to be less dense than the remaining lignocellulosic matrix, was removed. Dilute acid hydrolysis in the plug flow reactor (FRWA) produced substrates ranging in color from reddish (180 °C) to very dark brown (220 °C) with increasing temperature (Figure 4.2(b)). As temperature was increased, the pretreated slurries formed emulsions which were very difficult to filter. The solids pretreated at 200 °C (a) (b) Figure 4.2: Comparison of the physical appearances of pretreated mixed hardwood (wet) with Native wood: (a) Autohydrolysis, (b) Acid hydrolysis, 1 % H,SO,. 65 and 220 °C resembled a thick, muddy substance. 4.2 Substrate Chemical Compositions Results of the carbohydrate recovery experiment are tabulated in Table 4.1, including polymerization factors for each wood sugar. These values were used with equations (3-3) and (3-4) to calculate carbohydrate compositions in their anhydro- and free sugar forms from measured concentrations as described in Section 3.6. The carbohydrate, lignin and extractives contents of Native mixed hardwood are given in Table 4.2. All components were present in amounts that are typical for this type of hardwood”. ..The mass balance accounts for only 85.1 % of the total 100 % dry weight of mixed hardwood; the remaining 14.9 % is attributed to unknown glucuronic and galacturonic acid contents (from the hemicellulose) and to recovery errors in the technique. The mass balance, however, is comparable to other published data". The compositions of all pretreated substrates are listed in Tables 4.3(a) through 4.3(d). The ratios of glucan/lignin (G/L) and glucan/hemicellulose (G/H; hemicellulose was calculated as the sum of the compositions of sugars other than glucose) ratios for all pretreatments are given in Table 4.4. For the alkaline peroxide pretreatment (Table 4.3(a)), glucan content increased steadily with pretreatment time, while hemicellulose decreased (increasing G/H) slowly at short treatment time, then faster as treatment time was increased. Lignin content decreased (increasing G/L) more quickly in 66 Table 4.1: Fractional recoveries (R,) and polymerization factors (F,) for the wood carbohydrates. Wood Sugar (1') R, F, Glucose 0.927 0.90 Xylose 0.788 0.88 Mannose 0.904 0.90 Galactose 0.882 0.90 Arabinose 0.888 0.88 Table 4.2: Composition of Wilner 060 Native mixed hardwood. Potential Composition' Component (%) Carbohydrates Glucan 38.6 i 0.4 Xylan 15.5 :1: 0.1 Mannan 1.9 :l: 0.0 Galactan 1.1 i 0.1 Arabinan 0.6 i 0.1 Other Extractives 2.5 :l: 0.1 Klasson lignin 24.9 :i: 0.2 with ash SUM 85.1 i 0.5” a Based on 100 % dry weight of material. b Remaining fraction attributed to unknown uronic acid content and to recovery errors in analysis techniques. 67 Table 4.3(a): Compositional analyses of alkaline peroxide pretreated wood. Composition' (%) Potential Component 1/2 hr 5 hr 19 hr Glucan 41.6 :t 0.4 46.6 :i: 0.2 49.7 :i: 0.7 Xylan 16.8 i 1.4 16.6 i 0.3 12.7 i 0.3 Mannan 2.0 :l: 0.1 1.9 :I: 0.1 0.9 :t 0.0 Galactan 1.0 i 0.0 0.9 :l: 0.1 0.9 :t 0.0 Arabinan 0.6 :i: 0.1 0.5 :l: 0.0 0.5 i 0.1 Extractives N.A.” N.A. 0.5 i 0.1 Lignin with ash 25.6 :i: 0.4 24.3 i 0.4 24.5 :1: 0.1 & extractives SUM° 87.6 :1: 1.5 90.8 :i: 0.6 90.1 :1: 1.2 Table 4.3(b): Compositional analyses of EDA pretreated wood. Composition' (%) Potential Component 10% EDA 50% EDA 70% EDA Glucan 43.3 :1: 0.6 48.2 :t 1.1 51.8 :1: 2.5 Xylan 16.3 :i: 0.2 12.0 :t 0.4 9.0 :l: 0.7 Mannan 1.9 :l: 0.1 1.9 :i: 0.2 2.0 :l: 0.2 Galactan 1.4 :1: 0.1 0.8 :l: 0.5 1.4 :1: 0.1 Arabinan 0.5 :t 0.2 0.4 :I: 0.0 0.8 :l: 0.0 Extractives N.A.b N.A. 0.4 i 0.3 Lignin with ash 24.9 :1: 0.3 24.4 :1: 0.2 24.9 :i: 0.1 & extractives SUM“ 88.3 :I: 0.7 87.7 :t 1.3 89.9 :1: 2.6 a Based on 100 % dry weight. b Extractives not determined. c Remaining fraction attributed to unknown uronic acid content and to recovery errors in analysis techniques. 68 Table 4.3(c): Compositional analyses of autohydrolysis (FR) pretreated wood. Composition' (%) Potential Component 240 °C 260 °C 280 °C Glucan 43.8 i 1.9 60.0 :t 0.3 66.3 :1: 3.0 Xylan 12.7 i 0.7 7.9 i 0.0 3.9 i 0.1 Mannan 1.4 i 0.1 1.4 :l: 0.1 0.9 :I: 0.1 Galactan 1.2 :i: 0.5 1.5 :l: 0.1 0.7 i 0.2 Arabinan 0.4 :i: 0.0 0.5 i 0.1 0.4 :l: 0.0 Extractives N.A.b N.A. 4.8 i 0.5 Lignin with ash 26.0 i 0.2 22.8 i 0.1 19.2 :1: 0.1 & extractives SUM° 85.5 i 2.1 94.1 i 0.4 91.4 :i: 3.0 Table 4.3(d): Compositional analyses of acid hydrolysis (FRWA) pretreated wood (1 % H,SO,). Composition' (%) Potential Component 180 °C 200 °C 220 °C Glucan 46.8 :I: 1.1 52.5 i 2.0 57.6 :1: 2.5 Xylan 9.8 :l: 0.4 3.1 :i: 0.3 1.8 :i: 0.4 Mannan 0.8 :l: 0.2 0.6 :l: 0.1 0.4 :1: 0.0 Galactan 1.0 :l: 0.1 1.3 i 0.2 0.6 :l: 0.2 Arabinan 0.4 :l: 0.0 0.2 :t 0.1 0.2 i 0.0 Extractives N.A.” N.A. 13.9 :1: 0.2 Lignin with ash 28.0 :i: 0.3 29.7 i 0.3 33.0 :1: 0.2 & extractives SUM° 86.8 i 1.2 87.4 :1: 2.1 93.6 :t 2.5 a Based on 100 % dry weight. b Extractives not determined. c Remaining fraction attributed to unknown uronic acid content and to recovery errors in analysis techniques. 