Enzyme Hydrolysis of Unbleached Kraft Pulp Fiber Fines Using Membrane Bioreactor and Recycle

SUNY College of Environmental Science and Forestry Digital Commons @ ESF

Dissertations and Theses

Spring 4-9-2020

Surya Jampana SUNY College of Environmental Science and Forestry, [email protected]

Follow this and additional works at: https://digitalcommons.esf.edu/etds

Part of the Bioresource and Agricultural Engineering Commons

Recommended Citation Jampana, Surya, "Enzyme Hydrolysis of Unbleached Kraft Pulp Fiber Fines Using Membrane Bioreactor and Recycle" (2020). Dissertations and Theses. 166. https://digitalcommons.esf.edu/etds/166

This Open Access Dissertation is brought to you for free and open access by Digital Commons @ ESF. It has been accepted for inclusion in Dissertations and Theses by an authorized administrator of Digital Commons @ ESF. For more information, please contact [email protected], [email protected]. ENZYME HYDROLYSIS OF UNBLEACHED KRAFT PULP FIBER

FINES USING MEMBRANE BIOREACTOR AND RECYCLE

Surya R Jampana

A dissertation submitted in partial fulfillment of the requirements for the Doctor of Philosophy Degree State University of New York College of Environmental Science and Forestry Syracuse, New York April 2020

Department of Chemical Engineering

Approved by: Bandaru V. Ramarao, Major Professor Christina Limpert, Chair, Examining Committee Bandaru V. Ramarao, Department Chair S. Scott Shannon, Dean, The Graduate School Gary Scott, Director, Division of Engineering

Acknowledgements My sincere gratitude to Dr. Bandaru V Ramarao for his continuous guidance and support throughout this doctoral journey. Thank you for numerous discussions on science, and its philosophy, they broadened my thoughts and rationality. I am honored to graduate under your tutelage.

Dr. Deepak Kumar, thank you for your support and guidance over the past year. I really appreciate your feedback on experiments and thesis.

Dr. Rengasamy Kasinathan, thank you for economic feasibility studies of the wastes fines hydrolysis project.

Dr. Thomas Amidon and Dr. Shijie Liu, thank you for your support and being part of my steering committee.

Thanks to Mr. David Kiemle for providing assistance with NMR analysis, and Dr. Eric Finkelstein for letting me use the Syracuse Biomaterials Institute facilities at Syracuse University.

Dr. Byeong Cheol Min, thank you for teaching me the basics of waste fines enzyme hydrolysis. I learned a lot from the creative ways you set up your experiments. I miss all the fun we had in and outside the lab. Linjing Jia, thank you for helping me with the experiments, that too in a short notice.

Christopher Thomas and Dwyer Stuart, thank you so much for all the research and non-research related discussions, you enriched my thoughts and made working in the lab so lively. Many thanks to people who made my stay at Syracuse so wonderful: Dr. Bhavin Bhayani, Justin Wilson, Arjun Harur Muralidharan, Sergiy Lavrykov, Dr. Rakesh Yasarla, Jennifer Puthota, Vignesh Gnanavel, Aditi Nagardeolekar, Jinxia Yuan.

My wife Chaitu, parents, grandparents, brother, sister-in-law, Dhruv, in-laws, Rajesh, cousins in USA, thank you all for your unlimited love and affection.

Last but not the least, a big thanks to all the faculty and staff at chemical engineering department, SUNY ESF for providing the best possible working environment.

iii

TABLE OF CONTENTS

LIST OF TABLES ...... vii

LIST OF FIGURES ...... viii

ABSTRACT ...... x

CHAPTER I: INTRODUCTION ...... 1 1.1 Research Goals ...... 3 1.2 Thesis Outline ...... 3

CHAPTER II: BACKGROUND ...... 5 2.1 Lignocellulosic Biomass ...... 5 2.1.1 Cellulose ...... 6 2.1.2 Hemicellulose ...... 7 2.1.3 Lignin ...... 7 2.2 Waste Fines Rejects ...... 8 2.3 Enzymatic Hydrolysis ...... 10 2.3.1 Enzymes ...... 10 2.3.2 Hydrolysis Mechanism ...... 13 2.3.3 Factors Effecting Hydrolysis ...... 16 2.4 Membrane Bioreactor ...... 19 2.4.1 Membranes ...... 20 2.4.2 Concentration Polarization ...... 22 2.4.3 Enzyme MBR ...... 23 2.5 Zeolite...... 25 2.5.1 Zeolite β ...... 27 2.5.2 Properties of Zeolite ...... 30

CHAPTER III: UNBLEACHED KRAFT PULP HYDROLYSIS USING A MEMBRANE BIOREACTOR ...... 32 3.1 Introduction ...... 32 3.2 Materials and Methods ...... 35

3.2.1 Materials ...... 35 3.2.2 Membrane Bioreactor Experimental Setup ...... 35 3.2.3 Enzyme Hydrolysis ...... 37 3.2.4 Analytical Methods ...... 40 3.2.5 Experimental Runs ...... 41 3.3 Results and Discussion ...... 41 3.3.1 Composition of the Pulp ...... 41 3.3.2 Preliminary Experiments ...... 42 3.3.3 Effect of Dilution Rate on Hydrolysis ...... 44 3.3.4 Effect of Calcium Ions on Hydrolysis ...... 47 3.3.5 Effect of Membrane MWCO on hydrolysis ...... 50 3.4 Conclusions ...... 53

CHAPTER IV: ENZYME RECYCLE USING ZEOLITE ...... 55 4.1 Introduction ...... 55 4.2 Materials and Methods ...... 61 4.2.1 Materials ...... 61 4.2.2 Protein Quantification ...... 62 4.2.3 Adsorption of CTec2 by Zeolite ...... 62 4.2.4 Desorption of CTec2 from Zeolite ...... 63 4.2.5 Reducing Sugar Determination ...... 63 4.2.6 Enzyme Activity Assay ...... 64 4.2.7 Statistical Analysis ...... 65 4.2.8 Zeolite and Enzyme Surface Charge Measurements ...... 65 4.3 Results and Discussion ...... 65 4.3.1 Protein Content ...... 65 4.3.2 Adsorption Isotherms ...... 65 4.3.3 Enzyme-Zeolite Interaction ...... 67 4.3.4 Adsorption Kinetics ...... 71 4.3.5 Enzyme Desorption by PEG ...... 72 4.3.6 Enzyme Desorption by Alkali Elution ...... 76 4.3.7 Enzyme Desorption using NaCl ...... 77

4.4 Conclusions ...... 79

CHAPTER V: ENZYME RECYCLE MATHEMATICAL MODEL ...... 80 5.1 Introduction ...... 80 5.2 Background ...... 81 5.2.1 General Principles ...... 81 5.2.2 Langmuir Equation Models ...... 82 5.2.3 Random Sequential Adsorption (RSA) Models ...... 85 5.2.4 Kinetics of Batch Adsorption ...... 86 5.3 Results ...... 89 5.3.1 Adsorption Kinetics of Cellulase on Zeolite - Modeling of Experimental Data ...... 89 5.3.2 Adsorption Kinetics – Single Component Cellulase on Zeolite Using Random Sequential Adsorption Model ...... 91 5.3.3 Bi-component Adsorption Equilibrium ...... 95

CHAPTER VI: CONCLUSIONS ...... 99

CHAPTER VII: RECOMMENDATIONS FOR FUTURE WORK ...... 101

REFERENCES ...... 102

BIOGRAPHICAL SKETCH ...... 115

LIST OF TABLES

Table 1. Characteristics of fines rejects from containerboard ...... 10 Table 2. MBR operational strategies ...... 24 Table 3. Zeolite classification based on Si/Al ratio ...... 26 Table 4. MBR hydrolysis conditions ...... 38 Table 5. Enzyme recovery from lignocellulosic hydrolysates...... 58 Table 6. Langmuir isotherm constants at different pH ...... 66 Table 7. Effect of pH on adsorption of enzyme onto zeolite...... 68 Table 8. Langmuir isotherm constants at pH 5 and PEG 200 ...... 69 Table 9. Langmuir isotherm constants at pH 5 and PEG 20000 ...... 70 Table 10. Hydrolysis performance of recovered and unrecovered enzyme ...... 75 Table 11. Adsorption Kinetics for cellulase and PEG system on zeolite β ...... 89 Table 12. Parameters from RSA model – Best fitting ...... 92 Table 13. Kinetic and Equilibrium Parameters for calculations ...... 94 Table 14. Adsorption equilibrium for cellulase and PEG system on zeolite β ...... 96

vii

LIST OF FIGURES

Figure 1. Hydrocarbon and polyester building blocks from cellulosic biomass...... 2 Figure 2. Molecular structure of cellulose ...... 6 Figure 3. Destination of recovered paper and paperboard (in million tons) ...... 8 Figure 4. Old corrugated containers (OCC) recovery trends in USA ...... 9 Figure 5. An overview of Trichoderma reesei enzyme system ...... 11 Figure 6. Enzymatic degradation of cellulose ...... 12 Figure 7. The modular structure of CBH I from Trichoderma reesei ...... 13 Figure 8. Mechanisms of A) inverting, and B) retaining glycosidases ...... 15 Figure 9. Retention mechanism of glycosidases ...... 16 Figure 10. Overview of cellulose hydrolysis and product inhibition ...... 17 Figure 11. Alternative membrane bioreactor configurations ...... 19 Figure 12. Normal flow filtration (dead-end) Vs Tangential flow filtration (crossflow) ...... 21 Figure 13. Bronsted acid sites in zeolite β ...... 27 Figure 14. Composite building units (a, b, c) and framework (d) of zeolite β ...... 28 Figure 15. SEM image of zeolite β at 5000X magnification ...... 29 Figure 16. SEM image of zeolite β at 20000X magnification ...... 29 Figure 17. Ionization state of amino acid ...... 31 Figure 18. Membrane bioreactor experimental setup ...... 36 Figure 19. Schematic representation of membrane bioreactor ...... 37 Figure 20. Hyrdolysis yield as a function of time...... 43 Figure 21. Productivity as a function of time ...... 43 Figure 22. Reducing sugar concentration as a function of time...... 44 Figure 23. Hydrolysis yield and productivity as a function of dilution rate ...... 45 Figure 24. Reducing sugar concentration as a function of dilution rate ...... 46 Figure 25. Effect of calcium ions on hydrolysis yield and productivity ...... 47 Figure 26. Effect of surfactant on hydrolysis yield and productivity ...... 48 Figure 27. Effect of surfactant and dilution rate on hydrolysis yield and productivity ...... 49 Figure 28. Effect of membrane MWCO on hydrolysis yield and productivity at 5 FPU/g substrate...... 51 Figure 29. Effect of membrane MWCO on hydrolysis yield and productivity at 10 FPU/g substrate...... 52 Figure 30. Effect of membrane MWCO on hydrolysis yield and productivity at 15 FPU/g substrate ...... 53 Figure 31. Enzyme recycle strategies ...... 56 Figure 32. Adsorption Isotherm as a function of pH ...... 67 Figure 33. Adsorption Isotherm at pH 5 and PEG 200 ...... 69 Figure 34. Adsorption Isotherm at pH 5 and PEG 20000 ...... 70 Figure 35. Adsorption Kinetics of CTec2 on Zeolite...... 72 Figure 36. Enzyme desorption using PEG ...... 73 Figure 37. Enzyme recycle yield using PEG ...... 74

viii

Figure 38. Hydrolysis performance of the recycled enzyme ...... 75 Figure 39. Enzyme desorption as a function of pH...... 76 Figure 40. Enzyme recycle yield as a function of pH ...... 77 Figure 41. Enzyme desorption using NaCl ...... 78 Figure 42. Enzyme recycle yield using NaCl...... 78 Figure 43. Adsorption kinetics for batch adsorption of cellulase onto zeolite β (Langmuir) ...... 90 Figure 44. Adsorption kinetics for batch adsorption of cellulase onto zeolite β (RSA model) ... 92 Figure 45. Adsorption kinetics for batch adsorption of cellulase and PEG onto zeolites ...... 94 Figure 46. Adsorption Equilibrium for cellulase on zeolite β, modeled using the bi-component Langmuir isotherm ...... 96 Figure 47. Loading of PEG on zeolite β ...... 97

ABSTRACT

S. R. Jampana. Enzyme Hydrolysis of Unbleached Kraft Pulp Fiber Fines Using Membrane Bioreactor and Recycle, 127 pages, 14 tables, 47 figures, 2020.

Enhancement of hydrolysis yield is achieved by minimizing product inhibition using a membrane bioreactor. A membrane bioreactor was able to mitigate the negative effects of calcium ions on hydrolysis of waste fines and fiber rejects. Enzyme loading and membrane molecular weight cutoff (MWCO) influenced the performance of hydrolysis. At an enzyme loading below 7.5 FPU/ g substrate, a membrane bioreactor didn’t improve the hydrolysis. Better hydrolysis performance is achieved using a membrane with lower MWCO. However, at higher enzyme loadings, membrane MWCO didn’t influence the hydrolysis performance in MBR.

Enzyme in the liquid phase of the hydrolysate was efficiently recovered using zeolite. Enzyme adsorption on zeolite was driven mostly by hydrophobic interactions. Polyethylene glycol (PEG) was very effective at eluting the adsorbed enzyme. A very low concentration of high molecular weight PEG was enough to elute the proteins. The recovered enzyme retained its activity. The adsorption and desorption processes of the enzymes on zeolite β using PEG as an eluent was evaluated using mathematical modeling.

Keywords: Enzyme recycle, Membrane, Membrane bioreactor, Zeolite β, PEG

S. R. Jampana Candidate for the degree of Doctor of Philosophy, April 2020 Bandaru V. Ramarao, Ph.D. Department of Chemical Engineering State University of New York College of Environmental Science and Forestry Syracuse, New York

CHAPTER I: INTRODUCTION

The growing interest in green and sustainable technologies for the conversion of biomass to biofuels, commodity chemicals, and new bio-based materials such as bioplastics is driven by pressing need to achieve energy independence and mitigate climate change (A. Sheldon, 2014;

Lippke et al., 2011; Nanda et al., 2015; Perea-Moreno et al., 2019). The switch to renewable biomass as a feedstock can benefit the environment by reducing the carbon footprint of chemicals and liquid fuels (A. Sheldon, 2014). First-generation biomass feedstocks like sugarcane and corn interfere with the food sector, and its water consumption (especially for corn cultivation), as well as economic sustainability (Miret et al., 2016). Second-generation biomass is available year-round, and the cost incurred to collect, and store lignocellulosic biomass is low.

However, the transformation of second-generation biomass to fuels and biomaterials is economically not viable due to its recalcitrance (Martin Alonso et al., 2010). Therefore, new raw materials, like cellulosic biomass in the waste rejects from recycled linerboard mills, which circumvent the problems associated with first- and second-generation biomass is an attractive choice (Min, 2015).

Irrespective of the biomass type, cellulose chain needs to be broken into individual sugars, which then can be fermented into biofuels, and bioplastics, or transformed into hydrocarbon, and polyester building blocks by fermentation and chemocatalytic conversion

(Amidon et al., 2008; A. Sheldon, 2014; Christensen et al., 2008; Clark, 2007; Jong et al., 2012;

Tuck et al., 2012). Figure 1 depicts the valorization products of biomass to hydrocarbon and polyester building blocks. The deconstruction of cellulose chains into free monomeric sugars is achieved by a combination of pretreatment and enzymatic hydrolysis. The absence of

1 recalcitrance in waste fines avoids the requirement for a capital intensive pretreatment step before enzymatic hydrolysis (Min et al., 2015).

The enzyme system used in the deconstruction of cellulosic chain consists of a cellulase cocktail mixture that act synergistically to cleave the glycosidic bond. Despite the fact that research creating efficient enzyme systems has been accomplished, the amount of enzyme used in biomass valorization still makes the process economically infeasible. Therefore, finding new approaches that improve biocatalytic productivity by enzyme recovery and reuse appears to be an attractive option to pursue (Jørgensen and Pinelo, 2017).

Cellulosic Biomass

C and C Sugars 5 6

Fermentation Catalytic Conversion

Alcohols & Diols Isoprene Polyhydroxy- Benzene HMF

alkanoates Isobutene Toulene Furfural

Butadiene Xylene Ethylene

Farnesene

Propylene Polyesters

Butene

Butadiene

Xylene

Figure 1. Hydrocarbon and polyester building blocks from cellulosic biomass. Adapted from

(A. Sheldon, 2014)

1.1 Research Goals

The main goal of this research was to improve the biocatalytic productivity of the enzymes used in biorefinery. To accomplish this goal, the work was divided into two topics:

• Investigate the use of membrane bioreactor (MBR) for hydrolysis of waste fine rejects

• Recover enzymes in the liquid fractions of the hydrolysate using zeolite β as a carrier

1.2 Thesis Outline

This thesis comprises of seven chapters, summarized as follows:

❖ Chapter I introduces the motivation of the research, research goals, and thesis outline

❖ Chapter II discusses the substrates in biorefinery, hydrolysis of biomass by enzymes, and

strategies to optimize hydrolysis by enzyme recovery

❖ Chapter III investigates the use a reactor-separator combination (membrane bioreactor)

for the enzymatic hydrolysis of waste fines and fiber rejects from recycled linerboard

paper mills. The effect of dilution rate, Tween 80, and membrane molecular weight cutoff

(MWCO) on hydrolysis performance was investigated

❖ Chapter IV explores the potential of zeolite β to recycle cellulases from the liquid fraction

of the hydrolysate

❖ Chapter V gives a short review of adsorption models for proteins and macromolecules on

surfaces such as the zeolites

❖ Chapter VI presents the conclusions drawn from this thesis

❖ Chapter VII presents the future work

CHAPTER II: BACKGROUND

Lately, the search of sustainable resources that can reduce the dependence on fossil resources for energy demands and mitigate global environmental problems has gained importance. In this regard, replacing fossil fuels with renewable energy sources is a common goal of mankind. At present, renewable energy mainly includes biomass, geothermal energy, solar energy, wind energy, marine energy, etc. The potential ecological and environmental risks of nuclear energy, and large-scale hydropower, and the regional specificity of wind energy and geothermal energy, limits the practicality of these resources. However, biomass energy has been recognized for its ubiquity, richness, and reproducibility.

According to the International Energy Agency's 2019 Global Bioenergy statistics report, the total global energy consumption in 2017 was 370 Exa Joule (EJ), of which renewable energy accounted for 17.7% of the total energy consumption. Biomass accounted for 70% of the total renewable energy supply. Hydropower accounted for 18% of the total supply. Wind energy and solar energy technologies make little contribution to energy supply, accounting for 4% and 5% respectively, and geothermal energy only accounts for 1% (WBA, 2019).

2.1 Lignocellulosic Biomass

Biomass existing in nature can be categorized into five types: wood and woody biomass, herbaceous biomass, aquatic biomass, animal and human waste biomass, and biomass mixtures

(Tursi, 2019). Lignocellulosic biomass, comprising of woody biomass and herbaceous biomass, is an attractive renewable resource due to its abundance and nonfood use. Lignocellulosic biomass contains three major components: cellulose, hemicellulose, and lignin.

2.1.1 Cellulose

Cellulose is a homopolymer containing repeated units of cellobiose. Each cellobiose unit contains two anhydrous glucose molecules joined together by β-1,4-glycosidic linkage. The glucose units in cellobiose are rotated by 180° relative to each other providing the conformational stability to the cellulose structure. The number of repeating units in a cellulose can vary from 100 to more than 10,000. Numerous interchain and intrachain hydrogen bonds exists imparting crystallinity to cellulose. This crystalline structure makes cellulose insoluble in water. The degree of crystallinity is measured in terms of crystallinity index. An inverse relation exists between crystallinity index and digestibility of cellulose.

However, the rate of solubility can be increased by subjecting the cellulose to high temperatures and thereby breaking the hydrogen bonds. Cellulose also contains amorphous region that is relatively more susceptible to digestion. In order to achieve the effective hydrolysis preliminary pretreatment is needed prior to enzymatic hydrolysis (Kumar and Murthy, 2016; Sjostrom, 1993). The molecular structure of cellulose is represented in Figure 2.

Figure 2. Molecular structure of cellulose. Reprinted with permission from (Wustenberg, 2014)

2.1.2 Hemicellulose

Hemicellulose is the second most abundant compound which accounts 20-35 % of lignocellulosic biomass. They are linked to cellulose by hydrogen bonding, and lignin by covalent bonds. They are highly branched with a degree of polymerization ranging from 150 –

200 monomer units (Sun et al., 2003). Hemicelluloses are heterogeneous polysaccharides made up of pentoses (D-xylose, L-arabinose), hexoses (D-glucose, D-mannose, D-galactose), uronic acids (D-glucuronic, 4-O-methyl-D-glucuronic and D-galacturonic acids), and to a lesser extent,

L-rhamnose, and L-fucose. The hydrolxyl groups of the sugars can be partially substituted with acetyl groups. Xylans make up most of the hardwood and herbaceous hemicellulose. Mannans make up most of the softwood hemicellulose. The by-products of wood and pulp and paper industries are rich in xylans (Gírio et al., 2010).

