UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Understanding and modelling softwood hemicellulose hydrolysis and its adsorption to pulp fibres Chen, Jingqian 2020

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Notice for Google Chrome users:
If you are having trouble viewing or searching the PDF with Google Chrome, please download it here instead.

Item Metadata

Download

Media
24-ubc_2020_may_chen_jingqian.pdf [ 8.91MB ]
Metadata
JSON: 24-1.0388441.json
JSON-LD: 24-1.0388441-ld.json
RDF/XML (Pretty): 24-1.0388441-rdf.xml
RDF/JSON: 24-1.0388441-rdf.json
Turtle: 24-1.0388441-turtle.txt
N-Triples: 24-1.0388441-rdf-ntriples.txt
Original Record: 24-1.0388441-source.json
Full Text
24-1.0388441-fulltext.txt
Citation
24-1.0388441.ris

Full Text

Understanding and Modelling Softwood HemicelluloseHydrolysis and Its Adsorption to Pulp FibresbyJingqian ChenM.Sc.Eng., University of Michigan, 2012B.Sc.Eng., Shaanxi University of Science and Technology, 2011A THESIS SUBMITTED IN PARTIAL FULFILLMENTOF THE REQUIREMENTS FOR THE DEGREE OFDOCTOR OF PHILOSOPHYinTHE FACULTY OF GRADUATE AND POSTDOCTORAL STUDIES(Chemical and Biological Engineering)THE UNIVERSITY OF BRITISH COLUMBIA(Vancouver)January 2020c© Jingqian Chen, 2020The following individuals certify that they have read, and recommend to the Faculty ofGraduate and Postdoctoral Studies for acceptance, the thesis entitled:Understanding and Modelling Softwood Hemicellulose Hydrolysis andIts Adsorption to Pulp Fibressubmitted by Jingqian Chen in partial fulfillment of the requirements for the degree ofDOCTOR OF PHILOSOPHY in Chemical and Biological Engineering.Examining Committee:Heather L. Trajano, Chemical and Biological Engineering, The University of British ColumbiaCo-SupervisorRodger P. Beatson, Chemical and Environmental Technology, British Columbia Instituteof TechnologyCo-SupervisorD. Mark Martinez, Chemical and Biological Engineering, The University of British ColumbiaSupervisory Committee MemberPeter Englezos, Chemical and Biological Engineering, The University of British ColumbiaUniversity ExaminerFeng Jiang, Wood Science, The University of British ColumbiaUniversity ExaminerAdditional Supervisory Committee Members:Martin Lawoko, Wallenberg Wood Science Centre, KTH, Royal Institute of TechnologyExternal ExaminerPaul Bicho, Canfor Pulp InnovationSupervisory Committee MemberiiAbstractPulp and paper facilities are transforming into innovative biorefineries producing chem-icals and materials in order to enhance competitiveness and environmental performance.Hemicellulose fractionation and its integration with the pulp mill are crucial to successfulbiorefinery processes. Fundamental insight is needed to support technology growth.Hemicellulose oligomers were extracted by hydrolysis of softwood chip fines usinghot water; conditions were optimized for high yield and high molar mass. Comprehen-sive characterization of hydrolysate and hydrolyzed solids is reported. A two-dimensionalcalibration method enabled measurement of oligomer molar mass and concentration si-multaneously by size exclusion chromatography. A population balance model describ-ing evolution of hemicellulose molar mass during hydrolysis was posed. The modeldescribes the full evolution of oligomers from initial softwood solubilization, depoly-merization to ever-smaller molecules until final generation of degradation products. Amaximum yield and corresponding treatment condition for a specific molar mass couldbe predicted. Likely modes of hemicellulose bond breaking within the wood matrix andbulk solution are proposed and physical insights are explained. This work provides fun-damental insights into the relative reactivity of hemicellulose intermediates to facilitatefuture conversion technologies.Two hemicellulose hydrolysis integration methods to kraft pulping are proposed. First,adsorption of locust bean gum (LBG, model compound) to Northern Bleached SoftwoodKraft (NBSK) pulp was shown to improve paper tensile and burst strength and loweriiirefining time by strengthening inter-fibre bonding. LBG adsorption to NBSK pulp fibreis dependent on electrostatic forces, and high salt addition at low pH facilitates adsorp-tion. The adsorption followed pseudo-second-order kinetics and the Langmuir adsorp-tion isotherms, indicating a reversible, monolayer, homogenous adsorption to a finitenumber of sites on the fibre surface with chemisorption as the rate determining step.Second, mild hydrolysis was combined with kraft pulping as pre-treatment to producehemicellulose oligomer and kraft pulp. Particle size effects on hydrolysis and subsequentkraft pulping were assessed. Kraft pulping of pre-hydrolyzed softwood chips enhancesdelignification, reduces fibre yield and chemical consumption, producing fibres with de-creased fibre dimension but increased water retention value. The advantages and dis-advantages of pre-hydrolysis integration to kraft pulping and associated challenges andfuture recommendations are discussed.ivLay SummaryBiorefinery is a conceptual response to the oil refinery in which lignocellulosic biomass isfractionated and upgraded to produce multiple chemicals and materials. Wood, as one ofthe most abundant lignocellulosic biomass types, is a versatile feedstock.Biomass is a complicated matrix of cellulose, hemicellulose, and lignin. The matrixserves as a physical barrier and hinders fractionation of biomass components to valu-able bioproducts. My population balance model of wood hydrolysis answers this chal-lenge by tracking the hemicellulose oligomer transformation from initial solubilization,depolymerization to smaller molecules until final generation of degradation products.The pulp and paper industry is energy intensive partially due to refining to develop fi-bre properties. Hemicellulose oligomer can adsorp to pulp fibre as strength additive. Ourexperiments have shown that small amounts of hemicellulose can improve pulp strengthand lower the refining energy consumption by approximately half.vPrefaceA version of Chapter 2 has been published. Chen, J., Yuan, Z., Zanuso, E., & Trajano, H.L. (2017). Response of biomass species to hydrothermal pretreatment. In HydrothermalProcessing in Biorefineries (pp. 95-140). Springer, Cham.I prepared the content outline, wrote the wood hydrothermal pretreatment section,ultrastructure and mechanism section, introduction and summary. Dr. Zhaoyang Yuanwas responsible for bamboo and agriculture residue hydrothermal pretreatment sectionand reviewed the manuscript. Elisa Zanuso was responsible for preparation of the agavehydrothermal pretreatment section. Dr. Heather L. Trajano provided guidance on themanuscript preparation, reviewed and edited the manuscript.A version of Chapter 3 will be submitted for publication.I designed and conducted experiments, processed data and wrote manuscript. Xue-feng Chang conducted HPLC analysis for hemicellulose composition. Co-op studentsXinyi Chen and Rylan Martinez conducted part of hydrolysis and compositional analy-sis. Dr. Heather L. Trajano advised on experiment design, results interpretation, reviewedand edited the manuscriptA version of Chapter 4 and Chapter 5 will be submitted for publication. Evolution ofHemicellulose Molar Mass During Softwood Hydrolysis.I designed and conducted experiments, developed protocol, processed data, devel-oped model and wrote manuscript. Dr. Heather L. Trajano advised on experiment de-visign, data analysis, and results and model interpretation. Dr. D. Mark Martinez advisedon model development and model interpretation. Xuefeng Chang conducted HPLC anal-ysis for hemicellulose composition. Dr. Heather L. Trajano, Dr. D. Mark Martinez, Dr.Rodger P. Beatson and Xuefeng Chang reviewed and edited the manuscript.A version of Chapter 6 will be submitted for publication. Locust Bean Gum Adsorp-tion onto Softwood Kraft Pulp Fibres.I designed and conducted experiments, developed protocol, processed data and wrotemanuscript. BCIT CENV4400 project students Vicki Chen, Nick Park, Carrie Wang, highschool volunteers Bruce Li and Bruce Qin, and Co-op student Miaoran Li conducted partof adsorption experiments under my supervision. Dr. Heather L. Trajano and Dr. RodgerP. Beatson provided guidance on experimental design, results interpretation, reviewedand edited the manuscript.A version of Chapter 7 will be submitted for publication. Kraft Pulping of SoftwoodChips with Mild Hot Water Pre-Hydrolysis.This work was a collaboration with Canfor Pulp Innovation. I designed and con-ducted experiments, processed data and wrote manuscript. Dr. Paul Bicho provided labfacility access and advised on data interpretation. Kenny Tam contributed to design andperformance of experiments. Co-op student Hong Ma conducted some kappa numberand black liquor measurements. Dr. Heather L. Trajano and Dr. Rodger P. Beatson pro-vided guidance on experimental design, results interpretation, reviewed and edited themanuscript.viiTable of ContentsAbstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . iiiLay Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . vPreface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . viTable of Contents . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . viiiList of Tables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xviList of Figures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xxList of Abbreviations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .xxviiiList of Symbols . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xxxAcknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .xxxvDedication . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .xxxvii1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11.1 Motivation and Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11.1.1 Renewable energy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11.1.2 Pulp and paper industry . . . . . . . . . . . . . . . . . . . . . . . . . . 21.1.3 Biorefinery . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3viii1.1.4 Hemicellulose hydrolysis . . . . . . . . . . . . . . . . . . . . . . . . . 41.1.5 Hemicellulose hydrolysis integration to kraft pulping . . . . . . . . . 61.1.5.1 Hemicellulose as a paper strength additive . . . . . . . . . . 61.1.5.2 Pre-hydrolysis kraft pulping . . . . . . . . . . . . . . . . . . 71.2 Research Hypotheses . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71.3 Thesis Organization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92 Response of Biomass Species to Hydrolysis . . . . . . . . . . . . . . . . . . . . . 112.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112.1.1 Motivation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112.1.2 History, current state-of-the-art and future development . . . . . . . 122.1.3 Feedstock crop, production and utilization . . . . . . . . . . . . . . . 132.1.3.1 Wood . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 132.1.3.2 Bamboo . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 142.1.3.3 Agricultural residues . . . . . . . . . . . . . . . . . . . . . . 152.1.3.4 Agave and agave bagasse . . . . . . . . . . . . . . . . . . . . 162.2 Structure and Chemical Composition Analysis of Raw Biomass . . . . . . . 172.2.1 Ultrastructure of lignocellulosic biomass . . . . . . . . . . . . . . . . . 172.2.2 Compositional analysis of biomass . . . . . . . . . . . . . . . . . . . . 192.2.2.1 Wood . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 202.2.2.2 Bamboo and agricultural residues . . . . . . . . . . . . . . . 202.2.2.3 Agave and AGB . . . . . . . . . . . . . . . . . . . . . . . . . 212.3 Response of Biomass to Hydrothermal Treatment . . . . . . . . . . . . . . . . 252.3.1 Reactions in acidic condition . . . . . . . . . . . . . . . . . . . . . . . 262.3.1.1 Hydrolysis mechanism in acidic condition . . . . . . . . . . 272.3.1.2 Hemicellulose . . . . . . . . . . . . . . . . . . . . . . . . . . . 282.3.1.3 Cellulose . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 332.3.1.4 Lignin . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35ix2.3.1.5 Ash . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 362.3.1.6 Extractives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 372.3.1.7 Ultrastructure . . . . . . . . . . . . . . . . . . . . . . . . . . 372.3.1.8 Summary of acid pretreatment on biomass solids . . . . . . 382.3.2 Reactions in alkaline condition . . . . . . . . . . . . . . . . . . . . . . 392.3.2.1 Hydrolysis mechanism in alkaline condition . . . . . . . . . 392.3.2.2 Hemicellulose . . . . . . . . . . . . . . . . . . . . . . . . . . . 412.3.2.3 Cellulose . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 442.3.2.4 Lignin . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 452.3.2.5 Ash . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 472.3.2.6 Extractives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 472.3.2.7 Ultrastructure . . . . . . . . . . . . . . . . . . . . . . . . . . . 472.3.2.8 Summary of alkaline pretreatment on biomass solids . . . . 482.4 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 523 Softwood Hemicellulose Hydrolysis by Hot Water: Characterization and Op-eration Guideline . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 533.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 533.1.1 Motivation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 533.1.2 Softwood hemicellulose . . . . . . . . . . . . . . . . . . . . . . . . . . 543.1.2.1 Composition and structure . . . . . . . . . . . . . . . . . . . 543.1.2.2 Acidic groups of hemicellulose . . . . . . . . . . . . . . . . . 553.1.3 Wood hydrolysis factors influencing yield and molar mass . . . . . . 563.1.3.1 Effects of temperature and time . . . . . . . . . . . . . . . . 563.1.3.2 Effects of particle size and solids loading . . . . . . . . . . . 573.1.3.3 Effects of pH . . . . . . . . . . . . . . . . . . . . . . . . . . . 583.1.3.4 Effects of reactor type . . . . . . . . . . . . . . . . . . . . . . 593.1.4 Severity factor and pseudo-time . . . . . . . . . . . . . . . . . . . . . 60x3.2 Goals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 613.3 Experimental Design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 623.3.1 Material and milling . . . . . . . . . . . . . . . . . . . . . . . . . . . . 623.3.2 Hydrolysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 623.3.3 Compositional analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . 653.4 Results and Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 663.4.1 Particle size distribution . . . . . . . . . . . . . . . . . . . . . . . . . . 663.4.2 Severity factor and pseudo-time calculation . . . . . . . . . . . . . . . 663.4.3 Operation factor analysis . . . . . . . . . . . . . . . . . . . . . . . . . . 693.4.3.1 Agitation rate effect . . . . . . . . . . . . . . . . . . . . . . . 693.4.3.2 Solids consistency effect . . . . . . . . . . . . . . . . . . . . . 693.4.3.3 Particle size effect . . . . . . . . . . . . . . . . . . . . . . . . 703.4.3.4 Effect of composition . . . . . . . . . . . . . . . . . . . . . . . 713.4.4 Hydrolysate characterization . . . . . . . . . . . . . . . . . . . . . . . 753.4.4.1 Hemicellulose oligomer and monomer yield . . . . . . . . . 753.4.4.2 Hemicellulose composition . . . . . . . . . . . . . . . . . . . 763.4.4.3 Hemicellulose component yield . . . . . . . . . . . . . . . . 783.4.5 Hydrolyzed solids characterization . . . . . . . . . . . . . . . . . . . . 833.4.5.1 Acid insoluble residue and acid soluble lignin . . . . . . . . 833.4.5.2 Solubilized lignin . . . . . . . . . . . . . . . . . . . . . . . . . 843.4.5.3 Solubilized solids . . . . . . . . . . . . . . . . . . . . . . . . . 853.4.6 Chip fines vs. hog fuel . . . . . . . . . . . . . . . . . . . . . . . . . . . 863.4.6.1 Feedstock composition comparison . . . . . . . . . . . . . . 863.4.6.2 Hemicellulose yield comparison . . . . . . . . . . . . . . . . 863.5 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 883.6 Future Work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89xi4 Hemicellulose Molar Mass Evolution: Methods Development and Analysis . . 904.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 904.2 Goals and Hypotheses . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 924.3 Two-dimensional Calibration Method . . . . . . . . . . . . . . . . . . . . . . 924.4 Results and Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 974.4.1 Molar mass distribution analysis . . . . . . . . . . . . . . . . . . . . . 974.4.2 Molar mass influencing factor . . . . . . . . . . . . . . . . . . . . . . . 994.5 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1014.6 Future Work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1025 Kinetics of Softwood Hemicellulose Hydrolysis by Population Balance Model 1035.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1035.1.1 Motivation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1035.1.2 Classic carbohydrates hydrolysis model . . . . . . . . . . . . . . . . . 1035.1.3 Population balance model . . . . . . . . . . . . . . . . . . . . . . . . . 1055.1.3.1 Origins as a comminution model . . . . . . . . . . . . . . . . 1055.1.3.2 Population balance models of carbohydrate . . . . . . . . . 1075.2 Goals and Hypotheses . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1105.3 Experimental Design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1115.3.1 Material . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1115.3.2 Softwood hydrolysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1115.3.3 Two-dimensional calibration molar mass analysis . . . . . . . . . . . 1125.3.4 Compositional analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . 1125.4 Model Development . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1125.4.1 Pseudo-time calculation . . . . . . . . . . . . . . . . . . . . . . . . . . 1125.4.2 Molar mass distribution analysis . . . . . . . . . . . . . . . . . . . . . 1125.4.3 Population balance model . . . . . . . . . . . . . . . . . . . . . . . . . 1125.5 Results and Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 122xii5.5.1 Model fitting and optimization . . . . . . . . . . . . . . . . . . . . . . 1225.5.2 Parameter determination . . . . . . . . . . . . . . . . . . . . . . . . . . 1295.5.3 Activation energy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1355.5.4 Physical insights . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1395.6 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1405.7 Future work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1416 Locust Bean Gum Adsorption to NBSK Pulp: Isotherms and Kinetics . . . . . . 1426.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1426.1.1 Motivation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1426.1.2 Polysaccharides as strength additives . . . . . . . . . . . . . . . . . . 1436.1.3 Factors influencing hemicellulose adsorption . . . . . . . . . . . . . . 1456.1.4 Adsorption kinetics and isotherms . . . . . . . . . . . . . . . . . . . . 1496.2 Goals and Hypotheses . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1536.3 Experimental Design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1546.3.1 Material . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1546.3.2 LBG adsorption . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1546.3.3 LBG solution compositional analysis . . . . . . . . . . . . . . . . . . . 1566.3.4 LBG molar mass determination . . . . . . . . . . . . . . . . . . . . . . 1576.3.5 LBG adsorption for strength analysis . . . . . . . . . . . . . . . . . . . 1586.3.6 Freeness testing, handsheet preparation and strength analysis . . . . 1606.4 Results and Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1606.4.1 Adsorption kinetics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1606.4.2 Adsorption isotherms . . . . . . . . . . . . . . . . . . . . . . . . . . . 1646.4.3 Factors influencing adsorption . . . . . . . . . . . . . . . . . . . . . . 1686.4.3.1 Effect of temperature . . . . . . . . . . . . . . . . . . . . . . . 1686.4.3.2 Effect of refining . . . . . . . . . . . . . . . . . . . . . . . . . 1716.4.3.3 Effect of salt addition . . . . . . . . . . . . . . . . . . . . . . 172xiii6.4.3.4 Effect of pH . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1736.4.4 Paper strength enhancement by LBG adsorption . . . . . . . . . . . . 1746.5 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1786.6 Future Work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1797 Kraft Pulping of Softwood Chips with Mild Hot Water Pre-Hydrolysis . . . . . 1817.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1817.1.1 Motivation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1817.1.2 Pre-hydrolysis with pulping . . . . . . . . . . . . . . . . . . . . . . . . 1827.1.3 Particle size effects on hydrolysis and pulping . . . . . . . . . . . . . 1847.2 Goals and Hypotheses . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1857.3 Experimental Design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1867.3.1 Material . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1867.3.2 Pre-hydrolysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1867.3.3 Hydrolysate compositional analysis . . . . . . . . . . . . . . . . . . . 1887.3.4 Hydrolysate molar mass determination . . . . . . . . . . . . . . . . . 1887.3.5 Kraft pulping . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1887.3.6 Kappa number . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1897.3.7 Fibre compositional analysis . . . . . . . . . . . . . . . . . . . . . . . . 1897.3.8 Pulp properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1917.3.9 Effective alkali . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1917.4 Results and Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1927.4.1 Pre-Hydrolysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1927.4.1.1 Hemicellulose yield . . . . . . . . . . . . . . . . . . . . . . . 1927.4.1.2 Hemicellulose composition in hydrolysate . . . . . . . . . . 1947.4.1.3 Pre-hydrolyzed chips composition . . . . . . . . . . . . . . . 1967.4.2 Kraft pulping . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1977.4.2.1 Fibre yield and kappa number . . . . . . . . . . . . . . . . . 197xiv7.4.2.2 Fibre carbohydrates . . . . . . . . . . . . . . . . . . . . . . . 2007.4.2.3 Fibre properties . . . . . . . . . . . . . . . . . . . . . . . . . . 2027.4.2.4 Practical implications . . . . . . . . . . . . . . . . . . . . . . 2067.5 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2097.6 Future Work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2108 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2128.1 Conclusion and Contribution . . . . . . . . . . . . . . . . . . . . . . . . . . . 2128.2 Future Work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 217Bibliography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 221A Supporting Materials . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 249A.1 LBG Adsorption Kinetics Equation Derivation . . . . . . . . . . . . . . . . . 249A.2 Softwood Hemicellulose Hydrolysis Mass Transfer Model . . . . . . . . . . 252A.2.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 252A.2.2 Goals and hypotheses . . . . . . . . . . . . . . . . . . . . . . . . . . . 253A.2.3 Experimental design . . . . . . . . . . . . . . . . . . . . . . . . . . . . 254A.2.4 Model development . . . . . . . . . . . . . . . . . . . . . . . . . . . . 254A.3 Softwood Hemicellulose Hydrolysis Model Activation Energy Calculation . 260A.4 Softwood Hemicellulose Hydrolysis Model Calculation Codes . . . . . . . . 262A.4.1 Dominant equations . . . . . . . . . . . . . . . . . . . . . . . . . . . . 262A.4.2 Model validation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 262A.4.3 Parameter fitting . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 262xvList of TablesTable 2.1 Net primary productivity of different types of vegetation. The table isregenerated according to Sands. . . . . . . . . . . . . . . . . . . . . . . . . 14Table 2.2 Composition of hardwoods and softwoods (wt%). *Others: containingmainly acetyl and uronic acid groups of xylan, some pectins and otherpolysaccharides and inorganics . . . . . . . . . . . . . . . . . . . . . . . . 22Table 2.3 Composition of agricultural residues (wt%). *Others: containing mainlyacetyl and uronic acid groups of xylan, some pectins and other polysac-charides and inorganics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23Table 2.4 Composition of bamboo, agave and AGB (wt%). *Others: containingmainly acetyl and uronic acid groups of xylan, some pectins and otherpolysaccharides and inorganics . . . . . . . . . . . . . . . . . . . . . . . . 24Table 2.5 Effect of different chemical pretreatment technologies on the structure oflignocellulose. H: high effect, M: moderate effect, L: low effect. . . . . . . 49Table 2.6 Summary of pretreatment conditions and resulting yields/conversions . 50Table 2.7 A summary of pretreatment conditions and yields for bamboo and agri-culture residues . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51Table 3.1 Experimental design of softwood hydrolysis conditions (the time listedis isothermal time without conversion to pseudo-time). . . . . . . . . . . . 63xviTable 3.2 Experimental design of factors influencing softwood hydrolysis (the timelisted is isothermal time without conversion to pseudo-time). . . . . . . . 64Table 3.3 Experimental design of hog fuel hydrolysis (the time listed is isothermaltime without conversion to pseudo-time). . . . . . . . . . . . . . . . . . . 64Table 3.4 Milled softwood chip fines particle size distribution . . . . . . . . . . . . 66Table 3.5 Severity factor (calculated from Equation 3.2) and pseudo-time for hy-drolysis of milled softwood chip fines at varying temperature and time. . 68Table 3.6 Compositional analysis of softwood chips, chip fines (Fines), and milledchip fines (Mill) by shipping year and processing year. Weight fraction(%) of components is presented on an moisture free (o.d. softwood) ba-sis; error limits are standard deviation calculated from replicates. *Acidinsoluble residue is the sum of ash and acid insoluble lignin. . . . . . . . 72Table 3.7 Polysaccharides analysis of water extract generated by Dionex ASE 350(% yield of o.d. softwood) supplementary to Table 3.6. Note: S2014M2017 represents shipping year 2014 and milling processed year 2017. . . 73Table 3.8 Compositional analysis of hog fuel (wt% o.d. biomass) with standarddeviation of 2-4 replicates after ±. . . . . . . . . . . . . . . . . . . . . . . . 86Table 4.1 Calibration standard retention time and molar mass. . . . . . . . . . . . . 93Table 5.1 Carbohydrates hydrolysis by population balance model (k and h are hy-drolysis rate constant; a, b are constant). . . . . . . . . . . . . . . . . . . . 108Table 5.2 Softwood hydrolysis condition and use. . . . . . . . . . . . . . . . . . . . 111Table 5.3 Molar mass and SEC retention time of corresponding DP. . . . . . . . . . 114Table 5.4 Composition of softwood on o.d. biomass basis and average hemicellu-lose monomer molar mass. Note: hemicellulose content was correctedby soluble polysaccharides content from extraction. . . . . . . . . . . . . . 116xviiTable 5.5 Breakage function (bi,j) distribution matrix for hemicellulose interval 1(degradation products) to 9 (solids). . . . . . . . . . . . . . . . . . . . . . . 121Table 5.6 The Marquardt parameter (λ) and corresponding weighting factors (wi)for interval 1 to 9 of last regression iteration. . . . . . . . . . . . . . . . . . 131Table 5.7 Parameter regression results of population balance model. Note: 95%confidence limits after ±. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 132Table 5.8 Selection function results of interval 3 to 8. Note: 95% confidence limitsafter ±. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 133Table 5.9 Activation energy and pre-exponential factor for each soluble hemicel-lulose oligomers and solids interval. Note: 95% confidence limits after±. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 136Table 5.10 Comparison to hemicellulose/wood hydrolysis activation energies re-ported in literature. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 138Table 6.1 LBG adsorption isotherm and kinetics experimental design. . . . . . . . . 155Table 6.2 Factors investigated for LBG adsorption: LBG dosage, temperature, NaCladdition and pH and refining. . . . . . . . . . . . . . . . . . . . . . . . . . 156Table 6.3 Experimental design for investigation into the effects of LBG dosage andpulp refining on paper strength. . . . . . . . . . . . . . . . . . . . . . . . . 159Table 6.4 LBG adsorption capacity at equilibrium (qe) and adsorption rate con-stant (k) at temperature range of 25-45 oC as determined by fitting pseudo-second-order kinetics. Note: 95% confidence limits after ±. . . . . . . . . 163Table 6.5 Summary of adsorption capacity and isotherms of polysaccharides ad-sorption to pulp fibre. Note: 95% confidence limits of this work areprovided after ±. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 167Table 6.6 Fibre surface site coverage with varying LBG dosage of o.d. pulp fibreat 25 oC for NBSK pulp refined at 3000 rev calculated from Langmuiradsorption model. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 172xviiiTable 7.1 Experimental conditions for softwood chip pre-hydrolysis and kraft pulp-ing. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 187Table 7.2 Fibre property definition and parameter description tested by PulpEye. . 191Table 7.3 Fraction of solubilized solids after pre-hydrolysis and composition ofpre-hydrolyzed chips (% o.d. chips, average ± standard deviation).Note: no standard deviation for Fines due to inaccurate operation forone replicate. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 196Table 7.4 Summary of pre-hydrolysis combination to kraft pulping Pros and Consand particle size influence on pre-hydrolysis and kraft pulping. . . . . . . 208xixList of FiguresFigure 1.1 Share of global industrial energy consumption by sector, cited fromNapp et al. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2Figure 1.2 Pulp and paper production from 1998-2016 in (a) Canada and (b) Brazil,data from FAO [59] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3Figure 1.3 Block flow diagram of proposed softwood hemicellulose hydrolysis,kraft pulping and oligomers adsorption process. . . . . . . . . . . . . . . 9Figure 2.1 Simplified structure of a woody cell, showing the middle lamella (ML),the primary wall (P), the outer (S1), middle (S2), and inner (S3) layersof the secondary wall, and the warty layer (W) (Adapted from Sjostrom). 17Figure 2.2 Distribution of hemicellulose, cellulose, and lignin through the cell wall(Adapted from Liu). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18Figure 2.3 Categories of hydrothermal pretreatment methods. . . . . . . . . . . . . 26Figure 2.4 Hydrolysis mechanism in acidic condition. Methyl -D-glucopyranoside(1) to D-glucose (5) involving conjugate acid (2 and 4) and cyclic carbo-nium ion (3) intermediates (Adapted from Sjostrom). . . . . . . . . . . . 27xxFigure 2.5 Generalized reaction pathway for glucose decomposition in subcriti-cal and supercritical water conditions (Reprinted with permission fromKabyemela et al.. Glucose and fructose decomposition in subcriticaland supercritical water: detailed reaction pathway, mechanisms, andkinetics. Ind Eng Chem Res 38: 28882895. Copyright 1999 AmericanChemical Society. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34Figure 2.6 Peeling reaction of 1,4-β-D-glucan (cellulose). R=glucan (cellulose) chain.Reaction steps: isomerization (1→2), enediol formation (2→3), β-alkoxyelimination (3→4), tautomerization (4→5), and benzilic acid rearrange-ment (5→6) to epimeric 3-deoxy-2-C-hydroxymethylpentonic acids (glu-coisosaccharinic acid) (6) (Adapted from Sjostrom) . . . . . . . . . . . . . 40Figure 2.7 Stopping reaction. Reaction steps: 1,2-Enediol formation (1→2), β-hydroxy elimination (2→3), tautomerization (3→4), and benzilic acidrearrangement (4→5) to epimeric 3-deoxyhexonic acid end groups (glu-cometasaccharinic acid) (5) (Adapted from Sjostrom) . . . . . . . . . . . 41Figure 3.1 Chemical structure of GGM from softwood (Fengel and Wegener [62]) . 54Figure 3.2 Chemical structure of arabino-4-O-methylglucuronoxylan from softwood(Fengel and Wegener [62]) . . . . . . . . . . . . . . . . . . . . . . . . . . . 55Figure 3.3 A schematic illustration of pseudo-time calculation method. . . . . . . . 67Figure 3.4 Oligomer and monomer yield of o.d. softwood at 160 oC for 28.16 min-utes as a function of agitation rates from 90 r.p.m. to 270 r.p.m. . . . . . . 69Figure 3.5 Oligomer and monomer yield of o.d. softwood at 180oC for 32.19 min-utes as a function of consistency. . . . . . . . . . . . . . . . . . . . . . . . 70Figure 3.6 Oligomer and monomer yield of o.d. softwood for milled chips (Mill) asSection 3.4.1, chip fines with with a thickness less than 2 mm (Fines) andchips with a thickness from 2-6 mm (Chips) at 170oC for 12.09 minuteswith solids consistency of 5wt% as a function of particle size. . . . . . . . 71xxiFigure 3.7 Oligomer and monomer yield of o.d. softwood at 170 oC for 12.09minute and at 180 oC for 23.19 minutes from same batch of chip finesprocessed in 2015 and 2017. M2015 is a short storage batch and M2017is a long storage batch. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74Figure 3.8 (a) Oligomer yield and (b) monomer yield of 25g o.d. softwood at 140-200 oC. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75Figure 3.9 Oligomer composition in hydrolysate at (a) 160oC, (b) 180oC and (c)200oC from 25g o.d. softwood. . . . . . . . . . . . . . . . . . . . . . . . . 76Figure 3.10 Monomer composition in hydrolysate at (a) 160oC, (b) 180oC and (c)200oC from 25g o.d. softwood. . . . . . . . . . . . . . . . . . . . . . . . . 78Figure 3.11 (a) Total mannan yield (oligomer and monomer) and (b) mannose yieldrelative to mannan in 25g o.d. untreated softwood mill at 160-200 oC. . . 79Figure 3.12 (a) Total galactan yield (oligomer and monomer) and (b) galactose yieldrelative to galactan in 25g o.d. untreated softwood mill at 160-200 oC. . . 79Figure 3.13 (a) Total glucan yield (oligomer and monomer) relative to glucan in-cluding cellulose and hemicellulose and (b) hemicellulose glucan rela-tive to potential hemicellulose glucan in 25g o.d. untreated softwoodmill at 160-200 oC. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80Figure 3.14 (a) Total xylan yield (oligomer and monomer) and (b) xylose yield rela-tive to xylan in 25g o.d. untreated softwood mill at 160-200 oC. . . . . . . 81Figure 3.15 (a) Total arabinan yield (oligomer and monomer) and (b) arabinoseyield relative to arabinan in 25g o.d. untreated softwood mill at 160-200 oC. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82Figure 3.16 Hydrolyzed solids (a) AIR relative to AIR of untreated 25 g of o.d. soft-wood mill and (b) ASL relative to ASL of untreated 25 g of o.d. soft-wood mill at 160 oC, 180 oC and 200 oC as a function of pseudo-time. . . 83xxiiFigure 3.17 (a) Hemicellulose yield of o.d. softwood including oligomer and monomer(b) ratio of solubilized ASL to hemicellulose (wt/wt) compare to solu-bilized ASL of o.d. softwood chip fines at 160 oC to 200 oC. . . . . . . . . 84Figure 3.18 Solubilized solids of 25 g o.d. softwood at 160oC, 180 oC and 200 oC asa function of pseudo-time. . . . . . . . . . . . . . . . . . . . . . . . . . . . 85Figure 3.19 Oligomer yield from o.d. chip fines and hog fuel of hydrolysis (a) at160oC and (b) at 200oC; Monomer yield from o.d. chip fines and hogfuel of hydrolysis (c) at 160oC and (d) at 200oC as a function of time. . . 87Figure 4.1 Logarithm of calibration standard molar mass plotted against SEC re-tention time. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 94Figure 4.2 Concentration calibration curve of (a) Std 1, (b) Std 3, (c) Std 5, (d) Std7, and (e) Std 9. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95Figure 4.3 SEC chromatograph of RI detector signal of (a) calibration standardwith varying concentration of Std 1, Std 3, Std 5, Std 7, and Std 9 (b)hydrolysate at 180 oC for varying residence time. . . . . . . . . . . . . . . 96Figure 4.4 Number-average molar mass (Mn, square black dot), weight-averagemolar mass (Mw, round red dot) and z-average molar mass (Mz, trian-gle blue dot) at: (a) 160oC (b) 180 oC (c) 200 oC (d) Polydispersity index(Mw/Mn) at 160oC (square black dot), 180 oC (round red dot) to 200 oC(triangle blue dot). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 98Figure 4.5 Weight-average molar mass (Mw) of hemicellulose from 160oC to 200oC as a function of hydrolysate pH. . . . . . . . . . . . . . . . . . . . . . . 100Figure 4.6 Chromatogram of hydrlysate as represented by SEC coupled with RIdetector. (a) Hydrolysis was conducted at 180 oC for 74.19 min and at200 oC for 18.39 min; these conditions have equal severity as calculatedby Equation 3.2. (b) Hydrolysis was conducted at 180 oC and residencetime was varied. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100xxiiiFigure 5.1 Mechanism of hemicellulose molar mass evolution in the range of 9intervals including soluble oligomer with DP1-140, insoluble softwoodhemicellulose, and degradation products. P is a primary pathway toproduce soluble hemicellulose from insoluble wood; S is a secondarypathway to produce soluble hemicellulose from a primary oligomer. . . 113Figure 5.2 SEC chromatograph of RI detector signal with model interval divisionof hydrolysate at 180 oC for varying residence time. . . . . . . . . . . . . 117Figure 5.3 Comparison of model fit (black and red line) to experimental data (blackand red dot) at 160 oC for softwood hemicellulose including solubleoligomer of DP 1-140, solids hemicellulose and degradation products. . 123Figure 5.4 Comparison of model fit (black and red line) to experimental data (blackand red dot) at 180 oC for softwood hemicellulose including solubleoligomer of DP 1-140, solids hemicellulose and degradation products. . 124Figure 5.5 Comparison of model fit (black and red line) to experimental data (blackand red dot) at 200 oC for softwood hemicellulose including solubleoligomer of DP 1-140, solids hemicellulose and degradation products. . 125Figure 5.6 Maximum yield (red line and round hollow dot) and correspondingtreatment time (black line and square solid dot) from model predictionof oligomer with DP 21-140 at (a) 160 oC (b) 180 oC and (c) 200 oC. . . . . 126Figure 5.7 Validation of model fit to experimental data at 170 oC. . . . . . . . . . . . 127Figure 5.8 Optimization of treatment time and predicted yield of oligomer by modelat each interval at 160 oC (black line), 180 oC (red dot line), and 200 oC(blue dash line). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 128Figure 5.9 Hemicellulose component mass recovery from hydrolysate and hydrolyzedsolids at 160 oC to 200 oC (a) galactan (b) arabinan (c) xylan (d) glucan(e) mannan. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 134Figure 6.1 Structure of LBG cited from de Jong and van de Velde [48]. . . . . . . . . 144xxivFigure 6.2 The fraction of LBG absorbed to NBSK pulp at 25 oC as function of time.LBG dosage was 0.2 wt% relative to o.d. pulp and fibre consistency was0.5 wt%. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 161Figure 6.3 Linear fit of (a) pseudo-first-order kinetics and (b) pseudo-second-orderkinetics of LBG adsorption to NBSK pulp at 25 oC. (c) Linear fit ofpseudo-second-order kinetics of LBG adsorption to NBSK pulp at 35oC and (d) at 45 oC. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 162Figure 6.4 Linear regression of Arrhenius equation to determine activation energyof LBG adsorption to NBSK pulp at 25 oC to 45 oC based on the pseudo-second-order kinetic model. . . . . . . . . . . . . . . . . . . . . . . . . . . 164Figure 6.5 LBG adsorbed to unrefined NBSK pulp fibre as a function of LBG dosageto o.d. fibre after 10 minutes at 25 oC. (a) The fraction of LBG absorbedand (b) the absolute mass of LBG adsorbed. . . . . . . . . . . . . . . . . . 165Figure 6.6 Linear fit of Langmuir model (a) and Freundlich model (b) at 25 oC ofLBG adsorption to unrefined NBSK pulp. . . . . . . . . . . . . . . . . . . 166Figure 6.7 (a) Amount of LBG (mg·g−1 o.d. pulp) absorbed to NBSK pulp after10 minutes as a function of temperature at two dosage levels (0.2 wt%and 4.4 wt% of o.d. fibre). (b) Linear fit of Langmuir model for LBGadsorption at 35 oC of LBG adsorption to NBSK pulp. . . . . . . . . . . . 169Figure 6.8 Fibre surface site coverage (θs) calculated from Langmuir model at 25oC and 35 oC for unrefined NBSK pulp. . . . . . . . . . . . . . . . . . . . 170Figure 6.9 Linear fit of Langmuir model at 25 oC for LBG adsorption to NBSK pulprefined at 3000 rev after 10 minutes. . . . . . . . . . . . . . . . . . . . . . 171Figure 6.10 The adsorbed fraction of LBG on NBSK pulp fibre with varying sodiumchloride concentration (a) 0-0.0125 mol·L−1 (b) 0.05-1 mol·L−1 after 10minutes at 25 oC, LBG dosage 0.2wt% of o.d. fibre. . . . . . . . . . . . . . 173xxvFigure 6.11 The adsorbed amount of LBG on NBSK pulp fibre with varying pH ofpulp suspension after 10 minutes at 25 oC, LBG dosage 0.2wt% of o.d.fibre. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 174Figure 6.12 NBSK pulp and paper properties as a function of LBG dosage and PFIrefining level: (a) tensile index, (b) burst index, (c) tear index, (d) free-ness, (e) brightness and (f) scattering coefficient. . . . . . . . . . . . . . . 175Figure 7.1 Hemicellulose yield from o.d. chips as a function of particle size. . . . . 192Figure 7.2 Weight-average molar mass of hemicellulose oligomers as a function ofparticle size. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 193Figure 7.3 Yield of hemicellulose oligomers in the hydrolysate varying with parti-cle size (a) degradation products and DP 1-20, (b) DP21-40 and DP41-60, (c) DP 61-80 and DP 81-100, (d) DP 101-120 and DP 121-140, and (e)pre-hydrolyzed solids. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 194Figure 7.4 Hydrolysate hemicellulose monomer yield from o.d. chips (arabinose,galactose, glucose, xylose and mannose) as a function of particle size. . . 195Figure 7.5 Hydrolysate hemicellulose oligomer yield of o.d. chips a) arabinan, xy-lan and galactan b) glucan and mannan as a function of particle size. . . 195Figure 7.6 (a) Screened fibre yield from o.d. raw chips, (b) kappa number of kraftpulp, and (c) percent rejects after pulp screening relative to o.d. chipsas a function of particle size. Pulping was conducted to an H-factor of1600 and 1900. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 198Figure 7.7 (a) Alpha-cellulose content of screened pulp fibre (wt%) and (b) totalsoluble hemicellulose content (wt%) relative to o.d. chips as a functionof particle size. Total soluble hemicellulose is the sum of soluble arabi-nan, xylan, galactan, mannan and glucan. . . . . . . . . . . . . . . . . . . 201xxviFigure 7.8 Dimensions of kraft pulp fibres: (a) mean fibre length (mm) (b) meanfibre width (µm) (c) aspect ratio, and (d) kraft pulp fines content (wt%)as a function of particle size. . . . . . . . . . . . . . . . . . . . . . . . . . . 203Figure 7.9 Kraft pulp fibre kinks number per mm as a function of particle size. . . . 205Figure 7.10 WRV of kraft pulp as a function of particle size. . . . . . . . . . . . . . . 206Figure 7.11 EA consumption and sulphidity consumption of black liquor from blend10wt% softwood chips. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 207Figure A.1 Schematic illustration of hemicellulose mass transfer of softwood chipimmersion in water. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 254Figure A.2 Linear regression of activation energy determination. . . . . . . . . . . . 261xxviiList of AbbreviationsAGB Agave bagasseAGX arabino-4-O-methylglucurono-xylansAIR acid insoluble residueASL Acid soluble ligninDP degree of polymerizationEA effective alkaliEJ ExajoulesGGM O-acetyl-galactoglucomannansHMF hydroxymethylfurfuralHPLC high-performance liquid chromatographyLAP Laboratory Analytical ProceduresLBG locust bean gumLCC lignin-carbohydrate complexesNBSK Northern Bleached Softwood KraftNREL National Renewable Energy LaboratoryxxviiiO.D. oven-driedPCC-1600 pre-hydrolyzed chips cooked to an H-factor of 1600PDI polydispersity index , a representation of abdsPEG polyethylene glycolRCC-1600 raw chips cooked to an H-factor of 1600RCC-1900 raw chips cooked to an H-factor of 1900RI Refractive IndexSEC size exclusion chromatographySS sum of squaresSTD standardWRV water retention valuexxixList of SymbolsA Pre-exponential factor (L·mol−1 ·min−1)Ap Area beneath the heating and cooling temperature-time profiles (oC ·min)A∗ Interfacial area of softwood chip (m2)a Pre-exponential factor of selection function (min−1)Bi,j Cumulative breakage functionbi,j Breakage functionBi Biot numberC0L Initial LBG concentration (mg·L−1)C0P Initial o.d. pulp fibre concentration (g·g−1)CSL Soluble LBG concentration in supernatant (mg·L−1)Ce Equilibrium concentration of LBG in the aqueous phase (mg·L−1)CeSe Equilibrium concentration of adsorped LBG on fibre surface (mg·g−1)Ci Normalized concentration in a size interval iCsL LBG concentration in supernatant after adsorption (mg·L−1)D Diffusion coefficient (m2 · s−1)xxxDa Damkohler numberEa Activation energy (kJ ·mol−1)fL Fraction of LBG adsorbed to the pulp (%)h Mass transfer coefficient (m·s−1)Ke Equilibrium constantki Rate constant (min−1 or g·mg−1 ·min−1)L Half-width of the softwood chip (m)Mn Number-average molar mass (kg·mol−1)Mw Weight-average molar mass (kg·mol−1)Mz Z-average molar mass (kg·mol−1)mL Absolute amount of LBG adsorped to the pulp (mg·g−1)N Total number of intervalsn Freundlich constantnj Number of experimental data points for data set jP Primary pathwayQmax Maximum adsorption capacity (mg·g−1)qe Equilibrium adsorption capacity (mg·g−1)qt Concentration of LBG adsorbed at time t (mg·g−1)R Gas constant (J ·K−1 ·mol−1)R0 Severity factorxxxiS Secondary pathwaySe Equilibrium empty site concentration on the surface of pulp fibres (mg·g−1)Sei Selectivity of oligomer with molar mass interval iSi Selection function in a size interval i (min−1)T Temperature (oC)t Time (min)tp Pseudo-time (min)Wi Weight fraction of material in a size interval iwi Weighting factor for interval iV Volume (m3)Xd Oven dried fibre weight (g)Xi Initial o.d. pulp samples (g)Xk Klason lignin content of fibre (g)Xresidual Residual lignin content of pulp fibre after post-hydrolysis (g)Xs Acid soluble lignin content of pulp fibre (g)Xα Alpha-cellulose content in pulp fibre (%)xi Particle size of interval ixxxiiGreek symbolsα Order of selection function dependence on molar massη Dimensionless lengthθ Dimensionless concentrationθs Substrate surface coverageλ Marquardt parameter valueσi Variance for interval iτ Dimensionless timeΩ1 Wood chips diffusion domainΩ2 Bulk solution diffusion domainSubscripts and superscripts0 Intitalh Hemicellulosei Indexj IndexN Total number of intervalsp PseudoR ResidualS Solublexxxiiisoluble Soluble oligomers in the aqueous phasesolid Insoluble hemicellulose in wood matrix* Interface∞ Bulk solutionchip Wood chipsolution Bulk solutionxxxivAcknowledgementsThis work was supported by Canfor Pulp Products, the Natural Science and Engineer-ing Research Council of Canada, and Domtar Corporation. UBC Chemical and BiologicalEngineering (CHBE), Pulp and Paper Centre (PPC), British Columbia Institute of Tech-nology (BCIT), and Canfor Pulp Innovation provided the research facilities, training andcollaborations.There are many helps that have pushed me through the whole way. I would like togive special thanks to Dr. Heather L. Trajano, who offered me the opportunity to work onthe project and guided me since day one, helping me to pick up the frustrated heart dur-ing setback and advising me to keep the scientific judgment during success. Also, I offergratitude to my supervisory committee members, Dr. Rodger P. Beatson, Dr. D. MarkMartinez and Dr. Paul Bicho, who inspired me when I was facing pulping or modelingor chemistry problems and provided many collaboration and training opportunities. Iwould also like to thank my coworkers for their generous help during the past five years,Xuefeng Chang, Kenny Tam, George Soong, Dr. Zhaoyang Yuan, Reanna Seifert, RichardRyoo, Robertus Putra, Varun Rangu, Lingxiu Zhu, Miguel Villalba, Masoud Daneshi, andmany other PPC and CHBE staff.Many Co-op students, BCIT CENV4400 project students and high school volunteerstudents contributed to the thesis and worked together with me. I would like to acknowl-edge their hard work. Xinyi Chen and Rylan Martinez worked on softwood hydroly-sis. Miaoran Li, Yu Wu and John Shadarevian worked on LBG adsorption and micro-xxxvfibrillated cellulose production. Vicki Chen, Nick Park, Carrie Wang, Chelsea Park, Kim-berly Hoang, Samantha Yip, Sahar Shahabian, Nelson Fu, Brendan Confrey, Bruce Qin,Bruce Li and Roy Hung worked on LBG adsorption and paper strength. Hong Ma workedon pre-hydrolysis pulping.It is such a rewarding journal for me to walk out of the program with hope to be-come a better person and better engineer, looking at problem with much more flexibleperspective and open mind. Many thanks to my beloved parents and friends, who giveme courage, persistence, love and hope to explore the world and chase my dream.xxxviDedicationTo My ParentsXiumin Wang and Yanchang ChenxxxviiChapter 1Introduction1.1 Motivation and Background1.1.1 Renewable energyCarbon, the fourth most abundant element in the universe, creates life on earth (He et al.[89], Levi and Rosenthal [149]). Human combustion of fossil fuels, accumulated overmillions of years underground, has released carbon into atmosphere and disrupted theinherently slow carbon cycle (Riebeek [225]). Global warming caused by vast emissionsof carbon jeopardize the living environment on the planet (Huisingh et al. [100], Vitousek[311]).Renewable energy sources and sustainable technology are emerging as responses toglobal warming and climate change (Clark [38]). Currently, fuels, chemicals, and syn-thetic materials are primarily generated by the petro-chemical industry (Poliakoff and Li-cence [214]). There is strong interest in the production of materials, chemicals, and fuelsfrom renewable resources such as biomass. The envisioned biorefinery will a sustainableproducer of traditional products such as pulp and paper, but also valuable compoundssuch as esterified fermentation acids (Clark [38]).11.1.2 Pulp and paper industryPulp and paper industry is energy intensive, and produces approximately 40 Mt of CO2per year in Europe (Huisingh et al. [100]). Pulp, paper and wood processing accounted for7% of total global energy consumption in 2010, ranking it fifth among industrial sectors(Figure 1.1, Napp et al. [195]).Figure 1.1: Share of global industrial energy consumption by sector, cited from Nappet al.North American pulp and paper producers face significant challenges from develop-ing countries in tropical areas (Van Heiningen [305]). Low labor costs and low feedstockcosts threaten the profitability of North American pulp mills. Pulp and paper produc-tion in Canada began to decline in 2004 (Figure 1.2a). Meanwhile, pulp and paper pro-duction in Brazil has increased rapidly (Figure 1.2b); recent chemical pulp productionin Brazil is almost double that in Canada. Brazilian chemical pulp is mainly hardwoodkraft pulp from Eucalyptus while Canada produces substantial volumes of softwood kraftpulp from spruce, pine, and fir. Against this backdrop, it has become crucial to developtechnical and economic strategies to relieve competition. Integrating new value-addedproducts with traditional pulp and paper operations could address these issues.21998 2000 2002 2004 2006 2008 2010 2012 2014 20160246810121416182022 Mechanical wood pulp Chemical wood pulp Dissolving wood pulp Newsprint Printing and writing papers Household and sanitary papers Packaging and paperboardCanada production (106 tonnes)Year(a)1998 2000 2002 2004 2006 2008 2010 2012 2014 20160246810121416182022(b) Mechanical wood pulp Chemical wood pulp Dissolving wood pulp Newsprint Printing and writing papers Household and sanitary papers Packaging and paperboardBrazil Production (106 tonnes)YearFigure 1.2: Pulp and paper production from 1998-2016 in (a) Canada and (b) Brazil,data from FAO [59]1.1.3 BiorefineryThere have been significant efforts devoted to the conversion of wood into biofuels andbiochemicals as part of the biorefinery (Amidon and Liu [9], Liu et al. [167], Ruiz et al.[236]). The biorefinery can be integrated with existing pulp and paper mills to producebiochemicals and biomaterials such as ethanol, lactic acid, and aromatics (Amidon andLiu [9], Moshkelani et al. [191], Van Heiningen [305]).Lignocellulosic biomass is the most abundant raw biomass material on Earth (Han-nuksela et al. [88]). Forests account for 89.3% of the total standing biomass (Kajaste [117])and are the primary resource for cellulosic fibre production. The primary forest resourcein British Columbia is softwood and this material is used to produce Northern BleachedSoftwood Kraft (NBSK); the significant tensile strength of NBSK pulp distinguishes BritishColumbian producers in the global market.Biomass is a complicated matrix of cellulose, hemicellulose, and lignin (Galbe andZacchi [68]). Cellulose accounts for 35-50 wt% of wood, and is a linear crystalline poly-mer composed of glucose units that is difficult to hydrolyze (Smook et al. [271], Taarninget al. [282]). Lignin, accounting for 10-25 wt% of wood, is an amorphous polyphenolcomposed of aromatic alcohols, and is resistant to hydrolysis (Smook et al. [271], Taarn-3ing et al. [282]). Hemicelluloses are branched, amorphous polysaccharides composed ofdifferent monosaccharides and functional groups (Scheller and Ulvskov [250]), and ac-count for roughly 25-35 wt% of total organic material in wood (Ren and Sun [221]). Thematrix serves as a physical barrier and hinders fractionation of biomass components tovaluable bioproducts (Altaner and Jarvis [7], Cosgrove and Jarvis [44], Dammstro¨m et al.