Understanding Cellulose Primary and Secondary

UNDERSTANDING CELLULOSE PRIMARY AND SECONDARY PYROLYSIS REACTIONS TO ENHANCE THE PRODUCTION OF ANHYDROSACCHARIDES AND TO BETTER PREDICT THE COMPOSITION OF CARBONACEOUS RESIDUES

ZHOUHONG WANG

A dissertation submitted in partial fulfillment of the requirements for the degree of

DOCTOR OF PHILOSOPHY

WASHINGTON STATE UNIVERSITY Department of Biological Systems Engineering

DECEMBER 2013

To the Faculty of Washington State University:

The members of the Committee appointed to examine the dissertation of ZHOUHONG WANG find it satisfactory and recommend that it be accepted.

______Manuel Garcia-Pérez, Ph.D., Chair

______Armando G. McDonald, Ph.D.

______Su Ha, Ph.D.

______Shyam S. Sablani, Ph.D.

ACKNOWLEDGMENT

I would like to thank Dr. Manuel Garcia-Perez for his support and advice during my PhD studies.

It’s my great honor to learn from his passion on research and his oversight of my research.

I would like to thank my committee members: Dr. Armando McDonald, Dr. Su Ha, and Dr.

Shyam Sablani for their kind guidance and continued participation.

I would like to thank Prof. Sascha Kersten for the support during my visit to his group

(Sustainable Process Technology (original Thermal Chemical Conversion of Biomass),

University of Twente, the Netherlands). Thanks to all the members in SPT group (Prof Wim van

Swaaij, Prof. Jean-Paul Lange, Prof. van den Berg, Dr. Wim Brilman, Dr. Louis van der Ham,

Dr. Guus van Rossum, Dr. BoeloSchuur, Dr. Roel Westerhof, Dr. Michal Gramblicka, Johan

Agterhorst, Karst van Bree, Erna Fränzel-Luiten, Benno Knaken, Yvonne Bruggert-ter Huurne,

Maria Castellvi Barnes, Laura Garcia Alba, Stijn Oudenhoven, Rens Contact, Ying Du, Xiaohua

Li, Jeroen de Graaf, Joram Boegborn, Tim Hilbert, Michiel van Kuppevelt) for the help and joy they kindly offered. Special thanks to Dr. Roel Westerhof for all his help and recommendations when writing my papers.

Also I would like to thank all the staff members in our department (Wayne DeWitt, Vincent

Himsl, Joan Hagedorn, Jonathan Lomber, John Anderson, Dorota Wilk, Pat Huggins and Pat

iii

King), who helped me in the lab with the instruments, dealing with financial problems, and on the administrative work. Really thank you for all the support and help you offered.

My colleagues (Shuai Zhou, Shi-Shen Liaw, Brennan Pecha, Matt Smith, Jieni Lian, Robert

Johnson, Filip Stankovikj, Jesus A Garcia, Raul Pelaez, Waled Suliman), thank you for all those nice and enlightening discussions during my whole time at Washington State University. And also thank you for those great hang outs. I really enjoyed my time here in both studying and living.

UNDERSTANDING CELLULOSE PRIMARY AND SECONDARY PYROLYSIS REACTIONS TO ENHANCE THE PRODUCTION OF ANHYDROSACCHARIDES AND TO BETTER PREDICT THE COMPOSITION OF CARBONACEOUS RESIDUES

Abstract

by Zhouhong Wang, Ph.D. Washington State University December 2013

Chair: Manuel Garcia-Perez

Pyrolysis is a promising method to convert biomass into an oil that can be refined into valuable fuels and chemicals as well as a carbonaceous residue that can be converted into adsorbents. In pyrolysis, cellulose, the most abundant biomass constituent, is thermally converted into mono- and oligo-anhydrosugars (chiefly levoglucosan and cellobiosan) in high yields with small quantities of a carbonaceous “bio-char” residue. However, the reaction mechanisms for cellulose pyrolysis have not been well investigated.

The major goal of this dissertation is to advance our knowledge on the pyrolysis mechanisms responsible for the formation of anhydrosugars and carbonaceous residues under slow and fast pyrolysis. Our hypothesis is that under fast heating rate conditions, cellulose forms a very reactive liquid intermediate of cellulose primary products (levoglucosan, cellobiosan) which, if

v not evaporated fast enough, will undergo cross-linking reactions. Cross-linked product formation is a critical step for the formation of carbonaceous residues under slow heating rate conditions.

It was found that cellulose crystallinity can affect the formation rate of this liquid intermediate, which is promoted by amorphous state. Our research shows that this liquid state is a temperature controlled step that enhances dehydration reactions and cross-linking reactions if persistent. We also determined that secondary reactions in the liquid phase derive not from levoglucosan remaining in the liquid phase, but from oligo-anhydrosugars like cellobiosan.

Extended studies found sulfuric acid inducing dehydration and cross-linking reactions during pyrolysis of cellulose and levoglucosan. In pyrolysis of cellobiosan, sulfuric acid increased the yield of levoglucosan. The concentration of sulfuric acid at which the yield of levoglucosan from the pyrolysis of Douglas fir wood is maximized near the yield was found from cellulose.

The changes occurred during slow heating of cellulose were studied using 13C-NMR, FTIR, IEC and SEM. A new reaction mechanism that includes the formation of cross-linked sugars was developed to explain the evolution of carbonaceous residues from cellulose pyrolysis at slow heating rates.

The results in this dissertation provide fundamental insights into cellulose thermal reactions for developing new strategies to enhance yield of anhydrosugars or valuable chemicals and to better predict the composition of carbonaceous residues formed.

Table of Contents

ACKNOWLEDGMENT...... iii

Abstract ...... v

Table of Contents ...... vii

List of Tables ...... x

List of Figures ...... xi

Chapter 1 Introduction ...... 1

1.1 Multiscale Structure of Lignocellulosic Materials ...... 3

1.2 Cellulose ...... 4

1.3 Cellulose Characterization ...... 5

1.4 Cellulose Thermochemical Reactions ...... 8

1.5 Conclusions of literature review ...... 32

1.6 Dissertation objective ...... 33

1.7 Methodology ...... 34

1.8 Publications ...... 36

Reference ...... 40

Chapter 2 Effect of Cellulose Crystallinity on the Formation of a Liquid Intermediate and on Product Distribution during Pyrolysis ...... 49

2.1 Introduction ...... 51

2.2 Materials and Methods ...... 53

2.3 Results and Discussions ...... 55

2.4 Conclusions ...... 73

vii

Acknowledgements ...... 74

Reference ...... 75

Chapter 3 Effect of Pyrolysis Temperature on the Formation of Anhydrosugars during the Fast Pyrolysis of Cellulose in a Wire Mesh Reactor at Atmospheric Pressure ...... 80

3.1 Introduction ...... 83

3.2 Materials and Methods ...... 86

3.3 Results and Discussions ...... 92

3.4 Conclusions ...... 106

Acknowledgement ...... 107

Reference ...... 108

Chapter 4 Understanding the Effect of Sulfuric Acid on the Pyrolysis of Acid Washed Douglas fir in an Atmospheric Wire Mesh Reactor...... 112

4.1 Introduction ...... 115

4.2 Materials and Methods ...... 117

4.3 Results and Discussions ...... 121

4.4 Conclusions ...... 135

Acknowledgement ...... 136

Reference ...... 138

Chapter 5 Effect of Cellulose Crystallinity on Solid/Liquid Phase Reactions Responsible for the Formation of Carbonaceous Residues during Slow Pyrolysis ...... 142

5.1 Introduction ...... 144

5.2 Materials and Methods ...... 147

5.3 Results and Discussions ...... 151

5.4 Conclusions ...... 184

viii

Acknowledgements ...... 186

Reference ...... 187

Chapter 6 Conclusions and Recommendations ...... 193

6.1 General Conclusions ...... 193

6.2 Recommendations ...... 198

6.3 Scientific contributions ...... 199

Appendix ...... 202

Appendix A: Understanding Lignin-Cellulose Interactions during the Pyrolysis of their Blends ...... 203

A.1 Introduction ...... 205

A.2 Materials and Methods ...... 207

A.3 Results and Discussions ...... 211

A.4 Conclusions ...... 236

Acknowledgement ...... 237

Reference ...... 238

List of Tables

Chapter 2

Table 2.1 Compounds identified by Py-GC/MS of cellulose ...... 67

Chapter 3

Table 3.1 Values of Activation Energy (Ea) and pre-exponential factor (A) obtained for control and ball-milled cellulose using a fast speed camera ...... 95

Chapter 5

Table 5.1 FTIR peak assignments [43-48] ...... 149

Table 5.2 Overall kinetic parameters for the pyrolysis of cellulose ...... 154

Table 5.3 Kinetic parameters and stoichiometric coefficients obtained for cellulose pyrolysis in the 260-300 and 300-400 oC temperature ranges ...... 180

Appendix A

Table A.1 Compounds identified in Py-GC/MS with their major ion and retention time ...... 223

List of Figures

Chapter 1

Figure 1.1 Biofuel production from biomass in a sustainable way [1] ...... 2

Figure 1.2 Structure of lignocellulose [9] ...... 3

Figure 1.3 Cellulose structure with hydrogen bonds. Redrawn from [11] ...... 4

Figure 1.4 Microcrystalline cellulose model with crystalline and amorphous structure. Redraw from [12] ...... 5

Figure 1.5 Cellulose thermal reaction models: (A) Broido-Shafizadeh model [58], Ei: -1 - 242.7 kJ/mol Log Ai: 21.2 log(min ); Ev: 197.9 kJ/mol, Log Av: 16.3 log(min 1 -1 ); Ec: 153.1 kJ/mol, Log Ac: 11.9 log(min ); (B) Waterloo model [59] (C) -1 Varhegyi-Antal model [60], Ev: 238 kJ/mol Log Ai: 16.3 log(min ); Ec: 147 -1 -1 kJ/mol, Log Ac: 7.7 log(min ); Ed: 174 kJ/mol, Log Ad: 11.0 log(min ); Ee: -1 250 kJ/mol, Log Ae: 15.9 log(min ). Redrawn from [62] ...... 10

Figure 1.6 A scheme of intra-molecular dehydration reaction at 220 oC [70] ...... 11

Figure 1.7 Inter-molecular dehydration pathway proposed by Kilzer and Broido [70] ...... 12

Figure 1.8 Model of cellulose thermal degradation with high boiling point tar trapped in reacted space [73]. NR: non-reducing end; R: reducing end; LG: levoglucosan end...... 13

Figure 1.9 Cross-linking reactions caused by pretreatment and product from 670 oC pyrolysis [68] ...... 15

Figure 1.10 Cellulose fibril shrinking scheme during pyrolysis [80] ...... 16

Figure 1.11 Scheme of double internal nucleophilic attack by hydroxyl group [82] ...... 17

Figure 1.12 Scheme of radical-chain decomposition of cellulose [71] ...... 18

Figure 1.13 Concerted unzipping mechanism proposed by Mayes and Broadbelt [83] ...... 20

Figure 1.14 Levoglucosan fragmentation scheme proposed by Shafizadeh [84] ...... 22

Figure 1.15 Mechanism of glycolaldehyde direct cleavage from cellulose chain [85] ...... 22

Figure 1.16 Scheme of carbonium ion mechanism of levoglucosenone polymerization [82] ...... 24

Figure 1.17 Scheme of cellulose pyrolysis mechanism [86]...... 25

Figure 1.18 Scheme of the formation of ion m/z=191 in the presence of KOH [94]...... 28

Figure 1.19 Scheme of MgCl2 catalyzed chain scission (a) and dehydration (b) [98]...... 29

Figure 1.20 Interaction of cellulose and lignin pyrolytic products [111] ...... 32

Figure 1.21 Scheme of dissertation ...... 36

Chapter 2

Figure 2.1 XRD of Avicel with ball-milling time from 0 (bottom) to 48 h (top) ...... 56

Figure 2.2 Cellulose crystallinity (determined by XRD) as a function of ball-milling time (top) ...... 57

Figure 2.3 FTIR spectra of cellulose after ball-milling (0-48 h)...... 58

Figure 2.4 Crystallinity indexes obtained by FTIR spectroscopy (1372/2899) versus crystallinity obtained by XRD ...... 58

Figure 2.5 Scanning electron micrographs of control (left) and ball-milled cellulose (right)...... 59

Figure 2.6 TGA thermograms of control cellulose and ball-milled (24h) cellulose...... 60

Figure 2.7 DTG thermograms at different heating rates of control cellulose (left) and after ball-milling (24 h) (right)...... 61

Figure 2.8 Graphs showing Ea of control cellulose (left) and ball-milled (24 h) cellulose (right) as a function of conversion obtained by ASTM method...... 62

Figure 2.9 Scanning electron micrographs of TGA derived char for control (left) and ball- milled (right) cellulose. From top to bottom: unheated cellulose, residue full view, 500X, and 5000X...... 64

Figure 2.10 Total ion chromatograms of pyrolyzed control cellulose (left) and ball-milled cellulose (right) at 300 and 400 oC...... 66

Figure 2.11 Effect of cellulose pyrolysis temperature on the yield of levoglucosan, levoglucosenone, 1,4:3,6-dianhydro-glucopyranose, and total sugars...... 68

xii

Figure 2.12 Effect of cellulose pyrolysis temperature on the yield of carbon dioxide...... 69

Figure 2.13 Effect of pyrolysis temperature on the yield of 5-hydroxymethyl furfural, 5- methyl-furfural, and furfural...... 70

Figure 2.14 Effect of pyrolysis temperature on the yield of 2(5H)-furanone, 2(3H)- furanone, 2-propyl furan and furan...... 71

Figure 2.15 Effect of pyrolysis temperature on the yield of 3-methyl-1,2- cyclopentanedione, 2-hydroxy-3-methyl-2-cylcopenten-1-one, 2-hydroxy-2- cyclopenten-1-one, and phenol...... 72

Figure 2.16 Effect of pyrolysis temperature on the yield of A (light compounds) and B (acetol)...... 73

Chapter 3

Figure 3.1 Mesh reactor illustration and sample loading pattern...... 89

Figure 3.2 The working status of fast speed camera...... 90

Figure 3.3 Reaction time vs. effective diameter of control and ball-milled cellulose at temperatures between 300 and 500 oC...... 93

Figure 3.4 Reaction time vs. temperature of control and ball-milled cellulose ...... 93

Figure 3.5 Reaction rate vs. 1/T ...... 94

Figure 3.6 Residue of control (left) and ball-milled (right) cellulose under SEM...... 97

Figure 3.7 HPLC chromatograms of products from mesh reactor, in terms of heating temperature (results normalized, Control and Ball-milled cellulose). From left to right: cellobiosan, unknown compounds, levoglucosan...... 99

Figure 3.8. Chromatogram of control cellulose products from pyrolysis at 500 oC ...... 100

Figure 3.9 Yield of levoglucosan (by HPLC and GC/MS) and unknown compound(s) (by HPLC) from pyrolysis of control cellulose and ball-milled cellulose ...... 101

Figure 3.10 GC chromatogram of (A) levoglucosan and (B) cellobiosan pyrolysis products at 300 oC ...... 102

Figure 3.11 Mass fragmentation pattern of the second unknown peak in cellobiosan pyrolysis products ...... 102

xiii

Figure 3.12. HPLC chromatogram of products from pyrolysis of levoglucosan and cellobiosan ...... 103

Figure 3.13 Yield of levoglucosan (by HPLC and GC/MS) and unknown compound(s) (by HPLC) of cellobiosan, with area/mass of products from cellobiosan (by GC/MS) ...... 104

Figure 3.14 SEM pictures of mesh exhibit a thin layer of carbonaceous residue after cellobiosan pyrolysis (contrast and brightness adjusted) ...... 106

Chapter 4

Figure 4.1 Ash content of Douglas fir particle, after ball-milling and after ball-milling and washing ...... 118

Figure 4.2 Residence time of HPLC of standards. As marked, 1: cellobiosan (CL), 2: 1,6- anhydro-β-D-glucofuranose (AGF), 3: 1,4;3,6-dianhydroglucopyranose (DGP), 4: levoglucosenone (LVS) (or its hydrolyzed product in water), 5: levoglucosan (LVG)...... 120

Figure 4.3 Effect of temperature on the yield of levoglucosan from Douglas fir, Avicel and Ball-Milled Avicel pyrolysis by mesh reactor (pyrolysis result of Avicel and ball-milled Avicel were adapted from our previous work [4])...... 122

Figure 4.4 GC/MS of Douglas fir Pyrolysate at different sulfuric acid concentrations. Where: AGF, DGP, LVS (or its hydrolyzed product in water)...... 124

Figure 4.5 HPLC of Douglas fir with different sulfuric acid concentration impregnation ...... 125

Figure 4.6 Effect of sulfuric acid concentration on the yield of levoglucosan (by Douglas fir and cellulose) and other products from Douglas fir pyrolysis at 500 and 300 oC by mesh reactor (detected by HPLC) ...... 127

Figure 4.7 Effect of sulfuric acid concentration on the yield of dehydrated sugar products (by Douglas fir) and char yield from Douglas fir pyrolysis at 500 and 300 oC by mesh reactor (detected by GC/MS)...... 127

Figure 4.8 HPLC of cellulose and levoglucosan pyrolysate (left) and the evaluation of their yields (right) ...... 129

Figure 4.9 Effect of sulfuric acid concentration on the yield of residue and dehydrated products (by GC/MS) from Avicel, ball-milled Avicel, and levoglucosan pyrolysis at 500 oC on mesh reactor ...... 131

Figure 4.10 HPLC of unpyrolyzed cellobiosan samples with sulfuric acid (in mass %) ...... 133

xiv

Figure 4.11 GC/MS of unpyrolyzed cellobiosan samples with sulfuric acid ...... 133

Figure 4.12 HPLC of cellobiosan pyrolysate (left) and the evaluation of their yields (right) ...... 134

Figure 4.13 GC/MS chromatograph of cellobiosan with sulfuric acid pyrolysate ...... 135

Chapter 5

Figure 5.1 Residue yields from a spoon reactor at different temperatures and reaction time ...... 152

Figure 5.2 Graph of cellulose conversion rate versus (1/reaction temperature) to determine Ea for control and ball-milled cellulose...... 154

Figure 5.3 A comparison of control cellulose (left) and ball-milled cellulose (right) (a), and after pyrolysis for 120 min at 240 oC (b) and 280 oC (c) and for 60 min at 400 oC (d). The melted char of ball-milled cellulose was crushed for sampling...... 156

Figure 5.4 Graph of cellulose residue recovered after pyrolysis as a function of time for (left) control cellulose and (right) ball-milled cellulose...... 157

Figure 5.5 Yield of non-hydrolysable material (charcoal + cross-linked saccharides) from cellulose ((left) control and (right) ball-milled cellulose) after pyrolysis as a function of reaction time...... 159

Figure 5.6 FTIR spectra of pyrolyzed ball-milled cellulose for 30, 60, or 120 min at 240, 260, 280, 300 and 320 oC...... 160

Figure 5.7 Crystalline cellulose/cellulose ratio (I1420/I899) from FTIR for (left) control and (right) ball-milled cellulose at differing temperatures as a function of reaction time...... 162

Figure 5.8 Crystalline cellulose/biomass ratio (I1376/I2900) from FTIR for (left) control and (right) ball-milled cellulose at differing temperatures as a function of reaction time...... 162

Figure 5.9 C=O/C-O ratio from FTIR for (left) control and (right) ball-milled cellulose at differing temperatures as a function of reaction time...... 163

Figure 5.10 The C=C/C-O ratio from FTIR for (left) control and (right) ball-milled cellulose at differing temperatures as a function of reaction time...... 164

Figure 5.11 The OH/C-O ratio from FTIR (I3100-3600/I990-1070) for (left) control and (right) ball-milled cellulose at differing temperatures as a function of reaction time...... 165

Figure 5.12 13C-NMR spectra of control and ball-milled cellulose ...... 166

Figure 5.13 13C NMR Spectra for pyrolyzed control cellulose at differing conditions ...... 167

Figure 5.14 Graph showing the yield of carbohydrate, aliphatic and aromatic groups for (left) control and (right) ball-milled cellulose at differing pyrolysis temperatures as a function of reaction time...... 169

Figure 5.15 Comparison of cellulose content determined by hydrolysis and 13C-NMR...... 170

Figure 5.16 The yield of furanyl and carbonyl groups for (left) control and (right) ball- milled cellulose at differing pyrolysis temperatures as a function of reaction time ...... 171

Figure 5.17 Behavior of non-hydrolysable cross-linked saccharides ...... 172

Figure 5.18 Plots of Ln k vs. 1/T for the reactions 1-6 for the pyrolysis of cellulose ...... 179

Figure 5.19 Graphs showing calculated vs. experimental values of yield of fractions for (left) control and (right) ball-milled cellulose at low pyrolysis temperatures as a function of reaction time.( A, D, F, D+F are experimental results, A’, D’, F’, D’+F’ are calculated) ...... 183

Figure 5.20 Calculated vs. experimental values of yield of fractions for (left) control and (right) ball-milled cellulose at high pyrolysis temperatures as a function of reaction time. (A+B, E, H, J are experimental results, A’+B’, E’, H’, J’ are calculated) ...... 184

Chapter 6

Figure 6.1 Scheme of cellulose thermal reactions under slow pyrolysis condition ...... 194

Figure 6.2 Scheme of cellulose thermal reactions under fast pyrolysis condition ...... 195

Figure 6.3 Scheme of cellulose thermal reactions on the existence of sulfuric under fast pyrolysis condition ...... 196

Appendix A

Figure A.1 TG and DTG of crystalline (CC, left) and amorphous (AC, right) cellulose, lignin and cellulose lignin blend with heating rate of 10 oC/min. Dotted line is the prediction of TG and DTG by their pure compound...... 212

Figure A.2 Images of residue obtained from samples TGA analysis...... 214

xvi

Figure A.3 Char yield of cellulose-lignin blend with their predicted yield at heating rate of 50 and 10 oC/min in TGA...... 215

Figure A.4 SEM of residues from original compounds (from left to right: crystalline cellulose, amorphous cellulose, and organosolv lignin) ...... 218

Figure A.5 SEM images of 20% organosolv lignin (OL) with 80% crystalline cellulose (CC, left) and amorphous cellulose (AC, right) ...... 219

Figure A.6 SEM images of 50% organosolv lignin (OL) with 50% crystalline cellulose (CC, left) and amorphous cellulose (AC, right) ...... 220

Figure A.7 SEM images of 80% organosolv lignin (OL) with 20% crystalline cellulose (CC, left) and amorphous cellulose (AC, right) ...... 221

Figure A.8 Py-GC/MS of crystalline cellulose/organosolv lignin blend ( 50:50 wt. %) at 500 oC. Compounds were identified by NIST 08 library (See Table A.1) ...... 222

Figure A.9 Effect of lignin blend on yield of levoglucosan (LVG) and 1, 6-anhydro-α-D- glucofuranose (AGF) from crystalline cellulose (CC) and amorphous cellulose (AC) ...... 225

Figure A.10 Effect of lignin content on the yields of levoglucosenone and 1,4;3,6- dianhydro-α-D-glucopyranose at 350 and 500 oC ...... 226

Figure A.11 Effect of lignin blend on yield of 5-hydroxymethyl-2-furancarboxyaldehyde (5-HMF), furfural and 5-methylfurfural (MEF) at 350 and 500 oC ...... 227

Figure A.12 Effect of lignin blend on yield of 2-furanone and 2-propylfuran at 350 and 500 oC...... 228

Figure A.13 Effect of lignin blend on yield of 2-hydroxy-2-cyclopenten-1-one, methyl- cyclopentanedione, and 2-hydroxy-3-methyl-2-cyclopenten-1-one at 350 and 500 oC...... 229

Figure A.14 Effect of cellulose blend on yield of phenol, 2-methoxyphenol, 2-methoxy-4- methylphenol, and 3-methoxy-1,2-dibenzediol at 350 and 500 oC ...... 231

Figure A.15 Effect of cellulose blend on yield of 2-methoxy-4-vinylphenol, 2,6- dimethoxy-methylphenol, 3,4-dimethoxyphenol and 3,5- demethoxyacetophenone at 350 and 500 oC ...... 232

Figure A.16 Effect of cellulose blend on yield of another four compounds from lignin at 350 and 500 oC ...... 233

Figure A.17 Compounds not enhanced by cellulose existence...... 234

xvii

Figure A.18 Effect of lignin blend on the yield of 2-oxo-propanoic acid, hydroxyl- acetaldehyde, and acetol at 500 oC ...... 235

Figure A.19 Effect of lignin-cellulose interaction on yield of carbon dioxide and acetone...... 236

xviii

Dedication

This dissertation is dedicated to my wife who has been waiting and standing behind me from her

youth. It is also dedicated to my parents who provided both emotional and financial support.

xix

Chapter 1 Introduction

Before the industrial production of liquid transportation fuels from petroleum, which resulted from the development and deployment of the crude oil industry in the 19th century, most consumer energy and chemicals were obtained from plant biomass [1]. With increasing petroleum prices and the gradual depletion of fossil fuel resources, biomass is regaining interest for its potential to produce green liquid fuel and highly valued chemicals. With proper processes, bio-fuels and bio-chemicals derived from biomass could result in less carbon dioxide emissions than their equivalents derived from fossil resources [1-3].

Figure 1.1 shows a biofuel production process concept [1]. As the biomass grows by photosynthesis, it fixes carbon from the atmosphere which can then be harnessed utilized in a bio-fuel intermediate by thermal or bio-chemical conversion. The intermediates (bio-oil, syngas, and sugars) can be further upgraded to produce transportation fuel. The carbon dioxide released during the combustion of this fuel can be fixed again as the biomass grows. The whole cycle could be carbon neutral if no fossil fuel is used during transportation and processing.

Recycle Fuel Energy CO2, H2O Utilization

Recycle Fuel CO2, H2O, Energy Input Production Nutrients

CO2, H2O, Biomass Light, Energy, Transport Separation Growth Nutrients

Nutrients Recycle

Food and Feed

Figure 1.1 Biofuel production from biomass in a sustainable way [1]

In general the biomass conversion technologies can be classified in: (1) Chemical, (2)

Biochemical, and (3) Thermo-chemical. Among all the thermo-chemical processes, fast pyrolysis, with processing temperatures between 300 and 500 oC in the absence of oxygen and with short vapor residence times (around 2 s), is receiving significant attention in the United States [4-6].

With this technology it is possible to achieve bio-oil yields as high as 75 wt. % [7]. However, the design of the reactor, operating conditions and composition of the feedstock can significantly affect the yield and composition of the resulting oils and is a constraining factor for the technology’s advancement [4, 8].

1.1 Multiscale Structure of Lignocellulosic Materials

Figure 1.1 shows structure of plant biomass [9] at five different levels: whole plant, plant cell, macrofibril (cell wall), and major constitutions of microfibril. Cellulose microfibril provides the major structure of primary and secondary cell walls. Hemicellulose connects cellulose with tethers to form cross-linked network embed in pectin matrix. Lignin fills the space between cellulose, hemicellulose and pectin to provide waterproofing and mechanical strength to plant cell wall structure [10].

Figure 1.2 Structure of lignocellulose [9]

1.2 Cellulose

Figure 1.3 shows the chemical structure of cellulose [11], which is composed of β-D- glucopyranose units linked by 1, 4 - glycosidic bonds. Cellulose is mostly found in secondary cell wall and typically account for 40-45 % of the dry wood mass. The molecules of cellulose are linear and have high potential to form both intermolecular and intramolecular hydrogen bond

(Figure 1.3). The number of glucose units in one cellulose chain is called degree of polymerization (DP) which can vary significantly between materials (example, cotton: 15,000, and wood: 10,000) [10].

Figure 1.3 Cellulose structure with hydrogen bonds. Redrawn from [11]

In plants, cellulose forms microfibril with both highly ordered (crystalline) and less ordered

(amorphous) regions (See Figure 1.4) [12]. Six polymorph structures (I, II, IIII, IIIII, IVI, and IVII) of cellulose have been discovered and documented [13]. The chains of native cellulose (cellulose

I) parallel each other and the hydrogen bonds (O (6) to adjacent O (3) intermolecular) hold them

into a layer [10]. Evidence of two polymorphs of cellulose has been revealed [14-16]. Depending on the crystalline structure native cellulose can be classified in two types: I and I. Iα has a triclinic structure and Iβ has a monoclinic structure [15, 17] Cellulose Iβ, which is found mostly in higher plants, has more intermolecular hydrogen bonds than cellulose Iα (found mostly in primitive organisms) [10].

Amorphous Crystalline

Figure 1.4 Microcrystalline cellulose model with crystalline and amorphous structure. Redraw from [12]

1.3 Cellulose Characterization

1.3.1 Crystallinity

Native cellulose has both crystalline and amorphous structure. The definition of crystallinity is the weight fraction of crystalline material in the sample [13]. The most common methods for crystallinity determination are X-ray diffraction (XRD) [18-28], Carbon-13 Cross Polarization /

Magic Angle Sample Spinning (13C CP/MAS) [14, 17, 18, 22, 27, 29-33], and FTIR [19, 22, 34-

43].

All materials can exhibit peaks in X-ray scattering. In contrast with the crystalline cellulose which generates a specific sharp peak, the amorphous fractions, (amorphous cellulose, hemicellulose, and lignin) will give broad peaks [39].

The three most common methods used in XRD crystallinity determination are as follows. (1) The

o Segal Method [19, 44] compares the height of peak at 200 (I200, 2q = 22.7 ) and the minimum

o between 200 and 110 peaks (IAM, 2q = 18 ). (2) The Ruland-Vonk Method [28, 45] can separate the amorphous and crystalline cellulose by measuring compounds in amorphous form and scale it by a factor. (3) Rietveld refinement [46] uses the full diffraction pattern in a least-squares fitting procudure to concurrently fit the diffraction curve with elements. Of the three, Rietveld refinement is the most commnly used method for powder diffraction structural determination

[47].

The solid state 13C NMR spectra of native cellulose shows a single peak at ~106 ppm of C1, a splitted peak at 95-80ppm of C4, a splitted peak at 80-68 of C2, 3 and 5, and a splitted peak at

68-56 ppm of C6 [48]. Usually, a C4 seperation method is utilized to describe the CrI [30, 49, 50] by dividing the peak into an ordered region (~89 ppm, x) and a disordered region (~83 ppm, y).

CrI is calculated by dividing x by (x+y) [27].

Infrared spectroscopy (usually combined with Fourier transform, FTIR) utilizes an infrared source to get the vibration mode of the sample. Instead of giving a direct crystallinity value,

FTIR gives a relative value called crystallinity index (CrI). In 1964, Nelson and O’Connor [38] reported a prevention method with cellulose II which determines the crystallinity by measuring the peaks at 1420, 893-897, and 1111 cm-1. They later provided a new infrared ratio by

- -1 measuring I1372 cm 1 /I2900 cm [39]. FT-IR spectroscopy has been recently successfully used to determine the relative cellulose Iα content by modeling a system of two different cellulose [51].

1.3.2 Sugar Content

Measurement of sugar content can give the amount of glucose units contained in the form of poly-saccharides. According to the ASTM method [52], cellulose samples need hydrolysis before sugar content measuring. The hydrolysis procedure starts with a one hour 72% sulfuric acid treatment at 30 oC. Further hydrolysis is typically carried out at 125 oC (15 psi) in an autoclave for one hour. Liquid samples after hydrolysis are diluted, filtered, and analyzed by a liquid chromatography technique (typically ion-exchange).

1.3.3 Degree of Polymerization (DP)

The DP of cellulose is the number of glucopyranose units in cellulose chains, which multiplied by the molecular weight of glucose can be used to quantify the molecular weight of cellulose chains. Typical methods to determine this value are by viscosimetry in solvents [53, 54], or by size exclusion chromatography (SEC) after derivatization [55-57]. In the first method, cellulose samples are derivatized (mostly nitrated) and dispersed in solution (e.g. ethyl acetate). Then the

viscosity of the solution measured. Solvent and aggregation can affect the result [53]. In the case of the SEC method, cellulose samples are derivatized (typically carbanilated, or acetylated), and then dissolved in a solvent (e.g. THF, chloroform, etc.), which can be analyzed by SEC. The analysis is carried out with reflective index detector or differential light scattering detector and calibrated by polystyrene standards.

1.4 Cellulose Thermochemical Reactions

1.4.1 Lumped Mechanisms of Cellulose Thermochemical Reactions

There are many schemes reported in the literature to describe cellulose pyrolysis reactions [58-

61]. The Broido-Shafizadeh model [58], Waterloo model [59], and Varhegyi-Antal model [60] are among the most cited mechanisms (Figure 1.5 A, B, and C [58-60, 62]). Although all three models agree that the formation of anhydrosugars through depolymerization is competing with char formation reactions, difference could still be found among three mechanisms. Broido-

Shafizadeh model and Waterloo model stressed on the importance of active cellulose (low DP cellulose) as a key controlling intermediate. The “active cellulose” concept is based on the observation that after heating a cellulose sample at low temperature (150-200 oC), and slow heating rate, cellulose’s degree of polymerization decreases until reaching a constant value. The

Varhegyi-Antal model does not consider active cellulose as a controlling intermediate [60].

Although the original Broido model contains char evolution pathway, which was also adapted by

Varhegyi-Antal model, little attention was put into the reactions leading to the formation of char

[63].

Other authors highlight the importance of levoglucosan (a major product of cellulose pyrolysis)

[64-67]. According to Lin et al., levoglucosan is converted under the pathway from levoglucosan, levoglucosenone, further rearrangement to furans and then fragmentation, to the final polymerization to char [64]. Brown’s group on the other hand proposed the idea that levoglucosan could react to form sugars with higher DP (e.g. cellobiosan, cellotriosan, etc.) [65-

67].

Chaiwat et al. [68, 69] highlighted the importance of the formation of cross-linked sugars as a needed intermediate for char production, which can be enhanced by hot water treatment.

Figure 1.5 Cellulose thermal reaction models: (A) Broido-Shafizadeh model [58], Ei: 242.7 -1 -1 kJ/mol Log Ai: 21.2 log(min ); Ev: 197.9 kJ/mol, Log Av: 16.3 log(min ); Ec: 153.1 kJ/mol, Log -1 Ac: 11.9 log(min ); (B) Waterloo model [59] (C) Varhegyi-Antal model [60], Ev: 238 kJ/mol -1 -1 Log Ai: 16.3 log(min ); Ec: 147 kJ/mol, Log Ac: 7.7 log(min ); Ed: 174 kJ/mol, Log Ad: 11.0 -1 -1 log(min ); Ee: 250 kJ/mol, Log Ae: 15.9 log(min ). Redrawn from [62]

Formation of low DP active cellulose

Native cellulose is sensitive to heat and mechanical treatment. It has been observed that DP

reached 200-400 before noticeable weight loss is observed. This is due to the cleavage of very

weak glycosidic bond (usually in the amorphous section) of cellulose. This cellulose of low degree of polymerization is called in the literature as “active cellulose” [82].

Dehydration Reactions

When cellulose is heated at 220 oC, it is possible to observe a decrease in FTIR peak intensity at

990 cm-1 (C-O skeletal vibration) and two new peaks indicating C=C and C=O stretching at 1620 and 1700 cm-1 are typically observed [70]. Figure 1.6 shows the scheme of this intra-molecular dehydration. In this reaction mechanism, dehydration reaction starts from the C3 hydroxyl group.

Figure 1.6 A scheme of intra-molecular dehydration reaction at 220 oC [70]

Figure 1.7 [70] shows a pathway of inter-molecular dehydration during cellulose thermal treatment. This reaction was first proposed by Klizer and Broido [70] and is in part responsible for the formation of cross-linked structures and furanic ends in cellulose chain. This reaction is also believed to be the dominant dehydration reaction over intra-molecular dehydration. The double bond formation cannot explain the large quantity of water formed during pyrolysis [70].

Figure 1.7 Inter-molecular dehydration pathway proposed by Kilzer and Broido [70]

Formation of a liquid intermediate

During cellulose pyrolysis, monomeric and oligomeric products of cellulose may form a liquid intermediate [71]. Haas et al. [72] showed the presence of liquid intermediates on light micrographs and transmission electron micrographs (Figure 1.8). The authors found that this liquid intermediate is a major precursor for the formation of bio-char. The importance of this liquid intermediate was highlighted by Mamleev [73, 74]. The authors argue that dehydration and β-elimination, especially the elimination reaction needs an acidic catalytical environment which cannot exist in or introduced into cellulose solid polymer during the pyrolysis, but can be built in high boiling point tar formed by transglycosylation [73].

Figure 1.8 Model of cellulose thermal degradation with high boiling point tar trapped in reacted space [73]. NR: non-reducing end; R: reducing end; LG: levoglucosan end.

Dauenhauer et al [75-77] has found that the formation of liquid intermediates is critical for the thermal ejection of heavy anhydrosugars produced from cellulose primary and secondary pyrolysis reactions, where the temperature of reaction were not high enough for these sugar dimers and oligomers to evaporate.