69 pretreated substrate after pretreatment. Table 4.4: Relative amounts of lignin and hemicellulose in each Pretreatment G/H‘ G/Lh Native 2.02 1.55“ H203 : 0.5 hr 2.04 1.63 5 hr 2.34 1.92 19 hr 3.13 2.03 EDA : 10 % 2.15 1.74 50 % 3.19 1.98 70 % 3.92 2.08 FR : 240 °C 2.79 1.68 260 °C 5.31 2.63 280 °C 11.2 3.45 FRWA: 180 °C 3.90 1.67 200 °C 10.1 1.77 220 °C 19.2 1.75 a G/H a Glucan/Hemicellulose ratio. b G/L a Glucan/Lignin ratio. c Extractives have been subtracted out for Native. All others include extractives. 70 the initial stages of treatment, slowing at longer treatment times. For the EDA pretreatment (Table 4.3(b)), glucan increased with EDA concentration, while hemicellulose removal increased steadily with increasing EDA, and lignin removal increased up to 50-70 % EDA and appeared to level off around 70 %. Autohydrolysis (Table 4.3(c)) produced the substrates with the highest glucan content, probably due to the large amounts of hemicellulose and lignin removed; G/H was nearly four times those for peroxide and EDA at the most harsh conditions, while G/L was 1.5 times higher at the same conditions. Acid hydrolysis (Table 4.3(d)) removed nearly all the hemicellulose at 220 °C; this removal decreased with temperature. Only a little lignin was removed in the acid hydrolysis pretreatments. Extractives were determined only for the most harsh level of each pretreatment; the highest amounts of extractives were for FR (280°C) and FRWA (220°C), but the increase in extractives was attributed to the production of smaller molecular weight, organic solvent extractable lignin and was thus ignored in the acid insoluble portion from the quantitative saccharification procedure (which represents lignin with extractives and ash). Ash content was also ignored, since very little ash is generally present in hardwoods. Extractives in the EDA and peroxide pretreatments were very small (Tables 4.3(a) and (b)), and thus were ignored. Values for pOtential glucose, which will be used later, have 71 been tabulated for all substrates in Table 4.5. Recall from Section 3.6 that these are the free sugar forms, so when combined with the Other compositions, they need not add to 100 %. 4.3 Substrate Crystallinities The X-Ray diffractogram for Native mixed hardwood (wet cake) is shown in Figure 4.3. The intensities I, and I2 are labeled, and are measured as the absolute intensities at the points shown. The CrI for Native wood is thus 71.6. Values for Cd for all pretreated substrates and Native are given in Table 4.6. The trends for all substrates with increasing pretreatment harshness were to increase CrI, with the exceptions of EDA (70 %). H202 (0.5 hr) and H,O, (5 hr), which were all slightly lower than the CrI for Native (69.0, 68.4, and 69.0, respectively). The highest CrI values were for FR (260 °C) and FR (280 °C), at 85.2 and 85.8, respectively, and closely followed by FRWA (200 °C) and FRWA (220 °C), at 82.0 and 82.3, respectively. This was expected because of the nature of acid pretreated cellulose being more susceptible to recrystallization. Compared to the crystallility of Native wood, however, these represent only modest increases in CrI. 4.4 Pore Size Distributions and Surface Areas The pore size distribution for Native mixed hardwood, determined as described in Section 3.7, is shown in Figure 4.4. All data points are shown; results from both refractive index measurements and optical rotation measurements (different Table 4.5: Potential glucose amounts in each pretreated substrate. 72 Potential Pretreatment Glucose (%) Native 42.9 i 0.4 H,O2 : 0.5 hr 46.2 :I: 0.4 5 hr 51.8 i 0.2 19 hr 55.2 :1: 0.8 EDA : 10 % 48.1 :i: 0.7 50 % 53.6 i 1.2 70 % 57.6 i 2.8 FR : 240 °C 48.7 d: 2.1 260 °C 66.7 :t 0.3 280 °C 73.4 i 3.3 FRWA: 180 °C 52.0 :1: 1.2 200 °C 58.3 :I: 2.2 220 °C 64.0 i 2.8 73 500 ., NATIVE MIXED Hanouooo 05-.5. ss-s. 95-.15. nsn-.s C .. xv-4s. HA-tl N rs-tas.5 ; J 12-51.54 400 t l s E 0 c 300 0 1: 20° 0- too .. l ,4,me ‘2 15 18 ' 20 22 24 26 28 2 “TA RIGAKU / USA Figure 4.3: X-Ray diffractogram for Native mixed hardwood. Table 4.6: Measured values of Cd for all substrates. 74 I, I2 CrI‘ Pretreatment (22.8°) (18°) (%) Native 181.5 51.54 71.6 H,O2 : 0.5 hr 231.5 73.08 68.4 5 hr 255.4 79.23 69.0 19 hr 219.2 50.00 77.2 EDA : 10 % 238.5 65.38 72.6 50 % 294.6 73.08 75.2 70 % 161.5 50.00 69.0 FR : 240 °C 217.7 47.69 78.1 260 °C 296.9 43.85 85.2 280 °C 336.9 47.69 85.8 FRWA: 180 °C 251.5 53.08 78.9 200 °C 243.1 43.85 82.0 220 °C 316.9 56.15 82.3 a CrI = {(I, - I,)/I,} x 100 75 1.50 A Q _J 1.25 E v 1;,- 1.00 4.4 O 3: _,_, 0.75 :9 U) 8 0.50 O 0 O E 0.25 .0". 0000' I I IUIjTI‘Il I I IIIIIII U Tfi fIIIl .1 10 100 1000 Molecular Diameter (A) Figure 4.4: Pore size distribution for Native mixed hardwood. 