2.1.3 Lignin

Lignin is a natural amorphous polymer consisting of phenylpropane units. The phenylpropane units include para-coumaryl alcohol, coniferyl alcohol, and sinapyl alcohol. The space between cellulose and hemicellulose is filled up by lignin, which attaches through hydrogen and covalent bonds. Composition and amount of lignin varies from hardwood to softwood, species to species, and even part of the plant. Softwoods contain more lignin compared to hardwood and herbaceous plants (Pandey and Kim, 2011). Overall, the structure of lignin makes it resistant to enzymatic degradation (Sjostrom, 1993).

2.2 Waste Fines Rejects

Paper and paperboard market is broadly classified into three categories: packaging, graphic paper, and others. Packing market consists of paperboard and containerboard. Newsprint, printing and writing paper are categorized under graphic paper market. Others include tissue and other products. Global paper and paperboard industry is growing despite the shrinking graphic paper segment. This growth is primarily driven by the packaging market. In 2018, the paper and paperboard industry in USA supplied 77.3 million tons of paper and paper-based products.

About 68.1 % of these products are recovered, with 19 million tons of these recovered products ending up in containerboard. The destination of the recovered paper and paperboard products are depicted in Figure 3 (AFPA, 2018).

Tissue, 4.4

Net Exports, 20.2

Containerboard, 19.1

Newsprint & Other, 2.9 Boxboard, 6.1

Figure 3. Destination of recovered paper and paperboard (in million tons). Source: American Forest and Paper Association

Over the last three decades the containerboard industry grew consistently, and the supply reached an all-time high of 35.9 million tons (AFPA, 2018). The containerboard industry is predicted to see an compound annual growth rate exceeding 2 % in North America (McKinsey,

2019). Interestingly, the recovery rate of the containerboard is very high consistently over the past decade. In 2018, the recovery rate of containerboard reached a staggering 96.4 % i.e., the paper used in the containerboard goes through the paper machine multiple times over its lifetime.

This repeated recycling shortens the paper fiber, losing their ability to bind to paper sheet. These short fibers, also called fines, need to be rejected as their net contribution to the paper become negative. The composition of these fines rejects reported by (Min et al., 2015) is summarized in

Table 1.

35 100 Recovered amount Recovery rate 90 30 80 25 70 60 20 50 15 40

10 30 20 5 recovery rate Percent million tons recovered tons million 10

0 0

1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 Year

Figure 4. Old corrugated containers (OCC) recovery trends in USA. Source: American Forest and Paper Association

Table 1. Characteristics of fines rejects from containerboard. Adapted from (Min, 2017) Component % oven dry weight Cellulose 30 Hemicellulose 10 Lignin 4 Fats 5 Ash 41 Others (plastics, synthetic fibers, other organics) 10

These fines rejects, along with plastics, sand, clay, and other paper constituents are either incinerated or disposed of in landfill (Monte et al., 2009). In case of landfill, the average tipping fees is $ 34 per ton, with highs reaching $ 100 per ton (Scott et al., 1995). Alternatively, these fines rejects are rich in cellulose and can be a good source for producing sustainable products like biofuels, platform chemicals and bioplastics. These fines rejects have advantage of the cellulose being easily accessible to enzymes for quick hydrolysis.

2.3 Enzymatic Hydrolysis

2.3.1 Enzymes

In nature, microorganisms produce different kind of enzymes to degrade biomass for producing monomeric sugars for their survival (Balan, 2014). These enzymes vary structurally from α-helical to β-sheet with the size ranging from 14,000 – 170,000 Daltons (Withers, 2001).

Microbes having these enzymes act upon biomass in two ways: cellulosomal enzyme system and free enzyme system. In cellulosomal enzyme system, as in the case of Clostridium thermocellum, hydrolysis is performed by a complex mixture of enzymes that are docked to cohesive and dockerin domains anchored on the surface of the organisms (Bayer et al., 1998). Cellulosomal

10 enzyme systems are predominantly observed in anaerobic bacteria and fungi. They contains cellulases, hemicellulases, pectinases, and chitinases that can efficiently degrade biomass (Doi and Kosugi, 2004). In free enzyme system, as in the case of Trichoderma reesei, enzymes secreted as individual components act on biomass substrates (Martinez et al., 2008). An overview of the Trichoderma reesei enzyme system is represented in Figure 5. The free enzyme system is easy to duplicate and hence, widely employed in biomass conversion in biorefineries

(Balan, 2014).

Figure 5. An overview of Trichoderma reesei enzyme system. Reprinted with permission from (Kumar and Murthy, 2016)

Enzymes degrading biomass can be classified into three major categories: cellulases, hemicellulases, and pectinases. Cellulases degrade cellulose, hemicellulases degrade

11 hemicellulose, and pectinase degrade pectin (Bayer et al., 1998; Zhang et al., 2012). These enzymes act in synergy to break down the cellulose polymer (Figure 6).

Traditionally, cellulolytic enzymes are classified into three categories (Jørgensen et al.,

2007).

i. Exo-1,4-β-D-glucanases or Cellobiohydrolases (CBH): They cleave off the cellobiose

units at the ends by moving along the cellulose chain. Cellobiohydrolase I (CBH I) acts

on the reducing end of the cellulose chain whereas cellobiohydrolase II (CBH II) acts on

the non-reducing end of the cellulose chain.

ii. Endo-1,4-β-D-glucanases (EG): Hydrolysis of internal β-1,4-glycosidic bonds along the

cellulose chain is performed by these enzymes. However, they act in a random order. iii. 1,4-β-D-glucosidases (βG): They act by hydrolyzing the cellobiose units into glucose

units.

Figure 6. Enzymatic degradation of cellulose. Reprinted with permission from (Horn et al., 2012). Abbreviations: EG, endoglucanase; CBH, cellobiohydrolase; CDH , cellobiose-dehydrogenase; CBM, carbohydrate-binding module; C1GH61, C1 oxidizing EG; C4GH61, C4 oxidizing EG

2.3.2 Hydrolysis Mechanism

The hydrolysis of the cellulose chain is aided by three separate domains present in the enzyme molecule. Carbohydrate binding module (CBM), the smaller globular domain, facilitates the attachment of enzyme to the cellulose substrate. Catalytic domain (CD), the larger globular domain, has an active site inside a tunnel formed by a loop of protein chain that is responsible for hydrolysis of the glycosidic bond. These two globular domains are linked by the linker segment, the third domain (Zhong et al., 2009). These domains are illustrated in Figure 7: bigger domain on right side is CD, smaller domain on left side is CBM, linker segment in the middle connects CBM and CD.

Figure 7. The modular structure of CBH I from Trichoderma reesei. Reprinted with permission from (Zhong et al., 2009)

The fragmentation of cellulose chain by enzymes into glucose monomers involves six main steps

(Bansal et al., 2009):

1) Enzyme adsorb onto cellulose with the help of binding domain

2) The bond susceptible to hydrolysis is located on the cellulose surface. Exoglucanases

locate bond at the chain ends, while endoglucanases locate bond randomly across the

cellulose chain

3) The cellulose chain end is moved into the catalytic tunnel, resulting in the formation

of enzyme-substrate complex, to initiate hydrolysis

4) Enzyme slides along the cellulose chain upon hydrolysis of the β-glycosidic bond

5) Enzyme is desorbed or Step 4 is repeated. Step 2 or step 3 is repeated if catalytic

domain is detached form the cellulose

6) The available cellobiose units are hydrolyzed into glucose monomers in the presence

of β-glucosidase

Based on their mechanism, enzymes are classified into glycosidases and glycosyl transferases. Enzymes that hydrolyze the glycosidic bond with the retention of anomeric configuration are called glycosidases and those that hydrolyze by inversion of anomeric configuration are called glycosyl transferases. Hydrolysis mechanism of these two classes of enzymes are presented in Figure 8.

In inverting mechanism, one carboxylic group acts as an acid catalyst, while the other one acts as a base catalyst. In Figure 8 (A), the top carboxyl group protonates the glycosidic oxygen, while the water molecule is attacked by the bottom carboxyl group, that acts as a nucleophile.

Both carboxyl groups attain opposite protonation state to that before hydrolysis (Davies and

Henrissat, 1995; Payne et al., 2015; Withers, 2001).

Figure 8. Mechanisms of A) inverting, and B) retaining glycosidases. Adapted from (Payne et al., 2015)

In the retention mechanism, the hydrolysis of glycosidic bond proceeds in two steps.

However, the acid/base catalysis is done by a single carboxyl group. In the first step, the glycosidic oxygen is protonated by the acid catalyst resulting in the formation of a glycosyl- enzyme intermediate (glycosylation). In the second step, carboxyl group acts as a base catalyst resulting in the deglycosylation. Both these steps proceed through a transition state that forms a oxocarbenium ion (Bravman et al., 2003). Unlike in the inverting mechanism, both carboxyl groups retain the same protonation state to that before hydrolysis (Davies and Henrissat, 1995;

Withers, 2001).

Figure 9. Retention mechanism of glycosidases. Reprinted with permission from (Bravman et al., 2003)

2.3.3 Factors Effecting Hydrolysis

Product Inhibition

Like most enzyme-catalyzed reactions, the presence of a high enough concentrations of products (cellobiose and glucose) acts as an inhibitor in cellulolytic hydrolysis reaction.

Cellobiose inhibits CBH and EG for hydrolysis of cellulose to cellobiose. Glucose, the product of βG acts as an inhibitor for hydrolysis of cellobiose to glucose. Also, βG indirectly inhibits the

16 hydrolysis of cellulose to cellobiose by accumulation of cellobiose (Andrić et al., 2010). The schematic inhibition of cellulolytic enzymes on cellulose hydrolysis is presented in Figure 10.

Figure 10. Overview of cellulose hydrolysis and product inhibition: 1&2 are main reactions, 3,4&5 are inhibition reactions. Reprinted with permission from (Andrić et al., 2010a)

Biomass Composition

Composition of biomass influences cellulase hydrolysis performance. Substrates with higher content of lignin tend to show an inverse relation to hydrolysis performance (Vinzant et al.,

1997). It is hypothesized that lignin hinders the hydrolysis in two ways (Berlin et al., 2005):

1) Preferential interaction of enzyme to lignin due to hydrophobicity inducing non-

productive binding

2) Reduced enzyme access to cellulose chain due to physical blocking

Hydrophobic, electrostatic, and hydrogen bonding interactions are proposed to be the binding force for enzyme-lignin adsorption (Li et al., 2016; Pareek et al., 2013). This extent of adsorption depends both the type of biomass, and the kind of enzyme. Lignin in pretreated biomass is more inhibitory than non-pretreated biomass, and lignin from herbaceous plants is more inhibitory than wood (Li et al., 2016). The negative effect of lignin on cellulose hydrolysis is countered by the addition of surfactant. Nonionic surfactants, like tween 80 and PEG, compete

17 with the enzyme for lignin binding sites, and reduce the non-productive binding of enzyme to the lignin (Qing et al., 2010; B. Sipos et al., 2010). Nonionic surfactants are generally considered to be more benign to enzymes than ionic surfactants. Nonionic surfactant interact with enzymes through hydrophobic interactions. Whereas the ionic surfactants, like sodium dodecyl sulfate, bind to enzymes through strong electrostatic interactions, changing the conformation leading to reduced enzyme activity (Eriksson et al., 2002; Holmberg, 2018).

Temperature and pH Effect

As discussed before, a mixture of enzymes is usually required to hydrolyze biomass.

Besides the type of substrate, activity of these enzymes varies depending on temperature and pH.

Enzyme activities are found to be higher in the range of 37 °C to 60 °C and pH 4 to 5.5 (Farinas et al., 2010). The activity of enzyme is dependent on both temperature and pH. For instance, the enzyme mixture Celluclast exhibits higher activity up to almost 60 °C at pH 4.5, whereas at pH

6.5, the enzyme activity starts to decrease after 50 °C. Under optimum temperature and pH conditions, the activity of endoglucanase Cel8A is higher on barley-β-glucan than azo-CM- cellulose. The substrates dissimilar structure might affect the binding of enzyme and hence this variation (Herlet et al., 2017). Therefore, for a given substrate, the enzyme mixture must be tested for optimum temperature and pH to attain maximum hydrolysis.

2.4 Membrane Bioreactor

Membrane bioreactors (MBR) are reactor-separator units where the separation unit is configured either internally or externally. The schematic of MBR configuration is presented in

Figure 11. In a submerged configuration, the separation unit (membrane) is integrated within the reactor. The separation unit is coupled externally with the bioreactor in an external or sidestream configuration (Guglielmi and Andreottola, 2010). Typically, the separation unit is a membrane that allows only some components of the feed material to pass through. The components that pass through the membrane are called permeate and those retained by the membrane are called retentate. Membrane permeability depends on the membrane pore size and characteristic of the feed stream (such as shape of the molecule and inter-molecular and membrane-molecular interactions) (Judd, 2010, and Ramaswamy et al., 2013).

Figure 11. Alternative membrane bioreactor configurations

2.4.1 Membranes

Basically, there are four different types of membrane separation processes ranging from: microfiltration (MF) to ultrafiltration (UF) to nanofiltration (NF) to reverse osmosis (RO).

Microfiltration retains suspended solids and colloid matter (such as extractives) in the range of

0.1 to 1 µm. UF membranes retains macromolecules (such as hemicelluloses, proteins) in the range of 5 to 100 nm. Concentration of monosaccharides, multivalent inorganic ions and low- molar-mass lignin can be achieved by NF having a pore size of 1 – 5 nm. RO can reject singly charged (i.e. monovalent) ions, such as sodium (Na+) and chloride (Cl-). MF and NF membranes are classified based on pore size and molecular weight cut-off (MWCO) respectively. UF membrane has the capability of rejecting molecules in the range of 1-1000 kDa. The driving force is usually 2-10 bar during UF and less than 2 bar during MF (Judd, 2010 and Ramaswamy et al., 2013).

Membranes are usually made up of polymeric or ceramic material. Although metallic membranes are used in membrane separations, they are not widely used in MBR technology. The most common polymeric membrane materials include polysulfone (PS), polyethersulfone (PES), polyvinylidene fluoride (PVDF), polyethylene (PE) and polypropylene (PP) and regenerated cellulose. Ceramic membranes are usually made of α-Al2O3 and TiO2. By and large, all membranes comprise of a thin layer surface layer on the top of the more open, thicker porus support. The top layer provides the selective permeability of the components while the bottom porus support provides mechanical stability. The mechanical and chemical stability of the membranes are specified in terms of the maximum pressure and temperature it can withstand.

The chemical stability is given as the pH range and the resistance to solvents. Temperature and

pH resistance are generally higher for ceramic membranes, whereas the maximum pressure is

usually higher for polymeric materials (Judd, 2010; Ramaswamy et al., 2013).

The configuration of the membrane and the way in which the membranes are housed to

produce membrane modules play a crucial role in determining the overall process performance.

The primary membrane configurations employed in MBRs are plate-and-frame/flat sheet (FS),

hollow fibre (HF) and (multi) tubular (MT). HF is comprised of cylindrical bundles with a high

packing density, while MT has one or more membrane tubes. FS are usually produced in

rectangular panels. MT module operate with flow passing from inside (lumen-side) to outside

(shell-side) whereas HF operates in both inside to outside and outside to inside. The narrow

diameter of the HF modules in MBR is susceptible to clogging in case of inside to outside flow

pattern (Judd, 2010).

Figure 12. Normal flow filtration (dead-end) Vs Tangential flow filtration (crossflow). Source: Millipore

Conventional pressure-driven membrane processes operate in two modes: dead-end and crossflow filtration. In the dead-end filtration, the feed flow is perpendicular to the membrane and is characterized by the absence of retentate stream. The resistance of the membrane increases according to the thickness of the cake layer formed. In the crossflow filtration, the feed flow is parallel to the membrane and characterized by the presence of retentate stream. In crossflow filtration, the deposition continues until the adhesive forces binding the cake to the membrane are balanced by the scouring forces of the fluid passing over the membrane (Ramaswamy et al.,

2013).

2.4.2 Concentration Polarization

The overall resistance at the membrane-solution interface is increased by a number of factors such as concentration polarization and fouling. Concentration polarization (CP) is the accumulation of solute at the membrane-solution creating a liquid film. CP can eventually result in fouling of the membrane pores. CP and CP-related fouling can be reduced by increasing the turbulence. For a submerged MBR, increased membrane aeration is employed to decrease the CP and CP-related fouling to an extent. In MBRs, physical cleaning is achieved either by backflushing or relaxation. In backflushing, the direction of permeate is reversed for a small interval whereas in relaxation, the permeation is ceased with simultaneous scouring of membrane with air bubbles. Physical cleaning is employed to eradicate reversible fouling. Chemical cleaning using mineral or organic acids, caustic soda, or sodium hypochlorite is employed to decrease otherwise irreversible fouling. However, original permeability of the membrane is not achieved either by physical or chemical cleaning (Judd, 2010; Ramaswamy et al., 2013)

2.4.3 Enzyme MBR

Minimizing product inhibition is the major goal of the enzyme membrane bioreactors reported in literature. Both submerged and external MBRs configurations were studied for this purpose. An approach to re-use the enzyme separated from products was first initiated by (Ghose and Kostick, 1970). Their design of membrane reactor used a pressure filtration vessel provided with a membrane for simultaneous removal of glucose from the reacting system. They reported that there was an increase in the substrate conversion ratio when sufficient time was given for enzymatic hydrolysis. Since then, both submerged and external MBRs have been operated using a combination of substrate loading, enzyme loading, and product removal strategies. These MBR operational strategies were summarized in Table 2.

Table 2. MBR operational strategies Substrate Substrate MBR configuration Product Substrate Enzyme Reference feeding removal conc. loading, per (%w/v) g substrate Solca Floc Continuous Submerged Continuous 10 300 mg (Ghose and Kostick, 1970) Solca Floc Start Separate Continuous 1 22 mg (Klei et al., 1981)

α-cellulose Start Submerged Intermittent 2.5 4 -10 mg (Gan et al., 2002)

Corn stover Start Separate Intermittent 15 20 FPU (Knutsen and Davis, 2004)

Rice straw Start Separate Continuous 18.5 20 FPU (Yang et al., 2006)

Solca Floc Continuous Submerged Continuous 2.5 15 FPU (Bélafi-Bakó et al., 2006)

Amino acid- Fed-batch Submerged Intermittent 2 60 FPU (Zhang et al., 2011) soaked corn stover Spruce chips Continuous Separate Continuous 14.4 35 FPU (Ishola et al., 2013)

α-cellulose Continuous Submerged Continuous 10 50 mg (Malmali et al., 2015)

Milled corn Continuous Separate Continuous 2.5 - 5 20 mg (Stickel et al., 2018) stover

Dilute acid Continuous Submerged continuous 15.6 FPU (Mahboubi et al., 2020) pretreated wheat straw

2.5 Zeolite

Zeolite is a combination of two Greek words, zeo and lithos. Zeo means boil and lithos means stone. Zeolites are three-dimensional crystalline aluminosilicates made up of SiO4 and

AlO4 tetrahedra units linked to each other by oxygen atoms. Each AlO4 tetrahedron in the framework bears a negative charge which is balanced by extra-framework cation. The channels

+ + + + and voids in the framework are occupied by cations such as H , Na , K or NH4 and water molecules. These cations are mobile and provide zeolite its ion exchange capabilities. The water in the framework may make up to 50 % of the crystal volume. This water is generally removed by application of heat. The structural formula of zeolite crystallographic unit cell is given by

Mx/n[(AlO2)x(SiO2)y] (H2O)Z

Where, n is the valence of cation M

- x is the number of AlO4 tetrahedra per unit cell

y is the number of SiO4 tetrahedra per unit cell z is the number of water molecules per unit cell

The ratio of y to x varies with the type of zeolite. However, the ratio is always ≥ 1. While

- zeolites with pure SiO4 in the framework can be synthesized, constructions with AlO4 alone in the framework do not exist. Besides Al and Si, the tetrahedral position in the framework can consist of P (Phosphorus), Ge (Germanium), B (Boron), Mg (Magnesium), Zn (Zinc), Ga

(Gallium), Be (Beryllium).

- The primary structural units (PSU), SiO4 and AlO4 tetrahedra, are assembled to form secondary building units (SBU). The SBU units are polyhedra and their structure can vary from cube to hexagonal to cubo-octohedra. These SBU units combine to form the composite building

unit (CBU) (McCusker and Baerlocher, 2001). Based on the CBU framework, currently, there

are 237 zeolites identified by the Structure Commission of the International Zeolite Association

(IZA) (DZS, 2020).