It was not possible to find any research reporting how cellulose crystallinity affects the formation of liquid intermediates during pyrolysis under slow and fast heating rate conditions.

Cross-linking Reactions

In 1992, Mok et al. [78] reported the catalyzing effect of water on char production with heat release during cellulose pyrolysis. Following this work, Richard and Antal [79], in 1994, reported an increase of char by reducing the purge gas flow. Chaiwat et al. conducted studies to better understand the cross-linking reactions [68]. The authors introduced a new index relating the degree of cross-linking with the glucose produced from sample hydrolysis (Equation 1).

(1)

Results indicate that cross-linking reactions result in the loss of hydroxyl groups (inter-molecular dehydration) with water treatment at 200-260 oC and low tar yield in later flash pyrolysis [68].

The cross-linking reactions were also found to be suppressed when temperature rapidly (3000

K∙min-1) surpassed 360 oC [69], or when treated with phosphoric acid [68] (Figure 1.9).

Figure 1.9 Cross-linking reactions caused by pretreatment and product from 670 oC pyrolysis [68]

It was not possible to find any study on the effect of cellulose crystallinity on cross-linked reactions. None of the models so far proposed in the literature takes into account the formation of cross-linked structures as a critical step for charcoal formation (see models listed in section

1.4.1).

1.4.2 Cellulose thermal reactions at high temperatures (above 300 oC)

Unzipping

The depolymerization reactions of cellulose fibrils do not occur at once. Figure 1.10 shows the size reduction during cellulose thermal reactions [80], where cellulose microfibrils peels in width first and then in length. This means this reaction evolves gradually with certain order, the structure of cellulose “unzips”.

Figure 1.10 Cellulose fibril shrinking scheme during pyrolysis [80]

Pyrolysis of cellulose usually gives a high yield of levoglucosan, which has the exact molecular weight of one cellulose glucopyranose unit. The production of levoglucosan was once believed to react through a glucose intermediate, but this was disproven when experimental results showed that yield of levoglucosan below 420 oC from glucose pyrolysis was much lower than that from cellulose [81, 82]. Figure 1.11 shows one of the pathways suggested in the literature. It

is described as a combination of “internal chain cleavage and end-terminal cleavage of a single unit” [82].

Figure 1.11 Scheme of double internal nucleophilic attack by hydroxyl group [82]

In this mechanism, the cellulose chain needs a hydrolysis step with water assisting to release the non-reducing levoglucosan end unit. However, the dehydration product of water is known to be a catalyst for cross-linking reactions that leads to more char or cross-linked product instead of more levoglucosan; char production is typically low under fast pyrolysis, so this mechanism is likely not predominant [68]. Since the yield of levoglucosan can be enhanced with a high heating rate to high temperature (>360 oC), this mechanism could be a source of levoglucosan in the low temperature region, where both intra- and inter-molecular dehydration are still very active.

Golova suggested another radical mechanism called “unzipping”, which starts from an active end of cellulose chain and consequently splitting off glucopyranose units to form levoglucosan until the whole chain is decomposed [71]. Figure 1.12 shows this mechanism. In this scheme, an active end at C4 produced by random cleavage was transferred to C6, which later weakened the glycosidic bond on this unit and thus tear this unit off in the form of levoglucosan and transfer the active end to the adjunct glucopyranose unit.

Figure 1.12 Scheme of radical-chain decomposition of cellulose [71]

A recent molecular modeling study found that a concerned unimolecular mechanism for cellulose pyrolysis is favored over the ionic and free radical mechanisms [83]. This mechanism proposes the production of levoglucosan via an initiation step and a depropagation step typically

described in free-radical reactions (Figure 1.13). The H6 from an approaching glucose unit is attracted to O1 on the same glucose unit, and then transferred to the C4 of adjacent glucose unit together with O1 as a hydroxyl group; O6 connects to C1 then a non-reducing end is formed and the initiation step is finished. Following this step, same procedure happens on the glucose unit next to the non-reducing end and releases the non-reducing end as a levoglucosan. None of the depolymerization mechanisms reported in the literature explains the formation of heavier anhydrosugars found during cellulose pyrolysis: cellobiosan, cellotriosan and heavier anhydrosugars.

Figure 1.13 Concerted unzipping mechanism proposed by Mayes and Broadbelt [83]

Frangmentation

Although levoglucosan is the major product in cellulose pyrolysis, under certain circumstances smaller compounds can also be formed. Glycolaldehyde, acetol and acetic acid are commonly found in pyrolysis liquids derived from lignocellulosic materials. The reactions producing these small compounds are called fragmentation reactions, which means they are the result of the cleave of cellulose or levoglucosan carbon-carbon bonds to produce compounds with less than 5

carbons. These reactions compete with levoglucosan production (depolymerization) and lead to a lower yield of levoglucosan.

Shafizadeh and Stevenson in 1982 [84] proposed a mechanism (Figure 1.14) explaining the formation of fragmentation products from levoglucosan. The authors argued that the levoglucosan was first hydrolyzed to glycopyranose and then fragmented into a four carbon unit with glycolaldehyde production. Further reaction of the four carbon unit resulted in carbon monoxide, acetol, acetaldehyde and other small compounds. Richards [85] later suggested a different pathway (Figure 1.15), where glycolaldehyde was cleaved directly from glycopyranose units in cellulose chain instead of free levoglucosan units.

Figure 1.14 Levoglucosan fragmentation scheme proposed by Shafizadeh [84]

Figure 1.15 Mechanism of glycolaldehyde direct cleavage from cellulose chain [85]

It was not possible to find any study on fragmentation reactions of the heavy products of cellulose pyrolysis reactions (cellobiosan, cellotriosan, etc.)

Polycondensation

Any organic chemical compound subjected to thermal treatment for a long period of time will result in the formation of aromatic compounds [82]. Products of cellulose pyrolysis are highly reactive and tend to further react by either fragmenting into small compounds, or growing in size to form carbonous residues. This polymerization like reaction is usually called polycondensation.

Halpem et al. [76] proposed a carbonium ion mechanism for levoglucosan pyrolysis. This mechansim was further adapted into a complex polymerization of levoglucosenone (Figure 1.16) as one of the pathway to polycondensed aromatic residue [82].

Figure 1.16 Scheme of carbonium ion mechanism of levoglucosenone polymerization [82]

Similar to this mechanism, Kawomoto et al. [86], proposed a mechanism of char production following the route presented in Figure 1.17. They found carbonization reactions gradually transformed levoglucosan to methanol and water soluble residue, and eventually into a fraction totally insoluble to water and methanol. Combined with Figure 1.16, it is logical to conclud that further dehydration of levoglucosan leads to the polycondensation reactions of products and eventually to solid residue (char).

Figure 1.17 Scheme of cellulose pyrolysis mechanism [86].

It was not possible to find any study describing the direct polymerization and polycondensation of the heavy products of cellulose pyrolysis (cellobiosan, cellotriosan, etc.)

1.4.3 The effect of crystallinity on cellulose thermal reactions

The effect of cellulose crystallinity (CrI) on cellulose thermal reactions has been poorly studied.

It was early found that CrI remained constant until 50 wt. % of weight loss [87]. The thermal treatment at early stage of pyrolysis can restore some crystallinity in amorphous cellulose [87-

89]. Matthews et al. [90] found the formation of inter-layer hydrogen bonds not found in native cellulose after thermal treatment. This inter-layer bond can potentially increase the crystallinity of cellulose during the early stage of pyrolysis. Longer heating (30 minutes), on the other hand, turned cellulose (Avicel) into a totally disordered structure (most probably dehydrated) at 300 oC

[62]. This disordered structure was found to form through a crystalline intermolecular cross- linking reaction [87].

The change of DP as a function of thermal treatment is also affected by cellulose crystallinity

[88]. In the case of crystalline cellulose, the DP decreases rapidly without any measurable weight loss change till reaching a DP close to 400. This crystalline cellulose with low DP is often called in the literature as “active cellulose”. Amorphous cellulose, on the other hand gradually decreased its DP till mono or oligo-anhydrosugars are formed. Yu and Wu reported that amorphous part was more reactive towards oligomers and more accessible during the hot compressed water hydrolysis due to low hydrogen bond concentration [91].

Although there is some information on the effect of cellulose crystallinity on cellulose thermal behavior, the influence of this parameter on the formation of liquid intermediates and the product distribution of pyrolysis reactions is very poorly understood.

1.4.4. Interactions between cellulose and other compounds

Due to the native complexity of biomass, it is hard to imagine an undisturbed environment for cellulose pyrolysis. As discussed earlier (see Cross-linking Reactions under 1.4.1), the water evolved from dehydration reactions of lignin, hemicellulose, or cellulose can catalyze the cross- linking reactions and thus reduce the yield of levoglucosan from cellulose pyrolysis. In this section the known interactions between cellulose salts, acids, and lignin are discussed.

Cellulose-alkalines interactions

Generally, the yield of char and gases is found to increase with the addition of inorganic salt [82,

92]. For example, the addition of 1 % of CH3COOK, tripled char yield of both cellulose and levoglucosan [93]. Use of 0.1% potassium hydroxide is reported to increase the production of an ion of m/z = 191 with two C1s [94] by the mechanism shown in Figure 1.18.

Figure 1.18 Scheme of the formation of ion m/z=191 in the presence of KOH [94].

Both alkali metal chlorides and alkaline earth metal chlorides were reported to reduce the yield of levoglucosan and promote production of smaller molecules [95, 96] alkali metal chlorides promoted the formation of glycolaldehyde and acetol, which are products from fragmentation

reactions; alkaline earth metal chlorides promoted the dehydration in low temperature and the formation of furfural [95-97].

Figure 1.19 shows the effect of Mg2+ on the chain scission and dehydration. The Mg2+ is understood to couple to an oxygen atom and cause this oxygen to peel off from the carbon, leaving a positive charge at that position. Then adjacent hydrogen or H6 takes that charge, transfers it to the Mg+OR (R can be H or another glucose ring) system, and releases Mg2+ with formation of HOR. The chain losing HOR then either becomes a non-reducing end or a dehydrated glucose ring.

Figure 1.19 Scheme of MgCl2 catalyzed chain scission (a) and dehydration (b) [98].

Cellulose-acids interactions

The effect of acid on cellulose thermal reactions has been also investigated [99-101]. Tsuchiya and Sumi first observed the decrease in the yield of levoglucosan and the formation of a new compound when pyrolyzing cellulose with sulfuric acid [99]. Halpern et al. later identified this compound as levoglucosenone [100]. This effect was found to be caused by the aceleration of dehydration reactions (associated to the pH and binding of sulfate) when under slow pyrolysis condition (10 oC/min to 380 oC) [101], and resulted in a high yield of water, char, at the expense of levoglucosan.

A series of experiments were done by European groups showing that yield of levoglucosenone could be enhanced by phosphoric acid impregnating under both fast and slow pyrolysis conditions [102-105]. At slow pyrolysis conditions, levoglucosan yield was inhibited by phosphoric acid addition under slow heating rate conditions [102] but was increased under fast pyrolysis condition [102-105]. Phosphoric acid catalyzed the esterification of cellulose hydroxyl groups and dehydration reactions [102]. High acid concentration of sulfuric acid promotes levoglucosanone yield while suppressing levoglucosan formation [102-105]. Similar results were found by Japanese researchers [106], who also reported that the levoglucosenone could further react into furfural and 5-HMF in the presence of water.

Shafizadeh and Stevenson [84] reported an increase of levoglucosan yield in wood pyrolysis with additional sulfuric acid, which was not found in cellulose fiber and holocellulose alone.

Similar result was also reported when poplar wood was treated with 0.1% of sulfuric acid [107].

This effect could be attributed neither to the impact of the acid on cellulose pyrolysis nor to a simple cleavage of lignin-carbonhydrate bonds [84].

Cellulose-lignin interactions

Cellulose-lignin interactions are not well known. Studies on levoglucosan pyrolysis in the presence of lignin have found that lignin generally reduces the recovery of levoglucosan at ~250 oC [108, 109].

On the contrary, in certain aromatic structures levoglucosan was stablized and had high solubility

[110]. Further research on interactions between milled wood lignin and cellulose revealed that lignin could suppress the re-polymerization of levoglucosan at 800 oC [111]. This effectively reduced the char formation from both cellulose and lignin, and changed the composition of pyrolysis char towards more levoglucosan, small molecule compounds and catechols (Figure

1.20) [111, 112].

Figure 1.20 Interaction of cellulose and lignin pyrolytic products [111]

1.5 Conclusions of literature review

As a result of our literature review we have identified as the following areas needing further investigation:

1) The thermal behaviors of amorphous and crystalline cellulose have not investigated

separately. The effect of cellulose crystallinity on the formation of liquid intermediates

under slow and heating rate conditions is not known. Cellulose crystallinity effects on the

distribution of pyrolysis products are also poorly understood.

2) The composition of the liquid intermediate formed when cellulose is heated at very fast

heating rate conditions is unknown. Most of the authors consider this liquid intermediate

to be formed mainly by levoglucosan and as consequence the secondary reactions of

cellulose pyrolysis are described as reactions of levoglucosan. It was not possible to find

any paper highlighting the importance of cellulose oligomeric primary products

(cellobiosan, cellotriosan, etc.) as critical intermediates and participants of cellulose

secondary reactions (fragmentation, dehydration, cross-linking and polycondensation).

3) The nature of the interactions between cellulose and the lignocellulosic matrix is not

known. The literature reports the beneficial effect of sulfuric acid addition to acid wash

lignocellulosic materials on the yield of levoglucosan, but the mechanism of action is still

poorly studied.

4) None of the lumped models reported in the literature to describe cellulose pyrolysis

reactions (under slow heating rate conditions) include the formation of cross-linked

cellulose as a critical step for charcoal formation.

1.6 Dissertation objective

The objective of this work is to understand the mechanisms of cellulose thermal reactions and the effect of crystallinity, additives and pyrolysis conditions on the yield of cellulose derived products and reaction phenomena. The following 4 specific objectives are proposed:

1) To investigate the effect of cellulose crystallinity on the formation of “molten cellulose”

(liquid intermediate), yield of products and kinetic data during cellulose pyrolysis.

2) To investigate the effect of temperature and crystallinity of the formation of liquid

intermediate and yield of products during pyrolysis of cellulose and cellulose derived

products and the importance of cellobiosan as intermediate of cellulose secondary

reactions.

3) To investigate the effect of sulfuric acid as an additive on the yield of products from

Douglas fir and to explain the mechanism responsible for the increase observed in the

yield of levoglucosan.

4) To propose a new lumped reaction model that includes the formation of cross-linked

sugars as an important intermediate for charcoal production under slow heating rate

conditions and to investigate the effect of temperature and cellulose crystallinity on

reactions on cellulose solid/liquid phase reactions.

1.7 Methodology

To accomplish the objectives indicated in the previous section, the dissertation will be organized in the following sections:

Part 1: Literature review

Part 2: Understanding the effect of cellulose crystallinity on the formation of a liquid

intermediate and on product distribution during pyrolysis. Py-GC/MS, TGA, XRD, FTIR

and SEM will be utilized.

Part 3: Understanding the effect of pyrolysis temperature on the formation of levoglucosan and

oligoanhydrosugars during the fast pyrolysis of cellulose and Douglas fir (in wire mesh

reactor under atmosphere). Fast speed camera will be attached to analyze the kinetics of

cellulose thermal reactions. SEM, GC/MS and HPLC will be utilized for products

analysis.

Part 4: Understanding the effect of sulfuric acid on the pyrolysis of acid washed Douglas fir (in

wire mesh reactor under atmosphere). GC/MS and HPLC will be utilized for products

analysis purpose.

Part 5: Understanding the effect of cellulose crystallinity on solid/liquid phase reactions

responsible for the formation of carbonaceous residues during slow pyrolysis (in spoon

reactor). NMR, FTIR, IEC and SEM will be utilized.

Part 6: Conclusions

Figure 1.21 shows the scheme of this dissertation.

Figure 1.21 Scheme of dissertation

1.8 Publications

The following publications resulted from my PhD studies at WSU:

1. Zhouhong Wang, Armando G. McDonald, Roel J.M. Westerhof, Sascha R.A. Kersten,

Manuel Garcia-Perez: Effect of Cellulose Crystallinity on Cellulose Secondary Reactions in

Solid Phase — a Spoon Reactor Research (Submitted to Industrial and Engineering

Chemistry Research, 2013)

2. Zhouhong Wang, Armando G. McDonald, Roel JM Westerhof, Sascha RA Kersten,

Christian M. Cuba-Torres, Su Ha, Brennan Pecha, and Manuel Garcia-Perez: Effect of

Cellulose Crystallinity on the Formation of a Liquid Intermediate and on Product

Distribution during Pyrolysis. Journal of Analytical and Applied Pyrolysis 100 (2013): 56-

66.

3. Zhouhong Wang, Brennan Pecha, Roel J.M. Westerhof, Sascha R.A. Kersten, Manuel

Garcia-Perez, Effect of Pyrolysis Temperature on the Formation of Anhydrosugars during the

Fast Pyrolysis of Cellulose in a Wire Mesh Reactor at Atmospheric Pressure. (In preparation,

to be submitted to Energy & Fuels)

4. Zhouhong Wang, Shuai Zhou, Brennan Pecha, Roel J.M. Westerhof, Sascha R.A. Kersten,

Manuel Garcia-Perez. Effect of Sulfuric Acid on the Pyrolysis of Acid Washed Douglas fir in

an Atmospheric Wire Mesh Reactor. (In preparation, to be submitted to Energy & Fuels)

5. Shi-Shen Liaw, Zhouhong Wang, Pius Ndegwa, Craig Frear, Su Ha, Chun-Zhu Li, and

Manuel Garcia-Perez: Effect of pyrolysis temperature on the yield and properties of bio-oils

obtained from the auger pyrolysis of Douglas fir wood. Journal of Analytical and Applied

Pyrolysis 93 (2012): 52-62.

6. Roel JM Westerhof, D. Wim F. Brilman, Manuel Garcia-Perez, Zhouhong Wang, Stijn RG

Oudenhoven, and Sascha RA Kersten: Stepwise Fast Pyrolysis of Pine Wood. Energy &

Fuels 26, no. 12 (2012): 7263-7273.

7. Roel JM Westerhof, D. Wim F. Brilman, Manuel Garcia-Perez, Zhouhong Wang, Stijn RG

Oudenhoven, Wim PM van Swaaij, and Sascha RA Kersten: Fractional condensation of

biomass pyrolysis vapors. Energy & Fuels 25, no. 4 (2011): 1817-1829.

8. Daniel Mourant, Zhouhong Wang, Min He, Xiao Shan Wang, Manuel Garcia-Perez,

Kaicheng Ling, and Chun-Zhu Li.: Mallee wood fast pyrolysis: Effects of alkali and alkaline

earth metallic species on the yield and composition of bio-oil. Fuel 90, no. 9 (2011): 2915-

2922.

9. Jieni Lian, Shulin Chen, Shuai Zhou, Zhouhong Wang, James O’Fallon, Chun-Zhu Li, and

Manuel Garcia-Perez: Separation, hydrolysis and fermentation of pyrolytic sugars to produce

ethanol and lipids. Bioresource Technology 101, no. 24 (2010): 9688-9699.

10. Shuai Zhou, Zhouhong Wang, Shi-Shen Liaw, Chun-Zhu Li, Manuel Garcia-Perez: Effect

of Sulfuric Acid on the Pyrolysis of Douglas fir and Hybrid Poplar Wood: Py-GC/MS and

TG Studies. Submitted to the Journal of Analytical and Applied Pyrolysis, 2013

11. Tim Hilbers, Zhouhong Wang, Roel J.M. Westerhof, Sascha R.A. Kersten, Manuel Garcia-

Perez: Understanding Lignin-Cellulose Interactions during the Pyrolysis of their Blends

(Paper to be submitted to The Journal of Analytical and Applied Pyrolysis 2013)

Reference

[1] G.W. Huber, S. Iborra, A. Corma, Synthesis of transportation fuels from biomass: chemistry, catalysts, and engineering, Chemical reviews, 106 (2006) 4044-4098.

[2] H.S. Kheshgi, R.C. Prince, G. Marland, The potential of biomass fuels in the context of global climate change: Focus on Transportation Fuels1, Annual Review of Energy and the Environment, 25 (2000) 199-244.

[3] G. Berndes, M. Hoogwijk, R. van den Broek, The contribution of biomass in the future global energy supply: a review of 17 studies, Biomass and Bioenergy, 25 (2003) 1-28.

[4] A.V. Bridgwater, D. Meier, D. Radlein, An overview of fast pyrolysis of biomass, Organic Geochemistry, 30 (1999) 1479-1493.

[5] D. Meier, O. Faix, State of the art of applied fast pyrolysis of lignocellulosic materials — a review, Bioresource Technology, 68 (1999) 71-77.

[6] A.V. Bridgwater, Review of fast pyrolysis of biomass and product upgrading, Biomass and Bioenergy, 38 (2012) 68-94.

[7] S. Czernik, A.V. Bridgwater, Overview of Applications of Biomass Fast Pyrolysis Oil, Energy & Fuels, 18 (2004) 590-598.

[8] A.J. Toft, A comparison of integrated biomass to electricity systems, in: Chemical Engineering and Applied Chemistry, Aston University, Birmingham, UK, 1996.

[9] E.M. Rubin, Genomics of cellulosic biofuels, Nature, 454 (2008) 841-845.

[10] E. Sjöström, Wood chemistry: fundamentals and applications, ACADEMIC PressINC, 1993.

[11] B. Hinterstoisser, L. Salmén, Application of dynamic 2D FTIR to cellulose, Vibrational Spectroscopy, 22 (2000) 111-118.

[12] J. Hearle, The development of ideas of fine structure, Fibre Structure, (1963).

[13] A. O'Sullivan, Cellulose: the structure slowly unravel, Cellulose, 4 (1997) 173-207.

[14] D.L. VanderHart, R.H. Atalla, Studies of microstructure in native celluloses using solid- state carbon-13 NMR, Macromolecules, 17 (1984) 1465-1472.

[15] J. Sugiyama, R. Vuong, H. Chanzy, Electron diffraction study on the two crystalline phases occurring in native cellulose from an algal cell wall, Macromolecules, 24 (1991) 4168-4175.

[16] K. Mazeau, L. Heux, Molecular Dynamics Simulations of Bulk Native Crystalline and Amorphous Structures of Cellulose, The Journal of Physical Chemistry B, 107 (2003) 2394-2403.

[17] Y. Nishiyama, Structure and properties of the cellulose microfibril, Journal of Wood Science, 55 (2009) 241-249.

[18] R. Teeaar, R. Serimaa, T. Paakkarl, Crystallinity of cellulose, as determined by CP/MAS NMR and XRD methods, Polym Bull, 17 (1987) 231-237.

[19] L. Segal, J.J. Creely, A.E. Martin, C.M. Conrad, An Empirical Method for Estimating the Degree of Crystallinity of Native Cellulose Using the X-Ray Diffractometer, Textile Research Journal, 29 (1959) 786-794.

[20] L. Segal, J.J. Creely, A.E. Martin, C.M. Conrad, An empirical method for estimating the degree of crystallinity of native cellulose using the x-ray diffractometer, Tex Res J, 29 (1962) 786-794.

[21] L. Segal, An X-Ray Diffractometer Specimen Holder for Wet Cellulosic Materials, Textile Research Journal, 32 (1962) 702-703.

[22] R. Evans, R.H. Newman, U.C. Roick, Changes in cellulose crystallinity during kraft pulping. Comparison of infrared, x-ray diffraction and solid state NMR results, Holzforschung, 49 (1995) 498-504.

[23] Y. Cao, H. Tan, Study on crystal structures of enzyme-hydrolyzed cellulosic materials by X-ray diffraction, Enzyme Microb Tech, 36 (2005) 314-317.

[24] C.J. Garvey, I.H. Parker, G.P. Simon, On the interpretation of X-ray diffraction powder patterns in terms of the nanostructure of cellulose I fibres, Macromol Chem Phys, 206 (2005) 1568-1575.

[25] S.Y. Oh, D.I. Yoo, Y. Shin, H.C. Kim, H.Y. Kim, Y.S. Chung, W.H. Park, J.H. Youk, Crystalline structure analysis of cellulose treated with sodium hydroxide and carbon dioxide by means of X-ray diffraction and FTIR spectroscopy, Carbohydrate Research, 340 (2005) 2376- 2391.

[26] C. Kennedy, G. Cameron, A. Šturcová, D. Apperley, C. Altaner, T. Wess, M. Jarvis, Microfibril diameter in celery collenchyma cellulose: X-ray scattering and NMR evidence, Cellulose, 14 (2007) 235-246.

[27] S. Park, J. Baker, M. Himmel, P. Parilla, D. Johnson, Cellulose crystallinity index: measurement techniques and their impact on interpreting cellulase performance, Biotechnology for biofuels, 3 (2010) 10.

[28] W. Ruland, X-ray determination of crystallinity and diffuse disorder scattering, Acta Cryst, 14 (1961) 1180-1185.

[29] K. Wickholm, P.T. Larsson, T. Iversen, Assignment of non-crystalline forms in cellulose I by CP/MAS 13C NMR spectroscopy, Carbohydrate Research, 312 (1998) 123-129.

[30] R.H. Atalla, D.L. VanderHart, The role of solid state 13C NMR spectroscopy in studies of the nature of native celluloses, Solid State Nuclear Magnetic Resonance, 15 (1999) 1-19.

[31] T. Liitiä, S.L. Maunu, B. Hortling, Solid state NMR studies on cellulose crystallinity in fines and bulk fibres separated from refined kraft pulp, Holzforschung, 54 (2000) 618-624.

[32] E.L. Hult, P.T. Larsson, T. Iversen, A comparative CP/MAS 13C-NMR study of cellulose structure in spruce wood and kraft pulp, Cellulose, 7 (2000) 35-55.

[33] T. Liitiä, S.L. Maunu, B. Hortling, T. Tamminen, O. Pekkala, A. Varhimo, Cellulose crystallinity and ordering of hemicelluloses in pine and birch pulps as revealed by solid-state NMR spectroscopic methods, Cellulose, 10 (2003) 307-316.

[34] A. Watanabe, S. Morita, Y. Ozaki, Temperature-Dependent Structural Changes in Hydrogen Bonds in Microcrystalline Cellulose Studied by Infrared and Near-Infrared Spectroscopy with Perturbation-Correlation Moving-Window Two-Dimensional Correlation Analysis, Applied Spectroscopy, 60 (2006) 611-618.

[35] R.T. O'Connor, E.F. DuPre, E.R. McCall, Infrared Spectrophotometric Procedure for Analysis of Cellulose and Modified Cellulose, Analytical Chemistry, 29 (1957) 998-1005.

[36] R.T. O'Connor, E.F. DuPré, D. Mitcham, Applications of Infrared Absorption Spectroscopy to Investigations of Cotton and Modified Cottons, Textile Research Journal, 28 (1958) 382-392.

[37] C.Y. Liang, R.H. Marchessault, Infrared spectra of crystalline polysaccharides. II. Native celluloses in the region from 640 to 1700 cm.−1, Journal of Polymer Science, 39 (1959) 269-278.

[38] M.L. Nelson, R.T. O'Connor, Relation of certain infrared bands to cellulose crystallinity and crystal latticed type. Part I. Spectra of lattice types I, II, III and of amorphous cellulose, Journal of Applied Polymer Science, 8 (1964) 1311-1324.

[39] M.L. Nelson, R.T. O'Connor, Relation of certain infrared bands to cellulose crystallinity and crystal lattice type. Part II. A new infrared ratio for estimation of crystallinity in celluloses I and II, Journal of Applied Polymer Science, 8 (1964) 1325-1341.

[40] L.M. Ilharco, A.R. Garcia, J. Lopes da Silva, L.F. Vieira Ferreira, Infrared Approach to the Study of Adsorption on Cellulose: Influence of Cellulose Crystallinity on the Adsorption of Benzophenone, Langmuir, 13 (1997) 4126-4132.

[41] A. Watanabe, S. Morita, Y. Ozaki, Study on Temperature-Dependent Changes in Hydrogen Bonds in Cellulose Iβ by Infrared Spectroscopy with Perturbation-Correlation Moving-Window Two-Dimensional Correlation Spectroscopy, Biomacromolecules, 7 (2006) 3164-3170.

[42] A. Watanabe, S. Morita, Y. Ozaki, Temperature-Dependent Changes in Hydrogen Bonds in Cellulose Iα Studied by Infrared Spectroscopy in Combination with Perturbation-Correlation Moving-Window Two-Dimensional Correlation Spectroscopy: Comparison with Cellulose Iβ, Biomacromolecules, 8 (2007) 2969-2975.

[43] H. Yang, R. Yan, H. Chen, D.H. Lee, C. Zheng, Characteristics of hemicellulose, cellulose and lignin pyrolysis, Fuel, 86 (2007) 1781-1788.

[44] A. Thygesen, J. Oddershede, H. Lilholt, A.B. Thomsen, K. Stahl, On the determination of crystallinity and cellulose content in plant fibres, Cellulose, 12 (2005) 563-576.

[45] C. Vonk, Computerization of Ruland's X-ray method for determination of the crystallinity in polymers, Journal of Applied Crystallography, 6 (1973) 148-152.

[46] H. Rietveld, A profile refinement method for nuclear and magnetic structures, Journal of Applied Crystallography, 2 (1969) 65-71.

[47] Y. Nishiyama, J. Sugiyama, H. Chanzy, P. Langan, Crystal Structure and Hydrogen Bonding System in Cellulose Iα from Synchrotron X-ray and Neutron Fiber Diffraction, Journal of the American Chemical Society, 125 (2003) 14300-14306.

[48] R.H. Atalla, J.C. Gast, D.W. Sindorf, V.J. Bartuska, G.E. Maciel, Carbon-13 NMR spectra of cellulose polymorphs, Journal of the American Chemical Society, 102 (1980) 3249-3251.

[49] R. Teeäär, R. Serimaa, T. Paakkarl, Crystallinity of cellulose, as determined by CP/MAS NMR and XRD methods, Polymer Bulletin, 17 (1987) 231-237.

[50] H. Lennholm, T. Larsson, T. Iversen, Determination of cellulose Iα and Iβ in lignocellulosic materials, Carbohydrate Research, 261 (1994) 119-131.

[51] M. Akerholm, B. Hinterstoisser, L. Salmen, Characterization of the crystalline structure of cellulose using static and dynamic FT-IR spectroscopy, Carbohydr Res, 339 (2004) 569-578.

[52] ASTM Standard D5896-96, Standard Test Method for Carbohydrate Distribution of Cellulosic Materials, in, ASTM International, West Conshohocken, PA, 2007.

[53] E.O. Kraemer, Molecular weights of celluloses and cellulose derivates, Industrial & Engineering Chemistry, 30 (1938) 1200-1203.

[54] R. Evans, A.F.A. Wallis, Cellulose molecular weights determined by viscometry, Journal of Applied Polymer Science, 37 (1989) 2331-2340.

[55] N. Osman, Chemistry of Hot-Pressing Hybrid Poplar Wood, in: Natural Resources, University of Idaho, 2010.

[56] L. Schroeder, F. Haigh, Cellulose and wood pulp polysaccharides. Gel permeation chromatographic analysis, TAPPI [Technical Association of the Pulp and Paper Industry], (1979).

[57] I. Suckling, R. Allison, S. Campion, K. McGrouther, A. McDonald, Monitoring cellulose degradation during conventional and modified kraft pulping, Journal of pulp and paper science, 27 (2001) 336-341.

[58] A.G. Bradbury, Y. Sakai, F. Shafizadeh, A kinetic model for pyrolysis of cellulose, Journal of Applied Polymer Science, 23 (1979) 3271-3280.

[59] J. Piskorz, D.S.A.G. Radlein, D.S. Scott, S. Czernik, Pretreatment of wood and cellulose for production of sugars by fast pyrolysis, Journal of Analytical and Applied Pyrolysis, 16 (1989) 127-142.

[60] G. Varhegyi, E. Jakab, M.J. Antal, Is the Broido-Shafizadeh Model for Cellulose Pyrolysis True?, Energy & Fuels, 8 (1994) 1345-1352.

[61] J.P. Diebold, A unified, global model for the pyrolysis of cellulose, Biomass and Bioenergy, 7 (1994) 75-85.

[62] J.B. Wooten, J.I. Seeman, M.R. Hajaligol, Observation and Characterization of Cellulose Pyrolysis Intermediates by 13C CPMAS NMR. A New Mechanistic Model, Energy & Fuels, 18 (2003) 1-15.

[63] I. Pastorova, R.E. Botto, P.W. Arisz, J.J. Boon, Cellulose char structure: a combined analytical Py-GC-MS, FTIR, and NMR study, Carbohydrate Research, 262 (1994) 27-47.

[64] Y.-C. Lin, J. Cho, G.A. Tompsett, P.R. Westmoreland, G.W. Huber, Kinetics and Mechanism of Cellulose Pyrolysis, The Journal of Physical Chemistry C, 113 (2009) 20097- 20107.

[65] F. Ronsse, X. Bai, W. Prins, R.C. Brown, Secondary reactions of levoglucosan and char in the fast pyrolysis of cellulose, Environmental Progress & Sustainable Energy, 31 (2012) 256-260.

[66] X. Bai, P. Johnston, R.C. Brown, An experimental study of the competing processes of evaporation and polymerization of levoglucosan in cellulose pyrolysis, Journal of Analytical and Applied Pyrolysis, 99 (2013) 130-136.

[67] X. Bai, P. Johnston, S. Sadula, R.C. Brown, Role of levoglucosan physiochemistry in cellulose pyrolysis, Journal of Analytical and Applied Pyrolysis, 99 (2013) 58-65.

[68] W. Chaiwat, I. Hasegawa, J. Kori, K. Mae, Examination of Degree of Cross-Linking for Cellulose Precursors Pretreated with Acid/Hot Water at Low Temperature, Industrial & Engineering Chemistry Research, 47 (2008) 5948-5956.

[69] W. Chaiwat, I. Hasegawa, T. Tani, K. Sunagawa, K. Mae, Analysis of Cross-Linking Behavior during Pyrolysis of Cellulose for Elucidating Reaction Pathway, Energy & Fuels, 23 (2009) 5765-5772.

[70] J. Scheirs, G. Camino, W. Tumiatti, Overview of water evolution during the thermal degradation of cellulose, European Polymer Journal, 37 (2001) 933-942.

[71] O.P. Golova, Chemical Effects of Heat on Cellulose, Russian Chemical Reviews, 44 (1975) 687.

[72] T.J. Haas, M.R. Nimlos, B.S. Donohoe, Real-Time and Post-reaction Microscopic Structural Analysis of Biomass Undergoing Pyrolysis, Energy & Fuels, 23 (2009) 3810-3817.

[73] V. Mamleev, S. Bourbigot, M. Le Bras, J. Yvon, The facts and hypotheses relating to the phenomenological model of cellulose pyrolysis: Interdependence of the steps, Journal of Analytical and Applied Pyrolysis, 84 (2009) 1-17.

[74] V. Mamleev, S. Bourbigot, J. Yvon, Kinetic analysis of the thermal decomposition of cellulose: The main step of mass loss, Journal of Analytical and Applied Pyrolysis, 80 (2007) 151-165.

[75] P.J. Dauenhauer, J.L. Colby, C.M. Balonek, W.J. Suszynski, L.D. Schmidt, Reactive boiling of cellulose for integrated catalysis through an intermediate liquid, Green Chemistry, 11 (2009) 1555-1561.

[76] M.S. Mettler, A.D. Paulsen, D.G. Vlachos, P.J. Dauenhauer, Pyrolytic conversion of cellulose to fuels: levoglucosan deoxygenation via elimination and cyclization within molten biomass, Energy & Environmental Science, 5 (2012) 7864-7868.

[77] A.R. Teixeira, K.G. Mooney, J.S. Kruger, C.L. Williams, W.J. Suszynski, L.D. Schmidt, D.P. Schmidt, P.J. Dauenhauer, Aerosol generation by reactive boiling ejection of molten cellulose, Energy & Environmental Science, 4 (2011) 4306-4321.

[78] W.S.L. Mok, M.J. Antal, P. Szabo, G. Varhegyi, B. Zelei, Formation of charcoal from biomass in a sealed reactor, Industrial & Engineering Chemistry Research, 31 (1992) 1162-1166.

[79] J.-R. Richard, M. Antal, Jr., Thermogravimetric Studies of Charcoal Formation from Cellulose at Elevated Pressures, in: A.V. Bridgwater (Ed.) Advances in Thermochemical Biomass Conversion, Springer Netherlands, 1993, pp. 784-792.

[80] G.A. Zickler, W. Wagermaier, S.S. Funari, M. Burghammer, O. Paris, In situ X-ray diffraction investigation of thermal decomposition of wood cellulose, Journal of Analytical and Applied Pyrolysis, 80 (2007) 134-140.