[46], Zhang et al. [334]). Mechanistic insights into processes, systematically applied en-gineering fundamentals, and chemical characterization are needed to understand andharness biomass fractionation. The primary focus of this thesis was the recovery andutilization of hemicellulose from Canadian softwood residues.1.1.4 Hemicellulose hydrolysisHydrolysis is hydrothermal treatment to break a carbohydrate into monosaccharides bycleavage of glycosidic bonds (Smook et al. [271]).Multiple technologies have been ex-plored using liquid water as the media, with or without addition of chemicals (acids oralkalis).Treatment with water under high temperature and pressure without other chemicalsis called autohydrolysis (Ruiz et al. [236]). Autohydrolysis is desirable as it does notrequire corrosive chemicals, expensive materials of construction, and avoids generationof neutralization salts. Moreover, due to the nature of lignocellulosic biomass, naturalacids (acetic or uronic acid) are released causing pH to decrease (Leppa¨nen et al. [148], Puet al. [216], Rissanen et al. [227, 228]). The dissociation of water at high temperature alsogenerates protons and reduces pH. Despite an extensive literature, there is no optimized,universal autohydrolysis process due to the heterogeneity of biomass.This thesis begins by reviewing the literature of biomass hydrolysis (Chapter 2). Thecomposition and response to hydrolysis of four major biomass types: wood, bamboo,agricultural residues and agave, are discussed. The reaction pathways of individualbiomass components (hemicellulose, cellulose, lignin, extractive and ash) under different4treatment conditions (acidic and alkaline) are also comparatively reviewed. This reviewserves as a guide for subsequent chapters.Autohydrolysis of softwoods is generally less studied and the body of literature iseven smaller for softwood residues from pulp mills. Large quantities of residues aregenerated by daily mill operations and most of the residues are burned to supply processheat and electricity. This thesis investigated autohydrolysis of chip fines and hog fuel.Chip fines are undersized chips screened out prior to being fed to the digester. Hogfuel is bark generated by sawmills which is bought and used as a fuel. The influence oftemperature, time, solids consistency, agitation rate, and particle size on autohydrolysisof chip fines and hog fuel was investigated in Chapter 3. These results also informedselection of operating conditions for subsequent chapters.Of the three components of lignocellulose, hemicellulose is the most easily hydrolyzedto soluble oligomers and monomers (Fengel and Wegener [62], Ren and Sun [221], Schellerand Ulvskov [250]). Oligomers or oligosaccharides are short-chain polymers of monosac-charides linked by α and/or β glycosidic bonds and have a degree of polymerization(DP) ranging from 2 to 40 (Wyman [322]). Several studies (Garrote et al. [72], Jacob-sen and Wyman [106], Li et al. [150]) have shown that increasing severity (i.e. highertemperature, longer residence time, lower pH) leads to degradation of hemicelluloseoligomers to monomers, furfural, hydroxymethylfurfural (HMF) and light organics (Puet al. [216]). The concentration and size of oligomers is not easily quantified therefore atwo-dimensional calibration method for size exclusion chromatography (SEC) of oligomerswas established in Chapter 4. The diversity of hemicellulose oligomers enables many po-tential applications but controlling yield and oligomer molar mass is challenging there-fore a population balance model of softwood hemicellulose hydrolysis was developed inChapter 5.51.1.5 Hemicellulose hydrolysis integration to kraft pulpingThe integration of hydrolysis to kraft pulping and the role of hemicellulose oligomers tokraft pulp are proposed in this thesis. One is hemicellulose adsorption to NBSK pulp asstrength additive. Another is pre-hydrolysis prior to kraft pulping to produce kraft pulpand hemicellulose oligomers.1.1.5.1 Hemicellulose as a paper strength additiveRefining is an effective and the most common method to improve paper strength butconsumes high energy (Lindstro¨m et al. [160], Niskanen [197], Swanson [280]). Strengthadditives were developed to enhance paper strength and reduce energy consumption(Bhaduri et al. [19], Lindstro¨m et al. [160], Silva et al. [256], Swanson [280]). Bhaduriet al. found that adsorption of rhamnose- and galactose-based polysaccharides to jutestick pulp increased tensile and burst strength of paper. A maximum 36% energy savingwas obtained by 2 wt% hemicellulose dosage in the pulp (Bhaduri et al. [19]). Silva et al.reported a similar energy reduction, around 40%, by applying a relatively high xylandosage.Hunter was the first to propose that Abelmoschus manihot, a carbohydrate, couldbe used to enhance the strength of paper made from long-fibred pulp (Swanson [280]).The earliest paper strength additives, also called beater additives, were polysaccharideswith structural affinity for cellulose such as locust bean gum (LBG) and guar gum (Leech[143], Lindstro¨m et al. [160], Most [194], Russo [237], Swanson [280]). Extensive study ofstarch followed and many fundamental papers were published (Hedborg and Lindstrom[92], Shirazi et al. [255], van de Steeg [301], van de Steeg et al. [302, 302, 302], Zakrajsˇeket al. [333]). With the advent of the biorefinery, research on utilization of hemicellulose re-covered from wood, such as O-acetyl-galactoglucomannans (GGM) and xylan, as strengthadditive has renewed (Hannuksela et al. [87, 88], Linder et al. [158], Silva et al. [256], Su-urnakki et al. [279]). However, there is less fundamental understanding of application of6hemicellulose polysaccharides such as GGM and LBG (Hannuksela et al. [87, 88], Leech[143], Most [194], Russo [237], Silva et al. [256], Suurnakki et al. [279], Swanson [280]).This thesis selected LBG as a model compound to investigate its adsorption to NBSKpulp from a fundamental perspective in Chapter 6.1.1.5.2 Pre-hydrolysis kraft pulpingKraft pulping separates cellulosic fibres from wood by dissolving hemicellulose and ligninfrom fibre walls and middle lamella (Mimms et al. [185]). Under typical kraft pulping con-ditions, approximately 75% of hemicellulose and 80-90% of lignin are removed (Ek et al.[53], Mimms et al. [185]). Hemicellulose extracted prior to kraft pulping could be utilizedto produce valuable chemicals.Hot water hydrolysis or autohydrolysis is an environmental-friendly approach forhemicellulose extraction prior to pulping relative to alkaline, acidic or solvent extraction.Since water is the only chemical applied, requirements for corrosion-resistance, chemicalpost-treatment, and hydrolysate management are minor and inexpensive (Fisˇerova´ et al.[63], Liu et al. [163], Lu et al. [171]). During hydrolysis, most of the hemicellulose anda portion of lignin will dissolve, cellulose crystallinity may be reduced and wood chipporosity will increase (Chen et al. [29], Lu et al. [171], Mosier et al. [193]). These structuralchanges facilitate subsequent pulping by increasing chemical mass transfer and reducingenergy requirements (Liu et al. [165]). Hemicellulose is easily removed by hydrolysis andkraft pulping but its important role in final fibre properties is not thoroughly understood.There is limited fundamental understanding of hemicellulose mass transfer within woodchips. In this thesis, pre-hydrolysis and kraft pulping of commercial pulp chips was in-vestigated in Chapter 7.1.2 Research HypothesesThe proposed work will enable hemicellulose oligomer production from readily availablepulp mill residues such as chip fines and hog fuels. The utilization of non-white wood7material is consistent with the ultimate biorefinery goal of generating renewable biochem-icals and biomaterials. The use of hemicellulose oligomers as to improve paper strengthwill enhance the competitive advantage of NBSK pulp in the global marketplace.The envisioned process raises a number of research questions:1. What is the response of different biomass species to hydrolysis?(a) What is the correlation between biomass composition and the response to hy-drolysis?2. How can hemicellulose extraction from softwood be controlled?(a) What are favourable conditions for producing high yields of soluble hemicel-lulose derivatives or high molar mass hemicellulose oligomers?(b) How does the composition of hydrolysate and hydrolyzed solids change withchanges to hydrolysis operating conditions?(c) How does hemicellulose molar mass evolve during hydrolysis?3. How can hydrolysis be integrated into current kraft pulping process?(a) What are the effects of pre-hydrolysis on subsequent kraft pulping?(b) If chip fines are used, what is the particle size effect on pre-hydrolysis and kraftpulping?(c) If pre-hydrolyzed chips are kraft pulped, will the pulp fibre exhibit the sameproperties as pulp produced from untreated chips?4. How to optimize hemicellulose adsorption and effects on paper strength?(a) How to control the hemicellulose adsorption to NBSK pulp: adsorption isothermsand kinetics, favorable conditions?(b) What are the effects of hemicellulose adsorption to pulp strength?81.3 Thesis OrganizationThe envisioned process of hemicellulose oligomer production, adsorption, and integra-tion to kraft pulping is illustrated in Figure 1.3. Pulp mill residues will be hydrolyzedwith water to produce hemicellulose oligomers. Solid residues separated from the hy-drolysis step will be burned or subjected to kraft pulping. Oligomers will be adsorped toNBSK pulp to enhance paper strength. The essential processes are softwood hydrolysis,oligomer adsorption, and kraft pulping of pre-hydrolyzed softwood chips.SoftwoodHydrolysisPre-hydrolyzed solidsHydrolysateAdsorption Locust bean gumKraft pulpingNBSK pulpPaperFigure 1.3: Block flow diagram of proposed softwood hemicellulose hydrolysis, kraftpulping and oligomers adsorption process.9This thesis is divided into several chapters:1. Chapter 1 introduces the background, motivation, research questions and thesisstructure.2. Chapter 2 is the literature review of hydrolysis of wood, bamboo, agriculture residueand agave under acidic and alkaline conditions.3. Chapter 3 presents a study of softwood hydrolysis to produce hemicellulose oligomers.Comprehensive characterization of hydrolysate and pre-hydrolyzed solids providesa guide for selection of hydrolysis conditions for Chapter 5 and Chapter 7.4. Chapter 4 describes the method development for molar mass characterization ofhemicellulose oligomers produced by softwood hydrolysis in Chapter 3.5. Chapter 5 develops a population balance model of hemicellulose hydrolysis thatcan describe molar mass evolution by processed results from Chapter 4.6. Chapter 6 investigates the adsorption of hemicellulose model compound LBG toNBSK pulp fibre. Adsorption isotherms, kinetics, influencing factor analysis, andpaper strength modification are presented.7. Chapter 7 presents kraft pulping of commercial softwood chips with pre-hydrolysisand compares the changes in chip and fibre composition and pulp fibre properties.The analysis of hemicellulose hydrolysis of varying particle size is also reported.8. Chapter 8 summarizes the primary thesis conclusions, contributions and recom-mendations for future work.10Chapter 2Response of Biomass Species toHydrolysis2.1 Introduction2.1.1 MotivationConsiderable effort has been devoted to the development of biofuels from renewable lig-nocellulosic biomass. To resolve the challenges associated with the structural barrier oflignocellulosic biomass, hydrothermal pretreatment is applied to alter the structure andimprove the accessibility of carbohydrate sugars to microorganisms or chemicals in thesubsequent conversion processes. Hydrothermal pretreatment takes advantage of highmoisture content of biomass and efficiently converts polysaccharides into monomericsugars and their corresponding degradation products. To achieve this goal, multiple tech-nologies have been explored using liquid water as the media, with or without addition ofchemicals (acids or alkalis). However, there are difficulties developing an optimized anduniversal treatment approach due to the heterogeneity of biomass.In this chapter, four major biomass types: wood, bamboo, agricultural residues andAgave, are discussed and compared with respect to feedstock composition and response11to the various hydrothermal pretreatments. Moreover, the reaction pathways of indi-vidual biomass components (hemicellulose, cellulose, lignin, extractive and ash) underdifferent treatment conditions (acidic and alkaline) are also comparatively reviewed.2.1.2 History, current state-of-the-art and future developmentRecent concerns over climate change and diminishing fossil fuel reserves have broughtrenewed attention to lignocellulosic biomass as a source of renewable energy and sus-tainable fuels. Lignocellulosic feedstocks, such as wood, grass, and agricultural andforest residues, are the most abundant resources on the earth, with approximately 200billion tons generated annually (Zhang and Cai [335], Zhang [336]). In 2008, the globalenergy supply from biomass was approximately 50 Exajoules (EJ), equivalent to approxi-mately 10% of the worlds annual primary energy consumption (Dornburg et al. [49], FAO[60], Sands [245], WEC [318]). It has been predicted that bioenergy will provide 200 to 500EJ · year−1 in 2050 when the global primary energy demand will be 600 to 1000 EJ · year−1Dornburg et al. [49], Sands [245].It is desirable to produce biofuels from low-cost, inedible lignocellulosic materials butthere are a number of challenges associated with using lignoellulose: feedstock hetero-geneity, high moisture content and low energy density (Balat and Balat [15], Burkhardt[27]). To efficiently convert biomass to biofuels and biochemicals, three approaches, hy-drothermal treatment, pyrolysis and gasification, have been investigated and developed.Hydrothermal treatment takes advantage of the moisture content of biomass by usingwater in the process and is an inexpensive and environmental friendly option (Silveiraet al. [257]). Moreover, the hemicellulose and organics dissolved in the liquid phase canbe recovered for high-value products (Jun et al. [114], Liu et al. [162], Shen et al. [254]).122.1.3 Feedstock crop, production and utilization2.1.3.1 WoodWoody biomass from forests is the most plentiful biomass feedstock on earth. Forestsaccount for 89.3% of the total standing biomass, equivalent to 73 billion tons (Amidon andLiu [9], Klass [129] ). Forest productivity is determined mainly by temperature and wateravailability (Sands [245]). As shown in Table 2.1, forests have the largest productivity,at a maximum of 2.8 kg·m−2 · year−1. Approximately 45% of this inventory is tropicalrainforest (Sands [245]).Wood can be divided into two categories: gymnosperms and angiosperms. Conifer-ous woods or softwoods belong to the first and hardwoods to the second (Sjostrom [262]).Gymnosperm forests containing Pinus (pines), Picea (spruce) and Abies (fir) dominatethe cool temperate latitudes and the higher elevations of the northern hemisphere (Galbeand Zacchi [68]). Conifers are long-lived species, but have a lower productivity than an-giosperms. Consequently, angiosperms have progressively displaced conifers, except forthe cooler parts of the northern hemisphere (Sands [245]).Wood, as a solid fuel, has been a major energy source throughout the history of hu-man society. Wood is also used for timber and production of pulp and paper (Pelaez-Samaniego et al. [211]). Energy, specifically heat and electricity, continues to be the majorapplication of wood, accounting for up to 54.7% of the 2011 round wood harvest (Sands[245]). There have been significant efforts devoted to the conversion of wood into biofuelsand biochemicals as part of the biorefinery (Amidon and Liu [9], Liu et al. [167], Ruiz et al.[236]). The biorefinery can be integrated into existing forestry operations such as pulp andpaper mills to produce high-value biochemicals and biomaterials such as ethanol, lacticacid, and aromatics (Amidon and Liu [9], Moshkelani et al. [191], Van Heiningen [305]).The major drawbacks of using woody biomass in the biorefinery are the high energy de-mand for size reduction, the high transportation costs, and the need for high chemical13Table 2.1: Net primary productivity of different types of vegetation. The table is re-generated according to Sands.Land type Vegetation type Net primaryproductivity(kg·m−2 · year−1)ForestsTropical rainforest 1.75-2.8Tropical deciduous forest 1Temperate humid forest 1Temperate dry forest 0.8Boreal forest 0.65WoodlandDwarf and open scrub 0.6Tundra 0.16Desert scrub 0.07GrasslandTropical grassland 0.8Temperate grassland 0.8DesertDry desert 0.003Ice desert 0Cultivated land 0.65Fresh waterSwamp and marsh 2Lake and stream 0.5charges.2.1.3.2 BambooBamboo is a perennial species belonging to Graminease family but has a chemical compo-sition comparable to wood. It is highly abundant in tropical, subtropical, and temperateregions worldwide and encompasses over 1250 species within 75 genera (Scurlock et al.[251]). Bamboos are highly abundant in tropical and subtropical areas around the world,for example, about 7.6 million hectares of bamboo forests in China alone (Littlewood et al.[161]). Bamboo plantations also have several advantages such as limiting soil erosion incropping systems, improving water quality, and having lower chemical and nutrient re-14quirements (Garcı´a-Aparicio et al. [69]). Moreover, compared to woods, most bamboospecies need less time (3-5 years) to mature (Gratani et al. [79], Krzesin´ska et al. [134], Luoet al. [173]). Therefore, bamboo is a promising species for cultivation on marginal landfor bio-based products (Littlewood et al. [161]).Traditionally, bamboo is used for production of handicrafts, furniture, kraft pulp, re-inforcing fibre and high purity dissolving pulp (Batalha et al. [17], He and Yue [90], Luoet al. [174], Ma et al. [176], Salmela et al. [243], Vu et al. [313]). Recently, there have beenefforts to determine the potential of bamboo as a feedstock for production of second andthird generation biofuels such as cellulosic ethanol and hydrogen (Li et al. [154], Sathit-suksanoh et al. [248]). However, bamboo has a very dense structure, which makes it moredifficult to pretreat than other feedstocks such as agricultural residues and some woods.2.1.3.3 Agricultural residuesAgricultural residues, obtained from annual crops, have a high cellulose content and arethus also promising feedstocks for biorefinery applications. There is a wide range ofagricultural residues, but this review will focus on corn stover, sugarcane bagasse, ricestraw, and wheat straw due to their sustainability and availability throughout the year.Asia is the major producer of wheat straw and rice straw while corn stover and sugarcanebagasse are primarily produced in America (Sarkar et al. [246]).Conventionally, agricultural residues are used for livestock feed, bedding or housing,combustion to generate energy, or directly ploughed into the field as a fertilizer (Powlsonet al. [215], Rexen et al. [223], Staniforth et al. [276]). In developing countries where forestresources are insufficient, agricultural residues have also been used for the production ofpulp and paper (Jahan et al. [108], Kim and Holtzapple [125], Paavilainen [202]). Thereis a growing interest in expanding the use of agriculture residues for the production ofethanol and chemicals (Hsu et al. [99], Linde et al. [157], Lloyd and Wyman [170], Van Zylet al. [306]). However, agricultural residues are highly localized resources and have low15density, resulting in high collection and transportation costs.2.1.3.4 Agave and agave bagasseBelonging to the family Agavaceae, Agave is a genus of around 200-300 species. Mxicohas the greatest diversity of species, however Agave have been distributed world-widein arid and semi-arid regions as Agave water requirements are low (300-800 mm·year−1)(Davis et al. [47], Garcia-Moya et al. [70]). Mxico, Australia and South Africa are all suit-able for Agave production. Agave often grows on rocky soils of poor quality where fewother crops can easily grow. Precipitation is a key determinant of the productivity ofAgave. In areas with moderate annual rainfall (427 mm·year−1), A. lechuguilla had aproductivity of 3.8 ton·ha−1 · year−1 whereas in regions with 848 mm of annual rainfall,A. mapisaga produced 32 ton·ha−1 · year−1 (Davis et al. [47]). The estimated productivityof Agave word-wide is 10-34 megaton·ha−1 · year−1 (Somerville et al. [272]). A. tequilanais a perennial crop with at least 6-year life cycle before harvesting. The production oftequila in Mexico is estimated to use one million tons of agave plants yearly, thus, about400,000 tons of agave bagasse are generated annually (Corona-Gonza´lez et al. [42]).Disposal of agave bagasse is problematic as the fibers are slow to degrade; alter-natively agave bagasse is burned in the tequila distillery to generate heat and steam.In˜iguez-Covarrubias et al. demonstrated that agave bagasse can be used as raw mate-rial for livestock feed. Other common agave bagasse products include filters, geotextiles,packing, and sorbents. Corona-Gonza´lez et al. demonstrated that it is feasible producesuccinic acid from agave bagasse.162.2 Structure and Chemical Composition Analysis of RawBiomassDepending on the type of plant, the soil and the growing conditions, the compositionof lignocellulosic feedstocks varies widely (Xu and Huang [326]). Table 2.2, Table 2.3,and Table 2.4 summarizes the compositional ranges of hardwoods, softwoods, bamboo,agricultural residues and Agave.2.2.1 Ultrastructure of lignocellulosic biomassLignocellulose, the primary building block of plant cell walls, is a composite of cellulose,hemicellulose and lignin. Cellulose serves as the skeleton surrounded by other substancesfunctioning as matrix (hemicellulose) and encrusting (lignin) materials (Fengel and We-gener [62], Sjostrom [262]). The cell wall, as shown in Figure 2.1, is composed of severallayers: middle lamella (M), primary wall (P), outer layer of the secondary wall (S1), mid-dle layer of the secondary wall (S2), inner layer of the secondary wall (S3), and wartylayer (W) (Sjostrom [262]).Figure 2.1: Simplified structure of a woody cell, showing the middle lamella (ML),the primary wall (P), the outer (S1), middle (S2), and inner (S3) layers of thesecondary wall, and the warty layer (W) (Adapted from Sjostrom).17The structure and chemical composition of each layer differ (Figure 2.2). The pri-mary wall is a thin layer consisting of cellulose, hemicelluloses, pectin, and protein andis covered by lignin. In the outer portion of the primary wall, cellulose microfibrils areirregularly distributed to form a network, but in the interior they are oriented nearly per-pendicularly to the cell axis. The secondary wall is built up by three layers: thin outerand inner layer and a thick middle layer. These layers consist of lamellae composed ofalmost parallel microfibrils, and lignin and hemicelluloses filling the spaces between mi-crofibrils. The outer layer (S1) contains 3-4 lamellae and the inner layer (S3) consists ofseveral lamellae. The middle layer (S2) is the main portion of the cell wall. It might con-tain as few as 30-40 lamellae or more than 150 lamellae. The warty layer (W) is a thinamorphous membrane of unknown composition located in the inner surface of the cellwall in all conifers and some hardwoods (Fengel and Wegener [62], Sjostrom [262]).Figure 2.2: Distribution of hemicellulose, cellulose, and lignin through the cell wall(Adapted from Liu).182.2.2 Compositional analysis of biomassCellulose is a linear polymer of glucose units linked by β-1,4 glycosidic bonds (Smooket al. [271]). Due to intramolecular and intermolecular hydrogen bonding, cellulose chainsform crystalline structures interspersed with amorphous regions. The crystalline regionsresist hydrolysis (Smook et al. [271], Taarning et al. [282]). Hemicelluloses are branched,amorphous polysaccharides composed of short lateral chains consisting of several differ-ent monosaccharides and functional groups (Scheller and Ulvskov [250]). Lignin, which ishydrophobic in nature, is an amorphous polyphenol of the primary monolignols: coumaryl,coniferyl, and sinapyl alcohol (Saake and Lehnen [238]). Within the cell wall, lignin istightly bounded to cellulose and hemicellulose through hydrogen bonds and covalentbonds to form lignin-carbohydrate complexes (LCC).Biomass also contains inorganic ash components such as potassium, calcium, sodium,magnesium and silica (Fengel and Wegener [62], Sluiter et al. [264], Torelli and Cˇufar[292]), and includes both plant structural components and inorganic materials such as soilpicked up during harvesting and storage operations. In woody biomass, the ash contentis usually low (Table 2.2) while in non-woody species such as bamboo and agriculturalresidues (Table 2.3), the structural mineral content may account for as much as 16% (w/w)(Wyman [321]). For non-woody feedstocks, the major component of ash is silica (couldbe more than 50% of total ash).Biomass also contains some compounds known collectively as extractives, which aresoluble in water or organic solvent. In bamboo and agricultural residues, the extractivesare mainly composed of resins, fats, nonstructural sugars, nitrogenous material, chloro-phyll, and waxes (He and Yue [90]). Common extractives in wood include phenolics,terpenes, aliphatic acids, and alcohols (Fengel and Wegener [62]). Extractives often haveprotective biological and anti-microbial activities (Torssell [294]).192.2.2.1 WoodThough hardwoods and softwoods developed from the same ancestor, there are key dif-ferences in structure and composition. Softwood has a simpler structure consisting of90-95% tracheids. Tracheids are long, slender cells with flattened or tapered closed edges,which provide strength and conduct water and minerals within the tree. Hardwood hasstrength enhancing tissue containing libriform fibres and fibre tracheids. The dimensionsof the hardwood fibres are smaller than those of the softwood tracheids with thicker cellwalls and smaller lumina (Fengel and Wegener [62]). Vessels in hardwood might be sev-eral metres in length and are more effective for water transport than softwood tracheids(Sjostrom [262]).Within wood, cellulose accounts for approximately 45% (w/w), hemicellulose for 25-35% (w/w) and lignin for 21-25% (w/w) (Smook et al. [271]). Cellulose and lignin aremore resistant to hydrolysis than hemicellulose (Smook et al. [271], Taarning et al. [282]).Softwoods typically contain more lignin than hardwoods. The major components of hard-wood hemicellulose are arabinoxylan and acetyl groups. Softwood hemicellulose consistsprimarily of mannan, glucan, and galactan (Fengel and Wegener [62]).2.2.2.2 Bamboo and agricultural residuesBelonging to the grass family, bamboo and agricultural residues such as cereal straws,corn stover, and sugarcane bagasse are categorized as monocotyledons. Bamboo andcereal straws such as wheat straw and rice straw have a similar anatomy: a stem structurewith numerous vascular bundles scattered in a matrix of parenchyma storing cells, all ofwhich is surrounded by a strong, dense epidermis (Scurlock et al. [251]). Just inside theepidermis is a layer of hypodermis, mostly made up of sclerenchyma cells. The structureof bamboo and agricultural residues are similar to the fibre fraction of hardwoods. Toour knowledge, the structure and surface characterization of sugar cane bagasse has notbeen studied extensively, but some works have found that bagasse has a rigid, compact20morphology and fibres with an orderly distribution (Corrales et al. [43]).The hemicellulose of bamboo and agricultural residues is primarily glucuronoara-binoxylan which has a xylose backbone decorated with arabinose, glucuronic acid andacetyl side-groups (Pauly et al. [207], Scheller and Ulvskov [250]).2.2.2.3 Agave and AGBAgave is approximately 1.2-1.8 m tall with long leaves attached to a central pinecone.The leaves of the plant are removed in order to recover the pinecone for tequila produc-tion. Mature Agave leaves contain up to 42% carbohydrates and 12% lignin (Rijal et al.[226]). The pinecone is cooked to produce syrup for tequila production leaving a fibrousresidue known as agave bagasse (In˜iguez-Covarrubias et al. [103]). Agave bagasse (AGB)represents approximately 40% of the Agave plant (In˜iguez-Covarrubias et al. [103]). Thecellulose content of Agave and agave bagasse (73.6% and 43-45.6% respectively) is com-parable to that of sugarcane bagasse or corn cob (34.1-49% and 34.3-36.5% respectively).21Table 2.2: Composition of hardwoods and softwoods (wt%). *Others: containing mainly acetyl and uronic acid groupsof xylan, some pectins and other polysaccharides and inorganicsYear Author Feedstocks Carbo-hydratesCellulose Hemi-celluloseArabinose Galactose Glucose Mannose Xylose Lignin KlasonligninAcid-solubleligninExtractives Acetyl Ash Others*Wood2013Lehto andAle´nBirch (Betula pen-dula)64.1 0.4 1.3 39.7 1.6 21.1 22.1 17.3 4.8 3.1 10.72010Walton et al.Mixed southernhardwood0.49 1.28 46.58 2.19 17.4 26.8 2.64 2.85 0.34 2.32016Hundt et al.Birch (Betula pen-dula)40.6 21.6 22 3.1 0.7Beech 39 18 21.3 1.5 0.5Black locust 42.1 15.9 23.4 3.8 1.7Pine 42.9 15.6 24.3 27.9 2.6 0.42015Lehto andAle´nScots pine 60.4 1.4 1.8 39.3 13.8 4.1 27.7 27.2 0.5 3.7 8.22013Pu et al.Softwood chips 41 21 29 91992Mok andAntal JrEucolyptus gurn-rnifera38 16 37Papulus deltoides 39 21 26Luecnenn hybridKX-343 17 25Eucolyptus soligm 45 15 25Silver maple 40 23 22Sweet gum 40 23 192013Vena et al.E. grandis 15.8 52.7 21.1 4.2 1.22012Park andKimEucalyptus 41.8 18.7 30.1 9.4Larix leptolepis 43.4 24.4 28.9 3.3Pinus rigida 43.1 23.7 29 4.22010Jin et al.Mixed hardwood 66.4 47.8 2.2 16.3 22.7 4 222Table 2.3: Composition of agricultural residues (wt%). *Others: containing mainly acetyl and uronic acid groups ofxylan, some pectins and other polysaccharides and inorganicsYear Author Feedstocks Carbo-hydratesCellulose Hemi-celluloseArabinose Galactose Glucose Mannose Xylose Lignin KlasonligninAcid-solubleligninExtra-ctivesAcetyl Ash Others*Cereal straw2007Kabel et al.Wheat straw 58 31 2.5 25 1.7 202012Egu¨e´s et al.Corn stover 70.79 49.22 25.57 17.18 1.08 9.272006Jin and ChenRice straw 36.5 25.6 12.82006Karimi et al.Rice straw 39 37.5 27 12 11 20.62009Abedinifar et al.Rice straw 38 24 8 152012Jahan et al.Rice straw 38.2 38.2 23.5 22.1 4.1 14.62008Rodrı´guez et al.Rice straw 41.2 41.2 21.9 9.22008He et al.Rice straw 33.4 28.2 7.4 10.4 12.81997Peiji et al.Wheat straw 40.6 18 12.5 3.1 8.81997Peiji et al.Corn straw 42.6 21.3 8.2 5.2 10.12002Laser et al.Sugar cane bagasse 2 44 23 262009Peng et al.Sugar cane bagasse 43.6 33.5 18.1 2.32006Saha and CottaWheat straw 44.24 25.232010Yu et al.Wheat straw 3.6 1.3 36.7 12.5 9 21.51997Esteghlalian et al.Corn stover 2.8 1.3 36 19.7 17.8 1.9 7.2 13.22002Kim and LeeCorn stover 0.68 40.19 0.29 18.53 2.3 2.9 2.2 7.08 1.511996Ibrahim et al.Sugar cane bagasse 42.2 27.3 20.3 9.2 1.92000Pandey et al.Sugar cane bagasse 50 30 2.423Table 2.4: Composition of bamboo, agave and AGB (wt%). *Others: containing mainly acetyl and uronic acid groupsof xylan, some pectins and other polysaccharides and inorganicsYear Author Feedstocks Carbo-hydratesCellulose Hemi-celluloseArabinose Galactose Glucose Mannose Xylose Lignin KlasonligninAcid-solubleligninExtractives Acetyl Ash Others*Bamboo2010Yamashita et al.Bamboo 68.3 45.5 22.82010Leenakul and Tip-payawongBamboo 40.7 1.1 1.2 40.7 0.6 23.6 27.1 1.22011Ma et al.Bamboo 49.6 17.45 23.12 5.65 1.782015Mohan et al.Bamboo 38.3 25.512000Scurlock et al.Bamboo 40-48 1.0-2.0 0.5-1.2 40-48 0.4-0.6 25-30 11-27.01.5-3.0 0.2-22004Vu et al.Bamboo 68.6 25.8 0.8 2.22013Littlewood et al.Bamboo 64.2 1.8 3.6 38.4 20.8 13.5 3 0.92011Batalha et al.Bamboo 0.8 0.6 49.3 0.3 19.5 22.4 16.2 3 1.52011Garcı´a-Aparicioet al.Bamboo 3.1 41 16.7 30.7 20.8 9.9 7 2.42013Ma et al.Bamboo 50.8 21.8 23.7Agave/AGB2013Flores-Sahagunet al.A. tequilanaWebber var azul73.6 21.1 5.32016Davis et al.AGB from A.tequilana43 19 152016Rijal et al.A. tequilana leaf 38.85 18.31 14.32242.3 Response of Biomass to Hydrothermal TreatmentThe hemicellulose-lignin matrix in the biomass serves as a physical barrier and adheresto and tethers cellulose macrofibrils through hydrogen bonds and van der Waals inter-actions (Altaner and Jarvis [7], Cosgrove and Jarvis [44], Dammstro¨m et al. [46]). Thebarrier has high stability and hinders degradation of biomass components. For example,hemicellulose prevents enzyme access to the cellulose surface (Zhang et al. [334]). In addi-tion, enzymes preferentially attack the amorphous region of the cellulose structure. Thus,altering the biomass structure, decreasing cellulose crystallinity and increasing cell wallaccessibility to micro-organisms or chemicals is essential to the conversion of biomass tofuels and chemicals. Pretreatment methods can solubilize and remove hemicellulose anda fraction of lignin, and increase pore volume and surface area.Many pretreatment methods have been developed and can be classified by severalcriteria. Commonly, pH is applied as a standard to separate low, neutral and high pHpretreatment methods. This classification is shown in Figure 2.3. Pretreatment methodscan also be divided into physical, biological, chemical and hydrothermal (Xu and Huang[326]). Hydrothermal pretreatment uses liquid water as the media with/without additionof chemicals (acid or alkali) to extract hemicellulose or lignin from biomass and improvethe accessibility of residual cellulose to enzymes and chemicals during subsequent pro-cessing steps (Bobleter et al. [22]). The solid and liquid products can be recovered andfurther converted to valuable derivatives (Silveira et al. [257]).Treatment with water under high temperature and pressure without the addition ofany other chemicals is called autohydrolysis (Ruiz et al. [236]). Autohydrolysis is de-sirable as it does not require corrosive chemicals or expensive materials of constructionand avoids generation of neutralization salts. Moreover, due to the nature of lignocellu-losic biomass, natural acids (acetic or uronic acid) are released under high temperaturetreatment with water, resulting in an acidic liquid phase. During autohydrolysis, the dis-sociation of water at high temperature generates protons, which also reduces pH.25Figure 2.3: Categories of hydrothermal pretreatment methods.To increase treatment severity and promote the dissolution of biomass components,acid or alkali are added. Acids such as sulfuric acid, hydrochloric acid, acetic acid, formicacid, and phosphoric acid have been used for hydrolysis (Fengel and Wegener [62], Kanget al. [118], Kemppainen et al. [123], Larsson et al. [139], Song et al. [274], Tunc et al. [297]).Alkalis used for pretreatment include sodium hydroxide (NaOH), potassium hydroxide(KOH), calcium hydroxide (Ca(OH)2), green liquor (NaCO3 and Na2S) and ammonia (Jinet al. [111], Jun et al. [114], Walton et al. [317]).2.3.1 Reactions in acidic conditionAcidic pretreatment, including autohydrolysis and dilute acid hydrolysis, is normallyconducted at 120-280 oC (Borrega et al. [23], Rissanen et al. [227, 228], Sanchez and Car-dona [244], Yan et al. [328]). During autohydrolysis and dilute acid hydrolysis of lig-nocelluloses, low pH facilitates significant hemicellulose and partial cellulose removal.However, drawbacks of acid addition include the need for subsequent detoxification andneutralization as well as the need for high-cost corrosion resistant equipment (Xu andHuang [326]).262.3.1.1 Hydrolysis mechanism in acidic conditionThe fundamental hydrolysis mechanism is the breaking of glycosidic linkages betweenmonomers in the polymeric chains of hemicellulose and cellulose (Fengel and Wegener[62]). Hydrolysis of the glycosidic linkage is initiated by rapid protonation of the aglyconoxygen atom in Figure 2.4, followed by a breakdown of the conjugate acid to the cycliccarbonium ion. Subsequently a water molecule is added to form two monomers andrelease a proton (Sjostrom [262]). During acidic pretreatment, the biomass componentsundergo several parallel reactions such as deacetylation of hemicellulose, hydrolysis ofhemicellulose, chain cleavage of cellulose, degradation and solubilization of lignin, acidneutralization by ash, and removal of extractives.Figure 2.4: Hydrolysis mechanism in acidic condition. Methyl -D-glucopyranoside(1) to D-glucose (5) involving conjugate acid (2 and 4) and cyclic carboniumion (3) intermediates (Adapted from Sjostrom).272.3.1.2 HemicelluloseHemicellulose is readily solubilized from biomass during autohydrolysis. Mok and An-tal Jr studied six woody and four herbaceous biomass species in a percolating reactorwith water for 0-15 min at 200-230oC. A total of 40 to 60% of solids were dissolved; 100%hemicellulose, 4 to 22% cellulose and 35 to 60% lignin were solubilized. There was no cor-relation between feedstock composition and the fractional solubilization of cellulose andlignin suggesting that other factors such as structural differences influence the hydrol-ysis process. However, a strong negative correlation between the acid-insoluble ligninand hemicellulose solubilization was observed Mok and Antal Jr suggesting that lignin issolubilized as a LCC with hemicellulose. Increasing temperature and/or time increasesbiomass solubilization. For example, Pu et al. investigated autohydrolysis of mixed soft-wood chips and achieved solids recovery of 96% at 150oC for 1 hour, but solids recoverydecreased to 76% when temperature and time were increased to 180 oC and 2 hours. Au-tohydrolysis of bamboo with liquid hot water has been conducted over a wide range ofconditions: 140-220 oC for up to 240 min (Littlewood et al. [161], Ma et al. [178], Timunget al. [291], Xiao et al. [323]). Very little hemicellulose is removed from bamboo by auto-hydrolysis for short reaction times or low temperatures (Littlewood et al. [161], Xiao et al.[323]). By increasing temperature to 200oC for 120 min, Xiao et al. was able to remove99% of hemicellulose from bamboo; the need for high temperature and time is likely dueto bamboo’s dense epidermal layer.The addition of acid increases hydrolysis severity by providing more protons for thereaction; therefore, the reaction temperature can be lowered to achieve the desired hemi-cellulose and solids removal (Gu¨tsch et al. [84], Larsson et al. [139]). Pretreatment ofwoody biomass species including spruce, Eucalyptus, pine and mixed southern hard-wood has been investigated at 150-240oC for 1-118 minutes using sulfuric acid, acetic acid,oxalic acid, formic acid and maleic acid (Gu¨tsch et al. [84], Larsson et al. [139], Lim andLee [155], Tunc et al. [297]). Tunc et al. reported a linear relationship between dissolution28yield of hardwood and acetic acid concentration during treatment at 160oC for 90 min-utes with acetic and formic acid. Lim and Lee were able to remove 20 to 25% solids fromsoftwood by using acid at 160-180oC for 30-118 minutes; in contrast, Pu et al. achieved24% softwood solid dissolution only after autohydrolysis at 180oC for 120 minutes. Acidhydrolysis of bamboo has been carried out at temperatures similar to autohydrolysis,150-240oC, but reaction times are much shorter, normally less than 30 min (Leenakul andTippayawong [144], Li et al. [154]). Near complete hemicellulose removal from bamboowas achieved using 2% sulfuric acid at 180oC for 30 min (Li et al. [154]). Herna´ndez-Salaset al. Hrnandez-Salas et al. (2009) performed hydrochloric acid hydrolysis of sugarcanebagasse and agave residues at 121oC for 4 h; reducing sugar production from sugarcanewas approximately seven times greater than production from agave residues.Lim and Lee compared the effects of maleic, oxalic, and sulfuric acid on mixed soft-wood hydrolysis. Lim and Lee found sulfuric acid increased the rate of hydrolysis ofsoftwood relative to the other acids. They concluded pH and temperature had a greatereffect on biomass degradation than did acid type.The soluble products of acidic hydrolysis are complex and change as reaction severityincreases. Oligomers or oligosaccharides are one of the major hydrolysate components.These oligomers may be further hydrolyzed to monomers. As reaction severity increasesdue to increasing temperature and/or time or acid addition, the oligomers and monomerswill degrade to products such as furfural, HMF and other light organics (Pu et al. [216]).Hemicellulose also undergoes deacetylation during hydrolysis.The composition of soluble saccharides depends upon the hemicellulose compositionof the raw material. Hemicellulose in hardwoods, bamboo, and agricultural residuesis mainly arabinoxylan with a xylan backbone and arabinose branches. Low amountsof galactan and mannan may also be present depending on species. Thus soluble sac-charides in acidic hydrolysates are mainly xylose-derived with relatively low concentra-tions of arabinose, mannose, and galactose (Kaparaju et al. [119], Li et al. [154], Mosier29et al. [192], Saha and Cotta [240], Saha et al. [241], Xiao et al. [323], Yu et al. [331]). Dur-ing hydrolysis, arabinose branches are much easier to remove than other hemicellulosesugars, hence, arabinose concentration typically peaks before other sugars (Xiao et al.[323]). In contrast, softwood hemicellulose is primarily galactoglucomannan. The major-ity of soluble poly- and monosaccharides recovered from Norway spruce were mannose,glucose and galactose; mannose and glucose were present primarily as oligomers afterautohydrolysis at 150-180oC (Leppa¨nen et al. [148]). Galactose is a side chain on galac-toglucomannan thus up to 25% of extracted galactan was hydrolyzed into monomer form(Leppa¨nen et al. [148]). Arabinose is an easily hydrolysable furanosidic side chain, 50-75%of solubilized arabinose was in monomer form (Leppa¨nen et al. [148], Sjostrom [262]).Similarly, arabinose, glucose and xylose from mixed softwood were primarily detected asmonomers after hydrolysis at 170oC and 180oC (Pu et al. [216]).Typically soluble hemicellulose products are reported as monomers or total oligomersby conducting post-hydrolysis of the hydrolysate (Trajano et al. [296]). Temperature, time,and acid addition all affect the final ratio of monomer to oligomer. For example, Gu¨tschet al. concluded that the main difference between autohydrolysis and acid hydrolysiswas the ratio of xylose and xylooligomers in hydrolysate. However, some studies havemeasured the molar mass of oligomers produced during acidic hydrolysis (Borrega et al.[23], Kilpela¨inen et al. [124], Leppa¨nen et al. [148], Rissanen et al. [227, 228], Song et al.[273]). Borrega et al. found that oligomers recovered at temperatures above 200oC had alow average molar mass varying from 0.3-1.5 kDa. Leppa¨nen et al. concluded that 170-180oC is the optimal range to isolate high molar mass polysaccharides from spruce duringautohydrolysis (Leppa¨nen et al. [148]). Other authors (Rissanen et al. [227, 228], Songet al. [273]) have reported that temperatures of 160oC or lower favour higher molar massoligomers. The effect of pH on oligomer production from spruce wood and birch woodhydrolysis were studied by Song et al. and Kilpela¨inen et al. using phthalate buffers andacetic acid/sodium acetate buffer respectively. The temperature range was 160-180 oC30with pH levels from 3.8 to 4.6. The highest oligomer average molar mass, approximately14 kDa, was obtained at pH of 4.4, which was significantly higher than 7.5 kDa fromautohydrolysis (Song et al. [274]). Compared with autohydrolysis, at pH 4, delignificationdecreased and a lower total yield of sugar achieved (Song et al. [274]).Due to the dehydration reactions described below, total carbohydrates yield frombiomass has a maximum. A maximum yield of 10-13% soluble sugar from softwoodchips was achieved at 170 oC for 1.5-2 hours (Kang et al. [118], Pu et al. [216]). Completehemicellulose removal from Eucalyptus and maple was obtained at 200-230 oC for 0-15minutes, 90% of which was in a monomer form (Mok and Antal Jr [189]). For bamboohydrolysis, maximum reported yields of soluble hemicelluloses, including xylose andxylooligomers, are approximately 50-70% (Sella Kapu and Trajano [253]). Soluble hemi-cellulose sugar yields greater than 90% have been achieved from agricultural residues(Lloyd and Wyman [170], Pe´rez et al. [213], Thomsen et al. [287]). Pe´rez et al. studied theeffects of temperature (170-200oC) and time (0-40 min) on autohydrolysis extraction ofhemicelluloses from wheat straw; they recovered more than 90% of hemicelluloses dur-ing the pretreatment of wheat straw at 170oC for 40 min at solid loading of 10%. Solublexylose and xylooligomers yields over 90% have been achieved from hydrolysis of sugar-cane bagasse with the utilization of the flow through reactor (Allen et al. [6], Jacobsen andWyman [107]). The maximum reported xylose yield from corn stover was 93% after treat-ment with 0.92% H2SO4 at 160oC for 5-10 min (Torget et al. [293]). A xylose yield greaterthan 85% was achieved from rice straw after pretreatment with 1.6% H2SO4 at 121oC for30 min (Roberto et al. [230]).Pentoses and hexoses released during acidic pretreatment may undergo subsequentdehydration reactions to furfural and HMF, respectively (Rogalinski et al. [232]). Arabi-nose is particularly susceptible to dehydration relative to xylose and galactose (Xiao et al.[323]). Furfural and HMF may further react to generate formic acid and levulinic acid (Puet al. [216]). This trend was observed by Littlewood et al.; by increasing the temperature31of bamboo autohydrolysis from 170oC to 190oC, xylose yield in the liquid increased bymore than 60%. However, when treatment time was increased from 10 to 30 min at 190oC,the xylose content in the liquid phase decreased (Littlewood et al. [161]). Borrega et al.reported a 7-10% yield of furfural during autohydrolysis of birch wood between 200-240oC. Dilute acid addition to the hydrolysis of woody biomass also decreases monomeryield in the hydrolysate, while increasing degradation products concentration such asfurfural and HMF (Gu¨tsch et al. [84], Larsson et al. [139], Lim and Lee [155]). Lim and Leefound that the type of acid used has some influence on hydrolysis; the use of oxalic acidor maleic acid resulted in greater production of glucose and mannose and concurrently,lower production of HMF from softwood than sulfuric acid (Lim and Lee [155]).During autohydrolysis of lignocellulosic biomass regardless of the species, pH dropssignificantly mainly due to liberation of acetic and uronic acid from hemicelllulose, gen-eration of formic acid, and the demethylation of pectin to form pectic acid (Leppa¨nenet al. [148], Pu et al. [216], Rissanen et al. [227, 228]). Hardwood hemicellulose containsmore acetyl groups than softwood therefore hardwoods are better suited autohydrolysis.Borrega et al. investigated the hot water extraction of xylan-derivatives from birchwoodat 180 to 240 oC. Rapid deacetylation took place at the beginning of hydrolysis and a largeamount of acetyl groups were released to the liquid. Borrega et al. proposed that duringhardwood hydrolysis oligomers with intact acetyl groups are first solubilized. The solu-ble xylan oligomer is deacetylated and then hydrolyzed to ever smaller oligomers. Songet al. found that maintaining pH at approximately 4 caused deacetylation to decrease by40% relative to autohydrolysis.322.3.1.3 CelluloseThe 1,4-glycosidic bonds covalently linking the monomer units of cellulose are also cleavedduring pretreatment (Hallac and Ragauskas [85], Ma et al. [177]). In addition, acidic pre-treatment has been shown to affect the degree of polymerization and crystallinity of theresidual solid cellulose.Due to cellulose’s crystalline nature, it is more difficult to remove during acidic hy-drolysis (Fitzpatrick [65], Renders et al. [222]) and it is likely that the amorphous regionof cellulose is hydrolyzed first (Mok and Antal Jr [189]). Compared to hemicellulose,much harsher conditions, such as very low pH, high temperatures, or longer times, arerequired for cellulose hydrolysis (Fitzpatrick [65]). For example, 73% hemicellulose wasremoved but only 9% of cellulose was removed from dry poplar wood particles mixedwith 40 mL of MeOH, 0.2 g of Pd/C (5 wt %) and 5 g·L−1 H3PO4 after treatment at 200oCfor 3 hours (Renders et al. [222]). Similarly, Leppa¨nen et al. concluded that completehemicellulose extraction from spruce occurred at 220 oC but cellulose hydrolysis was notobserved until 240 oC. Several studies have investigated the degradation and removal ofcellulose from bamboo during pretreatment with water or 0.6-5% sulfuric acid (H2SO4) at120-200 oC and biomass consistency of 5-33% for 20-240 min (Leenakul and Tippayawong[144], Ma et al. [177], Sindhu et al. [259], Xiao et al. [324]). These treatments resulted in0.7-19% removal of initial cellulose. Liu et al. reported the addition of acids acceleratedthe removal of glucan from agricultural residues. Sasaki et al. conducted hydrolysis ofmicrocrystalline cellulose in subcritical and supercritical water at 25 MPa, 320-400 oC for0.05-10.0 s. At 320-350 oC, aqueous decomposition products of glucose were the mainproducts since the rate of cellulose hydrolysis was slower than the rate of glucose andcellobiose decomposition. However, at temperatures above 350 oC the cellulose hydroly-sis rate significantly increased relative to the glucose and cellobiose decomposition rates(Sasaki et al. [247] indicating a shift in reaction mechanism around 350 oC. At 400 oC,cellulose conversion reached approximately 100% after 0.05 s and the hydrolysis product33yield was 76.5%.Like soluble hemicellulose products, glucose generated from cellulose degrades toproducts such as HMF and levulinic and formic acid under high severity conditions(Fitzpatrick [65], Girisuta et al. [75], Sakaki et al. [242], Sasaki et al. [247]). The glucosedegradation pathway proposed by Kabyemela et al. in sub- and supercritical water forresidence times up to 2 s is illustrated in Figure 2.5. Glucose is isomerized to fructosewhich may further degrade to 5-HMF through the fructofuranosyl intermediate. Dehy-dration of glucose produces 1,6-anhydroglucose. Decomposition of 1,6-anhydroglucoseand erythrose generates mainly acids (Kabyemela et al. [116]).Figure 2.5: Generalized reaction pathway for glucose decomposition in subcrit-ical and supercritical water conditions (Reprinted with permission fromKabyemela et al.. Glucose and fructose decomposition in subcritical and su-percritical water: detailed reaction pathway, mechanisms, and kinetics. IndEng Chem Res 38: 28882895. Copyright 1999 American Chemical Society.During dilute acid pretreatment or autohydrolysis with high severity, the degree ofpolymerization of cellulose remaining in the solids decreases. Ma et al. observed thatthe degree of polymerization of cellulose in bamboo increased slightly after hydrolysisat temperatures below 150 oC (Ma et al. [177]); at temperatures above 170 oC, the degree34of polymerization initially decreased rapidly and then stabilized at a level-off degree ofpolymerization as treatment time increased (Ma et al. [177]). Ma et al. proposed that theinitial increase in degree of polymerization at lower treatment temperature (150 oC) wasdue to fast hydrolysis of short cellulose chains while at higher temperatures (>170 oC)the decrease of degree of polymerization was because of rapid hydrolysis of amorphouscellulose (Ma et al. [177], Stephens et al. [277]). Kumar et al. subjected corn stover tohydrolysis with dilute acid (H2SO4), sulfur dioxide (SO2), steam, and liquid hot water;they reported 3.1-12.1% of glucan was removed and that the degree of polymerizationwas reduced by 65-85%.Finally, hydrolysis affects the crystallinity of residual solid cellulose. Liu et al. foundthat the ratio of amorphous to crystalline components decreased after even mild acidicand alkali pretreatments. The increase in crystallinity index may therefore reflect removalof amorphous components such as hemicellulose and lignin from biomass and not anincrease in cellulose crystallinity.2.3.1.4 LigninCompared to hemicellulose and cellulose, lignin removal during acid hydrolysis in batchreactor is typically low regardless of biomass type (Mok and Antal Jr [189], Xiao et al.[324]). When Leppa¨nen et al. conducted autohydrolysis of spruce at 180 to 240 oC, 9%to 21% of lignin dissolved. The constant ratio of dissolved hemicellulose and lignin sug-gested the presence of LCC (Leppa¨nen et al. [148], Sjostrom [262]). Low pH facilitateslignin removal (Lim and Lee [155], Tunc et al. [297], Yan et al. [328]). Sindhu et al. treatedbamboo with the relatively high acid concentration of 5% (w/w) at 121 oC and 15 lb pres-sure to remove 36% of lignin. The use of sulfuric, oxalic or maleic acid did not changethe extent of lignin removal (Lim and Lee [155]). The removal of lignin is accompa-nied by the generation of aromatic monomers in the liquid hydrolysate, and the typeand amount of phenols varies with both the biomass treated and hydrolysis conditions35(Du et al. [50], Martin et al. [181]). For example, after hydrolysis of poplar wood with hotwater and dilute acid, vanillin and syringaldehyde were the dominant aromatics in thehydrolysate (Yan et al. [328]). The main lignin degradation product from poplar observedby Pecina et al. was coniferyl alcohol.Some cross-linking reactions, such as condensation of aromatic rings, may also occurat high temperature (Leppa¨nen et al. [148]) therefore the limited removal of lignin frombiomass during acid hydrolysis may be deceiving as it has been shown that the lignin thatremains in the solids following acidic pretreatment is modified in several ways (Effland[51]).2.3.1.5 AshAsh will neutralize some proportion of acid during autohydrolysis or dilute acid hydrol-ysis (Kang et al. [118]) thus the neutralization capacity of non-woody materials such asbamboo and agricultural residues is higher than that of woody materials (Esteghlalianet al. [57], Kim and Lee [126], Sella Kapu and Trajano [253]). Silica, the primary ash com-ponent in bamboo, does not react with common acids except for hydrofluoric acid. It isdifficult to predict the extent of neutralization by ash as it depends on ash content andcomposition as well as the solid to liquid ratio, temperature and chip size (Kang et al.[118], Rissanen et al. [227, 228]). Researchers have observed that pH is lower after hydrol-ysis of larger particles (Krogell et al. [133], Song et al. [274]). It is likely that the accessiblefraction of ash and thus neutralization capacity decreases as particle size increases. Rissa-nen et al.a and Rissanen et al.b concluded that overall biomass conversion contributes tovariation in pH and pH change is a consequence of extraction, not an influencing factoron extraction. However, Sella Kapu and Trajano recently found that proton concentrationduring autohydrolysis and acid hydrolysis was the result of competition between initialacid level, deacetylation, and ash neutralization.362.3.1.6 ExtractivesDuring acid pretreatment process, extractive components are effectively solubilized. Phe-nols are divided into three groups by their degree of methoxylation (hydroxyl, guaiacyl,syringyl) and their functionality (aldehydes, ketones, acids, other). Softwood materialsalmost exclusively produce guaiacyl phenols, while hardwoods and herbaceous materialsproduce hydroxyl, guaiacyl, and syringyl phenols consistent with biomass composition.The presence of hydroxyl phenol groups in the acid hydrolysates of willow and poplaris attributed to benzenediols (Jo¨nsson et al. [113]) and 4-hydroxybenzaldehyde and 4-hydroxybenzoic acid (Ando et al. [10]), respectively. These compounds are thought to beextractive components rather than lignin components (Jo¨nsson et al. [113]). The extrac-tives of non-woods such as bamboo and agricultural residues are also dissolved duringacid pretreatment. During acid pretreatment of wheat straw, fatty compounds and resinare solubilized (Talebnia et al. [283]).2.3.1.7 UltrastructureThe ultrastructure of biomass undergoes several changes during acidic pretreatment. Anincrease in fibre surface area following hydrolysis has been observed with sugar canebagasse (Corrales et al. [43]), bamboo (Sindhu et al. [259]), spruce (Rissanen et al. [227,228]) and wheat straw (Hsu et al. [99]). Pore volume increases have been observed forbamboo (Sindhu et al. [259]) and wheat straw (Hsu et al. [99]). Hsu et al. found thatpore volume of bamboo increased with increasing pretreatment temperature and acidconcentration. This is likely due to the removal of hemicellulose and acid soluble lignin(Hsu et al. [99], Zhu et al. [339]) as well as the partial removal of cellulose and lignin. Ithas also been shown that biomass particle sizes decrease during the acid pretreatmentand that, as the intensity of the treatment increases, the generated percentage of smallparticles and fines will increase (Chen et al. [31]).37Blumentritt et al. and Ma et al. both conducted topochemistry analysis of poplar sub-jected to autohydrolysis at 160-180 oC for 5-80 minutes. The degradation of cell wallvaried among the sublayers of cell wall and alterations depended on pretreatment time.Blumentritt et al. observed cell wall distortion and shifting of the middle lamella; thiswas partially attributed to lignin softening and removal. Ma et al. demonstrated byconfocal Raman microscopy that lignin concentration throughout the cell wall decreasesduring pretreatment; the largest concentration decrease occurred in the secondary cellwall. However, Ma et al. found that the loss of hemicellulose from the compound middlelamella was more than that from the secondary cell wall.2.3.1.8 Summary of acid pretreatment on biomass solidsThe effects of hydrothermal acidic pretreatment on biomass ultimately depend on theconcentration of the acidifying agent, temperature, and time of the reactions. In gen-eral, after acidic pretreatment, residual biomass solids rich in cellulose and lignin areproduced.Autohydrolysis and dilute acid hydrolysis break the glycosidic linkages between monomersin the hemicellulose and cellulose, thus solubilizing polysaccharides as oligomers andmonomers. Soluble oligomers and monomers reflect the composition of the raw biomass.The production of oligomers and monomers initially increases with increasing tempera-ture and residence time until production of furfural and HMF due to dehydration reac-tions become significant. Due to high crystallinity and degree of polymerization, low pH,high temperatures, or long residence times, are required for cellulose hydrolysis. Ligninis solubilized as a LCC , and removal during acid hydrolysis is typically low. The ultra-structure of the cell wall is disrupted.382.3.2 Reactions in alkaline conditionAlkali treatment of lignocellulose disrupts the cell wall by dissolving hemicelluloses,lignin, and silica, hydrolyzing uronic and acetic esters, swelling cellulose, and decreas-ing the crystallinity of cellulose (Jackson [104]). Compared with acid and autohydroly-sis, lower temperatures and pressures are effective for alkaline treatment (Ga´spa´r et al.