The pore size of the framework ranges from 0.3 nm to 1 nm. The pore volumes range

from 0.10 cc/g to 0.35 cc/g. Each framework in the database is given a three letter code such as

FER, *BEA, -CHI etc.,. The asterisk “*” symbol preceding the three-letter code indicates that

the zeolite structure is partially disordered. A “-” symbol preceding the three-letter code indicates

that the framework is interrupted. Irrespective of the composition, the three-letter code is

assigned to the zeolite solely based on the framework topology (Flanigen, 2001).

Based on the ratio of Si to Al in the framework, the zeolites are classified into four

categories (Flanigen, 2001). These categories are presented in Table 1. With increasing Si to Al

ratio, the thermal stability, and hydrophobicity of the zeolite increases .

Table 3. Zeolite classification based on Si/Al ratio Name Si : Al Examples

Low Si/Al zeolites 1 – 1.5 zeolite A, zeolite X

Intermediate Si/Al zeolites ~2 - 5 clinoptilolite, zeolite Y

High Si/Al zeolites ~10 - 100 mordenite, erionite

Silica Molecular Sieves silicalite

2.5.1 Zeolite β

Zeolite β is a synthetic zeolite with a partially disordered framework. It is highly hydrophobic and thermally stable due to its high Si/Al ratio. The Si/Al ratio of Zeolite β is greater than 5. The framework of Zeolite β has two 12-membered ring pores. The channels of the framework have a diameter of 0.56 x 0.56 nm and 0.77 x 0.66 nm (Bárcia et al., 2005). Zeolite β used in the chapter 4 has an ammonium ion neutralizing the zeolite framework. The acid sites and scanning electron microscopy (SEM) images of the Zeolite β used in this study is presented in Figure 13 and Figure 15, respectively

Figure 13. Bronsted acid sites in zeolite β

(a) (b) (c)

(d)

Figure 14. Composite building units (a, b, c) and framework (d) of zeolite β. Source: Structure Commission of the International Zeolite Association

Figure 15. SEM image of zeolite β at 5000X magnification. Courtesy of N. C. Brown Center, ESF

Figure 16. SEM image of zeolite β at 20000X magnification. Courtesy of N. C. Brown Center, ESF

2.5.2 Properties of Zeolite

Ion Exchange

The isomorphous replacement of the tetravalent framework cation (i.e., Si) by a cation of lower charge (normally Al) creates a net negative charge on the zeolite framework. This net negative charge is neutralized by the cations present in the pores. These weakly bonded cations are exchanged by washing with a solution containing another cation in excess explaining the ion exchange capabilities of the zeolite (Townsend and Coker, 2001).

Adsorption

Zeolites are used as adsorbents in several domains of applications. The ability to adsorb depends on the physical/chemical characteristics of the zeolite, such as Si to Al ratio, framework cation, presence of defects in the framework etc. In the past decades, zeolites were used to remove contaminants like ammonium and heavy metals in the waste water by adsorption

(Hedström Annelie, 2001; Kesraoui‐Ouki et al., 1994; Wang and Peng, 2010). Recent applications include removal of micropollutants like tris-2-chloroethyl phosphate (TCEP), an organo-phosphate ester used as a plasticizer and flame retardant, from the aqueous solutions

(Grieco and Ramarao, 2013).

Amino acids, the building blocks of proteins are produced in a commercial scale by chemical or enzymatic synthesis and microbial fermentation. All these methods involve the separation of amino acids from aqueous phase. Amino acid separation and recovery is usually done by chromatography (Munsch et al., 2001). But in the past decade, zeolites were tested as a chromatographic carrier to separate and recover amino acids. Adsorption was the basic principle employed to separate amino acids from the solution using zeolites.

Amino acids are organic compounds consisting of a basic amino and acidic carboxyl group along with a characteristic side chain. The basic amino group and acidic carboxyl group condense to a from a peptide bond. The ionization state of an amino acid varies with the pH. At high pH, the carboxyl group tends to dissociate, giving the amino acid a negative charge. At low pH, the amine group tends to acquire positive charge. At an interim pH, the net charge on the amino acid becomes neutral. This pH is called isoelectric point (Krohn and Tsapatsis, 2005). The ionization state of an amino acid is represented in Figure 17

Figure 17. Ionization state of amino acid. Reprinted with permission from (Krohn and Tsapatsis, 2005)

The ionization behavior is exploited to separate amino acids from aqueous solutions using zeolites. Basic amino acids (lysine, arginine), acidic amino acids (glutamic acid), non-polar amino acid (phenylalanine) were all adsorbed onto zeolite β by varying pH. Amino acids adsorbed onto negatively charged zeolite through electrostatic and hydrophobic interactions

(Chiku et al., 2003; Krohn and Tsapatsis, 2005; Munsch et al., 2001).

CHAPTER III: UNBLEACHED KRAFT PULP HYDROLYSIS USING A MEMBRANE BIOREACTOR

3.1 Introduction

There is a growing interest in the utilization of non-fossil-based resources, like lignocellulosic biomass, to create sustainable products like biofuels, platform chemicals, and bioplastics. To produce these products, the structural carbohydrates (cellulose and hemicellulose) of lignocellulosic biomass need to be transformed into fermentable sugars via hydrolysis (acid or enzymatic). As these carbohydrates are embedded in a complex matrix protected by lignin, a pretreatment step is necessary to break down the biomass recalcitrance and improve enzyme accessibility to these carbohydrates. Due to the advantages of mild operating conditions, low inhibitory compound formation, enzymatic hydrolysis is preferred over the acid hydrolysis process. Each of these steps has some bottlenecks for the cost-effective utilization of lignocellulosic biomass. Biomass that avoids any one of these steps will be a better source to pursue the production of fermentable sugars.

Waste fines and fiber rejects from recycled linerboard paper mills, is one such source which contains about 40 % cellulose . Unlike wood and agricultural waste biomass like corn stover or straw, the cellulose and hemicellulose in waste fines are easily accessible to enzymes and hence, does not require pretreatment. Currently, some paper companies dispose of these waste fines at the cost of $ 25 - $ 80 per wet ton (Min et al., 2015). The consistent composition of these waste fines from the paper mill, year-round availability, and negative cost associated with acquiring these waste fines make this a potential substrate for developing cost-effective fermentable sugar production.

Hydrolysis of the cellulose chain is performed by a mixture of enzymes namely: endoglucanases, exoglucanases, and β-glucosidases. These enzymes, with a molecular weight ranging from 15,000 to 170,000 Da, act on cellulose chain synergistically to produce sugar monomers (Withers, 2001). However, the cost of the enzymes and the slow reaction rate due to product inhibition is one of the major obstacles for the cost-effective hydrolysis of cellulose (Al-

Zuhair et al., 2013; Gan et al., 2002; Liu et al., 2016; Pino et al., 2018). Several strategies, like enzyme recycling and enzyme immobilization, were implemented to improve the enzyme catalytic activity and reduce the enzyme costs (Jørgensen and Pinelo, 2017). Although enzyme recycling by ultrafiltration is efficient, studies supporting the economic feasibility of this process is not reported with an accounting for the costs associated with membranes and energy consumption. Enzyme immobilization induces mass transfer limitations hindering the adsorption of enzyme to the cellulose. Simultaneous saccharification and fermentation (SSF), where enzymatic hydrolysis and fermentation is performed in the same reactor, is a common strategy employed to mitigate product inhibition (Hahn-Hägerdal et al., 2006). But the lower temperature requirement for fermentation is not optimum for the enzymes and compromises the rate of enzyme hydrolysis. Therefore, a certain degree of separate enzymatic hydrolysis that combines enzyme recycle and reduces product inhibition might be the most feasible approach for cost- effective production of sugars (Andrić et al., 2010b).

Selective retention of the enzyme with simultaneous removal of the product by attaching a membrane to the hydrolysis reactor will be a preferred approach that can possess the capabilities of the enzyme recycle strategy while overcoming the limitations associated with product inhibition. The attachment of the membrane to the reactor is achieved in two configurations: internal and external. In an internal membrane bioreactor (MBR), also called

33 submerged MBR, the membrane unit is placed inside the reactor to form a single unit (Alfani et al., 1982; Gan et al., 2002; Ghose and Kostick, 1970). In an external MBR, the membrane unit is operated outside of the reactor unit (Ishola et al., 2013; Knutsen and Davis, 2004; Yang et al.,

2006). Initial research using a membrane reactor was carried out in an internal MBR fitted with ultrafiltration membrane using pure cellulose as the substrate. In the later research, external

MBR is predominantly used for hydrolysis of lignocellulosic biomass where undigestible lignin components are separated initially using microfiltration and permeate containing enzyme was cutoff and recycled into reactor using ultrafiltration. In the current research, synthetic mixture mimicking the qualities of the waste fines rejects from old corrugated was chosen as a substrate.

Although the composition of the fines includes 40 % of the easily accessible sugars, significant quantities of mineral filler interfere with the hydrolysis. Mainly, precipitated calcium carbonate (PCC) provides a surface for competitive and non-productive binding of enzymes, hindering the hydrolysis (Min et al., 2015). Tween 80, a nonionic surfactant can be used to minimize the non-productive binding and promote enzymatic hydrolysis enhancement (Chu and

Feng, 2013; Eriksson et al., 2002; Kim et al., 2007, 2006; Min et al., 2015; Wu and Ju, 1998).

(Min et al., 2015) reported that PCC inhibited the hydrolysis by altering the pH from the optimal value of the hydrolysis. The inhibitory effect of PCC shifting the pH from optimal hydrolysis conditions could be improved by pH adjustment (PA) with acetic acid (Min et al., 2018).

However, calcium ions created in the PA method promoted enzyme aggregation, another type of inhibition related to enzyme deactivation hindering the hydrolysis (Min, 2018; Wang et al.,

2012). By reducing the concentration of calcium ions, there is a potential to reduce negative effect of calcium ions on enzyme hydrolysis. The operational characteristics of MBR can facilitate the removal of calcium ions and thus, improve hydrolysis.

The primary goal of this study is to improve hydrolysis yield and productivity of waste fines and fiber rejects from recycled linerboard paper mills using internal MBR. The proposed membrane bioreactor configuration is anticipated to reduce product inhibition and calcium ion concentration, thereby increasing the hydrolysis efficiency (yield and rate of hydrolysis). The effect of dilution rate, Tween 80, and membrane molecular weight cutoff (MWCO) on hydrolysis performance was investigated.

3.2 Materials and Methods

3.2.1 Materials

A synthetic mixture representing the composition of fines was made by mixing unbleached softwood kraft pulp (USKP) and precipitated calcium carbonate (PCC, Speciality

Minerals Inc., Bethlehem, PA) in the ratio of 7:3. This synthetic mixture was used as a model of paper fines for all hydrolysis experiments. Commercial cellulase enzymes, Cellic CTec2 kindly provided by Novozymes, USA, was used in all enzyme hydrolysis experiments. Sodium Acetate was purchased from Sigma-Aldrich, St. Louis, MO. Acetic acid was purchased from Fischer

Scientific, Fair Lawn, NJ. Sulfuric acid was purchased from Pharmco Products Inc., Brookfield,

CT. Tween 80 was purchased from Amresco, Solon, OH. The unbleached softwood kraft pulp

(USKP) sample was milled in Wiley Mill and screened through a 35-mesh sieve. Milled sample was stored in air-tight plastic bottles.

3.2.2 Membrane Bioreactor Experimental Setup

As shown in Figure 18, the membrane bioreactor experimental setup consisted of a membrane bioreactor, a container to enclose membrane bioreactor, a water bath, a peristaltic pump to exchange water between the water bath and the container, and a compressed air tank to

35 pressurize the membrane bioreactor. A dead-end ultrafiltration stirred cell, Amicon Model 8400

(Millipore Sigma, Burlington, MA) with a maximum holding volume of 350 mL was used as a membrane bioreactor to perform enzymatic hydrolysis. A flat sheet polyethersulfone (PES) membrane (Sartorius Stedim Biotech, Goettingen, Germany) with an effective area of 4.53*10-3 m2 was employed to retain enzymes and remove product. A stir plate was used to mix the contents of the membrane bioreactor by the action of the magnetic stirrer suspended in the reactor.

Figure 18. Membrane bioreactor experimental setup

Figure 19. Schematic representation of membrane bioreactor (A: Amicon cell, B: Membrane, C: Permeate collection, D: Water bath, E: Pressure inlet, F: Magnetic stirrer, G: Stir plate)

3.2.3 Enzyme Hydrolysis

Enzymatic hydrolysis was carried out in a membrane bioreactor with a working volume of 150 mL at a substrate loading of 5 % (w/v) at 50 °C. The solution was prepared using 50 mM sodium acetate buffer to maintain pH at 5. To determine the hydrolysis efficiencies under various conditions, experiments were carried with an enzyme loading of 5 FPU, 10 FPU and 15 FPU per gram of substrate, and PES membrane with a molecular weight cutoff (MWCO) of 10,000 Da,

30,000 Da, and 50,000 Da. Experimental setup conditions are presented in Table 4.

Table 4. MBR hydrolysis conditions Parameter Value Units

Solid loading rate 5 % (w/v)

Enzyme loading rate 5 / 7.5 / 10 / 15 FPU/g substrate

Working volume 150 mL

Membrane MWCO 10 / 30 / 50 kDa

pH 5

Temperature 50 °C

Dilution rate 0.25 / 0.5 h-1

Except for control, all experiments were carried out in a fed-batch mode with the supplementation of buffer only every 1 hour. All of the substrate and enzyme were loaded at the start of the hydrolysis. The hydrolysis products were removed by applying pressure at the intervals of 1 hour, and an equal amount of buffer was loaded to maintain a constant volume.

The amount of product removed can be defined by dilution rate, equation (1). Hydrolysis experiments were carried out for 12 hours at dilution rates of 0.25 h-1 and 0.5 h-1. Hydrolysis yields and productivity at each hour was calculated according to the equation 2 and equation 3, respectively.

푄퐷 ( 1 ) 푓푄 = 푉푅

Here, 푓푄 is a dilution rate given as the ratio of the dilution flow, 푄퐷 (L/h) to the reactor volume,

푉푅, (L).

푚 ∗ 0.9 ( 2 ) 푌 = 푅푆 푚푆

Where, Y is the hydrolysis yield, given as a ratio of cumulative mass of released reducing sugar,

mRS (g) to mass of substrate, mS (g).

푑 1 ( 3 ) 푃 = ( 푚푃) 푑푡 푉푅

Productivity, P is the rate of sugar formation in the reactor. The units of productivity are mg L-1

-1 min . mp is the mass of sugar released, VR is the volume of the reactor, and t is the time. The time derivative can be approximated by a finite difference equation and the reactor volume (VR) is usually a constant. This gives

1 ∆푚 ( 4 ) 푃 = 푃 푉푅 ∆푡

with mP being the incremental mass of sugar produced over a time interval t.

3.2.4 Analytical Methods

The pulp composition was analyzed using biomass analytical procedures of US National

Renewable Energy Laboratory (NREL) (Sluiter et al., 2008). 300 ± 10 mg of wheat straw was dissolved in 3 ± 0.01 mL of 72% sulfuric acid and incubated at 30 °C for 60 minutes. The acid concentration was diluted to 4% by addition of 84 ± 0.04 g distilled water and tubes were inverted several times to eliminate phase separation. Hydrolysis solution was vacuum filtered to collect the acid insoluble lignin. The hydrolysate was collected and measured for acid soluble lignin. Solids remaining in the pressure tubes were carefully transferred to the filter crucible by adding the required amount of distilled water. The acid insoluble lignin was measured by drying the crucibles containing solid in the oven at 105 °C. The acid soluble lignin was measured in UV

- Visible spectrophotometer (GENESYS 50) at 278 nm. Structural carbohydrates and acetic acid were determined by NMR.

Reducing sugars, glucose and xylose in the hydrolysate were quantified using Nuclear

Magnetic Resonance (NMR) spectroscopy. Glucosamine HCl (0.2 M) dissolved in deuterium oxide was used an internal standard. Internal standard (100 µL) and hydrolysate (900 µL) were added to the NMR tube and mixed thoroughly by vortexing. The concentrations of the individual sugars were measured by comparing with the calibration curves of glucose and xylose.

Cellulase activity in terms of filter paper units (FPU) was measured according to the

National Renewable Energy Laboratory (NREL) method (Adney and Baker, 2008). Whatman

No. 1 filter paper strip (50 mg) used as a substrate was saturated with 1 mL 0.05 M sodium acetate buffer, then equilibrated to 50 °C. A set of dilute enzyme solutions were made, with one dilution targeted to release slightly more than 2 mg of glucose and one slightly less than 2 mg of glucose. Simultaneously, glucose standards and enzyme blanks were prepared. 0.5 mL of these

40 diluted enzyme solution were added to saturated solution and were incubated at 50 °C for 60 minutes along with glucose standards and enzyme blanks. The reaction was terminated by adding

3 mL of DNS reagent, boiled for 5 minutes in vigorously boiling water to develop color formation followed by ice cooling. Absorbance was measured at 540nm. Using the standard curve, amount of glucose released from each enzyme assay tube were determined. Filter paper activity was calculated using equation (4)

0.37 퐹푃푈 ( 5 ) 퐹푃푈 = , 퐸푛푧푦푚푒 푐표푛푐푒푛푡푟푎푡푖표푛 푟푒푙푒푎푠푖푛푔 2 푚푔 표푓 푔푙푢푐표푠푒 푚퐿

3.2.5 Experimental Runs

All experiments were conducted in duplicate. The mean values are reported in results.

3.3 Results and Discussion

We first investigated the feasibility of reducing product inhibition using the membrane bioreactor configuration (MBR) and its dependence on the dilution flow as the key variable. After establishing the feasibility, we studied the impact of other reactor and separation variables including the substrate, enzyme loadings and different separation membranes.

3.3.1 Composition of the Pulp

The composition of the unbleached softwood kraft pulp was determined based on a two- stage acid hydrolysis, and the carbohydrates were analyzed using HPLC. From the compositional analysis, the amount of individual carbohydrates per gram of pulp was determined to be in the range of 363 – 547 mg glucose, and 101 – 134 mg xylose.

3.3.2 Preliminary Experiments

A preliminary set of hydrolysis experiments were conducted with the unbleached softwood kraft pulp as the substrate at a dilution rate of 0.5 h-1 and an enzyme loading rate of 7.5

FPU/ g substrate to check if product inhibition could be minimized by the dilution. A control experiment without any dilution was conducted at the same time. The conditions for this experiment are summarized in Table 4. Samples of the hydrolysate were drawn at even intervals

(from 1 to 2 h) and the composition of reducing sugars were measured. Hydrolysis yield and productivity were calculated using equation 2 and equation 3 respectively. It can be observed that dilution resulted in an increase in the hydrolysis yields at various times. After 5h, the yield increase is 15.6% and after 11 h, this has increased significantly by 43.2% over the case where there is no dilution. Thus, the operation of the reactor with continuous membrane separation appears to boost hydrolysis yields significantly. This increase in hydrolysis yield resulted in reduction of product inhibition due to the continuous flush of the glucose from the reactor. Figure

21 shows the reactor productivity, P measured at different times, calculated using equation (3).

The productivity of the reactor in the MBR configuration is consistently higher than that with no dilution, although the increase is small. Figure 22 shows the concentration of reducing sugars in the reactor measured at different times. In the control experiment, this concentration is 7 g/ L and

11 g/L after 5 hours and 11 hours of hydrolysis, respectively. Smith et al., (2010) reported that when the glucose concentration increases from 7.5 g/L to 11 g/L, the product conversion rate was reduced by about 30 to 35 % due to product inhibition. The current results agree with their findings and further confirm that MBR can be used to achieve efficient hydrolysis by reducing product inhibition. It is important to note that, although the hydrolysis yield and reactor productivity were improved using the MBR configuration, however, the sugar concentration is

42 lower (2 g/L in the MBR vs. 11 g/L in the undiluted case). Therefore, an additional concentration step using RO may be necessary.

D= 0 D= 0.5 30

15 Y, % Y,

0 0 2 4 6 8 10 12 Time, h Figure 20. Hyrdolysis yield as a function of time. Effect of dilution rate. Squares are control experiment (no dilution), and triangles are dilution rate experiment. Other conditions as in Table 4.

D= 0 D= 0.5 70

1 -

.min 1 - 40

30 P, mg.L P,

0 0 2 4 6 8 10 12 Time, h

Figure 21. Productivity as a function of time. Effect of dilution rate. Squares are control experiment (no dilution), and triangles are dilution rate experiment. Other conditions as in Table 4

D= 0 D= 0.5 10

2 Reducing sugar g/L sugar concentration, Reducing

0 0 2 4 6 8 10 12 Time, h

Figure 22. Reducing sugar concentration as a function of time. Effect of dilution rate. Squares are control experiment (no dilution), and triangles are dilution rate experiment. Other conditions as in Table 4.