[81] D. Gardiner, The pyrolysis of some hexoses and derived di-, tri-, and poly-saccharides, Journal of the Chemical Society C: Organic, 0 (1966) 1473-1476.

[82] P.M. Molton, T.F. Demmitt, Reaction mechanisms in cellulose pyrolysis: a literature review, in, 1977, pp. Medium: ED; Size: Pages: 90.

[83] H.B. Mayes, L.J. Broadbelt, Unraveling the Reactions that Unravel Cellulose, The Journal of Physical Chemistry A, 116 (2012) 7098-7106.

[84] F. Shafizadeh, T.T. Stevenson, Saccharification of douglas-fir wood by a combination of prehydrolysis and pyrolysis, Journal of Applied Polymer Science, 27 (1982) 4577-4585.

[85] G.N. Richards, Glycolaldehyde from pyrolysis of cellulose, Journal of Analytical and Applied Pyrolysis, 10 (1987) 251-255.

[86] H. Kawamoto, M. Murayama, S. Saka, Pyrolysis behavior of levoglucosan as an intermediate in cellulose pyrolysis: polymerization into polysaccharide as a key reaction to carbonized product formation, Journal of Wood Science, 49 (2003) 469-473.

[87] M. Weinstetn, A. Broido, Pyrolysis-Crystallinity Relationships in Cellulose, Combustion Science and Technology, 1 (1970) 287-292.

[88] A. Broido, A.C. Javier-Son, A.C. Ouano, E.M. Barrall, Molecular weight decrease in the early pyrolysis of crystalline and amorphous cellulose, Journal of Applied Polymer Science, 17 (1973) 3627-3635.

[89] M.T. Bhuiyan, N. Hirai, N. Sobue, Changes of crystallinity in wood cellulose by heat treatment under dried and moist conditions, Journal of Wood Science, 46 (2000) 431-436.

[90] J.F. Matthews, M. Bergenstr hle, G.T. Beckham, M.E. Himmel, M.R. Nimlos, J.W. Brady, M.F. Crowley, High-Temperature Behavior of Cellulose I, The Journal of Physical Chemistry B, 115 (2011) 2155-2166.

[91] Y. Yu, H. Wu, Significant Differences in the Hydrolysis Behavior of Amorphous and Crystalline Portions within Microcrystalline Cellulose in Hot-Compressed Water, Industrial & Engineering Chemistry Research, 49 (2010) 3902-3909.

[92] F.H. Holmes, C.J.G. Shaw, The pyrolysis of cellulose and the action of flame-retardants. I. significance and analysis of the tar, Journal of Applied Chemistry, 11 (1961) 210-216.

[93] D.J. Nowakowski, J.M. Jones, Uncatalysed and potassium-catalysed pyrolysis of the cell- wall constituents of biomass and their model compounds, Journal of Analytical and Applied Pyrolysis, 83 (2008) 12-25.

[94] R.J. Evans, D. Wang, F.A. Agblevor, H.L. Chum, S.D. Baldwin, Mass spectrometric studies of the thermal decomposition of carbohydrates using 13C-labeled cellulose and glucose, Carbohydrate Research, 281 (1996) 219-235.

[95] N. Shimada, H. Kawamoto, S. Saka, Different action of alkali/alkaline earth metal chlorides on cellulose pyrolysis, Journal of Analytical and Applied Pyrolysis, 81 (2008) 80-87.

[96] P.R. Patwardhan, J.A. Satrio, R.C. Brown, B.H. Shanks, Influence of inorganic salts on the primary pyrolysis products of cellulose, Bioresource Technology, 101 (2010) 4646-4655.

[97] A. Khelfa, G. Finqueneisel, M. Auber, J. Weber, Influence of some minerals on the cellulose thermal degradation mechanisms, Journal of Thermal Analysis and Calorimetry, 92 (2008) 795-799.

[98] D. Domvoglou, R. Ibbett, F. Wortmann, J. Taylor, Controlled thermo-catalytic modification of regenerated cellulosic fibres using magnesium chloride Lewis acid, Cellulose, 16 (2009) 1075-1087.

[99] Y. Tsuchiya, K. Sumi, Thermal decomposition products of cellulose, Journal of Applied Polymer Science, 14 (1970) 2003-2013.

[100] Y. Halpern, R. Riffer, A. Broido, Levoglucosenone (1,6-anhydro-3,4-dideoxy-.DELTA.3- .beta.-D-pyranosen-2-one). Major product of the acid-catalyzed pyrolysis of cellulose and related carbohydrates, The Journal of Organic Chemistry, 38 (1973) 204-209.

[101] S. Julien, E. Chornet, R.P. Overend, Influence of acid pretreatment (H2SO4, HCl, HNO3) on reaction selectivity in the vacuum pyrolysis of cellulose, Journal of Analytical and Applied Pyrolysis, 27 (1993) 25-43.

[102] G. Dobele, G. Rossinskaja, G. Telysheva, D. Meier, O. Faix, Cellulose dehydration and depolymerization reactions during pyrolysis in the presence of phosphoric acid, Journal of Analytical and Applied Pyrolysis, 49 (1999) 307-317.

[103] G. Dobele, D. Meier, O. Faix, S. Radtke, G. Rossinskaja, G. Telysheva, Volatile products of catalytic flash pyrolysis of celluloses, Journal of Analytical and Applied Pyrolysis, 58–59 (2001) 453-463.

[104] G. Dobele, T. Dizhbite, G. Rossinskaja, G. Telysheva, D. Meier, S. Radtke, O. Faix, Pre- treatment of biomass with phosphoric acid prior to fast pyrolysis: A promising method for obtaining 1,6-anhydrosaccharides in high yields, Journal of Analytical and Applied Pyrolysis, 68–69 (2003) 197-211.

[105] G. Dobele, G. Rossinskaja, T. Dizhbite, G. Telysheva, D. Meier, O. Faix, Application of catalysts for obtaining 1,6-anhydrosaccharides from cellulose and wood by fast pyrolysis, Journal of Analytical and Applied Pyrolysis, 74 (2005) 401-405.

[106] H. Kawamoto, S. Saito, W. Hatanaka, S. Saka, Catalytic pyrolysis of cellulose in sulfolane with some acidic catalysts, Journal of Wood Science, 53 (2007) 127-133.

[107] D. Radlein, J. Piskorz, D.S. Scott, Fast pyrolysis of natural polysaccharides as a potential industrial process, Journal of Analytical and Applied Pyrolysis, 19 (1991) 41-63.

[108] T. Hosoya, H. Kawamoto, S. Saka, Different pyrolytic pathways of levoglucosan in vapor- and liquid/solid-phases, Journal of Analytical and Applied Pyrolysis, 83 (2008) 64-70.

[109] H. Kawamoto, H. Morisaki, S. Saka, Secondary decomposition of levoglucosan in pyrolytic production from cellulosic biomass, Journal of Analytical and Applied Pyrolysis, 85 (2009) 247-251.

[110] T. Hosoya, H. Kawamoto, S. Saka, Thermal stabilization of levoglucosan in aromatic substances, Carbohydrate Research, 341 (2006) 2293-2297.

[111] T. Hosoya, H. Kawamoto, S. Saka, Cellulose–hemicellulose and cellulose–lignin interactions in wood pyrolysis at gasification temperature, Journal of Analytical and Applied Pyrolysis, 80 (2007) 118-125.

[112] T. Hosoya, H. Kawamoto, S. Saka, Solid/liquid- and vapor-phase interactions between cellulose- and lignin-derived pyrolysis products, Journal of Analytical and Applied Pyrolysis, 85 (2009) 237-246.

Chapter 2 Effect of Cellulose Crystallinity on the Formation of a

Liquid Intermediate and on Product Distribution during Pyrolysis

Journal of Analytical and Applied Pyrolysis

Volume 100, March 2013, Pages 56–66

Zhouhong Wang1, Armando G. McDonald2, Roel J.M. Westerhof3, Sascha R.A. Kersten3,

Christian M. Cuba-Torres4, Su Ha4, Brennan Pecha4, Manuel Garcia-Perez1*

1 Biological Systems Engineering, Washington State University, Pullman, WA 99164, USA

2Department of Forest, Rangeland and Fire Sciences, University of Idaho, Moscow, ID 83844,

USA

3Thermo-Chemical Conversion of Biomass Group, Faculty of Science and Technology,

University of Twente, Postbus 217, 7500AE Enschede, The Netherlands

4 The Voiland School of Chemical Engineering and Bioengineering, Washington State University,

Pullman, WA, 99164, USA

Abstract: The effect of cellulose crystallinity on the formation of a liquid intermediate and on its thermal degradation was studied thermogravimetrically and by Py-GC/MS using commercial cellulose (control, crystallinity at 60.5 %) and ball-milled cellulose (low cellulose crystallinity at

6.5%). The crystallinity of the materials studied was quantified by XRD and FTIR.

Thermogravimetric analyses (TGA) show the samples with lower crystallinity starts to degrade at lower temperatures, exhibiting sharper DTG curves and lower thermal degradation activation

energies. Scanning electron microscopy (SEM) studies of the solid residues formed in TGA tests showed that, while the conversion of the ball-milled cellulose (mostly amorphous cellulose) occurs through the formation of a liquid intermediate, in the conversion of Cellulose the fibrous structure is conserved. Py-GC/MS studies showed major differences in the thermal behavior of the samples studied. At 300 oC, amorphous cellulose yielded more levoglucosan. At temperatures between 350 and 450 oC, higher yields of mono-anhydrosugars (levoglucosan and levoglucosenone) were obtained with the samples with higher crystallinity (control). The ball- milled cellulose produced more 5-(hydroxymethyl) furfural, 5-methylfurfural and furfural. The higher yields of these compounds are due to the acceleration of dehydration reactions when a liquid phase intermediate was formed. Fragmentation reactions responsible for the formation of light compounds (glycolaldehyde, acetic acid, methyl-vinyl-ketone and acetol) and the reactions responsible for the formation of cyclopentane do not seem to be affected by cellulose crystallinity and by the formation of a liquid intermediate.

Key words: Cellulose crystallinity, TGA, Py-GC/MS, depolymerization, pyrolysis

Corresponding Author:

Dr. Manuel Garcia-Perez

Associate Professor

LJ Smith Hall, Room 205

Pullman, WA, 99164-6120

Phone: 509-335-7758, Fax: 509-335-2722 e-mail: [email protected]

2.1 Introduction

The increasing demand of fossil transportation fuels, its limited supply, environmental issues associated with its use, and price volatility are the main causes for the growing interest in developing advanced bio-fuels [1-3]. Cellulose, the most abundant terrestrial bio-molecule, has been widely studied due to its potential to serve as a cheap resource for the production of sugars either through thermochemical or biochemical conversion pathways [4]. However, compared with the large number of papers published on the enzymatic hydrolysis of cellulose to produce glucose, the body of information on the production of anhydrosugars via fast pyrolysis is limited

[5-22]. The anhydrosugars (mainly levoglucosan) can be hydrolyzed into glucose and fermented to produce ethanol [4, 23] or metabolized directly by genetically modified organisms containing levoglucosan kinase for direct ethanol or lipid production [4].

Levoglucosan yields, as high as 60 wt. %, have been reported for the pyrolysis of pure cellulose

[7, 8, 14, 15, 21, 24]. There is considerable controversy in the literature about the mechanism of levoglucosan formation from cellulose (homolytic vs. heterolytic; free radical, ionic vs. concerted) [25, 26]. The two free radical mechanisms proposed are based on the pioneering works of Broido and Golova et al. [5, 6]. While the first free radical mechanism [5], suggests that the depolymerization of cellulose occurs by random cleavage of glycosidic linkages, the second mechanism [6] states that once a chain reaction is initiated in the non-reducing end, the whole chain “unzips” [9, 10, 22, 27-29]. Ponder et al. [30] propose the presence of ionic mechanisms which involve the formation of a glycosyl cation end group by ionic glycosidic bond cleavage

(GBC) [30]. Based on the density functional theory (DFT) calculations, Mayes and Broadbelt

[26] proposed a concerted GBC mechanism. The authors found that the kinetics favor the

“concerted GBC mechanism” over free radical and ionic conditions. The concerted mechanism was used in the kinetic model of cellulose pyrolysis proposed by Vinu and Broadbelt [19].

It is worth noting that the effect of cellulose crystallinity has been very poorly studied in the literature [25, 28, 29, 31-34]. It has been postulated that amorphous cellulose leads to more char and gas formation while crystalline cellulose contributes more to the formation of levoglucosan

[25, 31]. Weinstein and Broido [32, 33] found that the crystallinity index remained at a fairly constant even at weight loss over 50 wt. % before dropping rapidly. Cabradilla and Zeronian [31] found that levoglucosan was formed in both crystalline and amorphous cellulose, but those competitive dehydration reactions only occurred in the amorphous fraction.

Although there are some studies devoted to the formation of liquid intermediates from cellulose

[35-37], the relationship between cellulose crystallinity and the formation of liquid intermediates is still poorly studied [37]. Mamleev et al. [38] argued that a liquid state is needed to justify the dramatic dehydration and cross-linking reactions happening at temperatures over 300 oC.

Furthermore, Mamleev et al. [38] suggested that because dehydration belongs to the E-1 (first order elimination) reaction, this mechanism has to be invariably connected with an acid catalyst in liquid phase. Because cellulose crystallinity affects the formation of liquid intermediates, it could also affect dehydration reactions. According to Lede [37] the theoretical meting point (Tm)

o of crystalline cellulose should be between 417 and 478 C. The glass transition temperature (Tg)

of amorphous cellulose has been reported between 243 and 307 oC [39]. The formation of oligosugars by transglycosylation could contribute to the formation of a liquid phase [20, 38] at even lower temperatures. The melting points of levoglucosan and cellobiosan are 180 and 240 oC, respectively [40]. Thus the main purpose of this paper is to study the effect of cellulose crystallinity on the formation of liquid intermediates and the yield of products during cellulose pyrolysis.

2.2 Materials and Methods

Cellulose Sample: Control cellulose (Avicel PH-101, ~50 µm) was purchased from Sigma-

Aldrich. Amorphous cellulose was obtained by ball-milling control cellulose at 300 rpm for up to

48 h, using an Across International PQ-N2 ball mill (ceramic 100 mL jar and balls).

Cellulose Crystallinity: Cellulose crystallinity is the mass fraction of crystalline cellulose in the sample studied. In this work, cellulose crystallinity index (CrI) (calculated by XRD and FTIR) was used to characterize this crystalline fraction.

XRD: A Philips diffractometer using Co Kα radiation with an iron filter and the Bragg–Brentano optical configuration. The instrument scanned 2θ from 10 to 50o. The data was later converted into Cu Kα radiation range with 2θ from 8.6 to 42.7o. The peaks in the spectra were assigned according to the literature (101, 10ī, 021, 002 and 040) and assumed Gaussian functions with

peak deconvolution method [41]. CrI was calculated as the ratio of peak areas assigned to crystalline cellulose to the total peak area [41].

FTIR: A Shimadzu FTIR-8400S equipped with Pike MIRacle ATR and Germanium optic lens was used. The spectra (3 replicates) obtained were ATR and baseline corrected, normalized, and averaged. The ratio of peak intensity at I1372/I2899, was also used to determine following the method described elsewhere [42].

Thermogravimetric analyses: TGA was performed with a Mettler Toledo TGA/SDTA 851e instrument at temperatures between 25 and 600 oC with heating rates of 2, 5, 10, 20, 30, 40, and

o 50 C/min. The kinetic parameters of activation energy (Ea) and pre-exponential factor describing the thermochemical reactions were estimated by the ASTM E1641-07 method described elsewhere [43].

Scanning Electron Microscope: SEM was performed on an FEI 200F SEM system with Large

Field Detector and a low vacuum environment (130 Pa). High voltage was set at 20kV (30kV for raw material) and the magnification was set from 50 to 5,000X.

Py-GC/MS: A CDS Pyroprobe 5,000 with a platinum coil probe (Calibrated with type K thermocouple) and a CDS 1,500 valve interfaced with an Agilent 6890N/Agilent 5975B GC/MS was used. The GC/MS was equipped with a HP-5MS capillary column (30 m x 250 m, 0.25 micron) and an inert XL MSD with EI ionization. The pyro-probe oven and the GC liner were

kept at 270 oC. The cellulose was weighed in a Mettler Toledo thermo balance and loaded in a quartz tube, held at 270 oC (1 min) before heating to between 300 and 500 oC at a heating rate of

100 oC per second, and kept at this temperature for 1 min. The pyrolysis products were transferred to the GC/MS for separation (column heated from 40 to 280 oC at a heating rate of 6 oC per minute. The identification of each compound was achieved based on retention times and matching the mass spectrum recorded with those in the spectral library (NIST/EPA/NIH Mass

Spectral Library Version 2.0d, FairCom Corporation).

2.3 Results and Discussions

2.3.1 Ball-milling and Cellulose Characterization

Cellulose samples were ball-milled using different ball-milling times and the XRD results are shown in Figure 2.1. In this figure the existence of five major peaks (101, 10ī, 021, 002 and 040) is a characteristic of crystalline cellulose and an overlapped shallow peak is assigned to the amorphous contribution. A peak fitting method, assuming Gaussian distributions, was employed.

The areas of fitted peaks were used to quantify cellulose crystallinity. The XRD analysis of control cellulose gave a crystallinity value of 60.5%, which is close to the range reported in previous studies: 60.6% measured by solid state 13C NMR, 71% measured by water-activated cellulose-based electro rheological fluids, and 63% by Herman’s method [44, 45]. The XRD shown in Figure 2.1 clearly indicates that the intensity of peaks assigned to crystalline cellulose was reduced and a broad peak indicating amorphous cellulose increased with the increasing of ball-milling time.

021 002 101 10ī

040 48 h

24 h

8 h

4 h

2 h

1 h

Control

Figure 2.1 XRD of Avicel with ball-milling time from 0 (bottom) to 48 h (top)

Figure 2.2 shows the changes in cellulose crystallinity as a function of ball-milling time. CrI was determined by taking a ratio between the peak area of crystalline cellulose and total area. The crystallinity of cellulose decreased gradually as ball-milling progressed, reaching a final crystallinity of 6.5% at the ball-milling time of 24 h. Ball-milling for longer times (48 h) did not result in any further reduction in crystallinity. TGA of ball-milled samples at 24 and 48 hours were conducted (data not shown). No major differences between these two samples were observed. This result suggests that the changes in structure (crystallinity and degree of

polymerization (DP)) within the temperature range studied do not influence the outcome of the

TGA experiments.

6.5 %

Figure 2.2 Cellulose crystallinity (determined by XRD) as a function of ball-milling time (top)

Figure 2.3 shows the FTIR spectra obtained for cellulose samples with different ball-milling times. The ratio of IR bands’ intensity at two different wavenumbers (i.e., 1372/2899) was used to estimate a crystalline index [44]. In Figure 2.4, the values obtained for this index were plotted versus the crystallinity obtained by XRD. A good correlation was obtained. Based on the results obtained by both the XRD and the FTIR, 24 h of ball-milling was chosen for the preparation of a sample rich in amorphous cellulose (crystallinity: 6.5 %).

48 h

24 h

8 h 4 h

2 h

1 h Control

Figure 2.3 FTIR spectra of cellulose after ball-milling (0-48 h).

Figure 2.4 Crystallinity indexes obtained by FTIR spectroscopy (1372/2899) versus crystallinity obtained by XRD

Figure 2.5 shows Scanning electron micrographs (SEM) of control and 24 h ball-milled cellulose.

Ball-milling reduces the size of the cellulose fiber from approximately 100 m to less than 10

m but in all cases a solid like material was observed.

A B

Figure 2.5 Scanning electron micrographs of control (left) and ball-milled cellulose (right).

2.3.2 Thermogravimetric Analyses

Figure 2.6 shows the thermograms obtained for samples heated at 5 and 50 oC/min till 600 oC. It can be observed that, independently of the heating rate used, the ball-milled samples start to degrade at a lower temperature. The heating rate used in our TG experiments was relatively low and the particles used were also very small. Under these conditions the thermal lag is very small and the thermal degradation is kinetically controlled. In kinetically controlled regimes a further reduction of particle size due to ball milling is unlikely to change the thermal behavior of the

samples studied. Thus, the change in thermal behavior observed is due to the higher content of amorphous cellulose in the ball-milled cellulose. This sample does not have to overcome the energy barrier of the crystal structure to degrade. Weinstein and Broido [32, 33] found a similar result but using ammonia swelled cellulose which has a crystallinity comparable to ball-milled cellulose. Ball-milling does not seem to significantly affect the amount of solid residue (charcoal) produced (See Figure 2.6).

Figure 2.6 TGA thermograms of control cellulose and ball-milled (24h) cellulose.

The DTG curves of control and ball-milled (24 h) cellulose are shown in Figure 2.7. The DTG curves obtained for the ball-milled cellulose was sharper and the maximum degradation rate

shifted to lower temperatures. Ball-milled cellulose showed weight losses concentrated in a much narrower range of temperature than the control cellulose.

50 oC/min

40 oC/min

50 oC/min 30 oC/min

40 oC/min 30 oC/min 20 oC/min o 20 C/min o o 10 C/min 10 C/min o o 5 C/min 5 C/min o 2 oC/min 2 C/min

Figure 2.7 DTG thermograms at different heating rates of control cellulose (left) and after ball- milling (24 h) (right).

The thermal degradation apparent activation Energy (Ea) for the control and 24 h ball-milled cellulose were estimated using the ASTM method [43]. In Figure 2.8, conversion is defined as the amount of cellulose converted into volatiles (vapor and gas), which escaped from the pan and caused weight reduction, divided by the weight loss achieved at the maximum temperature studied [43]. At very low conversions (0-0.25), the ball-milled cellulose showed considerably lower apparent activation energy Ea (162 kJ/mol) than the control (222 kJ/mol). This result may be due to the effect of crystalline structure on the thermal stability of cellulose or due to the effect of small particle sizes. The Ea for the ball-milled sample increases as a function of conversion, reflecting the concentration of the crystalline cellulose in the partially converted samples. The value of Ea calculated for control is similar to the Ea (231 kJ/mol) calculated for cellulose, based on the concerted mechanism proposed by Mayes and Broadbelt [26].

Figure 2.8 Graphs showing Ea of control cellulose (left) and ball-milled (24 h) cellulose (right) as a function of conversion obtained by ASTM method.

2.3.3 SEM

Further analysis of the TGA derived solid residue by SEM proves the existence of very different thermal state between the samples studied. The charcoal residue obtained in the TGA (final temperature 600 oC) for the ball-milled cellulose seems to be derived from a liquid intermediate

(Figure 2.9). Unheated cellulose samples show clear fiber/particle structure (Figure 2.5). In the case of the control cellulose, the fibers conserved their integrity through the thermal degradation process with a much narrower fiber dimension, indicating that the thermal degradation happens mostly on the surface of the fiber [27] (See Figure 2.9 A1, A2, A3). Ball-milled cellulose, on the other hand, seems to go through the formation of a liquid intermediate with bubbles formed from liquid boiling (See Figure 2.9 B1, B2 and B3). A cellular foam structure (pointed out in 500X) and unburst bubbles, respectively, were observed in the SEM (See Figure 2.9 B2 and B3). This is a clear indication that melting has occurred in ball-milled (amorphous) cellulose. Small crystallites were observed in the liquid intermediate which are perhaps associated to the

crystalline fraction (6.5 %, determined by XRD) of ball-milled cellulose (See Figure 2.9 B3).

The formation of this liquid intermediate is believed to be involved in dehydration and cross- linking reactions, which will be discussed later. Dehydration reactions of the E1-elimination type are invariably connected with acid catalysts that are viable in liquid phase [36, 46].

A1 B1

A2 B2

Bubble

A3 B3 Crystal

Figure 2.9 Scanning electron micrographs of TGA derived char for control (left) and ball-milled (right) cellulose. From top to bottom: unheated cellulose, residue full view, 500X, and 5000X.

2.3.4 Py-GC/MS analyses

Pyrolysis GC/MS studies were conducted to gain insight on the distribution of products at different pyrolysis temperatures as a function of cellulose. Figure 2.10 shows the Py-GC/MS chromatographs for control and ball-milled cellulose at 300 and 400 oC. The heating rate of the biomass sample was calculated to be approximately 100 oC per second. The compounds assigned to each of the peaks are shown in Table 2.1. The area of each of the peaks listed in table 1 was divided by the initial sample loading and plotted in Figure 2.11 to Figure 2.15. These figures show the (semi-quantitative) yield of each of the compounds formed as a function of temperature for the control and ball-milled cellulose.

o 300 C 300 oC

Control Ball-milled

o 400 C 400 oC

Control Ball-milled

Figure 2.10 Total ion chromatograms of pyrolyzed control cellulose (left) and ball-milled cellulose (right) at 300 and 400 oC.

Table 2.1 Compounds identified by Py-GC/MS of cellulose Peak No. Retention Time Name Formula (minutes) 1 carbon dioxide 1.6 CO2 2 furan 1.8 C4H4O 3 light compounds (glycolaldehyde, 2.06-2.5 methyl-vinyl-ketone, and acetic acid) 4 acetol 2.72 C3H6O2 5 2(5H)-furanone 4.6 C4H4O2 6 furfural 5.3 C5H4O2 7 2-propyl-furan, 6 C7H10O 8 dihydro-4-hydroxy-2(3H)-furanone, 6.1 C4H6O3 9 2-hydroxy-2-cyclopenten-1-one 7.5 C5H6O2 10 5-methyl-2-furfural, 8.3 C6H6O2 11 phenol 8.9 C6H6O 12 2-hydroxy-3-methyl-2-cyclopenten-1- 9.6 C H O one 6 8 2 13 3-methyl-1,2-cyclopentanedione 10 C6H8O2 14 levoglucosenone 12.15 C6H6O3 15 1,4:3,6-dianhydro- glucopyranose 14.7 C6H8O4 16 5-(hydroxymethyl)-2-furfural 15.4 C6H6O3 17 levoglucosan 21 C6H10O5

Figure 2.11 shows the effect of pyrolysis temperature on the yield of anhydrosugars (levoglucosan, levoglucosenone, 1,4:3,6-dianhydro-glucopyranose). The yield of anhydrosugars obtained was considerably larger for the sample with higher crystallinity, when the temperature was higher than 350 oC. Although, both the control and ball-milled cellulose started to produce levoglucosan at 300 oC, the yield of levoglucosan was higher for amorphous cellulose. This result suggests that the cellulose crystalline structure may hinder levoglucosan production at low temperature. Note, that the control cellulose still contains a significant amount of amorphous cellulose.

Figure 2.11 Effect of cellulose pyrolysis temperature on the yield of levoglucosan, levoglucosenone, 1,4:3,6-dianhydro-glucopyranose, and total sugars.

Levoglucosan is the main product of cellulose depolymerization reactions via transglycosylation

[6 - 8, 46]. The dehydration of levoglucosan leads to the formation of levoglucosenone and

o 1,4:3,6-dianhydro-glucopyranose [47]. The yield of CO2 (Figure 2.12) at 350 C was higher for the ball-milled cellulose. At 400 oC the yield of this compound was comparable with the ball- milled cellulose, but at higher temperatures does not seem to be affected by cellulose crystallinity. According to Mamleev et al. [38, 46] the dehydration of the non-reducing end of cellulose is the main source of H2O and CO2.

Figure 2.12 Effect of cellulose pyrolysis temperature on the yield of carbon dioxide.

Figure 2.13 shows the effect of pyrolysis temperature on the yield of larger furanic compounds

(5-hydroxymethyl furfural, 5-methyl-furfural, and furfural). It is interesting to note that, in the whole range of temperature studied, the yield of these larger furanic compounds was considerably higher for the ball-milled cellulose samples (with lower crystallinity) than for the control cellulose.

Figure 2.13 Effect of pyrolysis temperature on the yield of 5-hydroxymethyl furfural, 5-methylfurfural, and furfural.

Scheirs et al. [48, 49] proposed a mechanism for the formation of these compounds that considers an acid catalyzed ring contraction step followed by two consecutive water elimination steps and elimination of formaldehyde [48, 49]. The formation of higher quantities of these larger furanic compounds when ball-milled samples (lower crystallinity) are pyrolyzed could be explained by the formation of a liquid intermediate which is needed for the dehydration step to occur [46]. These liquid intermediates (mainly oligo-anhydrosugars) will enhance dehydration reactions of the E1 elimination type as well as the formation of higher concentrations of larger furanic compounds. Possibly, this effect is enhanced by a decrease in degree of polymerization

(DP) for the ball-milled cellulose.

Figure 2.14 shows the effect of pyrolysis temperature in the yield of smaller furanic compounds

(2(5H)-furanone, 2(3H)-furanone, 2-propyl furan, and furan) for control and ball-milled cellulose as a function of pyrolysis temperatures. At higher temperatures the yield obtained for the control compounds is slightly higher than the one obtained for the ball-milled samples. These results suggest that the reaction mechanism responsible for the formation of 2(5H)-furanone, 2(3H)- furanone, 2-propyl furan and furan is not the same with larger furanic compounds, or is affected dramatically by the presence of liquid intermediates.

Figure 2.14 Effect of pyrolysis temperature on the yield of 2(5H)-furanone, 2(3H)-furanone, 2- propyl furan and furan.

Figure 2.15 shows the effect of pyrolysis temperature on the yield of cyclopentenes and phenol.

Cyclopentene formation mechanism proposed by Paine et al [50] occurs through an intra- molecular cyclicization and then a biradical rearrangement mechanism to create the precursor

bicyclopentane structure. Phenol is formed by a mechanism that considers the rearrangement and condensation reactions of the light compounds [51]. The yield of these compounds does not seem to be significantly affected by cellulose crystallinity.

Figure 2.15 Effect of pyrolysis temperature on the yield of 3-methyl-1,2-cyclopentanedione, 2- hydroxy-3-methyl-2-cylcopenten-1-one, 2-hydroxy-2-cyclopenten-1-one, and phenol.

Figure 2.16 shows the effect of pyrolysis temperature on the production of light compounds (A: light compounds; B: acetol). These are the main products of fragmentation reactions [11]. It is interesting to note that none of these compounds are produced at temperatures below 350 oC, suggesting that cellulose fragmentation reactions only happen at higher temperatures. These reactions are at least partly responsible for the reduction in the yield of levoglucosan and other sugars observed at temperatures over 400oC. What was interesting is that these light compound oriented reactions (mostly fragmentation) were not influenced by the cellulose crystallinity. This may suggest that the liquid intermediate caused by amorphous state is not required in their

formation. These compounds are more likely to be the fragments of retro-aldol reactions in the gas phase [49, 51-54].

A B

Figure 2.16 Effect of pyrolysis temperature on the yield of A (light compounds) and B (acetol).

2.4 Conclusions

The effect of crystallinity on cellulose thermochemical reactions was studied by comparing the behavior of cellulose after ball-milling with varying CrI. Ball-milling effectively decreased the

CrI of cellulose. A good linear correlation was found between the CrI determined by XRD and

FTIR. Ball-milled cellulose was less thermally stable than the control cellulose and had a lower

Ea at low conversions. As the conversion increases, the samples with higher content of amorphous cellulose exhibit apparent activation energy (Ea) comparable to those obtained for the sample with higher crystallinity. The ball-milled samples (with low crystallinity) seem to have thermally degraded via the formation of a liquid intermediate involved in dehydration reactions, while the control samples (with significantly higher crystallinity) did not reveal clear evidence of the formation of liquid intermediates. Although higher yields of anhydrosugars are obtained for

the sample with higher crystallinity, high yields of levoglucosan are also obtained with high levels of amorphous cellulose. Samples with lower crystallinity yield more of the larger furanic compounds (5-HMF, 5-methyl-furfural, and furfural) formed by an acid catalyzed ring contraction mechanism and consecutive water-elimination steps. The higher yields of these compounds can be explained by the formation of a liquid phase intermediate needed for the dehydration reactions in the amorphous cellulose. Crystallinity does not seem to affect fragmentation reactions that form lighter organic molecules.

Acknowledgements

This project was financially supported by the US National Science Foundation (CBET-0966419), the Sun-Grant Initiative (Interagency Agreement: T0013G-A), and the Washington State

Agricultural Research Center. The authors are very thankful for their support. The authors are also very thankful to the help of Franceschi Microscopy and Imaging Center of Washington State

University and the kind support of Dr. Valerie Lynch-Holm.

Reference

[1] E. Mészáros, G. Várhegyi, E. Jakab, B. Marosvölgyi, Thermogravimetric and Reaction Kinetic Analysis of Biomass Samples from an Energy Plantation, Energy & Fuels, 18 (2004) 497-507.

[2] D.J. Hayes, An examination of biorefining processes, catalysts and challenges, Catalysis Today, 145 (2009) 138-151.

[3] M.R. Peláez-Samaniego, M. Garcia-Perez, L.B. Cortez, F. Rosillo-Calle, J. Mesa, Improvements of Brazilian carbonization industry as part of the creation of a global biomass economy, Renewable and Sustainable Energy Reviews, 12 (2008) 1063-1086.

[4] L. Jarboe, Z. Wen, D. Choi, R. Brown, Hybrid thermochemical processing: fermentation of pyrolysis-derived bio-oil, Applied Microbiology and Biotechnology, 91 (2011) 1519-1523.

[5] A. Broido, A.C. Javier-Son, A.C. Ouano, E.M. Barrall, Molecular weight decrease in the early pyrolysis of crystalline and amorphous cellulose, Journal of Applied Polymer Science, 17 (1973) 3627-3635.

[6] O.P. Golova, Chemical Effects of Heat on Cellulose, Russian Chemical Reviews, 44 (1975) 687.

[7] F. Shafizadeh, R.H. Furneaux, T.G. Cochran, J.P. Scholl, Y. Sakai, Production of levoglucosan and glucose from pyrolysis of cellulosic materials, Journal of Applied Polymer Science, 23 (1979) 3525-3539.

[8] F. Shafizadeh, T.T. Stevenson, Saccharification of douglas-fir wood by a combination of prehydrolysis and pyrolysis, Journal of Applied Polymer Science, 27 (1982) 4577-4585.

[9] J. Piskorz, D. Radlein, D.S. Scott, On the mechanism of the rapid pyrolysis of cellulose, Journal of Analytical and Applied Pyrolysis, 9 (1986) 121-137.

[10] J. Piskorz, D.S.A.G. Radlein, D.S. Scott, S. Czernik, Pretreatment of wood and cellulose for production of sugars by fast pyrolysis, Journal of Analytical and Applied Pyrolysis, 16 (1989) 127-142.

[11] D. Radlein, J. Piskorz, D.S. Scott, Fast pyrolysis of natural polysaccharides as a potential industrial process, Journal of Analytical and Applied Pyrolysis, 19 (1991) 41-63.

[12] D.S.A.G. Radlein, J. Piskorz, A. Grinshpun, D.S. Scott, Fast pyrolysis of pre-treated wood and cellulose, 1987.

[13] D.S.T.A.G. Radlein, A. Grinshpun, J. Piskorz, D.S. Scott, On the presence of anhydro- oligosaccharides in the sirups from the fast pyrolysis of cellulose, Journal of Analytical and Applied Pyrolysis, 12 (1987) 39-49.

[14] D.S. Scott, J. Piskorz, Process for the production of fermentable sugars from biomass, in, Google Patents, 1989.

[15] D.S. Scott, J. Piskorz, D. Radlein, P. Majerski, Process for the production of anhydrosugars from lignin and cellulose containing biomass by pyrolysis, in, Google Patents, 1995.

[16] T. Hosoya, H. Kawamoto, S. Saka, Different pyrolytic pathways of levoglucosan in vapor- and liquid/solid-phases, Journal of Analytical and Applied Pyrolysis, 83 (2008) 64-70.

[17] T. Hosoya, H. Kawamoto, S. Saka, Solid/liquid- and vapor-phase interactions between cellulose- and lignin-derived pyrolysis products, Journal of Analytical and Applied Pyrolysis, 85 (2009) 237-246.

[18] H. Kawamoto, H. Morisaki, S. Saka, Secondary decomposition of levoglucosan in pyrolytic production from cellulosic biomass, Journal of Analytical and Applied Pyrolysis, 85 (2009) 247- 251.

[19] R. Vinu, L.J. Broadbelt, A mechanistic model of fast pyrolysis of glucose-based carbohydrates to predict bio-oil composition, Energy & Environmental Science, 5 (2012) 9808- 9826.

[20] M.S. Mettler, D.G. Vlachos, P.J. Dauenhauer, Top ten fundamental challenges of biomass pyrolysis for biofuels, Energy & Environmental Science, 5 (2012) 7797-7809.

[21] P.R. Patwardhan, J.A. Satrio, R.C. Brown, B.H. Shanks, Influence of inorganic salts on the primary pyrolysis products of cellulose, Bioresource Technology, 101 (2010) 4646-4655.