76 experiments) were used. The curve shown is the best fit curve according to equations (3-10) through (3-14). The constants from equation (3-10) for each substrate, as well as their respective mean square residual errors, are given in Table 4.7. In all cases the residual errors were small. The exceptions were FRWA (200 °C) and FRWA (220 °C), in which the data showed too long of a lag for the type of curve used to fit the data for the other pretreatments. In these two cases, the curves were hand-drawn and the mean square residual errors calculated as before. The residual errors for the hand-drawn curves were in the range of those for the best fits, so the curves were presumed to be adequate. A solute of 560 A was assumed in Section 3.7 to be completely inaccessible to any of the pores. Thus the curve in Figure 4.4 represents a cumulative pore volume distribution; the total pore volume (found at 560 A) is 0.764 mL/g wood. It is evident that the majority of the pore volume in Native mixed hardwood is in small pores. Since it is the surface area in these pores that the enzyme must get to in order to catalyze the reaction, it is reasonable to expect that any pretreatment which increases the accessibility of this surface area to the enzyme must increase the pore size. This effect was observed for all of the different pretreatments, as shown in the following figures. In Figure 4.5, the porevolume distributions for the three levels of alkaline H202 pretreatment are shown. The effect of early Table 4.7: Best fit curve parameters and mean square residual errors for all substrates. MSE' Pretreatment K C 1 (x102) Native 0.774 2.300 0.111 0.516 H30, : 0.5 111‘ 0.883 1.004 0.319 0.513 5 hr 1.491 1.002 0.432 1.769 19 hr 1.490 1.005 0.461 0.949 EDA : 10 % 1.106 1.191 0.276 1.542 50 % 1.046 1.147 0.432 1.034 70 % 1.694 1.002 0.611 0.883 FR : 240 °C 0.873 1.034 0.389 0.538 260 °C 1.066 1.779 0.241 0.498 280 °C 1.193 1.251 0.366 0.655 FRWA:180 °C 0.785 1.029 0.343 0.490 200 °C N.A.” N.A. N.A. 1.130 220 °C N.A. N.A. N.A. 0.771 a MSE a Mean Square Residual Error, and is defined as: MSE = [2(y’ - y.)’]/[n - pl where y’ is the predicted response, the y, are the data points, it is the number of points and p is the number of constants in the model. b Not applicable since these curves were hand-drawn. 78 1 .25 (a ) 1.00 0.75 0.50 0.25 d, lnaccessible Water (mL/g) 0.00. - - .--.-.-, . .- 1 10 100 1000 Molecular Diameter (A) (b) 1.00 0.50 0.25 d. lnacceesible Water (mL/g) 0.00 v vvva v v v v vv‘vv‘ v v v vvv‘v' 10 1 W 1 000 Molecular Diameter (A) ‘d 1.50 e 1.23 (c) 1.00 0.73 0.50 0.23 d, Inaccessible Water (mL/g) a.” V V V 'V V V V V'V' V V V V 'V ' V "l V ' I 10 100 1000 Molecular Diameter (A) Figure 4.5: Pore size distributions for the alkaline H20, pretreat- ments. (a) 0.5 hr, (b) 5 hr, (c) 19 hr. 79 delignification from 0.5 hours of reaction to 5 hours is seen as a large increase in pore volume, especially above 12 A pore diameter. Not much change occurs in the hemicellulose removal portion of the reaction (5 to 19 hours). Thus there is not much difference in the pore volume distributions of 5 hour and 19 hour alkaline H202 pretreatments, except for a slight decrease in the amount of pore volume in the smaller pores. Figure 4.6 shows the pore volume distributions for the three levels of EDA pretreatment. Trends in the pore volume distributions for the EDA pretreatments are not very similar to those for alkaline H302. Pretreatment with 10 % EDA, which resulted in removal of both lignin and hemicellulose in comparable amounts (Table 4.4) increased pore volume markedly. However, pretreatment with 50 % EDA (greater hemicellulose removal) actually decreased tatal pore volume while increasing the volume in larger pores relative to the 10 % EDA treatment. With 70 % EDA, total pore volume was increased to a great extent, and the majority of pore volume was then in large pores. For the FR (autohydrolysis) pretreatments, Figure 4.7, pore volume did not increase as much as would have been expected, considering the amounts of hemicellulose and lignin removed, and in light of the results for the alkaline H,O2 and EDA pretreatments. A steady increase in pore volume is seen at all pore diameters with increasing temperature of pretreatment, but not to the degree of that 80 1.50 1.25 (a) 1.00 0.75 0.50 0.25 d, inaccessible Water (mL/g) V v VVVVV . . 00m ? vavvvt v v v vvvv 10° 1000 Molecular Diameter (A) ‘- 1.50 1.15 1.00 (b) 0.75 d, lnacceesible Water (mL/g) 1.50 1.2: (e) ,,,, 0.73 0.50 0.25 d. lnaccessible Water (mL/g) Molecular Diameter (A) Figure 4.6: Pore size distributions for the EDA pretreatments. (a) 10 %, (b) 50 %, (c) 70 %. 