[73], Xu and Huang [326]). Alkali pretreatment also cleaves the R-ether linkages betweenlignin and hemicelluloses and the ester bonds between lignin and/or hemicelluloses andhydroxycinnamic acids, such as p-coumaric and ferulic acids (Spencer and Akin [275]).In addition, cellulose swelling after dilute sodium hydroxide treatment also increases theinternal surface area, decreases the degree of polymerization, decreases the crystallinity,and disrupts the lignin structure (Fan et al. [58], Hsu et al. [99]). Calcium hydroxide re-moves the acetyl groups from hemicelluloses, reduces steric hindrance of enzymes andincreases cellulose digestibility (Mosier et al. [193]). However, alkali treatment generateslarge amounts of salts and precipitation of salts such as calcium oxalate on processingequipment may cause downstream processing problems (Xu and Huang [326]).2.3.2.1 Hydrolysis mechanism in alkaline conditionHydrolysis under alkaline conditions cleaves lignin bonds and glycosidic hemicellulosebonds and disrupts ester bonds crosslinking lignin and hemicellulose. The most im-portant alkali-catalyzed reactions include polysaccharides dissolution, deacetylation ofhemicelluloses, and peeling reactions of carbohydrates (Jin et al. [111], Lehto and Ale´n[145]). The alkaline peeling reaction removes terminal anhydro-sugar units to generatenew reducing end groups until a competitive stopping reaction begins and forms a stablecarboxylic acid end group. Simultaneously, dissolution and/or degradation of lignin, re-moval of extractives, and saponification of esters (fats and waxes) occurs (Lehto and Ale´n[145]). Alkaline pretreatment also removes acetyl and uronic substitutions on hemicellu-lose by alkaline saponification (Zhang and Lynd [337]).39The peeling reaction (Figure 2.6) starts with the isomerization of the terminal carbonylgroup to a ketose, which is unstable in alkali and removed by the cleavage of the gly-cosidic bond. The eliminated end group is tautomerized to a 4-deoxy-2,3-glycodiulose,then undergoing benzylic acid rearrangement to form epimeric isosaccharinic acids. Thestopping reaction (Figure 2.7) undergoes a direct β-hydroxy elimination from the C-3 po-sition, then the end group goes through a benzilic acid rearrangement to an alkali-stablemetasaccharinic acid end group (Sjostrom [262]).Figure 2.6: Peeling reaction of 1,4-β-D-glucan (cellulose). R=glucan (cellulose) chain.Reaction steps: isomerization (1→2), enediol formation (2→3), β-alkoxy elim-ination (3→4), tautomerization (4→5), and benzilic acid rearrangement (5→6)to epimeric 3-deoxy-2-C-hydroxymethylpentonic acids (glucoisosaccharinicacid) (6) (Adapted from Sjostrom)40Figure 2.7: Stopping reaction. Reaction steps: 1,2-Enediol formation (1→2), β-hydroxy elimination (2→3), tautomerization (3→4), and benzilic acid rear-rangement (4→5) to epimeric 3-deoxyhexonic acid end groups (glucometasac-charinic acid) (5) (Adapted from Sjostrom)2.3.2.2 HemicelluloseAlkaline pretreatment solubilizes hemicellulose, part of the cellulose, and lignin. Increas-ing severity by increasing temperature, time, or alkali charge facilitates solids dissolu-tion from woody biomass (Vena et al. [308], Yoon and van Heiningen [330]). Solublehemicellulose-derived products reflect the composition of the raw biomass; there are keydifferences between the alkali hydrolysis of different types of hemicelluose. As underacidic conditions, high severity alkali pretreatment produces degradation products suchas furans and carboxylic acids. Finally, the release of acetyl groups from hemicelluloseaffects the final system pH.A wider range of temperatures, 20-190 oC, and times, from seconds to days, havebeen used for alkali pretreatment of biomass compared to acidic pretreatments (Egu¨e´s41et al. [52], McIntosh and Vancov [182], Mosier et al. [192, 193], Sun et al. [278], Yoon andvan Heiningen [330]). Higher temperature treatments are applied to woody biomass andincreasing temperature increases solid dissolution. For example, Yoon and van Heiningenreported that solid dissolution from Loblolly pine chips increased from 4.2% at 170 oC for15 min to 15.8% at 190 oC after 90 min green liquor hydrolysis. Vena et al. investigated thealkaline pretreatment of hardwoods prior to pulping. Glucan and lignin removal variedfrom 4.6-7.2% and 6.2-24.2%, respectively. The maximum soluble xylan recovery yield,16% of oven-dry xylan in untreated eucalyptus, was obtained at 90 oC for 240 minuteswith 2M NaOH concentration. Yuan et al. subjected bamboo to alkaline pretreatment with6-18% NaOH at 80-100 oC for 1-5 h. These treatments removed up to 50% of hemicellulosefrom bamboo. Chen et al. conducted alkaline pretreatment of corn stover with 4-10% ofNaOH at 74-130 oC for 48-120 min resulting in up to 23.5% of xylan removal; increasingNaOH charge increased xylan removal. Non-woody biomass can also be treated at lowertemperatures; Sun et al. found that 80% of hemicellulose could be released from wheatstraw using 1.5% NaOH for 144 h at 20 oC. Similarly, Zhang and Cai treated rice strawwith 2% NaOH at 85 oC for 1 h and removed 61% of the hemicellulose. In general, alkalinepretreatment is more effective for hemicellulose removal from low-lignin biomass such ashardwood, herbaceous crops, and agricultural residues than from softwoods with highlignin content (Bjerre et al. [20]).Under alkaline conditions, solubilized xylan remains as oligomers in solution butmannan is observed in monomer form; this may be a result of hemicellulose side groups(Jin et al. [111], Van Heiningen [305], Yoon and van Heiningen [330]). Xylan has a 4-O-methylglucoronic acid group side chain at the C-2 position which slows the peelingreaction and favors the stopping reaction (Jin et al. [111], Sjostrom [262]). When Jin et al.studied green liquor hydrolysis of mixed hardwood, they reported an initial fast xylanand mannan removal. This rapid removal was followed by stabilization of xylan andmannan content even when alkali charge was increased from 8% to 20% (Jin et al. [111]).42As discussed previously, carbohydrates are degraded by peeling reactions during alkalitreatments thus accounting for the initial rapid loss of both xylan and mannan. As thestopping reaction begins, carbohydrates become stable.The monomer concentration in alkali hydrolysate is relatively low (Lehto and Ale´n[145], Yoon and van Heiningen [330]) and is favored by low alkaline charge. At 130-150oC, particularly with low alkali charge, high molar mass carbohydrates were solubi-lized from Scots pine(Lehto and Ale´n [146]). Yoon and van Heiningen compared greenliquor hydrolysis and autohydrolysis of Loblolly pine chips. They found that the xyloseand mannose yields were both lower with alkali addition; the mannose yield after greenliquor hydrolysis was approximately 92.7% lower than the autohydrolysis yield. Waltonet al. compared hydrolysis of mixed southern hardwood chips with hot water and alka-line addition of 2-8% at 160 oC for 1 to 2 hours. Xylose was the dominant monosaccharidepresent in the hydrolysates from water-only and alkali conditions; xylose concentrationincreased with increasing severity for autohydrolysis; this correlation was not as strongcorrelation for alkaline hydrolysis (Walton et al. [317]). The arabinose concentration waslow in the liquor produced from autohydrolysis; arabinose concentration was even lowerin the alkaline hydrolysates most likely due to degradation by peeling reactions (Wal-ton et al. [317]). Finally, Walton et al. observed that glucose concentration in the alkalinehydrolysate was lower than the water-only hydrolysate. This is most likely due to thedegradation of glucan by alkaline peeling reactions and increased hydrolysis of gluco-mannan and glucan at low pH.As under acidic conditions, high severity treatments result in the degradation of pen-toses into furfural, hexoses and HMF, and eventually to formic and levulinic acids. Birchand Scots pine were hydrolyzed with alkaline aqueous solution of 1-8% NaOH for 30-120minutes at 130-150 oC (Lehto and Ale´n [145, 146]). Total solid dissolution was 2.1-16.5% ofdry birch and 2-13.6% of dry Scots pine. Maximum furan generation was observed dur-ing pretreatment at 150 oC with 1% NaOH. No quantitative amount of furfural and HMF43was detected in the hydrolysates produced at 130 oC (Lehto and Ale´n [145, 146], Lehtoet al. [147]). During green liquor pretreatment of mixed southern hardwood chips, lacticacid was formed by the alkaline peeling reaction and formic acid was formed by waterextraction and the cleavage of acetyl groups (Walton et al. [317]).The final pH of the hydrolysate obtained from alkaline pretreatment of lignocellulosicbiomass is determined by the competition between alkali charge and acidic groups, suchas acetyl and uronic acid groups. If the pH is less than 4.8, acetic acid is generated fromthe release of acetyl groups; if the final pH is above 4.8 then the acetyl groups are releasedas acetate by the reaction with alkali reagents (Walton et al. [317]). Increasing hydrolysisseverity by increasing temperature, prolonging reaction time or increasing alkalinity in-creases the release of acetyl groups (Lehto and Ale´n [145, 146], Walton et al. [317]). Forexample, when Walton et al. investigated hydrolysis of mixed southern hardwood chipsby autohydrolysis and green liquor, they found that the final pH of autohydrolysate wasapproximately 3.5, addition of 2% green liquor resulted in a pH near 5.7, and addition of6% green liquor resulted in a pH of approximately 7. A similar study of 2% green liquorextraction of hardwood showed a rapid drop in final pH from alkaline to neutral duringthe early stages of hydrolysis before final pH leveled off at approximately pH 4.5 (Yoonand van Heiningen [330]). Due to higher acetyl group content in hardwoods, the pH ofsoftwood hydrolysates tends to be more alkaline (Lehto and Ale´n [145, 146]).2.3.2.3 CelluloseDuring alkaline pretreatment, glucose units are liberated by alkaline hydrolysis or peel-ing reaction from cellulose chains. The released glucose undergoes degradation and re-arrangement, known as Lobry de Bruyn-Alberda van Ekenstein rearrangement, to formlactic acid (Richards and Sephton [224]). Celluloses crystallinity and degree of polymer-ization make it more resistant to alkaline media thus alkaline hydrolysis of cellulose typi-cally starts at 140 oC while the peeling reaction of cellulose and hemicellulose can occur at44approximately 80 oC (Sixta [260]). Renders et al. found that up to 18% of cellulose but upto 39% hemicellulose was removed from poplar by alkaline treatment, probably due toamorphization and/or biomass swelling creating a more open structure within the woodmatrix. Sun et al. observed that 21-41% of wheat straw was solubilized during alkalinepretreatment with 1.5% NaOH at 20 oC for 0.5-144 hr while the yield of cellulose in theresidual solids remained nearly constant.A decrease in cellulose crystallinity index after alkaline pretreatment has been ob-served in many biomass species. Renders et al. found that alkaline treatment caused thecrystallinity index of poplar to decrease with increasing basicity. Mirahmadi et al. studiedNaOH pretreatment of spruce and birch at -15 to 100 oC for 2 hours and found that allpretreatments decreased the crystallinity of cellulose. Oh et al. reported that the biggestreduction in crystallinity of cellulose powder occurred when treating at 60 oC with 9-12wt% NaOH.2.3.2.4 LigninThe main purpose of alkaline pretreatment is to remove lignin and thus increase the ac-cessibility of residual carbohydrates to enzymes and chemicals. Alkaline pretreatmentdisrupts the lignin structure and breaks the bonds between lignin and carbohydrates.Softwoods and hardwoods have been pretreated with various alkali (sodium carbon-ate, sodium sulfide, sodium hydroxide, potassium hydroxide and aqueous ammonia) at40-190 oC for 15 minutes to 24 hours (Jin et al. [111], Lehto and Ale´n [145], Lehto et al.[147], Mirahmadi et al. [186], Park and Kim [205], Vena et al. [308]); total lignin removalranged from 9.6-35%. Jin et al. green liquor hydrolysis of mixed hardwood chips resultedin 35% lignin removal. Park and Kim removed 16.8%, 15.1% and 9.6% from Eucalyptusresidue, Larix leptolepis, and Pinus rigida, respectively, using soaking and percolationpretreatments at 60oC for 24 hr with alkaline solutions. Several researchers (Chang et al.[28], Kim and Holtzapple [125], Yuan et al. [332], Zhang and Cai [335]) treated bamboo45and agricultural residues with 0.5-18% NaOH or 0-30% Ca(OH)2 for up to 4 weeks. Thesetreatments resulted in 3-80% removal of lignin. Chang et al. subjected wheat straw tolime pretreatment at 50-135 oC for 1-24 h. They found that for short pretreatment times,higher temperatures (85-135 oC) were required to remove lignin (about 14%) and reachhigh sugar yields, whereas for long treatment times (e.g. 24 h), lower temperatures (50-65oC) were more effective. Alkaline pretreatment of chopped rice straw with 2% NaOHand 20% solid loading at 85oC for 1 h decreased the lignin content by 36% ([335]). Duringlime pretreatment of corn stover with aeration, up to 87.5% of lignin could be removedat 55oC after four weeks (Kim and Holtzapple [125]). McIntosh and Vancov were able toremove 72% lignin from wheat straw with 2% NaOH at 121 oC for 90 min, while only 33%of lignin was removed from the same wheat straw pretreated with 0.75% NaOH at 121oC for 30 min. Vela´zquez-Valadez et al. reported 83% delignification of Agave tequilanabagasse following a 2-stage alkaline oxidation: stage 1- 6% NaOH at 120 oC and 2 atm for1 h, stage 2- 6 % H2O2 at 30 oC for 24 h.High alkaline charges favor removal of high molar mass lignin from both hardwoodand softwood while higher temperatures and longer residence times lower the weightaverage molar mass of dissolved lignin. Lehto et al. investigated birch and Scot pinehydrolysis with up to 8% NaOH charge at 130 oC and 150 oC for up to 2 hours. Scotpine produced slightly higher concentrations of lignin (1.5-9.6 g·L−1) than birch (1.2-6.9g·L−1) (Lehto et al. [147]). The weight-average molar mass of pine lignin ranged from2260 Da to 7050 Da and the molecular weight of birch lignin varied from 2200 Da to 5550Da. The polydispersity of dissolved lignin from pine was higher than birch, indicatingbirch lignin had a more uniform molar mass distribution. Increasing the severity of thereaction reduced the polydispersity of both feedstocks and produced more uniform-sizedlignin.462.3.2.5 AshSilica, the main ash component in bamboo and agricultural residues, is dissolved intosoluble silicates during alkaline pretreatment (Pekarovic et al. [210], Yuan et al. [332]).The dissolved silicates in the alkaline hydrolysate can be recovered by lowering the pHof the liquor. Other metals such as calcium, potassium, and sodium in the ash will alsobe transformed to their alkali forms.2.3.2.6 ExtractivesAs discussed in Section 2.3.1.6, a large proportion of extractives are removed after pre-treatment. Under high alkaline charges (6 and 8% NaOH), as much as 78% of extractiveswere removed from birch and Scots pine (Lehto and Ale´n [145, 146]) but high alkali chargeprevented further extractives removal and extractives content stabilized at 0.8% to 1.4%of wood. After NaOH pretreatment, cold- and hot-water extractives of rice straw signif-icantly increased likely due to the degradation of hemicellulose and partial degradationof cellulose and lignin to smaller, more soluble molecules (He and Yue [90]). Benzene-ethanol extractives were removed from rice straw during pretreatment (He and Yue [90]).2.3.2.7 UltrastructureWith removal of lignin and hemicellulose from biomass, the vascular bundles of cornstover, rice straw and wheat straw exhibit a rougher and generally more textured cell wallsurface while crack formation is observed in wood cell walls (Ji et al. [109], Selig et al.[252]). Renders et al. observed damage to the microstructure and micromorphology ofpoplar using scanning electron microscopy. Separated and exposed micro-fibrils increaseexternal surface area and the porosity of lignocellulosic materials with the removal oflignin-carbohydrate crosslinks (Li et al. [151], Tarkow and Feist [285]).Li et al. used confocal Raman microscopy to study Eucalyptus after hydrolysis with70% ethanol with 0.4-5% NaOH for 2 h at 80 oC. Li et al. found that hemicellulose wasprimarily dissolved from the secondary cell wall. A similar study of alkaline pretreatment47of poplar wood was conducted at 121 oC with 2% NaOH charge (Ji et al. [109]). Resultsshowed that lignin was mainly removed from the middle layer of the secondary walland that pretreatment enlarged the microfibril angle, particularly in the outer layer of thesecondary wall layer. Scanning electron microscopy analysis revealed formation of cracksin cell wall surfaces, which could favor enzyme accessibility to cellulose.2.3.2.8 Summary of alkaline pretreatment on biomass solidsCompared with acid and autohydrolysis, alkaline treatment is more effective at break-ing ester bonds between lignin, hemicellulose and cellulose, and limiting fragmentationof hemicellulosic polymers. Similar to autohydrolysis and acid hydrolysis, increasingseverity by increasing temperature, time, or alkali charge facilitates solids dissolutionfrom lignocellulosic biomass, and promotes the generation of carboxylic acids. Alkalinepretreatment can be performed at low temperatures but this requires a relatively longtime and high concentration of base. Under alkaline conditions, softwood glucomannanis rapidly degraded by the peeling reaction but xylan in hardwood, bamboo and cerealstraws is solubilized in oligomer form. The monomer concentration in alkali hydrolysateis relatively low probably due to degradation. Glucose units are liberated from cellulosechains by alkaline hydrolysis or the peeling reaction. Celluloses crystallinity makes itmore resistant to alkaline media; crystallinity does decrease with increasing basicity. Themain purpose of alkaline pretreatment is to remove lignin. High alkaline charges favoredthe removal of high molar mass lignin.A summary of the structural change of biomass with different pretreatments is shownin Table 2.5. A summary of hydrolysis conditions (acidic and alkaline) and yield/conver-sion for various biomass is listed in Table 2.6 and Table 2.7.48Table 2.5: Effect of different chemical pretreatment technologies on the structure of lignocellulose. H: high effect, M:moderate effect, L: low effect.PretreatmentmethodIncrease ofaccessiblesurface areaCellulose de-crystallizationHemicellulosesolubiliza-tionLigninremovalGenerationof inhibitorcom-poundsLigninstructurealterationDAPa H L H L H HLHWb H L M L L MNaOH/ H L M M L HCa(OH)2Note: a: dilute acid pretreatment (DAP); b: liquid hot water (LHW). This table is regeneratedaccording to Alvira et al., Brodeur et al., Xu and Huang.49Table 2.6: Summary of pretreatment conditions and resulting yields/conversionsYear Author Biomass Particlesize(mm)Solid toliquidratio(wt/wt)Temperature(oC)Time(min)Charge Catalyst Reactor Flowrate(mL/min)Yield/conversion2011Borregaet al.Silver birch 0.6-0.25 0.025(g/mL)180-240 190 Liquid hot water Batch reactor Maximum oligomer yield 15%2014Rissanenet al.Spruce sapwood 1.25-2and10*10180 120-170 250 Liquid hot water Cascade reactor 2500 Maximum conversion of 80% at180oC for 60min1992Mok andAntal JrSix woody and fourherbaceous biomassspecies1 0.238(g/mL)200-230 10 to 15 Liquid hot water A tubular percolating reactor Dissolution of 100% hemicellu-lose, 4 to 22% of cellulose and 35to 60% of lignin2013Pu et al.Softwood chips 30*30*2.5 0.25 150, 160,170, 18060, 120 Liquid hot water A lab scale batch digester Maximum sugar yield 13% ofoven-dried wood at 170 C for 2 h2011Leppa¨nenet al.Norway spruce < 2 0.07(g/mL)120-240 10 to 50 Liquid hot water A laboratory-made extractionreactor1 4-22% hemicellulose solubiliza-tion to monomer 15% of lignin re-moval Maximum oligomer yield7% of wood2012Kang et al.Southern pine 6 to 8 0.17 170 0 to 90 Liquid hot water Cylindrical stainless steelbomb container10% of dry wood or 48.5% of ini-tial hemicellulose conversion tooligomers and monomers2015Lehto et al.Scots pine and birch 7*13*7 0.2(g/mL)130 and15030,60,90,1201, 2, 3,4, 6, 8%on wooddry mat-terNaOH A laboratory-scale oil bathdigester (CRS CAS 420)equipped with 2L rotatingstainless-steel autoclaves0.84.4% and 0.63.4% lignin extrac-tion of pine and birch2015Lehto andAle´nScots pine < 1.0 0.2(g/mL)130 and15030,60,90,1201,2,3,4,6,8%NaOH A laboratory-scale oil bathdigester (CRS CAS 420)equipped with 2L rotatingstainless-steel autoclaves2.013.6% wood dissolution2013Vena et al.E. grandis chips 0.42 0.25(g/mL)40-90 120-240 12 M NaOH Schott bottles 12.4% of xylan extraction2012Park andKimEucalyptus residue,Larix leptolepis, Pinusrigida, rice straw andbarley straw0.595and0.2970.1 60 1440 1M NaOH, KOH andNa2CO3Percolation apparatus Delignification of eucalyptusresidue, P. rigida and L. leptolepisof 16.8%, 15.1% and 9.6%2010Yoonand vanHeiningenLoblolly pine chips 0.074 0.22 170, 180,19015, 45,902, 4, 6% Green liquor(Na2CO3 and Na2S)Eight rocking bomb digestersin an oil bathMaximum yields of arabinoglu-curoxylan and galactoglucoman-nan about 2 and 1%,2010Jin et al.Mixed hardwood chipmostly oak and sweetgum< 6 0.25 80 30 4% to20%(w/w)Green liquor(Na2CO3 and Na2S)A lab-scale M/K pulping di-gesterDelignification 35% of the ligninnearly 100% cellulose and 75% xy-lan preserved2010Waltonet al.Mixed southern hard-wood chipsScreen atthe mill0.25 160 60-120 0, 2, 4,6%Kraft green liquor andsodium carbonateStainless steel digester Highest sugar concentation of30g/L50Table 2.7: A summary of pretreatment conditions and yields for bamboo and agriculture residuesYear Author Biomass Particlesize(mm)Solid toliquidratio(wt/wt)Temperature(oC)Time(min)Charge Catalyst Reactor Flowrate(mL/min)Yield/conversion2010Leenakul andTippayawongBamboo < 0.3 0.1 120 and14030,60,90 0.6,0.9,1.2% Sulfuric acid Autoclave batch 15% xylan conversion2010Yamashita et al.Bamboo 50-80 ∗30-500.01 121 60 0.5-10%wt.NaOH Autoclave batch Maximum glucose and reducingsugar production with 1%(v/v)hydrogen peroxide and 1 wt.% ofsodium hydroxide2014Sindhu et al.Bamboo < 1 0.01-0.3 121 30-90 1-5%w/wSulfuric acid Batch Reducing sugar 31.9%1996Abbi et al.Rice straw 5 to 6 0.1 100 60 4.40% Sulfuric acid Boiling water bath Total sugars 20 g/L2008Thomsen et al.Wheat straw 50 0.2 80-195 3-20 Water/steam Pilot-scale three-step heating reactor300-1200(kg/h)33-83% hemicellulose recovery al-most 100% cellulose recovery2011McIntosh andVancovWheat straw < 1.5 0.1 60-121 30-90 0.75, 1.0,2.0%NaOH Water bath/auto-clave33% of hemicellulose solubiliza-tion1995Sun et al.Wheat straw < 0.25 0.025 20 30-8640 1.50% NaOH Batch 80% of hemicellulose removed for144h2013Chen et al.Corn stover 6 0.11 74-130 48-120 4-10%w/wNaOH Stainless batch reac-tor70% glucan conversion2005Mosier et al.Corn stover 6.4 0.16 170-200 5 to 20 Liquid hot water Stainless steel, plug-flow reactor40% of biomass solubilization2002Laser et al.Sugar cane bagasse < 1.41 0.01-0.08 170-230 1-46 Liquid hot water A custom-built 25 Lbatch reactorMore than 80% xylan recovery2008Cheng et al.Sugar cane bagasse 0.45-0.9 0.1 121 120 1.25%w/wSulfuric acid Autoclave batch 59.1 g/L total sugars in the hy-drolsate2009Peng et al.Sugar cane bagasse 0.04 50 180 3% NaOH Batch 74.9% hemicellulose solubiliza-tion512.4 ConclusionThe intrinsic structure of biomass determines its response to pretreatment. Cellulose crys-tallinity, accessible surface area, protection by lignin, and sheathing by hemicellulose allcontribute to the resistance of cellulose in biomass to hydrolysis. The conditions em-ployed during pretreatment will affect various substrate characteristics, which in turn,govern the susceptibility of the substrate to enzymatic hydrolysis and the subsequentfermentation of the released sugars.Acid treatment severity is typically higher than alkaline, can almost completely dis-solve hemicellulose but produces more degradation byproducts. The acidic liquid streamor hydrolysate generated during pretreatment is mainly composed of hemicellulose-derivedmonomers and oligomers, and associated degradation products such as furfural andHMF. Cellulose degradation products such as levulinic and formic acid and lignin-derivedproducts such as vanillin and syringaldehyde may also be present. Crystallinity of cellu-lose increases and a small amount of lignin is removed from the solids. The pore volumeand surface area of the solids are significantly increased.Alkaline pretreatment removes lignin and hemicellulose from biomass matrix whilepreserving cellulose. Alkaline treatment decreases cellulose crystallinity and swells thefibre structure thus improving porosity and surface area of feedstock.Although biomass species display incredible variety in composition and structure, theresponse of all species follow broadly similar trends to acidic and alkali pretreatment.However, as this review demonstrates the precise fate of hemicellulose, cellulose, lignin,ash, and extractives can vary substantially between species. Without a fundamental un-derstanding of biomass response to pretreatment, extensive experimentation is requiredto develop biorefinery technologies. Ultimately, the success of the biorefinery concept willrest upon a detailed understanding of biomass response to pretreatment and harnessingthat knowledge to efficiently utilize every component of the raw biomass.52Chapter 3Softwood Hemicellulose Hydrolysis byHot Water: Characterization andOperation Guideline3.1 Introduction3.1.1 MotivationHydrolysis is the hydrothermal treatment of biomass to break a carbohydrate into monosac-charides by cleavage of glycosidic bonds (Fengel and Wegener [62]). Hemicellulose is eas-ily hydrolyzed to soluble oligomers and monomers; a portion of lignin is also removedduring hydrolysis (Garrote et al. [71], Jacobsen and Wyman [106], Leppa¨nen et al. [148], Liet al. [150], Mok and Antal Jr [189]). Optimization of oligomer generation with minimalsugar degradation is essential to the production of pulp strength additives.In this chapter, the effect of pretreatment variables, such as temperature, time, solidloading, and particle size, are determined. Compositional analysis of hydrolysate andhydrolyzed solids will be discussed. The response of two pulp mill residues to the samehydrolysis condition are also reported.533.1.2 Softwood hemicellulose3.1.2.1 Composition and structureSoftwood hemicellulose contains arabino-4-O-methylglucurono-xylans (AGX) (up to 10%of oven-dried (O.D.) softwood) and GGM (14-20% of o.d. softwood) (Fengel and Wegener[62], Hon and Shiraishi [97], Kim et al. [127], Sjostrom [262]). Hardwood mainly containsO-acetyl-4-O-methylglucuronoxylan (20-35% of o.d. hardwood) and trace amounts ofglucomannan (2-4% of o.d. hardwood) (Hon and Shiraishi [97], Timell [289]).GGM is composed of (1-4)-β-D-mannopyranosyl units alternating with some (1-4)-β-D-glucopyranosyl units, and side groups of (1-6)-α-D-galactopyranosyl units (Fengel andWegener [62], Lindstro¨m et al. [160]) (Figure 3.1). The ratio of mannose to glucose units inGGM has been reported to vary from 1:1 to 3:1 depending on wood species, and branchesare distributed randomly along the backbone (Gı´rio et al. [74], Gullichsen and Fogelholm[83], Mimms et al. [185], Willfo¨r et al. [319]). The proportion of galactose units depends onisolation method. Water-soluble GGM has a Man: Glu: Gal ratio of 3:1:1 while alkaline-soluble GGM has a ratio of 3:1:0.2 (Fengel and Wegener [62]).Figure 3.1: Chemical structure of GGM from softwood (Fengel and Wegener [62])Softwood xylans are AGX, which has a homopolymer main chain of xylose unitslinked by β-(1-4)-glycosidic bonds (Fengel and Wegener [62]), substituted with α-(1-2)-linked glucuronosyl and 4-O-methyl glucuronosyl residues (Scheller and Ulvskov [250]),54and side groups of arabinofuranose units linked by α-(1-3)-glycosidic bonds (Figure 3.2).The ratio of Xyl:Me-GluU:Ara was reported as 8:1.6:1 (Fengel and Wegener [62]).Figure 3.2: Chemical structure of arabino-4-O-methylglucuronoxylan from softwood(Fengel and Wegener [62])3.1.2.2 Acidic groups of hemicelluloseThe main acidic groups of hemicellulose are O-acetyl groups and 4-O-methylglucuronicacid groups.The content of O-acetyl groups in hardwood (3 to 5wt% of o.d. hardwood) with O-acetyl-4-O-methylglucuronoxylan is higher than softwood (1wt% of o.d. softwood) withGGM (Hon and Shiraishi [97], Katz [121]). In hardwood O-acetyl-4-O-methylglucuronoxylan,70% of xylose units are substituted by O-acetyl groups at C2 and C3 positions (Bouveng[25], Timell [289]). For softwood GGM, mannose and glucose distribute randomly on themain chain (Fengel and Wegener [62], Timell [288], Timell et al. [290]). Approximately 33to 65% of mannose in the glucomannan backbone are substituted with O-acetyl groups(Fengel and Wegener [62], Lindberg [156], Lundqvist et al. [172], Timell [288], Willfo¨r et al.[319]). A equal distribution of O-acetyl groups were reported on the C2 and C3 of man-nose (Fengel and Wegener [62], Lindberg [156], Lundqvist et al. [172]).A regular distribution pattern of one 4-O-methylglucuronic acid residue for every 7 to8 xylose units was identified in softwood xylans. However, the distribution is irregular55within hardwood xylans (Fengel and Wegener [62], Jacobs et al. [105]).3.1.3 Wood hydrolysis factors influencing yield and molar massTemperature, time, pH, particle size, solids loading and reactor type are the primary fac-tors affecting hydrolysis reactions.3.1.3.1 Effects of temperature and timeAs described in Chapter 2, temperature and time have similar effects on biomass underboth acid and alkali conditions. Increasing temperature and time facilitates dissolutionof solid polysaccharides into soluble oligomers and monomers. However, high pretreat-ment severity also consumes a significant amount of energy and results in severe sugardegradation, generating large amount of volatile byproducts such as furfural, HMF, andother light organics (Pu et al. [216], Yan et al. [328]).Hydrolysis has been conducted at 100 oC to 240 oC for 0 to 120 minutes (Kang et al.[118], Kemppainen et al. [123], Larsson et al. [139], Leppa¨nen et al. [148], Pu et al. [216],Song et al. [273]). A maximum yield of soluble sugar from mixed softwood was achievedat 170 oC for 2h by Pu et al. (Pu et al. [216]). Increasing temperature to 180 oC and in-creasing the residence time led to lower solids recovery and higher yields of degradationbyproducts such as furfural and light organics. Song et al. reported GGM removal as highas 90% from ground wood of spruce at 170 oC for 60 minutes, but did not report on theyield of soluble oligomers. Leppa¨nen et al. found that hemicellulose yields from the hy-drolysis of Norway spruce at 120-160 oC were low but increased significantly at 170-240oC. Complete hemicellulose extraction occurred at 220 oC and cellulose degradation wasobserved after hydrolysis at 240 oC.For hydrolysis of woody biomass, temperature also has a strong effect on the aver-age molar mass of soluble products. Temperatures lower than 160 oC and short reactiontime favor generation of high molar mass oligomers (Song et al. [273]). High tempera-ture reduces hemicellulose oligomer molar mass, and compared to pretreatment temper-56ature of 160 oC, smaller molar mass was obtained at 170 oC (Rissanen et al. [227, 228]).Oligomers recovered at temperatures above 200 oC had a low average molar mass varyingfrom 0.3-1.5 kDa due to the severe hydrolysis conditions (Borrega et al. [23]). Leppa¨nenet al. conducted autohydrolysis of spruce saw meal and reported 170-180 oC as the opti-mized range to isolate high molar mass polysaccharides; the highest average molar massachieved was 31 kDa (Leppa¨nen et al. [148]). Xu et al. studied acid hydrolysis of spruceup to 90 oC with pH of 1-3. The molar mass of soluble galactoglucomannan droppedsignificantly when the temperature was increased above 70 oC or the pH was decreasedto less than 2.The overall objectives of this thesis require high oligomer yields and high oligomermolar mass. In the literature, high oligomer yields were commonly reported at high tem-perature (> 170 oC) and long residence time. Low temperature (< 160 oC) with shortresidence time facilitated generation of high molar mass oligomers. Therefore, a largetemperature range of 140 oC to 200 oC with residence time varying from 0 minute to 120minutes will be explored in this Chapter.3.1.3.2 Effects of particle size and solids loadingConcerns about heat transfer and mass transfer in and out of wood particles have drivenstudies of the effect of particle size on treatment efficiency. The decrease of particle sizecan reduce heat and mass transfer limitations during thermal pretreatment.Hydrothermal pretreatment of woody biomass had been conducted with particle sizesvarying from less than 1 mm to 10 mm by 10 mm with solids to liquid ratio (wt/wt) of2.5 to 25% (Borrega et al. [23], Kang et al. [118], Leppa¨nen et al. [148], Mok and Antal Jr[189], Pu et al. [216]). Smaller particle size and lower solid to liquid ratios favor hemicel-lulose yield from wood. Song et al. concluded that carbohydrate yields were 67% higherfrom ground wood than wood chips under the same reaction conditions. Rissanen et al.aand Rissanen et al.b compared pretreatment of 1.25-2 mm and 10×10 mm chips and con-57cluded that the small chip size favored overall extraction rate and production of largehemicellulose oligomers. The highest molar mass of hemicelluloses from large chips wasapproximately 20 kDa at 130 oC for over 200 minutes, whereas values over 60 kDa wereobtained with the smaller chips at 170 oC for 10 minutes. Pu et al. conducted hydrolysisof softwood chips with water in a lab scale batch digester at a solid to liquid ratio of 20wt% and obtained 13% sugar from wood. A similar yield of 10% was achieved from thehydrolysis of Southern pine chips using a lower solid to liquid ratio of 15 wt% (Kang et al.[118]).Small biomass particle size and low solid loading facilitate mass and heat transfer dur-ing treatment and favor the dissolution of biomass solids. Low temperature with longresidence time (130 oC for 200 minutes) favors production of high molar mass oligomersfrom large biomass particles; high temperature with short residence time (170 oC for 10minutes) promotes production of high molar mass oligomers from small biomass parti-cles. This chapter, we further investigate particle size effects on hemicellulose yield andmolar mass to identify hydrolysis conditions for Chapter 7 kraft pulping of softwoodchips.3.1.3.3 Effects of pHThe use of inorganic acids such as sulfuric acid, hydrochloric acid, and phosphoric acidfor hydrolysis has been investigated (Fengel and Wegener [62]). Sulfuric acid is often usedas a catalyst for softwood hydrolysis. Softwood was impregnated with 0.5-5 wt% sulfu-ric acid at room temperature for 30 minutes to overnight before hydrolysis (Kemppainenet al. [123], Larsson et al. [139]). Kemppainen et al. concluded that acid catalyzed hydrol-ysis led to more complete hemicellulose removal compared to treatment without acid.Arabinose groups in AGX were extremely reactive in acid hydrolysis and depolymerizedinto monomers almost completely (Kang et al. [118]). Significant pH changes occur dur-ing hydrolysis as a result of neutralization by the ash content of wood and release of acetic58acid (Kang et al. [118], Song et al. [274]). Song et al. found the degree of polymerizationof generated oligomers increases with increasing pH; an initial pH of approximately 4efficiently inhibited complete degradation of hemicellulose (Song et al. [274]).Low pH increases the removal rate but degrades hemicellulose into lower molar massoligomers, corrodes equipment, and requires neutralization prior to downstream chem-ical or biological treatment. To simplify processing and reduce chemical usage as dis-cussed in Chapter 2, no pH adjustment will be applied. This thesis will only considerautohydrolysis.3.1.3.4 Effects of reactor typeReactor type and the corresponding mass and heat transfer also play an important role inthe fractionation of biomass. Batch reactors are among the most common type of reactorsused in the laboratory (Borrega et al. [23], Kang et al. [118], Pu et al. [216]).Recently, studies of biomass have used systems such as the Dionex Accelerated Sol-vent Extractor (Song et al. [274]), flow-through reactor (Yan et al. [328]) and custom re-actors (Leppa¨nen et al. [148], Trajano et al. [295]) to provide a continuous flow of waterto biomass to achieve high hemicellulose yields in the hydrolysate. These flow-throughsystems improve monitoring of hydrolysis products as a function of time, limit side anddegradation reactions as well as product precipitation during reaction quenching. A sig-nificant amount of lignin and carbohydrates are removed but the concentration of sugarsin the hydrolysate is low. It has been shown that lignin removal from poplar is increasedby using a flow-through reactor (Yan et al. [328]). When poplar was subjected to flow-through autohydrolysis at 220-280 oC or dilute acid hydrolysis (0.05% (w/w) sulfuricacid) at 200-250 oC, 87% of lignin was removed (Yan et al. [328]).High sugar concentrations will favor subsequent adsorption operations and reducehandling volumes. Thus, a batch, or Parr, reactor will be used for the hemicellulose ex-traction from softwood chips.593.1.4 Severity factor and pseudo-timeThe severity factor, or H-factor (Equation 3.1) was developed by Vroom and expressescooking time and temperature as a single variable for kraft pulping. Vroom emphasizedthat relative reaction rate was not the absolute reaction rate, but the representation oftemperature dependence of these rates. The model predicts the compensation in timerequired for changes in temperature, or vice-versa.H =∫ t0exp(43.2− 16113T)d t (3.1)Overend and Chornet adapted the concept of severity factor to un-catalyzed hard-wood hydrolysis. In this model, temperature and time are combined in a single parame-ter (Equation 3.2) indicating that the overall kinetics obey first-order kinetics and the rateconstant has an Arrhenius dependence on temperature. Chum et al. and Abatzoglou et al.combined the severity factor with an acidity function, to model acid-catalyzed aqueousand organosolv treatments of biomass.R0 = exp(T − 10014.75)t (3.2)Hydrolysis involves non-isothermal heating and cooling periods during which biomasscan react. However, the effect of the non-isothermal process has seldom been evalu-ated. In this study, the non-isothermal heating and cooling periods were converted intopseudo-isothermal reaction time Section 3.4.2.603.2 GoalsThis chapter targets calibration of the operating space for feedstock and reactor, gener-ating data for subsequent modeling, and providing a guide for work in Chapter 5 andChapter 7.To produce high yields of high molar mass hemicellulose oligomers from hot waterhydrolysis of softwood chip fines, a wide range of temperatures (120-200 oC) and resi-dence times (0-120 minutes) will be investigated. Softwood chips will be milled to 0.4 mmparticles to facilitate mass and heat transfer and increase dissolution of biomass solids. Aparticle size effect of raw softwood chips, chip fines and milled chips will also be ex-plored. Solids consistency of 5 wt%, 10 wt% and 15 wt% will be tested. The impact ofagitation rate (90 r.p.m., 180 r.p.m., and 270 r.p.m.) will be assessed. Conditions identifiedin this chapter will be used for pre-hydrolysis kraft pulping in Chapter 7.A thorough compositional analysis of hydrolysate and hydrolyzed solids is provided.Sugar species and molar mass are reported. The evoluation of molar mass during hydrol-ysis will support model development in Chapter 5.Another pulp mill residue, hog fuel, will be subjected to identical hydrolysis condi-tions and compared to softwood chip fines. The diversity of feedstock composition maychange hydrolysate characteristics and possible applications.613.3 Experimental Design3.3.1 Material and millingChip fines, a mixture of spruce, pine, and fir, were obtained from Canfor Pulp Products.The raw biomass was air-dried and milled into a uniform particle size by Wiley Millcoupled with a screen of 1 mm. Particle size distribution (Section 3.4.1) was determinedusing 40 mesh (0.42 mm) and 60 mesh (0.25 mm) sieves. The milled softwood was usedfor conditions labelled ”mill”.Particle size effects were investigated using softwood chips provided by Canfor PulpProducts. The chips were pre-screened into fines with a thickness less than 2 mm, andchips with a thickness from 2-6 mm. Chips and chip fines were compared with the milledchips at the same hydrolysis conditions. Results are discussed in Section 3.4.3.3.3.3.2 HydrolysisBiomass solids and water were mixed with deionized water to generate 500 g slurry ina stainless steel pressurized reactor (Parr Instrument Company 4843). pH was measuredbefore and after hydrolysis. The reactor was heated using a coil jacket. Agitation of 180r.p.m. was applied through the entire heating and reaction process. Since the heating timeto achieve the target temperature was approximately 40-60 minutes, a residence time of 0minute (i.e. the end of heat-up period) was investigated at each temperature. The reactorwas quenched in ice water when residence time was achieved. Heating and cooling timeswere recorded. Two to three replicates were conducted for each condition.The softwood hydrolysis experimental design is provided in Table 3.1 and is orga-nized by temperature. Three levels of agitation rate, solids consistency, and particle sizeare tested as in Table 3.2. Hog fuel hydrolysis is compared with chip fines at the condi-tions in Table 3.3. The maximum hemicellulose oligomer yield error was 2.3% relative too.d. biomass (average 1%); the maximum monomer yield error was 1.4% relative to o.d.62biomass (average 0.3%).Table 3.1: Experimental design of softwood hydrolysis conditions (the time listed isisothermal time without conversion to pseudo-time).ParticlesizeTemperature(oC)Time(min)Consistency(%)Agitation(r.p.m.)ReplicatesMill 140 0 5 180 2Mill 4.5 5 180 2Mill 9 5 180 2Mill 27 5 180 2Mill 160 0 5 180 2Mill 9 5 180 2Mill 27 5 180 3Mill 40 5 180 2Mill 100 5 180 2Mill 170 0 5 180 2Mill 9 5 180 2Mill 27 5 180 2Mill 180 0 5 180 2Mill 9 5 180 2Mill 27 5 180 2Mill 60 5 180 3Mill 200 0 5 180 2Mill 4.5 5 180 3Mill 9 5 180 263Table 3.2: Experimental design of factors influencing softwood hydrolysis (the timelisted is isothermal time without conversion to pseudo-time).Factor ParticlesizeTemperature(oC)Time(min)Consistency(%)Agitation(r.p.m.)ReplicatesAgitation Mill 160 18 5 90 1Mill 5 180 1Mill 5 270 1Consistency Mill 180 18 5 180 2Mill 10 180 1Mill 15 180 1Particle size Fines 170 0 5 180 1Fines 10 180 1Chip 5 180 1Chip 10 180 1Table 3.3: Experimental design of hog fuel hydrolysis (the time listed is isothermaltime without conversion to pseudo-time).Temperature(oC)Time(min)Consistency(%)Replicates160 0 5 118 5 168 5 1200 0 5 14.5 5 118 5 1643.3.3 Compositional analysisThe carbohydrate and lignin composition of the untreated solids, hydrolyzed solids, andliquid hydrolysate were determined according to the National Renewable Energy Lab-oratory (NREL) Laboratory Analytical Procedures (LAP) (Hames et al. [86], Sluiter et al.[265, 266]). Total oligomer yield was calculated from the difference between monomerconcentration before and after acid hydrolysis of the reactor hydrolysate. Acid solublelignin (ASL) was measured by UV-Vis spectroscopy (Shimadzu UV-1800 spectrophotome-ter) at wavelength of 240 nm and acid insoluble residue (AIR) (acid insoluble lignin andash) was determined by gravimetric analysis.Monomer concentration was analyzed with a Dionex ICS-50000 high-performance liq-uid chromatography (HPLC) (Thermo Scientific) coupled with CarboPac SA 10 columnand pulsed amperometric detector. Process samples were filtered through a 0.22 µmnylon syringe filter before injection. A volume of 10 µL was injected into the columnequilibrated with 66 mM KOH solution and eluted with 1 mM KOH at a flow rate of 1.5mL/min.653.4 Results and Discussion3.4.1 Particle size distributionThe particle size distribution was determined by sizing the milled solids with 40 mesh(0.42 mm) and 60 mesh (0.25 mm) screens (Conversion [41]). The total recovered solidswere calculated relative to the wood mass prior to screening. The distribution was de-termined from six replicates. More than half of the particles were smaller than 60 mesh(Table 3.4). Less than 10% of the particle size was above 40 mesh and under 1 mm.Table 3.4: Milled softwood chip fines particle size distributionParticle size Average (wt%) Standard deviation (wt%)> 40 mesh 9.31 1.5840-60 mesh 33.18 1.74< 60 mesh 57.50 2.80Mass balance 97.63 0.473.4.2 Severity factor and pseudo-time calculationPseudo-time was adapted from the severity factor in order to compensate for heatingand cooling periods (Abatzoglou et al. [1], Chum et al. [36], Overend and Chornet [201],Vroom [312]). In the calculation (Equation 3.3 and Equation 3.4), time and temperatureare converted to an isothermal condition. In this work, 100 oC is used as the temperaturethat hemicellulose hydrolysis begins. A schematic illustration is shown in Figure 3.3.66Figure 3.3: A schematic illustration of pseudo-time calculation method.Ap =∫ trxnt100Tdt (3.3)tp =ApT − 100 (3.4)where t100 is the time to heat up to 100 oC in minutes, trxn is the time to heat up to isother-mal temperature in minutes. The value of Ap is the area beneath the heating and coolingtemperature-time profiles, T is the reaction temperature in oC, t is the heating or coolingtime in minutes and tp is the calculated pseudo-time. Polymath was used to calculate thepseudo-time in minutes.The average pseudo-time for heating and cooling is presented as a function of temper-ature in Table 3.5.67Table 3.5: Severity factor (calculated from Equation 3.2) and pseudo-time for hydrol-ysis of milled softwood chip fines at varying temperature and time.Temperature(oC)Time(min)Severityfactoraverage(logR0)SeverityfactorstandarddeviationPseudo-timeaverage forheating andcooling (tp,min)Pseudo-timestandarddeviation forheating andcooling (min)140 0 2.05 0.007 6.70 0.594.5 2.22 0.0169 2.38 0.00327 2.70 0.001160 0 2.80 N.A. 10.16 0.709 3.00 0.04818 3.22 N.A.27 3.21 0.205100 3.80 N.A.170 0 3.13 0.046 12.09 0.389 3.36 0.01427 3.65 0.008180 0 3.53 0.058 14.19 0.699 3.69 0.01527 3.94 0.02060 4.21 0.015200 0 4.21 0.000 18.39 0.544.5 4.26 0.0529 4.35 0.002683.4.3 Operation factor analysis3.4.3.1 Agitation rate effectThe effect of agitation was examined at 90 r.p.m., 180 r.p.m. and 270 r.p.m. at 160 oC for28.16 minutes. From Figure 3.4, the oligomer and monomer yields were independent ofagitation rate. All subsequent hydrolysis experiments were conducted at 180 r.p.m.90 RPM 180 RPM 270 RPM024681012Yield of o.d. biomass (%)Agitation rate Oligomer yield Monomer yieldFigure 3.4: Oligomer and monomer yield of o.d. softwood at 160 oC for 28.16 minutesas a function of agitation rates from 90 r.p.m. to 270 r.p.m.3.4.3.2 Solids consistency effectThe effect of milled softwood solids consistency was compared at 5wt%, 10wt% and15wt% at 180 oC for 32.19 minutes (Figure 3.5). Oligomer yield decreased 20.4% as consis-tency increased from 5wt% to 10wt%. Monomer yields were unaffected by solids consis-tency. Oligomer production was favoured by 5wt% consistency therefore this was appliedin all other experiment.695% 10% 15%02468101214Yield of o.d. biomass (%)Solids consistency (%) Oligomer yield Monomer yieldFigure 3.5: Oligomer and monomer yield of o.d. softwood at 180oC for 32.19 minutesas a function of consistency.3.4.3.3 Particle size effectThe influence of biomass particle size on hemicellulose yield and molar mass was as-sessed by hydrolyzing chips, chip fines, and milled chips at 5wt% solids consistencyat 170 oC for 12.09 min. Due to mechanical limitations of the Parr reactor, the reactorcould not be agitated when treating fines and chips. Figure 3.6 shows the monomer andoligomer yield from different particle sizes. Oligomer yield decreased with increasingchip size. The larger particles, chips and chip fines, resulted in much lower yields com-pared to the milled chips indicating that caution must be exercise when making conclu-sions based on work conducted with milled chips regarding large-scale chip hydrolysisperformance.70Mill Fines Chips012345Yield of o.d. biomass (%)Softwood particle size Oligomer yield Monomer yieldFigure 3.6: Oligomer and monomer yield of o.d. softwood for milled chips (Mill) asSection 3.4.1, chip fines with with a thickness less than 2 mm (Fines) and chipswith a thickness from 2-6 mm (Chips) at 170oC for 12.09 minutes with solidsconsistency of 5wt% as a function of particle size.3.4.3.4 Effect of compositionThe feedstocks were characterized according to the NREL LAP (Hames et al. [86], Sluiteret al. [265, 267, 268, 269]). The detailed composition of oven-dried chip fines, chips andmilled chip fines shipped and processed in different years is provided in Table 3.6. Chipsand chip fines were shipped and processed in 2017. Milled chip fines were shipped in2014 and processed from 2014 to 2017.The composition of softwood chip fines milled in 2014 and 2015 were similar, espe-cially the extractives content and total carbohydrates content (Table 3.6). The materialmilled in 2016 and 2017 came from the original 2014 shipment. The material milled in2016 had a higher extractive content and lower carbohydrate content than material milledin 2014 and 2015. This trend was more noticeable when additional material was milled in2017: the extractives content was more than four times larger than the extractives contentmeasured in 2014 and 2015 and the total carbohydrate content was 26% lower than thecarbohydrate content measured in 2014 and 2015.71Table 3.6: Compositional analysis of softwood chips, chip fines (Fines), and milled chip fines (Mill) by shipping yearand processing year. Weight fraction (%) of components is presented on an moisture free (o.d. softwood) basis;error limits are standard deviation calculated from replicates. *Acid insoluble residue is the sum of ash and acidinsoluble lignin.Shippingyear, sub-strate andprocessingyearExtractives Totalcarbohy-dratesArabinan Galactan Glucan Xylan Mannan Acid in-solubleresidue*AcidsolubleligninTotalmassbalance2017 Chips 2017 2.06 64.8 1.3 2.28 44.14 6.11 10.97 26.13 4.97 97.96Standard deviation 0.05 0.71 0.01 0.03 0.5 0.11 0.06 0.55 0.06 0.682017 Fines 2017 1.2 65.85 1.28 2.09 45.52 6.01 10.94 25.82 5.2 98.07Standard deviation 0.01 0.71 0.03 0.02 0.5 0.1 0.1 1.21 0.31 1.562014 Mill 2017 18.79 47.51 0.91 1.47 34.34 4.38 6.41 25.81 4.94 97.05Standard deviation 0.01 0.54 0.02 0.02 0.35 0.11 0.04 0.2 0.15 0.92014 Mill 2016 6.88 62.11 1.28 2.16 43.45 5.28 9.94 25.99 5.21 100.19Standard deviation 0.15 2.2 0.04 0.06 1.37 0.16 0.59 1.91 0.03 1.832014 Mill 2015 4.01 65.08 1.41 2.42 45.03 5.65 10.58 26.67 5.24 101.01Standard deviation 0.19 1.3 0.05 0.09 1.07 0.36 0.54 1.25 1.95 1.72014 Mill 2014 4.88 63.99 1.4 2.53 44.26 6 9.8 27.31 4.04 100.22Standard deviation 0.56 2.65 0.04 0.04 1.79 0.2 1.07 0.45 1.31 1.2372The material taken from the stock bag in 2016 and 2017 represented the lower layers ofthe original shipment. Multiyear storage, even at 7 oC, may have enabled microorganismgrowth which would decrease carbohydrate content.Water extraction was conducted with a Dionex ASE 350 accelerated solvent extractoras part of compositional analysis. The extraction was performed at 100 oC for 21 min-utes with reaction cell pressure of 1600 psi. The polysaccharides content of this extract isreported in Table 3.7.Table 3.7: Polysaccharides analysis of water extract generated by Dionex ASE 350(% yield of o.d. softwood) supplementary to Table 3.6. Note: S2014 M2017represents shipping year 2014 and milling processed year 2017.Biomass S2014M2017S2014M2016S2014M2015S2014M2014Total extractive 18.79 6.88 4.01 4.88Total sugar monomers 0.10 0.00 0.00 0.00Total sugar oligomers 1.84 1.16 0.68 0.79Galactan oligomer 0.25 0.26 0.34 0.79Glucan oligomer 0.65 0.32 0.10 0.00Xylan oligomer 0.20 0.11 0.03 0.00Mannan oligomer 0.63 0.38 0.16 0.00Arabinan oligomer 0.11 0.08 0.05 0.00Polysaccharides removal occurred mainly as oligomers, and specifically galactan oligomersas in S2014 M2014 (Table 3.7), indicating galactan is more susceptible than other polysac-charides components to water extraction. No monomer content was detected for M2014to M2016 due to either low concentration within extraction liquor or no monomer gen-eration under extraction condition. The monomer content was detected for M2017 likelyextracted from soluble carbohydrates in the cell wall converted by microbial growth.73The distribution of oligomers within extract also varied by batch. Galactan contentcontinuously decreased while the fraction of other polysaccharides increased. The possi-ble reason could be that galactan was susceptible for degradation to non-sugar productsduring storage, maybe due to microbial growth. Glucan and mannan are typically muchmore stable components than xylan and arabinan. The presence of glucan oligomers sug-gests possible degradation of cellulose within wood matrix.To compare the influence of raw biomass composition on oligomer and monomeryield, hydrolysis was conducted at 170 oC for 12.09 minute and 180 oC for 23.19 minuteswith two different batches M2015 and M2017 (Table 3.7, Figure 3.7). At 170 oC, oligomerand monomer yield from high extractive content material (long storage batch M2017)were 58% and 67% greater than from the low extractive content material (short storagebatch M2015). However, at 180 oC the oligomer yield from both materials were compa-rable while the monomer yield from the high extractive content material (long storagebatch M2017) was greater than the yield from the low extractive content material (shortstorage batch M2015). This may be due to oligomer depolymerization to monomer.170C 12.09min 180C 23.19min0246810121416Yield of o.d. biomass (%)Hydrolysis condition Oligomer yield M2015 Oligomer yield M2017 Monomer yield M2015 Monomer yield M2017Figure 3.7: Oligomer and monomer yield of o.d. softwood at 170 oC for 12.09 minuteand at 180 oC for 23.19 minutes from same batch of chip fines processed in 2015and 2017. M2015 is a short storage batch and M2017 is a long storage batch.74The raw biomass composition had significant impact on oligomer yield. High extrac-tive content correlated to increased oligomer and monomer yield. Consequently, only re-sults from M2014 and M2015 are discussed in Section 3.4.4, Section 3.4.5, and Section 3.4.6.3.4.4 Hydrolysate characterization3.4.4.1 Hemicellulose oligomer and monomer yieldThe oligomer and monomer yields from oven-dried chip fines are shown in Figure 3.8 asa function of residence time. Generally, increasing temperature facilitated oligomer pro-duction. Longer times led to higher oligomer yields at 160 oC and 180 oC, and a loweryield at 200 oC. The decline of yield at 200oC suggested degradation of hemicellulose.Similar trends have been reported by many others (Pu et al. [216], Yan et al. [328]). Amaximum oligomer yield of 13.8% of oven-dried chip fines was achieved at 200 oC and4.5 minutes. At all conditions, the monomer yields were less than 5% indicating the hy-drolysis conditions were relatively mild.0 20 40 60 80 100 120024681012141618 140 oC 160 oC 170 oC 180 oC 200 oCOligomer yield (%)Pseudo-time (min)(a)0 20 40 60 80 100 1200123456 140 oC 160 oC 170 oC 180 oC 200 oCMonomer yield (%)Pseudo-time (min)(b)Figure 3.8: (a) Oligomer yield and (b) monomer yield of 25g o.d. softwood at 140-200oC.753.4.4.2 Hemicellulose compositionThe mass of oligomers produced by species are shown in Figure 3.9. Softwood hemicel-lulose contains AGX and GGM (Fengel and Wegener [62], Sjostrom [262]). It had beenreported that the majority of soluble poly- and monosaccharides recovered from Norwayspruce were mannose, glucose and galactose; mannose and glucose were present primar-ily as oligomers after autohydrolysis at 150-180 oC (Leppa¨nen et al. [148]).20 40 60 80 100 120 1400.00.20.40.60.81.01.21.41.61.82.02.2   Glucan Xylan Mannan Arabinan GalactanOligomer (g)Pseudo-time (min)(a)10 20 30 40 50 60 70 80 90 100 1100.00.20.40.60.81.01.21.41.61.82.02.2   Glucan Xylan Mannan Arabinan GalactanOligomer (g)Pseudo-time (min)(b)16 18 20 22 24 26 28 30 320.00.20.40.60.81.01.21.41.61.82.02.2   Glucan Xylan Mannan Arabinan GalactanOligomer (g)Pseudo-time (min)(c)Figure 3.9: Oligomer composition in hydrolysate at (a) 160oC, (b) 180oC and (c) 200oCfrom 25g o.d. softwood.As the Man:Glu:Gal ratio in water-soluble GGM is 3:1:1 (Fengel and Wegener [62]),calculation of the mannan to glucan ratio can provide an indirect measure of degree ofcellulose hydrolysis. At all tested conditions at 160 oC, the ratio of mannan to glucanwas 2.5-2.9:1 (Figure 3.9), which could due to incomplete release of mannan and glucan.76At 180 oC, the mass of mannan to glucan to galactan is approximately 3:1:1 from 14.19minute to 41.19 minutes. Using this ratio, it is estimated that GGM represents 77%-86%of the total oligomer content. The 3:1 ratio of mannan to glucan was maintained at 200oC and 18.39 minutes, but decreased to 2.1-2.3:1 after 22.89 minutes and 27.39 minutes.When the mass of mannan oligomer relative to glucan decreased below 3:1, the degreeof glucan oligomer extraction has increased and/or the degree of mannan oligomer re-covery has decreased. From Figure 3.11 and Figure 3.13, neither mannan oligomer norglucan oligomer degraded at 160-200 oC. As a result, at 200 oC, a small amount of cellu-lose may hydrolyze but the majority of glucan oligomer products are likely derived fromhydrolysis of hemicellulose.The negative mass of arabinan oligomer is an artefact of the calculation method, whichis the difference in monomer content between pre-hydrolyzed liquor and post-hydrolyzedliquor. If large amounts of monomer degrade during post-hydrolysis, a negative value isobtained.The mass of monomers produced during hydrolysis are reported in Figure 3.10 byspecies. Arabinose and xylose production is more sensitive to temperature; at 160-200 oC,arabinose and xylose are the major monomers. Arabinoxylan, the pentose-based polymer,was easily hydrolyzed to monomers while GGM, which is hexose-based, was hydrolyzedto oligomers. Degradation of monomeric arabinose was observed at 180 oC and 200 oC.7720 40 60 80 100 120 140 160-0.10.00.10.20.30.40.50.6   Glucose Xylose Mannose Arabinose GalactoseMonomer (g)Pseudo-time (min)(a)10 20 30 40 50 60 70 80 90 100 110-0.10.00.10.20.30.40.50.6   Glucose Xylose Mannose Arabinose GalactoseMonomer (g)Pseudo-time (min)(b)16 18 20 22 24 26 28 30 32-0.10.00.10.20.30.40.50.6   Glucose Xylose Mannose Arabinose GalactoseMonomer (g)Pseudo-time (min)(c)Figure 3.10: Monomer composition in hydrolysate at (a) 160oC, (b) 180oC and (c)200oC from 25g o.d. softwood.3.4.4.3 Hemicellulose component yieldThe resource of hemicellulose monomer generation within hydrolysate could be compli-cated, not just from hemicellulose, but also non-structural polysaccharides, such as pectin.Therefore, the monomer composition discussion in this section only considers the maincontributing source of hemicellulose.Mannan yield from the mannan content in untreated softwood mill is shown in Fig-ure 3.11. The total yield of mannan (oligomer and anhydro-monomer) exceeds 80% in-dicating the majority of mannan is hydrolyzed (Figure 3.11a). Mannose yield is low; themaximum yield was 1.5-8.5% at 200 oC (Figure 3.11b).780 20 40 60 80 100 120020406080100   160oC 180oC 200oCTotal mannan yield (%)Pseudo-time (min)(a)0 20 40 60 80 100 120-202468101214(b)   160oC 180oC 200oCMannose yield (%)Pseudo-time (min)Figure 3.11: (a) Total mannan yield (oligomer and monomer) and (b) mannose yieldrelative to mannan in 25g o.d. untreated softwood mill at 160-200 oC.Galactan yield followed a similar trend to mannan yield as shown in Figure 3.12. Hy-drolysis at 180 oC generated yields near 100% but some degradation was observed at 200oC. Galactan is found as branches on the glucomannan backbone of GGM (Fengel andWegener [62]) and is thus more reactive compared to mannan and glucan. The maximumgalactose yield was 35% at 200 oC (Figure 3.12b).0 20 40 60 80 100 120020406080100120140   160oC 180oC 200oCTotal galactan yield (%)Pseudo-time (min)(a)0 20 40 60 80 100 120051015202530354045(b)   160oC 180oC 200oCGalactose yield (%)Pseudo-time (min)Figure 3.12: (a) Total galactan yield (oligomer and monomer) and (b) galactose yieldrelative to galactan in 25g o.d. untreated softwood mill at 160-200 oC.79Glucan content in Table 3.6 includes hemicellulose and cellulose. The hemicellulose-specific glucan was assumed to be proportional to the mannan content in GGM; thecommonly-reported Man:Glu:Gal ratio of 3:1:1 was used to distinguish between hemicellulose-specific glucan and cellulose-specific glucan (Fengel and Wegener [62]). The glucan yieldin Figure 3.13b was calculated relative to the potential hemicellulose glucan content as13 of mannan content. In Figure 3.13a, the glucan yield is low compared to the mannanand galactan yields, as previously discussed. This is a result of the presence of cellu-lose, which is difficult to hydrolyze. However, glucan yield when calculated relative tohemicellulose-specific glucan in the raw biomass (Figure 3.13b) is comparable to mannanyields at (Figure 3.11). This further supports the previous conclusion that the majorityof glucan oligomers were derived from hemicellulose GGM and only a small amountof cellulose hydrolyzed. However, it should be noted that glucose might also derivefrom starch within softwood, which accounts for 1.1% of o.d. wood in species such asAbieslasiocarpa and Abiessibirica (Willfo¨r et al. [320]).0 20 40 60 80 100020406080100120   160oC 180oC 200oCTotal glucan yield (%)Pseudo-time (min)(a)0 20 40 60 80 100 120020406080100120   160oC 180oC 200oCTotal glucan yield (%)Pseudo-time (min)(b)Figure 3.13: (a) Total glucan yield (oligomer and monomer) relative to glucan includ-ing cellulose and hemicellulose and (b) hemicellulose glucan relative to poten-tial hemicellulose glucan in 25g o.d. untreated softwood mill at 160-200 oC.80Total xylan yield and xylose yield are shown in Figure 3.14. The total yield is lowerthan mannan and galactan (Figure 3.14a). Some degradation is observed at 200 oC, simi-lar to galactan (Figure 3.12). Monomer yields from xylan are high compared to monomeryields from mannan or galactan while oligomer yields from xylan are low. It was pre-viously found that arabinose and xylose from mixed softwood were primarily detectedas monomers after hydrolysis at 170 oC and 180 oC (Pu et al. [216]). The structure ofxylan is likely responsible for this observation. Xylan chains are highly substituted with4-O-methylglucuronic acid and arabinofuranose units (Fengel and Wegener [62]). The av-erage ratio for Xyl:Me-GluU:Ara is given as 8:1.6:1, which means there is one side groupfor every two or three of xylose units in the backbone (Fengel and Wegener [62]). Thehigh frequency of acidic side groups likely increases the probability of backbone scissioninto monomers. The linkages between xylose and the 4-O-methylglucuronic acid moietyis known to be more stable than most glycosidic bonds under acidic conditions, whichsuggests that the pH near xylose units is lower due to the retained 4-O-methylglucuronicacid moiety. This would be especially true at 200 oC when a reduction in oligomer yieldwas observed.0 20 40 60 80 100 120020406080100   160oC 180oC 200oCTotal xylan yield (%)Pseudo-time (min)(a)0 20 40 60 80 100 12001020304050   160oC 180oC 200oCXylose yield (%)Pseudo-time (min)(b)Figure 3.14: (a) Total xylan yield (oligomer and monomer) and (b) xylose yield rela-tive to xylan in 25g o.d. untreated softwood mill at 160-200 oC.81Arabinan total yield was fairly high. The majority was produced as monomer and ev-idence of severe degradation at 180 oC and 200 oC is shown in Figure 3.15a. Arabinan ispresent with galactan or xylan in pyranose or furanose form (Fengel and Wegener [62]).Inboth cases, arabinan is a side group attached to the backbone thus is more vulnerable tothermal treatment, and readily degrades to oligomer and monomer. The ease of arabi-nan hydrolysis is also reported by other authors; up to 75% of solubilized arabinan wasdetected in monomer form (Leppa¨nen et al. [148], Sjostrom [262]).Arabinan is only 1.28-1.41% of the raw softwood chip fines (Table 3.6). Low concentra-tion and high reactivity are the sources of the large error bars in Figure 3.15. Yields greaterthan 100% are primarily due to arabinose concentration approaching the lower detectionlimits of the Dionex ICS-50000 high performance liquid chromatography (Thermo Scien-tific).0 20 40 60 80 100 120020406080100120140(a)   160oC 180oC 200oCTotal arabinan yield (%)Pseudo-time (min)0 20 40 60 80 100 120020406080100120   160oC 180oC 200oCArabinose yield (%)Pseudo-time (min)(b)Figure 3.15: (a) Total arabinan yield (oligomer and monomer) and (b) arabinose yieldrelative to arabinan in 25g o.d. untreated softwood mill at 160-200 oC.823.4.5 Hydrolyzed solids characterization3.4.5.1 Acid insoluble residue and acid soluble ligninAIR includes acid insoluble lignin and ash, which was determined by gravimetric analy-sis of post-hydrolyzed solids before and after oven-drying. From Figure 3.16a, AIR con-tent of hydrolyzed solids relative to AIR of untreated softwood was approximately 1,suggesting hot water hydrolysis does not effectively remove insoluble lignin and ash.0 20 40 60 80 100 1200.00.20.40.60.81.01.2 160oC 180oC 200oCHydrolyzed solids AIR of initial AIRPseudo-time (min)(a)0 20 40 60 80 100 1200.00.20.40.60.81.01.2 160oC 180oC 200oCHydrolyzed solids ASL of initial ASLPseudo-time (min)(b)Figure 3.16: Hydrolyzed solids (a) AIR relative to AIR of untreated 25 g of o.d. soft-wood mill and (b) ASL relative to ASL of untreated 25 g of o.d. softwood millat 160 oC, 180 oC and 200 oC as a function of pseudo-time.ASL content of hydrolyzed solids decreased with residence time at 160 oC and 180 oC,but plateaued (0.47-0.60) for times greater than 40 minutes (Figure 3.16b). Lignin removalduring acid hydrolysis in batch reactor is typically low (Leppa¨nen et al. [148], Mok andAntal Jr [189]). Leppa¨nen et al. conducted autohydrolysis of spruce at 180 to 240 oC andreported 9% to 21% of lignin dissolved.833.4.5.2 Solubilized ligninSolubilized ASL was determined by the ASL content difference of biomass before andafter hydrolysis. From Figure 3.17a, a positive correlation between solubilized ASL andhemicellulose yield could be found at 160-200 oC. The ratio of solubilized ASL to hemi-cellulose also demonstrates a positive correlation with solubilized ASL (Figure 3.17b).When total hemicellulose yield including oligomer and monomer increased from 3.2% to12.7% of o.d. biomass at 160 oC, the ratio increased from 0.04 to 0.17. This ratio stabilizedaround 0.13-0.14 at 180oC and 200oC, when hemicellulose yield stabilized around 14.8-18.1% (Figure 3.17b). This correlation agrees with previous discussion of LCC (Leppa¨nenet al. [148], Mok and Antal Jr [189], Sjostrom [262]). Lignin was removed from wood withcarbohydrate, mainly hemicellulose (Giummarella [76], Tarasov et al. [284]). Therefore,at 200 oC, lignin could not be further removed when hemicellulose removal maintains atrelatively constant range.0.0 0.5 1.0 1.5 2.0 2.5246810121416182022(a) 160 oC 170 oC 180 oC 200 oCHemicellulose yield of o.d. biomass (%)Solubilized ASL of o.d. biomass (%)0.0 0.5 1.0 1.5 2.0 2.50.040.060.080.100.120.140.160.180.20(b) 160 oC 170 oC 180 oC 200 oCRatio of lignin to hemicellulose (wt/wt)Solubilized ASL of o.d. biomass (%)Figure 3.17: (a) Hemicellulose yield of o.d. softwood including oligomer andmonomer (b) ratio of solubilized ASL to hemicellulose (wt/wt) compare tosolubilized ASL of o.d. softwood chip fines at 160 oC to 200 oC.843.4.5.3 Solubilized solidsSolubilized solids were determined by the o.d. weight difference of biomass before andafter hydrolysis. Increasing temperature and/or time increases biomass solubilization.When hydrolysis temperature increased from 160oC to 180oC, the solubilized solids in-creased from 5% to more than 35% (Figure 3.18). Similarly reported 4% softwood soluil-ization after autohydrolyzsis at 150 oC for 1 hour and 24% soluilization after autohydrol-ysis at 180 oC for 2 hours Pu et al..At 200oC, the percentage solubilized solids were comparable to 180oC (Figure 3.18),suggesting the near complete removal of hemicellulose, negligible removal of celluloseand constant removal of lignin at high temperature. Complete removal of hemicellulosewas reported previously by Mok and Antal Jr. They studied six woody and four herba-ceous biomass species in a percolating reactor with water for 0-15 min at 200-230 oC. Atotal of 40 to 60% of solids were dissolved; 100% hemicellulose, 4 to 22% cellulose and 35to 60% lignin were solubilized.0 20 40 60 80 100 1200510152025303540 160oC 180oC 200oCSolublized solids of o.d. biomass (%)Pseudo-time (min)Figure 3.18: Solubilized solids of 25 g o.d. softwood at 160oC, 180 oC and 200 oC as afunction of pseudo-time.853.4.6 Chip fines vs. hog fuel3.4.6.1 Feedstock composition comparisonThe composition of hog fuel are provided in Table 3.8. Compared to Table 3.6 2014 Mill2014, hog fuel has slightly lower total carbohydrates content than chip fines but overall,the two feedstocks are quite similar.Table 3.8: Compositional analysis of hog fuel (wt% o.d. biomass) with standard de-viation of 2-4 replicates after ±.Composition (%) Hog fuelExtractive 5.65 ± 1.09Total carbohydrates 59.84 ± 0.50Arabinan 1.78 ± 0.03Galactan 3.00 ± 0.01Glucan 40.43 ± 0.26Xylose 6.61 ± 0.05Mannan 8.02 ± 0.31AIR 28.29 ± 0.63ASL 5.49 ± 0.36Total mass balance 99.27 ± 0.403.4.6.2 Hemicellulose yield comparisonChip fines and hog fuel were hydrolyzed at 160 oC and 200 oC for 0 minute to 100 minutesat 5wt% solid to liquid ratio. Corresponding oligomer and monomer yields are shown inFigure 3.19.At 160 oC, oligomer and monomer yields from hog fuel and chip fines increasedwith increasing time. At 200 oC, oligomer yields decreased with increasing time while86monomer yields increased before ultimately decreasing. This indicates increased hydrol-ysis of oligomers to monomers followed by monomer degradation into light organicssuch as furfural (Pu et al. [216], Rivas et al. [229]). The total oligomer yield from hog fuelwas greater than that from chip fines at 160 oC. However, oligomer yield from chip finessurpassed that from hog fuel at 200 oC suggesting greater degradation of oligomers tomonomers. Further analysis of hog fuel hydrolysis by hot water could be found fromRangu.0 20 40 60 80 100024681012(a) Hog fuel Chip finesOligomer yield of o.d. feedstock (%)Time (min)-2 0 2 4 6 8 10 12 14 16 18 200246810121416(b) Hog fuel Chip finesOligomer yield of o.d. feedstock (%)Time (min)0 20 40 60 80 1000.00.51.01.52.02.53.03.5(c) Hog fuel Chip finesMonomer yield of o.d. feedstock (%)Time (min)-2 0 2 4 6 8 10 12 14 16 18 20012345(d) Hog fuel Chip finesMonomer yield of o.d. feedstock (%)Time (min)Figure 3.19: Oligomer yield from o.d. chip fines and hog fuel of hydrolysis (a) at160oC and (b) at 200oC; Monomer yield from o.d. chip fines and hog fuel ofhydrolysis (c) at 160oC and (d) at 200oC as a function of time.873.5 ConclusionHemicellulose oligomer production from low cost, readily available pulp mill residueswas examined. Operating guidelines for hydrolysis were developed and data was ob-tained for subsequent Chapter 5 kinetic model and Chapter 7 kraft pulping.High temperature, long residence time and low consistency favor oligomer produc-tion. Some hemicellulose degradation occurred at 200 oC. A maximum oligomer yieldof 13.8% of oven-dried chip fines was achieved at 200 oC and 4.5 minutes. Increasingparticle size lowered the yield of oligomer. Monomer yield was not influenced by parti-cle size or consistency. The composition of raw biomass had significant influence on theyield of hemicellulose in the hydrolysate. High extractive content increases oligomer andmonomer yield.In Chapter 7, softwood chips will be subjected to hydrolysis followed by kraft pulping.Hydrolysis will be conducted at low consistency (5-10wt%), low temperature (130-150oC) and long residence time (1-2 h) conditions. These conditions were selected based oncurrent results and literature review.Galactoglucomannan oligomers were the largest fraction of produced oligomers. Glu-can oligomers appear to derive primarily from galactoglucomannan, not cellulose. Nearcomplete removal of galactoglucomannan from chip fines was achieved at 180 oC. Pen-tose sugars were easily hydrolyzed to monomers while hexose sugars were stable asoligomers.Lignin removal increased with hydrolysis temperature and residence time, but plateauedat approximately 55% under high severity conditions.Hog fuel is more reactive to hydrolysis than chip fines. Hog fuel polysaccharidesdegraded at 200 oC.883.6 Future Work1. Characterization of lignin monomers and oligomers and extractives in hydrolysatewould generate insight into interactions with hemicellulose and potential for impacton downstream operations.2. Chemical treatment incorporation with mechanical treatment awaits further explo-ration. Currently, hydrolysis serves as pre-treatment step for wood processing. Butthe mass transfer property and kinetics are influenced by particle size as discussedin this chapter. How to better combine mechanical treatment with hydrolysis to useless energy for higher yield will be a critical question.3. Hydrolysis represents an extra processing step before traditional pulp operations.Thus additional energy and processing costs are incurred. Economic analysis isneeded to balance these costs against the value of new products.4. Pulp mill residues include diverse feedstocks such as hog fuel and sludge. Furtherinvestigation of these these materials could expand the range of potential applica-tions.89Chapter 4Hemicellulose Molar Mass Evolution:Methods Development and Analysis4.1 IntroductionThe twin concerns, global petroleum depletion and global warming, have generated stronginterest in the production of materials, chemicals, and fuels from renewable resourcessuch as biomass. Significant efforts have been devoted to the production of fuels andchemicals as biorefinery alongside existing pulp and paper operations from wood. (Ami-don and Liu [9], Liu et al. [167], Moshkelani et al. [191], Ruiz et al. [236], Van Heiningen[305]).Cellulose, hemicellulose and lignin form a very stable wood matrix that resists frac-tionation (Altaner and Jarvis [7], Cosgrove and Jarvis [44], Dammstro¨m et al. [46]). Hemi-cellulose’s structural heterogeneity creates diverse opportunities for deriving chemicalsand products for a range of industries including paper and pulp, pharmaceuticals, food,cosmetics and mining industries (Gı´rio et al. [74], Lindqvist et al. [159], Ma¨ki-Arvela et al.[179], Sella Kapu and Trajano [253], Silva et al. [256]). It can be removed from the woodmatrix via hydrolysis. Hydrolysis of wood is one pathway for lignocellulosic biomassfractionation and has a substantial literature (Chen et al. [29], Sella Kapu and Trajano90[253], Wyman [322]).The soluble products of hydrolysis are complex and change as reaction severity in-creases. Oligomers, or oligosaccharides, are one of the first products of hydrolysis andare one of the major hydrolysate components. These oligomers may further hydrolyze tomonomers. As reaction severity increases due to increasing temperature and/or time oracid addition, oligomers and monomers will degrade to products such as furfural, HMFand other light organics (Pu et al. [216]).Molar mass or molecular weight, affects hemicellulose oligomers’ processing and in-teraction with other molecules (Mori and Barth [190]) as well as properties such as solu-bility, optical properties, viscosity and biological activity vary with molar mass (Gridnevet al. [81], Liu et al. [164], Oberlerchner et al. [198], Xu et al. [325]). Utilization of hemi-cellulose oligomers requires selection and control of molar mass, but high molar massoligomers are hard to obtain due to the high reactivity of hemicellulose. Industrial-scalesize fractionation is time and energy consuming, consequently, process optimization forproduction of specific molar mass will be critical. Many kinetic models have been devel-oped, therefore, to describe hydrolysis process and facilitate yield control.Analytical molar mass characterization studies of oligomers are relatively rare, in partdue to a lack of uniform calibration standards (Oberlerchner et al. [198]). For carbohy-drates extracted from biomass, the chromatography signal is typically continuous andbroadly distributed with multiple peaks. Therefore, concentration and molar mass is dif-ficult to determine.The current study addresses this challenge by SEC narrow span standard two-dimensionalcalibration. This application opens the door for degraded carbohydrate polymer charac-terization and utilization in the future.914.2 Goals and HypothesesThis chapter targets for calibration methods development for SEC analysis of broad dis-tributed peak. The average molar mass and concentration of specific molar mass willbe determined. This approach will be applied to softwood hydrolysis data obtained inChapter 3. Molar mass evolution of hemicellulose oligomers will be discussed in thischapter and a kinetic model will be developed in Chapter 5.The hypotheses of this chapter include:1. Wood hydrolysate will exhibit broadly-distributed peaks.2. Molar mass and concentration distribution could be simultaneously determined bySEC.4.3 Two-dimensional Calibration MethodHemicellulose molar mass was analyzed by SEC on a Waters Alliance HPLC coupledwith Refractive Index (RI) detector. Ultrahydrogel 120, 250, and 1000 columns were cou-pled with sodium nitrate eluent (0.1M) at 0.5 mL ·min−1. A narrow standard calibrationmethod was applied using polyethylene glycol (PEG) standards with molar masses from106 g·mol−1 to 20,600 g·mol−1 (Table 4.1). Pullulan is a more commonly used calibrationstandard for hemicellulose, but the molar mass of available standards (6.1 kDa - 642 kDa)is much higher than the molar mass of oligomers produced by hydrolysis. Pullulan poly-mers are thus unsuitable for hydrolysate calibration (Section 6.3.4). Process samples werefiltered through a 0.45 µm nylon syringe filter before injection. The injection volume was100 µL.92Table 4.1: Calibration standard retention time and molar mass.Calibrationstandard #Retentiontime (min)Mw (Da) LogMwStd1 42.14 20600 4.31Std2 43.63 12600 4.10Std3 45.73 6690 3.83Std4 47.50 4290 3.63Std5 51.45 1400 3.15Std6 52.88 1030 3.01Std7 55.01 633 2.80Std8 56.85 430 2.63Std9 60.29 202 2.31Std10 61.84 106 2.03A two-dimensional calibration method was applied to determine the concentrationof specific molar mass from a broad distributed sample. Each PEG calibration standardwas prepared as a standard solution of 8 mg·mL−1, and further diluted to a minimum of0.0625 mg·mL−1.The first calibration dimension was conducted by varying PEG molar mass whilekeeping concentration constant (1 mg·mL−1) (Figure 4.1). Linear regression was con-ducted in OriginLab2016 and the resulting equation was used to determine hemicelluloseoligomer molar mass from retention time.9340 45 50 55 60 652.02.53.03.54.04.5 logMw Linear fitlogMwRetention time (min)y= - 0.112x+8.956R2 = 0.996Intercept = 8.956  0.284Slope = - 0.112  0.00595% confidence limits after Figure 4.1: Logarithm of calibration standard molar mass plotted against SEC reten-tion time.The second calibration dimension for each PEG polymer were generated by varyingconcentration (0.0625-8 mg·mL−1). The concentration calibration was conducted usingPEG standard (STD) 1, 3, 5, 7, and 9 (Table 4.1 and Figure 4.3a). Chromatograph peak areawas integrated in OriginLab2016 and plotted as a function of concentration (Figure 4.2).The linear regression equation was used for to determine sample concentration at spe-cific molar mass. The high values of 95% confidence limits are due to limited number ofconcentration levels of the linear regression (Figure 4.2).940 2 4 6 80100020003000400050006000y= 659.416x- 7.750R2= 0.999Intercept = - 7.750  13.955Slope = 659.416  3.35595% confidence limits after  Std 1 Linear fitAreaConcentration (mg/mL)(a)0 2 4 6 80100020003000400050006000y= 662.123x- 4.939R2 = 0.999Intercept = - 4.939  11.901Slope = 662.123  2.86295% confidence limits after  Std3 Linear fitAreaConcentration (mg/mL)(b)0 1 2 3 4 5 601000200030004000y= 642.047x+ 1.429R2 = 0.999Intercept = 1.429  19.263Slope = 642.047  6.24695% confidence limits after  Std 5 Linear fitAreaConcentration (mg/mL)(c)0 2 4 6 80100020003000400050006000y= 630.190x+7.472R2 = 0.999Intercept = 7.472  10.893Slope = 630.190  2.62095% confidence limits after  Std 7 Linear fitAreaConcentration (mg/mL)(d)0 2 4 6 801000200030004000y= 476.388x + 110.721R2= 0.977Intercept = 110.721  347.259Slope = 476.388  83.50195% confidence limits after  Std 9 Linear fitAreaConcentration (mg/mL)(e)Figure 4.2: Concentration calibration curve of (a) Std 1, (b) Std 3, (c) Std 5, (d) Std 7,and (e) Std 9.95A SEC chromatograph of calibration standard with varying concentration is shownin Figure 4.3a, the calculation range division is labelled. Smaller intervals using morecalibration standards could improve the accuracy of concentration calculations. The mo-lar mass of hemicellulose oligomer process samples were calculated first and then theoligomer concentration was calculated.40 45 50 55 6005001000150020002500300035004000Std9Std7Std5Std3RI signal (mV)Retention time (min)43.94 48.54 53.21 57.6Std1(a)30 40 50 600200400600800RI signal (mV)Retention time (min) 14.19 min 23.19 min 41.19 min  74.19 min(b)Figure 4.3: SEC chromatograph of RI detector signal of (a) calibration standard withvarying concentration of Std 1, Std 3, Std 5, Std 7, and Std 9 (b) hydrolysate at180 oC for varying residence time.Raw data was extracted from Empower2 software and plotted in OriginLab2016. Oneexample of hydrolysates produced at 180 oC with varying residence time is provided inFigure 4.3b. With increasing residence time, the peaks shifted to low molar mass (long re-tention time), indicating greater amounts of small molar mass oligomers being generated.Retention time for specific DP was calculated from the molar mass calibration curve (Ta-ble 4.1). The chromatogram area for a molar mass interval was integrated with OriginLaband interpreted using the corresponding concentration calibration curve. This two-stepprocess provided oligomer molar mass distribution and concentration simultaneously.964.4 Results and Discussion4.4.1 Molar mass distribution analysisThe average molar mass of soluble oligomers can be reported as the number-averge(Mn,n = 0), weight average (Mw,n = 1) or Z-average (Mz,n = 2) according to Equa-tion 4.1.M = ∑Ni Mn+1∑Ni Mn(4.1)All three metrics at 160-200 oC are plotted in Figure 4.4. At 180-200 oC, Mw initiallydropped rapidly but then plateaued at approximately 0.97-1.5 kDa. This result agreeswith a previous study in which oligomers recovered at temperatures above 200 oC hada low average molar mass varying from 0.3-1.5 kDa (Borrega et al. [23]). Leppa¨nen et al.concluded that 170-180 oC is the optimal range to isolate high molar mass polysaccharidesfrom spruce during autohydrolysis (Leppa¨nen et al. [148]). Other authors (Rissanen et al.[227, 228], Song et al. [273]) have also reported that temperatures of 160 oC or lower favourhigher molar mass oligomers. In this work, the maximum weight-average molar masswas obtained by hydrolysis approximately 8.7 kDa at 140 oC for 6.7 minutes.From Figure 4.4, Mz is initially large and then drops rapidly. Mw exhibits a similartrend, although it is initially less than Mz. As Mz and Mw are sensitive to the presenceof large molar mass oligomers, this indicates that large oligomers form at the start ofhydrolysis and then rapidly depolymerize. As reaction severity increases, Mn, Mw andMz converged indicating small molar mass oligomers are the dominant form of solublehemicellulose products in the system. Therefore, hemicellulose hydrolysis occurs via se-ries reactions to produce large soluble oligomers and then small ones. This mechanism issupported by previous research(Jacobsen and Wyman [106, 107]).970 20 40 60 80 100 12002468101214 Mn Mw MzAverage molar mass (kDa)Pseudo-time (min)(a) 160oC0 10 20 30 40 50 60 70 80024681012(b) 180oC Mn Mw MzAverage molar mass (kDa)Pseudo-time (min)16 18 20 22 24 26 28 300123456(c) 200oC Mn Mw MzAverage molar mass (kDa)Pseudo-time (min)0 20 40 60 80 100 1201234567 160oC 180oC 200oCPolydispersity Mw/MnPseudo-time (min)(d)Figure 4.4: Number-average molar mass (Mn, square black dot), weight-average mo-lar mass (Mw, round red dot) and z-average molar mass (Mz, triangle blue dot)at: (a) 160oC (b) 180 oC (c) 200 oC (d) Polydispersity index (Mw/Mn) at 160oC(square black dot), 180 oC (round red dot) to 200 oC (triangle blue dot).Further insight into oligomer and monomer evolution can be gleaned from the ratio ofMw/Mn, or polydispersity index (PDI) (Figure 4.4d). The PDI represents polymer unifor-mity. For a homogeneous polymer solution, PDI = 1. At 180 oC, the initial PDI is 4.4 andthen decreases to approximately 2.1. Similar behaviour is observed at 160 oC: the initialPDI is 5.5 and then decreases to a final value of 3.0. At 200 oC, the PDI is approximately2.1-2.8 at all residence times. A high initial PDI indicates a set of heterogeneous oligomerproducts that are depolymerized to produce a set of oligomer products with increasinglyuniform molar mass.98Figure 4.4 provides guidance to the selection of chain scission mechanism for modeldevelopment (Emsley and Heywood [54], Xu et al. [325]). Random breaking lowers themean value of molar mass, but the PDI will remain constant for short residence times.Center-favored breaking reduces mean molar mass and rapidly unifies chain length thuscausing PDI to approach unity(Xu et al. [325]). End-wise bond breaking will generate amix of small and large polymers. The effect on mean molar mass is small and the PDIwill increase. Basedow et al. observed increasing trends of PDI and concluded a end-wise bond breaking mode when studying dextran. In this study, the rapid decrease inmean molar mass, especially of Mz and Mw, coupled with the decrease in PDI suggestshemicellulose hydrolysis occurs via center-favoured breaking.4.4.2 Molar mass influencing factorWeight-average molar mass is plotted as a function of final hydrolysate pH in Figure 4.5,and increases with increasing hydrolysate pH. For 2.5 <pH< 3.5, the molar mass is in-sensitive to pH variation and remains constant at approximately 1-4.5 kDa. Higher molarmass values greater than 6 kDa were achieved at 4 <pH< 5. Increasing temperatureincreases water dissociation as well as proton generation from the wood matrix thus de-creasing pH (Figure 4.5). Low pH, in turn, increases hemicellulose hydrolysis rate anddecreases the oligomer molar mass (Kang et al. [118], Kemppainen et al. [123], Larssonet al. [139]). This conclusion also agrees with Song et al.’s research that the hemicellu-lose DP increased with increasing pH, and a starting pH of approximately 4 efficientlyinhibited complete degradation of hemicellulose (Song et al. [274]). pH is a key factor forcontrolling the molar mass of hemicellulose oligomers.992.5 3.0 3.5 4.0 4.5 5.0 5.50246810 140 oC 160 oC 170 oC 180 oC 200 oCWeight-average molar mass (kDa)pH of hydrolysateFigure 4.5: Weight-average molar mass (Mw) of hemicellulose from 160oC to 200 oCas a function of hydrolysate pH.Softwood was subjected to equal severity hydrolysis conditions, severity factor of 4.21:200 oC for 18.39 minute and 180 oC for 74.19 minutes (Equation 3.2). The molar massdistributions from these conditions are presented in Figure 4.6. The hydrolysate producedat 200 oC contained more long-chain oligomers since a greater fraction of material hadshort retention times.30 40 50 600200400600800(a)RI signal (mV)Rentention time (min) 200C 18.39 min pH 3.2  180C 74.19 min pH 3.030 40 50 600200400600800RI signal at 180oC (mV)Retention time (min) 14.19 min pH 3.3 23.19 min pH 3.2 41.19 min pH 3.1 74.19 min pH 3.0(b)Figure 4.6: Chromatogram of hydrlysate as represented by SEC coupled with RI de-tector. (a) Hydrolysis was conducted at 180 oC for 74.19 min and at 200 oC for18.39 min; these conditions have equal severity as calculated by Equation 3.2.(b) Hydrolysis was conducted at 180 oC and residence time was varied.100The influence of time on molar mass distribution was determined by conducting hy-drolysis at 180 oC for 14.19 to 74.19 minutes (Figure 4.6b). At 180 oC, short residence timesproduced large hemicellulose oligomers as evidence by the shift in distribution to shortresidence times. Monomers elute at approximately 60 retention minutes. The low molarmass peak around 60 minutes retention time was obvious that larger peak area demon-strated higher concentration of molecules in that molar mass: 74.19 minutes hydrolysissample held the maximum peak area and 14.19 minute showed the minimum area.The pH effect on molar mass distribution was also demonstrated in Figure 4.6. Forsame severity or at same temperature, low pH hydrolysate showed greater peak area atlow molar mass retention time, and no large molar mass oligomer peak was observed atpH 3.0.4.5 ConclusionSoftwood hydrolysis and the resulting hemicellulose molar mass distribution was in-vestigated by size exclusion chromatography. The two-dimensional calibration methodenables measurement of oligomer molar mass and concentration simultaneously. Thisapproach enables detection of broad-distributed peaks, especially for wood hydrolysateanalysis.From average molar mass and PDI analysis, the soluble hemicellulose oligomers tendto break at the middle range of the chain to produce smaller molar mass hemicellulose.Hydrolysis time and temperature determined the hemicellulose molar mass distribu-tion. Long residence time reduced the concentration of high molar mass oligomers. Thefinal average molar mass was related to the hydrolysate pH. High molar mass oligomerswere produced when final hydrolysate pH was greater than 4.1014.6 Future Work1. More hemicellulose structurally relevant calibration standards could represent SECmolar mass separation more accurately.2. More calibration standards at a a greater number of concentration levels could im-prove concentration calibration accuracy.3. Lower hydrolysis temperature experiments are worthy of further investigation.4. Dissolve lignin during hydrolysis could influence the molar mass detection andmodel results, separation of lignin before SEC could be applied.5. The SEC could be coupled with RI and UV detectors simultaneously to assess ligninlinkages to carbohydrates. However, caution must be exercised when interpretingresults from hydrolysate produced at high temperatures due to the potential forpseudo lignin.6. Distinguishing lignin, carbohydrates and LCC remains a challenge for analysis ofhydrolysates. These differences will have implications for future models of ligno-cellulose hydrolysis.102Chapter 5Kinetics of Softwood HemicelluloseHydrolysis by Population Balance Model5.1 Introduction5.1.1 MotivationKinetic model of biomass hydrolysis was developed to understand the yield evolutionwith treatment severity and offer physical insights of the process to better control andoptimize hydrolysis up-scale design. Several types of model are discussed in this chapter,and a population balance model of softwood hydrolysis with the corresponding hemicel-lulose molar mass evolution will be proposed.5.1.2 Classic carbohydrates hydrolysis modelThere are two main categories of kinetic models for carbohydrate hydrolysis: severityfactor models and pseudo-homogeneous kinetic models.The severity factor model was discussed in Section 3.1.4, and pseudo-time calculationwas discussed in Section 3.4.2.Saeman studied hydrolysis of Douglas fir, and proposed the first cellulose hydrolysis103kinetic model using first-order reactions in series with rate constants having Arrheniustype temperature dependence (Equation 5.1). This model has been the basis of manysubsequent pseudo-homogenuous hydrolysis models.Cellulose ka−→ Reducing sugar kb−→ Sugar degradation products (5.1)The next development was the introduction of fast and slow reacting hemicellulosefractions as a result of the dramatic drop in rate after approximately 70% conversion (Ja-cobsen and Wyman [106], Kobayashi and Sakai [131]) and many others have used thisapproach (Conner [39], Garrote et al. [71], Maloney et al. [180], Root [234]). The bi-phasichydrolysis model had been frequently employed to explain the high reactivity of a por-tion of hemicellulose (Borrega et al. [23], Chen et al. [30], Greenwood et al. [80], Visuriet al. [310]). However, there is no physical or chemical standard to distinguish betweenthe two fractions.[Fast reaction xylan k f−−−−−−−−−−−−→Slow reaction xylan ks] →Oligomers k1−→ Xylose k2−→ Degradation products (5.2)Identification of soluble xylooligomers led to the incorporation of intermediate oligomers(Equation 5.2, Chen et al. [30]). Garrote et al. proposed a similar pseudo-homogeneouskinetic model with oligomer intermediate of high and low molar mass for the study ofEucalyptus wood hydrolysis. This differentiation of oligomer molar mass improves themodel fit to experimental data (Gonza´lez-Mun˜oz et al. [77]). Direct formation of xylosefrom sugar maple wood was added to the pseudo-homogeneous model to better describemonomer production (Mittal et al. [187]). Intermediates must be better characterized inorder to more deeply understand hydrolysis.SEC development enabled the characterization of oligomer intermediates and kinetic104models tracking the evolution of molar mass have been published in the literature. InBasedow et al.’s study of acid hydrolysis of dextran, the rate constant was proportional tothe oligomer molar mass raised to the power of 23 . Xu et al. assumed the hydrolysis ratewas a function of total number of bonds, and proposed a rate constant composed of bothtemperature independent and temperature dependent constants (Table 5.1). For Basedowet al. and Xu et al., hydrolysis of intermediates does not address oligomer formation fromlarger oligomers. Information on the formation of oligomers from the insoluble woodmatrix and largest soluble oligomers is missing. Generally, the pseudo-homogeneousmodels provide limited insight into the evolution of intermediates.5.1.3 Population balance model5.1.3.1 Origins as a comminution modelA population balance style model provides the opportunity to describe evolution of oligomersize. The population balance concept originates from comminution model of size reduc-tion process in the milling industry and treats the grinding mill as an analogue to a chem-ical reactor, in which large particles react to smaller particles (Austin et al. [14], Epstein[55, 56]). The size reduction process of solids has been studied by many scholars (Austin[12], Austin and Luckie [13], Chimwani et al. [34], Klimpel and Austin [130], Koka andTrass [132], Meyer et al. [183], Olson et al. [200], Reid [219]).Epstein’s statistical model established the concepts of breakage mechanisms and abreakage process of solids (Epstein [55, 56]). The dominant equation relies on mass bal-ance. The breakage process depends on two functions: Si, selection function which de-scribes the rate of breakage of a particle of size i and Bi,j, which is the cumulative distri-bution, by weight, of particles size i < j arising from the breakage of a particle of size j.It can be proven that the distribution function Si after n steps in the breakage process isasymptotically logarithmic-normal (Epstein [55, 56]).105Klimpel and Austin defined an equation for batch grinding using a first-order law(Equation 5.3), similar to classic hemicellulose hydrolysis kinetics (Austin [12], Klimpeland Austin [130]).dWidt= −SiWi (5.3)where Wi is the weight fraction of material in a closed size interval i.However, a complete mass balance must include material entering the size interval asa result of breakage of larger particles (Equation 5.4). The accumulation of oligomers ininterval i is the difference between the sum of material entering and the material exitingin time dt.dWidt= −SiWi +i−1∑j=1i>1bi,jSjWj (5.4)where Wi is the weight fraction of material in a closed size interval i. Si is selection func-tion which describes the rate of breakage of a particle of size i. bi,j is breakage functiondescribing fraction of material broken out of size interval j and entering interval i.The selection function is a simple power function of particle size xi (Equation 5.5,Klimpel and Austin [130]). This has not been adequately explained on a theoretical basis,but it has been repeatedly demonstrated through experimentation (Austin and Luckie[13], Austin et al. [14], Koka and Trass [132]).Si = a(xi)α (5.5)The value of α is a positive number, normally in the range 0.5 to 1.5, and is character-106istic of material while the value of a varies with mill conditions (Austin et al. [14]). Theunits of a will vary for different values of xi. When xi is dimensionless, a has dimensionsof time−1.The population balance model has since been applied to describe pulp and paper fibrelength reduction, depolymerization, granulation, dissolution (Meyer et al. [183], Olsonet al. [200], Ramkrishna and Singh [217]), and carbohydrate hydrolysis, oxidation, soni-cation and enzymatic treatment (Ahmad et al. [4, 5], Greenwood et al. [80], Kusema et al.[136], Lebaz et al. [141], Tayal and Khan [286], Visuri et al. [310]). The assumed mode ofbond breaking is central to final model and varies by study (Table 5.1).5.1.3.2 Population balance models of carbohydrateRandom bond breaking is the most common assumption for hemicellulose hydrolysis(Ahmad et al. [4], Greenwood et al. [80], Visuri et al. [310]). Visuri et al. assumed randombond breaking but added the phenomenon of fast and slow hydrolyzing bonds. This im-plies two kinds of glycosidic bonds and results in a selection function with competingslow and fast reactions. Greenwood et al. assumed that sugarcane bagasse contained afraction of xylan that did not undergo hydrolysis and added a parameter to describe this.Therefore, the hemicellulose bonds demonstrated varying rate to hydrolysis, making ran-dom scission an inappropriate assumption for all bonds. In contrast, Basedow et al. [16]found that chain ends were more reactive creating a parabolic dependency of individualrate constants. Since bond breakage is related to oligomer polydispersity, further discus-sion was provided in Section 4.4.1.107Table 5.1: Carbohydrates hydrolysis by population balance model (k and h are hydrolysis rate constant; a, b are con-stant).Year 1978 2008 2012 2015 2015AuthorBasedow et al. Xu et al. Visuri et al. Greenwood et al. Ahmad et al.Substrate Dextran O-Acetyl Galactogluco-mannansGalactoglucomannan Sugarcane bagasse Eucalyptus kraft pulp celluloseHydrolysis condi-tion0.12 N sulfuricacid1 M HCl pH of 1-2 Hot water pH of 3.84.2 0.5% sulfuric acid Alkaline solution with oxygenTemperature (oC) 80 90, 70, 50, 37, and 25 150, 160, 170 110 - 170 50Model Pseudo-homogeneousPseudo-homogeneous Population balance model Population balancemodelPopulation balance modelBond breakingmodeEnd wise Random Random/fast-slow bondcompetitionRandom RandomOligomer charac-terizationNumber ofbondsNumber of bonds Number of bonds Oligomer concen-trationParticle concentrationRate equation Rate=k(DP)a Rate=kw0kw(T)Nt(DP− 1) N.A. N.A. N.A.Selection function N.A. N.A. S = h[H+](DP− 1)h1 is constant orh2 = akfast + (1− a)kslowS = k[H+] S1 = k1(DP− 1)S2 = k2(DP− 1)exp(−a2t)S3 = k3(DP− 1)b3S4 = k4(DP− 1)b4 exp(−a4t)108The dependence of rate constant or selection function on carbohydrate polymer sizetakes different forms in the literature (Table 5.1). Visuri et al. combined hydrogen ionconcentration with selection function to represent proton effects for hydrolysis at low pH.Ahmad et al. proposed a selection function with an exponential dependence on carbohy-drate DP making the rate constant a function of hemicellulose molar mass. Ahmad et al.compared several cases of selection function for kraft pulp cellulose ageing, and foundthe exponential form had the best agreement with experimental results. Comminutionmodel analysis has shown an exponential relation between the size of reactants and thesize reduction process, suggesting rate varies as reactant size evolves. Therefore, in thispaper, we correlate the selection function to molar mass and the breakage function to themolar mass distribution.Several studies (Table 5.1) applied number of bonds as an oligomer characterizationapproach in kinetic model (Basedow et al. [16], Visuri et al. [310], Xu et al. [325]). Thenumber of bonds was assumed to be related to DP, which is the ratio of total molar massto monomer molar mass. However, for wood hydrolysate containing hexose and pentoseoligomers, accurate DP calculation is difficult due to differences in monomer molar mass.SEC of wood hydrolysates produce chromatograms with broad, overlapping peaks andpoorly resolved baselines, as demonstrated in Section 4.3. Greenwood et al. used HPLC todetermine the concentration distribution of xylan oligomers, but was only able to quantifythose with a DP of 2-6. In this study, oligomers with a broad range of molar masses wereidentified and quantified. Concentration and molar mass calibration methods and linearcurves were provided in Section 4.3.1095.2 Goals and HypothesesThe current study provides insight into the evolution of oligomers during hemicellu-lose hydrolysis by addressing two major challenges. First, SEC is applied using narrowspan standard two-dimensional calibration to quantify oligomer concentration and molarmass. Second, a population balance model is developed for hydrolysis of mixed softwoodspecies to track the evolution of hemicellulose molar mass and identify treatment condi-tions to maximize oligomer yield at a targeted molar mass.This objective of this chapter is kinetic model development for softwood hemicellu-lose hydrolysis. Oligomer intermediate molar mass and concentration evolution will becombined in the model.The hypotheses of this chapter include:1. Softwood hemicellulose oligomers evolve with hydrolysis condition.2. Maximum oligomer intermediates concentration could be determined.3. The kinetic model could predict softwood hemicellulose molar mass and concentra-tion evolution.4. The model provides physical insights of hemicellulose hydrolysis process.1105.3 Experimental Design5.3.1 MaterialRefer to Section 3.3.1.5.3.2 Softwood hydrolysisBiomass solids and water were mixed with deionized water to generate 500 g slurry ina stainless steel pressurized reactor (Parr Instrument Company 4843). The solids consis-tency is 5wt%. pH was tested and recorded before and after hydrolysis. The reactor washeated using a coil jacket. Agitation at 180 r.p.m. was applied during the entire heatingand reaction process. Hydrolysis was conducted at four temperatures. Results collectedat 160oC, 180oC, and 200oC was used to estimate model parameters while data collectedat 170oC was used to verify the model’s fit. Conditions are summarized in Table 5.2. Ittook approximately 40-60 minutes to reach target reaction temperature therefore a resi-dence time of 0 minute (i.e. the end of heat-up period) was investigated at each tempera-ture. The reactor was quenched in ice water when the desired isothermal residence timeachieved. Heating and cooling times were recorded.Table 5.2: Softwood hydrolysis condition and use.Temperature (oC) Time (min) Use160 0-100 Model development170 0-9 Validation180 0-60 Model development200 0-9 Model development1115.3.3 Two-dimensional calibration molar mass analysisRefer to Section 4.3.5.3.4 Compositional analysisRefer to Section 3.3.3.5.4 Model Development5.4.1 Pseudo-time calculationRefer to Section 3.4.2.5.4.2 Molar mass distribution analysisRefer to Section 4.4.1.5.4.3 Population balance modelIn order to reduce the computational load, the molar mass distribution was divided into9 intervals (Figure 5.1): 7 intervals containing soluble oligomers, one interval of degrada-tion products and one interval of insoluble hemicellulose. Soluble oligomer intervals de-scribed oligomers with DP from 1 to 140 and each interval corresponded to a DP changeof 20 (Table 5.3). The DP range corresponds to the SEC molar mass calibration range.Any oligomers with DP>140 were assumed insoluble and allocated to interval 9, whileany products with a DP less than 1 were allocated to interval 1, degradation products.112Solids softwood hemicelluloseDP61-80DP121-140DP101-120DP81-100DP41-60DP21-40DP1-20Degradation productsPPPPPPPSSSSSSFigure 5.1: Mechanism of hemicellulose molar mass evolution in the range of 9 intervals including soluble oligomerwith DP1-140, insoluble softwood hemicellulose, and degradation products. P is a primary pathway to producesoluble hemicellulose from insoluble wood; S is a secondary pathway to produce soluble hemicellulose from aprimary oligomer.113Table 5.3: Molar mass and SEC retention time of corresponding DP.IntervalnumberDegree of poly-merizationMolar mass(Da)Retentiontime (min)1 <1 <173.1 >60.12 1 173.1 60.110 1568.9 51.620 3119.8 48.93 30 4670.7 47.340 6221.6 46.24 50 7772.5 45.360 9323.4 44.65 70 10874.3 44.080 12425.2 43.56 90 13976.1 43.1100 15527.0 42.77 110 17077.9 42.3120 18628.8 41.98 130 20179.7 41.6140 21730.6 41.39 >140 >21730.6 <41.3A two-stage hydrolysis pathway was proposed (Figure 5.1). The first stage assumesrandom bond breaking of insoluble hemicellulose within the wood matrix to producesoluble hemicellulose, named primary pathways (P) in Figure 5.1. The second stage de-scribes the cascading reduction of hemicellulose molar mass by center-favoured breaking(Section 4.4.1). S represents secondary pathways which originate from primary oligomers.114Oligomers with DP1-20 can degrade to generate degradation products, providing a sinkin the hydrolysis pathway.Softwood hemicellulose contains AGX and GGM (Fengel and Wegener [62]). The ratioof glucose to mannose in GGM has been reported to vary from 1:1 to 1:3 depending on theraw wood resource and species (Gı´rio et al. [74], Gullichsen and Fogelholm [83], Mimmset al. [185]). Determination methods using NSERC LAPs do not differentiate betweenglucose associated with hemicellulose and that associated with cellulose (Hames et al.[86], Sluiter et al. [265, 266]). In order to calculate the hemicellulose yield, it is necessary toattribute a fraction of measured glucose to hemicellulose. Cellulose is more difficult to behydrolyzed than hemicellulose due to cellulose’s crystalline nature (Fitzpatrick [64], Ren-ders et al. [222]) and it is likely that the amorphous region of cellulose is hydrolyzed first(Mok and Antal Jr [189]). As in past studies (Gı´rio et al. [74], Gullichsen and Fogelholm[83], Mimms et al. [185]), it was assumed that hemicellulose-derived glucose in the rawbiomass is equal to the mass of measured mannose. Under this assumption, 23.3-23.8%of glucose in the biomass (Table 5.4) was attributed to hemicellulose.115Table 5.4: Composition of softwood on o.d. biomass basis and average hemicellulose monomer molar mass. Note:hemicellulose content was corrected by soluble polysaccharides content from extraction.SoftwoodbatchExtractive(%)Sugar(%)Arabinan(%)Galactan(%)Glucan(%)HemicelluloseGlucan (%)Xylan(%)Mannan(%)Acid in-solubleresidue(%)Acidsolublelignin(%)Totalmassbalance(%)Hemicellulosemonomermolar mass(g·mol−1)Average hemi-cellulosemonomer molarmass (g·mol−1)2014 Mill 2014 average 3.92 63.03 1.49 3.43 42.30 9.84 5.96 9.84 27.45 4.35 98.74 172.84 173.09Standarddeviation0.53 3.28 0.13 0.17 3.13 0.18 0.87 0.42 1.15 2.322014 Mill 2015 average 3.33 65.76 1.45 2.77 45.13 10.75 5.67 10.75 26.67 5.24 101.01 173.34Standarddeviation0.19 1.30 0.05 0.09 1.07 0.36 0.54 1.25 1.95 1.70116The hemicellulose monomer molar mass was defined to be the weighted average ofhexose (180.156 g·mol−1) and pentose (150.13 g·mol−1); the weighting factor was deter-mined from the compositional analysis of softwood. The retention time for each molarmass interval was calculated from calibration logarithm curve and target molar mass asshown in Table 5.3. The calculated average hemicellulose monomer molar mass is 173.09g·mol−1 (g·mol−1 equivalent to Da). The corresponding interval division and calibrationstandard calculation range are demonstrated in Figure 5.2.35 40 45 50 55 60 650200400600800std1 std3std5std7std9Int9Int8Int7Int6Int5Int4Int3Int2RI signal (mV)Retention time (min) 14.19 min 23.19 min 41.19 min 74.19 min43.94 48.5453.21 57.660.148.946.244.643.542.74241.4Int1Figure 5.2: SEC chromatograph of RI detector signal with model interval division ofhydrolysate at 180 oC for varying residence time.After hydrolysis, the acid soluble lignin content of hydrolyzed solids varied between2.9-4.9% (Figure 3.16b). The mass fraction of acid soluble lignin in the raw biomass was4.35-5.24% (Table 5.4). The equivalent acid soluble lignin in hydrolysate is 0-2.8% of rawbiomass. This amount is negligible relative to hemicellulose 30.56-31.39% of raw biomass117(Table 5.4). Acid insoluble residue in hydrolyzed solids was appoximately constant com-pared to raw softwood composition (Figure 3.16a). It was concluded that lignin degrada-tion products were limited and therefore no lignin degradation reactions were includedin the model.The initial hemicellulose concentration, Ch,0, in softwood was determined by compo-sitional analysis. The residual solids hemicellulose concentration after hydrolysis, CR,was defined by Equation 5.6, where Ch,0 is initial hemicellulose concentration and CS isthe concentration of soluble hemicellulose as measured by SEC. Oligomer concentrationin each interval CS,i was normalized by the initial softwood hemicellulose concentration(Equation 5.7). This normalized concentration Ci is equivalent to percent yield.CR = Ch,0 − CS (5.6)Ci =CS,iCh,0(5.7)The following assumptions were made to develop the population balance model:1) Material loss follows a first-order hydrolysis law (Equation 5.8) that can be de-scribed by selection and breakage functions. Rate constants have Arrhenius temperaturedependence.2) Polymer does not break into more than two pieces per reaction.3) There is no condensation of polymers.4) The breakage functions, bi,j, are independent of time.5) The selection functions, Si, are independent of time and can be written as a functionof molar mass interval size (Equation 5.9).6) The internal mass transfer is negligible due to the small particle size and the externalmass transfer is negligible due to vigorous agitation within the reactor.118Equation 5.8 reflects the conservation of mass for soluble hemicellulose in a single in-terval: material loss by hydrolysis to generate smaller oligomers and material gain due toproducts generated from larger molar mass intervals. For solids dissolution, only mate-rial loss is considered while only material gain was considered for degradation products.Material loss was assumed to have first order dependence on concentration. The rateconstant of hydrolysis is described by selection function Si (Equation 5.9), which has anArrhenius temperature dependence. The fraction of material which exits interval j andenters interval i is described by the breakage function, bi,j. Bi,j is the cumulative weightfraction of material exiting from interval j. At constant j, Bi,j should equal 1 indicating thatthe summation of breakage probabilities from one interval is 1. Given these constraints,the mass balance for soluble hemicellulose in interval 36 i6 8 is:dCidtp= −SiCi +N=9∑j>ibi,jSjCj = −k1(i− 1)k2Ci +N=9∑j>ibi,jSjCj (5.8)where Ci is the hemicellulose oligomer concentration of interval i. Both i and j vary from1 to 9. Si or Sj is the selection function, or rate constant for the hydrolysis of molar massinterval i or j. N refers to the total number of intervals.For intervals 3 6 i 6 8, the selection function is assumed to depend on the intervalmolar mass (DPmax = 20(i − 1)) and two constants, k1 and k2, as in Equation 5.9. Theselection function’s dependence on molar mass interval size is determined by the valueand sign of k1 and k2.Si = k1(i− 1)k2 (5.9)The selection function for insoluble hemicellulose (i=9) is given by rate constant k3to reflect the associated phase transition (Equation 5.21). The selection function for the119production of degradation products (e.g HMF or furfural, i=1) from monomers (i=2) isgiven by rate constant k4 (Equation 5.13). There are a total of 4 unknown parametersassociated with selection functions: k1, k2, k3, k4.From previous discussion, it was determined that insoluble hemicellulose likely expe-riences random bond breaking. The probability that random breaking of insoluble hemi-cellulose will generate a soluble oligomer of size i was assigned to be 17 (Equation 5.10).This implies there is an equal opportunity for a primary oligomer generated from insolu-ble hemicellulose to enter one of the soluble oligomer intervals.bi,9,solid =1N− 2 =17(5.10)A soluble oligomer exiting interval j will enter interval i as a secondary oligomer. Fromthe previous analysis of molar mass evolution, it is assumed that oligomer of interval jbreaks at the mid-point. This behaviour is captured by periodic functions Equation 5.11and Equation 5.12. The complete breakage function distribution matrix is shown in Ta-ble 5.5.For soluble hemicellulose when j = odd and 1< j < 9:bi,j,soluble =0 if i 6= j+121 if i = j+12(5.11)For soluble hemicellulose when j = even and 2< j < 9:bi,j,soluble =0 if i 6= j+221 if i = j+22(5.12)120Table 5.5: Breakage function (bi,j) distribution matrix for hemicellulose interval 1(degradation products) to 9 (solids).j 9 8 7 6 5 4 3 2 1i Solid DP121-140 DP101-120 DP81-100 DP61-80 DP41-60 DP21-40 DP1-20 Degradation1 Degradation 0 02 DP1-20 1/7 1 03 DP21-40 1/7 1 1 04 DP41-60 1/7 1 1 05 DP61-80 1/7 1 06 DP81-100 1/7 07 DP101-120 1/7 08 DP121-140 1/7 09 Solid 0The full set of equations for the population balance model is provided in Equation 5.13to Equation 5.21.dC1dt= k4C2 (5.13)dC2dt= −k4C2 + 17k3C9 + k1(2)k2C3 (5.14)dC3dt= −k1(2)k2C3 + 17k3C9 + k1(3)k2C4 + k1(4)k2C5 (5.15)dC4dt= −k1(3)k2C4 + 17k3C9 + k1(5)k2C6 + k1(6)k2C7 (5.16)dC5dt= −k1(4)k2C5 + 17k3C9 + k1(7)k2C8 (5.17)dC6dt= −k1(5)k2C6 + 17k3C9 (5.18)dC7dt= −k1(6)k2C7 + 17k3C9 (5.19)dC8dt= −k1(7)k2C8 + 17k3C9 (5.20)dC9dt= −k3C9 (5.21)121The initial condition when tp = 0 is provided in Equation 5.22.C1 = 0; C2 = 0; C3 = 0; C4 = 0; C5 = 0; C6 = 0; C7 = 0; C8 = 0; C9 = 1(5.22)5.5 Results and Discussion5.5.1 Model fitting and optimizationThe population balance model agreed with the experimental results very well for most in-tervals at 160 oC, 180 oC and 200 oC (Figure 5.3, Figure 5.4 and Figure 5.5) with parameterregression sum of squares less than 3.8E-04 (Table 5.7).Similar trends are visible at all three temperatures (Figure 5.3, Figure 5.4, Figure 5.5).Softwood hemicellulose disappears monotonically reflecting Equation 5.21. For intervalsfrom DP21-40 to DP121-140, oligomers initially acculmulate, reach a maximum yield, andthen decrease. The maximum yield achieved in each interval decreases with increasingmolar mass. For example, the maximum yield in interval DP21-40 was 0.1 (dimension-less) while the maximum yield in interval DP121-140 was 0.01 (dimensionless) at 160 oC(Figure 5.3). DP1-20 oligomers and degradation products accumulate monotonically in-dicating the dominance of accumulation terms in Equation 5.13 and Equation 5.14. Ifhydrolysis times were extended at 160 oC, the model predicts that the yield in intervalDP1-20 will reach a maximum after 400 minutes and diminish to zero after 50,000 min-utes. Similar parabolic behaviour on