3.3.3 Effect of Dilution Rate on Hydrolysis

From the earlier set of experiments, the dilution rate has been established to increase yield and productivity with reduced sugar concentration within the reactor. Since the dilution rate is a key variable, we studied the impact of varying this parameter on the hydrolysis of the pulp fibers.

The experiments were conducted at dilutions of 0, 0.25 and 0.5 h-1. All experiments were performed for 11 hours at 5% solid concentrations, and 7.5 FPU/ g enzyme loadings. The conditions are summarized in Table 4. Figure 23, depicts the hydrolysis yield as function of time for different dilution rates. The results show that the product yield increased from 17.8 % to 25.5

% as the dilution rate increased from 0 h-1 to 0.5 h-1 after 11 hours of hydrolysis. This

44 corresponds to an increase in hydrolysis yield by 24.1 % and 43.2 % at 0.25 h-1 and 0.5 h-1, respectively, compared to batch hydrolysis.

Although 0.5 h-1 resulted in the higher yield, product removal, and buffer replenishment in the fed-batch operation consumed 10 – 15 min of the hydrolysis time every hour. This not only compromised the hydrolysis time but is also susceptible to errors in hydrolysis yield and productivity calculations. Hence, the dilution rate of 0.25 h-1 was chosen for conducting the further MBR hydrolysis experiments.

D= 0 D= 0.25 D= 0.5

30 D= 0 D= 0.25 D= 0.5 70

60 25

50 1

20 -

.min 1 15 - Y, % Y, 30

10 mg.L P, 20

5 10

0 0 0 2 4 6 8 10 12 Time, h

Figure 23. Hydrolysis yield and productivity as a function of dilution rate. Solid shapes are hydrolysis yield, hollow shapes are productivity. Squares are control experiment (no dilution), diamonds are 0.25 h-1 dilution rate experiment, and triangles are 0.5 h-1 dilution rate experiment . Other conditions as in Table 4.

) D= 0 D= 0.25 D= 0.5 1 - 10 9 8 7 6 5 4 3 2

Reducing Sugar Concentration, (g.L Concentration, Sugar Reducing 1 0 0 2 4 6 8 10 12 Time, h

Figure 24. Reducing sugar concentration as a function of dilution rate. Squares are control experiment (no dilution), diamonds are 0.25 h-1 dilution rate experiment, and triangles are 0.5 h-1 dilution rate experiment . Other conditions as in Table 4.

It can be observed from Figure 24, that a higher dilution rate resulted in lower reducing sugar output concentration, and peak concentrations occurred between 2 to 4 hours at a dilution rate of 0.25 h-1. The trend is similar to those obtained for hydrolysis of sodium-hydroxide- pretreated sallow and steam-exploded rice straw (Ohlson et al., 1984; Yang et al., 2006). Yang et al. (2006) reported that both hydrolysis yield and reducing sugar output concentration increased as the dilution rate increased from 0.057 h-1 to 0.075 h-1. However, when the dilution rate was increased from 0.075 h-1 to 0.25 h-1, although hydrolysis yield increased, reducing sugar output concentration decreased.

3.3.4 Effect of Calcium Ions on Hydrolysis

The combination effect of dilution rate and the presence of calcium ions on hydrolysis yields at an enzyme loading of 7.5 FPU/ g substrate are presented in Figure 25. As evident from

Figure 25, the presence of calcium ions decreased the hydrolysis yield by 19.3 % and 28.6 % in the batch process and fed-batch process, respectively. The rate of hydrolysis and productivity also decreased in the presence of calcium ions for both cases of control and MBR experiment.

The negative effects of calcium ion on hydrolysis yield and productivity is due to the binding of calcium ions to cellulases, which promotes the aggregation of proteins (Min, 2017; Pal et al.,

2001). This aggregation might have decreased the access of free protein to the cellulose chain.

D= 0, PCC= N D= 0.25, PCC= N D= 0, PCC= Y D= 0.25, PCC= Y

25 70

60 20

1 - 15

.min

1 -

Y, % Y, 30 10

20 mg.L P, 5 10

0 0 0 2 4 6 8 10 12 Time, h

Figure 25. Effect of calcium ions on hydrolysis yield and productivity. Solid shapes are hydrolysis yield, hollow shapes are productivity. Squares are pulp hydrolysis experiment (no dilution), diamonds are 0.25 h-1 dilution rate pulp hydrolysis experiment, triangles are pulp hydrolysis experiment in presence of calcium ions (no dilution), and circles are 0.25 h-1 dilution rate pulp hydrolysis experiment in the presence of calcium ions. Other conditions as in Table 4.

D= 0, PCC= N, S= Y D= 0, PCC= Y, S= Y D= 0, PCC= N, S= N D= 0, PCC= Y, S= N

25 70

60 20

1 - 15

.min

1 -

Y, % Y, 30

10 P, mg.L P, 20 5 10

0 0 0 2 4 6 8 10 12 Time, h

Figure 26. Effect of surfactant on hydrolysis yield and productivity. Solid shapes are hydrolysis yield, hollow shapes are productivity. Squares are pulp hydrolysis experiment with tween addition, diamonds are pulp hydrolysis experiment in the presence of calcium ions with tween addition, triangles are pulp control experiment, and circles are pulp control experiment in the presence of calcium ions. Other conditions as in Table 4.

Previous studies have shown that addition of nonionic surfactants like Tween 80 enhanced the hydrolysis performance of mixed office waste newspaper by minimizing nonproductive and irreversible binding of cellulases to lignin (Chu and Feng, 2013; Eriksson et al., 2002; Kim et al., 2007, 2006; Wu and Ju, 1998). Min (2016) reported that the addition of

Tween 80 alleviated the inhibitory effect of calcium carbonate on waste fines hydrolysis.

Therefore, hydrolysis performance of USKP in the presence of Tween 80 and calcium ions were also studied. From Figure 26, the addition of Tween 80 to USKP devoid of calcium ions improved the hydrolysis yield by 32.2 %, whereas in the presence of calcium ions, hydrolysis yield improved by just 11.9 %. This increased hydrolysis yield in the presence of Tween 80 might be due to the prevention of lignin binding sites to cellulases, thus increasing the

48 accessibility of enzymes to the substrate. However, although lignin binding sites are covered by

Tween 80, the presence of calcium ions at higher concentrations promoted aggregation, reducing the availability of enzyme to the substrate (Min et al., 2018).

D= 0, PCC= Y, S= N D= 0, PCC= Y, S= Y D= 0.25, PCC= Y, S= N D= 0.25, PCC= Y, S= Y

D= 0, PCC= Y, S= N D= 0, PCC= Y, S= Y D= 0.25, PCC= Y, S= N D= 0.25, PCC= Y, S= Y 25 60

50 20

1 -

.min 1

30 - Y, % Y, 10

20 mg.L P,

5 10

0 0 0 2 4 6 8 10 12 Time, h

Figure 27. Effect of surfactant and dilution rate on hydrolysis yield and productivity. Solid shapes are hydrolysis yield, hollow shapes are productivity. Squares are pulp hydrolysis experiment in the presence of calcium ions (no dilution), diamonds are pulp hydrolysis experiment in presence of calcium ions with tween addition(no dilution), triangles are 0.25 h-1 dilution rate pulp hydrolysis experiment in the presence of calcium ions, and circles are 0.25 h-1 dilution rate pulp hydrolysis experiment in the presence of calcium ions with tween addition. Other conditions as in Table 4.

Figure 26, shows the impact of the addition of Tween 80 on hydrolysis yield. It appears that the surfactant did not improve the hydrolysis yield in the presence of calcium ions.

Therefore, the effect of another variable, the dilution rate on hydrolysis in the presence of calcium ions was explored. The results shown in Figure 27, indicate that the addition of Tween

80 improved the hydrolysis yield by 10.1 % in the control experiment. At the dilution rate of

0.25 h-1, the addition of Tween 80 improved the hydrolysis yield by 26.5 %. The combination of decreasing product inhibition and calcium ion concentration depletion at 0.25 h-1 resulted in the higher hydrolysis yield.

3.3.5 Effect of Membrane MWCO on hydrolysis

The molecular weight of enzymes in the cellulase cocktail mixture approximately range from 46 to 120 kDa (Haven and Jørgensen, 2013; Jäger et al., 2010; Kumar and Murthy, 2017).

Most of the membrane bioreactors reported in the literature operated with polyethersulfone membrane with a MWCO range of 10 kDa – 50 kDa (Alfani et al., 1983; Bélafi-Bakó et al.,

2006; Knutsen and Davis, 2004; Lee and Kim, 1993; Smith et al., 2010). Membranes with lower

MWCO are energy intensive and susceptible to membrane fouling. Therefore, hydrolysis as a function of MWCO was explored to determine the optimal membrane. Hydrolysis profiles using different MWCO (10, 30, 50 kDa) at an enzyme loading of 5, 10 and 15 FPU/g substrate are illustrated in Figure 28, Figure 29 and Figure 30, respectively. From Figure 28, at an enzyme loading rate of 5 FPU/g substrate, the dilution rate negatively affected the hydrolysis yield.

Membrane with a lower MWCO produced better hydrolysis yields. Although it is anticipated that the dilution rate reduces the product inhibition, at lower concentrations, the filtration operation might have promoted binding of the enzyme to the membrane surface, reducing the availability of enzyme in the bulk solution. The hydrolysis yields were 15.5 % and 17.6 % for 10 kDa and control experiments, respectively after 11 hours of hydrolysis. The membrane with higher

MWCO resulted in a 30.2 % less hydrolysis yield compared to the control experiments. This might be due to the leakage of the enzyme during filtration (i.e. incomplete retention of the enzyme by the membrane).

D=0 10 KDa 30 KDa 30 0 KDa 10 KDa 30 KDa 90

80 25 70

20 60

1 -

.min 1 15 -

Y, % Y, 40 P, mg.L P, 10 30

20 5 10

0 0 0 2 4 6 8 10 12 Time, h

Figure 28. Effect of membrane MWCO on hydrolysis yield and productivity at 5 FPU/g substrate. Solid shapes are hydrolysis yield, hollow shapes are productivity. Squares are control experiment (no dilution), diamonds are dilution rate experiment with 10 kDa MWCO membrane, triangles are dilution rate experiment with 30 kDa MWCO membrane, and circles are dilution rate experiment with 50 kDa MWCO membrane. Other conditions as in Table 4.

From Figure 29, the membrane with lower MWCO favored the hydrolysis yield at 10

FPU/ g substrate. This is due to the higher enzyme rejection coefficient for the membrane with lower MWCO. The hydrolysis yield increased by 12.8 % and 24.8 % with 30 kDa and 10 kDa membranes, respectively. At 10 FPU/ g substrate enzyme loading, the amount of enzyme adsorbed to the membrane surface might be trivial compared to the bulk solution. At 15 FPU/ g substrate enzyme loading, the hydrolysis was not affected by the membrane MWCO. The hydrolysis yield improved by 19.7 % irrespective of the membrane MWCO.

30 90 D=0 10 KDa 30 KDa 80 0 KDa 10 KDa 30 KDa 25 70

20 60

1 -

.min 1 15 -

Y, % Y, 40

10 30 mg.L P,

20 5 10

0 0 0 2 4 6 8 10 12 Time, h Figure 29. Effect of membrane MWCO on hydrolysis yield and productivity at 10 FPU/g substrate. Solid shapes are hydrolysis yield, hollow shapes are productivity. Squares are control experiment (no dilution), triangles are dilution rate experiment with 10 kDa MWCO membrane, and diamonds are dilution rate experiment with 30 kDa MWCO membrane. Other conditions as in Table 4

0 KDa 10 KDa 30 KDa

35 0 KDa 10 KDa 30 KDa 120

30 100

25 1 80 -

20 .min 1 60 -

Y, % Y, 15 40 10 mg.L P,

5 20

0 0 0 2 4 6 8 10 12 Time, h Figure 30. Effect of membrane MWCO on hydrolysis yield and productivity at 15 FPU/g substrate. Solid shapes are hydrolysis yield, hollow shapes are productivity. Squares are control experiment (no dilution), diamonds are dilution rate experiment with 10 kDa MWCO membrane, and triangles are dilution rate experiment with 30 kDa MWCO membrane. Other conditions as in Table 4

3.4 Conclusions

This investigation showed that the application of an integral membrane reactor configuration is likely to give higher yields and reactor productivity in terms of conversion rate of the substrate per unit reactor volume as compared to a standard batch or continuous stirred tank configuration. The principal advantage appears to be the reduction if not elimination of product inhibition by the dilution flow of the MBR configuration. The kinetic data on yield, product concentration and reactor productivity increases are similar to those obtained for such membrane bioreactor configurations applied to enzymatic and other reaction processes.

An interesting feature of the fines and fiber hydrolysis system is the presence of surfactants, calcium carbonate mineral solid particles and high calcium ion concentrations in the

53 reaction suspension. This would necessitate a more complex separation sequence, but our experimental results show the advantages of the application of ultrafiltration membranes to separate the enzyme and surfactants to reduce product inhibition. We briefly recapitulate some of the key results found with this experimental program. At an enzyme loading rate beyond 7.5

FPU/ g substrate, hydrolysis performance increased in membrane bioreactor. The improvement in hydrolysis performance is proportional to the dilution rate. Membrane with lower MCWO provided better hydrolysis performance. At a higher enzyme loading rate of 15 FPU/ g substrate, hydrolysis performance is indistinguishable between 10 kDa and 30 kDa membranes. Tween 80 didn’t improve the hydrolysis yield in presence of calcium ions.

The experimental data can be used to construct a model for the reaction-separation sequence and optimization of the hydrolysis process can be achieved using the data established in this chapter. This is left for future work in this area. The next two chapters focus on the development of a complete enzyme recycle process that can be integrated with the membrane bioreactor in a sequential manner for achieving robust economic operation of the entire process.

CHAPTER IV: ENZYME RECYCLE USING ZEOLITE

4.1 Introduction

Production of fuel or chemicals from lignocelluloisc biomass via biological conversion consists of three critical steps: pretreatment of biomass (decrease the structure recalcitrance), hydrolysis of sugar polymers (conversion to sugar monomers) and lastly conversion of sugar monomers to fuel or chemicals. The cost of enzymes involved in hydrolysis of sugar polymers to sugar monomers is substantial (Johnson, 2016). For instance, the contribution of enzyme costs to the production cost of cellulosic ethanol varies from $ 0.39 to $ 0.78 per gallon, an equivalent of

15% to 28% of the selling price (Humbird et al., 2011; Johnson, 2016; Klein-Marcuschamer et al., 2012). Therefore, the industrial and academic researchers have engaged in identifying approaches, such as improving hydrolysis efficiency through use of additives or accessory enzymes, improve enzyme activities, design synthetic cocktails, etc., to reduce these enzymes costs. Improving the biocatalytic productivity by recycling of the enzymes in the biorefinery remains another attractive option to pursue (Jørgensen and Pinelo, 2017). A variety of enzyme recycling strategies, particularly for cellulases applied in the hydrolysis of the lignocellulosic biomass have been investigated in the past (Gomes et al., 2015; Jørgensen and Pinelo, 2017).

Some of the common enzyme recycle strategies are shown in the flow scheme given below in

Figure 31.

Figure 31. Enzyme recycle strategies. Adapted from (Jørgensen and Pinelo, 2017)

Commercial enzymes are formulations of several enzymes tailored to specific classes of

biomass and to accomplish cellulose or hemicellulose deconstruction along precise pathways.

These cellulase formulations deploy a broad spectrum of enzymes: endoglucanases (EG; EC

3.2.1.4), cellobiohydrolases (CBH; EC 3.2.1.91), and β-glucosidases (BG; EC 3.2.1.21), that

work synergistically (Horn et al., 2012; Kumar and Murthy, 2016; Rodrigues et al., 2015). After

hydrolysis, cellulases are either released into the liquid fraction, or remain adsorbed on the solid

residue (Pribowo et al., 2012; Shang et al., 2014). Soluble cellulases in the liquid fraction have

been efficiently recovered using ultrafiltration or by adsorbing the enzymes onto the fresh

substrate. Lu et al. (2002) employed ultrafiltration to recover enzyme in the liquid fraction and

reported that the enzyme remained relatively active for three rounds of recycle. Using

ultrafiltration, Qi et al. (2012) reported 73.9% of cellulose protein recovery from the liquid

56 fraction after hydrolysis using ultrafiltration. However, operational costs and membrane fouling are the limitations for this strategy (Mores et al., 2001). Chen et al. (2013) applied an electric field to depolarize the membranes and increased the permeate flux. Another method is to simply add fresh substrate to the filtrate containing the cellulases (Ouyang et al., 2013; Tu et al., 2007a).

After conducting hydrolysis of biomass, the solids are separated using centrifugation or filtration, and are suspended in a buffer solution for the next round of hydrolysis. It was found however that the β-glucosidases, needed to be added owing to their lower affinity to adsorb onto the substrates. About 88% of the total protein content in the liquid phase was recovered employing this method (Tu et al., 2007b).

Recent research on the efforts and challenges of recycling of cellulase enzymes from biomass hydrolysate solutions has been summarized in Table 5 below.

Table 5. Enzyme recovery from lignocellulosic hydrolysates. Reference Substrate Enzyme Recovery method Protein Recovery Liquid Recycle Tu et al. (2007b) Organosolv pretreated Celluclast Exposure to fresh 88% Douglas fur substrate 85% protein activity recovered Qi et al. (2012) Steam exploded wheat straw Cellulase Ultrafiltration 74%

Rodrigues et al. Hydrothermally pretreated Celluclast 1.5 FG L Exposure to fresh 80% (2014) wheat straw substrate

Shang et al. Corncob Spezyme CP Exposure to fresh 50% (2014) substrate Solid Recycle Tu et al. (2009) Ethanol pretreated Mixed Celluclast 0.5% Tween 80 90% softwood Zhu et al. (2009) Dilute acid pretreated corn Spezyme CP Alkali elution 56% stover 72% ethylene glycol 76 %

1M NaCl 6%

Shang et al. Milled Corncob Spezyme CP Alkali elution 83% (2014) 80% cellulase activity Rodrigues et al. Hydrothermally pretreated Celluclast 1.5 FG L and Alkali elution 16% of Cel7A and 13% (2015) wheat straw Cellic of Cel7B for Celluclast

58% of Cel7A and 35% of Cel7B for Cellic

The techniques cover the spectrum of solution separations including membrane filtration, ultra centrifugation and different adsorption techniques using the biomass itself as the substrates.

Cellulases bound to the solid residue can also be recovered either by shifting the pH or by the addition of displacing solutes as in chromatography. An increase in the pH of the solution to 10 can release upto 94% of the bound enzymes to the solid residues as was shown by (Du et al.,

2012; Otter et al., 1989, 1984; Shang et al., 2014; Zhu et al., 2009). Although the conformational changes in the structure of Cel7A (cellobiohydrolases I) were determined to occur at pH 10, the changes were reversed when the pH of the solution was changed back to 4.8 (Rodrigues et al.,

2012). Surfactants like Tween 80, Tween 20, Triton X-100, other solutes like polyethylene glycol (PEG) and glycerol compete with cellulases for adsorption sites in lignin-rich residues and were reported to enhance the desorption of cellulases (Börjesson et al., 2007; Eriksson et al.,

2002; Pribowo et al., 2012; Bálint Sipos et al., 2010; Tu et al., 2009, 2007a; Zhu et al., 2009).

Deshpande and Eriksson (1984) investigated the use of phosphate buffer and urea to desorb cellulases bound to the lignocellulosic biomass.

The present study was conducted to assess the separation of cellulase enzymes from the residual suspension of waste paper fines after their hydrolysis. The separation of unadsorbed enzyme from the residual solids of waste paper hydrolysis can allow the enzyme to be recycled in a controlled manner.

Since most of the earlier research has concentrated on membrane filtration which is always hampered by flux declines and lost productivity by membrane fouling and concentration polarization effects, the present objective was to identify a suitable adsorbent which could bind the cellulases from solution and also a displacer or elutant solute that could displace previously

59 bound cellulases to make them available for reuse. The residual activity of the enzyme is also of a concern and will be investigated.

Cellulases have been adsorbed on ferric and manganese nanoparticles for immobilization

(Cherian et al., 2015; Ingle et al., 2017). A reduction in enzyme activity has been found.