[22] J. Piskorz, P. Majerski, D. Radlein, A. Vladars-Usas, D.S. Scott, Flash pyrolysis of cellulose for production of anhydro-oligomers, Journal of Analytical and Applied Pyrolysis, 56 (2000) 145-166.

[23] J. Lian, S. Chen, S. Zhou, Z. Wang, J. O’Fallon, C.-Z. Li, M. Garcia-Perez, Separation, hydrolysis and fermentation of pyrolytic sugars to produce ethanol and lipids, Bioresource Technology, 101 (2010) 9688-9699.

[24] F. Shafizadeh, Introduction to pyrolysis of biomass, Journal of Analytical and Applied Pyrolysis, 3 (1982) 283-305.

[25] M. Antal, Jr., Biomass Pyrolysis: A Review of the Literature Part 2—Lignocellulose Pyrolysis, in: K. Böer, J. Duffie (Eds.) Advances in Solar Energy, Springer US, 1985, pp. 175- 255.

[26] H.B. Mayes, L.J. Broadbelt, Unraveling the Reactions that Unravel Cellulose, The Journal of Physical Chemistry A, 116 (2012) 7098-7106.

[27] G.A. Zickler, W. Wagermaier, S.S. Funari, M. Burghammer, O. Paris, In situ X-ray diffraction investigation of thermal decomposition of wood cellulose, Journal of Analytical and Applied Pyrolysis, 80 (2007) 134-140.

[28] D.-Y. Kim, Y. Nishiyama, M. Wada, S. Kuga, T. Okano, Thermal Decomposition of Cellulose Crystallites in Wood, Holzforschung, 55 (2001) 521-524.

[29] U.-J. Kim, S.H. Eom, M. Wada, Thermal decomposition of native cellulose: Influence on crystallite size, Polymer Degradation and Stability, 95 (2010) 778-781.

[30] G.R. Ponder, G.N. Richards, T.T. Stevenson, Influence of linkage position and orientation in pyrolysis of polysaccharides: A study of several glucans, Journal of Analytical and Applied Pyrolysis, 22 (1992) 217-229.

[31] K. Cabradilla, S. Zeronian, Influence of crystallinity on the thermal properties of cellulose, in: Symposium on Thermal Uses and Properties of Carbohydrates and Lignins. San Francisco, Calif.(USA). 1976., 1976.

[32] A. Broido, M. Weinstein, Thermogravimetric Analysis of Ammonia-Swelled Cellulose, Combustion Science and Technology, 1 (1970) 279-285.

[33] M. Weinstetn, A. Broido, Pyrolysis-Crystallinity Relationships in Cellulose, Combustion Science and Technology, 1 (1970) 287-292.

[34] J. Zhang, J. Luo, D. Tong, L. Zhu, L. Dong, C. Hu, The dependence of pyrolysis behavior on the crystal state of cellulose, Carbohydrate Polymers, 79 (2010) 164-169.

[35] J. Schroeter, F. Felix, Melting cellulose, Cellulose, 12 (2005) 159-165.

[36] P.J. Dauenhauer, J.L. Colby, C.M. Balonek, W.J. Suszynski, L.D. Schmidt, Reactive boiling of cellulose for integrated catalysis through an intermediate liquid, Green Chemistry, 11 (2009) 1555-1561.

[37] J. Lédé, Cellulose pyrolysis kinetics: An historical review on the existence and role of intermediate active cellulose, Journal of Analytical and Applied Pyrolysis, 94 (2012) 17-32.

[38] V. Mamleev, S. Bourbigot, M. Le Bras, J. Yvon, The facts and hypotheses relating to the phenomenological model of cellulose pyrolysis: Interdependence of the steps, Journal of Analytical and Applied Pyrolysis, 84 (2009) 1-17.

[39] K. Mazeau, L. Heux, Molecular Dynamics Simulations of Bulk Native Crystalline and Amorphous Structures of Cellulose, The Journal of Physical Chemistry B, 107 (2003) 2394-2403.

[40] V. Oja, E.M. Suuberg, Vapor Pressures and Enthalpies of Sublimation of d-Glucose, d- Xylose, Cellobiose, and Levoglucosan, Journal of Chemical & Engineering Data, 44 (1998) 26- 29.

[41] S. Park, J. Baker, M. Himmel, P. Parilla, D. Johnson, Cellulose crystallinity index: measurement techniques and their impact on interpreting cellulase performance, Biotechnology for biofuels, 3 (2010) 10.

[42] M.L. Nelson, R.T. O'Connor, Relation of certain infrared bands to cellulose crystallinity and crystal lattice type. Part II. A new infrared ratio for estimation of crystallinity in celluloses I and II, Journal of Applied Polymer Science, 8 (1964) 1325-1341.

[43] M. Garc a-Pèrez, A. Chaala, J. Yang, C. Roy, Co-pyrolysis of sugarcane bagasse with petroleum residue. Part I: thermogravimetric analysis, Fuel, 80 (2001) 1245-1258.

[44] P. Hermans, A. Weidinger, Quantitative X‐Ray Investigations on the Crystallinity of Cellulose Fibers. A Background Analysis, Journal of Applied Physics, 19 (1948) 491-506.

[45] Y. Nakai, E. Fukuoka, S. Nakajima, J. Hasegawa, Crystallinity and Physical Characteristics of Microcrystalline Cellulose, Chemical & pharmaceutical bulletin, 25 (1977) 96-101.

[46] V. Mamleev, S. Bourbigot, J. Yvon, Kinetic analysis of the thermal decomposition of cellulose: The main step of mass loss, Journal of Analytical and Applied Pyrolysis, 80 (2007) 151-165.

[47] P.M. Molton, T.F. Demmitt, Reaction mechanisms in cellulose pyrolysis: a literature review, in, 1977, pp. Medium: ED; Size: Pages: 90.

[48] J. Scheirs, G. Camino, M. Avidano, W. Tumiatti, Origin of furanic compounds in thermal degradation of cellulosic insulating paper, Journal of Applied Polymer Science, 69 (1998) 2541- 2547.

[49] J. Scheirs, G. Camino, W. Tumiatti, Overview of water evolution during the thermal degradation of cellulose, European Polymer Journal, 37 (2001) 933-942.

[50] J.B. Paine Iii, Y.B. Pithawalla, J.D. Naworal, Carbohydrate pyrolysis mechanisms from isotopic labeling: Part 3. The Pyrolysis of d-glucose: Formation of C3 and C4 carbonyl compounds and a cyclopentenedione isomer by electrocyclic fragmentation mechanisms, Journal of Analytical and Applied Pyrolysis, 82 (2008) 42-69.

[51] I. Pastorova, R.E. Botto, P.W. Arisz, J.J. Boon, Cellulose char structure: a combined analytical Py-GC-MS, FTIR, and NMR study, Carbohydrate Research, 262 (1994) 27-47.

[52] J.B. Paine Iii, Y.B. Pithawalla, J.D. Naworal, C.E. Thomas Jr, Carbohydrate pyrolysis mechanisms from isotopic labeling: Part 1: The pyrolysis of glycerin: Discovery of competing

fragmentation mechanisms affording acetaldehyde and formaldehyde and the implications for carbohydrate pyrolysis, Journal of Analytical and Applied Pyrolysis, 80 (2007) 297-311.

[53] J.B. Paine Iii, Y.B. Pithawalla, J.D. Naworal, Carbohydrate pyrolysis mechanisms from isotopic labeling: Part 2. The pyrolysis of d-glucose: General disconnective analysis and the formation of C1 and C2 carbonyl compounds by electrocyclic fragmentation mechanisms, Journal of Analytical and Applied Pyrolysis, 82 (2008) 10-41.

[54] J.B. Paine Iii, Y.B. Pithawalla, J.D. Naworal, Carbohydrate pyrolysis mechanisms from isotopic labeling: Part 4. The pyrolysis of d-glucose: The formation of furans, Journal of Analytical and Applied Pyrolysis, 83 (2008) 37-63.

Chapter 3 Effect of Pyrolysis Temperature on the Formation of

Anhydrosugars during the Fast Pyrolysis of Cellulose in a Wire

Mesh Reactor at Atmospheric Pressure

(Paper to be submitted to Energy and Fuels)

Zhouhong Wang 1, Brennan Pecha 1, Roel J.M. Westerhof 2, Sascha R.A. Kersten 2, Manuel

Garcia-Perez 1*

1Biological Systems Engineering, Washington State University, Pullman, WA 99164, USA

2Thermo-Chemical Conversion of Biomass Group, Faculty of Science and Technology,

University of Twente, Postbus 217, 7500AE Enschede, The Netherlands

Abstract: This paper reports the effect of pyrolysis temperature on the production of levoglucosan and cellobiosan during the pyrolysis of control cellulose (Avicel, microcrystalline cellulose), ball-milled cellulose and their main two primary depolymerization products

(levoglucosan, and cellobiosan) in a wire mesh reactor (300 - 500 oC) at atmospheric pressure with the aid of a fast speed camera. Under fast heating rate conditions, both control and ball- milled cellulose formed observable liquid intermediates. Ball-milled cellulose pyrolyzed and evaporated more rapidly than the control cellulose but do not seem to affect much the yield of products obtained. The activation energy of the ball-milled cellulose and control cellulose was estimated to be 81.3 and 37.7 kJ/mol, respectively. The yield of levoglucosan and cellobiosan, was measured by HPLC and GC/MS with a maximum yield of levoglucosan close to 60 wt. % at

300 oC for both samples studied; very little cellobiosan was detected. HPLC and GC/MS analysis

of pyrolysate from control and ball-milled cellulose confirm the formation of levoglucosan, and an unknown compound with HPLC residence time similar than glycolaldehyde, 1,6-anhydroglucofuranose. No cellobiosan formation was observed from cellulose. Pyrolysis of levoglucosan at 300 oC yielded 92.5 wt. % of the starting material. Pyrolysis of cellobiosan at 300 oC resulted in the production of relatively large quantities of the unidentified peak (estimated in more 30 wt. %), small quantities of levoglucosan, glycolaldehyde, other unidentified compounds and relatively large quantities of solid residue (18.2 wt. %). A second unidentified compound was detected by GC/MS. Compared with vacuum pyrolysis, in atmospheric pressure, the high boiling point of cellobiosan and presumably the other heavier anhydrosugars (products of cellulose primary reactions) cannot be evaporated and further react lowering the yield of hydrolysable sugars. Cellobiosan thermal reactions are the main source for the unknown peak observed in the cellulose HPLC chromatogram and a major contributor to the formation of solid reside.

Understanding and controlling the mechanisms responsible for the formation of these unknown compounds is critical to enhance the production of fermentable sugars from atmospheric pressure fast pyrolysis.

Key words: Levoglucosan, cellobiosan, cellulose pyrolysis

Corresponding Author:

Dr. Manuel Garcia-Perez

Associate Professor

LJ Smith Hall, Room 205

Pullman, WA, 99164-6120

Phone: 509-335-7758, Fax: 509-335-2722 e-mail: [email protected]

3.1 Introduction

Biomass is a viable source for the production of fuel and chemicals that are currently produced from fossil fuels. While, most of the biomass is currently consumed as traditional fuel wood at approximately 38 EJ/y (EJ = 1018) using very polluting technologies in developing nations, barely 7 EJ/y is used for the production of transportation fuels and electricity using contemporary clean, green technologies. Increasing biomass use for fuels and chemicals production in the next

50 years will depend on the development of both legislative policies and advancing technologies

[1-3].

Fast pyrolysis is a promising method to convert up to 75 wt. % of lignocellulosic materials into a crude bio-oil that can be further refined for the production of fuels and chemicals [4, 5].

Although fast pyrolysis results in a high yield of oil, in most cases a significant fraction of cellulose is fragmented to produce C1-C4 oxygenated molecules with limited market. Under very specific reaction conditions cellulose can be converted into levoglucosan which is intensively studied as a fermentable sugar for the production of fuels and chemicals [4, 6-9].

The yield of levoglucosan obtained from the pyrolysis of lignocellulosic materials reported in the literature varies widely. It is known that even small quantities of alkaline metals like sodium and potassium naturally found in biomass can decrease the yield of usable sugars due to the acceleration of fragmentation reactions [10-13]. The catalytic effect of alkalines can be eliminated by acid washing the feedstock. The aqueous phase of pyrolysis oil is rich in acetic

acid and has been proven to effectively remove alkalines to increase levoglucosan production

[14]. Early on, researchers found that adding small amounts of sulfuric acid to lignocellulosic materials (cellulose with lignin) causes a 2-fold increase in levoglucosan yield [15].

Although cellulose pyrolysis has been extensively studied for more than 50 years [16-19], there is little understanding of the factors affecting the yield of fermentable anhydrosugars

(levoglucosan and cellobiosan) . Recent molecular modeling studies [20, 21] based on density functional theory (DFT) have found a highly viable concerted mechanism (glycosidic bond cleavage and levoglucosan formation) for levoglucosan production that is more favorable than previously proposed mechanisms based on the formation of radical or ionic intermediates.

The yield of “cellulose primary products” has been recently reported by Westerhof et al. in vacuum studies in a wire mesh reactor [41]. The authors found that under extremely fast heating rate conditions (up to 7,000 oC/s), deep vacuum (<20 mbar), small samples (close to 0.05 g) and residence time of the vapors in hot environment <15-25 ms (conditions that minimize the effect of secondary reactions), at ~500 oC it is possible to obtain oil yield close to 95 % (majorly anhydro-(oligo-)sugars (levoglucosan, cellobiosan, cellotriosan, etc) from cellulose.

Surprinsingly at 500 oC under vacuum yields of levoglucosan was found only about 4-6 wt.% respectively. The yield of cellobiosan and cellotriosan produced was found 7-8 and 5-8 wt. % respectively. Higher yields of levoglucosan (11-25 wt. %) were obtained at 300 oC. Westerhof et al [41] also found that high cellulose crystallinity prevented the yield of sugar dimers and oligomers which was favored in ball-milled cellulose.

In 1979 Shafizadeh et al. [22] reported a fermentable sugar yield (as the D-glucose product of cellulose primary products from levoglucosan, cellobiosan, and heavier anhydrosugars hydrolysis) from the slow pyrolysis of acid washed cellulose under vacuum close to 77 mass %.

The yield of levoglucosan reported was 58 mass %. The same author, in 1983, conducted vacuum pyrolysis studies of cellulose (Whatman CF 11 powder) under vacuum at temperatures between 300 and 500 oC [25] and reported yields of hydrolysable sugars (as D-glucose) close to

60 %. The yield of levoglucosan was between 34 and 40 wt. %. The reason for the difference in the yield of hydrolysable sugars between slow and fast heating rate conditions is not fully undertood, but seems to be associated to secondary reactions of cellulose primary products in the solid or liquid phase [23, 24]. Kawamoto et al. [24] studied the polymerization of levoglucosan into polysaccharides as a key reaction for the formation of carbonized products [24]. Bai et al.

[25, 26] also studied the role of levoglucosan physico-chemistry in cellulose pyrolysis and also concluded that the polymerization of this compound is responsible for the formation of extra char.

It was not possible to find any research reporting secondary reactions of cellobiosan and the other heavy anhydrosugars produced during cellulose pyrolysis.

Shafizadeh and Fu (1973) [16] conducted cellulose (Whatman chromatographic cellulose powder

CF11) pyrolysis tests under vacuum and atmospheric pressure in a pyrex tube partly covered by a furnace. The yield of hydrolizable sugars under vacuum was 20.9 wt.% and at atmospheric presure was 6.08 wt.%. The trimethylsilated tar sample collected was analysed by GC/MS. The authors found that the polysacharide fraction had no reducing end-group, was randomly linked,

containing some furanoid rings, similar to products of levoglucosan condensation reactions. In

2006, Kwon et al. [27] also compared the yield of levoglucosan obtained under vacuum and atmospheric pressure from cellulose for the purpose of practical production of levoglucosan. In glass-bottle pyrolysis tests under vacuum the authors were able to obtain levoglucosan yields of

50-55 % between 350 and 410 oC. By contrast, glass-bottle pyrolysis under nitrogen gave low yields 17-20 %. The reasons for the dramatic differences between the yield of anhydrosugars in atmospheric and vacuum pyrolysis are still unknown. This information is critical to increase the production of fermentable sugars in most of the existing commercial fast pyrolysis reactors which typically operate at atmospheric pressure. The effect of cellulose crystallinity on the formation of anhydrosugars and on their secondary reactions is also poorly understood.

Thus, the main purpose of this paper is to investigate the effect of crystallinity on cellulose thermochemical reactions at atmospheric pressure with the aid of a wire mesh reactor and advance our understanding the main secondary reactions affecting the yield of fermentable sugars when the pyrolysis of cellulose and its main primary products (levoglucosan and cellobiosan) is conducted at fast heating rates under atmospheric pressure.

3.2 Materials and Methods

3.2.1 Materials

The cellulose samples (control and ball-milled) used in this work have been described elsewhere

[19, 28]. Control cellulose (Avicel PH-101 ~50µm) was purchased from Sigma-Aldrich and ball-

milled cellulose was obtained by ball-milling control cellulose at 300 RPM for 24 hours. Control cellulose had a crystallinity of 60.5% and the ball-milled cellulose had a crystallinity of 6.5% by

XRD measurement [19]. Levoglucosan was purchased from Carbosynth (for reaction in mesh reactor) and from Sigma-Aldrich (as standard in GC/MS, IEC and HPLC). Cellobiosan was purchased from Carbosynth. Both anhydrosugars (levoglucosan and cellobiosan) are primary products of cellulose pyrolysis [29].

3.2.2 Reactor

Mesh Reactor

Figure 3.1 shows the scheme of the wire mesh reactor used in our studies. Wire meshes (TWP

200 Mesh T304 Stainless 0.0533mm Wire Dia., cut into 50.8 mm wide and 63.5 mm long) were used to hold and heat samples. A welder was used to provide controlled current. An infrared sensor (Raytek Marathon MT) was used to measure the temperature of the mesh. A PID controller was utilized to control the heat rate (maximum 120 oC/s based on current setting) and hold temperature. The actual temperature was calibrated by sandwiching an insulated thermal couple (Omega XC series, type K) between two piece of mesh and heat for 30 s when temperature reading was stable.

Samples (~40 mg) were sandwiched between two pieces of meshes and tightly clamped on the reactor. Then they were heated to designed temperature (300, 375, 450 and 500 oC) with heating rates of 100 oC/s since the actual heating rate of samples in pyroprobe were no higher than this

value. The temperature was held for 10 seconds (for 375, 450 and 500 oC) or until the reactions finish (30 s for 300 oC). Limit to the efficiency of the controller, the actual heating curve always had an overshot of ~15-30 oC (depending on the final temperature), then quickly cooled to and waved around the designed temperature (±3 oC). With infrared sensor measuring, the temperature difference between sampling and blank area was ~50 oC. The reactor chamber was flushed with nitrogen (0.4 L/min) continuously and a vacuum pump was utilized to help evacuate the vapor through the condenser when heating is on. A condenser was filled with liquid nitrogen in the outer tube so the pyrolysis vapors could be condensed on a mesh filter and a paper filter in the inner tube. The products collected on the condenser (cooled tube) were washed with HPLC grade methanol. The weight of the mesh was carefully measured before and after each experiment. The resulting methanol solution was analyzed by HPLC and GC/MS. Yield of levoglucosan, acetol and glycolaldehyde was calculated by dividing the concentration obtained in the methanol solution from HPLC and GC/MS analyses with the concentration of raw material pyrolyzed.

The validation of recovery was done by evaporating pyrocatechol and decanoic acid at both 300 and 500 oC. The pyrolyzed gas samples were collected with a methanol wash and analyzed in the

GC/MS that was calibrated with these compounds. The resulting average recovery rate for the mesh reactor setup was ~72%; this value was used as a correction factor when calculating the production yields.

Figure 3.1 Mesh reactor illustration and sample loading pattern

Fast Speed Camera

Figure 3.2 shows a fast speed camera coupled with the mesh reactor. This set up allows recording the changes in particle size during the fast heating of the samples. The camera is set at

60o to the reaction surface. The chamber removed the top lid and changed the flushing gas from nitrogen to argon to protect the samples from oxidation. A piece of quartz glass was placed on the mesh to prevent small particle from falling through mesh openings, so the reaction takes place on the quartz rather than on the wire mesh itself. If the small particles or the liquid sunk below the mesh, accurate reaction times cannot be taken.

A constant light source (USHIO’s Sōlarc® Fiber-Optic Illuminators LB50) was used to obtain better images. Meshes and quartz glass were pre-heated for 10 s. Camera then started recording and samples were sprinkled from top. The videos were recorded at 250 frames/s. The videos were analyzed in Midas 4.0 express to determine reaction time of the samples by counting the particle reacting frames on quartz glass.

Figure 3.2 The working status of fast speed camera.

3.2.3 Analytical Methods

Scanning Electronic Microscope (SEM)

SEM of the solid residue obtained was performed on an FEI 200F SEM system with large field detector and a low vacuum environment (130 Pa). High voltage was set at 30kV and the magnification was set from 1000X. Solid residues obtained from the pyrolysis of cellulose at 300 and 375 oC were analyzed.

GC/MS

The samples dissolved in methanol were first analyzed by GC/MS (Agilent 7890A equipped with HP-5 MS column, 30m×0.250mm, 0.25 micron and 5975C inert XL EI MS detector). 2 µL of sample was injected every experiment. The split ratio was set at 5 and 1 mL/min of He was used to flush the column. The inlet was heated to 200 oC. Column was kept at 40 oC for 1 minute and then heated to 190 oC at a heating rate of 3 oC/min. Then the temperature of the oven was ramped up to 280 oC with heating rate of 20 oC/min and then kept for 10 minutes to remove any product left over in the column. Concentrations of levoglucosan, acetol and glycolaldehyde were calibrated with standards.

HPLC

Products dissolved with solvent methanol were vacuum dried and carefully weighed. The dried samples were re-dissolved in E-pure water and injected in a Varian Prostar 230 HPLC with

Varian Prostar 350 RI (refractive index) as described elsewhere [25]. The water flow rate used with 0.4 mL/min on a Bio-Rad Aminex HPX-87P column. Levoglucosan and cellobiosan were calibrated with standard samples.

3.3 Results and Discussions

3.3.1 Fast speed camera experiments in the wire mesh reactor.

The formation of liquid intermediates at very high heating rate conditions of small particles in the wire mesh reactor was observed with a fast speed camera. The time the biomass particle was pyrolyzed at temperatures between 300 and 500 oC was measured by counting frames. The plot of reaction time vs. effective diameter is shown in Figure 3.3. The calculation of effective diameter followed the method described by Dauenhauer et al. [30]. As shown in Figure 3.3, for particle size below 0.2 mm the total conversion time for both samples was much less dependent of the particle size (especially clear in 300 oC). This result indicates kinetically controlled reaction conditions. The total conversion time of particles with effective diameter smaller than

0.2 mm was plotted in Figure 3.4 as a function of the wire mesh reactor temperatures (300, 375,

450 and 500 oC). The ball-milled cellulose pyrolyzed much faster than the control cellulose

(except 500 oC).

Figure 3.3 Reaction time vs. effective diameter of control and ball-milled cellulose at temperatures between 300 and 500 oC.

Figure 3.4 Reaction time vs. temperature of control and ball-milled cellulose

The data shown in Figure 3.4 was used to estimate the kinetic parameters of the pyrolysis reaction at the fast heating rates conditions studied. For the kinetic analysis we supposed a first order reaction order in which the rate of reaction (d/dt) is a function of the conversion (α) and the constant rate of reaction (k):

Equation 1

The integral solution of this equation between conversion 0 and 0.95 is shown in Equation 2.

Equation 2

The solution of this equation for full conversion (α=1) result in an infinite reaction time. In order to avoid this result we decided to consider that our visual inspection was only equivalent to 95 % of conversion since it was very difficult to distinguish the black char from the background.

Under these conditions it is possible to estimate constant rate of reaction at different temperatures using the following equation:

Equation 3

The values of constant rate of reaction as a function of reaction temperature are plotted in Figure

3.5. This data was used to calculate activation energy (Ea) and pre-exponential factor for both control and ball-milled cellulose (Table 3.1).

Figure 3.5 Reaction rate vs. 1/T

Table 3.1 Values of Activation Energy (Ea) and pre-exponential factor (A) obtained for control and ball-milled cellulose using a fast speed camera Control Ball-milled

Ea (kJ/mol) 81.3 37.7 log A (log (s-1)) 5.8 2.7

Compared to the literature by other authors (180-245 kJ/mol [31], 140-218 kJ/mol [32], 222 kJ/mol for control and 162 kJ/mol for ball-milled [19], 231 kJ/mol [20], 220 kJ/mol [33]) these values are relatively low, but are actually quite close to the values of our previous work (300-400 oC, 86.0 kJ/mol and 5.1 log(min-1) for control cellulose and 52.4 kJ/mol, 3.3 log(min-1) for ball- milled cellulose) [28]. Since the test here could actually be considered as isothermal condition with very small particle size (<50µm), the thermal lag between heating surface and particle should be much smaller than in the spoon reaction we used.

One assumption made was that below 0.2 mm particle size the reaction is conducted under a kinetically controlled regime. However, it is possible that other controlling phenomena are in play. For example, in the video recording of the reaction, those small particles clearly bounce around and that bouncing was a source of significant variation in that small particle region, as seen in Figure 3.3. These phenomena are well studied in the transport phenomena field as interactions between droplets and hot walls [34].

3.3.2 Analysis of solid residue

Figure 3.6 shows the solid residue obtained from control cellulose and ball-milled cellulose Control Ball-milled obtained from 300, 375, 450 and 500 oC. At 300 oC, residue of control cellulose was prepared by

sticking it off the mesh with the SEM stud. The fibril structure was maintained after the heat

treatment. At 375 oC, the residue had a bubbled structure in it indicating the formation of a liquid

intermediate.

Ball-milled cellulose, instead, melted from 300 oC. Bubbles could be found at the openings of the

mesh and small residues of crystalline kernels were observed. Different from the residue of

control cellulose, these residue bubbles were tightly stuck to the wires and could not be

transferred on to the stud. As was observed under fast speed camera, ball-milled cellulose rapidly

transformed into liquid intermediate and boiled along the mesh. The difference between 300 and

375 oC of ball-milled residue was that fewer and smaller bubbles could be found.

Control cellulose at 300 oC Ball-milled cellulose at 300 oC

Bubble

Crystallite

375 oC 375 oC

Bubble Bubble

Crystallite

450 oC 450 oC

Crystallite Crystallite

Figure 3.6 Residue of control (left) and ball-milled (right) cellulose under SEM.

With temperature reaching 450 oC, both cellulose samples left small crystallites residue with no residue bubbles. Same situation could also be found in 500 oC samples. A competition between residue-leaving reactions (deep cross-linking/dehydration and polycondensation) and weight loss reactions could be concluded. Compounds too heavy to be evaporated were left on hot surface and continued their pathway towards heavier and more stable carbonaceous residue.

3.3.3 Analysis of Volatile Products

Figure 3.7 shows the HPLC chromatographs of condensates derived from the pyrolysis of control and ball-milled cellulose between 300 and 500 oC. In the chromatograph it was possible to observe the formation of levoglucosan and an unknown peak triplet at residence time of ~24.2 min which had a same retention time as 1,6-anhydro-glucofuranose. Although the unknown peak has the same residence time that the 1,6-anhydro-glucofuranose, it cannot be assigned to this compound because we were not able to find it in the GC/MS chromatogram (Figure 3.8). Under the analytical conditions used the 1,6-anhydro-glucofuranose is typically found at 38 min.

Surprisingly, no cellobiosan formation was observed. This result contrasts with a yield of cellobiosan close to 10-16 wt. % reported in vacuum pyrolysis tests by Westerhof et al [41] and earlier result of 6-15 wt% [35, 36].

The GC/MS chromatogram of cellulose control pyrolysis (500 oC) products washed with methanol can be seen in Figure 3.8. The chromatograms for other temperatures and for the ball- milled had the same peaks with different heights. Note that the cellobiosan and the 1,6-anhydroglucofuranose peaks did not show up here. The absence of the cellobiosan peak corresponds well with the HPLC. However, the absence of the unknown peak in GC/MS confirms that it is not the

1,6-anhydroglucofuranose and that it should be a relatively heavy compound that can be evaporated at the pyrolysis temperatures studied (300-500 oC) but not at the analytical injection conditions (around 200 oC) with limited amount of methanol flushing through.

Figure 3.8. Chromatogram of control cellulose products from pyrolysis at 500 oC

The yields of levoglucosan obtained by GC/MS and by HPLC, the unknown compound(s) by

HPLC and the solid residue obtained as a function of pyrolysis temperature are shown in Figure

3.9. The maximum yield of almost 60% levoglucosan was achieved by both control and ball- milled cellulose at 300 oC. The unknown compound(s) (estimated with the calibration of levoglucosan) was found slightly decrease with temperature increasing (from 22 to 19 wt. %) in control cellulose, but slightly increased in ball-milled cellulose (from 21 to ~25 wt. %). The yield of products obtained from control cellulose and from ball-milled cellulose were very similar suggesting that cellulose crystallinity has very limited effect on the outcome of fast pyrolysis under atmospheric pressure. In both cases the yield of levoglucosan decreases as the pyrolysis temperature increases. Minor peaks could be found as dehydrated compounds, which, by using the calibration of levoglucosan, accounted for less than 1% of total yield. The yield of solid residue obtained in both cases was lower than 6% within the temperature range tested, which is usually believed to be higher within the region of 300-375 oC.

100

Figure 3.9 Yield of levoglucosan (by HPLC and GC/MS) and unknown compound(s) (by HPLC) from pyrolysis of control cellulose and ball-milled cellulose

3.3.4 Pyrolytic behavior of Cellobiosan and Levoglucosan

In order to understand the mechanism of responsible for the formation of the unknown peak detected in the HPLC studies of cellulose pyrolytic condensate and the origin of the solid residue formed which was observed by SEM, we decided to study the pyrolysis of two of the main cellulose primary products (levoglucosan and cellobiosan). One of the major products of cellulose pyrolysis (levoglucosan), has a boiling point of ~300 oC at atmospheric pressure, as reported [37-39]. The boiling point of the dimer (cellobiosan) is ~581 oC, and the trimer ~792 oC

[39, 40]. Although in most pyrolysis cases, the temperature will not exceed the range of 500-600 oC, cellobiosan and larger oligomers could still be found with fast pyrolysis [23, 29] which were probably released from the liquid intermediate through thermal ejection (spitting) [29].

Figure 3.10 shows the GC chromatograms of cellobiosan and levoglucosan products. The

GC/MS chromatogram of levoglucosan shows the peak of levoglucosan only. In addition to

101

levoglucosan; glycolaldehyde, acetol, and cellobiosan were also observed in the pyrolysis of cellobiosan. A second unknown compound which was not found in either cellulose or levoglucosan pyrolysis products GC/MS chromatograms was found to be derived from the pyrolysis of cellobiosan. The mass spectra of this unknown compound are shown in Figure 3.11.

The fragmentation pattern of the unknown peak looks similar to that of levoglucosenone and may be a partially fragmented oligomer.

A B

Figure 3.10 GC chromatogram of (A) levoglucosan and (B) cellobiosan pyrolysis products at 300 oC

“Unknown” peak

fragmentation pattern

Figure 3.11 Mass fragmentation pattern of the second unknown peak in cellobiosan pyrolysis products

102

The HPLC chromatogram obtained for the pyrolytic products of levoglucosan and cellobiosan is shown in Figure 3.12. Levoglucosan is the well-known major product from cellulose pyrolysis and is detected in the chromatograph. However, levoglucosan samples revealed no cellobiosan formation (retention time: 20.8 min). Instead, the peak of first unknown compound(s) was clearly observed in the range of 21-27 min.

Figure 3.12. HPLC chromatogram of products from pyrolysis of levoglucosan and cellobiosan

While the pyrolysis of levoglucosan results in the formation of very little quantities of this unknown compound(s) and no cellobiosan; the pyrolysis of cellobiosan results in the formation of very small quantities of levoglucosan, large quantities of the unknown compound(s), small quantity of cellobiosan and small quantities of other compounds that are likely produced by dehydration reactions of glucose ring. The HPLC result also shows the presence of a second unknown peak (that may or not be) the compound identified by GC/MS.

The yields of products obtained from levoglucosan and cellobiosan are shown in Figure 3.13.

The yields reported were obtained by calibrating the HPLC with standards of levoglucosan (~70 min) and cellobiosan (20.8 min). At 300 oC 92.5 wt. % of the levoglucosan is recovered as levoglucosan, 4.4 wt. % is recovered as the unknown sugar and 2.3 wt. % is recovered as a solid

103

residue. In the case of cellobiosan, 8.3 wt. % is recovered as levoglucosan, ~30 wt. % is recovered the unknown compound(s) and 18.2 wt. % is recovered as a solid residue.

Figure 3.13 Yield of levoglucosan (by HPLC and GC/MS) and unknown compound(s) (by HPLC) of cellobiosan, with area/mass of products from cellobiosan (by GC/MS)

The yield of levoglucosan decreased as the temperature increased, while the yield of the unknown compound(s) remained nearly constant with temperature. Levoglucosan had the lowest yield of the first unknown compound(s), while cellobiosan had highest yield of this compound(s).

In all cases, our experiments did not reveal significant amounts of products from fragmentation reactions (<2 wt. % in total). Although our system uses liquid nitrogen as a cooling agent, the efficiency of trapping light compounds from fragmentation reactions might still be low due to the purge gas nitrogen and vacuum pull. The decrease in levoglucosan yield is perhaps due to the fragmentation or the dehydration of this molecule. Water was another big portion of product which could not be measured in our current setting. The results obtained suggest that cellobiosan is too heavy to be rapidly evaporated at atmospheric pressure with temperature up to 500 oC.

This keeps it on the hot surface where it undergoes secondary reactions (crosslinking, dehydration, fragmentation) to form secondary products, which could not be identified in our

104

experiment. The cellobiosan found in cellobiosan pyrolysate were most probably results of forming an azeotrope or thermal ejection [29, 30].

Figure 3.14 shows the SEM picture of the solid residue formed during the pyrolysis of cellobiosan. The pictures at low temperature show the formation of a liquid intermediates and bubbles. As the temperature increased, the liquid layer could still be found covering the wire.

Considering the high boiling point predicted at ~581 oC [39, 40], it’s logical to conclude that 1.

Un-boiled liquid has a high tendency to cross-link/dehydrate/polycondense, thus forms char and unknown compounds (liked products of dehydration reactions); 2. Fiber structure of control cellulose was actually protected by the low temperature of liquid intermediate on the surface. 3.

Thermal ejection of oligo-sugars needs low boiling point compounds (in this work, levoglucosan did the job) involved to form an azeotrope to be released. In all the samples, we observe the formation of crystal residues. This result suggests that the crystals observed during the pyrolysis of cellulose are products of cellobiosan secondary reactions.

105

300 oC 375 oC

Crystal Crystal

Bubble Bubble

450 oC 500 oC Bubble Crystal

Bubble

Crystal

Figure 3.14 SEM pictures of mesh exhibit a thin layer of carbonaceous residue after cellobiosan pyrolysis (contrast and brightness adjusted)

3.4 Conclusions

The primary products of cellulose pyrolysis (levoglucosan, cellobiosan, and heavier anhydrosugars) form a liquid intermediate under high heating rates that evaporates, be thermally ejected from the solid/liquid phase, or further cross-link and polycondense to form bio-oil and a solid residue. Ball-milled cellulose pyrolyzed and evaporated much faster than the control

106

cellulose; however, these phenomena do not significantly affect the yield of levoglucosan produced. Our experimental results highlight the importance of the thermal behavior of cellobiosan (and likely the heavier anhydrosugars derived from cellulose) under atmospheric pressure. At atmospheric pressure, cellobiosan cannot evaporate as they do under vacuum. The formation of levoglucosan from cellobiosan (and perhaps from the other anhydrosugars) may explain why the yield of levoglucosan obtained under vacuum is much lower than the yield obtained at atmospheric pressure. Most of the levoglucosan formed in the primary reactions can be evaporated at atmospheric pressure in the wire mesh reactor with a very small fraction converted into the unknown compound and solid residue. Higher yields of hydrolysable

(fermentable) sugars could be obtained if cellobiosan secondary reactions can be mitigated using vacuum or by other methods.

Acknowledgement

This project was financially supported by the US National Science Foundation (CBET-0966419,

CAREER CBET-1150430), the Sun-Grant Initiative (Interagency Agreement: T0013G-A), and the Washington State Agricultural Research Center.

107

Reference

[1] M. Hoogwijk, A. Faaij, R. van den Broek, G. Berndes, D. Gielen, W. Turkenburg, Exploration of the ranges of the global potential of biomass for energy, Biomass and Bioenergy, 25 (2003) 119-133.

[2] M. Hoogwijk, A. Faaij, B. Eickhout, B. de Vries, W. Turkenburg, Potential of biomass energy out to 2100, for four IPCC SRES land-use scenarios, Biomass and Bioenergy, 29 (2005) 225-257.