81 (a) 0.50 0.25 d. lnaccessible Water (mL/g) 0.00 ' V""'I ' ‘ w""" a ' """I too 1000 Molecular Diameter (A) .e O 1.25 (b) 1.00 0.73 0.50 d. lnacceaaible Water (mL/g) .1 ' ' V "'10 ' ' ""160 V ' ""1300 Molecular Diameter (A) 1.25 (C) 1.00 0.50 d. lnaccessible Water (mL/g) 1 1 - ”'mt'tl ' ' “"700 . . mtooo Molecular Diameter (A) Figure 4.7: Pore volume distributions for the autohydrolysis (FR) pretreatments. (a) 240 °C, (b) 260 °C, (c) 280 °C. 82 observed for the previous cases. Perhaps the removal of both components in such large amounts (Table 4.4) caused the collapse of the remaining matrix, or the swelling aetion of the alkali in the H02 and EDA pretreatments (EDA is strongly basic) was enough to increase the pore volumes to those observed. Pretreatments with dilute H380, in the flow reactor resulted in substrates with the pore volume distributions shown in Figure 4.8. At 180 °C, pore volume was not increased much over that for Native wood. At 200 and 220 °C, however, a dramatic increase in pore volume in pores larger than around 12 A was observed. In both cases, the great majority of pore volume is in large pores and thus more accessible to the enzyme. Practically all of these increases were due to hemicellulose removal alone; FRWA 220 °C had‘ nearly complete hemicellulose removal. Tatal surface areas calculated using parallel plate geometry are given in Table 4.8, along with total pore volumes for each (fiber saturation point, at 560 A). Since the cellulase size range extends from 24 to 77 A, surface areas in pores larger than these nominal diameters were calculated and are given in Table 4.9. Steady increases in surface area are seen in pores larger than each assumed enzyme diameter with increasing pretreatment severity for all pretreatments. As expected, the FRWA pretreatments at 200 and 220 °C provide large increases in surface area in larger pores, while total surface area is actually decreased. EDA 70 % also markedly 83 (a) d, inaccessible Water (mL/g) v‘v' v 1 v v ' V ' ' V ' ' ' ‘ ‘ I V ' ' ' ' ' V '1 1 10 100 1000 Molecular Diameter (A) 1.50 1.25 (b) 1.00 0.75 0.50 d, lnacceesible Water (mL/g) l ' fi'm'i'o ' "Tmt'bo ' mib‘oo Molecular Diameter (A) (c) d, lnaccessible Water (mL/g) 0.00 - - . - . .fi-.. - - --.--. 1 1'0 100 1000 Molecular Diameter (A) Figure 4.8: Pore volume distributions for the acid hydrolysis (FRWA) pretreatments. (a) 180 °C, (b) 200 °C, (c) 220 °C. 84 Table 4.8: Fiber saturation points (FSP) and total specific surface areas (SSA) for all substrates. FSP Total SSA Pretreatment (mL/g) (m’Ig) Native 0.764 1577 H,O2 : 0.5 hr 0.874 1374 5 hr 1.428 1758 19 hr 1.407 1636 EDA : 10 % 1.083 1781 50 % 0.960 1149 70 % 1.465 1366 FR : 240 °C 0.843 1118 260 °C 0.989 1514 280 °C 1.108 1466 FRWAl:80 °C 0.771 1134 200 °C 1.059 944 220 °C 1.331 923 85 Table 4.9: Specific surface area (SSA) in pores larger than various enzyme sizes over the range 24 to 77 A. SSA (m’Ig) i 38 A n pores larger than: Pretreatment 24 A 51 A 61 A 77 A Native 61.9 26.1 14.8 10.4 6.65 H203 : 0.5 hr 107 44.4 24.5 16.8 10.4 5 hr 231 110 66.7 48.6 33.7 19 111' 232 115 70.8 52.2 35.4 EDA : 10 90 124 53.4 30.7 21.7 14.1 50 % 152 75.5' 47.3 35.2 24.2 70 % 265 144 94.5 72.4 51.3 FR : 240 °C 125 58.2 34.7 25.0 16.5 260 °C 124 60.9 38.0 28.3 19.5 280 °C 160 78.0 48.4 35.8 24.5 FRWA: 180 °C 103 45.0 25.8 18.1 11.6 200 °C 211 128 85.7 65.7 44.9 220 °C 295 183 128 100 72.0 86 increased the surface area in large pores. The increases in surface area for the FR pretreatments were small, in keeping with their small increases in pore volume. Alkaline H,O2 pretreatment did not provide much of an increase in surface area in large pores after 5 hours of pretreatment. Also, the water absorbing capabilities spoken of in Section 4.1 now become clear; the fiber saturation points for 5 and 19 hours of pretreatment are quite high, signifying a large potential for absorbing water. 4.5 Enzymatic Hydrolysis 4.5.1 Choice of Enzyme Level A series of experiments were performed with the most accessible pretreated substrate, FRWA 220°C, to determine the amount of cellulase needed to completely saturate the active sites. This "saturation" level corresponds to the maximum cellulase activity needed to ensure that no enzyme limitations were present. Because only a limited supply of Rutgers C30 cellulase was available, Novo CCN 3000 cellulase (also from T. reesei) was used. The variation of initial glucose yield with specific enzyme activity is shown in Figure 4.9; saturation is achieved by 50 FPU/g wood. To convert this activity to units of C30 enzyme, FRWA 220 °C wood was subjected to C30 hydrolysis at various concentrations. It was found that the initial yields were practically identical for CCN 3000 at 49.6 FPU/g wood and C30 at 38.8 FPU/g wood, as shown in Figure 4.10. Thus an A activity of 40.8 FPU/g wood (slightly larger than 38.8) was used for 87 Glucose Yield at 2 hr (%) 0 .,.,.,.,., o 10 20 30 40 50 CCNSOOO Activity (FPU/g wood) Figure 4.9: Determination of maximum CCN 3000 activity for FRWA 220 °C wood. 88 1001 - g 80- A . N 1 v Cl 2 60- 9 . 5. o .. 8 404 o . 2 o -1 20- 0E.., .r...,...,...,..., 0. 