Redissolution i.e. elution of the enzyme was not determined. Cellulases are known to adsorb strongly onto both organic and metal surfaces. Under acidic conditions (pH 5), cellulases are negatively charged but their hydrophobic and van der Waals interactions dominate in most adsorption phenomena. Thus, strong adsorption onto negatively charged surfaces may also be found. Adsorbents which can attach hydrophobic molecules, in particular the recently investigated class of zeolites (known as zeolite-β) offer an interesting possibility for cellulase adsorption and thereby recycling.

Several studies separating amino acids (L-arginine, L-phenylalanine, lysine, glutamic acid) from aqueous solutions using zeolite β have been reported (Chiku et al., 2003; Krohn and

Tsapatsis, 2005; Munsch et al., 2001). Proteins (bovine serum albumin, cytochrome c, hemoglobin, γ-globulin) adsorbed at their isoelectric points to zeolite were displaced by PEG

(Chiku et al., 2003), where conventional eluents like 2.5 M NaCl, 2 % Nonidet P-40, 1 % Triton

X-100, 0.5 % Tween 20, 0.2% Sodium dodecyl sulfate were much less effective. It was also reported that proteins retained their activities during this PEG facilitated recovery process.

Zeolite β was also employed for selective adsorption of microbes, removal of emerging contaminants like tris-2-chloroethyl phosphate from aqueous solutions, potentiometric biosensors development by coimmobilization of enzymes with zeolite on pH-ion-sensitive field- effect transistor (Grieco and Ramarao, 2013; Kubota et al., 2008; Soldatkin et al., 2015).

This study explores the scope of zeolite β to recycle cellulases from the liquid fraction of the hydrolysate. The objectives of this research include establish cellulase adsorption isotherms at working temperature and pH conditions of typical hydrolysate solutions. The purpose is to provide an understanding of the prominent factors affecting adsorption including electrostatic or other interactions (hydrophobic, steric and van der Waals). Other objectives include

• Recover adsorbed enzyme by different methods including (a) introducing a pH shock,

(b) the use electrolytes (NaCl) as an elutant, (c) The use of Poly-Ethylene Glycol (PEG)

as an elutant

• Evaluate the activity of the recovered enzyme

4.2 Materials and Methods

4.2.1 Materials

Zeolite β donated by Zeolyst International, Kansas City, KS, was used as an adsorbent.

2 The ratio of SiO2/Al2O3 and surface area of Zeolite β was 25 and 680 m /g, respectively. CTec2 kindly provided by Novozymes, USA, was used in all enzyme hydrolysis experiments. Bovine

Serum Albumin (BSA), Coomassie Brilliant Blue G-250, Avicel PH-101, Sodium Acetate, and

2,5-Dinitro salicylic acid were purchased from Sigma-Aldrich, St. Louis, MO. Ethanol 190 proof and hydrochloric acid (HCl) were purchased from Pharmco, Shelbyville, KY. Phosphoric acid was purchased from Macron Fine Chemicals. Polyethylene glycol (PEG) having molecular weighs 200, and 20000 Da were purchased from Alfa Aesar, Ward Hill, MA. Sodium hydroxide was purchased from J. T. Baker, Phillipsburg, NJ. Sodium potassium tartrate and acetic acid were purchased from Fischer Scientific, Fair Lawn, NJ. Tris base was purchased from Chem-

Impex International, Wood Dale, IL. Unbleached softwood kraft pulp (USKP) sample was

61 milled in Wiley Mill and screened through a 35-mesh sieve. Milled sample was stored in air-tight plastic bottles.

4.2.2 Protein Quantification

Bradford assay was used to quantify the protein content of the enzyme solution. Bradford reagent was prepared by dissolving 100 mg of Coomassie Brilliant Blue G-250 in 50 mL of 95% ethanol. The solution was then mixed with 100 mL of 85% phosphoric acid and volume adjusted to 1 L with distilled water (Kruger, 1994).

A BSA stock solution of 1 mg/mL was prepared by dissolving 10 mg of BSA protein in

10 mL of distilled water. For calibration curve, 200, 400, 600 and 800 µL of BSA stock solution was pipetted and mixed with 800, 600, 400 and 200 µL of distilled water to make 0.2, 0.4, 0.6 and 0.8 mg/mL of BSA protein solutions. The samples were mixed in a centrifuge by gentle vortexing. 60 µL of these sample solutions were added to 3 mL of Bradford reagent in a cuvette and the contents were mixed by inversion. After 20 min, absorbance of the mixture was measured using GENESYS 50 UV-Vis spectrophotometer at 595 nm. A calibration curve is also constructed for Cellic Ctec2 to quantify protein in adsorption and desorption experiments .

4.2.3 Adsorption of CTec2 by Zeolite

All the adsorption experiments were conducted in 125 mL Erlenmeyer flasks. For adsorption isotherm experiments, 50 mL of CTec2 solutions (0.7, 1.43, 2.15, 2.61, 3.39 and 8.12 mg/mL) were prepared by adding appropriate amount of original CTec2 solution to 0.05 M sodium acetate buffer (pH 5). In all experiments, 500 mg of zeolite was added to the enzyme solution and the contents were mixed for 15 minutes using Precision shaker bath adjusted to 25

°C and 120 rpm. At the end of mixing, the mixture was centrifuged at 4000 rpm for 5 min to

62 separate liquid and solid fractions. The protein content in supernatant (liquid) was determined using Bradford assay, as described in the previous section. The amount of protein adsorbed on to zeolite is calculated by subtracting the amount of protein in supernatant from total protein used in the experiment. The supernatant was discarded, and the zeolite-enzyme complex precipitated to the bottom of the centrifuge tube was used in desorption experiments. For all other adsorption experiments, 500 mg of zeolite was suspended in 50 mL of enzyme solution.

4.2.4 Desorption of CTec2 from Zeolite

To separate enzyme from the zeolite, desorption experiments were carried out by suspending the Zeolite-enzyme complex in 50 mL of desorption solution. The desorption solutions tested in this work include: PEG 200 (0.05 M, 0.25 M, 0.50 M), PEG 20000 (0.05 mM,

0.25 mM, 0.50 mM), 50 mM tris HCl (pH- 7, 8, 9), NaCl solution (0.01, 0.1, 1) M (Rodrigues et al., 2012). For PEG and NaCl solutions, appropriate amount of PEG and NaCl were added to

0.05 M sodium acetate buffer adjusted to pH 5. For tris HCl buffer, pH was adjusted using HCl.

The suspension was carried by vigorous mixing of contents in a 50 mL centrifuge tube. Upon vortex, the contents were left undisturbed for 30 minutes. The mixture was centrifuged and the protein content in the supernatant was quantified using Bradford assay. Enzyme recycle yield defined as the amount of protein recovered from the enzyme solution was reported.

4.2.5 Reducing Sugar Determination

Concentrations of reducing sugars in the hydrolysate were determined using DNS assay.

Test solution and DNS reagent were added to the test tubes in the ration of 1:4. The test tubes were placed in boiling water bath for exactly 5 min. The tubes were then transferred to ice bath to cool down the contents rapidly (Saqib and Whitney, 2011). 0.2 mL of the reacted solution was

63 mixed with 2.5 mL of distilled water and absorbance was measured at 540 nm using GENESYS

50 UV-Vis spectrophotometer.

DNS reagent was prepared as follows: 10 g of 2,5-dinitro salicylic acid and 300 g of sodium potassium tartrate was dissolved in 800 mL of 0.5 N NaOH by gentle heating. The solution volume was then adjusted to 1000 mL using distilled water (Coughlan and Moloney,

1988; Saqib and Whitney, 2011).

4.2.6 Enzyme Activity Assay

Cellulase activity (FPU) was measured according to the National Renewable Energy

Laboratory (NREL) method (Adney and Baker, 2008). 50 mg Whatman No. 1 filter paper strip used as a substrate was saturated with 1 mL 0.05 M sodium acetate buffer, then equilibrated to

50 °C. A set of dilute enzyme solutions were made, with one dilution targeted to release slightly more than 2 mg of glucose and one slightly less than 2 mg of glucose. Simultaneously, glucose standards and enzyme blanks were prepared. 0.5 mL of these diluted enzyme solution were added to saturated solution and were incubated at 50 °C for 60 minutes along with glucose standards and enzyme blanks. The reaction was terminated by adding 3 mL of DNS reagent, boiled for 5 minutes in vigorously boiling water to develop color formation followed by ice cooling. Absorbance was measured at 540nm. Using the standard curve, amount of glucose released from each enzyme assay tube were determined. Filter paper activity was calculated using equation

0.37 퐹푃푈 ( 6 ) 퐹푃푈 = , 퐸푛푧푦푚푒 푐표푛푐푒푛푡푟푎푡푖표푛 푟푒푙푒푎푠푖푛푔 2 푚푔 표푓 푔푙푢푐표푠푒 푚퐿

4.2.7 Statistical Analysis

All experiments were conducted in triplicate, and the error bars represent 95 % confidence interval. In some cases, the standard deviation was very small and the error was insignificant.

4.2.8 Zeolite and Enzyme Surface Charge Measurements

Zeta potential of zeolite and enzyme were measured at pH 4, 5, and 6 using surface charge analyzer (Zetasizer Ultra, Malvern Panalytical) at Syracuse Biomaterials Institute,

Syracuse University.

4.3 Results and Discussion

4.3.1 Protein Content

The protein content and enzyme activity of the CTec2 used in this work was measured to be 60 ± 5 mg/mL and 94 ± 4 FPU/ mL. The measured protein concentration is in agreement with the values reported by other researchers (Cai et al., 2019).

4.3.2 Adsorption Isotherms

Enzyme solutions with protein concentrations ranging from 0.7 to 8.12 g/L were tested in adsorption experiments to construct adsorption isotherm. The amount of protein adsorbed onto zeolite had attained equilibrium in 15 minutes. At 15 min, adsorption isotherm illustrated in

Figure 32 was constructed by plotting equilibrium adsorption (qe) against protein concentration

(c) at pH 4, 5 and 6. The adsorption isotherm data were fitted to the Langmuir isotherm model, equation 2 to calculate maximum adsorption (qm) and equilibrium constant (KL) (Ayawei et al.,

2017; Grieco and Ramarao, 2013).

퐾퐿푐 ( 7 ) 푞푒 = 푞푚 ( ) 1 + 퐾퐿푐

Equation 7 can be linearized into the following form for the purpose of fitting to experimental data.

퐶 1 1 ( 8 ) = 푐 + 푞푒 푞푚 퐾퐿푞푚

-1 From Table 6, the adsorption affinity, KL of the enzyme increased from 0.19 L.g to 0.28

-1 -3 L.g as pH is decreased from 6 to 4. The maximum binding, qm values are 210*10 g/g and

181*10-3 g/g at pH 4 and pH 6, respectively. To understand the adsorption isotherm behavior, the interaction between enzyme and zeolite were studied by measuring surface charge at different pH.

Table 6. Langmuir isotherm constants at different pH pH Maximum binding Equilibrium constant, R2

-3 capacity , qm x 10 g/g K L/g

4 210 0.28 0.9795

5 173 0.22 0.8436

6 181 0.19 0.9157

200

180

160

140

120

, g/g , 3

- 100 x 10 x

e 80 q pH 4 pH 5 pH 6 60

40 pH 4 pH 5 pH 6

0 0 2 4 6 8 10 12 14 16 18 20 c, g/L

Figure 32. Adsorption Isotherm as a function of pH. Squares are experimental data at pH 4, triangles are experimental data at Ph 5, and diamonds are experimental data at pH 6. Smooth curves are best fits of Langmuir Isotherm according to Eq. (2).

4.3.3 Enzyme-Zeolite Interaction

Enzyme-Zeolite interaction at pH 4, 5, and 6 was studied to understand the physicochemical properties of adsorption. Although a comprehensive model of adsorption equilibrium based on bi-component Langmuir equation and random sequential adsorption model is ideal, the Langmuir curves given here are meant as a reference for visual guidelines.

It can be observed from Table 7, at higher pH, the charge on the zeolite increased due to the dissociation of ammonium ions from the surface, creating a higher net negative charge. Also, enzymes tend to acquire more anionic charges due to higher dissociation of the acids at higher pH. Although both zeolite and enzyme are anionic at pH 6, strong adsorption occurred.

Therefore, this adsorption is not driven by electrostatic interactions. It is likely that enzyme adsorption is driven mostly by hydrophobic interactions, and also by some van der Waal’s

67 effects. This is also supported by the fact that increasing salt concentration did not desorb the enzyme significantly as compared to PEG.

Increased electrical double layer repulsions might have accounted for a reduction in adsorption as pH is increased from 4 to 6 (Chiku et al., 2003). The amount of protein adsorbed onto zeolite decreased from 46 ± 0.4 to 33 ± 3.3 mg per gram of adsorbent.

Table 7. Effect of pH on adsorption of enzyme onto zeolite.

pH Zeta potential, mV Equlibrium adsorption, qe

Enzyme Zeolite x 10-3 g/g

4 -2.56 -25.1 46 ± 0.4

5 -5.26 -27.7 37 ± 1.1

6 -7.56 -29.3 33 ± 3.3

Various studies have reported that the addition of PEG reduced the unproductive binding of cellulase to the substrate (Börjesson et al., 2007; Qing et al., 2010; Bálint Sipos et al., 2010).

\Chiku et al. (2003) recovered BSA protein adsorbed onto zeolite with SiO2/Al2O3 ratios ranging from 5.7 to 27.4 using PEG. They also reported that lower concentration of higher molecular weight PEG resulted in higher desorption compared to a higher concentration of lower molecular weight PEG. Therefore, PEG 200 with concentrations 0.05 M, 0.25 M and 0.5 M, and PEG

20000 with concentrations 0.05 mM, 0.25 mM, and 0.5 mM were used to construct adsorption isotherms as illustrated in Figure 33 and Figure 34.

180

0.5 M 0.25 M 0.05 M No PEG 160

0.5 M 0.25 M 0.05 M No PEG 140

120

100

, g/g ,

3 -

x 10 x e

q 60

0 0 1 2 3 4 5 6 7 8 9 10 c, g/L Figure 33. Adsorption Isotherm at pH 5 and PEG 200. Squares are experimental data at 0.5 M PEG, triangles are experimental data at 0.25 M PEG, diamonds are experimental data at 0.05 M PEG, and circles are experimental data at no PEG. Smooth curves are best fits of Langmuir Isotherm according to Eq. (2).

Table 8. Langmuir isotherm constants at pH 5 and PEG 200 PEG 200 Maximum binding Equilibrium R2

-3 Concentration, M capacity, qm x 10 g/g constant, K L/g

0 173 0.22 0.8436

0.05 157 0.21 0.9988

0.25 97 0.34 0.9972

0.50 145 0.18 0.9625

200 0.5 mM 0.25 mM 0.05 mM No PEG

0.5 mM 0.25 mM 0.05 mM No PEG

150

100

, g/g ,

- x 10 x

e 50 q

0 0 1 2 3 4 5 6 7 8 9 10

-50 c, g/L

Figure 34. Adsorption Isotherm at pH 5 and PEG 20000. Squares are experimental data at 0.5 M PEG, triangles are experimental data at 0.25 M PEG, diamonds are experimental data at 0.05 M PEG, and circles are experimental data at no PEG. Smooth curves are best fits of Langmuir Isotherm according to Eq. (2).

Table 9. Langmuir isotherm constants at pH 5 and PEG 20000 PEG 20000 Maximum binding Equilibrium R2

-3 Concentration, capacity, qm x 10 g/g constant, K L/g

0 173 0.22 0.8436

0.05 141 0.16 0.9541

0.25 109 0.06 0.9983

0.50 18 0.66 0.9979

From Figure 33, Figure 34, Table 8, and Table 9, it is clear that PEG has strong affinity to the zeolites as was demonstrated by its ability to dislodge the enzymes. PEG adsorbs competitively with the enzymes and is able to displace them, due to higher binding affinities.

Higher molecular weight PEG is much more effective at desorption than the lower molecular weight polymer. This is likely a consequence of smaller polymer chains being more soluble in water and are therefore not as easily adsorbed, reducing the adsorption affinities.

The amount of protein adsorbed on to zeolite decreased as the PEG concentration and molecular weight increased. In the presence of 0.25 M PEG 200 and 0.50 mM PEG 20000, the maximum binding capacity of zeolite reduced by 44% and 89.6%, respectively. Given the large size of the enzymes compared to the pore sizes in the zeolite, the enzyme and the PEO adsorption is expected on the surface and not on the internal pores.

4.3.4 Adsorption Kinetics

The amount of protein adsorbed onto zeolite was measured every 5 minutes up to 30 minutes, at 45 min, and at 60 min to determine the adsorption kinetics at 15 °C and 25 °C. It can be observed from Figure 35 that increasing temperature led to lower adsorption. This again indicates that hydrophobic interactions are the likely cause for enzyme zeolite interaction. At 15

°C, the mean value of protein adsorbed was between 36 to 42 mg per gram of zeolite, which corresponds to 38 to 44 percent of the total protein. At 25 °C, the mean value of protein adsorbed was between 35 to 38 mg per gram of zeolite, which corresponds to 36 to 39 percent of the total protein. The amount of protein adsorbed on to zeolite as a function of time is presented in Figure

35. Enzyme adsorbs to the zeolite rapidly, and the amount of enzyme adsorbed onto zeolite was insignificant between 5 min and 60 min. Interestingly, there is a small peak in adsorption at very

71 short times (~ 5 min) which may correspond to a rapid but weak cellulase adsorption on the surface followed by a slower relaxation to a stronger bound form, reflecting the so-called

Vroman effect (Rabe et al., 2011).

50 15 °C 25 °C

, g/g , 30

3 -

q x 10 x q 20

0 0 10 20 30 40 50 60 70 Time, min

Figure 35. Adsorption Kinetics of CTec2 on Zeolite. Squares are experimental data at 15 °C, and diamonds are experimental data at 25 °C

4.3.5 Enzyme Desorption by PEG

Enzyme desorption studies showed that even at a concentration 0.5 mM, PEG 20000 showed better desorption than PEG 200 at 0.5 M. The amount of protein desorbed from zeolite in the presence of 0.5 mM PEG 20000 and 0.5 M PEG 200 are 72% and 50 %, respectively . In other words, PEG 20000 at 0.5 mM desorbed 44 % more protein than PEG 200 at 0.5 M.

Enzyme recycle yields were 17.8 % and 33.2 % for 0.5 M PEG 200 and 0.5 mM PEG 20000 respectively. Figure 36 and Figure 37 show the desorption profile and enzyme recycle yield of

PEG 200 and PEG 20000. To further enhance desorption, hydrophobic interactions can be destabilized by the addition of salt to PEG (Chiku et al., 2003).

PEG is a benign polymer and does not cause any interactions with biological systems besides osmotic effects. PEG does not interact with either the pH or electrolytes since it is a neutral polymer (uncharged). Therefore, it is more attractive as an eluent as compared to various cationic surfactants like CTAB (cetyl trimethyl ammonium bromide), polyelectrolytes like pDADMAC (poly diallyl dimethyl ammonium chloride) and anionic surfactants like SDS

(sodium dodecyl (lauryl) sulfate). The experimental evidence presented here shows the efficacy of PEG and also establishes it as a preferred eluent for this application.

% Enzyme % desorbed from Adsorbent 10

0 0.00 0.10 0.20 0.30 0.40 0.50 0.60 CPEG

M, PEG 200 mM, PEG 20000

Figure 36. Enzyme desorption using PEG. Squares are experimental data with PEG 200 in moles, and diamonds are experimental data with PEG 20000 in milli moles.

% Enzyme % Recycle 15

0 0.00 0.10 0.20 0.30 0.40 0.50 0.60 CPEG M, PEG 200 mM, PEG 20000

Figure 37. Enzyme recycle yield using PEG. Squares are experimental data with PEG 200 in moles, and diamonds are experimental data with PEG 20000 in milli moles.

The performances of the recovered enzymes were evaluated by conducting enzymatic hydrolysis of Avicel and USKP. In Table 10, sample A represents Avicel in 0.5 g. L-1 enzyme solution; sample AP represents Avicel in 0.5 g. L-1 enzyme solution and 0.25 mM PEG 20000; sample U represents USKP in 0.5 g. L-1 enzyme solution; sample UP represents USKP in 0.5 g.

L-1 enzyme solution and 0.25 mM PEG 20000; sample AS represents Avicel in unrecovered enzyme solution; sample AR represents Avicel in recovered enzyme solution; sample US represents USKP in unrecovered enzyme solution; sample UR represents USKP in recovered enzyme solution.. Original enzyme solution represents 0.5 g. L-1 of enzyme in the hydrolysis reactor. The recovered enzyme refers to the enzyme desorbed from the zeolite using 0.25 mM

PEG 20000. Unrecovered enzyme represents the enzyme not adsorbed onto zeolite in the first step of recovery. Enzyme hydrolysis was carried out with 5 % (w/v) of the substrate at 50 °C for

24 hours.