[3] M. Wei, S. Patadia, D.M. Kammen, Putting renewables and energy efficiency to work: How many jobs can the clean energy industry generate in the US?, Energy Policy, 38 (2010) 919-931.

[4] S. Kersten, M. Garcia-Perez, Recent developments in fast pyrolysis of ligno-cellulosic materials, Current Opinion in Biotechnology, 24 (2013) 414-420.

[5] D. Mohan, C.U. Pittman, P.H. Steele, Pyrolysis of Wood/Biomass for Bio-oil: A Critical Review, Energy & Fuels, 20 (2006) 848-889.

[6] N.M. Bennett, S.S. Helle, S.J.B. Duff, Extraction and hydrolysis of levoglucosan from pyrolysis oil, Bioresource Technology, 100 (2009) 6059-6063.

[7] L. Jarboe, Z. Wen, D. Choi, R. Brown, Hybrid thermochemical processing: fermentation of pyrolysis-derived bio-oil, Applied Microbiology and Biotechnology, 91 (2011) 1519-1523.

[8] J. Lian, S. Chen, S. Zhou, Z. Wang, J. O’Fallon, C.-Z. Li, M. Garcia-Perez, Separation, hydrolysis and fermentation of pyrolytic sugars to produce ethanol and lipids, Bioresource Technology, 101 (2010) 9688-9699.

[9] J. Lian, M. Garcia-Perez, S. Chen, Fermentation of levoglucosan with oleaginous yeasts for lipid production, Bioresource Technology, 133 (2013) 183-189.

[10] D.J. Nowakowski, J.M. Jones, Uncatalysed and potassium-catalysed pyrolysis of the cell- wall constituents of biomass and their model compounds, Journal of Analytical and Applied Pyrolysis, 83 (2008) 12-25.

[11] H. Kawamoto, D. Yamamoto, S. Saka, Influence of neutral inorganic chlorides on primary and secondary char formation from cellulose, Journal of Wood Science, 54 (2008) 242-246.

[12] N. Shimada, H. Kawamoto, S. Saka, Different action of alkali/alkaline earth metal chlorides on cellulose pyrolysis, Journal of Analytical and Applied Pyrolysis, 81 (2008) 80-87.

108

[13] C. Di Blasi, A. Galgano, C. Branca, Influences of the Chemical State of Alkaline Compounds and the Nature of Alkali Metal on Wood Pyrolysis, Industrial & Engineering Chemistry Research, 48 (2009) 3359-3369.

[14] S.R.G. Oudenhoven, R.J.M. Westerhof, N. Aldenkamp, D.W.F. Brilman, S.R.A. Kersten, Demineralization of wood using wood-derived acid: Towards a selective pyrolysis process for fuel and chemicals production, Journal of Analytical and Applied Pyrolysis.

[15] F. Shafizadeh, T.T. Stevenson, Saccharification of douglas-fir wood by a combination of prehydrolysis and pyrolysis, Journal of Applied Polymer Science, 27 (1982) 4577-4585.

[16] F. Shafizadeh, Y.L. Fu, Pyrolysis of cellulose, Carbohydrate Research, 29 (1973) 113-122.

[17] F. Shafizadeh, Introduction to pyrolysis of biomass, Journal of Analytical and Applied Pyrolysis, 3 (1982) 283-305.

[18] J. Piskorz, D. Radlein, D.S. Scott, On the mechanism of the rapid pyrolysis of cellulose, Journal of Analytical and Applied Pyrolysis, 9 (1986) 121-137.

[19] Z. Wang, A.G. McDonald, R.J.M. Westerhof, S.R.A. Kersten, C.M. Cuba-Torres, S. Ha, B. Pecha, M. Garcia-Perez, Effect of cellulose crystallinity on the formation of a liquid intermediate and on product distribution during pyrolysis, Journal of Analytical and Applied Pyrolysis, 100 (2013) 56-66.

[20] H.B. Mayes, L.J. Broadbelt, Unraveling the Reactions that Unravel Cellulose, The Journal of Physical Chemistry A, 116 (2012) 7098-7106.

[21] R. Vinu, L.J. Broadbelt, A mechanistic model of fast pyrolysis of glucose-based carbohydrates to predict bio-oil composition, Energy & Environmental Science, 5 (2012) 9808- 9826.

[22] F. Shafizadeh, R.H. Furneaux, T.G. Cochran, J.P. Scholl, Y. Sakai, Production of levoglucosan and glucose from pyrolysis of cellulosic materials, Journal of Applied Polymer Science, 23 (1979) 3525-3539.

[23] W. Chaiwat, I. Hasegawa, T. Tani, K. Sunagawa, K. Mae, Analysis of Cross-Linking Behavior during Pyrolysis of Cellulose for Elucidating Reaction Pathway, Energy & Fuels, 23 (2009) 5765-5772.

[24] H. Kawamoto, M. Murayama, S. Saka, Pyrolysis behavior of levoglucosan as an intermediate in cellulose pyrolysis: polymerization into polysaccharide as a key reaction to carbonized product formation, Journal of Wood Science, 49 (2003) 469-473.

[25] X. Bai, P. Johnston, S. Sadula, R.C. Brown, Role of levoglucosan physiochemistry in cellulose pyrolysis, Journal of Analytical and Applied Pyrolysis, 99 (2013) 58-65.

109

[26] X. Bai, P. Johnston, R.C. Brown, An experimental study of the competing processes of evaporation and polymerization of levoglucosan in cellulose pyrolysis, Journal of Analytical and Applied Pyrolysis, 99 (2013) 130-136.

[27] G.-J. Kwon, S. Kuga, K. Hori, M. Yatagai, K. Ando, N. Hattori, Saccharification of cellulose by dry pyrolysis, Journal of Wood Science, 52 (2006) 461-465.

[28] Z. Wang, B. Pecha, R.J.M. Westerhof, S.R.A. Kersten, C.-Z. Li, A.G. McDonald, M. Garcia-Perez, Effect of Cellulose Crystallinity on Solid/Liquid Phase Reactions Responsible for the Formation of Carbonaceous Residues during Slow Pyrolysis, (Submitted to Industrial & Engineering Chemistry Research).

[29] A.R. Teixeira, K.G. Mooney, J.S. Kruger, C.L. Williams, W.J. Suszynski, L.D. Schmidt, D.P. Schmidt, P.J. Dauenhauer, Aerosol generation by reactive boiling ejection of molten cellulose, Energy & Environmental Science, 4 (2011) 4306-4321.

[30] P.J. Dauenhauer, J.L. Colby, C.M. Balonek, W.J. Suszynski, L.D. Schmidt, Reactive boiling of cellulose for integrated catalysis through an intermediate liquid, Green Chemistry, 11 (2009) 1555-1561.

[31] J.J.M. Orfão, F.J.A. Antunes, J.L. Figueiredo, Pyrolysis kinetics of lignocellulosic materials—three independent reactions model, Fuel, 78 (1999) 349-358.

[32] I. Milosavljevic, E.M. Suuberg, Cellulose Thermal Decomposition Kinetics: Global Mass Loss Kinetics, Industrial & Engineering Chemistry Research, 34 (1995) 1081-1091.

[33] M. Grønli, M.J. Antal, G. Várhegyi, A Round-Robin Study of Cellulose Pyrolysis Kinetics by Thermogravimetry, Industrial & Engineering Chemistry Research, 38 (1999) 2238-2244.

[34] A. Karl, A. Frohn, Experimental investigation of interaction processes between droplets and hot walls, Physics of fluids, 12 (2000) 785.

[35] D.S.T.A.G. Radlein, A. Grinshpun, J. Piskorz, D.S. Scott, On the presence of anhydro- oligosaccharides in the sirups from the fast pyrolysis of cellulose, Journal of Analytical and Applied Pyrolysis, 12 (1987) 39-49.

[36] J.A. Lomax, J.M. Commandeur, P.W. Arisz, J.J. Boon, Characterisation of oligomers and sugar ring-cleavage products in the pyrolysate of cellulose, Journal of Analytical and Applied Pyrolysis, 19 (1991) 65-79.

[37] W. Feng, H.J. van der Kooi, J. de Swaan Arons, Application of the SAFT equation of state to biomass fast pyrolysis liquid, Chemical Engineering Science, 60 (2005) 617-624.

[38] I. Milosavljevic, V. Oja, E.M. Suuberg, Thermal effects in cellulose pyrolysis: Relationship to char formation processes, Industrial and Engineering Chemistry Research, 35 (1996) 653-662.

110

[39] J. Lédé, J.P. Diebold, G.V.C. Peacocke, J. Piskorz, The nature and properties of intermediate and unvaporized biomass pyrolysis materials, in: A.V. Bridgwater, D.G.B. Boocock (Eds.) Developments in Thermochemical Biomass Conversion, 1997, pp. 27-42.

[40] V. Mamleev, S. Bourbigot, M. Le Bras, J. Yvon, The facts and hypotheses relating to the phenomenological model of cellulose pyrolysis: Interdependence of the steps, Journal of Analytical and Applied Pyrolysis, 84 (2009) 1-17.

[41] R.J.M. Westerhof, Z. Wang, S.R.G. Oudenhoven, M. Garcia-Perez, S.R.A. Kersten, Primary products from pyrolysis of cellulose in a vacuum wire mesh reactor. (To be submitted to ChemSusChem, 2013)

111

Chapter 4 Understanding the Effect of Sulfuric Acid on the

Pyrolysis of Acid Washed Douglas fir in an Atmospheric Wire Mesh

Reactor

(Paper to be submitted to Energy and Fuels)

Zhouhong Wang 1, Shuai Zhou1, Brennan Pecha 1, Roel J.M. Westerhof 2, Sascha R.A. Kersten

2, Manuel Garcia-Perez 1*

1Biological Systems Engineering, Washington State University, Pullman, WA 99164, USA

2Thermo-Chemical Conversion of Biomass Group, Faculty of Science and Technology,

University of Twente, Postbus 217, 7500AE Enschede, The Netherlands

Abstract: This paper reports the differences in yield of levoglucosan between acid washed

Douglas fir, Avicel (control), and ball-milled Avicel in the range of temperature between 300 and 500 oC. At 300 oC the yield of levoglucosan obtained from Avicel and ball-milled Avicel were higher (close to 60 wt. %) than the yield obtained from Douglas fir (close to 45 wt. % on cellulose basis). The main reason for the lower levoglucosan yield obtained when processing lignocellulosic materials is unknown but some authors suggest it is due to undesirable interactions between cellulose and the other organic constituents in lignocellulosic materials. The effect of sulfuric acid on levoglucosan production from acid washed Douglas fir was studied at two temperatures 300 and 500 oC and sulfuric acid concentrations up to 0.6 wt. %. A very important finding of this study is the existence of a sulfuric acid concentration (0.04 wt. %) at which the yield of levoglucosan reaches the expected value for cellulose (close to 60 wt. % at

112

300 oC and close to 50 wt. % at 500 oC). Pyrolysis studies in the presence of sulfuric acid were conducted with Avicel, ball-milled Avicel, and the two main primary products of cellulose pyrolysis (levoglucosan and cellobiosan) to investigate the cause of this increase in levoglucosan yield. GC/MS and HPLC were employed to analyze the liquid produced. For all the studies (with

Avicel, ball-milled Avicel, levoglucosan and cellobiosan) sulfuric acid acted as a dehydration agent contributing to the reduction of levoglucosan yield and the formation of 1,6-anhydro-β-D- glucofuranose. The increase in sulfuric acid concentration also contributed to the formation of an unknown compound quantified by HPLC. The results obtained with Avicel, ball-milled Avicel,

Levoglucosan and cellobiosan confirm that the increase in levoglucosan yield observed when small quantities of sulfuric acid was added to acid wash Douglas fir was not due to its effect on the cellulose depolymerization mechanism. The increase observed could be due to its effect on the poorly known interactions between cellulose and the other organic constituents of lignocellulosic materials or due to the passivation of the remaining alkalines (0.013 wt. %) in the biomass.

Key words: Levoglucosan, cellobiosan, cellulose pyrolysis

Corresponding Author:

Dr. Manuel Garcia-Perez

Associate Professor

LJ Smith Hall, Room 205

Pullman, WA, 99164-6120

Phone: 509-335-7758, Fax: 509-335-2722

113

e-mail: [email protected]

114

4.1 Introduction

Biomass is an attractive sustainable source for the production of transportation fuels [1]. Fast pyrolysis is among the most attractive strategies for converting lignocellulosic materials into green fuels and chemicals. This technology is able to convert up to 75 wt. % of lignocellulosic materials into a crude bio-oil [2, 3] that can be further refined to produce gasoline and chemicals.

A recent fast pyrolysis study in a vacuum mesh reactor has shown [40] that almost all the cellulose can be converted into hydrolysable fermentable sugars. At 500 oC, 4-6 wt. % is converted into levoglucosan; 7-8 wt. % is converted into cellobiosan; 5-8 wt. % is converted into cellotriosan; and more than 70 wt. % is converted into heavier anhydrosugars. We have recently proven [4] that under atmospheric pressure, the cellobiosan and probably also the heavier anhydrosugars produced by cellulose primary reactions, are converted by secondary reactions to levoglucosan as well as to other poorly unknown secondary products and char that reduce the yield of fermentable sugars. A small fraction of levoglucosan is also lost reducing its yield to close to 60 wt. %.

Similar levoglucosan yields (close to 60 %) have also been reported when cellulose is pyrolyzed at slow pyrolysis conditions under vacuum [5]. The levoglucosan and other hydrolysable sugars can be utilized as a nutrition source for microbe fermentation, and produce lipid and ethanol [2,

6-9].

115

Much lower yields of levoglucosan are obtained from lignocellulosic materials (1.89 wt. % [10],

17.3 wt. % [11], 2-3 wt. % [12], 1.8-3.9 wt. % [13], 4-10 wt. % [14]). One major reason for the reduced yield in lignocellulosic materials is the presence of alkali and alkaline earth metal

(AAME) which induces fragmentation reactions even at very small amount [15-18]. So far two strategies have been proposed to mitigate the undesirable effect of AAMEs. The first one consists on the removal of these AAME by washing with aqueous solutions containing [19]. The second strategy consists on the passivation of the catalytic effect of AAME by adding small quantities of strong acid able to form stable salts [20, 21]. These strategies result in increases in levoglucosan yields from lignocellulosic materials but the yields obtained are still lower than those obtained from pure cellulose. The reasons for this phenomenon are not well known but there is abundant information in the literature suggesting that levoglucosan formation is affected by the presence of other organic compounds in lignocellulosic materials [16, 22-26].

In 1982 Shafizadeh and Stevenson [21] observed an important increase in the yield of levoglucosan when acid wash pine wood was pyrolyzed in the presence of sulfuric acid. The author noted that this increase in levoglucosan yield only happened when lignin was present however, he recognizes that the mechanism for these phenomena is not clear [21]. Radlein et al

(1991) [27] also studied the effect of sulfuric acid on the yield of sugars and observed a similar phenomenon. Cellulose conversions into levoglucosan as high as 53.9 wt. % was obtained when

0.1 wt. % of sulfuric acid was used. In a preliminary semi-quantitative Py-GC/MS study we [28,

29] also observed an important increase in the yield of levoglucosan when sulfuric acid was

116

added [29]. The main cause for the phenomena is not known. The effect of crystallinity on the response of cellulose to acids is also very poorly understood.

Thus, the main purpose of this paper is to understand the effect of sulfuric acid on the pyrolysis of acid wash Douglas fir and the effect of this additive on the pyrolysis of amorphous and crystalline cellulose as well on its effect on cellulose primary depolymerization products

(levoglucosan and cellobiosan).

4.2 Materials and Methods

4.2.1 Materials

The Douglas fir feedstock was harvested from the cascade mountain range in Washington State,

USA, and kindly provided by Herman Brothers Logging & Construction (Port Angeles, WA).

The feedstock was grounded by hammer mill (Model number 400 HD, serial 2404, Bliss

Industries) in the Composite Material and Engineering Center at Washington State University and sieved to a 2 mm fraction or less. This woody material was then ball-milled at 300 rpm for

30 hours (Across International PQ-N2, ceramic 100 mL jar and balls). Then, the ball-milled

Douglas fir powder was immersed in 0.1% H2SO4 (w:w=1:10) over night to dissolve the ash.

Then the liquid was disposed and the sample was washed by E-pure water (w:w=~1:200-300) to remove acid and dissolved ash. The acid level was monitored by electric conductivity meter until the conductivity remained constant after three washes. This feedstock has ~46% of cellulose as reported elsewhere [29, 30]. Ash content of Douglas fir particles was 0.24 wt. % (Figure 4.1).

117

After ball-milling, this value increased to 0.76 wt. % with ceramic particles introduced. With acid and water washing, this value decreased to 0.52 wt. %, which removed almost all the ash, and remained almost only the stable ceramic particles. ICP-MS test confirmed that the ash remaining had 0.44 wt. % of Na, 0.18 wt. % of Mg, 0.29 wt. % of K and 0.72 wt. % of Ca, which converted to their contents in washed biomass were all less than 0.01 wt. % of washed powder and added up to 0.013 wt. %.

Figure 4.1 Ash content of Douglas fir particle, after ball-milling and after ball-milling and washing

Cellulose samples used were described in previous studies [31, 32]. Briefly, the Avicel (PH-101) was purchased from Sigma-Aldrich and the ball-milled Avicel was obtained by ball-milling control cellulose for 24 hours. Levoglucosan and cellobiosan samples were purchased from

Carbosynth.

118

Samples (Avicel and ball-milled Avicel and levoglucosan) were impregnated with sulfuric of different weight. 1 wt. % of sulfuric acid solution were prepared for this addition to prevent fast dehydration from sulfuric acid. Approximately 0.07, 0.13, 0.26, 0.40 and 0.60 wt. % of acid was added into samples. Tests with cellobiosan were conducted at four levels: 0, 0.07, 0.13, 0.32 wt. %. Acid added samples were left under vacuum to dry for a week.

4.2.2 Mesh Reactor

The mesh reactor has been described in our previous work [4]. Briefly, samples were sandwiched between two meshes with one additional mesh in the bottom, and then clamped between two electrodes. The acid added samples were heated at 300 and 500 oC with a heating rate of 100 oC/s. Collected pyrolysate was washed out by methanol and subjected to GC/MS and HPLC analyses.

4.2.3 Analytical Methods

The GC/MS and HPLC used for analysis were described in our previous work [4]. Pyrolysate dissolved in methanol were injected directly into GC/MS with levoglucosan calibrated.

Levoglucosenone (LVS), 1,4;3,6-dianhydro-glucopyranose (DGP), 1,6-anhydro-β-D- glucofuranose (AGF) and one unknown peak were monitored. Yield of these compounds found in GC/MS were calculated supposing that their response factor was similar to the one obtained with the levoglucosan calibration.

119

Samples in methanol were dried in the hood and then re-dissolved in water. Both amount of samples in methanol and amount of water were recorded for concentration calculation. These re- dissolved samples were injected into HPLC. Levoglucosan (LVG) (5), cellobiosan (CL) (1), 1,6- anhydro-β-D-glucofuranose (AGF) (2), 1,4;3,6-dianhydro-α-D-glucopyranose (DGP) (3), levoglucosenone (LVS) (4) and one unknown peak (6) presumably a product of dehydration reactions were monitored. Yield of these compounds were calibrated separately (1, 5 were calibrated together and 6 was assumed same response factor as levoglucosan). Figure 4.2 shows the position of standards. In HPLC of pyrolysate, at the same position than 2 we always observed

3 or more peaks with one of them being 1,6-anhydroglucofuranose.

120

4.3 Results and Discussions

4.3.1 Effect of temperature on the yield of levoglucosan from Douglas fir, Avicel and ball- milled Avicel

Figure 4.3 shows the HPLC chromatogram for the pyrolysate obtained at temperatures between

300 and 500 oC. Two main peaks observed in all the samples were: (1) levoglucosan and an unknown peak. Levoglucosan is a primary product of cellulose pyrolysis; the second peak is unknown but was found to be derived from cellobiosan secondary reactions [4]. Temperature didn’t change the yield of the unknown peak too much. Cellobiosan was not found in the products. The effect of temperature on the yield of levoglucosan and the unknown peak are also shown in Figure 4.3. The yield of levoglucosan obtained by HPLC and GC/MS were very similar and decreased for all the samples studied (Douglas fir, Avicel and Ball Milled Avicel). The levoglucosan yield (on cellulose base) was much lower (26-44 wt. %) in Douglas fir than for cellulose (36-58 wt. %). The yield of unknown peak also on cellulose basis (43-51 wt. %) for

Douglas fir was twice the yield estimated for cellulose (18-27 wt. %). The main cause for the lower levoglucosan yield and the higher yield of the unknown compound(s) in the lignocellulosic material is not known but the literatures suggest it may be due to cellulose-lignin interactions [21,

27].

121

Figure 4.3 Effect of temperature on the yield of levoglucosan from Douglas fir, Avicel and Ball- Milled Avicel pyrolysis by mesh reactor (pyrolysis result of Avicel and ball-milled Avicel were adapted from our previous work [4]).

4.3.2 Effect of sulfuric acid concentration on the yield products from Douglas fir

Figure 4.4 shows the GC/MS chromatogram (normalized) of the pyrolysate at different sulfuric acid concentrations. LVG, LVS, DGP and AGF were quantified at different sulfuric acid concentrations. The yield of LVG increased with 0.04-0.08 wt. % of sulfuric acid added. Further additions of sulfuric acid suppressed the formation of this compound. Yield of AGF was not

122

significant in the pyrolysate derived from raw Douglas fir. Addition of sulfuric acid enhanced the production of this compound, as in the case of LVG, at high sulfuric acid concentrations its content decreased. LVS content increased as the concentration of sulfuric acid increased becoming the most significant product in chromatographs. DGP followed a similar trend.

123

Figure 4.4 GC/MS of Douglas fir Pyrolysate at different sulfuric acid concentrations. Where: AGF, DGP, LVS (or its hydrolyzed product in water).

Figure 4.5 shows the HPLC chromatogram of Douglas fir pyrolysate obtained with different sulfuric acid concentrations at 300 and 500 oC. In addition to the two peaks (unknown and LVG)

124

observed in the pyrolysis of cellulose and Douglas fir three new peaks (DGP, LVS and a new unknown) were observed when sulfuric acid was added.

AGF and AGF and Unknown 2 Unknown 2 Unknowns Unknowns LVG DGP LVS DGP LVS LVG

CL CL

Figure 4.5 HPLC of Douglas fir with different sulfuric acid concentration impregnation

The effect of sulfuric acid on the yield of levoglucosan and other products from pyrolysis at 300 and 500 oC is shown in Figure 4.6. Levoglucosan yields obtained by GC/MS are very similar to those obtained by HPLC. A peak in the yield of Levoglucosan was observed when 0.04 wt. % was added. Any further increase in sulfuric acid concentration resulted in a decrease in the yield of this compound. Levoglucosan yield improvement was much higher at 500 oC than at 300 oC.

Considering the cellulose content of this biomass (46.3 wt. %), the yield of levoglucosan actually reached 55 and 50 wt. % at 300 and 500 oC with sulfuric acid concentration at 0.04 wt. %. This value is comparable to that from cellulose (~57 wt. %) at 300 oC and much higher than the value obtained at 500 oC (~37 wt. %). The higher yield compared to the one obtained from cellulose at

500 oC could be due to the hydrolysis of some of the heavier sugars to produce extra levoglucosan. The increase in levoglucosan yield could be due to the passivation of the small quantities of alkalines remaining in the acid wash Douglas fir [33], due to the effect of the acid in

125

the poorly known cellulose-lignin interactions [21, 29] or due to the hydrolysis of some heavy sugars leading to the formation of extra quantities of levoglucosan [34-37].

The yield of residue increased with acid addition (Figure 4.7). The yield of LVS and DGP also followed this increase while yield of AGF followed the trend of levoglucosan. It is worthy noticing that, the AGF was not a noticeable product when Douglas fir was pyrolyzed without acid impregnation. The formation of AGF is also a key route towards furanic compounds. AGF + unknown compounds in HPLC decreased with acid addition while another three peaks increased.

Clearly, sulfuric acid catalyses dehydration reactions. On the other hand, Figure 4.7 shows that, monitered by GC/MS, the yield of AGF also had a peak (~4 wt. %) when lowest sulfuric acid concentration was applied, while higher amount of acid changed lowered this value just like that of LVG. Compared to the value of “unknown peak” obtained by HPLC, obviously, yield of the unknown compound(s) was greatly suppressed even with the addition of AGF after applying acid.

Yield of residue was also increased with additon of sulfuric acid. Early researchers reported a binding effect of sulfate to cellulose and thus enhanced the early dehydration of cellulose which leads to char and low yield of LVG [38]. But comparing to our result, this effect might have been blocked by the existence of lignin and/or hemicellulose which resulted in more LVG production.

Higher concentration of acid (>0.2 wt. %) didn’t seem to increase the yield of char significantly in high temperature experiments, indicating reaction with sulfuric acid might be a catalyzed dehydration/cross-linking reaction which occurs majorly in liquid intermediate [31].

126

127

To investigate the main causes for the maximum yield of levoglucosan at sulfuric acid concentrations of 0.04-0.08 wt. %, in the next section, we study whether this increase in yield can be associated to the behavior of cellulose or its primary products (levoglucosan or cellobiosan).

4.3.3 Effect of sulfuric acid concentration on yield and composition of cellulose and levoglucosan pyrolysis products

Avicel, ball-milled Avicel and levoglucosan were pyrolyzed in mesh reactor and their pyrolysis products analyzed by GC/MS and HPLC. Since pyrolysis of Douglas fir had a much bigger increment (from 12.3 wt. % to 23.6 wt. %) in yield of levoglucosan at 500 oC, the pyrolysis tests on model compounds were conducted at 500 oC. Figure 4.8 shows the HPLC chromatogram and the yield of Levoglucosan, the unknown compound and the DGP, LVS and LVG as a function of sulfuric acid addition. Levoglucosan yields obtained by HPLC and GC/MS were very similar and in all cases decreased as sulfuric acid concentration increases even with less than 0.1 wt. % of sulfuric acid added. Ball-milled Avicel had a more drastic change in yield of levoglucosan than

Avicel indicating stronger action of sulfuric acid either in the primary reaction or on the secondary reactions in the liquid intermediate. The yield of the unknown compound also decreased as sulfuric acid concentration increases, which might indicated the production of these compounds are related to liquid intermediate or even yield of levoglucosan. The yield of DGP and LVS gradually increased as sulfuric acid concentration increases.

128

AGF and Unknown 2 DGP Unknowns LVS LVG

Figure 4.8 HPLC of cellulose and levoglucosan pyrolysate (left) and the evaluation of their yields (right)

The yield of GC/MS detectable compounds is shown in Figure 4.9. LVS, DGP and one unknown

GC/MS peak which was not obvious in Douglas fir pyrolysis was found increasing with more sulfuric acid added. Yield of AGF followed the same trend found in Douglas fir. These results confirm that the addition of sulfuric acid can enhance cellulose dehydration reactions responsible for the formation of anhydrosugars and extra-char. This effect was also discussed under the condition of slow pyrolysis [38] that sulfate could actually bind onto cellulose and push the

129

reactions towards water, char and small molecules. Comparing Avicel, ball-milled Avicel and levoglucosan, the yield of levoglucosan was much less affected in Avicel indicating that the sulfate might attacked the hydroxyl groups on cellulose and levoglucosan, but the existence of hydrogen-bond protected the major structure of cellulose.. On the other hand, the fact that the behavior of cellulose samples (Avicel and ball-milled Avicel) was similar to the behavior observed for levoglucosan (the main product of cellulose degradation reactions) suggest that the intra-molecular dehydration reactions will also happened after the levoglucosan intermediate is formed.

130

Figure 4.9 Effect of sulfuric acid concentration on the yield of residue and dehydrated products (by GC/MS) from Avicel, ball-milled Avicel, and levoglucosan pyrolysis at 500 oC on mesh reactor

Our experimental results also confirm that the increase in levoglucosan yield observed at sulfuric acid concentrations of 0.04-0.08 wt. % is not due to this additive working directly on crystalline and amorphous cellulose (here represented by Avicel and ball-milled Avicel). This effect is different from earlier enhancing effect found with phosphoric acid [35, 37]. Regarding to the chemical property of sulfuric acid, it can be concluded that sulfuric acid acted like a strong

131

dehydrator that could hydrolyze levoglucosan it also seems to catalyze crosslinking and polycondensation reactions leading to the formation of extra char [39].

4.3.4 Effect of sulfuric acid concentration on yield and composition of cellobiosan pyrolysis products

The thermal behavior of cellobiosan in the presence of different concentrations of sulfuric acid was also pyrolyzed in the mesh reactor. Our previous result showed that cellobiosan didn’t produce a lot of levoglucosan, but gave one unknown compound shown in GC/MS and big amount of another unidentified mixture peak that can be clearly shown in HPLC [4]. Here we tested three different sulfuric acid concentrations with cellobiosan. Cellobiosan samples were dissolved in water during the sample preparation and vacuum dried after applying acid.

The potential hydrolysis of cellobiosan in the presence of sulfuric acid was studied by analyzing the unpyrolyzed cellobiosan by GC/MS and HPLC (Figure 4.10 and Figure 4.11). The HPLC chromatograms (Figure 4.10) clearly show that the simple addition of small quantities of sulfuric acid did not resulted in the production of levoglucosan. However, a heavier compound seems to be produced even at the very low concentration of sulfuric acid studied. When the cellobiosan with different concentrations of sulfuric acid was dissolved in methanol and then injecting into

GC/MS (inlet temperature 200 oC), the chromatography showed the formation of levoglucosan and an unknown peak. The apparent contradiction between the HPLC and GC/MS analysis suggests that the cellobiosan containing small quantities of sulfuric acid can be hydrolyzed at the

132

GC/MS inlet temperatures to form levoglucosan and an unknown product previously identified in cellobiosan pyrolysis.

LVG

Figure 4.10 HPLC of unpyrolyzed cellobiosan samples with sulfuric acid (in mass %)

Unknown Without Sulfuric Acid 0.07 mass % product CL

LVG

Unknown 0.13 mass % Unknown 0.32 mass % product product

LVG LVG

Figure 4.11 GC/MS of unpyrolyzed cellobiosan samples with sulfuric acid

133

Cellobiosan pyrolysate analyses in HPLC are shown below (Figure 4.12). After applying acid, yield of residue was not visible for acid concentration at 0.07 and 0.13 wt. %, while 0.32 wt. % of acid increased the char yield to 6.2 wt. % (originally 4.3 wt. %). The yield of levoglucosan was tripled from ~5 wt. % to ~15 wt. % as the sulfuric acid concentration increases perhaps due to the acceleration of acid hydrolysis reactions. The yield of mixture peak (2) was not affected with 0.07 wt. % of sulfuric acid usage but decreased with higher sulfuric acid concentration.

Yields of LVS and DGP were only detected when acid concentration was at 0.32 wt. %.

AGF and Unknowns

LVG Unknown 2 DGP LVS

Figure 4.12 HPLC of cellobiosan pyrolysate (left) and the evaluation of their yields (right)

Figure 4.13 shows the GC/MS chromatography of pyrolysate from cellobiosan with different concentration of acid. It could be observed that the peak of cellobiosan was totally gone after adding acid, while the peak of unknown compound gradually decreased with increasing acid concentration. In agreement with the HPLC findings levoglucosan yield increased with increasing sulfuric acid concentration. Yield of AGF, LVS and DGP was only found with highest concentration tested, indicating the unknown peak in HPLC has a big fraction not detectable in

GC/MS (probably dehydrated or cross-linked dimer/oligo-sugar).

134

Unknown Without Sulfuric Acid Unknown 0.07 mass %

CL LVG LVG

0.13 mass % 0.32 mass % Unknown

LVG

LVS

LVG DGP AGF

Figure 4.13 GC/MS chromatograph of cellobiosan with sulfuric acid pyrolysate

The increase in levoglucosan yield observed when sulfuric acid was added to Douglas fir, could be due to the effect of this additive on cellulose poorly known interactions with organics forming lignocellulosic materials [21], due to its passivation effect on the alkalines remaining after acid hydrolysis or due to the hydrolysis of cellobiosan and heavy anhydrosugars to form extra levoglucosan.

4.4 Conclusions

Our results confirm that the yield of levoglucosan obtained when acid washed Douglas fir is pyrolyzed at 300 and 500 oC in a wire mesh reactor is ~13 and ~11 wt. % respectively lower than the yields obtained for Avicel (control), and ball-milled Avicel at the same temperature. The

135

main reason for the lower levoglucosan yield obtained with acid wash Douglas fir is unknown.

But it can be due to undesirable cellulose-lignin interactions or due to the presence of very small quantities of alkalines. There is an optimum concentration of sulfuric acid at which the yield of levoglucosan increases 11.3 and 7.1 wt. %. Furthermore, the yield by its cellulose content obtained under these conditions is even higher than those obtained from pure cellulose at 500 oC.

This increase in levoglucosan yield cannot be explained by the effect of sulfuric acid on cellulose or its major pyrolysis product levoglucosan. However, the tests with cellobiosan showed that the levoglucosan yield could be increased in the existence of small amount sulfuric acid and high temperature. Analysis showed that cellobiosan was modified by sulfuric acid before pyrolysis, and depolymerization reactions are possible to happen even at the GC/MS inlet temperature at

200 oC. The increase in the yield observed when small quantities of sulfuric acid were added to

Douglas fir could be due to (1) the effect of this additive mitigating the poorly known cellulose- lignin interactions allowing achieve the maximum yield expected for cellulose under the same reaction conditions, (2) its passivating effect on the remaining alkalines or (3) the modification of the heavy anhydrosugars (cellobiosan in this experiment) derived from cellulose primary reactions.

Acknowledgement

This project was financially supported by the US National Science Foundation (CBET-0966419,

CAREER CBET-1150430), the Sun-Grant Initiative (Interagency Agreement: T0013G-A), and

136

the Washington State Agricultural Research Center. The authors are very thankful for their support.

137

Reference

[1] M. Hoogwijk, A. Faaij, R. van den Broek, G. Berndes, D. Gielen, W. Turkenburg, Exploration of the ranges of the global potential of biomass for energy, Biomass and Bioenergy, 25 (2003) 119-133.

[2] S. Kersten, M. Garcia-Perez, Recent developments in fast pyrolysis of ligno-cellulosic materials, Current Opinion in Biotechnology, 24 (2013) 414-420.

[3] D. Mohan, C.U. Pittman, P.H. Steele, Pyrolysis of Wood/Biomass for Bio-oil: A Critical Review, Energy & Fuels, 20 (2006) 848-889.

[4] Z. Wang, B. Pecha, R.J.M. Westerhof, S.R.A. Kersten, M. Garcia-Perez, Effect of Pyrolysis Temperature on the Formation of Anhydrosugars During the Fast Pyrolysis of Cellulose in a Wire Mesh Reactor at Atmospheric Pressure, (In Preparation).

[5] F. Shafizadeh, R.H. Furneaux, T.G. Cochran, J.P. Scholl, Y. Sakai, Production of levoglucosan and glucose from pyrolysis of cellulosic materials, Journal of Applied Polymer Science, 23 (1979) 3525-3539.

[6] N.M. Bennett, S.S. Helle, S.J.B. Duff, Extraction and hydrolysis of levoglucosan from pyrolysis oil, Bioresource Technology, 100 (2009) 6059-6063.

[7] L. Jarboe, Z. Wen, D. Choi, R. Brown, Hybrid thermochemical processing: fermentation of pyrolysis-derived bio-oil, Applied Microbiology and Biotechnology, 91 (2011) 1519-1523.

[9] J. Lian, M. Garcia-Perez, S. Chen, Fermentation of levoglucosan with oleaginous yeasts for lipid production, Bioresource Technology, 133 (2013) 183-189.

[10] A. Demirbaş, Mechanisms of liquefaction and pyrolysis reactions of biomass, Energy Conversion and Management, 41 (2000) 633-646.

[11] R.J.M. Westerhof, N.J.M. Kuipers, S.R.A. Kersten, W.P.M. van Swaaij, Controlling the Water Content of Biomass Fast Pyrolysis Oil, Industrial & Engineering Chemistry Research, 46 (2007) 9238-9247.

[12] E. Hoekstra, R.J.M. Westerhof, W. Brilman, W.P.M. Van Swaaij, S.R.A. Kersten, K.J.A. Hogendoorn, M. Windt, Heterogeneous and homogeneous reactions of pyrolysis vapors from pine wood, AIChE Journal, 58 (2012) 2830-2842.

138

[13] M. Garcia-Perez, S. Wang, J. Shen, M. Rhodes, W.J. Lee, C.-Z. Li, Effects of Temperature on the Formation of Lignin-Derived Oligomers during the Fast Pyrolysis of Mallee Woody Biomass, Energy & Fuels, 22 (2008) 2022-2032.