4 8 12 16 20 24 time(hr) Figure 4.10: Comparison of the time course of enzymatic hydrolysis of FRWA 220 °C wood with (A) Novo CCN 3000 cellulase and (O) Rutgers C30 cellulase. 89 all substrates. 4.5.2 Enzymatic Hydrolysis of Pretreatcd Wood The time courses of the enzymatic hydrolysis of wood pretreated at the three levels of alkaline H,O, are shown in Figure 4.11. Glucose yields at 24 hr are increased markedly over that for Native wood, from about 14.7 % for Native to 47.1, 59.9 and 68.2 % for 0.5, 5, and 19 hours of pretreatment, respectively. In the initial period of hydrolysis, however, the differences are not so large; the 5 and 19 hour pretreatments are nearly indistinguishable from one another within the first hour of enzymatic hydrolysis. Considering the results of the surface area analysis of the last section, this is not unexpeCted. The enzymatic hydrolysis of EDA pretreated wood fractions are shown in Figure 4.12. Again, increasing pretreatment severity increases glucose yields at 24 hr quite markedly. In the case of EDA, however, initial glucose yields are quite different for the different levels of pretreatment.» For EDA 70 %, 77 % of the 24 hr yield was obtained in only 2 hours of enzymatic hydrolysis. This is quite an improvement over Native wood, in which the hydrolysis was, for all practical purposes, finished at 4 hours of hydrolysis with only 10.5 % glucose yield. For the FR (autohydrolysis) pretreatments, the enzymatic hydrolysis time profiles (Figure 4.13) appear much different from those for alkaline H20, and EDA. The initial increase in glucose 90 Glucose Yield (%) 0-9...,...,...,...,...,..., 0 4 8 12 16 20 24 time (hr) Figure 4.11: Time course of the enzymatic hydrolysis of Native (0), and alkaline H30, pretreated wood; (I) 0.5 hr, (A) 5 hr, and (V) 19 hr. 91 100- 3 8 8 eleneleeelee Glucose Yield (%) l. N 0 else U fr 1 U r'i‘r Y 1' U T‘r . . f. . . . , O. 4 8 12 16 20 24 time (hr) Figure 4.12: Time course of the enzymatic hydrolysis of Native (0), and EDA pretreated wood; (I) 10 %, (A) 50 %, and (V) 70 %. 92 100- so- A ' ——-—9 is v d 19.60- 2 . 5- . 8 .1 40 8 1 2 o l 20- 0 .ef,...,...,.e.,...,..., 0 4 a 12 15 20 24 time (hr) Figure 4.13: Time course of the enzymatic hydrolysis of Native (0), and FR (autohydrolysis) pretreated wood; (I) 240 °C, (A) 260 °C, and (V) 280 °C. 93 yield is much slower, and the curves do not level off completely by 24 hr, as with the others. Perhaps this is the effect of crystallinity in the latter stages of the enzymatic hydrolysis. As before, increasing pretreatment severity increased 24 hr glucose yields; although their initial rates were lower than those for alkaline H202 and EDA, the 24 hr yields for the FR pretreatments were as good or better than those for peroxide and EDA. As shown in Figure 4.14, dilute acid hydrolysis (FRWA) produced the greatest increases in enzymatic glucose yields of any of the pretreatments. The lone exception was FRWA 180 °C, which had the lowest final yield, with about three times that for Native wood. FRWA 200 °C and FRWA 220 °C pretreatments provided final glucose yields at 24 hr of 92.1 and 97.7 %, respectively. Their initial rates (yields) were also very high, at 51.1 and 71.0 % at 2 hours, respectively. With only four hours of enzymatic hydrolysis, these two pretreated substrates had already equalled or surpassed the 24 hr yields of all the other pretreated substrates. Defining a hydrolysis time for the average initial rate is difficult; thus, up to 4 hours of hydrolysis was considered to represent an average initial rate (for example, average initial rate at 0.5 hr, 1 hr, etc.). The yields at these times are given in Table 4.10, and are used separately as measures of the average initial rates. 94 80-1 A . N . V I E 60- 2 ., 5- . o . s 40- o . 2 ‘f (D 20- Oa..,s..,...r..4=,fi.fi,fi.., 0 4 8 12 16 20 24 time (hr) Figure 4.14: Time course of the enzymatic hydrolysis of Native (0), and FRWA (acid hydrolysis) pretreated wood; (I) 180 °C, (A) 200 °C, and (V) 220 °C. 95 Table 4.10: Glucose yields from enzymatic hydrolysis at times ranging from 0.5 to 4 hours. Glucose Yield (%) at : Pretreatment 1/2 hr 1 hr 2 hr 4 hr Native 6.59 7.42 9.83 10.47 H,O, : 0.5 hr 16.50 21.24 26.35 32.93 5 hr 23.48 30.76 37.20 45.66 19 hr 23.87 32.43 41.98 49.33 EDA : 10 % 19.47 24.55 31.10 39.90 50 % 23.08 31.72 41.77 49.99 70 % 30.77 42.01 54.48 59.77 FR : 240 °C 7.45 10.95 17.45 26.77 260 °C 10.57 15.92 25.21 35.10 280 °C 13.78 21.27 32.51 47.01 FRWA:180 °C 8.58 11.87 16.49 23.23 200 °C 27.95 40.74 51.05 73.58 220 °C 33.59 50.03 70.99 85.56 96 4.6 Surface Area Correlations The values for surface area available to solutes ranging in size from 24 to 77 A (Table 4.9) were combined with the initial rate (glucose yield) data from Table 4.10 to derive linear relationships between glucose yield (at a certain hydrolysis time) and surface area available