Table 10. Hydrolysis performance of recovered and unrecovered enzyme Avicel USKP

Original Enzyme solution A U

Original Enzyme solution AP UP

with 0.25 mM PEG 20000

Recovered Enzyme AR UR

Unrecovered Enzyme AS US

30 of Substrate) of

10 Yield, Yield, of (g % Reducing Sugar Produced/g 0 A AP U UP US UR AS AR

Figure 38. Hydrolysis performance of the recycled enzyme. A-Avicel; AP-Avicel and PEG; U- USKP; UP-USKP and PEG; US-USKP and unrecovered enzyme; UR-USKP and recovered enzyme; AS-Avicel and unrecovered enzyme; AR-Avicel and recovered enzyme

From Figure 38, hydrolysis yields of the recovered enzyme were 35.5 % and 35.8 % for

Avicel and USKP, respectively. The hydrolysis yields of the original enzyme were 53.3 % and

56.3 % for Avicel and USKP, respectively. In other words, recovered enzyme yields were 65.6

% and 63.5 % of the original enzyme after 24 hours of hydrolysis. This illustrates that the recovered enzymes retained activity during the recovery process.

4.3.6 Enzyme Desorption by Alkali Elution

At pH 9, alkali elution resulted in complete desorption of the enzyme from the zeolite.

This result is in accordance with the literature - more that 90 % of the cellulase were desorbed at pH 9 (Zhu et al., 2009). Shang et al. (2014) reported 83 % desorption at pH 10. In this study, 83

% desorption happened at pH, 8. The enzyme desorption decreased with further decrease in pH.

At pH 7, 59 % of the enzyme is desorbed. The amount of protein desorbed, and enzyme recycle yields by alkali elution are presented in Figure 39 and Figure 40. The enzyme recycle yields varied from 25.2 % to 46.1 % as pH increased from 7 to 9.

140

120

100

20 % Enzyme % desorbed from Adsorbent 0 6.5 7.0 7.5 8.0 8.5 9.0 9.5 pH

Figure 39. Enzyme desorption as a function of pH. Square is experimental data at pH 7, diamond is experimental data at pH 8, and triangle is experimental data at pH 9.

60.0

50.0

40.0

30.0

20.0 % Enzyme % Recycle

10.0

0.0 6.5 7.0 7.5 8.0 8.5 9.0 9.5 pH

Figure 40. Enzyme recycle yield as a function of pH. Square is experimental data at pH 7, diamond is experimental data at pH 8, and triangle is experimental data at pH 9.

4.3.7 Enzyme Desorption using NaCl

Enzyme desorption as a function of ionic strength was studied to measure the effective of

NaCl in enzyme desorption. Increasing the concentration of salt from 0.01 M to 1 M lead to increased protein desorption. Displacement of protein occurs due to the increased hydrophilicity of the protein with the addition of salt. Compared to PEG and alkali elution, ionic strength alone did not have much effect on enzyme desorption. The enzyme desorption and enzyme recycle yields were 24 % and 9.3 % in the presence of 1 M NaCl.

25 24

15 13 11 10

5 % Enzyme % desorbed from Adsorbent 0 0.0 0.2 0.4 0.6 0.8 1.0 1.2 CNaCl, M

Figure 41. Enzyme desorption using NaCl.

12.0

10.0 9.3

8.0

6.0 5.1

4.0 4.3 % Enzyme % Recycle

2.0

0.0 0.0 0.2 0.4 0.6 0.8 1.0 1.2 CNaCl, M

Figure 42. Enzyme recycle yield using NaCl.

4.4 Conclusions

The study demonstrated that enzyme in the liquid phase of the hydrolysate can be efficiently recovered by zeolites. Zeolite had a maximum binding capacity of 173*10-3 g/g. Even at very low concentrations, 0.5 mM, PEG 20000 acts as an effective desorbent to recover enzymes from zeolite. This process is an alternative to harsh chemical methods such as alkali elution and salting. Hydrolysis performance demonstrated that enzyme retained its activity during the recovery process. In the next chapter, we present a comprehensive mathematical analysis of the adsorption and desorption processes of the enzymes on zeolite β using PEG as the eluent. This will enable the rapid evaluation and design of batch and continuous adsorption desorption systems for controlled and careful recycling of the enzymes.

CHAPTER V: ENZYME RECYCLE MATHEMATICAL MODEL

5.1 Introduction

The earlier chapter (4) gave experimental data on the adsorption and desorption of cellulase enzymes on zeolite β. A preliminary analysis shows that the conventional Langmuir equation can be used to represent the adsorption isotherm data at different pH conditions. Further, the addition of the elutant, polyethylene glycol (PEG) results in competitive adsorption and displacement of the cellulase enzymes.

The experimental data and evidence presented in chapter 4 can be used to develop, design and optimize an enzyme recycling process. For this purpose, equilibrium and kinetic models of the adsorption and desorption processes are necessary. It is the purpose of the present chapter to provide analyses and mathematical models of both the equilibrium and kinetic behaviors in batch adsorption which can be easily extended to continuous adsorption in stirred reactors and in column operations. The availability of reliable models for kinetics and equilibrium can enable rapid quantitative analysis and predictions of advanced separation schemes such as simulated moving beds (SMB) and extended bed adsorbers (EBA).

In this chapter, we will give a short review of adsorption models for proteins and macromolecules on surfaces such as the zeolites used in our present work. From the literature survey, it is clear that competitive adsorption models using the Langmuir isotherm equation as a basis are quite successful at representing experimental data over a wide range of conditions.

Kinetic models based on this form are also accurate and provide the ability to represent experimental data. The theoretical basis of the Langmuir equation is that the availability of adsorption sites of equal energies that are distributed uniformly on the adsorbent’s surface. The

80 adsorption is assumed to be reversible with an equilibrium between the forward and backward reactions attained under steady conditions. Such assumptions are not necessarily valid for the adsorption of proteins where the backward desorption tends to be substantially weak if at all.

Other models are necessary in such cases (Liu, 2015a).

5.2 Background

Cellulases are mixtures of proteins of a similar class and therefore can often be considered as a single pseudo-species for the purpose of thermodynamic and kinetic representations. Several cellulase reactivity models are available based on this pseudo- component concept. Therefore, for the purpose of further discussions, we will consider the cellulase complex to be represented by a single species in solution. The macromolecular nature of proteins and also the fact that their conformation impacts their charge state are important factors for adsorption. Thus, cellulase adsorption and desorption can be analyzed by considering the generalized adsorption theories for proteins and other macromolecules.

5.2.1 General Principles

The cellulase mixture is considered as being composed of a single species with a MW of

69 kDa. The molecule is assumed to be spherical in aqueous solution with a diameter of approximately 50 nm. When adsorbed onto the zeolite surface, cellulase may or may not transition into a flatter conformation. For the first part of the analysis, we neglect any secondary transitions and other equilibrium states and assume the cellulase is adsorbed as a rigid sphere.

Since protein binding to surfaces is almost irreversible, especially to those of high energy like the zeolite surface, the desorption reaction is assumed to be small.

5.2.2 Langmuir Equation Models

Among the simplest models for protein adsorption and desorption are models based on the Langmuir equations. The Langmuir model is based on the assumptions that the surface binding sites are uniformly distributed and have identical binding energies to the solute. It is also assumed that the coverage of the surface is by a single solute layer, i.e. a monolayer is formed

(Kadam et al., 2004; Liu, 2015a; Rabe et al., 2011). Another assumption for Langmuir adsorption is that it is reversible and that the desorption is generally to first order kinetics

(Kadam et al., 2004; Liu, 2014). The adsorption process is denoted by the following reaction between the solute, denoted E (for the cellulase enzyme) and Z for the adsorption site on the zeolite surface. EZ* represents a bound state of the enzyme at a zeolite site.

E + Z ⇌ EZ∗ ( 9 )

The above is translated into a kinetic equation for the rate of adsorption of the enzyme as

dq ( 10 ) = k (q − q). c − k q dt 1 mE 2

When adsorption and desorption are at equilibrium the rate vanishes and the following equation known as the Langmuir isotherm equation is obtained.

∗ KLEc ( 11 ) q = qm 1 + KLEc

The maximum loading of the solute on the adsorbent is qm and the fraction of this θ, is defined as the ‘coverage’.

q θ = ( 12 ) qm

The coverage is used as the dependent variable in theories based on random sequential adsorption (RSA) concepts.

Some cautions are necessary to identify here. Although the Langmuir equations (both the kinetic and equilibrium forms) have been widely used to model protein adsorption, they must be understood as empirical fits or descriptions of experimental data and physical significance cannot be inferred (Liu, 2015b). Protein adsorption is mostly irreversible and secondary stages of conformational changes of the adsorbed proteins are common. The value of qm is interpreted as the maximum binding capacity and KLE, the slope of the adsorption isotherm at infinite solution dilution, is called the binding affinity of the solute molecule. In the case of proteins, this is understood as a pseudo-affinity constant.

Since the cellulase recycling process should also incorporate a technique of removing the cellulase from the adsorbent’s surface, the reverse reaction i.e. elution step is also important.

This means a suitable elutant molecule is necessary for desorption and from prior literature, polyethylene glycol has been identified for this purpose. If one imagines that desorption occurs in a solution of both cellulase and PEG, the situation becomes akin to the adsorption equilibrium of two solute components. In this case, the process may be described by the following reactions,

E + Z ⇌ EZ∗ ( 13 )

S + EZ∗ ⇌ SZ∗ + E ( 14 )

S + Z ⇌ SZ∗ ( 15 )

The second reaction denotes the displacement of adsorbed enzyme, similar to reactions used for ion exchange processes. The third reaction allows for uncovered sites of the substrate to be adsorbed by the elutant PEG molecule, denoted as S. During simple elution process, this is not expected to be a significant reaction and therefore is neglected in further modeling, although allowance is made for it when considering simultaneous adsorption of both the cellulase and

PEG from solutions.

The kinetic equations for the above reactions are given below, as

dq ( 16 ) = k ([q − (q + q )])c − k q − k qc + k q c dt 1 mE S 2 3 s 4 S

dq ( 17 ) S = −k ([q − (q + q )])c + k q − k qc + k q c dt 5 mS S S 6 s 3 s 4 S

When k5 and k6 are sufficiently small, the above equation simplifies to

dq ( 18 ) S = −k qc + k q c dt 3 s 4 S

Under equilibrium conditions, the time derivatives vanish and the equation for a bi- component Langmuir isotherm is obtained shown below.

∗ KLEc ( 19 ) q = qmE 1 + KLEc + KLEKLScs

∗ KLScS ( 20 ) q = qmS 1 + KLEc + KLSKLEcs

5.2.3 Random Sequential Adsorption (RSA) Models

Recognizing that the Langmuir models shown above are mechanistically flawed when applied to protein adsorption, a second class of models have been applied. The so-called random sequential adsorption concept conceives protein molecules as adsorbing sequentially and irreversibly onto the adsorbent surface resulting in a gradual occlusion of the surface. This is obviously governed by the geometry of the surfaces and filling by the protein. As molecules adsorb onto the surface, it becomes progressively more difficult for new molecules to adsorb, and the coverage reaches a limit known as the ‘jamming’ limit. For spherical particles depositing onto a flat surface, the jamming limit under RSA has been determined by simulations to be 0.547

(Rabe et al., 2011). In practice secondary diffusion of the solute molecules along the surface may cause this jamming limit to be much higher and sometimes approach complete coverage. The

RSA model for protein and spherical particle deposition has been well investigated. Recent reports showing its success for representing proteins have been given by (Brash and Horbett,

1995; Rabe et al., 2011; Talbot et al., 2000). A compact summary equation for RSA kinetics is given by (Johnson and Elimelech, 1995). This is applied to RSA of the cellulase molecules, considered as a single component. The kinetic equation for RSA is

dq ( 21 ) = q αΚ(πa2)nB(θ) dt mE p

85 where B(θ) is known as the blocking function and describes the amount of available surface area on the adsorbent that is free for the new solute molecule to occupy. The surface projection of the solute molecule of radius ap is assumed to be circular and Κ is a known constant equal to

0.2348 (Kurrat et al., 1997). Furthermore, the parameter α is known as the capture coefficient, translating into the fractional of successful collisions for deposition. The variables θ and n are the fractional coverage (q/qmE) and the number concentration of the enzyme in solution respectively.

5.2.4 Kinetics of Batch Adsorption

A kinetic study of adsorption based on the simple competitive adsorption model i.e. the bi-component Langmuir model can be written as follows. The main equations describing the kinetics are reproduced below for convenience, with the understanding that the rate of the third reaction is negligible and therefore its contributions are ignored.

dq ( 22 ) = k ([q − (q + q )])c − k q − k qc + k q c dt 1 mE S 2 3 s 4 S

dq ( 23 ) S = −k qc + k q c dt 3 s 4 S

For the case of batch adsorption in a stirred vessel, clean adsorbent is first loaded into a fixed volume of the solution containing both components. It is assumed that the vessel is stirred rapidly such that any mass transfer limitations due to diffusion of the solute molecules to the surface are negligible. In the case of significant resistances, the model needs to be adjusted appropriately through the use of mass transfer coefficients with suitable correlations. Mass

86 balance on the cellulase and PEG yield two more equations which can be used to solve the above equations as two non-linear ordinary differential equations.

c = c0 − βq ( 24 )

cs = cs0 − βqs ( 25 )

We consider three different cases which may be encountered during adsorption desorption operations.

5.2.4.1 Adsorption of cellulase as a single component

It is necessary to first adsorb the cellulase onto the zeolite as a first step during the separation. The kinetics of this process can be modeled using the first equation along with the initial condition that

c(t = 0) = c0; q(t = 0) = 0 ( 26 )

5.2.4.2 Adsorption of both cellulase and PEG together from a solution

Fresh zeolite is brought into contact with a solution of both cellulase and PEG in this case. This would be used for baseline kinetics determinations and other experimental analyses but would not be used in process operations directly.

c(t = 0) = c0; q(t = 0) = 0; ( 27 )

And

cs(t = 0) = cs0; qs(t = 0) = 0; ( 28 )

5.2.4.3 Desorption or elution of loaded adsorbents

In this case, zeolite preloaded with cellulase is brought in contact with a fresh solution of the elutant, PEG. The initial condition is

c(t = 0) = 0; q(t = 0) = q0; ( 29 )

cs(t = 0) = cs0; qs(t = 0) = 0; ( 30 )

5.3 Results

5.3.1 Adsorption Kinetics of Cellulase on Zeolite - Modeling of Experimental Data

Adsorption kinetics were first modeled by using the Langmuir kinetic model. This model consisted of the differential equation, (10), the mass balance for the enzyme, equation (24) and the initial condition given by equation (26). All the input variables are shown in Table 11.

Table 11. Adsorption Kinetics for cellulase and PEG system on zeolite β Symbol Value Units

Mass of zeolite m 10 g

Volume of solvent (~solution) V 1 L

Initial concentration of cellulase c0 1 g/L

Initial concentration of PEG cs0 0 g/L

PEG MW 20 kDa

Maximum loading of cellulase and PEG qmE 140 mg/g

Enzyme Binding Affinity KLE 0.7

PEG Binding Affinity KLS 4.0

Rate constant for adsorption k1 0.004

The results of the calculations are shown in Figure 43 as the dynamic loading of the enzyme as a function of time. Experimental data are also shown. The 95% confidence intervals for the experimental data points are given in Figure 35 of chapter 4. It is clear from this figure that the simple Langmuir kinetic model agrees with the experimental data very well.

Interestingly, there is a small peak in adsorption at very short times (~ 5 min) which may correspond to a rapid but weak cellulase adsorption on the surface followed by a slower

89 relaxation to a stronger bound form, reflecting the so-called Vroman effect (Rabe et al., 2011).

The experiments were done in triplicate and all the data points showed this minor peak relative to adsorption at later times. More investigations are necessary to confirm this effect.

We can hypothesize that the cellulase molecule adsorbs initially in a spherical conformation but transitions quickly to a flatter ellipsoidal conformation as more groups interact with the surface. In this case, it is possible to determine an approximate aspect ratio of the ellipsoid using random sequential adsorption theory and models (RSA).

Figure 43. Adsorption kinetics for batch adsorption of cellulase onto zeolite β (Langmuir). Initial enzyme concentration 1 g/L with a loading of 10 g/L zeolite. Experimental data shown and correspond to Figure 35 of chapter 4. For error bars, see chapter 4.

5.3.2 Adsorption Kinetics – Single Component Cellulase on Zeolite Using Random Sequential

Adsorption Model

The principal parameters in RSA are the collision efficiency and the maximum coverage parameter. RSA models idealize the adsorbing molecule as a sphere and therefore, the cellulase enzyme radius is necessary. Although the enzyme itself is small, some literature suggests that the size of enzyme aggregates in aqueous solution are typically about 50 nm in radius (Rabe et al.,

2011). This was used as an approximate size in our modeling. Slight deviations from this value may be expected to change the optimal parameters but should not impact the quality of the fit in a qualitative sense. The model consists of equations (21) and (24) subject to the initial condition of (26). Figure 44 shows the results of the calculation for different values of these two parameters, i.e. alpha and K. The optimal parameters can be observed by inspection as 0.8e-04 and 0.075 g/g respectively which are consistent with expectations. Although the value for qmE is lower than expected, for the case of RSA this value may be because of packing differences between actual molecules and the model assumptions. The low value of the collision efficiency is not surprising given that not all collisions between the solute molecule and the surface would be successful in binding, for energetically asymmetric molecules and the zeolite surface. We observe that with these best fitting parameters, the RSA model provides a good description of cellulase adsorption.

This indicates some advantages of the RSA model. This is a more accurate topographic representation of the enzyme adsorption process than the abstract and highly ideal form of the

Langmuir isotherm. Furthermore, it is nicely amenable to extensions for mixtures of asymmetric protein mixtures and also proteins and other solutes of differing sizes. Further, reconformation of macromolecular chains on the surface can lead to molecular spreading which can be described

91 nicely by extended RSA models in the literature. These issues are relegated to future work and in this work, we conclude by pointing out the usefulness of the RSA approach in this case.

Table 12. Parameters from RSA model – Best fitting Parameter Symbol Value Units

Maximum coverage qm 0.075 g Collision Parameter α 0.8x10-4 L

K K 0.2387

Maximum Blocking Area θm 0.537

Protein radius ap 50 nm

Figure 44. Adsorption kinetics for batch adsorption of cellulase onto zeolite β (RSA model). Legend denotes different parameter values. Experimental data are shown as blue circles.

5.3.2.1 Adsorption Kinetics of Cellulase and PEG Loading Simultaneously (Vroman Effect)

Next, we consider two cases of adsorption and desorption. In the first, we consider the adsorption of both cellulase and PEG together onto a fresh batch of zeolite in a batch stirred vessel. The initial enzyme dosage is assumed to be at 10 g/L and the other constants are given in

Table 11. The kinetic model for this case consists of Equations 22-25 along with the initial conditions given in equation 26. The results of the calculations are shown in Figure 45. The vertical axis in this figure represents the loading of the enzyme and PEG. Results for three different cases are shown. The first condition represents simple loading of the zeolite with cellulase at an initial concentration of 10 g/L. Under this condition, the PEG loading is identically zero as represented by the dots in the figure. A second condition is considered (blue) where the initial concentration of enzyme and PEG are each at 10 g/L. Under this condition, the cellulase enzyme quickly adsorbs to a peak but the adsorption reduces with increasing time and settles to a lower level. The lower level of adsorption is caused by the competitive adsorption from the PEG which simultaneously increases as shown by the curve E10S10 in this figure. The development of a peak adsorption at short times is a consequence of the balance between the reaction rates shown in Table 13. This is a direct prediction of the Vroman effect in the cellulase

PEG system which should be confirmed by experimental investigations.

Table 13. Kinetic and Equilibrium Parameters for calculations Variable Symbol Value Units

Maximum loading of cellulase and PEG qmE 140 mg/g

Enzyme Binding Affinity KLE 0.7

PEG Binding Affinity KLS 4.0

Rate constant for adsorption k1 0.004

Rate constant k2 k1/ KLE

Rate constant k3 0.004

Rate constant k4 k3/ KLS

Figure 45. Adsorption kinetics for batch adsorption of cellulase and PEG onto zeolites. The solution enzyme concentration is assumed to be 10 g/L and the PEG concentration varies from 0 through 10. qE denotes cellulase loading and qS denotes PEG loading. Legend denotes initial enzyme and PEG concentrations. The condition marked by * corresponds to desorption i.e. elution or unloading phase for column operations. The Vroman effect is shown by the solid blue curve E10S10 with a peak in adsorption for qE.