[14] D. Mourant, Z. Wang, M. He, X.S. Wang, M. Garcia-Perez, K. Ling, C.-Z. Li, Mallee wood fast pyrolysis: Effects of alkali and alkaline earth metallic species on the yield and composition of bio-oil, Fuel, 90 (2011) 2915-2922.

[15] D.J. Nowakowski, J.M. Jones, Uncatalysed and potassium-catalysed pyrolysis of the cell- wall constituents of biomass and their model compounds, Journal of Analytical and Applied Pyrolysis, 83 (2008) 12-25.

[16] H. Kawamoto, D. Yamamoto, S. Saka, Influence of neutral inorganic chlorides on primary and secondary char formation from cellulose, Journal of Wood Science, 54 (2008) 242-246.

[17] N. Shimada, H. Kawamoto, S. Saka, Different action of alkali/alkaline earth metal chlorides on cellulose pyrolysis, Journal of Analytical and Applied Pyrolysis, 81 (2008) 80-87.

[18] C. Di Blasi, A. Galgano, C. Branca, Influences of the Chemical State of Alkaline Compounds and the Nature of Alkali Metal on Wood Pyrolysis, Industrial & Engineering Chemistry Research, 48 (2009) 3359-3369.

[19] S.R.G. Oudenhoven, R.J.M. Westerhof, N. Aldenkamp, D.W.F. Brilman, S.R.A. Kersten, Demineralization of wood using wood-derived acid: Towards a selective pyrolysis process for fuel and chemicals production, Journal of Analytical and Applied Pyrolysis.

[20] N. Kuzhiyil, D. Dalluge, X. Bai, K.H. Kim, R.C. Brown, Pyrolytic Sugars from Cellulosic Biomass, ChemSusChem, 5 (2012) 2228-2236.

[21] F. Shafizadeh, T.T. Stevenson, Saccharification of douglas-fir wood by a combination of prehydrolysis and pyrolysis, Journal of Applied Polymer Science, 27 (1982) 4577-4585.

[22] T. Hosoya, H. Kawamoto, S. Saka, Different pyrolytic pathways of levoglucosan in vapor- and liquid/solid-phases, Journal of Analytical and Applied Pyrolysis, 83 (2008) 64-70.

[23] T. Hosoya, H. Kawamoto, S. Saka, Solid/liquid- and vapor-phase interactions between cellulose- and lignin-derived pyrolysis products, Journal of Analytical and Applied Pyrolysis, 85 (2009) 237-246.

[24] T. Hosoya, H. Kawamoto, S. Saka, Cellulose–hemicellulose and cellulose–lignin interactions in wood pyrolysis at gasification temperature, Journal of Analytical and Applied Pyrolysis, 80 (2007) 118-125.

[25] T. Hosoya, H. Kawamoto, S. Saka, Thermal stabilization of levoglucosan in aromatic substances, Carbohydrate Research, 341 (2006) 2293-2297.

139

[26] O.P. Golova, Chemical Effects of Heat on Cellulose, Russian Chemical Reviews, 44 (1975) 687.

[27] D. Radlein, J. Piskorz, D.S. Scott, Fast pyrolysis of natural polysaccharides as a potential industrial process, Journal of Analytical and Applied Pyrolysis, 19 (1991) 41-63.

[28] S. Zhou, N.B. Osman, H. Li, A.G. McDonald, D. Mourant, C.-Z. Li, M. Garcia-Perez, Effect of sulfuric acid addition on the yield and composition of lignin derived oligomers obtained by the auger and fast pyrolysis of Douglas-fir wood, Fuel, 103 (2013) 512-523.

[29] S. Zhou, D. Mourant, C. Lievens, Y. Wang, C.-Z. Li, M. Garcia-Perez, Effect of sulfuric acid concentration on the yield and properties of the bio-oils obtained from the auger and fast pyrolysis of Douglas Fir, Fuel, 104 (2013) 536-546.

[30] S.-S. Liaw, Z. Wang, P. Ndegwa, C. Frear, S. Ha, C.-Z. Li, M. Garcia-Perez, Effect of pyrolysis temperature on the yield and properties of bio-oils obtained from the auger pyrolysis of Douglas Fir wood, Journal of Analytical and Applied Pyrolysis, 93 (2012) 52-62.

[31] Z. Wang, A.G. McDonald, R.J.M. Westerhof, S.R.A. Kersten, C.M. Cuba-Torres, S. Ha, B. Pecha, M. Garcia-Perez, Effect of cellulose crystallinity on the formation of a liquid intermediate and on product distribution during pyrolysis, Journal of Analytical and Applied Pyrolysis, 100 (2013) 56-66.

[32] Z. Wang, B. Pecha, R.J.M. Westerhof, S.R.A. Kersten, C.-Z. Li, A.G. McDonald, M. Garcia-Perez, Effect of Cellulose Crystallinity on Solid/Liquid Phase Reactions Responsible for the Formation of Carbonaceous Residues during Slow Pyrolysis, (Submitted to Industrial & Engineering Chemistry Research).

[33] P.R. Patwardhan, J.A. Satrio, R.C. Brown, B.H. Shanks, Influence of inorganic salts on the primary pyrolysis products of cellulose, Bioresource Technology, 101 (2010) 4646-4655.

[34] H. Kawamoto, S. Saito, W. Hatanaka, S. Saka, Catalytic pyrolysis of cellulose in sulfolane with some acidic catalysts, Journal of Wood Science, 53 (2007) 127-133.

[35] G. Dobele, T. Dizhbite, G. Rossinskaja, G. Telysheva, D. Meier, S. Radtke, O. Faix, Pre- treatment of biomass with phosphoric acid prior to fast pyrolysis: A promising method for obtaining 1,6-anhydrosaccharides in high yields, Journal of Analytical and Applied Pyrolysis, 68–69 (2003) 197-211.

[36] G. Dobele, G. Rossinskaja, T. Dizhbite, G. Telysheva, D. Meier, O. Faix, Application of catalysts for obtaining 1,6-anhydrosaccharides from cellulose and wood by fast pyrolysis, Journal of Analytical and Applied Pyrolysis, 74 (2005) 401-405.

140

[37] G. Dobele, G. Rossinskaja, G. Telysheva, D. Meier, O. Faix, Cellulose dehydration and depolymerization reactions during pyrolysis in the presence of phosphoric acid, Journal of Analytical and Applied Pyrolysis, 49 (1999) 307-317.

[38] S. Julien, E. Chornet, R.P. Overend, Influence of acid pretreatment (H2SO4, HCl, HNO3) on reaction selectivity in the vacuum pyrolysis of cellulose, Journal of Analytical and Applied Pyrolysis, 27 (1993) 25-43.

[39] D.-Y. Kim, Y. Nishiyama, M. Wada, S. Kuga, High-yield Carbonization of Cellulose by Sulfuric Acid Impregnation, Cellulose, 8 (2001) 29-33.

[40] R.J.M. Westerhof, Z. Wang, S.R.G. Oudenhoven, M. Garcia-Perez, S.R.A. Kersten, Primary products from pyrolysis of cellulose in a vacuum wire mesh reactor. (To be submitted to ChemSusChem, 2013)

141

Chapter 5 Effect of Cellulose Crystallinity on Solid/Liquid Phase

Reactions Responsible for the Formation of Carbonaceous Residues

during Slow Pyrolysis

(Paper submitted to Industrial and Engineering Chemistry Research)

Zhouhong Wang1, Brennan Pecha1, Roel J.M. Westerhof2, Sascha R.A. Kersten2, Chun-Zhu Li3,

Armando G. McDonald4, Manuel Garcia-Perez1*

1 Biological Systems Engineering, Washington State University, Pullman, WA 99164, USA

2Thermo-Chemical Conversion of Biomass Group, Faculty of Science and Technology,

University of Twente, Postbus 217, 7500AE Enschede, The Netherlands

3Fuels and Energy Technology Institute, GPO Box U1987, Curtin University of Technology,

Western Australia, 6845, Australia

4Department of Forest, Rangeland and Fire Sciences, University of Idaho, Moscow, ID 83844,

USA

Abstract: This study reports changes in solid phase composition when samples of control and ball-milled microcrystalline cellulose (Avicel) were subjected to slow pyrolysis in a spoon reactor at temperatures between 240 and 400 oC for up to 120 min. The differences observed between control (crystallinity: 60.5%) and ball-milled cellulose (crystallinity: 6.5%) were used to investigate the effect of crystallinity on solid/liquid phase reactions. The evolution of the morphology of the solid residue collected after each experiment was studied by SEM. The content of cellulose, non-hydrolysable saccharides (cross-linked saccharides) was measured by

142

acid hydrolysis followed by ion exchange chromatography (IEC). The aliphatic, aromatic, furanyl and carbonyl groups as a function of reaction time and temperature were measured by

FTIR, and 13C-NMR spectroscopies. It was found that a liquid layer was formed on the surface of both samples; this caused particle agglomeration at temperatures below 300 oC. At higher temperatures, while the ball-milled cellulose melted completely, the control cellulose conserved its fibrous structure. The formation of C=O and C=C groups typically associated with intra-ring dehydration and associated cross-linking were accelerated by the presence of liquid intermediates derived from the amorphous cellulose. Consequently, a new reaction mechanism and its associated kinetic parameters to describe the changes in the solid residue composition at different reaction conditions are proposed.

Key words: Cellulose, crystallinity, cross-linked saccharides, char, pyrolysis

Corresponding Author:

Dr. Manuel Garcia-Perez

Associate Professor

LJ Smith Hall, Room 205

Pullman, WA, 99164-6120

Phone: 509-335-7758, Fax: 509-335-2722 e-mail: [email protected]

143

5.1 Introduction

With the increasing demand of liquid transportation fuel and the gradual depletion of crude oil, the production of fuels and chemicals from lignocellulosic materials is re-gaining societal interest. An additional advantage of using lignocellulosic materials for bio-fuel production is the potential to reduce greenhouse gases due to its participation in the natural atmospheric carbon cycle [1-3]. The pyrolysis of lignocellulosic materials followed by the refining of the bio-oil produced is one of the most promising alternatives currently studied for the production of transportation bio-fuels [4-6].

Cellulose is the most abundant carbohydrate biopolymer in nature and represents approximately

40 to 45 mass % in dry wood [7]. There is growing interest to increase the pyrolytic production of anhydrosaccharides from cellulose [8]. Pyrolytic anhydrosaccharides (chiefly levoglucosan) can then be hydrolyzed to obtain glucose or can be directly fermented to obtain ethanol or as source for chemicals like furans and levulinic acid [9, 10]. The formation of these saccharides, however, is largely influenced by the process conditions like temperature and heating rate [8]. In addition, understanding how pyrolysis conditions affect the structure of the remaining bio-char is also important for understanding inter-particle reactions responsible for possible decrease in the production of these saccharides and for the development of carbonaceous adsorbents for multiple industrial and environmental applications.

144

Although many researchers have studied cellulose pyrolytic reactions [11-18], there are very few studies on the changes occurring in solid phase [19]. The formation of a liquid intermediate, directly observed by the group of Prof. Lede [20, 21]and Dauenhauer et al, 2009 [22], has an important effect on the formation of carbonaceous materials [23]. The major cellulose pyrolysis product, levoglucosan, has a boiling point between 304.5 and 340 oC [23-25]. Thus, if not removed fast enough, it can remain in liquid phase during pyrolysis and lead to polycondensation towards char formation [26]. The formation of this liquid intermediate is also critical for the ejection of heavy products (oligosaccharides) during the pyrolysis process [27].

The role of cellulose crystallinity on pyrolysis reactions has received limited attention [28-31].

Wang et al. [32] found that low cellulose crystallinity does not affect much levoglucosan production but it affects the formation of liquid intermediates. The effect of liquid intermediates on the formation of char is still poorly understood.

The Broido-Shafizadeh mechanism is one of the most widely accepted cellulose pyrolysis scheme reported in the literature [16, 33]. This mechanism considers the formation of active cellulose as the reaction rate limiting step followed by the simultaneous formation of char and volatile compounds. Other important mechanisms reported in the literature include the Waterloo

[34], the Diebold [35], the Wooten-Seeman-Hajaligol [36] and the Varhegyi-Antal [37, 38] mechanisms. Among all the mechanisms only a few considered a multi-step (sequence of reactions) mechanism: i) the Varhegyi-Antal mechanism [37, 38], which considers a sequence of reactions responsible for changes in the structure of the carbonaceous residue formed. ii) Lin at

145

al [13] proposed a mechanism going first from anhydrosaccharides, with degree of polymerization (Dp) up to 7 and subsequently the formation of mono-saccharides derivatives followed by furans. Eventually the mono-saccharides and furans can form char. iii) Westerhof

[39](biomass) and Lewellen [40] (cellulose) discussed a mechanism in which some compounds were formed but could not leave the particle due to mass transfer limitations. Once trapped inside the particle the newly formed products could react into new compounds, which again could either escape from the particle or get trapped in the particle. Eventually, these trapped products grew long inside the particle and formed char. Among the models so far developed, only the

Antal model [41] takes into account the effect of cellulose crystallinity on the formation of carbonaceous solid residues. This model however, does not explicitly consider the formation of liquid intermediates or the formation of cross-linked saccharides as important steps in the pyrolysis process influenced by both temperature and cellulose crystallinity. Developing models predicting how pyrolysis conditions affect the structure of the product bio-char is critical to control the production of anhydrosaccharides and for the development of carbonaceous based products.

The main goal of this paper is to study the effect of cellulose crystallinity on the inter-particle reactions responsible for char formation and to propose a new scheme describing the experimental changes observed. Cellulose samples were pyrolyzed in a spoon reactor and the changes in cellulose solid residue structures were analyzed by several analytical techniques (acid hydrolysis followed by IEC, FTIR, 13C-NMR). A new cellulose reaction mechanism that describes the observed changes in the solid phase is proposed.

146

5.2 Materials and Methods

5.2.1 Materials

Microcrystalline (control) cellulose (Avicel, PH-101, ~50 µm, Sigma-Aldrich) was used after drying. Amorphous cellulose was obtained after ball-milling control cellulose at 300 rpm for 24 hours (Across International PQ-N2, ceramic 100 mL jar and balls). The characterization of both samples has been reported elsewhere [32]. Avicel had a crystallinity of 60.5% and the ball-milled cellulose had a crystallinity of 6.5%.

5.2.2 Spoon Reactor

A Spoon reactor similar to the one described by Guillain et al. [42] was used. Briefly, this set up has a Lindberg/Blue 1100 tube furnace (single point) as the heating source. The heating chamber is a stainless steel tube (25.4 mm ID). Approximately 400 mg of sample was evenly distributed

(by visual estimate) on the spoon. The stainless steel spoon was put on a holder and was introduced into water cooling chamber under nitrogen for 20 minutes. The purpose of this step is to remove all the oxygen in contact with the sample. It was later pushed into heating chamber and kept at that temperature for a designated time. The experimental temperature was calibrated by thermocouple (type K, armored and grounded) on the spoon at the position where the cellulose sample was heated. The temperature of the oven was varied between 240 and 400 oC and the cellulose particles were kept at the reaction temperature up to 120 min. The treatment time includes a temperature increasing period (from room temperature to experimental

147

temperature) which was typically between 5 and 10 min which is small compared with the overall reaction times studied (up to 120 min). A constant nitrogen flow of 900 mL/min (150 mL/min through char cooling chamber and 750 mL/min preheated by furnace) prevented oxidation. After the experiment the residue was cooled in the water cooled chamber for 5 min.

The mass of the solid residue obtained was reported for all the experiments.

5.2.3 Analysis of Solid Residues

FTIR spectroscopy

The solid residue samples obtained in the spoon reactor were analyzed with a Shimadzu FTIR with attenuated total reflection (ATR, MIRacle equipped with a Ge crystal, PIKE Technology).

Spectra were recorded for each sample in triplicate using 64 scans. The spectra were ATR and automatic baseline corrected and averaged. Bands were assigned according to the information listed in Table 5.1. Band height was used for comparison.

148

Table 5.1 FTIR band assignments [43-48] Wavenumber Band Assignment (cm-1) 895-898 Glucose ring stretch, C1-H deformation, C-H deformation in cellulose 990 C-O (secondary alcohols skeletal vibrations) 1020-1050 C-O stretching (C-6 skeletal vibrations) 1050-1070 C-O stretching (C-3 skeletal vibrations) 1080 C-O-C (pyranose ring skeletal vibrations) 1110 C-OH (skeletal vibrations) 1160 C-O-C (antisymmetrical bridge stretching at b-glucosidic linkage) 1200 C-O (stretching in pyranose ring) 1310-1360 OH in plane bending, C-C and C-O (skeletal vibrations) 1372 C-H symmetric deformation 1360 O-H (bending) 1425 CH2 scissoring 1597-1620 C=C stretching 1694-1710 C=O stretching 2800-3000 C-H stretching 3100-3600 O-H stretching

13C Nuclear Magnetic Resonance (NMR) spectroscopy

The 13C CP-MAS NMR spectral analysis was performed on a Bruker DRX 400. Samples were packed in zirconia rotors (5 mm Ø, 160 µl.) and spun at 6 kHz using a Chemimagnetics solid state probe (relaxation time 10 ms) and obtained 4 096 or 16 384 scans. The spectra were processed using ACD lab software. Peaks were assigned according to the literature [19, 49]. The area of peaks was evaluated by a deconvolution method applying Gaussian and Lorentzian lines previously used in describing cellulose structures [50-56]. The areas of the peaks were considered the molar fraction of the carbon assigned. In this paper we considered that the aromatic C is in the form of benzene (C6H6); that the aliphatic carbon is in the form of methylene

(-CH2); that the oligosaccharide carbon is in the form of levoglucosan (C6H10O5); and the furanyl

149

carbon is in the form of furan (C4H4O). Based on this supposition we calculated the mass fraction of each of the functional groups. The mass fraction of the group was then multiplied by the yield of the solid residue to obtain the yield of the group expressed on initial cellulose basis.

Hydrolysis and Ion Exchange Chromatography

The solid samples were hydrolyzed following the ASTM D5896-96 standard. Briefly, approximately 100 mg of sample was weighed in a culture tube (25 x 150 mm). 1 mL of 72 %

o H2SO4 was added to the tube and mixed with glass rod at 30 C for 1 h. Then water (28 mL) was carefully added (to flush the glass rods) and were sealed and autoclaved (125 oC for 1 h). The hydrolysates were diluted 250 times and filtered for ion exchange chromatographic (IEC, Dionex

ICS-3000 Ion Chromatograph, equipped with CarboPac PA20 3 x 150mm column) analysis.

Results collected were multiplied by the yield of residue to reflect the evolution of the solid phase on initial cellulose basis.

Scanning Electron Microscopy (SEM)

SEM analyses were performed on an FEI Quanta 200F with a Large Field Detector and a low vacuum of 130 Pa. Samples were distributed evenly (by visual estimation) on adhesive tape on a metal stub.

150

5.3 Results and Discussions

5.3.1 Yield of Solid residue

Figure 5.1 shows the evolution of the solid phase residue and conversion () as a function of pyrolysis conditions (time and temperature) for the control and ball-milled cellulose. The conversion was calculated as: 

(t) = (m cellulose (t=0) - mresidue (t)) / (mcellulose (t=0) – mresidue (t=∞)) Equation 1

Control cellulose remained relatively unconverted at temperatures below 260 oC. Under the same conditions (260 oC, 120 min) ball-milled cellulose lost approximately 40 % of its weight. A linear relationship was observed at pyrolysis temperatures below 260 oC. The depolymerization of the control cellulose accelerates at 280 oC. Remarkably, this is the huge weight loss in a relatively small temperature regime. At temperatures above 300 oC, both samples reached similar weight losses (>85%) after 30 min of thermal treatment.

151

Control cellulose Ball-milled cellulose

Figure 5.1 Residue yields from a spoon reactor at different temperatures and reaction time

Reaction kinetic analysis was performed on the slow pyrolysis results and the data obtained is shown in Figure 5.1. There was a noticeable difference between the two cellulose types and this may be a result of the formation of a liquid intermediate. The kinetic analysis was conducted using the method described by Thurner and Mann [57] but only considering conversion of cellulose to volatiles. For this analysis we assumed the Broido and Nelson’s model [33, 58] and simplified it as shown in Scheme 5.1:

k cellulose volatile + char + gas

Scheme 5.1 Scheme of cellulose overall thermal reactions in the spoon reactor Where, residue mresidue (t) =mcellulose (t) + mchar (t) = mcellulose (t=0) – (mgas (t) + mvolatile (t)) Assuming first order equations, the rates of this reaction can be expressed as:

Equation 2

152

Where: k is the overall conversion rate constant. By integrating this equation, it is possible to find the relationship between conversion () and reaction time (t):

Equation 3

The values of k in equation 2 were obtained using Solver in MS Excel. The plot of the values of the log of the constant rate of reaction (ln k) vs. 1/T is shown in Figure 5.2. The plot obtained shows two linear regions indicating the existence of two reaction zones. The first zone was

o between 240 and 300 C. The activation energies (Ea) and pre-exponential factors obtained are listed in Table 5.2. Overall, control cellulose has a higher Ea and pre-exponential factor than ball- milled cellulose, which may due to the fast formation of the liquid intermediate in ball-milled cellulose. Literature reported values of Ea for cellulose rich materials between 48.1 and 282 kJ/mol and values of pre-exponential factor between 4 and 24.9 min-1 [59], while cellulose

(Avicel) usually falls between 185-240 kJ/mol [60]. Typical kinetic values obtained from thermogravimetric analysis (TGA) are between 217 and 231 kJ/mol [14, 32, 61, 62]. The spoon reactor gave considerably lower overall activation energy, which is probably due to its overall lower heat transfer achieved in the cellulose sample [58].

Remarkably, the kinetic values in our experiment clearly differed by separating the kinetic evaluation into two temperature regions: 240 to 300 oC and between 300 and 400 oC (Table 5.2).

o Below 300 C, the Ea of control cellulose was in good agreement with TG results and ball-milled cellulose was relatively lower. In the higher temperature region, both cellulose samples showed much lower Ea. The control cellulose gave an Ea of 86.0 kJ/mol, which was even lower than the range mentioned by Milosavljevic and Suuberg [14], showing the heat transfer barrier has an

153

important effect as indicate by Antal and Varhegyi [58]. Meanwhile, ball-milled cellulose had an

-1 even lower Ea of 52.4 kJ/mol with a pre-exponential factor at 4.0 log (min ). We concluded that this difference between control and ball-milled cellulose was due to the formation of a liquid intermediate as a result of lower crystallinity of the ball-milled sample. At 300 oC was the starting point for the liquid intermediate to become a dominant factor.

300 oC

Figure 5.2 Graph of cellulose conversion rate versus (1/reaction temperature) to determine Ea for control and ball-milled cellulose.

Table 5.2 Overall kinetic parameters for the pyrolysis of cellulose Overall ≤ 300 oC 300-400 oC

Ea logA Ea logA Ea logA (kJ/mol) log(min-1) (kJ/mol) log(min-1) (kJ/mol) log(min-1) Control 135.3 8.1 207.0 12.6 86.0 5.1 Ball-milled 92.0 6.7 155.1 12.8 52.4 3.3

154

Pyrolyzed cellulose samples were examined by SEM. Figure 5.3 shows micrographs of control and ball-milled cellulose at native state (Figure 5.3a) and after being treated for 120 min at 240

(Figure 5.3b), 280 and 400 oC (60 min). Large agglomerates were formed from the sticky liquid on the surface of both samples pyrolyzed at 240 oC even when the weight loss was limited. The formation of liquid intermediates appears to happen at low temperatures. Considering the crystallinity of our control cellulose (60%), it is reasonable to assume the amorphous zones of cellulose started to melt at 240 oC. At 300 oC (120 min), the difference between the control and ball-milled cellulose was more evident because the liquid intermediate transferred the ball-milled cellulose to a second regime (see Figure 5.2). Although both cellulose samples had a weight loss >50% and shrunk in particle size (Figure 5.3c), ball-milled cellulose (with lower crystallinity) started to coalesce after pyrolysis for 120 min. At 320 oC, the ball-milled cellulose samples started to transform into a single piece of char (not shown). Figure 5.3d clearly shows that ball-milled cellulose after pyrolysis at 400 oC for 60 min lost its particle structure and was totally melted. However, the control cellulose retained its fibrous structure but of much smaller particle size.

155

Control Ball-milled a

o 240 C, 120 min Agglomerate Agglomerate b

o 280 C, 120 min Agglomerate Agglomerate

400 oC, 60 min d

156

5.3.2 Quantification of hydrolysable saccharides

Figure 5.4 shows the content of glucose recovered, determined by IEC, after solid residue hydrolysis as a function of reaction time and temperature. The formation of glucose is related to the amount of hydrolysable saccharides which come from the non-reacted (original) cellulose.

The content of hydrolysable sugar decreased following a trend similar to the one reported for the weight loss (Figure 5.1) but at higher rate. This result confirms the formation of a non- hydrolysable fraction as the pyrolytic conversion of cellulose advances. With temperature increasing, the treated cellulose samples went from yellow to brown and finally to black. When the residual carbonaceous solids were less than 20%, the residue was difficult to hydrolyze.

Control Ball-milled

cellulose cellulose

Figure 5.4 Graph of cellulose residue recovered after pyrolysis as a function of time for (left) control cellulose and (right) ball-milled cellulose.

The difference between the yield of the solid residue recovered and the left over unconverted cellulose (measured as hydrolysable saccharides) is reported as “non-hydrolysable” fraction. The yields of non-hydrolysable material in the various samples are shown in Figure 5.5. The graphs depict the same parameters on the y-axis but are split in two temperature regimes for

157

convenience. The content of non-hydrolysable saccharides was very pronounced in the early stages of thermal treatment (weight loss < 50%). These non-hydrolysable saccharides are formed from the dehydration, cross-linking and polycondensation of cellulose [63-65]. For example, at

240 oC and 120 min, 20% of ball-milled cellulose is non-hydrolysable while only 12% of the control cellulose was non-hydrolysable. Under pyrolysis at 280 oC for 120 min, residue of ball- milled cellulose had ~85% of non-hydrolysable saccharides and reached ~17% in yield, while residue of control cellulose had ~25% of non-hydrolysable saccharides (~10% in yield). Ball- milled cellulose produced these non-hydrolysable saccharides and started to show a reduction in this yield under less extreme conditions. A higher max yield could be found in ball-milled cellulose until temperature above 340 oC. This phenomenon may be due to the formation of liquid intermediates that accelerate dehydration and cross-linking reactions.

158

Control Ball-milled

cellulose

Figure 5.5 Yield of non-hydrolysable material (charcoal + cross-linked saccharides) from cellulose ((left) control and (right) ball-milled cellulose) after pyrolysis as a function of reaction time.

5.3.3 FTIR spectroscopy

Figure 5.6 shows the FTIR spectra of control and ball-milled cellulose residues obtained after pyrolysis at 240, 260, 280, 300 or 320 oC for 30, 60, or 120 min.

159

Figure 5.6 FTIR spectra of pyrolyzed ball-milled cellulose for 30, 60, or 120 min at 240, 260, 280, 300 and 320 oC.

It should be noted that temperatures above 320 oC and 30 min produced a black residue, which were giving bad response on our FTIR setting. So only residue formed at temperatures below

320 oC were analyzed. The spectra collected were used to qualitatively track the evolution of

160

functional groups in the solid samples. Both cellulose samples followed similar trends in their functional group evolution. The structure of cellulose did not change at temperatures below 240 oC. Evidence of cellulose intra-ring dehydration was indicated by the formation of C=O (1710 cm-1) and C=C (1620 cm-1) bands at temperatures close to 260 oC that increasing as the severity

(temperature and time) increased [63]. Obvious weakening of the C-O band at 1060 cm-1 was observed above 280 oC. This C-O band disappeared at 320 oC under 120 min of treatment, indicating that cellulose structure was completely modified under these conditions.

A detailed analysis of the FTIR band height was used to follow changes in chemical characteristics of the solid residue. Figure 5.7 and Figure 5.8 show the change of two

-1 crystallinity indexes: I1420/I892 (in this work, band 892 shifted to 899 cm , so here we use

I1420/I899 to represent this crystallinity index) and I1376/I2900 obtained from the FTIR spectra [44,

46]. In this work, the first index (I1420/I899) shown in Figure 5.7 is interpreted as proportional to the ratio between crystalline cellulose over remaining cellulose. The second index (I1376/I2900) shown in Figure 5.8 is proportional to the ratio between crystalline cellulose over total residue.

Figure 5.7 shows that the ratio between crystalline cellulose over remaining cellulose was lower for the ball-milled cellulose than the control but increased very rapidly when heated. This result was also seen by Wang et al [32] and clearly suggests that most of the weight loss observed for ball-milled cellulose around 280 oC was due to the conversion of the amorphous cellulose and a concomitant increase in crystalline concentration. Furthermore, Figure 5.8 shows that the actual

161

content of crystalline cellulose in the residue was initially higher for the control but then decreased with process severity.

Control Ball-milled

Figure 5.7 Crystalline cellulose/cellulose ratio (I1420/I899) from FTIR for (left) control and (right) ball-milled cellulose at differing temperatures as a function of reaction time.

Control Ball-milled

Figure 5.8 Crystalline cellulose/biomass ratio (I1376/I2900) from FTIR for (left) control and (right) ball-milled cellulose at differing temperatures as a function of reaction time.

162

Figure 5.9 shows the ratio between C=O/C-O ratio (I1694-1710/I990-1070) for control and ball-milled cellulose. The C=O groups are formed by the intramolecular dehydration reactions [63]. Below

280 oC these reactions seem to be favored in amorphous cellulose, perhaps due to the acceleration of dehydration reactions when the liquid intermediate is formed [23]. It is known that dehydration reactions occur via E1-elimination mechanism and are mediated by an acid catalyst [23]. The existence of a melting step at around 280 oC is required to form the C=O group.

The formation of agglomerates in this range of temperature is a clear indication of the presence of liquid intermediates on the surface of the converted cellulose particles.

Control Ball-milled

Figure 5.9 C=O/C-O ratio from FTIR for (left) control and (right) ball-milled cellulose at differing temperatures as a function of reaction time.

163

At higher temperatures the dehydration reactions are more pronounced (notably so for the control) as seen by the dramatic increase in C=O/C-O. The evolution of the C=C/C-O ratio (I1597-

1620/I990-1070) is shown in Figure 5.10. The figures were divided in two graphs for convenience.

The trends observed for the C=C/C-O were very similar to that observed for the C=O/C-O ratio both of which are products of intra-ring dehydration reactions. At higher temperatures the solid residue formed from the control cellulose seems to have higher content of C=C groups.

Control Ball-milled

Figure 5.10 The C=C/C-O ratio from FTIR for (left) control and (right) ball-milled cellulose at differing temperatures as a function of reaction time.

Figure 5.11 shows the OH/C-O ratio (I3100-3600/I990-1070) for control and ball-milled cellulose. This ratio did not change significantly below 300 oC and was very similar for both the control and

164

ball-milled samples. This ratio dramatically increased > 300 oC, indicating the dehydration of hydroxyl group on C-3 (1060 cm-1) and C-6 (1030 cm-1) positions on cellulose.

Control Ball-milled

Figure 5.11 The OH/C-O ratio from FTIR (I3100-3600/I990-1070) for (left) control and (right) ball- milled cellulose at differing temperatures as a function of reaction time.

5.3.4 13C-NMR spectroscopy

Solid state CP-MAS 13C-NMR spectroscopy was used to monitor C functional groups in cellulose. Figure 5.12 shows the difference between control and ball-milled cellulose. Spectra for control cellulose is consistent with those reported by Pastorova et al [19]: C-6 at ~66 ppm (62 ppm for ball-milled); C-2, C-3, C-5 at 74-76 (doublet) ppm (75 ppm for ball-milled); anomeric

165

C-1 peak at 107 ppm (104 ppm for ball-milled); C-4 at 90 (crystalline) and 86 (amorphous) ppm

(for ball-milled at 85 ppm (amorphous)).

Figure 5.12 13C-NMR spectra of control and ball-milled cellulose

Figure 5.13 shows the evolution of the 13C-NMR spectra for cellulose pyrolyzed between 260 and 400 oC. As the temperature increased the peaks broadened and were assigned to aliphatic groups (10-60 ppm), aromatic structure (100-155 ppm), furanyl structure (150-170 ppm) and carbonyl structure (175-230 ppm) [19].

166

C2, C3, C5

C6 C1 C4 Side Spin Side Spin

Figure 5.13 13C NMR Spectra for pyrolyzed control cellulose at differing conditions

167

The 13C spectra were peak fitted following the method described for cellulose crystallinity and allomorph analysis [50-56]. The method was modified to accommodate the thermally derived structures after pyrolysis. The molar fraction of C associated to the different structures was calculated from the mass fraction of all the functional groups (aromatics, aliphatic, carbohydrate and furanyl) and multiplied by the yield of solid residue to obtain the functional group yield on initial cellulose basis (Figure 5.14).

168

Control Ball-milled

The content of the carbohydrates decreased gradually following a trend similar to that measured by acid hydrolysis. Aliphatic groups were formed faster in ball-milled cellulose above 300 oC. At

300 oC, control cellulose products showed a higher presence of aliphatic groups than for ball- milled cellulose. Above 300 oC, trends were similar. Aromatic groups were formed more rapidly in the ball-milled samples than the control samples. The formation of aromatic groups is mainly

169

due to polycondensation reactions responsible for charcoal formation. These groups seem to be stable in the conditions studied.

Figure 5.15 shows a linear relationship between the cellulose content measured by acid hydrolysis to that determined by quantitative 13C-NMR. The values of cellulose determined by acid hydrolysis were lower than those determined by 13C-NMR. The slope of the correlation

(hydrolysis vs. NMR) were 0.86 and 0.798 respectively for control and ball-milled cellulose. It means that the 13C-NMR quantification method used, accounted the very poorly modified cross- linked saccharides as carbonhydrate (cellulose).

Figure 5.15 Comparison of cellulose content determined by hydrolysis and 13C-NMR.

Figure 5.16 shows the evolution of furanyl and carbonyl groups as a function of pyrolysis conditions. The furanyl and carbonyl groups are products of dehydration reactions. In general, higher yields of these two groups were formed in the ball-milled cellulose. This may be attributable to the more facile formation of liquid intermediates from amorphous cellulose.

170

Control Ball-milled

Figure 5.16 The yield of furanyl and carbonyl groups for (left) control and (right) ball-milled cellulose at differing pyrolysis temperatures as a function of reaction time

It is important to note that the addition of aliphatic, aromatic, furanyl and carboxyl fractions was considerably lower that the content of non-hydrolysable saccharides shown in Figure 5.5. This difference clearly indicates that the 13C-NMR method used accounted the poorly modified cross- linked saccharides as cellulose. So in this paper we will use the difference between the cellulose content measured by 13C-NMR and the cellulose content measured by acid hydrolysis as an estimate for the yield of cross-linked saccharides. The same results are obtained if the yields of furanyl, carbonyl, aliphatic and aromatic are subtracted from the yield of non-hydrolysable saccharides. The yield of cross-linked saccharides obtained for the different reaction conditions studied are shown in Figure 5.17. So the cross-linked saccharides seem to be formed by relatively unmodified glucose units linked by ether bonds.

171

Control Ball-milled

Figure 5.17 Behavior of non-hydrolysable cross-linked saccharides

5.3.5 Reaction Mechanism

The data presented in Figure 5.15, Figure 5.16 and Figure 5.17 was used to develop a new reaction mechanism explaining the behavior of cellulose during slow pyrolysis conditions. The yields of aromatics and aliphatic was added to calculate the bio-char yield. The yield of highly dehydrated cross-linked saccharides was calculated by adding the yield of furanyl and carbonyl.

So the reacting solid was formed by cellulose (A), depolymerized cellulose (B) vapor products of cellulose depolymerization (C), water vapor (D), cross-linked saccharides (E), depolymerization products of cross-linked saccharides (aerosol and vapors) (F) water vapor (G), dehydrated saccharides (H), water vapor and gases (CO, CO2, etc) (I) char (polyaromatic compounds) (J).

172

Based on the behavior observed in Figure 5.4, Figure 5.14, Figure 5.16 and Figure 5.17 we decided to propose the following reaction mechanism.

Scheme 5.2 Reaction mechanism proposed for cellulose slow pyrolysis

The authors recognize that the experimental data obtained under an initial slow heating does not allow obtaining real “kinetic” constants. The kinetic constants that will be obtained are basically to prove that the changes in the solid residue composition can be predicted at least qualitatively by the new reaction scheme proposed. Experiments under very high heating / mass transfer rates, and rapid product quenching conditions will have to be conducted to find the kinetic parameters of the chemical reactions. Hopefully, the same scheme will be valid at other heating rates but forming part of more complex models with actual kinetic constants obtained in near isothermal conditions (at very high heating rates) and explicitly accounting for the effect of heat and mass transfer phenomena. The data obtained is however very useful to better understand cross-linking, dehydration and polycondensation reactions which are better studied under slow heating rate conditions.