to solutes of sizes ranging from 24 to 77 A. The correlation for the 2 hour glucose yield and the surface area available to a solute of 51 A diameter is shown in Figure 4.15. The slope of this correlation is 0.605 %lm‘I and its correlation coefficient (r’) is 0.843; the correlation, therefore, is fairly good (r2 of 1 is a perfect correlation and an r1 of 0 shows no correlation). The slopes and correlation coefficients for the other fits are given in Tables 4.11. It can be seen from Table 4.11(b) that fairly good correlations exist for a fairly wide range of assumed enzyme sizes. The correlation coefficients, however, should be higher at the true (or "average") enzyme size. Figure 4.16 shows the variation of the correlation coefficients with minimum accessible pore diameter. The best range of correlations occurs with the 2 hour yield, so this is probably the correct time to use for the average initial rate. For all but the 0.5 hour series, a maximum r2 occurs at (or around) 38 A, suggesting that the average size of the component enzymes of the T. reesei cellulase complex may be near 38 A. In any case, the existence of a linear correlation between glucose yield and available surface area tends to suggest that surface area is the major limiting 97 Glucose Yield at 2 hr (%) I 1 I I l U j I I U I I 0' ' ' '2'; ' ' '5'0 75 100 ‘125' ' ' '150 Surface Area Available To 51A Solute (ma/g) Figure 4.15: Linear correlation between glucose yield at 2 hr and specific surface area available to a solute of 51 A diameter: Native (0), alkaline H302 (Cl), EDA (I), FR (0). and FRWA (A). 98 Table 4.11(a): Slopes of the correlations derived for glucose yields at various times and surface areas available to solutes of various sizes. Minimum Pore Correlation Slope (%lmz) for Yield at: Diameter (A) 0.5 hr 1 hr 2 hr 4 hr 24 0.112 0.157 0.210 0.263 38 0.210 0.296 0.395 0.494 51 0.321 0.452 0.605 0.756 61 0.421 0.594 0.796 0.993 77 0.598 0.846 1.134 1.413 Table 4.11(b): Correlation coefficients for the correlations given in Table 4.11(a). Minimum Pore Correlation Coefficient (r2): Diameter (A) 0.5 hr 1 hr 2 hr 4 hr 24 0.800 0.829 0.842 0.770 38 0.765 0.849 0.881 0.807 51 0.683 0.798 0.843 0.753 61 0.629 0.759 0.808 0.629 77 0.568 0.710 0.764 0.643 99 0.90- “E .9 0.804 .2 a: 0) 8 c 0.704 .9 4.1 2 d.) t: o 0.60“ o 0050 I . I I I I I I 1 fi ‘ I I 20 :50 4o 50 50 70 80 Minimum Accessible Pore Diameter (A) Figure 4.16: Variation of the correlation coefficients from the linear correlations in Table 4.11(a): (O) 0.5 hr yields, ([3) 1 hr yields, (I) 2 hr yields, and (Q) 4 hr yields. 100 factor in the hydrolysis. 4.7 Empirical Correlations Empirical correlations similar to those presented in Section 2.4.1 (equations (2-1) and (2-2)) were derived to determine whether crystallinity and/or lignin content has any effect on the initial glucose yield. Since 2 hours was determined to be the best time for measurement of the initial rate, these correlations were derived only for 2 hour yields. The correlations were of the form: GY, = a (100 - CrI)" (G/L)‘ (SSA,)‘ (4-1) where GY, is the glucose yield at 2 hours, Cd is the crystallinity index, G/L is the glucan to lignin ratio, SSA, is the surface area in pores larger than size 1', and a, b, c, and d are constants. Equation (4-1) was linearized to the form: 1n(GY,) = ln(a) + b ln(100 - CrI) + c ln(G/L) + d ln(SSA,) (4-2) Linear regression was then performed and the constants a, b, c, and d determined. The values for the constantsusing the glucan/lignin ratios given in Table 4.4, the CrI given in Table 4.6, and the available surface areas given in Table 4.9 are given in Table 4.12. The correlation for each was tested by performing linear regression on the actual versus predicted yields. The slopes and correlation coefficients for these tests are given in Table 4.13. If the correlations are "good," the best fit lines should have slopes of 1. For all the correlations, the slope of the best fit was nearly one. In addition, the predicted yields are plotted in Figure 4.17 againSt the 101 Table 4.12: Constants a, b, c, and (1 determined by linear regression for the empirical fits to equation (4-2). Minimum Pore Empirical Constantsz’ Diameter (A) a b c d 24 0.1080 -0.00806 0.03378 1.124 38 0.2637 0.2171 0.2108 0.9142 51 0.4437 0.2932 0.2470 0.8270 61 0.6113 0.3213 0.2487 0.7840 77 0.9798 0.3217 0.2174 0.7387 * The empirical correlations are of the form: Yield (%) = a (100 - CrI)” (GIL)° (SSA)‘ Table 4.13: Slopes and correlation coefficients for the correlations derived for actual glucose yield at 2 hr vs predicted yields from the empirical correlations. Minimum Pore Best Fit Diameter (A) Slope” r2 24 1.005 0.842 38 0.999 0.870 51 0.996 0.875 61 0.996 0.875 77 0.996 0.872 ** A perfect empirical fit would have a slope of 1 and an r2 of 1. 102 80- 70- . ‘ 60- 50- ‘ 30- ' 20- 10.. 0 Actual Glucose Yield at 2 hr (7.) 