5.3.2.2 Desorption of Cellulase Using PEG Elutant (Adsorbent Column Unloading)

In practical operation, cellulase is expected to be first loaded from solution, onto the zeolite adsorbent and is followed by desorption using a displacer such as PEG. This could be done either in a chromatographic column, as a part of a desorber column in a continuous simulated moving bed process (SMB), or in batch in a stirred vessel. We consider the last operation to show the significance of the equilibrium and kinetic models studied in this chapter.

Referring to Figure 45 again, the results of the calculations for this case are displayed as the curves E0S10*. The solid curve shows the cellulase enzyme loading on the adsorbent (qE) as a function of time. It decreases uniformly from its initial value of 0.01 g/g, a consequence of the increasing displacement loading of the elutant, PEG. The loading curve for PEG shows a simultaneous increase with time, reflecting the desorption process.

5.3.3 Bi-component Adsorption Equilibrium

Adsorption equilibrium experiments described in the previous chapter were modeled using the bicomponent Langmuir equation. Table 13 and Table 14 list the conditions and other variables assumed for the calculations of the model. Figure 46 and Figure 47 show calculations of the equilibrium loading of cellulase and PEG onto the zeolite with different concentrations of the elutant (from 0 to 0.5 mM, corresponding to 10 g/L). The total coverage of both the cellulase and the PEG molecule is assumed to be equal. The experimental data appear to indicate a maximum of 140 mg/L for the cellulase which was used for further calculations (see Table 13).

The best fitting binding affinities are also shown in this Table 12.

Table 14. Adsorption equilibrium for cellulase and PEG system on zeolite β Variable Symbol Value Units Mass of zeolite m 10 g Volume of solvent (~solution) V 1 L

Initial concentration of cellulase c0 10 g/L

Initial concentration of PEG cs0 10 g/L PEG MW 20 kDa

Figure 46. Adsorption Equilibrium for cellulase on zeolite β, modeled using the bi- component Langmuir Isotherm (Eq. 19-20). Second component is PEG20000. PEG concentrations shown as in legend. Experimental adsorption data for the enzyme loading are shown as individual data points.

It is clear from Figure 46 that the bi-component Langmuir model can be used to fit the data quite well. It appears that as the concentration of the elutant, PEG is increased in solution

96 from 0 through 500 µM, the adsorption of the cellulase is effectively inhibited. Thus, it is possible to achieve significant unloading of the zeolite using PEG as an elutant and also the equilibrium model represented by equations 19-20 can be used effectively for representing adsorption and desorption processes.

Figure 47. Loading of PEG on zeolite β. Other data as in Figure 46. PEG concentrations shown as in legend.

Figure 47 shows the loading of PEG onto the zeolite at different enzyme concentrations.

As the cellulase concentration increases, competition for the adsorption sites leads to reduction in the loading of the PEG. The highest loading of PEG is more resistant to this competition while at low loadings, the enzyme also adsorbs to the surface.

In conclusion, the predictions of the model appear to be in agreement with experimental observations and also the expected behavior of the adsorption as a competitive process on the zeolite surface.

CHAPTER VI: CONCLUSIONS

This study demonstrated that the biocatalytic activity of the enzyme can be improved by enzyme retention and recycle using membrane bioreactor (MBR) and zeolite adsorption. By using MBR, the hydrolysis yield improved by 24.1 % and 43.2 % at a dilution rate of 0.25 h-1 and

0.5 h-1, respectively after 11 hours of hydrolysis. The productivity is consistently higher when hydrolysis is performed in MBR. This increased yield and productivity is due to reduction of product inhibition. However, it should be noted that concentration of the reducing sugar is low and got even lower when dilution rate increased from 0.25 h-1 to 0.5 h-1.

Tween 80 was added to the substrate to minimize the effects of calcium ion on hydrolysis.

Tween 80 addition improved the hydrolysis yield by 10.2 % in the calcium ions. However, when operated in MBR, the hydrolysis yield increased by 26.5 %. The increased yield might be due to the combination of decreasing product inhibition and calcium ion concentration depletion.

MBR was effective only when enzyme loading is higher than 7.5 FPU/ g substrate.

Membrane MWCO had a significant effect on the hydrolysis. In general, lower MWCO resulted better hydrolysis. This might be due to the effective retention of enzymes by the membrane. At higher enzyme loading of 15 FPU/ g substrate, there was no significant effect of membrane

MWCO. The hydrolysis improved by 19.7 % in both cases where MBR is fitted with 10 kDa and

30 kDa MWCO membrane.

The enzyme in the liquid phase of the hydrolysate was effectively recovered using zeolites.

Adsorption isotherms were constructed at varying pH and temperatures to understand factors affecting adsorption. Adsorption isotherms indicate that enzyme adsorption on zeolite is driven mostly by hydrophobic interactions and also by some van der Waal’s effects. At pH 4,

99 maximum binding capacity of the zeolite is 210*10-3 g/g. The adsorption affinity increased as the pH decreased. Even at higher concentrations, salt was not effective at desorbing enzymes from the zeolite. At 1 M NaCl, the amount of desorbed protein accounted for 24 %.

PEG was very efficient at displacing the enzymes from zeolite. The higher molecular weight

PEG is much more effective at desorption than the lower molecular weight polymer. At a concentration of 0.25 mM, PEG 20000 was able to desorb about 75 % of the adsorbed protein.

PEG does not interact with either the pH or electrolytes since it is a neutral polymer (uncharged).

Therefore, it is more attractive as an eluent as compared to various cationic surfactants like

CTAB (cetyl trimethyl ammonium bromide), polyelectrolytes like pDADMAC (poly diallyl dimethyl ammonium chloride) and anionic surfactants like SDS (sodium dodecyl (lauryl) sulfate).

The data from adsorption and desorption processes of the enzymes on zeolite using PEG as an eluent can be fit to Bi-component Langmuir model. This model can evaluate batch and continuous adsorption desorption systems for controlled recycling of the enzymes.

100

CHAPTER VII: RECOMMENDATIONS FOR FUTURE WORK

Further improvements in hydrolysis of synthetic mixture, composed of pulp fiber fines and calcium ions, using membrane bioreactor (MBR) can be achieved by reducing the concentration of calcium ions. Therefore, strategies to reduce the effect of calcium ions on the hydrolysis process or removal of calcium ions from the synthetic mixture need to be explored.

An additional concentration step is necessary to increase the concentration of the sugars for downstream fermentation. Alternatively, performing enzymatic hydrolysis at a higher solid loading, and changing the product removal strategies need to be investigated for improving the output sugar concentration.

Conversion of sugars produced in MBR into value added products such as biofuels and polyhydroxy alkanoates need to be explored. A continuous saccharification and fermentation process – where the sugars separated from hydrolysis mixture are fed to the fermentation unit – can be beneficial for process efficiency. Therefore, it is recommended to explore this continuous saccharification and fermentation process.

Enzyme recycle using other adsorbents need to be investigated. An advanced separation scheme such as simulated moving bed need to be developed using the adsorption models to improve the efficiency of the recycle process.

Finally, it might be worth investigating the efficiency of the process by conducting a comprehensive techno-economic analysis.

101

REFERENCES

Adney, B., Baker, J., 2008. Measurement of Cellulase Activities: Laboratory Analytical Procedure (LAP); Issue Date: 08/12/1996. Technical Report. National Renewable Energy Laboratory (NREL), Colorado, USA.

AFPA, 2018. American Forest and Paper Association [WWW Document]. URL https://afandpa.org/media/statistics (accessed 3.23.20).

Alfani, F., Albanesi, D., Cantarella, M., Scardi, V., Vetromile, A., 1982. Kinetics of enzymatic saccharification of cellulose in a flat-membrane reactor. Biomass 2, 245–253.

Alfani, F., Cantarella, M., Scardi, V., 1983. Use of a membrane reactor for studying enzymatic hydrolysis of cellulose. Journal of Membrane Science 16, 407–416.

Al-Zuhair, S., Al-Hosany, M., Zooba, Y., Al-Hammadi, A., Al-Kaabi, S., 2013. Development of a membrane bioreactor for enzymatic hydrolysis of cellulose. Renewable Energy, The International Conference on Renewable Energy: Generation and Applications 56, 85–89.

Amidon, T.E., Wood, C.D., Shupe, A.M., Wang, Y., Graves, M., Liu, S., 2008. Biorefinery: Conversion of Woody Biomass to Chemicals, Energy and Materials. Journal of Biobased Materials and Bioenergy 2, 100–120.

Andrić, P., Meyer, A.S., Jensen, P.A., Dam-Johansen, K., 2010a. Reactor design for minimizing product inhibition during enzymatic lignocellulose hydrolysis: I. Significance and mechanism of cellobiose and glucose inhibition on cellulolytic enzymes. Biotechnology Advances 28, 308–324.

Andrić, P., Meyer, A.S., Jensen, P.A., Dam-Johansen, K., 2010b. Reactor design for minimizing product inhibition during enzymatic lignocellulose hydrolysis: II. Quantification of inhibition and suitability of membrane reactors. Biotechnology Advances 28, 407–425.

A. Sheldon, R., 2014. Green and sustainable manufacture of chemicals from biomass: state of the art. Green Chemistry 16, 950–963.

Ayawei, N., Ebelegi, A.N., Wankasi, D., 2017. Modelling and Interpretation of Adsorption Isotherms. Journal of Chemistry.

102

Balan, V., 2014. Current Challenges in Commercially Producing Biofuels from Lignocellulosic Biomass. ISRN Biotechnology. Hindawi Publishing Corporation

Bansal, P., Hall, M., Realff, M.J., Lee, J.H., Bommarius, A.S., 2009. Modeling cellulase kinetics on lignocellulosic substrates. Biotechnology Advances, Biotechnology for the Sustainability of Human SocietyInvited Papers from IBS 2008 27, 833–848.

Bárcia, P.S., Silva, J.A.C., Rodrigues, A.E., 2005. Adsorption equilibrium and kinetics of branched hexane isomers in pellets of BETA zeolite. Microporous and Mesoporous Materials 79, 145–163.

Bayer, E.A., Chanzy, H., Lamed, R., Shoham, Y., 1998. Cellulose, cellulases and cellulosomes. Current Opinion in Structural Biology 8, 548–557.

Bélafi-Bakó, K., Koutinas, A., Nemestóthy, N., Gubicza, L., Webb, C., 2006. Continuous enzymatic cellulose hydrolysis in a tubular membrane bioreactor. Enzyme and Microbial Technology 38, 155–161.

Berlin, A., Gilkes, N., Kurabi, A., Bura, R., Tu, M., Kilburn, D., Saddler, J., 2005. Weak lignin- binding enzymes. Appl Biochem Biotechnol 121, 163–170.

Börjesson, J., Engqvist, M., Sipos, B., Tjerneld, F., 2007. Effect of poly(ethylene glycol) on enzymatic hydrolysis and adsorption of cellulase enzymes to pretreated lignocellulose. Enzyme and Microbial Technology 41, 186–195.

Brash, J.L., Horbett, T.A., 1995. Proteins at Interfaces, in: Proteins at Interfaces II, ACS Symposium Series. American Chemical Society, pp. 1–23.

Bravman, T., Zolotnitsky, G., Belakhov, V., Shoham, G., Henrissat, B., Baasov, T., Shoham, Y., 2003. Detailed kinetic analysis of a family 52 glycoside hydrolase: a beta-xylosidase from Geobacillus stearothermophilus. Biochemistry 42, 10528–10536.

Cai, C., Bao, Y., Zhan, X., Lin, X., Lou, H., Pang, Y., Qian, Y., Qiu, X., 2019. Recovering cellulase and increasing glucose yield during lignocellulosic hydrolysis using lignin- MPEG with a sensitive pH response. Green Chem. 21, 1141–1151.

Chen, G., Song, W., Qi, B., Lu, J., Wan, Y., 2013. Recycling cellulase from enzymatic hydrolyzate of acid treated wheat straw by electroultrafiltration. Bioresource Technology 144, 186–193.

103

Cherian, E., Dharmendirakumar, M., Baskar, G., 2015. Immobilization of cellulase onto MnO2 nanoparticles for bioethanol production by enhanced hydrolysis of agricultural waste. Chinese Journal of Catalysis 36, 1223–1229.

Chiku, H., Matsui, M., Murakami, S., Kiyozumi, Y., Mizukami, F., Sakaguchi, K., 2003. Zeolites as new chromatographic carriers for proteins—easy recovery of proteins adsorbed on zeolites by polyethylene glycol. Analytical Biochemistry 318, 80–85.

Christensen, C.H., Rass‐Hansen, J., Marsden, C.C., Taarning, E., Egeblad, K., 2008. The Renewable Chemicals Industry. ChemSusChem 1, 283–289.

Chu, K.H., Feng, X., 2013. Enzymatic conversion of newspaper and office paper to fermentable sugars. Process Safety and Environmental Protection 91, 123–130.

Clark, J.H., 2007. Green chemistry for the second generation biorefinery—sustainable chemical manufacturing based on biomass. Journal of Chemical Technology & Biotechnology 82, 603–609.

Coughlan, M.P., Moloney, A.P., 1988. Isolation of 1,4-β-d-glucan 4-glucanohydrolases of Talaromyces emersonii, in: Methods in Enzymology, Biomass Part A: Cellulose and Hemicellulose. Academic Press, pp. 363–368.

Davies, G., Henrissat, B., 1995. Structures and mechanisms of glycosyl hydrolases. Structure 3, 853–859.

Deshpande, M.V., Eriksson, K.-E., 1984. Reutilization of enzymes for saccharification of lignocellulosic materials. Enzyme and Microbial Technology 6, 338–340.

Doi, R.H., Kosugi, A., 2004. Cellulosomes: plant-cell-wall-degrading enzyme complexes. Nature Reviews Microbiology 2, 541–551.

Du, R., Su, R., Li, X., Tantai, X., Liu, Z., Yang, J., Qi, W., He, Z., 2012. Controlled adsorption of cellulase onto pretreated corncob by pH adjustment. Cellulose 19, 371–380.

DZS, 2020. Database of Zeolite Structures [WWW Document]. URL http://www.iza- structure.org/databases/ (accessed 3.27.20).

Eriksson, T., Börjesson, J., Tjerneld, F., 2002. Mechanism of surfactant effect in enzymatic hydrolysis of lignocellulose. Enzyme Microb Technol 31.

104

Farinas, C.S., Loyo, M.M., Baraldo Junior, A., Tardioli, P.W., Neto, V.B., Couri, S., 2010. Finding stable cellulase and xylanase: evaluation of the synergistic effect of pH and temperature. New Biotechnology, Papers from Symbiosis - The 14th European Congress on Biotechnology (Part 1), Barcelona, September 2009. 27, 810–815.

Flanigen, E. M., 2001. Zeolites and molecular sieves: An historical perspective, in: van Bekkum, H., Flanigen, E. M., Jacobs, P.A., Jansen, J.C. (Eds.), Introduction to Zeolite Science and Practice. Elsevier, pp. 11–35.

Gan, Q., Allen, S.J., Taylor, G., 2002. Design and operation of an integrated membrane reactor for enzymatic cellulose hydrolysis. Biochemical Engineering Journal 12, 223–229.

Ghose, T.K., Kostick, J.A., 1970. A model for continuous enzymatic saccharification of cellulose with simultaneous removal of glucose syrup. Biotechnol. Bioeng. 12, 921–946.

Gírio, F.M., Fonseca, C., Carvalheiro, F., Duarte, L.C., Marques, S., Bogel-Łukasik, R., 2010. Hemicelluloses for fuel ethanol: A review. Bioresource Technology, Special Issue on Lignocellulosic Bioethanol: Current Status and Perspectives 101, 4775–4800.

Gomes, D., Rodrigues, A.C., Domingues, L., Gama, M., 2015. Cellulase recycling in biorefineries—is it possible? Appl Microbiol Biotechnol 99, 4131–4143.

Grieco, S.A., Ramarao, B.V., 2013. Removal of TCEP from aqueous solutions by adsorption with zeolites. Colloids and Surfaces A: Physicochemical and Engineering Aspects 434, 329–338.

Hahn-Hägerdal, B., Galbe, M., Gorwa-Grauslund, M.F., Lidén, G., Zacchi, G., 2006. Bio-ethanol – the fuel of tomorrow from the residues of today. Trends in Biotechnology 24, 549–556.

Haven, M.Ø., Jørgensen, H., 2013. Adsorption of β-glucosidases in two commercial preparations onto pretreated biomass and lignin. Biotechnol Biofuels 6, 165.

Hedström Annelie, 2001. Ion Exchange of Ammonium in Zeolites: A Literature Review. Journal of Environmental Engineering 127, 673–681.

Herlet, J., Kornberger, P., Roessler, B., Glanz, J., Schwarz, W.H., Liebl, W., Zverlov, V.V., 2017. A new method to evaluate temperature vs. pH activity profiles for biotechnological relevant enzymes. Biotechnology for Biofuels 10, 234.

105

Holmberg, K., 2018. Interactions between surfactants and hydrolytic enzymes. Colloids and Surfaces B: Biointerfaces, Honoring Piero Baglioni 168, 169–177.

Horn, S.J., Vaaje-Kolstad, G., Westereng, B., Eijsink, V., 2012. Novel enzymes for the degradation of cellulose. Biotechnol Biofuels 5, 45.

Humbird, D., Davis, R., Tao, L., Kinchin, C., Hsu, D., Aden, A., Schoen, P., Lukas, J., Olthof, B., Worley, M., Sexton, D., Dudgeon, D., 2011. Process Design and Economics for Biochemical Conversion of Lignocellulosic Biomass to Ethanol: Dilute-Acid Pretreatment and Enzymatic Hydrolysis of Corn Stover (No. NREL/TP-5100-47764). National Renewable Energy Lab. (NREL), Golden, CO (United States).

Ingle, A.P., Rathod, J., Pandit, R., da Silva, S.S., Rai, M., 2017. Comparative evaluation of free and immobilized cellulase for enzymatic hydrolysis of lignocellulosic biomass for sustainable bioethanol production. Cellulose 24, 5529–5540.

Ishola, M.M., Jahandideh, A., Haidarian, B., Brandberg, T., Taherzadeh, M.J., 2013. Simultaneous saccharification, filtration and fermentation (SSFF): A novel method for bioethanol production from lignocellulosic biomass. Bioresource Technology 133, 68– 73.

Jäger, G., Wu, Z., Garschhammer, K., Engel, P., Klement, T., Rinaldi, R., Spiess, A.C., Büchs, J., 2010. Practical screening of purified cellobiohydrolases and endoglucanases with α- cellulose and specification of hydrodynamics. Biotechnology for Biofuels 3, 18.

Johnson, E., 2016. Integrated enzyme production lowers the cost of cellulosic ethanol. Biofuels, Bioproducts and Biorefining 10, 164–174.

Johnson, P.R., Elimelech, M., 1995. Dynamics of Colloid Deposition in Porous Media: Blocking Based on Random Sequential Adsorption. Langmuir 11, 801–812.

Jong, E. de, Higson, A., Walsh, P., Wellisch, M., 2012. Product developments in the bio-based chemicals arena. Biofuels, Bioproducts and Biorefining 6, 606–624.

Jørgensen, H., Kristensen, J.B., Felby, C., 2007. Enzymatic conversion of lignocellulose into fermentable sugars: challenges and opportunities. Biofuels, Bioprod. Bioref. 1, 119–134.

Jørgensen, H., Pinelo, M., 2017. Enzyme recycling in lignocellulosic biorefineries. Biofuels, Bioproducts and Biorefining 11, 150–167.

106

Judd, S., 2010. The MBR Book: Principles and Applications of Membrane Bioreactors for Water and Wastewater Treatment. Second Edition. Elsevier.

Kadam, K.L., Rydholm, E.C., McMillan, J.D., 2004. Development and Validation of a Kinetic Model for Enzymatic Saccharification of Lignocellulosic Biomass. Biotechnology Progress 20, 698–705.

Kesraoui‐Ouki, S., Cheeseman, C.R., Perry, R., 1994. Natural zeolite utilisation in pollution control: A review of applications to metals’ effluents. Journal of Chemical Technology & Biotechnology 59, 121–126.

Kim, H.J., Kim, S.B., Kim, C.J., 2007. The effects of nonionic surfactants on the pretreatment and enzymatic hydrolysis of recycled newspaper. Biotechnol. Bioprocess Eng. 12, 147– 151.

Kim, S.B., Kim, H.J., Kim, C.J., 2006. Enhancement of the enzymatic digestibility of waste newspaper using tween. Appl Biochem Biotechnol 130, 486–495.

Klei, H.E., Sundstrom, D.W., Coughlin, R.W., Ziolkowski, K., 1981. Hollow-fiber enzyme reactors in cellulose hydrolysis. Biotechnol. Bioeng. Symp.; (United States) 11.

Klein-Marcuschamer, D., Oleskowicz-Popiel, P., Simmons, B.A., Blanch, H.W., 2012. The challenge of enzyme cost in the production of lignocellulosic biofuels. Biotechnol Bioeng 109.