similarly long time scales is predicted for intervalDP1-20 at 180 oC and 200 oC. Extrapolating the models to times significantly longer thanthe experimental data should be done cautiously.1220 20 40 60 80 100 1200.0000.0050.0100.0150.020 Degradation Model ExperimentalDegradation yieldPseudo-time (min)0 20 40 60 80 100 120-0.10.00.10.20.30.40.50.60.70.8 DP=1-20 Model ExperimentalDP=1-20 yieldPseudo-time (min)0 20 40 60 80 100 1200.000.050.100.150.20 DP=21-40  Model Experimental DP=41-60  Model ExperimentalDP=21-60 yieldPseudo-time (min)0 20 40 60 80 100 1200.000.010.020.030.040.05 DP=61-80 Model Experimental DP=81-100 Model ExperimentalDP=61-100 yieldPseudo-time (min)0 20 40 60 80 100 1200.000.010.02 DP=101-120 Model Experimental DP=121-140 Model ExperimentalDP=101-140 yieldPseudo-time (min)0 20 40 60 80 100 1200.00.20.40.60.81.01.2Solids yieldPseudo-time (min) Solids Model ExperimentalFigure 5.3: Comparison of model fit (black and red line) to experimental data (blackand red dot) at 160 oC for softwood hemicellulose including soluble oligomerof DP 1-140, solids hemicellulose and degradation products.Given the mechanism pathway proposed in Figure 5.1, oligomers with a DP greaterthan 80 (66 i6 8) can only be generated as primary oligomers. The maximum yields ofthese intervals range from 0.010 to 0.0117 (dimensionless, Figure 5.3). Further at t=10.16min for all soluble oligomer intervals (2 6 i 6 8), yields varied over a small range 0.009to 0.025 (dimensionless, Figure 5.3). These observations indicate that the rate of oligomergeneration from insoluble hemicellulose of solids is comparable for short residence timeand supports the assumption of equal probability of bond breakage. Kusema et al.’s acidhydrolysis of galactoglucomannan study likewise reported that initial oligomer forma-tion rates were approximately equal. It can further be inferred that the reactivity of insol-123uble hemicellulose is uniform, which is in contrast to past studies which assumed slow-and fast-reacting fractions (Garrote et al. [71], Greenwood et al. [80], Jacobsen and Wyman[106], Kobayashi and Sakai [131]).0 20 40 60 80 1000.000.010.020.030.04 Degradation  Model ExperimentalDegradation yieldPseudo-time (min)0 20 40 60 80 1000.00.20.40.60.81.01.2 DP=1-20  Model ExperimentalDP=1-20 yieldPseudo-time (min)0 20 40 60 80 1000.000.050.100.150.20 DP=21-40  Model Experimental DP=41-60  Model ExperimentalDP=21-60 yieldPseudo-time (min)0 20 40 60 80 1000.000.010.020.030.040.05 DP=61-80  Model Experimental DP=81-100  Model ExperimentalDP=61-100 yieldPseudo-time (min)0 20 40 60 80 1000.0000.0050.0100.015 DP=101-120  Model Experimental DP=121-140  Model ExperimentalDP=101-140 yieldPseudo-time (min)0 20 40 60 80 1000.00.20.40.60.81.01.2Solids yieldPseudo-time (min) Solids Model ExperimentalFigure 5.4: Comparison of model fit (black and red line) to experimental data (blackand red dot) at 180 oC for softwood hemicellulose including soluble oligomerof DP 1-140, solids hemicellulose and degradation products.Yields of oligomers with DP 6 80 were greater than yields of large oligomers. Theincrease in yield increased rapidly as molar mass decreases indicating a high rate of pro-duction, likely due to the existence of multiple source intervals. The low molar massintervals (2 6 i 6 5) contain primary and secondary oligomers. If the primary oligomercontent is similar to DP > 80 (66 i6 8), secondary oligomers (i.e. those generated fromlarge, soluble hemicellulose oligomers) account for the majority of material in the low DP124intervals.0 5 10 15 20 25 30 35 400.0000.0050.0100.0150.020 Degradation  Model ExperimentalDegradation yieldPseudo-time (min)0 5 10 15 20 25 30 35 400.00.20.40.60.81.0 DP=1-20  Model ExperimentalDP=1-20 yieldPseudo-time (min)0 5 10 15 20 25 30 35 400.000.050.100.150.20 DP=21-40  Model Experimental DP=41-60  Model ExperimentalDP=21-60 yieldPseudo-time (min)0 5 10 15 20 25 30 35 400.000.010.020.030.04 DP=61-80  Model Experimental DP=81-100  Model ExperimentalDP=61-100 yieldPseudo-time (min)0 5 10 15 20 25 30 35 400.0000.0050.010 DP=101-120  Model Experimental DP=121-140  Model ExperimentalDP=101-140 yieldPseudo-time (min)0 5 10 15 20 25 30 35 400.000.250.500.751.00Solids yieldPseudo-time (min) Solids  Model ExperimentalFigure 5.5: Comparison of model fit (black and red line) to experimental data (blackand red dot) at 200 oC for softwood hemicellulose including soluble oligomerof DP 1-140, solids hemicellulose and degradation products.The first derivative of the yield curves in Figure 5.3, Figure 5.4, and Figure 5.5 wasused to predict the maximum yield by interval and the time to achieve this maximumby hydrolysis temperature. These predictions are plotted in Figure 5.6. If high molarmass oligomers are targeted, residence time must be limited. Time to maximum yielddecreases as hydrolysis temperature increases. Maximum yields of large oligomers wereachieved at 180 oC due to incomplete hydrolysis at 160 oC and rapid depolymerization at200 oC. For example, the total yield of oligomer with DP80-140 was 3.26% after 20 min at160 oC, 3.73% after 6 minutes at 180 oC, and 2.60% after 3 minutes at 200 oC. If smaller125oligomers are desired, such as DP21-40, 200 oC is the preferred hydrolysis temperature ashigher yields are achieved with reduced residence time. The maximum yield of DP21-40,13.09%, was achieved at 200 oC after 8 minutes but the corresponding maximum yield at160 oC was 9.63% after 50 minutes.DP21-40 DP41-60 DP61-80 DP81-100 DP101-120 DP121-1401015202530354045505560 Time YieldOligomer intervalTreatment time at 160oC (min)Primary oligomer(a)0.000.020.040.060.080.100.12Predicted yield by modelDP21-40 DP41-60 DP61-80 DP81-100 DP101-120 DP121-14005101520253035 Time YieldOligomer intervalTreatment time at 180oC (min)Primary oligomer(b)0.000.020.040.060.080.100.12Predicted yield by modelDP21-40 DP41-60 DP61-80 DP81-100 DP101-120 DP121-1400246810 Time YieldOligomer intervalTreatment time at 200oC (min)Primary oligomer(c)0.000.020.040.060.080.100.120.14Predicted yield by modelFigure 5.6: Maximum yield (red line and round hollow dot) and corresponding treat-ment time (black line and square solid dot) from model prediction of oligomerwith DP 21-140 at (a) 160 oC (b) 180 oC and (c) 200 oC.The kinetic model was validated by experiments conducted at 170 oC. The rate con-stant at 170oC was calculated from Equation 5.26, solved by Matlab ode45, and comparedwith experimental results in Figure 5.7. The general trends agree with model prediction.1260 10 20 300.0000.0050.010 Degradation Model ExperimentalDegradation yieldPseudo-time (min)0 10 20 300.00.10.20.30.40.50.6 DP=1-20 Model ExperimentalDP=1-20 yieldPseudo-time (min)0 10 20 300.000.050.100.15 DP=21-40 ModelExperimental DP=41-60 Model ExperimentalDP=21-60 yieldPseudo-time (min)0 10 20 300.000.010.020.030.040.050.06 DP=81-100 Model Experimental DP=101-120 Model ExperimentalDP=61-100 yieldPseudo-time (min)0 10 20 300.0000.0050.0100.0150.0200.0250.030 DP=101-120 Model Experimental DP=121-140  Model ExperimentalDP=101-140 yieldPseudo-time (min)0 10 20 300.00.20.40.60.81.01.2Solids yieldPseudo-time (min) Model ExperimentalFigure 5.7: Validation of model fit to experimental data at 170 oC.It is more difficult to recover high molar mass oligomers than low molar mass prod-ucts. This challenge is related to the natural DP of hemicellulose within wood. The maxi-mum molar mass of GGM extracted with good solubility in water is approximately 21-35kDa, which corresponds to DP of 130-525 (Kishani et al. [128]). However, multi-filtrationand ultrafiltration processes are required to recover specific molar mass of oligomers (Kis-hani et al. [128]). Our model offers one approach to control hydrolysis conditions in orderto tune hydrolysis products and targets for specific molar mass from extraction itself.The selectivity of oligomer with specific molar mass interval (i) was calculated asEquation 5.23. The results of three temperature of DP 1-140 were plotted in Figure 5.8.127Sei =Ci∑82 Ci(5.23)0 20 40 60 80 100 1200.20.40.60.81.01.2 160 (oC) 180 (oC) 200 (oC)Selectivity DP 1-20Pseudo-time (min)0 20 40 60 80 100 1200.00.10.20.30.4 160 (oC) 180 (oC) 200 (oC)Selectivity DP 21-40Pseudo-time (min)0 20 40 60 80 100 1200.000.050.100.150.200.25Selectivity DP 41-60Pseudo-time (min)0 20 40 60 80 100 1200.000.050.100.150.20Selectivity DP 61-80Pseudo-time (min)0 20 40 60 80 100 1200.000.050.100.15Selectivity DP 81-100Pseudo-time (min)0 20 40 60 80 100 1200.000.050.100.15Selectivity DP 101-120Pseudo-time (min)0 20 40 60 80 100 1200.000.020.040.060.080.100.120.14Selectivity DP 121-140Pseudo-time (min)Figure 5.8: Optimization of treatment time and predicted yield of oligomer by modelat each interval at 160 oC (black line), 180 oC (red dot line), and 200 oC (bluedash line).Selectivity demonstrated a similar trend as yield for soluble intervals at 160-200 oC.For small hemicellulose with DP1-20, selectivity monotonicaly increases with residencetime. At all three temperatures, the value approaches 1 with increasing residence time,suggesting more depolymerization products arriving from higher DP intervals. DP21-80demonstrates maximum selectivity at 200 oC. The corresponding peak time decreases128with rising temperature. For instance, DP21-40 oligomer selectivity maximizes approxi-mately 0.268 at 160 oC after 25 minutes. The maximum value was 0.28 at 180 oC after 8minutes and 0.325 at 200 oC after 4 minutes. The trend is consistent with the previousdiscussion of maximum yield that small molar mass hemicellulose are favoured at hightemperature with short residence time. For large molar mass hemicellulose of DP81-140,maximum selectivity was quickly achieved and then declined with increasing residencetime.5.5.2 Parameter determinationThe parameters k1 to k4 were estimated by minimization the sum of squares (SS) basedon Marquardt method and the ordinary differential equations were solved by Matlabfunction ode23 (Section A.4, Constantinides et al. [40]). The objective function for ss isgiven by Equation 5.24.ss =9∑i=1wi(Ci∗ − Ci)′(Ci∗ − Ci) (5.24)where wi is the weighting factor for interval i, Ci∗ is the experimental concentration andCi is the calculated concentration using estimated parameter k for interval i.The weighting factor (wi) given by Equation 5.25 is applied to achieve a unbiasedweighted sum of squares (Constantinides et al. [40]). The weighting factor is assumed tobe 1 as initial guess and optimized by minimization the sum of squares.wi =1σi21∑9i=1 nj[∑9i=1∑njl=11σj2](5.25)where σi is the variance for interval i, σj is the variance for data set j (i.e. the experi-mental points within interval i), and nj is the number of experimental data points for data129set j (i.e. the number of time steps at reaction temperature).The iteration process contains two loops (Constantinides et al. [40]). The inner loopiterates parameters while keeping the weighting factor constant. The outer loop reevalu-ates weighting factor and repeats the inner loop until the ∆ss of iteration step is less than1.0E-06.For Marquardt method, the Marquardt parameter value (λ) is adjusted according tothe ss value. The initial value of λ is 1000, and reduces by a factor of 4 if ss decreases;increases by a factor of 2 if ss increases or remains constant (Section A.4, Constantinideset al. [40]). The weighting factor for each interval and λ value of last iteration are summa-rized in Table 5.6. At 160 oC to 200 oC, the λ value of last iteration was 250.Oven dried fibre weight (g) (xd) Initial o.d. pulp samples (g) (xi) Klason lignin contentof fibre (g) (xk) Residual lignin content of pulp fibre after post-hydrolysis (g) (xresidual)Acid soluble lignin content of pulp fibre (g) (xs) Alpha-cellulose content in pulp fibre(%) (xα) Particle size of interval i (xi) Order of selection function dependence on mo-lar mass (α) Dimensionless length (η) Dimensionless concentration (θ) Substrate surfacecoverage (θs) Marquardt parameter value (λ) Variance for interval i (σi) Dimensionlesstime (τ) Wood chips diffusion domain (Ω1) Bulk solution diffusion domain (Ω2) Inti-tal (0) Hemicellulose (h) Index (i) Index (j) Total number of intervals (n) Pseudo (p) Resid-ual (r) Soluble (s) Soluble oligomers in the aqueous phase (soluble) Insoluble hemicellu-lose in wood matrix (solid) Interface (*) Bulk solution (∞) Wood chip (chip) Bulk solu-tion (solution)130Table 5.6: The Marquardt parameter (λ) and corresponding weighting factors (wi) for interval 1 to 9 of last regressioniteration.Temperature(oC)λ of last itera-tionw1 w2 w3 w4 w5 w6 w7 w8 w9160 250 8.10E+00 1.44E-04 5.31E-03 8.75E-02 8.82E-02 2.84E-02 3.28E-01 3.67E-01 1.58E-04180 250 2.61E+00 2.01E-03 1.71E-01 2.45E-01 3.52E-01 3.16E-01 1.63E+00 3.67E+00 1.43E-03200 250 9.13E-03 5.44E-04 3.79E-04 4.93E-03 7.27E-02 1.32E+00 4.98E+00 2.62E+00 1.19E-03131The regressed parameters k1 to k4 are shown in Table 5.7. The selection function foreach soluble oligomer interval was calculated from parameters k1 and k2 and molar mass(Equation 5.9) and summarized in Table 5.8.Table 5.7: Parameter regression results of population balance model. Note: 95% con-fidence limits after ±.Temperature (oC) k1 (min−1) k2 (dimen-sionless)k3 (min−1) k4 (min−1) Sum ofsquares160 4.46E-02 5.59E-01 1.18E-02 1.28E-04 3.65E-05Standard deviation 5.26E-03 7.97E-02 5.39E-04 1.10E-0595% Confidence limits ± 1.06E-02 1.60E-01 1.08E-03 2.22E-05180 1.19E-01 6.75E-01 4.57E-02 2.69E-04 3.80E-04Standard deviation 1.21E-02 9.11E-02 2.81E-03 4.03E-0595% Confidence limits ± 2.45E-02 1.84E-01 5.68E-03 8.14E-05200 1.44E-01 1.15E+00 8.42E-02 4.95E-04 9.65E-06Standard deviation 3.02E-02 1.26E-01 2.48E-03 3.32E-0495% Confidence limits ± 6.16E-02 2.57E-01 5.05E-03 6.76E-04In previous studies, k2 was found to vary from 0.5 to 1.5 representing the dependenceof polymer reactivity on molar mass and structure, while k1 is related to reaction condition(Austin et al. [14], Basedow et al. [16]).Ahmad et al. assumed the same selection function form as Equation 5.9 when study-ing ageing of Eucalyptus kraft pulp cellulose and obtained k2 value of 0.89 (Table 5.1).Basedow et al. proposed a similar form of hydrolysis rate equation for dextran, and deter-mined k2 to be 23 (Table 5.1). Dextran is a hyperbranched glucose polymer, while celluloseis a linear glucose polymer. It may be that linear carbohydrates will have a higher orderof dependence on molar mass during hydrolysis compared to branched carbohydrates.In this work, k2 ranged from 0.56 to 1.15. Branched hemicellulose and linear cellulose arehydrolyzed in this study to varying degrees thus yielding a range of k2 values.132Table 5.8: Selection function results of interval 3 to 8. Note: 95% confidence limitsafter ±.Temperature (oC) S3 (min−1) S4 (min−1) S5 (min−1) S6 (min−1) S7 (min−1) S8 (min−1)160 0.07 0.08 0.10 0.11 0.12 0.1395% Confidencelimits ±0.02 0.02 0.03 0.04 0.05 0.05180 0.19 0.25 0.30 0.35 0.40 0.4495% Confidencelimits ±0.05 0.07 0.10 0.13 0.15 0.18200 0.32 0.51 0.71 0.91 1.12 1.3495% Confidencelimits ±0.15 0.26 0.39 0.54 0.71 0.88The change in k2 with temperature is likely a result of the changes in the compositionof hemicellulose hydrolyzed (Figure 5.9). Arabinan, xylan and galactan recovery was lowat 180 oC and 200 oC. Longer residence time intensified this trend. Arabinan, xylan andgalactan are the most reactive softwood hemicellulose species. Pu et al. and Leppa¨nenet al. reported arabinose and xylose from mixed softwood were primarily detected asmonomers after hydrolysis at 170 oC and 180 oC. They also reported that high severityhydrolysis degraded arabinose, xylose and galactose into furfural, HMF and other lightorganics (Leppa¨nen et al. [148], Pu et al. [216]). Arabinan and galactan are side chains ofAGX and GGM (Fengel and Wegener [62]). As hydrolysis proceeds, side chain removalwill produce more linear polymer structure of hemicellulose. Given the previous discus-sion, it is logical that k2 increase with increasing temperature.1330 20 40 60 80 100 1200.00.20.40.60.81.01.21.4 160oC 170oC 180oC 200oCGalactan mass recoveryPseudo-time (min)(a)0 20 40 60 80 100 1200.00.20.40.60.81.01.21.4 160oC 170oC 180oC 200oCArabinan mass recoveryPseudo-time (min)(b)0 20 40 60 80 100 1200.00.20.40.60.81.01.21.4 160oC 170oC 180oC 200oCXylan mass recoveryPseudo-time (min)(c)0 20 40 60 80 100 1200.00.20.40.60.81.01.21.4(d) 160oC 170oC 180oC 200oCGlucan mass recoveryPseudo-time (min)0 20 40 60 80 100 1200.00.20.40.60.81.01.21.4 160oC 170oC 180oC 200oCMannan mass recoveryPseudo-time (min)(e)Figure 5.9: Hemicellulose component mass recovery from hydrolysate and hy-drolyzed solids at 160 oC to 200 oC (a) galactan (b) arabinan (c) xylan (d) glucan(e) mannan.134From Table 5.8, temperature dependence of reaction rate is primarily captured by k1and therefore this parameter and Si increases as temperature increases from 160 oC to 200oC. Since selection function depends exponentially on molar mass, the rate of materialdisappearance also increases with molar mass. Hydrolysis rate of linear carbohydrates athigh temperature is more dependent on molar mass compared to branched carbohydratesat low temperature.Parameter k3 represents the solid hemicellulose solubilization rate constant, and k4 isthe degradation products generation rate constant (Table 5.7). Both parameters increasewith temperature. Higher temperature facilitates acid group dissolution into aqueousphase and the water protonation process. Increased proton concentration will acceleratethe rate of glycosidic bond breaking in solids and oligomers.5.5.3 Activation energySelection functions S3 to S8 and parameters k3 and k4 were assumed to follow Arrhenius-temperature dependence. Activation energy and pre-exponential factors were determinedby linear regression (Figure A.2) of the classic Arrhenius equation as Equation 5.26, whichis written for selection function S3. The calculated values for each interval are shown inTable 5.9. The high values of 95% confidence limits are due to small number of tempera-ture levels used to calculate activation energy (160 oC, 180 oC, and 200 oC).lnS3 = lnA− EaRT (5.26)Pre-exponential factor (L·mol−1 ·min−1) (a) Activation energy (kJ ·mol−1) (ea) Areabeneath the heating and cooling temperature-time profiles (oC ·min) (ap) Interfacial areaof softwood chip (m2) (a∗) Pre-exponential factor of selection function (min−1) (a) Cumu-lative breakage function (bi,j) Breakage function (bi,j) Biot number (bi)135Table 5.9: Activation energy and pre-exponential factor for each soluble hemicellu-lose oligomers and solids interval. Note: 95% confidence limits after ±.Interval DP1-20 DP21-40 DP41-60 DP61-80 DP81-100 DP101-120 DP121-140 SolidsEa (kJ ·mol−1) 57.5 67.6 77.6 84.7 90.3 94.8 98.6 84.295% Confidence limits± (kJ ·mol−1)21.0 145.7 97.5 63.2 36.7 15.0 3.4 205.9A (L ·mol−1 · s−1) 1.85E+01 1.65E+05 3.30E+06 2.77E+07 1.44E+08 5.53E+08 1.73E+09 3.14E+0695% Confidence limits± (L ·mol−1 · s−1)1.03E+02 6.41E+06 8.57E+07 4.65E+08 1.40E+09 2.20E+09 1.55E+09 1.72E+08The activation energy of solids hydrolysis was moderate (84.2 kJ ·mol−1) and similarto the activation energy of interval DP61-80, suggesting that hemicellulose dissolutioninto aqueous phase is not the most difficult step of hydrolysis process.The activation energy of soluble oligomer hydrolysis monotonically decreased withdecreasing molar mass from 98.6 kJ ·mol−1 to 57.5 kJ ·mol−1. Activation energy calcu-lated in this work describes material loss rate in each interval. It represents the reactionrate dependence on temperature for varying hemicellulose molar mass. From Table 5.9,high molar mass hemicellulose is sensitive to temperature variation as reflected by highactivation energy.The comparison of activation energy with previous literature is shown in Table 5.10.Xu et al. studied acid hydrolysis of GGM, which was extracted from thermomechanicalpulp with molar mass of 21.5 kDa, and reported activation energy of 150.3 kJ ·mol−1(Table 5.10). Visuri et al. applied hot water hydrolysis to GGM, extracted from spruceground wood with molar mass of 8 kDa, but obtained a much lower activation energyof 102 kJ ·mol−1 (Table 5.10). Despite the temperature and pH difference, the reactionrate of high molar mass GGM was more dependent on temperature than low molar massmaterial. Greenwood et al. modelled sugarcane bagasse acid hydrolysis and obtained thesolids dissolution activation energy of 191.3 kJ ·mol−1, which is much higher than thiswork (Table 5.10). This could be a result of Greenwood et al. treating large oligomers as136solids while only considering low DP xylan oligomers (DP=2-6) thus the potential barrierto produce small oligomers is high. The activation energy for soluble oligomers in thiswork (67.6-98.6 kJ ·mol−1) was comparable to Greenwood et al.’s activation energy forhydrolysis of soluble xylan oligomers, 93.4 kJ ·mol−1. The overall hemicellulose hydrol-ysis activation energy is lower than purified hemicellulose such as GGM (Xu et al. [325]).Softwood contains both pentose and hexose-derived hemicellulose. Pentose polymers aremore active and susceptible to hydrolysis than hexose as previously discussed in hemi-cellulose component mass recovery section (Figure 5.9 and Chapter 3) and reported byothers (Leppa¨nen et al. [148], Pu et al. [216]). Therefore, the mixture of sugars likely re-sults in a lower aggregate activation energy than for purified galactoglucomannan. Themost comparable experimental conditions are reported by Visuri et al., who reported anactivation energy of 102.2 kJ ·mol−1. This is similar to the activation energy of the DP121-140 interval, 98.6 kJ ·mol−1.Diffusion coefficient (m2 · s−1) (d) Damkohler number (da) Fraction of LBG adsorbedto the pulp (%) ( fl) Mass transfer coefficient (m·s−1) (h) Equilibrium constant (ke) Rateconstant (min−1 or g·mg−1 ·min−1) (ki) Half-width of the softwood chip (m) (l) Number-average molar mass (kg·mol−1) (mn) Weight-average molar mass (kg·mol−1) (mw) Z-average molar mass (kg·mol−1) (mz) Absolute amount of LBG adsorped to the pulp (mg·g−1)(ml) Total number of intervals (n) Freundlich constant (n) Number of experimental datapoints for data set j (nj) Primary pathway (p) Maximum adsorption capacity (mg·g−1)(qmax) Equilibrium adsorption capacity (mg·g−1) (qe)137Table 5.10: Comparison to hemicellulose/wood hydrolysis activation energies reported in literature.Year Author Substrate Hydrolysis condi-tionTemperature(oC)BondbreakingmodeEa (kJ ·mol−1) A (L ·mol−1 · s−1)2008 Xu et al. O-acetyl galac-toglucomannans1 M hydrochloricacid pH of 1 to 225 - 90 Random 150.3 6.09E+212012 Visuri et al. Galactogluco-mannanHot water pH of3.8 to 4.2150, 160, 170 Random 102±20 3.74E+112015 Greenwood et al. Sugarcanebagasse0.5% sulfuric acid 110 - 170 Random solid: 191.29; soluble:93.42; degradation:154.68N.A.2016 Ahmad et al. Birch wood Hot water 210 Random 134.4 3.7E+15This work Softwood Hot water 160 - 200 Middle solid: 84.2; soluble: 67.6 -98.6; degradation: 57.5solid: 3.14E+06; soluble:1.65E+05 - 1.73E+09; degra-dation: 1.85E+011385.5.4 Physical insightsFrom the model setup and analysis of the model results, hemicellulose hydrolysis beginswith random bond breaking. The random process could result from proton distribu-tion. During autohydrolysis of woody biomass, pH drops significantly due to liberationof acetic and uronic acid from hemicellulose, generation of formic acid, and demethyla-tion of pectin to form pectic acid (Leppa¨nen et al. [148], Pu et al. [216], Rissanen et al.[227, 228]). In poplar wood, acetic acid formation is the main cause for decreasing hy-drolysate pH (Li et al. [152]). Heating of the softwood slurry initiates release of acidicgroups, mainly O-acetyl groups and 4-O-methylglucuronic acid groups from hemicellu-lose. As the proton dissociates from the acidic group, the local pH drops and facilitatesthe hydrolysis reaction in localized regions.The random distribution of O-acetyl groups and regular distribution of 4-O-methylglucuronicacid (Section 3.1.2.2, Fengel and Wegener [62], Jacobs et al. [105], Lindberg [156], Lundqvistet al. [172], Timell [288], Willfo¨r et al. [319]) create random bond breaking opportuni-ties for hemicellulose within wood matrix. Therefore, the early stage of hydrolysis, ona macroscale, is characterized by random bond breaking to generate soluble primaryoligomers. The primary oligomers are generated from wood directly, therefore, the con-centration is relatively low. The soluble oligomer yield at short residence times is compa-rable for all molar mass intervals (Figure 5.3).The released acid groups and oligomers diffuse into the bulk solution where uniformmixing occurs due to vigorous agitation of the batch reactor. Proton concentration re-sulting from water dissociation is small compared to acid dissociation. Under these con-ditions, the probability of bonds breaking is now controlled by polysaccharide structureinstead of the spatial distribution of acid groups.From the molar mass analysis in this work (Section 4.4.1), the soluble hemicelluloseoligomers tend to break at the middle point of the chain, leading to a process that rapidlyunifies molar mass. Continuing from the suggested early stage hydrolysis mechanism,139the acid groups remaining on hemicellulose oligomers would likely be near the middleof the polymer resulting in production of small oligomers. Small oligomers are producedfrom both wood and large oligomers.5.6 ConclusionA population balance model was developed to describe the evolution of hemicelluloseoligomer molar mass. The general trend of hemicellulose molar mass evolution withreaction severity is concentration accumulation to a maximum followed by depletion,and reduction of oligomer size. It is likely that the bonds of insoluble hemicellulose breakrandomly to form soluble oligomers. This may be related to the distribution of acidicgroups within the softwood matrix. The soluble oligomers tend to break near the middleof the chain to produce smaller oligomers.The yield and selectivity of oligomer production by molar mass was calculated at160 oC to 200 oC. Medium temperature (180 oC) with short residence time favours largemolar mass hemicellulose generation. High temperature (200 oC) with short residencetime favours small molar mass hemicellulose generation. Hydrolysis rate of linear carbo-hydrates is more dependent on molar mass compared to branched carbohydrates. Highmolar mass oligomers are more sensitive to temperature variation as evidenced by higheractivation energies compared to smaller oligomers.This model provides new insights into the relative reactivity of hemicellulose inter-mediates. These insights are valuable to efforts to control molecular size and yield ofhemicellulose oligomers and open the door for future biorefinery chemical and materialconversion technologies.Initial LBG concentration (mg·L−1) (c0l) Initial o.d. pulp fibre concentration (g·g−1)(c0p) Soluble LBG concentration in supernatant (mg·L−1) (csl) Equilibrium concentrationof LBG in the aqueous phase (mg·L−1) (ce) Equilibrium concentration of adsorped LBGon fibre surface (mg·g−1) (cese) Normalized concentration in a size interval i (ci) LBG140concentration in supernatant after adsorption (mg·L−1) (csl)5.7 Future work1. This molar mass characterization and model approach could further be applied toother chemical reaction process, such as lignin extraction or cellulose depolymeriza-tion.2. Broader range of hydrolysis temperature and time could be explored to test themodel robustibility.3. Solids dissolution could be better modelled with more control parameters such ascore shrinking model.4. Mass transfer analysis could be combined into the model to capture hemicellulosediffusion process if large wood chips are used.5. Better characterization of hydrolysate composition and lignin and extractive re-moval by hydrolysis could be combined in the model.Concentration of LBG adsorbed at time t (mg·g−1) (qt) Gas constant (J · K−1 ·mol−1)(r) Severity factor (r0) Secondary pathway (s) Equilibrium empty site concentration onthe surface of pulp fibres (mg·g−1) (se) Selectivity of oligomer with molar mass intervali (sei) Selection function in a size interval i (min−1) (si) Temperature (oC) (t) Time (min) (t)Pseudo-time (min) (tp) Weight fraction of material in a size interval i (wi) Weighting factorfor interval i (wi) Volume (m3) (v)141Chapter 6Locust Bean Gum Adsorption to NBSKPulp: Isotherms and Kinetics6.1 Introduction6.1.1 MotivationPaper strength, essential to many applications, relies on the number of bonds, the strengthof bonds, fibre properties (e.g. fibre strength, length, and coarseness), and the distributionof fibres and bonds (sheet formation) (Leech [142, 143], Lindstro¨m et al. [160], Niskanen[197]). Paper strength can be modified through mechanical refining and chemical addi-tives.Refining or beating is a mechanical treatment designed to improve strength by en-hancing fibre swelling and inter-fibre bonding (Niskanen [197]). Laboratory refiningequipment, such as the PFI mill, imparts a mild treatment compared to industrial re-fining (Clark et al. [37]). Fibre response depends on refining severity. Paper strengthincreases with refining severity due to enhanced fibre swelling and straightening. Strongrefining causes kinks and dislocations of the fibres thus reducing paper strength (Niska-nen [197]). During refining, internal fibrillation and external fibrillation occur (Niskanen142[197]). Internal fibrillation causes delamination of the fibre wall creating a looser structurethat holds more water and produces a more flexible fibre. External fibrillation is due toabrasion of the fibre surface and fibril formation thus increasing inter-fibre contact andbonding. Refining also generates fibre fragments. Refining is an energy intensive processtherefore, strength improvement is sometimes achieved through application of additives.6.1.2 Polysaccharides as strength additivesPaper strength enhancement by hemicellulose adsorption primarily results from increasednumber of bonds and bonded area plus increased bond strength (Leech [142]). Bondingdepends on the fibre surface chemistry and physical structure of the fibre. Inter-fibrebonding refers to the zone where two fibres are sufficiently close for chemical bonding,van der Waals force, or molecular entanglement to occur (Niskanen [197]). The primarybonding between fibres and between hemicellulose and fibres is hydrogen bonds, whichform by mutual attraction of polar hydroxyl groups of cellulose and hemicellulose (Han-nuksela et al. [87], Leech [142], Niskanen [197]). The bonding energy is approximately8-32 kJ ·mol−1, and both the hydroxyl group and carboxyl group contribute to hydrogenbonding (Niskanen [197]). Hydrogen bonds can occur between fibres to establish the fi-bre network of paper and between micro-fibrils to create fibre structural rigidity. Van derWaals forces also contribute to the inter-fibre bonding, but with lower bonding energy, 2-8kJ ·mol−1, than hydrogen bonds (Niskanen [197]). Mechanical entanglement contributeslittle to paper strength (Leech [143]).GGM, the primary softwood hemicellulose, is not commercially available thereforegalactomannan is commonly used a substitute for GGM due to structural similarity andcommercial availability. Galactomannan is composed of (1-4)-β-D-mannopyranosyl unitswith side chains of α-D-galactopyranosyl units (Lindstro¨m et al. [160]). Natural galac-tomannans, such as guar gum and LBG, are primarily isolated from the seeds of the guarplant (Cyamposis tetragonoloba) and the carob tree (Ceratonia siliqua), respectively (Fen-143gel and Wegener [62]). Guar gum and LBG have a similar structure but LBG has a highermannose: galactose ratio (3:1-6:1) compared to guar gum (2:1) (Lindstro¨m et al. [160]).The greater mannan content was shown to cause greater improvements in pulp strength(Lindstro¨m et al. [160]). Studies from Suurnakki et al., Hannuksela et al.(2002) and Han-nuksela et al. (2004) also agreed that increasing mannose: galactose ratio increases ad-sorption to kraft fibres. For these reasons, LBG was selected as the model compound foradsorption to NBSK pulp in this study. The high molar mass of LBG and lack of O-acetylgroup substitutes are the main structural differences between LBG and GGM. Temper-atures greater than 80 oC are required to fully hydrolyze LBG; the hydrolysate is veryviscous (Roller and Jones [233]).The structure of LBG is mainly galactomannan-type polysaccharides: (1-4)-β-D-manno-pyranosyl units with side chains of (1-6)-α-D-galactopyranosyl units with a 1:4 ratio ofgalactose to mannose (BeMiller and Whistler [18], Roller and Jones [233]), as shown inFigure 6.1. It is a neutral polysaccharide with a reported molecular weight of 1500 kDa(de Jong and van de Velde [48]).Figure 6.1: Structure of LBG cited from de Jong and van de Velde [48].Leech concluded that 0.5 wt% LBG dosage doubled the bonding strength of paper.Swanson conducted LBG adsorption onto coniferous sulphite pulp and reported that ten-sile strength increased by 33% following sorption of 2 wt% LBG. Burst strength increased32% with 0.5 wt% LBG in beaten pulp; this corresponded to a 70% reduction in beating144time (Swanson [280]). Very little improvement in burst strength was observed after LBGaddition to unrefined pulp, which indicates that high surface area is necessary for LBGadsorption (Swanson [280]). Hannuksela et al. also demonstrated the importance of beat-ing and attributed increased LBG adsorption to higher degree of fibrillation and increasedtotal pore volume (Hannuksela et al. [87]).6.1.3 Factors influencing hemicellulose adsorptionThree factors have been shown to influence adsorption of hemicellulose to pulp fibre:adsorption conditions, fibre properties, and hemicellulose properties (Hannuksela et al.[87, 88], Leech [143], Most [194], Russo [237], Zakrajsˇek et al. [333]). Given the multitudeof factors and interactions, there are many contradictory reports regarding the effects ofchanging variables on adsorption results.Adsorption conditions include temperature, time, hemicellulose dosage, pH, salt ad-dition, fibre consistency and agitation rate. Temperature and time are the most commonlyexamined factors for hemicellulose adsorption, but the reported effects vary greatly. Ad-sorption of partially methylated LBG on bleached sulfite pulp increased with temperaturefrom 5 oC to 61 oC (Russo [237]). However, Gruenhut concluded that LBG adsorption tokraft fibre increased with decreasing temperature; maximum adsorption was observed at4.2 oC (Gruenhut [82]). Leech found that as the adsorption process continued, a greateramount of guar gum was retained by pulp fibre. However, the adsorption rate decreasedwith increasing contact time because adsorbed molecules partially blocked the fibre sur-face (Leech [143]). Most reported a similar trend in the dependence of adsorption rate ofpine hemicellulose to bleached sulfite pulp with respect to time, and further concludedthat adsorption equilibrium was not obtained even after 10 days. However, Swansonet al. reported equilibrium achievement within 30 minutes. The contrast might relate tothe hemicellulose dosage.Hemicellulose dosage level significantly impacts time required to achieve equilibrium145state. Most studied the adsorption of pine hemicellulose to bleached sulfite pulp withdosage of 3.0-6.4wt% of o.d. pulp, and concluded that adsorption equilibrium was notobtained even after 10 days. But Most also observed irreversible adsorption at low reten-tion of hemicellulose, thus, Most concluded that the adsorption mechanism might changeas greater amounts of hemicellulose adsorped on to pulp. Zakrajsˇek et al. studied cationicstarch adsorption at dosage of 1wt% of o.d. pulp and equilibrium was obtained in lessthan 200 seconds. Swanson et al. obtained 76-96% LBG adsorption to bleached sulfitepulp fibre for dosages of 1-3wt% of o.d. fibres and the equilibrium was reached within 30minutes.Adsorption occurs by electrostatic interaction of polyelectrolytes (polymers with elec-trolyte groups) with cellulose fibres (Hedborg and Lindstrom [92], Shirazi et al. [255])thus salts, process chemicals and pH strongly influence the process (Hedborg and Lind-strom [92], Shirazi et al. [255], van de Steeg [301], van de Steeg et al. [302, 302], Zakrajsˇeket al. [333]). Cellulose fibres are negatively charged due to carboxyl groups and hydroxylgroups (Niskanen [197], Sjostrom [261], van de Ven [304]). Adsorption of polyelectrolytessuch as cationic potato starch and cationic waxy maize starch (Shirazi et al. [255], Van deSteeg et al. [300], van de Steeg et al. [302]) have been extensively studied for retention offibre fines or mineral fillers (Hedborg and Lindstrom [92], Shirazi et al. [255], Van de Steeget al. [300], van de Steeg [301], van de Steeg et al. [302], Wa˚gberg and Ha¨gglund [316]).There are fewer reports examining adsorption of LBG or guar gum.Addition of salts such as NaCl or CaCl2 decreases the attractive electrostatic forces be-tween cationic starch and cellulose fibre thus adsorption decreases (Hedborg and Lind-strom [92], van de Steeg [301], van de Steeg et al. [302]). For example, adsorption ofcationic starch to microcrystalline cellulose was negligible in the presence of 0.02 mol·L−1NaCl (van de Steeg et al. [302]). This dependence also indicates that electrostatic inter-action is the main driving force for adsorption. Hedborg and Lindstrom also observedthat addition of 0.1 mol·L−1 CaCl2 prevented the adsorption of cationic potato starch to146bleached softwood kraft pulp (Hedborg and Lindstrom [92]). The influence of pulp sus-pension ionic strength on LBG adsorption to cellulose fibre is less well-studied. de Jongand van de Velde determined that the charge density, defined as mol negative charge/molof monosaccharide, of native LBG was less than 0.3. Thus, LBG is a weakly, negativelycharged polymer.pH also influences electrostatic forces between cellulose and adsorbate. When pHincreases, carboxyl groups deprotonate and generate more negative charge on fibre sur-face (Hedborg and Lindstrom [92]). As a result, adsorption of cationic polymers increaseswith rising pH (Shirazi et al. [255], Van De Steeg [299], van de Steeg et al. [302]). However,for polymers with negative charge, low pH facilitates adsorption process by convertingcarboxyl groups to their undissociated state (Scallan [249]). High pH leads to a highelectrostatic repulsion between fibre and negatively charged polymer, thus reducing theadsorption. Hemicellulose content within pulp fibre could dominate the adsorption re-sponse at varying pH. Gruenhut concluded that LBG adsorption to kraft pulp fibre washigher at pH 4 (Gruenhut [82]). Keen and Opie found that maximum guar gum adsorp-tion to bleached kraft pulp was obtained at pH 6.7 and minimum adsorption occurredat pH 11.5 (Keen and Opie [122]). In contrast, Most found hemicellulose from slash pineadsorped more quickly to bleached sulfite pulp at pH 10 than at pH 4.5. Kraft pulp typ-ically contains more hemicellulose than sulfite pulp. At higher pH, the carboxyl groupsof hemicellulose deprotonate to generate more negative charge, increasing the repulsionbetween hemicellulose and pulp fibres, leading to low adsorption at pH 11.5 (Keen andOpie [122]). Finally, Hannuksela et al. reported that adsorption of guar gum on bleachedkraft pulp is independent of refining severity, pH, temperature and salt concentration.Mass transfer is influenced by agitation rate and fibre consistency. A high agitationrate creates turbulence in pulp suspension. Russo found adsorption of partially methy-lated LBG to bleached sulfite pulp increased with agitation rate from 0 r.p.m. to 3000r.p.m. and concluded that external mass transfer is the rate determining step of adsorp-147tion. Turbulence reduces mass transfer resistance by disrupting the boundary layer at theinterface of the fibre and bulk solution (Russo [237]). Zakrajsˇek et al. reported a similartrend and further found that the increase in adsorption rate with agitation was greaterthan the increase in desorption rate. Fibre consistency negatively correlates to extent ofadsorption. Zakrajsˇek et al. showed 3wt% fibre consistency did not improve adsorptionof starch compared to 0.5wt% fibre consistency due to mechanical flocculation of fibres(Zakrajsˇek et al. [333]). Most also observed that low fibre consistency increased adsorp-tion of pine-derived hemicellulose to bleached sulfite pulp due to greater fibre surfaceavailability.Fibre properties such as surface area and fines content change the availability of ad-sorption sites (Zakrajsˇek et al. [333]). For example, Zakrajsˇek et al. found starch adsorp-tion increased due to refining-induced fibrillation and generation of fines. Adsorption ofpartially methylated LBG also increased with refining and the corresponding increase inspecific surface area (Russo [237]). Keen and Opie reported that guar gum adsorption isa function of freeness, which they took as an indirect measure of fibre surface area. Han-nuksela et al. attributed the increase of guar gum adsorption to refined, bleached kraftpulp to the changes in fibre wall structure such as increased surface area and total porevolume.Hemicellulose structure such as molar mass and side groups will change adsorptionrate and subsequent paper properties. Hannuksela et al. (2002) and Hannuksela et al.(2004) investigated guar gum adsorption to bleached softwood kraft pulp by using en-zymes to modify the guar gum. Mannanase was used to produce guar gum with varyingmolar masses and alpha-galactosidase was used to change the ratio of galactose to man-nose (Hannuksela et al. [87, 88]). Reducing galactose side groups facilitated adsorption ofguar gum to pulp but this did not change the tensile strength of handsheets (Hannukselaet al. [87, 88]). Molar mass modification had little effect on adsorption rate or the ten-sile strength of handsheets (Hannuksela et al. [87, 88]). The greatest degree of guar gum148adsorption occurred when there were approximately 4 galactose side groups per 10 man-nose units (Hannuksela et al. [88]). Lindqvist et al. modified GGM via cationization andcarboxymethylation and reported the subsequent effects on properties when the GGMswere added to peroxide-bleached thermomechanical pulp. The maximum tensile strengthincreased by 11% after adsorption of 10 mg native GGM per gram of dried fibre. Cation-ized GGM increased tensile strength by 4.7% at the same dosage level as native GGMand also increased water retention. Adsorption of carboxymethylated GGM slightly de-creased the strength relative to untreated pulp. Ultimately, functionalized GGM had littleimpact on paper strength relative to native GGM.6.1.4 Adsorption kinetics and isothermsAdsorption isotherms describe the adsorption of a substance to a solid surface from anaqueous phase under isothermal conditions (Foo and Hameed [67]). The isotherm repre-sents the equilibrium state when the rate of adsorption is equal to the rate of desorption.Langmuir isotherms and Freundlich isotherms are the most common models used to de-scribe dye or chemical adsorption to cellulosic fibres (Li et al. [153], Roy et al. [235], Urru-zola et al. [298], Vucˇurovic´ et al. [314], Zakrajsˇek et al. [333]).The Langmuir isotherm model, first proposed in 1916, assumes ideal monolayer chemisorp-tion on a smooth surface with a finite number of sites (Laidler [137], Langmuir [138]). Theadsorption sites are identical and have no lateral interaction or molecule rearrangement(Foo and Hameed [67], Langmuir [138], Shirazi et al. [255]) while adsorbate molecules areregarded as rigid particles with a fixed volume (Van De Ven [303]).The Langmuir isotherm is derived from the equilibrium adsorption reaction of LBGto substrate NBSK pulp fibre as Equation 6.1:Ce + Sek1k−1CeSe (6.1)where Ce is the equilibrium concentration of LBG (mg·L−1) in the aqueous phase, Se is149the equilibrium empty site concentration (mg·g−1 o.d. fibre) on the surface of pulp fibres,and CeSe is the equilibrium concentration of adsorped LBG (mg·g−1 o.d. fibre) on fibresurface.At equilibrium:k1[Ce][Se] = k−1[CeSe] (6.2)Defining substrate surface coverage as θs (Equation 6.3), Se could be expressed asEquation 6.4.θs =[CeSe][Se] + [CeSe](6.3)[Se] = (1− θs)([Se] + [CeSe]) (6.4)Equation 6.2 can then be rearranged as Equation 6.5:k1[Ce](1− θs) = k−1θs (6.5)The equilibrium constant Ke is defined as:Ke =k1k−1=[CeSe][Ce][Se](6.6)Equation 6.5 could be further rearranged (Equation 6.7) and solved for θs as Equa-tion 6.8:KeCe = (1+ KeCe)θs (6.7)θs =KeCe1+ KeCe=qeQmax(6.8)150where qe (mg·g−1 o.d. fibre) is the concentration of LBG adsorbed at equilibrium andQmax (mg·g−1 o.d. fibre) is the maximum adsorption capacity. The equation can be fur-ther linearized as:Ceqe=1KeQmax+CeQmax(6.9)Plotting Ceqe versus Ce enables calculation of maximum adsorption capacity, Qmax, fromthe slope and equilibrium constant Ke from the intercept.From Equation 6.8, when Ce is large, θs is approximately equal to 1, representing fullcoverage of substrate; while, when Ce is small, θs approaches zero, suggesting limitedadsorption and surface coverage.When the adsorption surface is non-ideal and there are interactions among adsorbedmolecules, multilayer adsorption can occur (Foo and Hameed [67], Laidler [137]); thiscondition is better described by the Freundlich isotherm model.The rate equation for Freundlich isotherm is:qe = kCe1n (6.10)where k is rate constant (g·mg−1 ·min−1), qe is the concentration of LBG (mg·g−1 o.d.fibre) adsorbed at equlibium state, n is the Freundlich constant and Ce is concentration ofLBG (mg·L−1) in the aqueous phase.Taking the logarithm of both sides allows the equation to be linearized:logqe = logk +1nlogCe (6.11)The plot of logqe as a function of logCe yields slope of 1n and intercept of logk.Adsorption kinetics describe the variation of amount adsorbed with residence timeand can instruct how to most effectively apply additives during papermaking (Zakrajsˇeket al. [333]). Adsorption rates of polymer are related to the collision rate, which is a func-151tion of Brownian motion and flow conditions (Van De Ven [303]). For small particles,collision rate is dependent on Brownian motion, while for large particles or systems withflow motion, the rate is dependent on flow conditions (Van De Ven [303]).Pseudo-first-order and pseudo-second-order are the most commonly used kinetic mod-els to describe adsorption of dyes or chemicals to pulp fibres (Li et al. [153], Roy et al.[235], Urruzola et al. [298], Vucˇurovic´ et al. [314]). The model was first proposed by Hoand McKay (Ho [95], Ho and McKay [96]) and applied in many subsequent cellulose /fibre adsorption studies (Li et al. [153], Pan et al. [203], Roy et al. [235], Vucˇurovic´ et al.[314]). LBG adsorption on NBSK pulp kinetics was studied by fitting pseudo-first-orderand pseudo-second-order model. The derivation of equations follows.Pseudo-first-order adsorption kinetics is described by:dqtdt= k(qe − qt) (6.12)where k is rate constant (g·mg−1 ·min−1), qe is the concentration of LBG (mg·g−1 o.d.fibre) adsorbed at equlibium, same as in Equation 6.8, and qt is concentration of (mg·g−1o.d. fibre) LBG adsorbed at any time, t (min).Integrating as shown in Section A.1, yields:ln(qe − qt) = −kt + lnqe (6.13)Plotting ln(qe − qt) as a function of t will yield a line with slope of k and intercept oflnqe.The rate equation for pseudo-second-order adsorption kinetics:152dqtdt= k(qe − qt)2 (6.14)where k is rate constant (g·mg−1 ·min−1), qe is the concentration of LBG (mg·g−1 o.d.fibre) adsorbed at equlibium state, same as in Equation 6.8, and qt is concentration of LBGadsorbed (mg·g−1 o.d. fibre) at any time, t (min).Integrating as shown in Section A.1, yields:tqt=1k(qe)2+tqe(6.15)Plotting tqt as a function of t will yield a line with slope1qe and intercept of1k(qe)2.Though starch adsorption to pulp fibre has been well studied (Hedborg and Lind-strom [92], Shirazi et al. [255], Van De Steeg [299], van de Steeg et al. [302], Wa˚gberg andBjorklund [315], Zakrajsˇek et al. [333]), there is a lack of theoretical understanding of theadsorption of LBG, a closely related additive. The mechanism, isotherms and kinetics arenot reported.6.2 Goals and HypothesesIt will be difficult to interpret the adsorption of the diverse oligomers produced by hydrol-ysis of softwood chip fines therefore adsorption experiments with LBG, a model hemicel-lulose compound, will be conducted in order to provide mechanistic insights.The goals of this chapter are to identify the favorable adsorption condition for LBGon NBSK pulp, understand the underlying mechanisms, and the resulting effects on pa-per properties. LBG adsorption kinetics are analyzed by pseudo-first-order and pseudo-second-order with respect to LBG concentration. The adsorption isotherms are analyzedusing the Langmuir model and the Freundlich model. Due to the conflicting reports , theeffects of temperature, refining, sodium chloride addition, and pH on LBG adsorption153are investigated. Changes in paper strength due to LBG adsorption to NBSK pulp werestudied; refining and dosage level of LBG were varied.Specific hypotheses are:1. LBG adsorption will display pseudo-first-order or pseudo-second-order kineticswith respect to LBG concentration.2. LBG adsorption isotherms can be described by the Langmuir adsorption model orby the Freundlich adsorption model.3. LBG adsorption is affected by temperature, time, sodium chloride addition and pHadjustment.4. NBSK paper strength can be enhanced by LBG adsorption and refining.6.3 Experimental Design6.3.1 MaterialNBSK pulp was kindly supplied by Canfor Pulp Products. LBG with purity greater than90% was purchased from Sigma Aldrich. Sulfuric acid (98 wt%) was purchased fromSigma Aldrich and diluted to desired concentration. The carbohydrates kit used to cal-ibrate the HPLC was purchased from Sigma Aldrich and contained mannose, glucose,galactose, xylose and arabinose. The purity of the standards was greater than 98%.6.3.2 LBG adsorptionLBG powder was hydrolyzed in deionized water at a concentration of 0.5 wt% at 98-100oC for 45 minutes with continuous agitation. Undissolved gum particles in hydrolyzedLBG stock solution were removed by two rounds of vacuum filtration. The filtrate wascentrifuged twice at 3500 r.p.m. for 15 minutes, and the supernatant was recovered for154adsorption experiments. The weight-average molar mass of hydrolyzed LBG was 1215.4kDa ± 88.9 kDa.Adsorption experiments were conducted in 150 mL Erlenmeyer flasks with rubberstopper. The kinetics study was conducted by adding 0.2 wt% LBG relative to o.d. pulpfibre; the fibre consistency of the slurry was 0.5 wt%. Adsorption of dye, starch or chem-icals to pulp fibre is usually tested at 20-40 oC (Zakrajsˇek et al. [333]). This temperaturerange captures typical mill operating conditions (T ≈ 35 oC). In this study, adsorptionwas conducted at 25 oC, 35 oC, and 45 oC in an incubator shaker with continuous agita-tion (150 r.p.m.) for 0.5 to 120 minutes. Control flasks containing only LBG or only fibrewere run in parallel. All conditions were tested in duplicate. After adsorption, all sam-ples were centrifuged at 3500 r.p.m. for 15 minutes. The supernatant was preserved forcompositional analysis.The experimental design for LBG adsorption isotherm and kinetics is summarizedin Table 6.1. Adsorption isotherm experiments were conducted by varying LBG dosagefrom 0.1 wt% to 2wt% of o.d. pulp fibre. Adsorption was conducted at 25 oC with contin-uous agitation (150 r.p.m.) for 10 minutes in the incubator shaker. Other conditions wereidentical to those in the protocol for the kinetics study.Table 6.1: LBG adsorption isotherm and kinetics experimental design.Dosage(wt%)Adsorption temper-ature (oC)Adsorption time(min)0.1-2 25 100.2 25, 35, 45 0.5-120Table 6.2 summarizes the conditions tested to determine the effect of LBG concen-tration, temperature, salt addition, refining and pH. The effect of temperature and LBGconcentration was determined by varying temperature from 25-80 oC at two dosage levels155(0.2 wt% and 4.4 wt%). The adsorption time was 10 minutes without pH adjustment orsalt addition. Salt addition and pH effects on LBG adsorption were investigated at 25 oCfor 10 minutes. Sodium chloride was varied from 0-1 mol·L−1 and pH was varied from 2to 13 in order to test the full range of conditions previously reported in the literature.Table 6.2: Factors investigated for LBG adsorption: LBG dosage, temperature, NaCladdition and pH and refining.Dosage(wt%)Adsorptiontemperature(oC)Adsorptiontime (min)NaCladdition(mol·L−1)pH of pulpsuspensionRefining(rev)0.2 25-45 10 N.A. N.A. N.A.4.4 25-80 10 N.A. N.A. N.A.0.2 25 10 0-1 N.A. N.A.0.2 25 10 N.A. 2 to 13 N.A.0.2 25 10 N.A. N.A. 3000The effects of refining and LBG adsorption were tested using NBSK pulp. For theseexperiments, the weight-average molar mass of hydrolyzed LBG was 1278.3 kDa ± 81.9kDa. The pulp was refined to 3000 rev and 0.2 wt% LBG was added. Adsorption condi-tions were the same as those used to determine the adsorption isotherm. All conditionswere tested in duplicates. After adsorption, all samples were centrifuged at 3500 r.p.m.for 15 minutes. The supernatant was preserved for compositional analysis.6.3.3 LBG solution compositional analysisThe galactomannans were hydrolyzed from oligosaccharides to monosaccharides withsulfuric acid in an autoclave at 121 oC for 1 hour according to NREL LAP (Hames et al.[86], Sluiter et al. [265, 266]). The galactomannan monomer content in the supernatant156was analyzed by Dionex AS50 HPLC (Thermo Scientific) coupled with an ion exchangePA1 column (Dionex), an ED50 electrochemical detector (pulsed amperometric detector)with a gold electrode, and an AS50 autosampler (Dionex). De-ionized water was usedas eluent with a flow rate of 1 mL·min−1. The auxiliary pump added 0.2 M NaOH at0.5 mL·min−1. The samples were filtered through a 0.22 µm nylon syringe filter beforeinjection. The injection volume was 10 µL.The fraction of LBG adsorbed to the pulp, fL, was determined by the difference ofgalactomannan content in supernatant, relative to LBG control after adsorption as Equa-tion 6.16.fL =C0L − CSLC0L× 100% (6.16)where C0L is initial LBG concentration (mg·L−1), calculated from LBG control flask andCSL is LBG concentration in supernatant (mg·L−1) after adsorption.The absolute amount of LBG, mL, adsorped to the pulp (mg·g−1 o.d. pulp) was calcu-lated as in Equation 6.17.mL =C0L − CSLC0P(6.17)where C0P is the initial o.d. pulp fibre concentration (g·L−1).6.3.4 LBG molar mass determinationRefer to Section 4.3. The calibration standard was a pullulan standard kit (WAT034207)with molar mass range of 6.1 kDa to 642 kDa.1576.3.5 LBG adsorption for strength analysisAn aqueous LBG solution of 0.5 wt% consistency was reacted at 85 oC for 10 minutes withconstant stirring to produce a transparent viscous solution. NBSK pulp of 1.5 wt% fibreconsistency was prepared in the pulp disintegrator for 600 counts, which is equivalent to1500 rev. The PFI mill was used to refine the pulp to 3000-9000 rev. The LBG solution wasthen added to NBSK pulp suspension to the desired dosage with manual stirring for 10minutes at 25 oC. Table 6.3 summarizes the combinations of LBG addition and refiningtested. The treated pulp was next diluted to a fibre consistency of 0.3 wt%; this reducedthe pulp suspension temperature to 20 oC. A 2 L sample was collected for freeness testing,and the remaining suspension was used for handsheet making. Two to three replicateswere conducted for each condition.158Table 6.3: Experimental design for investigation into the effects of LBG dosage andpulp refining on paper strength.Refining (rev) Dosage (wt%) Condition replicates0 0 20 0.1 20.5 21 23000 0 20.1 30.5 21 26000 0 30.1 30.5 31 29000 0 30.1 30.5 31 21596.3.6 Freeness testing, handsheet preparation and strength analysisFreeness (Canadian standard method) was tested according to Tappi Method T 227. Hand-sheets with an average grammage of 60 g·m−2 were prepared on a wire of 200 cm2 ac-cording to Tappi Method T 205. The following handsheet properties were tested: weightand thickness (L&W micrometer), tensile strength (L&W Tensile Strength Tester, TappiMethod T 494), tear index (Elmendorf Tearing Tester, Tappi Method T 414) and burst in-dex (Mullen Tester, Tappi Method T 403). Brightness and scattering coefficient were testedby Technidyne ColorTouch PC according to ISO 2470-1 and TAPPI T 525.6.4 Results and Discussion6.4.1 Adsorption kineticsThe fraction of LBG adsorbed to NBSK pulp (Equation 6.16) at 25 oC is plotted as a func-tion of time with LBG dosage of 0.2 wt% relative to o.d. pulp (Figure 6.2). The initial ad-sorption rate was high with more than 52% adsorption in 0.5 minutes and 82% adsorptionwithin 5 minutes. After 10 minutes, the adsorption fraction plateaued at approximately93%. Adsorption equilibrium was achieved in 10 minutes as the LBG adsorption fractionwas constant from 10 minutes to 120 minutes. This result is consistent with previous re-search that found initial adsorption is rapid, and achieves equilibrium in a few minutes(Swanson et al. [281], Zakrajsˇek et al. [333]) when low dosages are applied. Once free siteson surface of fibre are saturated, equilibrium has been reached and the adsorption frac-tion is constant (Leech [143]). Adsorption residence time was maintained at 10 minutesin all subsequent studies.1600 20 40 60 80 100 12030405060708090100110LBG adsorption fraction (fL, %)Time (min)Figure 6.2: The fraction of LBG absorbed to NBSK pulp at 25 oC as function of time.LBG dosage was 0.2 wt% relative to o.d. pulp and fibre consistency was 0.5wt%.Kinetic plots for LBG adsorption are presented in Figure 6.3. The poor fit of thepseudo-first-order model (R2=0.635) at 25oC suggests this model cannot describe LBGadsorption kinetics. The pseudo-second-order kinetics model fit well at all tested tem-peratures (R2 >0.997). This indicates that LBG adsorption is strongly influenced by theconcentration of LBG in solution. The second order reaction could be due to the potentialof a single, high molar mass LBG polymer to form multiple bonds. This also suggests thatchemisorption is the rate limiting step (Pan et al. [203], Vucˇurovic´ et al. [314]). Pseudo-second-order kinetics have also been observed for adsorption of several dyes and chem-icals to pulp fibres (Li et al. [153], Roy et al. [235], Urruzola et al. [298], Vucˇurovic´ et al.