3 l I U O U I j. U 1 1" l U Ififii l I I 0 10 2'0 30 40 5'0 60 70 80 Predicted Glucose Yield at 2 hr (%) Figure 4.17: Linear correlation between actual glucose yield at 2 hr and predicted glucose yield from the empirical correlation using surface area available to a solute of 51 A diameter: (0) Native, ([3) alkaline H203, (I) EDA, (0) FR, and (A) FRWA. 103 actual data for the correlation using surface area available to a 51 A solute. The best prediction of glucose yields at 2 hr is that using the surface area available to a 51 A solute. In all cases, the effects of CrI and G/L are small because the numbers themselves are small, in addition to their small exponents. Surface area, on the other hand, is the dominating factor in all cases, as evidenced by the large exponents for the surface area terms (the values for surface area are larger than those for Cd or G/L). Thus, the empirical correlatons tell pretty much the same story as the linear correlations between glucose yield and surface area alone: surface area is the dominating factor in the early stages of enzymatic hydrolysis. CHAPTER V: DISCUSSION 5.1 Significance The results of this study are significant because this is the first time that pretreatments have been compared on an equal basis. Past comparisons of pretreatment effectiveness have invariably been based on results from different substrates, different enzymes, and different analytical methods, some of which were more reliable than Others. The linear correlations developed by Stone et at.” Grethlein", and Grous er al.‘° have been tested with a common substrate which was modified by four different methods to produce a range of compositions of each wood component. The correlations were proven again, showing that surface area is definitely the controlling factor in the initial stages of enzymatic hydrolysis. The similarities of the resulting empirical correlations to those of Gharpuray er al." again shows that the enzymatic hydrolysis of lignocellulosics is fundamentally different than for cellulosics. Studies to determine the effects of various factors on any property of lignocellulosic materials should therefore be performed on lignocellulosic materials; the cellulosic materials differ from lignocellulosics so much in structure and chemical reactivity that they do not behave similarly at all. 5.2 Pretreatment Effectiveness The major point of this study was to prove or disprove the universality of the linear surface area correlation. The results, however, can also be used to compare the effectiveness of various 104 105 types of pretreatments. For all pretreatments, enzyme accessibility increased with pretreatment severity, but to different degrees, as evidenced by the trends in the accessible surface areas produced. A point to consider here is how the removal of the various components serves to increase this surface area. Alkaline peroxide pretreatment swells the cellulose matrix and removes moderate amounts of hemicellulose and lignin. EDA, which is also alkaline, swells the cellulose and also removes similar amounts of hemicellulose and lignin (slightly higher hemicellulose removal). Perhaps the reason that EDA at 70 % produced a substrate that was much more accessible than the 19 hr alkaline pretreatment was because of the relative concentrations of NaOH (about 22 wt. %) and EDA (70 vol. %). although swelling of the cellulose is not likely to completely account for the differences. The autohydrolysis (FR) pretreatment removed more lignin than any of the other pretreatments, and also more hemicellulose than either alkaline peroxide or EDA, yet its surface areas were smaller. This was probably due to a collapse of the matrix because so much of its supporting material (lignin and hemicellulose) was removed. The FRWA pretreatments, which removed mainly hemicellulose, had the highest yields at all times, and had very high yields in as little as 2 hr of enzymatic hydrolysis. Clearly this is the best pretreatment, since economics dictate that reaction times must be kept as short as possible or costs go up. In addition, the hydrolysis was nearly 106 completed in four hours for FRWA 200 °C and FRWA 220 °C, and 92-98 % yields of glucose were obtained in 24 hr. 5.3 Cellulase size The fact that the correlations for initial glucose yield versus available surface area are fairly linear over a wide range of assumed enzyme sizes (24 to 77 A) suggests that the endo- and exoglucanase components are distributed in size over this range. Since the hydrolysis of cellulose completely to glucose cannot occur without both of these activities, it stands to reason that if the rate is proportional to surface area at both 24 and 77 A, endo- and exoglucanases with sizes near these values must be present. The fact that the highest correlation occurs at an enzyme size of 38 A suggests that the average cellulase complex component