Knutsen, J.S., Davis, R.H., 2004. Cellulase Retention and Sugar Removal by Membrane Ultrafiltration During Lignocellulosic Biomass Hydrolysis, in: Finkelstein, M., McMillan, J.D., Davison, B.H., Evans, B. (Eds.), Proceedings of the Twenty-Fifth Symposium on Biotechnology for Fuels and Chemicals Held May 4–7, 2003, in Breckenridge, CO, Biotechnology for Fuels and Chemicals. Humana Press, pp. 585–599.

Krohn, J.E., Tsapatsis, M., 2005. Amino Acid Adsorption on Zeolite β. Langmuir 21, 8743– 8750.

Kruger, N.J., 1994. The Bradford Method for Protein Quantitation, in: Walker, J.M. (Ed.), Basic Protein and Peptide Protocols, Methods in Molecular BiologyTM. Humana Press, Totowa, NJ, pp. 9–15.

107

Kubota, M., Nakabayashi, T., Matsumoto, Y., Shiomi, T., Yamada, Y., Ino, K., Yamanokuchi, H., Matsui, M., Tsunoda, T., Mizukami, F., Sakaguchi, K., 2008. Selective adsorption of bacterial cells onto zeolites. Colloids and Surfaces B: Biointerfaces 64, 88–97.

Kumar, D., Murthy, G.S., 2017. Development and validation of a stochastic molecular model of cellulose hydrolysis by action of multiple cellulase enzymes. Bioresour. Bioprocess. 4, 54.

Kumar, D., Murthy, G.S., 2016. Chapter 7 - Enzymatic Hydrolysis of Cellulose for Ethanol Production: Fundamentals, Optimal Enzyme Ratio, and Hydrolysis Modeling, in: Gupta, V.K. (Ed.), New and Future Developments in Microbial Biotechnology and Bioengineering. Elsevier, Amsterdam, pp. 65–78.

Kurrat, R., Prenosil, J.E., Ramsden, J.J., 1997. Kinetics of Human and Bovine Serum Albumin Adsorption at Silica–Titania Surfaces. Journal of Colloid and Interface Science 185, 1–8.

Lee, S.-G., Kim, H.-S., 1993. Optimal operating policy of the ultrafiltration membrane bioreactor for enzymatic hydrolysis of cellulose. Biotechnology and Bioengineering 42, 737–746.

Li, Y., Sun, Z., Ge, X., Zhang, J., 2016. Effects of lignin and surfactant on adsorption and hydrolysis of cellulases on cellulose. Biotechnology for Biofuels 9, 1–9.

Lippke, B., Gustafson, R., Venditti, R., Volk, T., Oneil, E., Johnson, L., Puettmann, M., Steele, P., 2011. Sustainable Biofuel Contributions to Carbon Mitigation and Energy Independence. Forests 2, 861–874.

Liu, G., Zhang, J., Bao, J., 2016. Cost evaluation of cellulase enzyme for industrial-scale cellulosic ethanol production based on rigorous Aspen Plus modeling. Bioprocess Biosyst Eng 39, 133–140.

Liu, S., 2015a. Cooperative adsorption on solid surfaces. Journal of Colloid and Interface Science 450, 224–238.

Liu, S., 2015b. A mathematical model for competitive adsorptions. Separation and Purification Technology 144, 80–89.

Liu, S., 2014. A Visit on the Kinetics of Surface Adsorption. Journal of Bioprocess Engineering and Biorefinery 3, 100-114.

108

Lu, Y., Yang, B., Gregg, D., Saddler, J.N., Mansfield, S.D., 2002. Cellulase adsorption and an evaluation of enzyme recycle during hydrolysis of steam-exploded softwood residues. Applied Biochemistry and Biotechnology 14.

Mahboubi, A., Uwineza, C., Doyen, W., De Wever, H., Taherzadeh, M.J., 2020. Intensification of lignocellulosic bioethanol production process using continuous double-staged immersed membrane bioreactors. Bioresource Technology 296, 122314.

Malmali, M., Stickel, J., Wickramasinghe, S.R., 2015. Investigation of a submerged membrane reactor for continuous biomass hydrolysis. Food and Bioproducts Processing 96, 189– 197.

Martin Alonso, D., Q. Bond, J., A. Dumesic, J., 2010. Catalytic conversion of biomass to biofuels. Green Chemistry 12, 1493–1513.

Martinez, D., Berka, R.M., Henrissat, B., Saloheimo, M., Arvas, M., Baker, S.E., Chapman, J., Chertkov, O., Coutinho, P.M., Cullen, D., Danchin, E.G.J., Grigoriev, I.V., Harris, P., Jackson, M., Kubicek, C.P., Han, C.S., Ho, I., Larrondo, L.F., de Leon, A.L., Magnuson, J.K., Merino, S., Misra, M., Nelson, B., Putnam, N., Robbertse, B., Salamov, A.A., Schmoll, M., Terry, A., Thayer, N., Westerholm-Parvinen, A., Schoch, C.L., Yao, J., Barabote, R., Nelson, M.A., Detter, C., Bruce, D., Kuske, C.R., Xie, G., Richardson, P., Rokhsar, D.S., Lucas, S.M., Rubin, E.M., Dunn-Coleman, N., Ward, M., Brettin, T.S., 2008. Genome sequencing and analysis of the biomass-degrading fungus Trichoderma reesei (syn. Hypocrea jecorina ). Nature Biotechnology 26, 553–560.

McCusker, L.B., Baerlocher, C., 2001. Chapter 3 Zeolite structures, in: van Bekkum, H., Flanigen, E.M., Jacobs, P.A., Jansen, J.C. (Eds.), Studies in Surface Science and Catalysis, Introduction to Zeolite Science and Practice. Elsevier, pp. 37–67.

McKinsey, 2019. The packaging, pulp and paper industry in the next decade | McKinsey [WWW Document]. URL https://www.mckinsey.com/industries/paper-forest-products-and- packaging/our-insights/pulp-paper-and-packaging-in-the-next-decade-transformational- change (accessed 3.23.20).

Min, B., Jampana, S.N., Thomas, C.M., Ramarao, B.V., 2018. Study of Buffer Substitution Using Inhibitory Compound CaCO₃ in Enzymatic Hydrolysis of Paper Mill Waste Fines. JKTAPPI 50, 77–82.

Min, B.C., 2017. Enzymatic Hydrolysis of Cellulosic Fibers in Paper Mill Waste. State University of New York College of Environmental Science and Forestry. ProQuest Dissertations Publishing. 10270832.

109

Min, B.C., Bhayani, B.V., Jampana, V.S., Ramarao, B.V., 2015. Enhancement of the enzymatic hydrolysis of fines from recycled paper mill waste rejects. Bioresour. Bioprocess. 2, 1– 10.

Miret, C., Chazara, P., Montastruc, L., Negny, S., Domenech, S., 2016. Design of bioethanol green supply chain: Comparison between first and second generation biomass concerning economic, environmental and social criteria. Computers & Chemical Engineering 85, 16– 35.

Monte, M.C., Fuente, E., Blanco, A., Negro, C., 2009. Waste management from pulp and paper production in the European Union. Waste Management 29, 293–308.

Mores, W.D., Knutsen, J.S., Davis, R.H., 2001. Cellulase recovery via membrane filtration. Appl Biochem Biotechnol 91, 297–309.

Munsch, S., Hartmann, M., Ernst, S., 2001. Adsorption and separation of amino acids from aqueous solutions on zeolites. Chem. Commun. 1978–1979.

Nanda, S., Azargohar, R., Dalai, A.K., Kozinski, J.A., 2015. An assessment on the sustainability of lignocellulosic biomass for biorefining. Renewable and Sustainable Energy Reviews 50, 925–941.

Ohlson, I., Trägårdh, G., Hahn-Hägerdal, B., 1984. Enzymatic hydrolysis of sodium-hydroxide- pretreated sallow in an ultrafiltration membrane reactor. Biotechnol. Bioeng. 26, 647– 653.

Otter, D.E., Munro, P.A., Scott, G.K., Geddes, R., 1989. Desorption of Trichoderma reesei cellulase from cellulose by a range of desorbents. Biotechnology and Bioengineering 34, 291–298.

Otter, D.E., Munro, P.A., Scott, G.K., Geddes, R., 1984. Elution of Trichoderma reesei cellulase from cellulose by pH adjustment with sodium hydroxide. Biotechnol Lett 6, 369–374.

Ouyang, J., Liu, B., Zhang, M., Zheng, Z., Yu, H., 2013. Enzymatic hydrolysis, adsorption, and recycling during hydrolysis of bagasse sulfite pulp. Bioresource Technology 146, 288– 293.

Pal, G.P., Elce, J.S., Jia, Z., 2001. Dissociation and Aggregation of Calpain in the Presence of Calcium. J. Biol. Chem. 276, 47233–47238.

110

Pandey, M.P., Kim, C.S., 2011. Lignin Depolymerization and Conversion: A Review of Thermochemical Methods. Chemical Engineering & Technology 34, 29–41.

Pareek, N., Gillgren, T., Jonsson, L.J., 2013. Adsorption of proteins involved in hydrolysis of lignocellulose on lignins and hemicelluloses. Bioresour Technol 148.

Payne, C.M., Knott, B.C., Mayes, H.B., Hansson, H., Himmel, M.E., Sandgren, M., Ståhlberg, J., Beckham, G.T., 2015. Fungal Cellulases. Chem. Rev. 115, 1308–1448.

Perea-Moreno, M.-A., Samerón-Manzano, E., Perea-Moreno, A.-J., 2019. Biomass as Renewable Energy: Worldwide Research Trends. Sustainability 11, 863.

Pino, M.S., Rodríguez-Jasso, R.M., Michelin, M., Flores-Gallegos, A.C., Morales-Rodriguez, R., Teixeira, J.A., Ruiz, H.A., 2018. Bioreactor design for enzymatic hydrolysis of biomass under the biorefinery concept. Chemical Engineering Journal 347, 119–136.

Pribowo, A., Arantes, V., Saddler, J.N., 2012. The adsorption and enzyme activity profiles of specific Trichoderma reesei cellulase/xylanase components when hydrolyzing steam pretreated corn stover. Enzyme and Microbial Technology 50, 195–203.

Qi, B., Luo, J., Chen, G., Chen, X., Wan, Y., 2012. Application of ultrafiltration and nanofiltration for recycling cellulase and concentrating glucose from enzymatic hydrolyzate of steam exploded wheat straw. Bioresource Technology 104, 466–472.

Qing, Q., Yang, B., Wyman, C.E., 2010. Impact of surfactants on pretreatment of corn stover. Bioresour Technol 101.

Rabe, M., Verdes, D., Seeger, S., 2011. Understanding protein adsorption phenomena at solid surfaces. Advances in Colloid and Interface Science 162, 87–106.

Ramaswamy, S., Huang, H.-J., Ramarao, B.V. (Eds.), 2013. Front Matter, in: Separation and Purification Technologies in Biorefineries. John Wiley & Sons, Ltd.

Rodrigues, A.C., Felby, C., Gama, M., 2014. Cellulase stability, adsorption/desorption profiles and recycling during successive cycles of hydrolysis and fermentation of wheat straw. Bioresource Technology 156, 163–169.

111

Rodrigues, A.C., Haven, M.Ø., Lindedam, J., Felby, C., Gama, M., 2015. Celluclast and Cellic® CTec2: Saccharification/fermentation of wheat straw, solid–liquid partition and potential of enzyme recycling by alkaline washing. Enzyme and Microbial Technology 79–80, 70– 77.

Rodrigues, A.C., Leitão, A.F., Moreira, S., Felby, C., Gama, M., 2012. Recycling of cellulases in lignocellulosic hydrolysates using alkaline elution. Bioresource Technology 110, 526– 533.

Saqib, A.A.N., Whitney, P.J., 2011. Differential behaviour of the dinitrosalicylic acid (DNS) reagent towards mono- and di-saccharide sugars. Biomass and Bioenergy, ‘Biofuels and Bioenergy: Challenges and Opportunities’ Proceedings of a joint workshop of IEA Bioenergy Tasks 29, 31 and 39, August 2006, Vancouver, British Columbia, Canada 35, 4748–4750.

Scott, G.M., Abubakr, S., Smith, A., 1995. Sludge characteristics and disposal alternatives for the pulp and paper industry. Presented at the International environmental conference, Atlanta, GA, pp. 269–279.

Shang, Y., Su, R., Huang, R., Yang, Y., Qi, W., Li, Q., He, Z., 2014. Recycling cellulases by pH- triggered adsorption-desorption during the enzymatic hydrolysis of lignocellulosic biomass. Appl Microbiol Biotechnol 98, 5765–5774.

Sipos, B., Dienes, D., Schleicher, Á., Perazzini, R., Crestini, C., Siika-aho, M., Réczey, K., 2010. Hydrolysis efficiency and enzyme adsorption on steam-pretreated spruce in the presence of poly (ethylene glycol). Enzyme Microb Technol 47.

Sjostrom, E., 1993. Wood Chemistry: Fundamentals and Applications. Second Edition. Academia Press, San Diego, CA.

Sluiter, A., Hames, B., Ruiz, R., Scarlata, C., Sluiter, J., Templeton, D., Crocker, D., 2008. Determination of Structural Carbohydrates and Lignin in Biomass. Technical Report. National Renewable Energy Laboratory (NREL), Colorado, USA.

Smith, B.T., Knutsen, J.S., Davis, R.H., 2010. Empirical Evaluation of Inhibitory Product, Substrate, and Enzyme Effects During the Enzymatic Saccharification of Lignocellulosic Biomass. Appl Biochem Biotechnol 161, 468–482.

Soldatkin, O.O., Shelyakina, M.K., Arkhypova, V.N., Soy, E., Kirdeciler, S.K., Ozansoy Kasap, B., Lagarde, F., Jaffrezic-Renault, N., Akata Kurç, B., Soldatkin, A.P., Dzyadevych,

112

S.V., 2015. Nano- and microsized zeolites as a perspective material for potentiometric biosensors creation. Nanoscale Research Letters 10, 59.

Stickel, J.J., Adhikari, B., Sievers, D.A., Pellegrino, J., 2018. Continuous enzymatic hydrolysis of lignocellulosic biomass in a membrane-reactor system. Journal of Chemical Technology & Biotechnology 93, 2181–2190.

Sun, R., Sun, X.F., Tomkinson, J., 2003. Hemicelluloses and Their Derivatives, in: Hemicelluloses: Science and Technology, ACS Symposium Series. American Chemical Society, pp. 2–22.

Talbot, J., Tarjus, G., Van Tassel, P.R., Viot, P., 2000. From car parking to protein adsorption: an overview of sequential adsorption processes. Colloids and Surfaces A: Physicochemical and Engineering Aspects 165, 287–324.

Townsend, R.P., Coker, E.N., 2001. Chapter 11 Ion exchange in zeolites, in: van Bekkum, H., Flanigen, E.M., Jacobs, P.A., Jansen, J.C. (Eds.), Studies in Surface Science and Catalysis, Introduction to Zeolite Science and Practice. Elsevier, pp. 467–524.

Tu, M., Chandra, R.P., Saddler, J.N., 2007a. Recycling Cellulases during the Hydrolysis of Steam Exploded and Ethanol Pretreated Lodgepole Pine. Biotechnology Progress 23, 1130–1137.

Tu, M., Chandra, R.P., Saddler, J.N., 2007b. Evaluating the Distribution of Cellulases and the Recycling of Free Cellulases during the Hydrolysis of Lignocellulosic Substrates. Biotechnology Progress 23, 398–406.

Tu, M., Zhang, X., Paice, M., MacFarlane, P., Saddler, J.N., 2009. The potential of enzyme recycling during the hydrolysis of a mixed softwood feedstock. Bioresource Technology 100, 6407–6415.

Tuck, C.O., Pérez, E., Horváth, I.T., Sheldon, R.A., Poliakoff, M., 2012. Valorization of Biomass: Deriving More Value from Waste. Science 337, 695–699.

Tursi, A., 2019. A review on biomass: importance, chemistry, classification, and conversion. Biofuel Research Journal 6, 962–979.

Vinzant, T.B., Ehrman, C.I., Adney, W.S., Thomas, S.R., Himmel, M.E., 1997. Simultaneous saccharification and fermentation of pretreated hardwoods. Appl Biochem Biotechnol 62, 99–104.

113

Wang, G., Zhang, X., Wang, Li, Wang, K., Peng, F., Wang, Linsong, 2012. The activity and kinetic properties of cellulases in substrates containing metal ions and acid radicals. Advances in Biological Chemistry 2, 390–395.

Wang, S., Peng, Y., 2010. Natural zeolites as effective adsorbents in water and wastewater treatment. Chemical Engineering Journal 156, 11–24.

WBA, 2019. WBA - Worldbioenergy [WWW Document]. URL https://worldbioenergy.org/ (accessed 3.24.20).

Withers, S.G., 2001. Mechanisms of glycosyl transferases and hydrolases. Carbohydrate Polymers 44, 325–337.

Wu, J., Ju, L.-K., 1998. Enhancing Enzymatic Saccharification of Waste Newsprint by Surfactant Addition. Biotechnology Progress 14, 649–652.

Wustenberg, T., 2014. Cellulose, in: Cellulose and Cellulose Derivatives in the Food Industry. John Wiley & Sons, Ltd, pp. 91–142.

Yang, S., Ding, W., Chen, H., 2006. Enzymatic hydrolysis of rice straw in a tubular reactor coupled with UF membrane. Process Biochemistry 41, 721–725.

Zhang, M., Su, R., Li, Q., Qi, W., He, Z., 2011. Enzymatic saccharification of pretreated corn stover in a fed-batch membrane bioreactor. Bioenerg. Res. 4, 134–140.

Zhang, Z., Donaldson, A.A., Ma, X., 2012. Advancements and future directions in enzyme technology for biomass conversion. Biotechnology Advances, Biorefining: Thermochemical and Enzymatic Biomass Conversion 30, 913–919.

Zhong, L., Matthews, J.F., Hansen, P.I., Crowley, M.F., Cleary, J.M., Walker, R.C., Nimlos, M.R., Brooks, C.L., Adney, W.S., Himmel, M.E., Brady, J.W., 2009. Computational simulations of the Trichoderma reesei cellobiohydrolase I acting on microcrystalline cellulose Iβ: the enzyme–substrate complex. Carbohydrate Research 344, 1984–1992.

Zhu, Z., Sathitsuksanoh, N., Zhang, Y.-H.P., 2009. Direct quantitative determination of adsorbed cellulase on lignocellulosic biomass with its application to study cellulase desorption for potential recycling. Analyst 134, 2267–2272.

114

BIOGRAPHICAL SKETCH

Education Doctor of Philosophy in Bioprocess Engineering Aug 2013 - May 2020

State University of New York College of Environmental Science and Forestry, Syracuse, NY Master of Science in Industrial Biotechnology Aug 2010 - Nov 2012 University of Boras, Boras, Sweden Bachelor of Technology in Biotechnology Sep 2006 - May 2010

Chaitanya Bharathi Institute of Technology, Hyderabad, India

Publications Jampana, S.R., Jia, L., Ramarao, B.V., Kumar, D. Experimental Investigation of the Adsorption and Desorption of Cellulase Enzymes on Zeolite – β for Enzyme Recycling Applications. Bioprocess and Biosystems Engineering. (submitted)

Min, B., Jampana, S.N., Thomas, C.M., Ramarao, B.V., 2018. Study of Buffer Substitution Using Inhibitory Compound CaCO₃ in Enzymatic Hydrolysis of Paper Mill Waste Fines. J Korea TAPPI 50, 77–82.

Min, B.C., Bhayani, B.V., Jampana, V.S., Ramarao, B.V., 2015. Enhancement of the enzymatic hydrolysis of fines from recycled paper mill waste rejects. Bioresour. Bioprocess. 2, 1–10.

Conference Presentations Venkata Jampana and Bandaru V. Ramarao. 2018. Process Intensification of the Hydrolysis of Cellulose Fibers by Integration of Membrane Separations and Hydrolysis for Enzyme Recycle. American Institute of Chemical Engineers, Pittsburgh, PA, USA. N.V.R. Jampana, R. Kasinathan, B.C. Min and Bandaru V. Ramarao. 2016. Enzymatic Hydrolysis and Production of Polyhydroxyalkanoates from Papermaking Pulp Fines. 2016. 14th International Symposium on Bioplastics, Biocomposites and Biorefining, Guelph, Canada. Christopher Thomas, Byeongcheol Min, Venkata S.N.R. Jampana, Bhavin V. Bhayani and Bandaru V. Ramarao. 2015. Conversion of Cellulosic Waste to Glucose. The 15th Syracuse COE Symposium, Syracuse, NY, USA.

115