173

The scheme shown can be also presented in the form of the system of reactions that follow: k A 1 B Reaction 1 k2 B C Reaction 2 k3

B a1 E + (1- a1) D Reaction 3 k4 E F Reaction 4 k5

E a2 H + (1-a2) G Reaction 5 k6

H a3 J + (1-a3) I Reaction 6

Where,

A: cellulose

B: cellulose of low of Dp

C: volatile products of cellulose depolymerization (mainly levoglucosan)

D: water (vapor)

E: cross-linked saccharides (non-hydrolysable)

F: volatile products derived from the cross-linked saccharides (aerosols, vapor)

G: water (vapor)

H: dehydrated products

I: water (vapor) and gas (CO, CO2, etc.)

J: char ((poly-aromatic hydrocarbons)

The stoichiometric coefficients (a1, a2, and a3) were calculated based on the following assumptions:

174

The first reaction (1) describes the early stage depolymerization of cellulose (A) to produce a low Dp cellulose (B). The basic unit of crystalline cellulose morphology is the elementary fibril

(a nanocrystal) with approximately 300 Å length surrounded by amorphous zones that can be easily broken at low temperature (150-190 oC) [66]. The formation of a nano-crystal with molecular weight close to 300-400 g/mol when crystalline cellulose is heated has been reported by Shafizadeh [67]. The amorphous zones can be depolymerized all the way to produce oligomeric liquid intermediates. This cellulose with low Dp has the same formula as cellulose

(C6H10O5). This step does not lead to any weight loss.

The second reaction (2) is associated with the further depolymerization of low Dp cellulose to produce volatile anhydro-mono- and oligosaccharides (chiefly levoglucosan and cellobiosan)

(C). This reaction causes weight losses and has been extensively studied in the literature [62, 68-

71]. It consists of glycosidic bond cleavage at the ends of cellulose polymer chains. Once this reaction is initiated, the whole chain “unzips” to release monomers of levoglucosan and dimers of cellobiosan [62, 68-70, 72-74]. The evaporation of levoglucosan and other potential liquid intermediates has been extensively studied by Oja and Suuuberg [25].

The third reaction (3) corresponds to inter-ring dehydration reactions responsible for the formation of water and cross-linked saccharides [23, 63, 65]. In this reaction a slight inter-ring dehydration (cross-linking) of the low Dp cellulose (B) occur to release one unit of water (D) from every two unit of cellulose and give a cross-linked saccharide complex (E) with basic unit

175

as C6H9O4.5. This reaction has been extensively studied [63-65]. In order to balance this equation, a1 was 0.94. While Mamleev et al [23] argued this reaction is associated with the formation of liquid intermediate, Chaiwat et al [64, 65] presented their mechanism supposing the reaction happening in the solid phase. The existence of cellulose-derived liquid intermediates has been confirmed by our SEM studies and reported elsewhere in literature [20-22, 32]. This liquid intermediate seems to be formed by products of cellulose depolymerization (levoglucosan, cellobiosan, cellotriosan) and their cross-linking products [36, 64, 65, 75] which are not volatile enough to evaporate and escape the puddle [25]. It has been recently shown that this liquid intermediate is responsible for the thermal ejection of oligomeric products [27].

The fourth reaction (4) is associated with the cracking of cross-linked structures to form volatile products (F). This reaction was extensively studied by Chaiwat et al [64, 65].

The fifth reaction (5) describes the ring contraction and dehydration of cross-linked saccharides.

This reaction represents the dehydration of group E releasing one unit of water (G) every basic unit and giving dehydrated cross-linked saccharide complex (H, C6H7O3.5). In order to balance this equation in mass, a2 should be 0.88. This reaction has been extensively studied by Scheirs et al [63].

The sixth reaction (6) represents the polycondensation of dehydrated saccharides (H) to form char (J). This reaction has been poorly studied in the literature [19, 26]. According to Shafizadeh and Sekiguchi [76] and McGrath et al [77] can be represented by the formula (C5H3.5O). The

176

water and gases (I) produced in this reaction is represented by the formula (CH3.5O2.5). In order to balance the equation a3 should be 0.59.

Based on the reaction mechanism proposed above, the following system of equations describing the rates of reactions is proposed:

Equation 4

Equation 5

Equation 6

Equation 7

Equation 8

Equation 9

Equation 10

Equation 11

Equation 12

Equation 13

Each of the partial differential equations was converted into its final element solution and programmed in MS Excel. The MS Excel analysis tool “Solver” was used to calculate the values of the constant rates (k1, k2, k3, k4, k5 and k6) that minimized the error between the experimental values of Cellulose (A+B), cross-linked saccharides (E), dehydrated cross-linked saccharides (H)

177

and char (J) yield and those predicted by the model. The values of constant rates obtained for each reaction was potted vs. 1/T. The results are shown in Figure 5.18.

The values of k1 exhibited as a straight line showing this reaction not affected by the physical transition (forming of liquid intermediate). k2 also exhibited as straight line but only ball-milled cellulose was suggested by the model to have this reaction. This result suggest that under slow pyrolysis conditions the formation of cross-linked from crystalline cellulose is favored over its depolymerization to form levoglucosan. The values of k3, k4, k5 and k6 were found having a transition at 300 oC, which could be associated to the formation of liquid intermediates. This temperature was also predicted by the model considering conversion of cellulose (Figure 5.2).

This transition was less pronounced in most cases of control cellulose indicating a different physiochemical status of this high crystallinity cellulose during the thermal treatment in spoon reactor.

178

Figure 5.18 Plots of Ln k vs. 1/T for the reactions 1-6 for the pyrolysis of cellulose

The values of stoichiometric factors (a1, a2, and a3), activation energy and pre-exponential factor obtained for each of the reactions is shown in Table 5.3. The values of A and Ea are reported for temperatures below and over 300 oC according to the transition found in Figure 5.18.

179

Table 5.3 Kinetic parameters and stoichiometric coefficients obtained for cellulose pyrolysis in the 260-300 and 300-400 oC temperature ranges 260-300 oC 300-400 oC Control Ball-milled Control Ball-milled Ea logA Ea logA Ea logA Ea logA log(min- log(min- log(min- log(min- (kJ/mol) (kJ/mol) (kJ/mol) (kJ/mol) 1) 1) 1) 1) k1 121.9 9.7 103.7 8.4 70.3 5.1 48.9 3.4 k2 n.a. n.a. 71.6 4.8 n.a. n.a. 64.6 4.2 k3 39.7 2.4 56.5 3.9 60.2 4.3 44.7 2.8 k4 128.8 10.7 63.0 4.6 62.7 4.7 60.6 4.4 k5 165.4 13.1 160.1 12.8 65.1 4.0 74.4 5.0 k6 209.9 16.3 433.3 37.0 48.8 1.7 32.3 0.3 Stoichiometric coefficients: a1=0.94, a2=0.88, a3=0.59. “n.a.” indicates not available.

Overall, the reaction giving hydrolysable saccharides (k1, k2) had a much lower Ea and pre- exponential factor compared to values reported for this major weight loss reaction (238 kJ/mol,

~19.8 min-1 [38], 198 kJ/mol, 16.3 min-1, [16]). Considering a much larger sampling amount

(400 mg) in our spoon reactor, this may due to a larger heat transfer limitation within the sample as mentioned by Antal and Varhegyi [58]. Comparing to the result acquired earlier from the overall weight loss (Table 5.2), the first major decomposition (k1) of both cellulose samples was still accelerated at temperature 300 oC. Ball-milled cellulose, in this stage, was affected more by this temperature change.

Surprisingly, control samples were suggested by our model having no product of cellulose depolymerization releasing in the second reaction. Instead, the cross-linking reaction (k3) was preferred in this slow heating condition. Although the value of Ea and pre-exponential factor had minor difference, the plots of this reaction were found similar in the whole temperature regime.

Different from intra-ring dehydration reactions (k5), the cross-linking reaction (k3) was only

180

slightly enhanced in the ball-milled cellulose but not significantly changed in the performance of crystalline cellulose. By comparing the SEM results (Figure 5.3), this reaction was concluded as an early stage reaction during the slow pyrolysis, which may occurred preferably in solid phase and need the little involving of liquid intermediate. In the case of ball-milled cellulose, the liquid puddle formed rapidly at high temperature (>300 oC) and might immersed more solid surface for reaction.

Cracking and releasing of cross-linked saccharides was also predicted in this model (k4). This reaction is similar to the removing of hydrolysable saccharides (k2) but releasing non- hydrolysable saccharide vapor. This ‘reaction’, which might have included the evaporation and ejection of aerosol [22], together with cross-linking reaction (k3) was described in the literature

[36, 64, 65], but was never included in a cellulose model. This procedure is always showing lower Ea and pre-exponential factor in ball-milled cellulose. An obvious transition could be found in the control cellulose, while ball-milled cellulose maintained similar in two temperature regimes. With temperature > 300 oC, both samples were having similar rate of reaction.

The intra-ring dehydration reactions of cross-linked saccharides were found in similar rate in

o both samples. Below 300 C, the dehydration reaction (k5) of cross-linked saccharides is close to values in literature for both cellulose samples (153 kJ/mol, 13.7 log (min-1) [16], 147 kJ/mol, 7.7 log (min-1) [38]). As shown in Figure 5.3, both cellulose samples started to form agglomerates as the amorphous portion of cellulose started to melt and the liquid intermediate acted as adhesive to clump particles together. Ball-milled cellulose apparently had greater propensity to form this

181

intermediate (amorphous component: 93.5 wt. %), but it didn’t exhibit significantly more dehydration reactions. At temperatures above 300 oC, the liquid intermediate started to form in higher amounts in both cellulose samples, and the dehydration reaction rates were similar.

Polycondensation of cellulose turns dehydrated cross-linked saccharides to aromatic structures

o (k6). From the Ea and pre-exponential value we got, with temperature <300 C, the polycondensation reaction preferred crystalline structure. Ball-milled cellulose tended to maintain its dehydrated structure instead of polycondense to aromatic structure with its high Ea.

When temperature >300 oC, this reaction was rapidly enhanced in both samples. Both cellulose samples underwent similar strength of reaction which explained the similar amount of final residue found in both samples.

Figure 5.19 and Figure 5.20 show the yields of cellulose and hydrolysable saccharides (A+B), cross-linked saccharides (E), dehydrated cross-linked saccharides (H) and char (J) determined experimentally for the control and ball-milled cellulose and those fitted by the kinetic parameters obtained. A very good agreement was obtained between the values fitted with the kinetic parameters reported in Table 5.3 and those obtained experimentally.

182

Control Ball-milled

260 oC 260 oC

280 oC 280 oC

300 oC 300 oC

320 oC 320 oC

183

Control Ball-milled 340 oC 340 oC

o 360 C 360 oC

o 400 C 400 oC

5.4 Conclusions

Ball-milled cellulose (amorphous) was compared to control microcrystalline cellulose to quantify the effects of crystallinity during pyrolysis. A spoon reactor was used to perform the pyrolysis reactions. Below 260 oC, only the ball-milled cellulose presented significant weight losses. The

184

control cellulose underwent major weight loss at 280 oC. At temperatures over 300 oC, both samples showed similar weight loss. The carbonaceous residue from the ball-milled cellulose coalesced into large chunks, while control did not. Samples were hydrolyzed to glucose. The fraction that could not be hydrolyzed was called non-hydrolysable saccharides. This portion was found reduced in amount when temperature increased and heating length increased, showing it was acting as intermediate towards other products with further reaction (cracking, dehydration, and polycondensation). Ball-milled cellulose forms non-hydrolysable saccharides under less extreme conditions than the control cellulose. This suggests that inter-ring dehydration and cross-linking reactions are favored in liquid intermediates. FTIR spectral analysis of the residues shows that ball-milled cellulose forms C=O and C=C groups faster at temperatures from 240 to

280 oC. However, above 300 oC, control cellulose forms more C=C and C=O groups than ball- milled cellulose. Both C=O and C=C bonds formation are associated with the dehydration reactions accelerated by the formation of liquid intermediates. The slightly dehydrated cross- linked saccharides were quantified in 13C-NMR as cellulose. NMR showed carbohydrates disappearing faster, aliphatics forming faster, and aromatics forming faster in the ball-milled cellulose than control until 300 oC. Higher yields of furanyl and carbonyl groups are seen in the ball-milled cellulose than control. Based on this information, a new reaction scheme was presented with six reactions that for the first time explicitly includes in a cellulose Pyrolysis model the formation of cross-linked (non-hydrolysable) cellulose and its further depolymerization to produce non-hydrolysable saccharides. Reaction parameters were measured based on this reaction scheme and compared with those obtained experimentally. The new

185

reaction scheme proposed is able to satisfactorily describe the changes in cellulose composition during slow heating rate processes.

Acknowledgements

This project was financially supported by the US National Science Foundation (CBET-0966419,

CAREER CBET-1150430), the Sun-Grant Initiative (Interagency Agreement: T0013G-A), and the Washington State Agricultural Research Center. The authors are very thankful for their support.

186

Reference

[1] G. Berndes, M. Hoogwijk, R. van den Broek, The contribution of biomass in the future global energy supply: a review of 17 studies, Biomass and Bioenergy, 25 (2003) 1-28.

[2] G.W. Huber, S. Iborra, A. Corma, Synthesis of transportation fuels from biomass: chemistry, catalysts, and engineering, Chemical reviews, 106 (2006) 4044-4098.

[3] H.S. Kheshgi, R.C. Prince, G. Marland, The potential of biomass fuels in the context of global climate change: Focus on Transportation Fuels1, Annual Review of Energy and the Environment, 25 (2000) 199-244.

[4] D.C. Elliott, D. Beckman, A.V. Bridgwater, J.P. Diebold, S.B. Gevert, Y. Solantausta, Developments in direct thermochemical liquefaction of biomass: 1983-1990, Energy & Fuels, 5 (1991) 399-410.

[5] P. Grange, E. Laurent, R. Maggi, A. Centeno, B. Delmon, Hydrotreatment of pyrolysis oils from biomass: reactivity of the various categories of oxygenated compounds and preliminary techno-economical study, Catalysis Today, 29 (1996) 297-301.

[6] A.V. Bridgwater, Production of high grade fuels and chemicals from catalytic pyrolysis of biomass, Catalysis Today, 29 (1996) 285-295.

[7] E. Sjöström, Wood chemistry: fundamentals and applications, ACADEMIC PressINC, 1993.

[8] S. Kersten, M. Garcia-Perez, Recent developments in fast pyrolysis of ligno-cellulosic materials, Current Opinion in Biotechnology, 24 (2013) 414-420.

[9] J. Lian, S. Chen, S. Zhou, Z. Wang, J. O’Fallon, C.-Z. Li, M. Garcia-Perez, Separation, hydrolysis and fermentation of pyrolytic sugars to produce ethanol and lipids, Bioresource Technology, 101 (2010) 9688-9699.

[10] J. Lian, M. Garcia-Perez, S. Chen, Fermentation of levoglucosan with oleaginous yeasts for lipid production, Bioresource Technology, 133 (2013) 183-189.

[11] P.M. Molton, T.F. Demmitt, Reaction mechanisms in cellulose pyrolysis: a literature review, in, 1977, pp. Medium: ED; Size: Pages: 90.

[12] D.F. Arseneau, Competitive reactions in the thermal decomposition of cellulose, Canadian Journal of Chemistry, 49 (1971) 632-638.

[13] Y.-C. Lin, J. Cho, G.A. Tompsett, P.R. Westmoreland, G.W. Huber, Kinetics and Mechanism of Cellulose Pyrolysis, The Journal of Physical Chemistry C, 113 (2009) 20097- 20107.

187

[14] I. Milosavljevic, E.M. Suuberg, Cellulose Thermal Decomposition Kinetics: Global Mass Loss Kinetics, Industrial & Engineering Chemistry Research, 34 (1995) 1081-1091.

[15] F. Shafizadeh, A.G.W. Bradbury, Thermal degradation of cellulose in air and nitrogen at low temperatures, Journal of Applied Polymer Science, 23 (1979) 1431-1442.

[16] A.G. Bradbury, Y. Sakai, F. Shafizadeh, A kinetic model for pyrolysis of cellulose, Journal of Applied Polymer Science, 23 (1979) 3271-3280.

[17] G. Varhegyi, M.J. Antal, T. Szekely, P. Szabo, Kinetics of the thermal decomposition of cellulose, hemicellulose, and sugarcane bagasse, Energy & Fuels, 3 (1989) 329-335.

[18] F. Shafizadeh, Y.L. Fu, Pyrolysis of cellulose, Carbohydrate Research, 29 (1973) 113-122.

[19] I. Pastorova, R.E. Botto, P.W. Arisz, J.J. Boon, Cellulose char structure: a combined analytical Py-GC-MS, FTIR, and NMR study, Carbohydrate Research, 262 (1994) 27-47.

[20] O. Boutin, M. Ferrer, J. Lédé, Radiant flash pyrolysis of cellulose—Evidence for the formation of short life time intermediate liquid species, Journal of Analytical and Applied Pyrolysis, 47 (1998) 13-31.

[21] J. Lédé, J.P. Diebold, G.V.C. Peacocke, J. Piskorz, The nature and properties of intermediate and unvaporized biomass pyrolysis materials, in: A.V. Bridgwater, D.G.B. Boocock (Eds.) Developments in Thermochemical Biomass Conversion, 1997, pp. 27-42.

[22] P.J. Dauenhauer, J.L. Colby, C.M. Balonek, W.J. Suszynski, L.D. Schmidt, Reactive boiling of cellulose for integrated catalysis through an intermediate liquid, Green Chemistry, 11 (2009) 1555-1561.

[23] V. Mamleev, S. Bourbigot, M. Le Bras, J. Yvon, The facts and hypotheses relating to the phenomenological model of cellulose pyrolysis: Interdependence of the steps, Journal of Analytical and Applied Pyrolysis, 84 (2009) 1-17.

[24] W. Feng, H.J. van der Kooi, J. de Swaan Arons, Application of the SAFT equation of state to biomass fast pyrolysis liquid, Chemical Engineering Science, 60 (2005) 617-624.

[25] V. Oja, E.M. Suuberg, Vapor Pressures and Enthalpies of Sublimation of d-Glucose, d- Xylose, Cellobiose, and Levoglucosan, Journal of Chemical & Engineering Data, 44 (1998) 26- 29.

[26] H. Kawamoto, M. Murayama, S. Saka, Pyrolysis behavior of levoglucosan as an intermediate in cellulose pyrolysis: polymerization into polysaccharide as a key reaction to carbonized product formation, Journal of Wood Science, 49 (2003) 469-473.

188

[27] A.R. Teixeira, K.G. Mooney, J.S. Kruger, C.L. Williams, W.J. Suszynski, L.D. Schmidt, D.P. Schmidt, P.J. Dauenhauer, Aerosol generation by reactive boiling ejection of molten cellulose, Energy & Environmental Science, 4 (2011) 4306-4321.

[28] G. Dobele, D. Meier, O. Faix, S. Radtke, G. Rossinskaja, G. Telysheva, Volatile products of catalytic flash pyrolysis of celluloses, Journal of Analytical and Applied Pyrolysis, 58–59 (2001) 453-463.

[29] A. Basch, M. Lewin, The influence of fine structure on the pyrolysis of cellulose. I. Vacuum pyrolysis, Journal of Polymer Science: Polymer Chemistry Edition, 11 (1973) 3071- 3093.

[30] A. Broido, M. Weinstein, Thermogravimetric Analysis of Ammonia-Swelled Cellulose, Combustion Science and Technology, 1 (1970) 279-285.

[31] M. Weinstetn, A. Broido, Pyrolysis-Crystallinity Relationships in Cellulose, Combustion Science and Technology, 1 (1970) 287-292.

[32] Z. Wang, A.G. McDonald, R.J.M. Westerhof, S.R.A. Kersten, C.M. Cuba-Torres, S. Ha, B. Pecha, M. Garcia-Perez, Effect of cellulose crystallinity on the formation of a liquid intermediate and on product distribution during pyrolysis, Journal of Analytical and Applied Pyrolysis, 100 (2013) 56-66.

[33] A. Broido, M.A. Nelson, Char yield on pyrolysis of cellulose, Combustion and Flame, 24 (1975) 263-268.

[34] J. Piskorz, D.S.A.G. Radlein, D.S. Scott, S. Czernik, Pretreatment of wood and cellulose for production of sugars by fast pyrolysis, Journal of Analytical and Applied Pyrolysis, 16 (1989) 127-142.

[35] J.P. Diebold, A unified, global model for the pyrolysis of cellulose, Biomass and Bioenergy, 7 (1994) 75-85.

[36] J.B. Wooten, J.I. Seeman, M.R. Hajaligol, Observation and Characterization of Cellulose Pyrolysis Intermediates by 13C CPMAS NMR. A New Mechanistic Model, Energy & Fuels, 18 (2003) 1-15.

[37] G. Várhegyi, M.J. Antal Jr, E. Jakab, P. Szabó, Kinetic modeling of biomass pyrolysis, Journal of Analytical and Applied Pyrolysis, 42 (1997) 73-87.

[38] G. Varhegyi, E. Jakab, M.J. Antal, Is the Broido-Shafizadeh Model for Cellulose Pyrolysis True?, Energy & Fuels, 8 (1994) 1345-1352.

[39] R.J.M. Westerhof, D.W.F. Brilman, M. Garcia-Perez, Z. Wang, S.R.G. Oudenhoven, S.R.A. Kersten, Stepwise Fast Pyrolysis of Pine Wood, Energy & Fuels, 26 (2012) 7263-7273.

189

[40] P.C. Lewellen, W.A. Peters, J.B. Howard, Cellulose pyrolysis kinetics and char formation mechanism, Symposium (International) on Combustion, 16 (1977) 1471-1480.

[41] M. Antal, Jr., Biomass Pyrolysis: A Review of the Literature Part 2—Lignocellulose Pyrolysis, in: K. Böer, J. Duffie (Eds.) Advances in Solar Energy, Springer US, 1985, pp. 175- 255.

[42] M. Guillain, K. Fairouz, S.R. Mar, F. Monique, L. Jacques, Attrition-free pyrolysis to produce bio-oil and char, Bioresource Technology, 100 (2009) 6069-6075.

[43] S.Y. Oh, D.I. Yoo, Y. Shin, H.C. Kim, H.Y. Kim, Y.S. Chung, W.H. Park, J.H. Youk, Crystalline structure analysis of cellulose treated with sodium hydroxide and carbon dioxide by means of X-ray diffraction and FTIR spectroscopy, Carbohydrate Research, 340 (2005) 2376- 2391.

[44] C.Y. Liang, R.H. Marchessault, Infrared spectra of crystalline polysaccharides. II. Native celluloses in the region from 640 to 1700 cm.−1, Journal of Polymer Science, 39 (1959) 269-278.

[45] M.L. Nelson, R.T. O'Connor, Relation of certain infrared bands to cellulose crystallinity and crystal latticed type. Part I. Spectra of lattice types I, II, III and of amorphous cellulose, Journal of Applied Polymer Science, 8 (1964) 1311-1324.

[46] M.L. Nelson, R.T. O'Connor, Relation of certain infrared bands to cellulose crystallinity and crystal lattice type. Part II. A new infrared ratio for estimation of crystallinity in celluloses I and II, Journal of Applied Polymer Science, 8 (1964) 1325-1341.

[47] M. Schwanninger, J.C. Rodrigues, H. Pereira, B. Hinterstoisser, Effects of short-time vibratory ball milling on the shape of FT-IR spectra of wood and cellulose, Vibrational Spectroscopy, 36 (2004) 23-40.

[48] M. Kačuráková, A.C. Smith, M.J. Gidley, R.H. Wilson, Molecular interactions in bacterial cellulose composites studied by 1D FT-IR and dynamic 2D FT-IR spectroscopy, Carbohydrate Research, 337 (2002) 1145-1153.

[49] T. Liitiä, S.L. Maunu, B. Hortling, Solid state NMR studies on cellulose crystallinity in fines and bulk fibres separated from refined kraft pulp, Holzforschung, 54 (2000) 618-624.

[50] L. Heux, E. Dinand, M.R. Vignon, Structural aspects in ultrathin cellulose microfibrils followed by 13C CP-MAS NMR, Carbohydrate Polymers, 40 (1999) 115-124.

[51] E.-L. Hult, P. Larsson, T. Iversen, A comparative CP/MAS 13C-NMR study of cellulose structure in spruce wood and kraft pulp, Cellulose, 7 (2000) 35-55.

190

[52] C. Kennedy, G. Cameron, A. Šturcová, D. Apperley, C. Altaner, T. Wess, M. Jarvis, Microfibril diameter in celery collenchyma cellulose: X-ray scattering and NMR evidence, Cellulose, 14 (2007) 235-246.

[53] P.T. Larsson, U. Westermark, T. Iversen, Determination of the cellulose Iα allomorph content in a tunicate cellulose by CP/MAS 13C-NMR spectroscopy, Carbohydrate Research, 278 (1995) 339-343.

[54] Y. Pu, C. Ziemer, A.J. Ragauskas, CP/MAS 13C NMR analysis of cellulase treated bleached softwood kraft pulp, Carbohydrate Research, 341 (2006) 591-597.

[55] K. Wickholm, P.T. Larsson, T. Iversen, Assignment of non-crystalline forms in cellulose I by CP/MAS 13C NMR spectroscopy, Carbohydrate Research, 312 (1998) 123-129.

[56] H. Yamamoto, F. Horii, CPMAS carbon-13 NMR analysis of the crystal transformation induced for Valonia cellulose by annealing at high temperatures, Macromolecules, 26 (1993) 1313-1317.

[57] F. Thurner, U. Mann, Kinetic investigation of wood pyrolysis, Industrial & Engineering Chemistry Process Design and Development, 20 (1981) 482-488.

[58] M.J. Antal, Jr., G. Varhegyi, Cellulose Pyrolysis Kinetics: The Current State of Knowledge, Industrial & Engineering Chemistry Research, 34 (1995) 703-717.

[59] M. Garc a-Pèrez, A. Chaala, J. Yang, C. Roy, Co-pyrolysis of sugarcane bagasse with petroleum residue. Part I: thermogravimetric analysis, Fuel, 80 (2001) 1245-1258.

[60] J.J.M. Orfão, F.J.A. Antunes, J.L. Figueiredo, Pyrolysis kinetics of lignocellulosic materials—three independent reactions model, Fuel, 78 (1999) 349-358.

[61] M. Grønli, M.J. Antal, G. Várhegyi, A Round-Robin Study of Cellulose Pyrolysis Kinetics by Thermogravimetry, Industrial & Engineering Chemistry Research, 38 (1999) 2238-2244.

[62] H.B. Mayes, L.J. Broadbelt, Unraveling the Reactions that Unravel Cellulose, The Journal of Physical Chemistry A, 116 (2012) 7098-7106.

[63] J. Scheirs, G. Camino, W. Tumiatti, Overview of water evolution during the thermal degradation of cellulose, European Polymer Journal, 37 (2001) 933-942.

[64] W. Chaiwat, I. Hasegawa, J. Kori, K. Mae, Examination of Degree of Cross-Linking for Cellulose Precursors Pretreated with Acid/Hot Water at Low Temperature, Industrial & Engineering Chemistry Research, 47 (2008) 5948-5956.

[65] W. Chaiwat, I. Hasegawa, T. Tani, K. Sunagawa, K. Mae, Analysis of Cross-Linking Behavior during Pyrolysis of Cellulose for Elucidating Reaction Pathway, Energy & Fuels, 23 (2009) 5765-5772.

191

[66] R.J. Moon, A. Martini, J. Nairn, J. Simonsen, J. Youngblood, Cellulose nanomaterials review: structure, properties and nanocomposites, Chemical Society Reviews, 40 (2011) 3941- 3994.

[67] F. Shafizadeh, Introduction to pyrolysis of biomass, Journal of Analytical and Applied Pyrolysis, 3 (1982) 283-305.

[68] J. Piskorz, D. Radlein, D.S. Scott, On the mechanism of the rapid pyrolysis of cellulose, Journal of Analytical and Applied Pyrolysis, 9 (1986) 121-137.

[69] D. Radlein, J. Piskorz, D.S. Scott, Fast pyrolysis of natural polysaccharides as a potential industrial process, Journal of Analytical and Applied Pyrolysis, 19 (1991) 41-63.

[70] R.J. Evans, T.A. Milne, Molecular characterization of the pyrolysis of biomass, Energy & Fuels, 1 (1987) 123-137.

[71] R. Vinu, L.J. Broadbelt, A mechanistic model of fast pyrolysis of glucose-based carbohydrates to predict bio-oil composition, Energy & Environmental Science, 5 (2012) 9808- 9826.

[72] F. Shafizadeh, T.T. Stevenson, Saccharification of douglas-fir wood by a combination of prehydrolysis and pyrolysis, Journal of Applied Polymer Science, 27 (1982) 4577-4585.

[73] O.P. Golova, Chemical Effects of Heat on Cellulose, Russian Chemical Reviews, 44 (1975) 687.

[74] V. Agarwal, P.J. Dauenhauer, G.W. Huber, S.M. Auerbach, Ab Initio Dynamics of Cellulose Pyrolysis: Nascent Decomposition Pathways at 327 and 600° C, Journal of the American Chemical Society, 134 (2012) 14958-14972.

[75] J. Lédé, Cellulose pyrolysis kinetics: An historical review on the existence and role of intermediate active cellulose, Journal of Analytical and Applied Pyrolysis, 94 (2012) 17-32.

[76] F. Shafizadeh, Y. Sekiguchi, Development of aromaticity in cellulosic chars, Carbon, 21 (1983) 511-516.

[77] T.E. McGrath, W.G. Chan, M.R. Hajaligol, Low temperature mechanism for the formation of polycyclic aromatic hydrocarbons from the pyrolysis of cellulose, Journal of Analytical and Applied Pyrolysis, 66 (2003) 51-70.

192

Chapter 6 Conclusions and Recommendations

6.1 General Conclusions

This dissertation describes several studies on the effect of cellulose crystallinity, pyrolysis temperature and the presence of additives on the thermal behavior of cellulose under fast and slow heating rate conditions. These studies advanced our understanding on several primary and secondary reactions that should be controlled to maximize the yield of transportation fuels and chemicals. Based on the results, the general conclusions are listed below:

1. Studies in Py-GC/MS and TGA proved that, the formation of liquid intermediate is an

important phenomenon in the pyrolysis of cellulose, and the crystallinity of cellulose can

greatly affect the its formation particularly under slow heating rates (Figure 6.1).

Amorphous condition, which was achieved by ball-milling crystalline cellulose, can

significantly decrease the thermal stability of cellulose, and thus enhance the formation of

this liquid intermediate under slow heating rates. The formation of liquid intermediate

can enhance the dehydration reactions at the price of anhydrosaccharides (majorly

levoglucosan), but not totally eliminate its production. Also, formation of liquid

intermediate doesn’t affect the fragmentation reactions. Studies conducted in the wire

mesh reactor proved that under fast heating rates at temperatures over 350 oC, both

amorphous and crystalline cellulose form liquid intermediates. The total reaction time of

amorphous cellulose is much faster than the crystalline cellulose.

193

Figure 6.1 Scheme of cellulose thermal reactions under slow pyrolysis condition

2. Levoglucosan is the main product of cellulose pyrolysis under high heating rates (Figure

6.2). A large yield (close to 20 mass %) of an unknown molecule with the same residence

time than 1,6-anhydroglucofuranose was also obtained. The maximum yield of

Levoglucosan under atmospheric pressure was obtained at 300 oC. Higher pyrolysis

temperatures result in the fragmentation and dehydration of levoglucosan into

uncountable gas, water and small C1-C4 condensable molecules. Pyrolysis tests at

atmospheric pressure with cellobiosan (also a primary product of cellulose pyrolysis)

show that this molecule contributes to the formation of a liquid intermediate but cannot

be evaporated. Secondary reactions of this molecule are the responsible for the formation

of the unknown molecule detectable by HPLC at the same retention time than AGF that

was observed during cellulose pyrolysis. The pyrolysis secondary reactions of

cellobiosan also resulted in the formation of small quantities of levoglucosan and a high

yield of carbonaceous residues

194

Levoglucosan

Evaporation

Crystalline Levoglucosan Cellulose Liquid Depolymerization Intermediate Oligosaccharides Amorphous Cross-linked (Cellobiosan, Cross-linking Polycondenstaion Char Cellulose Oligosaccharides Cellotriosan, …)

Pyrolysis Pyrolysis

Volatile Volatile

Figure 6.2 Scheme of cellulose thermal reactions under fast pyrolysis condition

3. Addition of 0.04 % of sulfuric acid after ash removing can enhance the yield of

levoglucosan from Douglas fir to a level similar to the maximum yield obtained from

cellulose at the same pyrolysis condition. The maximum yield of levoglucosan (on

cellulose basis) obtained at 300 oC was 60 wt.%. The addition of similar concentrations

of sulfuric acid to both amorphous and crystalline cellulose reduced the yield of

levoglucosan with a significantly higher yield of solid residue, the unknown compound at

the same residence time than 1,6-anhydroglucofuranose and other dehydration products

(levoglucosenone) (Figure 6.3). The effect of sulfuric acid addition can be explained by

the acceleration of levoglucosan dehydration and polymerization reactions. The addition

of sulfuric acid to cellobiosan (another primary product of cellulose pyrolysis), on the

other hand, enhances the yield of levoglucosan (from 5 to ~15 wt. %) and suppressed in

yield of char. The presence of sulfuric acid modifies cellobiosan during the impregnating

procedure and thus changes its thermal behavior. GC/MS analysis of cellobiosan samples

195

impregnated with sulfuric acid show that the hydrolysis of this compound can happen

even at the temperature of the GC inlet (200 oC) to form levoglucosan. The unknown

compound(s) with the same HPLC residence time than 1,6-anhydroglucofuranosewas

found to decrease by the presence of sulfuric acid in cellulose, and cellobiosan. Our

experimental results conclude that the increase in the yield of levoglucosan observed

when small quantities of sulfuric acid were added to Douglas fir could be explained by

the deactivation of the catalytic effect of the alkalines left in the samples after acid wash,

by its effect on the interactions between cellulose and the rest of the other organic

fractions in lignocellulosic material or by the hydrolysis of the heavy anhydrosugars

(cellobiosan, cellotriosan, etc.) produced during cellulose pyrolysis.

Dehydrated Products Levoglucosan (Levoglucosenone,…)

H2SO4 Catalyzed Evaporation Dehydration Dehydration

Crystalline Levoglucosan Cellulose H2SO4 Catalyzed Depolymerization Hydrolysis Oligosaccharides Amorphous Cross-linked (Cellobiosan, Cross-linking Polycondenstaion Char Cellulose Oligosaccharides Cellotriosan, …)

Pyrolysis H2SO4 Pyrolysis Pyrolysis Catalyzed Dehydration Dehydrated Volatile Volatile Products

Figure 6.3 Scheme of cellulose thermal reactions on the existence of sulfuric under fast pyrolysis condition

4. The thermal behavior amorphous and crystalline cellulose under low heating rate

conditions in the range of temperature between 260 and 400 oC were tested in a in spoon

196

reactor. Amorphous cellulose was found to have significant weight loss at 260 oC or

lower while crystalline cellulose started its obvious weight loss from 280 oC (with 30

minutes of heating including a temperature ramp from room temperature). Both cellulose

samples can form a non-hydrolysable sugar, which shows same pattern as cellulose in

NMR spectra but cannot give glucose by hydrolyzing. This portion is highly related to

liquid intermediate and acts as intermediate towards other products with further reaction

(cracking, dehydration, and polycondensation). Crystalline cellulose shows a lower

tendency to form C=O and C=C bonds when temperature is lower than 300 oC, but has

higher yield of these double bonds when temperature is higher than 300 oC, which may

indicate a higher tendency of dehydration reaction due to the crystalline structure

(hydrogen bond system). Carbonyl and furanyl carbons are more abundant in amorphous

cellulose residue. Aromatic and aliphatic carbons are also formed faster in amorphous

cellulose (until 300 oC). Considering the liquid state of crystalline cellulose can be found

at 300 oC (Chapter 3), these are evidence that liquid intermediate is a key role in the

dehydration and cross-linking reactions.

5. A new reaction mechanism that introduced for the first time the formation of cross-linked

cellulose as an intermediate product for the formation of carbonaceous residues was

proposed. This mechanism and the associated mathematical model explain very well the

reactions happening in solid phase during cellulose pyrolysis.

197

6.2 Recommendations

The results obtained in this dissertation lead us to make the following recommendations:

1. To conduct experimental studies to better understand the composition of liquid

intermediate. Liquid intermediate is the important stage and reaction pool during the

cellulose pyrolysis. Its condition can affect the cellulose, and even biomass pyrolysis

reactions and related outcome. A better understanding of its composition could help to

better control the outcome of the pyrolysis reactions.