[314]).1610 1 2 3 4 5-2.0-1.5-1.0-0.50.00.51.0 Pseudo-first 25oC Linear fitln(qe-qt)Time (min)(a)y = - 0.408x - 0.016R2 = 0.635Intercept = -0.016  2.410Slope = -0.408  0.94195% confidence limits after 0 2 4 6 8 100123456(b) Pseudo-second 25oC Linear fitt/qtTime (min)y = 0.523x + 0.142R2 = 0.997Intercept = 0.142  0.273Slope = 0.523  0.05495% confidence limits after 0 2 4 6 8 10 12012345678910 Pseudo-second at 35oC Linear fitt/qtTime (min)(c)y = 0.860x+0.115R2 = 0.998Intercept = 0.115  0.331Slope = 0.860  0.07195% confidence limits after 0 2 4 6 8 10 120510152025 Pseudo-second at 45oC Linear fitt/qtTime (min)(d)y = 2.089x - 0.182R2 = 0.997Intercept = -0.182  0.900Slope = 2.089  0.16295% confidence limits after Figure 6.3: Linear fit of (a) pseudo-first-order kinetics and (b) pseudo-second-orderkinetics of LBG adsorption to NBSK pulp at 25 oC. (c) Linear fit of pseudo-second-order kinetics of LBG adsorption to NBSK pulp at 35 oC and (d) at 45oC.The adsorption rate constant and equilibrium adsorption amount at 25-45 oC are sum-marized in Table 6.4. The rate constant at 45 oC is 12 times larger than at 25 oC, and 3.7times larger than at 35 oC, suggesting that LBG adsorption to NBSK pulp is stronglytemperature dependent. The adsorption rate in an agitated pulp suspension dependson turbulent transport and Brownian motion (Van De Ven [303], Zakrajsˇek et al. [333]).When temperature increases, the collision frequency of particle and fibre increases thusincreasing adsorption rate with temperature (Table 6.4).162Table 6.4: LBG adsorption capacity at equilibrium (qe) and adsorption rate constant(k) at temperature range of 25-45 oC as determined by fitting pseudo-second-order kinetics. Note: 95% confidence limits after ±.Temperature(oC)Equilibrium ad-sorption capacity,qe, (mg·g−1 o.d.fibre)95% confidencelimits of qe ±(mg·g−1 o.d. fi-bre)Rate constant, k,(g·mg−1 ·min−1)95% confidencelimits of k ±(g·mg−1 ·min−1)25 1.91 0.05 1.93 0.1135 1.16 0.07 6.44 0.6145 0.48 0.16 24.03 12.06The equilibrium adsorption capacity (qe) , however, decreased with rising temperaturefrom 25 oC to 45 oC (Table 6.4). The amount at 45 oC is 25% of the amount at 25 oC.Since the amount adsorped is a result of competition between adsorption and desorption,increased desorption could lead to a lower adsorption amount at higher temperature. Tofurther investigate this competition, adsorption isotherms are discussed in Section 6.4.2and Section 6.4.3.1.The activation energy was determined by linear regression of the Arrhenius equation.For LBG adsorption at 25 oC to 45 oC, the activation energy was 99.33 kJ·mol−1 (± 56.28kJ·mol−1) with a pre-exponential factor of 4.73E+17 L·mol−1·min−1 (Figure 6.4). The highactivation energy suggests that LBG adsorption to NBSK pulp is a chemisorption process(Laidler [137]). Russo studied partially methylated LBG adsorption to bleached sulfitepulp and determined the activation energy of adsorption to be 18.4 kJ·mol−1 leadingRusso to propose that adsorption is a physical process dominated by diffusion or ad-sorption via van der Waals forces. The difference between this work and Russo’s mightlie in the agitation. Russo applied a low agitation rate of 12 r.p.m. while this study ap-plied an agitation rate of 150 r.p.m. Diffusion in the bulk solution might be rate limitingin Russo’s work.1630.00315 0.00320 0.00325 0.00330 0.003350.51.01.52.02.53.03.5y = -11946.953x + 40.697R2 = 0.998Intercept = 40.697  21.990Slope = -11946.953  6768.97495% confidence limits after  lnk Linear fitlnk1/TFigure 6.4: Linear regression of Arrhenius equation to determine activation energy ofLBG adsorption to NBSK pulp at 25 oC to 45 oC based on the pseudo-second-order kinetic model.6.4.2 Adsorption isothermsLBG adsorption isotherms were investigated by varying initial dosage of LBG relativeto NBSK pulp suspension. In Figure 6.5a the fraction of LBG adsorbed is plotted as afunction of dose while Figure 6.5b plots the mass of LBG absorbed as a function of dose.Increasing dosage causes the fraction of LBG adsorbed to decrease but a greater mass ofLBG is retained on the fibre up to a dosage of 2.1 wt%. Given the plateau in mass of LBGadsorbed for dosage between 0.5 wt% and 2.1 wt%, it can be inferred that there is finitenumber of adsorption sites on the fibre surface and that LBG adsorption is limited to thefibre surface. This inference is also supported by Wa˚gberg and Ha¨gglund’s conclusionthat polymers with molar mass greater than 48 kDa can only adsorb on the external fibresurface; the weight-average molar mass of LBG was 1215.4 kDa ± 88.9 kDa in this work.Hannuksela et al. also observed that the fraction of guar gum adsorbed decreased withincreasing concentration of guar gum; they attributed this to slow diffusion (Hannukselaet al. [87]).164-0.5 0.0 0.5 1.0 1.5 2.0 2.5 3.0020406080100120(a)LBG adsorption fraction ( fL, %)Dosage of LBG to o.d. pulp (%)-0.5 0.0 0.5 1.0 1.5 2.0 2.5 3.00123456(b)LBG adsorption amount (mL, mg/g pulp)Dosage of LBG to o.d. pulp (%)Figure 6.5: LBG adsorbed to unrefined NBSK pulp fibre as a function of LBG dosageto o.d. fibre after 10 minutes at 25 oC. (a) The fraction of LBG absorbed and (b)the absolute mass of LBG adsorbed.An adsorption isotherm model provides fundamental understanding of the adsorp-tion process and the relationship between the concentration of adsorbate and extent ofadsorption to the adsorbent (Roy et al. [235]). At 25 oC, the Langmuir model (Figure 6.6a)of Ceqe as a function of Ce exhibited a highly linear fit with R2=0.997. The Freudlich isothermmodel (Figure 6.6b) had a much lower R2, 0.566. It was concluded that LBG adsorptionto NBSK pulp at 25 oC obeys Langmuir adsorption principles, indicating that LBG ad-sorption is a monolayer chemisorption process limited to a finite number of sites (Fooand Hameed [67], Laidler [137], Langmuir [138], Shirazi et al. [255]).This implies that, un-refined NBSK pulp fibre has a smooth surface with identical sites that do not interact orrearrange.1650 20 40 60 80 100-5051015202530354045(a) Langmuir model 25oC Linear fitCe/qe (g/L)Ce (mg/L)y = 0.434x + 0.678R2 = 0.997Intercept = 0.678  1.711Slope = 0.434  0.03495% confidence limits after 0.0 0.5 1.0 1.5 2.00.050.100.150.200.250.300.350.400.45y = 0.097x + 0.179R2 = 0.566Intercept = 0.179  0.157Slope = 0.097  0.11895% confidence limits after (b) Freundlich model 25oC Linear fitlogqelogCeFigure 6.6: Linear fit of Langmuir model (a) and Freundlich model (b) at 25 oC ofLBG adsorption to unrefined NBSK pulp.Based on the Langmuir model calculation, the maximum adsorption capacity of thesubstrate (Qmax) is 2.31 mg·g−1 (± 0.03 mg·g−1) at 25 oC. Two cases of starch to pulpfibre adsorption kinetics and isotherm parameters are shown in Table 6.5 for comparison.The maximum adsorption capacity at 25 oC of this work, 2.31 mg·g−1, is much lowerthan cationic starch adsorption capacity 20-66 mg·g−1 (Shirazi et al. [255], Zakrajsˇek et al.[333]). However, the capacity for native corn starch adsorption to pulp fibre was reportedas 1.1-4.5 mg·g−1 (Cushing and Schuman [45]). Since the charge density of native starch isalmost as low as LBG, it can be concluded that positive charge leads to higher adsorptioncapacity while negative charge creates repulsion between fibre and adsorbate. Similar lowadsorption capacity to pulp fibre (Table 6.5) was observed for native LBG (1.8-5.0 mg·g−1,Gruenhut [82]) and partially methylated LBG (0.61-12.34 mg·g−1, Russo [237]). Addi-tional factors likely contribute to the differences in LBG adsorption and cationic starchadsorption. The first is adsorbate or/and fibre property diversity. Fibre properties suchas surface area, fines content, and refining level affect adsorption (Zakrajsˇek et al. [333]).The second source of variation is adsorption conditions, especially transport and thermalconditions (Van De Ven [303], Zakrajsˇek et al. [333]).166Table 6.5: Summary of adsorption capacity and isotherms of polysaccharides adsorption to pulp fibre. Note: 95%confidence limits of this work are provided after ±.Year Author Adsorbate Adsorbent Maximumadsorptioncapacity, Qmax(mg·g−1)Equilibriumrate con-stant, KeTemperature(oC)pH Agitationrate(r.p.m.)Isothermmodel2009Zakrajsˇeket al.CationicallymodifiedstarchHardwoodsulphate shortfibres66 40 25 7.5±0.2 500 ModifiedLangmuir2003Shirazi et al.CationicstarchUnbleachedblack sprucethermome-chanical pulp20 N.A. N.A. 5.2 200 Langmuir1959Cushing andSchumanNative cornstarchBleachedsulphite pulp1.1-4.5 N.A. 91-96 5.0-5.1 N.A. N.A.1959RussoPartiallymethylatedLBGBleachedsulfite pulp0.61-12.34 N.A. 25 6.5 12 N.A.1953GruenhutLocust beangumRosin sizedkraft pulp1.8-5.0 N.A. 22 N.A. N.A. N.A.2019 This work Locust beangumNorthernbleached soft-wood kraftpulp2.31 ± 0.03 0.64 ± 0.74 25 5.2-5.8 150 Langmuir167The equilibrium constant (Ke) calculated from the Langmuir model is approximately0.64 (± 0.74) at 25 oC in this work. This indicates that adsorped LBG concentration islower than the product of aqueous phase LBG concentration and free adsorption site con-centration (Equation 6.6). The equilibrium constant for cationic starch adsorption to pulpfibre was much higher, for example Zakrajsˇek et al. reported Ke= 40. This high equilib-rium constant could due to the high concentration of adsorped cationic starch resultedfrom the natural attraction between negatively charged pulp fibres and cationic starch.Some previous reports support the theory of irreversible adsorption of pulp fibres (Leech[142], Most [194]). Unlike cationic starch, LBG is natural carbohydrate polymer with anegative surface charge (de Jong and van de Velde [48]). The repulsive forces betweenLBG and NBSK pulp fibre will reduce adsorption capacity. Consequently, LBG adsorp-tion to unrefined NBSK pulp is characterized by low adsorption capacity and a relativelylow equilibrium constant.6.4.3 Factors influencing adsorption6.4.3.1 Effect of temperatureThe influence of temperature on LBG adsorption to NBSK pulp fibre at two dosage levels,0.2 wt% and 4.4 wt% on o.d. fibre, is reported in Figure 6.7a. For both dosage levels, LBGadsorption amount decreased when temperature increased from 25 oC to 80 oC. At 80oC, the amount of LBG absorbed to fibre is negligible making 25 oC the most favourabletested temperature. Polymer adsorption rates are related to the collision rate, a functionof Brownian motion and flow conditions (Van De Ven [303]).16820 30 40 50 60 70 80-4048121620(a) 0.2% dosage 4.4% dosageLBG adsorption amount (mL, mg/g pulp)Adsorption temperature (oC)-5 0 5 10 15 20 25 30 35 40 450246810 Langmuir model 35oC Linear fitCe/qe (g/L)Ce (mg/L)(b)y = 0.165x + 0.840R2 = 0.988Intercept = 0.840  0.631Slope = 0.165  0.03495% confidence limits after Figure 6.7: (a) Amount of LBG (mg·g−1 o.d. pulp) absorbed to NBSK pulp after 10minutes as a function of temperature at two dosage levels (0.2 wt% and 4.4wt% of o.d. fibre). (b) Linear fit of Langmuir model for LBG adsorption at 35oC of LBG adsorption to NBSK pulp.To understand decreasing LBG adsorption amount with increasing temperature (Ta-ble 6.4, Figure 6.7a), the Langmuir model was fitted to the LBG adsorption isotherm at 35oC (Figure 6.7b). At 35 oC, the fit of the Langmuir model was also very high (R2=0.988,Figure 6.7b) and the maximum adsorption capacity (Qmax) was calculated to be 6.07mg·g−1 (± 0.03 mg·g−1) o.d. pulp. This value is greater than the value determined at25 oC, 2.31 mg·g−1, suggesting that more surface sites are available at higher tempera-ture. As discussed previously, the frequency of LBG and fibre collisions increases withtemperature resulting in a higher maximum adsorption capacity.The equilibrium constant (Ke) was determined to be 0.20 (± 0.10) at 35 oC, which islower than 0.64 at 25 oC. This decrease could be caused by lower concentration of ad-sorped LBG or/and increased surface site concentration (Equation 6.6). From Table 6.4,the equilibrium adsorption amount (qe) decreased with increasing temperature, indicat-ing that the adsorped LBG concentration (CeSe) decreases with increasing temperature.This is in agreement with Figure 6.8 which shows that at 35 oC, site coverage decreases(θs, Equation 6.8) at equal LBG dosage level. When LBG dosage is approximately 0.13wt%, site coverage at 35oC was approximately one third of the site coverage at 25 oC169(Figure 6.8).0.0 0.5 1.0 1.5 2.0 2.50.00.20.40.60.81.01.2 25oC unrefined  35oC unrefined Site coverage, LBG dosage of o.d. pulp (%)Figure 6.8: Fibre surface site coverage (θs) calculated from Langmuir model at 25 oCand 35 oC for unrefined NBSK pulp.Site coverage on unrefined NBSK pulp increased rapidly at LBG dosages less than0.2wt% at 25 oC (Figure 6.8). However, the dependence on dosage diminishes as fullcoverage is approached. At 35 oC, the reduced dependence on dosage could result fromincreased maximum adsorption capacity (Qmax) and decreased LBG concentration adsor-ped on fibre (CeSe).Past studies have shown paper strength improvement is not proportional to hemi-cellulose dosage, and the increase is limited when high hemicellulose dosage is applied.Hannuksela et al. reported GGM dosage of 0.8 wt% o.d. fibre increased tensile strengthby 13% but a higher dosage of 1.6 wt% o.d. fibre caused only a small additional increasein tensile strength. The most significant improvement in tensile strength was achievedat GGM dosage less than 0.1 wt% dried fibre (Hannuksela et al. [88]). The results ofthis study likely explain Hannuksela et al.s observations. When LBG dosage approxi-mately 0.12wt%, the coverage of fibre site at 25oC was 0.52 (Figure 6.8). Full coveragewas obtained when the dosage increased to 2.13wt% of o.d. pulp fibre. In this study, thegreatest influence of dosage on LBG adsorption occurs under 0.20 wt% dosage at 25 oC170(Figure 6.8). As the fibre surface becomes saturated, it is likely that improvement in inter-fibre bonding will be limited leading to small gains in paper strength, even after additionof excess LBG. As a result of this observation, the strength analysis study (Section 6.4.4)was conducted using LBG dosage less than 1 wt% of o.d. NBSK pulp fibre.6.4.3.2 Effect of refiningTo further understand the effect of refining on LBG adsorption and subsequent changesin paper strength, LBG adsorption to NBSK pulp refined at 3000 rev was investigated at25 oC after 10 minutes. The linear fit of this data to the Langmuir model is presented inFigure 6.9. The calculated maximum adsorption capacity is 4.31 mg·g−1 (± 0.18 mg·g−1),almost double the capacity of the unrefined NBSK pulp (2.31 mg·g−1). This increase couldoriginate from the exposure of surface area due to fibrillation during refining.0 2 4 6 8 10 120.00.51.01.52.02.53.0 Langmuir model refined Linear fitCe/qe (g/L)Ce (mg/L)y = 0.232x + 0.133R2 = 0.996Intercept = 0.133  1.105Slope = 0.232  0.18095% confidence limits after Figure 6.9: Linear fit of Langmuir model at 25 oC for LBG adsorption to NBSK pulprefined at 3000 rev after 10 minutes.The equilibrium constant, Ke, is calculated to be 1.74 (± 0.27) with refined NBSK pulpfrom Figure 6.9, indicating that adsorped LBG concentration is greater at equilibriumstate, even with increased surface area of refined pulp (Equation 6.6). It can be furtherinferred that the LBG concentration in the aqueous phase should be very low. This is171in contrast to the equilibrium constant of unrefined NBSK pulp (Ke=0.64). It could beconcluded that when adsorption capacity is large, such as in refined pulp, most LBGis adsorped at equilibrium at low LBG dosage level. For unrefined pulp, however, theequilibrium state is characterized by high aqueous LBG concentration and free surfacesite concentration (Ke <1).Fibre surface site coverage is calculated from the Langmuir model and summarizedin Table 6.6. From Table 6.6 and Figure 6.8, site coverage of refined NBSK pulp at 25 oCis less than the site coverage of unrefined pulp due to the greater adsorption capacity ofrefined NBSK pulp.Table 6.6: Fibre surface site coverage with varying LBG dosage of o.d. pulp fibre at25 oC for NBSK pulp refined at 3000 rev calculated from Langmuir adsorptionmodel.LBG dosage of o.d. pulp (wt%) Site coverage (θs)0.19 0.440.34 0.690.62 0.956.4.3.3 Effect of salt additionLBG adsorption to NBSK pulp fibre in response to sodium chloride addition was inves-tigated at 25 oC for 10 minutes with LBG dosage 0.2wt% of o.d. fibre; sodium chlorideconcentration was varied from 0-1 mol·L−1 (Figure 6.10). The LBG adsorption fractionwas approximately 85-95% and independent of sodium chloride concentration at concen-trations less than 0.0125 mol·L−1 (Figure 6.10a). At all sodium chloride concentrationsgreater than 0.05 mol·L−1, the fraction of LBG adsorbed slightly increased to 100% (Fig-ure 6.10b). Native LBG is a negatively charged polymer (de Jong and van de Velde [48])therefore a high ionic strength solution will neutralize repulsive forces between LBG andfibre and cause increased adsorption. This trend is consistent with Hannuksela et al.’s172research that guar gum and GGM adsorption were unaffected by addition of sodiumchloride less than 0.1 mol·L−1.0.0000 0.0025 0.0050 0.0075 0.0100 0.0125 0.01502030405060708090100110LBG adsorption fraction (fL, %)NaCl concentration (mol/L)(a)0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.12030405060708090100110LBG adsorption fraction (fL, %)NaCl concentration (mol/L)(b)Figure 6.10: The adsorbed fraction of LBG on NBSK pulp fibre with varying sodiumchloride concentration (a) 0-0.0125 mol·L−1 (b) 0.05-1 mol·L−1 after 10 min-utes at 25 oC, LBG dosage 0.2wt% of o.d. fibre.6.4.3.4 Effect of pHLBG adsorption to NBSK pulp was investigated as buffer pH varied from 2 to 13 after 10minutes at 25 oC with LBG dosage 0.2wt% of o.d. fibre (Figure 6.11). The amount of LBGadsorbed ranged from 1.8-2.7 mg·g−1 of o.d. fibre with maximum standard deviation of0.26 mg·g−1 of o.d. fibre (Figure 6.11), indicating adsorption was not strongly affectedby pH. Adsorption was slightly higher at pH 2 and 5 due to the undissociated form ofhydroxyl group and carboxyl group on LBG and reduced repulsive forces. At high pH,hydroxyl group and carboxyl group deprotonate more easily increasing repulsive forcesbetween fibre and LBG thus lowering adsorption. The small difference likely reflects thelow negative charge density on LBG. Hannuksela et al. reported that pH (5 and 8) had noinfluence on guar gum adsorption to softwood kraft pulp fibre which is probably due toguar gum’s weak negative charge and the limited range of pH tested.1732 4 6 8 10 12 140.00.40.81.21.62.02.42.83.2LBG adsorption amount (mL, mg/g pulp)Adsorption pHFigure 6.11: The adsorbed amount of LBG on NBSK pulp fibre with varying pH ofpulp suspension after 10 minutes at 25 oC, LBG dosage 0.2wt% of o.d. fibre.6.4.4 Paper strength enhancement by LBG adsorptionPulp and paper properties including tensile, burst, tear index, brightness, and pulp free-ness were plotted as a function of PFI refining revolutions and LBG dosage in Figure 6.12.Refining and dosage positively influenced tensile strength and burst strength. Tear strength,freeness and brightness, in contrast, decreased with increasing refining or dosage level.It is well-established that refining increases NBSK paper strength (Figure 6.12). Tensileindex and burst index increased with refining level until 6000 rev and then plateaued.Refining to 9000 rev without LBG adsorption increased tensile index by 100.3%. However,as highlighted by Leech, paper strength does not increase continuously with bondingstrength as at higher levels of bonding, handsheet strength becomes more dependent onindividual fibre strength. Leech concluded that tensile and burst strength of bleachedsulfite pulp plateau after refining to approximately 6000 rev, which is consistent with theresults in Figure 6.12a and b.1740 2000 4000 6000 8000 100005060708090100110120130(a) 0% LBG 0.1% LBG 0.5% LBG 1% LBGTensile index (Nm/g)Refining revolutions0 2000 4000 6000 8000 10000345678910(b) 0% LBG 0.1% LBG 0.5% LBG 1% LBGBurst index (kPam2/g)Refining revolutions0 2000 4000 6000 8000 100008101214161820222426(c) 0% LBG 0.1% LBG 0.5% LBG 1% LBGTear index (mNm2/g)Refining revolutions0 2000 4000 6000 8000 10000350400450500550600650700750(d) 0% LBG 0.1% LBG 0.5% LBG 1% LBGFreeness (mL)Refining revolutions0 2000 4000 6000 8000 1000078798081828384858687(e) 0% LBG 0.1% LBG 0.5% LBG 1% LBGBrightness (%)Refining level (revolutions)0 2000 4000 6000 8000 1000016202428323640(f) 0% LBG 0.1% LBG 0.5% LBG 1% LBGScattering coefficient (m2/kg)Refining revolutionsFigure 6.12: NBSK pulp and paper properties as a function of LBG dosage and PFIrefining level: (a) tensile index, (b) burst index, (c) tear index, (d) freeness, (e)brightness and (f) scattering coefficient.175Mechanical refining had a much greater effect on tensile strength than LBG adsorp-tion. Miletzky et al. similarly reported that the effects of hemicellulose addition are mini-mal if high refining is applied. In this work, the maximum increase in tensile index due toadsorption of LBG to unrefined pulp was 20.1%. In comparison, the maximum increasein tensile index due to refining was 100.3%. Though refining improves paper strength toa greater extent, LBG adsorption can reduce the degree of refining and energy requiredto reach a target tensile index. For example, pulp must be refined to 2000 rev to reach atensile index of 80 N·m · g−1 without LBG adsorption but only 1000 rev were required toachieve the same index if the pulp is treated with 1wt% LBG. Others have made similarobservations. Swanson reported that refining time was reduced by approximately 70%with addition of 0.5 wt% LBG.Burst strength demonstrated a similar trend as tensile strength (Figure 6.12b). Themaximum burst index improved by 121.0% after applying 1 wt% LBG to highly refinedpulp (9000 rev). Enhanced tensile index and burst index could due to either increasedbonding area or an increased number of bonds and bond strength (Leech [142]). Similartrends were reported by Swanson.As expected, tear index decreased with refining and LBG addition possibly due tostronger bonding between fibres (Figure 6.12c). At 9000 rev, tear index decreased 62.8%after adsorption with 1 wt% LBG. A drop in tear index normally accompanies increasedtensile strength (Hannuksela et al. [88], Leech [142], Swanson [280]). Tearing occurs due tothe breakage of individual fibres or individual fibres being pulled from the sheet matrix.At higher bonding levels, individual fibres will break and this requires less energy thanremoving fibres from the sheet matrix (Hannuksela et al. [88], Leech [142], Swanson [280]).Freeness is a measurement of pulp drainage (Smook et al. [271]) and reflects fines con-tent, flexibility and the degree of external fibrillation (Niskanen [197]). With increasingrefining energy and LBG adsorption, freeness decreases (Niskanen [197]). NBSK pulp re-fined at 3000 rev with a tensile index of 108 N·m ·g−1 by LBG adsorption shows a freeness176of 531 mL. To achieve a comparable tensile index, 106 N·m · g−1, by refining required 9000rev and the freeness was 434 mL. LBG adsorption resulted in better drainage propertiesat equal tensile strength with less refining energy.Brightness decreased with increased refining and LBG dosage level (Figure 6.12e).However, the LBG solution is colourless. The main reason for brightness reduction islikely the decreased free, light-reflecting surface area of fibres, which is represented byscattering coefficient (Ek et al. [53]). Parsons reported that several factors contribute toscattering coefficient of paper including fibre size and shape, refractive index, and bond-ing in paper (Parsons [206], Simmonds and Coens [258]). If the same pulp is used toprepare handsheets, variation in scattering coefficient can be attributed to variation inbonding within the sheet.Scattering coefficient decreased with refining level and LBG dosage (Figure 6.12f),which suggests more bonding occurs after LBG adsorption and refining treatment. Thegreatest change in scattering coefficient, a 14% decrease, occurs between unrefined pulpwithout LBG and unrefined pulp with 1wt% LBG. Increasing refining level diminishedthe impact of LBG dosage on scattering coefficient (Figure 6.12f), indicating that refiningis the dominant factor for bonding. At 6000 rev and 9000 rev, the scattering coefficientsare independent of LBG dosage, thus indicating there is a finite degree to which bond-ing can be improved. The overall trends of scattering coefficient agree with the trend intensile strength with respect to refining and LBG adsorption. Consequently, NBSK papertensile strength enhancement is mainly due to bonding formation of LBG adsorption.1776.5 ConclusionLocust bean gum was studied as a strength additive to NBSK pulp. The kinetics andisotherms of the adsorption process were analyzed. The influence of temperature, time,salt addition and pH on amount of adsorption was explored. The effects of LBG adsorp-tion, refining and LBG dosage on pulp strength were assessed.The maximum fraction of LBG adsorption to NBSK pulp fibre was achieved in 10 min-utes. The adsorption rate followed pseudo-second-order kinetics, indicating chemisorp-tion is the rate determining step. The adsorption rate constant increased rapidly withtemperature from 25 oC to 45 oC, but the amount adsorbed at equilibrium decreased.The activation energy of LBG adsorption was approximately 99.33 kJ ·mol−1 with a pre-exponential factor of 4.73E+17 L·mol−1 ·min−1, which also suggests a chemisorption pro-cess.LBG adsorption at 25 oC and 35 oC was successfully modeled using the Langmuiradsorption model for LBG < 2.1 wt% of o.d. fibre, indicating monolayer, homogenousadsorption to a finite number of adsorption sites on the fibre surface. The maximumLBG adsorption capacity of NBSK pulp fibre was comparable to that of native starch topulp fibre. LBG adsorption is a reversible process with equilibrium rate constant lessthan one. The maximum adsorption capacity increases with higher temperature whilethe equilibrium constant decreases likely due to reduced LBG adsorption.Refining to 3000 rev increased the surface area of NBSK pulp leading to an increasedequilibrium constant and low LBG concentration in the aqueous phase. The maximumadsorption capacity is double that of unrefined NBSK pulp.Temperature negatively affected LBG adsorption at 25-80 oC. Low sodium chlorideaddition (less than 0.0125 mol·L−1) had little effect on adsorption, and high sodium chlo-ride addition (0.05-1 mol·L−1) resulted in slightly higher LBG adsorption. Adsorptionincreased slightly at pH 2-5. LBG adsorption is dependent on electrostatic forces, there-fore high salt addition at low pH facilitates adsorption.178Refining and LBG dosage increased NBSK paper tensile strength and burst strength.Tensile and burst strength plateaued when refining over 6000 rev, and strength gains weresmall for LBG dosage greater than 0.5 wt%. The effects of mechanical refining were muchgreater than that of LBG adsorption. However, addition of LBG enabled a reduction in re-fining revolutions to achieve a target tensile strength and preserved freeness. Tear index,brightness and scattering coefficient decreased, likely due to greater inter-fibre bonding.LBG adsorption enhances bonding formation of paper, leading to increasing NBSK paperstrength.6.6 Future Work1. The effects of pulp fibre consistency and agitation rate on mass transfer require clar-ification. Further isotherm and kinetic studies are needed.2. LBG adsorption conformation could be further characterized and experimental evi-dence of monolayer or multilayer adsorption will offer insights for mechanistic un-derstanding of the adsorption process.3. Further discussion of whether LBG can only access fibre surface or can diffuse intofibre pores is needed. Confocal microscopy could characterize diffusion throughfibre cross-section with fluorescent -labeled LBG.4. Broader range of pulp refining over 3000 rev could be tested as an influencing factorof LBG adsorption to better understand surface area relationship with adsorptionand could offer more understanding of paper strength variation with LBG dosagelevel.5. Adsorption of softwood hydrolysate to NBSK pulp requires investigation.(a) Due to the complexity of composition and low concentration of hemicellulosewithin hydrolysate, purification could be necessary prior to adsorption.179(b) Hydrolysate component charge is a critical property for adsorption, thus, char-acterization will be necessary to optimize the adsorption process.(c) Based on LBG adsorption results and assuming a negative charge on softwoodhemicellulose, the following condition is recommended for hydrolysate ad-sorption. A relatively low dosage level (<2wt%), lightly refined (<3000 rev)NBSK pulp at 25oC for 10 minutes with low fibre consistency (< 0.5 wt%), highagitation rate (> 150 r.p.m.) and acidic or neutral conditions (pH 2-7) withoutany salt addition.(d) Refining has the greatest influence on paper strength therefore it is suggestedto test hydrolysate adsorption to NBSK pulp at low refining level combinedwith low hydrolysate dosage to maximize paper strength. Since bonding isthe main reason for strength enhancement, the scattering coefficient test couldprovide bonding information, while the freeness test could provide comple-mentary drainage information.180Chapter 7Kraft Pulping of Softwood Chips withMild Hot Water Pre-Hydrolysis7.1 Introduction7.1.1 MotivationKraft pulping separates cellulosic fibres from wood by dissolving hemicellulose and ligninfrom fibre walls and middle lamella (Mimms et al. [185]). Lignin is fractionated intosmaller molecules of mono- or oligo-lignols and dissolves in black liquor due to cleavageof phenolic β-O-4 and α-O-4 linkages by presence of hydroxide and hydrosulfide ions(Arato et al. [11], Ek et al. [53], Mimms et al. [185]). Hemicellulose dissolves in blackliquor as complex sugar acids such as aldonic acids under oxidative conditions (Ek et al.[53], Sjostrom [263]).Carbohydrates consumes a large portion of alkali during kraft pulping. Alkali con-sumption occurs during both chip impregnation and the delignification process. Approx-imately 30-40% of total alkali is consumed during chip impregnation by neutralizationreactions with acidic groups from carbohydrates and is then unavailable for delignifica-tion reactions (Mimms et al. [185], Niskanen [197]). Delignification is limited during chip181impregnation (Ek et al. [53]). There are three delignification stages during kraft pulp-ing: initial, bulk and final stage. The initial delignification stage consumes an additional30-40% of total alkali mainly through carbohydrates degradation (20% of total carbohy-drates) and delignification (20% of total lignin) (Ek et al. [53], Niskanen [197]). Hemicel-lulose degrades more rapidly due to its amorphous structure while the crystallinity ofcellulose protects it from degradation (Mimms et al. [185]). Approximately 90% of totallignin dissolves during bulk delignification stage (Ek et al. [53]). Removal of the resid-ual lignin is very difficult and the required harsh conditions would lead to significantcarbohydrate loses (Ek et al. [53]).Extraction of hemicellulose prior to pulping could prove a source of valuable chem-icals without a major impact on energy generation. The concentrated black liquor com-poses of mainly lignin (19-30%), inorganic material (33%), and acids (hydroxy acids 17-23%, acetic and formic acid 8%) (Ek et al. [53]). The dissolved organic components ofblack liquor are burned in the chemical recovery furnace. It had been demonstrated thatthe heating value of dissolved lignin (≈27 MJ·kg−1) is double that of dissolved hemicellu-lose (≈13.6 MJ·kg−1) (Farhat et al. [61], Fisˇerova´ et al. [63], Lu et al. [171], Van Heiningen[305]). Further, pre-hydrolysis can also help reduce pulping energy requirements as dis-cussed below.7.1.2 Pre-hydrolysis with pulpingHot water hydrolysis or autohydrolysis has been applied as a pretreatment prior to pulp-ing to extract hemicellulose (Liu et al. [163, 165], Lu et al. [171], Yoon and Van Heiningen[329]). The removal of hemicellulose and lignin increases wood porosity and facilitatesubsequent pulping by increasing chemical mass transfer and reducing energy require-ments (Chen et al. [29], Liu et al. [165], Lu et al. [171], Mosier et al. [193]). It has beenshown that pre-hydrolyzed chips are more susceptible to pulping and bleaching, but re-finability and strength performance may be compromised (Fisˇerova´ et al. [63], Yoon and182Van Heiningen [329]). The drawbacks could be due to the absence of hemicelluloses,which contribute to fibre swelling (Yoon and Van Heiningen [329]).Previous research has shown that there is a two-stage response of lignin and cellu-lose during kraft pulping of pre-hydrolyzed wood (Smith [270]). Under mild hydrolysisconditions, a portion of lignin is removed from the wood matrix to increase porosity, de-creasing the pulping time required to reach a target kappa number (Smith [270]). Severehydrolysis conditions, such as high temperature or low pH, will cause lignin condensa-tion likely increasing the lignin recalcitrance in the wood matrix (Smith [270]) and makinglignin degradation more difficult resulting in longer pulping time. Similar trends werereported by other scholars studying pre-hydrolysis pulping (Borrega et al. [24], Lu et al.[171], Yoon and Van Heiningen [329]). The viscosity of pulp also exhibits a two-stagepattern. Mild pre-hydrolysis removes hemicellulose and low molecular weight celluloseresulting in higher pulp viscosity (Lu et al. [171]). Harsh conditions, however, causedegradation of cellulose and a significant loss of pulp viscosity (Borrega et al. [24], Vilaet al. [309]).A potential disadvantage of pre-hydrolysis is reduced fibre yield after pulping. Themore severe pre-hydrolysis conditions are, the lower the pulp yield (Borrega et al. [24],Smith [270], Yoon and Van Heiningen [329])]. Pre-hydrolysis ultimately influences pa-per strength. Tearing index increases or keeps constant with increasing severity of pre-hydrolysis (Helmerius et al. [93], Yoon and Van Heiningen [329]), and the increase coulddue to the reduced inter-fibre bonding. Tensile index and burst index decrease with ris-ing pre-hydrolysis severity, possibly due to the reduction of hemicellulose and weakenedinter-fibre bonding (Helmerius et al. [93], Liu et al. [163], Yoon and Van Heiningen [329]).After elemental chlorine free bleaching, however, the reduction of fibre yield and tensileindex are more comparable to control pulp without pre-hydrolysis, and the reason coulddue to full exposure of cellulose (Liu et al. [163]).From the literature it is clear that mild pre-hydrolysis conditions are preferable in183order to retain a portion of hemicellulose, minimize cellulose degradation and maintainpulp quality. In this paper, the effects of hot water hydrolysis of softwood chips at 140 oCfor one hour prior to kraft pulping is explored.7.1.3 Particle size effects on hydrolysis and pulpingAs discussed in Chapter 3 (hydrolysis of softwood), small biomass particle size and lowsolid loading facilitate mass transfer during hydrolysis and favor the dissolution of biomasssolids. Low temperature with long residence time favors high molar mass oligomer pro-duction from large biomass particles. High molar mass oligomers can also be producedfrom small biomass particles by employing high temperature with a short residence time.Particle size is an important factor influencing pulping process. Thickness is the masstransfer limiting dimension during chip impregnation and delignification of pulping pro-cess (Niskanen [197]). It is more difficult to achieve homogeneous white liquor impreg-nation of thick chips resulting in undercooking and non-uniform pulp properties. Thinchips typically overcook resulting in compromised fibre properties.After impregnation, mass transfer within wood chips during pulping is mainly bydiffusion (Grace and Malcolm [78]). The rate of diffusion is anisotropic and depends onliquor pH (Grace and Malcolm [78]). At low pH, diffusion in the longitudinal direction ismuch faster than in the radial or tangential directions. However, under basic conditions,diffusion rates are similar in all directions (Grace and Malcolm [78]) due to increasedporosity of the fibre wall. Under typical kraft pulping conditions the pH is greater than 13therefore the diffusion rate is independent of direction and chip thickness is the limitingfactor for mass transfer. In this study, chip thickness was varied from less than 2 mm togreater than 6 mm in order to assess the effect of chip thickness on pre-hydrolysis andkraft pulping.1847.2 Goals and HypothesesPre-hydrolysis kraft pulping is not new (Liu et al. [163, 165], Lu et al. [171], Yoon andVan Heiningen [329]), but the size effects of chips during pre-hydrolysis and in subse-quent processes are poorly understood. Chip size, a crucial pulping parameter, influencesdelignification, post-pulping fibre separation and subsequent fibre properties. Details ofhydrolysate and final fibre composition are also missing from literature reports and thisinformation is needed for process design, product control and economic assessment ofbiorefining opportunities.In this study, softwood chips with thickness varying from less than 2 mm to greaterthan 6 mm were hydrolyzed at 10 wt% consistency at 140 oC for 1 hr to produce highyields of large hemicellulose oligomers. The pre-hydrolyzed chips were then subjected tokraft pulping. Final fibre composition and pulp properties were evaluated.Hemicellulose, as an individual component of wood, has been primarily viewed as alow-grade fuel within black liquor. It is easily removed by hydrolysis and kraft pulpingbut its important role in final fibre properties is not thoroughly understood. The followinganalysis identifies the major challenges to pre-hydrolysis implementation and suggests amore promising direction for integration of pre-hydrolysis with kraft pulping.The following hypotheses will be tested:1. Softwood chip thickness influences yield and size of hemicellulose oligomers pro-duced during pre-hydrolysis.2. Pre-hydrolysis will affect kraft pulping process variables such as alkali usage andfibre yield.3. Pulp fibre properties will change as a result of combined pre-hydrolysis and kraftpulping.1857.3 Experimental Design7.3.1 MaterialSoftwood chips, a mixture of spruce, pine and fir, were manually screened to removeknots and other contaminates. All chips were then screened using a 16 mm round holetray. A portion of the screened chips were reserved for a limited number of tests to mimicindustrial conditions and will be referred to as the blend group. Chips were next sepa-rated by thickness: Fines (under 2 mm), 2 mm (2-4 mm), 4 mm (4-6 mm), and 6 mm (over6 mm). All chips were air dried and the moisture content was determined by NREL LAP([86]).7.3.2 Pre-hydrolysisHot water pre-hydrolysis experiments were conducted in 400 mL batch reactors at theCanfor Pulp Innovation Centre in Burnaby, British Columbia. A wire mesh was used tocompress the chips to the bottom of the reactor so that they were completely submergedin the liquor. Final slurry volume was 377 mL and the solids consistency was 10 wt% forall particle sizes. Pre-hydrolysis and kraft pulping condition are summarized in Table 7.1.The blend group materials were subjected to pre-hydrolysis at 10 wt% consistency. Eachexperiment was carried out in duplicates.The batch reactors were submerged in a 20 L digester with a hot water circulationsystem. The isothermal reaction time was 1 hour at 140 oC. It took approximately 2 hoursto heat the circulating water and digester from 20 oC to 140 oC. The time to cool the systemto 20 oC was approximately 30 min. The slurry was filtered using 20 µm porous filterpaper. The hydrolysate was stored at 4 oC for further analysis and the pretreated chipswere air dried and moisture content was determined in preparation for kraft pulping.186Table 7.1: Experimental conditions for softwood chip pre-hydrolysis and kraft pulp-ing.Particlesize (mm)Consistency(%)Pre-hydrolysisTemperature(oC)Pre-hydrolysisH factorKraft pulpingH-factorFines 10 140 76-78 16002 10 140 76-78 16004 10 140 76-78 16006 10 140 76-78 16002-6 blend 10 140 76-78 1600Fines 10 N.A. N.A. 16002 10 N.A. N.A. 16004 10 N.A. N.A. 16006 10 N.A. N.A. 16002-6 blend 10 N.A. N.A. 1600Fines 10 N.A. N.A. 19002 10 N.A. N.A. 19004 10 N.A. N.A. 19006 10 N.A. N.A. 19002-6 blend 10 N.A. N.A. 19001877.3.3 Hydrolysate compositional analysisRefer to Section 3.3.3.Monomer concentration was analyzed by Dionex AS50 HPLC (Thermo Scientific) cou-pled with an ion exchange PA1 column (Dionex), an ED50 electrochemical detector (pulsedamperometric detector) with a gold electrode, and an AS50 autosampler (Dionex). De-ionized water was used as eluent with a flow rate of 1 mL·min−1. The auxiliary pumpadded 0.2 M NaOH at 0.5 mL·min−1. The samples were filtered through a 0.22 µm nylonsyringe filter before injection. The injection volume was 10 µL.7.3.4 Hydrolysate molar mass determinationRefer to Section 4.3.7.3.5 Kraft pulpingKraft pulping was conducted using the same reactor system used for pre-hydrolysis.The liquor to solid ratio was 6.5 with 18 wt% effective alkali (EA). Chips were soakedovernight in white liquor before pulping.Two H-factors (Equation 3.4) were tested at 170 oC for kraft pulping: 1600 and 1900.Raw chips of same particle size and consistency were also pulped under the same condi-tions for comparison to pre-hydrolyzed chips.The cooked chips were washed by de-ionized water to remove residual alkali andtransferred to a disintegrator (British pulp evaluation apparatus by Mavis EngineeringLtd.) for 10 min (6000 revolutions) to separate the fibres. The fibre suspension was filteredwith Nylon filter wire. The fibre pad was washed with de-ionized water and oven-driedto determine the fibre pad weight.Oven-dried pulp fibres were blended in the disintegrator for 5 min. The fibre suspen-sion was transferred to the pulp screening machine (Voith, Inc.) with continuous waterinput. After all fibres passed through the screen to the bottom tray, residual shives on the188top screen (0.10 + 0.005/-0.01 mm) were collected, dried and massed to determine ovendried weight of rejects. All screened fibres were collected for further analysis.Fibre yield is defined as the pure fibre fraction to the total o.d. unprocessed, raw chipsinput before pre-hydrolysis/kraft pulping as Equation 7.1.Fibre yield =Fibre pad o.d.− Reject o.d.Initial o.d. chips× 100% (7.1)7.3.6 Kappa numberKappa number, used to indicate extent of delignification and bleaching requirement, isdefined as the volume of permanganate solution consumed by a controlled mass of pulp(Mimms et al. [185], Smook et al. [271]). In this paper, kappa number was determinedaccording to Tappi method T236 om-99.Klason lignin content (Xk) has a linear relationship with kappa number for pulps withless than 70% yield (Mimms et al. [185], Smook et al. [271]) as Equation 7.2:Xk = 0.147×Kappa number (7.2)7.3.7 Fibre compositional analysisFibre composition was determined using a modified alkaline hydrolysis procedure (Zhaoet al. [338]). One gram of air-dried pulp was mixed with 10 mL 17.5% NaOH and keptin a 20 oC water bath for 45 minutes. 50 mL de-ionized water was then added and theslurry was filtered in a 30 mL medium-porosity fritted glass crucible until at least 20-30mL liquor was recovered. This liquor was retained for further analysis. Then the solidfibre was sequentially washed in the crucible with 9.5% sodium hydroxide (NaOH), a189large amount of de-ionized water, and finally, 2 mol·L−1 acetic acid.The crucible and solid fibre pad were dried in the oven at 105 oC to determine fibremoisture content. The oven dried sample was weighed (Xd) to determine percentage ofalpha-cellulose content in pulp (Xα). The percentage of alpha-cellulose content in initialo.d. pulp samples (Xi) was calculated according Equation 7.3 and Equation 7.4. Xresidualis residual lignin content of pulp fibre after post-hydrolysis determined as Equation 7.4.Xα =Xd − XresidualXi× 100% (7.3)The carbohydrate content of 15 mL of the alkaline filtrate was determined. The pHof the liquor was slowly acidified and the amount of added acid was recorded. A 10 mLsample of the pH adjusted liquor was hydrolyzed in an autoclave at 121 oC for 1 h todetermine the monomer sugar content according to NREL liquid compositional analysis(Sluiter et al. [265]). Acid soluble lignin content (Xs) was determined by UV-vis spec-troscopy NREL biomass compositional analysis (Sluiter et al. [266]) and used to determinethe alpha-cellulose content.The residual lignin content of pulp fibre after post-hydrolysis and the acid solublelignin test, was:Xresidual = Xk − Xs (7.4)where Xk is Klason lignin content predicted from kappa number (Equation 7.2) and Xs isacid soluble lignin determined by UV-Vis spectroscopy.1907.3.8 Pulp propertiesThe water retention value (WRV) was measured according to SCAN-C 62:00. All the fi-bre properties were analyzed by PulpEye, which is an online pulp fibre analyzer fromEurocon Analyzer AB coupled with a camera speed of 60 frames per second. Table 7.2summarizes the properties measured in this study.Table 7.2: Fibre property definition and parameter description tested by PulpEye.Fibre property Parameter DefinitionFibre length (mm) Length-weighted average > 0.2mmFibre width (µm) Mean value of the fibre widthAspect ratio Length/width Ratio of fibre length to widthKinks Kinks/mm Number of local deformation per mmFines (%) Fines percent Fibre length 6 0.2mm7.3.9 Effective alkaliThe EA and sulfide in white liquor and black liquor were determined by T 624 cm-00and T625 wd-99. White liquor is a solution of sodium hydroxide (NaOH) and sodiumsulfide (Na2S), and some non-functional sodium carbonate (Na2CO3). The EA (NaOH+ 1/2Na2S), active alkali (NaOH + Na2S) and sulphidity (Na2S/[NaOH + Na2S] × 100)were determined for the white liquor before the kraft pulping. Determining these met-rics is challenging and time-consuming therefore only a limited number of samples couldbe tested. At the advice of Canfor Pulp personnel, EA consumption and sulphidity con-sumption of blended chips were determined. Blended chips were selected because CanforPulp screens softwood chips to 2-6 mm thickness in their industrial operations.1917.4 Results and Discussion7.4.1 Pre-Hydrolysis7.4.1.1 Hemicellulose yieldHemicellulose oligomer and monomer yield are shown in Figure 7.1. The maximum yieldof oligomer is produced from the Fines and oligomer yield decreases with increasingparticle size. This is consistent with previous studies that small particle size favors higholigomer yield (Rissanen et al. [227, 228], Song et al. [273]). Monomer yield, by contrast,is constant with varying particle size suggesting that monomer production is unaffectedby mass transfer limits within wood chips.Fines 2mm 4mm 6mm0.00.51.01.52.02.53.0 Monomer yield Oligomer yieldYield of o.d. chips (%)Particle sizeFigure 7.1: Hemicellulose yield from o.d. chips as a function of particle size.Hemicellulose oligomer weight-average molar mass decreases slightly with particlesize (Figure 7.2). Small particle size from Fines to 4 mm showed comparable weight-average molar mass of 5.5-5.8 kDa (Figure 7.2). The oligomer molar mass from largeparticles (6 mm) is lower than Fines, and only slightly lower than 4 mm. Overall, theoligomer weight-average molar mass decreases only slightly with particle size.192Fines 2mm 4mm 6mm2.02.53.03.54.04.55.05.56.06.5 MwWeight-average molar mass (kDa)Particle sizeFigure 7.2: Weight-average molar mass of hemicellulose oligomers as a function ofparticle size.Characterization and calculation methods described in Chapter 4 were used to de-termine oligomer yield by DP interval as described in Chapter 5 (Figure 7.3). The yieldof degradation products was independent of particle size. The yield of small oligomers,DP1-20 to DP41-60, from Fines was slightly higher as thickness was varied from 2 to6 mm. The yield of large oligomers, DP61-80 until DP121-140, decreased more clearlywith increasing particle size. The low diffusivity of large oligomers compounded withincreased chip thickness results in decreased yield of large oligomers. The yield of pre-hydrolyzed solids, Figure 7.3e, mirrors the trends found with the dissolved hemicellu-loses. The yield is lowest for the Fines and increases slightly with increasing particle size.193Fines 2mm 4mm 6mm0.000.010.020.030.040.050.06 Degradation products DP1-20Yield of hemicelluloseParticle sizeFines 2mm 4mm 6mm0.000.010.020.03 DP21-40 DP41-60Yield of hemicelluloseParticle sizeFines 2mm 4mm 6mm0.0000.0020.0040.0060.0080.0100.012 DP61-80 DP81-100Yield of hemicelluloseParticle sizeFines 2mm 4mm 6mm0.0000.0010.0020.0030.0040.005(e)(d)(b)(c) DP101-120 DP121-140Yield of hemicelluloseParticle size(a)Fines 2mm 4mm 6mm0.70.80.91.01.1 Pre-hydrolyzed solidsYield of hemicelluloseParticle sizeFigure 7.3: Yield of hemicellulose oligomers in the hydrolysate varying with particlesize (a) degradation products and DP 1-20, (b) DP21-40 and DP41-60, (c) DP61-80 and DP 81-100, (d) DP 101-120 and DP 121-140, and (e) pre-hydrolyzedsolids.7.4.1.2 Hemicellulose composition in hydrolysateParticle size effects on arabinose, xylose, galactose, glucose and mannose yield of hy-drolysate is negligible (Figure 7.4). This result further proves that monomer generationis independent of particle size. Arabinose has the greatest yield and thus is the most re-active hemicellulose component, as discussed in Chapter 3. Glucose and mannose arethe most stable hemicellulose component as demonstrated by the low monomer yields inFigure 7.4.194Fines 2mm 4mm 6mm-0.010.000.010.020.030.040.30.40.50.60.70.80.9   Arabinose Galactose Glucose Xylose MannoseMonomer yield of o.d. chips (%)Particle sizeFigure 7.4: Hydrolysate hemicellulose monomer yield from o.d. chips (arabinose,galactose, glucose, xylose and mannose) as a function of particle size.Oligomer yield by sugar species was influenced by particle size (Figure 7.5). Smallparticles favored oligomer production, regardless of sugar type, reflecting total oligomeryields (Figure 7.1). Unlike monomer production, mannan and galactan are present in thegreatest amount. This is a result of GGM being the primary hemicellulose in softwoodand the relatively high recalcitrance of the glucomannan backbone.Fines 2mm 4mm 6mm0.00.10.20.30.40.50.60.7   Arabinan Xylan GalactanOligomer yield of o.d. chips (%)Particle size(a)Fines 2mm 4mm 6mm0.00.10.20.30.40.50.60.7   Glucan MannanOligomer yield of o.d. chips (%)Particle size(b)Figure 7.5: Hydrolysate hemicellulose oligomer yield of o.d. chips a) arabinan, xylanand galactan b) glucan and mannan as a function of particle size.1957.4.1.3 Pre-hydrolyzed chips compositionThe solubilized fraction of pre-hydrolysis from softwood chips are listed in Table 7.3.The solubilized solids fraction from initial raw softwood chips decreases with increasingparticle size. The Fines lose twice the amount of material lost from the other particle sizes.The differences in degree of solubilization among 2 mm, 4 mm and 6 mm are small. FromFigure 7.1, the hemicellulose yield of Fines is less than 1.5 times of yield from 4 mm and6 mm, indicating that other components, such as extractives and lignin, are more readilyremoved during pre-hydrolysis of Fines.Table 7.3: Fraction of solubilized solids after pre-hydrolysis and composition of pre-hydrolyzed chips (% o.d. chips, average ± standard deviation). Note: no stan-dard deviation for Fines due to inaccurate operation for one replicate.Untreated chips Fines 2mm 4mm 6mmFraction solubilized N.A. 5.84 2.66 2.59 2.34Arabinan 0.44 ± 0.04 0 0.00 ± 0 0.00 ± 0 0.00 ± 0Galactan 1.65 ± 0.05 1.86 1.90 ± 0.02 1.88 ± 0.01 1.64 ± 0.03Glucan 49.40 ± 0.78 48.13 50.86 ± 0.11 49.49 ± 0.2 49.18 ± 0.38Xylan 2.11 ± 0.55 1.48 1.61 ± 0.20 1.85 ± 0.45 1.50 ± 0.12Mannan 9.19 ± 0.47 6.29 7.74 ± 0.15 7.56 ± 0.03 8.22 ± 0.27Acid soluble lignin 5.26 ± 0.43 5.1 5.19 ± 0.10 5.11 ± 0.15 5.38 ± 0.11Acid insoluble residue 26.16 ± 0.14 27.83 28.69 ± 0.13 27.46 ± 1.71 26.40 ± 0.51Extractive 3.04 ± 0.03Total 97.24 ± 2.49 96.45 98.45 ± 0.31 94.76 ± 1.78 94.48 ± 1.45The composition of raw and pre-hydrolyzed chips is summarized in Table 7.3. Themost striking difference is the disappearance of arabinan from the pre-hydrolyzed chips,which indicates that arabinan is the most reactive hemicellulose component within soft-wood. Mannan and xylan content decreased for all particle sizes. Glucan, which includes196cellulose and hemicellulose-derived glucan, was constant with respect to particle size andthe amount was comparable to the raw chips, which implies that the selected hydrolysiscondition does not substantially degrade glucan. The total carbohydrates content is con-sistent with oligomer and monomer yields reported in Figure 7.4 and Figure 7.5.7.4.2 Kraft pulpingIn the subsequent discussion, the abbreviation PCC-1600 represents pre-hydrolyzed chipscooked to an H-factor of 1600 (PCC-1600); RCC-1900 is raw chips cooked to an H-factorof 1900 (RCC-1900); RCC-1600 is raw chips cooked to an H-factor of 1600 (RCC-1600).7.4.2.1 Fibre yield and kappa numberFigure 7.6a presents the screened fibre yield (Equation 7.1) as a function of particle sizefor PCC-1600, RCC-1600 and RCC-1900 while Figure 7.6b presents kappa number.The screened fibre yield of RCC-1600 was greatest for all chip sizes (Figure 7.6a). RCC-1900 screened fibre yields were 0.7-2.0% lower and screened fibre yields from PCC-1600were lower still. As the discussion below will demonstrate, these differences are due tovariation in hemicellulose and lignin removal.There are two key features to note in Figure 7.6b. The first is that there is a slightincrease in kappa number of RCC-1600-4 mm followed by a larger increase in kappanumber of RCC-1600-6 mm indicating increasing lignin content. Interestingly, the kappanumber of the RCC-1900 and PCC-1600 series are comparable and show little variationwith chip thickness.197Fine 2mm 4mm 6mm4243444546474849Screened fibre yield of o.d. chips (%)Particle size PCC-1600 RCC-1600 RCC-1900(a)Fine 2mm 4mm 6mm24283236404448Kappa numberParticle size PCC-1600 RCC-1600 RCC-1900(b)Fines 2mm 4mm 6mm0.00.20.40.60.81.0Reject of o.d. chips (%)Particle size PCC-1600 RCC-1600 RCC-1900(c)Figure 7.6: (a) Screened fibre yield from o.d. raw chips, (b) kappa number of kraftpulp, and (c) percent rejects after pulp screening relative to o.d. chips as afunction of particle size. Pulping was conducted to an H-factor of 1600 and1900.The percentage of screened rejects (Figure 7.6c) exhibits similar behavior to kappanumber. Rejects occur when pulped wood chips are not fully separated into fibres bythe disintegrator. There is an increase in the percentage of rejects for RCC-1600-4 mmand a second increase in the percentage of rejects for RCC-1600-6 mm. The percent ofrejects from RCC-1900 is negligible for all sizes except RCC-1900-6 mm. The percentageof rejects from RCC-1900-6 mm chips is equal to the percentage of rejects from RCC-1600-6 mm. Finally, there is a negligible percentage of rejects from PCC-1600 treated chips atall sizes.198The differences in RCC-1600 and RCC-1900 will be discussed first so that the effectsof pre-hydrolysis relative to increasing pulping severity can be delineated. IncreasingH-factor from 1600 to 1900 causes an increase in lignin removal and thus a decrease inscreened fibre yield. Increasing H-factor also reduces the effect of chip size on the per-centage of rejects. The low percentage of rejects at an H-factor of 1900 for chips smallerthan 6 mm is due to better delignification and indicates complete pulping of the woodchips. The percentage of rejects was equal for chips with a thickness greater than 6 mmat H-factor of 1600 and H-factor of 1900. This equally high percent of rejects suggestsmass transfer limitations during pulping resulted in uneven cooking. Pulping under theconditions of RCC-1600 is undesirable despite the high yield because it resulted in a highkappa number and was accompanied by a high percentage of rejects. Canfor Pulp typi-cally targets kappa number lower than 30 for kraft pulping.The kappa numbers and lignin content of PCC-1600 and RCC-1900 are comparablealthough the pre-hydrolyzed chips were subjected to a milder pulping process. Despitesimilar lignin content, the screened fibre yield of PCC-1600 is 2.0-2.5% lower than