size is probably near 38 A. Cowling and Kirk‘ reported that T. reesei cellulases were near the upper limits of the 24 to 77 A size range; it would be interesting to collect gel filtration data and/or determine the molecular weight distribution of purified C30 cellulase to see if the suggestions from this study are in fact true. 5.4 Future Directions An important point should be checked concerning the nature of the surface area in the pores of the lignocellulosic matrix. The assumption was that the surface area in the pores represented cellulose surface area -- the solute exclusion technique has no way of determining whether this assumption is true or not. A way of 107 determining the effect of lignin on the reactive surface area might be to treat the wood with a mixture of enzymes that degrade hemicellulose (a hemicellulase) to degrade the hemicellulose while leaving the lignin intact. Most pretreatments which remove only hemicellulose use temperatures above the glass transition temperature of lignin, so even though the lignin is not removed, its spacial arrangement is modified. Once this substrate (hemicellulose removed, with lignin intact) is prepared, the surface area correlations could be used with enzymatic hydrolysis data to see if the lignin matrix, when intact, has any effect on the correlation prediction (I expect that it would). One problem with this method is that hemicellulases and ligninases have been shown to have accessibility problems similar to those for cellulases“. An excellent test of the suggestion of this study that the average size of T. reesei cellulase complex components is near 38 A would be to purify each separate component enzyme (including each endo- and exoglucanase), determine their individual sizes, and hydrolyze a lignocellulosic substrate (pretreated to get a large range of pore sizes) with mixtures of various components of the complex to see how the rate of hydrolysis varies with their individual sizes. If the suggestion of this study is in fact true, then the average of these ”values should be around 38 A. It is important to keep in mind that since the complex acts synergistically to degrade cellulose to glucose, both endo- and exoglucanases as well as B-glucosidases (these act on 108 soluble cellobiose and thus do not have an accessibility problem) must be included in the experiment. Otherwise a range of glucose oligomers, some soluble and some not, would have to be measured. Another area which merits attention is, in fact, the question of how ligninases and hemicellulases are limited by the matrix structure. Recent evidence"""”" has suggested that the enzymes penetrate the matrix in a synergistic manner; that is, neither enzyme would be very accessible without the action of the other. Finally, it would be interesting to hydrolyze the pretreated substrates from this study with the enzyme level used by Grethlein‘7 and Grous et at.“ (16.2 FPU/g wood) to see if the slope of the new correlation would be the same as their slopes (0.375 and 0.326 %lm’, respectively). APPENDIX 109 Appendix A Derivation of equation (3-8) A.1 Derivation Equation (3-8) was used to correct dry weights from solute exclusion experiments in which the filtration was too slow to wash the solutes from the wet presscake in a reasonable length of time. The form of equation (3-8) is: P = {P' - WCt(1 - lag/0111)}lll + (WCl/ai) (at/Phi...) (3-8) where p’ is the dry weight of presscake and residual solute (g), p is the dry weight of the wet presscake (g), at, is the optical ratation of the final (diluted) solution (°), 11, is the optical rotation of the initial solution (°), (oh/PL“...K is the average of (Gig/p) (correction factors for the blanks, °lg), w is the mass of solute solution added (g), and C, is the initial solute solution concentration (%). Additional parameters needed are: wwj, the mass of dextran left in the pulp (g); ww, the mass of dextran initially in the bottle (g); ww, the mass of dextran in solution after equilibration (g); and am", the optical rotation of the final solution, corrected for contributions of solubles (°). 1 The actual dry weight of pulp which contains residual dextrans is: 110 P, = P + WDJJ Now, the mass of dextran initially in the bottle is: WDJ = W C1 (A-l) (A-Z) In addition, the ratio of the initial and final optical rotations is equal to the ratio of the initial and final concentrations, or a,/a,,,,,, = C,/C, Similarly, Wot/Wot =-' Ct/Ct Combining equations (A-2), (A-3) and (A—4) gives: “’13,! = W C. [amine/01.] Now, the mass of dextran in the pulp is: “In.” 3 Won ' Wm: or w”, = w C, - w C, [awn/01,] Substitution of equation (A-7) into (A-l) gives: 9’ = p + w C. {1 - [cam/ad} Now, from equation (3-9), we know that “1.6m = a, " P (an/Phi.“ Then we have P’ = p {1 + w C. («Amen/at} + w C. {l - (cu/01.)] Solving for p then gives equation (3-8), or: p = 115' - wC.<1 - ia./a.1)1/{1 +