2. To study the composition of the unknown compound(s) found during the mesh reactor

pyrolysis of cellulose and that seem to be derived from the cellobiosan secondary

reactions. The structure of this compound could be critical to understand the secondary

reactions of the heavy anhydrosugars (cellobiosan, cellotriosan, etc.) leading to the

formation of carbonaceous residues and to understand the reduction of the yield of

fermentable sugars during pyrolysis.

3. To conduct experiments with sulfuric acid on lignocellulosic materials with no alkaline to

better explain the mechanism responsible for the increase in the yield of levoglucosan

observed in our studies.

198

4. To study detailed effect of other acid additives on the production of levoglucosan from

lignocellulosic materials.

5. To perform experiments on the effect of acid additives on cellobiosan and other

anhydrosaccharides with DP < 5 derived from cellulose pyrolysis to confirm the capacity

of these acids to hydrolyze these sugars and enhance levoglucosan yield.

6. To explore the potential of other additives to mitigate secondary reactions responsible for

the formation of the unknown intermediate from cellobiosan.

7. To develop a novel mathematical model that include all the phenomena and compounds

found in this dissertation (formation of liquid intermediate, evaporation of small

molecules, ejection of oligomers, cross-linking/dehydration in liquid intermediate, and

formation of levoglucosan from cellobiosan).

8. Conduct experimental studies in fluidized bed and Auger pyrolysis reactor to confirm the increase in the yield of fermentable sugars when sulfuric acid is used as additive.

6.3 Scientific contributions

This dissertation makes the following scientific contributions:

199

a. This is the first systematic study on the effect of cellulose crystallinity on cellulose thermal

degradation reactions. It was found that cellulose crystallinity has an important effect on the

formation of liquid intermediates under slow heating rate conditions as well as on some of

the reactions happening during cellulose pyrolysis affecting product composition. b. For the first time we proved the importance of the heavy products of cellulose pyrolysis

reactions (chiefly cellobiosan) as an important intermediate for cellulose secondary reactions

at atmospheric pressure and fast heating rate conditions. Previous studies only stressed the

importance of levoglucosan. c. We proved that at atmospheric pressure and under fast heating rate conditions, char

production was not a result of levoglucosan dehydration and re-polymerization but a

consequence of poorly known reactions of un-evaporated sugar dimers (cellobiosan) and

oligomers (cellotriosan, cellotetrasan etc.). d. The effect of sulfuric acid additive on the formation of levoglucosan from acid wash

lignocellulosic materials was systematically studied. It was found that there is an optimal

concentration at which the yield of levoglucosan reaches a maximum, likely because of the

effect of the acid on the interactions between cellulose and the lignocellulosic matrix. The

passivation effect of sulfuric acid on the catalytic properties of alkalines and its effect on the

hydrolysis of cellobiosan and the other heavy anhydrosugars cannot be ruled out. e. The effect of crystallinity on cellulose solid/liquid state reaction was systematically

investigated. The evolution of cellulose solid residue composition as a function of pyrolysis

temperature and time was systematically studied for the first time. A new lumped reaction

200

mechanism that for the first time includes cross-linked cellulose as a critical reaction intermediate for the formation of solid residues during cellulose pyrolysis was proposed.

201

Appendix

202

Appendix A: Understanding Lignin-Cellulose Interactions during

the Pyrolysis of their Blends

(Paper to be submitted to The Journal of Analytical and Applied Pyrolysis)

Tim Hilbers1, 2, Zhouhong Wang2*, Brennan M. Pecha2, Roel J.M. Westerhof 1, Sascha R.A.

Kersten 1, Manuel Garcia-Perez2

1Thermo-Chemical Conversion of Biomass Group, Faculty of Science and Technology,

University of Twente, Postbus 217, 7500AE Enschede, The Netherland

2Biological Systems Engineering, Washington State University, Pullman, WA 99164, USA

Abstract: Interactions between lignin and cellulose, the main components of lignocellulosic materials, were studied by thermogravimetric analysis (TGA), Scanning Electron Microscope

(SEM) and Pyrolysis-Gas Chromatography/Mass Spectroscopy (Py-GC/MS). The thermal behavior of a control cellulose (Avicel), a ball-milled cellulose (less crystalline), and their blends with organosolv lignin were compared. The DTG results revealed that the presence of lignin modifies the thermal behavior of crystalline cellulose, but its effect on amorphous cellulose was more modest. No effect was observed on the yield of bio-char produced. Cellulose - lignin interactions resulted in changes in the yields of some compounds detected by Py-GC/MS. The presence of lignin enhanced the yield of levoglucosan but decreased the yield their dehydration products (example levoglucosenone, 5-HMF, Furfural, MEF). Cyclopentens were enhanced at

350 oC, but mostly kept within expectation at 500 oC. Lignin favored the formation of cyclopentanedione. Some lignin derived compounds (e.g. 2-methoxyphenol, 2-methoxy-4-

203

methylphenol, 3-methoxy-1,2-debenzediol, 2-methoxy-4-vinylphenol, 2,6-dimethoxyphenol and

3,5-dimethoxyacetophenone) were enhanced at 350 oC by cellulose and slightly depressed at 500 oC. 1,2,3-trimethoxy-5-methylbenzene, 4-hydroxy-3,5-dimethoxybenzaldehyde, and 4-ethyl-2- methoxyphenol were not affected by the cellulose presence. Products from cellulose fragmentation reactions (hydroxyl-acetaldehyde and acetol) were enhanced by the presence of lignin.

Key words: Levoglucosan, cellobiosan, cellulose pyrolysis

Corresponding Author:

Zhouhong Wang

E-mail: [email protected]

204

A.1 Introduction

Fast pyrolysis of lignocellulosic materials has become increasingly attractive due to its sustainability and high potential pathway for transportation fuel production [1]. The main bio- polymers forming these materials are: cellulose, hemicelluloses and lignin. Fast pyrolysis of lignocellulosic materials typically resulting in crude bio-oil yields as high as 75 wt. % [2, 3].

Cellulose pyrolysis usually gives tar containing anhydrosugars (levoglucosan, cellobiosan, etc.), small fragmentation products (glycolaldehyde, acetol, etc.), furanic compounds (furfural, hydroxymethylfurfural, etc.). The distribution of these compounds is highly depending on reaction conditions and the composition of the lignocellulosic material processed [4]. It is known that small amount of alkaline metal found in biomass can significantly reduce the levoglucosan yield and enhance fragmentation reactions [5-8].

Several authors [9, 10] have also observed that the yield of levouglocosan (the main product of cellulose depolymerization reactions) obtained when acid washed lignocellulosic materials are pyrolysed is much lower than the yield obtained when cellulose is pyrolysed alone. This reduction in levoglucosan production could be due to very poorly interactions between cellulose and the other components of lignocellulosic materials. It is imaginable that during pyrolysis, the liquid intermediates of cellulose and the other biomass constituents (hemicelluloses, lignin) could interact with each other and induce secondary reactions. The interactions could also happen during the primary reactions and may be associated to the exchange of hydrogen or free

205

radicals between cellulose and lignin. Some authors [13, 14] have found holocellulose acted as H donor during lignin pyrolysis [11].

Hosoya et al. [12, 13] found that the presence of lignin enhanced levoglucosan production at gasification temperature (800 oC) [12].The author explained this result by the potential protection of levoglucosan in the presence of aromatic compounds. Golova at al [14] conducted pyrolysis experiments with cellulose in the antioxidant agents (di-2’-naphthylphenylene-diamine) and found that the presence of these compounds dramatically decreased the yield of levoglucosan from 34 to 4.4 %. Based on their experimental results the authors proposed a cellulose free radical depolymerization mechanism that can be affected by the presence of antioxidants. Lignin is an excellent antioxidant and as such could have an effect on cellulose depolymerization similar to the one found by Golova [14].

Shafizadeh et al [9, 15, 16] observed that the addition of sulfuric acid to acid wash lignocellulosic materials could double levoglucosan yields in materials. The same experiments conducted with cellulose resulted in reductions in levoglucosan yields. The authors [9, 15] suggested that the effect observed could be in fact due to effect of sulfuric acid on cellulose- lignin interactions. Interactions with hemicelluloses cannot be ruled out. Zhou et al. [9] and

Wang et al [17] conducted studies at different sulfuric acid concentrations on acid wash Douglas fir and found the presence of an optimal concentration at which the yield of this anhydrosugars reached a maximum. Other authors [18] argue that the effect observed could be due to the effect

206

of the acid on cellulose-lignin interactions, due to its passivation of alkaline catalytic properties or due to the hydrolysis of cellulose primary pyrolysis oligomeric products (chiefly cellobiosan).

The main purpose of this paper is to understand cellulose-lignin interaction during pyrolysis and the potential effect of cellulose crystallinity on these interactions. The effect of cellulose-lignin blends on product distributions is studied.

A.2 Materials and Methods

A.2.1 Materials

The Cellulose samples used has been described elsewhere [17, 19, 20]. Briefly, the crystalline cellulose used in powder was obtained from Sigma-Aldrich (11365 Fluka, Avicel® PH-101, ~50

μm particle size [CAS 9004-34-6]). It has a crystallinity of approximately 60%. Amorphous cellulose (herein call ball-milled cellulose) was prepared by ball milling for 24 hours the Avicel and has a crystallinity of approximately 6%. The organosolv lignin used in this study was obtained from Sigma-Aldrich (371017 Aldrich, organosolv lignin [CAS 8068-03-9]).

A.2.2 Sample Preparation

The studied lignin-cellulose blends were prepared by weighing different amounts of organosolv lignin, crystalline cellulose and amorphous cellulose. Blends with 20, 50 and 80% lignin were

207

prepared. The samples were prepared in 15 mL plastic centrifuge tubes with a mechanic shaker and stored overnight in a freezer to make physical blending easier.

A.2.3 Methods

Thermogravimetric Analysis (TGA)

The thermal degradation behavior of the blends studied was studied with a thermogravimetric analyzer (Mettler Toledo TGA/SDTA851e, In/Al calibrated, ceramic crucibles without lid, nitrogen purge gas). The samples were heated between 25 and 600 oC at heating rates of

10 °C/min and 50 °C/min. Approximately 9 mg of samples were used in each experiment.

Analysis of the data was carried out by exporting the sample temperature, mass percentage and time values to MS excel. Graphic results show the behavior of thermal degradation in the weight percentage as function of temperature. For the 50 °C/min results, the derivative weight percentage was calculated with the evaluation software from Mettler Toledo (1st derivative with

13 points). For the 10 °C/min data the mass percentage was first set to 100% at 150 °C to discard the presence of moisture in the samples by dividing all mass values by the value at 150 °C. The derivative was then calculated by dividing the difference between two data weight loss data points by their difference in time. The temperature of this derivative data point was set to the average temperature of the two points. The predictions of the thermal degradation behavior of the blends were calculated on mass basis by averaging the values of pure compound analysis. For

208

example the 80% crystalline cellulose with organosolv lignin: 80% of the water corrected mass % value of pure crystalline cellulose at a certain time was added up by 20% of the water corrected mass % value of pure organosolv lignin at the same time. The sample temperature can differ at the same data point (time), so the average temperature of the two values was also calculated on mass basis.

Char yield as percentage of the initial dry mass was also calculated to study interactions. The value was calculated by dividing two weight percentages: 1) the weight percentage when a temperature of 550 °C was reached for the first time by 2) the weight percentage at 150 °C to eliminate the presence of water. Char yield was plotted in two separated graphs for heating rates of 50 and 10 °C/min.

Scanning Electron Microscope (SEM)

Solid residues were obtained from the thermo-gravimetric analyzer was carefully removed from the crucibles and collected in the form of a disk. SEM images of the char were made with a FEI

200F SEM system with Large Field Detector and a low vacuum environment (130 Pa). High voltage was set at 20kV and the magnification was set from 50x to 5000x.

209

Pyrolysis-Gas Chromatography/Mass Spectrometry (Py-GC/MS)

The volatile products obtained under fast heating rate conditions were analyzed by pyrolysis-gas chromatography/mass spectrometry (Py-GC/MS). The apparatus (Agilent technologies 6890 N

Network GC system and an Inert XL MSD 5975B with EI ionization) was equipped with a pyroprobe (CDS pyroprobe 5000 series with a platinum coil probe and a 1500 valved interface) for manual sample injection. The GC/MS has an HP-5MS capillary column (30m x 0.250 mm,

0.25 μm). An amount of approximately 0.6 mg of sample was loaded in a cleaned quartz tube (2 mm × 25 mm), whereas the amount was weighed with the thermogravimetric analyzer using the weigh-function. The pyro-probe interface and GC liner were kept at 270 °C. After placing the pyro-probe with the quartz tube in the valved interface, the tube was held at 270 °C for 1 minute and at 350 or 500 °C for 1 minute, reached with a heating rate of 100 °C/s. The temperature in the probe was calibrated with an Omega type K thermal couple. The pyrolysis products from the sample were transferred to the GC/MS for separation at 40 °C with 6 °C/min to 280 °C and kept at that temperature for 5 minutes.

Peak studies were performed by analyzing as many peaks as possible derived from organosolv lignin and cellulose. Lignin and amorphous and crystalline cellulose were first analyzed at

500 °C and another sample at a temperature of 350 °C was used as control sample, to check if the same peaks were obtained. Every peak was analyzed based on the most abundant ion and a specific retention time. The peak area was always included without using a baseline (the whole area below a peak was included). The peak area (abundance) was divided by sample loading

210

(mass) and three or more samples were used to calculate an average value with an error based on the 95% confidence interval. Finally, the specific peak area (i.e. abundance/sample loading) was plotted against the organosolv lignin percentage for every compound. Crystalline and amorphous cellulose were plotted in the same graph, while separated graphs were made for the two different temperatures. In every graph a dashed line is plotted, which represents the predicted values of abundance per sample loading at blend concentrations.

A.3 Results and Discussions

A.3.1 TGA

Figure A.1 shows the TG and DTG of crystalline and amorphous cellulose with blending of organosolv lignin at 0, 20, 50, 80, 100 wt. % at heating rate of 10 oC/min. Comparing to amorphous cellulose, the thermal behavior of crystalline cellulose was more affected by the presence of lignin . The predicted DTG curves of crystalline cellulose-lignin blends shifted to higher temperature region while the influence on amorphous cellulose was much more limited.

Similar situation was also found with heating rate of 50 oC/min (not shown). As shown previously [19], crystalline cellulose does not melt in TGA tests, while amorphous cellulose did..

211

Figure A.2 shows pictures of TGA residues. It can be observed the presence of a liquid intermediate with bubbles from pure organosolv lignin. The solid residue looks like a foam.

Bubbling seems to be enhanced at higher heating rates (50 °C/min). Images of pure cellulose show a total different behavior. At a lower heating rate both crystalline and amorphous cellulose

212

volatile products seems to be released from the surface of the particle. The image of pure amorphous cellulose at a heating rate of 10 °C/min shows intermediate less intense bubbling and swelling. In short, crystalline cellulose degrades via a solid intermediate; amorphous cellulose and organosolv lignin via a liquid intermediate. The intensity of bubbling seems to be controlled by the heating rate.

At a composition of 80 wt. % organosolv lignin with cellulose the behavior of lignin seems to control the overall behavior of the blend. At a higher heating rate the boiling and melting effect seems to be enhanced. In the equal mass blends char formation is showing divergent behavior.

The behavior of blends with lower lignin content (20 wt. %) is similar to the behavior of cellulose,

Figure A.3 shows the char yield as a function of lignin content in the blend. The result shows under slow heating rate, the char yield simply followed a linear prediction and had no significant effect with the amount of lignin blended (Figure A.3).

213

CC (50 °C/min) AC (50 °C/min) CC (10 °C/min) AC (10 °C/min) Pure lignin

80% OL

50% OL

20% OL

10% OL

Pure cellulose

Figure A.2 Images of residue obtained from samples TGA analysis.

214

Figure A.3 Char yield of cellulose-lignin blend with their predicted yield at heating rate of 50 and 10 oC/min in TGA.

A.3.2 SEM

Figure A.4 shows the SEM images of residues from crystalline cellulose, amorphous cellulose and organosolv lignin TGA. Residue from crystalline cellulose was consisted of small fibrous char, whereas the amorphous cellulose and organosolv lignin showed the formation of a liquid intermediate. Amorphous cellulose had much more small bubbles than lignin. Lignin has a more layered structure instead of uniformly distributed bubbles. Small solids could be found in residue of amorphous cellulose.

Figure A.5 shows the cellulose samples blended with ~20 wt. % of organosolv lignin. Although the weight loss pattern was significantly influenced by lignin, the crystalline cellulose blend still had a residue structure similar to pure crystalline cellulose with crystallites started to be observed.

215

The fibril structure likely provided a bigger surface area for the lignin liquid intermediate to evaporate, while the lignin liquid intermediate acted as glue connecting fibers. Residue structure of amorphous became complex. Less small bubbles were formed with a much thicker residue layer. The amorphous cellulose and lignin tend to form a uniform liquid substance and “boil” together. Comparing to the pure amorphous cellulose much less debris particles could be observed in this residue.

Figure A.6 shows the cellulose samples blended with ~50 wt. % of organosolv lignin. The residue from the crystalline cellulose blend shows a relatively fibrous look structure with the main bone switched to a once melted liquid intermediate matrix. Enlarged image (1000X) showed that the cellulose fibers were totally “dissolved” and only crystallites were left. The amorphous cellulose blend exhibited a residue structure pronouncing more of organosolv lignin residue like structure. On the pores of the bubbles, thin films appeared, indicating higher surface tension strength was achieved from this blend. Crystallites could hardly be found from amorphous cellulose blend. Considering the conditions from crystalline cellulose, we suppose that the small crystallites already “dissolved” in lignin and formed that steady and viscous bubble film which was even preserved after TGA.

Figure A.7 shows the cellulose samples blended with ~80 wt. % of organosolv lignin. With this amount of lignin, both crystalline and amorphous cellulose had similar residue structure and the strong bubble film found in amorphous cellulose 50 wt. % blends could again be observed in this experiment. Crystallites could hardly be observed in both samples. These phenomena confirmed

216

our hypothesis that cellulose could actually “dissolve” in lignin liquid intermediate and form sticker and stronger liquid mixture. Hosoya et al. found levoglucosan could easily dissolve in environment with (poly-) benzene and phenolic compounds normally found in lignin pyrolysis, which prevented the thermal motion of levoglucosan [21]. It’s easy to imagine similar situation can be extended to cellulose. In this experiment, with the existence of lignin, cellulose fibers rapidly “dissolved”, lignin products might take the place of hydrogen bonds holding cellulose fibril structure and bind cellulose/oligosugar/levoglucosan with more motion limiting lignin products. This corrosion like behavior is similar to the particle shrinking model, or peeling effect

[22]. On the other hand, cellulose products stabled the existence of catechols from lignin which might have increased the viscosity of lignin liquid intermediate and thus left the unburst bubbles under slow pyrolysis.

217

Figure A.4 SEM of residues from original compounds (from left to right: crystalline cellulose, amorphous cellulose, and organosolv lignin)

218

Figure A.5 SEM images of 20% organosolv lignin (OL) with 80% crystalline cellulose (CC, left) and amorphous cellulose (AC, right)

219

Figure A.6 SEM images of 50% organosolv lignin (OL) with 50% crystalline cellulose (CC, left) and amorphous cellulose (AC, right)

220

Figure A.7 SEM images of 80% organosolv lignin (OL) with 20% crystalline cellulose (CC, left) and amorphous cellulose (AC, right)

221

A.3.3 Py-GC/MS

Figure A.8 shows a Py-GC/MS chromatography of control cellulose/ organosolv lignin blend at

50/50 wt. %. The peaks were identified with NIST 08 library. Typical lignin derived compounds were marked red while blue color marked the compounds from cellulose pyrolysis. Table A.1 gives the names of all the compounds identified (certain peaks with different retention time were identified as same compounds). Based on their origin and behavior, these products are discussed separately.

Figure A.8 Py-GC/MS of crystalline cellulose/organosolv lignin blend ( 50:50 wt. %) at 500 oC. Compounds were identified by NIST 08 library (See Table A.1)

222

Table A.1 Compounds identified in Py-GC/MS with their major ion and retention time Peak # Compound Retention Time Ion Formula (min.) (m/z)

1 Carbon dioxide 1.551 44 CO2

2 2-methyl-Furan 2.197 82 C5H6O

3 2,5-dimethylfuran 3.086 96 C6H8O

4 Vinylfuran 3.308 94 C6H6O

5 Phenol 8.968 94 C6H6O

6 2-methoxy-phenol 11.349 109 C7H8O2

7 2-methoxy-4-methyl-phenol 13.893 138 C8H10O2

8 3-methoxy-1,2,-dibenzediol 15.611 140 C7H8O3

9 4-ethyl-2-methoxy-phenol 15.897 137 C9H12O2

10 2-methoxy-4-vinylphenol 16.711 135 C9H10O2

11 2,6-dimethoxy-phenol 17.609 154 C8H10O3

12 3,4-dimethoxy-phenol 17.817 154 C8H10O3

13 1,2,4-trimethoxybenzene 19.667 168 C9H12O3

14 1,2,3-trimethoxy-5-methylbenzene 21.220 182 C10H14O3

15 3,5-dimethoxyacetophenone 22.022 165 C11H17O2

16 2,6-dimethoxy-4-(2-propenyl)-phenol 22.714 194 C11H14O3

17 2,6-dimethoxy-4-(2-propenyl)-phenol 23.637 194 C11H14O3

18 4-hydroxy-3,5-dimethoxy-benzaldehyde 23.942 181 C9H10O3

19 2,6-dimethoxy-4-(2-propenyl)-phenol 24.586 194 C11H14O3

20 1-(4-hydroxy-3,5-dimethoxyphenyl)-ethanone 25.246 181 C10H12O4

21 Desaspidinol 25.890 167 C9H10O4

22 1-(4-hydroxy-3,5-dimethoxyphenyl)-ethanone 26.803 181 C10H12O4

23 3,5-Dimethoxy-4-hydroxycinnamaldehyde 29.454 165 C11H12O4

24 Furanone 4.544 55 C4H4O2

25 Furfural 5.235 96 C5H4O2

26 2-propyl furan 5.923 81 C7H10O

27 dihydro-4-hydroxy-furanone 6.109 44 C4H6O3

28 2-hydroxy-2-Cyclopenten-1-one 7.381 98 C5H6O2

29 5-methyl-2-furancarboxaldehyde 8.281 109 C6H6O2 30 Unknown compound 1 (furanone-like) 9.171 114 -

31 2-hydroxy-3-methyl-2cyclopenten-1-one 9.637 112 C6H8O2

32 methyl-cyclopentanedione 9.943 112 C6H8O2

223

33 2,5-dimethyl-4-hydroxy-3-furanone 11.341 43 C6H8O3

34 Levoglucosenone 12.036 98 C6H6O3 35 Unknown compound 2 (sugar-like) 13.927 41 -

36 1,4:3,6-Dianhydro-à-d-glucopyranose 14.573 69 C6H8O4

37 5-hydroxymethyl-2-furancarboxaldehyde 15.165 97 C6H6O3 38 Unknown compound 3 (sugar-like) 17.000 87 -

39 1,6-Anhydro-β-D-glucopyranose (levoglucosan) 21-23 57 C6H10O5

40 1,6-Anhydro-α-d-glucofuranose 24.439 73 C6H10O5

41 Acetone 1.659 58 C3H6O

42 2-oxo-propanoic acid 2.072 68 C3H4O3

43 hydroxy-acetaldehyde 2.067 32 C2H4O2

44 Acetol 2.691 43 C3H6O2

Anhydrosugars and their dehydration products

Figure A.9 shows the effect of organosolv lignin on the yield of levoglucosan (LVG) and 1, 6- anhydro-α-D-glucofuranose (AGF) from crystalline cellulose and amorphous cellulose. Both anhydrosugars were significantly enhanced in yield. The increase of these two sugars can be concluded to the protection from lignin to further degradation [12, 13, 21] or because of enhancement of thermal ejection that the presence of lignin liquid intermediate could cause.

Lower temperature (350 oC) gave little more sugar.

224

Figure A.9 Effect of lignin blend on yield of levoglucosan (LVG) and 1, 6-anhydro-α-D- glucofuranose (AGF) from crystalline cellulose (CC) and amorphous cellulose (AC)

Figure A.10 shows levoglucosenone and 1,4;3,6-dianhydro-α-D-glucopyranose yield with existence of organosolv lignin. The yield of both compounds was significantly reduced. These are products of dehydration reactions. This reaction may be inhibited by the faster lease of cellulose primary products (cellulose, cellobiosan) by the help of the thermal ejection caused by w lignin liquid intermediate.

225

Figure A.10 Effect of lignin content on the yields of levoglucosenone and 1,4;3,6-dianhydro-α- D-glucopyranose at 350 and 500 oC

Furanic Compounds and cyclopentens

Figure A.11 shows the yield of 5-HMF, furfural and MEF under the effect of lignin blend at 350 oC and 500 oC. It is interesting to see that, lignin, temperature and crystallinity all had effect on their yield. Lignin enhanced the yield of 5-HMF in both cellulose and inhibited the yield of furfural in amorphous cellulose. Cellulose crystallinity limited the effect of lignin on furfural production but had no influence on the enhancement of 5-HMF. According to literatures, formation of furfural in cellulose is majorly from two pathways during pyrolysis: 1. Pyronose ring rearrangement on chain and then through 5-HMF intermediate [23, 24]; 2. Levoglucosan pyranose ring opening [25]. Lignin liquid environment in this case didn’t prohibit the

226

rearrangement of pyranose ring to form furanose ring and its dehydration to the yield of 5-HMF, which followed the trend of AGF.

Figure A.11 Effect of lignin blend on yield of 5-hydroxymethyl-2-furancarboxyaldehyde (5- HMF), furfural and 5-methylfurfural (MEF) at 350 and 500 oC

Figure A.12 shows the effect of lignin blend on yield of 2-furanone and 2-propylfuran at 350 and

500 oC. These two compounds were both reduced in yield with the presence of lignin, and both

227

pyrolysis temperature and cellulose crystallinity didn’t change its behavior. This result suggest that the reactions responsible for the formation of these compounds(which typically happen in the liquid intermediate) may be mitigated by the faster removal of levoglucosan and other cellulose primary products by thermal ejection enhanced by the presence of lignin.

Figure A.12 Effect of lignin blend on yield of 2-furanone and 2-propylfuran at 350 and 500 oC

Figure A.13 shows the effect of lignin on cyclopentanedione related compounds. 2-hydroxy-2- cyclopenten-1-one and methyl-cyclopentanedione follow a similar behaviour at both temperatures (350 and 500 oC). Here with the presence of lignin, amorphous cellulose obviously yielded more of these two compounds than crystalline cellulose at 350 oC. 2-hydroxy-3-methyl-

2-cyclopenten-1-one as the isomer of methyl-cyclopentanedione was found decreased with the presence of lignin.

228

Figure A.13 Effect of lignin blend on yield of 2-hydroxy-2-cyclopenten-1-one, methyl- cyclopentanedione, and 2-hydroxy-3-methyl-2-cyclopenten-1-one at 350 and 500 oC

Phenolic compounds from organosolv lignin

Figure A.14, Figure A.15, Figure A.16 and Figure A.17 show the changes in the yield of lignin derived products as a function of lignin content in a blend with amorphous and crystalline

229

cellulose. Figure A.14, Figure A.15 and Figure A.16 show compounds observed enhancement in yield with the existence of cellulose. Most of this increase was happening at temperature of 350 oC, while at 500 oC slight suppressing could be observed. As described by Hosoya et al. [12, 13] lignin monomers can be protected by levoglucosan in vapor phase in the form of catechol. But here obviously, this effect didn’t happen during the Py-GC/MS operation, but more likely happened during the not evaporated liquid state since a higher temperature didn’t enhance the yield of lignin monomers. Another interesting factor was that, this effect favored monomers with unsaturated substitutions. Figure A.17 shows the compounds significantly suppressed in yield with cellulose existence. The reason of this depression is still unknown.

230

Figure A.14 Effect of cellulose blend on yield of phenol, 2-methoxyphenol, 2-methoxy-4- methylphenol, and 3-methoxy-1,2-dibenzediol at 350 and 500 oC

231

Figure A.15 Effect of cellulose blend on yield of 2-methoxy-4-vinylphenol, 2,6-dimethoxy- methylphenol, 3,4-dimethoxyphenol and 3,5-demethoxyacetophenone at 350 and 500 oC

232

Figure A.16 Effect of cellulose blend on yield of another four compounds from lignin at 350 and 500 oC

233

Figure A.17 Compounds not enhanced by cellulose existence.

Light Compounds

Figure A.18 and Figure A.19 shows the yield of light compounds products of fragmentation reactions as a function of the content of lignin. Most light compounds (with 1-3 carbons) could only be found at 500 oC. The compounds shown in Figure A.18 are produced mostly from cellulose. These light compounds were found not to be influenced by cellulose crystallinity [19].

234

The yield of hydroxyacetaldehyde and acetol was enhanced by the presence of lignin. The yield of 2-oxo-propanoic acid decreased as lignin content in the blend increased. It is known that hydroxyl-acetaldehyde and acetol are products from fragmentation reactions of levoglucosan and also perhaps of the heavy anhydrosugars produced from cellulose pyrolysis. The 2-oxo- propanoic acid is not a product of sugar fragmentation reactions; it is related to char formation reactions.

Figure A.18 Effect of lignin blend on the yield of 2-oxo-propanoic acid, hydroxyl-acetaldehyde, and acetol at 500 oC

The yield of carbon dioxide and acetone as a function of lignin content are plotted in Figure

A.19. Cellulose-lignin blend had very limited influence on these two compounds except certain ratio (20 wt. % of lignin for carbon dioxide, and 80 wt. % of lignin for acetone).

235

Figure A.19 Effect of lignin-cellulose interaction on yield of carbon dioxide and acetone.

A.4 Conclusions

Crystalline and amorphous cellulose were blended with organosolv lignin at 0 (pure cellulose),

20, 50, 80, 100 (pure lignin) wt. %. TGA, SEM, and Py-GC/MS were performed to study the interaction between cellulose and lignin. TGA test showed that lignin had a significant effect on crystalline cellulose thermal behavior by shifting the weight loss temperature to a higher temperature region, but had a limited effect on amorphous cellulose. Char yield was not affected by cellulose-lignin interactions. Py-GC/MS results suggest that lignin presence enhanced the production of anhydrosugars (levoglucosan) but depressed the yield their dehydration products, which were probably related with the enhancement of the thermal ejection of these sugars and the consequent mitigation of anhydrosugars dehydration reactions. 5-HMF production in the

236

presence of lignin was enhanced at 350 oC in both cellulose samples, while at 500 oC, this effect was not observed. Furfural and MEF were depressed in amorphous cellulose. Lignin favored the formation of cyclopentanedione. Lignin derived compounds were mostly enhanced at 350 oC by cellulose and slightly depressed at 500 oC, though 1,2,3-trimethoxy-5-methylbenzene, 4- hydroxy-3,5-dimethoxybenzaldehyde, and 4-ethyl-2-methoxyphenol were not really affected by existence of cellulose.

Acknowledgement

This project was financially supported by the US National Science Foundation (CBET-0966419,

CAREER CBET-1150430), the Sun-Grant Initiative (Interagency Agreement: T0013G-A), and the Washington State Agricultural Research Center. The authors are very thankful for their support.

237

Reference

[1] X. Miao, Q. Wu, C. Yang, Fast pyrolysis of microalgae to produce renewable fuels, Journal of Analytical and Applied Pyrolysis, 71 (2004) 855-863.

[2] S. Kersten, M. Garcia-Perez, Recent developments in fast pyrolysis of ligno-cellulosic materials, Current Opinion in Biotechnology, 24 (2013) 414-420.

[3] D. Mohan, C.U. Pittman, P.H. Steele, Pyrolysis of Wood/Biomass for Bio-oil: A Critical Review, Energy & Fuels, 20 (2006) 848-889.

[4] I. Pastorova, R.E. Botto, P.W. Arisz, J.J. Boon, Cellulose char structure: a combined analytical Py-GC-MS, FTIR, and NMR study, Carbohydrate Research, 262 (1994) 27-47.

[5] D.J. Nowakowski, J.M. Jones, Uncatalysed and potassium-catalysed pyrolysis of the cell- wall constituents of biomass and their model compounds, Journal of Analytical and Applied Pyrolysis, 83 (2008) 12-25.

[6] H. Kawamoto, D. Yamamoto, S. Saka, Influence of neutral inorganic chlorides on primary and secondary char formation from cellulose, Journal of Wood Science, 54 (2008) 242-246.

[7] N. Shimada, H. Kawamoto, S. Saka, Different action of alkali/alkaline earth metal chlorides on cellulose pyrolysis, Journal of Analytical and Applied Pyrolysis, 81 (2008) 80-87.

[8] C. Di Blasi, A. Galgano, C. Branca, Influences of the Chemical State of Alkaline Compounds and the Nature of Alkali Metal on Wood Pyrolysis, Industrial & Engineering Chemistry Research, 48 (2009) 3359-3369.

[9] S. Zhou, D. Mourant, C. Lievens, Y. Wang, C.-Z. Li, M. Garcia-Perez, Effect of sulfuric acid concentration on the yield and properties of the bio-oils obtained from the auger and fast pyrolysis of Douglas Fir, Fuel, 104 (2013) 536-546.

[10] S. Zhou, N.B. Osman, H. Li, A.G. McDonald, D. Mourant, C.-Z. Li, M. Garcia-Perez, Effect of sulfuric acid addition on the yield and composition of lignin derived oligomers obtained by the auger and fast pyrolysis of Douglas-fir wood, Fuel, 103 (2013) 512-523.

[11] S. Omori, M. Aoyama, A. Sakakibara, Hydrolysis of Lignin with Dioxane-Water XIX. Reaction of β-O-4 Lignin Model Compounds in the Presence of Carbohydrates, Holzforschung, 52 (1998) 391-397.

[12] T. Hosoya, H. Kawamoto, S. Saka, Cellulose–hemicellulose and cellulose–lignin interactions in wood pyrolysis at gasification temperature, Journal of Analytical and Applied Pyrolysis, 80 (2007) 118-125.

238

[13] T. Hosoya, H. Kawamoto, S. Saka, Solid/liquid- and vapor-phase interactions between cellulose- and lignin-derived pyrolysis products, Journal of Analytical and Applied Pyrolysis, 85 (2009) 237-246.

[14] O.P. Golova, Chemical Effects of Heat on Cellulose, Russian Chemical Reviews, 44 (1975) 687.

[15] F. Shafizadeh, T.T. Stevenson, Saccharification of douglas-fir wood by a combination of prehydrolysis and pyrolysis, Journal of Applied Polymer Science, 27 (1982) 4577-4585.

[16] J. Piskorz, D.S.A.G. Radlein, D.S. Scott, S. Czernik, Pretreatment of wood and cellulose for production of sugars by fast pyrolysis, Journal of Analytical and Applied Pyrolysis, 16 (1989) 127-142.

[17] Z. Wang, B. Pecha, R.J.M. Westerhof, S.R.A. Kersten, M. Garcia-Perez, Effect of Pyrolysis Temperature on the Formation of Anhydrosugars During the Fast Pyrolysis of Cellulose in a Wire Mesh Reactor at Atmospheric Pressure, (In Preparation).

[18] P.R. Patwardhan, J.A. Satrio, R.C. Brown, B.H. Shanks, Influence of inorganic salts on the primary pyrolysis products of cellulose, Bioresource Technology, 101 (2010) 4646-4655.

[20] Z. Wang, B. Pecha, R.J.M. Westerhof, S.R.A. Kersten, C.-Z. Li, A.G. McDonald, M. Garcia-Perez, Effect of Cellulose Crystallinity on Solid/Liquid Phase Reactions Responsible for the Formation of Carbonaceous Residues during Slow Pyrolysis, (Submitted to Industrial & Engineering Chemistry Research).

[21] T. Hosoya, H. Kawamoto, S. Saka, Thermal stabilization of levoglucosan in aromatic substances, Carbohydrate Research, 341 (2006) 2293-2297.

[22] G.A. Zickler, W. Wagermaier, S.S. Funari, M. Burghammer, O. Paris, In situ X-ray diffraction investigation of thermal decomposition of wood cellulose, Journal of Analytical and Applied Pyrolysis, 80 (2007) 134-140.

[23] J. Scheirs, G. Camino, M. Avidano, W. Tumiatti, Origin of furanic compounds in thermal degradation of cellulosic insulating paper, Journal of Applied Polymer Science, 69 (1998) 2541- 2547.

[24] M.J. Antal Jr, T. Leesomboon, W.S. Mok, G.N. Richards, Mechanism of formation of 2- furaldehyde from d-xylose, Carbohydrate Research, 217 (1991) 71-85.

239

[25] F. Shafizadeh, Y.Z. Lai, Thermal degradation of 1,6-anhydro-.beta.-D-glucopyranose, The Journal of Organic Chemistry, 37 (1972) 278-284.

240