thatof RCC-1900. This difference can be attributed to the hemicellulose removal during pre-hydrolysis and this is verified by the equivalence of the RCC-1900 fibre yield to the sumof pre-hydrolysis hemicellulose yield and PCC-1600 fibre yield.Hemicellulose removal during pre-hydrolysis will have two effects during pulping.During pulping, hemicellulose is converted to organic acids which then neutralize al-kali thus reducing its availability for delignification reactions (Mimms et al. [185], Niska-nen [197]). This neutralization effect is reduced by removing hemicellulose during pre-hydrolysis. Hemicellulose removal also increases wood chip porosity (Chen et al. [29], Luet al. [171], Mosier et al. [193]), which improves chemical penetration during the subse-quent pulping step. These effects explain why pulping pre-hydrolyzed chips to an H-factor of 1600 results in the same kappa number as pulping raw chips to an H-factor of1900. As discussed in the introduction, there is a two-stage response to pre-hydrolysis199severity: enhanced lignin removal resulting in a decrease in H-factor to reach a targetkappa number and condensation of lignin resulting in the need to increase H-factor toreach a target kappa number. The equal kappa numbers for PCC-1600 and RCC-1900indicates that the selected pre-hydrolysis condition is sufficiently mild that lignin con-densation is minimal. Hemicellulose removal during pre-hydrolysis and enhanced delig-nification during pulping also explains why the screened fibre yields from PCC-1600 are2.8-4.0% less than those from RCC-1600. Similar observations have been reported byothers (Borrega et al. [24], Smith [270], Yoon and Van Heiningen [329]). Pre-hydrolysismakes it possible to decrease the H-factor required to meet a target kappa number whilehemicellulose removal is simultaneously increased. Energy and chemical savings may beattainable through reductions in H-factor or in application of pulping chemicals.Unlike the raw chips, the percentage of rejects produced from pre-hydrolyzed pulpwas negligible at all thicknesses. This is noteworthy as the residual lignin content ofRCC-1900-6 mm is only slightly higher than that of PCC-1600-6 mm but the percentageof rejects from RCC-1900-6 mm is more than 27 times higher. The low percentage ofrejects could relate to the greater porosity created during pre-hydrolysis and subsequentfacilitation of mass transfer during kraft pulping (Chen et al. [29], Lu et al. [171], Mosieret al. [193]).7.4.2.2 Fibre carbohydratesAlpha cellulose is composed of long chains of cellulose and is resistant to strong alka-line treatment (Smook et al. [271]). From Figure 7.7a, RCC-1600 has the highest alpha-cellulose content of the three materials. Pre-hydrolysis (PCC-1600) and high H-factorpulping (RCC-1900) decrease the alpha cellulose content of the final pulp relative to RCC-1600.200Fine 2mm 4mm 6mm323436384042444648  Alpha cellulose of o.d. chips (X, %)Particle size PCC-1600 RCC-1600 RCC-1900(a)Fine 2mm 4mm 6mm1.01.52.02.53.03.54.04.55.0(b)  Total soluble hemicellulose of o.d. chips (%)Particle size PCC-1600 RCC-1600 RCC-1900Figure 7.7: (a) Alpha-cellulose content of screened pulp fibre (wt%) and (b) total sol-uble hemicellulose content (wt%) relative to o.d. chips as a function of particlesize. Total soluble hemicellulose is the sum of soluble arabinan, xylan, galactan,mannan and glucan.Particle size did not affect the alpha-cellulose content of PCC-1600 and RCC-1900.This indicates that the mass transfer limitations during pre-hydrolysis and pulping donot influence cellulose content. The fibre from both conditions is sufficiently delignifiedfor all particle sizes. This conclusion agrees with kappa number (Figure 7.6b). The trendis less clear for RCC-1600 due to the size of error bars.The total soluble hemicellulose content, the sum of soluble galactan, glucan, mannan,xylan and arabinan, of RCC-1600 is unexpectedly equal to that of PCC-1600 (Figure 7.7b)as it was demonstrated in Figure 7.1 that hemicellulose is removed during pre-hydrolysis.It is also unexpected that the soluble hemicellulose content of RCC-1600 is less than thatof RCC-1900 (Figure 7.7b) as increasing H-factor will increase hemicellulose removal (Eket al. [53]). These observations are most likely an artefact of the methods used to pre-pare the samples for analysis. Residual lignin may tether hemicellulose through LCC inthe fibre during the alkaline extraction process. The elevated lignin content of RCC-1600may have hindered complete hemicellulose release from the fibre during sample prepa-ration. This restriction on hemicellulose release may be due to shielding of hemicellulosefrom sodium hydroxide and by making the hemicellulose insoluble through shared LCC201bonds.The decrease in soluble hemicellulose content of PCC-1600 relative to RCC-1900 isdue to the difference in acidic and alkali hydrolysis. Pre-hydrolysis removes hemicellu-lose and lignin from wood chips and increases wood chip porosity (Chen et al. [29], Hsuet al. [99], Lu et al. [171], Mosier et al. [193]) to enhance alkali diffusion and fibre libera-tion. High H-factor produces pulp with high delignification and readily separated fibres.Alkali conditions are more selective for delignification reactions while acidic conditionsare more selective for hemicellulose hydrolysis. By combining an acidic hemicelluloseremoval process with an alkali delignification process, it possible to produce a pulp withlignin content equal to RCC-1900 (Figure 7.6) but with lower soluble hemicellulose con-tent.There is little variation in hemicellulose content with chip size for all three pulps whichimplies even hemicellulose release during pulping and compositional analysis.7.4.2.3 Fibre propertiesFibre dimension and fines contentFibre length (length-weighted average) (Figure 7.8a) was strongly influenced by chipthickness. Pulping chip Fines produced 2.35-2.45 mm long fibres. Longer fibres were pro-duced from the thicker chips. Increasing chip thickness increased fibre length for RCC-1600 and RCC-1900; the longest fibres were produced by pulping to an H-factor of 1900.Pre-hydrolysis resulted in 0.04-4.15% shorter fibres relative to kraft pulping at the sameH-factor of 1600. Fibre length is a key predictor of paper strength therefore implementa-tion of pre-hydrolysis may not be favorable for the production of high-strength pulp.202Fines 2mm 4mm 6mm2.32.42.52.62.72.8  Fibre length (mm)Particle size PCC-1600 RCC-1600 RCC-1900(a)Fines 2mm 4mm 6mm262728293031  Fibre width (m)Particle size PCC-1600 RCC-1600 RCC-1900(b)Fines 2mm 4mm 6mm768084889296  Fibre aspect ratio (%)Particle size PCC-1600 RCC-1600 RCC-1900(c)Fines 2mm 4mm 6mm1.01.52.02.53.03.5(d)  Fines percent (%)Particle size PCC-1600 RCC-1600 RCC-1900Figure 7.8: Dimensions of kraft pulp fibres: (a) mean fibre length (mm) (b) mean fibrewidth (µm) (c) aspect ratio, and (d) kraft pulp fines content (wt%) as a functionof particle size.Average fibre width was independent of wood chip size for all conditions (Figure 7.8b).In addition, increasing H-factor from 1600 to 1900 had little effect on fibre width. How-ever, application of pre-hydrolysis resulted in an approximately 3.78-7.62% decrease infibre width relative to kraft pulping (Figure 7.8b). This reduction could relate to the re-moval of hemicellulose and lignin through pre-hydrolysis and subsequent kraft pulping.The fibre aspect ratio is defined as fibre length to fibre width and is presented in Fig-ure 7.8c. Pre-hydrolysis generated more slender pulp fibre, and the PCC-1600 aspect ratiowas higher than RCC-1600 (Figure 7.8c). The near parallel trends of PCC-1600 and RCC-1600 indicate that pre-hydrolysis does not change the effect of particle size on fibre aspect203ratio. The decreased aspect ratio of RCC-1600 paired with the comparable fibre lengthsof PCC-1600 and RCC-1600 indicates that the effects of pre-hydrolysis on the fibre mainlyoccurred in radial direction of the fibre wall.The fines content refers to the fraction of fibres with length equal to or less than 0.2mm. With increasing particle size, fines content decreased for RCC-1600 and RCC-1900(Figure 7.8d). Thin wood chips produces more fines compared to thick chips. Betterfibre separation due to high delignification at H-factor of 1900 resulted in less fibre dam-age and fines generation therefore RCC-1900 had the lowest fines content. The residuallignin of RCC-1600 hampers fibre separation resulting in maximum fines content. Thepre-hydrolyzed pulp had an intermediate content of fines reflecting the balance of en-hanced delignification but diminished width and cell wall thickness.KinksA kink is defined as the abrupt change in the axial direction of fibre curvature thatdoes not naturally exist (Niskanen [197]). As weak points in the fibre, kinks reduce pa-per tensile strength and elastic modulus but increase stretch by uneven bearing of tensilestress compared to paper formed by straight fibres (Johansson [112]). The pre-hydrolyzedpulp fibres had the greatest number of kinks per unit length (Figure 7.9). As discussedpreviously, hydrolysis mainly occurs in radial direction. The reduction of fibre widthmakes fibre more flexible and likely provokes subsequent mechanical treatment to gen-erate more kinks. Under similar mechanical forces, pre-hydrolyzed kraft pulp fibres areeasily damaged compared to traditional kraft pulp fibre.204Fines 2mm 4mm 6mm0.120.140.160.180.200.220.240.26  Kinks per mmParticle size PCC-1600 RCC-1600 RCC-1900Figure 7.9: Kraft pulp fibre kinks number per mm as a function of particle size.Water retention valueWRV indirectly describes the degree of swelling of pulp fibres by measuring the waterretained by the fibre wall after centrifugation (Niskanen [197]). One of the central goalsof refining is to enhance swelling by creating internal and external fibrillation (Niskanen[197]). The fibre wall structure delaminates leading to a loosened structure that can holdmore water. The fibre surface becomes hairier which promotes inter-fibre contact andbonding. The resulting fibre fragments and fibrils also contribute to increased swelling bycreating extended surface area. WRV is also influenced by chemical composition (Niska-nen [197]). Hydrophilic hemicellulose helps to retain water within the fibre wall whilehydrophobic lignin inhibits water retention. Hemicellulose removal may enhance inter-nal fibrillation by increasing porosity that can retain water after centrifugation, make thefibre more flexible and enhance inter-fibre bonding. However, excessive hemicelluloseremoval may result in reduced swelling due to excessive loss of hydrophilicity.RCC-1600 has the greatest lignin content (Figure 7.6b) and this likely explains whyRCC-1600 pulps have the lowest WRV (Figure 7.10). The lignin content of PCC-1600 andRCC-1900 is comparable therefore hemicellulose content and fibre wall structure must be205considered to explain differences in WRV. PCC-1600 has a lower hemicellulose contentbut a greater WRV than RCC-1900. It is likely that PCC-1600 pulp fibres have greaterporosity (Figure 7.10).Fines 2mm 4mm 6mm0.00.51.01.52.02.53.03.5  Water retention value (g/g)Particle size PCC-1600 RCC-1600 RCC-1900Figure 7.10: WRV of kraft pulp as a function of particle size.7.4.2.4 Practical implicationsDue to testing constraints, chemical consumption is only reported for the blended soft-wood chips. EA is the summation of sodium hydroxide plus half of sodium sulfide ex-pressed based on the equivalent amount of sodium oxide (Section 7.3.9). Pulping rawchips resulted in more than 95% of EA being consumed (Figure 7.11), this was indepen-dent of H-factor. Pre-hydrolysis reduced the EA consumption by approximately 9%. Thisdifference is likely due to diminished neutralization by acid groups present in wood (Gul-lichsen and Fogelholm [83], Yoon and Van Heiningen [329]).Sulphidity is the ratio of Na2S to active alkali (Section 7.3.9). Sodium sulfide facilitatesdelignification but has little effect on carbohydrates dissolution (Ek et al. [53], Mimmset al. [185]). Decreasing H-factor from 1900 to 1600 decreased sulphidity consumptionby 27.4% by decreasing residence time (Figure 7.11). PCC-1600 consumed approximately38% and 55% less sulphidity relative to RCC-1600 and RCC-1900, respectively. The re-206moval of lignin during pre-hydrolysis reduces sodium sulfide consumption. The reduc-tion relative to RCC-1900 is notable as the kappa numbers of PCC-1600 and RCC-1900pulps were comparable.EA Sulphidity020406080100120Consumption (%)Blend 10% PCC-1600 RCC-1600 RCC-1900Figure 7.11: EA consumption and sulphidity consumption of black liquor from blend10wt% softwood chips.The effects of pre-hydrolysis on subsequent kraft pulping and fibre properties aresummarized in Table 7.4. The main advantages are reduction of alkali consumption dur-ing pulping, kappa number reduction of kraft pulp fibre, and increased WRV. However,pre-hydrolysis results in lower pulp yield, reduced fibre dimensions, and increased kinks,which could lead to fibre strength loss. It is not possible to predict how final paper proper-ties will change due to contradictory influences of reduced fibre dimension and increasedWRV and there was insufficient fibre to prepare handsheets for direct testing. If strengthis an essential feature for the final pulp product, pre-hydrolysis is not recommended dueto potential fibre and paper strength losses.207Table 7.4: Summary of pre-hydrolysis combination to kraft pulping Pros and Consand particle size influence on pre-hydrolysis and kraft pulping.Pre-hydrolysisPros ConsSave EA Lower fibre yieldReduce kappa Reduced fibre dimensionIncreased WRV Increased kinksParticle size influencePre-hydrolysis Kraft pulpingNo influence on monomer yield Fibre yield little impactSmaller particle size, higheroligomer yield and molar massSmaller particle size, reduced fibre length,kappa number and reject, increased fines con-tentLittle influence on fibre width and kinksParticle size influence on pre-hydrolysis and kraft pulping are summarized in Ta-ble 7.4. Smaller particle size favors hemicellulose oligomer yield and higher molar mass,reduces kappa number, but also reduces fibre dimension and increases fines content.Thus, chip fines are beneficial for pre-hydrolysis and hemicellulose production, and thepre-hydrolyzed solids could be used for combustion. Large particle size chips are favor-able for pulp production.The main challenges associated with integrating pre-hydrolysis with kraft pulping arepulp yield reduction, modification of fibre property and potential loss of paper strength.It is critical to balance pre-hydrolysis severity and final fibre quality. Compared to severepre-hydrolysis with high hemicellulose yield but low pulp quality, mild pre-hydrolysisat relatively low temperature (lower than typical autohydrolysis temperature 160-200oC)208is more favorable to this purpose, and under the conditions applied in this work, alkaliconsumption and kappa number of pulp were reduced without serious loss of fibre qual-ity. Further economic analysis is needed to evaluate the differences in profits lost throughlower pulp yield and the value of hemicellulose obtained from the hydrolysates. If pulpproducts are sensitive to the strength variation, pulp mill waste streams such as chip finesand hog fuel could be an alternative feedstock. The extracted hemicellulose could furtherbe used as a strength additive to NBSK pulp.Pre-hydrolyzed kraft pulp fibre might be better utilized as micro-fibrillated cellulosedue to the reduced dimension and increased WRV. Micro-fibrillated cellulose is a mix-ture of varying scales of cellulose components including cellulose nanocrystals, cellu-lose nanofibrils, fibrillar fines, fibres and fibre fragments (Chinga-Carrasco [35], Nechy-porchuk et al. [196]). The high swelling property of pre-hydrolyzed fibre is also advanta-geous in liquid adsorbent paper applications, such as kitchen paper towel. The removalof hemicellulose could be advantageous for dissolving pulp and rayon pulp production.The extracted hemicellulose monomer could be converted to furfural, HMF and down-stream acids.7.5 ConclusionMild pre-hydrolysis was integrated with kraft pulping for softwood chips. Particle sizeeffects on hydrolysis and subsequent kraft pulping were assessed. Kraft pulp fibre com-position and properties were systematically discussed.Particle size effects on hemicellulose yield and molar mass were investigated. Lowwood chip thickness increased hemicellulose oligomer yield by reducing mass transferlimitations. Average oligomer molar mass and monomer yield were independent ofparticle size. Soluble hemicellulose products were primarily monomeric arabinose andoligomeric galactan and mannan.Pre-hydrolyzed chips and raw chips were subjected to kraft pulping with an H-factor209of 1600. Pulping of raw chips to H-factor of 1900 was also conducted and compared to thepre-hydrolyzed pulp. Pre-hydrolysis makes it possible to decrease the H-factor requiredto meet a target kappa number by hemicellulose removal and enhanced delignification.As a result, chemical consumption during pulping was reduced. The screened fibre yieldand fibre dimension were also reduced by implementing pre-hydrolysis. However, thepre-hydrolyzed kraft pulp fibres had a higher WRV, and contained fewer fines and rejects.Particle size variation showed little influence on fibre yield. But particle size had alarge impact on fibre length and fines content when the chip thickness was less than 2mm. Kappa number and fibre reject increased dramatically when chip thickness wasgreater than 6 mm.The main challenge of adding pre-hydrolysis to kraft pulping is the fibre yield lossand potential paper strength loss. Pulp mill waste stream could be used as an alternativefeedstock to solve this problem. Reduced fibre dimension and increased WRV could bean advantage in certain paper applications. Further subsequent application and economicanalysis are needed.7.6 Future Work1. Further study of mass transfer during hydrolysis requires:(a) The detailed model proposal for future work of hemicellulose hydrolysis com-bined with kinetic and mass transfer terms is in Section A.2.(b) Collection and analysis of hemicellulose oligomer and monomer compositionin hydrolysate from within fibre lumens.(c) Collection of data at multiple residence times in order to model transient be-havior.2. A more detailed understanding of the effects of pre-hydrolysis on pulping and re-sulting fibre properties will be determined by testing a greater number of pulping210conditions.3. Kraft pulp strength should be determined directly by making and testing hand-sheets. This will require preparation of larger masses of pulp.4. Microscopic analysis of kraft pulp fibre structure could provide more insight intothe effects of pre-hydrolysis on porosity.5. Economic analysis of pre-hydrolysis integration into kraft pulping process and cor-responding changes in energy balance of facility and profitability.6. Develop applications for pre-hydrolyzed kraft pulp fibres that capitalize on reducedfibre dimensions and increased WRV.211Chapter 8Conclusion8.1 Conclusion and ContributionThis thesis examined hemicellulose hydrolysis and its integration with kraft pulping ofsoftwood chips to produce valuable kraft pulp fibre and hemicellulose oligomers for useas a pulp strength additive. The utilization of pulp mill residues such as chip fines andintegration to current kraft pulping are consistent with the ultimate biorefinery goal ofgenerating renewable biochemicals and biomaterials alongside traditional pulp and pa-per products.First a comprehensive literature review of biomass hydrolysis in multiple species wasconducted to summarize hydrolysis principles and find the relationship between compo-sition and hydrolysis response. An experimental study of the conditions and factors in-fluencing softwood hemicellulose hydrolysis was conducted. Operation guidelines weredeveloped and hydrolysate and hydrolyzed solids were characterized to support subse-quent kinetic model and kraft pulping.To track hemicellulose oligomer molar mass evolution during hydrolysis, a two-dimensionalcalibration method was developed to simultaneously measure of oligomer molar massand concentration by size exclusion chromatography (SEC). This approach enables detec-tion of broadly-distributed peaks, which is essential for wood hydrolysate analysis. Fi-212nally, a population balance model was developed to describe evolution of oligomer molarmass during hydrolysis. This model revealed how high molecular weight oligomers ac-cumulate to a maximum and are then depleted and reduced in size. This work providesnew understanding of the relative reactivity of hemicellulose intermediates and enabledpredictions of specific hemicellulose oligomers molar mass at specified hydrolysis condi-tions.The role of hydrolysis and hemicellulose oligomers with respect to kraft pulp wereexamined from two viewpoints in this work. First, adsorption of a model compound forsoftwood hemicellulose, locust bean gum (LBG), to Northern bleached softwood kraft(NBSK) pulp was investigated. Adsorption kinetics and isotherms were examined in or-der to understand the physical mechanisms and identify favorable adsorption conditions.The adsorption effect on NBSK pulp properties were explored. The effects of mild pre-hydrolysis of softwood chips prior to kraft pulping were assessed by kraft pulping rawsoftwood chips and pre-hydrolyzed softwood chips at the same H-factor. Chip size effectson pre-hydrolysis and kraft pulping were analyzed by characterization of hydrolysate,pre-hydrolyzed chips and kraft pulp fibres. The advantages and disadvantages of pair-ing pre-hydrolysis with kraft pulping were discussed.Final conclusions and contributions are as follow:1. Hydrolysis is an effective and environmentally friendly approach to deconstructwood and extract hemicellulose oligomers.(a) The intrinsic structure of biomass determines its response to hydrolysis. Basedon the four major biomass types with respect to feedstock composition and re-sponse to the various hydrothermal pretreatments, the response of all types fol-low broadly similar trends. Overall acid treatment severity is typically higherthan alkali conditions, can almost completely dissolve hemicellulose but pro-duces more degradation byproducts. However, the precise fate of hemicel-213lulose, cellulose, lignin, ash, and extractives can vary substantially betweenspecies.(b) Hemicellulose oligomer production from low cost, readily available pulp millresidues was proposed; conditions were optimized for high yield and high mo-lar mass.i. High temperature, long residence time and low consistency favour oligomerproduction. Increasing particle size lowered the yield of oligomer. Monomeryield was not influenced by particle size or consistency. The compositionof raw biomass had significant influence on the yield of hemicellulose inthe hydrolysate.ii. Galactoglucomannan oligomers were the largest fraction of produced oligomers.Pentose sugars were easily hydrolyzed to monomers while hexose sug-ars were stable as oligomers. Lignin removal increased with hydrolysistemperature and residence time but approached constant value when highseverity condition applied.(c) Softwood hydrolysis and the resulting hemicellulose molar mass distributionwas investigated by SEC. Oligomer production by hydrolysis of mixed soft-woods was investigated and the evolution of molar mass was described by apopulation balance model.i. Hydrolysis time and temperature determined the hemicellulose molar massdistribution. The soluble hemicellulose oligomers tend to break at the mid-dle range of the chain to produce smaller molar mass oligomers. This maybe related to the distribution of acidic groups within the softwood matrix.The final average molar mass was related to the hydrolysate pH.ii. The model describes the full evolution of oligomers from initial softwoodsolubilization, depolymerization to ever-smaller molecules until final gen-214eration of degradation products. Maximum yield, selectivity and the corre-sponding reaction condition for every molar mass interval was predicted.iii. Medium temperature with short residence time favours large molar massoligomer generation. High temperature with short residence time favourssmall molar mass oligomer generation.2. Locust bean gum was studied as a strength additive to NBSK pulp.(a) The adsorption rate followed pseudo-second-order kinetics, indicating chemisorp-tion is the rate determining step.i. The adsorption rate constant increased rapidly with temperature from 25oC to 45 oC, but the amount adsorbed at equilibrium decreased.ii. The activation energy of LBG adsorption was approximately 99.33 kJ ·mol−1with a pre-exponential factor of 4.73E+17 L ·mol−1 ·min−1.(b) LBG adsorption at 25 oC and 35 oC was successfully modeled using the Lang-muir adsorption model for NBSK unrefined and refined pulp with LBG < 2.1wt% of o.d. fibre, indicating a reversible, monolayer, homogenous adsorptionto a finite number of adsorption sites on the fibre surface.i. The maximum adsorption capacity increases with higher temperature whilethe equilibrium constant decreases likely due to reduced LBG adsorption.ii. Refining to 3000 rev increased the surface area of NBSK pulp leading toan increased equilibrium constant and low LBG concentration in the aque-ous phase. The maximum adsorption capacity is double that of unrefinedNBSK pulp.(c) LBG adsorption is dependent on electrostatic forces, and high salt addition atlow pH facilitates adsorption. Temperature negatively affected on LBG adsorp-tion at 25-80 oC.215(d) Refining and LBG dosage increased NBSK paper tensile strength and burststrength. LBG adsorption enhances bonding formation of paper, leading tostrength increase of NBSK paper.i. Tensile and burst strength plateaued when refining over 6000 rev, andstrength gains were small for LBG dosage greater than 0.5 wt%.ii. The effects of mechanical refining were much greater than that of LBG ad-sorption. However, addition of LBG enabled a reduction in refining revo-lutions to achieve a target tensile strength and preserved freeness.iii. Tear index, brightness and scattering coefficient decreased, likely due togreater inter-fibre bonding.3. Mild pre-hydrolysis was integrated with kraft pulping for softwood chips. Particlesize effects on hydrolysis and subsequent kraft pulping were assessed. Kraft pulpfibre composition and properties were systematically discussed.(a) Particle size effects on hemicellulose yield and molar mass, and the subsequentkraft pulping and fibre property were investigated.i. Low wood chip thickness increased hemicellulose oligomer yield by reduc-ing mass transfer limitations. Average oligomer molar mass and monomeryield were independent of particle size. Soluble hemicellulose productswere primarily monomeric arabinose and oligomeric galactan and man-nan.ii. Particle size variation showed little influence on fibre yield. But particlesize had a large impact on fibre length and fines content when chip thick-ness was lower than 2 mm. Kappa number and fibre reject increased dra-matically when chip thickness over 6 mm.(b) Pre-hydrolyzed chips and raw chips were subjected to kraft pulping with an H-factor of 1600 to investigate the hemicellulose removal effect on kraft pulping.216Pulping of raw chips to H-factor of 1900 was also conducted and compared tothe pre-hydrolyzed kraft pulp.i. Pre-hydrolysis makes it possible to decrease the H-factor required to meeta target kappa number. Hemicellulose removal by pre-hydrolysis led toreduced screened fibre yield and enhanced delignification during pulping.ii. Chemical consumption during pulping was reduced by implementing pre-hydrolysis.iii. Pre-hydrolysis decreased fibre dimensions. However, the pre-hydrolyzedkraft pulp fibres had a higher water retention value (WRV), and containedfewer fines and rejects.(c) The main challenge of adding pre-hydrolysis to kraft pulping is the fibre yieldloss and potential paper strength loss. Pulp mill waste stream could be used asan alternative feedstock to solve this problem. Reduced fibre dimension andincreased WRV could be an advantage in certain paper applications. Furthersubsequent application and economic analysis are needed.8.2 Future WorkFuture research directions can be categorized in two main directions: micro-scale studyand universal application guideline study.1. Micro-scale study of wood hydrolysis offers mechanistic insights to build relation-ship between biomass structures to reaction condition.(a) A comprehensive characterization of hydrolysate will be helpful to better un-derstand each components response to hydrolysis.i. Lignin dissolved during hydrolysis could influence the molar mass detec-tion and model results therefore removal of lignin before SEC should beapplied.217ii. Extractives characterization and relevant separation technique waits forfurther study.iii. Collection and analysis of hemicellulose oligomer and monomer composi-tion in hydrolysate from within fibre lumens.(b) Microscopic analysis of softwood chip and pulp fibre structure could providemore insight into the effects of pre-hydrolysis and pulping on wood/fibre poros-ity.i. LBG adsorption could occur only on the fibre surface or could diffuse intofibre pores. Confocal microscopy using fluorescence labeled LBG couldclarify this matter.ii. LBG adsorption conformation could be further characterized and experi-mental evidence of monolayer or multilayer adsorption will provide greatermechanistic understanding of the adsorption process.iii. Microscopic analysis of wood/fibre structure and morphology varying withthe hydrolysis and pulping condition.2. Guidelines for hemicellulose and cellulose utilization are needed to accelerate biore-finery integration into pulp mills and generate economic benefits.(a) Hydrolysis requires an extra processing step before traditional pulp and pa-per operations. Economic analysis will be helpful to optimize the technologyintegration for biorefinery.(b) A general model applied to broader range of biomass with composition varia-tion and more versatile hydrolysis conditions waits for development.i. Pulp mill residues include additional feedstocks such as hog fuel and sludge.Further investigation of these materials could expand the number of biore-fining opportunities.218ii. Calibration standards that more closely match hemicellulose structure wouldresult in more accurate SEC analysis.iii. Lignin extraction and cellulose depolymerization should be incorporatedin to the newly developed hemicellulose hydrolysis model.iv. Model robustness should be verified by testing a broader range of hydrol-ysis temperature and time conditions.v. Solids dissolution could be better modelled such as core shrinking model.vi. Mass transfer analysis should be incorporated with the hemicellulose hy-drolysis model in order to capture the hemicellulose diffusion process inlarge wood chips.(c) Development of applications for pre-hydrolyzed kraft pulp fibre that take ad-vantage of reduced fibre dimension and increased WRV are needed.i. Pulp mill residue application by pre-hydrolysis and hydrolysate hemicel-lulose application, such as pulp strength additive and micro-fibrillated cel-lulose.ii. A more detailed understanding of the effects of pre-hydrolysis on pulp-ing and resulting fibre properties will be determined by testing a greaternumber of pulping conditions. Kraft pulp strength should be determineddirectly by making and testing handsheets.(d) Adsorption of softwood hydrolysate to NBSK pulp as strength additive re-quires further investigation.i. Due to the complexity of composition and low concentration of hemicellu-lose within hydrolysate, purification is likely necessary prior to adsorption.ii. Hemicelluose oligomer charge is a critical property for adsorption, thus,characterization is needed in order to to optimize the adsorption process.iii. Based on LBG adsorption results and assuming a negative charge on soft-219wood hemicellulose, the following condition is recommended for hydrolysateadsorption: low dosage level (<2 wt%), lightly refined (<3000 rev) NBSKpulp at 25 oC for 10 minutes with low fibre consistency (< 0.5 wt%), highagitation rate (> 150 r.p.m.) and acidic or neutral conditions (pH2-7) with-out any salt addition.iv. Refining has the greatest influence on paper strength therefore it is sug-gested to test hydrolysate adsorption to NBSK pulp at low refining lev-els and low hydrolysate dosage to maximize paper strength. Since bond-ing is the main reason for strength enhancement, the scattering coefficienttest could be used to provide bonding information, while the freeness testwould provide complementary drainage information.220Bibliography[1] N. Abatzoglou, E. Chornet, K. Belkacemi, and R. P. Overend. Phenomenologicalkinetics of complex systems: the development of a generalized severity parameterand its application to lignocellulosics fractionation. Chemical Engineering Science,47(5):1109–1122, 1992. → pages 60, 66[2] M. Abbi, R. Kuhad, and A. Singh. Fermentation of xylose and rice strawhydrolysate to ethanol bycandida shehatae ncl-3501. Journal of industrialmicrobiology, 17(1):20–23, 1996. → page 51[3] S. Abedinifar, K. Karimi, M. Khanahmadi, and M. J. Taherzadeh. Ethanolproduction by mucor indicus and rhizopus oryzae from rice straw by separatehydrolysis and fermentation. Biomass and bioenergy, 33(5):828–833, 2009. → page 23[4] W. Ahmad, S. Kuitunen, H. Sixta, and V. Alopaeus. Population balance basedmodeling of changes in cellulose molecular weight distribution during ageing.Cellulose, 22(1):151–163, 2015. → pages 107, 108, 109, 132[5] W. Ahmad, S. Kuitunen, M. Borrega, and V. Alopaeus. Physicochemical modelingfor hot water extraction of birch wood. Industrial & Engineering Chemistry Research,55(42):11062–11073, 2016. → pages 107, 138, 252, 253[6] S. G. Allen, L. C. Kam, A. J. Zemann, and M. J. Antal. Fractionation of sugar canewith hot, compressed, liquid water. Industrial & Engineering Chemistry Research, 35(8):2709–2715, 1996. → page 31[7] C. Altaner and M. Jarvis. Modelling polymer interactions of the molecularvelcrotype in wood under mechanical stress. Journal of theoretical biology, 253(3):434–445, 2008. → pages 4, 25, 90[8] P. Alvira, E. Toma´s-Pejo´, M. Ballesteros, and M. Negro. Pretreatment technologiesfor an efficient bioethanol production process based on enzymatic hydrolysis: areview. Bioresource technology, 101(13):4851–4861, 2010. → page 49[9] T. E. Amidon and S. Liu. Water-based woody biorefinery. Biotechnology advances,27(5):542–550, 2009. → pages 3, 13, 90[10] S. Ando, I. Arai, K. Kiyoto, and S. Hanai. Identification of aromatic monomers insteam-exploded poplar and their influences on ethanol fermentation by221saccharomyces cerevisiae. Journal of fermentation technology, 64(6):567–570, 1986. →page 37[11] C. Arato, E. K. Pye, and G. Gjennestad. The lignol approach to biorefining ofwoody biomass to produce ethanol and chemicals. Applied biochemistry andbiotechnology, 123(1-3):871–882, 2005. → page 181[12] L. G. Austin. Introduction to the mathematical description of grinding as a rateprocess. Powder Technology, 5(1):1–17, 1971. → pages 105, 106[13] L. G. Austin and P. T. Luckie. The estimation of non-normalized breakagedistribution parameters from batch grinding tests. Powder Technology, 5(5):267–271,1972. → pages 105, 106[14] L. G. Austin, R. R. Klimpel, and P. T. Luckie. Process engineering of size reduction:ball milling. American Institute of Mining, Metallurgical, and Petroleum Engineers,1984. → pages 105, 106, 107, 132[15] M. Balat and H. Balat. Recent trends in global production and utilization ofbio-ethanol fuel. Applied energy, 86(11):2273–2282, 2009. → page 12[16] A. M. Basedow, K. H. Ebert, and H. J. Ederer. Kinetic studies on the acid hydrolysisof dextran. Macromolecules, 11(4):774–781, 1978. → pages 99, 105, 107, 108, 109, 132[17] L. A. R. Batalha, J. L. Colodette, J. L. Gomide, L. C. Barbosa, C. R. Maltha, andF. J. B. Gomes. Dissolving pulp production from bamboo. BioResources, 7(1):0640–0651, 2011. → pages 15, 24[18] J. N. BeMiller and R. L. Whistler. Industrial gums: polysaccharides and theirderivatives. Academic Press, 2012. → page 144[19] S. Bhaduri, I. Ghosh, and N. D. Sarkar. Ramie hemicellulose as beater additive inpaper making from jute-stick kraft pulp. Industrial Crops and Products, 4(2):79–84,1995. → page 6[20] A. B. Bjerre, A. B. Olesen, T. Fernqvist, A. Plo¨ger, and A. S. Schmidt. Pretreatmentof wheat straw using combined wet oxidation and alkaline hydrolysis resulting inconvertible cellulose and hemicellulose. Biotechnology and bioengineering, 49(5):568–577, 1996. → page 42[21] M. Blumentritt, D. J. Gardner, B. J. Cole, and S. M. Shaler. Influence of hot-waterextraction on ultrastructure and distribution of glucomannans and xylans inpoplar xylem as detected by gold immunolabeling. Holzforschung, 70(3):243–252,2016. → pages 37, 38[22] O. Bobleter, R. Niesner, and M. Ro¨hr. The hydrothermal degradation of cellulosicmatter to sugars and their fermentative conversion to protein. Journal of AppliedPolymer Science, 20(8):2083–2093, 1976. → page 25222[23] M. Borrega, K. Nieminen, and H. Sixta. Degradation kinetics of the maincarbohydrates in birch wood during hot water extraction in a batch reactor atelevated temperatures. Bioresource technology, 102(22):10724–10732, 2011. → pages26, 30, 32, 50, 57, 59, 97, 104[24] M. Borrega, L. K. Tolonen, F. Bardot, L. Testova, and H. Sixta. Potential of hotwater extraction of birch wood to produce high-purity dissolving pulp afteralkaline pulping. Bioresource technology, 135:665–671, 2013. → pages 183, 200[25] H. O. Bouveng. Phenylisocyanate derivatives of carbohydrates. Acta Chem. Scand,15(1), 1961. → page 55[26] G. Brodeur, E. Yau, K. Badal, J. Collier, K. Ramachandran, and S. Ramakrishnan.Chemical and physicochemical pretreatment of lignocellulosic biomass: a review.Enzyme research, 2011, 2011. → page 49[27] S. J. Burkhardt. Forest residues as a potential feedstock for a biorefinery: material balanceand pretreatment strategies. PhD thesis, University of British Columbia, 2013. →page 12[28] V. S. Chang, M. Nagwani, and M. T. Holtzapple. Lime pretreatment of cropresidues bagasse and wheat straw. Applied biochemistry and biotechnology, 74(3):135–159, 1998. → pages 45, 46[29] J. Chen, Z. Yuan, E. Zanuso, and H. L. Trajano. Response of biomass species tohydrothermal pretreatment. In Hydrothermal Processing in Biorefineries, pages95–140. Springer, 2017. → pages 7, 90, 182, 199, 200, 202[30] R. Chen, Y. Y. Lee, and R. Torget. Kinetic and modeling investigationon ontwo-stage reverse-flow reactoras applied to dilute-acid pretreatment ofagricultural residues. In Seventeenth Symposium on Biotechnology for Fuels andChemicals, pages 133–146. Springer, 1996. → page 104[31] W.-H. Chen, Y.-J. Tu, and H.-K. Sheen. Disruption of sugarcane bagasselignocellulosic structure by means of dilute sulfuric acid pretreatment withmicrowave-assisted heating. Applied energy, 88(8):2726–2734, 2011. → page 37[32] Y. Chen, M. A. Stevens, Y. Zhu, J. Holmes, and H. Xu. Understanding of alkalinepretreatment parameters for corn stover enzymatic saccharification. Biotechnologyfor biofuels, 6(1):8, 2013. → pages 42, 51[33] K.-K. Cheng, B.-Y. Cai, J.-A. Zhang, H.-Z. Ling, Y.-J. Zhou, J.-P. Ge, and J.-M. Xu.Sugarcane bagasse hemicellulose hydrolysate for ethanol production by acidrecovery process. Biochemical Engineering Journal, 38(1):105–109, 2008. → page 51[34] N. Chimwani, D. Glasser, D. Hildebrandt, M. J. Metzger, and F. K. Mulenga.Determination of the milling parameters of a platinum group minerals ore tooptimize product size distribution for flotation purposes. Minerals Engineering, 43:67–78, 2013. → page 105223[35] G. Chinga-Carrasco. Cellulose fibres, nanofibrils and microfibrils: themorphological sequence of mfc components from a plant physiology and fibretechnology point of view. Nanoscale research letters, 6(1):417, 2011. → page 209[36] H. L. Chum, D. K. Johnson, S. K. Black, and R. P. Overend. Pretreatment-catalysteffects and the combined severity parameter. Applied biochemistry and biotechnology,24(1):1, 1990. → pages 60, 66[37] J. d. Clark et al. Pulp technology and treatment for paper. Pulp technology andtreatment for paper., 1978. → page 142[38] J. H. Clark. Green chemistry for the second generation biorefinerysustainablechemical manufacturing based on biomass. Journal of Chemical Technology &Biotechnology: International Research in Process, Environmental & Clean Technology, 82(7):603–609, 2007. → page 1[39] A. Conner. Kinetic modeling of hardwood prehydrolysis. Part I. Xylan removal,1984. → page 104[40] A. Constantinides, N. Mostoufi, et al. Numerical methods for chemical engineers withMATLAB applications, volume 380. Prentice Hall PTR Upper Saddle River, NJ,1999. → pages 129, 130, 262[41] Conversion. Particle size conversion table, 2019. URLhttps://www.sigmaaldrich.com/chemistry/stockroom-reagents/learning-center/technical-library/particle-size-conversion.html. → page 66[42] R. I. Corona-Gonza´lez, K. M. Varela-Almanza, E. Arriola-Guevara, A´. de Jesu´sMartı´nez-Go´mez, C. Pelayo-Ortiz, and G. Toriz. Bagasse hydrolyzates from agavetequilana as substrates for succinic acid production by actinobacillus succinogenesin batch and repeated batch reactor. Bioresource technology, 205:15–23, 2016. →page 16[43] R. C. N. R. Corrales, F. M. T. Mendes, C. C. Perrone, C. SantAnna, W. de Souza,Y. Abud, E. P. P. da Silva Bon, and V. Ferreira-Leita˜o. Structural evaluation of sugarcane bagasse steam pretreated in the presence of co 2 and so 2. Biotechnology forbiofuels, 5(1):36, 2012. → pages 21, 37[44] D. J. Cosgrove and M. C. Jarvis. Comparative structure and biomechanics of plantprimary and secondary cell walls. Frontiers in plant science, 3:204, 2012. → pages4, 25, 90[45] M. Cushing and K. Schuman. Fiber attraction and interfiber bonding-the role ofpolysaccharide additives. Tappi, 42(12):1006–1016, 1959. → pages 166, 167[46] S. Dammstro¨m, L. Salme´n, and P. Gatenholm. On the interactions betweencellulose and xylan, a biomimetic simulation of the hardwood cell wall.BioResources, 4(1):3–14, 2009. → pages 4, 25, 90224[47] S. C. Davis, F. G. Dohleman, and S. P. Long. The global potential for agave as abiofuel feedstock. Gcb Bioenergy, 3(1):68–78, 2011. → pages 16, 24[48] S. de Jong and F. van de Velde. Charge density of polysaccharide controlsmicrostructure and large deformation properties of mixed gels. Food Hydrocolloids,21(7):1172–1187, 2007. → pages xxiv, 144, 147, 168, 172[49] V. Dornburg, A. Faaij, P. Verweij, H. Langeveld, G. Van de Ven, F. Wester, et al.Biomass assessment: Global biomass potentials and their links to food, water,biodiversity, energy demand and economy, main report (climate change scientificassessment and policy analysis). Netherlands Environmental Assessment Agency(MNP), WAB secretariat (ipc 90), PO Box, 303:3720, 2007. → page 12[50] B. Du, L. N. Sharma, C. Becker, S.-F. Chen, R. A. Mowery, G. P. van Walsum, andC. K. Chambliss. Effect of varying feedstock–pretreatment chemistry combinationson the formation and accumulation of potentially inhibitory degradation productsin biomass hydrolysates. Biotechnology and bioengineering, 107(3):430–440, 2010. →page 36[51] M. J. Effland. Modified procedure to determine acid-insoluble lignin in wood andpulp. Tappi;(United States), 60(10), 1977. → page 36[52] I. Egu¨e´s, C. Sanchez, I. Mondragon, and J. Labidi. Effect of alkaline andautohydrolysis processes on the purity of obtained hemicelluloses from cornstalks. Bioresource technology, 103(1):239–248, 2012. → pages 23, 42[53] M. Ek, G. Gellerstedt, and G. Henriksson. Pulping chemistry and technology,volume 2. Walter de Gruyter, 2009. → pages 7, 177, 181, 182, 201, 206[54] A. Emsley and R. Heywood. Computer modelling of the degradation of linearpolymers. Polymer Degradation and Stability, 49(1):145–149, 1995. → page 99[55] B. Epstein. Logarithmico-normal distribution in breakage of solids. Industrial &Engineering Chemistry, 40(12):2289–2291, 1948. → page 105[56] B. Epstein. Statistical aspects of fracture problems. Journal of Applied Physics, 19(2):140–147, 1948. → page 105[57] A. Esteghlalian, A. G. Hashimoto, J. J. Fenske, and M. H. Penner. Modeling andoptimization of the dilute-sulfuric-acid pretreatment of corn stover, poplar andswitchgrass. Bioresource Technology, 59(2-3):129–136, 1997. → pages 23, 36[58] L. Fan, M. Gharpuray, and Y. Lee. Biotechnology monographs. Cellulose Hydrolysis,3, 1987. → page 39[59] FAO. Food and agriculture organization of united nations forestry statistics, 2019.URL http://www.fao.org/forestry/statistics/en/. → pages xx, 3225[60] FAO. The state of food and agriculture, 2008 - fao, 2019. URLhttp://www.fao.org/forestry/statistics/en/. → page 12[61] W. Farhat, R. Venditti, A. Quick, M. Taha, N. Mignard, F. Becquart, and A. Ayoub.Hemicellulose extraction and characterization for applications in paper coatingsand adhesives. Industrial Crops and Products, 107:370–377, 2017. → page 182[62] D. Fengel and G. Wegener. Wood: chemistry, ultrastructure, reactions. Walter deGruyter, 2011. → pagesxxi, 5, 17, 18, 19, 20, 26, 27, 53, 54, 55, 56, 58, 76, 79, 80, 81, 82, 115, 133, 139, 144[63] M. Fisˇerova´, J. Gigac, E. Opa´lena´, et al. Reduction of yield loss in kraft pulping ofhot water pre-extracted beech wood. Wood Research (Bratislava), 59(5):781–792,2014. → pages 7, 182[64] S. Fitzpatrick. Commercialization of the biofine technology for levulinic acidproduction from paper sludge: Us department of energy. Technical report,DOE/CE/41178, 2002. → page 115[65] S. W. Fitzpatrick. Final technical report: Commercialization of the biofinetechnology for levulinic acid production from paper sludge. Technical report,BioMetics, Inc., Waltham, MA (US), 2002. → pages 33, 34[66] T. H. Flores-Sahagun, L. P. Dos Santos, J. Dos Santos, I. Mazzaro, A. Mikowski,et al. Characterization of blue agave bagasse fibers of mexico. Composites Part A:Applied Science and Manufacturing, 45:153–161, 2013. → page 24[67] K. Y. Foo and B. H. Hameed. Insights into the modeling of adsorption isothermsystems. Chemical engineering journal, 156(1):2–10, 2010. → pages 149, 151, 165[68] M. Galbe and G. Zacchi. A review of the production of ethanol from softwood.Applied microbiology and biotechnology, 59(6):618–628, 2002. → pages 3, 13[69] M. Garcı´a-Aparicio, W. Parawira, E. Van Rensburg, D. Diedericks, M. Galbe,C. Rosslander, G. Zacchi, and J. Go¨rgens. Evaluation of steam-treated giantbamboo for production of fermentable sugars. Biotechnology progress, 27(3):641–649, 2011. → pages 15, 24[70] E. Garcia-Moya, A. Romero-Manzanares, and P. Nobel. Highlights for agaveproductivity. Gcb Bioenergy, 3(1):4–14, 2011. → page 16[71] G. Garrote, H. Dominguez, and J. Parajo. Hydrothermal processing oflignocellulosic materials. European Journal of Wood and Wood Products, 57(3):191–202, 1999. → pages 53, 104, 124[72] G. Garrote, H. Domı´nguez, and J. C. Parajo´. Mild autohydrolysis: anenvironmentally friendly technology for xylooligosaccharide production fromwood. Journal of Chemical Technology & Biotechnology: International Research inProcess, Environmental & Clean Technology, 74(11):1101–1109, 1999. → pages 5, 104226[73] M. Ga´spa´r, G. Ka´lma´n, and K. Re´czey. Corn fiber as a raw material forhemicellulose and ethanol production. Process Biochemistry, 42(7):1135–1139, 2007.→ page 39[74] F. M. Gı´rio, C. Fonseca, F. Carvalheiro, L. C. Duarte, S. Marques, andR. Bogel-Łukasik. Hemicelluloses for fuel ethanol: a review. Bioresource technology,101(13):4775–4800, 2010. → pages 54, 90, 115[75] B. Girisuta, K. Dussan, D. Haverty, J. Leahy, and M. Hayes. A kinetic study of acidcatalysed hydrolysis of sugar cane bagasse to levulinic acid. Chemical EngineeringJournal, 217:61–70, 2013. → page 34[76] N. Giummarella. Fundamental Aspects of Lignin Carbohydrate Complexes (LCC):Mechanisms, Recalcitrance and Material concepts. PhD thesis, KTH Royal Institute ofTechnology, 2018. → page 84[77] M. J. Gonza´lez-Mun˜oz, S. Rivas, V. Santos, and J. C. Parajo´. Aqueous processing ofpinus pinaster wood: Kinetics of polysaccharide breakdown. Chemical engineeringjournal, 231:380–387, 2013. → page 104[78] T. M. Grace and E. W. Malcolm. Pulp and paper manufacture: Volume 5 Alkalinepulping. Joint Textbook Committee of the Paper Industry, 1989. → page 184[79] L. Gratani, M. F. Crescente, L. Varone, G. Fabrini, and E. Digiulio. Growth patternand photosynthetic activity of different bamboo species growing in the botanicalgarden of rome. Flora-Morphology, Distribution, Functional Ecology of Plants, 203(1):77–84, 2008. → page 15[80] A. A. Greenwood, T. W. Farrell, Z. Zhang, and I. M. OHara. A novel populationbalance model for the dilute acid hydrolysis of hemicellulose. Biotechnology forbiofuels, 8(1):26, 2015. → pages 104, 107, 108, 109, 124, 136, 137, 138, 252[81] A. A. Gridnev, S. D. Ittel, and M. Fryd. A caveat when determining molecularweight distributions of methacrylate oligomers. Journal of Polymer Science Part A:Polymer Chemistry, 33(7):1185–1188, 1995. → page 91[82] N. Gruenhut. Mannangalactan gums-the mechanism of retention by paper fibers.Tappi, 36(7):297–301, 1953. → pages 145, 147, 166, 167[83] J. Gullichsen and C.-J. Fogelholm. Chemical pulping, paper making science andtechnology, vol. 6, 2000. → pages 54, 115, 206[84] J. S. Gu¨tsch, T. Nousiainen, and H. Sixta. Comparative evaluation ofautohydrolysis and acid-catalyzed hydrolysis of eucalyptus globulus wood.Bioresource technology, 109:77–85, 2012. → pages 28, 30, 32[85] B. B. Hallac and A. J. Ragauskas. Analyzing cellulose degree of polymerizationand its relevancy to cellulosic ethanol. Biofuels, Bioproducts and Biorefining, 5(2):215–225, 2011. → page 33227[86] B. Hames, R. Ruiz, C. Scarlata, A. Sluiter, J. Sluiter, D. Templeton, et al. Preparationof samples for compositional analysis. Laboratory Analytical Procedure (LAP), 1617,2008. → pages 65, 71, 115, 156, 186[87] T. Hannuksela, M. Tenkanen, and B. Holmbom. Sorption of dissolvedgalactoglucomannans and galactomannans to bleached kraft pulp. Cellulose, 9(3-4):251–261, 2002. → pages 6, 7, 143, 144, 145, 147, 148, 164, 172, 173[88] T. Hannuksela, B. Holmbom, and D. Lachenal. Effect of sorbedgalactoglucomannans and galactomannans on pulp and paper handsheetproperties, especially strength properties. Nordic Pulp & Paper Research Journal, 19(2):237–244, 2004. → pages 3, 6, 7, 144, 145, 148, 149, 170, 176[89] M. He, Y. Sun, and B. Han. Green carbon science: scientific basis for integratingcarbon resource processing, utilization, and recycling. Angewandte ChemieInternational Edition, 52(37):9620–9633, 2013. → page 1[90] Y. He and Y. Yue. A review of the effective component and applications of extractsfrom bamboo leaves. Biomass Chemical Engineering, 3, 2008. → pages 15, 19, 47[91] Y. He, Y. Pang, Y. Liu, X. Li, and K. Wang. Physicochemical characterization of ricestraw pretreated with sodium hydroxide in the solid state for enhancing biogasproduction. Energy & Fuels, 22(4):2775–2781, 2008. → page 23[92] F. Hedborg and T. Lindstrom. Adsorption of cationic starch on bleached softwoodcellulosic fibres. Nordic Pulp & Paper Research Journal, 8(2):258–263, 1993. → pages6, 146, 147, 153[93] J. Helmerius, J. V. von Walter, U. Rova, K. A. Berglund, and D. B. Hodge. Impact ofhemicellulose pre-extraction for bioconversion on birch kraft pulp properties.Bioresource technology, 101(15):5996–6005, 2010. → page 183[94] J. Herna´ndez-Salas, M. Villa-Ramı´rez, J. Veloz-Rendo´n, K. Rivera-Herna´ndez,R. Gonza´lez-Ce´sar, M. Plascencia-Espinosa, and S. Trejo-Estrada. Comparativehydrolysis and fermentation of sugarcane and agave bagasse. BioresourceTechnology, 100(3):1238–1245, 2009. → page 29[95] Y.-S. Ho. Second-order kinetic model for the sorption of cadmium onto tree fern: acomparison of linear and non-linear methods. Water research, 40(1):119–125, 2006.→ page 152[96] Y.-S. Ho and G. McKay. The kinetics of sorption of divalent metal ions ontosphagnum moss peat. Water research, 34(3):735–742, 2000. → page 152[97] D. N.-S. Hon and N. Shiraishi. Wood and cellulosic chemistry, revised, and expanded.CRC press, 2000. → pages 54, 55228[98] S. A. Hosseini and N. Shah. Multiscale modelling of hydrothermal biomasspretreatment for chip size optimization. Bioresource Technology, 100(9):2621–2628,2009. → page 252[99] T.-C. Hsu, G.-L. Guo, W.-H. Chen, and W.-S. Hwang. Effect of dilute acidpretreatment of rice straw on structural properties and enzymatic hydrolysis.Bioresource technology, 101(13):4907–4913, 2010. → pages 15, 37, 39, 202[100] D. Huisingh, Z. Zhang, J. C. Moore, Q. Qiao, and Q. Li. Recent advances in carbonemissions reduction: policies, technologies, monitoring, assessment and modeling.Journal of Cleaner Production, 103:1–12, 2015. → pages 1, 2[101] M. Hundt, N. Engel, K. Schnitzlein, and M. G. Schnitzlein. The alkapolp process:Fractionation of various lignocelluloses and continuous pulping within anintegrated biorefinery concept. Chemical Engineering Research and Design, 107:13–23, 2016. → page 22[102] A. Ibrahim, A. Nada, U. Hagemann, and O. A. El Seoud. Preparation of dissolvingpulp from sugar cane bagasse, and its acetylation under homogeneous solutioncondition. Holzforschung-International Journal of the Biology, Chemistry, Physics andTechnology of Wood, 50(3):221–225, 1996. → page 23[103] G. In˜iguez-Covarrubias, R. Dıaz-Teres, R. Sanjuan-Duen˜as, J. Anzaldo-Herna´ndez,and R. M. Rowell. Utilization of by-products from the tequila industry. part 2:potential value of agave tequilana weber azul leaves. Bioresource Technology, 77(2):101–108, 2001. → pages 16, 21[104] M. Jackson. The alkali treatment of straws. Animal Feed Science and Technology, 2(2):105–130, 1977. → page 39[105] A. Jacobs, P. T. Larsson, and O. Dahlman. Distribution of uronic acids in xylansfrom various species of soft-and hardwood as determined by maldi massspectrometry. Biomacromolecules, 2(3):979–990, 2001. → pages 56, 139[106] S. E. Jacobsen and C. E. Wyman. Cellulose and hemicellulose hydrolysis modelsfor application to current and novel pretreatment processes. In Twenty-firstsymposium on biotechnology for fuels and chemicals, pages 81–96. Springer, 2000. →pages 5, 53, 97, 104, 124[107] S. E. Jacobsen and C. E. Wyman. Xylose monomer and oligomer yields foruncatalyzed hydrolysis of sugarcane bagasse hemicellulose at varying solidsconcentration. Industrial & engineering chemistry research, 41(6):1454–1461, 2002. →pages 31, 97[108] M. S. Jahan, M. Shamsuzzaman, M. M. Rahman, S. I. Moeiz, and Y. Ni. Effect ofpre-extraction on soda-anthraquinone (aq) pulping of rice straw. Industrial Cropsand Products, 37(1):164–169, 2012. → pages 15, 23229[109] Z. Ji, Z. Ling, X. Zhang, G.-H. Yang, and F. Xu. Impact of alkali pretreatment on thechemical component distribution and ultrastructure of poplar cell walls.BioResources, 9(3):4159–4172, 2014. → pages 47, 48[110] S. Jin and H. Chen. Superfine grinding of steam-exploded rice straw and itsenzymatic hydrolysis. Biochemical Engineering Journal, 30(3):225–230, 2006. → page23[111] Y. Jin, H. Jameel, H.-m. Chang, and R. Phillips. Green liquor pretreatment ofmixed hardwood for ethanol production in a repurposed kraft pulp mill. Journal ofwood chemistry and technology, 30(1):86–104, 2010. → pages 22, 26, 39, 42, 45, 50[112] A. Johansson. Correlations between fibre properties and paper properties, 2011. →page 204[113] L. Jo¨nsson, E. Palmqvist, N.-O. Nilvebrant, and B. Hahn-Ha¨gerdal. Detoxificationof wood hydrolysates with laccase and peroxidase from the white-rot fungustrametes versicolor. Applied microbiology and biotechnology, 49(6):691–697, 1998. →page 37[114] A. Jun, U. W. Tschirner, and Z. Tauer. Hemicellulose extraction from aspen chipsprior to kraft pulping utilizing kraft white liquor. Biomass and bioenergy, 37:229–236, 2012. → pages 12, 26[115] M. A. Kabel, G. Bos, J. Zeevalking, A. G. Voragen, and H. A. Schols. Effect ofpretreatment severity on xylan solubility and enzymatic breakdown of theremaining cellulose from wheat straw. Bioresource technology, 98(10):2034–2042,2007. → page 23[116] B. M. Kabyemela, T. Adschiri, R. M. Malaluan, and K. Arai. Glucose and fructosedecomposition in subcritical and supercritical water: detailed reaction pathway,mechanisms, and kinetics. Industrial & Engineering Chemistry Research, 38(8):2888–2895, 1999. → pages xxi, 34[117] R. Kajaste. Chemicals from biomass–managing greenhouse gas emissions inbiorefinery production chains–a review. Journal of Cleaner Production, 75:1–10, 2014.→ page 3[118] L. Kang, Y. Y. Lee, S.-H. Yoon, A. J. Smith, and G. A. Krishnagopalan. Ethanolproduction from the mixture of hemicellulose prehydrolysate and paper sludge.BioResources, 7(3):3607–3626, 2012. → pages 26, 31, 36, 50, 56, 57, 58, 59, 99[119] P. Kaparaju, M. Serrano, A. B. Thomsen, P. Kongjan, and I. Angelidaki. Bioethanol,biohydrogen and biogas production from wheat straw in a biorefinery concept.Bioresource technology, 100(9):2562–2568, 2009. → page 29[120] K. Karimi, S. Kheradmandinia, and M. J. Taherzadeh. Conversion of rice straw tosugars by dilute-acid hydrolysis. Biomass and Bioenergy, 30(3):247–253, 2006. →page 23230[121] G. Katz. The location and significance of the O-acetyl groups in a glucomannan fromParana pine. PhD thesis, Georgia Institute of Technology, 1964. → page 55[122] J. Keen and J. Opie. Colorimetric method as a means of studying carbohydratesorption on pulps. Tappi, 40(2):100–104, 1957. → pages 147, 148[123] K. Kemppainen, J. Inkinen, J. Uusitalo, T. Nakari-Seta¨la¨, and M. Siika-aho. Hotwater extraction and steam explosion as pretreatments for ethanol productionfrom spruce bark. Bioresource technology, 117:131–139, 2012. → pages 26, 56, 58, 99[124] P. Kilpela¨inen, S. Hautala, O. Byman, L. Tanner, R. Korpinen, M. K. Lillandt,A. Pranovich, V. Kitunen, S. Willfo¨r, and H. Ilvesniemi. Pressurized hot waterflow-through extraction system scale up from the laboratory to the pilot scale.Green Chemistry, 16(6):3186–3194, 2014. → page 30[125] S. Kim and M. T. Holtzapple. Delignification kinetics of corn stover in limepretreatment. Bioresource Technology, 97(5):778–785, 2006. → pages 15, 45, 46[126] S. B. Kim and Y. Lee. Diffusion of sulfuric acid within lignocellulosic biomassparticles and its impact on dilute-acid pretreatment. Bioresource technology, 83(2):165–171, 2002. → pages 23, 36[127] Y. Kim, T. Kreke, N. S. Mosier, and M. R. Ladisch. Severity factor coefficients forsubcritical liquid hot water pretreatment of hardwood chips. Biotechnology andbioengineering, 111(2):254–263, 2014. → page 54[128] S. Kishani, F. Vilaplana, W. Xu, C. Xu, and L. Wa˚gberg. Solubility of softwoodhemicelluloses. Biomacromolecules, 19(4):1245–1255, 2018. → page 127[129] D. L. Klass. Biomass for renewable energy, fuels, and chemicals. Elsevier, 1998. → page13[130] R. R. Klimpel and L. G. Austin. Determination of selection-for-breakage functionsin the batch grinding equation by nonlinear optimization. Industrial & EngineeringChemistry Fundamentals, 9(2):230–237, 1970. → pages 105, 106[131] T. Kobayashi and Y. Sakai. Hydrolysis rate of pentosan of hardwood in dilutesulfuric acid. Journal of the Agricultural Chemical Society of Japan, 20(1):1–7, 1956. →pages 104, 124[132] V. Koka and O. Trass. Estimation of breakage parameters in grinding operationsusing a direct search method. International Journal of Mineral Processing, 23(1-2):137–150, 1988. → pages 105, 106[133] J. Krogell, E. Korotkova, K. Era¨nen, A. Pranovich, T. Salmi, D. Murzin, andS. Willfo¨r. Intensification of hemicellulose hot-water extraction from spruce woodin a batch extractor–effects of wood particle size. Bioresource technology, 143:212–220, 2013. → page 36231[134] M. Krzesin´ska, J. Zachariasz, and A. Lachowski. Development of monolithiceco-composites from carbonized blocks of solid iron bamboo (dendrocalamusstrictus) by impregnation with furfuryl alcohol. Bioresource technology, 100(3):1274–1278, 2009. → page 15[135] P. Kumar, D. M. Barrett, M. J. Delwiche, and P. Stroeve. Methods for pretreatmentof lignocellulosic biomass for efficient hydrolysis and biofuel production.Industrial & engineering chemistry research, 48(8):3713–3729, 2009. → page 35[136] B. T. Kusema, T. To¨nnov, P. Ma¨ki-Arvela, T. Salmi, S. Willfo¨r, B. Holmbom, andD. Y. Murzin. Acid hydrolysis of o-acetyl-galactoglucomannan. Catalysis Science &Technology, 3(1):116–122, 2013. → pages 107, 123[137] K. J. Laidler. Chemical kinetics,Third Edition. A Pearson Education Company, 1987.→ pages 149, 151, 163, 165[138] I. Langmuir. The constitution and fundamental properties of solids and liquids.part i. solids. Journal of the American chemical society, 38(11):2221–2295, 1916. →pages 149, 165[139] S. Larsson, E. Palmqvist, B. Hahn-Ha¨gerdal, C. Tengborg, K. Stenberg, G. Zacchi,and N.-O. Nilvebrant. The generation of fermentation inhibitors during dilute acidhydrolysis of softwood. Enzyme and Microbial Technology, 24(3-4):151–159, 1999. →pages 26, 28, 32, 56, 58, 99[140] M. Laser, D. Schulman, S. G. Allen, J. Lichwa, M. J. Antal Jr, and L. R. Lynd. Acomparison of liquid hot water and steam pretreatments of sugar cane bagasse forbioconversion to ethanol. Bioresource technology, 81(1):33–44, 2002. → pages 23, 51[141] N. Lebaz, A. Cockx, M. Spe´randio, and J. Morchain. Population balance approachfor the modelling of enzymatic hydrolysis of cellulose. The Canadian Journal ofChemical Engineering, 93(2):276–284, 2015. → page 107[142] H. Leech. An investigation of the reasons for increase in paper strength whenlocust bean gum is used as a beater adhesive. Tappi, 37(8):343–349, 1954. → pages142, 143, 144, 168, 174, 176[143] H. J. Leech. An investigation of the reasons for increase in paper strength when beateradhesives are used. PhD thesis, Georgia Institute of Technology, 1953. → pages6, 7, 142, 143, 145, 160[144] W. Leenakul and N. Tippayawong. Dilute acid pretreatment of bamboo forfermentable sugar production. Journal of Sustainable Energy & Environment, 1(3):117–120, 2010. → pages 24, 29, 33, 51[145] J. Lehto and R. Ale´n. Alkaline pre-treatment of hardwood chips prior todelignification. Journal of Wood Chemistry and Technology, 33(2):77–91, 2013. →pages 22, 39, 43, 44, 45, 47232[146] J. Lehto and R. Ale´n. Alkaline pre-treatment of softwood chips prior todelignification. Journal of Wood Chemistry and Technology, 35(2):146–155, 2015. →pages 22, 43, 44, 47, 50[147] J. Lehto, H. Pakkanen, and R. Ale´n. Characterization of lignin dissolved duringalkaline pretreatment of softwood and hardwood. Journal of Wood Chemistry andTechnology, 35(5):337–347, 2015. → pages 44, 45, 46, 50[148] K. Leppa¨nen, P. Spetz, A. Pranovich, K. Hartonen, V. Kitunen, and H. Ilvesniemi.Pressurized hot water extraction of norway spruce hemicelluloses using aflow-through system. Wood science and technology, 45(2):223–236, 2011. → pages4, 30, 32, 33, 35, 36, 50, 53, 56, 57, 59, 76, 82, 83, 84, 97, 133, 137, 139[149] P. Levi and R. Rosenthal. The periodic table. Schocken Books New York, 1984. →page 1[150] H. Li, A. Saeed, M. S. Jahan, Y. Ni, and A. van Heiningen. Hemicellulose removalfrom hardwood chips in the pre-hydrolysis step of the kraft-based dissolving pulpproduction process. Journal of Wood Chemistry and Technology, 30(1):48–60, 2010. →pages 5, 53[151] H.-Y. Li, S.-N. Sun, X. Zhou, F. Peng, and R.-C. Sun. Structural characterization ofhemicelluloses and topochemical changes in eucalyptus cell wall during alkaliethanol treatment. Carbohydrate polymers, 123:17–26, 2015. → page 47[152] Y. Li, W. Liu, Q. Hou, S. Han, Y. Wang, and D. Zhou. Release of acetic acid and itseffect on the dissolution of carbohydrates in the autohydrolysis pretreatment ofpoplar prior to chemi-thermomechanical pulping. Industrial & EngineeringChemistry Research, 53(20):8366–8371, 2014. → page 139[153] Y. Li, H. Xiao, Y. Pan, and L. Wang. Novel composite adsorbent consisting ofdissolved cellulose fiber/microfibrillated cellulose for dye removal from aqueoussolution. ACS Sustainable Chemistry & Engineering, 6(5):6994–7002, 2018. → pages149, 152, 161[154] Z. Li, Z. Jiang, B. Fei, Z. Cai, and X. Pan. Comparison of bamboo green, timber andyellow in sulfite, sulfuric acid and sodium hydroxide pretreatments for enzymaticsaccharification. Bioresource technology, 151:91–99, 2014. → pages 15, 29[155] W.-S. Lim and J.-W. Lee. Influence of pretreatment condition on the fermentablesugar production and enzymatic hydrolysis of dilute acid-pretreated mixedsoftwood. Bioresource technology, 140:306–311, 2013. → pages 28, 29, 32, 35[156] B. Lindberg. Position of the o-acetyl groups in pine glucomannan. SvenskPapperstidnin, 76:383–384, 1973. → pages 55, 139[157] M. Linde, M. Galbe, and G. Zacchi. Bioethanol production from non-starchcarbohydrate residues in process streams from a dry-mill ethanol plant. BioresourceTechnology, 99(14):6505–6511, 2008. → page 15233[158] A˚. Linder, R. Bergman, A. Bodin, and P. Gatenholm. Mechanism of assembly ofxylan onto cellulose surfaces. Langmuir, 19(12):5072–5077, 2003. → page 6[159] H. Lindqvist, J. Holmback, A. Rosling, K. Salminen, B. Holmbom, M. Auer, andA. Sundberg. Galactoglucomannan derivatives and their application inpapermaking. BioResources, 8(1):994–1010, 2013. → pages 90, 149[160] T. Lindstro¨m, L. Wa˚gberg, and T. Larsson. On the nature of joint strength inpaper-a review of dry and wet strength resins used in paper manufacturing. In13th Fundamental research symposium, volume 1, pages 457–562. The Pulp and PaperFundamental Research Society Cambridge, UK, 2005. → pages 6, 54, 142, 143, 144[161] J. Littlewood, L. Wang, C. Turnbull, and R. J. Murphy. Techno-economic potentialof bioethanol from bamboo in china. Biotechnology for biofuels, 6(1):173, 2013. →pages 14, 15, 24, 28, 31, 32[162] H. Liu, H. Hu, M. M. Baktash, M. S. Jahan, L. Ahsan, and Y. Ni. Kinetics of furfuralproduction from pre-hydrolysis liquor (phl) of a kraft-based hardwood dissolvingpulp production process. biomass and bioenergy, 66:320–327, 2014. → page 12[163] J. Liu, M. Li, X. Luo, L. Chen, and L. Huang. Effect of hot-water extraction (hwe)severity on bleached pulp based biorefinery performance of eucalyptus during thehwe–kraft–ecf bleaching process. Bioresource technology, 181:183–190, 2015. →pages 7, 182, 183, 185[164] J. Liu, S. Willfo¨r, and C. Xu. A review of bioactive plant polysaccharides:Biological activities, functionalization, and biomedical applications. BioactiveCarbohydrates and Dietary Fibre, 5(1):31–61, 2015. → page 91[165] L. Liu, W. Liu, Q. Hou, J. Chen, and N. Xu. Understanding of ph value and itseffect on autohydrolysis pretreatment prior to poplar chemi-thermomechanicalpulping. Bioresource technology, 196:662–667, 2015. → pages 7, 182, 185[166] S. Liu. A synergetic pretreatment technology for woody biomass conversion.Applied Energy, 144:114–128, 2015. → pages xx, 18[167] S. Liu, H. Lu, R. Hu, A. Shupe, L. Lin, and B. Liang. A sustainable woody biomassbiorefinery. Biotechnology Advances, 30(4):785–810, 2012. → pages 3, 13, 90[168] Z. Liu, W. Ahmad, and V. Kuitunen. Modeling of mass transfer and degradation ofhemicelluloses in flow-through hot water extraction. Nordic Pulp & Paper ResearchJournal, 29(4):568–583, 2014. → pages 252, 253[169] Z.-H. Liu, L. Qin, B.-Z. Li, and Y.-J. Yuan. Physical and chemical characterizationsof corn stover from leading pretreatment methods and effects on enzymatichydrolysis. ACS Sustainable Chemistry & Engineering, 3(1):140–146, 2014. → pages33, 35234[170] T. A. Lloyd and C. E. Wyman. Combined sugar yields for dilute sulfuric acidpretreatment of corn stover followed by enzymatic hydrolysis of the remainingsolids. Bioresource technology, 96(18):1967–1977, 2005. → pages 15, 31[171] H. Lu, R. Hu, A. Ward, T. E. Amidon, B. Liang, and S. Liu. Hot-water extractionand its effect on soda pulping of aspen woodchips. Biomass and Bioenergy, 39:5–13,2012. → pages 7, 182, 183, 185, 199, 200, 202[172] J. Lundqvist, A. Teleman, L. Junel, G. Zacchi, O. Dahlman, F. Tjerneld, andH. Sta˚lbrand. Isolation and characterization of galactoglucomannan from spruce(picea abies). Carbohydrate Polymers, 48(1):29–39, 2002. → pages 55, 139[173] X. Luo, X. Ma, H. Hu, C. Li, S. Cao, L. Huang, et al. Kinetic study of pentosansolubility during heating and reacting processes of steam treatment of greenbamboo. Bioresource technology, 130:769–776, 2013. → page 15[174] X. Luo, J. Liu, H. Wang, L. Huang, and L. Chen. Comparison of hot-waterextraction and steam treatment for production of high purity-grade dissolvingpulp from green bamboo. Cellulose, 21(3):1445–1457, 2014. → page 15[175] J. Ma, X. Zhang, X. Zhou, and F. Xu. Revealing the changes in topochemicalcharacteristics of poplar cell wall during hydrothermal pretreatment. BioenergyResearch, 7(4):1358–1368, 2014. → page 38[176] X. Ma, L. Huang, Y. Chen, and L. Chen. Preparation of bamboo dissolving pulp fortextile production; part 1. study on prehydrolysis of green bamboo for producingdissolving pulp. BioResources, 6(2):1428–1439, 2011. → pages 15, 24[177] X. Ma, S. Cao, L. Lin, X. Luo, H. Hu, L. Chen, and L. Huang. Hydrothermalpretreatment of bamboo and cellulose degradation. Bioresource technology, 148:408–413, 2013. → pages 24, 33, 34, 35[178] X. Ma, X. Yang, X. Zheng, L. Lin, L. Chen, L. Huang, and S. Cao. Degradation anddissolution of hemicelluloses during bamboo hydrothermal pretreatment.Bioresource technology, 161:215–220, 2014. → page 28[179] P. Ma¨ki-Arvela, T. Salmi, B. Holmbom, S. Willfo¨r, and D. Y. Murzin. Synthesis ofsugars by hydrolysis of hemicelluloses-a review. Chemical reviews, 111(9):5638–5666, 2011. → page 90[180] M. T. Maloney, T. W. Chapman, and A. J. Baker. Dilute acid hydrolysis of paperbirch: Kinetics studies of xylan and acetyl-group hydrolysis. Biotechnology andbioengineering, 27(3):355–361, 1985. → page 104[181] C. Martin, B. Alriksson, A. Sjo¨de, N.-O. Nilvebrant, and L. J. Jo¨nsson. Dilutesulfuric acid pretreatment of agricultural and agro-industrial residues for ethanolproduction. In Applied Biochemistry and Biotecnology, pages 339–352. Springer, 2007.→ page 36235[182] S. McIntosh and T. Vancov. Optimisation of dilute alkaline pretreatment forenzymatic saccharification of wheat straw. Biomass and Bioenergy, 35(7):3094–3103,2011. → pages 42, 46, 51[183] R. Meyer, K. Almin, and B. Steenberg. Length reduction of fibres subject tobreakage. British Journal of Applied Physics, 17(3):409, 1966. → pages 105, 107[184] A. Miletzky, M. Punz, A. Zankel, S. Schlader, C. Czibula, C. Ganser, C. Teichert,S. Spirk, S. Zo¨hrer, W. Bauer, et al. Modifying cellulose fibers byadsorption/precipitation of xylan. Cellulose, 22(1):189–201, 2015. → page 176[185] A. Mimms, M. Kocurek, J. A. Pyatte, and E. E. Wright. Kraft pulping: a compilation ofnotes. Tappi, 1989. → pages 7, 54, 115, 181, 182, 189, 199, 206[186] K. Mirahmadi, M. M. Kabir, A. Jeihanipour, K. Karimi, and M. Taherzadeh.Alkaline pretreatment of spruce and birch to improve bioethanol and biogasproduction. BioResources, 5(2):928–938, 2010. → page 45[187] A. Mittal, S. G. Chatterjee, G. M. Scott, and T. E. Amidon. Modeling xylansolubilization during autohydrolysis of sugar maple wood meal: reaction kinetics.Holzforschung, 63(3):307–314, 2009. → pages 104, 252[188] M. Mohan, T. Banerjee, and V. V. Goud. Hydrolysis of bamboo biomass bysubcritical water treatment. Bioresource technology, 191:244–252, 2015. → page 24[189] W. S. L. Mok and M. J. Antal Jr. Uncatalyzed solvolysis of whole biomasshemicellulose by hot compressed liquid water. Industrial & Engineering ChemistryResearch, 31(4):1157–1161, 1992. → pages 22, 28, 31, 33, 35, 50, 53, 57, 83, 84, 85, 115[190] S. Mori and H. G. Barth. Size exclusion chromatography. Springer Science & BusinessMedia, 2013. → page 91[191] M. Moshkelani, M. Marinova, M. Perrier, and J. Paris. The forest biorefinery andits implementation in the pulp and paper industry: Energy overview. AppliedThermal Engineering, 50(2):1427–1436, 2013. → pages 3, 13, 90[192] N. Mosier, R. Hendrickson, N. Ho, M. Sedlak, and M. R. Ladisch. Optimization ofph controlled liquid hot water pretreatment of corn stover. Bioresource technology,96(18):1986–1993, 2005. → pages 30, 42, 51[193] N. Mosier, C. Wyman, B. Dale, R. Elander, Y. Lee, M. Holtzapple, and M. Ladisch.Features of promising technologies for pretreatment of lignocellulosic biomass.Bioresource technology, 96(6):673–686, 2005. → pages 7, 39, 42, 182, 199, 200, 202[194] D. S. Most. The sorption of certain slash pine hemicellulose fractions by cellulose fibers.PhD thesis, Georgia Institute of Technology, 1957. → pages6, 7, 145, 146, 147, 148, 168236[195] T. A. Napp, A. Gambhir, T. P. Hills, N. Florin, and P. S. Fennell. A review of thetechnologies, economics and policy instruments for decarbonisingenergy-intensive manufacturing industries. Renewable and Sustainable EnergyReviews, 30:616–640, 2014. → pages xx, 2[196] O. Nechyporchuk, M. N. Belgacem, and J. Bras. Production of cellulosenanofibrils: A review of recent advances. Industrial Crops and Products, 93:2–25,2016. → page 209[197] K. Niskanen. Papermaking science and technology: Paper physics. Fapet Ou,Helsink, Finland, pages 192–221, 1998. → pages6, 142, 143, 146, 176, 181, 182, 184, 199, 204, 205[198] J. Oberlerchner, T. Rosenau, and A. Potthast. Overview of methods for the directmolar mass determination of cellulose. Molecules, 20(6):10313–10341, 2015. → page91[199] S. Y. Oh, D. I. Yoo, Y. Shin, H. C. Kim, H. Y. Kim, Y. S. Chung, W. H. Park, and J. H.Youk. Crystalline structure analysis of cellulose treated with sodium hydroxideand carbon dioxide by means of x-ray diffraction and ftir spectroscopy.Carbohydrate research, 340(15):2376–2391, 2005. → page 45[200] J. A. Olson, J. Drozdiak, M. Martinez, R. Garner, A. G. Robertson, and R. Kerekes.Characterizing fibre shortening in low-consistency refining using a comminutionmodel. Powder technology, 129(1-3):122–129, 2003. → pages 105, 107[201] R. P. Overend and E. Chornet. Fractionation of lignocellulosics by steam-aqueouspretreatments. Philosophical Transactions of the Royal Society of London. Series A,Mathematical and Physical Sciences, 321(1561):523–536, 1987. → pages 60, 66[202] L. Paavilainen. European prospects for using nonwood fibers. PPI, Pulp & PaperInternational, 40(6):4, 1998. → page 15[203] Y. Pan, F. Wang, T. Wei, C. Zhang, and H. Xiao. Hydrophobic modification ofbagasse cellulose fibers with cationic latex: Adsorption kinetics and mechanism.Chemical Engineering Journal, 302:33–43, 2016. → pages 152, 161[204] A. Pandey, C. R. Soccol, P. Nigam, and V. T. Soccol. Biotechnological potential ofagro-industrial residues. i: sugarcane bagasse. Bioresource technology, 74(1):69–80,2000. → page 23[205] Y. C. Park and J. S. Kim. Comparison of various alkaline pretreatment methods oflignocellulosic biomass. Energy, 47(1):31–35, 2012. → pages 22, 45, 50[206] S. R. Parsons. Optical characteristics of paper as a function of fiber classification. PhDthesis, Georgia Institute of Technology, 1941. → page 177237[207] M. Pauly, P. Albersheim, A. Darvill, and W. S. York. Molecular domains of thecellulose/xyloglucan network in the cell walls of higher plants. The Plant Journal,20(6):629–639, 1999. → page 21[208] R. Pecina, P. Burtscher, G. Bonn, and O. Bobleter. Gc-ms and hplc analyses oflignin degradation products in biomass hydrolyzates. Fresenius’ Zeitschrift fu¨ranalytische Chemie, 325(5):461–465, 1986. → page 36[209] G. Peiji, Q. Yinbo, Z. Xin, Z. Mingtian, and D. Yongcheng. Screening microbialstrain for improving the nutritional value of wheat and corn straws as animal feed.Enzyme and Microbial Technology, 20(8):581–584, 1997. → page 23[210] J. Pekarovic, A. Pekarovicova, T. W. Joyce, et al. Desilication of agriculturalresidues-the first step prior to pulping. Appita Journal: Journal of the TechnicalAssociation of the Australian and New Zealand Pulp and Paper Industry, 58(2):130,2005. → page 47[211] M. R. Pelaez-Samaniego, V. Yadama, E. Lowell, and R. Espinoza-Herrera. A reviewof wood thermal pretreatments to improve wood composite properties. WoodScience and Technology, 47(6):1285–1319, 2013. → page 13[212] F. Peng, J.-L. Ren, F. Xu, J. Bian, P. Peng, and R.-C. Sun. Comparative study ofhemicelluloses obtained by graded ethanol precipitation from sugarcane bagasse.Journal of agricultural and food chemistry, 57(14):6305–6317, 2009. → pages 23, 51[213] J. Pe´rez, I. Ballesteros, M. Ballesteros, F. Sa´ez, M. Negro, and P. Manzanares.Optimizing liquid hot water pretreatment conditions to enhance sugar recoveryfrom wheat straw for fuel-ethanol production. Fuel, 87(17-18):3640–3647, 2008. →page 31[214] M. Poliakoff and P. Licence. Sustainable technology: green chemistry. Nature, 450(7171):810, 2007. → page 1[215] D. S. Powlson, M. J. Glendining, K. Coleman, and A. P. Whitmore. Implications forsoil properties of removing cereal straw: Results from long-term studies 1.Agronomy Journal, 103(1):279–287, 2011. → page 15[216] Y. Pu, T. Treasure, R. Gonzalez, R. A. Venditti, and H. Jameel. Autohydrolysispretreatment of mixed softwood to produce value prior to combustion. BioEnergyResearch, 6(3):1094–1103, 2013. → pages4, 5, 22, 28, 29, 30, 31, 32, 50, 56, 57, 58, 59, 75, 81, 85, 87, 91, 133, 137, 139[217] D. Ramkrishna and M. R. Singh. Population balance modeling: current status andfuture prospects. Annual Review of Chemical and Biomolecular Engineering, 5:123–146, 2014. → page 107[218] V. Rangu. Fractionation of pulp mill waste to produce hemicellulose oligomers foradsorption onto nbsk pulp. Master’s thesis, University of British Columbia, 2017.→ page 87238[219] K. J. Reid. A solution to the batch grinding equation. Chemical Engineering Science,20(11):953–963, 1965. → page 105[220] R. C. Reid, J. M. Prausnitz, and B. E. Poling. The properties of gases and liquids.1987. → page 253[221] J. Ren and R. Sun. Applications of hemicelluloses and of their derivatives inpapermaking: A review. Cellulose chemistry and technology, 40(5):281–289, 2006. →pages 4, 5[222] T. Renders, W. Schutyser, S. Van den Bosch, S.-F. Koelewijn, T. Vangeel, C. M.Courtin, and B. F. Sels. Influence of acidic (h3po4) and alkaline (naoh) additives onthe catalytic reductive fractionation of lignocellulose. ACS Catalysis, 6(3):2055–2066, 2016. → pages 33, 45, 47, 115[223] F. Rexen, L. Munck, et al. Cereal crops for industrial use in Europe. Commission of theEuropean Communities, 1984. → page 15[224] G. Richards and H. Sephton. 905. the alkaline degradation of polysaccharides. parti. soluble products of the action of sodium hydroxide on cellulose. Journal of theChemical Society (Resumed), pages 4492–4499, 1957. → page 44[225] H. Riebeek. The carbon cycle: Feature articles. 2011. → page 1[226] D. Rijal, T. Vancov, S. McIntosh, N. Ashwath, and G. A. Stanley. Process optionsfor conversion of agave tequilana leaves into bioethanol. Industrial Crops andProducts, 84:263–272, 2016. → pages 21, 24[227] J. V. Rissanen, H. Gre´nman, S. Willfor, D. Y. Murzin, and T. Salmi. Sprucehemicellulose for chemicals using aqueous extraction: kinetics, mass transfer, andmodeling. Industrial & Engineering Chemistry Research, 53(15):6341–6350, 2014. →pages 4, 26, 30, 32, 36, 37, 57, 97, 139, 192, 252, 253[228] J. V. Rissanen, H. Gre´nman, C. Xu, S. Willfo¨r, D. Y. Murzin, and T. Salmi. Obtainingspruce hemicelluloses of desired molar mass by using pressurized hot waterextraction. ChemSusChem, 7(10):2947–2953, 2014. → pages4, 26, 30, 32, 36, 37, 50, 57, 97, 139, 192[229] S. Rivas, M. J. Gonza´lez-Mun˜oz, V. Santos, and J. C. Parajo´. Acidic processing ofhemicellulosic saccharides from pine wood: Product distribution and kineticmodeling. Bioresource technology, 162:192–199, 2014. → page 87[230] I. C. Roberto, S. I. Mussatto, and R. C. Rodrigues. Dilute-acid hydrolysis foroptimization of xylose recovery from rice straw in a semi-pilot reactor. IndustrialCrops and Products, 17(3):171–176, 2003. → page 31[231] A. Rodrı´guez, A. Moral, L. Serrano, J. Labidi, and L. Jime´nez. Rice straw pulpobtained by using various methods. Bioresource technology, 99(8):2881–2886, 2008.→ page 23239[232] T. Rogalinski, T. Ingram, and G. Brunner. Hydrolysis of lignocellulosic biomass inwater under elevated temperatures and pressures. The Journal of SupercriticalFluids, 47(1):54–63, 2008. → page 31[233] S. Roller and S. A. Jones. Handbook of fat replacers. CRC Press, 1996. → page 144[234] D. F. Root. Kinetics of the acid catalyzed conversion of xylose to furfural. University ofWisconsin–Madison, 1956. → page 104[235] A. Roy, B. Adhikari, and S. Majumder. Equilibrium, kinetic, and thermodynamicstudies of azo dye adsorption from aqueous solution by chemically modifiedlignocellulosic jute fiber. Industrial & Engineering Chemistry Research, 52(19):6502–6512, 2013. → pages 149, 152, 161, 165[236] H. A. Ruiz, R. M. Rodrı´guez-Jasso, B. D. Fernandes, A. A. Vicente, and J. A.Teixeira. Hydrothermal processing, as an alternative for upgrading agricultureresidues and marine biomass according to the biorefinery concept: a review.Renewable and Sustainable Energy Reviews, 21:35–51, 2013. → pages 3, 4, 13, 25, 90[237] V. A. Russo. Sorption studies of a modified locust bean gum in a bleached sulfite pulp.PhD thesis, Georgia Institute of Technology, 1959. → pages6, 7, 145, 147, 148, 163, 166, 167[238] B. Saake and R. Lehnen. Lignin. Ullmann’s Encyclopedia of Industrial Chemistry,2000. → page 19[239] J. F. Saeman. Kinetics of wood saccharification-hydrolysis of cellulose anddecomposition of sugars in dilute acid at high temperature. Industrial &Engineering Chemistry, 37(1):43–52, 1945. → page 103[240] B. C. Saha and M. A. Cotta. Ethanol production from alkaline peroxide pretreatedenzymatically saccharified wheat straw. Biotechnology Progress, 22(2):449–453, 2006.→ pages 23, 30[241] B. C. Saha, T. Yoshida, M. A. Cotta, and K. Sonomoto. Hydrothermal pretreatmentand enzymatic saccharification of corn stover for efficient ethanol production.Industrial Crops and Products, 44:367–372, 2013. → page 30[242] T. Sakaki, M. Shibata, T. Miki, H. Hirosue, and N. Hayashi. Decomposition ofcellulose in near-critical water and fermentability of the products. Energy & Fuels,10(3):684–688, 1996. → page 34[243] M. Salmela, R. Ale´n, and M. T. H. Vu. Description of kraft cooking andoxygen–alkali delignification of bamboo by pulp and dissolving material analysis.Industrial crops and products, 28(1):47–55, 2008. → page 15[244] O. J. Sanchez and C. A. Cardona. Trends in biotechnological production of fuelethanol from different feedstocks. Bioresource technology, 99(13):5270–5295, 2008. →page 26240[245] R. Sands. Forestry in a global context. CABI, 2013. → pages xvi, 12, 13, 14[246] N. Sarkar, S. K. Ghosh, S. Bannerjee, and K. Aikat. Bioethanol production fromagricultural wastes: an overview. Renewable energy, 37(1):19–27, 2012. → page 15[247] M. Sasaki, Z. Fang, Y. Fukushima, T. Adschiri, and K. Arai. Dissolution andhydrolysis of cellulose in subcritical and supercritical water. Industrial &Engineering Chemistry Research, 39(8):2883–2890, 2000. → pages 33, 34[248] N. Sathitsuksanoh, Z. Zhu, T.-J. Ho, M.-D. Bai, and Y.-H. P. Zhang. Bamboosaccharification through cellulose solvent-based biomass pretreatment followed byenzymatic hydrolysis at ultra-low cellulase loadings. Bioresource Technology, 101(13):4926–4929, 2010. → page 15[249] A. Scallan. The effect of acidic groups on the swelling of pulps: a review. Tappijournal, 66(11):73–75, 1983. → page 147[250] H. V. Scheller and P. Ulvskov. Hemicelluloses. Annual review of plant biology, 61,2010. → pages 4, 5, 19, 21, 54[251] J. Scurlock, D. Dayton, and B. Hames. Bamboo: an overlooked biomass resource?Biomass and bioenergy, 19(4):229–244, 2000. → pages 14, 20, 24[252] M. J. Selig, T. B. Vinzant, M. E. Himmel, and S. R. Decker. The effect of ligninremoval by alkaline peroxide pretreatment on the susceptibility of corn stover topurified cellulolytic and xylanolytic enzymes. Applied biochemistry andbiotechnology, 155(1-3):94–103, 2009. → page 47[253] N. Sella Kapu and H. L. Trajano. Review of hemicellulose hydrolysis in softwoodsand bamboo. Biofuels, Bioproducts and Biorefining, 8(6):857–870, 2014. → pages31, 36, 90, 91[254] J. Shen, I. Kaur, M. M. Baktash, Z. He, and Y. Ni. A combined process of activatedcarbon adsorption, ion exchange resin treatment and membrane concentration forrecovery of dissolved organics in pre-hydrolysis liquor of the kraft-baseddissolving pulp production process. Bioresource technology, 127:59–65, 2013. →page 12[255] M. Shirazi, T. G. van de Ven, and G. Garnier. Adsorption of modified starches onpulp fibers. Langmuir, 19(26):10835–10842, 2003. → pages6, 146, 147, 149, 153, 165, 166, 167[256] T. C. F. Silva, J. L. Colodette, L. A. Lucia, R. C. d. Oliveira, F. N. Oliveira, andL. H. M. Silva. Adsorption of chemically modified xylans on eucalyptus pulp andits effect on the pulp physical properties. Industrial & Engineering ChemistryResearch, 50(2):1138–1145, 2010. → pages 6, 7, 90241[257] M. H. L. Silveira, A. R. C. Morais, A. M. da Costa Lopes, D. N. Olekszyszen,R. Bogel-Łukasik, J. Andreaus, and L. Pereira Ramos. Current pretreatmenttechnologies for the development of cellulosic ethanol and biorefineries.ChemSusChem, 8(20):3366–3390, 2015. → pages 12, 25[258] F. Simmonds and C. Coens. Determining scattering and absorption coefficients bydiffuse illumination. Technical report, FOREST PRODUCTS LAB MADISON WIS,1967. → page 177[259] R. Sindhu, M. Kuttiraja, P. Binod, R. K. Sukumaran, and A. Pandey. Bioethanolproduction from dilute acid pretreated indian bamboo variety (dendrocalamussp.) by separate hydrolysis and fermentation. Industrial Crops and Products, 52:169–176, 2014. → pages 33, 35, 37, 51[260] H. Sixta. Handbook of pulp. Wiley-vch, 2006. → page 45[261] E. Sjostrom. The origin of charge on cellulosic fibers. Nordic Pulp & Paper ResearchJournal, 4(2):90–93, 1989. → page 146[262] E. Sjostrom. Wood chemistry: fundamentals and applications. Gulf professionalpublishing, 1993. → pages xx, xxi, 13, 17, 18, 20, 27, 30, 35, 40, 41, 42, 54, 76, 82, 84[263] E. Sjostrom. Wood chemistry: fundamentals and applications. Elsevier, 2013. → page181[264] A. Sluiter, B. Hames, R. Ruiz, C. Scarlata, J. Sluiter, and D. Templeton.Determination of ash in biomass laboratory analytical procedure. NationalRenewable Energy Laboratory Analytical Procedure, Golden, CO, 2005. → page 19[265] A. Sluiter, B. Hames, R. Ruiz, C. Scarlata, J. Sluiter, D. Templeton, et al.Determination of sugars, byproducts, and degradation products in liquid fractionprocess samples. Golden: National Renewable Energy Laboratory, 11, 2006. → pages65, 71, 115, 156, 190[266] A. Sluiter, B. Hames, R. Ruiz, C. Scarlata, J. Sluiter, D. Templeton, and D. Crocker.Determination of structural carbohydrates and lignin in biomass. Laboratoryanalytical procedure, 1617:1–16, 2008. → pages 65, 115, 156, 190[267] A. Sluiter, B. Hames, R. Ruiz, C. Scarlata, J. Sluiter, D. Templeton, and D. Crocker.Determination of structural carbohydrates and lignin in biomass. Laboratoryanalytical procedure, 1617:1–16, 2008. → page 71[268] A. Sluiter, B. Hames, R. Ruiz, C. Scarlata, J. Sluiter, D. Templeton, et al.Determination of ash in biomass: laboratory analytical procedure (lap). NationalRenewable Energy Laboratory, 2008. → page 71[269] A. Sluiter, R. Ruiz, C. Scarlata, J. Sluiter, D. Templeton, et al. Determination ofextractives in biomass. Laboratory analytical procedure (LAP), 2008. → page 71242[270] A. J. Smith. Hot water extraction and subsequent Kraft pulping of pine wood chips. PhDthesis, 2011. → pages 183, 200[271] G. A. Smook, M. J. Kocurek, et al. Handbook for pulp & paper technologists. CanadianPulp and Paper Association, 1982. → pages 3, 4, 19, 20, 176, 189, 200[272] C. Somerville, H. Youngs, C. Taylor, S. C. Davis, and S. P. Long. Feedstocks forlignocellulosic biofuels. science, 329(5993):790–792, 2010. → page 16[273] T. Song, A. Pranovich, I. Sumerskiy, and B. Holmbom. Extraction ofgalactoglucomannan from spruce wood with pressurised hot water. Holzforschung,62(6):659–666, 2008. → pages 30, 56, 57, 97, 192[274] T. Song, A. Pranovich, and B. Holmbom. Effects of ph control with phthalatebuffers on hot-water extraction of hemicelluloses from spruce wood. Bioresourcetechnology, 102(22):10518–10523, 2011. → pages 26, 30, 31, 32, 36, 59, 99[275] R. R. Spencer and D. E. Akin. Rumen microbial degradation of potassiumhydroxide-treated coastal bermudagrass leaf blades examined by electronmicroscopy. Journal of Animal Science, 51(5):1189–1196, 1980. → page 39[276] A. R. Staniforth et al. Cereal straw. Oxford University Press., 1979. → page 15[277] C. H. Stephens, P. M. Whitmore, H. R. Morris, and M. E. Bier. Hydrolysis of theamorphous cellulose in cotton-based paper. Biomacromolecules, 9(4):1093–1099,2008. → page 35[278] R. Sun, J. M. Lawther, and W. Banks. Influence of alkaline pre-treatments on thecell wall components of wheat straw. Industrial crops and products, 4(2):127–145,1995. → pages 42, 45, 51[279] A. Suurnakki, T. Oksanen, S. Buchert, and H. Kettunen. The effect of mannan onphysical properties of ecf bleached softwood kraft fibre handsheets. Nordic Pulp &Paper Research Journal, 18(4):429–435, 2003. → pages 6, 7, 144[280] J. Swanson. The effects of natural beater additives on papermaking fibers. Tappi,33(9):451–462, 1950. → pages 6, 7, 144, 145, 176[281] J. W. Swanson, M. B. Webb, and R. C. Morse. # 1467. locust bean and guar as beateradditives. project reports 1949-50. 1949. → pages 145, 146, 160[282] E. Taarning, C. M. Osmundsen, X. Yang, B. Voss, S. I. Andersen, and C. H.Christensen. Zeolite-catalyzed biomass conversion to fuels and chemicals. Energy& Environmental Science, 4(3):793–804, 2011. → pages 3, 4, 19, 20[283] F. Talebnia, D. Karakashev, and I. Angelidaki. Production of bioethanol fromwheat straw: an overview on pretreatment, hydrolysis and fermentation.Bioresource technology, 101(13):4744–4753, 2010. → page 37243[284] D. Tarasov, M. Leitch, and P. Fatehi. Lignin–carbohydrate complexes: Properties,applications, analyses, and methods of extraction: A review. Biotechnology forbiofuels, 11(1):269, 2018. → page 84[285] H. Tarkow and W. Feist. A mechanism for improving digestibility oflignocellulosic materials with dilute alkali and liquid ammonia. Advances inChemistry Series, (95):197, 1969. → page 47[286] A. Tayal and S. A. Khan. Degradation of a water-soluble polymer: molecularweight changes and chain scission characteristics. Macromolecules, 33(26):9488–9493, 2000. → page 107[287] M. H. Thomsen, A. Thygesen, and A. B. Thomsen. Hydrothermal treatment ofwheat straw at pilot plant scale using a three-step reactor system aiming at highhemicellulose recovery, high cellulose digestibility and low lignin hydrolysis.Bioresource Technology, 99(10):4221–4228, 2008. → pages 31, 51[288] T. Timell. Wood hemicelluloses: part ii. In Advances in carbohydrate chemistry,volume 20, pages 409–483. Elsevier, 1965. → pages 55, 139[289] T. Timell. Recent progress in the chemistry of wood hemicelluloses. Wood Scienceand Technology, 1(1):45–70, 1967. → pages 54, 55[290] T. Timell et al. Isolation and properties of a glucomannan from the wood of whitebirch (betula papyrifera, marsh.). Tappi, 43(10):844–8, 1960. → page 55[291] R. Timung, M. Mohan, B. Chilukoti, S. Sasmal, T. Banerjee, and V. V. Goud.Optimization of dilute acid and hot water pretreatment of different lignocellulosicbiomass: a comparative study. biomass and bioenergy, 81:9–18, 2015. → page 28[292] N. Torelli and K. Cˇufar. Mexican tropical hardwoods. comparative study of ashand silica content. Holz als Roh-und werkstoff, 53(1):61–62, 1995. → page 19[293] R. Torget, P. Walter, M. Himmel, and K. Grohmann. Dilute-acid pretreatment ofcorn residues and short-rotation woody crops. Applied Biochemistry andBiotechnology, 28(1):75, 1991. → page 31[294] K. B. Torssell. Natural product chemistry: A mechanistic. Biosynthetic andEcological Approach, Second Edition, Apotekarsocieteten-Swedish Pharmaceutical Society(Stockholm), pages 313–347, 1997. → page 19[295] H. L. Trajano, N. L. Engle, M. Foston, A. J. Ragauskas, T. J. Tschaplinski, and C. E.Wyman. The fate of lignin during hydrothermal pretreatment. Biotechnology forbiofuels, 6(1):110, 2013. → page 59[296] H. L. Trajano, S. Pattathil, B. A. Tomkins, T. J. Tschaplinski, M. G. Hahn, G. J.Van Berkel, and C. E. Wyman. Xylan hydrolysis in populus trichocarpa× p.deltoides and model substrates during hydrothermal pretreatment. Bioresourcetechnology, 179:202–210, 2015. → page 30244[297] M. S. Tunc, J. Chheda, E. van der Heide, J. Morris, and A. van Heiningen.Pretreatment of hardwood chips via autohydrolysis supported by acetic andformic acid. Holzforschung, 68(4):401–409, 2014. → pages 26, 28, 35[298] I. Urruzola, L. Serrano, R. Llano-Ponte, M. A´. de Andre´s, and J. Labidi. Obtainingof eucalyptus microfibrils for adsorption of aromatic compounds in aqueoussolution. Chemical engineering journal, 229:42–49, 2013. → pages 149, 152, 161[299] H. Van De Steeg. Cationic starches on cellulose surfaces. PhD thesis, Ph. D. Thesis,Wageningen Agricultural University. Netherlands, 1992. → pages 147, 153[300] H. Van de Steeg, A. De Keizer, M. C. Stuart, and B. Bijsterbosch. Adsorption ofcationic amylopectin on microcrystalline cellulose. Colloids and Surfaces A:Physicochemical and Engineering Aspects, 70(1):77–89, 1993. → page 146[301] H. G. van de Steeg. The adsorption of cationic starch on microcrystalline cellulose.Nordic Pulp & Paper Research Journal, 4(2):173–178, 1989. → pages 6, 146[302] H. G. van de Steeg, A. de Keizer, M. A. C. Stuart, C. Stuart, and B. H. Bijsterbosch.Adsorption of cationic starches on microcrystalline cellulose. Nordic Pulp & PaperResearch Journal, 8(1):34–40, 1993. → pages 6, 146, 147, 153[303] T. G. Van De Ven. Kinetic aspects of polymer and polyelectrolyte adsorption onsurfaces. Advances in colloid and interface science, 48:121–140, 1994. → pages149, 152, 162, 166, 168[304] T. G. van de Ven. A model for the adsorption of polyelectrolytes on pulp fibers:relation between fiber structure and polyelectrolyte properties. Nordic Pulp &Paper Research Journal, 15(5):494–501, 2000. → page 146[305] A. Van Heiningen. Converting a kraft pulp mill into an integrated forestbiorefinery. Pulp and Paper Canada, 107(6):38–43, 2006. → pages 2, 3, 13, 42, 90, 182[306] C. Van Zyl, B. A. Prior, and J. C. Du Preez. Production of ethanol from sugar canebagasse hemicellulose hydrolyzate bypichia stipitis. Applied Biochemistry andBiotechnology, 17(1-3):357–369, 1988. → page 15[307] U. Vela´zquez-Valadez, J. C. Farı´as-Sa´nchez, A. Vargas-Santilla´n, and A. J.Castro-Montoya. Tequilana weber agave bagasse enzymatic hydrolysis for theproduction of fermentable sugars: oxidative-alkaline pretreatment and kineticmodeling. BioEnergy Research, 9(4):998–1004, 2016. → page 46[308] P. Vena, M. Garcı´a-Aparicio, M. Brienzo, J. Go¨rgens, and T. Rypstra. Effect ofalkaline hemicellulose extraction on kraft pulp fibers from eucalyptus grandis.Journal of Wood Chemistry and Technology, 33(3):157–173, 2013. → pages22, 41, 42, 45, 50245[309] C. Vila, J. Romero, J. Francisco, G. Garrote, and J. Parajo´. Extracting value fromeucalyptus wood before kraft pulping: Effects of hemicelluloses solubilization onpulp properties. Bioresource technology, 102(8):5251–5254, 2011. → page 183[310] J. A. Visuri, T. Song, S. Kuitunen, and V. Alopaeus. Model for degradation ofgalactoglucomannan in hot water extraction conditions. Industrial & EngineeringChemistry Research, 51(31):10338–10344, 2012. → pages104, 107, 108, 109, 136, 137, 138[311] P. M. Vitousek. Beyond global warming: ecology and global change. Ecology, 75(7):1861–1876, 1994. → page 1[312] K. Vroom. The” h” factor: a means of expressing cooking times and temperaturesas a single variable. PPMC, 58:228–231, 1957. → pages 60, 66[313] T. H. M. Vu, H. Pakkanen, and R. Ale´n. Delignification of bamboo (bambusaprocera acher): Part 1. kraft pulping and the subsequent oxygen delignification topulp with a low kappa number. Industrial Crops and Products, 19(1):49–57, 2004. →pages 15, 24[314] V. M. Vucˇurovic´, R. N. Razmovski, and M. N. Tekic´. Methylene blue (cationic dye)adsorption onto sugar beet pulp: equilibrium isotherm and kinetic studies. Journalof the Taiwan Institute of Chemical Engineers, 43(1):108–111, 2012. → pages149, 152, 161[315] L. Wa˚gberg and M. Bjorklund. Adsorption of cationic potato starch on cellulosicfibres. Nordic Pulp & Paper Research Journal, 8(4):399–404, 1993. → page 153[316] L. Wa˚gberg and R. Ha¨gglund. Kinetics of polyelectrolyte adsorption on cellulosicfibers. Langmuir, 17(4):1096–1103, 2001. → pages 146, 164[317] S. L. Walton, D. Hutto, J. M. Genco, G. P. v. Walsum, and A. R. v. Heiningen.Pre-extraction of hemicelluloses from hardwood chips using an alkaline woodpulping solution followed by kraft pulping of the extracted wood chips. Industrial& Engineering Chemistry Research, 49(24):12638–12645, 2010. → pages22, 26, 43, 44, 50[318] WEC. World energy insight 2010, world energy council, 2010. URLhttps://www.worldenergy.org/publications/2010/world-energy-insight-2010. → page 12[319] S. Willfo¨r, R. Sjo¨holm, C. Laine, M. Roslund, J. Hemming, and B. Holmbom.Characterisation of water-soluble galactoglucomannans from norway sprucewood and thermomechanical pulp. Carbohydrate Polymers, 52(2):175–187, 2003. →pages 54, 55, 139[320] S. Willfo¨r, A. Sundberg, J. Hemming, and B. Holmbom. Polysaccharides in someindustrially important softwood species. Wood Science and Technology, 39(4):245–257, 2005. → page 80246[321] C. Wyman. Handbook on bioethanol: production and utilization. CRC press, 1996. →page 19[322] C. E. Wyman. Aqueous pretreatment of plant biomass for biological and chemicalconversion to fuels and chemicals. John Wiley & Sons, 2013. → pages 5, 91[323] X. Xiao, J. Bian, X.-P. Peng, H. Xu, B. Xiao, and R.-C. Sun. Autohydrolysis ofbamboo (dendrocalamus giganteus munro) culm for the production ofxylo-oligosaccharides. Bioresource technology, 138:63–70, 2013. → pages 28, 30, 31[324] X. Xiao, J. Bian, M.-F. Li, H. Xu, B. Xiao, and R.-C. Sun. Enhanced enzymatichydrolysis of bamboo (dendrocalamus giganteus munro) culm by hydrothermalpretreatment. Bioresource technology, 159:41–47, 2014. → pages 33, 35[325] C. Xu, A. Pranovich, L. Va¨ha¨salo, J. Hemming, B. Holmbom, H. A. Schols, andS. Willfo¨r. Kinetics of acid hydrolysis of water-soluble spruce o-acetylgalactoglucomannans. Journal of agricultural and food chemistry, 56(7):2429–2435,2008. → pages 57, 91, 99, 105, 108, 109, 136, 137, 138[326] Z. Xu and F. Huang. Pretreatment methods for bioethanol production. Appliedbiochemistry and biotechnology, 174(1):43–62, 2014. → pages 17, 25, 26, 39, 49[327] Y. Yamashita, M. Shono, C. Sasaki, and Y. Nakamura. Alkaline peroxidepretreatment for efficient enzymatic saccharification of bamboo. CarbohydratePolymers, 79(4):914–920, 2010. → pages 24, 51[328] L. Yan, L. Zhang, and B. Yang. Enhancement of total sugar and lignin yieldsthrough dissolution of poplar wood by hot water and dilute acid flowthroughpretreatment. Biotechnology for biofuels, 7(1):76, 2014. → pages 26, 35, 36, 56, 59, 75[329] S.-H. Yoon and A. Van Heiningen. Kraft pulping and papermaking properties ofhot-water pre-extracted loblolly pine in an integrated forest products biorefinery.Tappi J, 7(7):22–27, 2008. → pages 182, 183, 185, 200, 206[330] S.-H. Yoon and A. van Heiningen. Green liquor extraction of hemicelluloses fromsouthern pine in an integrated forest biorefinery. Journal of Industrial andEngineering Chemistry, 16(1):74–80, 2010. → pages 41, 42, 43, 44, 50[331] G. Yu, S. Yano, H. Inoue, S. Inoue, T. Endo, and S. Sawayama. Pretreatment of ricestraw by a hot-compressed water process for enzymatic hydrolysis. AppliedBiochemistry and Biotechnology, 160(2):539–551, 2010. → pages 23, 30[332] Z. Yuan, N. S. Kapu, R. Beatson, X. F. Chang, and D. M. Martinez. Effect of alkalinepre-extraction of hemicelluloses and silica on kraft pulping of bamboo(neosinocalamus affinis keng). Industrial Crops and Products, 91:66–75, 2016. →pages 42, 45, 47247[333] N. Zakrajsˇek, S. Knez, D. Ravnjak, and J. Golob. Analysis of modified starchadsorption kinetics on cellulose fibers via the modified langmuir adsorptiontheory. Chemical and biochemical engineering quarterly, 23(4):461–470, 2009. → pages6, 145, 146, 148, 149, 151, 153, 155, 160, 162, 166, 167, 168[334] J. Zhang, M. Tang, and L. Viikari. Xylans inhibit enzymatic hydrolysis oflignocellulosic materials by cellulases. Bioresource technology, 121:8–12, 2012. →pages 4, 25[335] Q. Zhang and W. Cai. Enzymatic hydrolysis of alkali-pretreated rice straw bytrichoderma reesei zm4-f3. Biomass and Bioenergy, 32(12):1130–1135, 2008. → pages12, 42, 45, 46[336] Y.-H. P. Zhang. Reviving the carbohydrate economy via multi-productlignocellulose biorefineries. Journal of industrial microbiology & biotechnology, 35(5):367–375, 2008. → page 12[337] Y.-H. P. Zhang and L. R. Lynd. Toward an aggregated understanding of enzymatichydrolysis of cellulose: noncomplexed cellulase systems. Biotechnology andbioengineering, 88(7):797–824, 2004. → page 39[338] L. Zhao, Z. Yuan, N. S. Kapu, X. F. Chang, R. Beatson, H. L. Trajano, and D. M.Martinez. Increasing efficiency of enzymatic hemicellulose removal from bamboofor production of high-grade dissolving pulp. Bioresource technology, 223:40–46,2017. → page 189[339] L. Zhu, J. P. ODwyer, V. S. Chang, C. B. Granda, and M. T. Holtzapple. Structuralfeatures affecting biomass enzymatic digestibility. Bioresource technology, 99(9):3817–3828, 2008. → page 37248Appendix ASupporting MaterialsA.1 LBG Adsorption Kinetics Equation DerivationPseudo-first-order adsorption kinetics is described by:dqtdt= k(qe − qt) (A.1)where k is rate constant (g·mg−1 ·min−1), qe is the concentration of LBG (mg·g−1 o.d.fibre) adsorbed at equlibium, and qt is concentration of (mg·g−1 o.d. fibre) LBG adsorbedat any time, t (min).Separating variables yields:dqtqe − qt = kdt (A.2)Integrating using the boundary conditions:249t = 0 qt = 0 (A.3)t = t qt = qt (A.4)ln(qe − qt) = −kt + lnqe (A.5)Plotting ln(qe − qt) as a function of t will yield a line with slope of k and intercept oflnqe.The rate equation for pseudo-second-order adsorption kinetics:dqtdt= k(qe − qt)2 (A.6)where k is rate constant (g·mg−1 ·min−1), qe is the concentration of LBG (mg·g−1 o.d.fibre) adsorbed at equlibium state, and qt is concentration of LBG adsorbed (mg·g−1 o.d.fibre) at any time, t (min).Separating variables:dqt(qe − qt)2 = kdt (A.7)Integrating using boundary conditions:t = 0 qt = 0 (A.8)t = t qt = qt (A.9)1qe − qt −1qe= kt (A.10)250Equation A.10 can be re-arranged as:tqt=1k(qe)2+tqe(A.11)Plotting tqt as a function of t will yield a line with slope1qe and intercept of1k(qe)2.251A.2 Softwood Hemicellulose Hydrolysis Mass TransferModelA.2.1 IntroductionTwo types of hemicellulose hydrolysis model with mass transfer were developed. One isordinary differential equation porosity model, and another is partial differential equationdiffusion model.Porosity model lumps the spatial variation of mass transfer into temporal variationby effective diffusion coefficient term as Equation A.12 (Ahmad et al. [5], Mittal et al.[187], Rissanen et al. [227]). No concentration distribution information is obtained andan average concentration is applied instead (Mittal et al. [187]). This model can describeirregular geometries and reduces the calculation load.Di,eff =ετDAB(T) (A.12)where DAB is the diffusivity, ε is porosity and τ is tortuosity.The diffusion model applies fundamental equations of mass transfer and can predictmass distribution within certain regular geometry, but is difficult to solve due to multipleunknown parameters and simultaneous variation of spatial and temporal terms. Liu et al.developed model of hemicellulose hydrolysis and mass transfer in three domains: solidwood, stagnant liquor within pores of wood, and bulk liquid phase. A cylindrical masstransfer term was applied for the liquor within pores, but no mass transfer term was in-voked for the solid wood domain. The kinetics of hemicellulose hydrolysis was modeledby population balance model (Liu et al. [168]).The diffusivity for previous hemicellulose hydrolysis models was usually calculatedfrom approximation model (Ahmad et al. [5], Greenwood et al. [80], Hosseini and Shah[98], Rissanen et al. [227]). Greenwood et al. studied the dilute acid hydrolysis of hemi-252cellulose within sugarcane bagasse fibre, and applied a diffusion term. The diffusivitywas obtained from Stokes-Einstein approximation. A similar estimation of diffusion co-efficient was made by Rissanen et al.. Ahmad et al. and Liu et al. used the Wilke-Changmodel (Reid et al. [220]) to estimate the diffusion coefficient in Equation A.13. From Equa-tion A.13, the diffusion coefficient of solute A is a function of molar volume of solute Aand molecular weight of solvent B in an exponential form.DoAB =7.4× 10−8(ΦMB)0.5TηBV0.6A(A.13)where DoAB is mutual diffusion coefficient of solute A in cm2/s, MB is molecular weightof solvent B in g/mol, T is temperature in K, ηB is viscosity of solvent B in cP, VA is molarvolume of solute A in cm2/mol and Φ is association factor of solvent B, dimensionless.In this work, a hemicellulose hydrolysis model will be developed for two domains,softwood chip and bulk solution, using mass transfer and kinetic terms. The kineticsrate constant was obtained according to method in Chapter 4. Thus, softwood hydrolysisof chips with varying particle size could provide mass transfer coefficients and furtherphysical insights. The concentration distribution will be solved by fitting the diffusioncoefficient and mass transfer coefficient. A correlation of these coefficients with the molarmass of hemicellulose oligomers will be proposed.A.2.2 Goals and hypotheses1. Hemicellulose concentration distribution within softwood chips could be predictedby hydrolysis modeling combined with diffusion.2. The diffusion coefficient is an exponential function of hemicellulose oligomer molarmass.253A.2.3 Experimental designRefer to Chapter 7 pre-hydrolysis kraft pulping.A.2.4 Model developmentSoftwood chip is completely immersed in water during hydrolysis, a schematic illustra-tion is shown in Figure A.1.Figure A.1: Schematic illustration of hemicellulose mass transfer of softwood chipimmersion in water.There are two hydrolysis and mass transfer domains. One is within the wood chips,Ω1, with concentration varying from C0 to C∗, where C0 is the centre line concentrationof hemicellulose, and C∗ is the concentration of hemicellulose at the interface of chip withwater. The characteristic length L is the half-width of the softwood chip. The diffusion ofhemicellulose within Ω1 is represented by diffusion coefficient or diffusivity, D.The second domain isΩ2, which is a bulk solution at the interface with wood chip. Thecharacteristic length of Ω2 is unknown. The mass transfer from the surface of chip to thesurrounding water is described by the mass transfer coefficient, h. In the bulk solution,the hemicellulose oligomer concentration, C∞, is homogeneous.254The hemicellulose mass transfer is assumed to be one-dimensional diffusion withinthe wood chip (Ω1) along axis x (Figure A.1), and one-dimensional mass transfer acrossthe bulk solution (Ω2).For Ω1, 9 intervals of oligomers are defined in Chapter 5. The concentration variationwith time and space are as follows:∂C1∂t= D1∂2C1∂x2+ k4C2 (A.14)∂C2∂t= D2∂2C2∂x2− k4C2 + 17k3C9 + k1(2)k2C3 (A.15)∂C3∂t= D3∂2C3∂x2− k1(2)k2C3 + 17k3C9 + k1(3)k2C4 + k1(4)k2C5 (A.16)∂C4∂t= D4∂2C4∂x2− k1(3)k2C4 + 17k3C9 + k1(5)k2C6 + k1(6)k2C7 (A.17)∂C5∂t= D5∂2C5∂x2− k1(4)k2C5 + 17k3C9 + k1(7)k2C8 (A.18)∂C6∂t= D6∂2C6∂x2− k1(5)k2C6 + 17k3C9 (A.19)∂C7∂t= D7∂2C7∂x2− k1(6)k2C7 + 17k3C9 (A.20)∂C8∂t= D8∂2C8∂x2− k1(7)k2C8 + 17k3C9 (A.21)∂C9∂t= −k3C9 (A.22)Hemicellulose concentration and diffusivity within softwood chip are C(i, x, t) andD, respectively, where i is the DP interval ranging from degradation products (i = 1) toinsoluble chips (i = 9).255For Ω2, the dominant equations are following:dC1∞dt=h1A∗V∞(C∗1 − C1∞) + k4C2∞ (A.23)dC2∞dt=h2A∗V∞(C∗2 − C2∞)− k4C2∞ + k1(2)k2C3∞ (A.24)dC3∞dt=h3A∗V∞(C∗3 − C3∞)− k1(2)k2C3∞ + k1(3)k2C4∞ + k1(4)k2C5∞ (A.25)dC4∞dt=h4A∗V∞(C∗4 − C4∞)− k1(3)k2C4∞ + k1(5)k2C6∞ + k1(6)k2C7∞ (A.26)dC5∞dt=h5A∗V∞(C∗5 − C5∞)− k1(4)k2C5∞ + k1(7)k2C8∞ (A.27)dC6∞dt=h6A∗V∞(C∗6 − C6∞)− k1(5)k2C6∞ (A.28)dC7∞dt=h7A∗V∞(C∗7 − C7∞)− k1(6)k2C7∞ (A.29)dC8∞dt=h8A∗V∞(C∗8 − C8∞)− k1(7)k2C8∞ (A.30)dC9∞dt= 0 (A.31)The external solution is assumed to be homogeneous with hemicellulose concentra-tion C(i, t). The total aqueous phase solution has a volume V∞, and the interfacial area ofsoftwood chip is A∗.The initial condition is:C1 = 0 C2 = 0 C3 = 0 C4 = 0 C5 = 0 C6 = 0 C7 = 0 C8 = 0 C9 = C0,hwhere C0,h is the initial hemicellulose concentration at the centre line of wood chip.Boundary condition at x = 0:dC1dx= 0dC2dx= 0dC3dx= 0dC4dx= 0dC5dx= 0dC6dx= 0dC7dx= 0dC8dx= 0dC9dx= 0256Boundary condition at x = L follows Robin condition:D1dC1dx= h1(C∗1 − C1∞) D2dC2dx= h2(C∗2 − C2∞) D3dC3dx= h3(C∗3 − C3∞)D4dC4dx= h4(C∗4 − C4∞) D5dC5dx= h5(C∗5 − C5∞) D6dC6dx= h6(C∗6 − C6∞)D7dC7dx= h7(C∗7 − C7∞) D8dC8dx= h8(C∗8 − C8∞) D9dC9dx= 0The dimensionless analysis:θ =CC0,hτ =D1tL2η =xLDa1 =k4L2D1Bi1 =h1LD1Converting the dominant equations in Ω1 as following:∂θ1∂τ=∂2θ1∂η2+ Da1θ2 (A.32)∂θ2∂τ=D2D1∂2θ2∂η2− Da1θ2 + Da17k3k4θ9 + Da1k1(2)k2k4θ3 (A.33)∂θ3∂τ=D3D1∂2θ3∂η2− Da1 k1(2)k2k4θ3 +Da17k3k4θ9 + Da1k1(3)k2k4θ4 + Da1k1(4)k2k4θ5 (A.34)∂θ4∂τ=D4D1∂2θ4∂η2− Da1 k1(3)k2k4θ4 +Da17k3k4θ9 + Da1k1(5)k2k4θ6 + Da1k1(6)k2k4θ7 (A.35)∂θ5∂τ=D5D1∂2θ5∂η2− Da1 k1(4)k2k4θ5 +Da17k3k4θ9 + Da1k1(7)k2k4θ8 (A.36)∂θ6∂τ=D6D1∂2θ6∂η2− Da1 k1(5)k2k4θ6 +Da17k3k4θ9 (A.37)∂θ7∂τ=D7D1∂2θ7∂η2− Da1 k1(6)k2k4θ7 +Da17k3k4θ9 (A.38)∂θ8∂τ=D8D1∂2θ8∂η2− Da1 k1(7)k2k4θ8 +Da17k3k4θ9 (A.39)∂θ9∂τ= −Da1 k3k4 θ9 (A.40)257Converting the dominant equations in Ω2 as following:dθ1∞dτ= Bi1VchipVsolution(θ∗1 − θ1∞) + Da1θ2∞ (A.41)dθ2∞dτ= Bi1VchipVsolutionh2h1(θ∗2 − θ2∞)− Da1θ2∞ + Da1k1(2)k2k4θ3∞ (A.42)dθ3∞dτ= Bi1VchipVsolutionh3h1(θ∗3 − θ3∞)− Da1k1(2)k2k4θ3∞ + Da1k1(3)k2k4θ4∞ + Da1k1(4)k2k4θ5∞(A.43)dθ4∞dτ= Bi1VchipVsolutionh4h1(θ∗4 − θ4∞)− Da1k1(3)k2k4θ4∞ + Da1k1(5)k2k4θ6∞ + Da1k1(6)k2k4θ7∞(A.44)dθ5∞dτ= Bi1VchipVsolutionh5h1(θ∗5 − θ5∞)− Da1k1(4)k2k4θ5∞ + Da1k1(7)k2k4θ8∞ (A.45)dθ6∞dτ= Bi1VchipVsolutionh6h1(θ∗6 − θ6∞)− Da1k1(5)k2k4θ6∞ (A.46)dθ7∞dτ= Bi1VchipVsolutionh7h1(θ∗7 − θ7∞)− Da1k1(6)k2k4θ7∞ (A.47)dθ8∞dτ= Bi1VchipVsolutionh8h1(θ∗8 − θ8∞)− Da1k1(7)k2k4θ8∞ (A.48)dθ9∞dτ= 0 (A.49)Converting the initial conditions with dimensionless term:θ1 =C1C0,h= 0 θ2 =C2C0,h= 0 θ3 =C3C0,h= 0 θ4 =C4C0,h= 0 θ5 =C5C0,h= 0θ6 =C6C0,h= 0 θ7 =C7C0,h= 0 θ8 =C8C0,h= 0 θ9 =C9C0,h= 1258Boundary condition when η = 0:dθ1dη= 0dθ2dη= 0dθ3dη= 0dθ4dη= 0dθ5dη= 0dθ6dη= 0dθ7dη= 0dθ8dη= 0dθ9dη= 0Boundary condition when η = 1:dθ1dη= Bi1(θ∗1 − θ1∞)dθ2dη= Bi1h2h1D1D2(θ∗2 − θ2∞)dθ3dη= Bi1h3h1D1D3(θ∗3 − θ3∞)dθ4dη= Bi1h4h1D1D4(θ∗4 − θ4∞)dθ5dη= Bi1h5h1D1D5(θ∗5 − θ5∞)dθ6dη= Bi1h6h1D1D6(θ∗6 − θ6∞)dθ7dη= Bi1h7h1D1D7(θ∗7 − θ7∞)dθ8dη= Bi1h8h1D1D8(θ∗8 − θ8∞)dθ9dη= 0For the dominant equations, Vchip and Vsolution are constant for a given particle size.If diffusivity D and mass transfer coefficient h for all the oligomers are the same order ofmagnitude, the distribution of dimensionless concentration (θ) will be dependent on Daand Bi only.In diffusion, Biot number (Bi) is most often applied to represent the ratio of solid re-sistance to fluid resistance. When Bi is large (Bi1), the solid resistance dominates, andfluid concentration is almost constant; when Bi is small (Bi1), fluid resistance domi-nates, and the concentration of chip is almost constant.Damkohler number (Da) is the ratio of rate of reaction to rate of diffusion. WhenDa is large (Da1), reaction rate dominates; when Da is small (Da1), diffusion ratedominates.Da is determined by k, L and D; Bi is determined by h, L, D. Thickness of L andhydrolysis rate constant of k are constant for a given chip particle size therefore Da andBi will be determined by D and h. Molar mass is a factor influencing D and h (Equa-tion A.13), thus, hemicellulose oligomers with varying molar mass should have varying259D and h from interval 1 to interval 9.Since water is the solvent applied in our system and the molar volume of solute isrelated to its molar mass (Equation A.13), we assume the diffusion coefficient of hemi-cellulose oligomers will depend on the molar volume of itself in an exponential form.Assuming both D and h have similar exponential correlation formula with molar mass ofoligomers as below:D = a(M)b (A.50)h = c(M)d (A.51)where M is the molar mass of oligomer from i = 1 to i = 8.Therefore, the model is developed to describe hemicellulose concentration distribu-tion within softwood chips and bulk solution depending on Da and Bi. For same particlesize, Da and Bi are determined by D and h, which correlates to molar mass of hemicellu-lose oligomers in an exponential form.A.3 Softwood Hemicellulose Hydrolysis ModelActivation Energy CalculationThe hemicellulose hydrolysis activation energy linear regressions were calculated by Equa-tion 5.26 and plotted for 8 hemicellulose intervals in Figure A.2.2600.00210 0.00215 0.00220 0.00225 0.00230-4.5-4.0-3.5-3.0-2.5y= - 10131.860x + 19.054R2 = 0.964Intercept = 19.054  54.763Slope = - 10131.860  24767.28395% confidence limits after  lnk3 Linear fitlnk31/T(a)0.00210 0.00215 0.00220 0.00225 0.00230-9.0-8.8-8.6-8.4-8.2-8.0-7.8-7.6y= - 6913.884x + 7.012R2 = 0.999Intercept = 7.012  5.576Slope = - 6913.884  2522.04895% confidence limits after (b) lnk4 Linear fitlnk41/T0.00210 0.00215 0.00220 0.00225 0.00230-2.8-2.6-2.4-2.2-2.0-1.8-1.6-1.4-1.2-1.0y= - 8124.689x + 16.110R2 = 0.972Intercept = 16.110  38.759Slope = - 8124.689  17529.53295% confidence limits after  lnS3 Linear fitlnS31/T(c)0.00210 0.00215 0.00220 0.00225 0.00230-2.5-2.0-1.5-1.0-0.5y= - 9334.589x + 19.105R2 = 0.990Intercept = 19.105  25.924Slope = - 9334.589  11724.53895% confidence limits after (d) lnS4 Linear fitlnS41/T0.00210 0.00215 0.00220 0.00225 0.00230-2.5-2.0-1.5-1.0-0.50.0y= - 10193.026x + 21.230R2 = 0.997Intercept = 21.230  16.817Slope = - 10193.026  7605.82895% confidence limits after (e) lnS5 Linear fitlnS51/T0.00210 0.00215 0.00220 0.00225 0.00230-2.5-2.0-1.5-1.0-0.50.0y= - 10858.881x + 22.879R2 = 0.999Intercept = 22.879  9.753Slope = - 10858.881  4411.11095% confidence limits after (f) lnS6 Linear fitlnS61/T0.00210 0.00215 0.00220 0.00225 0.00230-2.5-2.0-1.5-1.0-0.50.00.5y= - 11402.925x + 24.225R2 = 0.999Intercept = 24.225  3.982Slope = - 11402.925  1800.83495% confidence limits after (g)  lnS7 Linear fitlnS71/T0.00210 0.00215 0.00220 0.00225 0.00230-2.0-1.5-1.0-0.50.00.5y= - 11862.907x + 25.364R2 = 0.999Intercept = 25.364  0.898Slope = - 11862.907  406.12295% confidence limits after (h) lnS8 Linear